Eur. J. Biochcm. IY7.203-207 (1991) FEBS 1991 001429569100227X

Studies on RNase T l mutants affecting enzyme catalysis Hans-Peter GRUNERT, Athina ZOUNI, Marc BEINEKE, Rainer QUAAS, Yannis GEORGALIS, Wolfram SAENGER and Ulrich HAHN Institut fur Kristallographie, Freie Universitat Bcrfin, Federal Republic of Germany (Received October 8, 1990/January 3, 1991)

-

EJB 90 1204

Using an Escherichia coli overproducing strain secreting Aspergillus oryzae RNase T I , we have constructed and characterized mutants where amino acid residues in the catalytic center have been substituted. The mutants are His40 Thr, Glu58 + Asp, Glu58 + Gln, His92 + Ala and His92 +Phe. His92 + Ala and His92 + Phe mutants are inactive. On the basis of their k,,,/K, values, the mutants Glu58 + Asp and Glu58 + Gln show 10% and 7% residual activity, relative to wild-type RNase T1, whereas the His40 + Thr mutant shows 2% activity. The effect of amino acid substitutions on the enzymatic activity of RNase T1 lends further support for a mechanism where Glu58 (possibly activated by His40 and His92 act as general base and acid respectively; this is discussed in terms of the known three-dimensional structure of the enzyme. --f

The mould fungus Aspergillus oryzae produces the most thoroughly investigated microbial RNase, RNase TI. The sequence of the 104 amino acids of this protein has been determined by Takahashi ( 3 971, 1985), and the molecular mass was calculated to be 11085 Da. The enzyme and its interaction with nucleic acids have been studied by applying a variety of techniques (for reviews: Egami et al., 1980; Takahashi and Moore, 1982; Heinemann and Hahn, 1989). Its three-dimensional structure is known from X-ray analysis to a resolution of 0.18 nm (Heinemann and Saenger, 1982; Arni et al., 1988; Koepke et al., 1989; Kostrewa et al., 1989) as well as from two-dimensional NMR spectroscopic studies (Hoffmann and Riiterjans, 1988). RNase T1 cleaves single-stranded RNA specifically after guanosine residues to yield 3‘-phosphorylated reaction products. The cleavage of the RNA chain follows a two-step mechanism. In the first transesterification step, the P-05’ bond is broken and a 2’,3’-cyclic phosphate is formed as an intermediate, which is hydrolysed in the second step to yield the free 3’-phosphate. As shown by Eckstein et al. (1972), the stereochemistry of the reaction corresponds to an in-line mechanism. According to Usher (19691, this implies that at least two different groups of the protein must be involved in the catalytic action in a simultaneous acid/base catalysis. Based on a model published by Takahashi (1970) and on the three-dimensional structure of RNase TI, Heinemann and Saenger (1982, 1983) proposed Glu58 and His92 to be the reactive base and acid, respectively. In the transphosphorylation step, the deprotonated carboxylate group of Glu58 (activated by His40) abstracts the proton from the sugar 02’, thus initiating a nucleophilic attack by this atom at the Correspondence to U . Hahn, Abteilung Saenger, Freie Universitiit Berlin, Institut fur Kristallographie, TakustraBe 6, W-I000 Berlin 33, Federal Republic of Germany Abbreviations. RNase TI (T2), RNase T1 (T2) from Aspergillus o v y a e ; RNasc A, bovine pancreatic RNase; RNase C2, RNasc from Aspergillus cluvurus. Enzyme. Ribonuclease Ti (EC 3.1.27.3).

