Structure and stability of phosphate-methylated DNA duplexes: model systems for specific DNA-protein interaction andconformational transmissionCitation for published version (APA):Genderen, van, M. H. P. (1989). Structure and stability of phosphate-methylated DNA duplexes : model systemsfor specific DNA-protein interaction and conformational transmission. Eindhoven: Technische UniversiteitEindhoven. https://doi.org/10.6100/IR300079

DOI:10.6100/IR300079

Document status and date:Published: 01/01/1989

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:openaccess@tue.nlproviding details and we will investigate your claim.

Download date: 27. Jun. 2020

STRUCTURE AND STABILITY OF PHOSPHATE-METHYLATED DNA DUPLEXES

MODEL SYSTEMS FOR SPECIFIC DNA- PROTEIN INTERACTION

AND CONFORMATIONAL TRANSMISSION

M.H.P. VAN GENDEREN

STRUCTURE AND STABILITY OF

PHOSPHATE-METHYLA TED DNA DUPLEXES

MODEL SYSTEMS FOR SPECIFIC DNA-PROTEIN INTERACTION

AND CONFORMATIONAL TRANSMISSION

STRUCTURE AND STABILITY OF PHOSPHA TE-METHYLA TED DNA DUPLEX.ES

MODEL SYSTEMS FOR SPECIFIC DNA-PROTEIN INTERACTION

AND CONFORMATIONAL TRANSMISSION

PROEFSCHRIFT

TER VERKRIJGING VAN DE GRAAD VAN DOCTOR AAN DE TECHNISCHE UNIVERSITEIT EINDHOVEN. OP GEZAG VAN DE RECTOR MAGNIFICUS. PROF. IR. M. TELS. VOOR EEN COMMISSIE AANGEWEZEN DOOR HET COLLEGE VAN DEKANEN IN HET OPEN­BAAR TE VERDEDIGEN OP VRIJDAG 24 FEBRUARI 1989 TE 16.00 UUR

door

MARCEL HENRI PAUL VAN GENDEREN

geboren te 6ndhoven

DIT PROEFSCHRIFT IS GOEDGEKEURD DOOR

DE PROMOTOREN

PROF. DR. H.M. BUCK

EN

PROF. DR. G. CHALLA.

Chapter 1

lntroduction

Chapter 2

CONTENTS

A parallel right-handed duplex of the hexamer d(TpTpTpTpTpT) with

phosphate triester linkages

Chapter 3 Molecular mechanics studies of parallel and antiparallel phosphate­

methylated DNA

Chapter 4 Parallel phosphate-methylated mini-duplexes for Sp- and Rp-d(TpT).

and exdusively for Sp-d{CpC) and Sp-d(TpC). lmplications for

stereospecific complexation of poly-L-lysine and poly-L-ornithine with natura! d(T10) and d(C10}. inducing parallel duplexes

Chapter 5

Hybridization of phosphate-methylated DNA and natural oligonucleo­tides. lmplications for protein-induced DNA duplex destabilization

Chapter 6

Protein complexation with DNA phosphates as a cause for DNA duplex destabilization. A thermodynamic model

Chapter 7 Determination of the nature of the conformational transmission effect

9

13

30

46

84

104

in pentacoordinated phosphorus compounds 117

Chapter 8

6

Conformational transmission in nucleotides containing trigonal bipyramidal phosphorus as the internucleoside linkage 129

Chapter 9 Conformational transmission in silatranes. Confirmation of the pentacoordinated structure in solution

Chapter 10 Application of the conformational transmission effect for the assignment of diastereotopic proton resonances

Chapter 11 The role of hydration and stereoelectronic effects in the hydrolysis of cAMP

Summary

Samenvatting

Curriculum vitae

Dankwoord

147

157

174

181

183

185

186

7

CHAPTER 1

lntroduction

DNA duplex structures

In all living organisms. the genetic information is encoded in deoxyribonucleic acid (DNA). a biopolymer consisting of a sugar-phosphate backbone. which carries a coding sequence of the four nucleobases adenine (A). thymine (T). cytosine (C). and guanine (G) (see Figure 1). The DNA molecule is a highly flexible system. which can adopt various conformations.1

G

5'

c

A

3'

Figure 1. Structure of a DNA strand with the four bases.

Usually. the DNA molecules are present in a double-stranded form. where the bases form hydrogen bonds in the interior of a helical structure, According to the model of Watson and Crick, 2 exclusive format ion of A-T and G-C base pairs is present. and both strands have an opposite 5'-+ 3' direction ( antiparallel}. This structural model

g

has been extensively verified in the solid state3 ?tnd in solution.4 Because the genetic information is locked inside the double helix. duplex dissociation is essential for expression of the hereditary material. Therefore. it is of great importance to under­stand how the stability of the DNA duplex structure can be influenced. In biological systems. this occurs by various types of protein molecules that internet with the DNA duplex. and can either stabilize or destabilize it. lt is generally recognized. that the mechanism of these interactions and their relation to the duplex stability is poorly understood.5 The exterior of a DNA duplex consists of the two intertwined backbones. which carry an array of negatively charged phosphate groups. lt is ob\li­ous. that these charges form a primary recognition site for a protein molecule. which in general contains various posjtive charges. Based on this key function of the phos­phate groups in nudeic acids. investigations have been performed on DNA model sys­tems with modified backbone phosphates.

In order to mimic the complexation between the opposite charges on protein and phosphate. we first studied phosphate-methylated DNA systems. where the phos­phate moiety is completely neutralized. This results in a marked decrease of electros­tatic repulsions between the phosphate groups. which will influence both the struc­ture and stability of DNA duplex structures. lt will be shown that removal of the interstrand repulsions contributes to a higher duplex stability. and can even lead to the formation of completely new DNA structures. On the other hand. phosphate­phosphate repulsions within one strand play a large role in determining the structure of the biopolymer. lndeed. it is observed that phosphate-methylation in one strand of a duplex leads to a structural stress in the system. which has a negative influence on the duplex stability. lt is possible to extrapolate these results on the phosphate­methylated duplexes to DNA-protein complexes. In this way. more insight is gained in the mechanism of stabilizing and destabilizing influences of protein-phosphate complexation in DNA.

The phosphate groups also harbour the possibility for inducing structural changes in the DNA duplex. since they can be activated by the binding of a fifth ligand. In the resulting pentacoordinated phosphorus structure (Pv) with a trigonal bipyramidal (TBP) configuration. a process called conformational transmission occurs.6 The bonding properties around phosphorus are changed in the TBP struc­ture. leading to enhanced electron density on the axially located oxygens bound to phosphorus.7 This can cause conformational changes in other parts of the molecule via increased electrostatic repufsions. The concept of conformational transmission has proven to be effective _in various phosphorylated biomolecules.8- 10 and also in fundamental studies on the behaviour of pentacoordinated phosphorus. 11.1 2 In recent investigations on phospholipids it has been seen that formation of a pentacoordinated structure can increase the packing density in a phospholipid membrane.13 This change in membrane fluidity may be a key process in the activation of ion transport proteins that are embedded in the membrane. The conformational transmission effect in DNA is studied in this thesis via single-stranded model systems with a stabilized

10

pV TBP structure. lt is seen that the standard DNA conformation is markedly changed by the presenc:e of pentacoordinated phosphorus. indicating the possible trigger function of a transient pV TBP for structural changes in natural DNA sys­tems.14 lt is further shown. that the concept of conformational transmission is also valid for other compounds with a pentacoordinated center such as silicon or ger­manium. and is therefore of importance outside the bounds of organophosphorus chemistry.

Outline of the thesis

For DNA strands containing only thymine bases. methylation of the phosphate groups results in the formation of duplexes with parallel backbones and thymine­thymine base pairs. This new structure is studied in detail in Chapter 2 with NMR and UV spectroscopie techniques. Molecular mechanics calculations on phosphate­methylated DNA duplexes. both parallel and antiparallel. are used in Chapter 3 to investigate the structure and stability of the neutral systems. These theoretica! results confirm the experimental finding that elimination of the interstrand phosphate-phosphate repulsions are essential for the formation of the parallel DNA structures. In Chapter 4. it is demonstrated that the new parallel duplex structure also occurs with cytosine-cytosine base pairs, although steric interactions of the phosphate methyl group allow duplex formation exclusively for the S-chirality on phosphorus. Complexation of polycationic proteins. such as poly-l-lysine or poly-L­ornithine. with natural oligopyrimidines (i.e" only T and C bases are present} can also give rise to parallel duplexes. For oligothymidines. both próteins are equally effective in inducing duplexes. whereas oligocytidines only form duplexes after com­plexation with poly-L-lysine. lt is shown that this is due to steric interactions of the protein ammonium groups with the DNA in the case of poly-L-ornithine.

Antiparallel hybridization of phosphate-methylated and natural DNA is investi­gated in Chapter 5. and is compared to hybridization of other neutra! DNA systems. Phosphate-methylated DNA is found to be most effective in hybridization with natural DNA. Also. both the stabilizing and destabilizing effects of methylation in one strand of the duplex are seen. This balance of two effects is extrapolated to a ther­modynamica! description for protein complexation in Chapter 6. and a mechanism for protein-induced duplex destabilization is given.

Chapter 7 describes a detailed study on the nature of conformational transmis­sion in several pentacoordinated phosphorus compounds. lt is seen that increased electrostatic interactions are the driving force for conformational transmission. This knowledge is used in Chapter 8 for dinucleotide model systems. where the effect of a pV TBP in natura! DNA is simulated. lt is found that both a hydrogen-bond breaking solvent and a location of the phosphate group between two nucleosides are essential

11

for the effectiveness of the conformational transmission. Chapters 9 and 10 describe

the more genera! use of conformational transmîssion in studying the structure and

magnetic resonance assignments of pentacoordinated silicon and germanium com­

pounds. Finally. in Chapter 11 the conformation of some nucleotide systems is inves­

tigated in various solvents to give more insight in the role of sólvation in the hydro-­

lysis of cyclic AMP. an important messenger molecule.

REFERENCES

1. W. Saenger. "Principles of Nucleic Acid Structure". Springer Verlag. New York.

1984. Chapters 9 and 11.

2. J.D. Watson and F.H.C. (riek. Nature 171, 737 (1953).

3. R. Wing. H. Drew. T. Takano. C. Broka. S. Tanaka. K. ltakura and R.E. Dicker­

son. Nature 287. 755 (1980).

4. M. Nilges. G.M. Clore. A.M. Gronenborn. A.T. Brunger. M. Karplus and L. Nils-

son. Biochemistry 26. 3718 (1987).

5. See ref. 1. Chapter 18.

6. L.H. Koole, Thesis. Eindhoven University of Technology, 1986.

7. R.R. Holmes. "Pentacoordinated Phosphorus". American Chemica! Society.

Washington DC. 1980. Vol 1-11.

8. L.H. Koole~ E.J. lanters and H.M. Buck. J. Am. Chem. Soc. 106. 5451 (1984).

9. G.H.W.M. Meulendijks. W. van Es. J.W. de Haan and H.M. Buck. Eur. J.

Biochem.157. 421 (1986).

10. N.K. de Vries and H.M. Buck. Reel. Trav. Chim. Pays-Bas 105. 150 (1986).

11. A.E.H. de Keijzer, L.H. Koole and H.M. Buék. J. Am. Chem. Soc. 110. 5995

(1988).

12. A.E.H. de Keijzer and H.M. Buck. J. Org. Chem. 53. 4827 (1988).

13. G.H.W.M. Meulendijks. J.W. de Haan. M.H.P. van Genderen and H.M. Buck.

Eur. J. Biochem" submitted for publication.

14. H .. M. Buck. Reel. Trav. Chim. Pays-Bas 99. 181 (1980).

12

CHAPTER 2*

A parallel right-handed duplex of the hexamer d(TpTpTpTpTpT) with phosphate triester linkages

ABSTRACT

In this chapter it is shown that stable parallel thymine-thymine (T-T) base pairs can be formed in aqueous solution. lnitially. this observation was made with 3'.5' -di-0-acetylthymidine in water. which showed an imino resonance at 13.45 ppm in the 1H NMR spectrum. Using the nucleoside diphosphate d(pTp). the formation of T-T base pairs could only be induced via methylation of the phosphate groups. This leads to the suggestion that intermolecular electrostatic phosphat&-phosphate repulsion predudes T-T base pairing for unmodified d(pTp). lt is shown that T-T pairing is also manifest on the dinudeotide level. provided that the phosphate groups are methylated. Using the dinudeoside phosphate d(TpT) which was separated in its diasteromeric forms. it was shown that the mini-duplex melts at T m """ 30"C. Furth­ermore. it was shown that the duplex of d(TpT) is parallel. From the detailed confor­mational analysis of the individual diastereomers it follows that the duplex has a right-handed h.elical sense. since the backbone bonds C4·-Cs· and C5·-05• are preferen­tially 'Y + and (3 1, and the furanoses reside primarily in the south conformation. With the hexamer d(TpîpTpTpTpT). it was shown that T-T pairing alsa occurs on the hexanudeotide level. after methylation of the phosphate groups. The resulting duplex has a T m value of approximately 65"( as was established with UV hyperchromicity and variable-temperature 500 MHz. 1H NMR. lt could be dearly established that the duplex is parallel. Molecular modelling studies on the duplex of phosphate-methylated d(TpîpTpTpTpT) yielded a remarkably slim. parallel structure with about eight resi­dues per turn. The possible relevance of these alternative DNA-like duplexes is briefly ment ion ed.

*L.H. Koole. M.H.P. van Genderen and H.M. Buck. J. Am. Chem. Soc. 109. 3916 (1987).

13

INTRODUCTION

Thymine-thymine base pairing was recently observed in the crystal structures of

3' .5'-di-O-acetylthymidine1 and cis-thymidine-3' .5'-N.N-dimethylphosphoramidate.2

Both crystal structures display ari approximate twofold rotational symmetry. which is due to the fact that the T-bases are linked via two virtually identical N3-H···04

hydrogen bonds (see Figure 1). Following the convention of Rose et al..3 it is easily seen that the a faces of the coupled bases are on the same side: i.e .. the T-bases are

parallel.

Figure 1. X:ray crystal structure of the 3' .5' -di-0-acetylthymidine dimer. The , atomie ~umbering is .indicated in ttie left-side monomer.

Therefore. the 5'-. 3' vectors' ruil in the same direction on both sides of the T~T pair. This chaptèr is focused on the formatlon of T-T base pairs in solution. lt was found

that T- T pairing ·reàdily occurs on. e.g .. the mono-. di-. and hexanucleotide levet. pro­vided tnat the backbone phosphate groups are triesterified. In a previous communication.ta we already published our preliminary r'e5ults on the phosphate­methylated hexanucleotide duplex. Various physico-chemical techniques (hîgh­resolution proton NMR, UV hyperchromicity) were used in order to characterize these non-Watson-Crick parallel duplexes.

14

RESUL TS AND DISCUSSION

T-T pairing on the mononudeotide level

We first studied T-T pairing using the acetylated nucleoside 3' .5'-di-0-acetylthymidinela (vide supra) and the nucleoside diphosphate d(pîp). In fact. 3'.5' di-0-acetylthymidine provided the first indication that T-T pairing may occur in aqueous solution. since the imino proton NMR signa! was found at a remarkably

low-field position (13.45 ppm4) in comparison with unlinked thymine bases (11.2 ppm5). Using the d(p Tp) system. which contains charged phosphomonoester moieties. no indication for T-T pairing was found. This suggests that electrostatic phosphate-phosphate repulsions reduce the propensity for T-T pairing. lndeed.

methylation of the phosphate groups with methyl methanesulfonate results in a dis­tinct imino resonance in water at 13.5 ppm.4·6 thus indicating T-T pairing. For

phosphate-rnethylated d(pTp). it was also found that the chemica! shift of the imino protons is strongly concentration-dependent. At higher dilution. broadening and upfield shifting are observed. which is consistent with gradual dissociation of the

dimeric structure. Figure 2 shows the imino chemica! shift as a function of C0• which

denotes the primitive concentration of phosphate-methylated d(p Tp).7

t 13.8

SNH

( ppm)

13,4

13.0

12.6

1 2.2

0 5 10 15 20

c0 !mM) -

Figure 2. lmino chemica! shift as a function of C0• the primitive concentration of phosphate-methylated d(pTp}.

From these data. it follows that the duplex dominates to a C0 value as low as 5 mM.

15

demonstrating that T-T pairing is very stable. We have used the data in Figure 2 also to establish a lower limit for the formation constant K of the T-T dimer. From the fact that K = c0-

1 at the midpoint of the dimerization equilibrium (which is be low C0 = 5 mM), it follows that K > 200 M- 1. As far as we are aware. this is the

first self-association constant that refers to mononucleotides in aqueous solution.8

Using the high-resolution 1H NMR spectra of the dimers of 3' .5' -di-0-acetylthymidine and phosphate-methylated d{pTp) in water. it could also be esta­blished that the T-T pairing is parallel. The observation of a single. degenerate spec­trum for both model compounds proves that the hydrogen-bonded nucleotides are symmetry-related and hence linked via two N3-H-"Q4 hydrogen bonds (vide supra). In the case of antiparallel T-T pairing via one N3-H".04 and one N3-H"·02 hydrogen

bond. far more complicated 1H NMR spectra would be obtained. since the coupled nueleotides would then reside in different magnetic environments (assuming the thy­mine bases have the usual anti orientation).

T-T pairing on the dinucleotide level

T-T base pair formation on the dinueleotide level was studied with the dinuclecr side phosphate 1. The system was synthesized via a slight modification of the com­mon phosphite triester method. devised originally by Caruthers et al.9

T

Originally. 1 was obtained as a mixture of the Rp and Sp diastereomers. In water.

this mixture corresponds to a twcrline 31 P NMR spectrum (2.14 and 2.04 ppm). and a highly crowded 1H NMR spectrum. The Rp/Sp mixture was separated by reversecl­

phase HPLC. according to the procedure of Stee et al.10 (see Experimental Section). The diastereomer which was eluted first corresponds to the downfield 31 P peak at

16

2.14 ppm. whereas the slower fraction is found upfield at 2.04 ppm.11 For each of the two diastereomers in water. at 20°C. the imino protons were found at 13.1 ppm in the 1H NMR spectrum.12 i.e" 1 is also present in the duplex form. We observed that increasing the sample temperature results in substantial broadening and upfield shift­

ing of the imino resonances toward 12.5 ppm, which indicated dissociation (melting) of the duplex structure. From the melting curves (imino chemica! shift vs. tempera­ture. not shown) we conduded that the melting temperature (T m) of 1 is roughly

30"C. The detailed 1H NMR spectra of the diastereomers of 1 at 20°C show that the coupled strands give rise to identical spectra. i.e" the structure is higly symmetrie. Therefore. the spectra are only consistent with a duplex with parallel T-T pairing (vide supra). For the individual diastereomers. it was possible to analyze the confor­

mations of the 2' -deoxyribose rings and. in part. the phosphate backbone in great detail by means of high-resolution 1H NMR. In order to obtain a complete set of vici­nal proton-proton coupling constants. we measured a two-dimensional J-resolved 1H NMR spectrum at 300 MHz13 for both structures. The results of these experiments

are summarized in T able 1.

Table 1. Spectra! data of both diastereomers of 1. as determined from the two-dimensional J-resolved 300 MHz 1H NMR spectra.

Slow fraction• 8(31P) 2.04

top · bottom

J1·2· 8.2 Jn" 6.0 Jn 6.2 h·3· 2.8 Jn· 2.6 J4•5• 4.0 J4•5" 4.4 JP3' 5.9 Jps· Jp5"

•See ref. 11.

7.2 6.6 7.2 3.6 2.8 2.6b 4.oh

bObtained by simulation.

Fast fraction• 8(31P) 214

top bottom

8.4 6.0 6.0 2.6 2.4 4.0 4.4 6.4

7.6 6.4 6.4 3.6 3.2 2.6b 4.0b

6.0b 5.6b

lt is well-known. that the 2' -deoxyribose ring in nucleosides and nucleotides is

involved in a rapid two-state conformational equilibrium between a south form (C2·­

endo/Crexo twist). and a north form (Cr-exo/Crendo twist).14 According to

Altona15.16 the population of the south form can be accurately estimated on the basis

of Jn" and Jr3·: \

17

x(south) {17.8 Jrr - Jn·}/10.9

As can be seen in T able 11. the south conformation dominates for the two furanose

rings in each of the diastereomers. The coupling constants hs· and Jn" were used

to analyze the conformation around the C4·-Cs· (y) bonds. The formula 15·16

was used for this purpose. Clearly. the y+ conformation (in which 0 5• is located

above the 2' 7deoxyribose ring) dominates in both the top (5' ~0-acety 1) and bottom

(5' -0-phosphate) residues in the diastereomers. The three-bond phosphorus-proton

coupling constants Jp5• and Jps" were used to describe the conformation around the

centra! Cs·-05· (13) bond of the Rp and Sp structures. For this. we used the formula 15

x(l3 1) = {23.9 - Jp5· - JPS"}/18.9

From Table ll. it appears that both diastereomers have a dominant contribution of 13 1

(in which the phosphorus is in a trans orientation with respect to C4·) to the confor­

mational equilibrium around the central C5·-05· bond.

Table ll. Conformational characteristics of the 2' -deoxyribose rings. and the f3 and ')' honds of both diastereomers of 1.

Slow fractiona 8(31P) 2.04

top bottom

x(south) 0.83 x(')' +) 0.53 x(f3')

•see ref. 11

0.70 0.70 0.65

Fast fraction• 8(31P) 2.14

top bottom

0.84 0.53

0.72 0.70 0.65

The data in T able Il do not show significant conformational differences between

the diastereomers. For both structures. the combination of south {2'-deoxyribose).

y+ (C4·-C5-). and (3 1 (centra! C5·-0s·) is preferred. lt should be r:nentioned that the

same conformation is encountered in standard right-handed B DNA structures. both

in solution and in the solid state. In summary, from the overall structural information

on the diastereomeric forms of 1. we conclude that both systems exist as stable sym­

metrie mini-duplexes with right-handed parallel phosphate-sugar backbone strands. lt

should be mentioned that one- and two-dimensional NOE spectroscopy cannot

18

bolster this structural model. The symmetry of the structure implies that any inter­strand NOE contact has a stronger and overlapping intrastrand counterpart. There­fore the use of NOEs for interatomic distance estimation is essentially exduded.

T-T pairing on the hexanucleotide level

In order to investigate T-T base pair formation also on the hexanudeotide level. we used the hexanucleoside pentaphosphate d(T p T p T p T p T p T). ia The methylation of the phosphate groups with methyl methanesulfonate6 was essentîally complete (99%). as was shown by precipitation of only 1% of the unmethylated starting material with ethanol/water (75:25 v /v). However, since the phosphate methylation is not stereospecific. a multicomponent mixture of diastereomers is obtained. which could not be separated by means of HPLC techniques. Nonetheless. it could be clearly shown that the phosphate-methylated d(TpTpTpTpTpT) is present as a stable duplex in aqueous solution. This conclusion was based on the observation that the imino protons resonate at 13.3 ppm in the 1H NMR spectrum.4 Furthermore. încreasing the sample temperature results in a double helix ..... coil transition. as was observed with UV hyperchromicity and variable-temperature NMR experiments. Using the UV hyperchromicity technique. we observed a reversible dissociation of the double helix.at a T mof approximately 67"C. for a substrate concentration of 1.3· 10-5

M in water.1a lt was found that the melting behaviour is identical in aqueous Tris/EDTA buffer solutions (pH 7.5) of 20 mM and 0.2 M. The fact that the T m is

not influenced by the ionic strength of the solution is consistent with the absence of a formal negative charge on the phosphate groups. The neutral character of the methylated substrate also enabled us to study the melting behaviour in less polar sol­vents. The sa me T m value of approximately 67°C was found in a 80:20 ( v /v) mix­

ture of ethanol and water. No melting point in the ternperature range 10-80-C was found using the hydrogen-bond breaking solvent hexamethylphosphoric triamide (HMPT).17 This means that in the latter case only the single strand is present. which is confirmed by an iminö chemical shift of 12.5 ppm in HMPT.4

In the case of phosphate-methylated d(TpTpTpTpTpT). it was not possible to determine the conformation of the phosphate-sugar backbones from the high­resolution 1H NMR spectrum. Evidently. this is due to the fact that the duplex con­tains 10 chiral phosphate groups. and therefore exists as a complex diastereomeric mixture with a highly crowded 1H NMR spectrum. However. the subspectra of the imino- and base-methyl protons are well-defined (see Figure 3). since these protons are located at an appreciabie distance from the chiral phosphate groups. In both sub­spectra. three peaks in the approximate ratio 1:1:4 are observed. Most likely. the ter­minal base pairs are associated with the fower peaks. whereas the imino- and base­methyl resonances of the inner base pairs practically coincide at 13.29 and 1.32 ppm,

19

13.SO 1uo 13.10 1.so uo --- &lppmJ l\(ppm J

Figure 3. Subspectra from the 500 MHz 1H NMR spectrum of phosphate­methylated d(TpTpTpTpTpT): left. imino resonances; right. base-methyl reso­nances.

130

respectively. We have used these simple subspectra in two ways for further charac­terization of the duplex structure of phosphate-methylated d(TpîpTpîpTpT). First. the chemica! shifts of the imino- and base-methyl protons were measured as a func­tion of the sample temperature. The melting curves that were obtained are given in Figure 4. All curves show a melting transition at T m = 64°C.18 Remarkably. the UV hyperchromicity measurements (vide supra) .resulted in virtually the same T m value. This means that dilution from 10 mM (NMR sample) to 13 µM (UV sample) does not induce a measurable degree of dissociation of the duplex, which reflects the marked stability of the parallel structure.

Secondly. the subspectra were used to discriminate between the symmetrie. parallel T-T and the asymmetrie. antiparallel T-T coupling (see Figure 5). For instance. the terminal· base· pairs are expeeted to eorrespond with four methyl and four imino resonanees, The faet that the phosphate-methylated d(TpTpTpTpîpî) exists as a complex mixture of diastereomers precludes the use of two-dimensional NOE spectroseopy as a reliable tool for struetural elucidation. However. we did per­form one-dimensional double-resonanee experiments in whieh one of the lower base methyl resonanees was specifically irradiated. No NOE effeets in the other base-

20

13.4 •

J 13.2 0

NH lppml

13.0

12.8

12.li

12.4

1.s

JMe 1.4

lppml : : ~ 1.3 0

1.2

20 30 40 50 60 70 80 90

îl°CI ---

Figure 4. Melting curves of phosphate-methylated d(TpTpTpTpTpT). as obtained with variable-temperature 500 MHz 1H NMR: upper part. imino resonances; lower part. base-methyl resonances. In both curves. the resonances of the inner (i) and terminal (t) base pairs are shown separately. Note that the terminal imino reso­nances coincide at temperatures higher than 30"C.

methyl peaks were observed. These results are consistent with a symmetrie parallel structure. In this case. the two methyl groups of the same terminal base pair are simultaneously saturated; i.e. no methyl-methyl contact within the T-T base pair is

seen. In the case of an antiparallel arangement, a methyl-methyl NOE effect within the same base pair is expected.

lt is tempting to raise the question if self-association is restricted to thymine

bases. lnterestingly. literature data for the crystal structure of 2' -deoxycytidine (2' -

dC) also show the formation of a C-C pair with a parallel arrangement.19 However. we found for 2' -dC in aqueous solution at 2ü°C no low-field resonances of the NH 2

protons.4 which implies that these protons are involved in a rapid exchange with the

solvent: i.e .. the nucleosides are not coupled. Also on the hexanucleotide level of

21

3' ~ s' o

#N~O··············HN~N.l Me~NH··············O~·

0 L Me ,,,;:,

j"~, .............. ,,~,J Me~NH··············O~

os' Me3'

Figure 5. Schematic representation of the geometries of the terminal base pairs in oligomeric duplex structures with T-T base pairing: left. antiparallel T-T pairing: right. parallel T-T pairing.

phosphate-methylated d(CpCpCpCpCpC) no NH 2 resonances could be detected in the low-field region of the 1H NMR spectrum4 recorded at 20 and 4"C. Additionally. no melting· transition was observed in a UV hyperchromicity experiment in the tempera­ture range 10-90"C. We considered it of interest to synthesize the phosphate­methylated dodecamer d(CpCpCpCpCpCpTpTpTpTpTpT). which represents a com­bination of phosphate-methylated d(CpCpCpCpCpC) and d(TpTpTpTpTpT). Duplex formation of the dodecamer would result in either a parallel duplex with six T-T and six C-C base pairs. or in a parallel duplex with six T-T base pairs and dan­gling. nonpaired C-bases. However. since no double helix;::t coil transition was found in the temperature range 10-00°C (based on UV hyperchromicity experiments). it can be conduded that the dodecamer is present in the single strand form: i.e .. the C­bases actually preclude the formation of a parallel duplex via T-T pairing.

Structural model

A molecular model of the phosphate-methylated d(T p T p T p T p T p T) was con­structed with computer graphics using the structural information provided by the X­ray data of 3' .5' -di-0-acetylthymidine. and the detailed NMR data on both diastereo­mers of. the dinudeoside monophosphate .1. Figure 6 shows a top and side view of the proposed structure. 20

We found that the parallel T-T base pairs indeed fit excellently in a right-handed double helix with y+ and ~' backbone torsion angles. south conformation of the 2' deoxyribose rings. and anti conformation of the T-bases. The inherent symmetry

22

Figure 6. Top and side view of the computer-generated structural model of the parallel duplex of phosphate-methylated d(TpTpTpTpTpT).

results in the formation of two identieal grooves, instead of the minor and major grooves that are found in right-handed B DNA. The structure has approximately eight base pairs per turn. and a rise per base pair of 3.6 A. The combination of two thymine bases results in a helix diameter of 15 A. whereas the purine-pyrimidine base pairs in B DNA correspond with the much greater helix diameter of 21 Á.8 lnterest­ingly. our structural model also provides a more plausible explanation for the fiber X­ray diffraction pattern of the dinudeoside phosphate d(TpT). as was recently observed by Tollin et al.21 lt was found that d(TpT) crystallizes as a helical structure with about seven units per turn and a rise per base pair of 3.8 À. The structural model as proposed by Tollin et al. essentially comprises antiparallel T-T base pairs. and a head to tail arrangement of d(T p T) residues:

TpT TpT TpT TpT TpT Tpî TpT TpT

However. it was necessary to invoke unlikely conformational characteristics as a syn orientation of the T-bases22 and a y- conformation around the C4·-Cs· bond in order to construct an antiparallel helix. On the basis of our present data. we feel that a head-to-tail structure with parallet T-T pairing gives a more adequate explánation for the fiber X-ray diffraction pattern of d(TpT). The resulting double helical structure has backbone strands that are regularly interrupted. Consequently. phosphate groups are not found in opposite positions. which minimizes electrostatic repulsion.

23

Essentially. this situation parallels the phosphate-methylated system d(TpTpT.pTpTpT) in which th~phospbate repulsions a~e completely eliminated.

CONCLUOING REMARKS

lt has been shown that the formation of T- T base pairs may occur readily in aqueous solution. provided that interstrand phosphate-phosphate repulsions are elim­inated via methylation of the phosphate groups. The resulting duplex structures are parallel and highly symmetrie. Detailed conformational analyses of phosphate­methylated d(TpT). for which the Rp and Sp diastereomers were separated. revealed

that the backbone conformations are standard: i.e .. a close resemblance with respect to right-handed B DNA is seen. The · phosphate-methylated hexamer d(TpTpTpTpTpT) exists as a duplex with exceptional stability in aqueous solution. Since the duplex is essentially uncharged. the stability of the structure is not influenced by the ionic strength of the medium. However. d(TpTpTpîpTpT) exists in the single strand form in the hydrogen-bond disrupting solvent HMPT. We feel that the present model systems may be useful as simple artificial probes for phos­phate backbone-protein interactions (recognition) in which the negative chàrges on the phosphate groups are (partially) neutralized. In this context it is therefore of interest that complex formation between poly-L-lysine or poly-L-arginine and DNA leads to an increased stability of the duplex as was reflected in the elevated values of T m with respect to uncomplexed DNA.23 Our results show that neutralization of the phosphate groups may also result in the formation of parallel DNA structures. Finally. it can be concluded that the proposed parallel structures for the. phosphate­methylated thymidine oligomers differ markedly from the familiar DNA systems. and may therefore be considered as a valuable contribution to the design of new DNA-like structures.

EXPERIMENTAL SECTION

Synthesis

3' ,5' -Di-O-acetylthymidine1a

This compound was prepared by adding acetic anhydride (16.5 mmol. 1.55 ml) to a solution of thymidine (8.2 mmol. 2.0 g) in 50 ml of anhydrous pyridine. The sol­vent was evaporated after 3 h. and the resulting viscous glass was chromatographed on a Woelm silica gel column. using dry 2-butanone as eluent (R1 == 0.65). The pro­

duct was obtained as a white crystalline solid in 41% yield. Anal. Calcd for

C14H1s01N2: C. 51.50: H. 5.50: N. 8.61. Found: C. 51.84: H. 5.62: N. 8.61.

24

Thymidine 3' ,5'-di(dimethylphosphate)la

Thymidine 3'.5'-diphosphate (sodium salt. purchased from P-l Biochemicals) was treated with methyl methanesulfonate according to Rhaese and Freese.6 Methy­lation of the phosphate groups was essentially complete. as was shown with 31 P NMR spectroscopy. The methylated structure corresponds with two 31 P NMR reso­nances at 0.4 and 0.6 ppm. whereas two signals. at 3.8 and 4.1 ppm are found for the unmethylated compound.

3' -0-Acetylthymidine

5' -O-Tritylthymidine24 (15.6 mmol. 7.53 g) and acetic anhydride (16 ml) were dissolved in 70 ml of anhydrous pyridine. This mixture was magnetically stirred overnight. After complete removal of the solvent (coevaporation with two 20-ml por­tions of water). the oily residue was chromatographed on a Woelm silica gel column. using dry 2-butanone as eluent. The yield of 5'-0-trityl-3'-0-acetylthymidine (Rf 0.52) was 8.Ql g (98%): 1H NMR (acetone-d6) S 1.48 (3H. s. CH3 base). 2.06 (3H. s. CH3 acetyl). 2.48-2.54 (2H. m. H2'/H2"). 3.46-3.50 (2H. m. Hs'/HS"). 4.18 (1H. m. H4·). 5.50 (1H. m. H3·). 6.38 (1H. dd. Hr). 7.24-7.56 (15H. m. trityl). 7.64 (1H. s.

H6)·

A solution of this compound (14.8 mmol. 7.8 g) in 25 ml of a mixture of acetic acid and water (4:1 v/v) was refluxed for 10 min. After complete evaporation of the acetic acid (coevaporation with two 20-ml portions of water). the white residue was chromatographed on a Woelm silica gel column. using dry 2-butanone as eluent. 3' -0-Acetylthymidine (R1 = 0.23) was obtained as a white foam: yield 3.03 g (72%): 1H NMR (acetone-d6) S 1.84 (3H. s. CH 3 base). 2.09 (3H. s. CH 3 acetyl). 2.36-2.40 (2H. m. Hr/Hr). 2.88 (1H. bs. OH). 3.83-3.88 (2H. m. H5·/H5"). 4.09 (1H. m. H.t'). 5.35 (1H. m. H3·). 6.32 (1H; dd. Hr}. 7.84 (1H. s. H6)·

3' -(Methoxy-N ,N-diisopropylaminophosphino )-5' -0-tritylthymidine

5' -O-Tritylthymidine24 (14.3 mmol. 6.9 g) was suspended in a mixture of anhy­drous chloroform (120 ml) and anhydrous N.N-diisopropylethylamine (10 ml). The suspension was magnetically stirred and kept under an atmosphere of dry argon. Ch.loro(N.N-diisopropylamino)methoxy-phosphoramidite25 (15.5 mmol. 3.1 ml) was added dropwise over 5 min. After the addition. the reaction mixture was stirred for 2 h. Ethyl acetate (300 ml. prewashed with NaHC03) was added. and the solution was

washed with saturated NaCI solution (4 x 150 ml) and water (50 ml) and dried over Na2S04. After evaporation of the solvent. a yellow oil was obtained, which was chromatographed on a Woelm silica gel column. The eluent was dichloromethane/hexane/triethylamine (45:45:10). The yield of the desired product (Rt = 0.43) was 5.18 g (58%): 1H NMR (acetone-d6) 8 1.16 (12H. s. CH3 isopropyl). 1.58 (3H. s. CH3 base). 2.52-2.56 (2H. m. H2'/Hr). 3.63-3.68 (2H. m. H5·/H5"). 3.32

25

and 3.42 (3H. d. POCH3• J 11 Hz). 4.24 (1H, m. H4·). 4.80 {1H: m. H3·); 6.44.(!H. dd. Hr). 7.32-7.60 (15H. m. trityl). 7.68 (1H. s. H6):. 31P NMR (acetone-d6). 8 154.6 and 154.1 (intensity ratio 1:1.09).

3' ,5' -Di-0-acetylthymidyf..(3' ....... 5')"0-methylphQsphate (1)

3'-0-Acetylthymidine (2.46 rnmol. 0.70 g) and 3'-(methoxy-N.N-diisopropylaminophosphino)-5' -0-tritylthymidine (2.11 rnmol. ·1.33 g) were dissolved in 15 ml of anhydrous pyridine. 1H-Tetrazole (6.4 mmol. 0.45 g). dissolved in 5 ml of anhydrous pyridine. was added. and the yellowish mixture was stirred at room temperature for 24 h. After complete evaporation of the pyridine (coevaporation with three 10-ml portions of dry dichloromethane). a yellow syrup was obtained. Chroma­tography on a Woelm silica gel column using dry 2-butanone as eluent afforded 0.73 g ( 43%) of. 3' -0-acetyl-5' -tritylthymidyl-(3' ....... 5')-0-methylphosphite as a slightly coloured foarn. The two diastereomers were seen separately on the analytica! thin layer chromatography plates at R1 = 0.35. and 0.31: 1H. NMR (acetone-d6) o 1.51 (6H,

s" CH 3 base). 1.85 (3H. s. CH 3 acetyl). 2.28-2.42 (4H. m. Hr/Hr). 3.46 (3H. d. POCH3• J 11 Hz). 3.38-3.50 (4H. m. Hs·/HS"). 4.10 {2H. m. H4·). 4.65 (2H. m. H3·). 6.35 (1H. dd. Hl'). 6.40 (1H. dd. Hl'). 7.25-7.52 (15H. m. trityl). 7.62 (2H. s. H6): 31 P NMR (DMSO~d6)8145.7 and 145:2 {intensity ratio 1:0.86).

The phosphite was dissolved in 9 ml of dry dioxane. and N02 gas was slowly bubbled through. After 10 min. thin layer chromatography indicated complete conver­sion into the phosphate. The reaction vessel was then sparged with dry nitrogen: after evaporation of the solvent. 20 ml of the detritylating reagent (a mixture of acetic acid and water. 4:1 v/v} was added. This mixture was then refluxed for 10 min. cooled to .room temperature. and concentrated. The last traces of acetic acid were removed by coevaporation with two 10-ml portions of water. The remaining yellow oil was chromatographed on a Woelm silica gel column. using a mixture of 2-butanone and triethylamine (95:5 v/v) as eluent. Concentration of the appropriate fractions (R1 = 0.13) afforded 0.25 g of a yellowish syrup: 31 P NMR (DMSO-d6) 8 3.9 and 3.7 ppm. Subsequently. the oil was mixed with acetic anhydride (0.8 ml) and stirred for 20 h. The excess of acetic anhydride was thoroughly evaporated. and the residue was carefully chromatographed on a Woelm silica gel column. A mixture of 2-butanone and triethylamine (95:5 v /v) was used as eluent. The desired product (0.17 g. 24%). having R1 = 0.38. was obtained as a colourless viscous oil: 1H NMR

(020). 8 1.89 (6H. s. CH3 base). 2.15 (6H. s. CH3 ~cetyl). 2.41-2.65 (4H. m. H2'/Ht'). 3.80-3,82 (2H. m. Hs·/H5"). 3.88 {3H. d. POCH3• J = 11 Hz). 4.28-4.51 (3H. m. 2H4-/Hs·/H5"). 5.07-5.13 (1H. m. H3·). 5.37-5.42 (1H. m. H3·). 6.24 (2H. m. Hr). 7.57 (1H. s. H6). 7.64 (1H. s. H6); 31P NMR (D20) o 2.14 and 2.04 (intensity ratio 1:1.09).

26

Chromatographic separation of the diastereomers of 1

The separation of the diastereomers of t was performed with a Dupont-830 HPLC system. which was equipped with a Nudeosil 100-7-C18 column (250 x 20 mm). and a Zeiss EM2D UV detector (set at 265 nm). The eluent was prepared from acetonitrile (13%. HPLC grade). glacial acetic acid (1%. aldehyde free). triethylamine (1%. Gold label grade). and deionized water. The flow rate was 20-25 ml/min. The elution times for the fractions were 38 and 24 min.

Oligonucleotides

The hexamers d(TpTpTpTpTpT)ta and d(CpCpCpCpCpC) were synthesized on a 10-µ mol scale with an Applied Biosystems 380A DNA synthesizer following a stan­dard phosphite (OCH3) triester protocol. The purity of the material was carefully checked with gel electrophoresis and HPLC. For d(TpTpTpTpTpT). methylation of the phosphate groups could be performed in a straightforward manner according to the procedure of Rhaese and Freese.6 The methylation was essentially complete (99%). as was shown by precipitation of only 1% of the unmethylated starting material with ethanol/water (75:25 v/v). In the case of d(CpCpCpCpCpC). phos­phate methylation was accomplished as follows. The product from the synthesizer was dissolved in t ml of dry pyridine: benzoyl chloride (10 equiv) was added. After stirring for 24 h. the pyridine was completely removed. The residue was dissolved in 1 ml of water. and methyl rnethanesulfonate (7 equiv) was added. After standing overnight. the solvent was thoroughly evaporated. and t.5 ml of a 5% solution of hydrazine in water was added. Base deprotection was complete after stirring at 30"C for 5 h. as was proven with thin layer chromatography. Finally. the phosphate­rnethylated substrate was purified by means of short-column chromatography using a 0.2 M Tris/EDT A buffer solution (pH 7.5) as eluent. The dodecamer d(CpCpCpCpCpCpTpTpTpTpTpT) was synthesized on a 1-µmol scale. also on an Applied Biosystems 380A DNA synthesizer. The purity of the material was checked with gel electrophoresis. Protection of the C-bases. methylation of the phosphate groups. and subsequent deprotection of the C-bases were performed as described above. UV hyperchromicity experiments on the phosphate-methylated dodecamer were performed without further purification.

Spectroscopy 1H NMR spectra were run in the FT mode at 200.12 300.13 or 5004 MHz on

Bruker spectrometers. Measurements in water refer to a 85:15 mixture of H20 and D20. in which deuterium provided the field-frequency loek. The technique described

by Haasnoot et al.26 was used to suppress the strong H20/HDO solvent signal. Pro­ton chemica! shifts were referenced against tetramethylammonium chloride (8 3.18 pj>m). In all one-dimensional spectra, appropriate spectra! windows (10-15 ppm) were

27

chosen and Fourier transformation was usually performed .with 32K dàta'points. The two-dimensional J-resoh1ed spectra wer:e run at 300 MHz .with a spec;tral window of 160Q Hz (8K data points) on the.chemica! shift (f2) axis and 30 Hz (128 data points) on the J (f1) axis. 31 P NMR spectra wer.e run in the FT mode at 36.4 or 80.9 MHz.

also .. on .Sruker spectrometers, Chemcial shifts are rel<1tiv:~ to 85% H3P04: they are designated · positive if ;down field with respect to the standard. The UV hyperchromi• city experiments were performed on a Perkin-Elmer 124.spectrometer. using 10-mm cuvettes and a wavelength of 260 nm.

REFERËNCES

1. (<!k L.H. Koole. M.H.P. yan Genderen. H. Frankena. H.J.M. Koeken.· J.A. Kanters and H.M. Buck. Proc. K. Ned. Akad. Wet .. Ser. B 89. 51 (1986) (com­municated by H.M. Buck at the meeting of Nov 25. 1985). (b) C.C. Wilson. J.N. Low. P. Tollin and H.R. Wilso.n. Acta Crystallogr .. Sect, C: Cryst. Struct. 40. 171?J1984).

2. W,G,. Bentrude. A.E, Sopchik and W.N. Setzer: Acta CrystaUogr .. Sect. C: Cryst. Struct. 42 .. 584 :(1986),

3. l.A. Rose. KR. Hanson. K.O. Wilkinsori ànd M.J. Wimmer. Proc. Natl. Acad. Sci. USA 77,.-243!>(1980). ..

4. . Meàsurëd at 500 MHz on the Bruker WM 500 spectrometer Óf the Dutch National hf NMRFaciiity at Nijmegen, The Netherlarids.

5. C.A.G. Haasnoot .. J.H.J. den Hartog. J.F.M. de Rooij. J.H. van Boom and C. Altona. N.ature (London) 281. 235 (1979). .

6. Exdusive met.hylation of the phosphate groups was accomplished according to: H.-J. Rhaese and E. Freese.Biochim. Biophys. Acta 190. 418 {1969).

7. The primitive concentration C0 equals the concentration of free monomer plus

twice the concentration of the duplex.

8. W: Saenger. "Principles of Nucleic Add Structure". Springer Verlag. New Vork. 1984.

9. {a) M.D. Matteucci and M.H. Caruthers. Tetrahedron Lett. 21. 719 (1980). (b) M.D. Matteucci and M.H. Caruthers, J. Am. Chem. Soc. 103. 3185 (1981). (c)

· R. T. Pon. M. Dahna and K.K. Ogilvie. Nudeic Acids Res. 13. 6447 (1985).

10. W~J. Stee. G. Zon and B. Uznanski. J. Chromatogr. 326, 263 (1985).

11. A tf:ln,taJive Rp /Sp as!,iignment can be made on the basis of the 31 P chemica! shifts, .See, e.g.: W. Herdering. A. Kehne and F. Seela. Helv. Chim. Acta 68. 2119 (1985)'.' This as~ignment results in Sp configuration for 8(31P) = 2.14 and Rp config~ràtion for S(31P) 2.04 ppm. ·

12 •. Measured on .a Bruker AC 200 spectrometer at. the Eindhoven Univérsity of Technology.

13. Measured on a. Bruker CXP 300 spectrometer at the Eindhoven University of Technology.

14. Nomenclature in this chapter fotlows the recent IUPAC-IUB recommendations. See: E._r. J. Biochem.131; 9 (1983)~

. .

15. C. Altona. Red. Trav. Chim. Pays-Bas 181. 413 (1982).

16 •. C.A.G. Haasnoot. F.A.A.M. de Leeuw and C. Altona. Tetrahedron 36. 2783 (1980) ..

11. (a} H: Normant. Angew. Chem. 23. 1029 (1967). · (b) H. Normant. Bull. Soc. Chim. Fr. 2. 791 (1968).

ll. The· ~chanp of the imino · protons with the solvent ·is remarkably slow: bri)adening of the irnino resonances is · only observed above 75°C. . Chemica! shifts could be determined accuratély for temperatures up.to as·c.

10. O.W. Youna and H.R. Wilson. Acta Crystallocr" Sect. B: Struct. Sci. 31. 961 (1975).

20. Molecular mechanics calc:ulations usin& the AMBER program confirm this struc­tural model. See: M.H.P. van Genderen. LH. Koole. O.M. Aagaard. C.E.J. van l•e and H.M. Buck •. Biopolymers 26.1447 (1987).

21. P. Tollin. R.T, Walker and H.R. Wilson. Nûdeic Acids Res. 12. 8345 (1984).

22. A syn conformation for the thymine base has been observed in the left-handed duplex of d(CGCGATCGCG). in which the C•bases are brominated. See: J. · Feigon. A.H.:.:.J. Wang. G. van der Marel. J.H. van Boom and A. Rich. Science 230. 82 (1985).

