Construction of deletion mutants in the riboflavin pathway of Msm and Mtb

In a recent study [20], we characterized the riboflavin biosynthesis pathway in Mtb H37Rv and Msm mc²155 and confirmed that it comprises a bifunctional GTP cyclohydrolase II/3,4-DHBP synthase (RibA2); riboflavin deaminase/5-amino-6-(5-phosphoribosylamino) uracil reductase (RibG); an unidentified phosphatase; lumazine synthase (RibH), and riboflavin synthase (RibC) (Fig 1A and 1B). In Msm and Mtb the ribC, ribA2 and ribH genes are operonic [20], similar to the genomic organisation in B. subtilis where the riboflavin biosynthesis genes are arranged on a five-gene operon [21]. In B. subtilis, FMN riboswitches regulate the expression of riboflavin biosynthesis and transport [22]; however these regulatory elements have not been identified in the genomes of mycobacteria. Riboflavin is the precursor of flavin mononucleotide (FMN) and flavin adenine dinucleotide (FAD), which are produced by the bifunctional enzyme, RibF. The biosynthetic pathway also leads to the production of the deazaflavins, F₀ and F₄₂₀, from the pathway intermediate, 5-A-RU, via the sequential action of the enzymes FbiC, FbiA and FbiB [23]. Together, the flavin and deazaflavin cofactors enable the activity of a myriad of enzymes in mycobacteria, organisms which adopt ‘flavin-intensive’ lifestyles [24,25].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Riboflavin biosynthesis and utilization pathways in mycobacteria.

(A) In the biosynthesis pathway, GTP and ribulose-5-P are converted to 2,5-diamino-6-(5-phospho-D-ribosylamino)-pyrimidin-4(3H)-one (DARPP) and 3,4-dihydroxyl-2-butanone 4-phosphate (3,4-DHBP), respectively, by the bifunctional GTP cyclohydrolase II/ DHBP synthase, RibA2. DARPP is deaminated and the side chain subsequently reduced by the bifunctional riboflavin deaminase/ 5-amino-6-(5-phosphoribosylamino) uracil reductase, RibG, to form 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione-5-phosphate (ArPP). Dephosphorylation of ArPP by an unknown phosphatase generates 5-amino-6-D-ribitylaminouracil (5-A-RU). 5-A-RU together with 3,4-DHBP are condensed by the lumazine synthase, RibH (RibH1/RibH2), to yield 6,7-dimethyl-8-ribityllumazine (DMRL). Two molecules of DMRL are converted to riboflavin and 5-A-RU by the riboflavin synthase, RibC via a dismutation reaction. In the utilization pathway, riboflavin is converted to FMN and FAD by the bifunctional kinase/ FAD synthetase, RibF. The deazaflavin, F₄₂₀, is produced from 5-A-RU by the sequential action of FbiC, FbiA and FbiB. The deletion mutations are shown in red. (B) Genomic context of the riboflavin biosynthesis genes in Msm and Mtb.

https://doi.org/10.1371/journal.ppat.1012632.g001

A notable difference between Msm and Mtb is the redundancy in lumazine synthase activity in Msm where this function is served by a canonical RibH (MSMEG_3073, designated herein as RibH1) – the homologue of the sole RibH in Mtb (Rv1416) – and a second homologue, MSMEG_6598, designated here as RibH2 [20]. In prior work, we used a panel of inducible CRISPRi hypomorphs in genes involved in riboflavin biosynthesis and utilization to confirm the essentiality of the riboflavin pathway for mycobacterial growth and survival in vitro. We also demonstrated that Mtb and Msm can assimilate exogenous riboflavin but have yet to establish whether this occurs through passive diffusion or facilitated transport via a non-canonical transporter/s [20]. However, residual expression of target genes in induced hypomorphs could confound the ability to differentiate the contributions of individual riboflavin pathway intermediates in MAIT cell recognition of mycobacteria. To avoid this potential complication, we created Msm and Mtb riboflavin auxotrophs carrying deletions in individual genes in the riboflavin biosynthesis pathway for the purpose of this study (S1 Fig). Importantly, the deletion mutations in ribA2 eliminated both GTP cyclohydrolase II and DHBP synthase domains of RibA2; the former catalyzes the first step in the pathway whereas the latter produces a substrate for the RibH enzyme, which catalyzes the penultimate step.

The growth phenotypes of the Msm and Mtb mutants and their complemented counterparts were assessed in liquid and on solid media, with or without riboflavin supplement (Fig 2,S2 and S3 Figs). The Msm ΔribA2, ΔribG, ΔribH2ΔribH1 and ΔribC mutants were auxotrophic for riboflavin whereas the Msm ΔfbiC mutant showed no growth phenotype. When plating the wildtype and genetically complemented Msm auxotrophs on 7H10 agar with or without riboflavin supplementation, there was no significant difference in total CFU counts (S2B Fig). However, colonies of the complemented derivatives, Msm ΔribA2::ribA2, ΔribH1::ribH1, ΔribH2ΔribH1::ribH2 and ΔribH2ΔribH1::ribH2 ribH1 emerged later than the other strains (S2A Fig) and were significantly smaller than wildtype (S2C Fig). The retarded growth of these strains on solid media was consistent with their retarded growth kinetics in liquid culture (Fig 2A, 2D and 2F) which contrasts that other complemented derivatives for which restoration of growth to a level approximating the wildtype was observed (Fig 2B and 2C). This effect could be attributed to comparatively weak expression of the complementing genes, ribA2 and ribH1, from nested promoters instead of the main operonic promoter located upstream of ribC. The ability of Msm ΔribH1 to grow in the absence of riboflavin supplement on solid media (S2A Fig) and in liquid culture (albeit significantly more slowly than wildtype; Fig 2D) confirmed the existence of an alternate source/s of lumazine synthase activity in Msm that can partly compensate for the loss of RibH1. The fact that the deletion of both ribH1 and ribH2 rendered Msm incapable of growth without riboflavin supplement on solid media (S2A Fig) or in liquid culture (Fig 2F) is consistent with the designation of RibH2 as an alternate lumazine synthase in Msm.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Growth kinetics of wildtype, knockout and complemented mutant strains of Msm (A-G) and Mtb (H-J) in 7H9 media.

