Previously, we used transcriptional mapping to assign rv1708, originally annotated as a putative initiation inhibition protein, as a septum formation regulatory protein [4]. To improve the functional assignment of rv1708, a bioinformatics approach using a consensus sequence derived from alignments of proteins with an annotated function of septum inhibition was employed. This substantiated the functional assignment of rv1708 as encoding a septum regulatory protein because of shared similarity with the septum regulatory protein MinD. Septum inhibiting proteins, including MinD and Soj proteins, belong to the same family of P-Loop ATPases [31]. To narrow the functional assignment of Rv1708 we used the Phyre2 protein homology engine, which successfully revealed that Rv1708 shares greater sequence identity to Soj proteins (46% identity with 78% coverage; E value 2e-82) than to true MinD proteins (27% identity with 80% coverage; E value 6e-18). Alignments of the Rv1708 protein with Soj and MinD consensus models, Orthologous Matrix (OMA) groups 95838 and 78690 respectively, revealed that Rv1708 shares Soj-specific DNA-binding residues while lacking the hydrophobic C-terminal aliphatic helical extension that is involved in membrane association and is characteristic of MinD proteins [31,32] (Figure 1). Together, these analyses improve the functional annotation and indicate that rv1708 encodes an alternative Soj protein, referred to here as Soj_Mtb.

To assess whether Soj_Mtb is involved in regulation of cell cycle progression, we performed gene dosage experiments. Overexpression of soj _Mtb caused a decrease in growth in Mycobacterium smegmatis and Mtb compared to the vector control strain (Figure 2). To visualize alterations in cellular morphology due to overexpression of soj _Mtb , mycobacterial cells were imaged using electron microscopy (Figure 3). These studies revealed that overexpression of soj _Mtb in M. smegmatis and Mtb produced similar elongated cell phenotypes with a predominant ultra-structural morphology absent of concentric rings indicative of septa. This elongated cell phenotype is consistent with previous studies performed in M. smegmatis[33]. The absence of septa is characteristic of regulation of FtsZ polymerization prior to Z-ring formation [4,5,16,34,35]. Together, these data indicate that Soj_Mtb reduces the frequency of septum formation and governs cell cycle progression prior to septum formation.

To assess the involvement of Soj_Mtb in eliciting alternative adaptive responses, the global transcriptional response of Mtb upon overexpression of soj _Mtb was investigated through whole genome DNA microarray analysis. A total of 1,368 genes from all functional categories displayed a 1.5 fold or greater change in expression (p values < 0.05) in the Mtb merodiploid strain compared to the vector alone control strain (Figure 4). The microarray data analysis revealed that Soj_Mtb is part of a large alternative stress response based on the induction of alternative sigma factors sigH, sigG, sigL, sigM, sigI in response to overexpression of soj _Mtb . The pleotropic nature of this transcriptional response is likely the result of the induction of these alternative sigma factors, several of which have been shown to regulate stress response pathways [36]. Notably, rv1991c encoding the MazF6 toxin of the MazEF6 TA loci showed significant change in expression.

The MazF6 toxin is the most well characterized out of nine predicted maz-family TA loci encoded in the Mtb genome [20]. Molecular studies have demonstrated that the MazF6 is an endoribonuclease that cleaves ACA mRNA sequences, and that this activity is inhibited by interaction with the MazE6 antitoxin [37]. MazF6 is thought to induce rapid changes in the transcriptome due to its endoribonuclease activity, allowing coupling of cell cycle inhibition by Soj_Mtb and the transcriptional remodeling associated with the transition into an alternative state such as NRP.

We compared the soj _Mtb response with previously described programs associated with NRP in order to determine if there was overlap in the transcriptional responses. Overlap was determined by taking into account the number of genes in the given transcriptional response in relation to the number of genes in the genome (4,124 as indicated from the Broad Institute) to determine the number of genes that would be shared only by chance, as described previously [38]. There was no significant degree of overlap between the up-regulated genes of the soj _Mtb response and the dos-response or EHR (Table 1), and in fact, the dos-response was actively down-regulated (Table 2). However, there was significant overlap between the soj _Mtb response and the “non-culturable” response, with 14 up-regulated genes in common, the mazEF6 transcript among them.

