A Point Mutation in cycA Partially Contributes to the D-cycloserine Resistance Trait of Mycobacterium bovis BCG Vaccine Strains

In mycobacteria, CycA a D-serine, L- and D-alanine, and glycine transporter also functions in the uptake of D-cycloserine, an important second-line anti-tubercular drug. A single nucleotide polymorphism identified in the cycA gene of BCG was hypothesized to contribute to the increased resistance of Mycobacterium bovis bacillus Calmette-Guérin (BCG) to D-cycloserine compared to wild-type Mycobacterium tuberculosis or Mycobacterium bovis. Working along these lines, a merodiploid strain of BCG expressing Mycobacterium tuberculosis CycA was generated and found to exhibit increased susceptibility to D-cycloserine albeit not to the same extent as wild-type Mycobacterium tuberculosis or Mycobacterium bovis. In addition, recombinant Mycobacterium smegmatis strains expressing either Mycobacterium tuberculosis or Mycobacterium bovis CycA but not BCG CycA were rendered more susceptible to D-cycloserine. These findings support the notion that CycA-mediated uptake in BCG is impaired as a result of a single nucleotide polymorphism; however, the partial contribution of this impairment to D-cycloserine resistance suggests the involvement of additional genetic lesions in this phenotype.


Introduction
Mycobacterium bovis bacillus Calmette-Guérin (BCG), is a live vaccine originally derived from a virulent isolate of Mycobacterium bovis, and has been used to immunize more than three billion people against tuberculosis [1]. Despite widespread use, the protective efficacy imparted by BCG wanes significantly with time, and the molecular mechanism for this is poorly understood [1,2]. Additionally, the mechanisms underlying the derivation of attenuated BCG from virulent M. bovis also remain incompletely understood [1,2]. To further complicate matters, 13 strains of BCG with documented differences in attenuation and protective efficacy currently exist, and several of these are in use in various parts of the world [3,4]. Clearly, the basic biology of BCG is far from being fully understood. The availability and simultaneous analyses of complete genome sequences of several M. bovis and BCG strains however have generated fresh insights into the biology of BCG [1,[4][5][6]. Indeed, the demonstration that loss of the RD1 locus by BCG was a key event leading to its attenuation, originated from such comparative genome studies [7]. D-cycloserine (DCS), a cyclic analog of D-alanine, is an important second-line antibiotic used to treat multi-drug-and extensively drug-resistant Mycobacterium tuberculosis infections [8,9]. Compared to wild-type M. tuberculosis and M. bovis, BCG has always been found to be more resistant to DCS [6,10]. Although the molecular basis for this phenotype is unknown, this feature is used to differentiate the vaccine strain from M. bovis and M. tuberculosis strains [10]. In vitro, DCS inhibits M. tuberculosis alanine racemase (Alr) which converts L-alanine to D-alanine, and Dalanine:D-alanine ligase (Ddl) which synthesizes D-alanine pentapeptides [8,11,12]. Both enzymes are required for the synthesis of peptidoglycan in the cell wall of mycobacteria [8,11,12]. Overproduction of Alr and Ddl in BCG and Mycobacterium smegmatis confers increased DCS resistance [13,14], while their genetic inactivation in M. smegmatis increases susceptibility to the drug [15,16]. The resistance determinants and the cellular targets of DCS in M. tuberculosis and M. bovis are presumed to be Alr and Ddl, both of which are essential in M. tuberculosis [17]. However, the nucleotide sequences and expression patterns of alr and ddl in M. bovis and BCG appear to be identical and as such, might not contribute to the DCS resistance of the vaccine strain [1,6].
A non-synonymous single nucleotide polymorphism (nsSNP) in the cycA gene of all BCG strains was recently identified and predicted to result in a glycine 122 to serine substitution (G122S) in the transporter [6]. CycA is a bacterial D-serine/L-and Dalanine/glycine/D-cycloserine: proton symporter, comprising 12 helical trans-membrane domains, of the amino acid transporter (AAT) family (TCID: 2.A.3.1.7) [18]. In the early 1970's, David isolated and characterised two low-level and equally DCS-resistant forms of M. tuberculosis mutants -one that was also D-serine, L-and D-alanine, and glycine uptake defective, and the other uptake competent [19]. Point mutations resulting in transitions and transversions, as well as small deletions and duplications within the cycA gene of Escherichia coli, have been found to confer resistance to DCS [20]. Inspired by these reports, it was hypothesized that the G122S change functionally impairs BCG CycA and contributes to the vaccine strain's characteristic resistance to DCS [6].
A bioinformatics analysis shows that the G122S change in BCG CycA lies in a presumably critical extracellular loop between the conserved third and fourth helical trans-membrane domains. Taking a genetic approach we show that BCG CycA is indeed impaired for DCS uptake, however the G122S change only partially contributes to DCS resistance in BCG, thus implicating additional mutations in this phenotype.

