The hisT-purF region of the Escherichia coli K-12 chromosome. Identification of additional genes of the hisT and purF operons.

A 9.7-kilobase pair segment of the Escherichia coli chromosome spanning the hisT and purF loci has been characterized. Six genes were identified in this region by complete DNA sequence analysis, in vivo expression in maxicells, and RNA transcript analysis. S1 nuclease analysis has demonstrated that some of these genes are part of the hisT or purF operons. Two of the newly identified genes, dedA and dedB, were localized immediately downstream of hisT in the hisT operon. Two other genes, denoted dedC and dedD, have been localized between the hisT and purF operons. The other two genes, dedE and dedF flank the purF gene. dedE has been previously described as the first gene in the purF operon (Makaroff, C.A., and Zalkin, H. (1985) J. Biol. Chem. 260, 10378-10387). dedF was localized downstream from purF and is part of the purF operon. In addition, dedF is homologous to the ubiX gene of Salmonella typhimurium. Adjacent to dedF is the E. coli homologue of the S. typhimurium argT locus encoding the lysine/arginine/ornithine-binding protein. All of the genes in this region of the chromosome were found to be transcribed in a counter-clockwise direction on the E. coli map which revises the direction of purF transcription.

The nucleotide sequence(s) reported  operons (5)(6)(7). The hisT gene has been cloned and demonstrated to be a component of a multigene operon (8). Sequence and genetic analysis of a cloned DNA segment containing the hisT gene demonstrated that hisT and a gene of unknown function (denoted usg) are translationally coupled (9). S1 nuclease mapping indicated the possible existence of additional genes in the hisT operon (9). Tightly linked to hisT is the purF locus, which codes for glutamine:5-phospho-ribosy1amine:pyrophosphate phosphoribosyltransferase (EC 2.4.2.14), the first enzyme in de novo purine biosynthesis (10). This gene, like hisT, is a component of a polycistronic operon 111).
We have examined the region between the hisT and purF genes in order to characterize additional genes of the hisT operon. This segment of the chromosome has been characterized by DNA sequence analysis, in vivo expression in maxicells, and S1 nuclease analysis. We have defined two additional genes for the hisT operon and one for the purF operon. These two operons are separated by two other genes dedC and dedD. Finally, this study identified the E. coli homologs for the Salmonella typhimurium ubiX and argT loci.
The DNA analyzed in these studies was obtained from the previously described plasmid \k210. This plasmid consists of a 9.7-kb' E. coli Hind111 DNA fragment (derived from the Carbon-Clarke plasmid pLC28-44) inserted into pBR322. The pLC28-44 plasmid complements hisT-, hisJ-, and purFphenotypes (8,12). The sequence of the first 2.3 kb of q210 which contains the hisT gene in a multigene operon (9) and an additional 2.7 kilobases containing the purF gene in a multigene operon (10)(11) has been reported. In this report we describe the characterization of the remaining portion of the q210 plasmid.

The hkT-purF Region
kanamycin-resistant cassette inserted into the usg gene in the chromosome which has a hisT-phenotype, have been described (9). Plasmid constructions were cloned in JM83 (16) and plasmid pLC28-44 was originally in strain JA200 (17). Cells were grown in yeasttryptone media or Vogel-Bonner minimal media (18) plus 0.5% glucose with supplements where indicated.
Southern Analysis of pLC28-44 Plasmid DNA was isolated from E. coli JA200 transformed with pLC28-44. DNA (0.3 pg) was digested with 20 units of NaeI for 24 h at 37 "C. The digestion products were separated on a 0.6% Trisphosphate agarose gel and blotted onto nitrocellulose. The blots were hybridized to nick-translated radiolabeled (19) DNA in a buffer containing 1 M NaCl, 50 mM Tris-HC1, pH 8.0, 1 mM EDTA, 5 X Denhardt's, 0.1 mg/ml sonicated denatured salmon sperm DNA, and 0.1% sodium dodecyl sulfate. The blots were washed for 15 min at 25 "C then for 3-8 h at 65 "C in 2 X SSC, 0.1% sodium dodecyl sulfate with two changes of wash solution. Radioautography of the dried blots was at -80 "C with an intensifying screen.

