Characterization of calcium-binding sites in development-specific

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 3, Issue of January 25, pp. 1199-1203,1988 Printed in U. S.A.

Characterization of Calcium-binding Sites in Development-specific Protein S of Myxococcus xanthus Using Site-specific Mutagenesis*

(Received for publication, August 31, 1987)

Martin TeintzeS, Masayori InouyeQ, and Sumiko InouyeQ From the $Department of Medicine, Division of Digestive Diseases and the Department of Cell Biology and Anatomy, Cornel1 University Medical College, New York, New York 10021 and the §Department of Biochemistry, University of Medicine & Dentistry of New Jersey, Robert Wood Johnson Medical School at Rutgers, Piscataway, New Jersey 08854

Protein S, the most abundant protein synthesized during development of the Gram-negative bacterium Myxococcu~ xanthus, assembles on the surface of the spores. It can be dissociated from the spores using divalent metal chelators and will reassemble on the spores in the presence of calcium. The amino acid sequence of protein S contains regions which have homology to the calcium-binding sites of calmodulin. Protein S was found to bind 2 mol of calcium/mol of protein with Kd values of 27 and 76 WM. Using oligonucleotide-directed site-specific mutagenesis, the gene coding for protein S was changed in each of two regions of homology to calmodulin (Ser4’ + Arg, Ser12’ + Arg), and a double mutant was also constructed. Each mutant gene was then transduced into the genome of a M. xanthus strain from which the wild-type genes had been deleted. All three mutants produced protein S normally during development. One of the mutants (Ser12’ -., Arg) had normal amounts of protein S on its spores, whereas the other (Ser4’ -., Arg) bound much less and the double mutant had virtually none. Analysis of the calcium binding affinities of the purified proteins showed that [Arg4’]protein S and [Arg4’, Arg’2’]pr~- tein S did not bind detectable quantities of calcium, whereas [Arg12’]protein S bound less calcium than the wild-type protein and with a reduced affinity.

Myxococcus xanthus is a rod-shaped Gram-negative bacterium that undergoes a unique developmental cycle: upon starvation on a solid surface, the cells aggregate to form fruiting bodies, and then some of them convert to round or ovoid spores while most of the cells lyse (for reviews, see Ref. 1). During differentiation, the most striking change in the pattern of protein synthesis is the appearance of protein S (2, 3), whose synthesis is induced early in development and increases until it reaches 15% of total protein synthesis a t the stage of mound formation (3). It accumulates in the developing cells until the onset of sporulation, after which most of it is found assembled on the surface of the myxospores and a smaller amount inside (3-5). Protein S can be removed from the surface of the spores by extraction with 1 N NaCl or 10 mM EDTA and can be reassembled on the spores by adding a 10 mM excess of Ca2+ ion or removing the NaCl by dialysis. Protein S has been purified and crystallized (6), and a prelim- inary x-ray crystallographic study has been performed (7).

* This work was supported in part by Research Grant GM-26843 (to S. I.) and National Research Service Award GM-09683 (to M. T.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The gene has been cloned, and it was found that there are two very homologous genes (designated ops and tps) in the same orientation separated by a short spacer region (8, 9). Comparisons of the amino acid sequences predicted from the DNA sequences with that of protein S have shown that only the downstream (tps) gene codes for protein S (8, 10). The ops gene is induced very late in development, after sporulation; and its product is found only inside the spores (4, 5). In addition to the homology between these two genes, there is also striking internal homology in the protein sequences; they can be divided into four homologous domains, with domains 1 and 3 as well as domains 2 and 4 having particularly extensive homologies (8). This pattern of internal homology is also found in calmodulin and the y-crystallins, two classes of proteins that have been identified only in eukaryotes. In fact, both bovine brain calmodulin and bovine eye lens y- crystallin I1 have some homology to protein S.

The regions of domains 1 and 3 of protein S shown in Fig. 1 are homologous to the proposed Ca2+-binding sequences of bovine brain calmodulin (8); the corresponding regions of domains 2 and 4 have lesser homology (4, 8). There are other similarities between calmodulin and protein S. (i) Both are acidic proteins with a PI of 4.2 for calmodulin and 4.5 for protein S (6). (ii) Both calmodulin and protein S are heat- stable proteins (6). (iii) The contents of hydrophobic amino acids are similar: 37% for bovine brain calmodulin (11) and 39% for protein S (8). (iv) Proteolysis with trypsin cleaves both proteins near the center of their amino acid sequence (6, 13), in a region that forms a long a-helical connection between domains 1 and 2 and domains 3 and 4 in calmodulin (14) and has a similar role in a proposed structure for protein S (12). On the other hand, this proposed structure, which is based on the extensive homology between protein S and y-crystallin (12), is comprised almost entirely of P-pleated sheets, as predicted by the circular dichroism spectrum of protein S (6). Protein S therefore does not have the helix-loop-helix conformation found in calmodulin (14), but the proposed structure based on y-crystallin does show possible calcium-binding sites involving the sequence from G l P to Ser4” and its equivalents in the other homologous domains (12).