In silico Analyses and Helical Trans-membrane Domain Modeling Predicts a Functionally Important Extracellular Loop in CycA
Comparison of CycA orthologues from six different mycobacteria revealed 72-99% identity at the amino acid sequence level. Multi-species protein sequence alignment shows that CycA from all mycobacterial strains, except BCG possess a conserved glycine residue in position 122 ( Figure 1A). Helical trans-membrane (HTM) domain modeling of the M. bovis AF 2122/97 CycA based on HMMTOP prediction [21], shows the transporter to be a cell membrane protein with 12 HTM domains ( Figure 1B). The G122S change in BCG CycA is predicted to lie in an extracellular loop between the third and fourth trans-membrane helices ( Figure 1B). Similar modeling of the HTM domains of M. tuberculosis H37Rv, M. marinum M and M. smegmatis mc 2 155 CycA also place the conserved glycine residue at the beginning of the extracellular loop between the third and fourth trans-membrane helices. Single point mutations in the cycA gene have been identified in DCS-resistant E. coli mutants [20]. Therefore E. coli CycA was also modeled and DCS-resistance-associated single point mutations were mapped to the model. Like the mycobacterial CycA proteins, E. coli CycA was predicted to form a membrane protein with 12 HTM domains. Of twenty previously identified point mutations, we were able to place a threonine 114 to proline substitution (from an adenine to cytosine transversion) [20], in the extracellular loop between the third and fourth transmembrane helices of E. coli CycA ( Figure S1).
Our in silico analyses suggest that the extracellular loop between the third and fourth trans-membrane helices of CycA may be sensitive to modifications and is functionally important for transport activity.
Merodiploid Expression of M. tuberculosis CycA in BCG Partially Restores Susceptibility to D-cycloserine A merodiploid strain of M. bovis BCG-Pasteur was generated by transformation with an integrating cosmid (I425) containing the M. tuberculosis cycA allele. Sequencing from the 59-end of the cycA PCR product amplified from the genomic DNA of a Pasteur::I425 knock-in clone showed both GGC (encoding Gly) and AGC (encoding Ser) sequences, indicating the successful integration of the cosmid into the chromosome (Figure 2A). In contrast, a Pasteur::pYUB412 control showed only the AGC (encoding Ser) sequence ( Figure 2A). 39-end sequencing of cycA PCR products from both clones confirmed these findings.
The susceptibility of the Pasteur::I425 knock-in strain to DCS was assessed and compared to that of the Pasteur::pYUB412 control strain. In 7H9 broth cultures containing increasing concentrations of DCS (0, 2, 4, 8, 16 and 32 mg/mL), the Pasteur::pYUB412 strain showed complete growth inhibition at .32 mg/mL of the antibiotic, which is consistent with published observations [10,13,22]. The Pasteur::I425 strain however, appeared to be more susceptible to DCS, exhibiting significant growth inhibition at 16 and 8 mg/mL DCS compared to the control strain -the colony forming units of the Pasteur::I425 strain recovered was almost 1-log less at these concentrations of DCS than that of the control strain ( Figure 2B). The difference in susceptibility to DCS between the two strains was even more pronounced in experiments measuring growth inhibition over time upon exposure to 32 mg/mL of the antibiotic. Pasteur::I425 showed more rapid inhibition of growth by DCS compared to the Pasteur::pYUB412 strain as the colony forming units of the knockin strain recovered at all time-points post-exposure was consistently 1-log lower than the control strain ( Figure 2C). However, we did not observe in the Pasteur::I425 strain, the restoration of wild-type M. tuberculosis or M. bovis levels of DCS susceptibility, which is reportedly between 4 and 5 mg/mL of DCS [10,19].