3‘-phosphate group, and the leaving 05’ group is protonated by His92. The 2’,3‘-cyclic phosphate is formed with a pentacovalent phosphate as transition state. In the second reaction step, the cyclic phosphate is hydrolysed by a reverse simultaneous acid/bdse catalysis. This mechanism is different from that proposed by Riiterjans et al. (1987) and Nishikawa et al. (1987), but similar to that found for pancreatic RNase (RNase A) where two histidines, one protonated (Hisl 19; comparable to His92 in RNase T1) and the other one deprotonated (Hisl 2 ; comparable to Glu58 in RNase Tl), function as acid and base, respectively (Wlodawer, 1985; Heinemann and Hahn, 1989). Two genes coding for RNase T1 have been chemically synthesized and overexpressed in Escherichia coli by lkehara et al. (1986) and Quaas et al. (1988a, b) following independent and different approaches. The constructs permit the production of mutants where amino acids in the active site of the wild-type enzyme are exchanged with others. In the present work we report a kinetic analysis of such RNase T1 mutants. MATERIALS AND METHODS Enzymes and Chemicals

Enzymes were obtained from Boehringer (Mannheim, FRG) or Bio-Rad (Richmond, USA). Bio-Gel P30 was purchased from Bio-Rad (Richmond, USA), Geneclean glass milk from United States Biochemical Corporation (Cleveland, USA) and T7 Polymerase Sequencing Kit, DEAE-Sepharose CL-6B from Pharmacia (Uppsala, Sweden). The radiochemicals were from Du Pont/NEN (Dreieich, FRG), yeast RNA and GpN from Sigma (St Louis, USA). All other reagents were obtained from Sigma (St Louis, USA) or MerckSchuchardt (Darmstadt, FRG) and were of analytical grade. Bnc t er ial strains, hac ter iop hages and p lasmids E. coli DH5a strain (for genotype see Quaas et al. 1988a) was purchased from GIBCO BRL (Eggenstein, FRG). For

204 site-directed mutagenesis we used the E. coli strains CJ 236 [dut-1,ung-1, thi- I , relA-1, pCJ 105(Cmr)] and MV 1190 [ A (lacproAB), thi, supE, A(srl-recA)306 :: TnlO(tet') [ F : tra 0 3 6 , pro A B , lac IqZAM15]], both part of the mutagene M13 in vitro mutagenesis kit (Bio-Rad; Richmond, USA), which also contained the bacteriophage M I 3mpl8 (Messing, 1983). Recombinant D N A techniques

Plasmid DNA and replicative M I 3 DNA were isolated from E. coli by the method of Birnboim and Doly (1979). Transformation with plasmid DNA and transfection with M13 were carried out following standard procedures (Mandel and Higa, 1970; Messing, 1983). Fragments to be ligated were separated on 0.8% agarose gels and isolated by the glass beads/NaI method (Vogelstein and Gillespie, 1979). All other preparation steps were carried out according to standard methods described by Sambrook et al. (1989). Site-directed mutagenesis

The RNase TI gene, obtained from the expression vector pA2T1 (Quaas et al., 1988b), was inserted between the XbaI and Hind111 restriction sites of MI 3mp18. Mutants were constructed according to Kunkel (1985). The mutagenic primers were synthesized on a DNA synthesizer 380A from Applied Biosystems (Foster City, USA). Primer purification was carried out by reversed-phase and anion-exchange chromatography on a Pharmacia FPLC system. The primers used for the mutations were as follows: His40+Thr, 5'-GTTGTATTT GGTTGGGTAAG-3' ;

';

Glu58+Asp,

5 -GGA TAG GCC AGT CGT AGT A G G - 3

Glu58+Gln,

5 ' -GGC CAT TGG TAG TAG G - 3

His92+Ala,

5 I -GCA CCA GTA GCA GTG A T A A C A C C - 3

HisgB-Phe,

5 ' -GCA CCA GTG AAA G T G A T A A C A C C - 3 ' .

'; I

;

