Deciphering the Growth Behaviour of Mycobacterium africanum

Background Human tuberculosis (TB) in West Africa is not only caused by M. tuberculosis but also by bacteria of the two lineages of M. africanum. For instance, in The Gambia, 40% of TB is due to infections with M. africanum West African 2. This bacterial lineage is associated with HIV infection, reduced ESAT-6 immunogenicity and slower progression to active disease. Although these characteristics suggest an attenuated phenotype of M. africanum, no underlying mechanism has been described. From the first descriptions of M. africanum in the literature in 1969, the time to a positive culture of M. africanum on solid medium was known to be longer than the time to a positive culture of M. tuberculosis. However, the delayed growth of M. africanum, which may correlate with the less virulent phenotype in the human host, has not previously been studied in detail. Methodology/Principal Findings We compared the growth rates of M. tuberculosis and M. africanum isolates from The Gambia in two liquid culture systems. M. africanum grows significantly slower than M. tuberculosis, not only when grown directly from sputa, but also in growth experiments under defined laboratory conditions. We also sequenced four M. africanum isolates and compared their whole genomes with the published M. tuberculosis H37Rv genome. M. africanum strains have several non-synonymous SNPs or frameshift mutations in genes that were previously associated with growth-attenuation. M. africanum strains also have a higher mutation frequency in genes crucial for transport of sulphur, ions and lipids/fatty acids across the cell membrane into the bacterial cell. Surprisingly, 5 of 7 operons, recently described as essential for intracellular survival of H37Rv in the host macrophage, showed at least one non-synonymously mutated gene in M. africanum. Conclusions/Significance The altered growth behaviour of M. africanum might indicate a different survival strategy within host cells.


Introduction
Mycobacterium africanum, a member of the Mycobacterium tuberculosis complex, was first described in 1968 in Dakar, Senegal [1]. Infections with M. africanum are generally geographically restricted to human populations in West Africa, and are not well understood [2]. Molecular techniques have since refined the classification of the sub-species M. africanum into M. africanum West African 1, common around the Gulf of Guinea, and M. africanum West African 2, mainly found in Western West Africa [3,4]. Although up to 30-40% of all human tuberculosis in West Africa is caused by either of the two M. africanum lineages [2], basic research on these clinically important mycobacteria was neglected to date. However, an improved understanding of the biology of this mycobacterial lineage will also give clues about genetic functions in the closely related M. tuberculosis.
The biochemical characteristics of M. africanum vary; at times, they resemble those of M. bovis, and, at times, M. tuberculosis [5]. Clinically and epidemiologically, M. africanum behaves very differently from M. tuberculosis. For instance, studies from The Gambia showed that M. africanum West African 2 is associated with HIV infection [2], reduced ESAT-6 immunogenicity [6] and a slower progression to active disease [7]. These features suggest an overall attenuation of the bacterium, yet no underlying mechanism has been identified to date. From the first descriptions of M. africanum, the time to detection on solid medium was known to be longer for M. africanum compared to M. tuberculosis. This delayed growth, which may explain the reduced virulence of M. africanum, has not previously been investigated.
We determined the bacterial growth rate of molecularly characterized lineages of the M. tuberculosis complex collected from The Gambia in two liquid culture systems, both directly from sputum and in carefully controlled growth experiments. M.
africanum West-African 2 (from now on referred to as M. africanum) grows significantly slower than M. tuberculosis in all of the culture systems we used. By comparisons of genetic sequence data, M. africanum strains have several mutations in genes that were previously associated with growth-attenuation in M. tuberculosis H37Rv. This high mutation frequency was also observed in functional groups of molecular membrane transport systems that translocate macromolecules and nutrients across the cell membrane into the bacterial cell. We conclude that the altered growth behaviour of M. africanum may be a different survival strategy within the host.

Mycobacterial growth curves from sputum and from standardized inoculum
In the context of several TB cohort studies, we collected clinical isolates from patients with smear positive pulmonary TB. Each TB patient submitted up to three sputum samples. Sputum was decontaminated using NALC-NaOH and inoculated into either BACTEC MYCO/F-Sputa vials (for the BACTEC 9000, BD) and/or BACTEC MGIT 960 Tubes supplemented with PANTA (for BACTEC MGIT 960, BD). The tubes were incubated at 37uC and the ''Time to Positivity'' (manufacturer-set threshold: 75 Growth Units) was recorded in days. Tubes were incubated for a maximum of 42 days.
