Macrophages and monocytes belong to the first line of immune defense. During infection by microorganisms, they can undergo physiological changes in order to better eradicate the invading pathogen. One of these changes is known as polarization. Macrophages are able to adapt either an M1 or M2 polarization state (Anderson and Mosser, 2002; Mosser, 2003; Mosser and Edwards, 2008; Rodríguez-Prados et al., 2010; Galván-Peña and O’Neill, 2014; Jha et al., 2015; Shi et al., 2019; Seto et al., 2022). Mycobacterium tuberculosis (M.tb) the causative agent of tuberculosis (TB) is able to induce the polarization state of macrophages. In an in vitro granuloma model, macrophages were able to switch from the M1 to the M2 physiological state upon M.tb infection (Le et al., 2020). Furthermore, M2 macrophages were found to be predominant in granulomas of TB patients, while both M1 and M2 macrophages were found in non-granulomatous lung tissues (Huang et al., 2015). Therefore, the physiological state of macrophages plays a critical role in the pathophysiology of TB (Huang et al., 2019). Polyamines (PAs), such as spermine (Spm) are also able to alter the polarization state of macrophages (Latour et al., 2020). They can be obtained either directly from food and/or synthesized in humans and some bacteria. De novo synthesis of PAs occurs in humans when they absorb arginine from food. The absorbed arginine is converted by arginase to ornithine which is decarboxylated (catalyzed by ornithine decarboxylase, ODC) to the diamine putrescine (Ptc), which is in turn converted by spermidine synthetase [in the presence of decarboxylated-S-adenosyl-L-methionine (Dc-SAM)] to spermidine (Spd). Then, Spm is generated through the carbonylation of Spd by Spd-synthetase in the presence of Dc-SAM (Weaver and Herbst, 1958a,b; Stewart et al., 2018; Sagar et al., 2021). The produced Spm can also undergo an oxidative decarboxylation catalyzed by Spm-oxidase (SPO) to produce again Spd (in addition to hydrogen peroxide and aminopropyl; Weaver and Herbst, 1958a,b; Stewart et al., 2018; Sagar et al., 2021). It is not known how much Spm is produced by infected and uninfected macrophages. However, it was shown that treatment of murine embryonic fibroblasts (MEFs) with 2-difluoromethylornithine, an inhibitor of ODC and diethylnorspermine, an inductor of SPO, led to a decrease in the production of Spm (Puleston et al., 2019). Since Spm has been previously demonstrated to be active against M.tb (Hirsch and Dubos, 1952; Sao Emani and Reiling, 2023), (in addition to its ability to alter the polarization state of macrophages during infection; Latour et al., 2020), it is possible that, upon phagocytosis, M.tb produces enzymes that are able to detoxify Spm, in order to better facilitate its survival in the host cell. Previous studies showed that Streptomyces coelicolor required GlnA3 _sc (SCO6962) for the detoxification of PAs (Krysenko et al., 2017). Furthermore, it was shown that GlnA3 _sc was able to catalyze the glutamylation of putrescine as a possible mechanism of polyamines detoxification (Krysenko et al., 2017). However, we found that the corresponding M.tb ortholog GlnA3 _Mt (Rv1878) was not essential for Spm detoxification (Krysenko et al., 2023). Moreover, GlnA3_sc shares only 40.09% similarities with GlnA3_Mt (by NCBI proteins-proteins amino acids alignment). In addition, the gene encoding the multi-drug efflux pump Rv3065 (de Rossi et al., 1998, 2002) was significantly upregulated during Spm stress in our previous study (Krysenko et al., 2023), while the gene encoding another multi-drug efflux pump Rv1877 (de Rossi et al., 2002; Adhikary et al., 2022) was marginally upregulated (Krysenko et al., 2023). However, since we aimed to follow-up on the investigation of the physiological role of Rv1878 while assessing if Spm is detoxified by multi-drug efflux pumps, we chose to investigate the role of Rv1877 in the context of Spm detoxification and in the context of any overlapping physiological role it may have with Rv1878 since they are encoded by neighborhood genes that are co-transcribed (Harth et al., 2005). In order to determine if tolerance of Spm by M.tb is supported by other efflux pumps, we also investigated the role of Rv0191 which encodes a more specific MFS-type efflux pump (de Rossi et al., 2002; Li et al., 2019).

Our results demonstrate that Rv1877 is able to detoxify Spm while Rv1878 and Rv1877 are involved in the regulation of iron homeostasis, the reconstitution of the cell wall, survival of M.tb in hypoxia and acidic stress. On the other hand, we found that Rv0191 contributes to the survival of M.tb during oxidative stress (OS).

