Indian Journal of Biochemistry & Biophysics Vol. 38, February & April 200 I, pp. 34-4 1

Enhancement of catalytic activity of enzymes by heating in anhydrous organic solvents: 3D structure of a modified serine proteinase at high resolution

Sujata Sharma' , Renu Tyagi 2, M N Gupta2 and T P Singh' -

'Department of Biophysics, All India Institute of Medical Sciences, New Delhi 110 029, India

2Department of Chemistry, Indian Institute of Technology, New Delhi 11001 6, India

Accepted 3 October 2000

For the first time, it is demonstrated that exposure of an enzy me to anhydrous orga ni c solvents at optimized high temperature enhances its catalytic power through local changes at the binding region. Six enzymes, namely, proteinase K, wheat ge rm acid phosphatase, a-alllylase, l3-g lucosidase, chymotrypsin and trypsi n were exposed to aeetonitrile at 70°C for three hr. The ac tivities of these enzymes were found to be considerably enhanced. In order to understand the basis of thi s change in the acti vity of these enzymes, proteinase K was analyzed in detail us ing X-ray diffraction method. The overall structure of the enzyme was found to be si milar to the nati ve structure in aqueous environment. The hydrogen bonding system of the catalytie tri ad remained intact after the treatment. However, the water structure in the substrate binding site underwent some rearrange ment as some of the water molecules were ei ther di splaced or completeiy absent. The most striking observation concerning the water structure was the complete deletion of the water molecule which occupied the position at the so-called oxyanion hole in the active site of the native enzyme. Three acetonitrile molecules were found in the present structure. All the acetonitrile molecules were located in the recognition site. In terlinked th rough water molecules, the sites occupied by acetonitrile molecu les were independent of water molecules . The acetonitrile molecules are in volved in ex tensive interactions with the protein atoms. The methyl group of one of the acetonitrile molecules (CCN I) interacts si multaneously with the hydrophobic side chains of Leu 96, lie 107 and Leu 133. The development of such a hydrophobic environment at the recognition site introduced a striking conformation change in lie 107 by rotat.ing its side chain about Ca­CI3 bond by J 80° to bring about the o-methyl group within the range of attractive van der Waals interac tions with the methyl group of CCN I. A similar change had earlier been observed in proteinase K when it was complexed to a substrate analogue, lac toferrin fragment.

Introduction The analysis of enzyme behaviour in anhydrous

media has important implications in biotechnology and organic chemistryl -4 . Some interesting phenomena like pH memor/·6 and molecular imprinting7

•8 utilizing such media have also been

reported. The key issue is to evaluate the effects of exposure to such media on the structure of the enzymes. There have been reports on enzyme structures by soaking the crystals of enzymes which were grown in aqueous media, in organic solvents using methods of X-ray diffraction9

-11

, NMR 12 and FfIR 13. The results of these investigations did not indicate any significant structural change in the protein molecule and both flipping rates and liberational motions in the enzyme were found much greater in aqueous crystals rather than in lyophilized

*Author for correspondence E-mail : tps @aiims.aiims.ac.in

powders suspended in anhydrous solvents 12. It may be recalled that the enzymes when placed in such media showed exceptionally high thermal stability which is monitored by their biological activit/ 4

• With the prevailing notion that nothing drastic seems to happen to the enzyme structures in organic solvents, we conducted experiments by placing the enzymes in anhydrous organic solvents at a rnoderately high temperature of around 70°C. We report here the changes in the kjnetic parameters of six different enzymes as a result of such an exposure and three­dimensional (3 D) structural changes in one of them viz proteinase K, a serine proteinase from fungus Trirachium album Limber l5 .

Materials and Methods Trypsin, wheat germ acid phosphatase, a-amylase,

~-glucosidase, p-nitrophenyl, ~-·glucopyranoside

(PNGP), N-a-benzoyl DL-arginine p-nitroanilide (BAPNA) and N-a-benzoyl arg in ine ethyl ester

SHARMA el 01.: 3D STRUCTURE OF MODIFIED SERINE PROTEINASE 35

(BAEE) were purchased from Sigma Chemical Co. USA. Chymotrypsin and p-nitrophenyl phosphate (PNPP) were purchased from Sisco Research Lab., India. Proteinase K was obtained from E.Merck, Germany. Organic solvents were of HPLC grade and llsed after drying over 4 A molecular sieves. All other reagents used were of analytical grade.

