T H E JOURNAL OF BIOLOGICAL CHEMISTRY 0 1987 by The American Society for Biochemistry and Molecular Biology, Inc.

DNase I Footprint of ABC Excinuclease”

Vol. 262, No. 27, Issue of September 25, pp. 13180-13187,1987 Printed in U.S.A.

(Received for publication, April 3, 1987)

Bennett Van HoutenSg, Howard Gamperll((**, Aziz SancarS, and John E. HearstllJJ $$ From the $Department of Biochemistry, University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, North Carolina 27514, the TDepartment of Chemistry, University of California, Berkeley, California 94720, and the ((Division of Chemical Biodynamics, Lawrence Berkeley Laboratory, Berkeley, California 94720

The incision and excision steps of nucleotide excision repair in Escherichia coli are mediated by ABC exci- nuclease, a multisubunit enzyme composed of three proteins, UvrA, UvrB, and UvrC. To determine the DNA contact sites and the binding affinity of ABC excinuclease for damaged DNA, it is necessary to en- gineer a DNA fragment uniquely modified at one nu- cleotide. We have recently reported the construction of a 40 base pair (bp) DNA fragment containing a psora- len adduct at a central TpA sequence (Van Houten, B., Gamper, H., Hearst, J. E., and Sancar, A. (1986a) J. Biol. Chem. 261, 14135-14141). Using similar meth- odology a 137-bp fragment containing a psoralen-thy- mine adduct was synthesized, and this substrate was used in DNase I-footprinting experiments with the sub- units of ABC excinuclease. It was found that the UvrA subunit binds specifically to the psoralen modified 137- bp fragment with an apparent equilibrium constant of K, = 0.7 - 1.5 X lo8 M-’, while protecting a 33-bp region surrounding the DNA adduct. The equilibrium constant for the nonspecific binding of UvrA was K , = 0.7 - 2.9 X lo5 M” (bp). In the presence of the UvrB subunit, the binding affinity of UvrA for the damaged substrate increased to K. = 1.2 - 6.7 X lo8 M” while the footprint shrunk to 19 bp. In addition the binding of the UvrA and UvrB subunits to the damaged sub- strate caused the 11th phosphodiester bond 5‘ to the psoralen-modified thymine to become hypersensitive to DNase I cleavage. These observations provide evi- dence of an alteration in the DNA conformation which occurs during the formation of the ternary UvrA. UvrB .DNA complex. The addition of the UvrC subunit to the UvrA. UvrB -DNA complex resulted in incisions on both sides of the adduct but did not cause any detectable change in the footprint.

Experiments with shorter psoralen-modified DNA fragments (20-40 bp) indicated that ABC excinuclease is capable of incising a DNA fragment extending either 3 or 1 bp beyond the normal 5’ or 3’ incision sites, respectively. These results suggest that the DNA be- yond the incision sites, while contributing to ABC ex- cinuclease-DNA complex formation, is not essential for cleavage to occur.

* This work was suppor+ed by National Institutes of Health Grants GM11180 and GM32833 and National Science Foundation Grant PCM8351212. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Recipient of National Institutes of Health Postdoctoral Fellow- ship GM11277.

** Postdoctoral fellow supported by National Institute of General Medical Sciences Grant GM11180.

$$ Support was received in part from the Department of Energy, Office of Health and Environmental Research Contract DE-ACO3- 76F0093.

In Escherichia coli, the initial steps of nucleotide excision repair are mediated by the enzyme ABC excision nuclease (ABC excinuclease) which is composed of three proteins, UvrA (Mr = 103,874), UvrB (M, = 76,118), and UvrC (M, = 66,038) (Husain et al., 1986; Arikan et al., 1986; Backendorf et al., 1986; Sancar, G. et al., 1984). These subunits function in a concerted manner to hydrolyze the 8th phosphodiester bond 5’ and the 4th or 5th phosphodiester bond 3‘ to a modified nucleotide(s). ABC excision nuclease has broad sub- strate specificity acting on UV-induced pyrimidine dimers and 6-4 photoproducts, cisplatin diadducts, N-acetoxyace- tylaminofluorene, and psoralen monadducts with the same incision motif, regardless of the size or the conformation of the modified nucleotides (Sancar and Rupp, 1983; Yeung et al., 1983; Sancar et al., 1985; Beck et al., 1985, Van Houten et al., 1986a). The notable exception to this rule is the mode of action of ABC excision nuclease on psoralen cross-linked DNA in which the dual incisions are made at the 9th and 3rd phosphodiester bonds 5‘ and 3‘ (respectively) to the thymine which is covalently attached to the furan-side of the psoralen molecule (Van Houten et al., 1986b). This highly conserved incision pattern, in addition to the broad substrate specificity of ABC excinuclease strongly suggests that the enzyme binds to a helical distortion common to most bulky DNA adducts and not to the particular modified nucleotide (Van Houten et al., 1986b).

A model for the action mechanism of ABC excinuclease has been proposed (Husain et al., 1985; Van Houten et al., 1986a). The UvrA subunit is an ATPase which uses the binding and/ or hydrolysis of ATP to facilitate specific binding to damage- induced deformities in the DNA helix. The UvrB subunit in association with the UvrA subunit forms a stable ternary complex with DNA. UvrC interacts with this complex to initiate the dual DNA incisions. Genetic and biochemical data suggest that other organisms including yeast, Drosophilin, and man, contain a nucleotide excision repair pathway analogous to that found in E. coli (reviewed by Friedberg, 1985). ABC excinuclease therefore represents a unique system for the investigation of protein-DNA interactions which occur during the process of DNA damage detection and repair.