Purijkation of wild-type RNase TI and its mutants E. coli DH5a cells were transformed with the expression vector including the wild-type or mutated RNase T I gene and were grown in Luria-Bertani liquid medium (Sambrook et al., 1989) containing 100 pg ampicillin/ml. The growth of bacteria was monitored photometrically at a wavelength of 600 nm. Gene transcription was induced at an A600 of 0.5 by adding isopropyl-thio-fl-D-galactoside to a final concentration of 0.1 mM. Cells were then allowed to grow overnight reaching a total A,,, varying between 4.0 and 5.0. Harvesting of cells and preparation of the periplasmic fraction was performed as previously described (Quaas et al., 1988a). The periplasmic fraction of the E. coli cells (1 1, when starting with a 6-1 liquid culture) was applied to a DEAESepharose CL-6B column (Pharmacia; 5 cm x 12 cm) equilibrated with 50 mM Tris, 10 mM EDTA and 100 mM NaCl, pH 7.5. After elution with a linear increasing NaCl gradient (100-700 mM), fractions containing RNase TI or mutants were pooled, lyophilized, resuspended in 10 mi 50 mM ammonium acetate buffer and finally separated by gel-filtration chromatography using a Bio-Gel P-30 column (1.5 cm x 70 cm). Fractions containing native RNase TI or its mutants were combined and lyophilized. All proteins were identified by enzyme activity, SDSjPAGE and/or binding to antibodies. The purity was checked by SDS/PAGE (Laemmli,

1970) and analytical FPLC on a Mono Q HR5/5 column (Pharmacia) using a linear NaCl gradient (50 - 350 mM NaCl in 20 mM Tris/HCl, pH 7.5). Rapid assay of RNase TI activitj

Yeast RNA was used to determine RNase TI activity as described by Fulling and Ruterjans (1978). RNA hydrolysis was followed using a Beckmann DU 6 spectrophotometer by diluting yeast RNA to A260 = 0.80 in 10 mM Tris/HCl and 1 mM EDTA, pH 7.5, and monitoring the resulting hyperchromicity at 280 nm. Estimates of enzyme activity were made by comparing the rate of RNA hydrolysis with rates obtained by known quantities of native RNase T I . Inactive mutants were identified by SDSjPAGE and immunoassay. Kinetic measurements

At least two individual preparations of each protein were used for the kinetic measurements which were performed with the Shimadzu UV-2100 dual-beam recording spectrophotometer operating in the kinetics mode. A buffer system containing 20 mM Tris/HCl and 20 mM EDTA, pH 7.5, was employed throughout. The temperature, 20 0.1 "C, was controlled with a Lauda RC 6 thermostat. During measurement, the solutions were stirred by externally driven mini-magnetic bars. A11 experiments were conducted employing either 3-ml or 1.5-ml quartz cuvettes with a path length of 10 mm. Enzyme concentrations were determined spectrophotometrically using the newly determined absorption coefficient (17300 M - ' cm-') at 278 nm. Shirley and Laurents (1990) have determined a similar value (17 200 M - cm- '). The usual amounts of enzyme employed for the kinetics ranged, for the wild-type enzyme, over 10- 15 nM, and for some of the less active mutants over the range 100- 300 nM. The hyperchromic effect associated with the cleavage of the substrate GpA, was monitored at 280 nm. Initial rates (0.3 - 0.5% of the reaction) and their standard deviations were deduced by weighted linear-least-squares fit to each data set. The weights were assigned to the data according to the displayed individual standard deviations in the linear time region of the hyperchromicity curves. 10- 15 substrate concentrations were measured as duplicates, and for low substrate concentrations as triplicates, in order to assure sufficiently good statistics for the evaluation of the kinetic parameters. Individual data sets were normalized to a single RNase TI and/or substrate concentration whenever necessary, and were further analyzed by weighted Eadie-Hofstee plots.

RESULTS AND DISCUSSION Construction of' RNase TI mutants

Starting material for mutagenesis was the chemically synthesized RNase T1 gene obtained from pA2T1 (Quaas et al., 1988b) and subcloned into M13mp18. In each case the mutagenesis efficiency was at least better than 85%, as verified by single-strand sequencing (Sanger et al., 1977). After mutagenesis, genes were subcloned into the same expression/ secretion vector which is used for high-level protein expression of wild-type RNase T I . The plasmids of each mutant expressing clone were then double-strand sequenced (Chen and Seeburg, 1985).

