Appendix. Cloning and sequence of the gene encoding enzyme E-1 from the methionine salvage pathway of Klebsiella oxytoca.

The methionine salvage pathway converts the methylthioribose moiety of 5'-(methylthio)-adenosine to methionine via a series of biochemical steps. One enzyme active in this pathway, a bifunctional enolase-phosphatase called E-1 that promotes oxidative cleavage of the synthetic substrate 2,3-diketo-1-phosphohexane to 2-keto-pentanoate, has been purified from Klebsiella pneumoniae and is characterized in the preceding paper (Myers, R., Wray, J., Fish, S., and Abeles, R. H. (1993) J. Biol. Chem. 268, 24785-24791). We synthesized degenerate oligonucleotides corresponding to portions of the amino terminus of E-1. These oligonucleotides were used as polymerase chain reaction primers on whole genomic DNA from Klebsiella oxytoca. This resulted in an 82-base pair DNA fragment that was used as a hybridization probe to obtain a clone of the E-1 gene from a K. oxytoca gene library. The DNA sequence of the E-1 coding region was determined, and the amino acid sequence of E-1 was deduced. E-1 appears to represent a novel class of enzymes since no homology to known enzymes was found. Cloning the gene from K. oxytoca on a multicopy plasmid leads to overproduction of E-1 enzyme that has properties indistinguishable from those of the enzyme from K. pneumoniae.

The so-called "methionine salvage pathway" converts the methylthioribose moiety of 5'-(methylthio)-adenosine to methionine in various microorganisms and higher organisms (2, 3). The enzymes and intermediates of this pathway have been most thoroughly studied in Klebsiella pneumoniae (4). 5'4Me-thy1thio)-adenosine is ordinarily derived from S-adenosylmethionine during polyamine biosynthesis. The pathway is of interest both for its novel enzymology (4) and for its potential use in the commercial biosynthesis of amino acids (5). In either context it is useful to obtain DNA clones of each gene involved in the pathway. The K. pneumoniae 1-phospho-2,3diketohexane-2-hydrate enolase-l-phospho-2-hydroxy-3-keto-1-hexene phosphatase (called "E-1"), which catalyzes the fifth step in the pathway, has been purified and characterized (1). fur elektrochemische Industrie, GmbH. The costs of publication of this * This work was supported in part by a contract from the Consortium article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U00148. $  1 To whom correspondence should be addressed: OmniGene, Inc., On the basis of a partial amino acid sequence of E-1 (l), we have synthesized DNA probes that allowed cloning of the gene that encodes this enzyme. We have sequenced the gene and deduced the entire amino acid sequence of the enolase-phosphatase enzyme.

MATERIALS AND METHODS
Bacteria were grown in LB media (6). Solid medium contained 1.5% agar. Strains containing plasmids were selected by addition of 20 pdml chloramphenicol to the growth medium. Although the amino-terminal protein sequence of E-1 was determined using enzyme isolated from K. pneumoniae strain CG253 (l), the DNAfor clone bank construction was from the closely related strain, Klebsiella ozytoca UNF932, an hsdR, r e d derivative of the widely used strain M5a1, formerly classified as K. pneumoniae (7). The sole taxonomic distinction between the two species is that K. ozytoca produces indole from tryptophan, whereas K. pneumoniae does not. We have determined that strain CG253 produces indole from tryptophan,' so we suggest that K. pneumoniae CG253 may be properly classified as K. oxytoca.
All DNA manipulation methods were standard procedures (6). Oligonucleotides were synthesized on an Applied Biosystems model 380A or 380B, according to the manufacturer's protocols.
Double-stranded plasmid used for sequencing was purified with Qiagen Maxi kits. DNA sequencing reactions were performed using a Sequenase version 2.0 kit (U. S. Biochemical Corp.). The double-stranded DNA templates were denatured using 1 N NaOH, and primers were annealed according to the method described (8). The labeling reactions and subsequent steps were carried out as described in the Sequenase version 2.0 protocol (5th edition) obtained from the manufacturer. DNA fragments were purified by preparative agarose gel electrophoresis using either Qiaex (Qiagen) or Gene Clean (Bio 101, La Jolla, CA), as instructed by the manufacturers. Polymerase chain reactions (PCR) were performed with the Perkin-Elmer Cetus PCR kit using a Perkin-Elmer Cetus DNA thermal cycler. For the first 4 cycles melting was at 94 "C for 30 s and annealing at 37 "C for 30 s, followed by linear temperature rise to 72 "C over 2 min, with polymerization continued at 72 "C for 1 min. For the subsequent 25 cycles the annealing temperature was changed to 50 "C for 30 s, rising to 72 "C over 30 s, with continued polymerization at 72 "C for 30 s, and a final extension for 10 min at 72 "C. Colony screening by PCR was done with the toothpick method (9).
Hybridization probe of high specific radioactivity was produced in a [CY-~~PI~ATP stock solution (3000 Ci/mmol; DuPont NEN NEG-Ol2H; 10-pl PCR reaction containing the following reagents: 5 pl of final dATP concentration, 1.7 p~) ; 1.5 pl of dTTP, dCTP, and dGTP mix, containing each at 125 PM; 1.0 p1 of Perkin-Elmer Cetus PCR buffer I (equivalent to buffer I1 plus 15 nm MgCl2); 1.0 pl of mixed primers, each at 10 m; 0.5 p1 of Taq polymerase mix, diluted to 1/10 concentration of stock (0.25 units); and 1.0 pl of template solution, containing about 0.1 ng of template. The reaction was covered with 10 pl of mineral oil and run with the program described above (29 cycles) (10). About threequarters of the radioactivity was incorporated into PCR product. When products from similar reactions (performed without radioisotope labeling) are visualized in ethidium bromide-stained agarose gels, a distinct band of full-length product is visible with a pronounced shadow of R. R. Yocum, unpublished data. smaller products. Both full-length and partial products should hybridize to the target sequence and are useful as reporters.
Initially, all cloning work involving Klebsiella DNA was performed under BL2 conditions. During this work, the NIH Recombinant DNA Guidelines were amended, and work with clones from strain M5al and its derivatives was classified as requiring only BL1 conditions.