An important step in determining the mechanism by which ABC excinuclease locates, binds to, and incises DNA sur- rounding an adducted nucleotide(s) is to determine the con- tact sites and binding affinities of the individual Uvr subunits to a damaged DNA substrate. DNase I-footprinting tech- niques have been widely exploited in determining these pa- rameters for several types of protein-DNA interactions (Galas and Schmidt, 1978; reviewed by Brenowitz, 1986). The main limitation of DNase I-footprinting experiments with DNA repair proteins has been the availability of a DNA fragment containing one DNA adduct located at a defined position. We have recently employed oligonucleotide synthesis and psora-

13180

Footprint of ABC Excinuclease 13181

len photochemistry to engineer a 40-bp' DNA fragment with a 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT) furan- side monoadduct at a central thymine (Van Houten et al., 1986b). This substrate was useful in an unequivocal determi- nation of the incision mechanism of ABC excinuclease for psoralen monoadducts and cross-links but proved to be of insufficient length for DNase I experiments. Using the same methodology we have constructed a 137-bp DNA fragment uniquely modified with psoralen and used this DNA as a substrate for ABC excinuclease binding in DNase I-footprint- ing experiments.

We report here that the UvrA protein is the DNA damage recognition subunit and that the binding of UvrB to the UvrA.DNA binary complex induces tighter binding and is accompanied by a change in the conformation of the DNA- enzyme complex. The addition of UvrC, while resulting in the dual DNA incisions, does not seem to change the binding specificity or nature of the UvrA-UvrB footprint. Additional data from experiments with short psoralen-modified DNA fragments (20-40 bp) suggest that the DNA flanking the DNase I footprint helps to stabilize the formation of the active ABC excision nuclease but is not absolutely required for DNA incisions to occur.

MATERIALS AND METHODS

DNA Oligonucleotides-Oligomers (30-60-mers) used for the con- struction of the 137-bp substrate were a gift from Applied Biosystems (Foster City, CA), while the oligomers used for modification by HMT and for the construction of the 40-bp substrate were synthesized by phosphotriester chemistry using a Biosearch instrument. Full length oligomers were purified on 7 M urea 12% or 20% polyacrylamide- sequencing gels.

HMT-modified Oligomers-The psoralen-monoadducted octamer or dodecamer were obtained as described previously (Van Houten et al., 1986a). Briefly, the 5'-phosphorylated 8-mer, TCGTAGCT (175 pg) and a complementary 5-phosphorylated 12-mer, GAAGCTAC- GAGC (250 pg) were added together in buffer (0.7 ml) containing 100 mM NaCl, 10 mM MgC12, 1.0% ethanol, and 21 pg of HMT (HRI Associates, Berkeley, CA). The two oligomers were cross-linked by exposure to 320-380 nm ultraviolet light (600 milliwatts cm-') for 3 min at 4 "C. A second aliquot of HMT was added, and the UV exposure was repeated. The reaction mixture was extracted with chloroform/isoamyl alcohol (19:l). followed by an ether extraction to remove the unreacted HMT, and the DNA was precipitated in ethanol. The two orientational isomers of the HMT cross-linked 8- mer and 12-mer were purified by electrophoresis on a 7 M urea 20% polyacrylamide gel. To obtain the furan-side-monoadducted octamer or dodecamer, the cross-linked oligomers were partially photoreversed by exposure to a Sylvania model G30T8 germicidal UV lamp (254 nm). The monoadducted oligomers were purified by gel electropho- resis as described above.

Preparation of the 40- and 137-bp DNA Substrates-The purified psoralen furan-side-monoadducted octamer or dodecamer was ligated with a series of complementary oligomers to obtain the 40- or 137-bp DNA fragments (respectively) shown in Fig. 1. The end labeling and ligation reactions were as reported previously (Van Houten et al., 1986a). Full length DNA duplexes were obtained by reannealing the appropriate length single strands (138 or 41 bases) which had been purified on denaturing gels. This purification step was absolutely required to obtain DNA molecules that were completely double stranded. The identity and integrity of the full length duplexed DNA were checked by restriction enzyme digestions and Maxam-Gilbert sequencing (1980). A routine yield of fully duplexed substrate was 3- 6 pg from a total of 36 pg of component oligomers. The specific activity of the DNA fragments used in these experiments was 0.5 - 5.0 X lo7 cpmlpg.

DNase I Footprinting of the Uvr Subunits-ABC excinuclease reactions were routinely performed in 50-rl reactions containing 50 mM KCI, 50 mM Tris-HC1, pH 7.5, 10 mM MgCl,, 10 mM dithio-

The abbreviations used are: bp, base pairs; HMT, 4'-hydroxy- methyl-4,5',8-trimethylpsoralen.

threitol, 2 mM ATP, and 100 pg/ml of bovine serum albumin ( D C buffer).

Maximum subunit binding and incision were obtained when the subunits (0.1-30 pmol) were preincubated for 5 min at 37 "C prior to the addition of substrate DNA (0.1-0.3 pmol). After the subunits were allowed to equilibrate with the DNA for 30 min at 37 "C, the reaction mixtures were cooled to room temperature and were made 2.5 mM in CaCl,. DNase I (0.5 ng, Bethesda Research Laboratories) was added, and after 5 min at room temperature the reactions were stopped by the addition of EDTA to a final concentration of 15 mM. After rapidly freezing, the mixtures were lyophilized to dryness, 50 pl of formamide-plus-dyes were added, and the samples were heated at 90 "C for 2 min followed by quick cooling on ice. Portions of the samples (2-5 pl) were applied to an 8% polyacrylamide-sequencing gel and electrophoresed for 2.5 h at 1200 V and 25 milliamps. The gel was dried and was exposed to x-ray film (Kodak, GB-X2 film) overnight a t -70 "C with an intensifying screen.

