Internal structural features of E. coli glycyl-tRNA synthetase examined ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY Vol. 261, No. 15, Issue of May 25, pp. 66434646,1986 0 1986 by The American Society of Biological Chemists, Inc.

Communication

Printed in U.S.A.

to-tail subunit fusion would retain activity. For example, if Internal Structural Featuresof the active site forms at the junction of the subunit chains, E. coli Glycyl-tRNA Synthetase then orientation factors would be crucial. This could explain Examined by Subunit Polypeptide why the chains are not joined, even though the arrangement of coding regions in the genome is suggestive of a single Chain Fusions* polypeptide chain. (Received for publication, February 18,1986) Matthew J. Toth andPaul Schimmel From the Department of Biology, Massachusetts Znstitute of Technology, Cambridge,Massachusetts 02139

To investigate this question, we used oligonucleotide-directed mutagenesis to fuse the chains with several different linker constructions. These constructionswere then examined for protein stability and tested for activity in vivo and in vitro. MATERIALS ANDMETHODS

Incontrast to mostaminoacyl-tRNAsynthetases Oligonucleotide-directed Mutagenesis-Oligonucleotide-directed which are monomers or oligomers ofsingle a polypepmutagenesis followed the procedure of Zoller and Smith (asmodified tide, Escherichia coli glycyl-tRNA synthetase has an by L.Marsh) (5). The 5-kilobase pairs HindIII fragment that contains a-2,&2 structure. The enzyme requires both subunits the glyS gene was cloned into the single-stranded DNA phage mp8. for catalysis of either adenylate or aminoacyl-tRNA Our initialoligonucleotide-directed mutagenesis used the permutated synthesis. The head-to-tail arrangement of thea- and cc &chain codingregions in the genomesuggests that the heptadecamer CCGCCTC’I’TGCTTATCT (synthesized on a Systec two-subunit protein maybetantamounttoa single Microsyn 1450A Automated DNA Synthesizer). Except for the perchain. We fused the carboxyl terminus of thea-chain muted positions, this oligonucleotide is complementary to thecoding to the amino terminus of the&chain, through a short strand of glyS that encodes the a-subunitstop codon and its5’- and peptide linker. Five different amino acid substitutions 3‘-flanking sequences (AGATAAGTAAGAGGCGG).The permuted were placed in the linker. In all instances, the fusion positions alter the TTA that is complementary to the TAA stop polypeptide is stable in maxicell extracts. In a glyS (whose position is underlined above). null strain, a gene encoding any of the fusion proteins Mutant mp8 plaques were detected a t a frequency of 16% by with the 5’-32P-labeledmutagenic oligonucleotide (5). substitutes for the wild-type gene. Assays confirm that, hybridization Dideoxy sequencing (as modified for [a-35S]dATP) (6) with an oliin vitro, the engineered polypeptide fusion is active to gonucleotide that primes close to the site of modification was perwithin 2- to 3-fold of the wild-type, unfused chains. formed to confirm the putative mutant sequences. Two mutants were Oligomers of the fusion protein are observed andmay recovered these are Fusion 1 and Fusion 2, where the TAA stop be required foractivity. Because the creation and lim-codon of the a-subunitwas replaced by glutamic acid and glutamine ited manipulation of the artificial peptide linker region codons, respectively (seebelow). The double-stranded DNA from does not destroy the activity, we also conclude that the these mutant mp8 clones was subcloned into the HindIII site of C-terminal part of the a-chain and the amino-terminal pBR322. The result isgZyS fusion-encodingplasmids pFNlOl (Fusion 1)and pFN201 (Fusion 2). part of the &chain are not important foractivity.