We have therefore determined the number of calcium- binding sites in protein S and studied the effects of mutations in the sequences predicted to be involved in calcium binding.

MATERIALS AND METHODS

Bacteria and Plasmids-M. ranthus FB (DZF1) was used as the wild-type strain. The three mutant strains were derived from the ops- tps double deletion mutant described previously (15, 16). The kanamycin resistance gene (17) which had been inserted in place of the ops and tps genes (15) was first replaced by a streptomycin gene. This was done using the same plasmid TK103 used previously (E), which contains portions of the M. xanthus genome upstream and down-

1199

1200 Protein S Calcium-binding Sites

32 46 Protein S-Domain 1: -Leu-Cly-Ile-Clu- -A.n-Aan-Thr-Ile-Ser-Ser-VAl-Lya-Val-Pro-Pro-

FIG. 1. Homology between pro- V tein S and calmodulin. The residues Ar8 - involved in binding to the Ca2+ ion in 121 135 calmodulin are ~ ~ ~ ~ ~ h & ~ show Protein S-Domain 3 : -Tyr-Gly-Ile-Asp- -Aan-Aan-Thr-Ile-Ser-Ser-Val-Lys-Pro-Cln-Cly-

I

the substitutions made in the Drotein S .l sequence by site-specific mutagenesis. arg

Bovine Calmodulin : - ~ ~ ~ L y ~ ~ C l ~ ~ C l y ~ I l e ~ ~ = - -Lyn@Leu-Cly- 33

A

B

EcoRl

EcoRl

FIG. 2. A, construction of double mutant protein S (tps) gene. The locations of the mutations are marked with an x. B, plasmid used for P1 transductions of the mutant tps genes into M. ranthus (see “Materials and Methods”).

stream of the ops-tps gene region, except that the 1.8-kb’ HindIII- BamHI fragment between them carrying the kanamycin resistance gene was replaced by a 2.3-kb HindIII-BamHI fragment containing the streptomycin gene from RP4 (18). This fragment was obtained by cloning a 2.3-kb HincII fragment from RP4 (18) into pUC19 (19) and excising it with Hind11 and BamHI. The resulting plasmid, pMX75, was then used to replace the kanamycin gene in the ops-tps deletion mutant by the method of specialized P1 phage transduction (15-17). Streptomycin-resistant transductants of M. ranthus were screened by colony hybridization using the nick-translated streptomycin gene as a probe, and positives (resulting from homologous recombination) were screened for kanamycin sensitivity (the result of double crossover).

Plasmids pMX61 and pMX62 (see Fig. 2 A ) were constructed by inserting the 2.2-kb BamHI fragment containing the tps gene (8, 9) into the BamHI site of pBR322 and changing a single base pair using oligonucleotide-directed site-specific mutagenesis as described below. Plasmid pMX63, containing the tps gene with both mutations, was constructed by combining fragments of pMX61 and pMX62 as shown in Fig. 2 A . The 2.2-kb BamHI fragments from plasmids pMX61, -62, and -63 containing the mutated tps genes were then cut out and inserted into the unique BamHI site of a 22.5-kb vector consisting of a 12.6-kb EcoRI fragment of myxophage Mx8 containing the attach- ment site (attP) homologous to the attB region of the M. ranthus chromosome (20), a 5.4-kb KpnI-EcoRI fragment (Plinc) encoding P1-specific incompatibility (17), and the EcoRI-KpnI fragment of pUC19 (19) containing a 1.8-kb HindIII-BamHI insert carrying the kanamycin gene from Tn5 (15-17), generating plasmids pMX661, -662, and -663, respectively (see Fig. 2B). These plasmids were then used to introduce each mutant tps gene together with the kanamycin resistance gene into the genome of the Aops-tps strain of M. ranthus at the attB site by the method of specialized P1 transduction described previously (15-17).