Determination of the Binding Affinities-The quantitation by op- tical density of individual DNase I bands was performed by densito- metric scanning of appropriately exposed autoradiograms. The auto- radiograms were scanned on an Optronics P-1000 film scanner with an AED graphics terminal. Data analysis was performed on a VACS 11/730 computer using an algorithim, Gelscan, developed by Frank Hage of the Protein Crystallography Facility at the University of North Carolina. This program is similar to the one recently described by Brenowitz et al. (1986). The percent saturation of UvrA binding was determined from the relative intensity of the integrated optical density for the bands of interest. The protein concentration which reduced the band intensity to 50% was taken as the K d -

Incision Efficiency on Minimal Length Substrates-Terminally la- beled 40-bp substrate or gel-purified restriction fragments of this substrate containing the psoralen-modified thymine (0.2-0.5 pmol) were treated with UvrA (5.0 pmol), UvrB (10 pmol), and UvrC (10 pmol) in ABC buffer for 30 min at 37 'C. The reaction mixtures were frozen, lyophilized, resuspended in formamide-plus-dyes and loaded onto 12% polyacrylamide-sequencing gels. The intact and the ABC excision nuclease-generated bands were located by autoradiography and excised from the gel. The amount of DNA in each band was determined by Cerenkov radiation, and the incision efficiency of ABC excinuclease was determined from the ratio of incised DNA compared to the total.

RESULTS

Design and Preparation of 137-bp HMT-modified Frag- ment-The construction of a DNA fragment containing a modified nucleotide at a defined position was essential for DNase I-footprinting experiments with ABC excinuclease. Psoralens exhibit a marked preference for photoreaction with DNA containing a TA sequence, and as reported previously, we have used this property of psoralens in the preparation of oligonucleotides modified at a central thymine with the furan- side monoadduct of HMT (Van Houten et al., 1986a). These oligonucleotides were used in ligation reactions to construct uniquely modified 40- and 137-bp DNA fragments (Fig. 1). Previous DNase I-footprinting experiments with ABC exci- nuclease and the 40-bp DNA fragment indicated that this substrate was of insufficient length to determine the contact sites of the Uvr subunits, and we therefore designed and constructed the 137-bp fragment. Restriction fragments of the 40-bp substrate were used to determine the minimum length substrate for ABC excinuclease digestion.

Psoralen-induced DNA Conformational Changes-DNase I has been used as a probe for conformational changes in the DNA helix which occur during the binding of certain drugs to DNA (Drew and Travers, 1985). The width of the minor groove is believed to be an important determinant in the selection of a cleavage site by DNase I, with an optimal width of 12 A (Drew and Travers, 1984). The binding of echinomycin and distamycin to DNA are known to alter the width of the minor groove and in turn have been shown to affect the DNase I cleavage pattern for certain sequences (Low et al., 1986). Modification of the thymine at position 74 by the

13182 Footprint of ABC Excinuclease A )

I IO 20 30 .O 5 0 6 0 7 0 BO 90 100 ,IO 120 130 I . . . . . . . . . l , . . l . . 1 . . ? . . . . . . . .

S ' - C C C I \ T C C C C C T C ~ G I \ C M T T M T C A T C C G C T C C T A T A A T C T C T C C A A ~ C C A C A C T C A G T ~ C G ~ ~ C C ~ G C ~ A C G ~ C C T C ~ C C C C ~ T C C ~ ~ ~ C G T C T ~ C A T C T C C C ~ A T A G T C ~ G T C C T ~ T T A A T T T C C ~ C C ~ - 3 '

3 ' - ~ G ~ ~ ~ ~ ~ ~ ~ ~ ~ C T G T T A A ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ G C A T A ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ G ~ C T C ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ T T C ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ T ~ C ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ - ~ ~ ~ ~ ~ ~ ~ . 5 , A A

81

1 10 20 30 4 0 . . . . v . . . .

A A 5 ' - CTATCCATGGCCTCCAGTCGTACCTGM~CCTACTGAGTC - 3 ' 3 ' - ATAGCTACCGGACGTCAG~~TCGAC=~CCI \TG*CTCAC~ - 5 '

FIG. 1. Nucleotide sequences of the psoralen-monoadducted DNA substrates. The DNA substrates were constructed from several component oligomers. The sites of ligation are indicated by arrows. The central top strand oligomer was modified with the furan-side monoadduct of HMT at an internal thymine, so that in both substrates the top strand of the full length duplex is always the psoralen-modified strand. The DNA fragments contained multiple restriction enzymes sites and 5' overhanging ends. A, the 137-bp DNA fragment composed of eight oligomers; the HMT-modified thymine is indicated by a filled circle at position 74. E , the 40-bp DNA fragment composed of six oligomers; the HMT-modified thymine is indicated by a filled circle at position 20.

furan-side monoadduct of HMT produced a significant change in the DNase I digestion pattern in the vicinity of the DNA adduct (compare positions 70-80, lanes 1 and 8 in Fig. 2, A and B). The most striking change was the complete disap- pearance of the DNase I cleavage site between C,, and T74 (band II, lanes 1 and 8, Fig. 2 A ) in the top strand. This particular site was predicted by Pearlman et al., (1985) to be unwound by -7.5" as compared to normal B-DNA, and could be expected to change the minor groove width making this an unfavorable cleavage site for DNase I. In addition the HMT- modified thymine also caused a reduction in the cleavage of the phosphate bonds at positions G73'-A74' and C77'-T7s' and an enhancement in DNase I cleavage at T75'-G76' in the bottom strand (bands I1 + 111, respectively, of lanes 1 and 8, Fig. 2B).

