INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The mycobacterial cell wall core consists of two layers. The highly impermeable outer layer is composed of mycolic acids, 70 to 90 carbon-containing lipids. The inner layer consists of peptidoglycan. These two layers are covalently tethered via the connecting polysaccharide arabinogalactan (2, 4, 12-14). Arabinogalactan itself (Fig. 1) contains three regions: the disaccharide linker, -L-rhamnosyl-(13)--D-GlcNAc-(1phosphate), which is attached to the peptidoglycan; a galactofuran [6)--D-Galf-(15)--D-Galf-(1]_~15 (where Galf is galactofuranose), which is attached to the linker (4); and finally a complex mycolic acid-bearing arabinan, which is attached to the galactofuran (4).

View larger version (32K): [in this window] [in a new window]   FIG. 1.   Formation of UDP-Galf (A) and galactofuran (B). The role of galactofuranosyl residues of linking peptidoglycan to mycolic acids via arabinan is also shown. From the galactofuran structure, it is estimated that four galactofuranosyl transferase activities (Gal Tran A to D) are needed to form the galactofuran (with the remaining residues being assembled by the activities Gal Tran C and D); it is possible that some of these activities are combined into one polypeptide. Rv3808c presumably encodes one or more of these galactofuranosyl transferase activities (15). Galp, galactopyranose; Rhap, rhamnosylpyranose.

Galactofuranosyl residues are formed in nature by the enzyme UDP-galactopyranose mutase (16, 23), which converts UDP-galactopyranose to UDP-Galf. Although this activity was shown some time ago in penicillin fungus (22), only recently has the enzyme been isolated and its activity directly demonstrated to occur in Escherichia coli (16), Klebsiella sp. (9), and mycobacteria (23). Methods to assay the activity of the enzyme have been developed (10), and crystallographic structural study of the enzyme has commenced (11).

The location of galactofuran between the peptidoglycan and the mycolic acids strongly suggests that galactofuran is essential for mycobacterial growth (Fig. 1). Direct evidence of such a role exists for the similarly located arabinan (Fig. 1), because ethambutol, an effective antituberculosis drug, inhibits its formation (21). However, there are not yet any drugs known to directly inhibit the formation of galactofuran and other direct evidence supporting an essential role for galactofuran is lacking.

The gene encoding UDP-galactopyranose mutase in Mycobacterium tuberculosis has been identified as Rv3809c (23). Directly downstream from Rv3809c and overlapping it by a single nucleotide is Rv3808c, which was recently identified as a galactofuranosyl transferase (15), although the specific substrate and product were not identified. Downstream from Rv3808c are three more open reading frames (ORFs), separated from their upstream neighbors by 7, 29, and 89 bp. Thus, glf (Rv3809c) is very likely the first gene in an operon containing at least two and possibly up to five ORFs. The functions of Rv3807c, Rv3806c, and Rv3805c are unknown. In designing and analyzing knockout mutations of glf, this genetic organization must be borne in mind. Herein we present experiments that demonstrate that glf and Rv3808c are essential in Mycobacterium smegmatis. Our basic strategy was to show that M. smegmatis chromosomal glf could be knocked out only in the presence of appropriate rescue plasmids and then that loss of either the glf or the Rv3808c rescue plasmid correlated with the loss of the ability of the bacterium to grow.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. E. coli Top 10 electrocompetent cells (Invitrogen, Carlsbad, Calif.) were used for propagating all plasmids except for pCG76 where chemically competent DH5 (Life Technologies, Inc., Grand Island, N.Y.) cells were used. The bacteria were grown in Luria-Bertani (LB) broth and LB agar with appropriate antibiotics and incubated at 37°C routinely. A fast-growing mycobacterium (referred to in this paper as "mycobacterial lab strain") was used to isolate the DNA in the glf region, the sequence of which was shown to be nearly identical to that of M. smegmatis mc²155 (see below for further details). M. smegmatis mc²155 was used for allelic-exchange experiments and was grown in LB broth with 0.05% Tween 80 or on LB agar plates. Appropriate antibiotics were included, and incubations were at 30, 40, and 42°C, depending on the experiment. The growth curves of various M. smegmatis mc²155 strains (see Fig. 6) were obtained by culturing the bacteria in 5 ml of LB broth containing 0.05% Tween 80 and monitoring the optical density at 600 nm. Antibiotics were as follows: for M. smegmatis mc²155 containing plasmids pMVHG1:Rv3808c and pCG76:Tbglf, hygromycin was included in the medium at both temperatures, with streptomycin also being used at 30°C only; for M. smegmatis FP102 and M. smegmatis FP103 (see Table 1 for the plasmids in these strains), kanamycin (KAN) and hygromycin were included in the medium at both temperatures, with streptomycin also being used at 30°C only. The concentrations of antibiotics when used were as follows: 100 µg/ml for ampicillin; 5 µg/ml for gentamicin; 50 µg/ml (E. coli) and 25 µg/ml (M. smegmatis) for KAN; 100 µg/ml (E. coli) and 50 µg/ml (M. smegmatis) for hygromycin; and 20 µg/ml (E. coli) and 10 µg/ml (M. smegmatis) for streptomycin. Ten percent sucrose was added to the solid medium when required.

