Exposure to nitric oxide drives transition to differential culturability in Mycobacterium tuberculosis

During infection Mycobacterium tuberculosis (Mtb) forms differentially culturable (DC) subpopulations that are recalcitrant to treatment and undetectable using standard diagnostic tools. DC Mtb are revealed in liquid media, their revival is often stimulated by resuscitation-promoting factors (Rpfs), secreted peptidoglycan-remodelling enzymes, and prevented by Rpf inhibitors. Here we investigated the role of nitric oxide (NO) in generation of Rpf- dependent DC Mtb, using murine macrophage infection models and treatment with a synthetic NO donor (NOD). Mtb subpopulations were assessed by colony-forming unit counting on agar or by limiting dilution Most Probable Number assays in liquid media with or without Rpf inhibitor. Rpf-dependent DC Mtb were detected following infection of interferon-γ induced macrophages capable of producing NO, but not when iNOS was inactivated. NOD treatment also induced transition to the Rpf-dependent DC phenotype which was accompanied by global transcriptomic changes resulting in the dramatic down-regulation of rpfA-E gene expression. Furthermore, the DC phenotype was partially reverted by artificial over-expression of Rpfs. This study elucidates molecular mechanisms underlying the generation of DC Mtb, which are the dominant population recovered from clinical tuberculosis samples, with implications for improving both tuberculosis diagnostics and treatments.


INTRODUCTION
Mycobacterium tuberculosis (Mtb) is the causative agent of tuberculosis (TB), which resulted in the death of 1.4 million people in 2019 [1]. Standard TB treatment lasts at least 6 months and significant resources are directed towards developing shorter and more effective treatments [2]. One barrier to achieving this goal is the generation of tubercule bacilli populations during infection that are not detectable by standard tools [3]. These bacilli do not produce colonies on agar and only grow in liquid medium, often require supplementation with culture supernatant (CSN) [3,4] and therefore called differentially culturable bacilli (DCB) [4]. In published literature they have been also defined as non-culturable cells [5], viable but non-culturable bacilli [6] and differentially detectable cells [7]. DCB have higher tolerance to some anti-TB drugs [3,8], and drug treatment may increase the proportion of DCB in patients [3,7,9,10].
These cells are believed to be in a special dormant state that requires resuscitation in liquid media or CSN, however, DCC are currently poorly characterised.
CSN contains resuscitation-promoting factor (Rpf) proteins, which are peptidoglycan remodeling enzymes [11] that are able to resuscitate DCB from the sputum of TB patients [3,12] and animal lungs [13]. While resuscitation of these Rpfdependent Mtb has been linked to TB relapse in mice [14], the importance of this process in human TB remains to be established.
The molecular mechanism of DCB generation is currently unknown, although the in vivo environment has been proposed as a key factor [13]. Nitric oxide (NO) is a wellcharacterized component of the innate immune system that is deployed to control intracellular pathogens [15]. While its precise role in controlling human TB is unclear, the inducible NO synthase (iNOS) of macrophages is critical for controlling TB in mice coding for Rpf proteins (rpfA, rpfB and rpfE). Importantly, NO activation of THP-1 macrophages with retinoic acid and vitamin D3 promoted NO production and formation of DCB. Taken together, these observations suggest that NO plays a role in generation of DCB during TB infection.

Assessment of viable counts
MPN counts were quantified as previously described [3,8]

Synthesis of NO donor (ND) and control compound (CC) and quantification of NO release
NO donor 3-cyano-5-nitropyridin-2-yl diethyldithiocarbamate (ND) and Control compound 3-cyano-4,6-dimethyl-5-nitropyridin-2-yl piperidine-1-carbodithioate (CC) were synthesized as described in the Online Supplement. The concentration of NO generated by ND and CC were quantified using Griess reagents as previously described [22]. ND or CC were incubated in 200 mM sodium phosphate buffer (pH 6.0) at a final concentration of 200 µM for 24 hours and absorbance at 540 nm was measured at different time points. Known concentrations of sodium nitrite were used to construct a calibration curve.