Plasmid Construction
The construction of pLC28-44 is described by Clarke and Carbon (17) and construction of 9210 is described by Marvel et al. (8). Standard recombinant DNA procedures were used as described by Maniatis et al. (20). Each clone is discussed below. The locations of restriction endonuclease sites used for the constructions are shown in Fig. 3. Verification of the constructions was performed by restriction enzyme digestions and/or sequence analysis.
pKK-The 2396-bp KpnI fragment of 9210 was inserted into pUC18. The insert was oriented such that dedB transcription is opposite hZ' transcription.
PEE-The 2636-bp EcoRI fragment of 9210 was inserted into pUC8. The orientation was the same as pKK.
PEN-The clone, PEE, was digested with PstI and NruI and the 2626-bp fragment inserted into pBR322, which had been digested with EcoRI, made blunt by incubation with the large fragment of DNA polymerase I, then digested with PstI. This inactivated the blu gene and allowed selection by tetracycline resistance. pP1065, pP1180, and pP1508-3210 was digested with PvuI and the ends made blunt as above. HindIII linkers (BRL/Gibco) were added and the fragments cloned into pUC8. The resultant clones were screened by their insert size, and their orientations were confirmed by restriction endonuclease digestions. The clone pP1180 was oriented as in PEE and pKK. pP1508+ was oriented in the opposite direction. pP1065 had the orientation such that lucZ' transcription was similar to that of dedE.
pCS-A 2252-bp CluI to SalI fragment of 9210 was inserted into pBR322 at the homologous sites.
pSA-9210 was digested with SstII, diluted, and religated. The resultant plasmid had a deletion from 2967 to 7296.
pSP"pP1508 was digested with SalI which deleted a fragment from the SalI site of the vector to the SalI site at 4571. This created an in-frame fusion between the dedC and lucZ' genes. pS5500 and pS9000-9210 was partially digested with SalI, diluted, and religated. Clones were selected according to size, and restriction mapping confirmed the inserts. pS5500 contained a deletion from 622 bp in the vector's tetracycline resistance gene to the Sal1 site at position 7897 of 9210. pS9000 was similar to pS5500 except the deletion extended only to the Sun site at position 4571. pBgH-9210 was digested with BglII and BamHI, diluted, and religated. The resulting clone contained a deletion of 346 bp of the tetracycline resistance gene to the BglII site at position 8953 bp in the 3210 insert.
pSStu and pStuPst-9210 was digested with StuI, and either PstI or SalI. The fragments from 4571 to 6059 (SalI to StuI) and from 6059 to 9667 (StuI to PstI) were purified and cloned into SalI-SmuI or PstI-SmuI digested pUC9, respectively. The orientation of pSStu was such that lucZ' transcription was the same direction as dedD, and pStuPst was oriented such that lacZ' was opposite purF.
pEBg-9210 was digested with EcoRI and EglII, and the fragment from 3857 to 7677 was inserted into BamHI-EcoRI-restricted pUC8.
pBgBg-BglII and PstI digested 9210 was ligated into BamHIdigested pUC8 and the clones which contained the BglII fragment from position 7677 to position 8953 were confirmed by sequencing. pBgBg(+) was oriented such that the dedF transcription was the same as the hZ' transcription. pBgBg(-) had the opposite orientation.

DNA Sequencing
DNA was sequenced by the method of Sanger et al. (21), using M13mp8, M13mp9, M13mp18, and M13mp19 (22,23). Regions with dG-dC compression were sequenced by the substitution of dGTP with 7-deaza-dGTP (24) in the second-strand synthesis. Some regions were sequenced by the method of Maxam and Gilbert (25).

