Inactivation of the Mycobacterial Rhamnosyltransferase, Which Is Needed for the Formation of the Arabinogalactan-Peptidoglycan Linker, Leads to Irreversible Loss of Viability*

Temperature-sensitive mutant 2-20/32 of Mycobacterium smegmatis mc2155 was isolated and genetically complemented with a Mycobacterium tuberculosis H37Rv DNA fragment that contained a single open reading frame. This open reading frame is designated Rv3265c in the M. tuberculosis H37Rv genome. Rv3265c shows homology to the Escherichia coli gene wbbL, which encodes a dTDP-Rha:α-d-GlcNAc-pyrophosphate polyprenol, α-3-l-rhamnosyltransferase. In E. coli this enzyme is involved in O-antigen synthesis, but in mycobacteria it is required for the rhamnosyl-containing linker unit responsible for the attachment of the cell wall polymer mycolyl-arabinogalactan to the peptidoglycan. The M. tuberculosis wbbL homologue, encoded by Rv3265c, was shown to be capable of restoring an E. coli K12 strain containing an insertionally inactivated wbbL to O-antigen positive. Likewise, the E. coli wbbL gene allowed 2-20/32 to grow at higher non-permissive temperatures. The rhamnosyltransferase activity of M. tuberculosis WbbL was demonstrated in 2-20/32 as was the loss of this transferase activity in 2-20/32 at elevated temperatures. The wbbL of the temperature-sensitive mutant contained a single-base change that converted what was a proline in mc2155 to a serine residue. Exposure of 2-20/32 to higher non-permissive temperatures resulted in bacteria that could not be recovered at the lower permissive temperatures.

The acid-fastness of all mycobacteria is based upon a shared universal cell wall core structure. The mycobacterial cell wall consists of an outer lipid layer and an inner peptidoglycan layer. The outer layer is highly impermeable and is composed of unique 70 -90 carbon-containing lipids, known as mycolic acids. The mycolic acids are esterified to the non-reducing terminal arabinosyl residues of the polysaccharide arabinogalactan (1)(2)(3)(4)(5). The reducing end of arabinogalactan is connected to the peptidoglycan via the disaccharide linker, ␣-L-Rha-(133)␣-D-GlcNAc-(13phosphate). Structural analyses showed that the integrity of the whole two-layer mycolic acid peptidoglycan assembly hinges on the presence of the rhamnosyl moiety as depicted in Fig. 1A. The complete structure of the linker is illustrated in Fig. 1B, and the reaction catalyzed by the enzyme, dTDP-Rha:␣-D-GlcNAc-pyrophosphate polyprenol, ␣-3-Lrhamnosyltransferase (referred to as rhamnosyltransferase in this study) is shown in Fig. 1C. The rhamnosyl residue and much if not all of the arabinogalactan polysaccharide are synthesized on GlcNAc-P-P-decaprenyl carrier lipid (6). The eventual transfer of the arabinogalactan-Rha-GlcNAc-phosphate unit to the O-6 of a muramic acid places the polysaccharide in mass onto the peptidoglycan. Finally, at some still to be defined point, the mycolic acids are attached to arabinofuranosyl residues at the non-reducing end of arabinogalactan.
To further define and characterize the essential steps involved in the synthesis of the mycobacterial cell wall core, the classic microbial approach of isolating conditional lethal mutants was undertaken. Our strategy was to isolate temperature-sensitive (TS) 1 mutants in the genetically amenable and relatively fast growing Mycobacterium smegmatis mc 2 155 (7). A preferred large temperature range that would support growth precluded Mycobacterium tuberculosis from serving as the host for the generation of TS mutants. TS mutants would be genetically complemented with M. tuberculosis genomic DNA in hopes of identifying essential genes encoding cell wall biosynthetic enzymes. Herein, we describe the isolation of a TS cell wall mutant and the independent genetic complementation of that mutant with a M. tuberculosis gene and an E. coli gene. We report biochemical characterization of the TS mutant, the deduced amino acid change due to the mutation, the genetic complementation of an E. coli mutant to confirm the function of a M. tuberculosis gene, and the effect of the mutation on mycobacterial viability after exposure to non-permissive temperatures.