205 0.6 - 1

800

05 600 04 0 to

cu

a

400

5 300

E 400

03 02

200

2

to

N

a

01

0 300

0 600

500

400

5E2 200 z % Z

0

100

Volume [mll

A

0 10

0

18

Volume [mll Fig. 3. Analytical FPLC on a Mono Q column. Purified proteins were rechromatographed on a Mono Q anion exchanger. In this example, wild-type recombinant RNase TI is shown. Solid lines indicate the absorbance monitored at 280 nm. Dashed lines show the linear salt gradient 50-350 mM NaCl in 50 mM Tris and 10 mM EDTA, pH 7.5

0.1

100

0

300

200

Table 1. Specificity of RNuse TI and its mutants f o r the dinucleoside phosphate GpA. The minus sign indicates no detectable activity down to 0.1 %

400

Volume Cmll

B

Fig. 1. Purificution ofthe RNuse TI mutant Glu58 4 Gln. (A) Elution profile after anion-exchange chromatography on DEAE-Sepharose CL-6B. (-) Absorbance monitored at 280 nm; (-----) linear gradient of 100-700 mM NaCl in 50 mM Tris and 10 mM EDTA, pH 7.5. (B) Profile after gel filtration on Bio-Gel P30 in 50 mM ammonium acetate buffer. Hatched areas in both figures represent the fractions containing the RNase TI mutant as determined by activity test and SDS/PAGE. Both chromatography steps were carried out at 4°C

1

2

3

4

5

6

RNase TI

Wild type Glu58 + Asp Glu58-tGln His40 Thr His92+Phe His92-tAla --f

K,,, x

k,,, x lo3

lo6 x k,,,/K,

Activity

mM

min-'

M - ' min-'

%

0.171 0.41 8 0.176 0.024

0.98 0.25 0.06 0.002

5.9 0.6 0.4 0.1

-

-

-

-

-

-

100 10 7 2 < 0.1 < 0.1

7 97.4 66.2

42.7 31.0

21.5 14.4 Fig. 2. 15% SDSjPAGE of purified RNuse TI and mutant enzymes. The following forms of RNase T1 were loaded on to the gel: wildtype (lane 1); His40-tThr (lane 2), Glu58-tAsp (lane 3); Glu58 -+ Gln (lane 4); His92 + Ala (lane 5); His92 + Phe (lane 6). Lane 7 contained low-molecular-mass standards (kDa; Bio-Rad) as indicated on the right. On SDS/PAGE, the apparent molecular mass of RNase T1 is about 17 kDa instead of 11 kDa, as determined from the amino acid composition (Quaas et al., 1988a). All lanes shown hcre are part of the same gel

Protein purification

RNase T1 and mutants were isolated from the periplasmic fraction of the overproducers as described (Quaas et al., 1988b). All proteins could be purified within two main steps.

The acid treatment, as reported by Quaas et al. (1988a) was omitted, because mild purification conditions were preferred for mutants not yet characterized. The two purification steps were anion-exchange and sizeexclusion chromatography. As a typical example, the corresponding chromatograms taken during the course of isolation of the RNase TI mutant E58Q are presented in Fig. 1. Although in this mutant a negatively charged residue is replaced by a neutral one, it elutes at nearly the same ionic strength as wild-type RNase T I , suggesting that the change of the protein net charge does not have a significant effect on its binding to the anion-exchange resin. The reason for this behaviour could be associated with the position of Glu58 in the interior of RNase T1, removed from the surface and probably charge-balanced by a cation, which is absent in the Gln58 mutant. For the same reasons, the active site mutants showed similar chromatography properties as did the other mutants where the charges were not changed. As indicated in Fig. 1 B, a symmetrical peak of the products was obtained after gel filtration. Each of the proteins could be purified to homogeneity as verified by SDSjPAGE (Fig. 2) and analytical FPLC, shown in Fig. 3 for native RNase T l . Fig. 2 shows that some of the analyzed proteins display slightly different migration behaviour on SDS/ PAGE. The yields of most of the mutant enzymes were significantly smaller (3 - 5 mg/l culture) than those of the wild-type ribonuclease (20 mg/l culture).

