Eur. J. Biochern. 201.417-420 (1991) < FEBS 1991 001429569100647B
Some properties of site-specific mutants of human carbonic anhydrase I1 having active-site residues characterizing carbonic anhydrase 111 Xilin R EN , Bengt-Harald JONSSON and Sven LINDSKOG Avdelningen for biokemi, Urnei Universitet, Sweden (Received April 4/June 27, 1991) - EJB 91 0442
Four amino acid residues, His64, Asn67, Leu198 and Va1207, in the active site of human carbonic anhydrase 11. have been replaced by Lys64, Arg67, Phe198 and Ile207, which are characteristic for the muscle-specific, lowactivity isoenzyme form, carbonic anhydrase 111. The aim of the investigation has been to test if any of these residues, o r a combination of them, is important for the low COz hydration activity, low esterase activity. low pK, for the pH/rate profile and low affinity for sulfonamide inhibitors characterizing carbonic anhydrases 111. However, n o evidence for such critical roles was found. A combination of Lys64 and Arg67 appears to result in a decrease in C 0 2 hydration activity, but even the quadruple mutant having all four changes is only eight times less active (kcat/Km) than unmodified isoenzyme 11, in contrast to isoenzyme 111 which is nearly 300 times less active than isoenzyme 11. The 4-nitrophenyl acetate hydrolase activity of the quadruple mutant is sevenfold lower than that of unmodified isoenzyme 11, while the active site of isoenzyme 111 hardly catalyzes the hydrolysis of this ester at all. The pK, controlling the esterase activity of the quadruple mutant is 6.2, which should be compared to a value of 6.8 for unmodified isoeniyme 11, and about 5 for isoenzyme 111. While isoenzyme 111 binds sulfonamide inhibitors lo3 - lo4 times less strongly than isoenzyme 11, only [Asn-67 + Arglisoenzyme I1 shows a weaker binding of the investigated sulfonamide, dansylamide, but only by a factor of two. Some of the other mutants show enhanced affinities, up to nearly fourfold for the double mutant with Phe198 and Ile207. It is speculated that additional differences between the active sites of isoenzyme I1 and 111 might be important for the precise orientations and interactions of the side chains of isoenzyme-111-specific amino acid residues.
The zinc-containing enzyme carbonic anhydrase, which catalyzes the simple, reversible reaction, CO, + H 2 0 HCO, H ' , occurs in several, genetically distinct isoenzyme forms in higher vertebrates [l]. In slow-twitch, red skeletal muscle fibers. isoenzyme 111 is the predominating cytosolic form [2]. It differs in several respects from the high-activity form, isoenzyme 11, which is present in erythrocytes, kidney and many other tissues (31. Thus, the CO, hydration activity (kc,.,/Km) of bovine isoenzyme 111 is only about 0.3% of that of human isoenzyme I1 [4]. The ionization of a zinc-bound water molecule, controlling the catalytic activity, occurs with a pK, near 7 in isoenzyme I1 but the corresponding pK, is probably about 5 in isoenzyme 111 [5]. While human isoenzyme 11 catalyzes the hydrolysis of 4-nitrophenyl acetate quite efficiently with an apparent kc,,/Km of 2.8 x lo3 M . s-' at pH 9 and 25°C [6]. the active site of isoenzyme I l l hardly hydrolyzes this ester at all [7]. Aromatic and certain heterocyclic sulfonamides are potent inhibitors of isoenzyme 11, whereas isoenzyme 111 was originally discovered as a sulfonamide-resistant form of carbonic anhydrase in livers of male rats [XI. These isoenzyme-specific functional differences are presumably related to isoenzyme-specific structural differences, Thus, it has been proposed [9, 101 that the low pK, value
+
_
~
~
_
Corresponrlcncr to S. Lindskog. Avdelningen for biokemi, UmeH Universitet. S-90187 Umel, Sweden E n z j ~ i eCarbonic . anhydrase (EC 4.2. I . 1).
of the activity-linked group in isoenzyme 111 results from electrostatic effects from some characteristic, basic active-site residues, Lys64 (or Arg64 in horse isoenzyme 111). Arg67 and Arg91. The corresponding residues in human isoenzyme I1 are His64, Asn67 and Ile91 [9]. It has also been proposed [ l l ] that the weak sulfonamide affinity characterizing all carbonic anhydrase 111 isoenzymes is due to the fact that the active-site cavity in isoenzyme I11 is smaller than that in isoenzyme 11. The extra bulk in isoenzyme 111 is mainly contributed by the side chains of Phel98 and Ile207, which correspond to Leu1 98 and Val207 in isoenzyme 11. but also by Arg67 Ill]. One aim of our studies of carbonic anhydrase is to test these and other hypotheses concerning the structural basis of the functional differences between isoenzyme forms. In this paper we present some properties of site-specific mutants of human carbonic anhydrase I1 where residues 64.67, 198 and/ or 207 have been replaced by their isoenzyme-I1 I-specific counterparts.
MATERIALS A N D METHODS Enzynr
The numbering system of human carbonic anhydrase 1 is used throughout this paper. In this system, residues 64, 67. 198 and 207 correspond to residues 63, 66, 196 and 205 in the numbering systems of isoenzymes I1 and 111 [9].