Growth of strains in liquid culture was monitored every 3 hours for 12 hours for Msm, and every day for 7 days for Mtb. Data are plotted as the mean, and error bars represent standard deviation from four biological replicates. Statistical comparisons against wildtype were performed at the 12-hr (for Msm) or 7-day (for Mtb) time point using area under the curve analysis (for Msm, panels A-G) or a one-way ANOVA and Dunnett’s multiple comparison test (for Mtb, panels H-J) whereby statistical significance is represented by p < 0.05, p < 0.001 shown by *, ** respectively. Only statistically significant results (p < 0.05) are shown. RF; riboflavin.

https://doi.org/10.1371/journal.ppat.1012632.g002

Mtb ΔribA2, ΔribH and ΔribC were auxotrophic for riboflavin (S3A Fig). In all cases, growth of the Mtb knockout mutants with riboflavin, and of the complemented strains in liquid media was equivalent to wildtype (Fig 2H-2J). Likewise, CFU counts of the knockout mutants grown on riboflavin, as well as of the complemented strains, were equivalent to wildtype (S3B Fig) although the colony size of the wildtype strain was on average smaller than the other strains (S3C Fig).

Riboflavin pathway transcript and protein levels in the Msm wildtype, knockout and complemented strains were measured by qRT-PCR and LC-MS/MS based Data Independent Acquisition (DIA) proteomics, respectively (Figs 3 and 4). As the auxotrophs required exogenous riboflavin for growth, we assessed the impact of riboflavin supplement on the levels of riboflavin pathway transcripts and proteins in wildtype Msm and found no significant effects. The absence of the cognate transcript and protein was confirmed in the knockout mutants with variable restoration thereof in the complemented derivatives in a manner consistent with the corresponding growth phenotypes (Fig 2 and S2 Fig). Certain knockouts also significantly affected the expression of other members of the pathway, e.g., a ~ 3-fold increase in ribH2 transcript and concomitant increase in RibH2 protein was observed in the Msm ΔribH1 mutant relative to the wildtype suggesting a regulatory interdependence of expression between the two lumazine synthases (Fig 4C and 4D). However, the ΔribH1 mutant was still significantly growth-impaired relative to wildtype (Fig 2D) indicating that RibH2 cannot fully compensate for the loss of RibH1 in Msm, and suggesting that RibH1 is the main catalytic enzyme for production of riboflavin. The ΔribH2 ΔribH1::ribH2 ribH1 mutant also showed significantly retarded growth kinetics (Fig 2F), which improved to some extent in the presence of riboflavin albeit not to the level of the wildtype. Poor complementation of the ΔribH1 deletion mutation was evidenced by the markedly lower levels of ribH1 transcript (~40-fold) and RibH1 protein (~3-fold) observed in the ΔribH1::ribH1 and ΔribH2ΔribH1::ribH2 ribH1 mutants compared to the wildtype (Fig 4A and 4B), which could potentially explain the comparative growth impairment of these complemented strains (Fig 2D and 2F). Whole genome sequencing (WGS) revealed only two non-synonymous SNPs in Msm ΔribH2ΔribH1::ribH2 ribH1 which were situated in a σ⁷⁰ family RNA polymerase sigma factor and an intergenic region, respectively. The contributions (if any) thereof to the growth phenotype of the complemented double mutant and incomplete growth restoration by exogenous riboflavin are unclear.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Quantification of ribA2, ribC and ribG transcript and protein levels in Msm strains.

(A, C, and E) Fold changes in sigA-normalized transcript levels relative to wildtype. Transcript levels of the target genes were normalized to the housekeeping gene sigA (an essential housekeeping gene which is stably expressed) and scaled to the average of wildtype Msm to calculate the ΔΔCt and determine the fold difference in gene expression (n = 2 biological replicates, 2 technical replicates). (B, D, and F) Protein abundance was measured using DIA proteomics (n = 3 biological replicates, 2 technical replicates). Data are shown as mean quantification ± SEM. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively. Only statistically significant relationships are shown. RF, riboflavin. Significance in comparison to wildtype (+/- RF) is shown as symbols as described in legend (p < 0.05). – and + indicates deletion and complementation of the corresponding gene, respectively.

https://doi.org/10.1371/journal.ppat.1012632.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Quantification of ribH1, ribH2 transcript and protein levels of lumazine synthase genes in Msm strains.

(A and C) Fold changes in sigA-normalized transcript levels relative to wildtype. Transcript levels of the target genes were normalized to the housekeeping gene sigA (an essential housekeeping gene which is stably expressed) and scaled to the average of wildtype Msm to calculate the ΔΔCt and determine the fold difference in gene expression (n = 2 biological replicates, 2 technical replicates). (B and D) Protein abundance was measured using DIA proteomics (n = 3 biological replicates, 2 technical replicates). Data are shown as mean quantification ± SEM. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively. Only statistically significant relationships are shown. RF, riboflavin. Significance in comparison to wildtype (+/- RF) is shown as symbols as described in legend (p < 0.05). – and + indicates deletion and complementation of the corresponding gene, respectively.

https://doi.org/10.1371/journal.ppat.1012632.g004

Statistically significant changes in transcript and/or protein levels were also observed in other Msm knockouts, specifically, ribC deletion was accompanied by a decrease in ribG transcript, RibG protein (Fig 3C and 3D) and RibA2 protein (Fig 3B), whereas ribA2 deletion resulted in an increase in the levels of ribH1 (Fig 4A) and ribC transcripts (Fig 3E) relative to wildtype in riboflavin-supplemented media. We also observed some differences in the effect of pathway gene knockout vs. knockdown on expression of other genes in the pathway, e.g., silencing of ribH1 led to an upregulation of ribG and ribF, and silencing of ribA2 suppressed the expression of ribG, ribC and ribH1 in Msm, demonstrated in our previous study [20] but such effects were not observed by knockout of ribH1 or ribA2. The differences may be due in part to the fact that the target proteins were only partially depleted in the knockdowns under the conditions of the expression analysis [20] but are entirely absent in the knockouts.

