The sulfate activation locus of Escherichia coli K12: cloning, genetic, and enzymatic characterization.

The sulfate activation locus of Escherichia coli K12 has been cloned by complementation. The genes and gene products of this locus have been characterized by correlating the enzyme activity, complementation patterns, and polypeptides associated with subclones of the cloned DNA. The enzymes of the sulfate activation pathway, ATP sulfurylase (ATP:sulfate adenylyltransferase, EC 2.7.7.4) and APS kinase (ATP:adenosine-5'-phosphosulfate 3'-phosphotransferase, EC 2.7.1.25) have been overproduced approximately 100-fold. Overproduction of ATP sulfurylase requires the expression of both the cysD gene, encoding a 27-kDa polypeptide, and a previously unidentified gene, denoted cysN, which encodes a 62-kDa polypeptide. Purification of ATP sulfurylase to homogeneity reveals that the enzyme is composed of two types of subunits which are encoded by cysD and cysN. Insertion of a kanamycin resistance gene into plasmid or chromosomal cysN prevents sulfate activation and decreases expression of the downstream cysC gene. cysC appears to be the APS kinase structural gene and encodes a 21-kDa polypeptide. The genes are adjacent and are transcribed counterclockwise on the E. coli chromosome in the order cysDNC. cysN and cysC are within the same operon and cysDNC are not in an operon containing cysHIJ.

Despite the importance of activated sulfate as an obligate intermediate in sulfate metabolism, little is known of the detailed enzymatic mechanisms which catalyze its' formation. ATP sulfurylase has not been purified from a procaryotic organism, and the partial purification of APS kinase activity from Escherichia coli has only recently been reported (5).
The genetics and metabolism of activated sulfate have been characterized in E. coli and Salmonella typhimurium. In E. coli, cysD mutants are devoid of ATP sulfurylase activity, whereas cysC mutants lack APS kinase activity (6). These mutants were isolated by virtue of their growth requirement for cysteine (7). The cysC and cysD genes are tightly clustered along with cysH, cysi, and cysJ between 59 and 60 min on the E . coli K12 chromosomal map (8). The cysH and cysiJ genes are responsible for PAPS reductase and sulfite reductase activities, respectively (8). Although the genes of the cysCDHIJ cluster show positive and coincident regulation by the cysB gene product (6, 9), information regarding gene organization and operon structure(s) within this cluster is lacking in E, coli.
In studies directed toward characterization of the genes and proteins involved in sulfate activation, we have cloned the E. coli ATP sulfurylase structural genes and purified the enzyme to homogeneity. In so doing, a previously unidentified gene, denoted cysN, was discovered. ATP sulfurylase is composed of two types of subunits which are encoded by cysD and cysN. We have also cloned the E. coli cysC gene which appears to be the structural gene for APS kinase. In addition, an operon which includes cysNC and excludes cysHIJ has been identified.

RESULTS
Cloning of cysC and cysD-The cysC and cysD mutations present in strains JM81A and TSL1, respectively, prevent the anabolic utilization of sulfate (18). Growth of these strains on media containing sulfate as the only sulfur source provides a positive selection for cys+ transformants. cysD was cloned by transforming strain TSLl to a cys+ phenotype with a bacteriophage X library of HindIII digested E. coli K12 chromosomal DNA. The isolated transforming phage, XTL1, complemented both the cysC and cysD mutations of strains JM81A and TSL1, but did not complement the cysH mutation of strain JM96. Analysis of HindIII digested XTLl DNA revealed that the phage DNA contained a 9.4-kb insert.
Gene and Gene Product Characterization-The strategy employed in mapping and characterizing the cloned sulfate activation genes and gene products involved correlating the complementation patterns, ATP sulfurylase and APS kinase activities, and polypeptides associated with various subclones of the 9.4-kb insert. Thus, the presence or absence of a particular cloned gene product may be associated with that of a specific enzyme activity, known genetic allele, and region of the 9.4-kb insert. The plasmids used in these studies are illustrated in Fig. 1. The associated complementation and enzyme activity data are presented in Tables I and 11. Lanes 1-6 of Fig. 2 show the results of studies in which [35S] methionine was incorporated in vivo into the polypeptides expressed from the plasmids pTL1-pTL6 depicted in Fig. 1. The plasmids listed in the left-hand column are composed of the insert shown and either vector pT7-5 (contained in pTL1,2,3,5,7, and pJT1) or pT7-6 (found in pTL4,6, and 8) (see "Materials and Methods''). The proximity and orientation of the T7@10 promoter with respect to the cys genes is indicated by a right angle arrow at an insert terminus. The approximate coding regions of the cys genes are shown as labeled black rectangles. The positions and sizes of the coding regions are estimated from restriction mapping and apparent molecular weights of the gene products.   Fig. 2 and associated text).