In a second experiment we compared the growth rates of M. tuberculosis laboratory strain Mt14323 [8] and clinical M. africanum isolate ITM 080552 in a controlled laboratory setting, using defined inocula. For each strain, a standardized inoculum was prepared from a fresh subculture of 21 days with a turbidity of McFarland Nu 0.5. The OD 492 nm and OD 595 nm were measured and the bacterial suspension was adjusted to OD = 0.01-0.03. A dilution of 1:10 was prepared in distilled water. From this dilution we made a half logarithmic dilution series and 100 ml of each dilution was inoculated in triplicate into MGIT960. To estimate colony-forming-units (CFU) each inoculum was plated on 7H11 plates. Growth curves were monitored using the BD Epicentre software, data were extracted, and the length of the lag phase or ''Time to Positivity'' for each strain to reach the ''positivity threshold'' of 75 growth units (GU) was measured. Furthermore, the actual growth rate or doubling time was determined as the time needed for a strain to grow from 5000GU to 10000GU. We used the non-parametric Wilcoxon rank sum test to compare the median time to positivity for M. tuberculosis versus M. africanum.

Ethical statement
The samples used are all from the MRC strain collection, which comprises strains from various studies that were conducted over the last years. All these studies obtained ethical approval, informed consent from patients and samples were anonymized.

Genotyping mycobacterial isolates
Genotyping was done using spoligotype analysis [9] and PCR for Large Sequence Polymorphisms [4] on the mycobacterial DNA from one isolate from each patient, with the assumption that all samples from the same patient contained the same mycobacterial isolate. Isolates were grouped in phylogenetically distinct lineages within the M. tuberculosis complex, as previously defined [4].

Next generation sequencing for whole bacterial genomes
We sequenced the genomes of four M. africanum West Africa 2 isolates, three that originated from The Gambia and one publicly available strain from Senegal. We re-sequenced the published strain GM041982 [10], two randomly selected Gambian isolates 03/03910 and 03/030671 from MRC's strain collection, and strain ATCC 35711.
Library preparation. 2.5 mg of genomic DNA products of each sample were combined and fragmented into a peak size range of 200-400 bp using the Covaris S2 (Covaris, Woburn, MA, USA) (shearing conditions -Duty cycle: 20%; Intensity: 4; Cycles per burst: 200; Time: 360 seconds). After fragmentation, the samples were purified using the Qiagen PCR purification kit (Qiagen, Valencia, CA, USA). Fragmented products were quality-checked (2100 Bioanalyzer on a DNA 1000 Chip, Agilent Technologies, Santa Clara, CA, USA). The NEB Next DNA Sample Prep Master Mix kit (New England Biolabs, Ipswich, MA, USA) was used. Library preparation entailed: end-repair, A-tailing and ligation of adapters according to the manufacturer's instructions. Size selection was conducted on a Pipen Prep from Sage Science, fragments in the range 300-500 bp were selected. Then a qualitycheck of the size selected product was run on the 2100 Bioanalyzer (DNA High Sensitivity DNA Chip). Finally, using the Multiplexing Sample Preparation Oligonucleotide Kit (Illumina, San Diego, CA, USA), samples underwent 14 PCR cycles to incorporate indexes followed by Agencourt AMPure XP magnetic bead (Beckman Coulter, Brea, CA, USA) clean up according to manufacturer's instructions. One more quality check was conducted using the Bioanalyzer with a run on a DNA 1000 Chip, and all samples were adjusted to a final concentration of 10 nM. A qPCR step was performed to ensure all material sent for sequencing contained the adaptors and indexes. We Multiplexed sequencing. Next generation sequencing was done using Illumina Hiseq 2000 flow cell, 2676 base pair-end runs. PhiX was used as control. Sequencing was carried out by

Author Summary
Mycobacterium tuberculosis and Mycobacterium africanum are the two major lineages within the M. tuberculosis complex that cause human tuberculosis in West Africa. Despite being closely related, the outcome after infection differs between these two pathogens. Although M. africanum has not yet been studied to the same extent as M. tuberculosis, M. africanum is less likely to stimulate the host immune system or to progress to active disease. We hypothesized that this somewhat attenuated phenotype is due to the slower growth of M. africanum within the host. Therefore, we analysed clinical isolates from 522 patients with tuberculosis in The Gambia. M. africanum West Africa 2 strains grew more slowly than M. tuberculosis. We sequenced four M. africanum strains and identified several candidate genes that may cause the growthattenuation of the bacteria. Describing the fundamental genomic and phenotypic differences between M. tuberculosis and M. africanum will enable us to better understand the virulence mechanisms that make M. tuberculosis one of the most successful bacterial pathogens, and to discover potential strategies to interfere with mycobacterial pathogenicity.