M. ranthus was grown in CYE medium (1) containing 40 pg/ml kanamycin in the case of the mutant strains, and development was induced on CF agar as described previously (2, 3).

Site-specific Mutagenesis-The two mutations in the t I j s gene (Ser“ + Arg and Serlzg + Arg) were made using the same oligonucleotide 5’-TCACCGATCTGATGG-3’ which is complementary to the coding sequence for both regions except for the central thymidine base, which results in the third base of the AGC codon for Ser“ or Ser”’ being changed to adenine. This substitution also creates a new Sau3A restriction site which can be used to confirm the mutation (see below). The mutagenesis was carried out on the tps gene cloned into pBR322 (see Fig. 2 A ) using the plasmid method (21). A portion of this plasmid was cut with PstI and treated with DNA polymerase I Klenow fragment to inactivate the ampicillin resistance gene, and another portion was cut with SmaI to remove a fragment within the tps gene containing both sequences to be mutagenized. Both forms of the plasmid were then denatured and reannealed in the presence of the synthetic oligonucleotide to form heteroduplexes. In these experiments, the optimal buffer concentration for this step was one half that used normally for mutagenesis of Escherichia coli genes (21). After filling in the single-stranded regions, ligating, and transforming E. coli competent cells, the resulting colonies were screened using the synthetic oligonucleotide labeled with [y-3ZP]ATP as described previously (21). Positive clones were retransformed and rescreened (211, and then plasmid DNA was isolated from positive clones and digested with Sau3A to determine which mutation had been generated. Each mutant has a unique restriction pattern on 5% acrylamide gels which exactly matches the fragment sizes predicted by computer analysis of the sequences of the plasmids. Approximately equal numbers of

The abbreviations used are: kb, kilobase pair; SDS, sodium do- decyl sulfate; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; HEPES, N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid.

Protein S Calcium-binding Sites 1201

mutations occurred at each of the two sites, but no double mutants were found. Therefore, the double mutant plasmid was constructed from the two single mutant plasmids as shown in Fig. 2A.

Purification of Protein S-The wild-type and mutant proteins were purified from late-stage developing cells as described previously (6) or from mature spores by washing the spores several times with 10 mM Tris-HC1, pH 7.6, and then dissociating the protein S from the spores with 10 mM EGTA, 10 mM Tris, pH 7.6. The proteins prepared by both methods were >99% pure by SDS-polyacrylamide gel electro- phoresis.

Calcium Binding-The affinities of each of the proteins for Ca2+ ions and the number of calcium-binding sites were determined by equilibrium dialysis using conditions similar to those used for calmodulin (22). Buffer A (10 mM HEPES, 0.1 N NaCl, pH 7.5) was treated batchwise with Chelex-100 resin (Bio-Rad) to remove contam- inating metals. HEPES was obtained from Behring Diagnostics and NaCl (<0.0005% calcium) from J. T. Baker Chemical Co. All experiments were done using only plastic containers treated by agitation overnight in deionized, distilled water containing Chelex resin, fol- lowed by rinsing with deionized, distilled water. Solutions were trans- ferred using metal-free plastic pipette tips (Bio-Rad). Dialysis tubing (Arthur Thomas Dl2 or Spectrapor 2 from VWR Scientific) was treated as described (22); these membranes did not bind significant amounts of calcium or protein S. The purified proteins were first dialyzed overnight against two changes of Buffer A containing Chelex resin to remove bound Caz+ ions. Then, 150 pl of the protein solution (0.3-1.0 mg/ml) was placed on one side of a microdialysis chamber separated by a dialysis membrane (cut from the tubing described above) from an equal volume of Buffer A containing various concentrations of unlabeled CaCIZ (ranging from 0 to lo-’ M) and 0.9-4.0 pCi “CaC12 (Du Pont-New England Nuclear, >4 Ci/mmol). The dialysis chambers were shaken at room temperature for 48 h; control experiments without protein showed that equilibrium was achieved in 30-36 h. Aliquots were removed from each side of the membrane in the chamber and counted in a liquid scintillation counter; additional aliquots were used to determine the protein concentration in each sample either by acid hydrolysis and amino acid analysis using an automated LKB amino acid analyzer or by reaction with bicin- choninic acid (Pierce Chemical Co.) and measurement of the absor- bance at 562 nm. The latter assay was always calibrated using protein S samples whose concentrations had been determined by amino acid analysis. Binding data were analyzed using BMDP software on a VAX 11/780 computer.