Plasmid-directed Protein Synthesis in Muxicells
Maxicells were prepared and analyzed according to Sancar et al. (14) with two modifications. Cells were labeled with a [14C]amino acid mixture instead of [35S]methionine and 0.1% phenylmethylsulfonyl fluoride was added to the cell lysis buffer to minimize degradation of labeled proteins. The protein products were electrophoresed on sodium dodecyl sulfate-polyacrylamide gels with a gradient of 12-16% in acrylamide. Gels were first stained with Coomassie Blue or soaked directly in dimethylsulfoxide for 30 min, in 20% 2,5-diphenyloxazole followed by dimethyl sulfoxide for 45 min and water for 60 min before drying and radioautography. SI Nuclease Mapping DNA from M13 single-stranded recombinant clones served as templates for the synthesis of radioactively labeled second strands complementary to transcripts from this region of the E. coli genome.
The oligonucleotide primer used for sequence determination was   (8). The arrows inside the circle show the locations of the mapped operons as described in the text. Panel C, hybridization of probes to NaeI-digested pLC28-44 DNA blotted onto nitrocellulose. The digestion yielded two major and one minor ethidium-bromidestaining fragments of 10, 9, and 8.5 kb, respectively. The smallest band is likely due to digestion at a refractory NaeI site external to the dotted region (panel B, in parentheses). Lane 1,250 ng of pLC28-44-digested DNA probed with radiolabeled 9210 insert DNA (corresponding to the dotted region shown in panel B). Hybridization was for 24 h a t 68 "C with 1 X lo6 cpm/ml of probe DNA. Lane 2, as in lune 1 except probe was a 730-bp EcoRI fragment shown as part of the hisT operon in panel B. Lane 3, as in lune 1 except probe was a HindIII fragment isolated from the S. typhimurium h i d gene. Hybridization was done at 60 "C.  annealed to single-stranded template DNA, and radiolabeled DNA was synthesized as described by Burgess and Penhoet (26). S1 nuclease analysis was performed using the method of Buchman et al.
(27) with the exception that the S1 nuclease digestion was incubated at 37 "C for 1 h. Each reaction consisted of 25 pg of E. coli RNA isolated by the hot phenol method (28) from strains grown in Vogel-Bonner minimal media, 200 ng of primer-extended radiolabeled probe DNA and 3000 units of S1 nuclease. Protected DNA was electrophoresed in 2.0% neutral Tris-borate agarose gels and 2.0% alkali agarose gels as described by Maniatis et al. (20).

Computer Analysis
Analysis and manipulation of sequence data were performed on a DEC VAX-780 and VAX-750 computers using sequence analysis software developed by Dr. Hugo M. Martinez (Biomathematics Computation Laboratory, Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco). Homology searches used the program "dbalign" which was run against GenBank, release 35.0, 1 August 1985.
Additional analysis was performed using the ANALSEQ package (Whitehead Institute for Biomedical Research, Cambridge MA) for running the positional base preference program of Staden (29).

Orientation of the q210 Insert on the E. coli Chromosome
In E. coli, purF and hisT have been genetically mapped near hisJ in the order of HisJ-purF-hisT (30). In addition, phage lambda integrations had been used to determine a clockwise direction of transcription for the purF gene (15). However, our sequence analysis demonstrated that purF was transcribed away from hisT, and thus counter-clockwise using the present gene order (Fig. 1, panel A ) . This contradiction could be resolved by assuming a revised gene order of hisJ-hisT-purF which would allow clockwise transcription of purF, away from hisT. T o resolve this issue a restriction digestion of pLC28-44 DNA was analyzed with hybridization probes for hisT and hisJ. Restriction digestion of pLc28-44 with NaeI yielded three fragments. The smaller two fragments must have derived from the same region of the plasmid as they both hybridized to a probe for hisJ from S. typhimurium, and the combined sizes of all three was greater than the size of W H O . This doublet is likely due to a site which is refractory to NaeI digestion, such as is found in pBR322. The other fragment of 10 kb hybridized to the hisT probe ( Fig. 1, panels B and C). T h e gene probes clearly hybridized to two separate fragments, and since a known NaeI site is present in purF (IO), the two genes must be separated by purF. Therefore, hisJ does not map between purF and hisT and the gene order is hisJ-purF-hisT, which forces the conclusion that purF is transcribed counter-clockwise on the E. coli map.

Sequencing of q210
The Sanger dideoxy DNA-sequencing method was used with M13 vectors t o sequence the 4.5 kb separatingpurF and hisT and the 1. The sequence determined on both sides of purF overlapped the published data (10,11).

Analysis of Coding Probabilities of Open Reading Frames Identified by DNA Sequencing
Demonstration that the open reading frames in the DNA sequence were consistent with protein-coding regions in E. coli was done by computer analysis. The first analysis was a positional base preference method described by Staden (29). This program identifies coding sequence by amino acid composition with no weighting for codon preference. Results of this analysis are shown in Fig. 4. Nine peaks representing

GCC CGA ~~C T~~~~~G~C~~~~C C~C T~G C P L ' P G C C C T ' I T T G C P A~C P C A C C G T A A C G A A~C A G C G W ' P A C C
CGT CTT GGC Grc CTG AAA CCA GCG CCA CTT GTG TIP ACC GTT GCG GGT ACG M r GGC AM GGC ACC ACC r G c CGT ACG crc GAG TCG wrr 4 1 0 6   (11). The braces between the same genes delineates a repetitive extragenic palindromic sequence (11,37) identified by a homology search in GenBank. The regions homologous to ubiX and argT are denoted by brackets. The numbers at the right correspond to the nucleotide positions of the entire 4210 insert beginning at the Hind111 site in the hisT operon of 9210. The sequence from position 1 to 2164 and between 6897 and 8246 has been previously published (9-11) and is omitted from the depicted sequence.