A second round of oligonucleotide-directed mutagenesis was performed essentially as described above. This mutagenesis was applied to the pFNlOl and pFN201 glyS fusions to eliminate the ATG We report here the examination of one structural parameter (complementary to CAT) codon of the 8-subunit. This was accomby a CAT + TGC change which results in a coding strand of an aminoacyl-tRNA synthetase that has an unusual qua- plished change of Met + Ala. Adjacent to this anAGA + AGC or AGA + ternary structure. This is Escherichia coli glycyl-tRNA syn- TTC change was introduced. These “second generation” fusion muthetase. In E. coli, only glycyl- and phenylalanyl-tRNA syn- tants encoded on pBR322 vectors are designated as pFNlO2, pFN202, thetases have 01-2, @-2quaternary structures. The rest of the and pFN203 (Fusions 3,4, and5, respectively). Analysis of the Fusion Mutants-To visualize the plasmid encoded synthetases are either a, a-2, or a-4 structures, that is, oligoproteins, the maxicell technique (7) was used to produce 35S-labeled meric forms of one polypeptide chain (1). InglyS, the a-(303 codons) and @-chain(689 codons) coding fusion and wild-type proteins. The labeled proteins were separated on a Laemmli gel (8) (3% stacking gel, 10% running gel), soaked in regions are arranged in tandem,with a space of 9 nucleotides Autofluor (National Diagnostics), dried onto filter paper, and exposed separating the stop of the a-chain (ochre codon) from the to x-ray film. start of the @-chain(2). Neither of the purified subunits has For biochemical analysis, deletion strains thatcontain fusion mueither glycyl adenylate synthesis or aminoacylation activity. tant or wild-type gZyS plasmids were grown to late log phase in 100 Both activities are obtained when the two subunits are mixed ml of LB medium. Cells were collected by centrifugation, washed with 0.7% saline, centrifuged, scraped into a chilled mortar and pestle, together (3). Keng and Schimmel (4)transformed ochre suppressor ground with approximately 2 packed cell volumes of alumina in 1ml of grinding buffer (50 mM potassium phosphate (pH 7.5), 15 mM 2strains of E. coli with aglyS containing plasmid and detected mercaptoethanol, and 200 pl ofethanol (saturated with phenylmethsubunit fusions mixed with a preponderance of unfused ylsulfonyl fluoride)/liter). The extract was centrifuged at 12,000 X g chains; they could not determine if the fusions possessed (Eppendorf) and the supernatant was used for analysis or stored activity. We wanted to testwhether or not thistype of head- frozen a t -20°C. Aminoacylation assays were performed at 37 “C as described by Schreier and Schimmel (9). The pyrophosphate ex* This work was supported in part by Grant GM 23562 from the change assays were performed as in Calendar and Berg (10) (as National Institutes of Health andby W. R. Grace and Company. The modified by Putney et al. (11)). 8-Lactamase activity of crude extracts costs of publication of this article were defrayed in part by the was determined spectrophotometrically according to Jansson (12). payment of page charges. This articlemust therefore be hereby To estimate the sizes associated with the activities of the fusion marked “aduertisement” in accordance with 18 U.S.C. Section 1734 proteins, crude extracts were applied to a Sephacryl S-300 gel filtrasolely to indicate this fact. tion column, and fractions from the elution were assayed.

6643

6644

Internal Structural

Features of E. coli Glycyl-tRNA Synthetase

RESULTS

Tester Strain for Protein Fusions: Creation of a glyS Null Allele-The glyS locus is essential in E. coli (see below). To test rigorously the efficacy of specific fusion constructions in uiuo, however, the chromosomal glyS locus must be deleted. For this purpose, we adapted the procedures of Jasin and Schimmel (13) that were developed for the alanyl-tRNA synthetase system. The procedure is based upon site-specific recombination with a linear DNA fragment that replaces the target gene with a kanamycin resistance marker thatis bounded by glyS 5' and 3' adjoining sequences (Fig. 1A). When the chromosomal copy of glyS is deleted to give the A glyS strain TM101, cell viability is maintained by plasmid pMT901. This plasmid contains the HindIII fragment that encodesglys (Fig. 1B) cloned into theHindIII siteof pMT101. The latter plasmid was constructed by inserting the chloramphenicol resistance locus from pBR325 into pPM103 (which encodes tetracycline resistance and a temperaturesensitive replicon). PlasmidpMT901 thus encodes the HindIII fragment that encodes glyS, chloramphenicol resistance, and a temperature-sensitivereplicon such that plasmid replication is blocked at therestrictive temperature (42 "C). Functional recA+ activity is required for the site-specific recombination of linear DNA fragments. After such recombination, however, we assume that a recA- allele is required to prevent recombination between plasmid-borne sequences and thechromosome. In thepresent study, it was not possible to inactivate recA simultaneously with the deletion of the target gene (as was done previously) (13). A two-step procedure was required in which the glyS deletion of TMlO1/

type Fuslon 1 Fuslon 2

GAT

A A G T A A G A G GCG GCT ATG T C T GAG AAP. A S D L v s Sioo - f-Me1 S e r G l u L v s