RESULTS
Primers-Based on the amino acid sequence of the first 26 residues of the purified E-1 from strain CG253 (11, two mixed oligonucleotide PCR primers were designed to amplify an 82base pair DNA region that corresponded to most of the known protein sequence (see Fig. 1). Each primer consisted of an 11base pair sequence corresponding to the protein sequence. At mixed base sites all relevant bases were incorporated in equimolar amounts. In addition, each primer featured an extension of six or seven bases, not based on protein sequence, to provide flanking restriction endonuclease cleavage sites to facilitate cloning. Similar primers that incorporated inosine at the ambiguous base positions did not amplify the target.
Clone Bank-Genomic DNA from strain UNF932 was cleaved partially with Sau3AI, and fragments of 5-10 kilobases were isolated by preparative agarose gel electrophoresis. The purified fragments were ligated into a phosphatase-treated, BamHI-digested pACYC184 plasmid vector (11). Escherichia coli strain "294 was transformed with the resulting DNA, and clones resistant to chloramphenicol (20 pg/ml) were selected. About 2,000 transformant colonies were pooled and stored in 50% glycerol at -80 "C.
Aliquots of the bank pool were diluted and plated at a density of about 1,000 coloniedplate. Replicas were made onto nitrocellulose filters, which were then used for 32P-labeled DNA probe hybridization (6).
Hybridization Probe-Using the oligonucleotide primers described above (Fig. l), PCR amplification was performed with l pg of UNF932 genomic DNA as target. The PCR products were separated electrophoretically on a 2% agarose gel stained with ethidium bromide. No amplification products were observed after 29 cycles of amplification, so the 29-cycle program was repeated. An 82-base pair fragment was then produced using UNF932 DNA as the target but not from E. coli DNA treated under the same conditions. The 82-base pair fragment was excised and purified with Gene Clean. A sample of the 82-base pair fragment was itself used as a target for PCR amplification. The reamplified fragment was purified by preparative acrylamide gel electrophoresis, cloned into the unique SmaI site of M13 mp18, and sequenced. The amino acid sequence translated from this DNA sequence was entirely consistent with the known portion of the enolase-phosphatase amino acid sequence (see Fig. 1).
Some of the purified 82-base pair fragment was also reamplified using [ c~-~~P ]~A T P to produce a radioactive probe specific for the 82-base pair sequence. This probe was hybridized to filter replicas ofE. coli "294 colonies containing the UNF932 clone bank (see above). Eight positively hybridizing colonies were picked from the master plate and named DIOX 1 through DIOX 8 (at that time, the encoded enzyme was thought to be a dioxygenase). The first four of these and one non-hybridizing colony (named X) were assayed for enolase-phosphatase activity as described in the accompanying paper (1). One isolate, DIOX 2, gave a level of enzyme activity more than 100-fold higher than that of the others and the negative control, and 20-fold higher than that of a Klebsiella strain (Table I). All eight isolates were streaked for single colonies, and three to eight colonies from each were screened by PCR using the toothpick method (8) for templates able to direct synthesis of the 82-base pair fragment. Only one colony from isolate DIOX 2 was positive. The plasmid isolated from this colony was named pDiox-2. The other isolates were further screened by hybridization of radioactive probe to filter replicas of single colonies. The colonies that gave the strongest positive signals were screened by PCR, but none was positive by that test.
Strain CG253 was transformed with pDiox-2 by electroporation using a Bio-Rad Electroporator according to the manufacturer's instructions. Colonies resistant to chloramphenicol (20 pg/ml) were selected. An extract of the transformed CG253 contained about 6 times more enolase-phosphatase activity per mg of protein than the untransformed strain (1). On the basis of these results, we concluded that pDiox-2 contains a gene that encodes the enolase-phosphatase.
Mapping by restriction endonuclease digestion showed that the Klebsiella DNA insert of pDiox-2 was about 6 kilobases (Fig. 2). pDiox-2 DNA was cleaved with various restriction enzyme combinations and analyzed by Southern blotting (6) using the 32P-labeled 82-base pair fragment described above, whose sequence represents a portion of the amino-terminal region of the protein. The probe hybridized to all DNA fragments that contained the end of the cloned insert nearest to the 5' end of the tetracycline resistance gene of pACYC184 (Fig. 2). The fragment generated by BglI cleavage was particularly informative, placing the 5' end of the enolase-phosphatase gene within about 800 base pairs of the end of the cloned insert.
Starting from the middle of the 82-base pair sequence, the entire enolase-phosphatase gene was sequenced by the "primer walking" strategy. The DNA sequence of the entire coding region was determined for both strands. The resulting sequence

