M. smegmatis mc²155 and derivative strains were grown at 37°C in Middlebrook 7H9 (Difco) liquid culture medium supplemented with 10% albumin-dextrose-saline (ADS), 0.5% glycerol and 0.1% Tween-80 or on Luria Broth (LB) agar solid culture medium. Frozen stocks were prepared by growing cultures to mid-exponential phase (OD₆₀₀ of 0.6–0.8), adding glycerol to 15% final concentration, and storing aliquots at -80°C. Antibiotics were used at the following concentrations: kanamycin (Kan) 15 μg/ml, hygromycin (Hyg) 50 μg/ml, rifampicin (RIF) 200 μg/ml, isoniazid (INH) 50 μg/ml, ethionamide (ETH) 200 μg/ml, ethambutol (ETB) 5 μg/ml.

All plasmids used for cloning and strain construction are listed in Supplementary Table S1. Constructs for deletion of phoU1 (Msmeg_5776) or phoU2 (Msmeg_1605) in M. smegmatis were generated in the allelic exchange vector pJG1100 (Kirksey et al., 2011). Genomic regions 500–800 bp upstream and downstream of the genes to be deleted were PCR-amplified from M. smegmatis mc²155 genomic DNA using the oligonucleotides listed in Supplementary Table S2. Reverse primers for amplification of the upstream regions were designed with an AvrII restriction site in-frame with the translation start codon; corresponding forward primers for amplification of the downstream regions were designed with an AvrII restriction site in-frame with the stop codon. PCR products were cloned in pCR2.1-TOPO (Invitrogen) and sequenced. Upstream and downstream regions were removed from pCR2.1 by restriction with PacI/AvrII and AvrII/AscI, respectively and ligated together in pJG1100 between the PacI and AscI sites to generate the in-frame deletion constructs pBE101 (ΔphoU1), pBE102 (ΔphoU2).

Vectors for complementation of the phoU deletions were constructed in the episomal plasmid pMV261 under the control of the vector-encoded strong constitutive hsp60 promoter. The M. smegmatis phoU1 and phoU2 genes were PCR-amplified using the primers indicated in Supplementary Table S2, cloned in pCR2.1-TOPO and sequenced. The cloned genes were removed from pCR2.1 by restriction with EcoRI and HindIII and ligated in similarly digested pMV261 to generate pMVphoU1 and pMVphoU2.

For construction of Tet-inducible phoU1, the phoU1 gene was PCR-amplified using the primers indicated in Supplementary Table S2, cloned in pCR2.1-TOPO and sequenced. phoU1 was removed from pCR2.1 by restriction with HindIII and EcoRI and ligated in similarly digested pTIC10a or pJT6a to generate pTICphoU1 and pJTphoU1.

M. smegmatis ΔphoU1 and ΔphoU2 deletion mutants were generated by a two-step homologous recombination method of allelic exchange, as previously described (Upton and McKinney, 2007) except that LB agar medium containing Kan and Hyg and LB agar medium containing 5% sucrose were used for selection of transformants and counter-selection of the pJG1100 vector, respectively. Integration of the vectors was confirmed with the following primer pairs, listed in Supplementary Table S2: ΔphoU1 upstream 5776F3/5776R4; ΔphoU1 downstream 5776F4/5776R3; ΔphoU2 upstream 1605F3/1605R4; ΔphoU2 downstream 1605F4/1605R3. Identification of deletion mutants was done with the following primer pairs: ΔphoU1 5776F3/5776R3; ΔphoU2 1605F3/1605R3. Complemented strains were constructed by electroporating the corresponding deletion mutant with the complementation plasmid pMVphoU1 or pMVphoU2, and selecting on LB containing Kan.