EXPERIMENTAL PROCEDURES
Isolation of TS Mutants-The strategy for the isolation and enrichment of bacterial TS mutants in a culture as outlined by A. Morris Hooke (8) was adapted for use in this study. M. smegmatis mc 2 155 (7) was inoculated into Middlebrook 7H9 with ADC supplement (Difco) (7H9) and grown at 37°C to ϳ10 8 colony-forming units/ml. Nitrosoguanidine (Sigma) was added to a final concentration of 0.1 mg/ml, and cultures were incubated at 37°C without aeration for 40 min. The cells were recovered by centrifugation and washed three times in fresh 7H9 medium to remove the mutagen. Samples of cells were suspended in at least 20 volumes of fresh 7H9 broth distributed into several separate flasks to help minimize the isolation of siblings and incubated with shaking at the permissive temperature of 30°C. When cell density reached ϳ5 ϫ 10 7 colony-forming units/ml cultures were transferred to a shaking water bath at the non-permissive temperature of 42°C for 1 h. To enrich the population of TS mutants in the cultures D-cycloserine (Sigma) was added to a final concentration of 0.4 mg/ml. Cells were incubated at 42°C for another 6 h, recovered by centrifugation, and washed 3 times in fresh medium to remove the antibiotic. Surviving cells were cultured in fresh broth at 30°C until density again reached ϳ5 ϫ 10 7 colony-forming units/ml. The cultures were subjected to another cycle of enrichment at the non-permissive temperature as described, except the D-cycloserine was replaced with ampillicin (Sigma) at a final concentration of 2.3 mg/ml. Ampillicin was used in the second cycle to avoid enriching for D-cycloserine-resistant mutants. The treatment with either D-cycloserine or ampillicin resulted in a 99 -99.9% reduction in colony-forming units. After the second enrichment the surviving cells were washed free of the antibiotic, diluted, plated on Luria Bertani agar supplement with 1% Tween 80 (Sigma), and incubated at 30°C. M. smegmatis mc 2 155 grown in the presence of Tween 80 produce mucoid colonies on Luria-Bertani agar, which make them readily transferable to other plates. Colonies that arose after incubation were replica-plated onto Luria Bertani agar supplement with 1% Tween 80 and incubated at 30°C and 42°C to identify TS mutants.
Genetic Complementation of TS Mutant Strain 2-20/32-M. tuberculosis H37Rv genomic DNA was isolated and partially digested with Sau3A (9). Gel-purified DNA fragments of about 30 kilobases were ligated into BamHI-digested pCB7 (Table I), packaged into particles using Gigapack III XL packing extract (Stratagene, La Jolla, CA), and propagated in E. coli (XL1-Blue MR, Stratagene) on LB containing 200 g/ml hygromycin B (Sigma). Although plasmids recovered contained significantly smaller M. tuberculosis DNA fragments than the desired 30-kilobase inserts, they were still used to attempt to complement some of the TS mutants. These smaller plasmids were probably the result of the instability of the larger constructs in the transduced E. coli host. The plasmid library was introduced into the TS mutants by electroporation (8). Transformants were selected on Luria Bertani agar supplement with 1% Tween 80 with 50 g/ml hygromycin B (Sigma) at 42°C. Fortuitously, one of the plasmids from this library, pCB120, allowed TS mutant 2-20/32 to grow at the non-permissive temperature of 42°C. Plasmid pCB120 was sequenced and shown to contain a 1764-bp M. tuberculosis H37Rv DNA insert. A BLAST search showed that this DNA contained the entire open reading frame (ORF) of Rv3265c (10) as well as incomplete portions of the two flanking genes. Thus, the complementing M. tuberculosis DNA fragment contained only one complete ORF. The predicted protein product of this intact ORF showed homology to what was originally described as rfb ORF 11 in E. coli K-12 (11) but now has been designated as wbbL (12).