23. M. Tsuboi. in •conformation of Biopolymers•. Academie Press. New York. 1967. Vol. ll. pp~ 689-702.

24. A.M. Michelson and A.R. Todd. J. Chem. Soc. 951 (1953).

25. S. Beaucage and M.H. Caruthers. Tetrahedron lett. 22. 1859 (1981).

26. C.A.G; Haasnoot and C.W. Hilbers. Biof>olymers 22. 1259 (1983)~

29

CHAPTER 3•

;.':::."

Molecular mechanics studies of parallel and antiparallet , , , " phosphate-methylated DNA . .;

. ,·..... () '

ABSTRACT '< ' ,.,"., :'•. 1. ',

Methylation of phosphate groups in oligothymidine strands leads to' a. p~rallel duplex· ~ith ·T-T basé p~i,rs'. 'Mo1ecular · mechanits ' calculatforis o~ parallèf

d(TpTpTpTpTpT)i show it to be asymmetrie right-handed helix with B DNAcon­

formationai ·characteristics·r. Ph~sphate. methylation stabilizes the. d\lplex,.by, rca. 41,'. k.cal/.mol. due ;to remova.l of the" interstrand phosph~te-phosphate electros~tjc repul­sions. The chirality that, is. intro<j4,ced with phosphate 11'.\ethy lat ion is, important for

the n:iolecular geometry, since Rp-!Tlethylation predominantly influences the,.confor­

~ati;ri" aroLlnd t~e ~ bond (P-d3:)'. while Sp-methylation mostly. chànges the'~· cok

formation (P-05-). This is also true in antiparallel helices with methyláted phos­

phates.'• as li's show'n. by mölècUlar mechanics calèulations on d(GpCp(;pCpGpC)i.

These ·tesult!' may'' !Jé 6f rEÎlevance to prbtein-E>NA interadions. where · phosphate

charges are also shielded. As 'tlie'pro-Sp oiygen is most aváilable in a'right-banded

helix. we ~uggest ·changes,.aro,trncj .J:he/O'-bon<;J to occur up~n. protein complexation;·

Jeading to awidening of the major groove in the d(GpCpGpCpGpC)i duplex (from 1.2 1

to' fa À) ·~ncf red~ceii'~inor gr<?ove (froiri 6 to 5 Ä). ' . . . . . " - ' > ·,, ( ' . ". ~· ' ' ,, . ' ' . ' .

,:· .. ~ ! ' , ..,~ .,

;' f'

*M.H.P. van Genderen. L.H. Koole. O.M. Aagaard. C.E.J. van Lare and H.M. Buck.

Biopolymers 26. 1447 (1987) .

. 30

INTRODUCTION

Recently. we observed that thymidine oligomers in aqueous solution can form duplexes after methylation of the phosphate groups.t.2 These duplex structures are based on symmetrie T-T base pairing via two N:;H···04 hydrogen honds. from which it follows that the 5'-+ 3' directions of the backbones are parallel instead of antiparal­lel (see Figure 1). Since unmethylated thyrnidine strands show no association in

3'

5'

3'

o ... , 1

. r1YO···•••"·HN-(--Ó> r<;\ - N '-.,_...-- NH ·" • • • • • • • • • 0 J _.-:::::::.) 5'

~r:~r /1 · "r 0 0 Me

Figure 1. Structure of the parallel T-T base-pair with 5'-+ 3' backbone directions running in the same direction.

water.3 the stability of the neutral phosphate-triesterified duplexes is most likely based on the absence of interstrand phosphate-phosphate repulsions. A detailed 1H NMR conforrnational analysis of the mini-duplex of phosphate-methylated d(TpT) was performed for both the Rp and Sp diastereomers.2 This revealed that parallel hel­ices are formed for both isomers which are right-handed, with predominantly the standard y+ and /3 1 conformations for the C4·-C5• and C5·-05• backbone bonds. and south sugar puckers. Via model building with computer graphics it was concluded that the helix is remarkably slim and has about eight residues per turn.

In order to study this remarkable structure. we performed molecular mechanics calculations on the phosphate-methylated duplex d(TpTpTpTpTpTh with both Rp and Sp configurations on phosphorus4 (see Figure 2). Since both Rp and Sp forms of

pro-Sp g___ /03• - ' -p

-·/ "' 0 05' pro-Rp

Figure 2. Definition of Rp- and Sp-methylated phosphate triesters. and of the pro-Rp and pro-Sp nomendature in the phosphate diester.

31

phosphate-methylated d(TpT)i have similar (3. y. and sugar conformations.2 the

difference in phosphate configuration should only influence the Öónds neài phos­phorus (e.g .. P-05- or P-03-). The stability of the parallel duplex in relation to methy­

lation was investigated by comparison with the unmethylated d(TpTpTpTpTpT)2

structure. lt should be noted. that this latter structure does not exist. and is only

used as a reference compound in the calculations.

The duplex formation of d(TpTpTpîpTpT)isuggests that neutralization of the

phosphate groups can influence DNA structures in general. Since such a neutraliza­

tion will also occur (partially) in DNA-protein interactions which proceed via electros­tatic interactions between amino-acid side chains (e.g .. arginine or lysine) and phos­phates.5 phosphate-methylated oligonucleotides can be used as simple models to

study conformational changes in DNA after protein complexation. Therefore. we investigated the effects of Rp and Sp phosphate-methylation on antiparallel DNA

duplexes with molecular mechanics calculations on phosphate-methylated d(GpCpGpCpGpC)i. These results were compared with unmethylated

d(GpCpGpCpGpC)i duplexes.

METHODS

Calculations were carried out using the AMBER molecular mechanics program.6

The energy function we used is conventional7:

Etptal = L KR(R-Re11)2 + L Ka(0-6eq}2

honds angles ·

+ L Vn [l+cos(n</.l-y)) + 1:( A;j - B;j + q;qj) + L (~-~) d. 2 . . . RIT RO ER. R" R IV 1hedrals 1<1 ij ij 11 Hbonds ij ij

All degrees of freedom were energy-refined until the rms of the energy gradient was less than or equal to 0.1 kcal/mol.À. For all calculations. the united-atom approach was used first. Subsequently. · hydrogen atoms were added and an all-atom energy refinement was carried out. The force constants and equilibrium values were initially taken as previously described.7 hut changes were made in input parameters and force

field for calculations on phosphate-methylated systems (vide infra). The dielectric constant E was set at R;; in all calculations. to mimic the damping effect of the sol­

vent on electrostatic interactions. Since we were especially interested in effects of phosphate neutralization. we also performed calculations with sodium counterions for

unmethy lated structures. The ions were located at a distance of 5 Ä ftom phos­phorus. and at equal distances from the two non-bridging phosphate oxygens.

32

Changes in input parameters

For phosphate-methylated substrates. we defined new phosphate groups with

the PREP module.6 incorporating a methyl group on one of the non-bridging oxygens.

The types and. charges of the atoms in the phosphate group were adjusted to reflect

this triesterification (see Table 1). The coordinates of all units are those of Arnott et

Table 1. Input parameters for unmethylated. Rp-methylated. and Sp­methylated phosphate groups.

At om Standard3

Type Charge Rp-methylated

Type Chargeh Sp-methylated

Type Chargeh

p 1.429 p 0.680 p 0.680 05· os -0.535 os -0.292 os -0.292 03· os -0.535 os -0.292 os -0.292 Ûpro-Sp 02 -0.850 0 -0.356 os -0.292

Ûpro-Rp 02 -0.850 os -0.292 0 -0.356

Cmethyl (3C 0.211 c3c 0.211

3Taken from ref. 7. . bBased on ab initio (Gaussian-80 ST0-3G) calculations of trimethylphosphate. <For the all-atom approach. replace by CT (charge -0.047) and three HC's (charge 0.091).

al..8 and a methyl group was added on one phosphate oxygen with a torsion angle of

+60" relative to the P=O bond, a C-0 bond length of 1.44 Ä. and a P-0-C angle of 120°. For the construction of the parallel helix geometry. we started by building an

antiparallel d(TpTpTpTpTpT)i helix with the NUCGEN module.6 Elimination of one

strand and doubling the remaining one by a rotation of 180" around the helix axis

afforded a reasonable starting geometry (manipulations performed with the Chem-X

program9). even though the thymine bases are separated by a large distance (N3-04

distance 4.70 A instead of the required 2.84 )V). This helix geometry was used instead of the standard NUCGEN output.

Changes in the force field

New potentials were defined for the phosphate triester group .. First. bond stretching and bond-angle deformation parameters for the nèw P=O bond were intro­

duced. The value of Req was taken as 1.435 A. based on X-ray data.10 Linear extrapo­

lation of the force constants of the P-Oester and P-0- bonds7 suggested a KR value

33

of 625.0 kcal/mol..!.2 for the P=O bond. The equilibrium bond angle 0-P=O was set at eeq 117.4°. based on our own ab initio cadulations on trimethylphosphate. A

force constant Ke of 125.0 kcal/mol.deg2 was derived by comparison with the other

0-P-O angles.7 For the torsion angles in the Ü=P-0-C fragments. specific three- and

twofold potentials were introduced. similar to the existing O-P-0-C potentials7 ( ~3

V2 0.25 kcal/mol. y 0: and T = 0.75 kcal/mol. y = 0). This results in a force

field (FF1) whkh was used initially (see Table Il}.

Table ll. Force field parameters for phosphate-methylated DNA.

Bonds ~ Kit' Angles 6' ... Ked

P-0 1.435 625.0 0-P-OS 117.4 125.0

V e Torsions1

v • Torsions n n

T n î' T n î'

O-P-OS-C3 0.25 3 0 OS-P-OS-C3 0.25 3 0 0-P-OS-Cl 0.75 2 0 OS-P-OS-C3 0 2 0 O-P-OS-C2 0.25 3 0 OS-P-OS-C2 0.25 3 0 O-P-OS-C2 0.75 2 0 OS-P-OS-C2 0 2 0 0-P-OS-CH 0.25 3 0 OS-P-OS-CH 0.25 3 0 0-P-OS-CH 0.75 2 0 OS-P-OS-CH 0 2 0 0-P-OS-CTg 0.25 3 0 OS-P-OS-CTg 0.25 3 0 0-P-OS-CTg 0.75 2 0 OS-P-OS-CTg 0 2 0

aaond length in 'A. bBond stretching parameter in kcal/mol.Jl.2. tBond angle in deg. dBond angle deformation parameter in kcal/mol.deg2.

'Torsional harrier in kcal/mol. 10nly in FF2. gOnly for all-atom approach.

T aking into account the well-known predominance of the P=O bond over P-0 honds in determining phosphate çonformations.11 a second change produced a force field (FF2) without specific twofold potentials for O-P-0-C fragments (see Table 11). This results in a tendency for gauche orientation relative to the P=O bond. instead of the earlier preference of gauche orientation to the P-Oester honds. However. the FF1

results already showed conformational effects due to the P=O bond. and the use of F F2 did not produce large changes in the geometries. For the FF2 calculations. we did

34

find a lower total energy (since one torsional energy term is removed). but stabilities remained essentially the same when expressed as the total interstrand energy. In the case of unmethylated systems. FF1 is identical to the initial force field.7 and we present therefore the FF1 results. We feel that the FF2 force field gives a more accu­rate description for the methylated duplexes, and discuss FF2 results for these sys­tems. lt must be stressed however. that both force fields have been used on all sys­tems. and that the resulting geometries and stabilities are virtually independent of the choice of force field.

RESUL TS AND DISCUSSION

Parallel d(TpTpTpTpTpTh

The parallel input structure contracted appreciably (ca. 25% in volume) during the energy optimization to fit the T-T hydrogen bonds. The resulting structures for the two unmethylated duplexes (with and without counterions) and the two methy­lated duplexes (Rp and Sp configurations) display conformations that are well within the ranges for standard B DNA. as can be seen in Table 111 and Figure 3. The back­bone conformations /3 1 and y+ are found, as well as Cr-exo deoxyribose puckers (P ::::::: 122°). which agrees with the experimental data. Aside from end-effects. the hel­ices are very regular and highly symmetrie with respect to the helix axis with two identical grooves. A rather large helical repeat angle of ca. 40". or a nine-fold helix symmetry. also agrees with the earlier preliminary model studies. which suggested an eight-fold symmetry in this system.

The energy contributions (see Table IV) clearly show a diminished electrostatic energy after methylation. Especially the long-distance term E~~L is responsible for this decrease. which reflects the removal of phosphate-phosphate repulsions. The interstrand interaction energy (Einter). which we use to quantify the stability of the duplex. decreases ca. 18 kcal/mol for the counterion systems. and ca. 41 kcal/mol after methylation of the phosphates. Therefore it can be concluded that shielding of the phosphate charges indeed leads to a more stable duplex. Though all four sys­tems have a negative interstrand energy. analysis of the stability with various values of E indicates that only the neutra! duplexes remain stable (see Table V) when elec­trostatic interactions are enhanced at low E. In the unmethylated systems. the increased electrostatic repulsion destabilizes the duplex. even when counterions are present. For the methylated systems. the stability even increases on going from E = 4 to E = 1. indicating that electrostatic attractions are dominant.

In an interaction diagram ( see Figure 4). it is seen that the T-T base pairs have a binding energy of 10 kcal/mol. and stacking accounts for 6 kcal/mol (intrastrand)

35

1 cl

Figure 3. Stereo-view of the minimized structures for the Rp-methylated (a). un­methylated (b). and Sp-methylated (c) d(TpTpTpTpTpT)i duplexes.

and 3 kcal/mol (interstrand). This does not change significantly upon methylation of

the phosphates. Addition of counterions has only a small effect on interstrand

phosphate-phosphate (P-P) repulsions directly across the duplex. but methylation

reduces P-P interactions significantly. Due to the specific geometry of this parallel

helix (helix diameter 18 A. rise per turn 29 A. 8-9 residues per turn. rotational two­

fold symmetry). P-P distances behave in an unusual way. Numbering the nucleotide units (with the 5' -phosphate included in the unit13) as follows:

1 T T 7

2 T T 8

3 T T 9

4 T T 10

5 T T 11

6 T T 12

it was seen that interstrand P-P distances decrease for the series 1-7. 1-8. 1-9. and

1-10. after which they increase for 1-11 and 1-12. This means that P-P repulsions

between 1 and 8. 9. 10 are more important than the 1-7 interaction on the same level

in the helix. In Fifure 4. it can be seen that the diagonal P-P repulsion is indeed usu­

ally stronger than the horizontal one. In standard B DNA (helix diameter 23 À. rise

36

Table 111. Conformational characteristics'l of the d(TpTpTpTpTpT)2 systems.

System

Unmethylated -69.5 Counterions -68.1 Rp-methylated -63.1 Sp-methylated -61.5

167.0 167.3 165.2 164.1

65.4 63.0 63.0 61.6

115.3 114.5 116.0 112.0

181.9 182.0 184.6 183.0

-89.5 -125.0 124.5 40 40.4 -88.4 -131.6 120.6 40 38.3 -95.7 -125.5 121.4 40 40.8 -88.4 -124.7 122.8 40 40.9

•Average of the fout inner base-pairs. bTorsion angles in degrees. Definitions are according to IUPAC-IUB recommenda­tions.12 'Pucker phase and pucker amplitude in degrees. according to the pseudorotation concept of Altona and Sundaralingam .13

dHelical repeat angle in degrees.

Table IV. Total energies and energy contributions in the d(TpTpTpTpTpT)2 systems.

System

Unmethylated -321.00 5.36 47.19 103.73 62.71 -177.52 -563.50 206.76 -5.75 -67.20 Counterions -519.52 5.21 46.44 105.06 62.37 -189.86 -562.85 18.95 -4.84 -95.27 Rp-methylated -343.28 5.39 46.94 122.89 66.16 -185.01 -428.45 34.63 -5.84 -108.83 Sp-methylated -334.97 5.30 46.87 128.87 64.91 -181.62 -428.45 35.97 -6.83 -107.89

•Total energy. All energies in kcal/mol. 0Bond stretching energy. 'Angle deformation energy. dT orsional energy. . •vicinal van der Waals energy.

1Non-bonded van der Waals repulsion energy. gVicinal electrostatic energy. hNon-bonded electrostatic energy. iHydrogen-bond energy. 11nterstrand energy.

per turn 33 A. 10 residues per turn. strands form a major and minor groove). the interstrand P- P distance is always smallest between units on the same level in the helix. and increases uniformly for phosphates in nucleotide levels on different levels. For the slim parallel helix it is therefore necessary to eliminate P-P repulsions over a length of many base-pairs to achieve stability.

Concerning the differences between Rp and Sp configuration on phosphorus. it can be seen that the stabilities are almost equal (see Table IV). However. certain conformational differences between the two methylated duplexes are apparent,

37

Table V; lnterstrand energy• of d{TpTpTpTpTpT)1 duplexes as a func­tion of e.

System E~ R;j e=4 E=2 E 1

Unmethylated -67.20 66.14 177.79 401.09 Counterions · ·-95.27 4.98 56.81 160.47 Rr-methylated ~108.83 -69.57 -93.39 -141.01 Sp-rilethylated -107.89 -65.84 -84.04 -120.44

•in kcal/mol

.s~:'?:.3~--~s-_.•/. L....;-.J'. ""' ·i__;._J "t" ;"Y . "" ,,,1.:" ' ~

P "e -ais: ."'~ '-&.20 :\ P

120/'~ s 076rr--i-::: .~·~:'~rf-,01s s~ : / --i..__:__r -9.S2 ~ '\. , 1.31 ' P:~::-·-- ------ ---······--P

- - ... """ - 1.72 / - - - - s lol ---------~P

" 5~:'!'-JL.~s·\> /y ~'""~,'>!~~~ ~ P ".> -6.2c: .·"<~ . :·6.01 ",} P :"'* . .· -~~· :'y

211' So.89~--·····' T ··'®s . . --i..__:__r -10.28 "

~ ( _ - - - - - - - - !·'!.3 - - - - - •••. - .•• p ---- /

--- ... "1.~a" s lbl ___ -----~p

s~·'• T -:èo:i'.•--~· T o.e•s b " ~' Y.e .:p "" '),,~~ {o

~ , -ö.J2: >::; :~a30 '" P ~ . . -~ . :r

095: oso.ea~: .•.. :x-~s . : / --i..__:__r -·.·." ~ '

1 004 " P - •••••• - ••• - ••• - • - •• - - - •• ·P

• - - - - - - - • - _o?~ • s/ [cl ·-·----.::,P

Figure 4. Energy interaction scheme for the four d(TpTpTpTpTpTh systems: unmethylated (a). unmethylatéd with counterions (b). Rp-methylated (c). and Sp­methylated {d). For the unmethylated structure with counterions. the Na+ ion is in­cluded in the phosphate group.

especially for the~ (P-05·) and~ (P-03·) honds (see Figure 5). In the all-atom calcu­lations. the tight parallel helix has little conformational freedom. and differences are small (up to ca. 8°). Still it is clear that Rp-methylation influences the~ and ~ con­formations. while in the Sp form mostly changes around the a-bond are seen. The

fact that the chirality of phosphorus is important in determining the conformation of the DNA helix led us to investigate the effects of methylation in d{GpCpGpCpGpC) 2.

We feel that methylation may be relevant in studying complexation of proteins with

38

0:

/3 }'

s t'

~ -8 0 8 -8 0 8

Rp Sp

Figure 5. Changes in the backbone torsion angles in Rp- and Sp-methylated d(TpTpTpTpTpT)2 relative to the unmethylated duplex.

backbone phosphates (vide supra). lf interaction proceeds with one phosphate oxygen only. this could then be translated into stereospecific conformational changes.

Antiparallel d(GpCpGpCpGpC)2

The antiparallel input structures did not change much during energy refinement. T able VI shows that conformations for all four systems are in the expected range for right-handed helices.

Table Vl. Conformational characteristics• of the d(GpCpGpCpGpC,)2 systems.

System

Unmethylated Counterions Rp-methylated Sp-methylated

-72.6 -69.7 -72.6 -57.9

178.2 174.5 171.4 172.4

61.5 58.5 60.7 57.9

143.1 127.4 136.9b 139.9

183.6 184.4 187.8 180.6

-113.7 -115.1 155.6 41 38.6 -99.3 -123.1 136.1 39 35.1

-102.2 -112.8 155.4b 39 36.8 -115.6 -120.0 154.0 40 41.9

"Average of all base-pairs. See also footnotes under Table 111. bThe cytosine sugar torsion angles are excluded. due to high variability (P and 8 range from 97" to 140").

Sugar puckers are C2·-endo (P :::::: 145°). and helical repeat angles indicate a nine- to ten-fold symmetry. The effect of phosphate-methylation is a decrease of non-bonded electrostatic energy and a stabilization of the helix of ca. 34 kcal/mol (see Table VII). Additlon of counterions leads' to à stabilization of ca. 8 kcal/mol. In an interaction

39

Table VII. Total energies and ene~gy contributions in the d(GpCpGpCpGpC)2 systems.

System

Unmethylated -523.55 4.60 50.23 110.57 59.29 -185.73 -1114.04 553.09 -1.57 -121.20 Counterions -710.63 4.29 46.05 109.79 60.93 -200.91 -1115.08 386.21 -1.92 -129.22 Rp-methylated -539.04 4.23 48.65 124.39 63.83 -196.99 -977.09 395.82 -1.88 -153.33 Sp-methylated -519.58 4.49 48.18 142.57 63.34 -197.45 -979.32 402.08 -1.46 -158.00

•Energies in kcal/mol. See also footnotes under. Table IV.

diagram (see Figure 6). it can be seen that the interstrand phosphate-

~ -2M3 .r-;:;-L__ s L...;'.......J' ·······L!-i s / . ··.,-;~· : "'

p -1~87· ~" 'q , ... 6.31 p

"'s~""' ~ -:~!·;;i:~~s(· • " ... ~~y\f. ~ "" ,~~ • {o

p "10 .• u: ·." ~ ,~ ~-H»-41 • P

: ~" 270 : ·" ·-~~o:=r-· l.9! o/" o ss • S

2"

0rc-i. · · · · · · ·' G S . : ~ -22.00 ""'

;.( .•• -- ••••• o;•~ ••••. --··/·· P

- - - • - - • - 0.12 s

~~1-····-- .. ~p Figure 6. Energy interac:tion scheme for the four d(GpCpGpCpGpC)2 systems: un­methylated (a). unmethylated with counterîons (b). Rp-methylated (c). and Sp­methylated (d). For the unmethylated structure with counterions. the Na+ ion is in­duded in the phosphate group.

phosphate repulsions decrease appreciably in the methylated systems. As was . dis­cussed before. P-P interactions on the same level are more important than P-P

40

repulsions between units on different levels (vide supra). The guanine and cytosine bases have an interaction energy of ca. 22 kcal/mol. and stacking interactions follow the usual pattern.14 Guanine on the 5' -side of cytosine gives an interaction of ca. 10 kcal/mol. and guanine on the 3'-side of cytosine gives ca. 7 kcal/mol. There is a difference between the Rp- and 5.,.-methylated forms concerning the stacking interac­tions. since in the Rp-form the intrastrand stacking is stronger by 1 kcal/mol. This was suggested earlier by Kan et al.15 for methyl phosphonate dinucleotides on the basis of model building: We can now offer an explanation for this effect (vide infra). Analysis of Einter with various values of E {see Table VIII) shows that only the methylated systems remain stable. while the unmethylated systems becorne unstable at low E. as was also seen for the parallel systems.

Table VIII. lnterstrand energya of the d(GpCpGpCpGpC)2 duplexes as a function of E.

System E= Rij E=4 E=2 E=1

Unmethylated -121.20 48.61 143.23 332.58 Counterions -129.22 18.64 81.73 207.90 Rp-methylated -153.33 -69.90 -94.47 -143.6 Sp-methylated -158.00 -71.71 -93.21 -136.15

aln kcal/mol.

Focusing upon the different conformations of the Rp and Sp-methylated

duplexes. a clear effect on ( resp. °' can be seen compared to the unmethylated sys­tem (see Figure 7). since in the d{GpCpGpCpGpC)2 duplexes more conformational

freedom is present cornpared to the parallel helices (changes up to ca. 16°). Again we see the correlations R,,-( and Sp-Oi. These results are in good accordance with the

expected conformational changes upon introducing a P=O bond. Depending upon which phosphate oxygen bears a methyl group. one of the P-Oester bonds tends to change its unfavourable conformation with the 0-C bond trans to the P=O bond11

(see Figure 8). lt is seen. that this causes a specific relation between chirality and conformational change. The overall effect of the conformational changes around Oi or ( bonds is obvious when one takes into account the fact that the (-bond is located along the helix axis, whereas the Oi-bond is nearly at right angles with the axis. Changes around ( will therefore lead to longitudinal motions. and variations around °' in transversal motions. As can be seen in Figure 9. these effects are indeed present. In the Rp-methylated system, a small transversal shift for each base-pair is seen. which is the reason for the differences in stacking interactions between the Rp and Sp forms (vide supra). A different and larger effect is observed in the 5.,.-methylated duplex.

41

-16 0 16 ~16

Rp Sp

Figure 1 · Changes in the backbone torsion angles in Rp- and Sp-methylated d(GpCpGp.CpGpC)2 relative to the. unmethylated. duplex.

Rp

/C3' 03• 1

rP-o--·C5' pro -Sp o·· J s·

pro -Rp o .

Sp

Figure 8 .. Expected changes in the a and t honds upon Rp- or Sp-triesterification of the phosphate.group. Both C3' and Cs· prefer a gauche orientation to the P=O bond. ··

where a longitudinal change in the backbonès occurs, The stran{ls approach each óther. since P-P repulsions. ;are virtually eliminated; and a widening of the major groove results from 12 )\. to 13 A.. The minor groove diminishes in size from 6 A. to 5 A.. because of this process. Schematic descriptions of these effects .are shown in Fig­ure 10.

In a model of right-handed DNA. the pro-Sp oxygen in the phosphate group is

seen to be located at the periphery of the helix; whereas the pro-Rp oxygen resides in

the major groóve. This suggests that the pro-Sp oxygen will be the specific recogni­

tion site for protein interactions. as is confirmed by experiments with the Ada

enzyme frorn E. coli.16 îhis protein rétnoves only methyl groups from Sp'"rnethylated

phosphates in ds-DNA. lf one assurnes that the pro-Sp oxygen will be shielded by a

protein in the d(GpCpGpCpGpC)i duplex. conformational changes in the a-bond are

expected. According to C>ur results, this wilt then lead to a widening· of the major groove of the helix.

42

lal lbl

Figure 9. Overall changes in the d(GpCpGpCpGpC)2 helix for Rp-methylation (a) and Sp-methylation (b). Drawn lines depict the unmethylated structure. while dashed lines refer to the methylated structures.

major groove

minor groove

3'

'

s·

'

s·

l 3•

' ' ' .

Figure 10. Schematic description of the changes in the grooves for the Sp­methylated helix structure. The positions of the phosphate groups are indicated with • for the unmethylated system, and 0 for the Sp-methylated duplex.

43

CONCLUDING REMARKS

The calculated structures and stabilities of the parallel helix of d(TpTpTpTpîpT)i are in good agreement with our earlier experimental data.

Methylation of the phosphate groups increases the stability of the slim helix by elimi­

nation of the interstrand phosphate-phosphate repulsions. The difference between Rp

and Sp configuration in the phosphate triester are mainly reflected in the ~ resp. a'

conformations. As we found experimentally. the stabilities and the (3. y. and sugar conformations (which can be determined with 1H NMR) are virtually not influenced by the orientation of the methyl group. A different situation is found for phosphate triesters with ethyl groups.17 where the orientation of the more bulky ethyl moieties does have a strong influence on the duplex formation.

In antiparallel DNA. phosphate-methylation also induces specific conformational changes dependent on phosphate configuration. Considering this neutralization of the

phosphate groups as a model for protein-DNA interactions .. we conclude that com­plexation with the (best available) pro-Sp oxygen results in a larger major groove (ca.

10%) and smaller minor groove (ca. 20%) in the d(GpCpGpCpGpC) 2. This process

may be important for DNA~protein recognition. since the nucleobases are now more

exposed to the surroundings of the helix. Oligonucleotide systems with different length and sequences wilt be investigated to see the genera! impact of phosphate­shielding on DNA conformation5,

REFERENCES

1. L.H. Koole. M.H.P. van Genderen. H. Frankena. H.J.M. Koeken. J.A. Kanters and H.M. Buck. Proc. !(. Ned. Akad. Wet" Ser. B 89, 51 (1986) (communicated by H.M. Buck at the meeting of Nov 25. 1985).

2. L.H. Koole. M.H.P. van Genderen and H.M. Buck. H.M" J. Am. Chem. Soc. 109. 3916 (1987). '

3. W. Saenger. "Principles of.Nudeic Acid Structure". Springer Verlag. New York. 1984. pp. 310-311.

4. R.S. Cahn. C. lngold and V. Prelog'.Angèw;·(hem. 78, 413 (1966).

5. Ref. 3" pp. 399-402.

6. P.K. Weiner and P.A. Kollman. J. Comp. Chem. 2, 287 (1981).

7. S.J. Weiner. P.A. Kollman. D.A. Case. U.C. Singh. C. Ghio. G. Alagona. S. Profeta. Jr. and P.K. Weiner. J. Am. Chem. Soc: 106, 765 (1984).

8. S. Arnott. P. Campbell-Smith and P. Chandresekaharan. in "CRC Handbook of Biochemistry and Molecular Biology". Nucleic Acids. Vol. 2. G.O. Fasman. Ed"

44

CRC. Cleveland OH. 1976. pp. 411-422.

9. Copyright: Chemica! Design Ltd .. Oxford.

10. "Molecular Structures and Dimensions". Vol. Al. 0. Kennard. D.G. Watson. F.H. Allen. N.W. lsaacs. W.D.S. Motherwell. R.C. Pettersen and W.G. Town. Eds .. A. Oosthoek's Uitgevers. Utrecht. 1972. pp. 274-276.

11. D.M. Hayes. P.A. Kollman and S. Rothenburg. J. Am. Chem. Soc. 99, 2150 (1977).

12. IUPAC-IUB Commission on Biochemica! Nomenclature. Eur. J. Biochem. 131, 9 (1983).

13. C. Altona and M. Sundaralingam. J. Am. Chem. Soc. 94, 8205 (1973).

14. P.A. Kollman. P.K. Weiner and A. Dearing. Biopolymers 20, 2583 (1981).

15. L.S. Kan. D.M. Cheng, P.S. Miller. J. Yano and P.O.P. Ts'o, Biochemistry 19, 2122 (1980).

16. M.R. Hamblin and B.V.L. Potter. FEBS Letters 189, 315 {1985).

17. R.C. Pless and P.O.P. Ts'o, Biochemistry 16, 1239 (1977).

45

CHAPTER4*

Parallel phosphate-methylated mini~duplexes for Sp- and Rp­d(T p T). and exclusively for Sp-d( CpC) and Sp-d(T pC}

lmplications for stereospecific complexation of poly-L-lysine and poly-L-ornithine with natura! d(T10)

and d(C10). inducing parallel duplexes

ABSTRACT

A conformational study of the phosphate-methylated dinucleotides d(CpC) and d(TpC) has revealed that only for the Sp diastereomer a parallel mini-duplex is

formed. whereas phosphate-methylated d(TpT) forms a parallel mini-duplex for both

Rp and Sp chirality. This difference is attributed to unfavourable s.teric jnteractions of

the Rp methyl group with the C-C base pair. which has a larger propellor twist angle

than the T-T base pair. Parallel duplexes could also be induced for the natural oli­

gothymidine d(T 10) by complexation with polycationic proteins. such as poly-L-lysine

and poly-L-ornithine. which partially shield the phosphate charges. lt was concluded

that poly-L-lysine uses only the sterically most available Os phosphate oxygens.

while poly-L-ornithine must also complexate with the sterically less available OR oxy­

gen in one of the DNA strands. The stability of the parallel duplex. expressed as enthalpy and entropy changes of dissociation. is not influenced by the different

stereospecificity of complexation. Both for poly-L-lysine and poly-L-ornithine it is

found in molecular mechanics and NOESY studies that the two d(T 10) strands in the

duplex have a different conformation and flexibility. For natural oligomers containing cytosine bases. viz. d(C10). d(C6T6). and d(T6C2T2). it is seen that duplex formation

occurs exclusively after complexation with poly-L-lysine. Based on the results obtained with the phosphate-methylated systems it appears that involvement of OR

in complexation with poly-L-ornithine causes unfavourable steric interactions between ammonium groups and C-C base pairs. which preclude duplex formation.

*M.H.P. van Genderen. M.P. Hilbers. P.J.L.M. Quaedflieg. L.H. Koole and H.M.

Buck. J. Am. Chem. Soc" submitted for publication.

46

INTRODUCTION

Re<::ently. we have demonstrated that methylation of the phosphate groups in oligothymidine fragments results in the formation of a duplex structure with thymine-thymine (T-T) base pairing.1 This duplex shows parallel backbones. i.e .. the 5' - 3' vectors run in the same direction. Experimental1 and theoretica12 studies have revealed that the parallel duplex is right-handed. and can only exist when phosphate­phosphate electrostatic repulsions are diminished. Atso. it was found that the chiral­ity of phosphorus. which arises from methylation. does not influence the structure or stability of the parallel duplex. For the phosphate-methylated dinucleotide d(TpT) it was determined that both diastereomers form parallel mini-duplexes that dissociate at 3Q°C.1b

lnterestingly. in the X-ray crystal structure of 2' -deoxycytidine3 a parallel dimer was found. involving two N_.- H··· N3 hydrogen honds between the cytosine bases. This suggested that parallel duplexes based on cytosine-cytosine ( C-C) base pairs are also feasible. We investigated this possibility with UV and 1H NMR spectroscopy on the phosphate-methylated dinucleotide d(CpC). which indeed forms a parallel mini­duplex in aqueous solution. but exclusively for the Sp configuration. 4 The Rp diastereomer was found to be present in the single strand form only. Molecular mechanics studies show. that the outward orientation of the phosphate methyl group in the Sp form of d(CpC) easily allows duplex formation. whereas the Rp methyl group encounters unfavourable steric interactions in a duplex with C-C base pairs. For parallel duplexes with T-T base pairs. it has been shown that both outward and inward location of the methyl group are free of steric hindrance.2 which leads to an equal stability for both Sp and Rp chiralities. For the phosphate-methylated dinucleo­tide d(TpC). it was established that duplex formation only occurs for the Sp

diastereomer. which shows that steric interactions of the Rp methyl group with the C-C base pair also diminish the duplex stability in this case.

Phosphate charges can also be shielded without chemica! modification by means of proteins. in which amino acids have side chains with a positive nitrogen atom that can complexate with one of the oxygen anions in the phosphate group (e.g., lysine. arginine. or histidine). We therefore investigated the connection between the phosphate-methylated systems and complexes of DNA with polycationic proteins. For these studies. it is ne<::essary to use oligo- or polypeptides instead of monomeric amino acids. since the longer proteins can use the DNA as a template. This is more favourable. since then a protein loses less entropy upon adopting a helical conforma­tion in complexation. We will show in this chapter. that the partial shielding by the polycationic protein poly-L-lysine indeed results in the formation of parallel duplexes with T-T and C-C base pairs for oligothymidines and oligocytidines. With poly-L­arginine. no duplex formation was observed for oligothymidines.5 This shows that

47

only localized positive charges. such as the terminal ammonium groups in lysine. can be effective for duplex formation. while the delocalized positive charge in the terminal

guanidinium fragment of arginine is not sufficient.

The impact of the phosphorus configuration on the duplex formation of

phosphate-methylated d{CpC) and d{TpC) suggested. that parallel duplexes with C­C base pairs can be used to investigate the possibfüty of stereoselectivity in protein

complexation with the two phosphate oxygens (denoted 0 5 and OR when attachment

of a methyl group leads to an Sp or an Rp configuration. respectively). For this pur­

pose. we selected two model proteins. poly-L-lysine and poly-L-ornithine. which pos­sess terminal ammonium groups in the side chains. lt is found with 13C NMR spec­

troscopy and molecular modelling studies that poly-L-lysine is located in the groove of the parallel duplex. and that the lysine side chains point toward both DNA strands. allowing the terminal ammonium groups to complexate with the oxygens of all phosphate groups in the duplex. lt will be. demonstrated with 31 P NMR spectros­copy and model building that this complexation ·· always occurs with the sterically

favourable Os atom. allowing the formation of parallel duplexes with both T-T and

C-C base pairs. Poly-L-ornithine. for which the side chains are one methylene unit

shorter. spans a smaller distance than poly-L-lysine. and therefore complexates with the Os atoms in one strand of the duplex. and with OR atoms in the other strand.

The latter interaction leads to a steric hindrance of the ammonium group in the

groove of the duplex. These steric interactions are analogous to the ones found for the methyl group in the Rp configuration of the phosphate-methylated systems

d(CpC) and d(TpC). The steric interactions of the ammonium groups. that are indirectly linked to OR, and the methyl groups directly bound to OR can only be

accommodated in parallel duplexes with T-T base pairs. and preclude parallel duplex formation with C-C base pairs.

RESUL TS AND DISCUSSION

1. Phosphate-methylated parallel mini-duplexes

Parallel mini-duplex for the Sp form of phosphate-methylated d(CpC)

The phosphate-methylated dinucleotide d(CpC) was selected in order to investi­gate whether C-C base pair formation also leads to a parallel duplex. We first per­formed UV hyperchromicity experiments. which show a sigmoidal curve for the Sp

form of d(CpC). whereas the Rp form corresponds with a continuously increasing

curve. both in a O.Q1 M TrisHCl/0.01 M MgCl2 (pH 7.5) buffer solution. This indi­

cates a duplex formation exclusively for the Sp chirality. The midpoint of the melting

48

transition (T m value) for the mini-dupex of Sp-d( CpC) was found at 31°C. An almost identical T m value of 3Cl°C has been found for both diastereomers of phosphate­methylated d(TpT).1b The slopes of the melting transitions of Sp-d(CpC) and Rp­and Sp-d(T p T) are also virtually identical. which indicates a similar enthalpy of dis­sociation. Therefore. it can be concluded that T- T and C-C base pairs are equally stable. even though different hydrogen bonds are involved (C=Û···H-N vs. N···H-N). Recent ab-initio calculations of Hobza and Sandorfy6 have shown. that base pair for­mation of cytosines is energetically more favourable than T-T pairing (-21.6 kcal/mol vs. -12.6 kcal/mol). which is mainly due to stronger electrostatic attractions in the C-C dimer. However. we have performed molecular mechanics calculations on T-T and C-C parallel duplexes (vide infra). which show a smaller base-base stacking interaction for C-C duplexes compared with T-T duplexes (for hexamer duplexes: -83.4 kcal/mol vs. -101.4 kcal/mol). The net enthalpy effect of the stronger base pairing and the weaker stacking for C-C base pairs with respect to T-T base pairs is apparently negligible.

The selective duplex formation was observed independently in 300 MHz 1H NMR measurements of the chemica! shift of the H6 proton in the cytosine bases as a function of the temperature.

C:

H~NH2 6 I Il ~

NyN / 3

0

The chemica! shift of the H6 proton is highly sensitive to changes in base stacking. which accompany a duplex4=lcoil transition.7 As is evident from Figure 1. both cytidine residues in Sp-d(CpC) show a sigmoidal curve with a T m of 33°C. whereas continuously decreasing shifts are found for the Rp form.

The presence of N4- H".N3 hydrogen bonds in the mini-duplex of Sp-d( CpC) was ascertained with 600 MHz 1H NMR experiments in which we monitored the amino protons of cytosine in the low-field region of the spectrum in H20. The Rp diastereomer has an amino chemica! shift of 8.0 ppm (sample temperature 4°C). while the Sp form has a more downfield amino resonance at 8.93 ppm, which indicates hydrogen bond formation.8 lncreasing the sample temperature for Sp-d(CpC) resulted in a gradual broadening and upfield shifting of this . resonance. which corresponds with breaking of the hydrogen bonds and increased solvent exchange of the amino

49

7.8 7.8

ö(ppf 1

~ ö(pT'

~ 7.7 ~ 7.7

SP Rp

7.6 7.6

10 30 50 70 10 30 50 70 -rr0 c·I ~r1°c1

Figure 1. 1 NMR chemica! shift vs. temperature profiles for the H6 protons of Sp­d(CpC) (left) and Rp-d(CpC) (right). For both diastereomers. the upper profile corresponds with the dCp residue and the lower profile with the pdC residue.

protons upon dissociation of the mini-duplex. The difference of ca. 1 ppm between

free amino protons and those in hydrogen bonds is in good agreement with the values

found for standard base pairs involving cytosine.9

In order to determine the conformation of the 2' -deoxyribose rings and the C4·­

C5· (y) and C5·-05· (13) bonds in the backbone in the mini-duplex of Sp- and Rp­

d(CpC). we first measured the full set of vicinal 1H-1H and 31 P-1H coupling con­

stants. Assignment of the protons of the backbones and 2' -deoxyribose rings was

based on homonuclear irradiation experiments. Subsequently. the non-exchangeable

base protons were assigned with a set of one-dimensional NOE measurements. The

complete coupling data (see Table 1) have led toa detailed conformational picture. by

using published procedures to calculate conformational equilibria from the coupling

constants.10 The results of the conformational analysis closely resemble those of the

parallel mini-duplex of phosphate-methylated d(TpT).1b i.e .. dominant populations

exist for y+ around the C4·-C 5· bonds. {3 1 around the central C5·-05· bond. and south

puckers for the sugar rings11 (see Table 1). In order to assess the orientation of the

cytosine bases with respect to the attached sugar rings. we performed one­

dimensional NOE experiments. where the increase in intensity of a resonance upon

irradiation of another peak .is indicative for the distance between the corresponding

protons. For both the dCp and pdC residues. it was found that the NOE effect on H6

was negligible upon irradiation of Hr. whereas irradiation of Hr (increase 12% for

dCp and 14% for pdC) and H3• (increase 6% for dCp and pdC) clearly led to NOE

effects. lt fol lows therefore that H6 is located on the endo side of the 2' -deoxyribose

ring. so the cytosine bases reside in the standard anti conformation.

50

Table 1. Vicinal 1H- 1H and 1H- 31P coupling constants (in Hz) measured at 600 MHz. and the calculated conformational equilibria of the )' (C4·-Cs'). Il (Cs·-Os·) bands and the 2' -deoxyribose units of bath diastereomers of d(CpC).

Sp Rp dCp pdC dCp pdC

Jn· 8.0 7.4 7.6 7.2 Jr:r 6.0 6.2 6.0 6.4 J2'3' 7.2 7.2 7.2 7.2 Jn· 2.6 3.1 3.0 3.0 J2'2" -14.8 -14.6 -14.4 -14.8 Jn 4.8 4.0 Jn· 3.2 2.3 4.0 3.0 J4·5· 4.0 2.2• 3.6 1.8• J4•5" 4.0 3.oa 3.6 4.2• Jp; 5.oa 6.3• Jp;· 7.ia 5.9" Js·s" -12.4 -11.0• -12.4 -10.9•

x(y +) 0.57 0.78 0.65 0.69 x(yt) 0.24 0.22 0.21 0.31 x(y-) 0.19 0.00 0.14 0.00 x(ll +) 0.23 0.17 x(tl') 0.64 0.64 x(1r) 0.13 0.19 x(south) 0.84 0.78 0.81 0.77

•obtained by iterative simulation of the non-first order pattern of l-J4·/H5·/Hs·"

Dimerization of cytosine bases can occur only via symmetrie N4-H··· N3 hydrogen bonds.8·12 and therefore always results in identical magnetic environments for the two strands. This can be seen in the 600 MHz 1H NMR spectrum. where single. degen­erate resonances are present for all protons. Due to the exact equivalence of the two strands. two-dimensional NOE spectroscopy (results not shown) cannot contribute to the structural information. since each interstrand NOE has a stronger and overlap­ping intrastrand contact.1b lt can be concluded from the present data. that the phosphate-methylated dinucleotide d{CpC) forms a parallel mini-duplex for the Sp

form only. An indication that this selective duplex formation also occurs in longer cytidine oligomers is found in our earlier observation that no duplex was formed for the phosphate-methylated hexamer d(C6). which was present as a random mixture of phosphate configurations.1b Methylation of the phosphate groups of d(C6) was per­formed by reaction of methyl methanesulfonate ~ith the single-stranded natural

hexanudeotide with base-protecting groups. and no diastereomeric separation in favour of Sp chirality was performed, Since no measurable fraction of the mixture

showed duplex formation. accommodation of several Rp phosphates is unfavourable

in the hexanucleotide duplex with C-C base pairs.

Parallel mini-duplex for the Sp form of phosphate-methylated d(TpC)

In the mini-duplex of the phosphate-methylated dinucleotide d(T pC). both a C­C and a T-T base pair can be present. We performed UV hyperchromicity measure­ments on both diastereomers. which clearly show that only the Sp form undergoes a

duplexpcoil transition with a Tm value of 25°(. This was independently seen in 200

MHz 1H NMR experiments. where the shift of the H6 base protons of both the thymi­dine and cytidine residue as a function of the temperature displays a characteristic sigmoidal shape with a T m value of 25°C (see Figure 2). Also here. the slope of the

transition and therefore the enthalpy of dissociation is similar to the ones found for the other dinucleotides. The Rp form on the other hand. gave no indication for duplex formation.

7.7

.l(ppm)

t 7.6

7.5

7.4

7.3

,_,.

Sp

s 15 25 35 ,5 55 ---r(°Cl

ó(ppml

t 7.7

7.6

7.5

Rp 7·4 '--~5--1~5--2~5--3-5--4.-5--55-­

- r (0Cl

Figure 2. 1H. NMR chemica! .shift vs. temperature profiles for the He; protons of Sp-d(TpC) (left) and Rp-d(TpC) (right). For both diastereomers. the upper profile corresponds with the pdC residue and the lower profile with the dTp residue.

Apparently. the methyl group in the Rp form is unfavourable enough to disrupt the

T • T base pair. This result èorresponds quite nicely with our earlier finding.1h that a diastereomeric mixture of the phosphate-methylated dodecamer d(C6T6) was present in the single strand form after methylation with methyl methanesulfonate without separation of diastereomers in favour of Sp chirality. (see also the phosphate­

methylated d(C6). vide supra). The presence of Rp configurations in the cytidine part of the dodecamer · predudes C-C pairing. and thereby prevents T-T base pair

52

formation for the six thymidine residues.

The conformational characteristics of the sugar rings and the /3 (C 5,-05,) and y

(C4·-Cs·) honds in both diastereomers of phosphate-methylated d(TpC) were obtained from the 500 MHz 1H NMR spectrum as described for phosphate­methylated d(CpC) (vide supra). The results (see Table 11) show clearly that the standard y+. /3 1• and south conformations are preferred.

Table ll. Vicinal 1H- 1H and 1H-31P coupling constants (in Hz) measured at 500 MHz. and the calculated conformational equilibria of the )' (C"·­C5·) and ~ (C5·-05·) honds and the 2' -deoxyribose units of both diastereo­mers of d(TpC).

Sp Rp dTp pdC dTp pdC

Jn· 6.1 6.7 6.1 6.7 Jrr 6.1 6.7 6.1 6.7 Jn· 6.7 6.7 6.7 6.7 J2"3' 2.4 4.9 2.4 5.1 Jzr -14.6 -14.0 -14.7 -13.4 Jn 3.7 2.7 Jn 2.4 2.7 5.2 3.1 J4·5· 4.9 2.8 4.3 2.4 J4•5" 4.9 5.5 4.3 5.5 Jp5' 6.1 5.5 Jp5'' 5.5 5.5 J5•5" -12.5 -12.0 -12.5 -11.6

X~j' +) 0.39 0.52 0.52 0.55 x y') 0.31 0.45 0.27 0.45 x(y-) 0.30 0.03 0.21 0.00 x(~+) 0.15 0.15 x(~') 0.67 0.70 x(~-) 0.18 0.15 x(south) 0.85 0.57 0.85 0.55

Molecular mechanics studies on phosphate-methylated parallel duplexes

We examined the differences between parallel duplexes with C-C and T-T base pairs closer with AMBER molecular mechanics studies13 on the parallel duplexes of phosphate-methylated oligonucleotides with thymine or cytosine bases. Input struc­tures were used with either Sp or Rp configurations on all phosphate groups. The

53

complete conformational characteristics of the energy-minimized parallel duplexes of phosphate-methylated d(T 6) have been published before.2 and it was found that the

parallel duplexes of d(C6) do not differ much in backbone conformation .. However.

comparison of the b.ase pairs in the energy-refined structures of the four parallel duplexes reveals less favourable stacking interactions (vide supra) and a much larger propellor twist angle in the C-C base pairs (43° for Sp and 39° for Rp) with respect to

the T-T base pairs (29° for Sp and 22° for Rp). As can be seen in Figure 3. the different geomc;itry leads to a narrower groove in the C-C systems (width of groove expressed as phosphorus-phosphorus distance: T-T. 14.4 Ä: C-C. 13.4 À). causing steric hindrance for the Rp methyl group which has an inward location.