Treatment of Mtb with ND and CC
Mtb was grown in 7H9 medium to OD580nm of 0.8-1.0. Cells were diluted with fresh 7H9 medium to an OD580nm of 0.1 in a total volume of 5 ml. ND or CC were dissolved in DMSO and added to cultures to a final concentration of 100 µM. Cultures were incubated at 37°C without shaking for 24 hours. In other experiments, mycobacteria were treated with 200 μM spermine-NONOate (equivalent to up to 400 μM NO; Sigma), or 5 mM NaNO2. LIVE/DEAD BacLight (ThermoFisher) stained M. bovis BCG were studied in FACSAria flow cytometer using BD FACSAria TM software (BD Bioscience). The following parameters were applied: SYTO 9 excitation/emission at 480/500 nm; PI excitation/emission at 490/635 nm; 10,000 events were recorded from each sample.

Flow cytometry staining and analysis
Single parameter fluorescence overlay plots and dual parameter dot plots were generated as previously described [24]. The gating strategy is summarized in the Online Supplement. Heat-killed M. bovis BCG were included as a control.

Transcriptomic analyses
Three biological replicates of RNA from Mtb were extracted 4 hours after treatment with either ND or CC using the GTC/Trizol method [25]. RNA (2 μg) was labeled with Cy3 fluorophore and hybridized to a Mtb microarray (ArrayExpress accession number ABUGS-41) as previously described [26]. Fully annotated microarray data are available in ArrayExpress (E-MTAB-10776).
For quantitative RT-PCR, RNA was reverse transcribed to cDNA using SuperScript II reverse transcriptase kit (ThermoFisher Scientific) with mycobacterial genome-directed primers [27]. The qPCRs were run in the Rotor-gene 600 instrument (Corbett Research) using the 2´SYBR green master mix (ThermoFisher Scientific) and primers (Table S1). Levels of expression were normalized to 16S rRNA.

Macrophage infection
Human THP-1 monocytes were seeded at 2´10 5 cells/ml in 24 well plates (Greiner) followed by activation either with retinoic acid (1 µM) and vitamin D3/cholecalciferol (1 µM) or 10 ng/ml PMA (phorbol myristate acetate) for 72 hours. The macrophages were infected with M. bovis BCG at an MOI of 0.5 and incubated at 37°C with 5% CO2 for 3 hours, followed by treatment with 200 µg/ml amikacin for 1 hour. For CFU and MPN counts, the infected macrophages were washed three times with PBS and lysed with 0.2% (v/v) Triton X-100.

Detection of Rpf peptides using mass-spectrometry
Mtb was grown in Sauton medium to OD580 0.8. Filtered CSN was enriched on DEAE Sepharose and digested with trypsin. LC-MS/MS was carried out using an RSLCnano HPLC system (Dionex, UK) and an LTQ-Orbitrap-los mass spectrometer (Thermo Scientific). The raw data were processed using Proteome Discoverer (version 1.4.0.288, Thermo Scientific), Mascot (version 2.2.04, Matrix Science Ltd).

Statistical analyses
Unpaired t-test (Excel) or one-way ANOVA (Prism 9) were used to analyze growth and flow cytometry datasets. Differentially expressed genes in ND-compared to CCtreated cultures, 4 hours after exposure, were identified using a modified t-test (GeneSpring 14.5; Agilent Technologies) with Benjamini and Hochberg multiple testing correction, and defined as those with p<1´10 -5 and minimum fold change of 2.0.