Construction of Plasmids-Plasmids with their relevant characteristic(s) are listed in Table I. AmpliTaq 2 DNA polymerase (Invitrogen) or Vent polymerase (New England Biolabs, Beverly, MA) were used according to the manufacturer's instructions to amplify fragments for cloning. All constructions were confirmed by sequencing.
M. tuberculosis wbbL was amplified using primers CCGAGCTCCT-GAAGTGACTGACGTCCTGCCGG and GAGGTACCTTATTCAGTGC-CGCCCTTCTACC from pCB120. The former primer included a SacI site and a stop codon designed to be in-frame with the ␤-galactosidase followed by the wbbL GTG start. The latter primer included a KpnI site. The PCR products were digested with appropriate enzymes and ligated into SacI/KpnI-digested pBluescript (Stratagene) to generate pCB220. M. tuberculosis wbbL was also cloned into the E. coli overexpression vector pET 23b (Novagen, Milwaukee WI) to generate pVV1. Primers GGAATTCCATATGACTGACGTCCTGCCGGT containing an NdeI site and CAGCTCGAGTCAGTGCCGCCCTTCTACCAGCT containing a stop codon and XhoI site were used. Amplified products from pCB120 were digested with NdeI and XhoI and ligated into NdeI/XhoI-digested pET 23b.
E. coli wbbL was amplified from E. coli strain WG1 (11) genomic DNA using PCR primers CGGAATTCAGAAACATGGGGATTGCCG and GAGGTACCTTATTACGGGTGAAAAACTGAT. EcoRI and KpnI sites in primers facilitated cloning into pBluescript. The resulting plasmid pKM1 produced a fusion protein formed between ␤-galactosidase and ORF 10 (10). The ATG start site in pKM1 for the E. coli wbbL followed two base pairs after the stop codon of ORF 10. Plasmid pCB227 was generated when E. coli wbbL was cloned into the mycobacterial expression vector pMX1 (20). Primers AAACTGCAGCCTGAAATGG-TATATATAATAATCG, which included a PstI site, and GAGGTACCT-TATTACGGGTGAAAAACTGAT (same primer used for pKM1) were used. PCR products amplified from E. coli WG1 were blunt end-ligated into SmaI-digested pUC18 (Invitrogen). The insert was then cut out with PstI and KpnI and ligated into PstI/KpnI-cleaved pMX1. (13) as well as visualization by silver stain were done as described (14). At the beginning of these studies E. coli K12-producing O-antigen was not available. Therefore, antibodies against the O-antigen of E. coli O-16, which were known to cross-react (11) against O-antigen of E. coli K12 were prepared; the two strains of E. coli are known to produce O-antigens of similar structure ( Fig. 2A) (11,15). Rabbit antibodies against O-antigen were raised against whole bacteria suspended in 0.5% formalin, phosphate-buffered saline. The resulting antiserum was cross-absorbed with E. coli K12 EMG2, which is devoid of any O-antigen. Western blots were done using the rabbit O-antigen antiserum and goat anti-rabbit secondary antibody conjugated to peroxidase (Roche Applied Science) and detected using a chemiluminescence kit (Roche Applied Science).

SDS-PAGE and Western Blots of LPS Produced by Various Bacteria-LPS solubilization and SDS-PAGE
Preparation of dTDP-[ 14  Then 700 l of absolute ethanol was added, and the precipitated protein was removed by centrifugation at 14,000 ϫ g for 5 min. The bulk of the ethanol was removed by evaporation, and the dTDP-[ 14 C]Rha was purified by high performance liquid chromatography as described (17).