3'

Figure 3. Energy-refined structures of the parallel duplexes of Rp and $p phosphate-methylated d(T6) (left) and d(C6) (right}. The methyl groups of the phosphates have been shaded.

Therefore. only the Sp configuration can result in duplex formation for the

phosphate-methylated systems with C-C base pairs. whereas for T-T base pairs both configurations can form stable duplexes.

54

ll. Molecular modelling studies on parallel duplexes of d(T10) with poly-l­lysine and poly-l-ornithine

Model building studies

For reasons of clarity. we will first present the molecular structure of the protein-complexated parallel duplex. as it follows from molecular modelling studies. lt will be shown that the modelling results are completely corroborated by experimental data. Preliminary studies have been performed with the oligonucleotide d(T 8) and

poly-L-lysine in a mixture of lengths.5 We will now use a well-defined model system consisting of the natural decamer d(T 10) and octadecamer proteins. since a duplex structure of two d(T 10) strands contains eighteen phosphate groups. Octadeca-L­

lysine (Lys1s) and octadeca-L-ornithine ( Orn18) are known to complexate with

antiparallel DNA duplexes.14 and differ only in the length of the side chains (one methylene shorter for ornithine). Model building studies show that Lys18 and Orn18

are also well-suited for complexation with the phosphate groups in a parallel arrange­ment. The repeat distance of the ammonium groups along the proteins is virtually identical to that of the phosphates in a DNA strand (both nitrogen-nitrogen and phosphorus-phosphorus distance ca. 7 À). so an arrangement with alternating side chain directions (see Figure 4) results in a precise fit between the ammonium groups and the phosphates in both strands. The DNA strands s1 and s2 will be dis­tinguished. since it can be seen in Figure 4 that complexation by the protein side chains occurs in the 5'-+ 3' -direction for strand s1• and in the 3'-+ 5' -direction for

strand s2. lt was obvious from model building studies that the complexation of the ammonium groups with the phosphate oxygens 0 5 and OR is determined by the

width of the peptide chain (twice the distance from the backbone to the terminal ammonium nitrogen. Lys1a: 13.5 À. Orn13: 10.8 À). Poly-L-lysine can associate with

Os oxygens in both strand s1 and s2 (distance ca. 16 À. see Figure 5). This is favour­able. since 0 5 is sterically the most available phosphate oxygen.2 Poly-L-ornithine. on

the other hand. cannot span a larger distance than from OR in strand s2 to 0 5 in

strand s1 (ca. 13 À. see Figure 5). This suggests a different stereospecificity of com­plexation with respect to strands s1 and s2 for poly-L-lysine ({05.05}) and poly-l­

ornithine ({Os.OR}).

For both model proteins. the backbone conformation must give rise to a helical structure that follows the DNA duplex in the proposed molecular model. With the PROTEIN program (see Experimental Section). we found that a protein backbone can form a helix with a diameter and pitch corresponding to parallel DNA (18 resp. 34 Á). when a two-amino acid fragment as repeating unit is used. in which the back­bone torsion angles have the following values: {$. ro. eb.$. ro. et>} = {173°. -170". -160°. -173°. -170". 160°1.15 In the standard graphics program Chem-X.16 a model of a

55

Figure 4. Molecular model of the parallel DNA duplex of two oligothymidine strands (s1 and s2) complexated with a poly-L-lysine strand.

+

[ ~"' NH3

Figure 5. Schematic representation of the model building studies on the stereospecificity of protein complexation with a parallel duplex. Geometries of the d(T 10)/Lys1s complex (left) and the d(T 10)/0rn13 complex (right).

protein molecule with these backbone torsion angles and completely stretched side chains was built. For this siructure, it is found that half of the side chains (pointing to strand s2) have an outward orientation, which is unsuitable for complexation with the phosphate groups. Bending of the aberrant side chains was performed for oligo­lysines by manipulating the side-chain torsion angles x1• x2• x3• and x4• which refer to the C-C honds in the side chain going from C" to C".15 lnitially, the torsion angles were all 18cr. but it was found that changing x2 and x3 to 60" in the deviating side

56

chains produced the correct positioning of all terminal ammonium groups relative to the phosphate groups. Similarly. for ornithine oligomers it was necessary to use x 1

= 180°. x2 = 60°. and x3 = 60° in the side chains pointing to strand s2. A question which is irrelevant fot antiparallel duplexes now arises. viz. whether the N ....... C orien-tation of the protein chain is parallel or antiparallel with respect to the 5' ....... 3' direc-tion in the parallel duplex. Model building studies have revealed that these two possi­bilities yield an essentially identical protein conformation. and similar DNA-protein contacts. Therefore. we have assumed for the interpretation of the experimental results that the two possible proteîn orientations are identical.

Molecular mechanics calculations on d(T10)/0rn18

The structures of the parallel duplexes complexated with protein as described in the previous section (with anti parallel N ....... C and 5' ....... 3' directions) have been sub­jected to an energy minimization. using the well-known AMBER force-field.13 Particu­

lar attention will be given to the energy-minimized d(T 1o)/ Orn1s structure. since it can be compared with NOESY measurements (vide infra). The conformation and energy interactions of the minimized structure are summarized in Table 111. The most striking feature in this structure is the difference in conformation for the two strands s1 and s2• especially with respect to the deoxyribose rings. In Figure 6. it can be seen

that this leads to a larger overlap for adjacent thymine bases in strand s2• which

results in stronger base-stacking interattions. This implies a lower mobility for the thymidine residues in strand s2• compared with the residues in strand s1. The

differente in statking for the two strands is found to be taused by a displatement of the T-T base pairs away from the helix axis of 0.31 À. As a result. the base methyl group of s1 points into the groove where the Orn18 is located. This structural feature will also be seen in the experimental studies.

The protein tonformation is also changed with respect to the input strutture. The values for © and (/>. differ markedly from the initial structure. while also the side chains are deformed. The peptide bond tends to a trans orientation. whith is known to be most favourable.17 Note that the stritt alternation in the sign of (/> has disap­peared. and that X 1 has actually adopted two alternating values. while the protein remains in a helical conformation.

An analysis of tontatts between the ammonium groups of the protein and the oxygen atoms of the phosphate groups revealed that the protein seletts the OR and

Os phosphate oxygens in the tomplexation with strands s1 and s2. In the energy­

minimized structure of d(T 10)/ Orn18• it is seen that str.and s1 is almost exclusively

tomplexated via Os (87% of the contatts). while in strand s2 a strong preferente

(65% of the tontacts) exists for OR. This is in good agreement with the model

57

Table 111. Energy contributions and conformational characteristics of the energy-minimized structure of the d(T10)/0rn1s complex with $ymmetric input.

Energy Term• h Protein Cónformationc

(kcal/mol) Strand s1 Strand s2 Toward s1 Toward s2

E1ot -2503.56 Q -90.9d -65.9 1/1 157.8 -175.4 Et>ood 16.78 13 170.4d 177.4 "' 173.1 173.1 Eangle 155.81 )' 57.ld 56.2 </> -152.1 -160.6 Edih•d 362.35 8 115.4d · 143,2d x1 -146.3 -169.2 EvdW.14 145.12 E -178.6d -171.0 x2 59.1 -164.5 EEEL.14 -2040.72 t -7.8.7 -77.6 x3 57.3 -130.5 EvdW.NB -410.90 x 240.2 219.4 EEEL.NB -724.15 p 122.1 166.0d EHbond -7.83 'llmax 0.35 0.35 E;n1 -60.55 hra 36.9

twist 19.6

"Energy terms are respeètively: total energy. bond stretching eriergy. angle deforma­tion energy. torsional energy. vicinal van der Waals energy. vicinal electrostatic en­ergy. non-bonded van der Waals .energy. non-bonded electrostatic energy. hydrogen-bond energy. and ioterstrand energy. bsee refS. 11 and 26 for definitions: hra and twist denote the helical repeat angle and the propellor twist angle. respectively. csee ref . .15. dThese values varied appreciably in the residues of the duplex.

building studies (see Figure 5). and supports the proposition that complexation of Orn18 proceeds in a stereospecific manner.

Since the calculations indicate a perturbed twofold axis of symmetry for the d(T 10)/ Orn18 complex. we generated an asymmetrie input structure by rotating one

of the DNA strands 12° around the helix axis. In this way. the groove containing the protein is narrowed. which facilitates complexation of Orn18. The energy-minimization

of the asymmetrie input structure results in a duplex with little residue-to-residue variation (see Table IV). compared with the one obtained from a symmetrie input structure. The conformation of the protein strand is quite simila·r to the previous cal­culations { see T able 111). Evidently. the asymmetrie input is more realistic since it yields a more regular structure. Although the total energy is somewhat higher for the asymmetrie input. it can be seen that the interaction energy between the DNA

strands (Emtl is more favourable by ca. 18 kcal/mol. Since E;01 is a measure of the

stability of the DNA duplex.2 the asymmetrie input structure results in a more stable duplex. and therefore seems a better description of the DNA-protein complex.

58

Figure 6. Top view of strands s1 and s2 in the energy-refined structure of the d(T 10)/0rn1s complex.

Table IV. Energy contributions and conformational characteristics of the energy-minimized structure of the d(T10)/0rn1a complex with asymmetrie in­put.

Energy Term• , DNA Conformation• Protein Conformation•

(kcal/mol) Strand s1 Toward s1 Toward s2

E1ot -2350.56 Cl<' -63.2 -62.3

"' 161.4 179.8

Ebood 16.30 fj 165.5 162.4 "' 178.9 178.9 Eanck' 167.93 ')' 57.4 55.1 </> -169.2 -153.6 ~ihed 330.12 a 119.4 117.4b x1 -151.3 -170.8 EvdW.14 133.90 E 178.3 -170.3 x2 46.0 -158.9 EEEL.14 -2048.26 t -86.3 -71.0 x3 52.4 -173.1 EvdW.NB -346.88 x 234.0 237.0 EEEL.NB -590.16 p 116.0 161.o'> EHbond -13.38 'llmax 0.37 0.37 Ei111 -78.35 hra 37.4

twist 24.7

asee notes under Table 111. bThese values varied appreciably in the residues of the duplex.

5g

111. Experimental studies on parallel duplexes of d(T 10) with poly-l-lysine and poly-L-ornithine

Duplex formation

In the presence of the proteins Lys18 and Orn18• a melting transition is dearly

seen in UV hyperchromicity experiments for the natura! decamer d(T 10) in a 10 mM

TrisHCl/1 mM EDTA (pH 7:5) buffer solutlon. with T m values of 21 and 22°C, respectively (see Table V). By varying the ratio of d(T10) and Lys18. it was deter­

mined that the largest UV transition (and hence maximum duplex formation) indeed occurs for the stoichiometry 2:1 (DNA strands:protein strands). Determination of the enthalpy (àH0) and entropy (àS0) changes in the melting transition (see Table V) shows that both Lys18 and Orn18 give rise to duplexes with almost identical stability.

which shows that complexation with either only Os (Lys18) or both 0 5 and OR (Orn18) has the same effect for T-T duplexes. This indicates that the binding of an

ammonium group to OR does not result in steric difficulties. For the analogous

phosphate-methylated system d(T p T). the same condusion was made for attach­

ment of a methyl group to OR·

Table V. Thermodynamic parameters for the melting transitions of protein-induced parallel duplexes.

System T m (°C)• àH8 (kcal/mol bp)• àS0 (cal/mol bp-K)•

d(T 10)/Lys1sb 21.0 10.0 31.4 d{T 10)/0rn1sb 22.3 10.5 33.0 d(C 10)/Lys18' 25.3 7.6 23.0 d(C 10)/0rn1s' d(C6T6)/Lys1l 20.1· 10.9 35.0 d{C6T6)/0rn1sc d(TGC2T2)/Lys1sb 19.7 8.4 26.0 d (T 6C2 T 2)/0rn18b

•Determined by computer-fitting of the melting curves (see Experimental Section). bMeasured in a 10 mM TrisHCl/1 mM EDTA buffer solution (pH 7.5). 'Measured in a 10 mM TrisHCl/1 mM EDTA buffer solution (pH 9.0).

The presence of a duplex structure was ascertained independently by measure­ments of the 1H NMR chemical shift of the imino proton resonances in d(T10). which

is indicative for hydrogen bonding of the thymine bases. For thymine bases in single strands. it is known that the imino chemica! shift is around 11 ppm.18 while in T-T base pairs chemica! shifts of 13.4 ppm have been determined.1b At 1CY'C. the imino

60

chemica! shift of the d(T 10)/ Lys18 system in aqueous solution is found at 13.0 ppm. which dearly indicates T-T base pairing. No reliable imino chemica! shift is deter­mined for the d(T 10) / Orn18 complex. because of its low solubility in water.

31 P NMR spectroscopy

A solution of d(T 10) in water at room temperature showed a sharp resonance at 1.93 ppm in the 31 P NMR spectrum. which indicates that all phosphate groups in the single-stranded system are identical. Upon addition of Lys18 to the NMR sample. a

more upfield resonance appears at 1.85 ppm.19 in an intensity ratio of 1:0.65 (upfield to downfield).20 The upfield signal has a slightly increased linewidth (5.5 Hz vs. 3.6 Hz for the downfield resonance). which indicates a more rigid structure.21 Therefore the signal at 1.85 ppm is indeed due toa d(T10)/Lys18 complex. The melting transi­tion of this complex could be independently verified with variable-temperature 31 P

NMR spectroscopy. Upon raising the sample temperature. the intensity ratio of upfield (complexed phosphates) to downfield (free phosphates) increases to nearly unity with a characteristic sigmoidal curve (T m = 30 °C, see Figure 7).

0.90 intensity

ratio

0.80

1 0.70

0.60

0 10 20 30 40 50 60 ---T(°C)

Figure 7. lntensity ratio of the upfield to the downfield 31P resonance in the d(T10)/lys18 complex as a function of the sample temperature. for a 2:1 stoichiometry.

Therefore. the protein-induced duplex dissociates in two d(T 10) strands. of which one remains bound to Lys18. The high T m value of 30"C with respect to the UV experi­

ments (21°C. see Table V) is due to the higher concentration (ca. 1 mM) in the NMR sample, which shifts the duplex+:lcoil equilibrium to the duplex form.22

Comparison of the 31 P NMR spectra of the d(T10)/Lys13 and d{T10)/0rn18 com­plexes shows a markedly higher linewidth for the latter complex (7.7 Hz vs. 5.5 Hz)

61

under identical experimental conditions. This finding supports our model of stereospecific protein complexation. because in the case of Orn18 phosphates are

bound via Os (strand si) or OR (strand s2). îhis results in two slightly different

magnetic environments for phosphorus. which give two overlapping 31 P NMR reso­nances. A chemica! shift difference of 0.04 ppm between two resonances with a width of 5.5 Hz results in the observed 7.7 Hz linewidth. lt is of interest that the chemica! shift difTerence between the 31 P signals of Rp and Sp phosphate-methylated dinucleo­

tides is only 0.1 ppm. Therefore the proposed difference of 0.04 ppm between com­plexation with OR or Os seems realistic.

13C NMR spectroscopy

The effects of complexation with the parallel duplex on the protein molecule were studied with 13C NMR. The 13C nucleus was most efficient. since in the 1H NMR spectrum no changes could be observed upon complexation. Changes in the 13C chemica! shifts of Lys18 (see Table VI) show that the backbone atom C°' is deshielded. while the side chain carbon atoms are shielded. The d(T10)/0rn18 com­plex cannot be studied on account of the low solubility (vide supra).

Table Vl. 13C chemica! shifts (ppm) in the Lys18 model protein. before and after complexation with d(T10) at 20"C and pH 7.

Lysis d(T 10)/Lys1s AS (ppm)

c" 54.52 54.71 0.21 cll 31.57 31.49 -0.08 C-y 23.11 23.08 -0.03 Ca 27.34 27.32 -0.02 c. 40.21 40.15 -0.06

While Cy and C8 show virtually no effect. the shielding effect is most pronounced for

Ce. because this nucleus is next to the ammonium group that binds the oxygen atom

in the phosphate group (for numbering scheme. see Figure 4). The deshielding of C°' and the shielding of C13 aré in goód agreement with their proposed locations in the

groove of the duplex (vide supra). Ring currents in the thymine bases create deshield­ing effects in cone-shaped areas (see Figure 8). and shielding effects outside the cones.23 The large downfield shift (deshielding) of C°' shows that it is located in the plane of the thymine ring. i.e" in the center of the groove. where the shielding effect is the largest. C13 is more to the edge o_f the groove (see Figure 8). and is near the

62

border of the cone where only a small shielding effect remains.

s·

l 3'

Figure 8. Zones of shielding (inside cones) and deshielding (outside cones) for py­rimidine bases. 23

These findings support the molecolar structure proposed in the modelling studies, consisting of a protein molecule in the duplex groove with side chains pointing to the phosphates.

One- and two-dimensional 1H NMR spectroscopy on d(T10)/0rn13

The structure of the protein-induced parallel duplex was studied in more detail with one- and two-dimensional 600 MHz 1H NMR spectroscopy. These studies were performed on the system d(T 10)/ Orn18• since it followed from the molecular rnodel­ling studies that in this system protein complexation occurs with both OR and Os atoms. and large differences between the d(T 10) strands are expected. The low con­centration of the complex was sufficient in this case. due to the high sensitivity of 1H NMR spectroscopy. Assignment of the proton resonances was obtained from the phase-sensitive NOESY spectrum in a standard fashion (see Table VII and Figure 9). For the 2' -deoxyribose rings. we followed the path from the Hr signal to thé H2• and Hr resonances (assigned with the fact that the NOE contact between H1• and Hr is stronger than that between H1• and H2·· regardless of sugar pucker: see Table VII). which in turn led to H3•• From there. H4• was identified. and subsequently the H5• and Hs" resonances, which were assigned according to the Remin ·and Shugar rule24

(os· > 8d. For the thymine base. it was found that the H6 protons had strong NOE contacts with the thymine methyl groups. The location of the protein

63

T able VII. Chemical shifts of the proton resonances in the d(Trn)/Orn18 complex for both DNA strands and the protein. as determined from the NOESY experiment.

Chemica! shifts Strand s1 Strand s2

H6 7.32 7.28 Hr 5.à6 5.84 H2· 2.05 1.97 Hr 2.05 2.15 H3· 4.23 4.52 H4·" 3.93 Hs·• 3.77 Hs"• 3.73 CH:rbase 1.51 1.49

3.99 1.43 2.69

•Not seen separately for strand s1.

resonances was already known from one-dimensional experiments. and was found to be virtually unchanged.

As expected from the modelling studies. the same protons have different chemi­ca! shifts in strand s1 or s2. Within one strand. however. all residues are virtually identical. as is seen from the complete overlap of resonances of similar protons even at 600 MHz. The most revealing aspect of the 1H NMR spectrum of the d(T10)/0rn18

complex was the appearance of two equally high H6 signals for the thymine bases with different chemica! shifts upon addition of protein. This suggests a difference in the stacking of the thymine bases in each DNA strand. since the chemica! shift of the H6 proton is highly sensitive for changes in overlap between adjacent bases.7 The strand with the upfield H6 resonance is indicated as s2. The H6 proton in this strand encounters the strongest shielding influence from neighbouring bases.23 which indi­cates a large overlap of thymine bases and therefore a pronounced stacking interac­tion in strand s2• A similar conclusion was reached on the basis of molecular mechan­ics calculations of this structure. so it seems reasonable to condude that the sifs2

assignment of the strands is identical in the model building and the experiment. The difference in stacking is due to a displacement of the base pairs from the helix axis (vide supra). We calculated this effect on the chemica! shift of the H6 proton in both strands. usiog the theory of Giessner-Prettre et al.23 which accounts for the

64

.. -~~.::--=::~.:::.:-::::~.:~r..,:::;:::"" ":-~ j. ·~·:::.

.·

... - - - - - .,.. - - - - ..,. ·- - - - - - - 1- -~ . . " . :· ... :,. : . . . ·:; " .. 1 1 1 1 .• - ·" ...... --· '•· •.• " •• ,,. •. ". " ......... "

.•·. l 1, 1

1

I'

••

1 t._ _..:.. -· - - .Jd------:1-1----::----i'~ 1 •

H4'/5'15"

7.0 6.0 5.0 4.0 3.0 2.0

ó(ppm)

HJ'

S2

1.0

1.0

2.0

3.0

4.0 j 5.0

ö(ppml

6.0

7.0

Figure 9. Assignment of the proton resonances in strand s1 (- - - ) and strand s2 (-)of the d{T10)/0rn18 complex via the NOE contacts.

paramagnetic and diamagnetic shielding as well as ring current effects. The contribu­tions of all base pairs in. a heptamer duplex were used. to get realistic values without excessively long computation times. A displacement of the base pairs of 0.3 Ä leads to the observed chemica! shift difference of 0.04 ppm ( ± 0.002 ppm). which is in perfect agreement with our AMBER calculations (displacement 0.31 Ä. vide supra). As an independent criterion. we also studied the chemica! shift difference between the base methyl groups in the d(T10)/0rn1s complex (0.02 ppm. see Table VII). From the shift calculations it follows that a 0.02 ppm difference for the methyl groups is also due to a 0.3 À. displacement of the base pairs.

Further evidence for a different stacking mode for the two DNA strands was found in the two-dimensional NOESY spectrum. in which it is seen that each H6

65

resonance has its own separate sequence of NOE contacts with other protons (see Figure 9).

Table VIII. Observed NOE contacts and strengths in the d(T10)/0rn1s complex.

NOE contacls DNA NOE contacts DNA

Strand s1 Strand sz Strand s1 Strand s2

Hi;-CH3 2.34 6.63 Hr-H4· 1.23 H6-H1· 0.62 3.07 H3·-H4· 3.95 H6-H2• 2.15 7.03 Hl·-Hs· 2.43 H6-H2" 2.15 4.16 Ha·-Hs" 3.90 H6-H3· 0.22 1.03 HrHs· 5.63 Hc,-H4· 0.58 H41-Hs" 8.15 Hc;-Hs" 0.37 H1·-CH3 0.72 NOE contacts protein Ht'-H2• 0.97 5.16 H1·-H2" 0.97 8.63 Strand St Strand s2 H1·-H4· 2.08 H1·-Hs" 0.85 CH3-H.,. 6.75 H2·-H2" 30.17 CH:r-Hs 1.64 H2·-H3· 1.26 4.40 H.,.-H11," 30.48 30.48 HrH4· 0.68 Hs-H111" 20.49 20.49 Hr-H3· 1.26 2.85

In Table VIII. all identified NOE contacts and their relative strengths are listed. lt is clear that in strand s2• the NOE contacts are consistently stronger. even when no variation in interproton distance is possible (e.g .. for 1H6-CH3 which is a fixed distance in the thymine base). The strength of NOE contacts is known to be determined by interproton cross-relaxation rates. which are related to the correlation time of the interproton vector.25 Therefore. a higher correlation time must be present in strand s2• which is due to a lower mobility of all bonds in this strand. This finding is in good agreement with the above observation that base stacking must be stronger in strand s2• since an increase of stacking interactions results in a more rigid structure. Both experimental and theoretica! data show that s1 and s2 have a different conformation and a different flexibility after protein complexation.

Beside the overall structural features of the complex. we studied the conforma­tions of the 2' -deoxyribose rings and the glycosidic bonds in detail. From the NOE contacts of the protons in the 2' -deoxyribose rings, a sugar pucker analysis was per­formed in terms of the standard south+:t north equilibrium.26 For each conformation.

66

a set of interproton distances was calculated from model building studies. Using the well-known relation between the strength of NOE contacts and the inverse of the sixth power of the interproton distance. we could obtain the equilibrium distribution from the ratios of NOE strengths.27 For these calculations. only NOE contacts were used that correspond to intranucleotide distances. since all internucleotide distances are irrelevant for the sugar pucker. Due to the weakness of the NOEs in strand s1•

the calculation has been confined to strand s2• Determination of the population of the south conformation (xs) from the experimental data resulted in a predominance of the south pucker (see Table IX). Also. the conformation around the glycosidic bond was determined from the NOE contacts with H6. Since the strongest contact is that with H2•. an anti orientation of the bases is virtually certain.

Table IX. Conformational analysis of the 2' -deoxyribose rings in the d(T 10)/0rn18 complex via intranucleotide NOE contacts.

NOE NOE Distance contact strength South (Á)

Distance x~ North {Ä)

H4·-Hr 1.23 3.8 2.3 H4·-H2· 0.68 3.6 3.5 0.90 HrHt' 2.08 2.9 3.3 0.92 HrH3· 3.95 2.6 2.8 0.90

acalculated relative to the H4-Hr NOE.

Both south sugar puckers and anti base orientations are found in the most common right-handed geometry. the B form. and are also seen in our molecular modelling stu­dies. In the other major right-handed form. the A geometry, north sugar puckers must occur.12 A further indication for the presence of a B-type helix geometry in the d(T 10)/ Orn1s complex is the fact that the H6~H3• contact is rather weak compared to the H6-H 2• contact. which is not the case for the A-type right-handed geometry.28

Beside the intra-DNA NOE contacts. we also found cross peaks between pro­tons of the protein and the DNA. The strongest of these was the contact between H°' and the thymine methyl group of strand s1 (see Figure 10). According to the molecular mechanics studies, this methyl group is indeed pointing into the groove of the duplex where Orn111 is located (vide supra). Therefore. this specific NOE contact in detail corroborates the structural model proposed in this paper.

In summary. the theoretica! and experimental results show that the d(T10)/0rn13 complex consists of a right-handed parallel duplex with T-T base pairs. The Orn18 molecule is located in one of the grooves of the duplex. and induces different flexibilities for the strands s1 and s2. Complexation occurs with the

67

~ @> ·c:::> 3.5

• -<J c:? <Z ~

;Q

~ j • ~ C1 ! ....... Ha 4.0

olppm) 4.2

4.4

1.8 1.7 1.6 1.5 1.4 1.3 1.2

o(ppm) ----

Figure 10. NOE contact between H"' of Orn1s and the base methyl group of thy­mine in strand s1.

phosphate oxygen Os in strand s1• and wîth OR in strand s2•

One-dimensional 600 MHz 1H NMR spectroscopy on d(T10)/Lys1s

The difference between the strands s1 and s2 is also studied in the d(T 10}/ Lys1s complex. The chemica! shifts of the two H6 protons in the thymine bases are found

0.028 ppm apart. which is smaller than the chemica! shift difference in d(T 10)/ Orn18

(0.04 ppm. vide supra). Using the chemica! shift calculations described before. it fol­lows that a base-pair displacement of 0.18 Ä must be present in the d(T10)/Lys18

complex. This small value compared with 0.3 Ä for d(T10)/0rn18 shows that the

strands s1 and s2 differ less after complexation with Lys18• which agrees with the fact

that only 0 5 is used in both strands.

An independent confirmation for the fact that complexation with Lys18 leads to

a different conformation and flexibility of the oligothymidine strands in a parallel duplex is found in the work of Bobst et al.29 on spin-labeled poly(dT). The spin label allows one to investigate the DNA strand with the EPR technique. and a mobility can be deducèd for the molecule from the intensity ratio of the resonances from the spin label. In their studies. Bobst et al. found two different mobilities in the poly(dT) after

68

cornplexation with poly-L-lysine. which they explained by the existence of regions with different mobility within one strand. lt now appears obvious that the different mobilities they observed reside in the two different strands of a protein-induced parallel duplex of poly(dT). which is formed upon addition of poly-L-lysine to poly(dT).

IV. Experimental studies on protein-induced parallel duplex.es with C-C base pairs

Parallel duplex for d(C10} with Lys13 exclusively

For the study of parallel duplexes with C-C base pairs after protein complexa­tion. we used the model compound d(C10). to allow comparison with the results from the d(T 10)/protein complexes. The natural decamer d(C10) was studied with UV hyperchromicity measurements at pH 9. where a single strand form was clearly observed.8·30 In contrast to the results with d(T 10). addition of the two model pro­teins to d(C10) resulted in a markedly different behaviour. Complexation with Lys18

gave rise to a melting transition at 25"C. while in the presence of Orn18 no indication for duplex formation was found. The enthalpy of the melting transition in d(C10)/ lys18 is slightly lówer than the one found for the d(T10)/ Lysis complex (see Table V). but it is compensated by a smaller entropy change. These results give strong support for the existence of a stereospecific protein complexation. Comparison with the absence of duplex formation for phosphate-methylated d(CpC) with the Rp chirality due to steric hindrance. leads to the conclusion that complexation with Lys18

can only involve contacts with Os. while complexation with Orn18 must indeed involve the close approach of ammonium to OR. An additional experimental indication for the exclusive use of Os by Lysis comes from the 31P NMR spectrum of d(C 10)/ Lys18. A broadening is seen (15.4 Hz. resonance at 1.87 ppm) with respect to single-stranded d(C10) (18.7 Hz. resonance at 1.73 ppm). which is just the effect of a higher rigidity in a duplex compared to a single strand. 21

The cause of the steric hindrance for the ammonium group close to OR is the high propellor twist angle in the C-C base pairs. as can be verified with 13C NMR spectroscopy. In Table X it can be seen that the shielding and deshielding of c" and Cll in Lys18 are substantially different from those found in T-T duplexes ( see Table VI). Since the location of the protein backbone is similar in both d(T 10)/ lys18 and d(C10)/ lys18. a difference in orientation between the thymine and cytosine bases must be present. A higher propellor twist angle in the C-C base pair will lead toa tilt­ing of the deshielding cones (see Figure 8), placing CIS in the plane of the cytosine ring (large deshielding) and C" near the edge of the cone (small effect). The influence

69

Table X. 13c NMR chemica! shifts (ppm) in the Lys 1s model protein: ' before and after complexation with d(C10) at 6°C and pH 9.

Lys18 d(C10)/Lys1s as (ppm)

c" 54.37 54.39 0.02 . cfl 31.24 31.48 0.24

Cy 22.85 22.85 0.00 có 27.18 27.14 -0.04 c. 4009 40.02 -0.07

of the .propellor twist angle on the shielding in the groove of the duplex is calculated for Co: and c13 (see Figure 11) .. by determining the diamagnetic and paramagnetic

shielding as well as ring current effects. according to the theory of GiessnercPrettre.23

The total effect of seven hase pairs was taken into account to prevent excessively

long computation times. lt can b.e seen cle.arly that .the obsèrved changes in chemica!

shift in .the complexes of Lys18 (T-T: C"' +0.21. and c13 .-0.08 ppm; Ç-C C"' +0.02.

and C13 +0.2.4 ppm) can only be explained with a higher propellor twist angle in the

C-C base pairs: For T-T duple.Jtes. C" must be deshielded n:iore stron~ly than CiJ.

indicating a propellor twist angle. between 0 and 20°. In (7( duplexes. the. shielding is ·

re\lersed. so the pr.opellor twist angle must exceed 20°. This COl'Jclusion corroborates

the molecula.r mechanics calculat.io.ns on the .phosphate-methylated systems. Fro.m ·

the differences in shielding effects of T- T and C-C base pairs it can also be seen that ,• " ' '

recognition çif bases inside a duplex by a protein may occ.ur via ring .cur.rent effects. ·

b~sides well-known hydrogen bond.· electrostatic and vàn der Waals interactions. '

The selectivity that is found for duplex formation of the diastereomers of

d(T pC) can also be seen in protein-induced duplexes. For this. we selected natural

oligomèrs with both thymine 'and cytosine bases. viz. d(C6 T 6) and d(T 6C2 T2). The

first system allows comparison with our results on phosphate-methylated d( C6 T 6)

(vide supra).1b while in the second oligomer the impact of only two cytidine residues

can be investigated. The dodecamer d(C 6T6) is a single strand at pH 9. but UV

hyperchromicity experiments show the presence of a duplex (T m = 22°C) after c:om­

plexation with Lys18• The enthalpy and entropy of the melting transition of this

duplex are similar to the values found for the other duplexes induced. by Lys18 (see

Table V). With Orn18 however. d(C6T6) remains in the single strand form. A similar

70

T-T HEPTAMER

-oD2

-0.03

!f -<l.04

~ iii -Q.o5

,,,. ... _ ... _ ---_...,_,.. __ _ ____ "

-0.06

-0.07 0 20 40 60

PAOPeLLOR-TWIST N>G..E

C-C HEPTAMER

-0.02

-O.o3

i -0.04

-O.o5

Figure 11. Calculated shielding effect of the thymine bases (top) and the cytosine bases (bottom) on the C" (-) and the C13 (- - -) nuclei in Lys18 as a function of the propellor twist angle in the base pairs.

behaviour was seen for the decamer d(T 6C2 T 2). which forms a duplex at neut ral pH (T m = 2TC. see Table V) exclusively after complexation with Lys18. This was observed in UV hyperchromicity experiments, and independently with 1H NMR meas­urement of the imino proton chemica! shift in the T-bases (13.0 ppm at 4°C. whereas uncoupled T-bases have imino resonances at 11 ppm18). Also. the 31P chemica! shifts of the d(T 6C2 T 2) f Lys1s complex are shifted upfield compared to uncomplexed deca­mer {for phosphates next to T-bases from 1.61 to 1.55 ppm; for phosphates next to C-bases from 1.85 and 1.78 ppm to 1.75 and 1.73 ppm. respectively). In these mixed oligomers. 31 P NMR measurements are hampered by the presence of various reso­nances in the spectrum. so no analysis of the resonance linewidth is performed. Like­wise. in the 13C NMR spectra signals are seen of Lys18 in the presence of both T- T

71

and C-C base pairs. which precludes a detailed shift analysis. However. it is seen that the presence of cytidine residues completely prevents duplex formation when com­

plexation with an OR atom (as is necessary for Orn1s) occurs. These results are in accordance with our observations for phosphate-methylated d(T pC). where duplex

formation is precluded for the Rp form because of steric hindrance between the

methyl group and the cytosine ba.ses.

CONCLUDING REMARKS

The present results show. that parallel duplex formation occurs for phosphate­methylated d(CpC) and d(TpC) with the Sp chirality exclusively. Therefore. both T­

T and C-C base pairing has now been established in phosphate-methylated systems. Up to now. all studies on phosphate-methylated systems with the purine bases aden­ine and guanine have shown no parallel duplex formation. No duplex;:::t coil transition

has for instance been seen for phosphate-methylated d(ApA) in aqueous solution. 31

In a more indirect way it was seen that adenine and guanine do not form dimers in the solid state. The X-ray crystal structures of 3' .5'-di-O-acetyl-2'-deoxyadenosine32

and 3' .5' -di-0-acetyl-2' -deoxyguanosine33 show the presence of intricate hydrogen

bond networks.

lt is further se~n trom t~e present results that complexation of poly-L-lysine and poly-L-ornithine can lead to the formation of parallel duplexes for natura! pyrimidine oligomers. Therefore. phosphate-methylated oligonucleotides are accurate model sys­

tems for protein complexation with DNA. When cytosine bases are present. parallel duplex formation in both protein-complexated and phosphate-methylated systems is only possible when the shîelding group (ammonium or methyl) binds the sterically most avaible 0 5 atom in the ph~sphate groups. The OR oxygen atom cannot be

shielded without strong steric interactions. as could be seen from molecular modelling studies. lt follows from our results that stereospecific complexation of proteins with

antiparallel DNA duplex.es may also be present. an aspect which has received little attention up to now. The present study can therefore lead to a better understanding of protein-:DNA interactions in which .complexation of the phosphates is involved. Such interactions .·are seen for various prqteins that stabilize duplex structures12 ·14

(such as histones). but also for enzymes which destabilize duplexes34 (such as heli­cases: RNA polymerase. and restriction endonucleases). For both classes. shielding of the phosphate charges is highly important for their function.

72

EXPERIMENTAL SECTION

Oligonudeotides

The oligonudeotides were synthesized on a 15-µmol scale with an Applied Biosystems 381A DNA synthesizer. using the standard ,B-cyanophosphoramidite pro­tocol.35 After removal from the column and deprotection in ammonia. the compounds were purified by alcohol precipitation at -20°C. For the short oligomers we use (deca­mers d(T10). d(C 10). and d(T6C2T2). and dodecamer d(C6T6)). the following approach proved to be the most efficient: to 80 µL of DNA solution. 20 µL of a 3 M sodium acetate/acetic acid buffer solution (pH 5.6) was added, and 900 µ,L 96% ethanol. Overnight precipitation. centrifugation at 14000 rpm for 15 min. and drying afforded DNA pellets in ca. 50% yield. based on UV absorption at 260 nm (assuming a single-strand concentration of 33 µ,g/ml to give an E260 of 136). The purity of the

oligomers was checked with 1H and 31 P NMR spectroscopy.

Proteins

Poly-l-lysine (average molecular weight: 3700. degree of polymerization: 18) and poly-l-ornithine ( average molecular weight: 6200. degree of polymerization: 32) were purchased as the hydrobromide salts from Sigma Chemica! Co. Proteins with an aver­age length of 18 residues were needed. since a duplex of the decamer oligonucleotides contains 18 phosphate groups to be shielded. The poly-L-lysine was used as received. and a stock solution was made by weighing. The ornithine mixture was fractionated on a Sephadex G25 superfine gel filtration column. using an 0.1 M HCI solution as eluent. The average polymer length in the fractions was deduced from the 1H NMR spectrum. by comparing the intensities of the H" resonances in the end groups and in the interior of the protein. Based on the UV absorption of a known solution of the poly-l-ornithine mixture at 209 nm. it was calci.Jlated that the concentration of ornithine residues equals 4.28-10- 4• E209 M. Using this relation. a stock solution was

prepared of an olîgo-ornithine with the correct length. The complexation of the oli­gonucleotides wîth the proteins is described under UV measurements.

Phosphate-methylated d{TpT)

The synthesis and diastereomeric separation are described in ref. tb.

Phosphate-methylated d( CpC)37

2'-Deoxycytidine (4.54 g. 20 mmol) was dried by coevaporation with dry pyri­dine and subsequently suspended in 100 ml of dry pyridine. 13.0 ml (100 mmol) of trimethylchlorosilane was added dropwise and the mixture was stirred for 15 min. Then 5.8 g (22 mmol) of 9-fluorenylmethoxycarbonylchloride was added and the reaction mixture was stirred for 2 h at room temperature. yielding a white suspension

73

in the yellow solution. Hydrolysis of the trimethylsilyl groups and unreacted chlorides was effected by addition of 20 ml water. After stirring for 15 min. a dear yellow solution was obtained. which was evaporated to near dryness by coevaporation with toluene. Upon addition of 300 ml water to the resulting oil. a white precipitate appeared. The mixture was shaken vigorously until no more yellow oil was visible. After addition of 150 ml of ethyl acetate and shaking. a precipitate formed on the separation layer. which was isolated by filtration and washed with ethyl acetate. After drying in vacuo the product was obtained as a white solid (8. 71 g. 97%). 1H NMR (acetone-d6): o 2.40 (1H. m. H2-). 2.59 (1H. m. H2"). 3.77 (2H. m. Hs-/HS").

4.03 (lH. m. H4'). 4.33 (1H. t. CH Fmoc). 4.49 (3H. m. H3·/CH 2 Fmoc). 6.23 (lH, t. Hr}. 7.12 (1H. d. H5). 7.37 (4H. m. aromatic Fmoc). 7.85 (4H. m. aromatic Fmoc).

8.50 (1H. d. H6).

Base-protected 2' -deoxycytidine (8.55 g. 19 mmol) was dried by coevaporation with dry pyridine and subsequently suspended in 200 ml of dry pyridine. After addi­tion of 6.5 g (21 mmol) of 4-monomethoxytritylchloride. the solution was stirred for 15 h. The mixture was then poured in a saturated aqueous sodium bicarbonate solu­tion (250 ml). and extracted three times with dichloromethane. The combined organic layers were washed with a saturated aqueous sodium bicarbonate solution and dried on magnesium sulfate. After filtration. the solution was concentrated in vacuo. Removal of all pyridine from the brownish oil was accomplished by coevapora­tion with toluene {twice) and 2-butanone. The resulting brown foam was purified by column chromatography on Woelm silica gel with 2-butanone as eluent R1 = 0.37).

yielding 8.90 g (65%) white solid product. 1H NMR (CDC13): o 2.25 (1H. m. Hr).

2.71 (1H. m. Hr). 3.47 (2H. m. H5·/Hd. 3.79 (3H. s. OCH3 trityl). 4.14 {1H. m.

H4'). 4.29 (lH. t. CH Fmoc). 4.50 (3H. m. H3·/CH2 Fmoc). 6.28 (1H. t. Hr). 6.85 (2H. m. trityl). 6.99 (1H. d. H5). 7.31 (12H. m. trityl). 7.41 (4H. m. aromatic Fmoc).

7.69 (4H. m. aromatic Fmoc). 8.22 (1H. d. H6).

The 5' -tritylated. base-protected 2'-deoxycytidine (5.00 g. 7 mmol) was stirred for 15 min in a mixture of 50 ml of dry pyridine and 5 ml of acetic anhydride. The mixture was then poured in 50 ml of a saturated aqueous sodium bicarbonate solu­tion. and extracted twice with dichloromethane. The combined organic layers were dried on magnesium sulfate. After filtration. the solution was concentrated in vacuo. The resulting oil was purified by column chromatography on Woelm silica gel with 2-butanone as eluent (R1 0.65). yielding a white solid. This was stirred for 15 h in a mixture of 160 ml acetic acid and 40 ml water. After addition of another 120 ml water. the reaction mixture was concentrated in vacuo to near dryness. The excess acetic acid and water were removed by coevaporation with 2-butanone. Crystallization from 2-butanone yielded the product as a white solid (2.04 g. 57%). 1H NMR

(CDCl3/CH30D 9:1 v/v): 8 2.12 (3H. s. CH 3 acetyl). 2.27 (1H. m. H2·). 2.66 (1H. m.

Hr). 3.86 (2H. m. H5'/H5"}. 4.20 (1H. m. H4·). 4.31 (1H. t. CH Fmoc). 4.55 (2H. d.

74

CH 2 Fmoc). 5.33 (1H. m. H3')· 6.29 (1H. t. Ht'). 7.36 (5H. m. H5/aromatic Fmoc). 7.74 (4H. m. aromatic Fmoc). 8.39 (1H. d. H6).

2.00 g (2.8 mmol) of 5' -tritylated. base-protected 2' deoxycytidine was dried by coevaporation with dry pyridine. and subsequently dissolved in 20 ml of dry pyridine. 0.10 g (1.4 mmol) of 1H-tetrazole and 0.86 g (3.3 mmol) bis-(N.N­diisopropylamino)-methoxyphosphine (dried by coevaporation with dry pyridine) were added. and the reaction mixture was stirred at room temperature for 15 min. Forma­tion of the phosphoramidite coupling compound in situ was evident from the 31 P NMR spectrum (CDCl3: 8 149.5 and 150.2). Then. 1.65 g (3.36 mmol) of 3'-0-acetyl-4-N-(9-fluorenylmethoxycarbonyl)-2' -deoxycytidine and 0.50 g (7 mmol) 1H­tetrazole (both dried by coevaporation with dry pyridine) were added. and the reac­tion mixture was stirred at room temperature for another 2 h. The obtained phos­phite (31 P NMR (CDC13): 8 140.4 and 140.8) was oxidized by addition of 1.5 ml of

tert-butylhydroperoxide. giving the phosphate (31P NMR (CDC13): 8 -0.33 and -0.44).

After stirring for 5 min. the mixture was concentrated in vacuo to near dryness and coevaporated twice with toluene and 2-butanone. The product was detritylated by addition of a mixture of 100 ml acetic acid and 25 ml water. After stirring fot 15 h. another 75 ml water was added. and the mixture was concentrated in vacuo to near dryness. The excess acetic acid and water were removed by coevaporation with 2-butanone and ethyl acetate. The product was purified by column chrornatography on

Woelm silica gel. First. impurities were eluted with ethyl acetate/dichloromethane (9:1 v/v); then the product was obtained with ethyl acetate/dichlorornethane/ethanol (85:10:5 v/v/v) as eluent (R1 = 0.25). with a yield of .1.25 g (44%). 1H NMR.

(CDC13): 8 2.10 (3H. s. CH 3 acetyl). 2.43 (2H. m. H2• of dCp and pdC). 2.77 (2H. m.

H2" of dCp and pdC). 3.84 (3H. d. POCH3• J = 11.0 Hz); 3.91 (2H. m. H5-jH5" of

dCp). 4.2-4.5 (10H. m. H4• of dCp/H4·/H5·/H5" of pdC/CH/CH2 Fmoc). 5.29 (2H. m. H3' of dCp and pdC). 6.24 (2H. m. Hr of dCp and pdC). 7.30 (10H. m. Hs of dCp

and pdC/aromatic Fmoc). 7.53 (4H. m. aromatic Fmoc). 7.77 (4H. m. aromatic Fmoc). 8.07 (1H. d. H6 of dCp). 8.20 (1H. d. H6 of pdC). 31P NMR (CDC13): S -0.35 and 0.07 ppm.

The base-protecting Fmoc groups were removed by treatment with a mixture of of triethylamine (20 mmol/mmol Fmoc) and of pyridine (10 mL/mmol Fmoc). After ca. 3 h. thin layer chromatography showed the reaction to be complete. The reaction mixture was concentrated in vacuo. and the excess pyridine and triethylamine were removed by coevaporating three times with. toluene. Extraction of an. aqueous solu­tion of the residue with dichloromethane removed virtually all split-off fluorene groups. 1H NMR (020): S 2.06 (3H. s. CH 3 acetyl). 2.30 {2H. m. H2• of dCp and

pdC). 2.56 (2H. m. Hr of dCp and pdC). 3.70 (2H. m. H5·/H5" of dCp). 3.78 (3H. d.

POCH3. J 11.4 Hz). 4.20 (1H. m. H.r of dCp). 4.32 (3H. m. H4·/H5·/H5" of pdC). 4.96 (1H. m. H3· of dCp). 5.28 (1H. m. H3• of pdC). 5.98 (1H. d. H5 of dCp). 5.99

75

(1H. d. H5 of pdC). 6.15 (1H. dd. Hr of dCp). 6,18 (1H. dd. Hr of pdC). 7.69 (1H. d. H6 of dCp). 7.73 (1H. d. H6 of pdC). 31 P NMR (D20): B 1.96 and 2.07.

The diastereomeric mixture was dîssolved in a solution of acetonitrile and 15 mM aqueous ammonium acetate (7:3 v/v). and a 0.5 ml portion was used for developing the separation. Then. the rest of the solution was diluted with 2 L of the aqueous part of the mobile phase containing half the percentage of organic modifier to be used in the preparative chromatography. The separation of the diastereomers was developed on a H P 1090 gradient H Pl C system. Preparative chromatography. was executed on a HPlC system consîsting of a Waters M590 solvent select valve module. an Alltech C18 10µm column (250x22 mm) and a Waters 480 detector. Frac­tions were checked on an analytica! isocratic HPlC comprîsing a Spark SPH 125 autosampler. a Gilson Model 302 pump. a Machery-Nagel Nucleosil 120-s C18 reversed-phase column (250x4 mm). and a Linear UVis 203 detector (set at 260 nm). Mobile phases were made from Milli-Q water containing 1 ml/L acetic acid (AnalaR. BDH). 200 µl/l triethylarnine (zur Synthese. Merck). and adjusted toa pH 3.5 with a 25% aqueous solution of ammonium hydroxide (Baker, Analysed Reagent). Aceton­itrile (Lichrosolv. Merck) was added (13%) as an organîc mobile phase rnodifier. The individual diastereomers were obtained in more than 98% purity and more than 70% chromatographic yield. Assignment of the phosphorus configuration in both diastereomers was perforrned according to the recent work of Sumrners et al. 38 Rp

exclusively corresponds with a NOE contact between H3• of the 5' -residue and the phosphate methyl group. The outcome of this assignment fits the conclusion of Seela et al. 39 that elution from a reversed phase column is faster for Sp than for Rp.