41 8 The preparation of [His64 + Lys]isoenzyme I1 has been described previously [6, 121. Other mutants were made by it7 vitro oligonucleotide-directed mutagenesis using singlestranded DNA prepared from the plasmid pACA by use of the helper phage M13K07 [13]. We employed the Muta-Gene system (Bio-Rad) which is based on the method of Kunkel (141. where the mutated strand is built on a parent strand containing U for T. pACA has been constructed in the laboratory of C. Fierke (Duke University) and consists of the human carbonic anhydrase I1 gene [I 51 behind a T7 KNA polymerase promoter [I61 in the pMa/c vector (131. Mutations were verified by dideoxy sequencing [17]. The mutant DNA was transformed into Escherichia coli strain BL21/DE3 [16]. The cells were grown at 37°C in Luria broth medium supplemented with 0.1 mM ZnSO,, 2 mM K 2 H P 0 4and 50 pg/ml ampicillin. Enzyme synthesis was induced by the addition of 0.5 niM isopropyl-thio-/?-D-gahctoside. All enzyme variants were purified by affinity chromatography essentially according to Khalifah et al. [18]. All purified variants showed a single band after SDSjPAGE in the same position as unmodified carbonic anhydrase 11. All variants appeared to be properly folded, as evidenced by their circular dichroism spectra which were very similar to that of unmodified human isoenzyme I1 in the investigated wavelength range, 200- 320 nm. Enzyme concentrations were estimated spectrophotometrically at 280 nm using c = 54.7 mM-' . cni-' [19] based on M , 29300 [20]. Kinetic tneasurernents
Initial rates of COz hydration were measured in a HiTech stopped-flow apparatus at 25 "C by the changing-pHindicator method [21, 221. Buffer/indicator pairs were 1,2dimethylimidazole/nz-cresol purple monitored at 578 nm. l-methylimidazole/4-nitrophenol or Mops/4-nitrophenol monitored at 400 nm and Mes/chlorophenol red monitored at 574 nm. The ionic strength was kept at 0.1 M using NazS04. Initial rates of 4-nitrophenyl acetate hydrolysis were monitored at 348 nm and 2 5 ' C using a substrate concentration of 0.4 mM which is well below the K , for human red cell carbonic anhydrase I1 [23]. Buffers were 50 mM Mes/ NaOH (pH 5.3 - 6.9), 50 mM Hepes/NaOH (pH 6.9- 8.6) or 'I'ris/HzS04 (pH 8.6 - 9.2). The ionic strength was maintained at 0.1 M with Na2SO4. The apparent second-order rate constants, k,,, (kca,/Kn,),for the enzymic reaction were calculated = 5.15 mM-' . cni-' [23]. using Fluoritnetric tneastirenzetits
A Shimadzu RF-500 spectrofluorophotometer was used
Table 1. Kinetic purarneters for C 0 2 hydration catulyzed by various mitarits of human carhonic cmlij3drase I1 in 50 n i M I ,2-ditnetli.r.liniidazolc hi4fer ut p H 8.8 and 25 'C. The ionic strength was maintained at 0.1 M with NaZSO4.The oncletter amino acid code i s used to designate mutants so that, for example. 1164K stands for (His64 -, Lyslcarbonic anhydrase 11. Data for unmodified isoenzyme I1 and [His64 -, Lyslcarbonic anhydrase I1 from [6]. Data for bovine isoenzyme 111 (BCA 111) were measured in 10 mM buffer, pH 8.9. and are from [25]. Parameter values and standard deviations were obtained by a non-linear least-squares procedure
mM
S-'
Unmodified H64K N67R L198F V207I H64K, N67R H64K. L198F N67R. L198F V2071, L198F H64K. N67R, L198F H64K. N67R. L198F. V2071 BCA 111
76 49 61 88 37 2.0 20 34 54 8.2
f8 &4 f2
11
f1
M-1. s- I
+
k8 f3 f 0.1 f1 f3 f3 f 0.4
8.0 1.7 6.3 k 1.0 6.3 f 0.5 6.7 f 1.4 8.1 f-1.5 1.7 f 0.2 5.1 k 0.5 10 f 2 9.9 f 1.0 5.6 k 0.7
8.8 f 1.3 15
0.6
25,
0
9.5 7.8 9.7 13 4.6 1.2 3.8 3.3 5.4 1.5 1.2 0.04
I
3
6
9 vo AS1
12
15
18
(5.')
Fig. 1. Eadie-Hofstee plots qf rates of' C 0 2 hydration cataljwd hv [His64 -, Lvs,Am67 + Arglcarhonic anhydrase I1 in 1 ,-7-diinethyliniidrrzole buffers at pH 8.8 and 25°C. The ionic strength was kept at 0.1 M with Na2S04. Indicator was 20 pM M-cresol purple. Enzyme concentration. 1.23 pM. ( 0 )50 mM buffer; ( A )5 mM buffer
to study the binding of dansylamide (5-dimethylamino-naph-
thalene-1-sulfonamide) to the enzyme [24]. A stock solution contained 1 mM dansylamide in 10 mM HCl. Experiments were performed at 25°C and pH 8.5 using 20 mM Tris/HzS04. triple mutant with Lys64, Arg67 and Phe198, and a quadruple The ionic strength was kept at 0.1 M by addition of NaZSO4. mutant containing all four changes, were prepared. Rates of COz hydration catalyzed by these 10 mutants The sample was excited at 320 nm and the emission spectrum were measured at 25 ' C in 50 mM 1,2-dimethylimidazole/ recorded. H z S 0 4 buffer, pH 8.8. Michaelis-Menten behavior was observed in all cases. Kinetic parameters are given in Table 1. The double mutant with Lys64 and Arg67 was selected for RESULTS further kinetic studies. As shown in Fig. 1, the COz hydration Carbonic-anhydrase-111-specificresidues were introduced activity depends strongly on the concentration of 1.2into four sequence positions of human isoenzyme 11. In ad- dimethylimidazole buffer, and a K , value of about 45 mM dition to the single mutants, His64 + Lys, Asn67 + Arg, with respect to buffer can be estimated. However. in Mops/ Leu198 + Phe and Val207 ---t Ile, four double mutants, one NaOH buffers at pH 7.2 ( 5 - 50 mM) and Mes/NaOH buffers
419 Table 2. C'ornpririson of kiiir.ticpriratireters.for C'Oz Iiydrcition c t i r a l j ~ ~ e t l DISCUSSION by bovine curhonic orrhjdrase [ [ I ( BCA I I l i cind the double rniiim~tqf hutnmi I,roc~ri:j~inc~ I1 wit17 Ly.64 and Arg67 ( H 6 4 K , N 6 7 R ) in 10 inM The replacement of His64, Asn67, Leu198 and Val207 Mops hufrer. p H 7.2, and I 0 niM Mes hufyer, p H 6.3. in the active site of human carbonic anhydrase I1 with the Temperature. 25 'C. Data for bovine carbonic anhydrase 111 arc from isoenzyme-111-specific residues Lys64, Arg67, Phel98 and [15]