The Mtb ΔribA2, ΔribH and ΔribC mutants likewise showed riboflavin auxotrophy with riboflavin-independent growth restored by genetic complementation (Fig 2H–2J and S3A Fig). Knockout and complementation of ribA2, ribH and ribC were further validated by transcript and protein quantification (Fig 5). Additionally, stable isotope-labeled standards were made for peptides derived from RibC and RibH with the goal of determining targeted peptide concentration and, by inference, a more accurate quantification of the intact protein using parallel reaction monitoring (PRM) proteomics (S6 and S7 Figs and S8 Table). Riboflavin supplementation reduced the expression of ribC and increased the expression of ribA2 in wildtype Mtb (Fig 5A and 5E) but had no significant impact on expression of ribH. Also, the variable transcript levels had no impact on the abundance of their protein products (Fig 5B, 5D and 5F). Knockout of other pathway genes had no effect on RibA2 abundance (Fig 5B). RibH transcript and protein levels were markedly lower than wildtype in the complemented Mtb ΔribH strain (Fig 5C and 5D) but still sufficient to restore growth to wildtype levels (Fig 2I). In contrast, ribH transcript and RibH protein levels were increased ~ 10–12-fold in the ΔribA2 mutant and its complement vs. wildtype (Fig 5C and 5D), a finding confirmed by absolute quantification of RibH using PRM.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Quantification of ribA2, ribH and ribC transcript and protein levels in Mtb strains.

(A, C and E) Fold changes in sigA-normalized transcript levels relative to wildtype. Transcript levels of the target genes were normalized to the housekeeping gene sigA (an essential housekeeping gene which is stably expressed) and scaled to the average of wildtype Mtb to calculate the ΔΔCt and determine the fold difference in gene expression (n = 3 biological replicates, 2 technical replicates). (B and D) Protein abundance was measured using DIA proteomics (n = 3 biological replicates, 2 technical replicates). Data are shown as mean quantification ± SEM. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively. Only statistically significant relationships are shown. RF; Riboflavin. Significance in comparison to wild type (+/- RF) is shown as symbols as described in legend (p < 0.05). – and + indicates deletion and complementation of the corresponding gene, respectively.

https://doi.org/10.1371/journal.ppat.1012632.g005

RibH1 and RibH2 are capable of synthesizing DMRL in Msm while Mtb depends on a single lumazine synthase, RibH

Having confirmed the loss and restoration of gene expression and protein abundance in the cognate mutants and complemented strains of Msm, we next examined how these genetic disruptions affected metabolic intermediates within the pathway. To assess this, we measured the intracellular levels of 5-A-RU, DMRL, and riboflavin using LC-MS/MS Multiple Reaction Monitoring (MRM) metabolomics. DMRL was detected in all Msm strains except for the ΔribA2, ΔribG, and ΔribH1ΔribH2 mutants (Fig 6A and 6B). Although RibA2 protein expression was reduced in the ΔribA2::ribA2 complemented strain (Fig 3B), DMRL production was equivalent to wildtype levels (Fig 6A). We previously hypothesized that RibH1 and RibH2 may both independently produce DMRL, as supported by our observation that the deletion of ribH1 did not completely abolish Msm viability (S2A Fig). Consistent with this hypothesis, we found that DMRL levels in individual ribH1 and ribH2 mutants were similar to those in wildtype, indicating both lumazine synthases can synthesize DMRL utilized for riboflavin biosynthesis (Fig 6B and S7A Fig). However, when comparing DMRL levels between lumazine synthase mutants, the ribH2 mutant exhibited significantly lower levels compared to the ribH1 mutant, and these reduced levels were not restored upon complementation (Fig 6B). In Mtb, deletion of ribA2 and ribH resulted in the complete loss of DMRL production (Fig 6C, S9A Fig), reflecting the reliance of Mtb on a single lumazine synthase for DMRL synthesis. This observation aligns with our initial finding that deletion of ribH severely compromises the viability of Mtb. Additionally, significantly elevated DMRL levels observed in the ribC mutant suggest accumulation due to stalled riboflavin synthesis, a phenomenon not observed in Msm (Fig 6C). Riboflavin was consistently detected in the wildtype strain and in all complemented strains for both Msm and Mtb (S7B, S8 and S9B Figs). Since the riboflavin auxotroph mutants were grown in riboflavin, the substrate was also detected in those strains. For Msm, riboflavin levels in complemented strains typically mirrored wildtype levels, except for the ΔribC::ribC and ΔfbiC::fbiC complemented strains, which exhibited slightly higher riboflavin levels (S8A Fig). Riboflavin quantification in the riboflavin prototrophic ΔribH1 and ΔribH2 mutants grown without riboflavin supplementation revealed that deletions of the lumazine synthase paralogs exerted distinct effects on riboflavin production: a minor albeit significant reduction in riboflavin production was observed in the ΔribH1 mutant, whereas riboflavin production in the ΔribH2 mutant was indistinguishable from wildtype (S8B Fig). In contrast, Mtb strains showed no significant differences in riboflavin levels between the wildtype and complemented strains (S8C Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Quantification of DMRL in Msm and Mtb strains.