60Ll OC TAA T A C G G T C . $ G C C P G A P G C C C~G G C G C G P C T P A T W~C A G G G G . P A W A C C G T A G G T~f f i A T A A G G C G T T T A C G C C G C A~C C G I C~G C A T T G C C C G A P G C C G C A M 6134
I" I t

T CAG ACC G T P R A P CGT G T P C'PT G I C CAG TTT GCG ATA ACC CTT CCT GAA GAP CTC TTP GCC CGC TGG CAG GGC GCA T M C T C C W 8989 01 II
GTTCCCCPGTTPCAGVICAAlT'PTGCMCCGCGAPCAMTCCN1GACATTTTG~TTTGCCATTCAA~GIMCGCTGCGA~AACCGCTAPACCTGCTATCTTCAACTTCAGGAC~AP 9109 "  Table I. The codon usage was expressed as a percentage of the most optimum codon usage possible for E. coli (31). All of the transcripts encoded on 9210 had values from 0.6 to 0.76. These values were similar to the lacy and trpB genes but less than the values for highly expressed genes. This analysis demonstrated that the open reading frames identified by DNA sequencing were probable protein-coding regions. Verification that these DNA regions actually code for proteins was done by analysis in a maxicell protein expression system (14).

Location of Nine Genes Expressed in Maxicells in Q210
Maxicell assays revealed at least nine proteins encoded on 9210 ranging in molecular weight from 19 to 56 kDa (Fig. 5 ,  panels A-C, lane 1). Of these nine, four have been described previously; hisT, usg (8,9), purF (10) and one upstream from purF (11). The genes characterized in this report have been designated dedA through dedF for downstream E. coli DNA (from hisT).
Extensive subcloning of Q210 followed by maxicell analysis confirmed the location, size, and orientation of each of the ded genes. The subclone designations used refer to a DNA fragment cloned into a plasmid vector and transformed into the maxicell strain CSR603. The insert DNA of plasmids used in these experiments are depicted by the solid bars in Fig. 2. dedA-The position of the dedA gene was determined by its presence in the maxicell assays of PEN and pSA shown in Fig. 5, panel A , lanes 5 and 7. The assay of pSA localized dedA between hkT and the SstII restriction site. The direction of transcription was indicated by the maxicell assay of pCS, which implicated a truncation of the dedA promoter. dedB-The position of the dedB protein is shown in panel A , lanes 5-7. dedB protein was present in maxicell assays of pCS and absent in maxicell assays of pSA. This located the gene between the SstII site and the SalI site. The assay of PEN revealed a truncation product of 27 kDa in place of the native dedB product. The direction of transcription for dedB gene was toward purF.
dedC-The position of dedC was defined by the maxicell assays of pP1508+ and pSP shown in panel B, lanes 3 and 4. The assay of pP1508+ contained a full-sized dedC gene product. This and dedCs absence in assays of pSA located the gene between dedB and the PuuI site at position 5392. Assays of pSP gave a smaller protein of 37 kDa. This provided evidence for the direction of transcription, as an in-frame fusion was created between the lacZ' reading frame of pUC8 and the dedC-coding region which accounted for the dedC* protein product.
dedD-The position of dedD was determined by maxicell assaying pS9000 and pEBg as shown in panel D, lanes 1 and