DNase I Footprint Titrations of UvrA-The UvrA subunit of ABC excinuclease is known to bind to damaged DNA preferentially and is therefore assumed to be the recognition subunit of the enzyme. (Seeberg and Steinum, 1982; Husain et al., 1985). To investigate the interactions of this subunit with DNA, the effect of UvrA concentration on the DNase I digestion pattern for the HMT-modified and nonmodified 137-bp DNA fragments was examined. The results of such an experiment are displayed in the autoradiograms in Fig. 2. The UvrA protein protected a 33-bp region, which is bracketed by arrows. For consistency we have marked these borders the same in all the figures, although some differences in the length of the footprints were seen from experiment to experiment, due to the partial protection observed at the borders of the footprint. At low concentrations (9.6-96 nM), the UvrA sub- unit bound specifically to the furan-side HMT-monoadducted DNA in a concentration-dependent manner, while at higher concentrations (100-300 nM) nonspecific binding occurred. The nonspecific binding resulted in a general disappearance of DNase I bands in both the modified and nonmodified DNA (band IV, lanes 6, 7, 13, 14, Fig. 2 A ) . UvrA binding was analyzed quantitatively by densitometric scanning of partic- ular bands (marked by Roman numerals, see Fig. 2 legend) from appropriately exposed autoradiograms. These UvrA sat- uration curves were in turn used to determine the equilibrium constants for specific and nonspecific binding. The results of such an analysis for the top strand are shown in Fig. 2C. A similar analysis for the bottom strand gave essentially the same results (data not shown), although the values were slightly higher due to the clearer footprint that was consist- ently obtained for the bottom strand of the psoralen-modified substrate. We therefore give a range of values for specific and nonspecific binding constants as K. = 0.7 - 1.5 X 10' M-',

and K, = 0.7 - 2.9 X lo6 M" bp, respectively. Based on these equilibrium constants an estimated 4.3 kcal increase in free energy is associated with the specific binding of UvrA to the substrate. Kinetic studies have suggested that ABC excinu- clease consists of one of each subunit (Husain et aZ., 1985), while recent evidence from DNA unwinding experiments (Oh et al., 1986) and hydrodynamic studies2 suggest that the active form of UvrA may be a dimer at the concentrations used in these experiments. Since the exact composition of the active form of the ABC excinuclease is not known we have reported the equilibrium constants of UvrA binding in terms of con- centration of UvrA as a monomer.

The footprinting experiments in Fig. 2 revealed two addi- tional features of Uvr A binding: 1) there was an apparent increase in the region protected with increasing Uvr A con- centration, 2) the footprints of the two strands were consist- ently different in that UvrA completely protected the bottom (nonadducted) strand while the top (adducted) strand was only partially protected at equivalent protein and DNA con- centrations. No estimate of the stoichiometry for UvrA spe- cific binding could be made due to the relatively low ratio of specific to nonspecific binding of UvrA.

The Effect of UvrB Binding on the UvrA Footprint-UvrB has little or no binding affinity for DNA in the absence of UvrA, but binds tightly to the UvrA.DNA binary complex. (Kacinski and Rupp, 1981; Yeung et al., 1983; Husain et al., 1985; Yeung et al., 1986). Having obtained the UvrA footprint, we next wanted to determine what effect the UvrB subunit had on the binding of UvrA to the HMT-modified substrate. The results of these experiments are shown in Fig. 3. There were three important effects of UvrB binding. First, the presence of UvrB induced an apparent conformational change in the UvrAeDNA complex such that binding of the UvrA and UvrB subunits to the psoralen-modified DNA caused the 11th phosphodiester bond 5' to the HMT-adducted thymine (between G,, and &) to become hypersensitive to DNase I cleavage (Band ZZI, lanes 8-14, Fig. 3A). Second, this hyper- sensitivity was accompanied by an apparent decrease in the size of the UvrA footprint from 33 to 19 bp. Third, the addition of UvrB increased the affinity of UvrA severalfold for the psoralen-modified DNA; this is depicted by the change in saturation curves (Fig. 3C), from which an equilibrium con- stant was derived. As with UvrA alone we observed more efficient binding of the UvrA.UvrB complex to the bottom strand of the psoralen-adducted fragment, and we therefore report the equilibrium binding constant as a range with K, = 1.2 - 6.7 x 10' M-'. It should be noted, however, that the

D. K. Orren and A. Sancar, unpublished results.

Footprint of ABC Excinuclease 13183

A TOP STRAND

r uvr A

IP

$ 3 - 90 !-

.. I .- " " . "_. < -

" BOTTOM STRAND 1 2 3 4 5 6 7 A G T C 8 9 1 0 1 1 1 2 1 3 1 4

I - U -m

-E

-I

"0 -D - m

TOP STRAND 1 2 3 4 5 6 7 8 9 1 o l 1 I 2 1 3 l 4 1 5 1 6

Uvr A

B

UW

L

l 70 4o ' 3 0 110 I40 160 !BO &

U v r A COYCENTRATION lnM1

FIG. 2. DNase I footprint of the UvrA subunit. In both panels A and R, [ones 1-7 contained HMT-modified DNA and lanes 8-14 contained unmodified DNA. Lunes A, G, 7'. and C refer to the Maxam- Gilbert chemical sequencing reactions for A + G, G , T + C, and C, respectively. The 137-bp DNA (2.2 nM) was digested with DNase I in the presence of various amounts of the UvrA subunit. Lunes 1-7, and 8-14 contained 0.0, 9.6, 19.2, 38.4, 96.2, 192, or 384 nM of UvrA, respectively. The region protected from DNase I digestion by the UvrA subunit is bracketed by arrows. A, DNase I footprint of the UvrA subunit bound to the 137-bp DNA fragment with a 5' terminally labeled top strand (HMT-modified strand). The position of the HMT- modified thymine is shown as a circled 7'. The bands used for densiometric scanning of the autoradiogram are indicated by Roman numerals. Note the complete absence of band I1 in the psoralen- modified DNA (compare h e I to lane 8). B, DNase I footprint of the UvrA subunit on the 137-bp DNA fragment containing a 5' terminally labeled bottom strand (nonmodified strand). The position of the adenine complementary to the HMT-modified thymine is shown as a circled A. R a d s I-IV were used for densiometric scanning. Note the differences in intensity of bandy I I and 111 in the modified DNA ( l a n e I ) compared to unmodified DNA (lane 8). C, data derived from scanning the autoradiogram in Fig. 2A and from similar exper- iments were used to determine the saturation curves for the binding of the UvrA subunit to psoralen-modified and unmodified DNA.