For construction of the ΔphoU1ΔphoU2 double deletion mutant, the ΔphoU2 mutant was electroporated with pTICphoU1 and transformants were selected on LB containing Kan. The resulting strain was then electroporated with the pBE101 (ΔphoU1) allelic exchange vector and transformants were selected on LB containing Kan and Hyg. Integration of the ΔphoU1 vector in the transformants was confirmed by PCR as described above. The resulting strain was then grown overnight in 7H9 medium containing 50 ng/ml anhydrotetracycline (ATc) to induce expression from the pTICphoU1 construct and subsequently plated on LB containing 5% sucrose and 100 ng/ml ATc to counterselect the pBE101 allelic exchange vector. Individual Hyg^S colonies were tested for the ΔphoU1 deletion by PCR using the 5776F3/5776R3 primer pair. The resulting ΔphoU1ΔphoU2pTICphoU1 strain was then electroporated with pJT6a and transformants were selected on LB containing Hyg. Individual Hyg^R transformants were patched to LB Kan and LB Hyg. Those that were Hyg^R and Kan^S were selected for PCR analysis with primers pTfor and pTIC6a_R that amplify a 1.3 kbp fragment from pTICphoU1 and a 500 bp fragment from pJT6a. Isolates from which only the 500 bp product was amplified were tested for the deletions of phoU1 and phoU2 by PCR as described above.

Cultures of M. smegmatis grown in complete 7H9 medium to mid-logarithmic phase were diluted to OD₆₀₀ 0.01 in 7H9 and 200 μl were inoculated in a 96 well plate (polystyrene rounded square well, Fisherbrand). Cultures were incubated at 37°C with shaking and OD₆₀₀ was monitored hourly for 36 h using a Synergy H1 hybrid reader (BioTek).

Cultures of M. smegmatis were grown at 37°C with shaking in either 14 ml polystyrene snap-cap tubes (Falcon) in 5 ml complete 7H9 medium to assess pellicle formation or a polystyrene flat-bottom 12-well plate (Falcon) in 4 ml complete 7H9 medium to test clumping. After 72 h of incubation, the tubes and plate were imaged.

Isolated single colonies of M. smegmatis were grown on LB agar for 4 days at 37°C. Plates were imaged with a FluorChem FC3 (Cell Biosciences) using the white light setting. Colony area for 50 representative colonies was measured on each of three plates grown from independent biological replicates using ImageJ.

Bacteria were grown to mid-exponential phase (OD₆₀₀ 0.5) in 7H9 broth and RNA was extracted as described (Tischler et al., 2013). Equivalent amounts of total RNA were treated with Turbo DNase (Ambion) and reverse transcribed to cDNA with the Transcriptor First Strand cDNA Synthesis Kit (Roche) using random hexamers for priming and the following cycling conditions: 10 min at 25°C for annealing, 60 min at 50°C for extension, 5 min at 85°C for heat inactivation. cDNA was stored at -20°C until real-time PCR reactions were performed. Quantitative Real Time PCR reactions were prepared with 2× FastStart Sybr Green Master Mix (Roche), 2 μl cDNA, and 0.3 μM primers and were run in absolute quantification mode on a LightCycler 480 (Roche). PCR cycling conditions were: 95°C 10 min; 45 cycles of 95°C for 10 s, 60°C for 20 s, 72°C for 20 s with data collection once per cycle during the extension phase; one cycle of 95°C for 5 s, 60°C for 1 min, 97°C for 15 s with a 0.11°C/s ramp rate during the final denaturation to generate melting curves for confirmation of product specificity. Mock reactions (no RT) were performed on each sample to confirm the absence of genomic DNA contamination. Cp values were converted to copy numbers using standard curves for each gene. Target cDNA was internally normalized to sigA cDNA. Primers for real-time quantitative reverse transcriptase (RT) PCR used in this study are listed in Supplementary Table S3. Primers were designed using the Roche online Universal ProbeLibrary assay design center tool and were tested in standard PCR reactions using 100 M. smegmatis genome equivalents as template.

Cultures of M. smegmatis grown in complete 7H9 medium overnight to late-logarithmic phase were diluted to OD₆₀₀ 0.05 in 20 ml complete 7H9 and antibiotics were added. Cultures were incubated at 37°C with aeration and viable CFU were determined at 0, 8, 24, 48, and 72 h by plating serially diluted cultures on LB agar medium. CFU were enumerated after 3–4 days of incubation at 37°C.