Heat treatment of enzymes in organic solvent The enzyme (l mg) was put in different 10 ml

screw capped vials. One ml organic solvent (aceto­nitrile/toluene/dimethyl fonnamide) was added to each vial and the enzyme solvent suspension was heated at 70°C for different time intervals (0-5 hr) in a water bath. The samples were cooled at 25°C and the solvent was removed using vacuum (Tables 1 and 2).

Enzyme assays Activities of proteinase K 16 and chymotrypsin 17

were measured using BTEE as substrate. Activities of acid phosphatasel 8

, a-amylase 17, ~-glucosidaseI9 and

trypsin20 were measured using PNPP, starch, PNGP and BAPNA as substrates respectively. All the activity measurements were repeated six times. The values showed a variation of less than 3%.

Table I-Effect of heat treatment on enzymes in acetonitrile

[The enzyme (1 mg) was incubated with acetonitrile (1 ml) at 70°C for varying time intervals (0-5 hr) and activities were measured after the removal of solvent as described. Control was also run where the organic solvent was removed after 30 seconds of incubation at 25°C. The initial rates were calculated with those samples where maximum increase in the activity was observed]

Enzymes Time V/ Vc# VTIVC (hr)*

Proteinase K 3 48 35 1.37

Chymotrypsin 2 60 48 1.25

Acid phosphatase 5 40 36 1.11

~-glucosidase 4 260 164 1.59

Trypsin 3 67 53 1.32

a-Amylase 3 42 36 1.16

*Time of incubation for maximum enzyme activity. #VT and Vc are the initial rates (in Jlmole min· t mg' t protein) of the product formation for treated and control samples respectively.

Crystal preparation Proteinase K (E.e. 3.4.21.14) from fungus

Tritirachium album Limbert5 obtained from Merck (Darmstadt, Germany) and treated with organic solvents as described earlier was purified by gel filtration using Sephadex G-75 column in 50 mM Tris.HCl, pH 7.5 containing 1 mM CaCI2. Fractions of highest activity were pooled, dialyzed exhaustively at 4°C against 1 mM calcium acetate and lyophilized. This treated and purified form of proteinase K was crystallized by dissolving the 10% (w/v) of lyophilized enzyme in 50 mM Tris.HCl, 1 mM CaCI2, H 6.5. This solution (25 III drops) were equilibrated in a microdialysis set-up against 1 M NaN03 in the same buffer at 4°e. The crystallization set-up did not contain organic solvents. Single crystals of size 0.3xO.2xO.2 mm3 grew within 6-7 days.

X-ray diffraction data For X-ray intensity data collection, one crystal was

mounted in a glass capillary. The X-ray intensities were collected to 2.2 A resolution at 12°C using a 300 mm MAR Research imaging plate scanner mounted on a RU-200 Rigaku rotating anode generator operating at a rate of 40 kY and 100 mAo The graphite monochromater was used to generate the CuKa radiation. Crystallographic data, data collection and processing details are given in Table 3. The data were processed using the HKL-package2t .

Results and Discussion

Analysis of kinetic data It has been shown that the enzymes in anhydrous

organic solvents show considerably higher themo­stability as compared to their stability in aqueous buffers 14. It has also been reported that at considerably high temperatures, viz. 11O-145°C, the enzymes such as ribonuclease, chymotrypsin and lysozyme lose their activity by following inactivation mechanisms similar to the ones established for thermo-inactivation in aqueous buffers22

. However, no data were available about the effect of heating on the enzymes in such media at temperatures which are moderately high but still not high enough to cause significant loss of enzyme activity.