RESULTS

Calcium Binding to Protein S-The calcium-dependent assembly of protein S on the myxospores (3) and the homologies between sequences in protein S and the calcium-binding sites of bovine brain calmodulin (8) suggested that there might be two Ca2’-binding sites in protein S (see Fig. 1) or possibly as many as four, if one includes the corresponding sequences in domains 2 and 4 of protein S, which have slightly less homology to calmodulin (8). We therefore determined the number and affinity of the Ca2+-binding sites in protein S by equilibrium dialysis. The results, shown in Fig. 3 (closed circles), indicated that protein S boupd 2 mol of Ca2+/mo1 of protein. When plotted in Scatchard form (Fig. 3, inset), the data points could be fitted to two straight lines, indicating one binding site with a Kd of 2.1 X M and one with a Kd of 7.5 X M, if one assumes that the sites are independent. The binding data could also be fitted to a curve by nonlinear regression using a sequential binding model, as has been described for calmodulin (22). This method yielded & values of 2.7 X and 7.6 X M, respectively, and is probably more accurate. The presence of two binding sites in protein S, in contrast to the four present in calmodulin, suggested that only the sequences in domains 1 and 3 (shown in Fig. l), which are the most homologous to the calmodulin sites, are capable of binding Ca”, whereas the corresponding sequences in domains 2 and 4 do not. This hypothesis is also consistent with the previous observation that the “tryptic core peptide” consisting of domains 1 and 2 was capable of Ca2+-dependent

-6 - 5 - 4 - 3 -2

log [Cat + 1 FIG. 3. Ca2+ binding to protein S . Equilibrium dialysis experi-

ments were carried out as described under “Materials and Methods.” 0, wild-type protein S; 0, [Arg’29]protein S. Error bars show standard error of the mean for experiments that were repeated two to five times. The inset shows data for wild-type protein plotted in Scatchard form. E, moles Ca2+ bound per mole of protein; c, concentration of free Ca”.

assembly on myxospores and competed with protein S for binding to the spores with a reduced affinity (6), indicating that at least one of the Ca2+-binding sites was within domains 1 and 2. We therefore decided to make mutations in the proposed Ca2+-binding sequences of domains 1 and 3 in protein S to see whether they would have the predicted effects on Ca2+ binding.

Mutagenesis of the Proposed Calcium-binding Sites-Ser4’ and Ser’” were selected as the sites for mutagenesis because of the alignment with the calmodulin Ca2+-binding sequence (see Fig. 1) and their location in the proposed three-dimensional structure based on homologies to y-crystallin, which predicts these residues to be facing possible Ca2+-binding sites on the surface of the molecule, whereas Ser41 and Ser130, for example, have their side chains buried (12). Ser4’ and SerlZ9 were changed to arginine residues because the presence of a positive charge at those locations was likely to interfere with the binding of the Ca2+ ions. Both mutations were made in the cloned tps gene using the same oligonucleotide as described under “Materials and Methods,” and a double mutant was constructed as shown in Fig. 2 A .

Expression of Mutant tps Genes-The mutant genes were each transduced into the chromosome of an M. xanthus strain from which the protein S gene had previously been deleted. The mutant strains grew and developed normally and ex- pressed the mutant protein S genes at the proper time in development. The mutant forms of protein S could each be immunoprecipitated from lysates of developing cells of the respective strains using rabbit antiserum to the wild-type protein S. As shown in Fig. 4, all three mutants produced the same amount of protein, which is comparable to the wild-type protein S production at that stage of development (2, 3).

Assembly of Mutant Proteins on Spores-The mutant M. xanthus strains were allowed to sporulate, and the proteins on the surface of the mature myxospores were analyzed by harvesting the spores from the agar plates and boiling them in SDS. This treatment removes the spore surface proteins, including protein S, but does not damage the spores (3-5). Fig. 5 shows that [Arg’29]protein S was assembled onto the spores in the same quantities as the wild-type protein S , whereas [Arg4’]protein S was found in much smaller amounts

1202 Protein S Calcium-binding Sites 1 2 3 4 5

4

FIG. 4. Immunoprecipitation of protein S from developing cells. Lysates were incubated with antiserum to wild-type protein S. Immunoprecipitates were boiled in SDS buffer and analyzed on 17.5% polyacrylamide gels. Lane 1, molecular weight markers; lane 2, mutant 661 (Ser4' + Arg); lane 3, mutant 662 ( S e P 4 Arg); lane 4, mutant 663 (double mutant); lane 5, parent strain (ops-tps deletion mutant). The arrowhead points to protein S.