TABLE I
Calculated optimal codon usage frequency of the genes on 9210 Codon usage frequency of the ded and other genes expressed on 9210, shown on the left columns, expressed as a percentage of the most optimal codon usage according to the rules of Ikemura (31). Values calculated for low and high expressed genes are shown in the right hand column for comparison. The analysis of the pSStu clone revealed the dedD product, whereas that of pStuPst revealed the presence of the dedE, purF, and dedF, but no products originating from dedD. This evidence located dedD between dedC and dedE. The results from the pStuPst clone showed that the genes of the purF operon were transcribed independently of dedD. Evidence for the direction of transcription of dedD toward purF was determined due to the singular open reading frame present in the DNA sequence (Fig. 4). The data do not allow an unambiguous prediction of the translational start site for the dedD protein. The open reading frame between the end of the dedC gene and the StuI site did not contain a AUG codon until position 5519. This AUG is not thought to be used, as the clone pP1065 did not express a protein from the transcript initiated from the lacZ' promoter of this plasmid. There exist several GTG codons in the dedD reading frame near the end of the dedC open reading frame. Two have conserved Shine-Dalgarno ribosome-binding sites (35) upstream from a GTG codon, although the furthest one would cause an overlap of four codons of the dedC gene (See Fig. 3). dedE-The position of dedE was defined as shown in panel D, lanes 1-6. The absence of dedD and the presence of dedE in the maxicell assay of the pStuPst clone (lane 5 ) localized the dedE gene between dedD and purF. The assay of pP1065 revealed a truncation product of 12 kDa (dedE*), which was consistent with a truncation of the dedE gene. This gene was identical to the first protein in the purF operon described previously (11).
dedF-The location of dedF was determined by maxicell assays of the clones shown in panel C, lanes 2-5. Expression of dedF was independent of purF as shown by assays of the pS5500 clone. This result localized the gene between the SalI site at position 7895 and the end of the q210 insert. The gene product was not expressed by the pBgH clone (lane 5 ) , which eliminated the possibility that the dedF product represented the amino terminus of a larger gene which extended outside of q210. The location and the direction of transcription were defined by assays of the clones pBgBg+ and pBgBg- (lanes 3  and 4 ) . Proteins of 25 kDa and a 23.5 kDa could be explained as fusion products between the dedF reading frame and the lacZ' polypeptide or an open reading frame on the opposite strand of the lacZ', respectively (dedF3). In this analysis the FIG. 5. Maxicell plasmid directed in vivo protein synthesis. The products were analyzed by electrophoresis on 12-16% sodium dodecyl sulfate-polyacrylamide gels. Marker proteins of known molecular weight were coelectrophoresed, and their positions are given at the right of panel A . The arrows to the left in each panel denote the positions of the major protein products seen in maxicell experiments using the entire 9210 plasmid ( l a n e 1, all panels). The products produced from selection markers on the vectors were either tet or bla gene products. The p-lactamase products can appear as one, two, or three different molecular species; 30 and 31 kDa mature products and a 32.5 kDa precursor. The tetracycline gene product is a protein migrating at 36 kDa when present as a selection marker (panel A  The asterisk (*) denoted truncated protein products and the f denotes fusion gene products. direction of transcription was the same as purF. Analysis of the DNA sequence of q210 predicted open reading frames coding for the ded proteins A-F, of molecular mass 24,500, 33,400, 45,500, 23,000, 17,900, and 20,800 daltons, respectively. The apparent molecular weights of the proteins based on sodium dodecyl sulfate-polyacrylamide gel electrophoresis were 19,500, 33,000, 49,000, 35,500, 16,500, and 21,000, respectively. Except for dedA and dedD the size of each open reading frame was within 10% agreement with its apparent molecular weight. Amino acid compositions may account for these differences, for instance, dedD is proline rich (14%). These effects are similar to what has been described for the usg protein of the hisT operon (8). The deduced amino acid sequence of each protein is found in Fig. 3 above the DNA sequence of their respective genes.
Arps et al. (9) have shown that usg and hisT genes are transcribed on a single RNA. In order to test if the dedA gene is coordinately transcribed with hkT, a single-stranded DNA clone bridging the hisT-dedA reading frames (CK-3) was used in S1 nuclease protection experiments (Fig. 6) FIG. 6. S1 nuclease mapping of the ded genes using polar mutants in hisT and purF. RNA (25 pg) from E. coli strains W3110 (wild type), NU399 (hisT mutant), TX140 (wild type), or "X158 (purF mutant) was used to protect radiolabeled M13 clones from S1 nuclease. The clones used are denoted at the top; CK-3 was constructed from a KpnI-ClaI fragment inserted at the KpnI-AccI sites of M13mp18, 3a was constructed by insertion of a PuuI-NruI fragment into the HindIII-SmI sites of M13mp18, BBF-342 was constructed by insertion of a BgDI-BgZII fragment into the BarnHI site of M13mp18 in the orientation such that the phage DNA is antisense of dedF. The positions of the fully protected fragments and their apparent sizes determined in alkali gel electrophoresis are denoted at the right. At the bottom is a schematic of the regions encoded in these clones and the expected sizes of fully protected DNA. There is a deletion in the map where there are dotted lines. DNA (200 ng) was radiolabeled to 4 X 10' cpm/pg and digested with S1 nuclease for 60 min at 37 "C. The control experiments used RNA from yeast (lanes y ) . The numbers, in bp, at the left are from coelectrophoresed size markers. Radioautography was from 4-36 h. The residual full-sized protected fragment using the 3a clone with the hisT polar mutant may indicate that this mutant is "leaky," or the existence of an additional internal promoter downstream from the mutation. In addition, the unexplained smaller protected fragments of less than full length in these experiments could reflect the heterogeneity of the RNAs due to a short half-life, artifacts due to possible secondary structures, or additional internal promoter/termination sites. ubiX-The amino acid sequence of dedF was 70% similar to the sequence identified in S. typhimurium as ubiX. 2 argT-Immediately following dedF and extending to the end of the 9210 insert was an open reading frame of 146 amino acids which was 89% similar to the argT locus of S. typhimurium, the gene for the lysine/arginine/ornithinebinding protein (32).