Genome Technology Biology team in Genome Institute of Singapore.
Analysis of whole genome sequences. Unix Korn Shell was used to access the server and perform the analysis. Scripts written by the bioinformatics team in GIS were used to do assembly, call SNPs and build a phylogeny tree for the sample set. CLC Genomics was used to visualize quality of the reads, mapped SNPs and align sequences.
Genetic comparison of M. tuberculosis H37Rv and M. africanum We compared selected genes from the M. tuberculosis H37Rv genome [11] with their respective homologues of the sequenced M. africanum strains. We only considered single nucleotide polymorphisms (SNPs) or deletion/insertion polymorphism (DIPs) that were common to all four M. africanum strains. Genes that were only mutated in some of the sequenced strains were considered to be uncommon to M. africanum and were considered wildtype genes. The analysed set of genes which is responsible for attenuated growth in vitro was extracted from a previous publication [12]. Genes and operons essential for in vivo growth within macrophages were recently published [13]. Genes encoding nutrient and macromolecule transport mechanisms were identified from the literature [14][15][16][17] and by a NCBI PubMed search. Additionally, genes annotated as transporters were identified in the TubercuList database (http://tuberculist. epfl.ch/). In both searches, genes encoding transport proteins with unknown substrate specificity or annotated drug/antibiotic efflux pumps were excluded from the analysis. To compare the proportion of genes carrying non-synonymous SNPs between groups the Fisher's exact test was conducted and the results were considered significant at the level of p#0.05, assuming the likelihood of a Type I error was a = 0.05.

Predicting the impact of non-synonymous mutations on protein function
To understand whether an amino acid substitution affected protein function in M. africanum, we conducted an analysis using the SIFT (''Sorting Intolerant from Tolerant'') Sequence algorithm (http://sift.bii.a-star.edu.sg/www/SIFT_seq_submit2.html) [18]. The parameters were used at their default setting and the gene sequences and respective substitutions were analysed using the ''UniProt-SwissProt+TrEMBL 2010_09 database'' as reference.

Growth curves from sputum
Among isolates from 552 TB cases the prevalence of M. africanum West Africa 2 (n = 223) was 40%, consistent with previously reported results [19]. From these 552 patients a total of 1333 positive cultures (M. tuberculosis n = 823, M. africanum n = 510) were obtained and analysed in the study (Figure 1). In both liquid culture systems, the median time to culture positivity was significantly shorter for M. tuberculosis (Bactec 9000: 13d, MGIT960: 9d) relative to M. africanum (Bactec 9000: 21d, MGIT960: 15d) (see Table 1 and Fig. 2). Growth attenuating genes. Based on transposon insertion and gene-disruption, a set of 42 growth-attenuating genes was previously identified [12]. Each of these genes resulted in growthattenuation when inactivated through transposon insertion. In our study, 12 of the 42 growth-attenuating genes had non-synonymous mutations, and 4 of these genes were affected in their protein function as predicted by SIFT analysis. For a detailed list of the genes and their mutations see table 2.
Essential genes. M. tuberculosis genes were previously classified into two groups: essential or non-essential for in vitro growth [12].We selected the 614 essential genes described for M. tuberculosis H37Rv and genetically compared them with their respective homologues in M. africanum. We found that 132 (21%) of 614 essential M. africanum genes contained non-synonymous mutation(s) which resulted in an amino acid change within the respective proteins. We considered the 21% percentage difference to be the baseline difference between the genomes of M. tuberculosis H37Rv and M. africanum (see Table 3).