To better understand the functional role of Maz-family TA pairs, we sought to characterize the functionality and interaction of MazE6 and MazF6. When expressed in vivo, the MazF6 toxin induces bacteriostasis, while expression of MazE6 causes no stalling of bacterial growth (Figures 4B and 5A, respectively). Importantly, this bacteriostatic phenotype can be rescued when the toxin and antitoxin are co-expressed thus demonstrating a functional interaction in vivo (Figure 5C). To show a physical interaction between these two recombinant proteins, co-purification was performed and protein identity was confirmed via Western blot and mass spectrometry. When both antitoxin and toxin are produced individually and subject to metal affinity chromatography, MazE6-HIS is present in the elution fraction, while MazF6-HSV is present in the unbound flow and wash fraction, indicating that it is not retained in the affinity column (Figure 5D). In contrast, when co-expressed, MazF6-HSV elutes primarily with MazE6-HIS indicating that they physically interact in vivo and that the bacteriostatic phenotype associated with overproduction of MazF6 can be prevented by its interaction with its cognate antitoxin, MazE6 (Figure 5). Western blot analysis was confirmed by in-gel trypsin digestion of the gel fragment corresponding to the recombinant MazEF6 complex and identification by LTQ mass spectrometry. MazEF6 were among the top hits returned when searched against the NCBInr database for eubacteria. MazE6 was identified with 70% coverage with 12 non-duplicate peptides, and MazF6 was identified with 95% coverage and 8 non-duplicate peptides.

To determine if there is any degree of functional redundancy among the numerous MazEF loci in Mtb, we assessed the interactions between MazF6 and non-cognate MazE antitoxins. Functional redundancy was assessed by the ability of non-cognate MazE antitoxins to prevent bacteriostasis induced by MazF6. Growth arrest due to MazF6-HSV production was not rescued by co-production of non-cognate MazE-HIS antitoxins (Table 3). The MazF8 antitoxin encoded by rv2274a was not tested because it is absent in some clinical strains of Mtb[20,39]. Co-purification was also performed and with all pairs tested, MazF6-HSV was only able to physically interact with its cognate antitoxin, MazE6 (Figures 5D and 6). Our data confirm that the MazF6 toxin does not interact with non-cognate MazE antitoxins in vivo indicating that there may be little functional redundancy across MazEF pairs.

Adaptations resulting in a NRP state have been associated with the execution of complex and complementary molecular programs designed to allow the cell to survive the stresses experienced within the host [40]. These include programs to slow bacterial growth, induce alternative metabolic pathways, and to withstand acid, hypoxic, nutrient and additional stresses. TA loci have been associated with stress responses in bacteria and are becoming the focus of increasing efforts to elucidate their contributions to alternative states in vivo[18].

In order to determine if Soj_Mtb and MazEF6 play a role during infection, we assessed their transcriptional activity during the infectious processes using the established mouse model of infection [41]. Quantitative-PCR was used to assess soj _Mtb and mazEF6 abundance in cDNA obtained from in vitro Mid-Log phase cultures, and the lungs and spleen at 20, 40 and 100 days post-infection (Figure 7AB). The analyses revealed an overall difference in expression patterns and overall abundance of soj _Mtb and the mazEF6 loci as compared to in vitro grown bacteria. soj _Mtb was abundant at day 20, 40, and 100. The transcriptional response in the lungs was characterized by a significant difference in the transcriptional abundance of mazE6 and mazF6 at day 20, while this difference in the abundance of mazE6 and mazF6 was not noted at days 40 and 100 (Figure 7A). Similarly to the transcriptional pattern observed at day 20 in the lungs, in the spleen soj _Mtb was abundant at day 20, 40 and 100, and there was a two log or greater difference in the abundance between mazE6 and mazF6 throughout the course of infection (Figure 7B). The observed transcriptional abundances observed between mazE6 and mazF6 is consistent with observations of TA loci in other systems [42]. Specifically, only the toxin component of several TA loci were identified in Mtb in a human macrophage model of late stage infection [43]. The transcriptional activity of soj _Mtb , and mazE6 and mazF6 in bacteria obtained from infected tissues indicate that MazF6 is important throughout the course of infection, and is uniquely temporally functional in facilitating adaptations to the alternative environments encountered during infection.

To identify putative cell division regulators in Mtb, consensus models were derived from septum formation inhibitors from OMA groups 78690, 63437, 84083, 90270, and 245137–40 from a variety of bacterial species using a MAFFT local multialignment tool. Resulting consensus sequences were used to BLAST the Mtb H37Rv genome to identity mycobacterial ORFs that encode septum regulatory proteins. BLAST searches were optimized for percent identity and score. Top hits were analyzed by the Phyre2 server to confirm or refute identification as putative septum formation inhibitors.