Transformation. Transformation of E. coli Top 10 and DH5 cells was done by following the protocol provided by the vendor. Electrocompetent M. smegmatis was made as described previously (18). Electroporation was done by setting the voltage and capacity of a Gene Pulser (Bio-Rad, Richmond, Calif.) to 2,500 V and 25 µF and the resistance of the pulse controller to 1,000 . To prepare M. smegmatis strains containing both pMVHG1:Rv3808c and pCG76:TBglf (and for strains containing both pMVHG1:Tbglf and pCG76:Rv3808c), both plasmids were electroporated into bacteria at the same time and selection was done using both streptomycin and hygromycin.

DNA extraction, Southern blot analysis, and DNA sequencing. Mycobacterial genomic DNA was extracted as described previously (1). Genomic DNA was digested overnight by appropriate enzymes and then loaded onto a 0.8% agarose gel. The gel was run at 30 V overnight (20 to 24 h). Then the DNA was transferred to a Nytran Plus membrane (Schleicher & Schuell, Keene, N.H.). The DNA was fixed to the membrane by using a Stratalinker 2400 (Stratagene, La Jolla, Calif.). For Southern blots, the DNA probe was generated by using DIG High Prime Labeling and Detection Starter Kit I (Boehringer Mannheim, Indianapolis, Ind.). The 1,595-bp SmaI fragment containing ~90% of glf and 535 bp upstream of glf of the mycobacterial lab strain was used as the probe template (see below for isolation of this DNA fragment). DNA hybridization and detection were performed as recommended by the vendor. Sequences of double-stranded plasmids were obtained by Macromolecular Resources (Colorado State University) using an ABI Prism 377 automated DNA sequencer.