Enolase-phosphatase activity
Enolase-phosphatase activity of transformants containing clones that hybridized to the E-1 specific probe. Transformant "x" is a non-hybridizing isolate taken as a negative control. UNF932 is the K. oxytoca strain from which the clone bank was constructed, taken as a positive control. The enzyme assay monitored release of formate from the artificial substrate, 2,3-diketo-l-phosphohexane, as described in the accompanying paper (1).  lines indicate restriction fragments that hybridized to the 82-base pair Just upstream from the coding region of enolase-phosphatase are sequence elements characteristic of typical Gramnegative bacterial promoters, namely a ribosome binding site (12), a "-10 region," and a "-35 region" (13) (see Fig. 3). Just downstream from the enolase-phosphatase coding region and overlapping the enolase-phosphatase coding region by 1 base pair is a second open reading frame that extends past the end of the DNA sequence that has been determined so far. This arrangement of genes is typical of translationally coupled genes in prokaryotic operons. acid sequence. Potential "-35" and "-10" promoter elements are un-

DISCUSSION
Despite the different source strains for the DNAclone and for the protein that was sequenced, the two determined sequences corresponded exactly within the region examined. Note that the gene sequence from the pDiox-2 plasmid differs from that of the cloned 82-base pair PCR product at 2 base pairs, both in regions corresponding to the PCR primers. It is likely that this resulted from priming by slightly mismatched primers contained in the mixture of primers. The properties of enolasephosphatase enzyme purified from a plasmid-containing overproducer were similar to those of the enzyme purified from untransformed CG253 (1). Thus, the enzymes from the two related strains are very similar, if not identical.
The E-1 nucleotide sequence was compared with the entire GenBank (release 69) and EMBL (release 28) data bases, and the deduced amino acid sequence was compared with the entire NBRF-PIR (release 30) and Swiss-Prot (release 19) data bases.
No significant homologies were found. Thus, E-1 may represent a novel class of enzymes. We have named the gene that encodes E-1 m a d , for methylthiogdenosine Salvage.
The cloning of the gene encoding enolase-phosphatase on a multicopy plasmid has allowed overproduction of the enzyme, which has aided in the biochemical characterization of this novel enzyme (1). The enzyme encoded by the open reading frame immediately downstream from the enolase-phosphatase gene might be related to the methionine salvage pathway. It is not yet known whether this enzyme is also overproduced from the multicopy plasmid pDiox-2. We propose the name masB for the gene that encodes the second open reading frame.