Assay for Rhamnosyltransferase-The assay is essentially that of Reeves and co-workers (11). The acceptor for the rhamnosyl residue, GlcNAc-P-P-decaprenol, is formed in situ from exogenously added UDP-GlcNAc and from undecaprenyl phosphate present in the membrane preparations by GlcNAc-1-phosphate transferase also present in the membrane preparations. Thus, bacteria, as indicated in Table II were disrupted by sonication, and cell debris were removed by centrifugation 8,000 ϫ g. The membrane pellets were then prepared by additional centrifugation for 1 h at 100,000 ϫ g and resuspended in 50 mM, pH 8, MOPS containing 5 mM mercaptoethanol and 10 mM MgCl 2 at protein concentrations in all cases of ϳ13 mg/ml. To 150 l of these membrane preparations were added 0.1 Ci of dTDP-[ 14 C]Rha (221 Ci/mol), 6 nmol of UDP-GlcNAc, 18 nmol of ATP, and additional MOPS buffer to total 320 l. After incubation for 1 h at 37°C, 200 l of water, 400 l of methanol, and 800 l of chloroform were added, and the organic layer removed and counted.
Combined Assay for Rhamnosyl and GlcNAc-1-phosphate Transferases-M. smegmatis mc 2 155, M. smegmatis 2-20/32, and M. smegmatis 2-20/32 complemented with pCB120 were grown at 30°C to approximately late log. For the experiment reported in Fig. 3A they were then harvested; for the experiment in Fig. 3B they were additionally treated for 6 h at 42°C and then harvested. Cell-free extract preparation, membrane isolation, and incubation with dTDP-Rha (non-radioactive) and UDP-[ 14 C]GlcNAc, extraction with organic solvent, TLC chromatography, and visualization by autoradiography were as described (6).

RESULTS
Isolation and Complementation of TS Mutant 2-20/32-Forty-seven M. smegmatis mc 2 155 TS mutants were isolated in 15 independent mutagenic experiments as described under "Experimental Procedures." The mutagenesis and enrichment procedure yielded an average frequency of 1 TS mutant per 420 colonies screened. One TS mutant 2-20/32 was genetically complemented by pCB120, i.e. the plasmid allowed the mutant to grow at the non-permissive temperature of 42°C. Plasmid pCB120 contained a 1764-bp M. tuberculosis H37Rv genomic DNA fragment. Sequence analysis revealed that pCB120 contained only one intact ORF. This ORF was translated and subjected to a BLAST (18) search that revealed it corresponded to M. tuberculosis Rv3265c and showed significant homology to the protein product of the wbbL gene of E. coli K12 (accession L19537; see also Ref. 12). This fragment DNA was known as ORF 264 in the original report (19). It was, therefore, postulated that the ORF in pCB120 was, in fact, the M. tuberculosis version of wbbL encoding the rhamnosyltransferase responsible for the synthesis of the ␣-L-Rha-(133)-␣-D-GlcNAc linker unit (Fig. 1C).
Complementation of E. coli K12 EMG2 with M. tuberculosis DNA Coding for Putative Rhamnosyltransferase-The work of Reeves and co-workers (11,19) allowed us to readily conduct complementation experiments that could determine whether the ORF in pCB120 was a rhamnosyltransferase. The ORF in pCB120 was amplified by PCR and inserted in pBluescript to form pCB220 (Table I). In an analogous fashion the E. coli wbbL was amplified by PCR from E. coli WG1 (19) and cloned in pBluescript to form pKM1 (Table I). Plasmids pCB220 and pKM1 were introduced into E. coli EMG2. EMG2 contains an insertionally inactivated wbbL (11,19) and, thus, is unable to produce O-antigen. LPS profiles by non-transformed EMG2, EMG2 (pCB220), and EMG2 (pKM1) were examined by SDS-PAGE. Visualization by silver staining (Fig. 2B) showed as expected that LPS complete with O-antigen was produced by EMG2 containing the plasmid encoding the E. coli rhamnosyltransferase (19), whereas EMG2 without plasmid or pBluescript was devoid of O-antigen. Remarkably, a plasmid encoding the putative rhamnosyltransferase gene from M. tuberculosis also resulted in O-antigen synthesis. Relative to the E. coli wbbL, the number of O-antigen units added to the LPS appeared to be diminished with the M. tuberculosis wbbL. Blot analysis (Fig. 2C) showed that the O-antigen produced from both plasmids reacted with antibody raised against E. coli O-16 O-antigen, but no reaction occurred without plasmid or pBluescript. Known cross-reactivity of antiserum raised against E. coli O-16 O-antigen with E. coli K12 O-antigen (11) allowed this antiserum to be used.