Cultures of M. smegmatis grown in complete 7H9 medium to mid-logarithmic phase (OD₆₀₀ of 0.5) were diluted to OD₆₀₀ of 0.002 in fresh complete 7H9 medium. Antibiotics were added to 2 ml aliquots of the dilute culture in twofold decreasing concentrations with the highest concentration at 10x the MIC for wild-type M. smegmatis mc²155 (RIF 200 μg/ml, INH 50 μg/ml, EMB 5 μg/ml, ETH 200 μg/ml). Cultures were incubated for 48 h at 37°C with shaking and the OD₆₀₀ was determined. The MIC₉₀ was the concentration of antibiotic that inhibited 90% of the growth of a no drug control.

Cultures of M. smegmatis were grown in complete 7H9 medium to mid-logarithmic phase (OD₆₀₀ 0.4–0.8) and centrifuged at low speed (150 × g for 5 min) to remove large clumps. The OD₆₀₀ of the de-clumped culture was determined, and 0.5 ml of bacteria from this culture were pelleted by centrifugation (16000 × g for 10 min). The culture supernatant was removed and bacteria were resuspended in 0.1 ml 1M Tris buffer, pH 8.1 ml of 2 mM p-nitrophenyl phosphate was added and samples were incubated at 37°C in the dark for 10 min. Bacteria were removed by centrifugation and 1 ml of the supernatant was transferred to a cuvette. The OD₄₂₀ of the sample was measured using 2 mM p-nitrophenyl phosphate as a blank. Alkaline phosphate units were calculated as (1000 × OD₄₂₀)/(minutes of incubation × OD₆₀₀ × 0.5 ml).

Cultures of M. smegmatis were grown to mid-logarithmic phase (OD₆₀₀ 0.5–1.0) in complete 7H9. Cells from 20 ml of culture were pelleted (4700 × g for 15 min), then resuspended and lysed in 4 M guanidine isothiocyanate, 50 mM Tris-HCl (pH 7) at 95°C for 30 min. Total protein was quantified using 10 μl of the lysed sample (Pierce BCA Protein Concentration Assay, Thermo Scientific). PolyP was isolated from the remaining cell lysate using glassmilk (GeneClean) and quantified by binding to toluidine blue O (Sigma) dye solution (6 mg/L in 40 mM acetic acid) as previously described (Namugenyi et al., 2017).

M. smegmatis strains were grown to mid-logarithmic phase (OD₆₀₀ 0.5–0.8) and bacteria from 20 ml of culture were pelleted by centrifugation (2850 × g for 10 min). Genomic DNA was extracted by the cetyltrimethylammonium bromide (CTAB) – lysozyme method as described (Larsen et al., 2007) except that bacteria were lysed by incubation with lysozyme for 3 h at 37°C. DNA was resuspended in 100 μl TE buffer and stored at 4°C overnight. Concentration and purity of the genomic DNA were determined using a Nanodrop spectrophotometer (Thermo Scientific).

Genomic DNA extracted from M. smegmatis mc²155 and five ΔphoU1ΔphoU2 isolates was diluted to 12 ng/μl in 25 μl DEPC-treated water and submitted to the University of Minnesota Genomics Center. Whole-genome sequencing libraries were created using the TruSeq Nano Library Preparation Kit (Illumina). Libraries were processed using the 350 bp shearing protocol according to the manufacturer’s instructions. Libraries were multiplexed and sequenced on 0.25 lane of a HiSeq 2500 (Illumina) in high-output mode using the cBot HiSeq PE Cluster Kit v4 (Illumina) to generate clusters, and HiSeq SBS Kit, v4 chemistry (Illumina) with paired-end 125 bp reads. Image analysis and base calling were done using the Illumina software package RTA v 1.18.62. De-multiplexing and fastq file creation were performed using bcl2fastq v2.17.1.14. Total sequence yields (in Gb) were: mc²155 = 13.6; ΔphoU1ΔphoU2 #504 = 5.1; ΔphoU1ΔphoU2 #518 = 9.1; ΔphoU1ΔphoU2 #521 = 4.4; ΔphoU1ΔphoU2 #625 = 8.4; ΔphoU1ΔphoU2 #664 = 4.8. Sequences from mc²155 were aligned to the reference M. smegmatis mc²155 sequence in Geneious version 10.0.9 software¹ (Kearse et al., 2012) using the Geneious mapper to generate a Tischler lab WT mc²155 consensus sequence. Sequences from each of the five ΔphoU1ΔphoU2 mutants were then aligned to the WT mc²155 consensus using the Geneious mapper to identify variants. Single nucleotide polymorphisms (SNPs) and short duplications in Msmeg_1387, pstS, pstC, and pstB were confirmed by PCR amplification and sequencing using the primers listed in Supplementary Table S2.