We have recently shown that GlnA3_Mt (Rv1878) was not required for the detoxification of Spm in M.tb (Krysenko et al., 2023), whereas other studies showed that the corresponding S. coelicolor ortholog (GlnA3_Sc) was required for the survival of this actinomycete in excess polyamines (Krysenko et al., 2017). In that same study we nevertheless found that rv1877 encoding a multi-drug transporter (Adhikary et al., 2022) was marginally upregulated during Spm stress (Krysenko et al., 2023). Moreover, genes encoding other multi-drug transporters such as rv3065 were significantly upregulated (Krysenko et al., 2023). Nevertheless, since we wanted to re-evaluate the physiological role of Rv1878, we generated a ∆rv1877 mutant. This was because rv1877 is located upstream rv1878, is co-transcribed with rv1878 and other genes in the cluster (Harth et al., 2005) and they may therefore have overlapping physiological roles though not necessarily during Spm stress. Furthermore, to investigate if tolerance to Spm was facilitated by any efflux pump, or may be specific to multi-drug efflux pumps such as Rv1877 (Adhikary et al., 2022) and Rv3065 (Rodrigues et al., 2013), we generated a ∆rv0191 mutant, where rv0191 codes for an efflux pump which seems not be a multi-drug pump, but is more specific to chloramphenicol (Li et al., 2019). We first examined their transcriptomic profile and found that the ∆rv1877 and ∆rv1878 mutants displayed almost similar transcriptomic profiles (Table 1), as genes encoding enzymes involved in dormancy, hypoxia, C/N, iron and lipid metabolism were upregulated in both strains (see Results section, Supplementary Tables S3, S4), suggesting they could have overlapping functions in these physiological conditions. Overall, the mutants displayed similar transcription profiles (see Results section, Supplementary Tables S3–S5) which could be a transcription signature for all M.tb deletion mutants.

In our previous studies, we showed that Spm was bactericidal against M.tb when tested in 7H9 media, yet with a very high MIC₉₀ of ~5 mM (MIC₅₀ ~ 3 mM), partially due to the conjugation of Spm to albumin found in the supplement of the 7H9 media (Krysenko et al., 2023; Sao Emani and Reiling, 2023). However, it had a much lower MIC in Sauton’s media (that does not contain albumin), yet with a wide range of MIC₉₀ ~ 150-320 μM (MIC₅₀ ~ 80; Krysenko et al., 2023; Sao Emani and Reiling, 2023). These MICs were determined from dose response curves usually performed over 2 weeks, and the broth microdilution (resazurin) assay, which also requires incubation of mycobacteria for at least 7 days (Krysenko et al., 2023; Sao Emani and Reiling, 2023). In the current study, we exposed M.tb for only 3 h to a high concentration (2 mM) of Spm in Sauton’s media. Therefore, the toxic effect of Spm was not expected for such a short incubation, even though it was a high concentration, unless the mycobacteria had lost a protein that supposed to enable it to survive a short burst of Spm stress, which is what we found with the ∆rv1877 mutant (Figure 2A). It was sensitive while the wild-type was not affected (Figure 2A). Suspecting that an extended exposure of the wild-type to Spm stress could affect it at concentrations ≥320 μM, we chose concentrations lower and concentrations higher than the MIC, and exposed the mycobacteria for a longer period (7 days) in a nutrient deprived buffer (PBS). As expected, the wild-type was indeed sensitive to Spm at 500 μM while it was unaffected at 250 μM, and the ∆rv1877 mutant displayed a marginal sensitivity in this case, while its complement survived better than the wild-type (Figure 2). It is worth clarifying that the previously determined MIC₉₀ used the visual scoring of the color change of resazurin in the broth microdilution assay and OD₆₀₀ values measured over time in the dose response curves for about 2 weeks (Krysenko et al., 2023; Sao Emani and Reiling, 2023). But in this study, we used the CFUs-based method, which could explain why, though we observed loss of viability of the wild-type at a concentration (500 μM) higher than its MIC₉₀, it was not a 90% loss of viability (Figure 2). This is expected when comparing different methods. In the case of the CFU-based method in this study, the mycobacteria were first exposed for 7 days, then serially diluted, and plated on 7H11 agar base plates, which were incubated for at least 2.5 weeks before the colonies were counted, giving room for the inhibited (but still viable) mycobacteria to recover, explaining the slight discrepancy between the two methods.

In our previous studies we found that the ∆rv1878 mutant, was not sensitive to Spm stress (Krysenko et al., 2023). Instead, at lower non-toxic concentrations of Spm, it seemed to multiply despite lacking nutrients (Krysenko et al., 2023). Therefore, the physiological role of Rv1877 seems to be antagonistic to that of Rv1878 in this condition (Figure 2). Furthermore, the expression of rv1878 is upregulated in the ∆rv1877 mutant (Table 1; Supplementary Figure S3C). Therefore, the expression of rv1878 is not affected by the in-frame unmarked deletion of rv1877, further showing that the sensitivity of the ∆rv1877 mutant is likely not due to a polar effect on downstream genes (Supplementary Figure S3) but probably due to its non-specific efflux activity. In addition, the homologous recombination templates (inserts) of the constructs used to generate each mutant in this study were sequenced to check for integrity during each stage of the cloning process. Thus, mutation in other genes besides the genes of interest is less likely, yet is possible in some circumstances. Moreover, the reversal of this phenotype in the complemented strain of the ∆rv1877 mutant further supports its role in Spm tolerance (Figures 2A,C).