Six enzymes, viz. Proteinase K, chymotrypsin, acid phosphatase, ~-glucosidase, trypsin and a-amylase were heated at 70°C in nearly anhydrous acetonitrile for a few hours. After this treatment, the enzymes were cooled to room temperature and organic solvent

36 INDIAN J BIOCHEM BIOPHYS, VOL. 38, FEBRUARY & APRIL 2001

was removed by placing the system under reduced pressure. The initial rates of respective substrate conversions are shown in Table 1. In all the cases, the "exposed" enzymes showed higher in itial rates of activities as compared to the "controls". These observations are in agreement with an earlier preliminary report on urease, N,N-dimethyl

formamidase and phospholipase A/3. So far,

chemical modifications24 and protein engineering25

had been used to alter the catalytic power of the enzymes. The semjnal work with triose phosphate isomerase26 has emphasized that the evolutionary forces tend to optimize kca/Km in thei r search for the perfect enzyme.

Table 2- Effect of heat treatment in organic solvents on VmaJKm of enzymes

Enzymes Organic Solvent Vm .. Kill Vma.tKm

Proteinase K -control Acetonitrile 0.34 5.43 0.063

-treated 0.46 5.48 0.084

-control Toluene 0.38 5.43 0.07

-treated 0.47 5.52 0.085

-control Dimethyl fo rmamide 0.36 5.25 0.069

-treated 0.43 5.23 0.077

Chymotrypsin -control Acetonitrile 0.24 0.98 0.245

-treated 0.31 1.02 0.304

-control Toluene 0.38 1.23 0.309

-treated 0.49 1.2 I 0.405

-control Dimethyl formamide 0.27 1.17 0.231

-treated 0.36 0.92 0.391

Acid phophatase -control Acetonitrile 9.32 0.29 32.14

-treated 10.02 0.23 43.57

-control Toluene 10.54 0.31 34.00

-treated 12.21 0.29 42.10

-control Dimethyl formamide 8.74 0.24 36.42

-treated 9.25 0.19 48.70

f3~lucosidase

-control Acetonitrile 0.11 2.89 0.038 -treated 0.17 3.25 0.052 -control Toluene 0.07 4.22 0.017 -treated 0.10 3.99 0.025 -control Dimethyl formamide 0.10 2.28 0.044 -treated 0.15 3.1 I 0.048

Trypsin -control Acetonitrile 0.13 0.91 0.143

-treated 0.18 1.20 0.150

-control Toluene 0.17 1.41 0.121

-treated 0.23 1.71 0.135

-control Dimethyl formamide 0.16 1.31 0.122

-treated 0.24 1.61 0.149

a-Amylase -control Acetonitrile 1.26 0.21 6.000 -treated 1.56 0.25 6.240 -control Toluene 2.07 0.47 4.404 -trcated 3.38 0.57 5.930 -control Dimethyl formamide 1.63 0.24 6.792 -treated 3.33 0.35 9.514

SHARMA et at.: 3D STRUCTURE OF MODIFIED SERINE PROTEINASE 37

Table 3-X-ray data collection statistics of proteinase K

Resolution (A)

Wavelength (A)

Space group

Cell dimensions

X-ray source

Collimation

Detector

Crystal to plate distance

Total number of observed

reflections

Number of independent

reflections

Completeness of all data (%)

Rsym* factor for all data (%)

Completeness of outer shell

2.30-2.2 (%)

Rsym in outer shell

2.30-2 .2 (%)

15.0-2.2

CuKa, 1.5418

P432,2

a=b=68.32, c=108.37

Rotating anode generator,

operating at 40 kY, 100 rnA

(focal size 0.3x3 mm2)

0 .3xO.3 mm2

MAR Research imaging plate scanner

100mm

89243

12156

97

4.5

94

11.6

The consistent improvement in catalytic efficiency observed by heating in nearly anhydrous organic solvent at 70°C led us to examine the details of the structural changes in one of the enzymes viz. Proteinase K as a result of such an exposure. The 3D structure of the exposed enzyme was determined by X-ray diffraction .