lle207 has not resulted in the appearance of kinetic and inhibitor-binding properties like those characterizing isoenzyme I l l . Enzyme Buffer x k,,, K, 10 Thus, we have not found any evidence for a key residue or set x kcailk, of residues responsible for the low C 0 2 hydration activity, low M - 1 , s - l esterase activity, low-activity-linked pK, and low sulfonamide SC1 in M affinity of carbonic anhydrase 111. Specifically, our results H64K. N67R Mops 1.8 f 0.05 0.8 f 0.2 2.1 BCA I I L Mops 2.7 78 0.34 suggest that the low pK, is not due to electrostatic interactions between the metal-bound O H - ion and the side chains of Mes H64K, N67R 0.44 0.01 1.0 f 0.2 0.44 BCA I l l Mes 1.7 5.5 0.31 Lys64 and Arg67 as previously hypothesized [9] and, furthermore, that the low sulfonamide affinity is not simply due to the presence of the bulky side chains of Arg67, Phe198 and Ile207. Table 3. Driici jiotn tnecisiiretiiiwts qf the y H depcnrlence of 4Indeed, the single mutations His64 + Lys, Asn67 + Arg nifroplretij.l cic'rtatc hydrolysis catalyzed hy various mutants Iiumun and Leu198 ---* Phe hardly result in any significant changes of c'orhonic anlijdraw II, rind data j i o m spc~ctrojlrrorinietriciitrtition.s of kca,/K,,,for C 0 2 hydration (Table l), which is the most relevant thc tnututits ii.ith tiunsylamide Temperature. 25 C. Mutants arc designated as in Table I . Values of' kinetic parameter in this context since it is buffer independent (cf. Fig. 1 and [9, lo]), whereas the dependence of k,,, on the k:'n"," and pk', were obtained by fitting the pH/rate profiles to simple litration curves. Data on esterasc activities of unmodified isoenzyme concentration and the chemical structure of the buffer might be different for different enzyme variants [26]. The effect of I1 and His64 -+ Lys are from [6].Values of the dissociation constants. &. and standard deviations were obtained by a least-squares pro- the single mutation Val207 + Ile on k,,,/Km is probably significedure assuming a 1 : 1 enzyme/dansylarnide stoichiometry. Ire,is the cant, but only twofold. With the exception of the combination relative lliiorcscence intensity of the enzyme-inhibitor complex a t Leu198 Phe and Val207 4 Ile, the effects of the various 440 nm. where contributions from unbound dansylamide were neglimutations on the apparent free energy of activation are not gible additive. The lowest COz hydration activities are found for mutants having Lys64 as well as Arg67, but it is striking that even for the quadruple mutant, k,,,/K,, is only a factor of eight smaller than for unmodified isoenzyme I1 at pH 8.8. Thus, m M C ' . sC1 M Yo although carbonic anhydrase I1 has evolved to be an almost Unmodified 2.8 6.8 3.3 _+ 0.4 100 perfect enzyme (cf. [27]), its active site i s not particularly H64 K 6.8 4.0 f 0.2 2.9 I 40 delicate, but can tolerate rather extensive modifications withN67R 6.8 6.2 f 0.8 0.8 40 L19XF 1.2 6.8 120 1.6 f 0.2 out drastic effects on the catalytic power. V2071 I .4 6.4 1 . 1 f 0.1 I x0 While it has previously been found that some mutations H64K, N67R 0.5 6.6 3.7 f 0.9 22 of human isoenzyme I1 give rise to enhanced esterase activities H64K. LlYXF 2.0 7.0 1.9 f 0.5 160 [28, 291, all mutants investigated in the present study (except N67R, LlOXF 6.7 1.4 f 0.4 0.8 85 [His64 ---* Lyslisoenzyme 11) have decreased activities (Table V2071, LlYXF 0.7 6.7 0.9 f 0.3 120 3). As in the case of C 0 2 hydration, most of the effects of the H64K. N67R. L198F 6.3 1.4 f 0.4 0.8 XO various alterations are not strictly additive. An analysis of H64K. N67R, L19XF. the effect of a particular alteration of the active sites of the V2071 120 0.4 6.2 2.3 f 0.6 - unmodified and variously modified forms of the enzyme, respectively, indicates that the largest relative decrease of k;,; (Table 3) occurs when Asn67 i s replaced by Arg. This alterat pH 6.3 (10 - 25 mM), no significant dependence on buffer ation changes kynz to 0.4 f 0.2 (means & SD) of its previous concentration was observed (data not shown). Kinetic par- value. The effect of all four alterations is a modest sevenfold ameters obtained for this double mutant in Mops and Mes decrease in esterase activity, a change of similar magnitude as buffers are compared with data for bovine isoenzyme 111 in observed for the C 0 2 hydration activity. The pKa derived from the pH/rate profiles of ester hydrolyTable 2. The pH dependences of rates of 4-nitrophenyl acetate hy- sis is not significantly affected by the single mutations His64 + drolysis catalyzed by the mutants were measured at 25'.C. In Lys, Asn67 + Arg and Leu198 + Phe. whereas the mutation all cases. the activities increased with pH. The pH/rate profiles Val207 ---* Ile results in a decrease of the pK, value by 0.4. The were generally reasonably well described as simple titration quadruple mutant has the lowest pK, value of the investigated curves. Maximal values of the catalytic, second-order rate con- variants, but it is still far from the pK, near 5 characterizing isoenzyme 111 [5].The effects of the individual alterations are stants, k,,,. and estimated pK, values are given in Table 3. Table 3 also shows results from spectrofluorimetric ti- not strictly additive in this case either and, thus, one cannot trations of the mutants with the fluorescent sulfonamide in- assign the overall pKa change of 0.6 in the quadruple mutant hibitor, dansylamide. Estimated dissociation constants. Kd. to any individual amino acid replacement. The C 0 2 hydration activity of the double mutant with and relative fluorescence intensities of the enzyme-inhibitor complexes are given. The wavelength of maximal fluorescence Lys64 and Arg67 has a similar magnitude as that of bovine emission was relatively constant. It varied from 458 nm for carbonic anhydrase 111 in Mes/NaOH buffer at pH 6.3 (Table Val207 + Ile to 465 nm for the double mutant with Arg67 2). This reflects a pronounced pH dependence of k , , , / K , for and Phe198 and the triple mutant. the double mutant. The data in Tables 1 and 2 indicate a pK;, --f