Student’s unpaired t-test (two-tailed) was used for pairwise comparisons between WT and mutant strains of M. tuberculosis. P values were calculated using GraphPad Prism 5.0 software (GraphPad Software, Inc.). P values < 0.05 were considered significant.

M. smegmatis encodes two putative orthologs of E. coli PhoU. To determine which of these proteins negatively regulates the M. smegmatis SenX3-RegX3 system when P_i is abundant, we constructed in-frame unmarked deletions of phoU1 (Msmeg_5776) and phoU2 (Msmeg_1605) by two-step allelic exchange (Upton and McKinney, 2007). The ΔphoU1 and ΔphoU2 deletions were confirmed by PCR (data not shown). Growth of each deletion mutant was monitored in complete 7H9 medium, which is P_i-rich. The ΔphoU2 mutant exhibited no change in replication rate or overall growth yield compared to wild-type (WT) M. smegmatis mc²155 in this medium (Figures 1A,B). Although the ΔphoU1 mutant grew at a similar rate as the WT control, it did not reach the same maximal optical density (Figure 1A). In addition, the ΔphoU1 mutant lost viability in stationary phase, with significantly fewer CFU recovered from cultures grown for 36 h as compared to the WT control (Figure 1B). The ΔphoU1 mutant also formed significantly smaller colonies than WT M. smegmatis on LB agar plates (Table 1). Colonies of the ΔphoU1 mutant had a sticky phenotype. The ΔphoU1 mutant also exhibited clumping phenotypes in liquid culture; when grown in tubes it formed a pellicle at the air-liquid interface (Figure 2A) and when grown in 12-well plates it formed large visible clumps with fewer dispersed cells compared to the WT control (Figure 2B) even in complete 7H9 medium, which contains Tween-80. Clumping may contribute to the reduced growth yield of the ΔphoU1 mutant. The growth yield, stationary phase viability, colony size and clumping phenotypes of the ΔphoU1 mutant were all complemented by providing the phoU1 gene in trans on pMVphoU1 (Figures 1A,B, 2 and Table 1), confirming that the phoU1 deletion is responsible for these phenotypes. These phenotypes were also unique to the ΔphoU1 mutant, as the ΔphoU2 deletion did not alter colony size (Table 1) or clumping in culture (Figure 2).

To determine if either phoU single deletion influences activity of SenX3-RegX3, we performed quantitative RT-PCR experiments on RNA extracted from M. smegmatis strains grown in P_i-rich 7H9 medium. We chose 3 known RegX3-dependent genes for analysis: regX3, pstS, which encodes the substrate-binding protein of the Pst system, and phoA, which encodes alkaline phosphatase (Glover et al., 2007). As controls, we tested expression of the phoU1 and phoU2 transcripts. The phoU1 and phoU2 transcripts were undetectable in the corresponding deletion mutant strains, verifying that the genes were deleted (Figure 3). Transcription of regX3, pstS, and phoA was not significantly altered by deletion of phoU2 (Figure 4). Similarly, the ΔphoU1 mutant showed no significant change in expression of either regX3 or phoA (Figures 4A,C). Consistent with these results, alkaline phosphatase activity was unchanged in the ΔphoU2 mutant and was increased less than twofold in the ΔphoU1 mutant (Table 2). Increased alkaline phosphatase activity of the ΔphoU1 mutant was not complemented by the pMVphoU1 plasmid (Table 2). In contrast, transcription of pstS was significantly increased in the ΔphoU1 mutant compared to the mc²155 control and this phenotype was complemented by pMVphoU1 (Figure 4B), suggesting that the phoU1 deletion is responsible for this phenotype. Because neither regX3 nor phoA transcription were altered in either the ΔphoU1 or ΔphoU2 mutants, our data suggest either that PhoU1 and PhoU2 have partially redundant function in regulating the activity of SenX3-RegX3, or that additional factors contribute to controlling SenX3-RegX3 activity in M. smegmatis.