' ATP sulfurylase and APS kinase activities were determined using the radioactive assays described in the text. The extracts assayed were prepared from the strain JM83 harboring the indicated plasmid and pGP1-2. The activity units are expressed as nmol/min/mg extract protein.
Methods." e The complementation protocol is described under "Materials and Plasmid pT7-5 contains no cloned DNA and was used as the control plasmid. e None detected.
ATP sulfurylase and APS kinase activities were determined using the radioactive assays described in the text. The extracts assayed were prepared from the strain CL510, the precursor of DM62, harboring the indicated plasmids. The activity units are expressed as nmol/min/mg extract protein.
* The complementation protocol is described under "Materials and Methods." Plasmid pT7-5 contains no cloned DNA and was used as the control plasmid.
The incorporation of radioactive amino acids was accomplished using the pT7/pGP1-2 dual vector system (see "Materials and Methods").
Polypeptides Expressed from the 9.4-kb Clone-Plasmids pTLl and pTL2, illustrated in Fig. 1, contain the 9.4-kb insert in opposite orientations with respect to the T7$10 promoter of the vector pT7-5. The radioactively labeled proteins expressed from pTL1 and pTL2 are shown in lanes 1 and 2 of Fig. 2. These proteins, indicated in Fig. 2 as CysN, CysD, CysC, and 1-6, appear using either [35S]methionine or a 14Camino-acid mixture as the radioactive label. The apparent molecular masses of cysN, cysD, cysC proteins and proteins 1-6 are 62, 27, 21 kDa and 39, 29, 26, 25, 20, and 15 kDa, respectively. Proteins 1-6 derive from coding regions in the vicinity of the cysCDHIJ gene cluster. Proteins 1-6 did not appear to be involved in sulfate activation and were not investigated further in the current studies. Lanes 3-6 of Fig.  2 show the labeled proteins expressed from the 9.4-kb subclones pTL3-pTL6 (see Fig. 1). The coding regions for cysN, cysD, cysC proteins and proteins 2-6 are on the same strand of the 9.4-kb insert; whereas, protein 1 is encoded on the complementary strand. Thus, the protein 1 coding region is transcribed antiparallel to that of the coding regions for the other labeled proteins. Maxicell labeling techniques were used to map protein 1 to the 3.2-kb HindIII-EcoRI fragment of the 9.4-kb insert (not shown). Mapping, Gene Product Identification, and Characterization of CysC-To map the position of cysC on the cloned 9.4-kb fragment, a series of subclones were constructed in pT7 vectors. The subclone inserts share a common PstI endpoint and were progressively deleted from the ClaI site of the PstI-ClaI fragment of the 9.4-kb clone using Bal31 exonuclease (see Fig.  1 and "Materials and Methods"). Two such plasmids, pTL3 and pTL4 (see Fig. l ) , differ in that the insert of pTL4 is -800 base pairs shorter than that of pTL3. ["SIMethionine labeling of the proteins encoded by these plasmids demonstrates that pTL4 expresses five of the six polypeptides expressed by pTL3 (see lanes 3 and 4 of Fig. 2). The polypeptide not expressed by pTL4 is of apparent molecular mass 21 kDa. Thus, some fraction of the coding region for the 21-kDa polypeptide has been deleted in pTL4. Table  I shows that some part of the -800 base pairs removed from pTL3 is required for complementation of the cysC allele of JM81A. The concomitant loss of -800 base pairs of DNA, expression of the 21-kDa polypeptide, and the ability to complement cysC suggests that the 21-kDa polypeptide is the cysC gene product. Plasmids pTL1, containing the entire cysC coding region, and pTL3 appear to express the same 21-kDa polypeptide (see Figs. 1 and 2). This indicates that the 21-kDa protein is not artificially truncated due to Bal31 deletion of the cysC coding region of pTL3. Extracts of cells harboring pTL3 exhibit a -32-fold increase in APS kinase specific activity as compared to control extracts (see Table I). Extracts of cells harboring pTL4 show approximately wild type levels of APS kinase activity (see Table I). In light of the mapping and complementation studies, the activity data suggest that cysC is the structural gene for APS kinase and that the 21-kDa protein is the APS kinase protomer (see "Discussion").