Structure determination

The initial phases were determined by molecular replacement using the program AMORE27 as implemented in the CCP4 package28

, using the previously determined structure of proteinase K at 1.5 A resolution in aqueous buffer as the search model29

. The rotation search using the 10.0-4.0 A data revealed one unique solution. The comparison of the molecular replacement solution to that of the original search model showed that the root mean squares (r.m.s.) displacement was only 0.32 A, suggesting that the drastic structural rearrangement did not take place as a result of the heat treatment of the enzyme in the anhydrous medium. A series of 2Fo-Fc maps were calculated using the phases determined from models that had sequential regions omitted (8% of the total

atoms). The omit maps showed clear density for all the omitted regions.

Refinement and model building After molecular replacement, the R-factor for the

search model was 0.296 in the range of 17.0 to 2.2 A resolution and the Rfree consisting of reflections that are left out of refinement (6% of total) was 0.34. The search model was first rebuilt residue by residue into the omitted regions by using the omit maps generated as above adjusting side-chain torsion angles and occasionally residue rotamers. Approximately 30% of the residues required some adjustments to centre them in the electron density. The side chains of three amino acid residues: Gin 103, Arg 167 and Gin 278 were not present in the original structures. These side chains were clearly visible to fit them into the electron densities well. The most striking observation was made in respect of lie 107 for which electron density suggested a rotation of the side chain by 180° about the Ca-C~ bond (Fig. 1). As this observation was rather striking and implications were strong, the structure of native proteinase K in aqueous buffer was

I-I g. I-Stereoview of a 2Fo-Fc electron density map calculated at 17.0-2.2 A resolution contoured at 1.0 (J with superimposed coordinates of treated proteinase K (red) and native proteinase K (blue) structures. [The orientation of He 107 side chain of treated proteinase K fits well in the electron density while that of nati ve proteinase K needs 1800 rotation about Ca-CI3 bond]

38 INDIAN J BIOCHEM BIOPHYS, VOL. 38, FEBRUARY & APRIL 2001

analyzed once again using freshly grown crystals with a similarly collected high quality data to confirm the conformation of lie 107 in the original structure. The new structure determination showed that the structure was identical to the one determined earlier (unpublished results). Several rounds of refinement by restrained parameter least-squares using PROTINI PROLSQ program28 followed by manual rebuilding using the program 0 30 were then performed on the protein structure. Initially, we used the model structure for 15 cycles of xyz refinement. Thereafter, the refinement was carried out for xyz and individual B factors for another 15 cycles. During the refinement, adjustments of side chains was carried out. The inspections of (2Fo-Fc) and (Fo-Fe) difference Fourier maps indicated the position of two Ca2

+ ions which were added in the subsequent cycles of refinement.

Solvent molecules Two rounds of manual building using Program 0

and refinement with PROTIN/PROLSQ resulted in the addition of 42 water molecules. Potential water molecules were located in the (Fo-Fc) electron density map at the 3.0 (J contour level. Initially, they were only built into the density peaks that clearly exhibited the appropriate shape. Any peak that was possibly due to an acetonitrile molecule was excluded. The refinement protocol using PROTINIPROLSQ was same as described above. Those water molecules were retained for which the electron density in the subsequent 2Fo-Fc maps persIsted after the refinement cycles and if they fulfilled the followi ng criteria: within 3.4 A of the enzyme oxygen or nitrogen atom (or bound water in the second round) with good hydrogen bonding geometry, B factors less than 45 A 2 and the real space correlation coefficients above 65%. After inclusion of these 42 water molecules, the R-factor and free R factor came down to 0.233 and 0.276 respectively for all the data (17.0-2.2A resolution). Further rounds of manual building in program 0 combined with several cycles of refinement with PROTIN/PROLSQ resulted in the inclusion of 40 additional water molecules for a total of 82 water molecules in the model. The refinement with further cycles and model building resulted in the inclusion of final 40 water molecules with an improvement in the R-factor to 0.208.

Acetonitrile molecules were introduced into the model through several rounds of manual building, using program 0 and refinement with PROTINI

Fig. 2-Ball and stick model of an acetonitrile molecule and water molecule with characteristic electron densities in Fo-Fc map.