Phosphate-methylated d(TpC)

Thymidine (5.0 g. 20.7 mmol) was suspended in 50 ml of dry pyridine. and 4-monomethoxy'tritylchloride (7.65 g. 24.8 mmol) was added. The solution was stirred ovèrnight in darkness. After evaporation of the py'ridine. the residue was coeva­porated with toluene. and then recrystallized from toluene. Filtration and drying at 50°C yielded 5'-0-(4-monomethoxytrîtyl)-thymidine as a white solid (9.97 g. 94%). 1H NMR (CDC13): B 1.45 (3H. s. CH 3 base). 2.41 (1H .. m. Hr). 2.88 (1H. m. Hr).

3.41 (2H. m. H5'1H5"). 3.78 (3H. s. OCH 3 trityl). 4.07 (1H. m. H4,). 4.58 (1H. m.

H3·). 6.43 (1H. t. Hr). 6.80-6.90 (2H. d. trityl). 7.12-7.49 (12H. m. trityl). 7.58 (1H. s. H6). 9.24 (1H. s. NH).

5' -0-(4-Monomethoxytrityl)-thymidine (6.18 g. 12.0 nimol) and 1H-tetrazole (0.42 g. 6.0 mmol) were dissolved in 60 ml of dry pyridine. Bis-(N.N­diisopropylamino)-methoxyphosphine (3.78 g. 14.4 mmol) was added. and the reac­tion mixture was stirred for 30 min. Complete formation of the phosphoram'idite was seen in the 31 P NMR spectrum (CDCl3: 8 149.8 and 149.1). Then a solution of 2'­

deoxyctidine (3.00 g, 13.2 mmol) and 1H-tetrazole (2.31 g. 33.0 mmol) in 30 ml of

76

dry pyridine were added. and the mixture was stirred for 1 h. 31 P NMR showed com­plete formation of the phosphite. almost exclusively with the 5' -OH of 2' -deoxycytidine (CDCl3: 8 141.2 and 140.9). The phosphite was oxidized with 7 ml of

tert-butylhydroperoxide at 0°C in 10 min. After evaporation of the volatile com­

ponents and coevaporation with toluene. the residue was chromatographed on a silica gel column with dichloromethane/methanol (9:1 v/v) as eluent· (R1 0.22). The

phosphate was obtained in a yield of 53% (5.2 g). 31 P NMR (CDC13): 8 -0.02 and

-0.04.

The monomethoxytrityl protecting group was removed by dissolving the phos­phate (1.09 g. 1.33 mmol) in 40 ml of an acetic acid/water (4:1_ v /v) mixture. and stirring overnight. After evaporation and coevaporation with water. the residue was

dissolved in 60 ml water and washed three times with 4 ml of dry diethyl ether. The aqueous layer was coocentrated. and the diastereomers of d(Ti>C) were separated with reversed-phase HPlC as described. Assignment of the phosphorus configuration

in both diastereomers was performed as above. Sp: 31 P NMR (D20) S 2.01: 1H NMR (D20) 8 1.80 (3H. s. CH3 base). 2.28 (1H. m.

Hr pdC). 2.37 (1H. m. H2· dTp). 2.42 (1H. m. Hr pdC). 2.49 (1H. m. Hr dTp).

3.71 (2H. m. Hs)H5", dTp). 3.77 (3H. d. POCH3. J 11.6 Hz). 4.11 (1H. m. H4·

dTp). 4.18 (1H. m. H4• pdC). 4.27 (1H. m. Hç pdC). 4.35 (1H. m. H5• pdC). 4.45

(1H. m. H3• pdC). 4.97 (1H. m. H3• dTp). 5.36 (1H. d. H5). 6.17 {1H. m. H1• pdC).

6.18 (1H. m. Hr dTp). 7.52 (1H. s. H6 dTp). 7.66 (1H. d. H6 pdC).

Rp: 31 P NMR (D20) 8 2.10: 1H NMR (D20) 8 1.81 (3H. s. CH 3 base). 2.28 (1H. m.

H2• pdC). 2.37 (1H. m. H2• dTp). 2.42 (1H. m. Hr pdC). 2.52 {1H. m. Hr· dTp).

3.71 {2H. m. H5·/H5". dTp). 3.78 (3H. d. POCH3. J = 11.6 Ht). 4.13 (1H. m. H4•

dîp). 4.18 (1H. m. H4• pdC). 4.27 (1H. m. H5" pdC). 4.35 (1H. m. H5· pdC). 4.46

(1H. m. H3• pdC). 4.98 (1H. m. H3• dTp). 5.99 (1H. d. H5). 6.18 (2x1H. m. Hr pdC and dTp). 7.54 (1H. s. H6 dîp). 7.69 (1H. d. H6 pdC).

UV measurements

All variable-temperature UV measurements were performed in 10-mm quartz cuvettes at 260 nm on a Perkin-Elmer 124 spectrophotometer. Samples of phosphate-methylated dinucleotides were made at a concentration of ca. 1 µ.M in a

0.01 M TrisHCl/0.01 M MgCl2 (pH 7.5) buffer solution. Samples of DNA/protein

complexes were prepared by diluting a DNA stock solution in the appropriate buffer

to an E260 of ca. 0.2. after which such a quantity of protein was added. that the

number of phosphates in the DNA equalled the number of terminal ammonium groups in the protein.The samples were then incubated for 15 min at room tempera­

ture. and after that for 15 min in ice. Usually. TE buffer (10 mM TrisHCl/1 mM EDTA pH 7.5) was used for these experiments. For the samples containing d(C10)

however. it was necessary to adjust the pH of the buffer to 9. since otherwise the

77

cytosine base is protonated at N3• and self-association spontaneously occurs.8·3° For

obtaining melting curves. sample temperatures were raised at l°C/min. and measure­

ments were performed at 2°C intervals. The UV hyperchromicity curves were

computer-fitted to obtain melting temperatures (T111 values). For this. we used the

following formula. which has a slightly different form than usual. 22 but is completely

equivalent to the standard equation for self-complementary strands. The fraction

single-strand form f is determined by:

f=

1+

2 1 2

where AH0 is the dissociation enthalpy of the DNA duplex. The entropy change is

then determined from:

o AH0 .:lS = -- - R·lnCr

Tm

where Cr is the total single strand concentration in the sample.

NMR measurements

One-dimensional 1H NMR measurements were performed at 200. 500. or 600

MHz on Bruker AC 200. AM 500. or AM 600 spectrometers. all interfaced with an

Aspect 3000 computer, or at 300 MHz on a Bruker CXP 300 spectrometer. interfaced

with an Aspect 2000 computer. Usually. a 15 ppm sweep width was used with 16K

data points. which were zero-filled to 32K points before Fourier transformation.

Chemica! shifts were referenced to the H20 resonance or the residual HDO peak in

D20. which were set at 4.68 ppm at room temperature and at 4.88 ppm at 4°C. Dur­

ing the measurement. these peaks were suppressed with the routine described by

Haasnoot et al.40 Phosphate-methylated dinudeotides and DNA/protein complexes

were dissolved in either D20. or in an 85:15 (v/v) H20/D20 mixture for measure­

ments of the exchangeable protons. with a concentration of ca. 1 mM. In some cases. partial precipitation of the DNA/protein complex occurred. and sainples were filtered

before use. All measurements were performed in 5-mm NMR tubes without addition

of buffers or.salts. since high salt concentrations disrupt the DNA-protein interaction.

Samples were thoroughly sparged with argon gas to remove dissolved oxygen. and

stored at 4°C.

Two-dimensionai 1H NMR experiments were performed at 6ClO MHz on a Bruker

AM 600 spectrometer. interfaced with an Aspect 3000 computer. The phase-sensitive

NOESY experiment was performed at 4°( in D20 with an frsize of 4K. and 1024

experiments in the f1 direction. Each experiment had 64 scans and a 90°-pulse of

11.56 µs. A mixing time of 300 ms was used. and a sweep width of 6000 Hz. The

residual H DO peak was suppressed by presaturation. 41 The obtained file was weighed

78

with a shifted squared-sine bell window in both directions before Fourier transforma­tion. Zero-filling to 4K points was performed in the fi-direction. and appropriate

phase corrections were applied. Cross peaks in the resulting 16M file were integrated by volume to assess the strength of the NOE contact.

31 P NMR measurements were performed at 80.8 MHz an a Bruker AC 200 spec­trometer. interfaced with an Aspect 3000 computer. Usually. a sweep width of 150

ppm was used with 16K data points. Before Fourier transformation, zero-filling to 64K points was performed. In some cases. it was necessary to apply a Gaussian win­dow function for resolution enhancement. Chemica! shifts were referenced to 85%

H3P04 (8 0 ppm). 13C NMR measurements were performed at 50.3 MHz on a Bruker AC 200 spec­

trometer. interfaced with an Aspect 3000 computer. A size of 16K data points was used. and a sweep width of 12 kHz. Chemica! shifts were related to tetramethylam­monium bromide (8 56.4 ppm).

Model building and energy refinement

lnitially. the protein backbone conformation that corresponds with DNA helical parameters was established with a self-written Pascal program called PROTEIN. which calculates atomie coordinates in a growing protein · chain · from standard bond lengths and angles. based on the X-ray structure of L-lysine hydrochloride.42 and vari­able torsion angles. The result of these calculations was used in model building stu­dies with the graphics program Chem-X. where the helical protein was fitted to a parallel DNA duplex by optimizing the conformation of the side chains. Parallel DNA duplexes were generated in Chem-X16 as described before. using standard B DNA coordinates.2 These DNA/protein structures were then used as input geometries for energy minimization in the molecular mechanics program AMBER.13 The coordinates of the input structures were energy-refined until the gradient was smaller than 0.1 kcal/mol.À. which on the average took 4000 steps. The geometries of the minimized structures were analyzed. and energy interactions between the two DNA strands were determined. All calculations were performed in the united atom approach. with a dielectric constant E r;j· While the lysine residue is directly available from the

AMBER database, the ornithine residue had to be defined with the PREP module of AMBER. The coordinates of ornithine were derived from lysine by removing one methylene group from the side chain. For phosphate-methylated DNA systems. we used the force field parameters and input geometries as described before. 2

79

REFERENCES

1. (a) L.H. Koole. M.H.P. van Genderen. H. Frankena. H.J.M. Koeken. J.A. Kanters and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 89. 51 (1986) (communicated by H.M. Buck at the meeting of Nov 25. 1985). (b) L.H. Koole. M.H.P. van Genderen. H.M. Buck. H.M" J. Am. (hem. Soc. 109. 3916 (1987).

2. M.H.P. van Genderen. L.H. Koole. O.M. Aagaard. C.E.J. van Lare and H.M. Buck. Biopolymers 26. 1447 (1987).

3. D.W. Young and H.R. Wilson. Acta Crystallogr .. Sect. B. Struct. Sci 31. 961 (1975).

4. L.H. Koole. N.L.H.L. Broeders. M.H.P. van Genderen. P.J.L.M. Quaedflieg. S. van der Wal and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 245 (1988) (communicated by H.M. Buck at the meeting of May 30. 1988).

5. M.H.P. van Genderen. L.H. Koole and H.M. Buck. Proc. K. Ned. Akad. Wet .. Ser. B 90. 181 (1987) (communicated by H.M. Buck at the meeting of Jun 22.

1987).

6. P. Hobza and C. Sandorfy. J. Am. Chem. Soc. 109. 1302 {1981).

7. D. Patel. In "Nucleic Acid Geometry and Dynamics". R.H. Sarma. Ed" Pergamon Press. New York. 1980.

8. K. Wüthrich. "NMR of Proteins and Nucleic Acids". Wiley & Sons. New York. 1986. A protonation of N3 can occur. leading to a different parallel C-C base pair.

See: M.H. Sarma. G. Gupta and R.H. Sarma. FEBS Letters 205. 223 (1986). However. at pH 7.5 we detected no N:;H imino resonance. even at low tempera­

ture.

9. R.H. Sarma. in "Nucleic Acid Geometry and Dynamics". R.H. Sarma. Ed" Per­gamon Press. New York. 1980. p. 143.

10. (a) C. Altona. Reel. Trav. Chim. Pays-Bas 101. 413 (1982). (b) P.P. Lan­khorst. C.A.G. Haasnoot. C. Erkelens and C. Altona. J. Biomol. Struct. Dyns. 1. 1387 (1984).

11. Nomenclature according to IUPAC-IUB recommendations. See: Eur. J. Biochem. 131. 9 (1983).

12. W. Saenger. "Principles of Nucleic Acid Structure". Springer Verlag. New York. 1984.

13. S.J. Weiner. P.A. Kollman, O.A. Case. U.C. Singh, C. Ghio. G. Alagona. S. Profeta. Jr. and P.K. Weiner. J. Am. Chem. Soc. 106, 765 (1984).

14. J.T. Shapiro. M. Leng and G. Felsenfeld. Biochemistry 8. 3219 (1969).

15. Nomenclature follows: IUPAC-IUB Commission on Biochemica! Nomenclature 1969. Biochemistry 9. 3471 (1970).

80

R H H 1 Ijl w 1 <P 1

-C -C-N-C -10:. Il 1 a. H 0 R

x' x2 X3 x• + lysine: R~ -CH2-CH2-CH2-CH2 -NH3

p y ó ~

x' x2 X3 + ornithine: R = -CH2-CH2 -CH2 -NH3 p y ó

16. Chemica! Design Ltd" Oxford.

17. G.E. Schulz and R.H. Schirmer. "Principles of Protein Structure". Springer Ver­lag. New York. 1979.

18. C.A.G. Haasnoot. J.H.J. den Hartog. J.F.M. de Rooij. J.H. van Boom and C.

Altona. Nature (London) 281, 235 (1979).

19. M.H.P. van Genderen. L.H. Koole and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 171 (1988) (communicated by H.M. Buck at the meeting of Nov 30. 1987). A similar upfield shift has been observed in the complexation of tetra-L­lysine with the antiparallel duplex of self-complementary d(A3GCT 3). See: P. Davanloo and D.M. Crothers. Biopolymers 18. 2213 (1979).

20. Determination of the surface areas of the two 31 P signals with a curve-fitting program revealed that 24% of the phosphates is not associated with protein (downfield resonance). Addition of extra Lys1s to the NMR sample did not change the 31 P spectrum. so the downfield signal can not be caused by remain­ing free d(T 10) strands. hut free phosphates must be present in the d(T10)/Lys18 system. Since eighteen phosphate groups are present in the com­plex. four of them (24%) are therefore not bound to protein. They most likely represent the relatively freely moving ends of the duplex. where association with protein is disturbed by a higher mobility of the phosphate groups. In each DNA strand. the two phosphates at each end remain uncomplexed. giving a total of four.

21. A reduced rotational motion of the molecule results in a shortened 31p transver­sal relaxation time, and hence in a broadened phosphorus resonance. See: D.G. Gorenstein. Ann. Rev. Biophys. Bioeng. 10. 355 (1981).

22. L.A. Marky and K.J. Breslauer. Biopolymers 26. 1601 (1987).

23. (a) C. Giessner-Prettre and B. Pullman. Biopolymers 15. 2277 (1976). (b) F.R. Prado and C. Giessner-Prettre. J. Mol. Struct. Theochem 76. 81 (1981).

24. M. Remin and D. Shugar. Biochem. Biophys. Res. Commun. 48. 636 (1972).

81

25. G.M. Clore and A.M. Gronenborn. FEBS Letters, 172. 219 (1984).

26. C. Altona and M. Sundaralingam. J. Am. Chem. Soc. 94. 8205 (1973).

27. We used the following formula for calculating x5 from the ratio of two NOE

contacts:

where 71 1 and 71 2 denote the strengths of two NOE contacts. and r15• rJN. r25•

and r2N the corresponding interproton distances in the south and north form.

28. D.R. Kearns. in "Two-Dimensional NMR Spectroscopy. Applications for Chem­

ists and Biochemists". W.R. Croasmun and R.M.K. Carlson. Eds" VCH Publish­

ers. New York. 1987.

29. A.M. Bobst. S.-C. Kao. E.V. Bobst and G.T. Pauly. J. Biomol. Struct. Dyns. 3.

249 {1985).

30. (a) A. Adler. L. Grossman and G.D. Fasman. Proc. Natl. Acàd. Sci. USA 57.

423 (1967). (b) W. Guschlbauer. Proc. Natl. Acad. Sci. USA 57. 1441 (1967).

31. L.H. Koole. unpublished results.

32. L .. H. Koole. H.M. Buck. J.A. Kanters and A. Schouten. Can. J. Chem. 65. 326

(1987).

33. L.H. Koole. H.M. Buck, J.A. Kanters and A. Schouten. Can. J. Chem. 66.

(1988). in press.

34. M.H.P. van Genderen and H.M. Buck, Biopolymers (1989). in press.

35. M.J. Gait. "Oligonucleotide Synthesis. A Practical Approach". IRL Press. Oxford.

1984. Chapter 3 and references cited therein.

36. T. Maniatis. E.F. Fritsch and J. Sambrook. "Molecular Cloning. A Laboratory

Manual". Cold Spring Harbor Laboratory. New York. 1982. p. 468.

37. (a) L.H. Koole. P.J.L.M. Quaedflieg. W.H.A. Kuijpers. N.L.H.L. Broeders. H.A.

Langermans. M.H.P. van Genderen and H.M. Buck. Proc. K. Ned. Akad. Wet ..

Ser. B 91. 205 (1988) (communicated by H.M. Buck at the meeting of Jan 25. 1988). (b) L.H. Koole. N.L.H.L. Broeders. P.J.L.M. Quaedflieg. M.H.P. v~n Gen­

deren .. S. van der· Wal and H.M. Buck. J. Org. Chem. (1989). in press. (c)

M.H.P. van Genderen. L.H. Koole and H.M. Buck. Reel. Trav. Chim. Pays-Bas

(1989). in press.

38. M.F. Summers. C. Powell. W. Egan, R.A. Byrd. D.W. Wilson and G. Zon.

Nucleic Ac.ids Res. 68. 7421 (1986).

39. W. Herdering. A. Kehne and F. Seela. Helv. Chim. Acta 68. 2119 (1985).

40. C.A.G. Haasnóot and C.W. Hilbers. Biopolymers 22. 1259 (1983) .

. 82 <

41. G. Bodenhausen. H. Kogler and R.R. Emst. J. Magn. Res. 58. 370 (1984).

42. DA Wright and R.E. Marsh. Acta Crystallogr. 15. 54 (1962).

83

CHAPTER5*

Hybridization of phosphate-rnethylatecl DNA and natura! oligonucleotides lmplic.ations for protein-induced DNA duplex destabilization

ABSTRACT

Duplex stabilities have been determined for several hybrids of phosphate­methylated oligonucleotides and natural DNA or RNA using UV hyperchromicity experiments. These hybrids have a higher stability than the corresponding natural duplexes. due to the absence of interstrand electrostatîc repulsions. Comparison with stability data for hybrids of other neutral oligonucleotides (with phosphate-ethylated and methyl phosphonate linkages) and natural DNA or RNA revealed that differences in stability could be attributed mainly to steric and stereoelectronic factors. For phosphate-ethylated oligonudeotides. hybridization with a natural strand is strongly influenced by steric interactions of the ethyl group. Hybridization with RNA. which requires a tight A-type conformation. is therefore difficult. and the ethyl orientation (înward or outward. depending on the phosphorus configuration) determines the strength of the association with natural DNA. In methyl phosphonate systems. it appears that the presence of a P-C bond disturbs the helix conformation for stereoelectronic reasons. This leads to a weaker hybridization with DNA and RNA for longer strands. Phosphate-methylated oligonucleotides are found to have an optimal combination of steric and stereoelectronic factors. and form the strongest hybrids with natural DNA. Absence of întrastrand phosphate-phosphate repulsions causes a slightly different conformation for phosphate-methylated DNA. which is evident in the cooperative character of the hybridization with natural DNA. A thermodynamic model for this cooperativity is presented. and the model studies are extended to protein-DNA complexes in which phosphate charges are also shielded. The prelim­inary results suggest that protein association can destabilize a DNA duplex. thus pro­viding a possible mechanism for the action of DNA unwinding enzymes.

*M.H.P. van Genderen. L.H. Koole and H.M. Buck. Reel. Trav. Chim. Pays-Bas (1989). in press.

84

INTRODUCTION

Synthetic DNA oligomers with backbone modifications are useful as model com­pounds for studying DNA structure and dynamics.1 and as probes for elucidating the stereochemistry in enzymatic cleavage of oligonucleotide strands.2 Systems in which the charge of the phosphate groups is eliminated. e.g. phosphate triesters and phos­phonates. have received particular attention. because of their resistance to breakdown by nucleases.3 and their easy transport through cell membranes as a result of their lipophilic character.4 The absence of the negative phosphate charge also minimizes electrostatic repulsions between the strands in a DNA duplex. thus raising the duplex stability.5 These properties have led to the suggestion that neutral DNA analogues are well-suited for anti-sense hybridization techniques. where a specific genomic sequence is blocked by strong association with a complementary probe.6

In recent years. we have studied phosphate-methylated DNA systems. which are now easily available via new synthetic routes.5·7 lt has already been demonstrated that methylation of the phosphate groups in oligothymidine strands gives rise to a new duplex structure. based on thymine-thymine pairing.8 In this duplex. the 5'-+ 3' vectors run in the same direction. i.e. parallel. due to the symmetrical base pairing. Since pyrimidine-pyrimidine association forces the backbones into close proximity. phosphate charge shielding is essential for the formation of the parallel duplex. Very recently. this parallel duplex was also observed with cytosine-cytosine base pairing in the phosphate-methylated dinucleotide d(Cp(OMe)C). for the Sp configuration on phosphorus exclusively.9 lt is of interest to note. that complexation of the phosphate groups with a polycationic protein (e.g. polylysine or polyornithine) can accomplish the same effect as methylation.9b.lO This provides the possibility of extrapolating the results with phoshate-methylated systems to protein-DNA interactions. which are important in understanding the action of specific enzymes which destabilize the DNA duplex.

In this chapter. we present our results on the hybridization of phosphate­methylated oligomers and natura! DNA or RNA. The duplex stabilities wilt be com­pared to those found for other types of neutral DNA. The discussion will focus on phosphate-ethylated and methyl phosphonate oligonucleotides. which have been extensively studied by Miller et al..3a.6c.ll and by Pless and Ts'o.12 For duplexes between phosphate-methylated and natura! DNA. a more detailed study has been car­ried out. which provides insight into the structural properties of the phosphate­methylated strand.

RESULTS

The various types of neutra! DNA oligomers are identified using the following notation: phosphate-methylated. p(OMe): phosphate-ethylated. p(OEt): methyl phos­phonate. p(Me). The synthesis of the phosphate-methylated systems was achieved

85

1 î î o~ /03· 0 03' o~ /03• ~/ /p'\_ /p'\. /p'\.

MeO 05' EtO 05• Me 05•

l l l p (OMe) p(OEt) p(Me)

using a modified phosphoramidite protocol. 13 New base-protecting groups such as amidine14 and 9-fluorenylmethoxycarbonyl7 .1 5 (Fmoc) were used. which can be removed under conditions that leave the phosphate triesters intact (see Experimental Section). Duplex µ coil melting temperatures (T m values) for hybrids between phosphate-methylated and natural DNA or RNA were established in all cases by means of UV hyperchromicity experiments. Measurements were performed in a salt­free aqueous solution. since it has been shown repeatedly that for these types of neutral/natural hybrids the T 01 value is independent of the ionic strength of the solution.6< For the system d(Ap(OMe)Ap(OMe)A}·poly(dT). the T m value was corro­borated using variable-temperature 500 MHz 1 H NM R spectroscopy in H20. by fol­lowing the chemica! shift and linewidth of the imino protons involved in the hydrogen bonding (see Figure 1).

15

~(ppm)

1 14

13

12

20 30 ~ T(OC) 40

Figure 1. 1H NMR chemica! shift of the imino protons in the system d(Ap(OMe)Ap(OMe)A)·poly(dT) vs. teinperature.

A clear melting transition occurs around 40°(, as is seen m the broadening and

86

upfield shifting of the imino resonance.16 All T m measurements were performed at a ratio of 1:1 for complementary bases. unless stated otherwise. Of all phosphate­methylated oligonucleotides. diastereomeric mixtures were used with respect to the phosphorus configuration. lt will be shown (vide infra). that the phosphorus chirality in phosphate-methylated DNA does not influence the stability of hybrids with natural systems. Table 1 lists the T m values of the studied systems. and the natura! refer­ence duplex d(A3)·poly(dT) has a T m value of 20"C.

Table 1. T m values for hybrids between phosphate-methylated oligonu­deotides and natura! DNA or RNA.

System T m (°C) System

d(Ap(OMe)A)·poly(dT) 30 d(Ap(OMe)A)·poly(U) d(Ap(OMe)Ap(OMe)A)-poly(dT) 41 d(Ap( OMe)Ap(OMe)A)- poly( U) d{[Ap(OMe))~)· poly(dT) 57 d(Cp(OMe)C)-poly(dG) 45 d(Cp(OMe)C)-poly(rG) d(Cp(OMe)Cp(OMe)C)-poly(dG) 55 d(Cp(OMe)Cp(OMe}C)-poly(rG)

T m (°C)

13 < 10

28 12

From the difference of 21°( in T m value between this reference system and the

corresponding phosphate-methylated system. it is obvious that a stabilization of the duplex structure indeed occurs upon neutralization. lt can also be clearly seen. that formation of G-C base pairs gives rise to a much higher duplex stability than found with A-T base pairs. Table Il contains the stability data of the hybrids between methyl phosphonate and natura! DNA or RNA. as reported by Miller et al. in several publications.6dl The results obtained on hybrids of phosphate-ethylated oligomers and natura! systems shown in Table Il come mainly from the work of Miller et al..3a

and Pless and Ts'o.12 As a consequence of the independence of ionic strength (vide supra). the results given in Tables 1 and Il can be compared to abstract the various factors which determine the duplex stabilities of the neutral/natural systems.

DISCUSSION

Steric factors

lt is obvious from molecular models that the various alkyl groups in the neutra! DNA systems. whether directly linked to phosphorus or attached to a phosphate oxy­gen. cause additional steric interactions compared to natural DNA. This is most obvi­ous when comparing duplexes with either natura! DNA or natural RNA. since in the

87

Table ll. T 01 values for hybrids between methyl phosphonate or phosphate-ethylated oligonucleotides and natural DNA or RNA.

System T 01 (°C) System

d([Tp(Me ))8 T)· poly(dA) 12• d([Tp(Me))s T)· poly( rA) d(Ap(Me)A)· poly(dT) 19 d(Ap(Me)A)· poly(U) d(Ap(Me)Ap(Me)A)·poly(dT) 37 d(Ap(Me)Ap(Me)A)· poly(U) d([Ap(Me)]3A)· poly(dT) 45 d([Ap(Me)]3A)· poly(U)

d([T p(OEt )), T)· poly( dA) ga.b d(T[p(OEt)),T)· poly(rA) d(Ap(OEt)A)· poly(U)

"Range from 0-25°C. bTemperature of 90% dissociation into single strand (T 09).

Tm (°C)

< 0 17 33 43

former duplex a B-type helical geometry is found. while in the latter an A conforma­

tion is required. 17 Steric interactions are of higher importance in a duplex with an A

geometry. since intrastrand phosphate-phosphate distances are strongly diminished.18

while substituents on phosphorus are in closer contact with other parts of the

DNA.12 In accordance with these facts. all systems listed in Tables 1 and Il display a

stronger hybridization with DNA than with RNA. The difference in T m value is

expected to follow the steric requirements of the phosphorus substituents. The

increase in steric size on going from methyl phosphonate to phosphate-methylated to

phosphate-ethylated is indeed reflected in the T m values of the adenosine dinucleo­

tides hybridized with poly(U): d(Ap(Me)A). 17°(; d(Ap{OMe)A). 13°C;

d{Ap(OEt)A), 12°C. In these systems. no interactions between neighbouring phos­

phate groups are present. For trinucleotides. these become more important and cause

stronger differences in T m value: d(Ap(Me)Ap(Me)A) hybridizes with poly(U) and

has a T m value of 33°(, while no duplex between d(Ap(OMe)Ap(OMe)A) and

poly(U) is found above 10°(.

Another factor that can influence steric interactions is the chirality of phos­

phorus which is present in all the phosphate-modified systems presented here. From

a molecular model abstracted from a B DNA duplex it follows that two orientations

of the phosphorus substituents are possible. designated outwards and inwards.

corresponding with the two chiral forms of the phosphorus atom. The Cahn-lngold­

Prelog R/S notation can be applied to phosphorus.19 but gives a different label to

spatially identical orientations of the alkyl substituent for the three systems. Model

studies show that the inwards orientation can cause stronger steric repulsions. and

hence a lower duplex stability is present for this phosphorus configuration.

88

Experimental data on this effect are limited. since it requires separation of the diastereomeric forms of the neutra! oligonudeotides. However. for phosphate­methylated DNA it has been found that the phosphorus configuration does not affect its ability to hybridize with natura! DNA. This is most evident for the trinudeotide d(Ap(OMe)Ap(OMe)A). which consists of four diastereomeric forms. Hybridization with poly(dT) gives rise to a single sharp melting transition at 41°C in both UV and NMR experiments. indicating that all diastereomers form an equally stable duplex. A similar behaviour was found for all the diastereomeric mixtures shown in Table 1. A more indirect proof of the similarity of the various diastereomeric forms was provided by the stability studies on the parallel duplexes formed by phosphate-methylated oli­gothymidines (vide supra). Even for double-stranded d([Tp(OMe)]sT). which con­

tains five chiral centres in each strand. a well-defined melting transition at 65°C was found in both UV8 and 1H NMR studies.Sa

For phosphate-ethylated triester systems. a clear difference has been found for the duplex stability of oligonucleotides containing one modified phosphate with a specific chirality. The self-complementary octamer d{GpGpApAp(OEt)TpTpCpC) forms a duplex with a T m value of 31°C for the outward orientation of the ethyl

group. and already melts at 25°C when an inward orientation is present.20 Evidently. the large ethyl group has a much stronger effect on the duplex stability in comparison with a methyl group.

In the case of methyl phosphonate DNA. it should be expected on the basis of steric factors that no difference between the two methyl orientations is present. since methyl is sterically even less demanding than methoxy (vide supra). For small frag­ments (up to four nucleotides) this is indeed the case. but longer fragments of methyl phosphonate DNA exhibit strong variations in their hybridization ability. For example. the hybrid of the decamer d(Tp(Me)[TpTp(Me)]4 T). with alternating phos­phate and methyl phosphonate groups, and poly(dA) has a T m value of 34°( when

one configuration is present. and dissociates at 2°C for the other configuration.1b A sîmilar behavîour is found for the hybridization of this decamer . with the RNA poly(rA): T m = 20°C for the outward form. and T m = ere for the inward form.

Furthermore. the nonamer d((Tp(Me)]8 T) shows a broad range of T m values (0 -25°C) upon hybridization with poly(dA) (see Table 11). which shows the importance of the diastereomeric composition of the oligomer for duplex stability.12 Evidently. other influences besides steric hindrance must cause a difference between the isomers of the chiral methyl phosphonate group. An explanation for this behaviour can be given on the basis of the stereoelectronic effects in phosphate and methyl phos­phonate groups.

89

Stereoelectronic factors

The difference between the hybridization properties of methyl phosphonate and phosphate-methylated DNA with natural DNA may be due to the fact that the methyl phosphonate strands cannot readily adopt the right-handed backbone confor­mation which is needed for the formation of a stable duplex. In particular. this con­cerns the backbone angles ~ (C3•-Ü3·-P-05·) and G' (03·-P-Os·-Cs·). which must reside in the Ç and G'- domains for right-handed DNA.21 In the case of a methyl phosphonate linkage. it can be shown that the combination {G'-. Ç} is energetically unfavoured. Figure 2 depicts the Newman projections of the tl:' - and Ç rotamers for the Sp configuration.

Figure 2. Newman projections of the ar- rotamer, and the~- and~+ rotamers for the methyl phosphonate group with the Sp configuration.

Clearly. both lone pairs on 0 5• have a favourable trans orientation with respect to a phosphorus-oxygen bond22 in the tl:'- rotamer. The Ç rotamer shows an undesirable trans location of the phosphorus-carbon bond and one of the lone pairs on 0 3·• which

can be relieved by a rotation toward the~+ domain. Figure 3 shows the Newman pro­jections of O!- and Ç for the Rp configuration. Now. it appears that the t- rotamer

C5• C3• C5•

Me.*03' 0*05• o*o . . . . 0 Me Me

a- ~ - a•

Figure 3. Newman projections of the ar- and a+ rotamers. and the ~- rotamer for the methyl phosphonate group with the Rp configuration.

corresponds with the advantageous trans orientation of the lone pairs on O:r with

respect to a phosphorus-oxygen bond. Clearly. the unfavoured trans location of one of the lone pairs on Os· and the phosphorus-carbon bond drives a rotation from Il' -

to (l'+. lt should be noted that the above reasoning cannot be verified on the basis of

90

NMR conformational analysis of the methyl phosphonate oligonucleotides in solution. since it is impossible to determine the conformational equilibria around the O' and { linkages from the available set of vicinal proton-proton and proton-phosphorus cou­pling constants.23 However. the crystal structure of the methyl phosphonate dinucleo­tide d(Ap(Me)T) with the Sp configuration in the phosphonate moiety 24 indeed clearly shows the distorted backbone geometry which is realized by a rotation around the P-03• bond from Ç to{+ (see Fîgure 4).

Figure 4. Left: Standard {a-.t-1 conformation as observed in the X-ray crystal structure of natural r(AU}. 25 Right: Distorted (a-,c+J conformation which is found in the X-ray crystal structure of d(Ap(Me)T) with the Sp configuration.24

We feel that the crystal structure of d(Ap(Me)T) strongly supports our argument. that phosphonate internucleotide linkages correspond with a backbone geometry which does not intrinsically match the right-handed DNA conformation. Notwîth­standing. it can be concluded that short methyl phosphonate oligonucleotides may adopt the unfavourable {Cl'-. {-} conformation, since stable hybrids with natura! DNAs are formed (vide supra). For longer strands of methyl phosphonate DNA. the right-handed backbone becomes highly unfavoured. which explains why only poor hybridization is found in these systems. In contrast. the {Cl'-. Ç} conformation is readily available for phosphate:methylated DNA. because trans location of the lone pairs on either 0 5• or 0 3• with a phosphorus-carbon bond is exduded. In agreement with this explanation is the fact that the phosphate-methy lated systems d(Ap(OMe)A). d{Ap(OMe)Ap(OMe)A). and d([Ap(OMe)hA) form more stable hybrids with poly(dT) (T m values 30. 41. and 57°C) than do the corresponding

methyl phosphonate systems d(Ap(Me)A). d(Ap(Me)Ap(Me)A). and d([Ap(Me)hA) (T m values 19. 37. and 45°C).

91

Cooperative effects

The present results show dearly. that phosphate-methylated oligonucleotides hybridize more strongly with natural DNA than do either methyl phosphonate or phosphate-ethylated systems. Nevertheless it must be assumed that the phosphate­methylated strand adopts a somewhat different geometry than natura! DNA. since all intrastrand phosphate-phosphate repulsions are eliminated. The situation may be similar to the complexation between DNA and histones (basic proteins). where a one-sided shielding of the phosphate charges causes a bending of the DNA duplex around the protein.26

We examined the effect of an altered geometry for the phosphate-methylated DNA on the hybridization process of the trinudeotide d(Ap(OMe)Ap(OMe)A) with natural oligo- and poly(dT). The T m value of the hybrid was measured for various lengths of the natura! strand. and different degrees of occupancy. The results of these experiments (see Table 111) clearly show that a certain number of phosphate­methylated trimers must be present to realize a stable duplex. suggesting a coopera­tive behaviour in the hybridization between phosphate-methylated and natura! DNA.

Table 111. T m values for hybrids between d{Ap(OMe)Ap(OMe)A) and natura! oligo- and poly(dT).

System Base Ratio (A:T)

d(Ap(OMe)Ap(OMe)A}poly(dT} 1:1 1:2

d(Ap(OMe)Ap(OMe)A)·d(T 30) 1:1 1:2

d(Ap(OMe)Ap(OMe)A}d(T10) 1:1 d(Ap(OMe)Ap(OMe)A}d(T s) 1:1

T rn ("C)

41 39 27 23

< 10 < 10

Since the length of the natural DNA is of importance. the T m value cannot

solely be determined by the changes in enthalpy and entropy caused by base pair dis­sociation {äH0

0 and as00 for n base pairs). We have therefore put forward 27 that the

natural DNA strand must adapt its conformation in the hybridization. introducing an extra chang~ of entropy &S0k' (for a length of k nudeotides). The cooperative

behaviour can be incorporated via a parameter <l'. ranging from 0 to 1. which decreases the impact of the extra entropy change by successive multiplications (yield­ing <l'i&s0k· for the i'h hybridization). Summing all contributions for p phosphate­

methylated n-mers hybridized with a natural k-mer. it is found that:

92

Tm= 0 - 0,

AS n + p(t _ Cl') AS k

Since AH03 has been previously determined5 to be 23.4 kcal/mol for A-T base pairs.

an analysis of the results in T able 111 in the light of this model can now be performed. lt is found that the values 01 ::: 0.89 . .AS0

3 74.1 cal/mol.K. AS030' = 6.2 cal/mol.K.

and .1.S0pol/ = 4.4 cal/mol.K are consistent with the observations. while ÀS05' and

.1.5°10' must be higher than 8 cal/mol.K.

From this model it appears that modification of the phosphates in one strand has two consequences. viz. (i) stabilization of the duplex due to elimination of elec­trostatic repulsions. giving a higher .AH0

0 • and (ii) destabilization of the duplex due to

structural changes in the modified strand. leading to a higher total entropy change. The balance of these effects has been found to be in favour of stronger duplex forma­tion for the presented results on phosphate-methylated/natural systems (vide supra). A similar shielding of the phosphate groups in one strand can also occur in protein­DNA complexes where cationic amino-acid side chains associate with the negative phosphate groups (vide supra). In most earlier studies. proteins were chosen that complexate with both DNA strands simultaneously.28 so only the stabilizing effect is present. since any stuctural modifications are identical in both strands. However. extrapolating the above results on hybrids between phosphate-methylated and natura! DNA. we may expect a different result from protein complexation with the phosphate groups in one strand only. A structural difference between the strands will occur. and the in enthalpy and entropy of hybridization will be different from the phosphate-methylated/natural systems. which may effectuate destabilization of the DNA duplex.

In order to study whether this type of duplex destabilization can occur. we have performed thermodynamic model studies on a DNA duplex with A-T base pairs which is complexed with protein in one strand only.29 For these calculations (see also Chapter 6}. we adapted the model of Breslauer et al.30 for determining the enthalpy and free energy of hybridization in natura! DNA duplexes. They found that summing fixed contributions (Ah and Ag) for all dinucleotide units in a duplex gave correct values for these thermodynamic quantitiés. Using different dimer contributions Ah* = 13· Àh and lig* = 8· Ag for the A-T base pairs in our model duplex. and the abovementioned formula for the T m value of shielded duplexes. we could evaluate the stability of a unilaterally protein-shielded duplex. The results of the calculations show.29 that the effect of protein shielding on the duplex stability depends mainly on the ratio 8/ 13. For 8/ /3 > 1. additional stabilization of the duplex is always found. independent of the number of shielded phosphates. When 8 / /3 < 1. an increase of the length of unilateral shielding results in a lower T m value for the duplex.

93

Th!s latter process may be of interest for understandîng the action of enzymes, whiçh çan open a DNA duple.11. for replication or transçrîption. Several structural

features known for these proteins coincide with our model assumptions. For

en.zymes that unwind the DNA dupleli. completely for repli~ation (heiicases). ît is

known that only a short stretch {ca. 8 base pairs) of the DNA is covered by the pro­

tein.31_ More specificalty, the so-called gene 5 proteîn of phage fd. is known to have a

high affinity for a single Strand of DNA. and to assocîate primarily with 5 adjacent

phosphate groups via lysir\e and argînîne residues.32 For the RNA polymerase

enzymes that partially unwlnd a DNA dup~ex fOr transcription, it is established that

e.g. E. coli RNA polymerase always remains assocîated with one of the DNA strands

over' a length of 14 base p.3trs'. 33 For many other RNA polymerases. it is reported that

lengths of 17 base pairs34 up to 70 base pairs31 ol the DNA duplex are covered by the

protein. Also, the interactîon between DNA and proteïn in this case is thought to be

mainly electrostatic. 36 In good agreement with the above facts, our results show2~ that destabilization of a DNA duplex by phosphate shielding in the interior of the

sequence (as in transcrîption) requires a modîfied stretch which is almost twîce as

long. compared to a similar duplex opening at the end of the sequence (as in replica­

tïon of !Jnear DNA). lt therefore may be possible. that unilateral protein compleKation

serves as an initiatîng step for processes that invofve the di~sociation of the DNA duplex.

EXPERIMENTAL SECTION

1H-NMR spectra were run in the FT mode at 200, 500 ar 600 MHz37 on Bruker

spectrometers. Measurements in water refer to an 85:15 (v/v) mixture of H'20 and

0 20. in wh•ch the deuterium provided the field-frequency tock. The strong water sig­

nal was suppressed with the technique as described by Haasnoot et atlli Proton

chernical shifts were referenced against HDO (S 4J18 ppm). llp NMR spectra were

run in the FT mode at 80.9 MHz on a Bruker AC 200 spectrometer. Chemica! shifts

are relativè to 85% H3P04: they are designated positive if downfield from the stan­

dard. The UV hyperchromicity measuréments were performed on a Perkin~Elmer 124 spectrophotometer. using 10-mm cuvettes and a wavelength of 260 nm. 'The natura!

oligonudeotides d(T 5). d(Trn)- and d(T 1-0) were prepared via the standard phos­

phorus amidite protocol13 on an Appiied Biosystems 381A DNA synthesizer.

Purification was performed With alcohol precipitation.39

94

d(Ap(OMe)A), d(Ap(OMe)Ap(OMe)A), and d((Ap(OMe)],A)

Starting compound 3' -0-acetyl-6-N- ( 1-( dimethylamino j-ethylidene)-2' -deoxyadenosine

5' -0-(4.4' -Dimetho•ytrityl)-6-N-(1-( dimethylamino)ethylidene)-2' -deoxy- adem>­sine14 (5 mmol, 3.2 g) was ·coevaporated three times with pyridine Lo remove water.

and dissolved in 15 ml pyridine. Acetic anhydride (12 mmol. 12.2 g) was added. and the solution was stîrred overnighL Pyridine was then removed by coevaporation with water. and the yellow residue was dissolved in SO ml chloroform, To eliminate e11.cess

acetîc anhydride. thfs solution was treated with a 5% aqueous sodium bicarbonate

soJution. and extracted with thr.ee 25 ml portions of water. Concentration of the organic layer afforded a white foam, which was dissolved in 30 ml of

nitromethane/methanol (95 : 5 v/v) and treated with anhydrous zine bromide.(31.1 mmol. 7J9 g} for 1 ~· To the red solution. 150 ml of a 5% aqueous ammonium ace­tate solution. \Vas added, followed by eAtractlon with four portions of 50 ml dichloromethaoe. The organic layer was dried· o~er sodium sulfate and concentrated. The brown residue was purified by column chromatography on a Woelm silica gel column. First. Împurities were eluted with dichloromethaoe. after whîch the product was taken from the column with dichloromethaoe/methanol (95:5 v/v) as elueot (R1

0.16). The product was obtained as a white loam in 78% yield (2.2S g). 1H NMR (CDCl,j: ó 2.11 {3H. s. N=C-CH 3). 2.16 (3H. s. CH, acetyl). 2.42 (1H. m. H,.-).

3.15 (6H. s. N(CHiJ,). 3.25 (!H. m. Hr)- 3.94 (2H. m. Hs·/H5"). 4.27 (lH. m. H").

5.58 (lH. m. H3·J. 6.30 (1H. dd. H,.j. 7.98 (1H. s. H2). 8.58 (1H. s. H8).

Coupling compound 3' -0-( (N.N-dii•opropylamioo )methoxyphosphino )-5' -0-( 4,4' -dimethoxy trityl)-6-N -( 1-( dimethylamino )·ethylidene)-2' - · deoxyadenosine40

5' -0-(4 ,4' -Dimethoxytrityl)-6-N-(1-( dimethylamino)ethylidene)-2' - deoxyadeoo­sine14 (19.62 mmol. 12.2 g) and diisopropylethylamine (80 mmol, 14 ml) were dis­solved in 60 ml chloroform under an argon atmosphere. and chloro-(N.N­diisopropyalmino)-methoxyphosphine•• (21.6 mmol. 4.32 ml) was added over 90 .s while stirring. Alter 1 h. the mixture was diluted with 500 ml ethyl acetate (pre­washed wîth sodium bïcarbonate). and washed wlth foor mml portions of a saturated sodium chloride soiution and 75 ml water. The solution was ·dried over sodiom sulf ate. and concentrated. The resulting foam was chromatographed on a Woelm silica gel column. using dichloromethane/hexaoe/triethylamine (5:4:1 v/v/v) as eluent (R1 = 0.40). The product was obtaioed as a yellowish foam in 70% yield

(10.76 g). and stored onder argon. 1H NMR (CD3CN): ó 1.12 '(12H, m. CH 3 isopro­

pyl). 2.01 (3H. s. N=C-CH 3). 2.53 {lH. m. Ht·). 3.01 (lH. m. H,-). 3.10 (12H. s.

N(CH3)i). 3.26 (2H. d, H5'fH 5--). 3.29 (3H. d. POCH 3• J = 13.2 Hz). 3.36 (3H. d.

95

POCH3• J 13.2 Hz). 3.55 (2H. m. H isopropyl). 3.73 (6H. s. OCH3). 4.17 (1H. m.

H4-}. 4.83 (1H. m. H3·). 6.38 (1H. dd. Ht'). 6.73-7.40 (13H. m. trityl). 8.05 (1H. s.

H2). 8.37 (1H. s. H8): 31 P NMR (CD3CN): 8 154.40 and 154.66.

Coupling reaction, oxidation, and detritylation to form 3'-0-acetyl-di-(6-N-(l­( dimethylamino) ethylidene) -2' -deoxyadenyl)-( 3' ...... 5' )-0-methylpho sphate13

The starting compound 3' -O-acetyl-6-N-(1-(dimethylamino)-ethylidene)-2' -

deoxyadenosine (1 mmol. 0.36 g) and the coupling compound (1 mmol. 0.78 g) were each coevaporated three times with pyridine. and then dissolved in pyridine. To start the reaction. 1H-tetrazole (5 mmol. 0.35 g) was added. The solution was stirred for 2

h. and concentrated. 31 P NMR (CD3CN): 8 146.13 and 145.91 (ratio 1.08:1).

The phosphite was then oxidized by addition of a solution of iodine (1 mmol.

0.26 g) in acetonitrile/lutidine/water (3:2:1 v/v/v). After 5 min. the brown solution was diluted with 100 ml ethyl acetate. and extracted with 20 ml of a 1 % aqueous sodium bisulfate solution. The organfc layer was dried on sodium sulfate. and concen­

trated. The resulting viscous oil was chromatographed on a Woelm silica gel column.

using dichloromethane/methanol (90:10 v /v) as eluent (R1 = 0.40). The yield was

30% (0.32 g}. 31 P NMR (CDC13): S 0.05 and -0.33 (ratio 1.09:1).

