Glutamyl-tRNA Reductase from Escherichia coli and Synechocystis 6803” GENE STRUCTURE AND EXPRESSION

In the cyanobacterium Synechocystis sp. PCC 6803 and in the enterobacterium Escherichia coli &amino-levulinic acid (ALA) is formed from glutamyl-tRNA by the sequential action of two enzymes, glutamyl-tRNA reductase (GluTR) and glutamate- 1-semialdehyde aminotransferase. E. coli has two GluTR proteins with sizes of 45 kDa (GluTR46) and 85 kDa (GluTR85) (Jahn, D., Michelsen, U., and 5611, D. (1991) J. Biol. Chem. 266, 2542-2648). The hemA gene, isolated from E. coli and several other eubacteria, has been proposed to encode a structural component of GluTR. Because of the inability to overexpress this gene in E. coli, we demonstrate directly GluTR function for the E. coli hemA gene product by its expression and func- tional analysis in yeast, which does not form ALA from Glu-tRNA. Gel filtration experiments demonstrated definitively that the yeast-expressed HemA protein corresponded to GluTR45. Furthermore, analysis of GluTR activity in an E. coli strain with a disrupted hemA gene displayed GluTR86, but not GluTR45 ac- tivity. The hemA gene from Synechocystis 6803 was cloned by functional complementation in E. coli. DNA se- quence analysis revealed an open reading frame capable of encoding a 427-amino alignment acid These experiments are the first direct demonstration of GluTR activity of the HemA protein and provide further evidence for two pathways of ALA formation in prokaryotes.

cursor of all tetrapyrroles and is synthesized in nature via two dissimilar routes. In the Shemin pathway (1) ALA synthase catalyzes the condensation of glycine and succinyl-CoA to form ALA in yeast, mammalian cells, and some bacteria. The C, pathway operates in plants, algae, cyanobacteria, and a variety of other bacteria (reviewed in Refs. 2 and 3). The starting substrate in this route is glutamyl-tRNA (Glu-tRNA) which also provides the source of glutamate for protein biosynthesis. In the presence of NADPH, glutamyl-tRNA reductase (GluTR) catalyzes the formation of glutamate-l-semialdehyde (GSA) from Glu-tRNA. ALA is then formed in a transamination reaction catalyzed by glutamate-l-semialdehyde aminotransferase.
Although GSA aminotransferase has been purified and characterized and the corresponding genes isolated from a number of organisms (4-7), GluTR has proven more refractory to similarly detailed analyses. This enzyme is of particular interest as it is the first enzyme unique to the C5 pathway and may provide a major regulatory point for heme and chlorophyll biosynthesis (reviewed in Ref. 2). In addition, its utilization of tRNA as a "cofactor" is unique in metabolism (8), and its mechanism of action is of special interest. Purification of GluTR from Chlamydomonas reinhardtii, Synechocystis sp. PCC 6803 (Synechocystis 6803), and Escherichia coli has been reported. The purified green alga enzyme (9) is a monomer of 130 kDa, whereas the native molecular mass of the cyanobacterial enzyme (10) was determined to be 350 kDa, with a subunit size of 39 kDa. In E. coli two distinct proteins possessing GluTR activity with sizes of 85 kDa (GluTR85) and 45 kDa (GluTR45) were identified (11).
The E. coli hemA gene (12-14) is essential for ALA synthesis from glutamate (15) and has been proposed to encode a structural component of GluTR (16). Genes that complement the E. coli hemA mutation have been cloned and characterized from Salmonella typhimurium (17), Bacillus subtilis (18), and Chlorobium uibrioforme (19). Despite the availability of cloned genes, the hemA gene product has not been overexpressed, purified, and unequivocally identified. Attempts to overexpress the E. coli or Bacillus subtilis hemA genes in the homologous systems have been unsuc~essful.~~~ In order to determine definitively the nature of enzymic activity of the protein encoded by the E. coli hemA gene and to determine its relationship to the two known GluTR activities in E. coli (11), we expressed this gene in yeast, which lacks the C5 pathway. In addition, in order to gain more insight into the structure of GluTR, we cloned by functional complementation of the 8276 Glutamyl-tRNA Reductase from E. coli and Synechocystis 6803

MATERIALS AND METHODS
General-Restriction endonucleases and DNA modification enzymes were obtained commercially and used according to the manufacturer's specifications. E. coli tRNAf'" was obtained from Boehringer Mannheim. ~-['~C]Glutamate (specific activity 266 mCi/mmol) was a product of Du Pont-New England Nuclear. Chemically synthesized GSA (20) was a gift of Dr. C. G. Kannangara (Carlsberg Laboratory, Copenhagen, Denmark). pMR57 containing the E. coli hemA gene was described earlier (14).