In E. coli, PhoU has previously been implicated as a “persister switch” that is required for formation of antibiotic-tolerant persister variants (Li and Zhang, 2007). Similarly M. tuberculosis requires either PhoY1 or PhoY2 to promote persister formation (Namugenyi et al., 2017). To establish if M. smegmatis similarly requires PhoU1 or PhoU2 for antibiotic tolerance, we determined the minimal inhibitory concentrations (MICs) of the anti-tubercular drugs rifampicin (RIF), isoniazid (INH), ethambutol (ETB), or ethionamide (ETH) against these mutants. Neither the ΔphoU1 nor the ΔphoU2 mutant displayed a substantial change in susceptibility to these drugs (Table 3). We also analyzed the rates of death of these mutants following antibiotic treatment to assess persister formation. For the ΔphoU2 mutant, we observed little change in the kinetics of bacterial death compared to the WT mc²155 control during treatment with EMB or ETH (Figures 5C,D). The ΔphoU2 mutant exhibited modestly reduced tolerance to RIF at 24 h compared to the WT control, though this difference was not maintained at later time points (Figure 5A). INH killed the ΔphoU2 mutant more rapidly than mc²155, and this phenotype was partially complemented by the pMVphoU2 plasmid (Figure 5B). The ΔphoU1 mutant was equally susceptible as the mc²155 control to RIF (Figure 5A). The ΔphoU1 mutant showed slightly increased susceptibility to INH at 48 h (P = 0.035) that was partially complemented by pMVphoU1 (Figure 5B). We also observed modest increases in the death rate of the ΔphoU1 mutant exposed to either EMB or ETH, though the differences compared to the WT control were significant only for ETH at 24 h (Figures 5C,D). These phenotypes were both complemented by pMVphoU1 (Figures 5C,D). These data suggest that PhoU1 and PhoU2 may have independent effects on M. smegmatis physiology that influence tolerance to the cell wall targeting antibiotics INH and ETH and to the transcriptional inhibitor RIF.

Our results suggest that PhoU1 and PhoU2 have partially redundant function in negative regulation of SenX3-RegX3 activity in P_i-rich conditions. To test this idea, we attempted to construct a ΔphoU1ΔphoU2 double deletion mutant. We introduced the ΔphoU1 allelic exchange vector into the ΔphoU2 mutant and obtained 4 independent transformants that contained the vector integrated via the cloned regions of homology either upstream or downstream of the phoU1 gene. After counter-selection against the allelic exchange vector by plating on sucrose, we screened a total of 172 sucrose resistant, kanamycin sensitive (Kan^S), hygromycin sensitive (Hyg^S) colonies for the ΔphoU1 deletion by PCR. None of these isolates harbored the ΔphoU1 deletion. These data suggested that M. smegmatis requires either PhoU1 or PhoU2 for viability under the growth conditions used.

To determine if phoU1 and phoU2 are jointly essential for M. smegmatis viability, we attempted to delete phoU1 in the ΔphoU2 background with a complementing copy of phoU1 provided in trans. We cloned phoU1 under the control of a tetracycline-inducible promoter on the integrating plasmid pTIC10a to enable conditional phoU1 expression. We created a ΔphoU2 pTICphoU1 strain and then electroporated this strain with the pJGΔphoU1 allelic exchange vector (Figure 6, steps 1–2). We obtained transformants in which the allelic exchange vector was integrated via either the upstream or downstream region of homology. After counter-selection on sucrose, we screened a total of 7 sucrose resistant, Hyg^S colonies for the ΔphoU1 deletion by PCR. Two of these isolates had the ΔphoU1 deletion (Figure 6, step 3). The ease with which we constructed the ΔphoU1ΔphoU2 double mutant when a complementing copy of one gene was provided in trans suggests that M. smegmatis requires phoU1 or phoU2 for in vitro replication. The ΔphoU1ΔphoU2 pTICphoU1 strain exhibited alkaline phosphate activity similar to the WT control, even in the absence of induction with anhydrotetracycline (ATc) (Table 2), suggesting that leaky expression of phoU1 from the pTICphoU1 plasmid was sufficient for negative regulation of SenX3-RegX3. In fact, phoU1 was expressed at a significantly higher level from the pTICphoU1 plasmid even in the absence of ATc compared to the mc²155 control (Supplementary Figure S1).