1 2 3 4 5 ri.. , ."rWp*P T5mi? I ^?...

4

FIG. 5. Protein S on the spore surface. Spores were boiled in SDS, and the supernatants were electrophoresed as described for Fig. 4. Lunes 1 and 5, wild-type strain (DZF1); lane 2, mutant 661 (Ser4' 4 Arg); lane 3, mutant 662 ( S e P + Arg); lane 4, mutant 663 (double mutant). The arrowhead points to protein S.

and [Are , Arg129]protein S did not assemble on the spores at all.

Calcium Binding of Protein S Mutants-The three mutant proteins were purified and assayed for Ca2+ binding by equilibrium dialysis with 45CaC12 as described above for the wild- type protein s. For [Ar$']protein S and [Are , Arg129]protein S, no significant Ca2+ binding could be detected. It is possible that small amounts of Ca2+ (cO.1 mol/mol of protein) are bound at Ca2+ concentrations of 1 mM or higher, but this would not be detectable because of the inaccuracy inherent in the assay method at high Ca2+ concentrations when the amount of 45Ca2+ bound is only a very small fraction of the total. [Arg'29]protein S bound Ca2+, but with a lower affinity than the wild-type protein S (see Fig. 3). The number of binding sites could not be determined unequivocally because the lower affinity again limited the accuracy of the binding experiments at the high Ca2+ concentrations. There may also be a problem with nonspecific binding of the Ca2+ ions to this rather acidic protein at high Ca2+ concentrations. This has been observed, for example, in human erythrocyte calmodulin (23). On the other hand, the mutated Ca2+-binding site in domain 3 may still be able to bind Ca2+ with a greatly reduced affinity; so that if the Ca2+ concentration were high enough (>>1 mM), both sites might be occupied. A Scatchard plot cannot be used to determine the number of binding sites from

these data because saturation of the sites was not achieved (24).

DISCUSSION

Our data are consistent with the hypothesis that the sequences in domains 1 and 3 of protein S which are homologous to the calmodulin Ca2+-binding sites (shown in Fig. 1) are in fact involved in Ca2+ binding by protein S. The decreased Ca2+ binding by [Arg'29]protein S probably reflects a greatly reduced affinity at the binding site in domain 3; this may also reduce the affinity of the other site due to cooperative interactions. The fact that this mutant protein continued to assemble normally on the spores was not surprising since the tryptic core fragment lacking domains 3 and 4 was also able to assemble on the spores (6). Thus, the binding of a Ca2+ ion to domain 1 is probably sufficient to induce a conformational change in the protein that results in its assembly on the spores. Unlike calmodulin, however, the binding of Ca2+ by protein S does not result in a conformational change that is detectable by a change in electrophoretic mobility.

The inability of [Ar$']protein S to bind Ca2+ or to assemble on the myxospores has two possible explanations. Either the mutated site in domain 1 was the high affinity site and the binding of Ca2+ to the remaining site in domain 3 was too weak to detect in the binding assay and did not change the conformation of the protein to permit assembly, or the mutation in domain 1 changed the conformation of the protein in such a way that no Ca2+ binding to either site was possible. However, the fact that this protein was still recognized by antibodies to the wild-type protein S and was not degraded in the developing cells indicates that [Ar$']protein S still resembles the wild-type protein.

The relationships between the prokaryotic protein S and the eukaryotic proteins calmodulin and y-crystallin are still not completely clear. Protein S binds calcium like calmodulin, but only in two of its four domains and with a lower affinity than the sites in calmodulin. Competition experiments with 45Ca2+ and unlabeled Mg2+ (not shown) indicated that the affinity of protein S for Mg2+ was an order of magnitude lower than for Ca2+. Calmodulin has two types of binding sites: one binds both Ca2+ and M$+, and the other is specific for Ca2+. Apart from the Ca2+-binding sites, protein S has no sequence homology to calmodulin (8) and does not appear to have the helix-loop-helix structure characteristic of calmodulin, tro- ponin C, and parvalbumin. It is therefore not surprising that protein S cannot stimulate 3',5'cyclic-nucleotide phosphodi- esterase and that calmodulin antagonists have no effect on the assembly of protein S (25). Recently, the structure of E. coli periplasmic D-galactose-binding protein was determined by x-ray crystallography (26). This protein has a single Ca2+- binding site, with a sequence very homologous to those of the calmodulin sites, that forms a calcium-binding loop like those in the calmodulin structure but without the helices on each side (26). These results provide additional support for an ancient origin for the calcium-binding structure and demon- strate that the helices in the calmodulins and related proteins are not essential for the Ca2+ binding function.