Construction of vectors. pFP101 was constructed as follows. A partial mycobacterial lab strain genomic DNA library was constructed by isolation of SmaI-digested fragments of DNA from the strain of approximately 1,600 bp, ligation into the pCR-Blunt vector (Invitrogen), and maintenance in E. coli Top 10 cells. The glf gene was found to be contained in a 1,595-bp fragment by colony hybridization (6) using the entire M. tuberculosis glf gene (23) as a probe. Sequencing revealed that the fragment contained 535 bp upstream of the ATG start site and 1,060 bp of the glf sequence (approximately 90% of the glf gene). Plasmid pBluescript II SK(+) (Stratagene) was treated sequentially with BamHI, mung bean nuclease, and T4 DNA polymerase to remove the BamHI site to facilitate later cloning operations. The 1,595-bp glf-containing fragment was cut out from the pCR-Blunt vector with EcoRI and inserted into the EcoRI site of pBluescript II SK(+) (which lacks a BamHI site) to yield plasmid pFP5. A 1.2-kb KAN resistance cassette was cut out with BamHI from plasmid pUC4K, filled in with the Klenow fragment of DNA polymerase I, and inserted into the Klenow fragment-filled BamHI site of the glf gene carried by pFP5, yielding plasmid pFP6. A 2.8-kb glf::kan fragment was cut by EcoRI from pFP6 and then moved to the EcoRI site of plasmid pXYL4, a plasmid carrying the xylE gene from Pseudomonas putida (3), which yielded plasmid pFP7. Finally, the 3.8-kb (glf::kan)::xylE fragment was cut by BamHI from pFP7 and put into the BamHI site of pPR27 (Table 1) to yield plasmid pFP101 (Fig. 2), the vector used to achieve allelic replacement at the glf locus of M. smegmatis. As shown in Fig. 2, pFP101 has the mycobacterial temperature-sensitive origin of replication from the parent plasmid pPR27. Thus, it can replicate at 30°C but is efficiently lost at 39°C and above (17). Plasmid pFP101 also harbors the counterselectable marker sacB from Bacillus subtilis (17) for use in selection of the double-crossover event in the presence of sucrose. To check the orientation of xylE, kan, and glf, pFP101 plasmid was digested with BamHI and a 3.8-kb fragment was purified by using a QIAEX II kit. The 3.8-kb fragment was digested with XbaI and SmaI, and analysis of the restriction fragments by gel electrophoresis showed that the orientations of kan, glf, and xylE were as shown in Fig. 2 and 3. The genes xylE, kan, and sacB are all transcribed from their own promoters.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Key plasmids (pFP101, pCG76:TBglf, and pMVHG1:Rv3808c) used in this study. The plasmids pCG76:Rv3808c and pMVHG1:TBglf are strictly analogous to CG76:TBglf and pMVHG1:Rv3808c. TS oriM, temperature-sensitive oriM; AmpR, ampicillin resistance cassette; GmR; gentamicin resistance cassette; Str/sp, streptomycin/spectinomycin resistance cassette; HygR, hygromycin resistance cassette.

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Two possible pathways for homologous recombination between pFP101 and the M. smegmatis chromosome. Crossover upstream of kan yields a functional glf gene (with its promoter) along with functional Rv3808c and any transcriptionally linked genes further downstream. Crossover downstream from kan yields no functional glf gene, and the interrupted glf gene upstream from Rv3808c is likely to inhibit transcription of Rv3808c and any transcriptionally linked genes downstream from it. If glf, Rv3808c, or any other downstream ORF expressed from the glf promoter is essential, only single-crossover events of type 1 should occur. Also illustrated are the NruI fragments used to distinguish which single-crossover event occurred.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA sequence of the mycobacterial lab strain glf. A DNA fragment hybridizing with M. tuberculosis glf from the mycobacterial lab strain was cloned as a 1,595-bp SmaI fragment as described under Materials and Methods. Sequencing of the fragment revealed that it contained 535 bp upstream and 1,060 bp downstream of the ATG start codon of glf (approximately 90% of the ORF). During this study, sequence data of glf and surrounding regions of M. smegmatis mc²155 appeared on The Institute for Genomic Research web site (http://www.tigr.org/). There were only eight base pair differences in the 1,595-bp fragments of DNA from the two bacteria, and the deduced amino acid sequences for Glf were identical. Thus, the 1,595-bp DNA fragment could be readily used for the homologous-recombinant experiments that are discussed below.