The product was then dissolved in 10 ml of acetic acid/water (4:1 v /v). and the solution was heated to 5G°C for 2 h. Then 25 ml cold water was added to precip­itate the 4,4' -dimethoxytrityl alcohol. Filtration and coevaporation with water yielded

the dinucleotide in a yield of 81% (0.18 g). R1(CH 2Cl2/CH 30H) = 0.26. 31 P NMR

(CD30D): 8 3.34 and 3.29.

d{Ap(OMe)A)

The final acetylation of the 5' -hydroxyl group was performed as described for the starting compound. Then. the N6-protecting groups were removed14 by dissolving

the material at 40"( in 1.2-ethanediamine/phenol/water (2:8:1 v/w/v). After 1 h, the mixture was coevaporated with water. and the product was isolated with two­

dimensional preparative thin layer chromatography: R1 = 0.23 (eluent

dichloromethane/methanol 70: 30 v/v). 1H NMR (D20): S 2.08 (3H. s. CH 3 acetyl).

2.12 (3H. s. CH3 acetyl). 2.33 (1H. m. Hr of dAp). 2.44 (1H. m. Ht of dAp). 2.62

(1H. m. Hr of pdA). 2.84 (1H. m. Hr of pdA). 3.49 (2H. m. H5'/Hs" of dAp). 3.76

(3H. d. POCH3. J = 11.4 Hz). 4.22 (1H. m. H4· of dAp). 4.39 (3H. m. H4-/Hs·/Hs" of

pdA). 5.00 (1H. m. H3• of dÀp). 5.45 (1H. m. H3• of pdA). 6.03 (1H. dd. Hl' of dAp).

6.26 (1H. dd. H1• of pdA). 7.88 (1H. s. H2 of dAp). 7.96 (1H. s. H2 of pdA). 8.02 (1H.

s. H8 of dAp). 8.12 (1H. s, H8 of pdA). 31 P NMR (020): 8 1.62 and 1.72. UV: "-ma• = 257 nm. A2so/ A260 = 0.83. A280/ A260 = 0.10.

96

d(Ap(OMe)Ap(OMe)A)

The sequence of coupling with the amidite. oxidation. and detritylation was repeated to add another residue. Coupling gave: 31 P NMR (CD3CN): o 146.34 and 145.68 (ratio 1.17:1) for the phosphite linkage. Oxidation: 31 P NMR (CD3CN): 8 4.43 and 4.33. After detritylation. the free 5' -hydroxyl group was acetylated as described for the starting compound. Then. the N6-protecting groups were removed according to the procedure for the dinucleotide. The product was isolated with two­dimensional preparative thin layer chromatography. using dichloromethane/methanol (70:30 v /v) as eluent (Rt = 0.12). 31P NivlR (D20): 8 3.87. UV: Àmax = 257 nm. A2sol A260 = 0.94. Awo/ A260 = 0.12. The 1H NMR spectrum is too crowded to give a complete assignment.

d([Ap(OMe)]JA)

After another coupling protocol the tetranucleotide was obtained with two­dimensional thin layer chromatography in a minute quantity. so no NMR data are available (R1 = 0.07. eluent dichloromethane/methanol 70:30 v /v}: UV: Àmax 260

nm. A250/ A200 = 0.83. A280/ A260 = 0.16.

d(Cp(OMe)C) and d(Cp(OMe)Cp(OMe)C)

4-N- (9-Fluorenylmethoxycarbonyl}-2' -deoxycytidine

2' -Deoxycytidine ( 4.54 g. 20 mmol) was dried by coevaporation with dry pyri­dine and subsequently suspended in 100 ml of dry pyridine. 13.0 ml (100 mmol) of trimethylchlorosilane was added dropwise and the mixture was stirred for 15 min.41

Then 5.8 g (22 mmol) of 9-fluorenylmethoxycarbonylchloride was added and the reaction mixture was stirred for 2 h at room temperature. yielding a white suspension in the yellow solution. Hydrolysis of the trimethylsilyl groups and unreacted chlorides was effected by addition of 20 ml water. After stirring for 15 min. a clear yellow solution was obtained, which was evaporated to near dryness by coevaporation with toluene. Upon addition of 300 ml water to the resulting oil. a white precipitate appeared. The mixture was shaken vigorously until no more yellow oil was visible. After addition of 150 ml of ethyl acetate and shaking. a precipitate formed on the separation layer. which was isolated by filtration and washed with ethyl acetate. After drying in vacuo the product was obtained as a white solid (8.71 g. 97%). 1H NMR (acetone-d6): 8 2.40 (1H. m. H2·). 2.59 (1H. m. Hr). 3.77 (2H. m. H5-fHS"). 4.03 {1H. m. H4·). 4.33 (1H. t. CH Fmoc). 4.49 (3H. m. H3'fCH2 Fmoc). 6.23 (1H. t. Hi'). 7.12 (1H. d. Hs). 7.37 (4H. m. aromatic Fmoc). 7.85 (4H. m. aromatic Fmoc).

8.50 (1H. d. H6)·

97

5' -0-{4-Monomethoxytrityl)-4-N-(9-fluorenylmethoxycarbonyl)-2' -deoxycytidine

Base-protected 2' -deoxycytidine (8.55 g. 19 mmol) was dried by coevaporation with dry pyridine and subsequently suspended in 200 ml of dry pyridine. After addi­tion of 6.5 g (21 mmol) of 4-monomethoxytritylchloride. the solution was stirred for 15 h. The mixture was then poured in a saturated aqueous sodium bicarbonate solu­tion (250 ml). and extracted three times with dichloromethane. The combined organic layers were washed with a saturated aqueous sodium bicarbonate solution and dried on magnesium sulfate. After filtration. the solution was concentrated in vacuo. Removal of all pyridine from the brownish oil was accomplished by coevapora­tion with toluene (twice) and 2-butanone. The resulting brown foam was purified by column chromatography on Woelm silica gel with 2-butanone as eluent R1 = 0.37).

yielding 8.90 g (65%) white solid product. 1H NMR (CDC13): 8 2.25 (1H. m. Hr).

2.71 (lH. m. Hr). 3.47 (2H. m. H5,/H5"). 3.79 (3H. s. OCH 3 trityl). 4.14 (lH. m.

H4·). 4.29 (1H. t. CH Fmoc). 4.50 (3H. m. H3'/CH2 Fmoc). 6.28 (1H. t. Hr). 6.85 (2H. m. trityl), 6.99 (1H. d. H5). 7.31 (12H. m. trityl). 7.41 (4H. m. aromatic Fmoc).

7.69 (4H. m. aromatic Fmoc). 8.22 (1H. d. H6).

Starting compound 3' -O-acetyl-4-N-(9-fluorenylmethoxycarbonyl}-2' -deoxycytidine

The 5' -tritylated. base..protected 2' -deoxycytidine (5.00 g. 7 mmol) was stirred for 15 min in a mixture of 50 ml of dry pyridine and 5 ml of acetic anhydride. The mixture was then poured in 50 ml of a saturated aqueous sodium bicarbonate solu­

tion" and extracted twice with dichloromethane. The combined organic layers were dried on magnesium sulfate. After filtration. the solution was concentrated in vacuo. The resulting oil was purified by column chromatography on Woelm silica gel with 2-butanone as eluent (R1 = 0.65). yielding a white solid. This was stirred for 15 h in a

mixture of 160 ml acetic acid and ~O ml water. After addition of another 120 ml w,ater. the reaction mixture was concentrated in vacuo to near dryness. The excess acetic acid and water were removed by coevaporation with 2-butanone. Crystallization from 2-butanone yielded the product as a white solid (2.04 g. 57%). 1H NMR (CDCl3/CH30D 9:1 v/v); 8 2.12 (3H. s. CH 3 acetyl). 2.27 (1H. m. H2·). 2.66 (1H. m.

Hr). 3.86 (2H. m. Hs·/HS"). 4.20 (1H. m. H4·). 4.31 (1H. t. CH Fmoc). 4.55 (2H. d.

CH 2 Fmoc). 5.33 (1H. m. H3·). 6.29 (1H. t. Hr). 7.36 (SH. m. H5/aromatic Fmoc).

7.74 (4H. m. aromatic Fmoc). 8.39 (1H. d. H~J

Bis-( N ,N-diisopropylamino )-methoxyphosph ine

139.8 g (1.02 mol) phosphorus trichloride was cooled to -8°C. and 32.53 g (1.02 mol) dry methanol was added dropwise over 1 h while stirring. The produced hydro­chloric acid was absorbed in a gas trap containing water. Pure methoxy-

98

dichlorophosphine was obtained by distillation at 100 mmHg (b.p. 35°C). 31 P NMR (CDC13): 8 181.5. A solution of N.N-diisopropylamine (162.1 g, 1.6 mol) in 500 ml

dry diethyl ether was cooled to 0°C. and a solution of methoxy-dichlorophosphine (53.37 g. 0.4 mol) in 20 ml dry diethyl ether was added dropwise over 1 h. The ammonium salt was then removed by filtration. and the solution was concentrated in vacuo. Pure bis-(N.N-diisopropylamino)-methoxyphosphine was obtained by distilla­tion of the residue at 1 mmHg (b.p. 115°C) in a yield of 20.9 g (20%). 1H NMR (CDC1 3): S 1.20 (24H. m. CH 3). 3.35 (3H. d. POCH3• J 11.2 Hz). 3.52 (4H. m.

CH). 31 P NMR (CDCl3): 8 131.7.

Coupling reaction, oxidation, and detritylation to form 3' -0-acetyl-di-(4-N­{9-fluorenylmethoxycarbonyl)-2' -deoxycytidyl)- (3' ......+ 5')-0-methylphosphate

2.00 g (2.8 mmol) of 5' -tritylated. base-protected 2' deoxycytidine was dried by coevaporation with dry pyridine. and subsequently dissolved in 20 ml of dry pyridine. Then. 0.10 g (1.4 mmol) of 1H-tetrazole and 0.86 g (3.3 mmol) bis-(N.N­diisopropylamino)-methoxyphosphine (dried by coevaporation with dry pyridine) were added. and the reaction mixture was stirred at room temperature for 15 min. Forma­tion of the phosphoramidite coupling compound in situ was evident from the 31 P NMR spectrum (CDCl3: S 149.5 and 150.2). Then. 1.65 g (3.36 mmol) of the starting

compound 3' -O-acetyl-4-N-(9-fluorenylmethoxycarbonyl)-2' -deoxycytidine and 0.50 g (7 mmol) 1H-tetrazole were added. and the reaction mixture was stirred at room tem­perature for another 2 h. The obtained phosphite (31P NMR (CDC13): S 140.4 and 140.8) was oxidized by addition of 1.5 ml of tert-butylhydroperoxide.42 giving the phosphate (31P NMR (CDC13): 8 -0.33 and -0.44}. After stirring for 5 min. the .mix­ture was concentrated in vacuo to near dryness and coevaporated twice with toluene and 2-butanone. The product was detritylated by addition of a mixture of 100 ml acetic acid and 25 ml water. After stirring for 15 h. another 75 ml water was added. and the mixture was concentrated in vacuo to near dryness. The excess acetic acid and water were removed by coevaporation with 2-butanone and ethyl acetate. The product was purified by column chromatography on Woelm silica gel. First. impurities were eluted with ethyl acetate/dichloromethane (9:1 v/v): then the product was obtained with ethyl acetate/dichloromethane/ethanol (85:10:5 v /v /v) as eluent (R1 = 0.25). with a yield of 1.25 g (44%). 1H NMR (CDCl3): 8 2.10 (3H. s. CH 3 acetyl).

2.43 (2H. m. H2• of dCp and pdC). 2.77 (2H. m. Hr of dCp and pdC). 3.84 (3H. d. POCH3• J = 11.0 Hz). 3.91 (2H. m. H5·/H5" of dCp). 4.2-4.5 (10H. m. H4• of

dCp/H 4-fH5-fH5" of pdC/CH/CH2 Fmoc). 5.29 (2H. m. H3• of dCp and pdC). 6.24 (2H. m. Hr of dCp and pdC). 7.30 (10H. m. Hs of dCp and pdC/aromatic Fmoc).

7.53 (4H. m. aromatic Fmoc). 7.77 (4H. m. aromatic Fmoc). 8.07.(1H. d. H6 of dCp). 8.20 (1H. d. H6 of pdC). 31P NMR (CDC13): 8 -0.35 and 0.07 ppm.

gg

d(Cp(OMe)C)

The base-protecting Fmoc groups were removed15h by treatment with a ipixture of triethylamine (20 mmol/mmol Fmoc) and of pyridine (10 ml/mmo! Fmoc}. After ca. 3 h. thin layer chromatography showed the reaction to be complete. The reaction mixture was concentrated in vacuo. and the excess pyridine and triethylamine were removed by coevaporating three times with toluene. Extraction of an aqueous solu­tion of the residue with dichloromethane removed virtually all split-off fluorene groups. 1H NMR (020): 8 2.06 (3H. s, CH 3 acetyl). 2.30 (2H. m. H2• of dCp and

pdC). 2.56 (2H. m. Hr of dCp and pdC). 3.70 (2H. m. H5-/Hs" of dCp). 3.78 (3H. d,

POCH3• J = 11.4 Hz). 4.20 (1H. m. H4• of dCp). 4.32 (3H. m. H4·/H5)H5" of pdC).

4.96 (1H. m. H3• of dCp). 5.28 (1H. m. H3· of pdC). 5.98 (lH. d. H5 of dCp). 5.99

(1H. d, H5 of pdC). 6.15 (1H. dd. Hr of dCp). 6.18 (1H. dd. Hr of pdC). 7.69 (1H. d. H6 of dCp). 7.73 (1H. d. H6 of pdC). 31P NMR (020): 8 1.96 and 2.07.

d( Cp( OMe)Cp( OMe )C)

The sequence of coupling with the amidite. oxidation. and detritylation was repeated to add another residue. Coupling gave: 31P NMR (CDC13): 8 140.7. 0.65. and -0.40. Oxidation: 31 P NMR (CDC13): 8 -0.02, -0.05. -1.00. and -2.00. After detrityla­

tion. the product was purified by column chromatography on Woelm silica gel with chloroform/methanol (9:1 v/v) as eluent (R1 = 0.11). 31 P NMR (CDC13): ó -0.13 and -0.27. The base-prötecting Fmoc groups were removed by treatment with pyridine and triethylamine.15h Evaporation of the pyridine yielded a crude product. which was purified by extraction of an aqueous solution with dichloromethane. The 1H NMR spectrum is too crowded for complete assignment. 31 P NMR (020): 8 1.9-2.3 and

3.2. UV: Àma~ = 268 nm. A2so/ A210 = 0. 76. A290/ A210 = 0.26.

REFERENCES

1. (a) L.H. Koole. E.J. Lanters and H.M. Buck. J. Am. Chem. Soc. 106. 5451 (1984). (b) P.S. Miller. N. Dreon. S.M. Pulford and K.B. McParland. J. Biol. Chem. 255. 9659 (1980). (c) J.W. Sugas and O.A. Taylor, Nucleic Acids Res. 13. 5707 (1985).

2. (a) P.A. Bartlett and F. Eckstein, J. Biol. Chem. 257. 8879 (1982). (b) R. Jar­vest. G. Lowe. J. Baraniak and W.J. Stee. Biochem. J. 203. 461 (1982). (c) P.A. Usher. D.J. Richardson and F. Eckstein. Nature {london) 228. 663 (1970). (d) R.S. Brody. S. Adler. P. Modrich. W.J. Stee. Z.J. Lesnikowski and P.A. Frey, Biochemistry 21. 2570 (1982).

3. (a) P.S. Miller, K.N. Fang. N.S. Kondo and P.O.P. Ts'o, J. Am. Chem. Soc. 93. 6657 (1971). (b) P.S. Miller. K.B. McParland. K. Jayaraman and P.O.P. Ts'o.

100

Biochemistry 20. 1874 (1980).

4. P.S. Miller. LT. Braiterman and P.O.P. Ts'o, Biochemistry 16. 1988 (1977).

5. L.H. Koole. M.H.P. van Genderen. R.G. Reiniers and H.M. Buck. Proc. K. Ned. Akad. Wet .. Ser. B 90. 41 (1987) (communicated by H.M. Buck at the meeting of Dec 15. 1986).

6. (a) H. Weintraub. J.G. lzant and R.M. Hartland. Trends Genet. 1. 22 (1985). (b} M. Lemaitre, B. Bayard and B. Lebleu. Proc. Natl. Acad. Sci. USA 84. 648 (1987). (c) P.S. Miller. C.H. Agris. K.R. Blake. A. Murakami. SA Spitz. P.M. Reddy and P.O.P. Ts'o. in "Nucleic Acids, The Vectors of Life". B. Pullman and J, Jortner. Eds" Reidel Publishing Company, Boston. 1983. pp. 521-535.

7. L.H. Koole. P.J.L.M. Quaedflieg. W.H.A. Kuijpers. N.L.H.L. Broeders. H.A. Langermans. M.H.P. van Genderen and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 205 (1988) (communicated by H.M. Buck at the meeting of Jan 25. 1988).

8. (a) L.H. Koole. M.H.P. van Genderen and H.M. Buck. J. Am. Chem. Soc. 109. 3916 (1987). (b} L.H. Koole. M.H.P. van Genderen. H. Frankena. H.J.M. Koeken. J.A. Kanters and H.M. Buck. Proc. K. Ned. Akad. Wet.. Ser. B 89. 51 (1986) (communicated by H.M. Buck at the meeting of Nov 25. 1985).

9. (a) L.H. Koole. N.L.H.L. Broeders. M.H.P. van Genderen. P.J.L.M. Quaedflieg. S.J. van der Wal and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 245 (1988) (communicated by H.M. Buck at the meeting of May 30. 1988). (b) M.H.P. van Genderen. M.P. Hilbers. P.J.L.M. Quaedflieg. L.H. Koole and H.M. Buck. J. Am. Chem. Soc" submitted for publication.

10. (a) M.H.P. van Genderen. L.H. Koole and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 90. 181 (1987) (communicated by H.M. Buck at the meeting of Jun 22. 1987). (b} M.H.P. van Genderen. L.H. Koole and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 171 (1988) (communicated by H.M. Buck at the meeting of Nov 30. 1987).

11. P.S. Miller, J. Yano. E. Yano. C. Carroll. K. Jayaraman and P.O.P. Ts'o, Biochemistry 18. 5134 (1979).

12. R.C. Pless and P.O.P. Ts'o, Biochemistry 16. 1239 (1977).

13. T. Atkinson and M. Smith. in "Oligonucleotide Synthesis. A Practical Approach". M.J. Gait. Ed" IRL Press Ltd" Oxford. 1984, Chapter 3 and refèr­ences cited therein.

14. L.J. McBride. R. Kierzek. S.L. Beaucage and M.H. Caruthers. J. Am. Chem. Soc. 108. 2040 (1986}.

15. (a) C. Gioeli and J. Chattopadhyaya, J. Chem. Soc" Chem. Commun. 263 (1982). (b) J. Heikkilä and J. Chattopadhyaya. Acta Chem. Scand. 837, 263 (1983}. {c} T.R. Webb and M.D. Matteucci. Nucleic Acids Res. 14. 7661 {1986).

101

(d} E. Happ. C. Scalfi-Happ and S. Chladek. J. Org. Chem. 52. 5387 (1987).

16. K. Wüthrich. "NMR of Proteins and Nucleic Acids". J. Wiley & Sons. New York. 1986. Chapter 2.

17. W. Saenger. "Principles of Nucleic Acid Structure". Springer Verlag. New York. 1984. Chapter 11.

18. See ref. 17. Chapter 9.

19. R.S. Cahn. C. lngold and V. Prelog. Angew. Chem. 78. 413 (1966).

20. M.F. Summers. C. Powell. W. Egan. R.A. Byrd. W.D. Wilson and G. Zon. Nudeic Acids Res. 14. 7421 (1986).

21. (a) See ref. 17. Chapter 4. (b) Nomenclature for nucleic acid conformations fol­lows the IUPAC-IUB recommendations. See: Eur. J. Biochem. 131. 9 (1983).

22. (a) S. Wolfe. Acc. Chem. Res. 5. 102 (1972). (b) A.J. Kirby. "The Anomeric Effect and Related Stereoelectronic Effects at Oxygen", Springer Verlag. New York. 1983.

23. C. Altona. Reel. Trav. Chim. Pays-Bas 131. 9 (1982).

24. K.K. Chacko. K. Lindner. W. Saenger and P.S. Miller. Nucleic Acîds Res. 11. 2801 (1983).

25. N.C. Seeman. J.N. Rosenberg. F.L Suddath. J.J.P. Kim and A. Rich. J. Mol. Biol. 104. 109 (1976).

26. See ref. 17. Chapter 19.

27. M.H.P. van Genderen. L.H. Koole. H.M. Moody and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 91. 59 (1988) (communicated by H.M. Buck at the meeting of Nov 30. 1987).

28. M. Tsuboi. in "Conformation of Biopolymers". Academie Press. New York. 1967. Vol. ll. pp. 689-702.

29. M.H.P. van Genderen and H.M. Buck. Biopolymers (1989). in press.

30. K.J. Breslauer. R. Frank. H. Blöcker and LA. Marky. Proc. Natl. Acad. Sci. USA 83. 3746 (1986).

31. (a} M. Szekely. "From DNA to Protein". McMillan Press Ltd" London. 1980. Chapter 2. (b} J.W. Chase. Ann. Rev. Biochem. 55, 103 (1986).

32. (a) G.O. Brayer and A. McPherson. J. Mol. Biol. 169. 565 (1983). (b) J.E. Cole­man. R.A. Anderson. R.G. Ratcliffe and I.M. Armitage. Biochemistry 15. 5419 (1976).

33. (a) D.C. Hinkle and M. Chamberlin. J. Mol. Biol. 70. 157 (1972). (b) W.F.

102

Mangel and M. Chamberlin. J. Biol. Chem. 249. 2995 (1974). (c) M. Chamber­lin. in "RNA Polymerase". R. Losick and M. Chamberlin. Eds" Cold Spring Har­bor Press. New York. 1976. p.159.

34. A. Schmitz and D.J. Galas. Nucleic Acids Res. 6. 111 (1979).

35. P.H. von Hippel. D.G. Bear. W.D. Morgan and J.A. McSwiggen. Ann. Rev.

Biochem. 53. 389 (1984).

36. S.L. Shaner. K.S. Lee. R.R. Burgess and M.T. Record. Jr.. Cold Spring Harbor Symp. Quant. Biol. 47. 463 (1983).

37. 200 MHz NMR spectra were measured on a Bruker AC 200 spectrometer at the

Eindhoven University of Technology; 500 and 600 MHz spectra were .measured

on Bruker AM 500 and AM 600 spectrometers at the Dutch National High Fre­

quency NMR Facility in Nijmegen, The Netherlands.

38. C.A.G. Haasnoot and C.W. Hilbers. Biopolymers 22. 1259 (1983).

39. T. Maniatis. E.F. Fritsch and J. Sambrook. "Molecular Cloning", Cold Spring

Harbor Laboratory. New York. 1982. pp. 461-463.

40. S.L. Beaucage and M.H. Caruthers. Tetrahedron Lett. 22. 1859 (1981).

41. G.S. Ti. B.L Gaffney and R.A. Jones. J. Am. Chem. Soc. 104. 1316 (1982).

42. J. Engels and A. Jäger. Angew. Chem. Suppl.. 2010 (1982).

103

CHAPTER6*

Protein complexation with DNA phosphates as a cause for DNA duplex destabilization. A thermodynamic model

ABSTRACT

Complexation of positively charged sites in a protein with the negative DNA phosphate groups shields the phosphate charges. When this occurs in both strands. interstrand electrostatic repulsions are diminished. which stabilizes the duplex. When phosphate shielding is present in one DNA strand only. the conformation of this strand changes due to a decrease of intrastrand phosphate-phosphate repulsions. This destabilizes the duplex since then the strands differ in conformation. A thermo­dynamic model is formulated to describe this stabilization/destabilization effect in terms of changed enthalpies and entropies of hybridization. lt is found that protein complexation with one DNA strand can indeed lower the T m value of a duplex. The model is applied to the action of helicases (replication). RNA polymerases (transcrip­tion). and restriction endonucleases. Mechanisms with unilateral charge shielding are proposed for their duplex-destabilizing properties.

*M.H.P. van Genderen and H.M. Buck. Biopolymers (1989). in press.

104

INTRODUCTION

The ability of enzymes to unlock the genetic information in the interior of double-stranded DNA is one of the most important processes in biochemistry. Fora good understanding of the way these enzymes work. it is necessary to realize that the exterior sugar-phosphate backbones are the primary recognition sites for a protein. One of the strongest and most obvious interactions between proteins and the DNA backbone is the electrostatic attraction between positively charged residues in the enzyme (e.g .. lysine or arginine) and the negatively charged backbone phosphates.1·2

At a first view. one would expect that this interaction. which shields the phosphate charges. always stabilizes the DNA duplex by decreasing interstrand electrostatic repulsions. lndeed. model studies with polycationic proteins such as polylysine usually feature a markedly higher duplex stability.3- 5 However. in these cases the protein always associates with both DNA strands simultaneously.

In this chapter. we wish to present a way in which protein complexation can lead to destabilization of a DNA duplex. even though phosphate charges are shielded. This idea is derived from our work on hybrids of phosphate-methylated and natura! DNA.6- 9 where a perfect charge shielding is present in one strand due to methylation of the phosphate groups. Stability studies on these neutral/natural hybrids have shown that unilateral charge shielding has. two effects. viz. (i) diminished interstrand repulsions give duplex stabilization. and (ii) diminished intrastrand repulsions are present in the shielded DNA strand. which changes its conformation. Duplex desta­bilization now occurs due to structural differences between the shielded and natural strands. In a thermodynamic description. the first effect yields a more favourable enthalpy of hybridization. whereas the second effect is evident in a less favourable hybridization entropy. So. the overall effect of unilateral charge shielding is deter­mined by the balance of these two effects. and is visualized in the melting tempera­ture (T m) of the duplex.

Using thermodynamic model calculations. it will be demonstrated that hybrids between phosphate-methylated and natural DNA are invariably more stable than their natura! counterparts. whereas complexation of a protein with one strand may result in duplex destabilization. The thermodynamic model will be applied to various types of enzymes that interact with duplex DNA.

METHODS

The duplex stability is evaluated as the duplexo;:t coil transition temperature T m·

which can normally be expressed as the ratio of the enthalpy and entropy of dissocia­.ó. Ho

tion: T m = 85

/ . where n represents the number of base pairs. In this equation. we n

neglect effects of DNA concentration and ionic strength. which are known to

influence duplex stability. We assume that these factors are equal for all studied

105

systems. and will evaluate onfy the consequences of unilateral charge shielding on the thermodynamic parameters. In our studies on hybrids between phosphate­methylated and natura! DNA.8 we found that an extra entropy term had to be added to the equation for Tm to account for the structural ditTerence between the natura! and the neutral strands: AS0k'. where k represents the length of the natural DNA that has to adapt itself to the modified strand. One then obtains:

T - AHon "' - AS0

0 + AS6k'

According to the work of Breslauer et al.. 10 the enthalpy and free energy of dissocia­tion fora natura! duplex (t.H0n and nG0

0 ) must be thought of as the algebraic sum of all possible dimer contributions (Ah and Ag). to account for hydrogen bonding and stacking interactions. For the free energy. it is necessary to indude a start value (Ag;). In our model calculations. we used a duplex consisting of a strand with A­bases and a strand with T-bases. both n nucleotides long. In such a duplex. n-1 M/TT dimers are present that contribute to the enthalpy and free energy. The Breslauer model now states that:

t.H00 = è.h(n - 1). and

t.G 00 = !i.g(n - 1) - lig;

For several natural duplexes. t.H 00 and t.G0

0 have been determined. based on the melting behaviour. From these experimental data. the M/TT dimer contributions Ah {9.1 kcal/mol) and Ag (1.9 kcal/mol). and the start value Ag; (6.0 kcal/mol) were abstracted.18 The standard entropy change can now be calculated from:

50 ..:_ AH0

0 - AG0n _ Ah - é.g ( -l) f.g;

A. n - ·298 - 298 n + 298

The adaptation entropy AS0k' for hybridization between phosphate-methylated and natural DNA has been determined8 from hybridization experiments for several values of k: t.S0

P01; = 4.4 cal/mol.K. t.S030' = 6.2 cal/mol.K. and t.S0

10' and t.5°5' > 8 cal/mol.K. A continuous function was fitted to these experimental data:

ASO , = 4.4k + 83.0 k k + 4.7

The behaviour of this function mirrors the observations that A.S0k' approaches 4.4 cal/mol.K for polynucleotides (large k). and rises sharply for shorter lengths (k < 30). Since our experimental data represent an average adaptation entropy in the interior and on the ends of a strand. we weighed AS0k· with a window function:

106

a-lp+l q

if 0 ~ p ~ q

h(p.a.q) = a if q<p<l-q

1 - a (p - 1) + 1 if 1-q~p~I q

where 1 is the length of the duplex. and p the number of nucleotides from the end of the duplex to the first shielded phosphate. The parameters a and q control the ampli­tude and width of the window. This window function (see Figure 1) introduces

h 0

t

0 q 1-q ---P

Figure 1. Window function h{p.a.q) that is applied to the adaptation entropy .6.S0k'·

the effect that • conformational changes in the interior of a duplex are more difficult then at the ends. so ASV will have a higher value in the interior. Summing all enthalpy and all entropy terms for a given duplex system and calculating the ratio of the resulting enthalpy and entropy will then give the duplex stability expressed as a T m value.

RESUL TS AND DISCUSSION

1. Shielded phosphates at the end of one strand

The situation in which a number of phosphates is shielded at the end of a duplex is depicted in Figure 2. and gives rise to the following expression for T m•

where an asterisk indicates a base pair with one shielded and one natura! phosphate:

AHo + AHO* T = n m m AS0

0 + AS0*m + AS0m· ·

We now define the dimer contributions for shielded phosphates in one strand as: Ah* =/:!·Ah. and Ag* = 8·Ag. Using 1 = m + n. and counting only one duplex initiation.

n m S' 1--------*'---*--'*--*--'-*-* .... 3·

3'1----------------<S'

Figure 2. Schematic representatîon of a DNA duplex with m shielded phosphates (*) at the end of one strand.

one can write:

7 _ T m _ àh(I - m - 1) + llh*m

- 298 - (àh - àg)(I - m - 1) +à~+ (llh* - llg*)m + 298 4.4m + 83·0 m + 4.7

Note that in this case h(p.a.q) = 1 always. since p = 0. Substituting all numerical values. it then follows that:

9.1(1 - m - 1) + 9.113m T = --------'-----'-----:-::-.,..----,-.......,...,..~

7.2(1 - m - 1) + 6 + (9.1/3 - 1.98}m + 1.31m + 24.73 m + 4.7

We have evaluated T as a function of m. the number of shielded phosphates. with f3. 8 and 1 as variable parameters. using an IBM PC-AT personal computer. Note that m ~ 2 must hold. since dinudeotide contributions are counted. The general result of these calculations is shown in Figure 3. First. we varied the thermodynamic parame­ters /3 and 8. lncreasing the value of 8 at a fixed value of /3 (see Figure 3a). and decreasing the value of fl at a fixed value of 8 {see Figure 3b) give the same result. viz. a continuously rising function r(m). At low values of 8 or high values of (3. a constantly decreasing graph is found. For intermediate values. the curves show a maximum for T. These three domains can be characterized roughly with the ratio 8/ {l. A value of 8/ (3 ~ 1 will always give a rising T-function. while values below ca. 0.85 give a decreasing stability for more shielded phosphates. The shape of the curves is virtually unaffected by variation of the duplex length 1 (see Figure 3c). although higher r values are of course found for longer duplexes.

From the experimental data6·8 on the phosphate-methylated/natural system d(Ap(OMe)Ap(OMe)A)·poly(dT). we can cadulate f3 = 1.29. and 8 1.93. lt is now evident. that since 8/ (3 = 1.50. duplex stabilization will always occur for this type of hybrid. In accordance with our earlier experimental work.8 the T m value goes up in the model when more phosphate-methylated trimers associate with poly(dT). giving a cooperative effect.

As we have remarked before. shielding of the phosphate charges can also occur in the case of protein complexation with DNA phosphates. The present resu/ts indicate. that protein complexation with DNA phosphates at the end of one strand can give destabilization of the duplex. as long as BI {3 < 1. This may be

108

1.26

t

lal t 1.24 1.25

t

10 20 30 Ic) î 1.20

---m

1.27 1.15

2 10 20 30

t --m

(bi t 1.25

10 20 30 --m

Figure 3. Graphs of the function T(m) with: (a) /3 = 1.10. 1 = 100. and 8 varying from 0.80 (lower trace) to 1.20 (upper trace): (b) 8 = 1.10. 1 = 100. and fJ varying from 0.90 (upper trace) to 1.30 (lower trace): (c) /3 = 1.10. 8 = 0.95. and 1 varying from 20 (lower trace) to 120 (upper trace).

the mechanism that DNA unwinding enzymes (helicases) use during the replication of DNA. In this process. the DNA strands are separated. and their genetic informa­tion is copied to new strands in a structure known as a replication fork.11 Several structural features are known for helicases that support our argument. In genera!. hel­icases cover about 8 base pairs at the end of a DNA duplex in the replication fork. 12·13 and cause the duplex to dissociate in to single strands. More specifically. the unwinding enzyme of phage fd (gene 5 protein) is known to bind electrostatically with 5 phosphates via lysine and arginine residues.14.15 and has a high affinity for a single strand. In fact. in model studies it has been found that complexation of gene 5 protein with the duplex of poly(dAT) decreases the T m value from 54°( to 11.5°C.16

Our model calculations now provide a possible mechanism for these protein-induced unwinding processes. Binding of a helicase enzyme to one of the strands in a duplex will deform this strand. and can therefore destabilize the duplex when the deforma­tion is not outweighed by diminished interstrand repulsions.

109

ll. Shielded phosphates in the interior of one strand

In the situation that a protein associates with the DNA phosphates in the inte­rîor of a duplex (see Figure 4). the effect of the window function h(p.a.q) becomes apparent. lt now holds for the T m value that:

AHo + AH0* + AH0 Tm = p m n

AS0P + AS0*01 + AS0n + AS0

0,'·h(p.a.q)

p m n S'>--~~~--~~~--<>--~~~--3'

3· 5'

Figure 4. Schematic representation of a DNA duplex with m shielded phosphates (*} in the interior of one strand.

With 1 = p + m + n. this can be written as:

- Tm -T-298-

Ah(p - 1) + Ah*m + Ah(I - m - p)

(Ah-Ag)(p-1)+Agi+(Ah*-Ag*)m+(Ah-Ag)(l-m-p)+298 4.4m \Si.O h(p.a.q) m+ .

lnserting numerical values then yields:

7 = 9.1(1 - m - 1) + 9.1.Bm

( } ( ;:,) 1.31m + 24.73 ( ) 7.2 1 - m - 1 + 6 + 9.1.B - 1.9o m + 4 7 h p.a.q

m+ .

Evaluation of r(m) now depends on the parameters .B. 8. and 1. and the window parameters a and q. The results are shown in Figures 5 and 6. The window parame­ter q defines the length of the end effect. and is only relevant in comparison with the value of p. An appropriate length should be the Debije-Hückel screening length K- 1•

which is the relevant distance for electrostatic interactions in solution. However. in these model calculations we neglect concentration effects. which are implicitly present in the screening length K-1. Therefore a fixed value of 10 is used throughout. which corresponds with one complete turn of the helix. lncreasing the window height (a) results in a decrease for the overall r-value. and especially disfavours short stretches of shielded phosphates (see Figure 5a). Normally, the value of parameter a was taken as 2. Variation of the distance from the duplex end (p) indeed gives an extra desta­bilization for charge shielding in the interior of the duplex (see Figure 5b). although no extra destabilization is present for p > q. Again. variation of the duplex length (1) does not alter the essential features of the curves (see Figure 5c). whereas the ratio

110

8 / f3 determines the characteristics of the curves ( see Figure 6a.b).

1.23

t

t1.19 1.25

{al

t 1.15 t 1.20

lel 2 10 20 30

-m

1.22 1.15 v-:-t

2 10 30

t 1.21

20

{bi

-m

10 20 30 -m

Figure 5. Graphs of the function r(m) with f3 1.10. 8 = 0.95. and: (a) 1 = 40. p q = 10. and a varying from 1 (upper trace) to 6 (lower trace); (b) 1 = 40. q =

10. a = 2. and p varying from 0 (upper trace) to 10 (lower trace): (c) p = q = 10. a = 2. and 1 varying from 20 {lower trace) to 120 (upper trace).

1.26

t

(al f 1.24

10 20 --m ----m

Figure 6. Graphs of the function r(m) with 1 = 100. p q = 10. a = 2. and: (a) f3 1.10 and 8 varying from 0.80 (lower trace) to 1.20 (upper trace): (b) 8 = 1.10 and f3 varying from 0.90 (upper trace) to 1.30 (lower trace).

30

111

Comparing the graphs in Figure 6a with those in Figure 3a. it is evident that for an equal value of of {3 (e.g., {3 1.10 and o 0.90). a longer stretch of shîelded phosphates (ca. 10 in Figure 6a vs. ca. 5 in Figure 3a) is needed in the interior of the duplex to achîeve the same destabilization. This corresponds nicely with the charac­

teristîcs of RNA polymerases. the enzymes that locally unwind two DNA strands in the interior of a duplex to start transcription (i.e .. copying of the DNA information to an mRNA molecule17). The RNA polymerase enzyme is always affixed to one strand of the duplex. 18- 20 Electrostatic attractions are thought to be the main protein-DNA interactions in this case. 21 and the length of duplex DNA covered by the protein is

35-70 base pairs in the initiation phase.22 and at least 17 base pairs in the elongation phase.23 The present model can adequately exp/ain why at least twice the length of duplex must be covered for RNA polymerases in comparison with he/icase enzymes (vide supra}.

Recently Schinkel et al. 24 have provided experimental support for the proposed conformational changes in a DNA duplex after protein complexation in the interior and at the end ( see previous section). They observed in gel retardation experiments that binding of yeast mitochondria! RNA polymerase to a DNA fragment containing the promoter either in the interior or near the end. induces a bending of the helix. This bending must be due to a difference in conformation between the two DNA

strands in the complex. which causes a local melting of the DNA duplex according to our model calculations. lt follows therefore. that protein complexation in the interior

can produce the open complex. which serves as a starting point for transcription.22 while formation of a replication fork can be induced by complexation at the end of a duplex.

111. Shielded phosphates in both strands symmetrically

The situation in which phosphate charges in both strand ends are symmetrically shielded by a protein (see Figure 7} has two modes with different mathematica!

expressions for T m· For mode A. one can easily derive with 1 = m + n: '..,.'f;f

T - 2AHo* m + AH01-2m m.A - 2AS0* + AS0 + 2AS0 ' m l-2m m

(Note that h{p.a.q) = 1). This means for r:

T . _ 2Ah*m + Ah{I - 2m - 1) A-

2(Ah* - Ag*)m + (t.h - t.g)(I - 2m - 1) + Ag + 298 8.8m + t66.0 1 m + 4.7

which becomes with numerical values inserted:

112

n m 5' * * * * 3'

A 3'

* * * * 5'

m n

l-2m

s· 3' 8

3' s· n

2m-l

Figure 7. Schematic representation of a DNA duplex with symmetrical regions of m shielded phosphates (*) in both strands.

9.1(1 2m - 1) + 18.2/3m TA= ~~~~~~~~~~~--~~~~=-::,.,.--,.-r:::-:-.:~ 7.2(1 - 2m - 1) + 6 + (18.28 - 3.88)m + 2.26m + 49.46

m + 4.7

This holds as long as m ~ n. since for longer shietded stretches an overlap will occur

(mode B). Then. the situation arises that two strands are simultaneously shielded. which is indicated with a double asterisk. The dimer contributions in this case are

written as: Ah** = y· Ah. and Ag** = e-Ag. For mode B. one can write with 1 m

+ n:

or:

_ 2Ah*(I - m - 1) + Ah**(2m - 1) r 8 - . -2-( l1-h-*--l1-g*_)_( 1,_-m---1-) _+_A_g;_,_· _+_(_l1_h_* .-_"-'1-g-*-*)-( 2-m"---1-) _+_.2_9_8=8.~8~~,-=-;::~l+-+4-~~766=.~o

In th.e case of protein complexation._ the v.alues of y and E could be abstracted from the thermodynamic studies of Fujioka et aJ. 25 on ONA-polylysine systems: y 0.83.

and E = 1.38. With all numerical values substituted. it follows that:

~ 18.213{1 - m ~ 1) + 7.55(2m - 1) r 6

- -(1-8-.2-8---. -3.-88-)-(1---m-. -~1)_+_6_+_4-'-.9-3-(2-m~--1)_+___,2.....,·6"""'2,..,.~1_---m-m....-~-c-~--,.j..,.,9....,..·46=-

which holds for m > n. The transition from mode A to mode B is discóntinuous in this description. A more detailed model will have to incorporate a better description

113

for the transition area (m~ n). thus yielding a smooth r-function. lnspection of r(m) as a function of ~. 8. and 1 shows that 1 only influences the halfway transition point between mode A and mode B. hut not the shape of the curves (see Figure 8a). The ratio 8/ ~ again determines the function's behaviour in mode A (see Figure 8b.c) For certain values of 8/ ~. destabilization occurs initially. but for m > n one always finds a strong stabilization. Thus. a minimum in duplex stability can be present in this symmetrical shielding situation.

lol

lbl

t40

t

îl.20 1.40 t

t 1.30 10 20 30 lel -m

1.20 1.40

t

tt30 1.1 0

2 10 20 30 -m

-m

Figure 8. Graphs of the function T(m) with: (a) 13 = 1.10. 8 = 1.00. and 1 varying from 10 (lower trace) to 50 (upper trace): (b) 13 = 1.10. 1 = 30. and 8 varying trom 0.80 (lower trace) to 1.30 (upper trace): (c) 8 = 1.00. 1 = 30. and fJ varying from 0.80 (upper trace) to 1.30 (lower trace).

This model is especially pertinent for restriction endonucleases. enzymes which can recognize and cut palindromic sequences in duplex DNA. lt is known for the enzyme EcoRI. that first an electrostatic complexation with DNA phosphates takes place via protonated amino-acid residues and polarized a-helices.26 This complexation locally unwinds the DNA duplex 25° by the formation of a so-called neokink. after which recognition of the base sequence occurs with a set of specific hydrogen bonds. Since EcoRI is a symmetrical dimeric protein. shielding of the phosphates will always be present on two DNA strands symmetrically.26 Therefore. the present model may account f or the generation of neokinks in duplex DNA by endonucleases. which is an essential mechanism f or the recognition process.

114

CONCLUDING REMARKS

The thermodynamic model presented above is dearly able to give a mechanism for the enzymatic DNA duplex destabilization. Moreover. with relatively few parame­ters it can account for certain salient features of enzymes that interact with DNA. In the present version. effects of ionic strength on the duplex stability are neglected. lt can not be exduded that inter- and intrastrand electrostatic repulsions are influenced in a different way by addition of salts. so in future studies this will have to be taken into consideration. From the present model. the difference in shielded length of DNA duplex between helicases and RNA polymerases is readily explained. Also. a better understanding of the relevance of symmetrical complexation in the case of endonu­deases can be obtained. We therefore feel. that unilateral shielding of the phosphate charges in DNA may be an important mechanism in protein-DNA interactions. Further experimental studies will be performed to establish the ratio o/ fj for protein complexation with one strand from simpte model compounds. and to verify the effect on duplex stability.

REFERENCES

1. W. Saenger. "P rindples of Nucleic Acid Structure". Springer Verlag. New York. 1984. Ch. 18.

2. D.L. Ollis and S.W. White, Chem. Rev. 87. 981 (1987).

3. M. Tsuboi. in "Conformation of Biopolymers", Academie Press. New York. 1967. Vol. ll. pp. 689-702.

4. P.H. von Hippel and J.D. McGhee. Ann. Rev. Biochem. 41. 231 (1972).

5. C. Hélène and J.-C. Maurizot. CRC Crit. Rev. Biochem. 10. 213 (1981 ).

6. LH. Koole. M.H.P. van Genderen. R.G. Reiniers and H.M. Buck. Proc. K. Ned. Akad. Wet .. Ser. B 90. 41 (1987) (communicated by H.M. Buck at the meeting of Dec 15. 1986).

7. M.H.P. van Genderen. L.H. Koole. K.B. Merck. E.M. Meijer, L.A.AE. Sluyterman and H.M. Buck. Proc. K. Ned. Akad. Wet" Ser. B 90. 155 (1987) (communi­cated by H.M. Buck at the meeting of Feb 23. 1987).

8. M.H.P. van Genderen. L.H. Koole. H.M. Moody and H.M. Buck. Proc. K. Ned. Akad. Wet .. Ser. B 91. 59 (1988) (communicated by H.M. Buck at the meeting of Nov 30. 1987).

9. M.H.P. van Genderen. L.H. Koole and H.M. Buck. Recl.Trav. Chim. Pays-Bas (1989). in press.

115

10. K.J. Breslauer. R. Frank. H. Blöcker and L.A. Marky. Proc. Natl. Acad. Sci. USA

83. 3746 (1986).

11. L. Stryer. "Biochemistry". Freeman and Co" New York. 1988. 3•d ed" Ch. 27.

12. M. Szekely. "From DNA to Protein''. McMillan Press Ltd .. London. 1980. Ch. 2.

13. J.W. Chase and K.R. William. Ann. Rev. Biochem. 55. 103 (1986).

14. G.D. Brayer and A. McPherson. J. Mol. Biol. 69. 565 (1983).

15. J.E. Coleman. R.A. Anderson. R.G. Ratcliffe and I.M. Armitage. Biochemistry

15. 5419 (1976).

16. B. Alberts. L. Frey and H. Delius. J. Mol. Biol. 68. 139 (1972).

17. See Ref. 11. Ch. 29.

18. D.C. Hinkle and M. Chamberlin. J. Mol. Biol. 70. 157 (1972).

19. W.F. Mangel and M. Chamberlin. J. Biol. Chem. 249. 2995 (1974).

20. M. Chamberlin. in "RNA Polymerase". R. Losick and M. Chamberlin. Eds" Cold

Spring Harbor Press. New York. 1976. pp. 159-191.

21. S.L. Shaner. P. Melancon. K.S. Lee. R.R. Burgess and M.T. Record. Jr" Cold

Spring Harbor Symp. Quant. Biol. 47. 463 (1983).

22. P.H. von Hippel. D.G. Bear. W.D. Morgan and J.A. McSwiggen. Ann. Rev.

Biochem. 53. 389 (1984).

23. A. Schmitz and D.J. Galas. Nucleic Acids Res. 6. 111 (1979).

24. A.H. Schinkel. M.J.A. Groot-Koerkamp. A.W.R.H. Teunissen and H.F. Tabak.

Nucleic Acids Res. 16. 9147 (1988).

25. K. Fujioka. Y. Baba and A. Kagemoto. Polymer J. 11. 509 (1979).

26. J.A. McClarin. C.A. Frederick. B.-C. Wang. P. Greene. H.W. Boyer. J. Grable

and J.M. Rosenberg. Science 234. 1526 (1986).

116

CHAPTER 7*

Determination of the nature of the conformational transmission effect in pentacoordinateQ phosphorus compounds

ABSTRACT

A high resolution 1 H NMR study of several four- and five-coordinated (TBP) phosphorus compounds enabled us to determine the nature of the conformational transmission effect. which describes the influence of the phosphorus coordination on the molecular conformation. Of great use was the accurate determination of the JPOMe coupling constant. by which the pseudorotational equilibrium around the pen-­tacoordinated phosphorus can be described. lt was determined that conformational transmission is purely electrostatic in origin. due to a larger charge density in the axis of the TBP structure. These findings were corroborated with quantum chemica! cal­culations.

*M.H.P. van Genderen. L.H. Koole. B.C.C.M. olde Scheper. L.J.IVI. van de Ven and H.M. Buck. Phosphorus Sulfur 32. 73 (1987).