Levels of intracellular DMRL were quantified in Msm (A and B) and Mtb (C) using MRM. Data are shown as mean quantification ± SEM for three biological replicates. Quantification was carried out using chromatographic peak area of the most intense transition ion (m/z 327.1 → 193) of DMRL. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively. Only statistically significant relationships are shown. RF, riboflavin; Sig., significance in comparison to wild type (+/- RF) is shown as symbols, as described in the legend (p < 0.05).

https://doi.org/10.1371/journal.ppat.1012632.g006

Despite including 5-A-RU in our MRM assay, we could not detect this intermediate in any tested strain, likely due to its highly labile nature, as noted during assay development (S10 Fig) and supported by published data [26]. However, we can infer from our proteomic and metabolic analysis that the loss of DMRL when ribA2 and ribG were deleted is due to the loss of the substrate, 5-A-RU, required for its synthesis.

Deletion of ribA2 or ribG selectively ablates MAIT cell activation by Msm

To assess the impact of disrupting the riboflavin biosynthesis pathway on MAIT cell activation, Msm cultures grown in the presence of riboflavin were used to assess the ability to activate a panel of three MR1 restricted T cell clones: D426-G11, D481-C7 and D481-F12, TRAV1–2 + MAIT cell clones with varied TCRα chains (S1 Table) [9]. These were selected to reflect the diversity of ligands seen by MR1Ts with D481-C7 uniquely able to detect the photolumazine antigens derived from Msm [9]. Each clone was tested by IFN-γ ELISPOT for its ability to recognize dendritic cells infected with mycobacteria at an MOI of 10 (Fig 7). Deletion of ribA2 or ribG resulted in an 82–99% reduction in T cell recognition of Msm for all clones tested, a phenotype reversed by complementation. Since we still observed minor MAIT cell activation for both the ribA2 and ribG Msm deletion strains, we evaluated if this residual activity was MR1-dependent. We observed that the residual MAIT cell activation against both strains was blocked by an anti-MR1 antibody (S11 Fig). In contrast, deletion of ribC had no significant impact on T cell activation by Msm (Fig 7A, 7C and 7E). Deletion of ribH1 and ribH2 alone or in combination had no impact on recognition of Msm by the three TRAV1–2 + clones, D426-G11, D481-C7 and D481-F12.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Impact of riboflavin pathway mutations on MR1T recognition of Msm.

MR1T cell clone (1e4) IFN-γ response to dendritic cells (1e4) incubated with Msm at a MOI of 10. Response was normalized to wildtype Msm. All strains were grown with riboflavin. Wildtype Msm and positive control (PHA) response are shown in S12 Fig. Data are representative of n = 3 independent experiments. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively.

https://doi.org/10.1371/journal.ppat.1012632.g007

In addition to its role as a substrate of RibH and precursor of DMRL, 5-A-RU is also required for the production of the deazaflavin MAIT cell antagonist, F₀ [23], via the action of FbiC. We therefore assessed the effect of blocking this route of 5-A-RU metabolism on MAIT cell activation using a fbiC deletion mutant of Msm. The ΔfbiC mutation had no effect on Msm viability (Fig 2) or recognition by the TRAV1–2 + T cell clones, D426-G11, D481-F12 and D481-C7 (Fig 7A, 7C and 7E).

Divergent effects of ribA2 vs. ribH deletion on MAIT cell activation by Mtb

The Mtb strains were tested for their ability to activate two of the MR1 restricted T cell clones described above, D426-G11, D481-C7, and D481-F12 (Fig 8). Each clone was tested by IFN-γ ELISPOT for its ability to recognize the dendritic cells infected with Mtb at an MOI of 10. As the strains varied in their ability to activate the MAIT cell clones, we normalized these responses to the HLA-B45, Mtb8.4-specific clone, a classically restricted CD8 T cell clone that recognizes Mtb8.4 [27]. As observed in Msm, the deletion of ribA2 in Mtb reduced T cell recognition of all clones by 87–97%, an effect restored by complementation (Fig 8). The residual activity seen in the Mtb ribA2 gene knockout mutant could be blocked using an anti-MR1 antibody, suggesting the possibility that a subset of ligands are 5-A-RU independent (S13 Fig). On the other hand, the deletion of ribH enhanced recognition of Mtb by all three MAIT cell clones (Fig 8), which was most notable when normalized to the classically restricted CD8 T cell clone response, a different trend compared to the response observed in the single ΔribH1, ΔribH2 or double ΔribH2 ΔribH1 Msm mutants. Furthermore, knockout of the gene ribC had no significant effect on T cell recognition of Mtb.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Differential impacts of riboflavin pathway knockouts on MR1T recognition of Mtb.

The data represent the MR1T cell clone (1e4) IFN-γ response to dendritic cells (1e4) infected at an MOI of 10 of the indicated Mtb strain. The response for each Mtb strain was normalized to the response of D454 H1-2, a classically restricted CD8 T cell clone that recognizes Mtb8.4 [22]. Data are representative of n = 3 independent experiments. Statistical comparisons were performed using a one-way ANOVA and Sidak’s multiple comparison test whereby statistical significance is represented by p < 0.05, p < 0.001, p < 0.0005 p < 0.0001, shown by *, **, ***, **** respectively.