E

4

0 4 I? '7 '6 2<! ?4 28 32 36 40

Uvr 4 CONCEYTRATION InMI

FIG. 3. Effect of the UvrB subunit on the binding specificity of the UvrA subunit to psoralen-modified DNA. 5' terminally labeled top strand ( A ) or bottom strand ( R ) 137-bp HMT-modified DNA (2.2 nM) was digested with DNase I in the presence of 0.0, 2.4, 4.8, 6.4, 9.6, 12.8, 38.4, or 96.2 nM UvrA (lanes 1-8, respectively) or together with 5.5 nM UvrB (lanes 9-16). The region of DNase I protection by the UvrA subunit or the UvrA and UvrB subunits is bracketed by arrows. Bands marked by Roman numerals were scanned and used to determine the saturation binding curves. C, data from scanning the autoradiogram in Fig. 3A and from several other experiments were used to plot the UvrA saturation curves. U, band II , lanes 1-8 U, b a d II , [ones 9-16; A-A, band I l l , lanes 9-16.

DNA concentration used in these experiments was only 2.5- fold lower than the UvrB concentration so that the equilib- rium constants given here are therefore an underestimate of the true value. Since the appearance of the DNase I hyper- sensitive band was diagnostic for binding of both the UvrA and UvrB subunits to the HMT-modified substrate, the emer- gence of this band was monitored a t saturating amounts of UvrA and increasing amounts of UvrB. In this way the

M, mean intensity of bands I + 111 of psoralen-modified DNA (lanes 1-7); A-A, band I of unmodified DNA (lanes 8-14); U, means of band IV for both the modified and unmodified DNA (lanes 1-7 and 8-19).

~ ~~

13184 Footprint of ABC Excinuclease

concentration of UvrB which was necessary to promote max- imum binding could be estimated. The results of such an experiment are shown in Fig. 4. Maximum binding appeared to occur at a concentration of UvrB which was equal to that of the DNA. Incision experiments in the presence of all three subunits yielded essentially the same results (data not shown). This suggests that the active form of the UvrA and UvrB complex contains one UvrB molecule.

Effect of the UvrC Subunit-Addition of the UvrC subunit to the UvrA .UvrB .DNA ternary complex results in DNA incisions at the 8th and the 5th phosphodiester bonds 5' and 3' (respectively) to the psoralen-modified thymine (Van Hou- ten et al., 1986a). Because of these dual incisions, it was not possible to observe an effect of the UvrC subunit on the UvrA- UvrB footprint on the damaged (top) strand. Surprisingly, even on the nondamaged strand, UvrC, over a wide range of concentrations, had no detectable effect on the size or quality of the DNase I footprint (Fig. 5). UvrC binds to both double and single strand DNA, with the latter type of binding being more efficient (Sancar et al., 1981). In DNase-footprinting experiments UvrC had no effect on the affinity of the UvrA or the UvrA and UvrB subunits to the psoralen-monadducted DNA (data not shown) and UvrC alone exhibited no specific or nonspecific binding to DNA (Figs. 5 and 6, A, lane 4, and B, lane 7 ) .

Summary of DNase I Footprints during the Assembly of A BC Excision Nuclease-The representative footprints for the assembly of the active form of the enzyme are shown in Fig. 6. At the concentrations used in these experiments (2-20 nm), there is little or no binding of any of the three subunits to the unmodified 137-bp DNA fragment (lanes 2-4, Fig. 6A). Furthermore, when UvrB or UvrC were added separately no detectable binding was observed (Fig. 6B, lanes 6 + 7, respec- tively).

UvrA binds to the psoralen-modified DNA fragment pro-

TOP STRAND 1 2 3 4 5 6 7 8

80

0 7c

6C

5c

1 Uvr A

JB Uvr A

1

FIG. 4. Stoichiometric binding of the UvrB subunit to UvrA.DNA binary complexes. 5' terminally labeled top strand 137-bp HMT-modified DNA (4.4 nM) was digested with DNase I either alone (lanes I ) or in the presence of 96.2 nM UvrA together with 0.0, 1.3, 1.8, 2.6, 5.3, 10.6, or 25.9 nM UvrB (lanes 2-8). The addition of UvrB produced a hypersensitive DNase I cleavage site, which resulted in the appearance of a new DNA band which is marked by I. The intensity of this band was used as an indicator for UvrB binding to the binary UvrA.DNA complex. The region of DNase I protection by the UvrA subunit or the UvrA and UvrB subunits is bracketed by arrows.

BOTTOM STRAND 1 2 3 4 5 6 7 8 -

5c

6 C

70'- - @" so. - 7 1 3 c'

a - "C" . _

Uvr A

9dSo ---"-- I - - - . . " " . _ _ - -

"""0.

- - - . ""- loo'.

FIG. 5. Effect of UvrC subunit on the UvrA-UvrB footprint. DNase I digestion of 5' terminally labeled bottom strand HMT- modified 137-bp DNA. Lane 1, DNase I alone; lane 2, DNase I plus 4.8 nM UvrA, lanes 3-8, DNase I plus 4.8 nM UvrA, 5.3 nM UvrB, and 0.0, 10, 15, 20, 28, and 49 nM UvrC, respectively.

tecting a 33-bp region in manner which is not dependent upon, but stimulated (5-10-fold) by the presence of ATP (Fig. 6A, lane 8, on 6B, lane 2, and data not shown). There is a characteristic decrease in the length of the DNase I footprint from 33 to 19 bp when both the UvrA and UvrB subunits bind to the psoralen-modified DNA (Fig. 6, A, lanes 7 and 8 and B, lanes 2 and 3) . The addition of all three subunits to the psoralen-modified 137-bp fragment resulted in dual DNA incisions a t positions Twi and AT,, and can be seen as two bands marked by the Roman numerals I1 and 111 (Fig. 6A, lane 10). Note that when the substrate was labeled on the 5' terminus of the strand, the band corresponding to incision on the 3' side of the thymine adduct could only be observed if 3' incision occurred in the absence of 5' incision. The ability to discern the 3' incision site on 5"labeled DNA indicates that ABC excinuclease occasionally cleaves on the 3' side of a DNA adduct in the absence of a 5' incision. We have found that both pH and ionic strength of the buffer can cause uncoupling of the 3' and 5' incisions.:'