I

tiAA

GIu GIu CAA

Ala Ala

Met

CIn

bln

Iln

Mrl

C1u

Fusion 3

GCA GCT

Fuslon 4 Fusion 5

FIG. 2. Depiction of the glyS intersubunit region that was modified by site-directed mutagenesis. The nucleotide sequence of the end of the a- and of the beginning of the @-subunit isshown together with the translatedamino acid sequence. Note that inall of the fusion constructions the stop codon of the a-chain has been replaced with either GAA (Glu) or CAA (Gln). Adjacent to thealtered stop codon, the sequence is the same for the next three codons in all of the constructions. Fusions 3,4,and 5 have changed the @-chain's initiator codon to GCA (Ala) in order to eliminate initiationof protein synthesis at this pointin the sequence (see text). Additional sequence alterations also have been placed in the second codon of the 8-chain region of three of the fusion protein genes. kb, kilobase pair.

pMT901 was first created; then the recA56 allele of GW554l wasmoved intothisstrain by P1 transduction (14). The resulting temperature-sensitive A glyS recA strain is designated TM102/pMT901. For reasons of convenience, the A glyS and recA alleles were then moved into a wild-type K12 A strainto give the temperature-sensitive TM202/pMT901 strain. Specific details of all of these constructions followed along lines described earlier (13). Construction of Simple In-frame Fusions of Subunit Coding Regions-Fig. 2 shows the glyS intersubunit region. The translation stop of the a-subunit coding region of 303 codons is separated by 9 nucleotides from the translation start of the 689 codons of the @-subunitopen reading frame. The intersubunit region encodes a ribosome binding site which enables independent translational initiation at the beginning of the P-chain (Fig. 2 and see below). Edman degradation of purified @-subunit has shown that the mature protein starts with Glu and lacks, therefore, the Met and Ser that are at the start of the coding region (2). This pointis important for the interpretation of data that are presented below, because the Met and Ser positions are part of the fusion polypeptides. pMT9OI Oligonucleotide mutagenesis was used initially to construct two simple in-frame fusions of the (Y and B polypeptides. The EcoRl l15Kb) Hpa I a-chain TAA stop codon is changed to GAA in Fusion 1 and to CAA in Fusion 2. These are Glu and Gln codons, respectively (Fig. 2). Eachof these fusion constructions on plasmids (Fig. 3) was transformed into TM202/pMT901. Elevation of Hind Ul the temperature to 42 "C gives stable colony growth of the FIG. 1. A , schematic representation of the deletion of glyS by sitetransformants, with loss of the temperature-sensitive plasmid. specific recombination with a linear DNA fragment. The E. coli glyS locus and adjacent sequences that extend to PstI sites are diagram- This establishes that the gene fusion per se supports viable cell growth. matically represented in the lower part. Note that, within the PstI sites, glyS is flanked by HindIII sites. A linear PstI fragment was To establish the effect of the fusion on the synthesis of the constructed in which the region between the Hind111 sites was re- gene product, maxicell analysis of protein products was carplaced with kanamycin resistance (KunR).Recombination between ried out as described under "Materials and Methods." The the chromosome and the fragment results in deletion of glyS and results for Fusion 1 and Fusion 2 are shown in Fig. 4. The insertion of KunR into the chromosome. B, schematic diagram of plasmid pMT901. This plasmid contains glyS and has a temperature- fusion proteins are clearly evident as the predominant glyS sensitive origin of replication. The plasmid also encodes a marker for chloramphenicol resistance (CmR).kb, kilobase pair.

G. Walker, personal collection.

Internal Structural Features of E. coli Glycyl-tRNA Synthetase

6645

TABLE I Relative enzyme activity in extracts of a glyS null strain (TM202) that harbors various plasmids Details of samplepreparationsand of assays aregivenunder "Materials and Methods." Plasmid pTK201 encodes wild-type glycyltRNA synthetase. Plasmids pFNlOl to pFN203 encode the fusions that are defined in Fig. 2 as Fusions 1 to 5.

FIG.3. Schematic representation of plasmid pFN101. This plasmid encodes Fusion 1 ( s e e Fig. 2). Note that it is constructed by joining plasmid pBR322 withthe H i d 1 1 fragment that encodesglyS (Fig. LA). Except for the nature of the sequence alteration in the intersubunit region, the other fusion protein encoding plasmids are the same as pFN101.