Construction of the glf replacement plasmid (pFP101) and obtaining the first homologous-recombination event. Essentially the same strategy was used to replace glf as was used recently to replace pgsA (8). Hence, plasmid pFP101 was constructed. This plasmid carried the 1,595-bp fragment described above that had been modified so that a KAN cassette disrupted the glf gene (glf::kan) (Fig. 2). The plasmid has a temperature-sensitive mycobacterial origin of replication that facilitates obtaining recombinant strains that have undergone a single homologous-recombination event at the glf locus. It also harbors the sacB counterselectable marker (17) and the xylE colored marker (3). Plasmid pFP101 was electroporated into M. smegmatis mc²155, and transformants were selected on LB broth-KAN plates at 30°C. One transformant was then propagated in LB broth-KAN at 30°C and then plated onto LB broth-KAN plates at 42°C. Since the temperature-sensitive plasmid is able to replicate at 30°C but not at 42°C, the KAN-resistant colonies that appear on plates have necessarily integrated part or all of the vector into their chromosome. Single homologous-recombination events upstream and downstream from the KAN resistance gene resulting in genotypes 1 and 2 are shown in Fig. 3. The important difference between the two genotypes is that genotype 1 (Fig. 3) results in an intact glf-containing operon and that genotype 2 (Fig. 3) results in both the lack of an intact glf gene (since the introduced 1,595-bp fragment does not encode the entire Glf protein) and the possibility that genes downstream of glf that are dependent on the natural promoter of glf are not expressed. Illegitimate recombination which would leave ORFs Rv3809c through Rv3805c fully intact may also occur. Finally, a double-crossover event would lead to the disruption of the glf gene and presumably would affect the expression of the downstream genes. Analysis of 17 colonies from the 42°C plate by Southern blotting after digestion with NruI (Fig. 4) revealed that 10 colonies were able to grow on KAN due to illegitimate recombination and that seven colonies came from homologous-recombination pathway 1 (Fig. 3 and 4 and the legend to Fig. 4). We detected no colonies that arose from homologous-recombination pathway 2 (Fig. 3) or from a double-crossover event. These results suggested that either glf and/or a gene(s) downstream from glf is essential. One of the seven colonies that came forth from homologous-recombination pathway 1 was propagated for further experiments and named M. smegmatis FP101 (Table 1).

View larger version (37K): [in this window] [in a new window]   FIG. 4.   Southern blot analysis of M. smegmatis of the single homologous-recombination event at the glf locus of M. smegmatis. Lane 1 is wild-type M. smegmatis; lanes 2 to 18 are 17 clones selected from the LB broth-KAN plates at 42°C. The DNA was cleaved with NruI, and the 1,595-bp glf-containing fragment was used as the probe template. The DNAs in lanes 3, 5, 9, 11, 13, 14, and 16 resulted from type 1 homologous recombination (Fig. 3), as bands at 1.2 and 15.6 kb are evident, whereas the DNAs in lanes 2, 4, 6, 7, 8, 10, 12, 15, 17, and 18 come from clones with illegitimate recombination, as wild-type glf at 2.4 kb is evident, as are the expected two bands of various sizes.

Construction of rescue plasmids. Before attempting the second crossover event, two sets of rescue plasmids were constructed. It could not be predicted if the KAN resistance cassette would introduce a polar effect (i.e., lack of transcription of Rv3803c) or not. It is possible that transcription of mRNA could continue on past the KAN resistance cassette since no obvious termination sequence is present as part of the cassette. However, there are also many ways that the presence of the KAN resistance gene might result in the downstream Rv3808c gene not being transcribed or translated. Therefore, the first set of rescue plasmids consisted of pCG76:TBglf and pMVHG1:Rv3808c (Table 1 and Fig. 2). The pMVHG1:Rv3808c was merely M. tuberculosis Rv3808c under the control of P_hsp60 (Table 1); pCG76:TBglf was glf under the control of P_hsp60 but in a plasmid with the same temperature-sensitive origin of replication used in pFP101, so that later experiments to cure the cells of this plasmid could be performed. The second set of rescue plasmids was complementary to the first set and consisted of pCG76:Rv3808c and pMVHG1:TBglf (Table 1). In this case, Rv3808c was in the plasmid containing the temperature-sensitive promoter and could be cured in later experiments. M. smegmatis FP101 bacteria (Table 1) containing various combinations of these plasmids were then prepared.