We also observed a low but detectable level of anomalous DNA incisions in the nonadducted strand (lane 5, Fig. 6, A and B). Longer exposures of these gels revealed that these incisions were consistent with ABC excinuclease-mediated DNA cleavage of the 8th phosphodiester bond 5' to guanines. I t is known that the phosphotriester method of oligonucleo- tide synthesis can lead to damaged guanines a t a low fre- quency, and we therefore concluded that these bands were due to incision of the damaged guanines by ABC excinuclease.

The results of the DNase I-footprinting experiments are shown schematically in Fig. 7. The DNA incision sites (shown by heavy arrows) prevented an analysis of the effect of UvrC on the top strand footprint. Analysis of the bottom strand footprints did not reveal any differences in the UvrA-UvrB and UvrA-UvrB-UvrC footprints. The results summarized in this figure indicated that ABC excinuclease contacted a rela- tively small region around the adducted nucleotide. To sup- port this conclusion we conducted the experiments described below.

Minimal Length Substrate-We also studied the interaction

B. Van Houten and A. Sancar, unpublished results.

Footprint of ABC Excinuclease 13185 TOP STRAND

YY). 90. 80'

&

@OTTOM STRAND

" "

Uvr A

FIG. 6. DNase I footprints of the UvrA subunit, the UvrA and UvrB subunits, and the complete ABC excinuclease. A, 5' terminally laheled top strand nonmodified (lanes 1-5) or HMT- modified (lanes 6-10) 13'7-bp DNA were treated with DNase I (lanes 1-4 and 6-9) in the presence of the subunits of ABC excinuclease. Lunes 1 and 6, DNase I alone; lanes 2 and 7,19 nM UvrA; lanes 3 and 8, 19 nM UvrA and 5 nM UvrB; lanes 4, 5, 9, and 10, 19 nM UvrA, 5 nM UvrB, and 20 nM UvrC. B, 5' terminally labeled bottom strand psoralen modified 137-bp DNA was digested with DNase I (lanes 1- 4, 6, and 7) in the presence of various amounts of the Uvr subunits. Lune 1, DNase I alone; lane 2, 19 nM UvrA; lane 3, 19 nM UvrA and 5.3 nM UvrB; lane 4, 19 nM UvrA, 5.3 nM UvrB, and 28 nM UvrC; lane 5 , as in lane 4 but in the absence of DNase I digestion; lane 6, 5.3 nM UvrB; lane 7,49 nM UvrC; lane R,S'-labeled top strand HMT- modified DNA treated as in lane 5. Lanes A, G, T, and C refer to the Maxam-Gilbert sequencing reactions for A + G , G , T + C, and C, respectively. The region of DNase I protection by the UvrA subunit or the UvrA and UvrB subunits is bracketed by arrows. The addition of the UvrC subunit to the UvrA and UvrB subunits produced a footprint that was not different from that seen with UvrA and UvrB.

of ABC excinuclease with DNA using the 40-bp substrate. We had engineered several convenient restriction sites into a 40-bp DNA fragment for two reasons. First, we wanted to determine if the binding of the Uvr subunits to the modified DNA fragment would protect the DNA fragment from cleav- age by a particular restriction enzyme. Secondly, by digesting the 40-bp fragment into shorter length fragments we hoped to delineate the minimal contact sites which are necessary to

- FIG. 7. Summary of the DNase I footprint of ABC excinu-

clease during various stages of assembly. For clarity only the central 55 bp of the 137-bp psoralen-modified DNA fragment are displayed. The psoralen-adducted thymine is circled. The boxes above and below the sequence indicate the minimum area of protection from DNase I digestion. The heauy arrows indicate the positions of the phosphodiester bonds which are cleaved by ABC excinuclease. The light arrow indicates the position of the phosphodiester bond which becomes hypersensitive to DNase I cleavage upon the forma- tion of the UvrA.UvrB.DNA ternary complex.

I . .

Hae III EcoRI DdeI Hinf I

percent incision: 2.96 015 460 6.30

FIG. 8. Efficiency of ABC excinuclease incisions on psora- len-modified DNA substrates of various lengths. Terminally labeled top strand 40-bp HMT-modified DNA was digested with the indicated restriction enzymes, and the fragments containing the HMT-modified thymine were purified on polyacrylamide gels. The purified restriction fragments were digested with ABC excision nu- clease and a portion of these reactions were loaded onto 12% polya- crylamide-sequencing gels. The extent of ABC excinuclease incision was determined by scintillation counting of the gel slice containing the appropriate DNA band. The percent incision given under the restriction enzyme indicates the extent of 3' incision by ABC exci- nuclease for a HaeIII-digested fragment and the extent of 5' incision for either the EcoRI; DdeI- or HinfI-digested fragments normalized for the incision observed with ARC excision nuclease on the 40-bp DNA fragment.

promote productive DNA binding and allow DNA cleavage to occur. Restriction enzyme protection experiments with both the UvrA and UvrB subunits bound to the 40-bp substrate were consistent with the binding domains identified by DNase I protection experiments with the 138-bp DNA fragment. Namely, the restriction enzymes, HinfI and HaeIII were par- tially inhibited by UvrA and UvrB binding, whereas DdeI was completely inhibited (data not shown). The positions of the restriction enzyme sites are shown in Fig. 8. The HaeIII site is 9-bp 5' and the HinfI site is 17-bp 3' to the HMT-adducted thymine; both sites are outside the regions protected against DNase I digestion by the UvrA and UvrB subunits. By con- trast the DdeI site is only 14 bp 3' to the modified thymine, just at the border of the UvrA-UvrB footprint. EcoRI diges- tions were partially inhibited by the psoralen-thymine adduct a t position 20 and were not suitable for analysis.