Mapping, Gene Product Identification, and Characterization of cysD-Plasmid pTL5 was constructed for the characteriza-tion of cysD (see "Materials and Methods"). This plasmid is composed of a -1.1-kb cysD complementing fragment of the 9.4-kb clone inserted into pT7-5 (see Fig. 1). ["S]Methionine incorporation studies indicated that the pTL5 insert encodes a single polypeptide which is of apparent molecular mass 27 kDa (see Fig. 2, lane 5). Plasmid pTL6 neither complements strain TSL3 nor expresses the 27-kDa polypeptide at levels detectable by [35S]methionine labeling. These data map cysD within a -1.1-kb region of the 9.4-kb clone and strongly suggest that the 27-kDa polypeptide is the cysD gene product.
The levels of ATP sulfurylase activity in extracts of cells containing plasmid-encoded cysD protein differ by as much as two orders of magnitude (see Tables I and 11). Although cysD protein is not visible in extracts of control strains electrophoresed on SDS polyacrylamide gels and stained with Coomassie Brilliant Blue, it is detected in extracts of strains containing plasmid-expressed cysD. The plasmids pTL5 and p J T l ( Fig. 1) express detectable levels of cysD protein, yet the ATP sulfurylase activity in extracts of strains containing these plasmids approximate the levels found in extracts of control strains (Tables I and 11). Plasmid pTL1, which contains the full 9.4-kb insert, expresses nearly a -100-fold increase in ATP sulfurylase activity over control (Table 11).
Thus, the entire insert contains the genetic determinants necessary for the overproduction of ATP sulfurylase activity and expression of cysD alone is inadequate for this purpose.
In attempting to isolate the sequences sufficient for overproduction of enzyme activity, numerous subclones of the 9.4kb insert were constructed, all of which contained cysD, the presumed ATP sulfurylase structural gene (18) (see "Materials and Methods"). Plasmid pTL4 (Fig. l ) , isolated from these constructions, is our smallest isolate which overproduces ATP sulfurylase activity. Extracts of strains cont,aining pTL4 show a 60 fold increase in ATP sulfurylase activity over the pT7-5 control (Table I). Plasmid pTL4 expresses only the cysD protein and the 62-kDa polypeptide. To verify that the 62-kDa polypeptide was not solely responsible for the enhanced enzyme activity associated with pTL4, pTL6 was constructed (Fig. 1). Plasmid pTL6 differs from pTL4 in that part of the cysD coding region has been deleted. Although extracts of cells harboring pTL6 show no apparent increase in ATP sulfurylase activity (Table I), expression of the 62-kDa protein is easily detected by Coomassie Brilliant Blue staining of acrylamide gels containing these extracts. Thus, the enhanced expression of both the cysD and 62-kDa proteins is required for overproduction of ATP sulfurylase activity. These observations do not discriminate between regulatory versus catalytic roles for these proteins; however, it is clear that the 62-kDa protein plays a critical role in the overproduction of ATP sulfurylase activity.
Identification of cysN, a Second Gene Required for ATP Sulfurylase Overproduction-To further investigate the function of the 62-kDa protein in the expression of ATP sulfurylase activity, a strain harboring a mutation in the 62-kDa coding region was constructed. In this strain, DM62, the gene which confers kanamycin resistance (kan) has been inserted into the 62-kDa coding region of the chromosome by homologous recombination using the PuuII fragment of p J T l (see "Materials and Methods" and Fig. 1). Strain DM62 is a cysteine auxotroph. The complementation studies, which characterize DM62 and address the essential role of the 62-kDa protein in sulfate activation, are shown in Table 11. The plasmids used in these studies express different subsets of the cloned genes (see Fig. 1 and "Materials and Methods"). Plasmid pTL3 which complements DM62 encodes only the cysCD and 62-kDa proteins (see lane 4, Fig. 2). Plasmid pTL7 which contains intact cysC and 62-kDa coding regions and a partially deleted cysD also complements DM62. Strain DM62 is not complemented by pTL6 which encodes only the 62-kDa protein (see lune 6, Fig. 2). Plasmid pTL8, containing the cysC coding region and lacking part or all of the cysD and 62-kDa coding regions, does not complement DM62. Taken together, these data indicate that both the cysC and the 62-kDa coding regions are necessary and sufficient for the cys complementation of DM62. That pTL8 complements the cysC allele of JM81A but does not complement DM62 verifies that the 62-kDa polypeptide is essential for sulfate activation in E. coli. As such, the gene encoding the 62-kDa polypeptide was previously unidentified in the sulfate activation locus of E. coli. The mnemonic cysN has been chosen to represent this gene.