However, it is worth noting that the difference between the wild-type and the ∆rv1877 mutant is not more than 50% (Figure 2A, though statistically significant). It is probably due to compensation by the other multi-drug efflux pumps such as Rv3065 (Krysenko et al., 2023; and/or similar efflux pumps). It is worth noting that, rv3065 was found to be upregulated in the ∆rv1877 and ∆rv1878 mutants (Table 1; Supplementary Tables S3, S4) in this study. This gene (rv3065) encodes a multi-drug efflux pump (Krysenko et al., 2023) that was found to be the most upregulated during Spm stress and therefore was speculated to be involved in Spm tolerance in our previous studies (Krysenko et al., 2023). Therefore, if the ∆rv1877 mutant still displays a sensitivity towards Spm despite the upregulation of rv3065 (that is supposed to detoxify Spm), then Rv1877 is indeed involved in the detoxification of Spm, though it is not the only protein involved (explaining the low sensitivity). Compensation can also occur by other mechanism of Spm inactivation/detoxification such as acetylation by SpeG as reported in other bacteria (Limsuwun and Jones, 2000; Filippova et al., 2019; Le et al., 2021; Kumar et al., 2022). In our previous studies, we found that M.tb has a SpeG ortholog that is Rv3034c (Sao Emani and Reiling, 2023). In this study, we discovered the upregulation of two genes that may play similar role (rv3535c and rv1323 that code for enzymes possessing acetylation activities (Platt et al., 1995)) in the ∆rv1877 mutant (Supplementary Table S3). All these findings justify the low but significant sensitivity of the ∆rv1877 mutant to Spm stress.

It was previously shown that over-expression of rv1877 in E. coli increased its resistance to a wide range of antibiotics (Adhikary et al., 2022) making it a multi-drug efflux pump, while over-expression of rv0191 in E. coli increased its resistance to only chloramphenicol (Li et al., 2019; making it a more specific efflux pump). Since M.tb is naturally resistant to chloramphenicol supposedly due to the inactivating enzyme chloramphenicol acetyltransferase (CAT; Shaw, 1983; Sohaskey, 2004), it is possible that Rv0191 likely plays a more specific physiological role in M.tb. This has been observed before with other efflux pumps. A few examples include LfrA which seems to be specific for fluoroquinolones (Takiff et al., 1996) and TaP (Rv1258c) that seems to be specific for tetracycline (Ramón-García et al., 2006). On the other hand, others can detoxify a wider range of antibiotics such as the multi-drug efflux pump Mmr (Rv3065; Takiff et al., 1996; de Rossi et al., 1998; Rodrigues et al., 2013). Moreover, some efflux pumps ensure transport or secretion of endogenous metabolites such as the LpqY-SugA-SugB-SugC system that transports trehalose from the cell wall to the cytosol (Kalscheuer et al., 2010; Sharma et al., 2022) and AsnP2 (Rv0346c) that imports asparagine (Gouzy et al., 2014). Therefore, the ability of Rv1877 to enable Spm tolerance could be due to its ability to transport a wide range of compounds (de Rossi et al., 2002; Adhikary et al., 2022), as opposed to Rv0191, which has been described to be more specific, though they are both MFS-type PMF driven transporters (de Rossi et al., 2002; Li et al., 2019). It is worth noting that the gene encoding galactose kinase (galk, rv0620) that was found to be significantly upregulated during Spm stress in our previous studies (Krysenko et al., 2023), was significantly downregulated in the ∆rv1877 mutant in this study (Table 2; Supplementary Tables S3). It is unclear why Spm stress will induce the expression of galK. It could be related to the effect of Spm on energy metabolism (Lüthi et al., 1999; Song et al. 2010; Sao Emani and Reiling, 2023), since galactose is imported into the cell in order to be converted through a series of reaction (one of them catalyzed by GalK) to glucose which produces ATP through glycolysis (Frey, 1996; Solopova et al., 2018). Therefore, if Rv1877 is involved in the import of galactose, the absence of Rv1877 in the corresponding mutant may explain the downregulation of galk. This suggests that Rv1877 may enable M.tb to tolerate Spm through the import of galactose, and therefore through the restoration of energy levels altered by Spm. However, this remains to be shown. On the other hand, Spm tolerance could also occur through the direct transport/export of Spm by Rv1877 and other multi-drug efflux pumps. The anti-mycobacterial activity of Spm and the ability of Spm to enhance the activity of some antibiotics (Sao Emani and Reiling, 2023) make it an attractive substrate for multi-drug efflux pumps. This is supported by previous studies that demonstrated transport of polyamines in eukaryotes by efflux pumps (Sala-Rabanal et al., 2013; Abdulhussein and Wallace, 2014; Moriyama et al., 2020). Eukaryotes are able to synthesize Spm or obtain it through their diet (Pegg, 2009). However, excess Spm can be toxic to eukaryotes (Tabor and Rosenthal, 1956), therefore the high MIC of Spm can cast doubts on the clinical relevance of this study. However, since eukaryotes can already produce Spm, it means they can tolerate a certain level of Spm, as opposed to other antibiotics. Moreover, the introduction of Spm as a food supplement to treat or prevent other disease conditions, further supports the ability of eukaryotes to tolerate an appreciable level of Spm (Senekowitsch et al., 2023). Lastly, because Spm can reduce the MIC of known TB drugs and vice versa (Sao Emani and Reiling, 2023), a high dose may not be required after all, if used in combination with these antibiotics. However, this requires further investigations in vivo, and in clinical settings, which is beyond the scope of this study.