DNA Methods-Southern hybridizations and random primer probe preparations were performed according to published protocols (22). Synechocystis 6803 genomic DNA was purified as described previously (40). DNA sequence was determined by the method of Sanger (23) with [(U-~'SS]~ATP. Nucleotide sequence data were analyzed by the University of Wisconsin Genetics Computer Group program (24).
Construction of the Yeast Overexpression Plasmid-Two primers covering the 5' and 3' region of the E. coli hemA gene were used to generate a DNA fragment by polymerase chain reaction containing the coding region of the gene of interest, appropriate restriction sites for cloning and a 5' region ensuring correct and efficient initiation. During the amplification reaction an initial denaturation phase of 5 min at 94 "C was followed by 40 cycles of 1 min at 94 "C, 1 min at 55 "C, and 1.5 min at 72 "C. The generated DNA fragment was purified by phenol extraction, gel filtration through Sephadex-G50, and subsequent ethanol precipitation before the correct size was analyzed by agarose gel electrophoresis. The fragment was cut with BamHI and purified by agarose gel electrophoresis before ligation into the yeast shuttle vector pVT103-U (26), previously linearized with BamHI. Clones with the desired structure were identified by restriction analysis, confirmed in their integrity by DNA sequence determination and functionally tested for their ability to complement the hemA strain SASX41B. Two plasmids, pMJ55 and pMJ3, containing the hemA gene in the correct and reversed orientation, respectively, were used for functional analysis in vitro.
Overexpression of hemA in Yeast and Its Partial Purification-Plasmids pVT103-U, pMJ55, and pMR3 were transformed into Saccharomyces cerevisiae strain 3A84 (ade 1-UGA, his4-260, leu2-2, urd-52) (27) by electroporation (28), and transformants were screened for ura+ phenotype. For the overexpression of the cloned E. coli hemA gene, yeast containing pMJ55, pMJ3, and the empty vector pVT103-U were grown under selective conditions to A~w ~1 . 5 and harvested by centrifugation. Spheroplast formation and cell lysis were performed as described (29). Disruption of the cells was checked by light microscopy. The crude extract was centrifuged for 90 min at 100,000 X g, and the supernatant (S-100) was purified further by chromatography on DEAE-cellulose as outlined earlier (11). The nucleic acid-free extract was dialyzed for 4 h against assay buffer.
Glutamyl-tRNA Reductase Assay-The standard assay involved the conversion of ["CIGlu-tRNA into [I4C]GSA by the action of the GluTR containing fraction and its subsequent conversion to ["C] ALA by the addition of pure recombinant GSA aminotransferase (30). Assays (0.1-0.22 ml) containing indicated amounts of protein fraction and purified recombinant E. coli GSA aminotransferase (10 pg) were incubated for 20 min at 30 "C under standard conditions described previously (11). The reaction products were isolated by chromatography on Dowex-1 as described earlier (11,31).
For the HPLC analysis of the GluTR activity derived from the E. coli hemA gene expressed in yeast, the reaction product (GSA) was analyzed by chromatography on a precalibrated Bondapak CIR reversed-phase column using ["C]Glu-tRNA as substrate (11, 32). Its identity as GSA was confirmed by its ability to serve as a substrate for GSA aminotransferase to produce ALA. Assays were carried out with GluTR activity-containing DEAE-cellulose fractions (0.2 ml) under standard conditions (0.3 ml) with the addition of recombinant GSA aminotransferase (0.1 mg) where indicated.
Purification of Glutamyl-tRNA Reductase Activity from E. coli-The E. coli strains SASX41B, K12, and K12/pMR57 were grown as outlined above. Cells were harvested and lysed by sonication as described earlier (11). After centrifugation of the crude extract for 1 h at 100,000 X g, the S-100 was chromatographed through DEAEcellulose as described above. Active fractions were pooled, dialyzed against Mono Q buffer (20 mM Hepes, pH 7.9, 10 mM MgC12, 10 mM KCl, 10% (v/v) glycerol, 3 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and loaded onto Mono Q 10/10 (10 mg of protein/ ml of column material). Proteins were eluted with a 40-ml linear KC1 gradient (10-400 mM) in Mono Q buffer.
Gel Filtration-GluTR activity-containing fractions (0.5 ml) from Mono Q chromatography (in the case of strain EV61 grown anaerobically and strains K12 and K12/pMR57 grown aerobically) or from DEAE-cellulose chromatography (in case of the yeast expression products) were dialyzed against Mono Q buffer containing 0.5 M KC1 before gel filtration through Superose 12 as described in detail previously (9). Fraction were dialyzed against assay buffer before enzyme activity was determined.