https://doi.org/10.1371/journal.ppat.1012632.g008

Media, culture conditions, strains, plasmids and chemicals

The plasmids used and constructs generated in this study are described in S2 Table. All strains generated were derivatives of either E. coli DH5α, Msm mc² 155 or Mtb H37RvMA, shown in S3 Table. E. coli DH5α was grown in Luria-Bertani (LB) medium (10g/L tryptone, 5g/L yeast extract and 10g/L NaCl) and plated onto Luria-Bertani agar (LB agar), containing LB media components supplemented with 15g/L agar. Mycobacterial strains were grown in Middlebrook 7H9 OADC medium (4.7 g/L 7H9 (Millipore Sigma), 2g/L glycerol (Millipore Sigma), 2.5 ml 25% Tween 80, and 100 ml of oleic acid-albumin-dextrose-catalase (OADC) enrichment (Millipore Sigma). Cells were plated onto Middlebrook 7H10 agar plates (19g/L 7H10 (Millipore Sigma), 100 ml OADC and 5 ml glycerol). Msm was grown in Erlenmeyer flasks in a rotating incubator set at 100 rpm at 37^o C and unless otherwise stated Mtb was cultured in sealed cell culture flasks with no agitation at 37^oC. Where required 7H9 broth or 7H10 solid media was supplemented with 50 µg/ml kanamycin, 50 µg/ml hygromycin or 2.5 µg/ml gentamicin. Riboflavin auxotrophs were supplemented with exogenous riboflavin at a concentration of 83 µM for Msm or 21 µM for Mtb based on the minimal concentration of riboflavin required to fully rescue the growth of CRISPRi hypomorphs in genes involved in riboflavin biosynthesis [20]. Media used to select for knockout mutants constructed by allelic exchange [33] or ORBIT [34] are described below.

Construction of mutant and complemented strains

Allelic exchange by homologous recombination was used to knockout ribC in Mtb and ribC, fbiC and MSMEG_6598 in Msm, as described [33]. In contrast, for reasons that are unclear, attempts to generate a knockout mutant of fbiC in Mtb were unsuccessful. Briefly, 1000–2000 bp fragments amplified up- and downstream of the gene of interest were ligated and cloned in p2NIL. The PacI cassette from either pGOAL19 or pGOAL17 cloned in the PacI site in the p2NIL (S2 Table). The resulting suicide construct was electroporated into the mycobacterial strain and single crossover clones selected by plating onto 7H10 media supplemented with 50 µg/ml kanamycin, 50 µg/ml hygromycin and 40 µg/ml X-gal (for pGOAL19 constructs) or 50 µg/ml kanamycin and 40 µg/ml X-gal (for pGOAL17 constructs). Double crossover recombinants were selected by plating onto media supplemented with 2% sucrose, 40 µg/ml X-gal and riboflavin at concentration of 83 µM for Msm or 21 µM for Mtb. The ORBIT system [34] was used to create individual gene knockouts of ribA2, ribG or ribH in Msm and ribA2 or ribH in Mtb although an attempt to construct a ribG knockout in Mtb using ORBIT was unsuccessful. Briefly, a targeting oligonucleotide containing 70 bp of homology to the target gene flanked by the BxbI phage attP sequence (S4 Table) was co-electroporated with pKM464 into the mycobacterial strain carrying pKM461 and selected on 7H10 agar containing 10% sucrose and 50 µg/ml hygromycin. To create complementation vectors, ca. 300–400 bp upstream of the gene along with its open reading frame were amplified and cloned into an integrating vector. All primers and amplicons are described in S4 Table. The complementation vectors were introduced into the knockout mutants by electroporation and selected on the appropriate antibiotic selection plates. DNA was extracted from all gene knockout and complement strains using the CTAB method [35] and the gene deletion/ gene restoration verified by whole genome sequencing.

RNA extraction

Mtb and Msm strains were inoculated in 80 ml of Middlebrook 7H9 OADC broth supplemented with 0.01% tyloxapol until mid-log phase (OD₆₀₀ between 0.4 and 0.6) was reached. Media was supplemented with riboflavin as necessary. Afterwards, cells were pelleted, resuspended in 1 ml of TriZOL (Thermo Scientific) and 5 µl of glycogen (Thermo Scientific) and then transferred to screw-cap homogenization microtubes prefilled with Lysing Matrix B (MP Biomedical). The cells were homogenized using the FastPrep-24 bead beater (MP Biomedical) at 6 m/s for 30 seconds (6 rounds), cooling on ice for 2 min between cycles. After centrifugation at 12,000 × g for 120 seconds, homogenate was collected in a new tube and treated with 200 µl of chloroform followed by vortexing for 15 seconds and incubation at room temperature for 5 minutes. Samples were centrifuged at 12,000 × g for 15 minutes at 4 ºC and upper aqueous phase was transferred into new tubes containing 500 µl of cold isopropanol for overnight precipitation. Samples were centrifuged at 12,000 × g for 10 min and supernatant disposed. Pellet was washed with 1 ml of cold 80% ethanol and then resuspended with nuclease free water. Extracted RNA was processed for DNase treatment and clean-up using the Rneasy Mini Kit (QIAGEN) and Rnase free Dnase Set (QIAGEN) according to manufacturer’s instruction to eliminate genomic DNA. RNA was quantified using a NanoDrop (Thermo Scientific) and Agilent TapeStation.

Gene expression analysis

Two hundred ng of RNA was retrotranscribed to complementary DNA (cDNA) using the SuperScript IV VILO Master Mix (Thermo Scientific) according to manufacturer’s instruction. PowerTrack SYBR green Master Mix (Thermo Scientific) with 250 nM of primer pair was used to amplify the target region of interest for 5 ng of cDNA following manufacturer’s instruction. Quantification was run on the QuantStudio 3 Real-Time PCR System (Thermo Scientific). A no reverse transcriptase (no-RT) control was included for all samples to ensure that there was no genomic DNA contamination. Amplicon size was confirmed by melting curves for each sample. All qRT-PCR experiments were performed with two technical replicates derived from three biological replicates for Mtb and two biological replicates for Msm. Transcript levels of the target genes were normalized to the housekeeping gene sigA (an essential housekeeping gene which is stably expressed) and calibrated to wildtype Mtb and Msm to calculate the ΔΔCt and determine fold difference in gene expression. The primers used are listed in S5 Table.