We next wanted to determine if the regions of DNA beyond the areas detected by DNase I footprinting contributed to the overall stability of the complex formation and could be re- flected in the extent of DNA incision. The terminally labeled 40-bp substrate was prepared as described under "Materials and Methods," digested with the appropriate restriction en- zyme and the top strand-labeled fragment containing the HMT-modified thymine gel purified. These labeled fragments were then digested with ABC excinuclease and the extent of 5' or 3' incision was quantitated by determining the radio-

13186 Footprint of ABC Excinuclease

activity in the gel slice containing the intact and the ABC excinuclease-generated incision bands. As shown in Fig. 8, removal of 9 bp from the 5' end by HaeIII digestion resulted in a decrease in the 3' incision efficiency to 3.0%. Similarly, digestion with HinfI, DdeI, or EcoRI reduced the 5' incision efficiency to 6.3, 4.6, and 0.15%, respectively. These results suggest that regions of DNA beyond the incision sites greatly stabilize the formation of the active ABC excision nuclease complex. Increasing the length of the substrate from 40 to 137 bp increased the overall incision efficiency from 30-40% to 50-60%.

DISCUSSION

We have used DNase I footprinting to study the assembly of ABC excinuclease at the site of a DNA adduct. The infer- ences drawn from these results, when viewed in the context of previously published experiments help to formulate a mech- anism of how ABC excinuclease selectively binds to and cleaves a DNA substrate containing a damaged nucleotide.

The UvrA subunit binds specifically to damaged DNA in an ATP-stimulated reaction (Seeberg and Steinum, 1982) protecting a 33-bp region from DNase I digestion with an equilibrium constant of K, = 0.7 - 1.5 X 10' M-'. An estimate of the nonspecific binding affinity was found to be K, = 0.7 - 2.9 X IO5 M-' (bp) such that the UvrA subunit bound to the damaged DNA with a specificity ratio of lo3 compared to nondamaged DNA. Another aspect of the DNase I experi- ments presented here is that protection obtained on the nonadducted (bottom) strand was consistently clearer than that observed for the adducted (top) strand. This inequality in the footprints might be indicative of a more intimate contact of UvrA with the nonadducted strand as compared to the adducted strand. A similar pattern has been observed for the binding of T7 RNA polymerase to the 410 promoter of T7 phage. The binding of T7 RNA polymerase to this pro- moter was shown to protect only the template strand from digestion by DNase I (Basu and Mitra, 1986).

It has been shown that UvrB alone does not bind to modi- fied DNA but forms a ternary complex with UvrA and DNA (Kacinski and Rupp, 1981; Yeung et al., 1983; Husain et al., 1985; Yeung et al., 1986). The addition of the UvrB subunit caused the following important changes in the mode of Uvr A binding: (i) the size of the footprint decreased to 19 bp; (ii) the binding affinity increased to 1.2 - 6.7 X 10' K'; (iii) a DNase I hypersensitive band appeared 11 phosphodiester bonds 5' to the modified thymine (Fig. 3, A and B) . These alterations indicate that the UvrA. UvrB . DNA complex is significantly different than the UvrA. DNA complex. Al- though kinetic experiments suggest that the active complex consists of one of each subunit (Husain et al., 1985), hydro- dynamic studies2 and DNA unwinding studies (Oh et al., 1986) with UvrA suggest that its active form may be a dimer at the concentrations of proteins used in these experiments (2-40 nM). The decrease in the size of the UvrA footprint with the addition of UvrB might indicate a displacement of one of the UvrA monomers, which is then free to participate in another round of substrate binding. Alternatively, UvrA binding might be imprecise, so that in a population of DNA molecules the overall footprint might appear larger. In this case addition of UvrB serves to promote binding of UvrA in a precise manner.

Yeung et al., (1986) have shown that the addition of UvrB stabilizes the binding of UvrA to damaged substrate. This enhanced stability was reflected in the increased binding affinity of the UvrA and UvrB subunits to the 137-bp sub- strate with a K, = 1.2 - 6.7 X 10' "'. UvrB is known to stimulate the ATPase activity of UvrA in the presence of

DNA and more so in the presence of damaged DNA (Thomas et al., 1985). The function of ATP hydrolysis is not well understood at present. Several proteins, including catabolite activator protein and RNA polymerase, have been shown to undergo a characteristic change in their interaction with DNA upon binding of nucleotide substrates or cofactors (Carpousis and Gralla, 1985; Straney and Crothers, 1985; McClure, 1985; Liu-Johnson et al., 1986). With these proteins, ligand binding promotes a shift from a less to a more specific DNA binding mode for a particular sequence.

The hypersensitive site created by UvrB binding was noted previously with the 40-bp substrate (Van Houten et al., 1986a). There the 10th and 11th phosphodiester bonds 5' to the psoralen-modified thymine were sensitized to DNase I cleavage. The appearance of DNase I hypersensitive sites have also been observed for the binding of the UvrA and UvrB subunits to a DNA fragment containing a pyrimidine dimer4 and may be responsible for the unique phenotype of the uurC-mutants (Van Houten et al., 1986).

The UvrC subunit, at the concentrations used in this study, displayed no affinity for the damaged DNA substrate and did not affect the UvrA or the UvrA and UvrB binding as deter- mined by DNase I protection experiments. Previous studies suggested that following the concerted DNA incision the three subunits stay bound at the site of the DNA damage; the dual action of helicase I1 and Pol I are needed to displace the subunits and allow them to act catalytically (Husain et al., 1985; Caron et al., 1985). The footprint data are consistent with the UvrA and UvrB subunits staying bound to the site after DNA cleavage, although no conclusion can be reached concerning the fate of the UvrC subunit.