For all mycobacteria experiments, Mtb strain H37Rv (ATCC 27294) and M. smegmatis mc²155 was cultured at 37°C with shaking in Middlebrook 7H9 liquid medium containing 0.4% glycerol, 10% OADC, and 0.05% Tween80 or on Middlebrook 7H10 agar containing 0.4% glycerol and OADC containing appropriate antibiotics as noted. For generation of recombinant mycobacterial strain overexpressing soj _Mtb , the rv1708 gene was PCR amplified from Mtb H37Rv genomic DNA using AccuPrimePfx DNA polymerase (Invitrogen) and cloned into the NdeI and HindIII of the pVV16 constitutive, extrachromosomal mycobacterial vector, which is a derivative of pMV261 [50,51] (Additional file 1). DNA constructs were transformed into One Shot® TOP10 chemically competent E. coli (Invitrogen) and selected by growth on LB agar containing 50 μg ml^-1 kanamycin. Sequence confirmed plasmid DNA was transformed into electrocompetent M. smegmatis or Mtb and selection was carried out using kanamycin at 25 μg ml^-1.

Recombinant M. smegmatis strains containing the bacterial shuttle vector pVV16 as a control and soj _Mtb ::pVV16 were grown to an O.D._600nm of 1.0 and collected by centrifugation. Cells were washed three times in PBS, pH 7.4, and fixed with 2.5% glutaraldehyde in Buffer A (0.1 M potassium phosphate (pH 7.4), 1 mM CaCl₂ and 1 mM MgCl₂) at 4°C for 48 hrs. The fixed bacterial cells were collected by centrifugation, washed three times in Buffer A and treated with 1% OsO₄ in Buffer A for 30 minutes at 4°C. Again, cells were washed three times with Buffer A_. and prepared for SEM with a graded series of ethanol treatments (20-100%). Ultrastructural examination was performed using a JOEL JEM -100CX electron microscope. Recombinant Mtb strains containing the bacterial shuttle vector pVV16 as a control and pvv16::soj _Mtb were prepared as described above. However, the pvv16::soj _Mtb culture did not reach an O.D._600nm of 1.0 and was prepared once the growth had plateaued.

At Mid-Log growth, Mtb cells were harvested by centrifugation and resuspended in TRIzol (Invitrogen). Cells were lysed by bead beating for one-minute increments for three minutes total, cooling the sample on ice between each step. RNA was separated from other cellular products by the addition of chloroform and centrifugation, according to the TRIzol protocol. The upper aqueous layer was combined with an equal volume of 70% ethanol, and the entire volume was applied to an RNeasy Mini kit (Qiagen) column for RNA cleanup.

The Mtb whole genome microarrays were obtained through the TB Research Materials and Vaccine Testing Contract at Colorado State University. Slides were post-processed using succinic anhydride as described previously [4]. Approximately 8 μg of total RNA was converted to cDNA in the presence of either Cy5- or Cy3-labeled nucleotides as previously described [4,6]. Hybridization was performed at 42°C for 12 hr. Slides were scanned using an Axon Genepix scanner. The final microarray data set (Additional file 2) resulted from combining independent biological replicates. Data reduction and global normalization was performed on the raw fluorescent intensities. The normalized intensity values of treated and control cultures were used to generate ratio and log₂ expression values for each gene.

For the Maz6 TA loci interaction studies, mazE and mazF gene fragments were PCR amplified from Mtb H37Rv genomic DNA using Phusion® High-Fidelity PCR Master Mix with GC Buffer (New England BioLabs Inc.). Generally, mazE antitoxins were cloned in the BamHI and XhoI restriction sites of pET28a (Novagen), and mazF6 toxin was cloned into the SphI and NotI restriction sites of pETcoco2 (Novagen) (Additional file 1). DNA constructs were transformed into One Shot® TOP10 chemically competent E. coli (Invitrogen) or Z-Competent E. Coli - Strain DH5α (Zymo Research) and transformants selected by growth on LB agar containing 50 μg ml^-1 kanamycin for pET28a and 50 μg ml^-1 ampicillin for pETcoco2. Sequence confirmed plasmid DNA was transformed into One Shot® BL21(DE3) pLysE Chemically Competent E. coli (Invitrogen) and selection was carried out overnight by growth in LB broth supplemented with 0.2% glucose, 34 μg ml^-1 chloramphenicol, and containing 50 μg ml^-1 kanamycin for pET28a selection or 50 μg ml^-1 ampicillin for pETcoco2 selection, or both in cases of co-transformation. Glucose was added to the media to maintain low basal expression from the pET vectors as previously described [52]. The overnight outgrowths were used for recombinant protein induction. Briefly, cultures were diluted back 1:50 in fresh media containing antibiotics without glucose. When mazF6::pETcoco2 construct was used, diluted cultures were grown for one hour with shaking at 37°C and L-arabinose was added to a final concentration of 0.01% to amplify plasmid copy number prior to protein induction. Once cultures reached Mid-Log, protein production was induced using 1 mM final concentration of isopropyl-beta-D-thiogalactopyranoside (IPTG)(Fermentas). Cultures were incubated for four hours with shaking at 37°C, with timepoints being taken every hour to monitor growth by O.D._600nm. After four hours, bacterial cell pellets were collected by centrifugation and stored at −80°C.