To confirm that at least one of the phoU genes is essential for M. smegmatis viability, we attempted to swap the pTICphoU1 plasmid with a non-compatible plasmid containing a Hyg^R marker, pJT6a (Rosen et al., 2017). The ΔphoU1ΔphoU2 pTICphoU1 strain was electroporated with pJT6a and transformants were selected on LB containing Hyg (Figure 6, steps 4–5). M. smegmatis mc²155 containing pTICphoU1 was similarly electroporated with pJT6a as a control. We obtained 4500 CFU/ml Hyg^R transformants from the WT pTICphoU1 control, but only 150 CFU/ml Hyg^R transformants from the ΔphoU1ΔphoU2 pTICphoU1 strain, suggesting that there is strong selective pressure against loss of the pTICphoU1 plasmid in the ΔphoU1ΔphoU2 background. The Hyg^R transformants were then screened for loss of the Kan^R marker on pTICphoU1 (Figure 6, step 6). For the WT control, all 10 Hyg^R colonies screened were sensitive to Kan, suggesting that they had lost the pTICphoU1 plasmid. PCR analysis on 8 of these Hyg^R Kan^S isolates confirmed that they all contained the pJT6a plasmid and not pTICphoU1 (data not shown). In contrast, for the ΔphoU1ΔphoU2 pTICphoU1 strain, of the 700 Hyg^R transformants that were screened, only 75 were sensitive to Kan, suggesting that the majority had not lost the pTICphoU1 plasmid. PCR analysis indicated that 63 of the Hyg^R Kan^S isolates still harbored the pTICphoU1 plasmid and did not contain pJT6a, suggesting that some recombination event had occurred to swap the antibiotic resistance markers in these strains. Using PCR, we ultimately identified only five Hyg^R Kan^S isolates that had successfully undergone the plasmid swap (Figure 6 and data not shown). All five of these isolates were also confirmed to have both the ΔphoU1 and ΔphoU2 deletions by PCR (data not shown).

To determine if the plasmid swap was inefficient in the ΔphoU1ΔphoU2 pTICphoU1 strain due to some deficiency in recombination as opposed to the loss of both phoU genes, we performed a similar plasmid swapping experiment using pJTphoU1 (Figure 6, steps 7–9). We observed a substantial increase in the number of transformants obtained with the pJTphoU1 plasmid compared to the empty pJT6a plasmid. We obtained 5800 CFU/ml Hyg^R transformants from the WT mc²155 pTICphoU1 background and 4200 CFU/ml Hyg^R transformants from the ΔphoU1ΔphoU2 pTICphoU1 strain. In addition, all 10 of the Hyg^R transformants that we screened on each strain background were sensitive to Kan, suggesting that they had replaced pTICphoU1 with pJTphoU1 (Figure 6). Taken together, these data suggest that M. smegmatis requires either phoU1 or phoU2 for viability under the in vitro culture conditions that we used and that the five ΔphoU1ΔphoU2 pJT6a strains we generated may harbor secondary suppressor mutations that enable their growth.

To determine if the ΔphoU1ΔphoU2 mutants that we isolated by the plasmid swap method contain suppressor mutations that enable their in vitro replication, we performed whole-genome sequencing on all five of the ΔphoU1ΔphoU2 pJT6a isolates. We compared the sequences of these strains to the sequence of our WT M. smegmatis mc²155 control, and identified two mutations in each strain. All five ΔphoU1ΔphoU2 pJT6a isolates harbored a G818T substitution in Msmeg_1387, which encodes a putative acyl-CoA dehydrogenase, resulting in a S273I transition in the Msmeg_1387 protein. We confirmed the G818T substitution in each ΔphoU1ΔphoU2 pJT6a isolate by PCR and sequencing, and determined that it was also present in the ΔphoU1ΔphoU2 pTICphoU1 parent strain, suggesting that this single nucleotide polymorphism (SNP) is not responsible for suppressing the essentiality of the phoU genes. Each ΔphoU1ΔphoU2 pJT6a strain also harbored an independent mutation in pstS, pstC or pstB, which encode components of the Pst P_i transporter (Table 4). These mutations were likewise confirmed by PCR and sequencing. Four of the strains (#504, 521, 625, and 664) contained SNPs that were predicted to severely disrupt the function of the encoded protein by causing a frame shift. The mutation in strain #518 was predicted to introduce a single glutamine residue into a region of PstB that contains primarily polar amino acids. This mutation may disrupt interaction of the PstB ATPase with other components of the Pst system. These data suggest that in the absence of PhoU1 and PhoU2, P_i import by the Pst system is toxic. All experiments described below used the ΔphoU1ΔphoU2 pJT6a #504 mutant, which harbors a frameshift mutation in pstS that is predicted to truncate more than 70% of the PstS protein (Table 4). Hereafter, we refer to this strain as the ΔphoU1ΔphoU2 pstS(Fs) triple mutant.