Transformation of TS Mutant M. smegmatis 2-20/32 with a Plasmid Containing the E. coli wbbL Gene Encoding for
Rhamnosyltransferase Allows for Growth at 37°C-Further evidence that the ORF on pCB120 encoded for cell wall linker rhamnosyltransferase was obtained by transforming M. smegmatis 2-20/32 with pCB227 a plasmid containing E. coli wbbL (Table I) under the control of the heat shock promoter found in pMX1 (20). It was found that although this plasmid did not The M. tuberculosis wbbL Gene Product Exhibits Rhamnosyltransferase Activity in E. coli-Direct evidence that the M. tuberculosis wbbL encodes for rhamnosyltransferase was obtained using the rhamnosyltransferase assay of Reeves and co-workers (11). Membranes were prepared from E. coli strains BW24599, a K12 strain repaired for rhamnosyltransferase (17), and a non-repaired K12 strain, HMS 174 containing the control vector with no insert (pET 23b) and HMS 174 with pVV1, a pET 23b vector with a M. tuberculosis wbbL insert. The membranes containing endogenous undecaprenyl phosphate were incubated with UDP-GlcNAc to form in situ the acceptor for rhamnosyltransferase, GlcNAc-P-P-decaprenol. dTDP-[ 14 C]Rha donor was added, and the production of product, [ 14 C]Rha-GlcNAc-P-P-undecaprenol was monitored by the production of organic soluble radioactivity. As shown in Table  II, the positive control, E. coli strain BW24599 and the E. coli HMS 174 strain containing pVV1 did show rhamnosyltransferase activity. E. coli HMS 174 without plasmid showed no activity.

TS Mutant M. smegmatis 2-20/32 Exhibits Diminished Rhamnosyltransferase Activity-Membranes prepared from
M. smegmatis strains mc 2 155, 2-20/32, and 2-20/32 containing pCB120 were tested for rhamnosyltransferase activity using the procedure developed by Mikusova et al. (6). This procedure, using radioactive UDP-GlcNAc instead of radioactive dTDP-Rha, allows direct comparison of the amounts of [ 14 C]GlcNAc-P-P-decaprenol and of Rha-[ 14 C]GlcNAc-P-P-decaprenol produced by the M. smegmatis membranes. Thus, the membranes were incubated with dTDP-Rha and UDP-[ 14 C]GlcNAc. Endogenous decaprenyl phosphate present in the membrane preparations was utilized by the M. smegmatis GlcNAc-1-P transferase to form [ 14 C]GlcNAc-P-P-decaprenyl, which was in turn utilized by the rhamnosyltransferase along with dTDP-Rha to form Rha-[ 14 C]GlcNAc-P-P-decaprenyl. The results, as monitored by TLC, are presented in Fig. 3A. Wild-type M. smegmatis mc 2 155 showed a strong Rha-[ 14 C]GlcNAc-P-P-decaprenyl band and a weak [ 14 C]GlcNAc-P-P-decaprenyl band (Fig. 3A, lane 1, in accordance with previous results (21). In contrast, strain 2-20/32 showed Rha-[ 14 C]GlcNAc-P-P-decaprenyl and [ 14 C]GlcNAc-P-P-decaprenyl in about equal amounts (Fig. 3A,  lane 3), suggesting that even with cell growth and enzyme incubation at 30°C, the rhamnosyltransferase activity was lower than in wild type. Complementation of M. smegmatis 2-20/32 with pCB120 restored the rhamnosyltransferase activity to wild-type levels (Fig. 3A, lane 2). When enzyme incubations were shifted from 30 to 42°C a similar TLC profile with no decrease of the rhamnosylated lipid was produced. Therefore, the effect of varying the growth temperatures of the cells was ascertained. Cells of all three strains were again grown at 30°C and then incubated at the non-permissive temperature of 42°C for another 6 h. It was expected that in strain 2-20/32 no new active rhamnosyltransferase would be formed at 42°C and the previously folded rhamnosyltransferases would be depleted by normal turnover events. Membranes were prepared and assayed for rhamnosyltransferase activity at 30°C (Fig. 3B). The results showed that under these conditions strain 2-20/32 produced very little Rha-[ 14 C]GlcNAc-P-P-decaprenyl in comparison to its precursor, [ 14 C]GlcNAc-P-P-decaprenyl (Fig. 3B,  lane 3), whereas both the wild type strain and the complemented TS mutant strain continued to produce major amounts of the rhamnosyl containing product (Fig. 3B, lanes 1 and 2). The fact that the rhamnosyltransferase activity of 2-20/32 was not dependent on the temperature of enzyme incubation but rather on the temperature of the growth of the bacteria suggests that the mutation affects the conformation of the protein as it is being synthesized (22).