ABC excinuclease acts on many dissimilar DNA adducts with essentially the same incision mechanism. What then is the recognition signal for the specific binding of ABC excision nuclease? Pearlman et al., (1985), have suggested that three common pertubations associated with bulky DNA damage are local DNA unwinding (20-90") helical bending or kinking, (25-45"), and displacement of the two helical axes relative to one another (2.5-3.5 A). Based on the asymmetric incision preference of ABC excinuclease for the furan-adducted strand of psoralen-cross-linked DNA as a model, we have previously suggested that ABC excinuclease binds to only one face of the DNA helix at the damage-induced kink (Van Houten et al., 1986b). This model is further supported by incision studies performed on UV-irradiated DNA with ABC excinuclease in the presence of DNA photolyase: photolyase binds to the damaged strand (Husain et al., 1987) and stimulates the binding of ABC excinuclease (Sancar et al., 1984; Myles et aL, 1987). The differences in the protection of the adducted and nonadducted strands by ABC excision nuclease reported here lends further support to this model. Verification of the model awaits the determination of more precise contact sites of the Uvr subunits on damaged DNA by conducting chemical foot- printing experiments.

Acknowledgments-We thank Drs. Gwen Sancar, Intisar Husain, and Gary Myles for their helpful discussions and critical reading of this manuscript. We also thank Suzanne Cheng for help in prepara- tion of the short oligonucleotides and Applied Biosystems for their gift of the long (30-60-mers) oligonucleotides.

REFERENCES Arikan, E., Kulkarini, M. S., Thomas, D. C., and Sancar, A. (1986)

Backendorf, C., Spaink, H., Barbeiro, A. P., and van de Putte, P.

B. Van Houten, 1. Husain, and A. Sancar, unpublished observa-

Nucleic Acid Res. 14,2637-2650

tions.

Footprint of ABC Excinuclease 13187

(1986) Nucleic Acid Res. 14, 2877-2890 Basu, S., and Maitra, U. (1986) J. Mol. Biol. 190 , 425-437 Beck, D. J., Popoff, S., Sancar, A., and Rupp, W. D. (1985) Nucleic

Brenowitz, M., Senear, D. F., Shea, M. A., and Ackers, G. K. (1986)

Carpousis, A. J., and Gralla, J. D. (1985) J. Mol. Biol. 183, 165-177 Caron, P. R., Kushner, S. R., and Grossman, L. (1985) Proc. Natl.

Drew, H. R., and Travers, A. A. (1984) Cell 37, 491-502 Drew, H. R., and Travers, A. A. (1985) Nucleic Acids Res. 13, 4445-

Friedberg, E. C. (1985) DNA Repair, Freeman Publications, San

Galas, D. J., and Schmitz, A. (1978) Nucleic Acids Res. 5, 3157-3170 Husain, I., Van Houten, B., Thomas, D. C., Abdel-Monem, M., and

Sancar, A. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 6774-6778 Husain, I., Van Houten, B., Thomas, D. C., and Sancar, A. (1986) J.

Biol. Chern. 261,4895-4901 Husain, I., Sancar, G., Holbrook, S. H., and Sancar, A. (1987) J. Biol.

Chem. 262, 13188-13197 Kacinski, B. M., and Rupp, W. D. (1981) Nature 294,480-481

Acids Res. 13, 7395-7412

Methods Enzyrnol. 130 , 132-181

Acad. Sci. U. S. A. 82,4925-4929

4467

Francisco

Liu-Johnson, H.-N., Gartenberg, M. R., and Crothers, D. M. (1986) Cell 47,995-1005

Low, C. M. L., Drew, H. R., and Waring, M. J. (1986) Nucleic Acids

Maxam, A. M., and Gilbert, W. (1980) Methods Enzymol. 6 5 , 499- Res. 14, 6785-6801

560

McClure, W. R. (1985) Annu. Reu. Biochern. 5 4 , 171-204 Myles, G. M., Van Houten, B., and Sancar, A. (1987) Nucleic Acids

Oh, E. Y., and Grossman (1986) Nucleic Acids Res. 14, 8557-8571 Pearlman, D. A., Holbrook, S. R., Pirkle, D. H., and Kim, S.-H.

Sancar, A., Kacinski, B. M., Mott, D. L., and Rupp, W. D. (1981)

Sancar, A., and Rupp, W. D. (1983) Cell 33 , 249-260 Sancar, A., Franklin, K. A., and Sancar, G. (1984) Proc. Natl. Acad.

Sancar, A., Franklin, K. A., Sancar, G., and Tang, M. (1985) J. Mol.

Sancar, G. B., Sancar, A., and Rupp, W. D. (1984) Nucleic Acids Res.

Seeberg, E., and Steinum, A.-L. (1982) Proc. Natl. Acad. Sci. U. S. A.

Straney, D. C., and Crothers, D. M. (1985) Cell 43,449-459 Thomas, D. C., Levy, M., and Sancar, A. (1985) J. Biol. Chern. 260,

9875-9883 Van Houten, B., Gamper, H., Hearst, J. E., and Sancar, A. (1986a)

J. Biol. Chern. 261 , 14135-14141 Van Houten, B., Gamper, H., Holbrook, S. R., Hearst, J. E., and

Sancar, A. (1986b) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 8077-8081 Yeung, A. T., Mattes, W. B., Oh, E. Y., and Grossman, L. (1983)

Proc. Natl. Acad. Sci. U. S. A. 80, 6157-6161 Yeung, A. T., Mattes, W. B., and Grossman, L. (1986) Nucleic Acids

Res. 14.2567-2583

Res. 15,1227-1243

(1985) Science 227 , 1304-1308

Proc. Natl. Acad. Sci. U. S. A. 78,5450-5454

Sci. U. S. A. 81, 7397-7401

Biol. 184, 725-734

12,4593-4608

79,988-992