117

INTRODUCTION

Recently. a group of studies has been presented of model systems for nucleo­tides.1 phospholipids. 2 and 6--phosphorylated pyranosides3 that incorporate a pen­tacoordinated phosphorus (Pv) with trigonal bipyramidal geometry (TBP). The pV TBP moiety in these compounds is a stabilized analogue of the transient pV struc­ture that can be formed in phosphate groups by temporary attachment of a fifth ligand (e.g .. a solvent molecule). Compared with their plV counterparts. the pV TBP systems show distinct conformational changes. and it was argued that transient pV formation can therefore have a trigger function for structural changes in e.g. nucleic acids and membranes. The conformational transmission effect on going from plV to pV TBP in the model systems is based on an increased conformational preference for trans location of the vicinal oxygens in the common P-0-C-C-O fragment. leading to the explanation that an enhanced charge repulsion between the oxygens is the cause of this effect. In order to further investigate this idea, we have synthesized com­pounds 1a,b-4a,b. and have determined their C1-C 2 conformations with 300 MHz 1H

NMR. These models have substituents with various electronic properties. which allows for the determination of the nature of the conformational transmission effect.

a

RESUL TS AND DISCUSSION

CrC2 conformational analysis

Me

O~ Me

MeO-._ 1 f __...,P-0

MeO 1

Ht~--01 H,, 1 2

H2 ---x H2'

b

1: X = OMe

2: X = NMe2

3: X = CH 2Me

4: X = SMe

The conformation around the C1-C 2 bond is an equilibrium between the three staggered rotamers. hut as two of these rotamers are mirror images and have identi­cal populations. we use a two-state description with a gauche (g) and a trans (t) state (see Figure 1). The population densities Xg and x1 of these states can be deter­

mined from the vicinal proton-proton coupling constant J12. which is a mixture of the

coupling constants in the two states: J12 = xgJg + x1J1• with Xg + x1 = 1. The values of Jg and J1 can be calculated with the semi-empirically modified Karplus relation as developed by Haasnoot et al:' (see Table 1). This relation explicitely accounts for

118

trans

Figure 1. Staggered rotamers around the CrC2 bond.

electronegativity and orientation of the no~-hydrogen substituents.

Table 1. Calculated coupling'.constants for 1~4.

Compound Jg (Hz) J1 (Hz)

1 X ~ Of\/le · 4.13 7.46 2X NMe2 4:69 ··1.66 3X CH2Me 4.83 7.91 4X SMe 4.83 7.91

plV Compounds ta-4a

The Ci-C2 conformations5 of 1a-4a (see Table 11) accurntely reflect the proper­ties of the substituents. ·

Table ll. CrC2 conformation of 1a-4a.

Compound J12 (Hz)

la X = OMe 4.6 2a X NMe2 5.8 la X CH2Me 6.5 4a X = SMe 6.8

0.84 0.16 0.61 0.39 0.47 0.53. 0.37 0.63

119

In 3a. the C,-C2 bond tends to the sterically favourable trans conformation. Replace­ment of CH 2Me by OMe or NMe2 in 1a and2a. respectively. causes a distinct prefer­ence for the gauche state. due to the well-known gauche effect.6 The electrostatic repulsion between the hetero atoms is fully expressed in 4a. since it is known that no gauche effect operates between 0 and S.6b Although sulphur nominally has the same electronegativity as carbon. 4 there is a charge density on sulphur due to its ability to use d-orbitals as polarization functions.7 Correspondingly. a strong preference for trans orientation of 0 1 and Sis found in 4a.

Pseudorotational analysis

In the pV compounds rapid pseudorotation6 around the pentacoordinated phos­phorus obscures the CrC2 conformational analysis. since an average conformation of axial and equatorial OCCX fragments is observed. The axial site in the TBP carries a higher negative charge than the equatorial site.9 so it is necessary to separate their contribution to the Ci-C2 conformation. In earlier workt.3 this was accomplished by placing three identical groups on the interchanging positions. We now introduce a method that uses the fact that equatorial or axial location of a methoxy group in a TBP influences the phosphorus-proton coupling constant JpoMe·rn An accurate11

measurement of the JpoMe coupling constant in 5 and 6 allowed us to determine the equatorial and axial phosphorus-proton coupling constants Jeq and Jax.

Me

o~ Me MeO. 1 f --p-o Meo"I

OMe

5

Me

o~ Me

Me0--.~-7 o...,....I lyo

6

Taking into account the strain rule for pV TBP systems. stating that five-membered rings preferentially span an axial-equatorial location.12 one sees that the three methoxy groups in 5 are distributed over two equatorial and one axial site. while in 6 the methoxy group is forced to remain in an equatorial site. From this it follows that JpoMe(5) = (J"" + 2Jeq)/3 and JpotJe(6) = Jeq· As we measured a JroMe coupling constant for 5 and 6 of 13.02 and 13.80 Hz. respectively. it can be derived that Jeq = 13.80 Hz and J3 " = 11.43 Hz. For the compounds 1b-4b. we are interested in the equilibrium between axial location of the OCCX fragment (JroMe = Jeq = 13.80 Hz) and equatorial location (JPOMe = Pax + Jeq)/2 = 12.62 Hz). The fraction y with OCCX axial can be determined from the measured JpoMe with:

120

Y (JPOMe - 12.62)/(13.80 - 12.62).

As a test for this new method for assessing the pseudorotational equilibrium. we performed a detailed conformational study of lb. The results of this study can be compared with earlier work on 7a,b. which have a similar conformation1 (xt(la) = 0.16 and x1(7a) = 0.17: x1{1b) = 0.28 and x1(7b) = 0.32).

7a 7b

Using variable-temperature experiments. the equatorial and axial conformations were separa~ed (see Appendix). The results (see Table 111) compare favourably with the axial and equatorial conformations of 7b (xt.ax = 0.65 and x1.eq = 0.27 at 295 K1).

Table 111. Full Ci-C2 conformational analysis of 1b at various temperatures.

T (K) JpoMe (Hz) y J12 (Hz) x; x,• Xg.eq Xt.eq Xg.ax Xt.ax

295 1306 0.37 5.03 0.72 0.28 0.86 0.14 056 0.44 285 13.06 0.37 5.01 0.74 0.26 0.87 0.13 0.55 0.45 267 13.04 0.36 4.93 0.76 0.24 0.88 0.12 0.53 0.47 249 13.06 0.37 4.84 0.79 0.21 0.90 Q.10 0.50 0.50 244 13.04 0.36 4.81 0.80 0.20 0.90 0.10 0.49 0.51

L!.H8i = -1.1 kcal/mol L!.H.~ = 0.8 kcal/mol L!.Seq = -1.3 calfmol.K L!.S.~ = 1.8 cal/mol.K

•Measured under rapid pseudorotation conditions.

Only in the axis of the TBP a change in the conformation occurs. while the equa­torial site behaves virtually identical to the plV situation. Therefo[e the further con-

formational analysis will be performed with Xg1i.eq xg/1.p1v.

121

pV TBP Compounds 1b-4b

Upon increasing the phosphorus coordination from four to five. a shift in confor­mational preference toward the trans state is observed for 1. 2. and 4. whereas in 3 no change occurs ( see T able IV).

Table IV. Full C1-C2 conformational analysis of 1b-4b at 300K.

Compound JpoMe (Hz) y J12 (Hz) x• g x,• Xg.eq Xt.eq Xg.ax Xt.ax

tb X = OMe 13.05 0.36 5.0 0.72 0.28 0.84 0.16 0.51 0.49 2b X = NMe2 13.02 0.34 6.2 0.47 0.53 0.61 0.39 0.21 0.79 3b X = CH2Me 12.93 0.26 6.4 0.48 0.52 0.47 0.53 0.50 0.50 4b X = SMe 13.12 0.42 7.0 0.28 0.72 0.37 0.63 0.17 0.83

•Measured under rapid pseudorotation conditions.

As the presence of the TBP geometry in 3b does not affect the C1-C 2 conformation compared to 3a. the conformational transmission is not based on steric interactions. A possible influence via a diminished gauche effect in tb and 2b is ruled out by the occurrence of conformational transmission in 4b. Especially this latter compound demonstrates that only electrostatic repulsion between the hetero atoms can be the cause of the change in conformation. This is consistent with the fact that only in the axis of the TBP (with a high charge density) conformational transmission is encoun­tered. Furthermore. the thermodynamic parameters that were obtained for tb (see Table 111) indicate that the effect is enthalpy-controlled. as is expected for an electros­tatic repulsion. Comparison of the driving forces for the conformational transmission in 1-4 is possible when we express each Ci-C2 conformation as the difference in free

energy (ÄG0) between the gauche and trans state (see Appendix). The change on going from the plV situation to axial location in the pV TBP compound (~~G0) reflects the magnitude of the conformational change. The results (see Table V) show indeed a greater change for the more electronegative substituents. In fact. a relation can be found with the charge density on the atom in the X substituent that is vicinal to 0 1 (vide infra). This again demonstrates the electrostatic nature of conformational transmission.

Quantum chemica! calculations

Calculations were performed with the MNDO method13 on the systems 1a-4a. and tb (with OCCX axial). For la and tb. enthalpies of formation (ÄH1) were

122

Table V. Driving force for the conformational transmission in 1-4.

AG0 (kcal/mol) Compound · plV

aG0 tkcal/mol) pV axial AAG0 (kcal/mol)

1 X = OMe -0.6 0.4 1.0 2 X = NMe2 0.1 1.2 1.1 3 X = CH~e 0.5 0.4 -0.1 4 X = SMe 0.7 1.4 0.7

obtained after relaxation of the molecular geometry in the gauche and trans states of the Ci-C2 bond. From these data we could extract enthalpy differences (AHgt) for the

gauche;::trans equilibrium (see Table VI).

Table Vl. Calculated enthalpies of formation of the states in la and tb. .

AHt (kcal/mol) AHgt (kcal/mol)

la gauche -203.78 -0.38 trans -203.40

tb ga uche -294.60 0.69 trans -295.29

lt is obvious that the gauche state is more favoured in la. while the trans state has the lowest enthalpy in lb. As we also found experimentally. there is a reversal in the sign of the enthalpy difference. The magnitude of the enthalpy change between la and 1b {1.1 kcal/mol) is somewhat smaller than the experimental value (1.8 kcal/mol).

On going from plV to axial location in pV TBP. the electron density on 0 1

increases from 0.48 e.u. to 0.59 e.u" consistent with the experimental findings {i.e" enhanced repulsion). The electron density on the equatorial oxygen remains 0.48 e.u. The susceptibility of the X substituent for electrostatic interactions is determined by the charge density on the atom that is vicinal to 0 1. Calculation of these charge den­sities in la-4a indeed yields a correspondence with the driving force of the conforma­tional change (see Table V): 0, -0.36: N. -0.33: C. -0.02: S, -0.15.

123

CONCLUDING REMARKS

The present conformational study on 1-4 has clearly shown that the conforma­tional transmission effect is due to an electrostatic repulsion, caused by the increased charge density on the axial oxygen in the TBP. This explanation is confirmed by quantum chemica! calculations. The accurate measurement of the phosphorus-proton coupling constant JpoMe has proven to be a useful technique for characterizing the pseudorotation around the pentacoordinated phosphorus.

APPENDIX

Defining h.G0 = G1° -G:. we get the Boltzmann distribution:

~ = 2exp(-h.G0/RT) X1

. x or: t:iG 0 = -Rîln(-g )

2x1

where the factor two accounts for the twofold degeneracy of the gauche state. With Xg + x1 1. we can write:

exp(-t.G6/ RT) Xg = l + exp(-t:iG0JRT)

2

Since this equation holds for both the axial and equatorial conformation. one can deduce with t.G0 t:iH0 -TAS0:

Xg = y•Xg.ax + (1-y)•Xg,eq =

exp(-l:iH3~/RT)·exp(l:iSa~/R) + (l-y) exp(-h.H~/RT)-exp{h.S.~/R) y} + exp(-h.Ha~IRT)·exp(h.Sa~/R) t + exp(-h.He~IRT)·exp(h.S.i/R)

As Xg and y are known for each temperature. four unknown quantities remain. allow­ing a set of four equations based on measurements at foor temperatures to be solved.14

EXPERIMENTAL SECTION

Solvents and materials were reagent grade. and were used as received or purified as required. Reactions were run onder an atmosphere of dry nitrogen. 1H NMR spec­tra were run in the FT mode on a Broker CXP 300 spectrometer at 300.1 MHz. For more details. see refs. 5 and 11. Samples were dissolved in acetone-d6• unless stated

124

otherwise. and chemica! shifts were ·relatéd to 2TMS (o·' =' 0). Coupling constants were taken from the expansions. and were simulated with the PANIC program. 15 31P NMR spectra were run at 36.4 MHz on a Bruker HX 90R spectrometer with a Digilab FT-NMR3 pulsing accessory. Chemica! shifts were referenced against 85% H3P04 (8 = 0). and were designated positive downfield of this standard. Compounds 5 and 6 were prepared acoording to Ramirez.16 Phosphites were prepare as described in ref. 1.

2-Methoxyethyl dimethyl phosphite

B.p. 68°C/30 mmHg. 31P NMR: 8 145.2. 1H NMR: 8 3.48 {3H. s; OCH3). 3.63

(6H. d. POCH3. JpoMe = 10.6 Hz). 3.68 (2H. m. H2/H2-). 4.05 (2H. m. Hi/HJ').

N,N-Dimethyl-2-aminoethyl dimethyl phosphite

B.p. 70-C/30 mmHg. 31P NMR: 8 145.0. 1H NMR: o 2.31 (6H. s. NCH3). 2.56

(2H. t. H2/Hr). 3.57 (6H. d. POCH3. JpoMe = 10.4 Hz). 3.95 (2H. dt .. Hi/Ht').

n-Butyl dimethyl phosphite

B.p. 80-C/40 mmHg. 31P NMR: o 145.0. 1H NMR: 8 1.02 (3H. t. CH3). 1.43-1.57

(2H. m. CH2). 1.62-1.73 (2H. m. H2/H2·). 2.55 {6H, d. POCH3. JpoMe = 10.4 Hz). 3.88 (2H. dt. HtfHt').

2-Methylthioethyl dimethyl phosphite

B.p. 72°C/30 mmHg. 31 P NMR: o 145.0. 1H NMR: o 2.21 (3H. s. SCH3). 2.79

(2H. t. H2/H2·). 3.57 (6H. d. POCH3. JpoMe = 10.5 Hz). 4.02 (2H. dt. H1/Ht').

Phosphates

All phosphates were obtained by oxidation of the corresponding phosphite. An ozone-oxygen stream was passed at -78°C through an NMR sample tube containing a solution of the phosphite in dry dichloromethane. until a blue colour appeared due to excess ozone. The reaction mixture was then sparged with oxygen. and allowed to warm to room temperature. The dichloromethane was evaporated by passing dry nitrogen through the tube. after which a deuterated solvent was added. Completion of the reaction was confirmed by 31 P NMR.

2-Methoxyethyl dimethyl phosphate (1a) 31 P NMR: o 6.5. 1H NMR: o 3.43 (3H. s. OCH3). 3.57 {2H. m. H2/H2·). 3.71

(3H. d. POCH3. JpoMe = 11.1 Hz). 4.20 (2H. m. H1/H,.).

125

N,N-Dimethyl-2-aminoethyl dimethyl phosphate (2a) 31 P NMR: 8 6.6. 1H NMR: 8 2.37 (6H. s. NCH3). 2.79 (2H. t. H2/H2'). 3.81 (6H.

d. POCH3. JpoM~ = 11.0 Hz). 4.33 (2H. dt. H1/Hr).

n-Butyl dimethyl phosphate (3a) 31 P NMR: 8 6.6. 1H NMR: 8 1.04 (3H. t. CH 3). 1.47-1.58 {2H. m. CH 2). 1.7()..

1.80 (2H. m. Hi/H2·). 3.81 (6H. d. POCH3• JpoMe = 12.0 Hz). 4.13 (2H. dt. Hl/Hr).

2-Methylthioethyl dimethyl phosphate (4a) 31 P NMR: 8 6.2. 1H NMR: 8 2.02 (3H. s. SCH3). 2.87 (2H. t. H2/H 2·). 3.88 (6H.

d. POCH3. JpoMe = 12.0 Hz). 4.40 (2H. dt, H1/Hr).

Phosphoranes

All phosphoranes were obtained by adding an equimolar amount of 2.3-butanedione at o·c to a solution of the corresponding phosphite in a deuterated sol­vent. After standing for 30 min at O"C. formation of the phosphorane was ascertained with 31P NMR.

2.2-Dimethoxy-2-{2-methoxyethoxy )-4,5-dimethyl- t,3,2-dioxaphos phol-4-ene (tb}

31P NMR: 8 -43.9. 1H NMR: 8 1.89 (6H. s. CH3 dioxalene ring). 3.40 (3H. s. OCH3). 3.57 (2H. m. H2/Ht). 3.62 (6H. d. POCH3. Jpol\Áe = 13;0 Hz). 4.03 (2H. m.

H1/H1').

2,2-Dimethoxy-2-( N ,N-dimethyl-2-aminoethoxy )-4,5-dimethyl-1.3.2-dioxaphosphol-4-ene (2b)

31P NMR: 8 -44.0. 1H NMR: 8 1.71 (6H. s. CH3 dioxalene ring). 2.37 (6H. s.

NCH3}. 2.51 (2H. t. Hi/H2'). 3.60 (6H. d. POCH3. JpoMe 13.0 Hz). 4.00 (2H. dt.

Hi/Hl'}.

2,2-Dimethoxy-2-( n-butoxy )-4,5-dimethyl-1,3,2-dioxaphosphol-4-ene (3b) 31 P NMR: ·8 -43.9. 1H NMR: 8 1.01 (3H. t. CH3). 1.50 (2H. m. CH2). 1.62 (2H.

m. H2/H2'). 1.89 (6H. s. CH3 dioxalene ring). 3.60 (6H. d. POCH3, JpoMe = 12.9 Hz). 3.95 (2H. dt. Ht/Hr).

2,2-Dimethoxy-2-(2-methylthioethoxy)-4,5-dimethyl-1,3,2-dioxaphosphol-4-ene (4b)

31P NMR: 8 -43.9. 1H NMR: 8 1.90 (6H. s. CH 3 dioxalene ring). 2.21 (3H. s.

SCH3). 2.74 (2H. t. H2/H2·). 3.62 (6H. d, POCH3. JpoMe = 13.1 Hz). 4.07 (2H. dt.

126

REFERENCES

1. L.H. Koole. E.J. Lanters and H.M. Buck. J. Am. Chem. Soc" 106, 5451 (1984).

2. G.H.W.M. Meulendijks. W. van Es. J.W. de Haan and H.M. Buck. Eur. J. Biochem. 157. 421 (1986).

3. N.K. de Vries and H.M. Buck. Red. Trav. Chim. Pays-Bas 105. 150 (1986).

4. C.A.G. Haasnoot. F.A.A.M. de Leeuw and C. Altona. Tetrahedron 36. 2783 (1980). The equation used is: JHH = 13.89 cos2c/> - 0.98 cos</> + E,t.x,{1.02 3.40 cos2(~ic/> + 14.9IAxïlH.

i

where t.x; = t.x;°' - 0.24E,.t.x 1~ is the group electronegativity corrected for fj-'

substituents. </> is the proton-proton torsion angle (taken as ± 60° or 180°). and ~ i is a substituent orientation parameter ( ± 1). The electronegativity of the ele­ments relative to hydrogen has been derived from the Huggins scale: Axo = 1.30. AxN = 0.85. and AXs = ÄXc 0.40. As the coordination of phosphorus is an effect in y-substituents only. it does not affect the calculated coupling con­stants. which can therefore be used for both piv and pV TBP compounds.

5. Using 32K points and a spectra! window of 3000 Hz. one obtaines an accuracy of 0.09 Hz in coupling constants. resulting in population fractions with an accu­racy of ca. 0.02. For the variable-temperature experiments described later. 16K points were used with a spectra! window of 300 Hz. yielding an accuracy of 0.018 Hz in coupling constants and 0.005 in population fractions.

6. (a) S. Wolfe. Acc. Chem. Res. 5. 102 (1972). (b) A.J. Kirby. "The Anomeric Effect and Related Stereoelectronic Effects at Oxygen". Springer Verlag. Berlin. 1983. p. 32-36.

7. (a) M.M.E. Scheffers-Sap and H.M. Buck. J. Am. Chern. Soc. 102. 6422 (1980). (b) H.S. Aldrkh. l.A. Alworth and N.R. Clement. J. Am. Chem. Soc. 100. 2362 (1978). (c) J.-M. Lehn and G. Wipff. J. Am. Chem. Soc. 98. 7498 (1976).

8. R. Luckenbach. "Dynamic Stereochemistry of Pentacoordinated Phosphorus and Related Elements". Georg Thieme Verlag. Stuttgart. 1973. p. 3-5.

9. ibid .. p. 10-11.

10. See for instance: {a) F. Ramirez. Acc. Chem. Res. 1. 168 (1968). (b) D. Goren­stein and F.H. Westheimer. J. Am. Chem. Soc. 92, 634 (1970).

11. Measurements of JpoMe were performed with 32K points and a spectra! window of 100 Hz. giving an accuracy of 0.003 Hz in coupling constants. Repeated meas­urements showed the coupling constants to be reproducible within 0.01 Hz.

127

12. See ref. 8. pp. 43-46 and 50-51.

13. Although the MNDO method does not include d-orbitals for phosphorus in its

parametrization. a fifth bond can be accommodated because the program uses

four bonding MOs and the first antibonding MO. In the latter MO. the AO of

phosphorus has the same symmetry as the necessary d-orbital. This is shown

by MN DO calculations on PH 5• and is confirmed by earlier ab initio work on the

same molecule: R.A.J. Janssen. G.J. Visser and H.M. Buck. J. Am. Chem. Soc.

106. 3429 (1984).

14. M.J.D. Powell. "A FORTRAN subroutine for solving systems of non-linear alge­

braic equations". Harvell Report. AERE-R5947. H.M.S.O" 1968.

15. PANIC program. copyright Bruker Spectrospin AG. Switzerland.

16. F. Ramirez. Bull. Chim. Soc. Fr. 1. 2443 (1966).

128

CHAPTER8*

Conformational transmission in riucleotides containing trigonal bipyramidal phosphorus as the internucleoside linkage

ABSTRACT

A set of nucleotide analogues contammg a stable trigonal bipyramidal phos­phorus (Pv TBP) moiety (5-11) has been developed. and their conformational pro­perties were studied with 300 and 500 MHz ~H NMR. In the solvent acetone-d6• it is found that the conformation of the model compounds is determined by a hydrogen bond between the backbone atom 0 5• and the base proton H6 (pyrimidine base) or H8

(purine base). resulting in a preference for the standard gauche(+l conformation around the C4·-C5· bond. In the hydrogen-bond disrupting solvent DMSO-d6• the pV

TBP nucleotides 5-8 clearly show conformational transmission. Le .. a preference for the unusual gauche(-) (g-) rotamer around the C4·-C5· bond is found. This structural distortlon. opposes stacking of the bases. as is confirmed by the observation that the preference for g- is strongest for 7, and 8. in which stackîng is eliminated. The present .results provide support to our earlier proposal that. formation of pV TBP locations in DNA can lead to a marked change of the secondary structure. (H.M. Buck. Reel. Trav. Chim. Pays-Bas 99. 181 (1980)).

*L.H. Koole. M.H.P. van Genderen and H.M. Buck. J. Org. Chem. 53. 5266 (1988).

INTRODUCTION

In the past years we developed and firmly established a concept for conforma­tional transmission in a variety of trigonal bipyramidal (Pv TBP) phosphorus com­pounds.1 tt was shown. that the construction of specific ligands directly linked to phosphorus as pV_O-C-C-Ö(R) makes it possible to select a different conformational behaviour around the C-C linkages for equatorial and axial positions in the TBP. Compounds 1 and 2 are typical model compounds used to study the conformational transmission effect in our previous work.

Me

o~ Me

MeO- f f ._.;;p-o MeO 1

0

~OMe 2

A pronounced trans orientation of both oxygens is found for the axial sites. whereas the well-known gauche arrangement has an equatorial preference.2 The introduction of the concept of conformational transmission is based on the observation that in the corresponding plV tetrahedrar compounds (P1v-O-C-C-O(R)) the gauche arrangement of both oxygens is unique. whereas after introduction of an extra (similar) ligand the pV TBP with its (chemically) different sites selects the conformational change from gauche to trans via exchange of axial and equatorial positions respectively. The addi­tion of an extra ligand which is reflected in the intrinsic chemica! bonding properties of a pV TBP configuration3 results in an enhanced electron density on the axial oxy­gens directly linked to phosphorus. In its turn this effect is transmitted in a confor­mational change around the C-C linkage via an increased Coulombic repulsion between both oxygens. leading to a trans orientation. Very recently. de Keijzer et al. investigated the impact of conformational transmission on the rate of intramolecular ligand exchange in pv TBP model systems (pseudorotation} .4 With variable­temperature 13C NMR on the monocyclic pV TBP compounds 3a,b and 4a,b it was established that pseudorotation in 3a and 4a is 2-4 times faster than in 3b and 4b. With the acceptance of a square pyramid in controlling the pseudorotation. it could be shown that conformational transmission in the basal ligands in the square pyramid is reponsible for lowering ·of the activation barrier for pseudorotation by 0.5-0.7 kcal/mol. In previous publications.1 we have regularly emphasized that the concept of conformational transmission might be of significance in activating phosphorylating biomolecules. A straightforward example has been given by Meulendijks et a1.tc.5 in their studies on conformational transmission in model systems for phospholipids. For monomeric phospholipid models in solution. it was found that going from plV toward

130

3a:X=0 3b: X = CH2

4a:X=0 4b: X = CH2

pV TBP resu.lts in a structural change in the glyceryl fragment leading toa stronger van der Waals interaction between the two acyl çhains.k Precise condusions could be drawn from a set of phospholipid analogues in the solid state. which have been stu­died with cross-polarization MAS 13C NMR. lt was observed that conformational transmission results in a more downfield 13C chemica! shift for the ro-methyl groups and a reduced cross-polarization optimal contact time .. which show that the chain ends are forced into a more proximale position. Based on th.ese results. the sugges­tion was put forward that conformational transmission might be of importance for controlling ion transport in phospholipid bilayers.5

Now we will offer a detailed study of the impact of pV TBP locations in the backbone of nucleotides for conformational transmission on the level of single­stranded phosphate-methylated DNAs in various solvents. The pV TBP nudeotides 5-t t ( see Chart 1) were chosen as representative model systems. The selection of phosphate-methylated DNAs is necessary to guarantee stable pV TBPs. The presen­tation of the results will be discussed with the different contributions of the bioor­ganic ligands leading toa relaxed pV TBP structure.

METHODS

Synthesis of 5-16

The model compounds 5-8 (Pv TBP) and 12 and 13 (P1v) were synthesized from the corresponding p!losphite triester (P 111) nudeotides via reaction with butanedione. and ozone/oxygen. respectively. The precursor P111 nucleotides were prepared from 5' -protected thymidine 3' -(methyldiisopropylphosphoramidite) (in the case of 5. 6. and 12) or 5' -protected 1' .2' -dideoxyribose 3' -(methyldiisopropylphosphoramidite) (in the case of 7. 8. and 13}. in a tetrazole­catalyzed reaction in dry pyridine. Standard column chromatography using Woelm sil­ica gel as the stationary phase and dry butanone as eluent afforded these compounds

13'1

5 : R = T: R' = Ac 6 : R = T: R' = CPh3

7 : R = H: R' = Ac 8 : R = H: R' = CPh3

~· 0 ,..o ~p

MeO ..... '-o

~T AcO

12: R = T 13: R = H

T

Me

~~Me MeO .:;.:P-O MeO 1

\:Y· AcO

9 : R = T 10: R = A 11 : R = N4-acetyl-C

0 MeO---p~ MeO,.- \

0

~· AcO

14: R = T 15: R = A 16 : R = N4-acetyl-C

Chart 1. pV TBP nucleotide structures studied in this chapter and their plV coun­terparts.

in the pure form in moderate yields (vide infra). In all cases. 31 P NMR clearly confirmed the formation of P 111 nucleotides, each of which exists as a mixture of two diastereomers. For 5-8. it was observed that the 31 P NMR spectrum consisted of a single line. This proves that stereomutation around the pV TBP is rapid on the NMR

time scale.6 The model compounds 9-11 and 14-16 were prepared by phosphoryla­tion of the corresponding 3' -0-acetylated nucleosides with dimethoxy-(N .N­dimethylamino)phosphine. leading to the 5'-P 111 precursors. Purification of these com­

pounds was also accomplished with chromatography on a silica gel column with dry butanone as eluent.

132

Conformational analysis

The structural aspects of 5-16 were investigated with 300 and 500 MHz 1H NMR. Conformational analysis was focused on the C4·-C5• bonds. as well as the sugar moieties. C4·-C5• conformations are described in terms of a time-averaged distribution over the staggered rotamers gauche(+) (g+). gauche(trans) (g1

). and gauche(-) (g-).

Os• Os• Os• 0*3' "*~· c*"" Hs• Hs" Hs• Hs" Hs• Hs"

H4• C31 041

g+ gt g-

The rotamer populations were calculated from the experimental coupling constants Jn· and J4·s" with the help of the empirically generalized Karplus equation of Altona et al.7 The conformation of the sugar rings in nudeotides is generally treated as a two-state equilibrium between a C2·-endo and a Crendo type puckered form.8 In principle, five vicinal proton-proton coupling constants are available to monitor the sugar conformation (Jn" Jrr. Jn·. Jrs·. and Jn-). In various cases. ho~ever. it proved impossible to determine accurate values for Jn" J2" 3•• and/or Jn" due to one of the following reasons: (i) collapse of the H2• and Hr in the NMR spectra: (ii) overlap of the H2• or Hr spectra! pattern with the residual signal of the solvent DMSO-d6: (iii) overlap of H4· and the Hs·/Hs" spectra! pattern. In order to arrive at a uniform treatment for all model compounds. we used the formula

x(Ct-endo) = lJrr + Jrr 9.8}/5.9.

as developed by Rinkel et al.9 This method allows one to estimate the conformational equilibrium of the sugar ring in deoxynucleotides with a fair accuracy. on the basis of Jn· and J1'2'· exclusively. For the nucleotides 5. 6. and 12. the assignment of the H1·

patterns of the upper and lower residues was performed with homonuclear decoupling experiments. based on the fact that the connectivity sequence phosphorus-Hr H2'/r-H1• only exists for the upper residue.

RESUL TS AND DISCUSSION

The solvents acetone-d6 and DMSO-d6 have been chosen to study the confor­mational aspects of the model systems 5-16. Acetone-d6 was found to be an

133

unsuitable solvent to study conformational transmission. since the hydrogen bonding between the backbone atom 0 5• and H6 of thymine or cytosine. or H8 of adenine.

strongly fixes the C4·-C5• conformation in the g+ rotamer (see Figure 1).10

Figure 1. Part of the X-ray crystal structure of 3' .5'-di-0-acetylthymidine.18 in which the g+ conformation is stabilized via hydrogen bonding between 0 5• and H<> Hetero atoms (N.O) are shaded. and hydrogen atoms other than H6 have been om­itted for clarity.

The formation of the 0 5·-base hydrogen bond was perfectly prevented in DMSO-d6•

which enabled us to establish the impact of conformational transmission on the molecular structure of our model systems in an unequivocal way.

Conformation of 5-16 in DMSO-d6

Table 1 (left) summarizes the experimental coupling constants Jn· and Jn" and

the calculated rotamer distribution around the C4·-C 5• bond for 5-16 in the solvent DMSO-d6. lnspection of these data show that the pV TBP nucleotides 5-8 have

dominant populations of g-. which corresponds with trans orientation of 0 5• and 0 4•

(vide supra): x(g-) varies from 0.48 to 0.65 for 5-8. The plV structures 12 and 13. on the other hand. display a clear preference for the well-known g+ conformation. in which 0 5• is gauche with respect to 0 4• (x(g+) = 0.71 and 0.63 for 12 and 13.

respectively). The occurrence of conformational transmission in 5-8 implies that 0 5•

is preferentially located in the axis of the pV TBP. i.e" structure 1 (Os· axial. 0 3·

equatorial) prevails over the two possible alternatives. Il (03• axial. Os· equatorial)

and 111 (03• and 05· equatorial). The preference of 1 over Il correlates with quantum

134

T able 1. Experimental coupling constants J.-s' and J4'5" measured in DMSO-d6 (left) or acetone-d6 (right) and the. calculated time-averaged rotamer populations around the C.i·-C5, bond.•

DMSO-dc, acetone-du

Compd Jn (Hz) j 4,5" (Hz) x(g+) x(g1) x(g-) J4'5' (Hz) j4's" (Hz) x(g"") x(g1) x(g-)

5 6.7 6.0 0.20 0.23 0.48 2.4 2.6 0.87 0.13 0.00 6 6.8 5.7 0.22 0.28 0.50 2.5 2.6 0.86 0.14 0.00 7 7.5 5.5 0.19 0.16 0.65 3.0 2.8 0.79 0.15 0.06 8 8.0 5.1 0.19 0.16 0.65 2.9 2.7 0.81 0.14 0.05 g 3.8 3.1 0.70 0.15 0.15 2.4 2.2 0.90 0.10 0.00

10 3.6 4.0 0.61 0.25 0.14 3.0 3.7 0.70 0.24 0.06 11 3.7 3.1 0.70 0.15 0.15 2.4 2.2 0.90 0.10 0.00 12 3.7 3.1 0.71 0.18 0.11 2.8 2.6 0.83 0.13 0.04 13 3.8 3.8 0.63 0.25 0.12 3.0 3.2 0.75 0.19 0.06 14 3.5 3.5 0.68 0.20 0.12 3.0 2.9 0.78 0.16 0.06 15 3.7 3.1 0.70 0.15 0.15 4.4 4A 0.50 0.27 0.23 16 3.4 3.5 0.68 0.20 0.12 2.9 2.9 0.80 0.14 0.06

aoata refer to the 3' -residue in the case of the nucleotides 5-8 and 12. 13.

chemica! calculations by van Lier et al.11 which showed that Os· axial. O:r equatorial is approximately 2 kcal/mol more stable than 03· axial. 0 5• equatorial. From Dreiding molecular models. it seems clear that 111 is unfavourable with respect to 1 and Il (no quantum chemica! calculations have been performed).

Me

Me ~o / \ 1 •. -03• o-~°'OMe

/ 05•

I

Me

Me ~o 1 \ ., .-05• o-r ..... <h·

OMe \

II I

These results provide strong support for our original proposition12 that formation of a pV TBP in the DNA backbone can substantially perturb the DNA secondary struc­ture via a rotation around the C4·-Cs· linkage from g+ toward g-. Tne pV TBP sys­tems 7 and 8. in which base stacking is eliminated since the 5' -base is replaced by hydrogen. are of further interest. Comparison with 5 and 6 reveals that the prefer~ ence for g- is most pronounced in the absence of stacking (7 and 8: x(g--:) 0.59 and 0.65, respectively: 5 and 6: x(g-) = 0.48 and 0.50. · respectively). i.e ..

135

conformational transmission opposes the regular stacking of adjacent bases.

The data on the pV TBP nucleotides 9-11 show that a high preference exists for the g+ conformation (see Table 1). The explanation for the absence of conforma­

tional transmission in these systems rests on the fact that 0 5· is preferentially

located in an equatorial position in the TBP.13 The similarity of the C4·-C5• rotamer

populations of 9-11 and the plV counterparts 14-16 is in line with our earlier work. in

which a close resemblance was found for 5'-P 1v tetrahydrofurfuryl. and tetrahydrofur­furyl in an equatorial location in a pV TBP .1a lt must be concluded that the 5' _pv

TBP nucleotides 9-11 are in fact inadequate models to study conformational transmission in DNA structures.

Table ll. Experimental coupling constants J1'2· and J1'2" measured in DMSO­d6 (left) and acetone-d6 (right) and the calculated populations of the C2·-endo puckered form of the 2' -deoxyribose ring.

DMSO-d6 acetone-d6

Conipd Jn· (Hz) Jn" (Hz) x(C2·-endo) Jn· (Hz) Jl'r (Hz) x(C2·-endo)

5 5' -residue 7.0 6.8 0.6~ 6.9 6.9 0.68 5 3' -residue 7.9 7.5 0.95 7.4 7.0 0.78 6 5' -residue 7.0 6.7 0.66 6.7 6.5 0.58 6 3' -residue 8.0 7.5 0.97 7.6 7.1 0.83 7 3' -resid ue 7.8 7.6 0.95 7.7 7.1 0.85 8 3' -residue 7.9 7.4 0.93 7.5 7.3 0.85 9 7.9 7.0 0.86 8.0 7.4 0.95 10 7.7 7.2 0.86 7.7 7.0 0.83 11 7.9 7.1 0.88 8.0 7.4 0.95 12 5' -residue 7.0 6.9 0.70 7.0 6.8 0.68 12 3' -residue 7.2 7.0 0.75 7.7 7.2 0.86 13 3' -résidue 7.2 7.0 0.75 7.4 7.4 0.85 14 7.6 7.0 0.81 7.5 7.0 0.80 15 7.4 6.9 0.76 7.6 7.0 0.81 16 7.5 7.0 0.80 7.5 7.0 0.80

The conformational data on the sugar rings in 5-16 are summarized in Table Il (left). These data clearly show a preference for the Ct-endo puckered form of the

ring. Conformational transmission upon going from plV (12, 13) toward pV TBP (5-

8) results in a slight increase of x(Ct-endo) for the 3'-residue. The apparent prefer­

ence for the conformational combination g- (C4·-C5• boncj) and C2·-endo (sugar ring)

corresponds with the conclusion of Remin14 that a g- /C3·-endo conformation is highly unfavourable.

136

Conformation of. 5-'16. in acetone-d6•

The experimental coupling constants J4·s· and Jn" rneasured in acetone-d6• as well as the calculated rotamer populations of g+. g1

• and g-. are listed in Table 1

(right). lnspection of these data shows that none of the pV TBP systems display conformational transrnission. In fact. it appears that increasing the phosphorus coor­dination frorn plV to pV TBP results in a slight increase of the g+ rotamer popula­tions. For example. it is found for the pV TBP systems 5-8 in acetone-d6 that x(g-) ranges from 0.85 to 0.91. while x(g+) = 0.70 and 0.72 fo~ the plV coiJnterpàrts 12 and 13. respectively. These data suggest that conformational transmission is prevented by the formation of a hydrogen bond between 0 5• and H6 of thymine (vide supra). The extreme situation is represented by the pV TBP compound g with x(g+) = 0.90. The data in Table Il (right) show that the conformational equilibria of the sugar rings in 5-16 in acetone-d6 are heavily biased toward the C;r-endo form.

CONCLUDING REMARKS

The results obtained with the model compounds 5-11 illustrate several novel and revealing aspects of conformational transmission in nueleotide structures. First. it is clear that the solvent is of importance in determining whether or not conforma­tional transmission wil!' occur. Apparently. it is a prerequisite for conformational transmission that a hydrogen-bond disrupting solvent such as DMSO is used.15 Oth­erwise. the C4·-Cs· conformatiOn is determined by an Os·-base hydrogen bond. leading to an exdusive preference for the g+ conformation. Secondly. it follows from com­parison of the data on 5, 6 and 12 with those of 7, 8 and 13 that conformàtional transmission opposes stacking of adjacent bases

The present results provide support for our earlier suggestion that conforma­tional changes in natural DNA can also be achieved by · activation of the backbone phosphates via a plV into pV TBP transition.ta.b.12 Two points must be made in extrapolating the present data to conformational transitions in natural DNA: (i) The pV TBP compounds 5-U are neutral species. whereas the transient pV TBP system formed in natural DNA has two negatively charged oxygens bound to phosphorus. Quantum chèmical calculations by van Lier et al..11 and more recently by de Keijzer et al..4 have shown that conformational transmission occurs in both charged and neu­tra! pv TBPs. These data strengthen our original point that neutral pV TBP struc­tures (which are stable enough for experimental studies) can be used as models for unstable transient pV TBPs as formed in our proposed mechanism for conforma­tional transmission in natural DNA. (ii) The present study refers to OMSO-d6 or acetone-d6 as the solvent. whereas natural DNA is usually in an aqueous environ-

137

ment. The instability of the systems 5-11 has precluded conformational studies in protic media (e.g .. CD30D or D20). However. since it is known that hydrogen­

bonding interactions in aqueous solutions are relatively weak due to competition of water molecules for hydrogen-bond donor and acceptor sites (ref. 8. p. 126). it must be expected that the conformational transmission effect is also operative in water as

the solvent.

EXPERIMENTAL SECTION

1H NMR spectra were run in the FT mode at 300 or 500 MHz on a Bruker CXP 30016 or a Bruker AM 50017 spectrometer. T etramethylsilane was used as the inter­nal standard. Appropriate spectra! windows (10-15 ppm) were chosen. and Fourier transformation was usually performed with 32K data points. 31 P NMR spectra were run in the FT mode on a Bruker HX 90 (36.4 MHz) or a Bruker AC 200 (80.9 MHz) spectrometer. Woelm silica gel was used for column chromatography. All melting and boiling points are uncorrected.

5' -0-Acetylthymidine

Acetic anhydride (2.45 g, 24 mmol) was added over 30 min to a magnetically stirred solution of thymidine (4.48 g, 20 mmol) in 150 ml of dry pyridine. The reac­tion mixture was stirred for 3 h. after which the solvent was evaporated under reduced pressure with moderate heating (4D°C). The last traces of pyridine were removed by coevaporation with toluene. Thin-layer chromatography (TlC) of the residual gum. using 2-butanone as eluent. revealed the presence of four different com­pounds. i.e .. 3'.5'-di-0-acetylthymidine (R1 = 0.51). 3'-0-acetylthymidine (R1 = 0.37). 5'-0-acetylthymidine (R1 = 0.17). and unreacted thymidine (R1 ~ 0).

Repeated column chromatography afforded 5' -0-acetylthymidine as a white solid in 28% yield (1.60 g): mp 134-137°(; 1H NMR (acetone-d6): 8 1.87 (3H. s. CH 3 base).

2.12 (3H. s. CH3 acetyl). 2.34-2.42 (2H. m. H2'fH2"). 4.02-4.12 (2H. m. H5·/H5").

4.19 (1H. m. H4·). 5.32 (1H. m. H3·). 6.33 (1H. dd. Hr). 7.66 (1H. s. H6). Anal. Calcd for C12H160 6Ni: C. 50.70: H. 5.63: N. 9.86. Found: C. 50.5: H. 5.8: N. 10.1.

3' -0-( ( N,N-Diisopropylamino )methoxyphosphino )-5' -0-acetylthymidine

5' -0-Acetylthymidine (1.42 g, 5 mmol) was added with stirring to a mixture of 100 ml of dry chloroform and 10 ml of dry diisopropylethylamine. After the addi­tion. the reaction flask was thoroughly flushed with argon and sealed with a rubber septum. After the mixture was stirred for 2 h. dropwise addition of chloro(N.N­diisopropylamino)methoxyphosphine18 (1.03 g, 5.2 mmol) was started. The resulting yellow solution was stirred for 2 h ánd diluted with 2500 ml of ethyl acetate

138

(prewashed with NaHC03) .. Repeated washing with 100 ml-portions or a saturated NaCI solution in water. and finally with pure water. drying on Na2S04• and evapora­

tion of all volatile mat~·:rial afforded a yellowish oil. which was transferred to a silica gel column. Elution with a mixture of n-hexane/dichloromethane/triethylamine (45:45:10 v /v /v) yielded an oily product with Rt 0.34. Coevap9ration with dry dichloromethane yielded the desired product as a slightly colo.ured foarn (1.52 g, 68%): mp 106-109"(: 1H NMR (acetone-d6) 8 0.90-1.25 (12H', m. CH3 diisopropyl).

1.58 (3H. s. CH 3 base). 2.12 (3H. s. CH3 acetyl). 2.52-2.56 (2H. m. Hr/Hr). 3.38 (3H. d. POCH3. J 11Hz). 3.96-4.09 (2H. m. H5)HS"). 4.24 (1H .. m. H4·). 4.80 (1H. m. H3·). 6.44 (1H. dd. Hr). 7.68 (1H. s. H6). 31 P NMR (aceto~e-d6) S 154.8 and

154.1 (intensity rati.o 1:0.91). Anal. Calcd for C19H32N3P07: C. 51.23; H" 7.19: N. 9.44. Found: C. 50.7: H. 7.2: N. 9.7.

5'.-0-(3' -0-Acetylthymidyl) 3' -0-(5' -0-acetylthymidyl) methyl phosphite

3' -O-Acetylthymidine19 (0.80 g, • 2.46 mmol) and .3' -0~( {l\l.N-di isopropylarnino)rnethoxyphosphino)-5' -0-acetylthymidine (0.94 g. 2.11 mmol) were dissolved in 15 ml of dry pyridine. 1H-Tetrazole (0.24 g, 3.2 mmol) was added. and the reaction mixt.ure was stirred for 4 h. Thorough evaporation of the pyridine afforded a yellow syrup. which was transferred to a 10-cm long silica gel column. Elution with 2-butanon~ yielded a slightly coloured foam (Rt = 0.32). 31 P NMR indi­cated the presence of two diastereomers with 8 145.8 and 145.2 (acetone-d6): 1H

NMR (acetone-d6) S 1.53 and 1.58 (2x3H. s. CH3 base), 1 .. 95 .and 2.00 (2x3H. s. CH3 acetyl). 2.08-2.35 (4H. m. Hr/Ht'). 3.32 (3H, d, POCH3• J 11 Hz). 3.36-3.52 (4H. m. Hs-/Hd. 4.06 and 4.16 (2x1H. m. H4·). 4.56 and 4.71. (2x1H, m. H3·). 6.35 and 6.42 (2x1H. dd. Hr). 7.58 and 7.62 (2x1H. s. H6). Anal. Calcd for C25H33N4P013: C.

52.08: H. 5.73: N. 9.72. Found: C. 51.9: H. 5.6 N. 9.6.

2-( 3' -0-( 5' -0-Acetylthymidyl} )-2-( 5' -( 3' -0-acethylthymidyl) )-2-methoxy-4,5-dimethyl-1,3.2À 5-dioxaphosphole (5)

This compound was prepared by the addition of 1 equiv of freshly distilled butanedione to a cooled (0°C) solution of 5' -0-(3' -0-acetylthymidyl) 3' -0-(5' -0-acetylthymidyl) methyl phosphite in a 5-mm NMR sample tube. After 30 min. 31 P NMR indicated complete conversion of the phosphite structure to 5: 31 P NMR (acetone-d6) S -48.3: 1H NMR (acetone-d6) S 1.86 (6H. s. CH 3 dioxaphosphole). 1.87

and 1.90 (2x3H. s, CH 3 base}. 2.10 and 2.12 (2x3H. s. CH 3 acetyl). 2.21-2.42 (4H. m. H2'/H2"}. 3.68 (2H. m. H5·/H 5" 5'-residue). 3.75 (3H. d. POCH3• J 12.9 Hz). 4.14-4.18 (2H. m. H5-jH5" 3'-residue). 4.23 (2H. m. H,r). 5.32 (1H. m. H3• 5'-residue). 5.80

(1H. m. H3· 3'-residue). 6.12 (1H. t. H1• 3'-residue). 6.20 (1H. t. H1• 5'-residue). 7.60 and 7.68 (2x1H. s. H6)· 8.4 (2H. bs. NH2).

139

5' -0-(3' -0-Acetylthymidyl) 3' -0-(5' -0-acetylthymidyl) methyl phosphate (12)

This compound was prepared by bubbling N0 2 gas through a cooled (0°C) solu­tion of 5' -0-(3' -0-acetylthymidyl) 3' -0-(5' -0-acethylthymidyl) methyl phosphite in a 5-mm NMR sample tube. 31 P NMR indicated complete conversion to the phosphate 12: 31 P NMR (acetone-d6) 8 0.2 and 0.8: 1H NMR (020) 8 1.89 (6H. s. CH 3 base).

2.15 (6H. s. CH 3 acetyl). 2.41-2.65 (4H. m. H2·/Hr). 3.80-3.82 (2H. m. H5-fH5").

3.88 (3H. d. POCH 3. J = 11 Hz). 4.28-4.51 (4H. m. 2H.r/H5-fH5"). 5.07-5.13 (1H. m. H3·). 5.37-5.42 (1H. m. H3·). 6.24 (2H. m. Hl'). 7.57 (1H. s. H6). 7.64 (1H. s. H6).