Growth kinetics of mycobacteria

Msm and Mtb growth kinetics in 7H9 media were monitored by OD₆₀₀ as previously described [20].

Growth and preparation of Mtb and Msm for targeted proteomics

Glycerol stocks of Mtb and Msm strains were used to start cultures on 7H11 agar plates (Millipore Sigma, Cat No.: M0428) supplemented with the appropriate antibiotics with or without riboflavin. Plates were incubated for 2 weeks for Mtb and 4–5 days for Msm at 37 ºC. Cells were scraped from plates into 7H9 media supplemented with OADC with or without riboflavin and grown at 37ºC until mid-log phase. Mtb cells from all strains were inactivated by γ-irradiation and inactivation was confirmed using the Alamar Blue assay. Cells were harvested and washed 3 times either with phosphate buffered saline (PBS) or water to remove media and stored at -80 ºC until further processing. Cell pellets were further processed as previously described [20].

Growth and preparation of Mtb and Msm for targeted metabolomics and untargeted proteomics

Glycerol stocks of Mtb and Msm strains were used to start cultures on 7H11 agar plates supplemented with the appropriate antibiotics with or without riboflavin. Plates were incubated for 2 weeks for Mtb and 4–5 days Msm at 37 ºC. Cells were scraped from plates into 7H9 media supplemented with OADC with or without riboflavin and grown at 37ºC until mid-log phase. Cells were collected via centrifugation, washed with mass spectrometry grade water and resuspended in Acetonitrile/Methanol/Water (2:2:1) and then transferred to screw-cap homogenization microtubes prefilled with 0.1 mm zircon beads. The cells were homogenized using the FastPrep-24 bead beater (MP Biomedical, SKU:116004500) at 6 m/s for 30 seconds (10 rounds), cooling on ice for 2 minutes between cycles. Homogenized cells were collected and spun down at 16,000 × g for 20 minutes and supernatant collected for metabolomics and pelleted cell lysate for proteomics. The pelleted proteins were quantified using Bicinchoninic acid assay (Thermo Scientific, Cat No.: 23225). Supernatant and pellet were then stored at -20ºC until further processed.

Liquid chromatography - MS/MS analysis (Proteomics)

Thirty µg of retentate from whole cell lysate from Msm and Mtb after ultrafiltration with 3 kDa filters (Millipore Sigma, Product Number: UFC9003), was subjected to in-gel trypsin digestion as described previously [36]. Digested samples were resuspended in 120 µl of loading buffer (3% acetonitrile, 0.1% formic acid in water). All samples were analyzed using nano-UHPLC (ultrahigh-pressure liquid chromatography) nanoElute (Bruker Daltonics, USA) coupled to a timsTOF Pro (Bruker Daltonics, USA) mass spectrometer. 0.25 µg of total peptides were loaded onto a trap cartridge using PepMap Neo Trap Cartridge (Thermo Scientific, Cat: 174500) (5 µm C18 300 µm X 5 mm) packed with spherical, fully porous, ultrapure silica and reverse phase UHPLC was performed in a PepSep Column (Bruker Daltonics, Part No:1895802) (1.5 µm C18 75 µm X 25 cm). Mobile phase A was 0.1% formic acid in water; mobile phase B was 0.1% formic acid in acetonitrile. Liquid chromatography (LC) separation was carried out at a flow rate of 0.5 µL/min using a linear gradient of 5% solvent B to 30% solvent B over a duration 17.8 minutes, then 30% solvent B to 95% solvent B for 2.5 minutes. Thereafter, 95% solvent B was held for 2.4 minutes for equilibration.

Mass spectra and tandem mass spectra were obtained using the parallel accumulation serial fragmentation (PASEF) acquisition method in positive ion mode. Captive spray source was operated with the following settings: 1600 V capillary voltage, 3.0 l/min of dry gas at 200ºC with the NanoBooster switched off.

A spectral library was generated by analyzing samples using data-dependent acquisition mode (DDA-PASEF). DDA was set to obtain high resolution MS scans over the m/z range from 100 to 1700 and ion mobility range from 0.70 Vs/cm² to 1.50 Vs/cm². Ramp time was set at 100 msec and accumulation time was set at 100 msec. The collision energy was set to follow a linear function starting from 20 eV for 0.6 V. s/cm² to 59 eV for 1.6 V. s/cm². The number of PASEF MS/MS was set to 6. For DIA-PASEF analysis, data was acquired with a method consisting of 24 mass steps per cycle (S6 Table). Mass width was set at 50 Da with 1 Da overlap between windows covering a mass range between 307.8 to 1484.8 Da. The ion mobility range (1/K₀) covered was between 0.71 to 1.44. The collision energy was set to follow a linear function starting from 20 eV for 0.6 V. s/cm² to 59 eV for 1.6 V. s/cm².

Data analysis of untargeted proteomics Data Independent Acquisition (DIA)

DIA-PASEF data was processed on FragPipe software v. 19.1 (MSFragger version 3.7, IonQuant version 1.8.10, Philosopher version 4.8.1) using the DIA_SpecLib-Quant workflow as described [37]. DIA raw files were loaded with DIA Quant selected as data type while DDA raw files were loaded for spectral library generation. A search was conducted using default settings with the FASTA sequence obtained from the Mycobrowser database (Release 4, 2021-03-23). Decoy and contaminant sequences were generated by FragPipe. Default settings of MSFragger was used, including oxidation of M (+ 15.9949 Da mass shift) acetylation at N-terminal (+ 42.0106) as variable modifications. Advanced output options were set to write calibrated mzML and MGF files. Tryptic searches were performed with 20 ppm precursor and fragment tolerance permitting up to 2 missed cleavages. Search results were filtered to 1% FDR at both protein and peptide levels using peptide prophet. Output files with quantification based on summed intensity of all ions for a single peptide and summed intensity of peptides for a single protein were generated. The output file containing summed intensity of peptides for a single protein was used for quantification of proteins and is provided in S6 Table.