Transporter genes. We identified a total of 132 genes that were common to both M. tuberculosis H37Rv and M. africanum and that were described or annotated as genes encoding structural components of membrane transporters from the literature and publicly available databases. The 132 genes included transporters specific for nutrients as well as for the transport of macromolecules. We found that M. africanum genes encoding sulphur transporters, ion transporters and fatty acids transporters were significantly more likely to carry a nonsynonymous mutation than essential M. africanum genes (see Table 3 and Figure 4). No such effect was observed in genes  encoding carbohydrate, amino acid or peptide transporters. Interestingly, we found transporter systems specific for nitrogen were hyper-conserved and were less likely to be mutated when compared with essential genes. Genes essential for intracellular survival. We studied a group of 7 operons that were previously described as crucial for the intracellular survival of M. tuberculosis H37Rv in macrophages [13]. We found that five out of seven operons had at least one mutated gene in M. africanum (see Fig. 5). Although the pstA1 gene of the phosphate transport operon was mutated in M. africanum GM041182 and ATCC35711, this was not a common trait of all analysed M. africanum strains.

Discussion
M. africanum grows slower than M. tuberculosis, with delayed culture positivity (by 4-6 days) when grown from sputum in modern liquid culture systems. Although these liquid culture systems are only indirectly measuring growth by detecting oxygen or radioactive precursor consumption as a proxy for growth, they are well suited to compare the growth behaviour of different bacterial isolates. The observed growth differences between the two lineages were further emphasized by a survival analysis which was adjusted for smear grade, and we estimated a Hazard ratio (HR) = 0.40 (95%CI 0.35-0.47, p,0.0005). Consistent with this diagnostic observation, we determined a longer doubling time of M. africanum in growth experiments in which the inoculum was carefully standardized by CFU. One limitation of this approach is that it is not clear whether cording differs between M. africanum and M. tuberculosis, which could potentially impact on CFU standardization. Also testing a wider range of isolates together with comparative genomics could enhance the power of the in vitro experiments in the future. However, both our findings on growth from sputum and standardized inoculum are consistent with other studies, as the measured doubling times for both of the lineages were in the same range as previously described by Bold et al. [20]. Our results also reproduced the initial observations from Senegal in 1968, when isolates, identified as M. africanum by biochemical methods, yielded growth on solid media later than M. tuberculosis isolates [1]. Therefore, Castets in 1979 recommended incubation for 90 days for the detection of M. africanum on solid media [21]. However, with the advent of modern automated liquid culture systems, the original recommendations need to be adjusted and redefined. For the current study, we incubated all samples for 42 days, as recommended by the manufacturer of the liquid culture systems. Although 42 days are sufficient to detect M. tuberculosis strains, the frequency distributions indicated that for M. africanum strains a longer incubation period might be advisable. For instance, in the Bactec 9000, only seven cultures turned positive on the very last day of the incubation period. For this reason, the Bactec MGIT960 seems preferable and better suited for the cultivation of M. africanum, as overall incubation times decreased and detection of culture positivity can be achieved faster. Therefore to evaluate the potential of these liquid cultures systems as diagnostic tools for M. africanum detection, further long-term growth studies need to determine the maximum M. africanumspecific incubation times. Such studies could also evaluate whether the currently applied incubation protocols resulted in an underestimation of M. africanum prevalence or failed to detect mixed infections.
The selective advantage of the growth delay of M. africanum is not clear. However comparison of the genome sequences of M. tuberculosis H37Rv and M. africanum give some clues to the underlying mechanisms. First, we investigated a group of genes, each of which has already been described to result in in vitro growth-attenuation of M. tuberculosis H37Rv upon transposon (TraSH) inactivation [12]. Of 42 growth-attenuating genes, 12 genes contained non-synonymous mutations or were pseudogenes due to frameshift mutations in M. africanum. In particular, 4 gene products (Rv2112c/MAF_21240, Rv0862c/MAF_08710, AceE, RecA) were predicted by SIFT analysis to be affected in their protein function. These four proteins are the most likely candidates responsible for the observed growth attenuation in M. africanum. A fifth identified gene, glpK, is a pseudogene in GM041182 or with non-synonymous mutations in the other 3 strains, yet SIFT analysis identified the amino acid substitution to be tolerated by the bacteria. Therefore the Glpk protein is most likely functional in 3 out of 4 M. africanum strains and is not a common cause for the observed, slower growth.