The insertion of h n into chromosomal cysN inactivated cysC (Table 11). This tzan-mediated polar inactivation of cysC suggested that cysN and cysC reside within the same operon. To exclude the possibility that cysC inactivation was an artifact of strain construction, enzyme activity and complementation studies were performed using strains containing p J T l (see Table I1 and Fig. 1). Plasmid pJT1, derived from pTLl by insertion of kan in cysN, was the plasmid used in the construction of DM62. Restriction mapping showed that Determined by Bradford analysis (see "Materials and Methods"). b A T P sulfurylase activity was determined using the continuous The crude extract was prepared from 12 g (wet weight) of E. coli sulfate sulfate S-300 assay (see "Materials and Methods").
JM83 containing pGP1-2 and pTL3 (see Fig. 1 the transcriptional orientation of kan in pJTl is parallel to that of cysDNC. The enzyme assays demonstrate that lzan insertion into cysN reduces the expression of APS kinase activity at least 127-fold compared to that associated with pTLl (see Table 11). The complementation studies (see Table  11) show that pJT1, which carries an intact cysC coding region, cannot complement the cysC allele of JM81A. The activity and complementation data attest a strong polarity between cysN and cysC. This polarity indicates that cysN and cysC lie within the same operon (35). It is significant that plasmid pTL8, which does not contain most or all of the cysD and cysN coding regions, complements the cysC allele of JM81A (see Fig. 2 and Table 11). This suggests that either an internal cysC promoter exists within the operon or that transcription of cysC initiates at promoter(s) within the pT7-5 vector of pTL8. That pJTl does not complement JM8lA argues against the possibility of a promoter located between cysN and cysC. cysD and cysN Proteins Are Subunits of ATP Sulfuryluse-Having established an essential role for the cysN protein in sulfate metabolism and the interdependence of the cysD and cysN proteins in overproducing ATP sulfurylase activity, it was of interest to more acutely probe the functional relationship of these polypeptides. The results presented thus far suggested that these polypeptides could be subunits of ATP sulfurylase. To address this possibility, and for future mechanistic studies, the enzyme was purified to apparent homogeneity (see "Materials and Methods"). The enzyme purification involved streptomycin sulfate and ammonium sulfate precipitation followed by gel filtration and anion exchange chromatography. The procedure resulted in a 40-fold increase in the specific activity of ATP sulfurylase over that found in extracts of strains which overproduce this enzyme (see Table   111 and "Materials and Methods"). A photograph of a COOmassie Blue-stained SDS polyacrylamide gel containing the purified enzyme and extracts of control and overproducing strains is shown in Fig. 3. The specific activity of the purified ATP sulfurylase was 0.23 pmol/min/mg (see "Materials and Methods"). Two peptides copurify with ATP sulfurylase activity through the steps of the purification protocol (Fig. 3,  lane 1). These proteins were identified as the cysD and cysN peptides by co-chromatography during SDS-PAGE with plasmid-expressed [35S]methionine-labeled cysD and cysN proteins.