420 of about 7.8 rather than the value of 6.6 obtained for the esterase activity (Table 3). Since there is strong evidence from previous measurements on Co(I1)-substituted carbonic anhydrases that the titration of the metal-bound water molecule coincides with the esterase pH/rate profile [9], these results suggest that the pK, derived from the pH dependence of k,,,/K,,, for COz hydration is perturbed by kinetic factors. This is in apparent disagreement with the current mechanism hypothesis [9, 101, but the consequences of this observation for the interpretation of available kinetic data are not yet clear. None of the various mutations has a drastic effect on the affinity of the sulfonamide inhibitor, dansylamide (Table 3). Only the single mutant with Arg67 binds the inhibitor significantly more weakly than unmodified isoenzyme 11. Although there are considerable uncertainties in the values of the dissociation constants, K d . it appears, however, that the average effects of the alterations His64 -+ Lys and Asn67 + Arg are small, while the alterations Leu198 + Phe and Val207 + Ile lead to somewhat stronger binding. Since non-polar interactions with active-site groups are known to be important for sulfonamide affinity [3], these results are not surprising, but they show that the increased bulk of residues 198 and 207 within the active-site framework of isoenzyme 11 has not resulted in steric hindrance of inhibitor binding. While the alteration Asn67 + Arg consistently leads to a decrease of the fluorescence intensity of the dansylamide complex, the increase of hydrophobic bulk resulting from the alterations Leu198 -+ Phe and Val207 --t Ile mostly gives rise to an increased fluorescence intensity which further emphasizes the non-polar nature of the inhibitor environment in the complex. I n contrast to our results, Silverman, Tu and LoGrasso (personal communication) have found that the replacement of Phe198 in human carbonic anhydrase 111 with Leu results in enhanced COz hydration activity, increased activity-linked pK, and enhanced sulfonamide binding towards the values characterizing isoenzyme 11. These findings strongly point at a role for Phe198 as a determinant of the isoenzyme-specific properties of carbonic anhydrase 111. We can only speculate as to why our results d o not mirror those of Silverman, Tu and Lo Grasso. Such speculations would be based on additional differences between the active site regions of the two isoenzymes. For example, Lys64 in bovine isoenzyme I11 presumably interacts with Glu4 via a salt link [ll], which cannot be present in [His64 + Lyslisoenzyme I1 because residue 4 is His [9]. In isoenzyme 111, the side chain of Arg67 points into the active site and the guanidinium group hydrogen bonds with some ordered water molecules [ll]. In [Am67 + Arglisoenzyme 11, the orientation of the side chain of Arg67 is unknown, but it might be pointing out from the active site forming a salt link with Glu69. Residue 69 is Val in isoenzyme 111. I n isoenzyme 111. the edge of the benzene ring of Phe198 is only 0.33 nm from the phenolic oxygen atom of Tyrl31 [Ill. which corresponds to Phel31 in isoenzyme 11. These considerations indicate that the overall active-site framework of isoenzyme 11 might be sufficiently different from that of isoenzyme Ill, so that the isoenzyme-Ill-specific residues, Lys64, Arg67, Phe198 and IIe207 have somewhat different orientations and/or different interactions with other residues in the two environments. Thus. it seems as if the fine tuning of the catalytic activity of carbonic anhydrase hinges on quite subtle features of the active-site structure.
We are grateful to Dr Carol A. Fierke for the generous gift of the plasmid pACA and to Ms Katarina Wallgren for excellent technical assistance. Financial support was obtained from the Swedish Natural Science Research Council (K 291 1).
REFERENCES 1. Tashian. R. E. (1989) Bioessays 10, 186-192. 2. Jeffery, S., Carter, N. D. & Smith. A. (1986) J . Histoehem. Cytochem. 34, 513-516. 3. Lindskog. S. & Wistrand, P. J. (1989) in Design ofenzyme inhibitors us drugs (Sandler, M. & Smith, H. J.. eds) pp. 698-723. Oxford University Press. Oxford. 4. Engberg, P.. Millqvist, E., Pohl. G. & Lindskog, S. (1985) Arch. Biochcm. Biophyx 241, 628-638. 5. Ren, X., Sandstrom, A . & Lindskog, S. (1988) E w . J . Biochem. 173, 73-78. 6. Forsman, C., Behravan, G., Jonsson, B.-H., Liang, Z.. Lindskog, S., Ren, X., Sandstrom. J. & Wallgren, K. (1988) FEBS Left. 22Y, 360- 362. 7. Tu, C., Thomas, H. G., Wynns, G. C. & Silverman, D. N. (1986) J . B i d . Chrm. -361, 10100-10103. 8. King, R. W., Garg, L. C., Huckson, J. & Maren, T. H. (1974) Mol. Pharmncol. 10. 335 -343. 9. Lindskog, S. (1983) in Zinc rnzyme.r (Spiro, T. G., ed) pp. 78121, Wiley. New York. 10. Silverman, D. N. & Lindskog, S. (1988) Aec. C'hem. Res. 21. 3036. 11. Eriksson. E. A. (1 988) Ph.D. thesis. Uppsala University. 12. Forsman, C., Behrdvan, G., Osterman, A. & Jonsson. B.-H. (1988) A c f a Cliem. Scand. B 4 2 . 314--318. 13. Stanssens, P., Opsomer, C., McKeown, Y . M., Kramer, W., Zabeau, M. & Fritz. H.-J. (1989) Nucleic Acids Res. 17.4441 4454. 14. Kunkel, T. A . (1985) Proc. Nail Acad. Sci. USA 82.488-492. 15. Murakami. H.. Marelich, G. P., Grubb. J. H., Kyle, J. W. & Sly, W. S. (1987) Genomics I , 159-166. 16. Rosenberg, A. H., Lade, B. N.. Chui. D.-S., Lin, S.-W., Dunn. J. J. & Studier, F. W. (1987) Gene ( A m s i . ) 56. 125-135. 17. Sanger, F.. Nicklen. S. & Coulson. A. R . (1977) Proc. Nut1 Acad. Sci. USA 74, 5463 - 5467. 18. Khalifah, R. G., Strader. D. J., Bryant. S. I l . & Gibson. S. M. (1977) Biochemistrj316. 2241 - 2247. 19. Nyman, P. 0. & Lindskog, S. (1964) Biochim. Biophj.s. Acta 85. 141 - 151. 20. Henderson, L. E., Henriksson, D. & Nyman, P. 0.(1971) J . Biol. Chetn. 251. 5457 -- 5463. 21. Khalifah. R. G. (1971) J . B i d . Chem. -746, 2561 -2573. 22. Steiner, H., Jonsson, B.-H. & Lindskog, S. (1975) Eur. J . Bioclzem. S9,253 - 259. 23. Thorslund. A. & Lindskog, S. (1967) Eirr. J . Biochmi. 3 , 117123. 24. Chen, R. F. & Kernohan, J. C. (1967) J . Biol. Chem. 242, 58135823. 25. Ren, X . , Jonsson, B.-H.. Millqvist. E. & Lindskog. S. (1988) Biochim. Biop/zjv. Acta 953, 79- 85. 26. Tu, C. K.. Silverman, D. N.. Forsman, C.. Jonsson, B.-H. & Lindskog, S. (1989) Biochemistry 28, 7913-7918. 27. Lindskog, S. (1985) in Gas enzynzology (Degn. H., Cox, R. P. & Toftlund, H.. eds) pp. 121 -133. Reidel. Dordreclit. 28. Behravan, G., Jonsson, B.-H. & Lindskog, S. (1990) Eur. J . Biocheni. 190, 351 -357. 29. Behravan, G.. Jonasson, P.. Jonsson, B.-H. & Lindskog. S . (1991) Eur. J.. Biochem. 198, 589 - 592.