We next wanted to verify that the higher susceptibility of the Pasteur::I425 strain to DCS was not due to the integration of an extra copy of the cycA gene. Although M. smegmatis mc 2 155 has a higher MIC for DCS [13,14] compared to members of the M. tuberculosis complex, we transformed this strain with I425 and pYUB412 to test if we could obtain a similar fold increase in susceptibility to DCS as seen with BCG. Using M. tuberculosis cycAspecific primers we confirmed the successful integration of the cosmid prior to testing with DCS (data not shown). Growth of wild-type M. smegmatis mc 2 155 is inhibited by DCS at $75 mg/mL [13,14], and consistent with this, the M. smegmatis::pYUB412 strain exhibited complete growth inhibition at $80 mg/mL DCS ( Figure 2D). The growth of the M. smegmatis::I425 knock-in strain was inhibited by DCS to the same extent as the M. smegmatis::-pYUB412 control strain ( Figure 2D). Likewise, time-course experiments measuring growth inhibition by DCS also did not show significant differences between the two strains ( Figure 2E). This shows that having one extra copy of M. tuberculosis cycA does not appreciably increase DCS susceptibility in M. smegmatis. All of these observations taken together indicate that the increased DCS susceptibility of Pasteur::I425 results from the production of a functional CycA in a genetic background expressing either a nonfunctional or impaired CycA.
M. smegmatis Overexpressing BCG CycA does not become More Susceptible to DCS Heterologous expression of M. tuberculosis proteins involved in efflux and drug resistance in M. smegmatis has been an effective approach to study their function [23][24][25]. We hypothesized that if M. bovis BCG CycA is defective, then M. smegmatis strains overexpressing functional M. tuberculosis and M. bovis CycA but not BCG CycA, would become more susceptible to DCSmediated growth inhibition. Working along these lines, M. tuberculosis, M. bovis and BCG cycA alleles and their promoter regions were individually cloned into the pMD31 plasmid for multi-copy ectopic expression in M. smegmatis mc 2 155. Colony PCR was used to discriminate bona fide transformants harbouring plasmids containing M. tuberculosis, M. bovis or BCG cycA alleles from those harbouring the empty pMD31 vector alone (data not shown).
The DCS sensitivities of these recombinant M. smegmatis strains were then tested in 7H9 broth containing increasing concentrations of DCS (0, 40, 80 or 160 mg/mL). M. smegmatis strains producing M. tuberculosis and M. bovis CycA exhibited significantly increased susceptibility to DCS-mediated growth inhibition compared to the vector-only control strain ( Figure 3A). In contrast, the susceptibility of M. smegmatis producing BCG CycA was similar to that of the vector-only control strain ( Figure 3A). Consistent with this observation, in the presence of 200 mg/mL DCS, only M. smegmatis strains producing M. tuberculosis and M. bovis CycA but not BCG CycA exhibited significantly faster growth inhibition compared to the vector-only control ( Figure 3B). These results indicate that raising the gene-dosage of cycA readily increases DCS susceptibility in M. smegmatis. More importantly it shows that BCG CycA is defective because even in the M. smegmatis genetic background, raising the gene-dosage of BCG cycA does not affect its susceptibility to DCS.