We further hypothesized that a reduction and/or deficiency of molecular membrane transporters could limit growth of M. africanum. For instance, the knock-out of outer membrane Msp porins and a subsequent reduced sugar and phosphate uptake led to a slower growth rate of M. smegmatis [22]. Similarly it was previously suggested that the slower growth of M. tuberculosis, when compared to the fast-growing Mycobacterium smegmatis, could be due to the loss of several sugar transporters [17]. Therefore we aimed to determine the status of known transport mechanisms in the sequenced M. africanum genomes and M. tuberculosis H37Rv.
We identified 132 membrane transporter genes common to the two mycobacterial lineages. In M. africanum, there were signifi- cantly more genes encoding sulphur/sulphate-, lipid/fatty-acid, and ion-transporter with non-synonymous or frameshift mutation than in essential genes. Although sulphur is an essential nutrient for mycobacterial survival and virulence (for review see [16,23]), we identified (protein function affecting, SIFT) mutations in the cysTWA/subI ABC-transporter of M. africanum. Since subI-knock out mutants of M. bovis were restricted in their sulphate uptake [24], it is conceivable that M. africanum strains are similarly impaired in their import of sulphur. Although there was speculation that Rv1739c, another predicted sulphate transporter [25] could compensate for the loss of the cysTWA/subI transport system [26], it is unlikely because this protein is likewise potentially inhibited in its protein function due to a SNP mutation. Whether the hypothetical sulphate-transporter Rv1707, which carries a tolerated amino acid change in M. africanum, is a functional sulphate transport mechanism still has to be experimentally confirmed.
A second group of highly mutated M. africanum genes encode for ion transport mechanisms. Unfortunately, the knowledge about this important group of proteins is still scarce [14]. However, KefB, a potassium/proton antiporter that controls the early acidification of the phagosome, was mutated and impaired in its protein function (SIFT) in M. africanum [27]. Also, knock-out mutants of the Mg 2+ -transporter MgtC, which is potentially affected in its protein function (SIFT) in M. africanum, had impaired growth under certain in vitro conditions [28]. Another group of heavy-metal ion transporter genes in M. tuberculosis, ctpA-ctpV, are very different in M. africanum. For instance ctpV, one of the best studied members of this family, yet with a tolerated (SIFT) mutation in M. africanum, is key for mycobacterial copper homeostasis and virulence [29]. Similarly, the iron-specific ABCtransporter IrtAB, in which the IrtA subunit has an intolerable amino acid substitution (SIFT) in M. africanum, is not only crucial for survival in iron-deficient conditions, but is also required to effectively establish infection in the experimental murine host [30]. Most importantly, all the above mentioned ion transporter genes have one thing in common: they were found to play key roles in the intracellular survival of the bacteria within the phagolysosome of macrophages [27][28][29][30]. It is surprising that genes important for this crucial step of mycobacterial pathogenesis were among the least conserved in M. africanum, which could indicate that M. africanum might pursue a different intracellular survival strategy than M. tuberculosis to cope with the harsh environment within a phagolysosome.