Some properties of site-specific mutants of human carbonic anhydrase I1 having active-site residues characterizing carbonic anhydrase 111 Xilin R EN , Bengt-Harald JONSSON and Sven LINDSKOG Avdelningen for biokemi, Urnei Universitet, Sweden (Received April 4/June 27, 1991) - EJB 91 0442
Four amino acid residues, His64, Asn67, Leu198 and Va1207, in the active site of human carbonic anhydrase 11. have been replaced by Lys64, Arg67, Phe198 and Ile207, which are characteristic for the muscle-specific, lowactivity isoenzyme form, carbonic anhydrase 111. The aim of the investigation has been to test if any of these residues, o r a combination of them, is important for the low COz hydration activity, low esterase activity. low pK, for the pH/rate profile and low affinity for sulfonamide inhibitors characterizing carbonic anhydrases 111. However, n o evidence for such critical roles was found. A combination of Lys64 and Arg67 appears to result in a decrease in C 0 2 hydration activity, but even the quadruple mutant having all four changes is only eight times less active (kcat/Km) than unmodified isoenzyme 11, in contrast to isoenzyme 111 which is nearly 300 times less active than isoenzyme 11. The 4-nitrophenyl acetate hydrolase activity of the quadruple mutant is sevenfold lower than that of unmodified isoenzyme 11, while the active site of isoenzyme 111 hardly catalyzes the hydrolysis of this ester at all. The pK, controlling the esterase activity of the quadruple mutant is 6.2, which should be compared to a value of 6.8 for unmodified isoeniyme 11, and about 5 for isoenzyme 111. While isoenzyme 111 binds sulfonamide inhibitors lo3 - lo4 times less strongly than isoenzyme 11, only [Asn-67 + Arglisoenzyme I1 shows a weaker binding of the investigated sulfonamide, dansylamide, but only by a factor of two. Some of the other mutants show enhanced affinities, up to nearly fourfold for the double mutant with Phe198 and Ile207. It is speculated that additional differences between the active sites of isoenzyme I1 and 111 might be important for the precise orientations and interactions of the side chains of isoenzyme-111-specific amino acid residues.
The zinc-containing enzyme carbonic anhydrase, which catalyzes the simple, reversible reaction, CO, + H 2 0 HCO, H ' , occurs in several, genetically distinct isoenzyme forms in higher vertebrates [l]. In slow-twitch, red skeletal muscle fibers. isoenzyme 111 is the predominating cytosolic form [2]. It differs in several respects from the high-activity form, isoenzyme 11, which is present in erythrocytes, kidney and many other tissues (31. Thus, the CO, hydration activity (kc,.,/Km) of bovine isoenzyme 111 is only about 0.3% of that of human isoenzyme I1 [4]. The ionization of a zinc-bound water molecule, controlling the catalytic activity, occurs with a pK, near 7 in isoenzyme I1 but the corresponding pK, is probably about 5 in isoenzyme 111 [5]. While human isoenzyme 11 catalyzes the hydrolysis of 4-nitrophenyl acetate quite efficiently with an apparent kc,,/Km of 2.8 x lo3 M . s-' at pH 9 and 25°C [6]. the active site of isoenzyme I l l hardly hydrolyzes this ester at all [7]. Aromatic and certain heterocyclic sulfonamides are potent inhibitors of isoenzyme 11, whereas isoenzyme 111 was originally discovered as a sulfonamide-resistant form of carbonic anhydrase in livers of male rats [XI. These isoenzyme-specific functional differences are presumably related to isoenzyme-specific structural differences, Thus, it has been proposed [9, 101 that the low pK, value
+
_
~
~
_
Corresponrlcncr to S. Lindskog. Avdelningen for biokemi, UmeH Universitet. S-90187 Umel, Sweden E n z j ~ i eCarbonic . anhydrase (EC 4.2. I . 1).
of the activity-linked group in isoenzyme 111 results from electrostatic effects from some characteristic, basic active-site residues, Lys64 (or Arg64 in horse isoenzyme 111). Arg67 and Arg91. The corresponding residues in human isoenzyme I1 are His64, Asn67 and Ile91 [9]. It has also been proposed [ l l ] that the weak sulfonamide affinity characterizing all carbonic anhydrase 111 isoenzymes is due to the fact that the active-site cavity in isoenzyme I11 is smaller than that in isoenzyme 11. The extra bulk in isoenzyme 111 is mainly contributed by the side chains of Phel98 and Ile207, which correspond to Leu1 98 and Val207 in isoenzyme 11. but also by Arg67 Ill]. One aim of our studies of carbonic anhydrase is to test these and other hypotheses concerning the structural basis of the functional differences between isoenzyme forms. In this paper we present some properties of site-specific mutants of human carbonic anhydrase I1 where residues 64.67, 198 and/ or 207 have been replaced by their isoenzyme-I1 I-specific counterparts.
MATERIALS A N D METHODS Enzynr
The numbering system of human carbonic anhydrase 1 is used throughout this paper. In this system, residues 64, 67. 198 and 207 correspond to residues 63, 66, 196 and 205 in the numbering systems of isoenzymes I1 and 111 [9].