206

4

'I 1.2

1.0-

\

\

m 0

1 0.8> -

0.6-

0.L0.2-

0.0

L 1.0

0.0

0.05

0.00

v /IS1 10-2 Fig. 4.Eudie-Hofsteeplot of' typical dutu sets for the transesterifiication reaction of GpA wirh wilcl-type RNase TI ( a ) and with the His40 + Thr mutant ( h ) . Data were deduced from initial velocity measurements of the reaction by monitoring the increase in hyperchromicity after GpA cleavage at 280 nm, as described in Materials and Methods. The vertical bars are indicative of an estimate for the data spread deduced from triplicate measuremcnts

Enzyme kinetics of wild-type RNuse TI and mutants

Kinetic data of the investigated proteins His40 + Thr, Glu58 Asp, Glu58 4 Gln, His92 Ala and His92 + Phe are summarized in Table 1. These data were deduced from EadieHofstee plots, two of which are presented as examples in Fig. 4. Remarkable differences in kinetic constants and relative activities compared to Ikehara et al. (1987) and Nishikawa et al. (1987) could be attributed to the use of different substrates and/or to the different methodology employed (radioactive assay). According to the mechanism proposed by Heinemann and Saenger (1983), deprotonated Glu58 as a base is directly involved in the acid/base-catalyzed cleavage of RNA by RNase T1. In the Glu58 +Asp mutant where a glutamic acid is replaced by aspartic acid and consequently the carboxyl group has moved by approximately 0.1 5 nm away from the catalytic centre, the activity drops to 10% compared to the wild-type enzyme (Table 1). In this mutant, the functional group is --f

--f

still available for the catalytic reaction, although not in the optimum position, and activation of 0 2 ' is possibly mediated by a water molecule. In the Glu58 + Gln mutant, the charge in this position is removed. There is 7% residual activity, with a K, similar in both enzymes, but a k,,, of about 6% in the mutant. Based on the mechanism of Heinemann and Saenger (1983), one could have expected an inactive [Glu58 + Gln] RNase T I . Amino acid sequence alignment of 13 procaryotic and eucaryotic microbial RNases belonging to the RNase T1 family (Heinemann and Hahn, 1989) revealed that Glu58 and His92 are present in all of these ribonucleases, whereas His40 is only found in the eucaryotic enzymes. In one of the four procaryotic representatives, RNase St from Streptornvces erythreus, His40 is substituted by threonine. To shed light on this problem we investigated the [His40 + ThrIRNase TI mutant, which shows a n about 500-fold lower k,,, and a sevenfold lower K, corresponding to 2% activity compared to the wild-type enzyme (Table 1). The substitution of His92 by Ala or Phe leads to inactive enzyme and emphasizes and verifies the role of this residue as a general acid during RNA cleavage. The story is more complex, however, as removal of the imidazole side chain deletes a His92 N'H . . . 0 = C Asn99 hydrogen bond and is associated with structural changes which render the loop segment between amino acids 90 and 100 so flexible that it isn't seen in the X-ray electron density (Koellner, G. unpublished results). In addition, the recent X-ray structure of RNase TI complexed with two guanosines suggests that the imidazole of His92 acts as subsite (Lenz, H. unpublished results). This also agrees with the prediction of additional binding sites for longer substrates by Ostermann and Walz (1979). The side chain of His92 appears to serve three functions: to act as an acid, to stabilize the residue 90- 100 loop and to permit subsite binding. Until1 now, all data show that His92 is not indispensable for RNase T I catalysis, whereas our results indicate that either Glu58 or His40 can be replaced by other amino acids yielding still active mutants. At first sight, this contradicts the mechanism proposed by Heinemann and Saenger (1983), but also that postulated by Nishikawa et al. (1987). Ikehara et al. (1987) and Nishikawa et al. (1987) have also analyzed the two mutants Glu58 +Asp and Glu58 + Gln and observed 10% and I YO residual activity. Replacing His40 for Ala yielded protein with very low activity: 0.1% (Ikehara et al., 1987) o r 0.01% (Nishikawa et ai., 1987). Based on the remaining activities for the Glu58 mutants, Nishikawa et al. (1987) concluded that His92 and His40, but not GIu58, are indispensable for RNase T I activity, and proposed a mechanism similar to that of RNase A, where two histidines are the acid/base pair catalyzing the RNA-cleavage reaction. N M R spectroscopic studies of three RNases, T I , T2 and A, revealed similar high pK, values of His40 and His92 in RNase T1,7.9 and 8.0, respectively, whereas the two histidines which are probably involved in catalysis in the other two mentioned enzymes differ remarkably in their pK, values (Arata et al., 1979; Kawata et al., 1990). These findings support the postulate that in RNases T2 and A two histidines are directly involved in catalysis, and point to Glu58 as the responsible base engaged in the phosphodiester hydrolysis by RNase T I . This is also supported by the above-mentioned amino acid sequence alignment of the microbial RNases belonging to the RNase T1 family, which shows that His92 and Glu58 are conserved in all these RNases whereas His40 is only