Previous studies have shown that Rv1877, Rv1878 and Rv0191 are not essential for the survival of M.tb in macrophages (Rengarajan et al., 2005), nor in the in vivo mouse model of infection (Sassetti and Rubin, 2003;Harth et al., 2005; Lee et al., 2006). However, this could be because these systems were not suitable for the elicitation of the role of these proteins. For instance, though Rv1878 was found not to be essential for the survival of M.tb in the mouse model of infection (Harth et al., 2005; Lee et al., 2006), it was found to be essential for the survival of M.tb in the primate model of infection (Lee et al., 2006; Dutta et al., 2010). Moreover, since Spm is known to induce the polarization state of macrophages (Huang et al., 2015; Le et al., 2020), it is possible that an in vivo model that presents granuloma features such as the C3HeB/FeJ mice (Kramnik et al., 1998) and/or the guinea pig (Larenas-Muñoz et al., 2023) model of infection and an ex vivo model that uses pre or post-activated macrophages may be suitable for the study of the role of Rv1877, Rv1878 and Rv0191. However, this remains to be shown in future studies. Moreover, it was previously indicated that the ∆rv1878 M.tb mutant was not sensitive to stress conditions such as microaerobic conditions, the utilization of different nitrogen sources, growth at different pH values and high-salt conditions (Lee et al., 2006). Here we expanded the investigation on the role of Rv1878 by characterizing the mutant in other physiological conditions, and observed that the ∆rv1878 mutant was sensitive to acidified 7H9 and the ∆rv1877 mutant was sensitive to acidified Sauton’s media (Figure 6). It was not indicated before at what pH values was the ∆rv1878 mutant tested and what media was used and for how long (Lee et al., 2006). Therefore, the discrepancy between the two results could be at the type of media tested, since in this study, this mutant was not sensitive to acidified Sauton’s. Moreover, it could have been due to the difference in the exposure period to acid stress (as we observed that at later time points, this sensitivity phenotype was lost, Supplementary Figure S7). In this study, we also observed that while two genes involved in nitrogen metabolism were differentially regulated in all strains, which were rv3012c encoding a glutamyl-tRNA(GLN) amidotransferase-subunit C (Wolfe et al., 2010; downregulated) and rv2780 encoding l-alanine dehydrogenase (Giffin et al., 2012; upregulated), more genes in this context were differentially regulated in the ∆rv1877 and ∆rv1878 mutants. These were rv0337c encoding an aspartate aminotransferase (AspC; Jansen et al., 2020), rv0013 encoding a glutamine amido-transferase (Bashiri et al., 2015) and rv3011c encoding a glutamyl-tRNA(GLN) amidotransferase-subunit A which were downregulated in the ∆rv1877 and ∆rv1878 mutants (Supplementary Tables S3, S4). Moreover, few other genes involved in nitrogen metabolism were differentially regulated only in the ∆rv1878 mutant. These were rv1652 (argC), rv1653 (argJ), rv1655 (argD) which are all involved in the biosynthesis of arginine from glutamate (Tiwari et al., 2018) and rv0858c encoding N-succinyldiaminopimelate aminotransferase (DapC; Weyand et al., 2006) which were downregulated. Also, rv3290c (encoding lysine 6-aminotransferase; Mani Tripathi and Ramachandran, 2006) that was shown to be contribute to the ability of M.tb to persist during hypoxia (Duan et al., 2016) was upregulated only in the ∆rv1878 mutant. Moreover, Rv1878 has been experimentally proven to be involved in nitrogen/glutamate metabolism (Harth et al., 2005). Therefore, these findings support the possibility that both Rv1877 and Rv1878 are implicated in nitrogen metabolism. And if that is the case, their phenotype during acid stress is justifiable since proteins involved in nitrogen metabolism have been implicated in the survival of M.tb during acid stress (Gouzy et al., 2014; Gallant et al., 2016). This is because of the release or consumption of ammonia (by these proteins), which is able to reduce/alter the pH of the environment (Gouzy et al., 2014; Gallant et al., 2016). It is intriguing that as opposed to the ∆rv1878 mutant, the ∆rv1877 mutant was sensitive to acidified Sauton’s but was not to acidified 7H9 (Figure 6). It is possible that the missing proteins in these strains play a protective role against acid stress during different environmental conditions: Rv1878, when there is abundance of nutrients (7H9) and Rv1877, when nutrients start depleting (Sauton’s) during the first 3 days of infection. Or because as opposed to 7H9, Sauton’s media does not contain glutamate, but contains asparagine as a C/N source. And the role of Rv1878 in acid stress could have been more pronounced in presence of glutamate (7H9), because it is a glutamine synthetase (Harth et al., 2005) and because few genes implicated in glutamate metabolism were downregulated in the ∆rv1878 mutant as discussed above (Supplementary Table S4), while that of Rv1877 could have been more pronounced in the presence of asparagine (Sauton’s) because an aspartate amino transferase (rv0337c) is downregulated in the ∆rv1877 mutant (Supplementary Table S3) for reasons that remained to be investigated.