41 8 The preparation of [His64 + Lys]isoenzyme I1 has been described previously [6, 121. Other mutants were made by it7 vitro oligonucleotide-directed mutagenesis using singlestranded DNA prepared from the plasmid pACA by use of the helper phage M13K07 [13]. We employed the Muta-Gene system (Bio-Rad) which is based on the method of Kunkel (141. where the mutated strand is built on a parent strand containing U for T. pACA has been constructed in the laboratory of C. Fierke (Duke University) and consists of the human carbonic anhydrase I1 gene [I 51 behind a T7 KNA polymerase promoter [I61 in the pMa/c vector (131. Mutations were verified by dideoxy sequencing [17]. The mutant DNA was transformed into Escherichia coli strain BL21/DE3 [16]. The cells were grown at 37°C in Luria broth medium supplemented with 0.1 mM ZnSO,, 2 mM K 2 H P 0 4and 50 pg/ml ampicillin. Enzyme synthesis was induced by the addition of 0.5 niM isopropyl-thio-/?-D-gahctoside. All enzyme variants were purified by affinity chromatography essentially according to Khalifah et al. [18]. All purified variants showed a single band after SDSjPAGE in the same position as unmodified carbonic anhydrase 11. All variants appeared to be properly folded, as evidenced by their circular dichroism spectra which were very similar to that of unmodified human isoenzyme I1 in the investigated wavelength range, 200- 320 nm. Enzyme concentrations were estimated spectrophotometrically at 280 nm using c = 54.7 mM-' . cni-' [19] based on M , 29300 [20]. Kinetic tneasurernents
Initial rates of COz hydration were measured in a HiTech stopped-flow apparatus at 25 "C by the changing-pHindicator method [21, 221. Buffer/indicator pairs were 1,2dimethylimidazole/nz-cresol purple monitored at 578 nm. l-methylimidazole/4-nitrophenol or Mops/4-nitrophenol monitored at 400 nm and Mes/chlorophenol red monitored at 574 nm. The ionic strength was kept at 0.1 M using NazS04. Initial rates of 4-nitrophenyl acetate hydrolysis were monitored at 348 nm and 2 5 ' C using a substrate concentration of 0.4 mM which is well below the K , for human red cell carbonic anhydrase I1 [23]. Buffers were 50 mM Mes/ NaOH (pH 5.3 - 6.9), 50 mM Hepes/NaOH (pH 6.9- 8.6) or 'I'ris/HzS04 (pH 8.6 - 9.2). The ionic strength was maintained at 0.1 M with Na2SO4. The apparent second-order rate constants, k,,, (kca,/Kn,),for the enzymic reaction were calculated = 5.15 mM-' . cni-' [23]. using Fluoritnetric tneastirenzetits
A Shimadzu RF-500 spectrofluorophotometer was used
Table 1. Kinetic purarneters for C 0 2 hydration catulyzed by various mitarits of human carhonic cmlij3drase I1 in 50 n i M I ,2-ditnetli.r.liniidazolc hi4fer ut p H 8.8 and 25 'C. The ionic strength was maintained at 0.1 M with NaZSO4.The oncletter amino acid code i s used to designate mutants so that, for example. 1164K stands for (His64 -, Lyslcarbonic anhydrase 11. Data for unmodified isoenzyme I1 and [His64 -, Lyslcarbonic anhydrase I1 from [6]. Data for bovine isoenzyme 111 (BCA 111) were measured in 10 mM buffer, pH 8.9. and are from [25]. Parameter values and standard deviations were obtained by a non-linear least-squares procedure
mM
S-'
Unmodified H64K N67R L198F V207I H64K, N67R H64K. L198F N67R. L198F V2071, L198F H64K. N67R, L198F H64K. N67R. L198F. V2071 BCA 111
76 49 61 88 37 2.0 20 34 54 8.2
f8 &4 f2
11
f1
M-1. s- I
+
k8 f3 f 0.1 f1 f3 f3 f 0.4
8.0 1.7 6.3 k 1.0 6.3 f 0.5 6.7 f 1.4 8.1 f-1.5 1.7 f 0.2 5.1 k 0.5 10 f 2 9.9 f 1.0 5.6 k 0.7
8.8 f 1.3 15
0.6
25,
0
9.5 7.8 9.7 13 4.6 1.2 3.8 3.3 5.4 1.5 1.2 0.04
I
3
6
9 vo AS1
12
15
18
(5.')
Fig. 1. Eadie-Hofstee plots qf rates of' C 0 2 hydration cataljwd hv [His64 -, Lvs,Am67 + Arglcarhonic anhydrase I1 in 1 ,-7-diinethyliniidrrzole buffers at pH 8.8 and 25°C. The ionic strength was kept at 0.1 M with Na2S04. Indicator was 20 pM M-cresol purple. Enzyme concentration. 1.23 pM. ( 0 )50 mM buffer; ( A )5 mM buffer
to study the binding of dansylamide (5-dimethylamino-naph-
thalene-1-sulfonamide) to the enzyme [24]. A stock solution contained 1 mM dansylamide in 10 mM HCl. Experiments were performed at 25°C and pH 8.5 using 20 mM Tris/HzS04. triple mutant with Lys64, Arg67 and Phe198, and a quadruple The ionic strength was kept at 0.1 M by addition of NaZSO4. mutant containing all four changes, were prepared. Rates of COz hydration catalyzed by these 10 mutants The sample was excited at 320 nm and the emission spectrum were measured at 25 ' C in 50 mM 1,2-dimethylimidazole/ recorded. H z S 0 4 buffer, pH 8.8. Michaelis-Menten behavior was observed in all cases. Kinetic parameters are given in Table 1. The double mutant with Lys64 and Arg67 was selected for RESULTS further kinetic studies. As shown in Fig. 1, the COz hydration Carbonic-anhydrase-111-specificresidues were introduced activity depends strongly on the concentration of 1.2into four sequence positions of human isoenzyme 11. In ad- dimethylimidazole buffer, and a K , value of about 45 mM dition to the single mutants, His64 + Lys, Asn67 + Arg, with respect to buffer can be estimated. However. in Mops/ Leu198 + Phe and Val207 ---t Ile, four double mutants, one NaOH buffers at pH 7.2 ( 5 - 50 mM) and Mes/NaOH buffers
419 Table 2. C'ornpririson of kiiir.ticpriratireters.for C'Oz Iiydrcition c t i r a l j ~ ~ e t l DISCUSSION by bovine curhonic orrhjdrase [ [ I ( BCA I I l i cind the double rniiim~tqf hutnmi I,roc~ri:j~inc~ I1 wit17 Ly.64 and Arg67 ( H 6 4 K , N 6 7 R ) in 10 inM The replacement of His64, Asn67, Leu198 and Val207 Mops hufrer. p H 7.2, and I 0 niM Mes hufyer, p H 6.3. in the active site of human carbonic anhydrase I1 with the Temperature. 25 'C. Data for bovine carbonic anhydrase 111 arc from isoenzyme-111-specific residues Lys64, Arg67, Phel98 and [15]