The wbbL Gene of M. smegmatis TS Mutant 2-20/32 Contains a T Rather than C at Position 364 -The wbbL genes from M. smegmatis mc 2 155 wild-type (GenBank TM accession number AF187550) and 2-20/32 (GenBank TM accession number AF187551) were sequenced, and a single base changed at position 364 from a C to T was found. This translates into a change from proline to serine at amino acid 122 in the WbbL polypeptide. The wild-type wbbL sequence agreed with the sequence for the incomplete M. smegmatis genome on Tigr (www.tigr.org). The nucleotide sequence of the M. tuberculosis wbbL from pCB120 was in complete agreement with the published Rv3265c sequence (10)  transformed with pCB120 were grown at 30°C. When the cultures reached A 600 of 0.2, one 2-20/32 culture and the 2-20/32 (pCB120) culture were shifted to 42°C, and the A 600 was monitored. After a brief growth spurt, 2-20/32 with no plasmid stopped growing (Fig. 4A). In contrast the 2-20/32 with no plasmid but maintained at 30°C as well as 2-20/32 (pCB120) shifted to 42°C continued to grow. After 24 h the untransformed 2-20/32 cultures incubated at 30°C and 42°C were subcultured to fresh media and incubated at the permissive temperature of 30°C. As can be seen (Fig. 4B) after the 17-h treatment at the non-permissive temperature, 2-20/32 was no longer viable. Additional experiments (data not presented) showed that an incubation of ϳ10 h was required before the TS mutant could no longer be recovered; this result is consistent with the mutation being the temperature-sensitive folding type (22). This observation also explains why 2-20/32 could be recovered after a 7-h incubation at 42°C during the TS enrichment procedure. DISCUSSION We have previously determined the structure of the linker region of arabinogalactan (Fig. 1A) (5) and predicted because of its key structural location (1,23) that the synthesis of this linker region would be essential for mycobacterial viability (16). In addition we have shown that RmlD, one of the enzymes required for the formation of linker, is essential in M. smegma-  a Substrate dTDP-͓ 14 C͔Rha was converted from aqueous to organic counts by the presumed attachment of the ͓ 14 C͔rhamnosyl residue to in situ GlcNAc-P-P-decaprenol. In total, 250,000 cpm of dTDP-͓ 14 C͔Rha were used, and hence, in the case of E. coli HMS 174 (pVV1), 2% of the donor was converted to product. In similar experiments (data not presented) where decaprenyl phosphate and/or detergent was added, the results were similar; the conversion of dTDP-͓ 14 C͔Rha into product ͓with E. coli HMS 174 (pVV1)͔ remained between 1.5 and 2%. With detergent there was some increase in background activity in the nonproducing strains and the zero time points presumably due to the solubilization by the detergent of some dTDP-͓ 14 C͔Rha into the organic phase.  (6), and the glycolipids produced by this strain were used as a standard for these experiments. The identity of the glycolipid indicated by an asterisk is unknown; it might be the C-35 polyprenyl version of Rha-GlcNAc-P-Ppolyprenyl. The identity of the radioactivity at the origin is also unknown. Li-PP, polyprenyldiphosphoryl.
tis (24). In the present study we took a classical microbial drug target identification approach that could define any type of essential gene and serendipitously isolated a TS mutation in wbbL, another gene involved in linker biosynthesis.