Purification of MazE-HIS in complex with MazF6-HSV was carried out under native conditions to maintain protein-protein interactions. Cleared lysates of bacterial cell pellets were obtained using BugBuster® including Benzonase® (Novagen) according to manufacturer’s instructions. Generally, cleared lysate was combined with one-fourth the volume of Ni-NTA His-Bind® Resin and mixed gently at 4°C for 1 hour before packing into a column. The column was washed with wash buffer 1 (500 mM NaCl, 20 mM Tris–HCl, 60 mM imidazole, pH 9.0), followed by wash buffer 2 (250 mM NaCl, 20 mM Tris–HCl, 100 mM imidazole, pH 9.0). Recombinant protein or protein complexes were eluted with elution buffer (250 mM NaCl, 20 mM Tris–HCl, 300 mM imidazole, pH 9.0). All fractions were collected and samples were resolved using pre-cast 12% NuPAGE® gels (Invitrogen), followed by transfer to 0.2 micron nitrocellulose membrane (BioRad) for Western blotting. Membranes were blocked in 4% BSA in TBST, incubated with primary Penta-His antibody (Qiagen) or anti-HSV antibody (Novagen), both diluted 1:10,000, followed by goat anti-mouse-alkaline phosphatase (Sigma) secondary antibody diluted 1:10,000. Membranes were developed using SIGMAFAST™ BCIP®/NBT tablet (Sigma) according to manufacturer’s instructions.

All use of vertebrate animals took place at Colorado State University, which is AAALAC approved and has an OLAW number of A3572-01. Care was provided by veterinarians in the Laboratory Animal Resources of Colorado State University under the supervision of the University Veterinarian. Six week-old female C57BL/6 mice (Jackson Laboratories) were infected via low dose aerosol with Mtb H37Rv in a Glas-Col Inhalation Exposure System (Glas-Col, Inc.). Exposure was conducted by aerosolizing approximately 2×10⁶ cfuml^-1 in a volume of 5 cubic feet over a period of 30 minutes, followed by a 20 minute period of cloud decay. Infected mice were housed in sterile micro-isolator cages in the ABSL-3 facility at Colorado State University and provided water and food ad libitum and monitored for morbidity. Five animals per time point were sacrificed at 20, 40, and 100 days post-infection by CO₂ inhalation and lungs and spleens were harvested. Organs were cut into halves, and half of each organ was homogenized in 2 mL sterile saline using an Omni Tissue homogenizer (Omni International). Bacterial burden in the lungs and spleen at each time point were assessed by plating 10-fold serial dilutions of organ homogenates on Middlebrook 7H11 agar containing 10% OADC and carbenicillin (Sigma) and cycloheximide (Sigma) at 50 μg ml^-1 and 10 μg ml^-1, respectively. Plates were incubated at 37°C for three weeks.

The remaining half of each lung and spleen were homogenized in 5 ml TRIzol by bead beating. Total RNA was isolated by organic phase separation. RNA samples were treated with DNAse (Fermentas) for 1 hour at 37°C, re-extracted in phenol/chloroform/isoamyl alcohol (25:24:1)(Sigma), and precipitated with ammonium acetate.

cDNA synthesis from total RNA from the lungs and spleen of infected mice was carried out using a gene-specific reverse primer cocktail with Transcriptor First Strand cDNA Synthesis Kit (Roche). The reverse primer cocktail contained reverse primers specific for mazE6, mazF6, and soj _Mtb in addition to random hexamers (Additional file 3). Reactions contained 1 μg total RNA, 2.5 μM gene-specific reverse primers, and 60 μM random hexamers. cDNA synthesis were carried out according to manufacturer instructions along with no reverse transcriptase control and no template control reactions. Quantative-PCR was carried out using 50 ng of cDNA generated from RNA isolated from Mid-Log phase Mtb, and RNA isolated from lungs and spleens of mice at Day 20, 40,and 100 post-infection. Calibration curves were generated using respective mazE6, mazF6, and soj _Mtb specific dsDNA amplicons generated from PCR using GoTaq®Green Master Mix (Promega). Calibration curves of respective dsDNA amplicons ranged from 1 ng to 1^-8 ng per reaction in 10-fold serial dilutions. Modeled linear fits and primer efficiencies were used to determine concentration of respective gene. Amplicons in experimental samples were converted to total mRNA copy number per mL. Each sample was normalized by taking total mRNA copy number per mL and dividing by the associated cfu mL^-1 recovered from each biological sample. No reverse transcriptase and no template controls confirmed no gDNA contamination or non-specific amplification.