Isolation of the Synechocystis 6803 hemA Gene-A library of 6-9kb Synechocystis 6803 genomic DNA fragments cloned into the EcoRI site of AZapII (Stratagene) was the gift of Dr. L. McIntosh (MSU-DOE Plant Research Laboratory, East Lansing, MI). Recombinant Bluescript phagemids were excised (33) and various dilutions of the rescued phagemid preparation were used to transform competent SASX4lB cells to ALA prototrophy. Samples were plated on LB/ ampicillin plates and incubated at 37 "C to select for colonies that contained putative hemA-harboring plasmids.

RESULTS
Overexpression of the E. coli hemA Gene in S. cerevisiae-Significant overexpression of the cloned hemA gene in E. coli was unsuccessful despite the use of overexpression systems with different promoters and RNA polymerases. An example is seen in Fig. 1B where the expression of pMR57, a pUCderived plasmid containing the hemA gene with its own promotor, resulted only in an approximately 4-fold increase in GluTR activity. Therefore we attempted to express this gene in the yeast S. cereuisiae, an organism which lacks the C5 pathway and utilizes the Shemin pathway for ALA synthesis (34). The E. coli hemA gene was cloned into the yeast shuttle vector pVT103-U. By the use of polymerase chain reaction the gene was engineered to ensure correct transcription and translation initiation in yeast (see "Materials and Methods"). S. cerevisiae cultures were transformed with plasmid containing the hemA gene in the correct (pMJ55) and reversed (pMJ3) orientation and with empty vector. Extracts prepared from the different cultures were partially purified by DEAEcellulose chromatography. Enzyme assays of the different preparations clearly demonstrated the presence of GluTR activity only in the extract from S. cerevisiae transformed with the hemA gene in the correct orientation (Table I). This result was confirmed by the product analysis on HPLC (Fig.  2) which showed that the extracts contained GluTR activity capable of transforming Glu-tRNA into GSA. The authentic-  ity of GSA was confirmed by its conversion into ALA by GSA aminotransferase (Fig. 2C). These results demonstrate that the E. coli hemA gene encodes GluTR.
Characterization of the E. coli hemA Gene Product Expressed in Yeast-The hemA gene encodes a protein of 46,312 kDa (14). However, as there are two GluTR activities in E. coli ( l l ) , the molecular weight of the overexpressed product was measured by gel filtration of the yeast expression product. As shown in Fig. lD, the relative molecular mass of the GluTR activity is approximately 45 kDa. Therefore, the E. coli hemA product is GluTR45. This is in line with the weak overexpression in the homologous system (see above) which also caused a slight increase of the GluTR45 activity (Fig. 1B).
Molecular Weight of the GluTR Activity Found in a hemA Strain of E. coli-Since the hemA gene product was identified as GluTR45 we wanted to ascertain that an E. coli strain in which the product of this gene was inactive lacked the GluTR45 activity. Therefore we analyzed an extract from E. coli strain EV61, which contains a gene disruption of hemA. Like other hemA strains (35), EV61 can grow under anaerobic conditions without ALA supplementation utilizing glucose as carbon source. Extracts from anaerobically grown E. coli EV61 contained GluTR activity. This activity was partially purified by chromatography on DEAE-cellulose and Mono Q before the preparation was subjected to gel filtration on Superose 12. As is clearly seen from Fig. 2C the GluTR activity of the h e &strain has a molecular mass of approximately 85 kDa and presumably corresponds to the GluTR85 enzyme.
Isolation and Characterization of the Synechocystis 6803 hemA Gene-The hemA mutation in E. coli strain SASX41B confers a growth requirement for ALA. In order to isolate the hemA homolog from Synechocystis 6803, we transformed a XZapII clone bank of Synechocystis genomic DNA into E. coli SASX41B and selected for ALA prototrophy of SASX41B by the recombinant phagemids on LB/ampicillin plates. Nineteen complemented transformants were obtained by this pro-Glutamyl-tRNA Reductase from E. coli and Synechocystis 6803 cedure. Plasmid DNA purified from these strains simultaneously conferred ampicillin resistance and ALA prototrophy to SASX41B cells upon re-transformation.
One plasmid, designated pSH11, was analyzed in detail, and a partial restriction endonuclease map of the 6.5-kb EcoRI insert of plasmid pSH 11 is depicted in Fig. 3. Total genomic DNA extracted from Synechocystis 6803 was double-digested to completion with BamHI and HindIII, size-fractionated on a 1% agarose gel, transferred to cellulose nitrate, and hybridized to the 32P-labeled DNA of the EcoRI insert of pSH11. As expected, this probe hybridized to two BamHI/HindIII fragments; a 2.2-kb fragment that corresponded to the fragment present on pSHll and an approximately 4.3-kb fragment that would correspond to a fragment delimited by the HindIII site present on the insert and the next BamHI or HindIII site present on the Synechocystis genome (data not shown). Since no other hybridizing bands could be observed, these data are consistent with the presence of one copy of this gene in the Synechocystis genome.