Targeted proteomics analysis by parallel reaction monitoring -PASEF Mass Spectrometry (PRM-PASEF MS)

DDA peptide search results from MSFragger version 3.7 were used to build a spectral library in Skyline [38]. The Ion Mobility library was generated on Skyline software from raw DDA files. The protein list was trimmed to contain only proteins from the riboflavin biosynthetic pathway. Tryptic peptides were filtered to have length only be between 8 and 30 amino acids. Tryptic peptides containing C, M, D-P, and W were excluded. Peptides for each protein were further screened based on the shape of the precursor chromatogram and absence of the precursor peak from the cognate knock-out sample. Ion types were limited to only precursor ions (2 and 3 ion charges) and 6 y ions (1 and 2 ion charges). The retention time window was set at 3 minutes and mass accuracy set at 10 ppm for both MS1 and MS2 filtering. The method was exported to timsTOF engine (Bruker Daltonics, USA) as an unscheduled method. 0.25 µg of each Mtb sample was loaded for reverse phase UHPLC and MS. performed. Analytical settings were kept the same as previously described (LC-MS/MS Section) with the inclusion of a target list containing the Mtb peptides to quantify in the PRM-PASEF method generated.

Raw data were exported into Skyline software and peptides with the best peak shape and intensity were selected for further optimization. At least, one peptide per protein was selected for the final method. For absolute quantification, peptides labeled with heavy C-terminal Lysine (K) or Arginine (R) were purchased (Biosynth Ltd, UK) to be used as internal standards. Limit of detection (LOD) and Limit of quantification (LOQ) were calculated using a linear regression method as described elsewhere [39]. Details of the peptide sequences and transition list used for the PRM-PASEF experiment are shown in S7 Table.

Targeted quantification of riboflavin pathway proteins by PRM-PASEF MS

A mix of the heavy labeled peptides containing 10^-1 – 10^-5 pmol/µl of each peptide was used to resuspend each of the digested samples at a final concentration of 250 ng/µl of digested protein (S8 Table). After resuspension, the sample peptide concentration was estimated using a Nanodrop 1000c and the concentration was adjusted to ensure equal loading of all samples. The developed PRM-PASEF MS method was exported as a scheduled method and used to obtain the Ratio to Heavy Standard for each of the monitored peptides. Briefly, raw data from each injection was imported into Skyline [38] and peak boundaries for each peptide were manually validated and adjusted if necessary. The Ratio to Heavy Standard for each peptide as calculated by Skyline was exported into Prism GraphPad for statistical analysis.

Liquid chromatography -MS/MS analysis (Metabolomics)

Supernatant collected for metabolomics was dried in savant and their weight determined. Dried samples were then resuspended in 3% Acetonitrile in Water (0.1% Formic Acid) to make a final concentration of 2 mg/ml for Mtb and 10 mg/ml for Msm. Samples were analyzed using the Agilent 1290 Infinity III Liquid Chromatography System attached to a 6470 Triple Quadrupole MS system. 10 µl of samples were loaded per injection. Reverse phase UHPLC was performed using InfinityLab Poroshell 120 EC-C18 Column (Agilent, Part Number: 693975–906) (4.6 x 150 mm, 2.7 µm) for Mtb samples and InfinityLab Poroshell 120 EC-C18 Column (Agilent, Part Number: 695975–906) (4.6 × 100 mm, 2.7 µm) for Msm samples. Mobile phase A was 0.1% formic acid in water; mobile phase B was 0.1% formic acid in acetonitrile. LC separation was carried out at a flow rate of 0.3 µL/min using a linear gradient of 2% solvent B held for 1.5 minutes and then to 20% solvent B over a duration 5 minutes, then 20% solvent B to 99% solvent B for 3 minutes. 99% solvent B was held for 0.5 minutes and then returned to 2% solvent B for 3 minutes for equilibration.

Mass spectra and tandem mass spectra were carried out using the dynamic multiple reaction monitoring (DMRM) acquisition method in positive ion mode. Column temperature was set at 45 ºC. Source was operated with the following settings: 10 l/min of gas flow at 325ºC. Nebulizer was set at 20 psi. Sheath gas temperature was set at 350 ºC with flow rate of 10 l/min. Capillary voltage was set at 3500 V and nozzle voltage set at 0 V.

Targeted metabolomics analysis by dynamic multiple reaction monitoring-mass spectrometry (DMRM-MS)

Agilent Masshunter Workstation was used to build MRM and DMRM methods for all compounds. Standards for DMRL, 5-A-RU, and riboflavin at 1 µg/ml were used to determine optimal fragmentor voltage and collision energy for each compound. Subsequently, observed product ion were filtered to keep the top three ions based on abundance. An unscheduled MRM was run to determine retention time for each compound. Final retention times were set in DMRM mode, and a retention time window was set at 2 minutes. See S9 Table for parent ion and product ion list used for DMRM. Mixtures containing 103 ng/ml to 10–3 ng/ml of the four compounds were analyzed using the DMRM method to determine the limit of detection (LOD) and limit of quantification (LOQ) as reported in S9 Table. The product ion for each parent ion with the highest abundance was selected as the quantitative ion while the two other product ions in addition to the most abundant product ion were used to qualitative ions (S9 Table).