2-(3' -0-(5' -0-Tritylthymidyl) )-2-(5' -0-(3' -0-acetylthymidyl) )-2-methoxy-4,5-dimethyl-1,3,211. 5-dioxaphosphole (6)

This compound was prepared according to the procedure described for 5 from 5' -0-(3' -0-acetylthymidyl) 3' -0-(5' -0-tritylthymidyl) methyl phosphite.19 31 P NMR (acetone-d6) 8 -50.3: 1H NMR (acetone-d6) 8 1.84 and 1.88 (2x3H. s. CH 3 base). 1.90

(6H. s. CH 3 dioxaphosphole). 2.10 (3H. s. CH 3 acetyl). 2.20-2.28 (4H. m. Ht/H2").

3.40 (2H. m. Hs-/Hs" 5'-residue). 3.72 (3H. d. POCH3. J = 12.8 Hz). 4.12-4.17 (2H. m. H5·/H5" 3'-residue). 4.20 (2H. m. H4·). 5.44 (1H. m. H3• 5'-residue). 5.78 (1H. m. H3· 3'-residue). 6.16 (1H. t. Hr 5'-residue). 6.21 (1H. t. H1• 5'-residue). 7.14-7.92 (15H. m. trityl). 7.51 and 7.52 (2x1H. s, H6). 8.20 (2H. bs. NH 2).

5' -0-Acetyl-1 ',2' -dideoxyribose

1' ,2' -Dideoxyribose20 (5.9 g. 50 mmol) was reacted with acetic anhydride (6.1 g. 60 mmol) according to the procedure that was described for 5' -0-acetylthymidine. The product (R1 = 0.38) was obtained in 17% yield (1.38 g). Detection on TLC was effected by exposure to iodine vapor: 1H NMR (acetone-d6) 8 1.58-2.35 (2H. m.

Ht/Hr). 2.13 (3H. s. CH 3 acetyl). 3.16-4.30 (6H. m. Hr/H 1·'/H3-/H 4·/Hs-/H5"). Anal. Calcd for C7H120f C. 52.52: H. 7.50. Found: C. 51.8: H. 7.6.

3' -0-((N,N-Diisopropylamino )methoxyphosphino )-5' -0-acetyl-1 ',2' -dideoxyribose

This compound was prepared from 5'-0-acetyl-1'.2'-dideoxyribose (1.20 g. 7.5 mmol) and chloro(N.N-diisopropylamino)methoxyphosphine18 (1.58 g. 8.0 mmol) according to the procedure that was described for 3' -0-( (N.N­diisopropylamino)methoxyphosphino)-5' -0-acetylthymidine. The desired product was obtained as a foam. mp 96-101°(. in 52% yield (0.78 g): 1H NMR (acetone-d6) 8

1.12 (12H. CH 3 diisopropyl). 1.40-2.40 (2H. m. H2)Hr). 2.10 (3H. s. CH3 acetyl).

3.00-4.40 (SH. m. H1-/H 1", H3·/H 4·/Hs·/Hs"/2xCH isopropyl): 31 P NMR (acetone-d6)

8 154.2 and 153.7 (intensity ratio 1:0.88). Anal. Calcd for C14H28PN05: C. 52.34: H. 8.72: N. 4.36. Found: C. 51.6: H. 8.4: N. 5.0.

140

5' -0-(3' -0-Acetylthymidyl) 3' -0-(5' -0-acetyl-1 ',2' -dideoxyribosyl) methyl phosphite

This compound was prepared in a coupling reaction of 3' -O-acetylthymidine19

(620 mg. 2.2 mmol) and 3'-0-((N.N-diisopropylamino)methoxyphosphino)-5'-0-acetyl-1'.2'-dideoxyribose {640 mg. 2 mmol) as described for 5'-0-(3'-0-acetylthymidyl) 3' -0-(5' -0-acetylthymidyl) methyl phosphite. The product was obtained as a colourless glass. 31 P NMR indicated the presence of two diastereomers with S 145.2 and 144.9 (acetone-d6): yield. 360 mg (36%); Anal. Calcd for C20H29N2P0 11 : C. 47.62: H. 5.75; N. 5.56. Found: C. 46.8: H. 5.7: N. 5.9.

2-(5' -0-(3' -O-Acetylthymidyl))-2-(3' -0-(5'-0-acetyl-1' ,2' -dideoxyribosyl))-2-methoxy-4,5-dimethyl-1,3,2X 5-dioxaphosphole (7)

This compound was prepared according to the procedure described for 5. 31 P NMR (acetone-d6) S -49.0; 1H NMR (acetone-d6) 8 1.82 (6H. s. CH3 dioxaphos­phole). 1.87 (3H. s. CH3 base). 2.05 and 2.13 (2x3H. s. CH 3 acetyl). 2.22-2.23 (4H. m. H2'/Hr). 3.68 (2H. m. Hs·/Hs" 3'-residue). 3.82 (3H. POCH3. J 13.0 Hz). 3.92-4.10 (2H. m. Ht'/Ht"). 4.18 (1H. m. H4• 3'-residue). 4.23 (1H. m; H4· 3'-residue). 5.28 (1H. m. H3• 5'-residue). 5.80 (1H. m. H3• 3'-residue). 6.19 (1H. t. H1• 3'-residue). 7.60 (1H. s. H6). 8.90 (2H. bs. NH2).

5' -0-(3' -0-Acetylthymidyl) 3' -0-(5' -0-acetyl-1 ',2' -dideoxyribosyl) · methyl phosphate (13)

This compound was obtained from 5' -0-(3' -0-acetylthymidyl) 3' -0-(5' -0-acetyl-1' .2' -dideoxyribosyl) methyl phosphite according to the procedure described for 12. with ozone/oxygen instead of N0 2. 31P NMR (acetone-d6) 8 0.1 and 0.6.

5' -0-(3' -0-Acetylthymidyl) 3' -0-(5' -0-trityl-1' ,2' -dideoxyribosyl) methyl phosphite

This compound was obtained from a coupling reaction of 3' -O-acetylthymidine19

(1.28 g. 4.5 mmol) and 3' -((N.N-dimethylamino)methoxyphosphino)-5' -0-trityl-1 '.2' dideoxyribose (2.00 g. 4.0 mmol) as described for 5' -0-(3' -0-acetylthymidyl) 3'-0-(5' -0-acetylthymidyl) methyl phosphite. The product was obtained as a foam in 37% yield (1.04 g): 31 P NMR (acetone-d6) Ö 145.9 and 145.0.

2-(5' -0-(3' -0-Acetylthymidyl) )-2-(3' -0-(5' -0-trityl-1' ,2' -dideoxyribosyl) )- 2-methoxy-4,5-dimethyl-1,3,2>.. 5-dioxaphosphole ( 8)

This compound was prepared from 5' -0-(3'-acetylthymidyl) 3' -0-(5' -0-trityl-1'.2' -dideoxyribosyl) methyl phosphite according to the procedure described for 5. 31P NMR (acetone-d6) 8 -50.3: 1H NMR (acetone-d6) 8 1.81 (6H. s. CH3 dioxaphos­phole), 1.88 (3H. s. CH 3 base). 2.19-2.41 (4H. m. HdHr). 3.41 (2H. m. H5·/H5" 5'-

141

residue). 3.70-3.78 (2H. m. Hr/Hr 5'-residue). 3.80 (3H. d. POCH3. J = 13.2 Hz).

4.22 (1H. m. H4• 3'-residue). 4.33 (1H. m. H4• 5'-residue). 5.22 (1H. m. H3· 5'­

residue). 5.72 (1H. m. H3· 3'-residue). 6.13 (1H. t. Hr 3'-residue). 7.20-8.12 (15H. m.

trityl). 7.58 (1H. s. H6). 8.70 (2H. bs. NH 2).

Oimethoxy(N,N-dimethylamino)phosphine

Phosphorus trichloride (0.5 mol. 69 g) was added over 30 min to trimethyl phos­phite (1 mol. 124 g) that was kept at 60°C. After completion of the addition the reac­

tion mixture was cooled to O"C and dliuted with 500 ml of sodium-dried diethyl ether. Dimethylamine (3 mol. 135 g) was bubbled through the reaction mixture. After

filtration of the dimethylamine hydrochloride. evaporation of the solvent afforded a yellowish oil that was distilled twice at 45 mmHg through a 20-cm Vigreux to afford 46 g (22%) of the desired product: bp 51-52"C: 1H NMR (C6D6) 8 2.63 (6H. d.

N(CH3)2. JPNCH = 8.8 Hz). 3.42 (6H. d. POCH 3• J = 12.0 Hz): 31 P NMR (C6D6) 8 147.6.

3'-0-Acetylthymidine 5' -( dimethylphosphite)

A solution of dimethoxy(N.N-dimethylamino)phosphine (1.49 mmol. 1.95 g) in

25 ml of dry 1.4-dioxane was added dropwise toa stirred and heated (80"C) solution of 3' -O-acetylthymidine18 (2.00 g. 7 .1 mmol) and lH-tetrazole (250 mg) in 50 ml of dry 1.4-dioxane. After 3 h. TlC using 2-butanone as eluent indicated complete

conversion into the product (Rf = 0.64). The reaction mixture was concentrated in

vacuo. and the resulting glass was chromatographed on a silica gel column. using dry

2-butanone/triethylamine (95:5 v/v) as eluent: yield. 1.6 g (60%): 1H NMR (acetone-d6) 8 1.87 (3H. s. CH 3 base). 2.12 (3H. s. CH3 acetyl). 2.34-2.42 (2H. m.

H2·/H2"). 3.58 (6H. d. POCH 3). 4.02-4.14 (2H. m. H5'/H5"). 4.19 (lH. m. H4·). 5.32

(lH. m. H3'). 6.33 (lH. t. Hr). 7.66 (1H. s. H6): 31 P NMR (acetone-d6) o 145.1.

2-(3' -O-Acetylthymidyl)-2,2-dimethoxy-4,5-dimethyl-1,3,2>.. 5-dioxaphos phole (9)

This compound was prepared from 3'-0-acetylthymidine 5'-(dimethylphosphite)

according to the procedure described for 5. 31 P NMR (acetone-d6) o -44.1: 1H NMR

(acetone-d6) o 1.82 (6H. s. CH 3 dioxaphosphole), 1.90 (3h. s. CH 3 base). 2.32-2.50

(2H. m. H2·/Hr). 3.80 (6H. d. POCH3. J = 13.2 Hz). 4.23-4.28 (2H. m. H5)Hs").

4.30 (1H. m. H4·). 5.30 (1H: m. H3'). 6.16 (1H. t. Ht'). 7.68 (1H. s. H6). 8.90 (2H. bs.

NH2).

142

3' -0-Acetylthymidine 5' -( dimethylphosphate) ( 14)

This compound was prepared from 3' -0-acetylthymidine 5' -{ dimethylphosphite) according to the procedure described for 13. in anhydrous dichloromethane: R1

0.30: 31 P NMR (acetone-d6) 8 6.9: 1H NMR (acetone-d6) 8 1.87 (3H. s. CH 3 base). 2.05 {3H. s. CH3 acetyl). 2.30..2.41 (2H. m. H2)H2"). 3.85 (6H. d. POCH3. J = 11.3 Hz). 4.06-4.18 (2H. m. H5·/HS"). 4.32 (1H. m. H.r). 5.32 (1H. m. H3·). 6.23 (1H. dd. Hl'). 8.15 (1H. s. H6). 8.80 (1H. bs. NH).

2' -Deoxy-3' -0-acetyladenosine

This compound was prepared according to the procedure described tor 5' -0-acetylthymidine (R1 = 0.32 in chloroform/ethanol 6:1 v /v). in a yield of 13%. M.p. 212-214°(. 1H NMR (acetone-d6} 8 2.18 (3H. s. CH 3 acetyl). 2.70 (1H. m. Hr). 3.31 (1H. m. H2·). 4.42 (3H. s. H4·/H5-JHS"). 5.60 (1H. m. H3·). 6.54 (1H. dd. Hr). 8.30 (2H. bs. NH 2). 8.30 (2H. s. Hs/H2). Anal. Calcd for C12H15N50.i: C. 49.14: H. 23.88: N. 23.88. Found: C, 48.26; H. 5.37: N. 22.33.

2' -Deoxy-3' -0-acetyladenosine 5' -(dimethylphosphite)

This compound was prepared frofl'! dimethoxy(N .N-dimethylamino)phosphine (0.35 g. 2.6 mmol) and 2' -deoxy-3' -0-acetyladen~sine (0:5 g. 1. 7 mmol) according to the procedure desctibed for 3'-0-acetylthymidine 5'~(dimethylphosphite). The pro­duct was obtained 'as a yellowish glass (R1 0.41): yield. 315 mg (48%); 1H NMR (acetone-d6) 8 2:11 (3H. s. CH3 acetyl). 2.65 (1H. m. H2"). 3.12 (1H. m. H2'), 3.38 (6H. d. POCH3). 4.10 (2H. m. Hs·/HS"). 4.30 (1H. m. H.i'). S.52 (tH. m. H3·). 6.54

, . , , " . , (1H. dd. Hl'). 8.36 (1H. s. H2). 8.40 (1H. s. H8). P NMR (acetone-d6) 8 145.5.

2-( 3' -0-Acetyl-2' -deoxyadenosyl)-2,2-dimethoxy-4,5-dimethyl-1,3,2À 5- dioxa-phosphole (10) '

This compound was prepared from 2' -deoxy-3' -0-acetyladenosine 5' (dimethylphosphite) accotding to the procedure described for 5. 31P NMR (acetone­d6) 8 -46.2; 1H NMR (àcetone-d6) 8 1.88 (6H. s. CH 3 dioxaphosphole),'2.1~2.31 (2H. m: Hr/Hr). 3.78 (6H, d. POCH3.,J = 13.0 Hz). 3.92-4.31 (2H. m. H5·/H5")."4.52 (1H. m. H4·). 5.22 (1H. m. H:r). 6.04 (1H. t, Hl'). 7.20 (2H'. bs. NH2):"8.10 and 8.20 (2x1H. Hi/tla)· . . . .

2' ~Deoxy-3' -0-acetyladènosine 5'-(dimethylphosphatè) (15)

This compound was prepared from 2' -deoxy-3' -0-acetyladenosine 5' -dimethylphosphite) according to the procedure described for 14: R1 = 0.14; 31 P NMR (acetone-d6) 8 6.7; 1H NMR {acetone-d6) o 2.18 (3H. s. CH3 acetyl). 2.27-2.38 (2H. m. H2'/Hr). 3.78 (6H, d. PÛCH3• J = 11.2 Hz). 4.07-4.20 {2H. m. H5·/H5"). 4.37

(tH. m. H4·). 5.55 (tH. m. H3·). 6.12 (1H. dd. Hl'). 7.04 (2H. bs. NH2). 8.25 and 8.30 (2x1H. s. H3/H2).

2' -Deoxy-3' -O,N4-diacetylcytidine 5' -{dimethylphosphite)

This compound was prepared from dimethoxy(N.N-dimethylarnino)phosphine (0.51 g. 3.8 mmol) and 2' -deoxy-3' -O.N4-diacetylcytidine (0.6 g. 1.9 rnmol) according to the procedure described for 3'-0-acetylthymidine 5'-(dimethylphosphite). The pro­duct (Rf = 0.46) was obtained as a colourless glass: yield 420 rng (55%).

2-(3' -O,N4-Diacetyl-2' -deoxycytidyl)-2,2-dimethoxy-4,5-dimethyl-1,3,2À 5-dioxaphosphole (11)

This compound was prepared from 2' -deoxy-3' -O.N4-diacetylcytidine 5' -(dimethylphosphite) according to the procedure described for 5. 31 P NMR (acetone­d6) 8 -45.7: 1H NMR (acetone-d6) 8 1.90 (6H. s. CH 3 dioxaphosphole). 2.28-2.41 (2H. m. Hi-/Hr). 3.80 (6H. d. POCH3• J = 13.0 Hz). 4.23 (1H. m. H4·). 4.30-4.35 (2H. m. H5·/H5"). 5.32 (1H. m. H3·). 5.83 (1H. d. H5). 6.18 (1.H. dd. Hl'). 7.80 (1H. d. H6).

8.00 (1H. bs. NH).

2' -Deoxy-3' -O,N4-diacetylcytidine 5' -(dimethylphosphate) (16)

This compound was prepared from 2' -deoxy-3' -O.N4-diacetykytidine 5' -(dimethylphosphite) according to the procedure described for 14: R, = 0.12: 31 P NMR (acetone-d6) 8 5.9: 1H NMR (acetone-d6) 8 2.17-2.41 (2H. m. H2• /Hr). 3.81 {6H. POCH3. J = 11.3 Hz). 4.06-4.19 (2H. m. H5-fH5"). 4.38 (1H. m. H4·). 5.42 (1H. m. H3·). 5.90 (1H. d. H5). 6.20 (1H. dd. Hr). 7.75 (1H. d. H6). 8.30 (1H. bs. NH}.

REFERENCES

1. (a) L.H. Koole. E.J. Lanters and H.M. Buck. J. Am. Chem. Soc. 106. 5451 (1984). (b) L.H. Koole. R.J.L. van Kooyk and H.M. Buck. J. Am. Chem. Soc. 107. 4032 (1985). (c) G.H.W.M. Meulendijks. W. van Es. J.W. de Haan and H.M. Buck. Eur. J. Biochem. 157. 421 (1886). (d) N.K. de Vries and H.M. Buck. Reel. Trav. Chim. Pays-Bas 105. 150 (1986). (e) M.H.P. van Genderen. L.H. Koole. B.C.C.M. olde Scheper. L.J.M. van de Ven and H.M. Buck. Phos­phorus Sulfur 32. 173 i1987). (g) M.H.P. van Genderen and H.M. Buck. Magn. Reson. Chem. 25. 872 (1987).

2. For 1. in acetone-d6 at 276 K. it was shown that axial and equatorial locations in the pV TBP oorrespond with 68 and 20% 0-0 trans. respectively. See ref. la.

3. See. for instance: (a) R.R. .Holmes. "Pentacoordinated Phosphorus". ACS Mono­graph 175. 176. American Chemical Society. Washington DC. 1980. Vol. 1. ll. (b)

' J.H,H. Hamerlinck. P. Schipper and H.M. Buck. J. Org. Chem. 48. 306 (1983}.

4. A.E.H. de Keijzer. L.H. Koole and H.M. Buck. J. Am. Chem. Soc. 110. 5995 (1988).

5. G.H.W.M. Meulendijks. Thesis. Eindhoven University of Technology. 1988. (b) 1.1. Merkelbach and H.M. Buck. Reel. Trav. Chim. Pays-Bas 102. 283 (1983).

6. (a) L.H. Koole. W.J.M. van der Hofstad and H.M. Buck. J. Org. Chem. 50. 4381 (1985). (b) L.H. Koole. H.M. Moody and H.M. Buck. Recl.Trav. Chim. Pays-Bas 105. 196 (1986).

7. C.A.G. Haasnoot. F.A.A.M. de leeuw and C. Altona. Tetrahedron 36. 2783

(1980).

8. W. Saenger. "Principles of Nucleic Acid Structure". Springer Verlag, New York. 1984.

9. L.J. Rinkel. Thesis. State University of leiden. 1987. (b) L.J. Rinkel and C. Altona. J. Biomol. Struct. Dyns. 4. 939 (1987).

10. lt is weH-known that the C4·-Cs· conformation in nucleotides is determined in part by hydrogen bonding between 0 5• and the base proton H6 (pyrimidine) or H8 (purine). See: (a) N. Yathindra and M. Sundaralingam. Biochemistry 12. 297 (1973). (b) J. Rubin. T. Brennan and M. Sundaralingam. Biochemistry 11. 3112 (1972). (c) M. Sundaralingam. "Structure and Conformation of Nudeic Acids and Protein-Nucleic Acids lnteractions". M. Sundaralingam and T. Rao. Eds .. University Park. Baltimore. 1975. p. 487. (d) G.l Amidon, S. Anik and J. Rubin. ibid .. pp. 729-744. (e) R. Taylor and 0. Kennard. J. Am. Chem. Soc. 104. 5063 (1982).

11. J.J.C. van Lier. L.H. Koole and H.M Buck. Reel. Trav. Chim. Pays-Bas 102. 148 (1983).

12. H.M. Buck. Reel. Trav. Chim. Pays-Bas 99. 181 (1980).

13. A single bulky substituent on a pV TBP structure prefers an equatorial location. See. for instance: R. luckenbach. "Dynamic Stereochemistry of Pentacoordi­nated Phosphorus and Related Elements", Georg Thieme Verlag. Stuttgart. 1973.

14. M. Remin. J. Biomol. Struct. Dyns. 2. 211 (1984).

15. Conformational transmission was also observed with the model compounds 5-8 in the hydrogen-bond disrupting solvent ((CH3hNhP=0. See: H. Normant. Bull. Chim. Soc. Fr. 2. 791 (1968).

16. NMR facility at the Eindhoven University of Technology.

17. Dutch National NMR Facility at Nijmegen. The Netherlands.

18. S.L. Beaucage and M.H. Caruthers. Tetrahedron Lett. 22. 1859 (1981).

19. LH. Koole. M.H.P. van Genderen and H.M. Buck, J. Am. Chem. Soc. 109. 3916 (1987).

20. M. Hoffer. Chem. Ber. 93. 2777 (1960).

146

CHAPTER9*

Conf ormational transmission in silatranes. Confirmation of the pentacoordinated structure in solution

ABSTRACT

A 300 MHz 1H NMR conformational analysis study of a set.of pentacoordinated silicon compounds (silatranes) and their tetracoordinated counterparts ,is presented. The solution conformation of the silàtranes shows a close. resemblance to known X­ray structures. although the silatrane conformation was found to"depend Opon .tbe solvent polarity. C~mpa;ison of the confOrmation of a 2-methoxyetho~y fr~gment in the tetra- and pentacoordinated systerns· showed' the existence of conformational transmission. similar to that known for phosphorus compounds. Enhanced electros­tatic repulsion between the vicinal oxygens in the 2-methoxyethoxy fragment pr~ duces a trans conformation around the C-C bond in the· pentacoordinated silatranes. The conformational transmision effect is smaller .in silatranes , than in phosphorus compounds. due to a telatively weak fifth bond.

*M.H.P. van Genderen and.H.M. Buck. Reel. Trav. Chim. Pàys-Bas 106. 449 (1987)':

147

INTRODUCTION

The concept of conformational transmission has been shown to be important in various organophosphorus systems.1- 4 In essence. this involves an increase in the coordination of phosphorus from four to five. which leads to a preference for trans location of the oxygens in a P-0-C-C-O fragment. when this is located in the axis of the trigonal bipyramid (TBP) structure around the pentacoordinated phosphorus. The driving force fot this conformational change is an increased electrostatic repul­sion between the vicinal oxygens. which is caused by the higher electron density on the axially located atom.1·4

We have now extended our studies on conformational transmission to silicon systems. which also have the capability of forming pentacoordinated structures. The model compounds used were the pentacoordinated 1-alkoxysilatranes5 1 and 2 and their tetracoordinated counterparts 3 and 4 (see Figure 1).

1~tf::N5~ 10 'o!. ~. l'

~51-0 o, I' 2

X:'' XMe

1 :X=O 2: X = CH2

3: x = 0 4: X = CH2

Figure 1. Model sytems 1-4 with the numbering scheme used.

Silatranes have received much attention. since they are biologically active compounds having a broad spectrum of activity. The TBP structure with a Si+- N transannular interaction.6 as shown in Figure 1. is based on data obtained from X-ray diffraction studies.7- 9 In solution. the possibility exists of an equilibrium between this structure and an exo structure. in which the lone pair of the nitrogen atom is directed away from silicon. Some indications for the presence of the pentacoordinated structure in solution have been found in spectroscopie studies (IR. 29Si and 15N NMR). but a definitive structural determination has not yet been reported. We have now used high resolution 1H NMR spectroscopy to analyze the molecular conformation of the sila­tranes in solution. Comparison of the Cr-C2' conformation in the penta- and tetracoordinated silicon compounds has shown that the Si+- N interaction is reflected in a conformational transmission effect. Furthermore. the conformational characteris­tics of the silatrane cage structure were determined in detail and compared with the X-ray structure.

148

RESUL TS AND DISCUSSION

Cr-Cr conformational analysis

We were able to observe the presence of a Si+- N bond in solution by means of 1H NMR conformational analysis of a suitably chosen substituent on silicon. The transfer of the nitrogen lone pair to the empty d-orbitals of silicon in silatranes changes the charge distribution around silicon. and can therefore influence the C1·-C2•

conformation in the 1-substituent through electrostatic interactions (vide supra}. For th~ freely rotating C1·-CL· bond. a conformational analysis was performed in acetone­d6. yielding population fractions for the rotamers with 0 1• and X located gauche (.xg) and with 01' and X located trans (x1} (see Figure 2).

~ gauche trans

Figure 2. Staggered rotamers around the C1·-C2· bond. The two rotamers with 0 1•

and X located gauche are mirror images. and are therefore combined for the confor­mational analysis.

These fractions (see Table 1) were obtained from the e.xperimental coupling constants Jr2' in the 300 MHz 1H NMR spectrum. and the calculated coupling constants Jg and J1 in the gauche and trans rotamers. with the relation Jn· = .xgJg + x1J1• The values of Jg and J1 were obtained as previously reported.4 using the Karplus relation developed by Haasnoot et al.11

For 1. we see an increased preference for trans location of 01' and 0 2• as com­

pared with 3. The control compounds 2 and 4 . .where electrostatic interactions are excluded by replacement of 02· by CH 2• have almostddentical C1·-C2• conformations. Thus, the silatrane cage per se has no influence on the C1·-C2• bond. and the trans location of Or and 0 2• in 1 must be the result of an enhanced 0 1·-0r charge repul­sion. due toa higher elettron density on Or. This reflects the electron donation from nitrogen via silicon to the axial ligand. as was also predicted by quantum chemica! 'calculations.12 lt is also in complete agreement with the mechanism found for confor­mational transmission in phosphorus compounds (vide supra).·

149

Table 1. C1·-C2· conformations of the silicon compounds 1-4 in acetone-d6.

Compound p g J1a Jnb Xg

Pentacoordinated 1 x 0 4.05 7.50 602 0.43 2 X= CH2 4.76 7.92 6.38 0.49 T etracoordinated 3 X=O 4.05 7.50 509 0.70 4 x CH2 4.76 7.92 6.32 0.51

a Calculated coupling constants for the gauche and trans conformations in Hz. b Measured coupling constant in Hz.

x,

0.57 0.51

0.30 0.49

A confirmation of the influence of this electrostatic repulsion on the Cr-C2• con­formation can be found in measurements of 1-4 in a series of solvents (see Figure 3). Lower solvent polarities. which enhance electrostatic repulsions. lead to larger trans populations in 1 and 3 (1. 020: x1 = 0.27. C6D6: 0.58; 3. D20: 0.13. C6D6: 0.32). As expected. in 2 and 4 the C1'-C2• conformation does not vary. since electrostatic interactions are absent.

Xf

t 0 0

0.5 0

0

• • • •

0.0 30 40 50

0

• 0

•

60 70 Er

• 0.5

0 Il • • •

0·0 '--3~0--4~0--5 .... 0---=6:'::0---=1=-o ----Er

Figure 3. Measured trans populations of the C1·-C:r bond in 1-4 as a function of the solvent polarity. left: 1 (•)and 3 (0). Right:.2 (•)and 4 (0).

· The conformational transmission in the silicon compounds 1-4 can be compared with earlier studies4 on tetra- and p.entacoordinated phosphorus systems (5-8). In these· latter compounds. we see more. pronounced conformational changes for the axial substituent upon increased coordination (see Table Il). This can be explained by. the fact that the Si+- N bond is only a partial bond (calculated strength ca. 25 kcal/mol13 vs. 105 kcal/mol for a Si-N single bond). which has a relatively small

150

Me

~~Me MeO--.P-O MeO.,....... I

,:co,· 2 XMe

5: x :=: 0 6: X = CH2

7:X 0 8: X = CH2

Table ll. Ci--C2, conformations for the phosphorus compounds 5-8• in acetone-~.

Compound

Pentacoordinated 5b x 0 0.49 0.51 6b x CH2 0.50 0.50 T etracoordinated 7 x 0 0.86 0.14 8 x CH2 0.48 0.52

• Data taken from ref. 4. b Fractions refer to axial location of the OCCXMe fragment.

influence on the charge distribution around silicon.

Conformation of the tricyclic cage

In the solid state. the five-membered rings in the cage were found in two puck­

ered forms, with the carbon next to N5 out of the plane of the four other atoms.1- 9

In solution. these two puckered forms interconvert rapidly. as shown by the presence of a simple triplet signa! in the 1H NMR spectrum for both the OCH2 and NCH 2 pro­

tons. In order to characterize the amplitude of the pucker. we have used the NCCO

torsion angle 4' (see Figure 4). Since the conformations around the C-C bond are

mirror images. the vicinal proton-proton coupling constant JHH is dependent only on

the value of tl>. and can therefore be used to monitor the ring pucker. The relation

151

eb ,- Ns

O*H H H

H

</>

Ns"'

H*O H H

H

Figure 4. Newman projections along a C-C bond in the silatrane cage for the two puckered forms.

between JHH and </> can be calculated using the modified Karplus relation. as pro­posed by Haasnoot et al..11 and is given by the solid line in Figure 5.

8

JHHIHil

1

7

6

s

4

3

30 40 so 60 70 80 90 </Jldegl

Figure 5. Calculated relation between JHH and 4> for silatranes (-) and phospha­tranes (- - -)

Precise determination of JHH in 1. 2. and 1-ethoxysilatrane in acetone-dt. (5.89. 5.87 and 5.87 Hz. respectively) gives a value for </> of 36°. comparing favourably with the solid state angle in. for example. 1-phenylsilatrane (39.8°7 and 38.2"8). This indicates that the cage structure in solution is very similar to that found in the crystalline state.

The conformation of the cage was found to vary with the solvent polarity. as is shown in Figure 6. In more polar solvents. JHH increases from 5.80 to 6.00 Hz and therefore </> decreases from 36° to 33°. This can be understood in terms of shielding of the electrostatic repulsion between 0 and N. which results in decreased cage puck­ering in polar solvents. The present results can be compared with earlier measure­ments on the phosphatranes 9-12 in solid state and in solution. Using the relation

152

6.00 a

0

• l!

0 0 Cl

0 • 0

0 Il I!

• s.eo Cl

30 40 50 60 70 Er

Figure 6. Measured coupling constants JHH for 1 (•). 2 (D) and 1-ethoxysilatrane ( O) as a function of the solvent polarity.

between JHH and <f> for these systems ( dashed line in Figure 3). we were able to analyse the 1H NMR data of van Aken14 (see Table 111).

t:s;-··· 0 1

y

g : Y = OMe 10: Y = OEt 11: Y = SMe 12: Y = SEt

This gives torsion angles of 43-50° in acetonitrile-d3. while in an X-ray study 15 of 12 a value of ca. 43° was found. Again, we see a good correlation between solid state and solution structures. The difference in cage puckering between silatranes and phosphatranes is obviously due to the fact that the P-N bond is stronger and shorter than the corresponding Si .... N bond. which is only a partial one (vide supra). There­fore. the phosphatrane cage is more strongly compressed.

CONCLUDING REMARKS

The present results show that in solution the silatrane structure comprises a Si .... N bond. and has conformational properties similar to the solid-state structures. Furthermore, conformational transmission appears to be a general phenomenon. and in future studies other pentacoordinated systems which can accommodate d-orbitals

153

Table 111. Cage conformations of the phosphatranes 9-123 in solution (CD3CN) and solid state.

will be investigated.

Compound

9 10 11 12

Y =OMe y OEt y SMe Y = SEt

crystal data of 12

JHHb

6.2 6.0 6.0 5.8

.:/>'

43 46 46 50

42.3 44.4

• Data taken from ref. 14 and 15. b Coupling constants of the cage in Hz. < NCCO torsion angle </> in deg.

EXPERIMENTAL SECTION

1H NMR spectra were recorded on a Bruker CXP 300 spectrometer at 300.1 MHz. with 32K points and a spectral window of 1 kHz. Tetramethylsilane (8 = 0 ppm) was used as internal standard. Reagents and solvents were dried on 4 À mol­sieves before use.

Silatranes16

A stirred mixture of 10.42 g (0.05 mol) tetraethoxysilane. 7.46 g (0.05 mol) triethanolamine. 0.05 mol appropriate alcohol. and 0.1 g KOH in 70 ml dry p-xylene was heated to reflux. Ethanol was distilled off during several hours (ca. 8 h afforded a good yield). After the reaction. the solution was filtered while hot, and the solvent was evaporated in vacuo. A yellow oil resulted. from which the silatranes could be obtained by repeated extractions with hot n-hexane. The white solid was dried under a stream of nitrogen for ca. 5 h.

1-(2-Methoxyethoxy )silatrane ( 1)

M.p. 115-116°C. C9H19N05Si (249.19): calcd. C. 43.37: H. 7.63: N.5.62; found C. 43.42: H. 7.54: N. 5.61. 1H NMR (acetone-d6): 8 3.00 (6H. t. NCH 2): 3.36 (3H. s. OCH3): 3.46 (2H. t. H2'/Hr): 3.78 (2H. t. H1'/Ht"): 3.81 (6H. t. OCH2).

154

1- ( n-B utoxy) silatrane · ( 2)

M.p. 113-114°C. C10H21 N04Si (247.20): calcd. C, 48.58: H. 8.50: N. 5.67: found C. 48.52: H. 8.30: N. 5.69. 1H NMR (acetone-d6): 8 0.95 (3H. t. CH3): 1.41 (2H. m. CH 2): 1.47 (2H. m. HdHr); 2.98 (6H. t, NCH 2): 3.65 (2H. t. Ht"fHt"): 3.79 (6H. t. OCHL).

1-E thoxysilatrane

M.p. 102-103°C. C8H17N0 4Si (219.18): calcd. C. 43.84: H. 7.76; N. 6.39: found C. 43.36: H. 6.24: N. 7.43. 1H NMR (acetone-d6): 8 1.12 (3H. t. CH3): 2.98 (6H. t. NCH 2): 3.71 (2H. q. CH 2): 3.79 (6H. t. OCH2).

Tetra-alkoxysilanes17

To 0.7 mol alcohol was added carefully while stirring 25.7 g (O.i5 mol) tetra­chlorosilane over 30 min. The liberated hydrochloric acid was absorbed in a gas trap. After addition. the mixture was stirred for 2 h. Distillation in vacuo yielded pure tetra-alkox ysilane.

Tetra(2-methoxyethoxy)silane (3)

Yield: 32.1 g. 65% B.p. 138°C/107 Pa. 1H NMR (acetone-d6): 8 3.41 (12H. s. CH 3): 3.56 (SH. t. HdHr); 3.99 (SH. t. Ht'/Hl").

Tetra(n-butoxy)silane (4)

Yield: 34.7 g. 72% B.p. 132°C/667 Pa. 1H NMR (acetone-d6): 8 1.02 (12H. t. CH 3): 1.49 (SH. m. CH2): 1.63 (SH. m. HdHr); 3.87 (SH. t. Hr/Hl").

REFERENCES

1. L.H Koole. E.J. Lanters and H.M. Buck, J. Am. Chem. Soc. 106. 5451 (1984).

2. G.H.W.M. Meulendijks. W. van Es. J.W. de Haan and H.M. Buck. Eur. J. Biochem. 157. 421 (19S6).

3. N.K. de Vries and H.M. Buck. Reel. Trav. Chim. Pays-Bas 105. 150 (1986).

4. M.H.P. van Genderen. B.C.C.M. olde Scheper. L.H. Koole. L.H. van de Ven and H.M. Buck. Phosphorus and Sulfur 32. 73 (1987).

5. Silatrane is a widely used abbreviation for 2.S.9-trioxa-5-aza-1-sila­tricyclo(3.3.3.01·5]undecane.

6. M.G. Voronkov. Topics Curr. Chem. 84. 77 (1979).

155

7. J.W. Turley and P.F. Boer. J. Am. Chem. Soc. 90. 4026 (1968).

8. L. Párkányi. K. Simon and J. Nagy, Acta Cryst. 830. 2328 (1974).

9. L. Párkányi. J. Nagy and K. Simon. J. Organomet. Chem. 101. 11 (1975).

10. M.G. Voronkov. V.M. Oyakov and S.V. Kirpichenko. J. Organomet. Chem. 233. 1 (1983). pp. 59 and 64.

11. C.A.G. Haasnoot. F.A.A.M. de Leeuw and C. Altona. Tetrahedron 36. 2783 (1980).

12. Yu.L. Frolov. S.G. Shevchenko and M.G. Voronkov. J. Organomet. Chem. 292. 159 (1985).

13. V.F. Sidorkin. V.A. Pestunovich. V.A. Shagun and M.G. Voronkov. Dokl. Akad. Nauk SSSR. Sekt. Khim. 233. 386 (1977).

14. D. van Aken. Thesis, Eindhoven University of Technology. 1981.

15. D. van Aken. 1.1. Merkelbach. A.S. Koster and H.M. Buck. J. Chem. Soc" Chem. Comm. 1045 (1980).

16. M.G. Voronkov and G.I. Zekhan. Khim. Geteros. Soed. 1. 210 (1965).

17. E.W. Abrahamson. 1. Joffe and H.W. Post. J. Chem. Soc. 13. 275 (1948).

1.56

CHAPTERtO•

Application of the coliformational transmission effect for the assignment of diastereotopic proton resonances

ABSTRACT

A new rnethod is presented for the assignment of NMR ~esonances of the diastereotopjc protons in tetrahydrofurfuryl and pyrrolidinylrnethyl groups, based on the conforrnational transmission effect. This effect describes the conformational changes produced by increase of coordination from four to five of a phosphorus. sili­con or germanium atom attached to the tetrahydrofurfuryl and pyrrolidinylmethyl moieties. The 1H NMR conformational analysis of these cydic four- and five­coordinated compounds yields two possible results. based on two assignments of the diastereotopic proton resonances. Comparison with the conformations of the analo-­gous tour- and five-coordinated systems with acyclic 2-methoxy-ethyl and dimethylamino-ethyl groups (which lack diastereotopic protons) directly shows which assignment is correct. since a similar conformational transmission occurs in both the cyclic and the acyclic compounds.

*M.H .P. van Genderen and H.M. Buck. Magn. Reson. Chem. 25. 872 (1987).

157

INTRODUCTION

High resolution NMR techniques have proved to be an invaluable tool in study­ing molecular conformations in solution. Especially the vicinal coupling constants in 1H NMR spectra provide detailed structural information. since their magnitude is related to the conformation around the centra! bond. as was first postulated by Karplus.1 With the Karplus equation. the coupling constants in each staggered rota-­mer can be calculated. and the rotameric contributions to the e.xperimental coupling constants can be determined. An essential requirement for this technique is the correct assignment of all proton resonances in the spectrum. This is usually a straightforward procedure. but difficulties are encountered for diastereotopic protons. since they have only small chemica! shift differences. This situation arises e.g. in the conformational analysis of nucleotide systems. where the assignment of H5• and Hs" is crucial for the discrimination between the g1 and g- rotamers around the C.r-C5•

bond (see Figure 1) on the basis of the proton-proton coupling constants J4•5• and J.rs"· An incorrect assignment will result in appro.ximately reversed populations for g1

and g-. The correct identification of Hs· and H5° in the NMR spectrum is also essen­tial for the conformational analysis of the 0 5·-C5• bond. based on the phosphorus­proton coupling constants Jp5• and Jps" (vide infra).

Os• Os• 05• 0*3' "*" c*"" base Hs• Hs" H5• Hs" Hs• Hs"

H4' C3• 041

g+ gt g-,,...,

Figure 1. Left: Structure of a nucleotide fragment with the diastereotopic protons Hs· and Hs". Right: Newman projections of the staggered rotamers around the C4·­

Cs· bond.

For nucleotide-like systems. various methods have been used to arrive at a correct Hs· /Hs" assignment. The most exact method is the stereospecific replacement

of one of the diastereotopic protons by deuterium. which can be achieved by enzy­matic reactions using a deuterium source. lnspection of the 1H NMR spectrum will then directly show the correct assignment. However. this procedure is quite elaborate and time-consuming. lt has been applied by Ritchie and Perlin2 for adenosine. by Gerit and Youngblood3 for tetrahydrofurfuryl alcohol. and more recently by Meulen­dijks et al:4 for phospholipid model systems. For DNA systems. Remin and Shugar5

158

derived an assignment by taking in to account the: specific .shielding effects on Hs~;and

H5" due to. the 3\•phospllate group, This was tonfirmed by work of Davies and Rabczènko.ti."The.fact that the gauche effeet7 fav,ours the,g1 conformation over the g:­conf<ll'rmátiorr in;nucleotide systems has been:used by, Altona;8 who derived :an.ass1gn,­ment from vaFiable-temperawre studies on several:di- .and trinucleotides, On·-r:aising the tempetature.: the .confor.mational preference: shifts from .the dominant g~trotamer mostly tow!lfd thlZ•g1; and less toward the g-. rot.amer: ltJ:a similar way. KQ<>le et al.9

compared tetrahydrofurfuryl and cyclopentanemethyl dimethyl phosphates. and assigned H5• and H5" to account for the presence of the gauche effect in the former systems. Also. Koole et al.rn have observed that lower solvent polarities. which enhance 0 5·-04• ch~rge repulsions. lead to the expected larger 11;- populations (with

0 5· and 0 4• trans) only for·or\è,assignment in an adenosine mod~I system.

All these techniques yielded the same assignmeot, which is generally used in conformational studies of this type. viz. H5• resonates downfield from H5". or

85· > 85". In the present work. we wish to presen{a heW method for assigning the

resonances of this type ''M·.:dî'astereotopic protons, ~h i~ based on conformational transmission.

CONFORMATIONAL TRANSMISSION

The conformational transmission effect describes the conformational change in a phosphorylated molecule, that occurs when the phosphórus atom raises its coordina­tion from four (P1v) to five (Pv).9•11'The resulting trigomÎLbipyramidal (TBP) struc-

. .. . ture has a higher electron density on the axial ligands, and the ehharked electrostatic repulsions cause changes in the conformational preferences. A clear example is found in l and 2. where it could be ascertained that the tetrahydrofurfuryl group in the axis of the TBP of 2 has a dominant g- population (x(g-) = 0.68 in acetone-d6). with 04· and 0 5• trans. while the equatorial tetrahydrofurfuryls (x(g-) = 0.20) and the tetrahydrofurfuryls in 1 (x(g'-) 0.13) prefer the g+ and g1 rotamers.9

l

169

. . , . . .

lnll)is analysis. thè standard assignment was used for the protons Hs·/Hs" (vide supra). For pv compe>unds pseudorotation occursP which resuks in a fast inter­c;:hang~ of áxial . and equat<>i'ial groups. Therefore. the C4·-C5• confc:>rmation that is obtained directly from the spectrum is an average of the axial and equatorial confor­mations. which causes tonfoimationaJ lransmission to be seen in an obscured way. However~_ comparing the average conformations of the tetrahydrofurfuryl plV and pV

compounds. 3 and 4 it is still obvióus that the g- population increases upon

0 EtQ__ ; ~p

EtO '

s~ 3

0 MeO-. · Q ·p MeO ....... '

0

CoMe

5

.Me

o~ .Me

EtO •• 1 y ·p 0

Et0__...1-

·(> 4

Me

o~ Me MeO-. 1 f ·p-o MeO__... I

0

(OMe

6

raising the phosphorus coordination9 (see Table 1). Conformational transmission has alsó been found on going from plV to pV in the acyclic 2-methoxy-ethyl compounds 5 a.nd 611 (see Table 1)~ which have the P-0-C-C-O fragment in common with the tetrahydrofurfuryl compounds.

Comparing the populations .in 3.4 and 5.6. we can conclude that the standard assignment for the .diastereotopic H5·/Hs" protons has indeed been correct. since the reverse assignment would have led to an increase of the g1 instead of the ,- popula­tion upon going from 3 (P1v) to 4 (Pv). This process of comparing conformational tninsmission in symmetrie acyclic and asymmetrie cyclic compounds to arrive at an Hs·/Hs" assignment is applied in the present Work for several other compounds.

160

T able 1. C4·-C5• rotamer populations for 3-6 in acetone-di;.

x(g+) . 0.41 x(g1

) 0.44 x(g-) 0.15

0.32 .0.43 0.35 0.43 0.33 0.14

•Data taken from ref. 9. bDifta taken fron;i ref. 11.

C4·-C5· CON FOR,MA TIONAL ANAL v,s1s.

0.36 0.36 0'.28

The conformational analyses of th.e C4·-C5• bond are based on the modified Karplus relation as developed by Haasnoot et al.13:

JHH = P1cos21P + P2cos<t> + P3 + r,A::dP4 + P5cos2(giq, ·+ P6IL\x;1 H i

with: ÀXj = Ax{" - P1r.Ax/ j

In this equation. </> is the proton-proton torsion angle. ÄX; is the Huggins electronega­tivity14 relative to hydrogen. corrected for 13-substit'ue~ts. and ei is a sub~tituènt orientation parameter. In Table ll. values of P1 through P7 are listed for both. cydic

. . \"· · .. ; }

and acydic systems.

Tabl~ ll. Values ~f the parameters in the Karplus e~uation.

Cydic Acyclic

P1 13.22 13.89 P2 -0.99 -0.98 P3 0 0 P4 0.87 1.02 Ps -2.46 -3.40. p6 19.9 14.9 P1 0 0.24

As can be seen. for cyclic systems the influence of 13-substituents is absent, since P7 = 0. In each staggered rotamer (see Figure 1) the values of Jn· and J4·s" can be cal­culated. and population densities are obtained from the equations:

161

with the normalization equation: x(g+) + x(g1) + x(g-) = 1. Of course. the two possible Hs'fH5" assignments lead to two possible solutions for the conformational equilibria. In the acyclic compounds. the conformational analysis is simplified by the absence of diastereotopic protons. and only one coupling constant J4·s· is found. Since the protons H4· and H4" are identical. as well as Hs· and Hs"• the rotamers g+ and g1 are symmetry-related (see Figure 2). and their populations are always equal in acyclic systems.

Os• Os• Os•

~H,· "*0.' "*"•' Hs• Hs" Hs• Hs" Hs• Hs" H4• H4" 04•

g+ gt g-

Figure 2. Newman projections of the staggered rotamers around the C4·-Cs· bond in the acyclic systems.

RESUL TS AND DISCUSSION

2-(1-Methyl-pyrrolidlnyl)methyl plV and pV compounds

The four- and five-coordinated phosphorus compounds 1 and 8 are derived from the amino-acid L-proline. The reduced and methylated form of L-proline. 2-(1-methyl-pyrrolidinyl)rnethanol. has recently been used to prepare highly efficient phos­phine ligands for catalytic asymmetrie reactions15 and has served as a building block in asymmetrie alkaloid synthesis.16 The compounds 1 and 8 are similar to the tetrahydrofurfuryl systems 3 and 4. with the endocyclic oxygen 04· replaced by N(CH3}. In the acyclic systems 9 and 10. it has been shown that conformational transmission is present (see Tables 111 and IV). due to an increased 0-N electrostatic repuls ion. 11 We have now synthesized the corresponding cyclic systems. and deter­mined their C4·-C5· conformations for the two possible H5·/H5" assignments (see Tables 111 and IV). The results dearly show that 85• < 85" is the correct assignment. since oonformational transmission (i.e" an enhanced g- population) is observed only then. Note that this assignment differs from the one found for tetrahydrofurfuryl systems. so replacernent of 0 by N(CH3) already changes the shift sequence of Hs·

and Hs"·

162

0 MeO-- ,; ·p MeO........ '-.,,.

(o N~Me 'Q>

1

9

Conformational analysis of the 0 5·-C5• bond

Me

O~Me MeO- --~ __:_/ Me0.........-,-

0

~ 8

10

The assignment 85• < 85" which we have now obtained. can be used for a detailed conformational analysis of the 05·-Cs· (13) bond in the plV compounds 7 and 9. using the vicinal coupling constants between phosphorus and H5·/H5". The values of Jps' and Jps" in the staggered rotamers 13+. /r, and /31 (see Figure 3) can be cal­culated with a Karplus relation. proposed by Lankhorst et al.17:

JpH = 15.3 cos2t/,l-6.1 cos<b+l.6

where eb is the torsion angle P-05·-Cs·-C4·. With these theoretica! values {see Table V). one can derive the 0 5·-C5• population densities from the set of equations:

Jp5• = 2.5x{/3+) + 21.4x(lr) + 2.5x{/31)

Jps" = 21.4x(l3+) + 2.5x(fr) + 2.5x(/31) x(/3+) + x(l3-) + x(/3 1

) = 1

163

164

J4•5· J4•5"

x(g+) x(g')

Table 111. Calculated C4·-C5· coupling constants (Hz) for 7-10.