In order to localize the h e d gene within the 6.5-kb EcoRI fragment, subclones were constructed that contained either the 4.0-kb EcoRIIHind I11 fragment (pSH12) or the 2.5-kb EcoRI/HindIII (pSH13) in pBluescript (Fig. 3). Neither plasmid complemented the h e d mutation, indicating that HindIII cleaves the DNA within the region required for expression of the putative Synechocystis hemA gene. A 1.4-kb TfiI fragment (Fig. 3), subcloned into the EcoRV site of pBluescript SK-to construct plasmid pSH17, complemented the hemA mutation and therefore contained the hemA gene.
DNA sequence analysis identified an open reading frame capable of encoding a polypeptide of 427 amino acids with a molecular weight of 47,525. The DNA sequence has been deposited with GenBank (accession number M84218). The deduced amino acid sequence exhibited substantial similarity to the HemA polypeptides reported from E. coli, B. subtilis, S. typhimurium, and C. uibrwforme (Fig. 4).
In order to characterize the product of the Synechocystis 6803 gene, GluTR assays were performed on protein extracts prepared from strains SASX41B/pBluescript (empty vector) and SASX41B/pSH17. The plasmid pSH17 contains the entire Synechocystis hemA gene with the upstream ribosome binding site and a 3' flank extending approximately 110 nucleotides downstream of the stop codon (Fig. 3). The gene is presumably transcribed from the lac2 promoter.
From the functional complementation results we expected that the Synechocystis 6803 enzyme was expressed in E. coli and therefore wanted to determine if GluTR activity could be demonstrated in uitro. Since the GluTR activity in E. coli consists of GluTR45 and GluTR85, we used strain SASX41B in which the hem4 gene, and therefore GluTR45 is inactive for these experiments. Using SASX4lB/pBluescript as control, the in uitro assays showed that 4 pmol of ALA above background were formed by extracts prepared from SASX41B/pSH17.

DISCUSSION
The knowledge that h e d encodes GluTR45 and that it can be overexpressed allows a direct experimental attack on two questions regarding this key enzyme in chlorophyll and heme synthesis. Obviously, the 46-kDa protein possesses reductase activity (including an NADPH binding site) and is able to recognize tRNAG'" in a sequence-specific manner (3). As both glutamyl-tRNA synthetase (36) and GluTR are able to recognize tRNAG'", it will be interesting to determine their respective tRNA identities (i.e. which nucleotides in tRNA are important for specific recognition by these proteins). The sequence comparison of the presently known GluTR enzymes (Fig. 4) reveals some highly conserved regions. For instance, there is a sequence of 23 amino acids (positions 99-121) in which 20 amino acids are conserved. Within this region, a mutation of Cys'" to Tyr in B. subtilis leads to loss of enzyme activity resulting in a h e d phenotype (18).
A second question concerns the regulation of this enzyme. Clearly, as demonstrated by the h e d phenotype, GluTR is the crucial enzyme for ALA formation in aerobically grown E. coli cells. There is currently no information on the extent of transcriptional and translational regulation of the h e & gene. A number of experiments have indicated that regulation by intermediates of later steps in the heme pathway is certainly a possibility (37). In addition, the fact that good overexpression in E. coli is not observed may indicate that proteolysis may regulate the level of GluTR as is the case for ALAsynthase, the corresponding enzyme in the Shemin pathway in mammals (38).
The different sizes of the known GluTR enzymes also merit further study. Although the five bacterial proteins for which the gene sequences are known are very similar (see Fig. 4) the enzyme recently purified from Synechocystis 6803 (10) shows a size inconsistent with that of the protein deduced from our DNA sequence. In addition, the 42-amino acid N-terminal sequence determined for the Synechocystis enzyme is not represented in the gene sequence reported here. It may also be pertinent to note that the Synechocystis 6803 GluTR is relatively abundant, as evidenced by a 370-fold purification factor (lo), whereas the enzymes from E. coli and C. reinhrdtii require a 5-10-fold higher factor for obtaining homogeneous preparations (9, 11). The difference in these results may be resolved by further characterization of Synechocystis GluTR enzyme preparations or may point to the existence of two reductase activities in this organism.
Why are there two GluTR activities in E. coli? The observation of two GluTR activities has so far only been reported for E. coli. The demonstration that hemA encodes GluTR45 and that a hemA-strain contains only GluTR85 suggests that there are two pathways for ALA formation in E. coli. In line with this notion is the fact that hemL strains of E. coli lacking GSA aminotransferase display a leaky phenotype regarding ALA auxotrophy, implying the presence of another aminotransferase which can utilize GSA or the existence of a compensatory pathway for ALA formation (30); It may also be pertinent that there are two pools of ALA found m cucumber chloroplasts (39), which may be due to different pathways for their formation.