Strain

Aminoacylation activity

ATP -.pyrophosphate

TM202/pTK201 TM202/pFN101 TM202/pFN201 TM202/pFN102 TM202/pFN202 TM202/pFN203

1.0 0.48 0.32 0.45 0.56 0.36

1.0 1.2 1.3 2.2 2.1 1.2

exchange ectivlty

of the second codonof the @-chainand of the a-subunitTAA terminator codon. Each of these three fusion constructs complements the glyS null strain TM202. Maxicell analysis of the gene products is shown in Fig. 4. These data show clearly that stable fusion protein is produced in all three instances and that neither anor @-chainis produced insignificant amounts. This confirms that internal initiation was responsible for synthesis of the kD F free @-chainobserved with plasmids encoding Fusion 1 and Fusion 2. More significantly, complementation of the gfyS JT null strain by the fusion protein, in the absence of significant amounts of either of the free subunits, suggests that the fusion protein itself is active. d67 In Vitro Activities and Stabilities of Fusion Proteins-The '60 enzymatic activities of extracts of strains bearing the fusions - 43 were compared with the wild-type protein. Two activities were a- 36 measured, the glycine-dependent ATP-PP, exchange, that monitors aminoacyl adenylate synthesis, andthe glycine8.-30 specific aminoacylation of tRNA. For this purpose, the activFIG. 4. Autoradiogram of s6S-labeled extracts of maxicells ities were normalized to the 8-lactamase activity that is enthat have been subjected to sodium dodecyl sulfate gel electro- coded by the multicopy plasmid whichalso encodes the conphoresis. Six extracts that were subjected to gel electrophoresisare shown. These extracts were prepared from maxicells that contained struct of interest. Data given inTable I show that, for each fusion, the glycinea plasmid that encodes the wild-type enzyme or one ofthe five fusion proteins. The position of migration of the fusion polypeptide and of dependent ATP-PPi exchange activity is within 1-2-fold of the individual subunits is indicated. Each gel shows the presence of that of the wild-type protein. The aminoacylation activities, the B-lactamase gene product at approximately M, = 30,000. Note while relatively lower, are nonetheless 35 to 65% of that of the absence of the a-chain in each of the extracts from cells that the wild-type enzyme. We concludethat the fusions are effecencode a fusion protein. The free p-chain is prominent in the lanes tive surrogates for glycyl-tRNAsynthetase activity. for wild-type, Fusion 1, and Fusion 2 proteins. Additional bands in The stabilities of the fusions were also investigated. A some of the lanes are probably dueto degradation products. maxicell extract of Fusion 2 was incubated (before electrophoresis) for 1 h at 37 "Cwith and without additional extract species in the system. This suggests that the fusion proteins from a nonirradiated K12 strain. This hadno discernible are stable, even though six amino acids have been inserted effect on the yield of fusion protein or on the appearance of between the endof the native a-chain and the beginning (Glu) a- or @-chainproducts that could result from breakdown of of the mature @-chain.Little or no a-chain is evident. There the fusion protein (data not shown). In addition to these are, however, significant amounts of @-subunitand this conexperiments, we determined that the aminoacylation activity fuses the interpretation of the complementation of the glyS of Fusion 3 was unchanged at room temperature for at least null allele by the fusion constructions. 3 h. The presence of the @-subunit,in the absence of significant amounts of the a-chain, suggests that independent initiation DISCUSSION of @-chainsynthesis has occurred in Fusion 1 and Fusion 2. To address this problem, additional fusions weremadeby These data imply that thetwo chains of glycyl-tRNA synoligonucleotide-directedmutagenesis. These were constructed thetase need not be separate in order to generate activity in so as toremove the potential for independent ribosome bind- uiuo or in uitro. Limitedmanipulations of the sequence of the ing at the beginning of the @-subunitcoding region. junction regiondid not lead to largeeffectaon catalytic Construction and Analysis of Additional Fusion Polypep- activity (Table I). The interpretation of the results with tides-Three additional gene fusions (Fusions 3, 4, and 5) Fusions 1 and 2 are complicated by small amounts of free 8were constructed (Fig. 2). Each of these has replaced the 8- subunit in admixture with the fusion proteins. This difficulty chain's methionine initiator codon with GCA (which encodes is overcome inthe case of Fusions 3.4, and 5. The latter three Ala). This change alone should eliminate internal initiation fusions have either Glu or Gln codons at the position of the of translation at the start of the @-portionof the fusion normal stop codon, Ala at the first codon of the &chain, and construct. This change has been combinedwith an alteration Ala or Glu at the second codon of the @-chain.These substi-