To quantitatively assess the requirement of phoU1 and phoU2 for M. smegmatis growth in vitro, we performed growth curves. Like the ΔphoU1 mutant, the ΔphoU1 ΔphoU2 pstS(Fs) mutant grew at a similar rate as WT M. smegmatis in liquid medium in vitro (Figure 1C), but formed significantly smaller colonies on LB agar plates (Table 1). Colonies of the ΔphoU1ΔphoU2 pstS(Fs) mutant typically did not appear until after 3–4 days of incubation at 37°C while the WT and single deletion mutant colonies were visible after only 2 days. This growth defect on plates was complemented by pMVphoU1 (Table 1). The ΔphoU1ΔphoU2 pstS(Fs) mutant also lost viability in stationary phase, though this phenotype was less pronounced as compared to the ΔphoU1 single mutant (Figure 1D). Only phoU1 provided in trans could complement this stationary phase survival phenotype (Figure 1D). Finally, the ΔphoU1ΔphoU2 pstS(Fs) mutant produced a pellicle at the air-liquid interface of cultures grown in tubes (Figure 2A), and aggregated in complete 7H9 liquid medium similar to the ΔphoU1 mutant (Figure 2B). These clumping phenotypes were both complemented by providing either phoU1 or phoU2 in trans (Figure 2). These data suggest that M. smegmatis requires either phoU1 or phoU2 for normal replication in culture.

To test if PhoU1 and PhoU2 function redundantly to regulate activity of SenX3-RegX3, we analyzed expression of known RegX3-regulated genes in the ΔphoU1ΔphoU2 pstS(Fs) mutant. By quantitative RT-PCR, the phoU1 and phoU2 transcripts were undetectable in this mutant (Figure 3), confirming that both phoU genes were deleted. Expression of regX3, pstS, and phoA was significantly increased 5-fold, 110-fold, and 158-fold, respectively, in the ΔphoU1ΔphoU2 pstS(Fs) triple mutant compared to the WT mc²155 control (Figure 4). Expression of pstS was also significantly increased in the ΔphoU1ΔphoU2 pstS(Fs) mutant relative to the ΔphoU1 single mutant (P = 0.004). We observed similar increases in the abundance of the regX3, pstS, and phoA transcripts in the #518 and #521 ΔphoU1ΔphoU2 pJT6a strains that each harbor an independent compensatory mutation in pstB (data not shown). We attempted to complement the gene expression phenotypes of the ΔphoU1ΔphoU2 pstS(Fs) mutant with either phoU1 or phoU2. The phoU1 and phoU2 transcripts were significantly elevated in the respective complemented strains compared to the WT mc²155 control (Figure 3). Complementation of the ΔphoU1ΔphoU2 pstS(Fs) mutant with pMVphoU1 restored expression of the regX3 and phoA transcripts to levels that were not significantly different from the WT mc²155 control (Figures 4A,C). Complementation with pMVphoU1 also reduced transcription of pstS to a level that was only 13-fold higher and not significantly different from the WT mc²155 control (Figure 4B). Complementation with pMVphoU2 only partially reversed the over-expression of the regX3, pstS, or phoA transcripts; levels of these transcripts were reduced compared to the ΔphoU1ΔphoU2 pstS(Fs) mutant but remained significantly elevated relative to the WT control (Figure 4). Consistent with these results, we observed a 8.5-fold increase in alkaline phosphatase activity in the ΔphoU1ΔphoU2 pstS(Fs) strain that was complemented by providing either phoU1 or phoU2 in trans, though full complementation was only observed with pMVphoU1 (Table 2). These data suggest that the PhoU1 and PhoU2 proteins have overlapping functions in negatively regulating SenX3-RegX3 activity in P_i-rich growth conditions. They further suggest that PhoU1 plays a more dominant role in fulfilling this P_i sensing signal transduction function. Since the gene expression phenotypes of the ΔphoU1ΔphoU2 pstS(Fs) mutant were complemented with pMVphoU1, our data also suggest that deletion of both phoU genes, not the pstS frameshift mutation in this strain, is responsible for activation of SenX3-RegX3.