These studies were aided enormously by the pioneering work of Peter Reeves and co-workers (11,19) on E. coli K12 Oantigen biosynthesis. Coincidentally and fortuitously the rhamnosyltransferase in E. coli K12 O-antigen biosynthesis utilizes the same substrates, dTDP-Rha and GlcNAc-diphosphoryl polyprenol, to form the exact same product, ␣-L-Rha-(133)-␣-D-GlcNAc-diphosphoryl polyprenol, as mycobacteria. This product is then utilized in very different cell wall structures by E. coli and by mycobacteria. In E. coli it becomes part of the moiety that forms the O-antigen that is attached to LPS core ( Fig. 2A), but in mycobacteria it is metabolized to generate the linker region that attaches the arabinogalactan to the peptidoglycan (Fig. 1A). Moreover, O-antigen is nonessential to the host, whereas the linker is essential.
Given the conservation of cell wall core structure in mycobacteria, it was presumed that M. tuberculosis cell wall-associated homologues would be able to complement TS mutations in M. smegmatis. Expecting E. coli and M. tuberculosis wbbL genes to function interchangeably to help confirm enzymatic roles was more uncertain. Fortunately, the E. coli wbbL did permit the growth of the TS mutant at the non-permissive temperature of 37°C, but not at 42°C, where M. tuberculosis wbbL still functioned. Similarly, the M. tuberculosis wbbL allowed O-antigen units to be added to the LPS core but not to same degree as the E. coli wbbL (Fig. 2). Although complementation was observed in both experiments, the foreign wbbL was not expressed as well and/or the WbbL protein did not interact as well as the endogenous WbbL with the native host environment to allow the same level of production. Nevertheless, these complementation results and biochemical evidence of linker-associated metabolites in E. coli (Table I) and M. smegmatis (Fig. 3) confirmed the putative WbbL to be the rhamnosyltransferase involved in synthesis of the linker region in mycobacteria.
In addition to wbbL, the M. tuberculosis genome has annotated sequence Rv1525 as wbbL2 (10). This gene shows no homology to the wbbL genes discussed herein, and there is no reason to suspect that it encodes a protein that catalyzes the reaction shown in Fig. 1C. Nor does M. leprae, M. avium paratuberculosis, or M. smegmatis carry a known homologue to wbbL2. At this time the role of wbbL2 is unknown. Other rhamnosyltransferase genes have been identified in mycobacteria that function to generate other cell wall-associated structures. The gene rtfA encodes a rhamnosyltransferase required for the synthesis of serovar 2-specific oligosaccharides in M. avium (25). Other putative rhamnosyltransferases in M. tuberculosis and M. leprae have been hypothesized to be involved in the synthesis of rhamnose-containing oligosaccharides on phenolic glycolipids (25,26). It is important to note that these various rhamnosyltransferases do not have the same substrates nor do they generate the same products as WbbL.
Because wbbL has been shown by a genomic approach (27) and RmlD by a biochemical approach (24) to be essential, it is not surprising that a TS mutation of wbbL could be obtained. However, it was difficult to predict whether the defect would be static or lethal. The behavior of the TS mutant 2-20/32 after treatment at the non-permissive temperature suggests the latter consequence. This phenomenon would predict that inhibitors of the rhamnosyltransferase (and inhibitors of dTDPrhamnose formation) will also be lethal rather than static. Further studies to purify M. tuberculosis WbbL enzyme and develop a direct assay to identify inhibitors of its activity are ongoing. Obstacles to be overcome include robust expression of the protein, production of required amounts of acceptor, and development of an assay amenable to screening.