Design, synthesis and testing of a novel NO donor
Exposure of mycobacteria to reactive nitrogen species lowers CFU counts [17, 28,29], which could indicate the presence of DCB. Mtb incubation with acidified nitrite in 7H9 medium at concentrations above 5 mM reduced Mtb CFU counts, but resuscitation in liquid medium was not achieved (Fig. S1A) S1A). We wondered if NO release might be too rapid for alteration of Mtb CFU counts and that sustained exposure to NO might be required for the generation of DCB. Therefore a novel NO donor (ND), 3-cyano-5-nitropyridin-2-yl diethyldithiocarbamate ( Figure 1A) and a structurally related control compound (CC), 3-cyano-4,6-dimethyl-5-nitropyridin-2-yl piperidine-1-carbodithioate ( Figure 1B) were designed and synthesized. The nitro group of CC was protected from hydroxyl or enzymatic attack by two methyl groups to impair release of NO in culture media. NO was progressively released from ND at 20°C, pH 6.0 with the highest concentration detected after 8 hours. NO was not produced from CC ( Figure 2A).

DAF-FM diacetate can be used to detect the presence of intracellular NO [31].
DAF-FM diacetate positive M. bovis BCG were significantly higher (p<0.0001) in NDtreated bacteria compared to CC-treated and untreated bacteria ( Figure 2B). Heatkilled M. bovis BCG did not stain with DAF-FM diacetate after exposure to ND. This suggested that NO released from ND entered mycobacteria and was therefore suitable for testing the effects of NO exposure on growth phenotypes using CC as a control.
After 24 hours there was no significant difference (P>0.05, t-test) in the percentage of live or dead bacteria for ND-treated or CC-treated M. bovis BCG ( Figure 2C), which suggested that NO was not directly bactericidal. More than 50% of cells in both samples were PI-, SYTO 9+, and likely remained viable.
While CC treatment for 24 hours had no effect on CFU counts ( Figure S3), exposure to ND resulted in a 1000-fold reduction in CFU ( Figure 3A). Furthermore 1.2´10 5 bacteria were recovered in 7H9 medium 24hour post-treatment with ND, and supplementation with CSN increased this to 1.7´10 6 ( Figure 3A). This suggested that the majority of the ND-treated cells were differentially culturable, in line with observations with Mtb from clinical TB samples [3,4,8]. Similar results were observed with M. bovis BCG; however, the number of resuscitated bacteria was generally lower and more variable ( Figure S4).

The addition of a Rpf inhibitor [19] abolished the resuscitation of ND-treated Mtb
in 7H9 medium, and significantly (p<0.05) impaired resuscitation in CSNsupplemented medium ( Figure 3B). In accordance with previous results [13,19], the Rpf-inhibitor had no effect on actively growing Mtb ( Figure S5). Proteomics analysis confirmed that RpfA and RpfB were present in the Mtb CSN ( Figure 3C). Together, these data suggest that Rpf proteins play an important role in resuscitation of NDgenerated DCB.

Generation of Rpf-dependent Mtb by ND does not require DosR
DosR is a response regulator that has been previously shown to respond to NO [33,34]. Treatment of the Mtb dosR deletion mutant, a complemented strain, and the parent Mtb with ND for 24 hours resulted in similar numbers of DCB ( Figure S6). All strains were equally resuscitated in CSN-supplemented medium ( Figure S6), indicating that transition to the Rpf-dependent differentially culturable state after ND exposure and subsequent resuscitation by Rpf proteins did not require DosR.