On the other hand, Rv1878 seems to play a role in the cell wall modeling of M.tb since the ∆ rv1878 mutant was sensitive to cell wall stress generated by 0.5% SDS (Figure 3). Poly-L-glutamate/glutamine cell wall structure accounts for 10% of the cell wall mass (Imaeda et al., 1968; Wietzerbin et al., 1975). Therefore, the glutamate/glutamine metabolomic function of Rv1878 (Harth et al., 2005) supports its putative role in the reconstitution of the cell wall of M.tb. Moreover, the ∆rv1877 mutant displayed a marginal sensitivity to cell wall stress (Figure 3). However, the lack of sensitivity of the ∆rv0191 mutant to the cell wall stress generated by SDS does not necessarily imply that it does not play any role in the cell wall reconstitution of M.tb. It could be simply be that the role of Rv0191 is not interfered by the effect of SDS. This hypothesis is supported by our previous study, where the ∆cysk₂ mutant was not sensitive to SDS, nor isoniazid (Sao Emani et al., 2022), yet, it was sensitive to vancomycin and displayed an altered cell wall lipid profile (Sao Emani et al., 2022). This is because vancomycin targets a lipid component (phthiocerol dimycocerosate: PDIM; Nieto and Perkins, 1971; Hammes and Neuhaus, 1974; Soetaert et al., 2015) of the cell wall of M.tb (related to the role of CysK₂), that is different to the cell wall lipid component targeted by isoniazid (mycolate; Takayama et al., 1972; Quémard et al., 1991). Moreover, while the expression of almost no gene involved in nitrogen metabolism was altered in the ∆rv0191 mutant, the only two genes which had an altered expression in this context in the ∆rv0191 mutant (including other strains) were rv3012c which encodes a glutamyl-tRNA(GLN) amidotransferase-subunit C that was identified to be a protein of the cell wal (Wolfe et al., 2010; downregulated) and rv2780 which encodes l-alanine dehydrogenase that catalyzes the oxidative deamination of l-alanine to pyruvate that is channeled towards the production of peptidoglycan (Giffin et al., 2012; upregulated). In addition, rv3574 (kstR) that regulates lipid metabolism (Kendall et al., 2007) was also upregulated in all mutants while rv0447c encoding a cyclopropane-fatty-acyl-phospholipid synthase UfaA (Meena and Kolattukudy, 2013) was also downregulated (Supplementary Tables S3–S5) in these mutants. Moreover, the expression of rv2911 that encodes a penicillin-binding protein (DacB2) was downregulated in all mutants and rv3330 that encodes another penicillin-binding protein (DacB1) was downregulated in only the rv1878 mutant (Supplementary Tables S3–S5) while penicillin itself is known to inhibit cell wall synthesis (Yocum et al., 1980). The cell envelope of M.tb consists of a plasma membrane, the myco-membrane, the cell wall and finally a capsule (Daffé et al., 2015). And the cell wall of M.tb consists of peptidoglycan, arabinoglycan, and mycolic acid forming a complex known as mycolyl-arabinogalactan-peptidoglycan (mAGP) complex (Daffé et al., 2015). In view of the finding that rv2780 that is involved the biosynthesis of peptidoglycans (Giffin et al., 2012; a component of the cell wall; Daffé et al., 2015) was upregulated in these strains, while rv3012c, rv2911 and rv3330 that are involved in the synthesis of the cell wall (Yocum et al., 1980; Wolfe et al., 2010) were downregulated in these strains, in addition to the prior knowledge that poly-L-glutamate/glutamine cell wall structure accounts for 10% of the cell wall mass (Imaeda et al., 1968; Wietzerbin et al., 1975), these results all together suggest putative roles of Rv1877, Rv1878 and Rv0191 in the reconstitution of the cell wall of M.tb.