Second crossover attempts and events. Single-colony isolates of M. smegmatis FP101 (Table 1) containing no rescue plasmid and various rescue plasmids were grown in LB medium (containing appropriate antibiotics as reported in Table 2) at 30°C and then plated onto LB broth-sucrose plates at 30°C. The resulting colonies (Table 2) were analyzed for their XylE phenotype (a yellow color develops in colonies expressing xylE when they are sprayed with catechol). Colonies that have undergone a second crossover should both be able to grow on sucrose and have lost the XylE marker; colonies that can grow on sucrose but still express xylE are likely to be sacB mutants rather than arising from the second crossover event. Thus, only the white colonies were candidates for the second crossover event occurring. Examination of Table 2 revealed that a small number (10%) of white colonies were formed when only pCG76:TBglf was present. Analysis of 12 of these colonies by Southern blot analysis showed that none of them resulted from the second crossover event (data not presented). In contrast, when one of the rescue plasmid pairs, pCG76:TBglf and pMVHG1:Rv3808c or pCG76:Rv3808c and pMVHG1:TBglf, was present, 100% of the colonies were white. Southern blot analysis (Table 2 and Fig. 5) revealed that, in these cases, genuine second crossover events occurred. This outcome strongly suggested that both glf and Rv3808c are essential and expressed from the same promoter in M. smegmatis and that the KAN resistance cassette for some reason introduces a polar mutation. No information on whether Rv3807c, Rv3806c, and Rv3805c are needed for growth of M. smegmatis was obtained in these experiments as these genes may be expressed from a different promoter than the glf promoter. One of the colonies showing the genuine second crossover event with the rescue plasmid pair pCG76:TBglf and pMVHG1:Rv3808c was named M. smegmatis FP102 and propagated for further experiments; one of the colonies showing the genuine second crossover event with the rescue plasmid pair pCG76:Rv3808c and pMVHG1:TBglf was named M. smegmatis FP103 and propagated for further experiments (Table 1).

View larger version (33K): [in this window] [in a new window]   FIG. 5.   Southern blot analysis of M. smegmatis FP102 (the glf knockout strain) containing plasmids pMVHG1:Rv3808c and pCG76:TBglf. The DNA was cleaved with NruI, and the 1,595-bp glf-containing M. smegmatis fragment was used as the probe template. Lanes 1 to 3 are controls; lanes 4 to 6 are positive for the second crossover event. Lane 1, plasmid pCG76:TBglf only; lane 2, M. smegmatis mc2155 wild type; lane 3, M. smegmatis FP101 (first single-crossover bacterium [see also Fig. 4]) with plasmids pMVHG1:Rv3808c and pCG76:TBglf before selection on sucrose for the second crossover event; lanes 4 to 6, three colonies of M. smegmatis FP102 (Table 1) containing plasmids pMVHG1:Rv3808c and pCG76:TBglf. The origins of the bands at 1.2 and 2.4 kb in M. smegmatis FP102 are illustrated. Both the wild type (lane 2) and M. smegmatis FP102 yield a band near 2.4 kb (2.39 kb for the wild type and 2.41 kb for M. smegmatis FP102). However, the band near 2.4 kb also hybridizes with a probe made from the KAN resistance cassette in the case of M. smegmatis FP102 (lanes 4 to 6) but does not hybridize in the case of wild-type M. smegmatis (lane 2) (data not presented). The bands in lanes 3 to 6 (M. smegmatis FP102) at 11 and 0.5 kb come from plasmid pCG76:TBglf (see lane 1) being present in M. smegmatis FP102.

Neither M. smegmatis FP102 nor M. smegmatis FP103 grow at 40°C. As a final experiment to confirm that UDP-galactopyranose mutase and the galactofuranosyl transferase are essential for growth, M. smegmatis FP102 containing plasmids pCG76:TBglf and pMVHG1:Rv3808c and M. smegmatis FP103 containing plasmids pCG76:Rv3808c and pMVHG1:TBglf were shown to be unable to grow at 40°C (Fig. 6), a temperature at which pCG76 and its insert are lost. These experiments were conducted at 40°C rather than at 42°C because it was found that M. smegmatis mc²155 containing pCG76:TBglf and pMVHG1:Rv3808c was unable to grow at 42°C, presumably due to the stress of harboring two plasmids.