ND-treatment induces SigH-and SigF-transcriptional alteration and lower rpf gene expression
Transcript profiling was used to investigate the transition of Mtb to differential culturability. Compared to treatment with CC, 640 (407 induced, 233 repressed) genes were differentially expressed in response to ND 4 hours after exposure ( Figure 4A, Table S2). Analysis of up-regulated genes (using Transcription Factor Over Expression (TFOE) tool [35]) indicated this transcriptional response was coordinated by the alternative sigma factors SigH (TFOE output significance of enrichment of the regulon 2.3´10 -10 ) and SigF (pvalue 1.2´10 -7 ), both of which genes were also upregulated in response to ND ( Figure 4B). SigH is activated by oxidation of its antisigma factor RshA, which can be triggered by NO, and by phosphorylation of RshA by the serine-threonine protein kinase, PknB [36]. SigF is associated with cell envelope integrity and is controlled by anti-sigma factor agonists, RsfA and RspB, which sequester the anti-sigma factor RsbW and also by the anti-sigma factor Rv1364c. In M. smegmatis the SigF regulon was induced by respiratory inhibitory conditions, such as might occur in the presence of NO, in a RsfB-dependent manner [37]. The upregulation of sigH (3.82 fold) and sigF (5.79 fold) was accompanied by downregulation of the primary sigma factor, sigA (4.56-fold), and 35 genes from the rpl, rpm and rps families, coding for ribosomal proteins, in accordance with impaired growth in the presence of ND (Table S2).
In contrast to other models of NO exposure [33, 34, 38], the DosR regulon was not induced after ND treatment ( Figure 4B). Moreover, cmr, coding for a repressor of dosR expression [39], was up-regulated. Thus, the transcriptional profiling is consistent with the conclusion from our DosR deletion experiments that transition to the Rpf protein-dependent differentially culturable state after ND exposure does not require DosR .
Mtb is regarded as an obligate aerobe using aerobic respiration to support growth and NO is an inhibitor of respiration [34]. Under stress conditions Mtb can deploy two alternative terminal electron acceptors, nitrate reductase (Nar) and fumarate reductase (Frd). In accord with the lack of induction of the DosR-regulon, narK1 was down-regulated in response to ND and expression of the narG, narH, narI, and narX genes, coding for nitrate reductase, was not significantly changed. In contrast, the frdABCD genes, coding for fumarate reductase, were up-regulated, suggesting utilization of fumarate reduction to maintain redox balance and membrane potential [40]. Interestingly, the serine-threonine protein kinase G (encoded by pknG), which contains a rubredoxin domain with two thioredoxin motifs and has been implicated in the regulation of intracellular NADH concentrations [41], was up-regulated. PknG controls central metabolism by sensing glutamate and aspartate availabilities and regulating activities of α-ketoglutarate dehydrogenase, glutamate dehydrogenase, and glutamine synthase [42]. Glutamate has been previously shown as a major carbon source for mycobacteria exposed to NO in vitro and during growth in macrophages [43]. In addition to these changes in operation of the Krebs cycle, the pta-ackA operon, coding for phosphotransacetylase and acetate kinase, was induced, consistent with impaired aerobic respiration and increased reliance on substrate-level phosphorylation.
Inhibition of bacterial growth by NO is linked to redox chemistry [44]. As a radical, NO itself reacts with transition metals and other molecules with unpaired electrons, such as superoxide. The latter reaction produces peroxynitrite, which can oxidize cysteine thiol to sulfenic acid. Hence, NO perturbs redox balance by damaging ironsulfur clusters and protein thiol groups [38,45]. Accordingly, exposure to ND induced expression of anti-oxidant genes, such as the SigH/AosR-activated non-canonical cysteine biosynthetic genes (mec-cysO-cysM), consistent with increased synthesis of cysteine-based anti-oxidants (e.g. up-regulation of ergothioneine biosynthesis genes; rv3700c and 3704c). In addition, other thiol-related anti-oxidant activities were upregulated, including sulfate-containing compound ABC transporter (cysA1), sulfate adenyltransferase (cysDN), methionine sulfoxide reductases (msrA and msrB), a rhodanese domain protein (rv1674c), a thioredoxin reductase and thioredoxin proteins (trxB2-C, thiX, trxB1). Damage to iron-sulfur proteins resulting in disrupted iron homeostasis was suggested by up-regulation of mycobactin biosynthetic genes (mbtE, mbtG and mbtH) and the Esx-3 genes required for siderophore-mediated iron uptake (rv0282-rv0292). These anti-oxidant responses are consistent with repair of NO-mediated oxidative damage.