PROLSQ. Water molecules were distinguished from acetonitriles in the electron density map because at 1.0 (J contour level, water molecule electron densities were spherical whereas for acetonitrile they were

. quite ellipsoidal (Fig. 2). The existence of proteinase K-bound acetonitrile molecules was further confirmed when we overlaid the (2Fo-Fc) electron density map for the enzyme in water with the refined coordinates of proteinase K in acetonitrile and found no electron density corresponding to acetonitrile. A total of three acetonitrile molecules were included in the final model. The occupancies of aceton itrile molecules were refined which finally optimized to 1.0 for all the nine atoms in three molecules. The positions which were occupied by acetonitrile in the present structure had no water molecules in the native structure of proteinase K in aqueous medium.

Model quality The final model contammg 2030 protein atoms,

two calcium atoms, 135 water molecules and three acetonitrile molecules (nine atoms) gi ves an R factor of 0.168 for all the reflections to 2.2 A resolution. The Rfrce factor was calculated to be 0.231 for 6% reflections. The model has stereochemistry with r.m.s. deviations of 0.008 A for bond lengths, 1.90 for bond angles and 18.90 for dihedral angles.

The main chain and side chain atoms have well defined electron densities. The distribution of main chain dihedral angles (</>,\jI) 3 I shows that 88% of the

SHARMA et al.: 3D STRUCTURE OF MODIFIED SERINE PROTEINASE 39

Table 4-Details of the refinement of proteinase K structure

Resolution range (A)

R factor for all data (%)

Rfree (%)

Number of reflections

Average B factor of protein atoms (A 2)

Average B factor of buried atoms (A 2)

Average B factor of main chain atoms (A 2)

Average B factor of side chain atoms (A 2)

Number of protein atoms

Number of calcium atoms

Number of acetonitrile atoms

R.m.s. deviation in bond lengths (A)

R.m.s. deviation in bond angles (0)

Overall G factor

Residues in the most favoured region

of Ramachandran plot (%)

Outliers in Ramachandran plot (%)

R-factor=Lhkll Fobs I-k I F e•1 II Lhkl l Fobs I

15 .0-2.2

16.8

23 .1

12156

17.1

13.2

15.6

18.7

2030

2

9

0.008

1.9

-0.08

88

None

Rfree is the R-factor for a set of 6% randomly chosen reflections that are not included in the refinement.

residues are located in the most favoured regions, 12% are in the additionally allowed regions and no residues are found in the generously allowed and disallowed regions of the plot as calculated using the program PROCHECK27

• The quality of the final model is summarized in Table 4.

Binding of acetonitrile molecules to the enzyme There are three acetonitrile molecules in the

structure. All of them are located in the binding domain of the enzyme. They occupy unique positions which were not occupied by solvent water molecules in the native structure of proteinase K. The structure of subtilisin carlsberg in 40% acetonitrile had also indicated the presence of three acetonitrile molecules in the active si te33

. As seen from Fig. 3, the acetonitrile molecules interact with each other through solvent molecules. CCN1 is linked to CCN3 through two interconnected water molecules, OW 576 and OW 629. They also interact extensively with protein atoms. CCN 1 forms a hydrogen bond with -NH of Gly 134 and its -CH3 group is involved in attractive hydrophobic interactions with lie 107, Leu 133 and Leu 96. A close-up view of the binding region with a hydrophobic surface is shown in Fig. 4. The CCN2 forms a hydrogen bond with Asn 161 and interacts with Ser224 via OW 642. Similarly, CCN3

forms a hydrogen bond with Asn 67 and OW 642. The water molecule which is known to occupy the position of oxyanion hole in the native proteinase K structure, and interacts with Asn 161 and Ser 224 is not present in the treated enzyme. Instead, a new water molecule, OW _ 642, which is involved in hydrogen bonded interactions with Asn 161, CCN2 and Ser 224 seems. to serve the role of oxyanion hole.

Structure comparison The structures of treated proteinase K and the

native enzyme29 were superimposed using LSQKAB (CCP4) and the r.m.s. deviation for the backbone was found to be 0.14 A. This indicated that the overall

ASN 161 l GLY, ..

SER224k:::t .. ~ }(l. lEU' .. (\"~ CCN1 ~l~U;07

CCN2 '.'............... -"OW578 , , ow, •..... \(LM.