lle207 has not resulted in the appearance of kinetic and inhibitor-binding properties like those characterizing isoenzyme I l l . Enzyme Buffer x k,,, K, 10 Thus, we have not found any evidence for a key residue or set x kcailk, of residues responsible for the low C 0 2 hydration activity, low M - 1 , s - l esterase activity, low-activity-linked pK, and low sulfonamide SC1 in M affinity of carbonic anhydrase 111. Specifically, our results H64K. N67R Mops 1.8 f 0.05 0.8 f 0.2 2.1 BCA I I L Mops 2.7 78 0.34 suggest that the low pK, is not due to electrostatic interactions between the metal-bound O H - ion and the side chains of Mes H64K, N67R 0.44 0.01 1.0 f 0.2 0.44 BCA I l l Mes 1.7 5.5 0.31 Lys64 and Arg67 as previously hypothesized [9] and, furthermore, that the low sulfonamide affinity is not simply due to the presence of the bulky side chains of Arg67, Phe198 and Ile207. Table 3. Driici jiotn tnecisiiretiiiwts qf the y H depcnrlence of 4Indeed, the single mutations His64 + Lys, Asn67 + Arg nifroplretij.l cic'rtatc hydrolysis catalyzed hy various mutants Iiumun and Leu198 ---* Phe hardly result in any significant changes of c'orhonic anlijdraw II, rind data j i o m spc~ctrojlrrorinietriciitrtition.s of kca,/K,,,for C 0 2 hydration (Table l), which is the most relevant thc tnututits ii.ith tiunsylamide Temperature. 25 C. Mutants arc designated as in Table I . Values of' kinetic parameter in this context since it is buffer independent (cf. Fig. 1 and [9, lo]), whereas the dependence of k,,, on the k:'n"," and pk', were obtained by fitting the pH/rate profiles to simple litration curves. Data on esterasc activities of unmodified isoenzyme concentration and the chemical structure of the buffer might be different for different enzyme variants [26]. The effect of I1 and His64 -+ Lys are from [6].Values of the dissociation constants. &. and standard deviations were obtained by a least-squares pro- the single mutation Val207 + Ile on k,,,/Km is probably significedure assuming a 1 : 1 enzyme/dansylarnide stoichiometry. Ire,is the cant, but only twofold. With the exception of the combination relative lliiorcscence intensity of the enzyme-inhibitor complex a t Leu198 Phe and Val207 4 Ile, the effects of the various 440 nm. where contributions from unbound dansylamide were neglimutations on the apparent free energy of activation are not gible additive. The lowest COz hydration activities are found for mutants having Lys64 as well as Arg67, but it is striking that even for the quadruple mutant, k,,,/K,, is only a factor of eight smaller than for unmodified isoenzyme I1 at pH 8.8. Thus, m M C ' . sC1 M Yo although carbonic anhydrase I1 has evolved to be an almost Unmodified 2.8 6.8 3.3 _+ 0.4 100 perfect enzyme (cf. [27]), its active site i s not particularly H64 K 6.8 4.0 f 0.2 2.9 I 40 delicate, but can tolerate rather extensive modifications withN67R 6.8 6.2 f 0.8 0.8 40 L19XF 1.2 6.8 120 1.6 f 0.2 out drastic effects on the catalytic power. V2071 I .4 6.4 1 . 1 f 0.1 I x0 While it has previously been found that some mutations H64K, N67R 0.5 6.6 3.7 f 0.9 22 of human isoenzyme I1 give rise to enhanced esterase activities H64K. LlYXF 2.0 7.0 1.9 f 0.5 160 [28, 291, all mutants investigated in the present study (except N67R, LlOXF 6.7 1.4 f 0.4 0.8 85 [His64 ---* Lyslisoenzyme 11) have decreased activities (Table V2071, LlYXF 0.7 6.7 0.9 f 0.3 120 3). As in the case of C 0 2 hydration, most of the effects of the H64K. N67R. L198F 6.3 1.4 f 0.4 0.8 XO various alterations are not strictly additive. An analysis of H64K. N67R, L19XF. the effect of a particular alteration of the active sites of the V2071 120 0.4 6.2 2.3 f 0.6 - unmodified and variously modified forms of the enzyme, respectively, indicates that the largest relative decrease of k;,; (Table 3) occurs when Asn67 i s replaced by Arg. This alterat pH 6.3 (10 - 25 mM), no significant dependence on buffer ation changes kynz to 0.4 f 0.2 (means & SD) of its previous concentration was observed (data not shown). Kinetic par- value. The effect of all four alterations is a modest sevenfold ameters obtained for this double mutant in Mops and Mes decrease in esterase activity, a change of similar magnitude as buffers are compared with data for bovine isoenzyme 111 in observed for the C 0 2 hydration activity. The pKa derived from the pH/rate profiles of ester hydrolyTable 2. The pH dependences of rates of 4-nitrophenyl acetate hy- sis is not significantly affected by the single mutations His64 + drolysis catalyzed by the mutants were measured at 25'.C. In Lys, Asn67 + Arg and Leu198 + Phe. whereas the mutation all cases. the activities increased with pH. The pH/rate profiles Val207 ---* Ile results in a decrease of the pK, value by 0.4. The were generally reasonably well described as simple titration quadruple mutant has the lowest pK, value of the investigated curves. Maximal values of the catalytic, second-order rate con- variants, but it is still far from the pK, near 5 characterizing isoenzyme 111 [5].The effects of the individual alterations are stants, k,,,. and estimated pK, values are given in Table 3. Table 3 also shows results from spectrofluorimetric ti- not strictly additive in this case either and, thus, one cannot trations of the mutants with the fluorescent sulfonamide in- assign the overall pKa change of 0.6 in the quadruple mutant hibitor, dansylamide. Estimated dissociation constants. Kd. to any individual amino acid replacement. The C 0 2 hydration activity of the double mutant with and relative fluorescence intensities of the enzyme-inhibitor complexes are given. The wavelength of maximal fluorescence Lys64 and Arg67 has a similar magnitude as that of bovine emission was relatively constant. It varied from 458 nm for carbonic anhydrase 111 in Mes/NaOH buffer at pH 6.3 (Table Val207 + Ile to 465 nm for the double mutant with Arg67 2). This reflects a pronounced pH dependence of k , , , / K , for and Phe198 and the triple mutant. the double mutant. The data in Tables 1 and 2 indicate a pK;, --f