DISCUSSION
The description of six genes downstream from the hisT operon, and evidence for the coordinate transcription of three of these genes with the hisT and purF operons, has been accomplished using three independent criteria. First, the DNA sequence data predicted coding regions with codon usage similar to that of moderately expressed E. coli proteins. In addition computer analysis identified likely protein-coding reading frames on 9210. Second, the proteins expressed in a maxicell system were consistent with each gene's orientation, position, and size. Third, S1 nuclease mapping identified additional genes which are coordinately transcribed with either hisT or purF.
The hisT gene has been previously determined to be part of an operon. In addition, previous S1 nuclease analysis had demonstrated that transcription continues past the hisT gene * G. F-L. Ames, unpublished data. (9). Furthermore, in this study, S1 nuclease mapping of transcripts from wild type and a hisT polar mutant has demonstrated that dedA and dedB are additional components of the hisT operon. Although the operon may extend further downstream from dedB, a consensus p-independent terminator sequence (33) is located immediately following the coding sequence for dedB (403-057).
This sequence could form an 11-base pair GC-rich stem followed by 4 thymidine residues and may be the terminator for the hisT operon, however, this has not been experimentally demonstrated. Taken together the hisT operon consists of at least four cistrons of which only one encodes a protein of known function. Immediately downstream of dedB two genes have been identified which are termed dedC and dedD. These genes may have been previously mentioned in earlier studies. Makaroff and Zalkin (11) reported that an upstream transcript terminated near thepurF operon promoter, and an uncharacterized protein (47 kDa) was produced from a 1.5-kb fragment of DNA that extends 5' upstream of purF. The immediate upstream transcript likely corresponds to the dedD gene, and the 47 kDa protein is probably the dedC gene product. In addition, Bogner et al. (34) have reported the cloning of the folC gene. This gene is present on the plasmid pLC28-44, has a restriction map which indicates that it lies in the region identified here as dedC, and the purified folC product (molecular mass approximately 47 kDa), folypolyglutamate synthe-tase-dihydrofolate synthetase, closely corresponds to the molecular mass of the dedC protein (45.5 kDa). It is possible that dedC and folC are equivalent loci.
Immediately downstream from dedD are the genes comprising the purF operon. The first two genes dedE and purF have been described previously (11). Downstream from the purF gene is dedF which has been identified as a component of the purF operon by S1 nuclease mapping and whose transcription is affected by a purF polar mutation. Moreover, in several maxicell clones dedF was transcribed independently of dedE, purF, or known vector promoters, which may indicate alternate expression independent of the operon promoter. The start of transcription for this operon has been determined (11) and is shown in Fig. 3 (position 6233). In addition, the sequence after dedF contains a consensus p-independent termination structure (8991+9013) which may be a terminator of the purF operon. However, this proposed terminator has not been experimentally defined. The purF operon appears to consist of the three genes, dedE, purF and dedF.
Previous genetic studies in this region are limited. Hong and Ames (36) have mapped five lethal temperature-sensitive mutations between hisT and purF in S. typhimurium. Some of these mutations map nearer to purF than hisT and could be localized in dedE. This is supported by the observation of Makaroff and Zalkin (11) that no polar purF mutations could be mapped to the dedE gene in E. coli. It is also possible that some of these mutations map to other ded genes. Further analysis of these previously isolated mutants and mutational analysis of this region will be required to define the function of the ded genes.