View larger version (16K): [in this window] [in a new window]   FIG. 6.   Growth curves of M. smegmatis strains at 30 and 40°C. Shown are results with M. smegmatis mc2155 containing plasmids pMVHG1:Rv3808c and pCG76:TBglf at 30°C () or at 40°C (), M. smegmatis FP102 (Table 1) at 30°C () or at 40°C (), and M. smegmatis FP103 (Table 1) at 30°C () or at 40°C (). The medium was LB broth in all cases, and antibiotics were present as detailed in Materials and Methods. M. smegmatis mc2155 containing plasmids pCG76:Rv3808c and pMVHG1:TBglf, a second control construct, grew at both 30 and 40°C (data not presented). The slight lag in growth seen for the two knockout strains of M. smegmatis at 30°C is likely due to a weaker inoculum, as evidenced by the optical density (OD) at 600 nm at time zero.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The inability to form a second recombination event, i.e., a knockout of glf in the absence of both glf and Rv3808c rescue plasmids (Table 2), coupled with the inability of M. smegmatis FP102 and FP103 to grow without pCG76:TBglf and pCG76:Rv3808c, conclusively demonstrates that UDP-galactopyranose mutase and the galactofuranosyl transferase are necessary for the viability of M. smegmatis. Even though glf was the only gene knocked out in strains FP102 and FP103, the fact that inserted DNA, kan, caused a polar effect on the gene Rv3808c allowed us to determine that Rv3808c is also essential. It is not surprising that our experiments demonstrate that kan causes a polar effect, as there are several possible ways that transcription or translation may be stopped and/or not initiated downstream of kan. However, the mechanism for the polar effect has not been determined. The experiments were done with M. smegmatis due to the fast-growth characteristics of this organism and the availability of a temperature-sensitive origin of replication for it. With respect to other mycobacteria, we have shown that the basic structure of the cell wall core of all mycobacteria is indistinguishable by ¹³C nuclear magnetic resonance and oligosaccharide profiling (5). In addition, glf and Rv3808c are found in the genomes of M. tuberculosis, Mycobacterium bovis, Mycobacterium avium, M. smegmatis, and Mycobacterium leprae. The presence of these genes in M. leprae is of special note due to the fact that the M. leprae genome is smaller than that of M. tuberculosis, many potential ORFs are pseudogenes, and half of the DNA is noncoding (7). It therefore seems likely that only more necessary genes are present in M. leprae. Thus, the similarity of the cell wall core structure along with the identical genetic organizations around glf argue strongly that UDP-galactopyranose mutase and the galactofuranosyl transferase encoded by Rv3808c are essential in all mycobacteria, including M. tuberculosis.

Demonstration that UDP-galactopyranose mutase and one of the galactofuranosyl transferases are essential for mycobacterial growth is part of a logical sequence of a long-range M. tuberculosis drug development program. Initially, the cell wall core arabinogalactan was structurally characterized (2, 4, 13, 14). This led to the recognition that inhibiting the formation of any of three fundamental structural components of the arabinogalactan, L-rhamnosyl, D-arabinofuranosyl, and D-galactofuranosyl residues, was a logical approach to developing new tuberculosis drugs because of the key structural roles of these components (4) and the lack of these three glycosyl residues in humans. The next logical step required determining how these components were biosynthesized and led, in the case of galactofuranosyl residues, to the expression and characterization of UDP-galactopyranose mutase (23) and recognition that Rv3808c encodes a galactofuranosyl transferase (15). The next step was to prove that the formation of galactofuranosyl residues is essential for growth of the mycobacterium, and this has now been accomplished. Following this, inhibitors of the mutase and/or galactofuranosyl transferases are being sought. The enzyme inhibitors that gain entry into the mycobacterium and thus inhibit the growth of M. tuberculosis will then be candidates for further development. To identify the enzyme inhibitors, facile assays amenable to a microtiter plate format for UDP-galactopyranose mutase are currently being developed based on the release of radioactive formaldehyde from UDP-6-[³H]Galf by periodate. Assays for the transferase will require determination of the exact substrates of the enzyme; then a scintillation proximity assay where the acceptor is attached to scintillation proximity assay beads may be feasible.