420 of about 7.8 rather than the value of 6.6 obtained for the esterase activity (Table 3). Since there is strong evidence from previous measurements on Co(I1)-substituted carbonic anhydrases that the titration of the metal-bound water molecule coincides with the esterase pH/rate profile [9], these results suggest that the pK, derived from the pH dependence of k,,,/K,,, for COz hydration is perturbed by kinetic factors. This is in apparent disagreement with the current mechanism hypothesis [9, 101, but the consequences of this observation for the interpretation of available kinetic data are not yet clear. None of the various mutations has a drastic effect on the affinity of the sulfonamide inhibitor, dansylamide (Table 3). Only the single mutant with Arg67 binds the inhibitor significantly more weakly than unmodified isoenzyme 11. Although there are considerable uncertainties in the values of the dissociation constants, K d . it appears, however, that the average effects of the alterations His64 -+ Lys and Asn67 + Arg are small, while the alterations Leu198 + Phe and Val207 + Ile lead to somewhat stronger binding. Since non-polar interactions with active-site groups are known to be important for sulfonamide affinity [3], these results are not surprising, but they show that the increased bulk of residues 198 and 207 within the active-site framework of isoenzyme 11 has not resulted in steric hindrance of inhibitor binding. While the alteration Asn67 + Arg consistently leads to a decrease of the fluorescence intensity of the dansylamide complex, the increase of hydrophobic bulk resulting from the alterations Leu198 -+ Phe and Val207 --t Ile mostly gives rise to an increased fluorescence intensity which further emphasizes the non-polar nature of the inhibitor environment in the complex. I n contrast to our results, Silverman, Tu and LoGrasso (personal communication) have found that the replacement of Phe198 in human carbonic anhydrase 111 with Leu results in enhanced COz hydration activity, increased activity-linked pK, and enhanced sulfonamide binding towards the values characterizing isoenzyme 11. These findings strongly point at a role for Phe198 as a determinant of the isoenzyme-specific properties of carbonic anhydrase 111. We can only speculate as to why our results d o not mirror those of Silverman, Tu and Lo Grasso. Such speculations would be based on additional differences between the active site regions of the two isoenzymes. For example, Lys64 in bovine isoenzyme I11 presumably interacts with Glu4 via a salt link [ll], which cannot be present in [His64 + Lyslisoenzyme I1 because residue 4 is His [9]. In isoenzyme 111, the side chain of Arg67 points into the active site and the guanidinium group hydrogen bonds with some ordered water molecules [ll]. In [Am67 + Arglisoenzyme 11, the orientation of the side chain of Arg67 is unknown, but it might be pointing out from the active site forming a salt link with Glu69. Residue 69 is Val in isoenzyme 111. I n isoenzyme 111. the edge of the benzene ring of Phe198 is only 0.33 nm from the phenolic oxygen atom of Tyrl31 [Ill. which corresponds to Phel31 in isoenzyme 11. These considerations indicate that the overall active-site framework of isoenzyme 11 might be sufficiently different from that of isoenzyme Ill, so that the isoenzyme-Ill-specific residues, Lys64, Arg67, Phe198 and IIe207 have somewhat different orientations and/or different interactions with other residues in the two environments. Thus. it seems as if the fine tuning of the catalytic activity of carbonic anhydrase hinges on quite subtle features of the active-site structure.
We are grateful to Dr Carol A. Fierke for the generous gift of the plasmid pACA and to Ms Katarina Wallgren for excellent technical assistance. Financial support was obtained from the Swedish Natural Science Research Council (K 291 1).
REFERENCES 1. Tashian. R. E. (1989) Bioessays 10, 186-192. 2. Jeffery, S., Carter, N. D. & Smith. A. (1986) J . Histoehem. Cytochem. 34, 513-516. 3. Lindskog. S. & Wistrand, P. J. (1989) in Design ofenzyme inhibitors us drugs (Sandler, M. & Smith, H. J.. eds) pp. 698-723. Oxford University Press. Oxford. 4. Engberg, P.. Millqvist, E., Pohl. G. & Lindskog, S. (1985) Arch. Biochcm. Biophyx 241, 628-638. 5. Ren, X., Sandstrom, A . & Lindskog, S. (1988) E w . J . Biochem. 173, 73-78. 6. Forsman, C., Behravan, G., Jonsson, B.-H., Liang, Z.. Lindskog, S., Ren, X., Sandstrom. J. & Wallgren, K. (1988) FEBS Left. 22Y, 360- 362. 7. Tu, C., Thomas, H. G., Wynns, G. C. & Silverman, D. N. (1986) J . B i d . Chrm. -361, 10100-10103. 8. King, R. W., Garg, L. C., Huckson, J. & Maren, T. H. (1974) Mol. Pharmncol. 10. 335 -343. 9. Lindskog, S. (1983) in Zinc rnzyme.r (Spiro, T. G., ed) pp. 78121, Wiley. New York. 10. Silverman, D. N. & Lindskog, S. (1988) Aec. C'hem. Res. 21. 3036. 11. Eriksson. E. A. (1 988) Ph.D. thesis. Uppsala University. 12. Forsman, C., Behrdvan, G., Osterman, A. & Jonsson. B.-H. (1988) A c f a Cliem. Scand. B 4 2 . 314--318. 13. Stanssens, P., Opsomer, C., McKeown, Y . M., Kramer, W., Zabeau, M. & Fritz. H.-J. (1989) Nucleic Acids Res. 17.4441 4454. 14. Kunkel, T. A . (1985) Proc. Nail Acad. Sci. USA 82.488-492. 15. Murakami. H.. Marelich, G. P., Grubb. J. H., Kyle, J. W. & Sly, W. S. (1987) Genomics I , 159-166. 16. Rosenberg, A. H., Lade, B. N.. Chui. D.-S., Lin, S.-W., Dunn. J. J. & Studier, F. W. (1987) Gene ( A m s i . ) 56. 125-135. 17. Sanger, F.. Nicklen. S. & Coulson. A. R . (1977) Proc. Nut1 Acad. Sci. USA 74, 5463 - 5467. 18. Khalifah, R. G., Strader. D. J., Bryant. S. I l . & Gibson. S. M. (1977) Biochemistrj316. 2241 - 2247. 19. Nyman, P. 0. & Lindskog, S. (1964) Biochim. Biophj.s. Acta 85. 141 - 151. 20. Henderson, L. E., Henriksson, D. & Nyman, P. 0.(1971) J . Biol. Chetn. 251. 5457 -- 5463. 21. Khalifah. R. G. (1971) J . B i d . Chem. -746, 2561 -2573. 22. Steiner, H., Jonsson, B.-H. & Lindskog, S. (1975) Eur. J . Bioclzem. S9,253 - 259. 23. Thorslund. A. & Lindskog, S. (1967) Eirr. J . Biochmi. 3 , 117123. 24. Chen, R. F. & Kernohan, J. C. (1967) J . Biol. Chem. 242, 58135823. 25. Ren, X . , Jonsson, B.-H.. Millqvist. E. & Lindskog. S. (1988) Biochim. Biop/zjv. Acta 953, 79- 85. 26. Tu, C. K.. Silverman, D. N.. Forsman, C.. Jonsson, B.-H. & Lindskog, S. (1989) Biochemistry 28, 7913-7918. 27. Lindskog, S. (1985) in Gas enzynzology (Degn. H., Cox, R. P. & Toftlund, H.. eds) pp. 121 -133. Reidel. Dordreclit. 28. Behravan, G., Jonsson, B.-H. & Lindskog, S. (1990) Eur. J . Biocheni. 190, 351 -357. 29. Behravan, G.. Jonasson, P.. Jonsson, B.-H. & Lindskog. S . (1991) Eur. J.. Biochem. 198, 589 - 592.