To determine if the M. smegmatis PhoU proteins are jointly required for antibiotic tolerance, similar to the M. tuberculosis PhoY proteins, we performed MIC and persister assays on the ΔphoU1ΔphoU2 pstS(Fs) mutant. MICs of the ΔphoU1ΔphoU2 pstS(Fs) mutant for INH, EMB and ETH, were unchanged relative to the WT control, but the MIC for RIF was reduced 32-fold (Table 3). Complementation with either phoU1 or phoU2 restored the RIF MIC to the WT level (Table 3). These data suggest that constitutive activation of SenX3-RegX3 contributes to increased susceptibility to RIF. The ΔphoU1ΔphoU2 pstS(Fs) mutant was also killed more rapidly by RIF, INH and ETH in liquid culture compared to the WT mc²155 control, suggesting that it forms fewer persister variants (Figures 7A,B,D). The ΔphoU1ΔphoU2 pstS(Fs) mutant also showed a trend toward decreased tolerance to EMB, though it was not statistically significant (Figure 7C). However, in most cases these phenotypes were only partially reversed by complementation with either phoU1 or phoU2. For RIF, complementation with pMVphoU1 restored tolerance at early time points, but failed to prevent the increased killing observed at later time points (Figure 7A). For INH, complementation with pMVphoU1 resulted in an intermediate antibiotic tolerance phenotype (Figure 7B), consistent with the fact that the ΔphoU2 single mutant exhibited a modest defect in survival of INH treatment (Figure 5B). Complementation with either pMVphoU1 or pMVphoU2 similarly led to intermediate tolerance to ETH (Figure 7D), consistent with the fact that the ΔphoU1 mutant was also more susceptible to this drug (Figure 5D). In contrast, both complemented strains exhibited further enhanced susceptibility to EMB (Figure 7C). It is possible that over-expression of phoU1 or phoU2 from the episomal plasmid is detrimental and prevents complete complementation of these antibiotic tolerance phenotypes. Alternatively, it is possible that the point mutation in pstS, which is predicted to disrupt P_i transport by the Pst system, is partially responsible for the decreased antibiotic tolerance.

Mutation of phoU has previously been reported to cause accumulation of polyphosphate (polyP) a linear polymer of phosphate residues linked by high-energy phosphoanhydride bonds (Kornberg et al., 1999), in a variety of bacterial species including M. marinum and M. tuberculosis (Morohoshi et al., 2002; Wang et al., 2013; Namugenyi et al., 2017). We therefore examined whether deletion of the M. smegmatis phoU genes altered polyP storage during growth in P_i-rich 7H9 medium. The ΔphoU2 mutant showed no significant alteration in polyP accumulation (Table 5). Intracellular polyP was increased approximately twofold in the ΔphoU1 mutant though this difference was not statistically significant and this phenotype was not complemented with the pMVphoU1 plasmid (Table 5). In contrast, the ΔphoU1ΔphoU2 pstS(Fs) triple mutant exhibited a statistically significant 10-fold increase in polyP storage relative to the mc²155 control (Table 5). Complementation with either phoU1 or phoU2 resulted in only a partial reduction in polyP accumulation, though pMVphoU1 complemented the phenotype more effectively than pMVphoU2 (Table 5). These data suggest that the M. smegmatis PhoU proteins regulate polyP storage. Since the ΔphoU1ΔphoU2 pstS(Fs) mutant accumulates polyP despite the mutation in pstS, our data further suggest that P_i uptake through the Pst system does not contribute to increased polyP storage.