Cyclic 7,8 J4•5• 2.36 3.76 10.96 J4'5" 1.58 10.96 4.53

Acyclic 9,10 J4•5• 4.69 4.69 7.66

Table IV. Measured C4·-C5· coupling constants (Hz) and C4·-C5· rotamer po­pulations for 7-10 in acetone-d6.

85· < 85"

7 8

4.85 5.09 6.05 6.62

0.39 0.31 0.24 0.23

7

6.05 4.85

8

6.62 509

9•

4.60

x(g-) 0.37 0.46

0.38 0.38 0.24

0.30 0.46 0.24

0.31 0.31 0.38

•Data taken from ref. 11.

p~ Hs• Hs"

rr Figure 3. Newman projections of the staggered rotamers around the 05·-C5· bond in 7 and 9.

10•

6.20

0.24 0.24 0.52

Table V. Calculated phosphorus-proton coupling constants (Hz) in 7 and 9.

13'

Jp5' 2.4 23.0 2.4 Jp5" 23.0 2.4 2.4

From the experimental data in Table VI it is obvious that in the acyclic compound 9. where the values of Jp5• and Jps" are necessarily equal. the 13' rotamer is dominant.

while 13+ and 13- have equal populations

Table Vl. Measured phosphorus-proton coupling constants (Hz) and 05·-C5· rotamer populations for 7 and 9 in acetone-d6.

7 9

Jp5· 7.21 7.96 Jp5" 6.71 7.96

x(f3+) 0.22 0.13 x(f3-) 0.25 0.13 x(f3 t) 0.53 0.74

In the cyclic system 7. the conformational equilibrium is also dominated. albeit less.

by 13 1• but now it is found that the 13+ and 13- rotamers have different population

densities. The correct assignment of the diastereotopic protons H5· /Hs" is now essen­tial to discriminate 13+ and 13- properly.

Tetrahydrofurfuryl Si1V and Siv compounds

A different type of compounds are the four- and five-coordinated silicon systems

11-14. where a Si+- N transannular interaction creates a TBP structure in the sila­tranes18 12 and 14. These latter systems are of considerable interest. since they are

biologically active with a broad spectrum of action.19 A conformational transmission effect has been established in the acyclic systems20 (see Tables VII and VIII). which is

165.

~") ~-

_,;.S•-O 0 1

0

\? 11 12

Me3Si-O~OMe

13 14

due to the transfer of electron density from the nitrogen lone pair via silicon to 05·.

The conformational transmission appears to be stronger than in the phosphorylated tetrahydrofurfuryl systems. but it must be taken into account that in the silatranes Os· is located in an axial position always. while in the pV systems pseudorotation dis­

tributes Os· over the equatorial and axial locations. Comparison with an axially

located tetrahydrofurfuryl group (x(g-) = 0.68. vide supra) shows that the confor­mational transmission effect is actually smaller than in the corresponding phosphorus compounds. This is due to the fact that the bond between silicon and nitrogen is only a partial one (estimated to be one fourth of a silicon-nitrogen single bond21).

lt should be noted. that four-coordinated systems with four Si-0 honds are in principle better for comparison with the silatranes. This was accomplished for 13 by studying 15. which was found to have an identical C4·-C5· conformation as 13. How­

ever. in the cyclic case we were unable to obtain 16 in a pure form. and we therefore used the trimethylsilyl systems. as no difference in C4·-Cs· conformation is expected.

For the cyclic systems 11 and 12. conformational analysis shows that only the assignment 85• < 85", or H5· upfield from H5" is consistent with the conformational

transmission behaviour (see Tables VII and VIII). Also here. a reverse assignment is found compared to the phosphorylated tetrahydrofurfuryl compounds.

166

J4•5• J4'5"

x(g+) x(g') x(g-)

15 .16

Table VII. Calculated C4·-C5· coupling constants (Hz) for 11~14.

Cyclic 11,12 J4•5• 2.84 3.07 10.68 J4•5" 0.90 10.68 5.01

Acyclic 13,14 J4'5' 4.05 4.05 7.50

Table VIII. Measured C4·-C5· coupling constants (Hz) and C4·-C5· rotamer populations for 11-14 in acetone-d6.

85· < 85" 85· > 85'"

11 12 11 12 13 14•

5.05 6.95 5.08 4.81 5.08 6.02 5.08 4.81 5.05 6.95

0.41 0.20 0.41 0.23 0.35 0.22 0.31 0.18 0.31 0.52 0.35 0.22 0.28 0.52 0.28 0.25 0.30 0.56

•oata taken from ref. 19.

167

Tetrahydrofurfuryl Ge1V and Gev compounds

The four- and five-coordinated germanium compounds 17 - 20 are related to the abovementioned silicon systems. For the five-coordinated germatranes. it is known from X-ray crystal structures that they also possess a cage structure with a TBP geometry around germanium. 22 The properties of these systems. however. have not yet been studied in great detail. The data that we report here is to our knowledge the first conformational study of germatranes in solution.

(L! 0 1

0

~ 17 18

19 20

Since the Huggins electronegativities of germanium and silicon are equal.14 the calcu­lated coupling constants of Table VII can also be applied here. In the acyclic system 19 and 20. the CrC5• conformational analysis dearly indicates conformational transmission (see Table IX). This is direct evidence that the Ge ..... N transannular interaction. which was found in the solid state. is also present in solution. The change in the population of the g- rotamer is similar to the one found in the silicon systems (see Table VIII). This indicates that the transfer of electron density to 0 5•

due to the transannular interaction is virtually unchanged by substitution of silicon by germanium. From the conformational analysis of the cyclic systems 17 and 18. and comparison with the acyclic compounds. it is obvious that also here the correct assignment must be 85· < 85" (see Table IX).

168

Table IX. Measured C4·-C5· coupling constants (Hz) and C4·-C5· rotamer po­pulations for 17-20 in acetone-du.

85• < 85" 85· > 85"

17 18 17 18 19 20

J4•5• 4.88 7.05 4.88 4.78 4.96 6.00 J4'5" 4.88 4.78 4.88 7.05

x(g+) 0.43 0.28 0.43 0.23 0.37 0.21 x(g') 0.32 0.18 0.32 0.54 0.37 0.21 x(g-) 0.25 0.54 0.25 0.23 0.26 0.58

CONCLUDING REMARKS

The present results show that the conformational transmission effect can be a useful tool for identifying 1H NMR resonances of diastereotopic protons. We were able to obtain an assignment for phosphorylated tetrahydrofurfuryl and pyrrolidinyl­methyl systems, and for silylated and germanylated tetrahydrofurfuryl compounds. For the DNA-like phosphorylated tetrahydrofurfuryl. the assignment is in accordance with earlier studies (vide supra). but for the other three species a reverse assignment is found. These results show the necessity of carefully checking the diastereotopic proton assignments in new types of compounds. The new method described here can be of assistance in this process.

EXPERIMENTAL SECTION

Spectroscopy 1H NMR spectra were recorded at 300.1 MHz in the FT mode on a Bruker CXP

300 spectrometer equipped with an Aspect 2000 computer. Usually, 32K data points and a 1.5 kHz spectra! window were used. resulting in an accuracy for line positions of 0.05 Hz. Samples were dissolved in acetone-d6 (unless indicated otherwise) at a

concentration of 5 mg/ml in 5-mm NMR tubes. and measured at 300 K. Chemica! shifts are related to tetramethylsilane (ö = 0). 31 P NMR spectra were run at 36.4 MHz on a Bruker HX 90R spectrometer with 85% H3P04 as external reference (ö = 0).

169

Synthesis

THF was carefully dried with lithium aluminium hydride. distilled. and stored on

4 À molsieves. Diethyl ether was dried on sodium metal. Benzene and toluene were

dried on 4 À molsieves. The syntheses of 3 and 4 are described in ref. 9. those of 5.

6. 9. and 10 in ref. 11. and of 14 and 15 in ref. 19.

Dimethyl 2-(1-methyl-pyrrolidinyl)methyl phosphite

2-Pyrrolidinemethanol. obtained by reduction of L-proline. 23 was methylated 24 to

form 2-(1-methyl-pyrrolidinyl) methanol. which was reacted with dimethoxychloro­phosphine.11 Distillation in vacuo yielded the product in 40% yield: b.p. 42-46°(/0.03

mmHg. 1H NMR (CDC13): S 1.5-1.8 (4H. m. Ht /Hr). 1.9-2.2 (2H. m, Hr}. 2.3 (3H.

s. NCH 3). 2.8--3.1 (1H. m. H4·). 3.6-3.8 (2H. m. H5·/H5"). 31 P NMR (CDC1 3): S 145.1.

Dimethyl 2-(1-methyl-pyrrolidinyl)methyl phosphate (7)

This compound was prepared by the reaction of dimethyl 2-(1-methyl­

pyrrolidinyl)methyl phosphite with ozone. according to the procedure described in ref.

11. 1H NMR: S 1.8 (4H. m. H2'/H:r). 2.0 (2H. m. Hl'}. 2.6 (3H. s. NCH3). 3.0 (1H.

m. H.r). 3.7 (6H. d. OCH3). 4.1 (2H, m. H5-/Hd. 31 P NMR: S 6.6.

2,2-Dimethyl-2-(2-(1-methyl-pyrrolidinyl)methyl)-4,5-dimethyl-

1,3,2-dioxaphosphol-4-ene (8)

This compound was prepared by addition of a few drops butanedione to dimethyl 2-(1-methyl-pyrrolidinyl)methyl phosphite at Q°C in an NMR sample tube.

After ca. 30 min. 31 P NMR showed the formation of 8 to be complete. 1H NMR: S 1.8 (4H. m. Hr/H3·). 2.0 (2H. m. Hr). 2.4 (6H. s. CH3 dioxaphospholene ring). 2.5

(3H. s. NCH 3). 3.3 (1H. m. H4·). 3.7 (6H. d. OCH3). 3.8--4.0 (2H. m. H5)H5"). 31 P NMR: S -44.1.

(2-Methoxy-ethoxy)trimethylsilane (13) 25

To a stirred solution of 7.6 g (0.1 mol) 2-methoxy-ethanol and 7.9 g (0.1 mol)

pyridine in 30 ml diethyl ether was added dropwise 10.9 g (0.1 mol) trimethylsilyl­

chloride. After addition. the mixture was stirred for 4 h. The pyridinium hydrochloride

was removed by filtration. and the filtrate was concentrated. Distillation yielded pure

13: b.p. 119-120°(. 1H NMR: s 0.2 (12H. s. SiCH3). 3.4 (3H. s. OCH3). 3.5 (2H. m. H4·/H,f'). 3.8 (2H. m. H5·/H5").

170

(Tetrahydrofurfuryloxy)trimethylsilane (11)

This compound was prepared from tetrahydrofurfuryl alcohol (10.2 g. 0.1 mol) and trimethylsilylchloride (10.9 g. 0.1 mol) according to the procedure described for (2-methoxy-ethoxy)trimethylsilane. B.p. 58-59°C/9 mmHg. 1H NMR: 8 1.7-1.9 (4H.

m. Ht/H 3·). 2.9 (2H. m. H5-fH5"). 3.0 (1H. m. H4'). 3.1 (2H. m. Hl').

1-Tetrahydrofurfuryloxy-2,8,9-trioxa-5-aza-1-silatricyclo[3.3.3.01·5]undecane

(12) This compound was prepared from tetrahydrofurfuryl alcohol (5.1 g, 0.05 mol).

tetraethoxysilane (10.42 g. 0.05 mol) and triethanolamine (7.46 g. 0.05 mol) accord­ing to the procedure described in ref. 26. 1H NMR: 8 1.8 (4H. m. H2·/H 3·). 2.9 (6H.

t. NCH 2). 3.4 (1H. dd. H5-). 3.6 (1H. dd. Hç). 3.7 (6H. t. OCH 2). 3.8 (3H. m.

H4-/Hr).

Tetra(2-methoxy-ethoxy)germane (19)

Through a mixture of 5 g (23 mmol) germanium tetrachloride and 9.65 g (127 mmol) 2-methoxy-ethanol in 50 ml dry benzene. 5 g (0.29 mol) ammonia was passed. The ammonium chloride was removed by filtration. and the filtrate was concentrated. Distillation in vacuo yielded 7.39 g (85%) pure 19: b.p. 12S°C/0.4 mmHg. 1H NMR: 8 3.32 (12H. s. OCH 3). 3.49 (8H. t. H4-/H4"). 3.9S (SH. t. Hs-/HS").

Tefra(tetrahydrofurfuryloxy)gerinane (17)

This compound· was prepared from germanium tetrachloride (1.06 g. 5 mmol) and tetrahydrofurfuryl alcohol (2.01 g. 20 mmol) in toluene. according to the pro­cedure described for 19. The crude product was pure enough to allow conformational

analysis. 1H NMR: 8 1.5-1.9 (16H. m. H2-fH3·). 3.5 (SH. d. H5-fH5"). 3.6-3.9 (12H.

m. H4• /Hl').

1-(2-Methoxy-ethoxy)-2,8,9-trioxa-5-aza-1-germatricyclo[3.3.3.01·5]undecane (20)

Tetra(2-methoxy-ethoxy)germane (5.52 g. 15 mmol) and triethanolamine (2.20 g. 15 mmol) were heated in 50 ml benzene. and the azeotrope of 2-methoxy-ethanol and benzene was removed during 3 h. After cooling. 20 precipitated as a white solid. Filtration and drying under a stream of nitrogen yielded 1.13 g (26%). 1H NMR: 8

2.95 (6H. t. NCH2). 3.2S (3H. s. OCH3). 3.3S (2H. t. H4·/H4"). 3.73 (6H. t. OCH2).

3.S1 (2H. t. Hs·/HS").

171

l-Tetrahydrofurfuryloxy-2,8,9-trioxa-5-aza-l-germatricyclo­[3.3.3.01.5]undecane (18)

A mixture of tetramethoxygermane27 (3.06 g. 16 mmol). triethanolamine (2.32 g.

16 mmol) and tetrahydrofurfuryl alcohol (1.59 g. 16 mmo!) in 50 ml toluene was heated. and methanol was removed by distillation during 3 h. After cooling. 18 pre­cipitated as a white solid. Filtration and drying yielded 2.78 g (54%). 1H NMR: S

1.7-1.9 (4H. m. H2'fH3·). 2.96 (6H. t. NCH 2). 3.50 (1H. dd. H5·). 3.60 (1H. m. Hr).

3.62 (6H. t. OCH2). 3.76 (1H. dd. HS"). 3.8-3.9 (2H. m. Hl').

REFERENCES

1. M. Karplus. J. Chem. Phys. 30. 11 (1959).

2. R.G.S. Ritchie and A.S. Perlin. Carbohydrate Res. 55. 121 (1977).

3. JA Gerit and A.V. Voungblood. J. Am. Chem. Soc. 102. 7433 (1980).

4. G.H.W.M. Meulendijks. W. van Es. J.W. de Haan and H.M. Buck. Eur. J.

Biochem. 157. 421 (1986).

S. M. Remin and D. Shugar, Biochem. Biophys. Res. Comm. 48. 636 (1972).

6. D.B. Davies and A. Rabczenko. J. Chem. Soc. Perkin Trans. ll. 1703 (1975).

7. (a) S. Wolfe. Acc. Chem. Res. S. 102 (1972). (b) A.J. Kirby. "The Anomeric Effect and Related Stereoelectronic EfTects at Oxygen". Springer Verlag. Berlin. 1983.

8. C. Altona. Reel. Trav. Chim. Pays Bas 101. 413 (1982).

9. LH. Koole. E.J. Lanters and H.M. Buck. J. Am. Chem. Soc. 106, 5451 (1984).

10. L.H. Koole. R.J.L van Kooyk and H.M. Buck. J. Am. Chem. Soc. 107. 4032 {1985).

11. M.H.P. van Genderen. B.C.C.M. olde Scheper, L.H. Koole. L.H. van de Ven and H.M. Buck. Phosphorus Sulfur 32. 73 (1987).

12. R. Luckenbach. "Dynamic Stereochemistry of Pentacoordinated Phosphorus and Related Elements". Georg Thieme Verlag, Stuttgart. 1973. p. 3-5.

13. C.A.G. Haasnoot. F.A.A.M. de Leeuw and C. Altona. Tetrahedron 36. 2783

(1980).

14. M.J. Huggins. J. Am. Chem. Soc. 75. 4123 (1953).

15. T. Hayashi. M. Konishi. M. Fukushima. K. Kanehira. T. Hioki and M. Kumada. J. Org. Chem. 48, 2195 (1983).

16. C.G. Chavdarian and E.B. Saunders. U.S. Patent 4 321 387 A, 1982. Chem. Abstr. 99, 22575d (1982).

172

17. For a review of silatrane (2.8.9-trioxa-5-aza-1-silatricyclo[3.3.3.01.5]undecane) structure and chemistry. see: M.G. Voronkov. V.M. Dyakov and S.V. Kirpi­

chenko. J. Organomet. Chem. 233. 1 (1983).

18. P.P. Lankhorst. C.A.G. Haasnoot. C. Erkelens and C. Altona. J. Biomol. Struct. Dyns. 1. 1387 (1984).

19. M.G. Voronkov. Top. Curr. Chem. 84. 77 (1979).

20. M.H.P. van Genderen and H.M. Buck. Reel. Trav. Chim. Pays Bas 106. 449

(1987).

21. V.F. Sidorkin. V.A. Pestunovich. V.A. Shagun and M.G. Voronkov. Dokl. Akad. Nauk SSSR Sekt. Khim. 223. 386 (1977).

22. L.O. Atovmyan. Va. Va. Bleidelis. A.A. Kernme and R.P. Shibaeva. Zh. Strukt.

Khim. 11. 318 (1970).

23. D. Enclers and H. Eichenhauer. Chem. Ber. 112. 2933 (1979).

24. Houben & Weyl. "Methoden der Organischen Chemie", Vol. 11-1. Georg Thieme Verlag. Stuttgart. 1957. p. 650.

25. A.E. Pierce. "Silylation of Organii:: Compounds". Pierce Chemica! Co" Rockford. 1968. Chapter 1.

26. M.G. Voronkov and G.I. Zelchan. Khim. Geter. Soed. 1. 210 (1965).

27. O.C. Bradley. L.J. Kay and W. Wardlaw. J. Chem. Soc" 4916 (1956).

173

CHAPTER 11*

The role of hydration and stereoelectronic effects in the hydrolysis of cAMP

ABSTRACT

With the help of 1H NMR measurements on a model system (either tetrahydro­furf uryl diphenyl phosphine oxide or 2' .3' -0-îsopropy lideneadeosine 5' -dimethylphosphate) representative for 5' -AMP. it was possible to obtain quantitative information on the regiospecific hydration between 0 5• and 0 4-. Furthermore. the

magnitude of the gauche effect influencing the Cr-Cs· conformation was evaluated.

Both aspects contribute to the large exothermic enthalpy of hydrolysis of cAMP.

*M.H.P. van Genderen. L.H. Koole, R.J.L. van Kooyk and H.M. Buck, J. Org. Chem.

50. 2380 (1985).

174

INTROOUCTION

lt is well-known. that the coenzyme cydic adenosine 3',5'-monophosphate1 is enzymatically hydrolyzed to adenosine 5' -monophosphate1 with a large exothermic enthalpy (-11.1 kcal/mol). in contrast to trimethykme phosphate (-3.0 kcal/mol). Experimental and theoretica! work has demonstrated that the large exothermic enthalpy of cAMP is caused by strain. stereoelectronic effects. and solvation effects. The various, contributions to the enthalpy difference between cAMP an~ trimethylene phosphate. i.e" 8.1 kcal/mol (vide supra). can be derived from experirriental work caried out by Gerit et a1.2- 4 Their calorimetrie measurements showed that. 4.6 kcal/mol is immlved for strain. caused by the trans fusion of a cyclopentane ring to trimethylene phosphate (see Figure 1).

0 :1

_:,'._P-o~ o"\~·

' 0 ,

ilH=-3.0kcal/mol a'

l!.H =-10.1 kcal /mol c

Á H=~7.6kcal /mol .b

,0 ' ' ,, '

è\;~tt>-.. /""' - Y ro1 · OH N.......,..N

l!.H=-11.1kcal/mol d

Figure 1. Enthalpies of hydrolysis of trimethylene phosphate (a). trans-2-hydroxycyclopentanemethanol cydic phosphate (b). trans-2-hydroxytetrahydro­furanmethanol cyclic phosphate (c). and cAMP1 (d).

lntroduction of an endocyclic oxygen results in a difference of 2.5 kcal/rnQI. due to stereoelectronic and solvation effects. The presence of the 2' -hydroxyl group and the adenine base on the 1'-location in cAMP is responsible for the remaining 1.0 kcal/mol. The aforementioned stereoelectronic effect disfavours the antiperiplanar arrangement of the phosphate oxygens 0 3• and 0 5• and the ribose oxygen 0 4•

(gauche effect5·6). The magnitude of this gauche effect was assessed with NMR measurements of the equilibrium between axial and equatorial methoxy in . 3-

175

methoxytrimethylene phosphate.4

,o ·I

ó;:'.\:::; OCH3

From these measurements it followed that the axial methoxy location {oxygens gauche) is 1.0 kcal/mol lower in energy than the equatorial methoxy location (oxy­gens trans). Therefore Gerit divided the enthalpy difference of 3.5 kcal/mol between the hydrolysis of trans-2-hydroxycyclopentanemethanol cydic phosphate and cAMP in 1.0 kcal/mol due to the gauche effect and 2.5 kcal/mol due to solvation effects. The experimental results are in fairly good agreement with quantum chemica! calcula­tions carried out by Scheffers-Sap and Buck. 7 They found that strain re lief in the ribose ring is responsible for 2.2 kcal/mol (strain in the phosphate ring was not taken into account) of the overall 4.6 kcal/mol difference between the hydrolysis of trimethylene phosphate and trans-2-hydroxycydopentanemethanol phosphate (vide supra). According to Scheffers-Sap and Buck. the solvation effect as suggested by Gerit is a specific hydration between Os· and 0 4-. which is absent in cAMP since the distance between the oxygens is too large. They obtained for this effect a value of 2-3 kcal/mol. which is in good agreement with the experimental value of 2.5 kcal/mol for solvation (vide supra). For the hydration between 0 5• and 0 4• two structures were proposed. viz. a five-membered ring structure and a seven-membered ring structure.

/-H.. H \

\o*"bs· C3•

Hs• Hs" H4•

H-0 \

H,,

/*'Os• 04' C3•

Hs• Hs" H4•

These structures were calculated to differ only 0.3 kcal/mol in favour of the five­membered ring.7 a difference too small to select one of the structures. Presently we report new experimental work concerning the magnitude of the gauche effect and the solvation structure. based on a conformational analysis of the model systems 1 and 2. In particular. we focused on the conformation around the C"r-Cs· linkage. which determines the position of Os· with respect to 0 4•. The C4·-Cs· conformation can be described as an equilibrium between the three staggered rotamers gauche(+). (g+). gauche(trans). (gt). and gauche(-). (g-). The population densities x(g+). x(g1

). and x(g-) of these rotamers were calculated trom the NMR spin-spin coupling constants

176

Ph ___ ~O p

Ph/ \ 0

5 '( / 0 -:::::;;..

4~

1

g+

0)(0 2

Me Me

gt g-

Jn· and J4·s" as described by Koole et al.8 Using this method. we wère able to obtain an independent value for the magnitude of the gauche effect of 1.0 kcal/mol (vide supra). In order to isolate the gauche effect from other factors. we .used the simplified modél system 1. The thermodynamic parameters of the conformational equilibrium around C4·-Cs· were determined with variable-temperature 1H NMR spectroscopy. Compilation of the conformational data at various temperatures in a van 't Hoff plot yielded 6H0(g- .g+) -0.9 kcal/mol. b.S0(g- .g+) = -1.3 cal/mol. K. 8H0(g- .g1) = -1.2 kcal/mol. and 6S0(g- .g1) = -2.0 cal/mol.K.9 Since in cAMP the C4·-Cs· bond ·is locked in the g- conformation, while in 5' -AMP the g+ rotamer is dominant. our value of 0.9 kcal/mol for the gauche effect is in excellent agreement with Gerlt's observation.

The solvation structure in 5'-AMP was elucidated by determination of the C4·­

C5· conformation of 2 in various solvents ( see T able 1). In Figure 2. the rótamer populations of g+, g1• and g- are presented as functions of the solvent polarity ET.10

lt appears that the population of the g- rotamer. in which 0 5• is trans to 0 4••

increases as the solvent polarity is lowered. This can be attributed to a charge repul­sion between Os· and 04·. which becomes more effective at lower polarities. No par­ticular trends are observed for the variations of the g+ and g1 rotamer populations with ET. However. it can be seen directly that g+ is the dominant C4·-Cs· rotamer in water. whereas g1 is clearly preferred in all other solvents. including methanol. This

117

Table 1. Measured population densities of the three rotamers around the C4·-C5· bond of 2 in various solvents.

Solvent x(g+) x(gt)

C6D6 34.5 0.25 0.41 CDCl3 39.1 0.33 0.38 (CD3)2CO 42.2 0.21 045 (CD3)~0 45.0 0.17 0.53 CD3CN 46.0 0.26 0.47 CD30D 55.5 0.27 0.46 D20 63.1 0.50 0.41

• 0.5 • 0.5 0.5

t 0.4 t 0.4 t 0.4 ••

• • x(g•l • xlg·l • • x(gll Q.3 0.3 ••• 0.3

• • • • 0.2 • 02 0.2

• 0.1 0.1 • 0.1

30 40 50 60 70 30 40 50 60 70 30 40 50

Ey-. Er --Figure 2. Population densities of the rotamers in solvents of various polaritiesrn

x(g-)

0.34 0.29 0.34 0.30 0.27 0.26 0.09

• •

60 70

Er -change in preference must be due to a specific solvation between 0 5• and 0 4• in water that favours the g+ rotamer. lt seems reasonable to assume that this solvation has a seven-membered ring structure that can exist in water only.

EXPERIMENTAL SECTION

Spectroscopy 1H NMR spectra were run in the FT mode at 300 MHz on a Bruker CXP 300

spectrometer and at 500 MHz on a Bruker WM 500 spectrometer. Both instruments are interfaced with an Aspect 2000 computer. A standard computer simulation­iteration procedure11 was employed to obtain accurate values for spin-spin coupling

178

constants. 31 P NMR spectra were run in the FT mode at .36.4 MHz on a Bruker HX 90R spectrometer with a Digilab FT,-NMR-3 pulsing atcessory. 31P chemica! shifts are related to 85% H3P04 as an external standard (S 0 ppm). ·.

Synthesis

Tetrahydrofurfuryl diphenyl phosphine (1)

This compound was prepared as described in ref. 8.

Dimethoxy(N,N-dimethylamino)phosphine

Phosphorus trichloride (0.5 mol. 69 g) was added over 30 min to trimethyl phos­phite (1 mol. 124 g) that was kept at60"C. After completion of the addition the reac­tion mixture was cooled to 0°C and dliuted with 500 ml of sodiurrf.dried diethyl ether. Dimethylamine (3 mol. 135 g) was bubbled through the reaction mixture. After filtration of the dimethylamine hydrochloride. evaporation of the sol.vent afforded a yellowish oil that was distilled twice at 45 mmHg through a 20-cm Vigreux to afford 46 g (22%) of the desired product: bp 51-52°C: 1H NMR (Cr,06) 8 .2.63 (6H. d. N(CH3)2. JPNCH 8.8 Hz). 3.42 (6H. d. POCH3. J = 12.0 Hz); 31 P NMR (C6D6) 8 147.6.

2' ,3' -0-lsopropylideneadenosine 5' -dimethylphosphlte

A magnetically stirred solution of 2' .3' -O-isopropylideneadenosine12 (6.51 mmol. 2.00 g) in 30 ml of dry 1,4-dioxane was kept at 85°C. A solution of dimethoxy(dimethylamino)phosphine (11.20 mmol. 1.53 g) in 10 ml of dry 1.4-dioxane was added over 3 h. After stirring for 15 h at 85°C. thin layer chromatogra­phy (TLC) with 2-butanone .as eluent showed the 2' .3' -0-isopropylideneadenosine (Rt = 0.21) to be completely converted in the product (Rt = 0.52). Evaporation of the solvent afforded a yellowish. viscous oil that was purified on a Woelm silica gel column using 2-butanone as eluent. Pure phosphite was obtained as a white crystal­line material in 79% yield: mp 164-165°(: 1H NMR (CDC13) 8 1.41 and 1.64 (6H. s. CH3 isopropylidene). 3.47 (6H. d. POCH3• J = 10.8 Hz). 4.00 (2H" m. H5·/H5"}. 4.48 (1H. m. H4·). 5.05 (1H. dd, H3·). 5.39 (1H. dd. H2·). 6.19 (3H. m. HrJHN2). 8.04 (1H. s. Ha). 8.36 (1H. s. H2): 31P NMR (CDCl3) & 141.1. Anal. Calcd for C1sN22Ns06P: C. 45.11: H. 5.55: N. 17.54. Found: C. 44.95: H. 5.68; N, 18.06.

2' ,3' -0-lsopropylideneadenosine 5' -dimethylphosphate (2)

2'.3'-0-lsopropylideneadenosine 5' -dimethylphosphite (1.13 mmol. 450 mg) was dissolved in 25 ml of dry dichloromethane and an ozone-oxygen (15:85) stream was bubbled through. After 35 min. TLC with 2-butanone as eluent showed the reaction

170

to be complete (R1 = 0.27). Evaporation of the solvent yielded the product as a hygroscopic white solid in 96% yield: 1H NMR (CDCl3) 8 1.40 and 1.62 (6H. s. CH 3

isopropylidene). 3.72 (6H. d. POCH3• J = 11.0 Hz). 4.28 (2H. m. Hs-/Hd. 4.51 (1H.

m. H.r). 5.13 (1H. dd. H3·). 5.46 (1H. dd. H2·). 6.19 {tH. d. H1·0. 6.42 (2H. bs. NH2). 8.08 (1H. s. H3). 8.48 (1H. s. H2): 31 P NMR (CDC13) ö 1.4. Anal. Calcd for C15H22 N50 7P: C. 43.37: H. 5.34: N. 16.87. Found: C. 43.41; H. 5.44; N. 17.09.

REFERENCES

1. The abbreviations used are: cAMP. cyclic adenosine 3' .5' -monophosphate: 5' -AMP. adenosine 5'-monophosphate.

2. J.A. Gerit. N.l. Gutterson. P. Datta, B. Belleau and C.L. Penney. J. Am. Chem. Soc. 102. 1655 (1980).

3. F.J. Marsh. P.Weiner. J.E. Douglas. P.A. Kollman. G.L. Kenyon and J.A. Gerit. J. Am. Chem. Soc. 102. 1660 (1980).

4. J.A. Gerit. N.l. Gutterson. R.E. Drews and J.A. Solokow. J. Am. Chem. Soc. 102. 1665 {1980).

5. S. Wolfe. Acc. Chem. Res. 5. 102 (1972).

6. A.J. Kirby. "The Anomeric Effect and Related Stereoelectronic Effects at Oxy­gen". Springer Verlag. Berlin. 1983. p. 36.

7. M.M.E. Scheffers-Sap and H.M. Buck. J. Am. Chem. Soc. 102. 6422 (1980).

8. {a) L.H. Koole. E.J. Lanters and H.M. Buck. J. Am. Chem. Soc. 106. 5451 (1984). (b) C.A.G. Haasnoot. F AA.M. de Leeuw and C. Altona. Tetrahedron 36. 2783 (1980).

9. 6H0(g- .g+) and 6H0(g- .gt) denote the enthalpy differences H0(g-) - H0(g+) and H0(g-) - H0(g1). respectively. while AS0(g- .g+) and AS0(g- .g1) denote the enropy differences s0(g-) - s0(g+) and s0(g-) - S0(g1

) respectively. The van 't

Hoff plots showed little scatter (straight lines with r2 = 0.997) .8 indicating the reliability of the numerical values.

10. The solvent polarity is not expressed as the bulk dielectric constant but as the empirically determined micropolarity Er. See: J. March. "Advanced Organic Chemistry". McGraw-Hill. New York. 1977, p. 335. and references cited there.

11. PANIC program: Bruker Spectrospin AG. Switzerland.

12. H.P.M. Fromageot. B.E. Griffin. C.B. Reese and J.E. Sulston. Tetrahedron 23. 2315 (1967).

180

SUMMARV

This thesis describes experimental and theoretica! investigations on the struc­ture and stabitity of DNA duplexes when either the charges of the phosphate groups are shielded. or the phosphate groups are activated by raising the coordination number of phosphorus from four to five. The charge shielding diminishes electrostatic phosphate-phosphate repulsions in the DNA duplex structure. both between the strands and within one strand. Methylation of the phosphate groups in oligothymi­dines results in the formation of a parallel duplex with thymine-thymine base pairing (Chapters 2 and 3). due to the absence of interstrand electrostatic repulsions. The parallel duplex can also occur for cytosine-cytosine base pairs. hut only for the Sp

chirality (Chapter 4). since the phosphate methyl group cannot be accommodated for the Rr chirality within the more dense duplex structure with C-C base pairs. Further­

more. it is shown that the phosphate-methylated systems are accurate models for the complexation of polycationic proteins (poly-L-lysine and poly-L-ornithine) with natura! oligopyrimidines (Chapter 4). Parallel duplexes are induced for natura! d(T10)

by both proteins. while d(C10) only forms a duplex after complexation with poly-L­lysine. This selective behaviour is again attributed to a more dense duplex in the case of C-C base pairs. which does not allow a complexation with the pro-R phosphate oxygen by poly-L-ornithine in the parallel duplex.

The stability of antiparallel hybrids of phosphate-methylated and natura! DNA is found to be higher than the stability of the corresponding natural DNA duplexes because of the absence of interstrand electrostatic repulsions (Chapter 5). However. it is also seen that the hybridization has a cooperative character. The cause of this cooperativity is a structural difference between the charged and the neutral strands. since intrastrand repulsions are absent in the neutral DNA. The structural difference represents an unfavourable entropy term in duplex formation. and therefore lowers the duplex stability. Based on a thermodynamic description of the stabilizing and destabilizing effects. it is predicted that protein complexation with one strand in a duplex can lead to destabilization (Chapter 6). This thermodynamic model can explain the mechanism of action of several enzymes (helicases. RNA polymerases. and restriction endonudeases) that are able to open the DNA duplex.

Activation of the phosphate groups to a pentacoordinated (Pv) form induces conformational changes in a phosphorylated molecule. lt is established that this con­formational transmission occurs exclusively via increased electrostatic repulsions between an oxygen in the axis of the trigonal bipyramidal structure and a vicinal hetero atom (Chapter 7). In dinudeotide model systems. it is seen that the presence of a stabilized pV in hydrogen-bond breaking solvents (e.g" DMSO) disturbs the

181

secondary structure by inducing the unusual g- rotamer around the C4·-C5· bond

(Chapter 8). This shows that the formation of a transient pV in the backbone of natura! DNA can lead to conformational transitions of the molecule. The process of conformationat transmission also occurs in pentacoordinated silicon (Chapter 9) and

germanium compounds (Chapter 10). which has given an independent proof for the structure of these compounds in solution. Furthermore. a new method is described

for assignment of proton resonances in 1H NMR spectra. employing the conforma­tional transmission effect ( Chapter 10). Finally. a conformational study of nucleotide model systems gives more insight in the effects of regiospecific hydration on the

hydrolysis of the cyclic nucleotide cAMP (Chapter 11).

182

SAMENVATTING

Dit proefschrift beschrijft een experimenteel en theoretisch onderzoek naar de struktuur en stabiliteit van DNA duplices. waarbij hetzij de negatieve ladingen van de fosfaatgroepen afgeschermd worden. of de fosfaatgroepen geactiveerd worden door het coördinatiegetal van fosfor te verhogen van vier naar vijf. De ladingsafscherming vermindert de elektrostatische fosfaat-fosfaat repulsies in de DNA duple11i .. zowel tussen de strengen onderling als binnen één streng. Methylering ,van de.J05fatim. in oligothymidines leidt tot de vorming v~n een parallelle duplexstruktuur met,thymine;­thymine basenparen. door de afwezigheid van elektrostatische. repulsfos tussen de strengen (Hoofdstuk. 2 en 3). Een vergelijkbare parallelle duplex k.an .ook gevormd worden met cytosine-cytosine basenpa,ren, exclusief voor de 5p ch,iraliteit in de fps­faatgroep {Hoofdstuk 4). Hieraan ligt het feit ten grondslag dat bij deze.chiraliteit de methylgroep van de .fosfaat naar buiten gericht is, terwijl de naar .binnen gekeerde methylgroep in. de Rp chiraliteit niet geag:ommodeerd kan ,wor~11m in de me~r com­pacte duplexstruktuur met C-C basenparen. Verder is aangetoond dat fosfaat­gemethyleerde systemen als modellen voor de complexatie van polykationische eiwit­ten (poly-L-lysine en poly-L-ornithine) met natuurlijk oligopyrimidines te gebruiken zijn (Hoofdstuk 4). Parallelle duplices worden door beide eiwitten veroorzaakt bij natuurlijk d(T10). terwijl natuurlijk d(C10) alleen een duplex vormt na complexatie met poly-L-lysine. Dit selectieve gedrag kan weer toegeschreven worden aan een meer compacte duplex in het geval van C-C basenparen. wat een complexatie met het pro­R fosfaat-zuurstofatoom door poly-L-ornithine in de parallelle duplex onmogelijk maakt.

De stabiliteit van antiparallelle hybriden van fosfaatgemethyleerd en natuurlijk DNA blijkt hoger te zijn dan de stabiliteit van de overeenkomstige natuurlijke DNA duplices door de afwezigheid van repulsies tussen de strengen (Hoofdstuk 5). Verder is gevonden dat de hybridisatie een coöperatief karakter heeft. De oorzaak van deze coöperativiteit is een struktuurverschil tussen de geladen en de neutrale streng. aangezien repulsies binnen de streng in het geval van neutraal DNA afwezig zijn. Het struktuurverschil leidt tot een ongunstige entropieterm bij duplexvorming. wat een verminderde duplexstabiliteit veroorzaakt. Gebaseerd op een thermodynamische beschrijving van de stabiliserende en destabiliserende effecten. blijkt dat de eiwitcom­plexatie met één streng in een duplex kan leiden tot destabilisatie van de duplex (Hoofdstuk 6). Deze wijze van destabilisatie is gebruikt om de werking te verklaren van een aantal enzymen (helicases. RNA polymerases en restictie endonudeases) dat de DNA duplex opent.

183

Activering van de fosfaatgroepen naar een vijfgecoördineerde (Pv) vorm met een trigonale bipyramidale (TBP) configuratie induceert conformationele veranderingen in een gefosforyleerd molekuul. Vastgesteld is. dat deze conformatie-transmissie uitslui­tend plaatsvindt via verhoogde elektrostatische repulsies tussen een zuurstofatoom in de as van de TBP struktuur en een vicinaal hetero-atoom (Hoofdstuk 7). In dinucleo­tide modelsystemen vervormt de aanwezigheid van een gestabiliseerde pV TBP in waterstofbrugbrekende oplosmiddelen (bijv. DMSO) de secundaire struktuur van het DNA door een ongebruikelijke g- conformatie van de C4·-C5• binding te induceren

(Hoofdstuk 8). Hieruit volgt. dat de vorming van een tijdelijke pV in de ruggegraat van DNA kan leiden tot conformationele overgangen in het molekuul. Het proces van conformatie-transmissie treedt ook op in vijfgecoördineerde silicium- (Hoofdstuk 9) en germaniumverbindingen (Hoofdstuk 10). wat tevens een bewijs heeft geleverd voor de struktuur van deze verbindingen in oplossing. Het feit dat conformatie-transmissie optreedt in bepaalde acyclische systemen. is gebruikt om protonresonanties in NMR spectra van analoge cyclische verbindingen toe te kennen (Hoofdstuk 10). Verder is de conformatie van een aantal nucleotide modelsystemen bestudeerd in verschillende oplosmiddelen. hetgeen meer inzicht heeft gegeven in het effekt van regiospecifieke hydratatie op de hydrolyse van het cyclische nucleotide cAMP (Hoofdstuk 11).

184

CURRICULUM VITAE

van Marcel van Genderen. geboren op 1december1961

Na het behalen van het gymnasium-tl diploma aan het van Maerlantlyceum te Eindhoven in 1979. werd in datzelfde jaar begonnen met de studie Scheikundige Technologie aan de Technische Hogèschool Eindhoven. Het kandidaatsexamen werd in 1982 met lof behaald. In 1983 werd mij de Unilever' chemieprijs toegekend: Het afstudeerwerk werd ver'richt in de vakgroep Organischè Chemie. en in 1984 werd het ingenieursexamen met lof afgelegd.

Het onderzoek. beschreven in dit proefschrift. werd gestart in september 1984 en stond onder leiding van prof. dr. H.M. Buck. Sinds. 1 september t984 ben ik als universitair docent in dienst van de Technische .Universiteit Eindhoven. en werkzáam bij de vakgroep Organische Chemie. In 1988 . werd mij de DSM prijs voor c~mie en technologie toegeke~d.

185

DANKWOORD

Aan de totstandkoming van dit proefschrift hebben velen een bijdrage geleverd. Speciale dank gaat uit naar mijn promoter prof. dr. Buck. voor immer stimulerende

ideeën. en naar mijn collega dr. ir. Leo Koole voor een grote bijdrage aan het hier beschreven onderzoek. Ook wil ik mijn collega dr. ir. René Janssen bedanken voor een goede samenwerking. Een grote hoeveelheid werk is verricht door mijn afstudeerders: ir. Bemd olde Scheper. ir. Olav Aagaard. ir. Coert van Lare. ir. Maurice Moody. ir.

Sjoerd Miesen. Martin Hilbers en Ron van Empel. Daarnaast wil ik ook diegenen noe­men die aan het onderzoek hebben meegewerkt in het kader van hun afstuderen bij dr. ir. Leo Koole: ir. Harold Moody. ir. Raymond van Kooyk. ir. Peter Quaedflieg en ir. Niek Broeders. Ook hebben vele studenten aan het onderzoek bijgedragen bij het

researchgedeelte van hun praktikum Organische Chemie. Daarnaast wil ik graag de prettige samenwerking noemen met alle collega's in de vakgroep.

Verder dank ik dr. T. Doornbos (Unilever Chemie. Vlaardingen) voor de synthese van de hexanucleotiden in Hoofdstuk 2. en dr. J. Kanters (Vakgroep Algemene

Chemie. Rijksuniversiteit Utrecht} voor de kristalstructuuranalyse in Hoofdstuk 2. Dr.

S. van der Wal (DSM Research, Geleen) heeft de HPLC-scheidingen verricht van de gebruikte dinucleotiden. en dr. S. Wijmenga en ing. J. Joordens (NMR Faciliteit. Nijmegen) hebben assistentie verleend bij het opnemen van de tweedimensionale

NMR spectra. Dr. J. Noordik (CAOS/CAMM Centrum. Nijmegen) en dr. M. Blom­mers (Katholieke Universiteit Nijmegen) hebben hulp verleend bij het opstarten van de AMBER berekeningen. De vele tekeningen in dit proefschrift werden verzorgd door

Henk Eding.

186

Stellingen

1. De spectroscopische gegevens van fosfaat-geëthyleerd d(GpA) zijn door Kan et al. onjuist geihterpreteerd bij de toekenning van de fosforconfiguratie. en geven onvoldoende ondersteuning voor de voorgestelde linkshandige struktuur.

L.S. Kan. O.M. Cheng. S. Chandrasegaran. P. Pramanik en P.S. Miller. J. Biomol. Struct. Dyns. 4 .. 785 (1987).

2. Het verschil tussen de experimentele en de door Freier et al. berekende waarde voor de associatie-entropie van het RNA tetrameer CGCG is geheel te wijten aan het weglaten van een initiatie-entropieterm. die uit basale thermodynamische relaties af te leiden is.

S.M. Freier. R. Kierzeck. J.A. Jaeger. N. Sugimoto. M.H. Caruthers. T. Neilson en D.H. Turner. Proc. Natl. Acad. Sci. USA 83. 9373 (1986).

3. Het vermelden van uitsluitend T m-waarden voor duplex+;t enkelstrengs over­gangen van DNA geeft geen volledig beeld van de veranderingen in de stabiliteit van de duplex na eiwitcomplexatie.

B. Albert en l. Frey. J. Mol. Biol. 68. 139 (1972).

4. Het is onduidelijk. hoe de interactie tussen aminozuren en hun coderende RNA nucleotidenvolgorde volgens het model van Shimizu ka.n leiden tot de vorming van de secundaire genetische code in tRNA's.

M. Shimizu. J. Phys. Soc. Jpn. 57. 54 (1988).

5. De suggestie van Hess et al. dat het cyclobutadieen-anion een C5-geometrie bezit. bestaande uit een negatief geladen koolstofatoom en een allylisch anion. is onvoldoende gefundeerd.

B.A. Hess. Jr" C.S. Ewig en L.J. Schaad. J. Org. Chem. 50 (1985).

6. Het is onverantwoord zonder gedegen onderzoek aan te nemen dat loco of gene­rieke preparaten identiek zijn aan merkpreparaten.

Th. Rentmeester. J. Doelman. J. Huisman en R. Heijnsbroek. British-Danish­Dutch Meeting on Epilepsy. September 1988. Heemstede.

7. De term "overexpressie" van para-aminobenzoaat synthase in E. coli wordt door de experimenten van Mcleish et al. niet gerechtvaardigd.

M.J. Mcleish. P.J. Wookey en K.G. Mortimer. Bioch. Int. 16. 727 (1988).

8. De bewering van Kemp et al. dat de Co-atomen in sulfidische Co-Mo katalysa­toren bij voorkeur een tetraëdrische zwavelomringing hebben. is in strijd met recente EXAFS en XANES metingen. waaruit blijkt dat de Co-atomen groten­deels een oktaëdrische struktuur hebben.

R.A. Kemp. R.C. Ryan en J.A. Smegal. Proceedings of the 9th International Congress of Catalysis. Calgary 1. 128 (1988). S.M.A.M. Bouwens. O.C. Koningsberger. V.H.J. de Beer en R. Prins. Catal. Lett. 1. 55 (1988).

9. Het argument van Chasman et al. om in de berekening van de fotodissociatie van ozon slechts twee van de drie interne coördinaten in beschouwing te nemen vanwege "the tremendous difficulties". is niet geheel gerechtvaardigd gezien de ontwikkelingen op het gebied van semi-klassieke dynamika.

D. Chasman. D.J. Tannor en O.G. lmre. J. Chem. Phys. 89. 6667 (1988). D. Huber en E.J. Heller. J. Chem. Phys. 89. 4752 {1988).

10. In de architectuur wordt tijdens het ontwerpproces te weinig gebruik gemaakt van proefmodellen op ware grootte. waardoor het totstandgekomen gebouw dikwijls verrassingen kan opleveren.

11. Een geringe afwijking van een statistisch verwacht resultaat bij een kansspel is een realiteit. een paranormale verklaring daarentegen is een doorgaans overbo­dige fictie.

12. Hulpverlening aan "slachtoffers" van dure 06-lijnen ondermijnt het verantwoor­delijkheidsgevoel van mensen.

M.H.P. van Genderen Eindhoven. 24 februari 1989