b

6646

Internal Structural Features of E. coli Glycyl-tRNA Synthetase

tutions do not greatly affect the activity (Table I). This observation and thefinding that introduction of the ‘‘spacer’’ of six amino acids connecting the subunits does not block activity suggest that the carboxyl terminus of the a-subunit and the amino terminus of the @-subunit are notcritical for function. In the fusion protein, our data do not distinguish whether the active sites are formed entirely within one chain or whether they are assembled from components of two or more fusion chains. Additional experiments have attempted to determine the size of the fusion protein on a Sephacryl s-300 gel filtration column, in crude extracts and in partially purified preparations. These experiments suggest that all fusion proteins associate into several aggregation states. These states have sizes corresponding to more than two fusion polypeptides. (A dimer would be expected if the fusion chain was associated analogously to thewild-type a-2, @-2protein.) This means that the catalytic site may form across fusion chains, rather than within a single chain, making it unclear as to what oligomeric state confers functionality. These data also provided no evidence for the presence of cleavage products that have catalytic activity. An increasing body of evidence suggests that theadenylate binding site is near the amino terminus of the respective synthetase polypeptides (15, 16). There are some data that also show that the tRNA binding site is proximal to and on the C-terminal side of the adenylate site (17). While no adenylate synthesis activity has been ascribed to theisolated a-chain of glycyl-tRNA synthetase, the @-subunitspecifically binds to tRNA(18).Moreover, like the alanyl- andmethionyltRNA synthetases in which C-terminal sequences are dispensable for activity (15, 19), deletion of 112 codons from the carboxyl terminus of the glycyl-tRNA synthetase fusion and wild-type proteins leaves a residual polypeptide that has catalytic activity and complements the deletion strain. These similarities of the glycine fusion protein to other single chain aminoacyl-tRNA synthetases suggest that the fusion could mimic in some way these better characterized aminoacyltRNA synthetases.

The rationale for discrete a- and @-chains is not known. We consistently note, however, that in maxicell synthesis of the plasmid-encoded wild-type protein, substantially more achain than @-chainis synthesized. This may reflect a requirement for the a-chain in addition to its role as a subunit of glycyl-tRNA synthetase. REFERENCES 1. Schimmel, P. R., and Soll, D. (1979) Annu. Reu. Biochem. 48, 601-648 2. Keng, T., Webster, T. A., Sauer, R. T., and Schimmel, P. R. (1982) J. Biol. Chem. 257,12503-12508 3. McDonald, T., Breite, L., Pangburn, K. L., Hom, S., Manser, J., and Nagel, G. M. (1980) Biochemistry 19,1402-1409 4. Keng, T., and Schimmel, P. (1983) J. Biomol. Struc. Dyn. 1,225229 5. Zoller, M., and Smith, M. (1983) Methods Enzymol. 100, 468500 6. Biggin, M. D., Gibson, T. J., and Hong, G. F. (1983) Proc. Natl. A c u ~Sci. . U. S. A. 80,3963-3965 7. Sancar, A., Hack, A. H., and Rupp, W. D. (1979) J. Bacteriol. 137,692-693 8. Laemmli, U. K. (1970) Nature 227,680-685 9. Schreier, A. A., and Schimmel, P. R. (1972) Biochemistry 11, 1582-1589 10. Calendar, R., and Berg, P. (1966) in Procedures in Nucleic Acid Research (Catoni, G. L., and Davies, D. R., e&) pp. 384-399,

Harper and Rowe, New York 11. Putney, S., Sauer, R. T., and Schimmel, P. R. (1981) J. Biol. Chem. 256, 198-204 12. Jansson, J. A. T. (1965) Biochim. Biophys. Acta 99, 171-172 13. Jasin, M., and Schimmel, P. R. (1984) J.Bacteriol. 159,783-786 14. Miller, J. (1972) in Experiments in Molecular Biology, pp. 201205, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 15. Jasin, M., Regan, L., and Schimmel, P. (1983) Nature 306,441447 16. Webster, R., Tsai, H., Kula, M., Mackie, G. A., and Schimmel, P. (1984) Science 226,1315-1317 17. Schimmel, P. (1986) in Robert A. Welch Symposium on Genetic

Chemistry, in press 18. Nagel, G. M., Cumberledge, S., Johnson, M. S., Petrella, E., and Weber, B. H. (1984) Nucleic Acids Res. 12,4377-4384 19. Cassio, D., and Waller, J. P. (1971) Eur. J. Biochem. 20, 283300