Differentially culturable mycobacteria were produced in activated THP-1 cells
Mycobacterial pathogens can survive and replicate in macrophages where they are exposed to NO [50]. The human THP-1 monocytic cell line has been shown to produce NO after exposure to retinoic acid and THP-1 cells treated with retinoic acid and vitamin D3 (RAVD) were proposed as a model for investigation of Mtb persistence [51].
We treated THP-1 cells with either PMA, the typical stimulant for differentiation of these cells into macrophages, or RAVD, and stained them with DAF-FM diacetate to detect NO ( Figure 5A). The number of DAF-FM positive macrophages was significantly higher in samples treated with RAVD compared to untreated and PMAtreated cells (p<0.05, Figure 5A).
Next, either PMA-or RAVD-derived macrophages were infected with M. bovis BCG. CFU and MPN counts of M. bovis (BCG) were determined during a 7-day infection ( Figures 5B, C). In both cases, there was a significant reduction of CFU after

DISCUSSION
During infection pathogenic mycobacteria produce heterogenous populations that differ in their metabolic activity, growth characteristics and tolerance to antimicrobials [8, 13,52]. Amongst these populations, differentially culturable mycobacteria are of particular interest, because they are difficult to detect and eradicate, but frequently represent the dominant Mtb population in clinical TB samples [3-4, 8-12, 53-55]. In the laboratory, DCB are induced in vitro by prolonged incubation in stationary phase [56], gradual acidification [57], alteration of sodium-potassium ratio [26], treatment with antimicrobials [14,58,59] or exposure to vitamin C [60]. Here we describe a new model system based on a novel NO donor (ND) that simulates the in vivo environment by generating a heterogenous population of Mtb thereby permitting investigation of the physiologically distinct states of Mtb produced during the pathogenesis of TB.
In our ND-induced system transition to the differentially culturable state did not require activation of the DosR regulon, but involved two alternative sigma factors, SigH and SigF and their regulons, previously implicated in adaptation to redox stress and respiration-inhibitory conditions [36,37,61]. Overall, the transcriptomic signature of ND-treated DCB was associated with inhibition of cell division, repair of NO-induced damage, metabolic reprogramming to restore redox balance and membrane potential ( Figure 6).
While there is an ongoing debate concerning the importance of NO in controlling Mtb infection in humans, NO is detected in TB patients and our data suggest that it contributes to the generation of DCB by causing the down-regulation of rpfA, rpfB and rpfE expression. Rpf are cell wall cleaving enzymes and their expression under conditions when the peptidoglycan producing machinery is likely to be damaged can be fatal for mycobacteria resulting in uncontrollable cell lysis and death. In addition, NO also has the potential to directly damage Rpf proteins, which require a disulfide bond for correct folding and activity [62]. Hence the simplest explanation for the observed Rpf-dependency of ND-treated mycobacteria is that DCC cells cannot initiate Rpf production when grown in standard media and require supplementation with recombinant Rpf or Rpf-containing CSN to restart growth and peptidoglycan biosynthesis.
Expression of Rpf coding genes is tightly controlled by multiple regulators [63], Fig. S7, including Lsr2 [64] and MtrA [65]. The precise mechanism underpinning the response to ND is unknown.
Exposure to various levels of NO drives Mtb heterogeneity in the infected host ( Figure 6). These heterogenous populations have different growth requirements and respond differently to drugs, may induce differential immune responses and create a unique microbial community (8,13). Our observation that NO down-regulates the expression of rpfA, rpfB and rpfE and generates Rpf-dependent DCB in vitro and intracellularly in macrophages provides the foundation for further investigations to understand the molecular mechanisms that underpin these adaptations and elucidate the precise role of DCB in TB pathogenesis and treatment outcomes.     genes repressed (blue) by ND compared to CC at 4 hours. Significantly differentially expressed genes were identified using a moderated t-test (p<1´10 -5 with Benjamini and Hochberg multiple testing correction) and fold change >2.0. (B) Box and whisker plots showing the distribution of expression ratios (log2) of the SigH regulon [66], aerobic respiration genes functional classification I.B.6.a [67] and DosR regulon [66] after 4 hours exposure to ND or CC. (C) Expression of selected genes was confirmed by qPCR from Mtb exposed to ND relative to CC at 4 hours normalised to 16S rRNA transcript abundance. Data shown as the log2 (relative fold change) ± SEM (N=3).