Samples for Mtb and Msm were loaded as described for LC-MS/MS analysis. Raw data for each injection was imported into Agilent Masshunter Qualitative Analysis Software version 10.0. Peaks for the three selected product ions were integrated for each compound and adjusted if necessary. Total Area Under Curve (AUC) was manually extracted for each product ion. To ensure accurate quantification, detected compounds were qualified by comparing the ratio of the peak area of the transition ions observed in the sample to the ratio observed for the standards. Ratios were compared as dot products and a cutoff of 0.98 was set for qualification. Extracted data that passed the qualification step was analyzed on GraphPad Prism. MRM data have been deposited at the Dryad Digital Repository [40] and are available at: https://doi.org/10.5061/dryad.z08kprrr1.

Whole genome sequencing (WGS)

Genomic DNA was analyzed at the Genomics and Microarray Core at the University of Colorado Anschutz using a NovaSEQ 6000 Instrument, paired end 150 cycles 2x150. Reads were subjected to quality control using FastQC before and after trimming. Trimming was performed using trimmomatic. Alignment of reads to the Mtb H37Rv genome (NC_000962.3) or Msm mc² 155 (CP000480.1)/ (NZ_CP054795.1) was performed using Bowtie2. Due to the high density of reads, the aligned sam files were subjected to subsample using samtools. Samtools were also used to determine coverage results. Subsample alignment was used to determine SNPs using lofreq with a minimum coverage of 10 and minimum frequency of 80. The bam file was visualized using the Geneious software. The expected deleted regions were confirmed by visual inspection and very few inadvertent second-site mutations were detected in the deletion mutants of Msm and Mtb relative to their parental wildtype strains (S10, S11 and S12 Tables). WGS data have been deposited at the at the Dryad Digital Repository [41,42] and are available at: https://doi.org/10.5061/dryad.4tmpg4fnv and https://doi.org/10.5061/dryad.vq83bk454

T cell clones

All T cell clones were expanded and maintained as previously described [7,27]. Briefly, a T cell clone is incubated with irradiated PBMC and LCL in addition to anti-CD3 for 12–17 days and frozen. The anti-CD3 is washed out on day 5. Four T cell clones were used in this work. All have been published previously: D426 G11, D481-C7, and D481-F12 are MAIT cell clones expressing TRAV1–2 [7,9], and D454 H1-2 is a CD8 T cell clone that recognizes Mtb8.4_33–43 and is restricted by HLA-B*15:01 [27] (see S1 Table for details of TCRs). D426 and D454 were isolated from a patient that was Mtb-infected (PPD+) and D481 from a patient that had previously had active TB. To generate these clones, CD8 + T cells were sorted from PBMC and incubated in a 96 well limited dilution assay with autologous Mtb-infected dendritic cells for 2 weeks. Wells with growth were screened against autologous and heterologous Mtb-infected dendritic cells. T cell clones that responded only to autologous dendritic cells were characterized to determine the Mtb antigen they recognize (D454 H1-2), and T cell clones that responded to both autologous and heterologous dendritic cells were blocked with various antibodies. D426 G11, D481 C7, and D481 F12 were blocked by anti-MR1 antibody and later found to stain the MR1/5-OP-RU tetramer.

IFN-γ ELISpot assays

All Msm and Mtb strains were grown in the presence of riboflavin with or without supplements as needed (see above) to an OD₆₀₀ ~ 0.5 and frozen down with 10% glycerol. All strains were titered on 7H10 agar plates to obtain CFU/ml. Monocyte derived dendritic cells were prepared from D520 PBMC, which expresses HLA-B*15:01, using plastic adherence and culturing for 5 days with IL-4 and GM-CSF [27]. For all ELISPOT assays, ELISPOT plates were coated with anti-IFN-γ antibody (Mabtech, clone 1-D1K, Cat No. 3420-3-1000, used at 10 μg/ml) as previously described [43]. After overnight incubation at 4°C, ELISPOT plates were washed three times with phosphate-buffered saline and then blocked with RPMI 1640 + 10% human serum for at least 1 hour. For experiments with Msm approximately 1 × 10⁴ dendritic cells were plated in the ELISPOT wells in RPMI+10% HuS. Msm bacilli were added to the wells at concentrations indicated for each experiment. PHA (Millipore) was used as positive control. After at least 1 hour, 1–2 × 10⁴ T cell clones that were freshly thawed from a cryopreserved expansion were added, and the plate was incubated overnight at 37°C. IFN-γ ELISPOTs were enumerated following development as previously described [43]. For experiments with live Mtb, 5 × 10⁵ dendritic cells were plated in each well of an ultralow adherence 24 well tissue culture plate. A vial of frozen Mtb was thawed, passaged through a tuberculin needle >20 times, and the appropriate volume was added to the dendritic cells at the MOI indicated and incubated overnight. Dendritic cells were harvested and 1 × 10⁴ uninfected or Mtb-infected dendritic cells were plated in the ELISPOT wells in RPMI+10% HuS. 1–2 × 10⁴ T cell clones were added, and the plate was incubated overnight at 37° C. IFN-γ ELISPOTs were enumerated following development.

Statistical analyses

Statistical calculations were performed using GraphPad Prism 10 statistical software (GraphPad Software, Inc., San Diego, CA). For mycobacterial growth on solid media, statistical comparisons against the corresponding wildtype strain were performed using a one-way ANOVA and Kruskal-Wallis comparison test. For growth in liquid culture, comparisons were performed using area under the curve (for Msm) or one-way ANOVA and either Sidak’s or Dunnett’s multiple comparison test (for Mtb). For proteomics, qRT-PCR and the EliSPOT assay, statistical comparisons were performed using a one-way ANOVA followed by Dunnett’s or Sidak’s multiple comparison test. The latter analyses were done using GraphPad Prism 9 and statistical significance was established at p < 0.05 or p < 0.001.