CCN3/j"" (. T \ GI.Y100

~ASN67

Fig. 3-Hydrogen interactions involving acetonitrile (blue), water (red) and protein (green). [OW 642 is located close to the oxyanion hole] .

Fig. 4-Acetonitrile molecules in the binding region of the enzyme. [CCN 1 interacts with Leu 96, lIe 107 and Leu 133. CCN I, CCN 2 and CCN 3 arc interlinked through hydrogen bonds]

40 INDIAN J BIOCHEM BIOPHYS, VOL. 38, FEBRUARY & APRIL 2001

Fig. 5-Superimposition of the active site residues of treated proteinase K (red) and native proteinase K (blue) showing a good match of their stereochemistry.

structure of the enzyme was unaffected even after such a strong treatment. Since the treated enzyme exhibited a markedly higher activity (1.4 times higher than the native proteinase K), attention was focussed on the substrate binding site. The substrate recognition site in proteinase K is formed by two peptide chains Asn 99-lIe 108 and Leu 131-Gly 136. The recognition site residues are accessible to the solvent on one side whereas, the other ends of these strands terminate at the catalytic residues. The residues of the catalytic triad display low thermal motions with a well defined geometry and are hardly accessible to the solvent molecules. These residues in the modified enzyme were found to be stereo­chemically identical to those of the native proteinase K (Fig. 5) and their hydrogen bonding distances compared well with the native proteinase K (Fig. 6), thus suggesting that the activity of the enzyme was not influenced due to any rearrangement of catalytic residues.

Indeed, the most striking change was observed in the side-chain of lie 107 where a 1800 rotation had occurred about the Ca-C~ bond. This change seems to have been induced by the interaction of acetonitrile molecule (CCN 1) with the side-chain of lie 107, thus optimising the structure of the recognition. site to facilitate the diffusion of the substrate smoothly. The regions of the recognition site of the treated enzyme were superimposed on the corresponding regions of proteinase K which was complexed with a substrate analogue lactoferrin fragment34

. It clearly showed that the two structures matched very well and lie 107 in the two structures adopted identical conformations.

\ .. fU4 ":"

< ..

·:i;u;}

~J , , .-

/~Uj4 . :

Fig. 6-The normal hydrogen bonding distances between the residues of the catalytic triad showing that exposure of the enzyme to acetonitrile at 70°C for three hr and subsequently, location of three acetonitrile molecules at the binding site did not affect the hydrogen bonding distances ..

Thus, it could be concluded that the structure of the treated enzyme resembled with that of the non­covalent intermediate in the enzyme·· ubstrate cascade reaction .

Conclusions It is clear that the enzymes in certain anhydrous

organic solvents behave differently. It has been reported that the enzyme crystals transferred to the organic solvents possess structures similar to the native structures in aqueous media9

-1I

. A preliminary study has indicated that the enzymes in anhydrous organic solvents show higher thermal stabili ty23. Our studies have shown that the enzymes heated at reasonably higher temperatures in organic solvents for a number of hours show substantially enhanced enzymatic activity. The nature of organic solvents and the range of temperature and the period of heating need to be optimized for di fferent enzymes. The treatment carried out by us is novel and enhances both the thermal stability and the enzymatic activity. The required changes in the enzyme for enhanced activity need both organic solvents and heating.

The 3-D structure determination of the treated proteinase K revealed that the heating of the enzyme at 70°C for three hours in acetonitrile did not perturb its overall structure but caused certain strong changes which turned out to be useful. Three acetonitrile molecules were located in the binding region. The water structure elsewhere in the molecules remained

SHARMA et al. : 3D STRUCTURE OF MODIFIED SERINE PROTEINASE 41

unchanged. The conformations of Leu 96 and Leu 133 side chains changed considerably. The most drastic change was observed in case of lie 107 in which the side chain had rotated about the Ca.-C~ bond by 180°.

Acknowledgement It is a pleasure to thank Dr. Christian Betzel for

very fruitful discussions on the structure of the enzyme. The early exploratory experiments related to this work were carried out by Dr. Renu Batra. The financial support from the Department of Science and Technology (DST, New Delhi) and the Council of Scientific and Industrial Research (CSIR, New Delhi) is acknowledged.