Protein S has homologies to the eye lens protein y-crystallin throughout its sequence (12), but the similarity is not very strong in the area of the proposed calcium-binding sites and y-crystallin is not known to bind calcium. Whether the three- dimensional structure of protein S resembles that of y-crystallin as has been proposed (12) and whether the calcium- binding sites are indeed located where we have suggested can

Protein S Calcium-binding Sites 1203

only be answered for certain with the solution of the x-ray 11. Watterson, D. M., Sharief, F., and Vanaman, T. C. (1980) J. Biol. crystal structure of protein S, which is now in progress.' Chem. 255,962-975

12. Wistow, G., Summers, L.. and Blundell. T. (1985) Nature 315 .

5.

6.

7.

8.

9.

10.

REFERENCES

Rosenberg, E. (ed) (1984) Myxobacteria: Development and Cell

Inouye, M., Inouye, S., and Zusman, D. R. (1979) Deu. BWZ. 6 8 ,

Inouye, M., Inouye, S., and Zusman, D. R. (1979) Proc. Natl. Acad. Sci. U. S. A. 76, 209-213

Teintze, M., Furuichi, T., Thomas, R., Inouye, M., and Inouye, S. (1985) in M o k c h r Biology of Differentiation (Hoch, J., and Setlow, P., eds) pp. 253-260, American Society for Microbiol- ogy, Washington, D. C.

Teintze, M., Thomas, R., Furuichi, T., Inouye, M., and Inouye, S. (1985) J. Bacteriol. 163 , 121-125

Inouye, S., Harada, W., Zusman, D., and Inouye, M. (1981) J. Bacterwl. 148,678-683

Inouye, S., Inouye, M., McKeever, B., and Sarma, R. (1980) J.

Inouye, S., Franceschini, T., and Inouye, M. (1983) Proc. Natl.

Inouye, S., Ike, Y., and Inouye, M. (1983) J. Biol. Chem. 2 5 8 ,

Takao, T., Hitouji, Y., Shimonishi, T., Tananbe, T., Inouye, S.,

Interactions, Springer-Verlag, New York

579-591

Bid. Chem. 255,3713-3714

Acad. Sci. U. S. A. 80,6829-6833

38-40

and Inouye, M. (1984) J. Mol. Biol. 259,6105-6109

771-773 . .

13. Newton, D. L.. Oldewurtel. M. D.. Krinks. M. H.. Shiloach. J..

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24. 25.

and Klee, C:B. (1984) J..Biol. C h m . 259,441914426

A. R., and Cook, W. J. (1985) Nature 315 , 37-40

S. (1984) J. Bacteriol. 158 , 1195-1197

Gen. Genet. 199,434-439

. .

Babu, Y. S., Sack, J. S., Greenhough, T. J., Bugg, C. E., Means,

Komano, T., Furuichi, T., Teintze, M., Inouye, M., and Inouye,

Furuichi, T., Komano, T., Inouye, M., and Inouye, S. (1985) Mol.

Shimkets, L. J., Gill, R. E., and Kaiser, D. (1983) Proc. Natl.

Furuichi, T., Komano, T., and Nisioka, T. (1984) J. Bacteriol.

Yanisch-Perron, C., Vieira, J., and Messing, J. (1985) Gene

Orndorff, P., Stellwag, E., Starich, T., Dworkin, M., and Zissler,

Inouye, S., and Inouye, M. (1987) in DNA and RNA Synthesis

Crouch, T. H., and Klee, C. B. (1980) Biochemistry 19, 3692-

Jarrett, H. W., and Kyte, J. (1979) J. Biol. Chem. 254 , 8237-

Klotz, I. M. (1982) Science 217 , 1247-1249 Teintze, M., Thomas, R., Inouye, S., and Inouye, M. (1984) Fed.

Acad. Sci. U. S. A. 80,1406-1410

158,997-1004

(Amst.) 33,103-119

J. (1983) J. Bacteriol. 154, 772-779

(Narang, S., ed) pp. 181-204, Academic Press, New York

3698

8244

Proc. 43. 1519 26. Vyas, N. K:, Vyas, M. N., and Quiocho, F. A. (1987) Nature 327,

R. Sarma, personal communication. 635-638

Documents

Characterization of calcium-binding sites in development-specific