Therefore we investigated 7 putative operons that were previously described as essential for the intracellular survival of M. tuberculosis H37Rv in macrophages [13]. Components of the sugar transport system (sugA/B/C/lpqY) were hyperconserved among the M. africanum isolates. Similarly, with the exception of pstA1, all M. africanum genes encoding phosphorous transporters were conserved amongst the sequenced isolates. Our results suggest that both pathways are equally important for M. tuberculosis and M. africanum. However, we found a remarkable difference between the two lineages. In the course of host macrophage infection, lipids become increasingly more important and replace carbohydrates as the major carbon source [17], and 3 operons of lipid metabolism (Rv3540c-Rv3545c, Rv3550-Rv3552, Rv3569c-Rv3570c) were described to be required for mycobacterial survival [13]. Interestingly, all of these operons have at least one mutated gene in M. africanum. Consistently, 8/14 mmpL genes, that likely transport lipids/fatty acids across the membrane, have altered amino acid sequences, including a frameshift mutation in mmpL13a that results in a pseudogene. Of note, mmpL3, which is one of 5 conserved mmpL genes in the obligate intracellular M. leprae, was Table 3. Comparison of certain gene groups in M. africanum compared to essential genes mapped against M. tuberculosis H37Rv. shown to be the only gene of this family to be essential for viability of M. tuberculosis [31]. However, mmpl3 is potentially affected in its protein function (SIFT analysis) in M. africanum. Other genes, essential for survival in macrophages, belong to a putative operon spanning from Rv3864 to Rv3878 [13], a genetic region that partially includes the RD1 locus encoding the ESX-1 secretion system and its virulence genes such as esxA (encoding ESAT-6) and esxB (encoding CFP-10). One of the genes, Rv3864 (espE), had a non-synonymous SNP in one M. africanum isolate and a frameshift mutation in the remaining three sequenced M. africanum strains. This is interesting as Rv3864 was associated with virulence, yet it is assumed that a loss of the gene can be compensated by its homologue Rv3616 (espA) [32]. However, as the Rv3616 (espA) homologue has a non-synonymous mutation in M. africanum as well (data not shown), it is possible that none of these genes is functional in M. africanum. This is supported by the previous finding that certain M. africanum isolates were less likely to induce an ESAT-6 dependent IFN-c host response and it was speculated that this was due to an ESAT-6 secretion impairment [6]. Combining the finding that the Rv3864/espE homologue in M. marinum is required for secretion of CFP-10 [32], and CFP-10 contains the secretion signal of the ESAT-6/CFP-10 dimer [33], an inactive Rv3864/ espE could therefore be the missing genetic link to explain the reduced ESAT-6 secretion of M. africanum. Finally, the seventh operon under study, Rv0169-Rv0178, is essential for entry into the mammalian cell and intracellular survival, yet several members are highly mutated in M. africanum. Interestingly, the overall regulator of this operon, Rv0165 (mce1R), has a frameshift mutation in M. africanum (data not shown) and recent studies suggest that Mce1R is part of a global genome-wide regulatory network which control cell growth [34].
In the present study we found that M. africanum strains are impaired in their capacity to grow. We identified several potential gene candidates and functional protein groups that might contribute to the observed growth defect. To unambiguously confirm causality, complementation experiments in which the M. . Schematic overview of molecular transport mechanisms present in M. africanum. Genetic information of 132 genes encoding various families of membrane transporters was compared between the sequenced M. africanum strains and M. tuberculosis H37Rv. The analysis comprised both, transport mechanisms specific to nutrients and macromolecules (upper and lower figure). Genes that are identical to the wildtype M. tuberculosis H37Rv gene homologue are displayed with a solid border line and white background. Genes, with non-synonymous mutation are displayed in blue with dashed lines. Genes with a frameshift mutation, are displayed with striped background. doi:10.1371/journal.pntd.0002220.g004 Figure 5. Status of putative M. africanum operons essential for intracellular survival. Seven operons were previously defined as essential for growth within the macrophage [13]. Genes that are identical to the wildtype M. tuberculosis H37Rv gene homologue are displayed with a solid border line and white background. Genes, with a non-synonymous mutation are displayed in blue with dashed lines. Genes with a frameshift mutation, are displayed with striped background. doi:10.1371/journal.pntd.0002220.g005 africanum mutant genes are replaced by wildtype H37Rv genes will have to be conducted. However, our genomic analysis suggests that the underlying genetic reason for the growth defect is rather complex. The growth attenuation might be a redundant result due to the loss of multiple genes. Moreover, large scale genomic analyses on additional M, africanum genomes will have to be conducted to confirm which of the described SNPs are really specific to all members of the M. africanum West Africa 2 lineage.
Taken together, from a genetic and phenotypic point of view M. africanum appears to be distinct from M. tuberculosis. M. africanum may have a modified, yet unknown, survival strategy of the bacterium within the human host. Future research on the lifestyle of M. africanum may lead to an improved understanding of growth promoting factors in M. tuberculosis and may ultimately reveal new strategies to interrupt bacterial growth and replication within the human host.