207 been found in the eucaryotic enzymes (Heinemann and Hahn, 1989). Studies on an RNase from the prokaryote Bacillus amyloliquefaciens with a tertiary structure similar to that of RNase T1 (Hill et al., 1983) corroborate the mechanism suggested by Heinemann and Saenger (1983). The RNase TI amino acids Glu58 and His92 have their B. amyloliquefaciens RNase counterparts in Glu73 and Hisl02. His40 in RNase T1 corresponds to Asp54 in B. amyloliquefaciens RNase (Heinemann and Hahn, 1989). Mutation of Glu73 or His102 to Ala yields inactive enzymes whereas replacement of Asp54 by Ala does not substantially reduce the catalytic activity (Massakowska et al., 1989). These investigations, which were supported by 'H-NMR titration experiments, identified key residues, analogues to that proposed for RNase T1 by Heinemann and Saenger (1983), at least for the wild-type enzyme. In the Glu58 --t Gln mutant, His40 could probably play the part of Glu58 yielding a less but still active enzyme. This interpretation of the catalytic mechanism of RNase TI is further supported by the results published very recently by Steyaert et al. (1990), who investigated the pH dependence of the kc,,/Km values of the wild-type RNase TI and the mutants His40 Lys and Glu58 + Ala. They found that His40 can take the part of the catalytically active residue Glu58, if replaced by Ala. --f

We thank Udo Heinemann and Olfert Landt for discussion and for critically reading the manuscript. The exellent technical assistance of Christiane Seidl, Claudia Alings, Maria Zirpel-Giesebrecht and Andreas Milde is deeply acknowledged. Thanks belong also to Sabine Schultze and Rolf Bald for synthesizing the oligodeoxynucleotides for the site-directed mutagenesis experiments. This work was supported by the Deutsche Forschungsgemeinschaft.

REFERENCES Arata, Y., Kimura, S., Matsuo & Narita, K. (1979) Biochemistry 18, 18-24. Ami, R., Heinemann, U., Tokuoka, R. & Saenger. W. (1988) J . Biot. Chem. 263, 15 358 - 15 368. Birnboim, H. C. & Doly, J. (1979) Nucleic Acids Res. 7, 1513-1523. Chen, E. J. & Seeburg, P. H. (1985) DNA 4 , 165-170. Eckstein, F., Schulz, H. H., Ruterjans, H., Haar, W. & Maurer, W. (1972) Biochemistry 11, 3507-3512. Egami. F., Oshima, T. & Uchida, T. (1980) Mol. Biol. Biochem. Biophys. 32, 250 - 271. Fulling, R. & Ruterjans, H. (1978) FEBS Lett. 88, 279-282. Heinemann, U. & Hahn, U. (1989) in Protein-nucleic acid interaction (Saenger, W. & Heinemann, U., eds) pp. 111 -141, Macmillan Press, London and Basingstoke, UK. Heinemann, U. & Saenger, W. (1982) Nature 299,27-31. Heinemann, U. & Saengcr, W. (1983) J. Biomol. Struct. Dyn. 1, 523538. Hill, C., Dodson, G., Heinemann, U., Saenger, W., Mitsui, Y., Nakamura, K., Borisov, S., Tischenko, G., Polyakov, K. & Pavlovsky, S. (1983) Trends Biochem. Sci. 8, 364-369.