Furthermore, we found that the ∆rv1878 mutant survived better in IS (Figure 5) indicating its role in the regulation of iron homeostasis. The ∆rv1877 and ∆rv0191 mutants displayed the same phenotype but later (5 days). However, these phenotypes were not fully complemented (only partially in the ∆rv0191c, Figures 5A,D). It is either because after a prolonged exposure to IS, a secondary general stress response is triggered, not necessarily related to only IS. Or, because of the general low expression of rv1877 in its complement (Supplementary Figure S8A), or this may be related to the specific role of Rv1877 and Rv0191 during IS, they may play dual functions, enabling M.tb to survive in IS and in excess iron (EI; Figure 5D). However, this remains to be shown. The transcription profile of the mutants revealed dysregulation of many iron-homeostasis-related genes with the highest regulations observed in the ∆rv1877 and ∆rv1878 mutants (Table 3). A few were rv3503c (fdxD; Ortega Ugalde et al., 2018) an iron–sulfur cluster protein, rv0282 encoding a type VII secretion system (Tufariello et al., 2016), and genes involved in the utilization/uptake of iron such as rv2123 encoding PPE37 (Tullius et al., 2019) and rv1037c encoding an ESAT-6-like protein (EsxL; Bukka et al., 2011; Kumar et al., 2013; Jha et al., 2020), rv3841 (brfB; Pandey and Rodriguez, 2012; Reddy et al., 2012; Khare et al., 2017), rv1786 (Choi et al., 2021), rv1177 (fdxC; Ortega Ugalde et al., 2018), rv1636 (Chakraborti et al., 2021), rv1349 (Rodriguez and Smith, 2006), rv3597c (Lsr2; Liu and Gordon, 2012), fdxD (Ortega Ugalde et al., 2018) and many others (Table 3). In addition, many other PE_PPE and PE_PGRS related genes were also differential regulated in these strains (Supplementary Table S3–S5; Table 3). However, it is worth noting that PE proteins are not all involved in the iron homeostasis of M.tb. Most are involved in the host pathogen interaction during infection, others in the secretion of other proteins, others in the regulation of apoptosis and others in the general stress response and many other physiological roles (Ates, 2020; De Maio et al., 2020). Moreover, alteration in the expression of iron-homeostasis-related gene can be induced by various stress conditions such as fatty acid synthesis inhibition, respiration inhibition, ATP synthesis inhibition, oxidative stress, DNA gyrase inhibition and others, as previously demonstrated by compounds generating these conditions (Boshoff et al., 2004). In addition, it was shown that while some PE genes were induced during IS, others were induced during EI (Gold et al., 2001; Rodriguez et al., 2002), which could explain the diversity in their regulation observed in the mutants in this study as some were upregulated while others were downregulated (Supplementary Tables S3–S5; Table 3). Therefore, relying on our RNA sequencing data alone was not enough to delineate the mechanistic roles of Rv1877, Rv1878 and Rv0191 during iron homeostasis. However, comparing our data to previously published (Gold et al., 2001; Rodriguez et al., 2002) transcriptomic profiles of M.tb during EI or IS, revealed the following. The genes mbtA (rv2384) and rv3402c that were previously shown to be upregulated during IS (Gold et al., 2001) are downregulated in the mutants (Table 3; Supplementary Tables S3–S5), suggesting that these mutants are not experiencing IS but the opposite, probably because they stored it in iron storage proteins such bacterioferritin, ferredoxin, since the related encoding genes were upregulated in these strains (Table 3; Supplementary Tables S3–S5). This could explain the accumulated iron levels in the ∆rv1878 mutant (Figure 5C). Further comparison of previously published (Gold et al., 2001; Rodriguez et al., 2002) transcriptomic profile to ours, revealed that rv2526, rv2549c, rv2550c, rv2927c and rv3246 (mtrA) and NADH dehydrogenase (Ndh (rv1854c) but not NuoA (rv3145) in our study) that were shown to be upregulated during EI (Rodriguez et al., 2002), are also upregulated in the mutants (Supplementary Tables S3–S5). These findings further support the theory that these mutants have high intracellular iron levels (Supplementary Tables S3–S5). Lastly, more comparison of previous findings to ours, revealed that rv0464c, rv0465c, rv1169c, rv1184c, rv1520 and monooxygenases (rv2378c, rv0385 and rv0793 in our study but not specifically rv3854c and rv1393c) that were found to be downregulated during EI (Rodriguez et al., 2002), were also downregulated in the mutants of our study (Supplementary Tables S3–S5). It is worth mentioning that there were some few discrepancies (between the previous data (Rodriguez et al., 2002) and ours), but this could be because of other physiological roles of Rv1877, Rv1878 and Rv0191 that may have altered the transcription profile of the respective mutants. An example of these discrepancies was on membrane associated proteins or lipid metabolism related protein, which is reasonable since these Rv1877, Rv1878 and Rv0191 seem to be involved as well in the cell wall reconstitution of M.tb. Nevertheless, these results altogether imply that at least Rv1878 and probably Rv1877 and Rv0191 are implicated in iron homeostasis in relation to the production of iron storage proteins, though the exact mechanism remains to be shown. The study of iron homeostasis in bacteria is complex because while some enzymes enable bacteria to survive in IS, others enable them to thrive in EI and still others fulfil both functions (Khare et al., 2017; Rodriguez et al., 2022; Richardson-Sanchez et al., 2023).