Extra Copies of cycA Alleles in Mycobacteria do not Confer an in vitro Growth Advantage
Since CycA is a D-serine, L-and D-alanine, and glycine transporter, we wanted to verify that the differences in DCS susceptibilities observed were not due to changes in the growth rates of the strains tested. Accordingly growth measurements were made for all of the strains in 7H9 broth. No difference in growth between the merodiploid Pasteur::I425 and the Pasteur::pYUB412 control strain were found (data not shown). Similarly no growth differences between M. smegmatis::I425 and M. smegmatis::pYUB412 were observed (data not shown). Consistent with the above results, M. smegmatis strains producing M. tuberculosis, M. bovis and BCG CycA, also grew at rates similar to the control strain in 7H9 media (data not shown). These findings indicate that expression of extra copies of cycA do not confer a growth advantage to the host strain in the culture media used to test for susceptibility to DCS.

Discussion
We employed bioinformatics as well as two complementary genetic approaches to investigate the involvement of the nsSNP in BCG cycA in DCS resistance. Evidence is provided that suggests this nsSNP causes a defect in the BCG CycA transporter but only partially contributes to DCS resistance in the vaccine strain.  Trans-membrane helical domain modeling places the BCG G122S mutation in an extracellular loop between the conserved third and fourth helical trans-membrane domains of CycA. This finding hints at possible substrate recognition or binding defects at the structural level in BCG CycA and warrants further study. Consistent with the notion that BCG CycA is defective for DCS uptake, heterologous overexpression of CycA from M. tuberculosis and M. bovis but not from BCG in M. smegmatis results in increased susceptibility to DCS. This effect of increased cycA gene-dosage on DCS susceptibility in M. smegmatis raises the question of whether a decrease in CycA expression in BCG might result in increased resistance. However, the identical expression patterns of cycA in M. bovis and BCG rule out this possibility [1,6]. More importantly, the G122S mutation in BCG CycA appears to be a minor contributor as we were unable to restore M. tuberculosis and M. bovis levels of DCS susceptibility to the merodiploid Pasteur::I425 strain. This is not surprising as the isolation of two equally DCS-resistant mutants of M. tuberculosis, one of which is transport defective and the other transport competent, has been reported [19]. This strongly suggests there are other genetic lesions besides the nsSNP in cycA that also contribute to DCS resistance in BCG. Indeed an M. smegmatis isolate, resistant to both DCS and vancomycin was found to have no mutations in the alr, ddl or cycA genes [26]. Only multi-copy transformation of a gene encoding a putative penicillin-binding protein (Pbp4) homolog restored DCS and vancomycin susceptibility to this strain [26]. Our analyses of pbp genes in BCG however did not reveal any nsSNPs that might explain the BCG phenotype. Moreover, a recent metabolomic study of M. smegmatis revealed that there may potentially be alternative routes to D-alanine synthesis in mycobacteria that could circumvent DCS inactivation of Alr and Ddl [12]. This notion is supported by the observation that alr mutants of M. smegmatis are not complete D-alanine auxotrophs [12,16]. As such, genetic mutations in these alternate alanine metabolic pathways could potentially augment CycA-mediated DCS resistance in BCG.
BCG underwent in vitro evolution during its initial derivation from a virulent M. bovis progenitor strain as well as during its subsequent global dissemination [1,2]. A consequence of this evolution was the acquisition of mutations causing defects in Lserine and L-alanine catabolism, dysregulated nitrogen metabolic pathways and in vitro growth defects [27]. CycA is the sole transporter for D-serine, L-and D-alanine, and glycine in mycobacteria, and their uptake by the transporter is induced by L-alanine [19]. Although in this study BCG and M. smegmatis strains bearing extra cycA alleles and their control strains grew similarly in 7H9 broth, growth differences may yet appear in Lserine and/or L-alanine-replete conditions not found in 7H9 broth. A link between defective L-serine and L-alanine catabolism and the acquisition of the CycA G122S mutation impairing Dserine, L-and D-alanine and DCS uptake is intriguing and merits additional investigation as this will add to our understanding of BCG nitrogen metabolism.
In conclusion, this study reveals the involvement of CycA in the resistance of BCG to DCS and provides fresh insights into BCG biology as well as clues for the improvement of DCS as an antimycobacterial agent.

Genetic Manipulations and Genotyping
An integrating cosmid (I425) [28], bearing the M. tuberculosis cycA allele (rv1704c) was transformed into M. bovis BCG-Pasteur 1173P2 and M. smegmatis mc 2 155 following standard electroporation procedures [29]. pYUB412, the cosmid backbone, was electroporated to obtain parental control strains. Genomic DNA was extracted from several M. bovis BCG colonies obtained after selection on 7H11 agar plates containing hygromycin. Positive BCG-Pasteur::I425 clones were verified by PCR amplification of the genomic DNA using cycA-specific internal primers (cycA-int-Forward: agctgctgctgtcgaacctg and cycA-int-Reverse: gttgggaagaacccgttgtc) and sequencing of the PCR products. M. smegmatis::I425 clones were obtained and identified by PCR amplification using M. tuberculosis cycA-specific primers described above. M. smegmatis::pYUB412 served as control. Table 1  cycA-PstI-Reverse: cgcctgcagcgtgggcgaggacatag introducing 59-HindIII (bold italics) and 39-PstI (bold italics) restriction sites. The PCR fragments obtained were digested with HindIII and PstI, purified and then ligated into HindIIIand PstIlinearized pMD31 [30]. Plasmid clones containing cycA alleles were verified by sequencing before electroporation into M. smegmatis mc 2 155. Positive transformants were identified by colony PCR using the primers cycA-HindIII-Forward and cycA-PstI-Reverse described above.

D-cycloserine Susceptibility Testing
DCS is unstable in aqueous solutions buffered at pH 7 including Middlebrook 7H9 broth [8,31] therefore working solutions of the antibiotic were prepared freshly in sodium phosphate buffer (pH 9.0) just before use. Two-fold serial dilutions of DCS covering the required range of concentrations were prepared in 7H9 broth (supplemented with ADC and 0.05% Tween-80) for all experiments. For assessing DCS-sensitivity over time, the antibiotic was added to 7H9 broth (supplemented with ADC and 0.05% Tween-80) at approximately 2X the reported MIC per strain for every experiment. Colony forming units at different time points were determined by 7H11 agar plate spreading of serially diluted aliquots of cells.

Statistics
The unpaired, two-tailed t test was used to assess the statistical significance in differences when comparing experimental groups using the GraphPad Prism program (http://www.graphpad.com).