To further verify that the cysD and cysN proteins are subunits of ATP sulfurylase, the co-migration of these proteins with enzyme activity was investigated. The A m and enzyme activity profiles obtained from gel filtration chromatography of the purified enzyme areghown in Fig. 4. The data suggest that both polypeptides co-migrate with the enzyme activity. The comparable Coomassie staining intensities of these polypeptides (Fig. 3, lane 1 ) indicate that separation of these proteins would have been detected. Furthermore, SDS-PAGE of the eluant fractions indicates that the polypeptides are found solely in fractions containing enzyme activity and always in a stoichiometry comparable to that seen in lane 1 of Fig. 3. Similar studies were performed using a variety of different chromatographic matrices including: phenyl-Sepharose CL-4B; the dye ligand matrices blue-Sepharose, red A, and green A (36); and the Q Sepharose anion exchange matrix. These studies also indicate that the cysD and cysN peptides co-migrate with ATP sulfurylase activity. Thus, genetic and activity studies have shown a requirement for enhanced expression of both cysD and cysN in attaining overproduction of ATP sulfurylase activity (see above), and chromatographic data suggest that both peptides co-migrate with enzyme activity on several types of chromatographic matrices. Taken together, these data indicate that cysD and cysN encode the subunits of ATP sulfurylase and are therefore the structural genes for this enzyme.

DISCUSSION
The current study offers considerable information regarding the genetic organization of the cysDCHZJ cluster located at -59 min on the E. coli chromosomal map (8). Previous studies of the cys cluster failed to identify cysN (18). It is likely that this was due to a paucity of well-defined mutants and the relative inaccuracy of traditional mapping techniques. Our studies map cysN between cysD and cysC thus redefining the cluster as cysDNCHZJ. Construction and characterization of a cysN mutant has revealed that cysN is essential for the utilization of sulfate as a sole sulfur source. The relative positions and transcriptional orientations of cysDNC have been determined. From the known positions of cysD and cysC on the E. coli chromosome (8), it is apparent that cysDNC is transcribed in a counterclockwise orientation on the chromosome. Insertion of kan into cysN dramatically decreases expression of cysC. This demonstration of polarity between cysN and cysC indicates that these genes belong to the same operon. The contiguous, parallel structure of cysDNC, the functional relatedness of cysD and cysN, and the coincident metabolic regulation of ATP sulfurylase and APS kinase suggest the possibility that cysD lies within the operon containing cysN and cysC. Although the literature generally supports the likelihood that enzyme subunit structural genes are contained within the same operon (37-41), exceptions exist (42, 43). Classical mapping studies of the cysCNDHIJ gene cluster suggest a genetically silent region between cysCD and CYSHIJ (18), similar to that identified in S. typhimurium (44).
The 39-kDa polypeptide expressed from pTLZ lies within this region and is transcribed antiparallel to cysDNC. Thus, cysHIJ, while clearly part of a regulon which includes cysCNDHIJ, is not within the operon which includes cysNC. Separate operons encoding cysCD and cysHIJ have been identified in S. typhimurium (45).
Purification of ATP sulfurylase has revealed that the enzyme is composed of two types of subunits. Furthermore, we have demonstrated that the polypeptides encoded by cysD and cysN are the subunits of ATP sulfurylase. These studies have defined a functional role for the cysN protein and verified that cysD and cysN are the structural genes for ATP sulfurylase. cysD and cysN mutants do not require plasmids containing both cysD and cysN for complementation. This indicates that the subunit interactions required for ATP sulfurylase activity occur when the wild type polypeptides are expressed from separate mRNAs. In contrast to the dissimilar subunit structure of ATP sulfurylase from E. coli, the enzyme from Penicillium chrysogenurn appears to be composed of identical 56-kDa subunits (46). The Saccharomyces cerevisiae ATP sulfurylase appears to be the product of a single gene consistent with identical subunits (47).
Plasmids which complement cysC confer high levels of APS kinase activity and express a 21-kDa peptide. These data suggest that cysC is the structural gene for APS kinase; however, they are also consistent with cysC encoding a trans activator of APS kinase. Although numerous mutants in the cysCNDHIJ cluster exist in S. typhimurium and E. coli, genetic and biochemical studies have failed to reveal an APS kinase regulatory gene in this cluster (18,48). APS kinase has been partially purified from E. coli (5). The authors suggest that the apparent molecular mass of the APS kinase subunit is 40 kDa and quote a specific activity of -600 nmol of PAPS produced/min/mg for the purified enzyme. Recently, the 21-kDa cysC polypeptide has been purified to near homogeneity and shown to have an APS kinase specific activity of -IO4 nmol of PAPS produced/min/mg,4 suggesting that cysC is the structural gene for APS kinase. of E. coli: Genetics and Enzymology 13. 14.