Evaluation of the transcriptomic profile of the mutants revealed many genes that displayed the same regulation across all mutants. However, genes involved the survival of M.tb during hypoxia and dormancy were mostly dysregulated in the ∆rv1877 and ∆rv1878 mutants (see results section, Table 1; Supplementary Tables S3–S5). Moreover, rv1877 and rv1878 were upregulated during hypoxia (Figure 4B) as previously shown (Voskuil et al., 2004). In line with that, we found that the ∆rv1877 and ∆rv1878 mutants were marginally sensitive to hypoxia (Figures 4C,D,F,G). The marginal or low sensitivity of these mutants is unlikely due to our hypoxia system, since it was validated by RT-PCR (Figure 4B) as genes previously shown to be upregulated during hypoxia were also upregulated in ours (Figure 4B). It was also validated by an oxygen meter that showed that it became hypoxic within 3 h (Figure 4A). Finally, it was also validated by the growth phenotype of mycobacteria in the system (Supplementary Figure S6). Therefore, it is possible that the upregulated hypoxia-related genes, compensated for the lack of rv1877 or rv1878, making the respective mutants only marginally sensitive to hypoxia. It is worth noting that, though the ∆rv1877 and ∆rv1878 mutants appeared to be sensitive to hypoxia, backed up by the expression of their respective missing genes during hypoxia (Figure 4) and the dysregulation of hypoxia-related genes in their transcriptomic profile.

Table 1; Supplementary Tables S3, S4, solidifying at least a marginal role of these proteins (Rv1877 and Rv1878) during hypoxia, the phenotypes of the mutants were not fully complemented (Figures 4C,D,F,G). It is either because the experimental conditions do not allow full expression of these genes under the hsp60 promoter as previously observed with other genes (Carroll et al., 2010; Kolbe et al., 2020) or because of the general low expression level of these genes in their complements (Supplementary Figure S8) since the expression of rv1877 and the expression of rv1878 were low while that of rv0191 was ~50-fold higher in their respective complements (Supplementary Figure S8). Therefore, it is possible that rv1877 and rv1878 may require regulatory elements found in their genomic location to ensure optimal expression, and could explain why there was a general low/partial complementation of the ∆rv1877 and ∆rv1878 mutants in most (but not all) phenotypes presented in this study. This is because the plasmid (pMV306hsp., single copy integrative vector) used to complement these strains is integrated at the attP site of the genome of M.tb, where each gene is transcribed independently (Lee et al., 1991; Saviola and Bishai, 2004) as opposed to the wild-type where they are co-transcribed in their original genomic location (Harth et al., 2005), probably subject to necessary regulatory elements. While complementation with pMVhsp60 in the manner presented in this study works for some genes (such as rv0191; Supplementary Figure S8), the expression of other genes is optimal only if they are re-introduced in their genomic location or when an episomal multi-copy plasmid is used.

On the other hand, the mutants show no sensitivity to nitrosative stress generated by TBN (relative to the wild-type, Supplementary Figures S5A–C). Since acr was upregulated under this condition, while the corresponding missing genes of the mutants were not (Supplementary Figure S5D), it is possible that Rv1877, Rv1878 and Rv0191 do not play any role in the defense of M.tb against nitrosative stress. Nevertheless, we previously noticed and discussed that the difference in the species of free radicals generated by various ROS or RNS donors, or the difference in their half-lives and/or structure may result to different effects on mycobacteria (Sao Emani et al., 2019). Ideally, it would be more reliable to test a wide range of RNS and ROS donors, nevertheless, the up regulation of acr (Supplementary Figure S5D), when mycobacteria were treated with TBN, supports its suitability for TB studies investigating the physiological roles of proteins during nitrosative stress, and therefore validates these findings. Nevertheless, the best way to show the clinical relevance of the role of these proteins during nitrosative stress and OS, would be to investigate the survival of the respective mutants in specific mice with altered OS or altered nitrosative stress responses such as the gp91phox^−/− (OS-deficient/phagocyte oxidase-deficient mice; Adams et al., 1997) and the NOS2^−/− (nitrosative stress deficient/inducible nitric oxide synthase deficient-mice; Adams et al., 1997) in future studies.