References I Gupta, M N (1992) Ellr J Biochem 203, 25-32 2 Hailing, P J (1997) Biochem Soc Trans 25, 170-174 3 Tuena de Gomez-Piyou M. & Gomez-Piyou A (1998) Crit

Rev Biocheln Mol Biol33, 53-89 4 Travais, J (1993) Science 262 , 1374 5 Klibanov, A M (1995) Nall/re. Londoll 374, 596 6 Ke, T & Klibanov, A M ( 1998) BiotecllllOl Bioeng 57, 746-

750 7 Braco, L, Dabulis, K & Klibanov, A M ( 1990) Proc Nat

Acad Sci 87, 274-277 8 Mingarro, I, Abad, C & Bmco, L ( 1995) Proc Nat Acad Sci

92,3308-3312 9 Fitzpatrick, P A, Steinmetz, A C, Ringe, D & Klibanov, A M

(1993) Proc Nat Acad Sci 87, 274-277 10 Fitzpatrick, P A, Ringe, D & Klibanov, A M (1993) Biochem

Biophys Res Commlill 198,675-681 11 Schmitke, J L, Stern , L J & Klibanov, A M (1997) Proc Nat

Acad Sci 94, 4250-4255 12 Desai, U R & Klibanov, A M (1995) JAm Chelll Soc 117,

3940-3945 13 Griebnow, X K & Klibanov, A M (1997) Biotecllllol Bioeng

56,485-490

14 Gupta, S & Gupta, M N (1994) in Mechanisms of irreversible thermoinactivation and medium engineering. In : Thermostability of enzymes (Gupta, M NEd) pp 114-1 22 Springer Verlag, Germany

15 Ebeling, W, Henrich, N, Klockow, M, Metz, H U & Lang, H (1974) EliI' J Biochem 47, 91-97

16 Broemme, D, Peters, K, Fink, S & Frittkau, K (1986) Arch Biochem Biophys 244, 439-446

17 Decker, L A (1977) in Worthington En zyme Manllal. Worthington Biochem Corp. NelV Jersey

18 Saini, M S & Etten, R L (1979) Biochim Biophys Acta 568, 370-376

19 Honda, H, Saito, T I & Kobayashi, T (1988) Enzyme Microb Technol 10, 559-560

20 Erlanger, B F, Kowowsky, K & Chen , W (1961 ) Arch Biochem Biophys 95, 271-278

21 Otwinowski, Z & Minor, W (1997) Methods En zymol 276, 307-328

22 Vol kin , D B, Staubli , A, Langer, R & Klibanov, A M (1991 ) Biotechnol Bioel/g 37, 843-853

23 Grunwald, P, Hansen, K & Remus, M (1993) Indian J Chem 32B,70-72

24 Tyagi, R & Gupta, M N (1998) Biochemistry (Moscow) 63, 334-344

25 Ulmer, K M (1993) Science, 219, 666-668 26 Knowles, J & Alberty, J A (1977) Aces. Che. Res. 10, 105 27 Navaza, J (1994) Acta Crystallogr. A50, 157-163 28 Collaborative Comutational Project No. 4 (1994) Acta

Crystallogr. D50, 760-763 29 Betzel, Ch, Pal, G P & Saenger, W ( 1988) EliI' J Biochem

178, 155-171 30 Jones, T A, Zou, J Y, Cowan, S W & Kjeldgaard, M ( 1991 )

Acta Crystallogr A47, 110-119 31 Ramachandran, G N & Sasisekharan, V (1968) Adv Prot

Chem 22, 283-437 32 Laskowski, R A, McArlhu, M W, Moss, D S & Thornton, J

M (1993) J Appl Crystallogr 26,283-291 33 Schmitke, J L, Stern, L J & Kilbanov, A M (1998) Biochem

Biophys Res COI11I11 248, 273-277 34 Singh, T P, Sharma, S, Karthikeyan , S, Betzel , Ch & Bhati a,

K L (1988) PROTEINS: Struct. Fill/ C. Gel/et. 33,30-38