Hoffmann, E. & Riiterjans, H. (1988) Eur. J . Biochem. 177,539-560. Ikehara, M., Ohtsuka, E., Tokunaga, T., Nishikawa, S., Ucsugi, S., Tanaka, T., Aoyama, Y., Kikyodani, S., Fujimoto, K., Yanase, K., Fuchimura, K. & Morioka, H. (1986) Proc. Nut1 Acrrd. Sci. USA 83,4695 -4699. Ikehara, M., Ohtsuka, E., Tokunaga, T., Nishikawa, S., Uesugi, S.. Tanaka, T., Aoyama, Y., Kikyodani, S., Fujimoto, K., Yanase. K., Fuchimura, K., Kim, H., Ueda, Y., Kimura, T. & Morioka, H. (1987) in Biophosphates and their analogues: synthesis, sfructure, metabolism andactivity (Bruzik, K. S. & Stec, W. J., eds) Elsevier, Amsterdam, 335 - 344. Kawata, Y., Sakiyama, F., Hayashi, F. & Kyogoku, Y. (1990) Eur. J . Biochem. 187.255 - 262. Koepke, J., Maslowska, M., Heinemann, U. & Saenger, W. (1989) J . M o ~Biol. . 206, 415-488. Kostrewa, D., Choe, H.-W., Heinemann, U. & Saengcr, W. (1989) Biochemistry 28, 7592 - 7600. Kunkel, T. A. (1985) Proc. Natl Acad. Sci. USA 82,488-492. Laemmli, U. K. (1970) Nature 227,680-685. Mandel, M. & Higa, A. (1970) J . Mol. Biol. 53, 159- 162. Massakowska, D. E., Nyberg, K . & Fersht, A. R. (1989) Biochemistry 28, 3843 -3850. Messing, J. (1983) Methods Enzymol. 101, 20-78. Nishikawa, S., Morioka, H., Kim, H. J., Fuchimura, K., Tanaka, T., Uesugi, S., Hakoshima, T., Tomita, K., Ohtsuka, E. & Ikehara, M. (1987) Biochemistry 26, 8620-8624. Ostermann, H. L. & Walz, F. G. Jr (1979) Biochemistry I8, 19841988. Quaas, R., McKeown, Y., Stanssens, P., Frank, R., Blocker, H. & Hahn, U. (1988a) Eur. J . Biochem. 173,617-622. Quaas, R., Grunert, H.-P., Kimurd, M. & Hahn, U. (1988b) Nucteosides & Nucleotides 7, 619 - 623. Riiterjans, H., Hoffmann, E., Schmidt, J. & Simon, J. (1987) in Metabolism and enzymology qfnucleic acids including gene manipulations (Zelinka, J. & Balan, J., eds) vol. 6, pp. 81 -96, Slovak Academy of Sciences, Bratislava. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular cloning: a laboratory manual, Cold Spring Harbor Laboratory Press, New York. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl Acad. Sci. USA 74, 5463 - 5467. Shirley, B. A. & Laurents, D. V. (1990) 1. Biochem. Biophys. Methods 20, 181 -188. Steyaert, J., Hallenga, K., Wyns, L. & Stanssens, P. (1990) Biochemistry 29,9064 - 9072. Takahashi, K. (1971) J. Biochem. (Tokyo) 70,617-634. Takahashi, K. (1985) J . Biochem. (Tokyo) 98, 815-817. Takahashi, K. (1970) J . Biochem. (Tokyo) 67, 833-839. Takahashi, K. & Moore, S. (1982) in The enzymes, 3rd edn (Boyer, P. D., ed.) vol. 15, pp. 435-468, Academic Press, New York, USA. Usher, D. A. (1969) Proc. Nut1 Acad. Sci. U S A 62, 661 -667. Vogelstein, B. & Gillespie, D. (1979) Proc. Natl Acad. Sci. USA 76, 61 5-619. Wlodawer, A. (1 985) in Biological macromolecules & assemblies (Jurnak, F. A. & McPhcrson, A,, eds) vol. 2, Wiley, New York, pp 393-439.