The ∆rv0191 mutant was the only mutant that was sensitive to OS (Figure 3). It is worth noting that the difference between the wild-type and the ∆rv0191 mutant was not very large, though was statistically significant. This is possibly due to the redundant ROS-detoxification system of M.tb (Wengenack et al., 1999; Chouchane et al., 2000; Buchmeier and Fahey, 2006; Jaeger and Flohé, 2006; Rho et al., 2006; Xu et al., 2011; Nambi et al., 2015; Saini et al., 2016; Sao Emani et al., 2018b,c, 2019), resulting to compensation by other enzymes. Moreover, the transcriptomic profile of the ∆rv0191 mutant revealed no significant alteration in the expression of genes involved in redox-homeostasis. It remains to be shown, if that is the case under OS, if the expression profile of the ∆rv0191 would reveal higher regulation of ROS-related enzymes relative to the wild-type. If that is the case, it is possible that under standard conditions, the ∆rv0191 mutant is not required for basic/intrinsic ROS detoxification, but it becomes important when the strain encounters external OS assault. This phenomenon was also observed in our previous study, where the ∆cysK₂ mutant of M.tb showed no increased levels of ROS under standard culture conditions, yet was sensitive to OS relative to the wild-type (Sao Emani et al., 2022). While, in another study, the mycothiol-deficient ∆mshA M.tb mutant, could not grow on agar plates that did not contain catalase (Sareen et al., 2003; Xu et al., 2011; Sao Emani et al., 2018b), indicating its requirement for basic redox homeostasis of M.tb (den Hengst and Buttner, 2008), though mshA (rv0486) is not upregulated during oxidative stress (Namouchi et al., 2016). Therefore, it is possible that some ROS-detoxification enzymes are essential to maintain a balanced redox state even under standard conditions, while other enzymes come to play only when the mycobacteria are experiencing unusual and elevated ROS assaults. This could be related to the mechanism of ROS-detoxification of the specific enzyme. In case of MshA, it is because it is the only enzyme that catalyzes the first step of mycothiol biosynthesis. Mycothiol is a low molecular weight thiol (LMWT), that is able to detoxify a wide range of ROS, RNS and other toxins, including some antibiotics (Wang et al., 2015; Sao Emani et al., 2019). In the case of CysK₂ (Rv0848), it is thought to be because it catalyzes the synthesis of cysteine-sulfate which serves as a signaling metabolite, that is able to activate the production of other molecules, when mycobacteria encounters stress conditions (Steiner et al., 2014). In case of Rv0191, it is possible that it enables the transport of LMWT. Therefore, as seen with genes involved in LMWT biosynthesis (including mshA), their anti-oxidative roles are not depicted at the transcriptomic profile but at their metabolomic profile of M.tb. This could explain why, the expression of rv0191 is not altered under OS as opposed to cysK₂ (Supplementary Figure S4) whose expression is altered by various stress conditions (Voskuil, 2004; Provvedi et al., 2009; Vilchèze et al., 2013; Kurthkoti et al., 2017) including when M.tb loses a gene that may affect its fitness, as seen in this study (Supplementary Tables S3–S5; Results section), because it catalyzes the synthesis of a stress signaling molecule. Therefore, since the transcriptomic profile of the ∆rv0191 mutant under standard growth conditions (Supplementary Table S5) did not give us a hint on its mechanistic role in the defense of M.tb against OS, a targeted metabolomic profile of intracellular and extracellular LMWT, under standard and OS stress conditions coupled with a proteomic and transcriptomic profile of the ∆rv0191 mutant during OS may shed light on the actual mechanistic role of Rv0191 during OS. The little information we could obtain that support the putative role of Rv0191 during OS, is the fact that it is located upstream a gene encoding an oxidoreductase (rv0197) that was shown to be upregulated during OS (Voskuil et al., 2011). Moreover, rv0192A, which is small gene overlapping the DS region of rv0191 encodes for a protein that is able to indirectly interact with MshA according to the string database.³ The role of efflux pumps in the protection of M.tb against OS is still an underexplored field. This is because, it was believed that the defense against OS, relied solely on ROS-detoxifying enzymes and LMWT (Voskuil et al., 2011; Sao Emani et al., 2013, 2018a,d, 2019). It was only recently that it was shown that LMWT could be secreted (Sao Emani et al., 2013, 2018d), thereby indicating that M.tb has both extracellular and intracellular ROS-defense mechanisms. It was shown that Salmonella enterica is able to secrete siderophore products through the MacAB efflux pump as a defense mechanism against oxidative stress (Bogomolnaya et al., 2020). Therefore, it is also possible that M.tb uses efflux pumps to secrete LMWT to detoxify extracellular ROS and RNS. To the best of our knowledge, only two studies, have reported the role of efflux pumps, p55 (Rv1410c; Ramón-García et al., 2009), Rv1258c (Sun et al., 2024) during OS in M.tb. Therefore, our results further support the possible role of specific efflux pumps during oxidative stress.

In brief, we have shown for the first time, that the multi-drug efflux pump Rv1877, enables M.tb to tolerate excess spermine. Furthermore, we identified a physiological role of Rv1878 during iron starvation and cell wall stress and the roles of Rv1877 and Rv1878 during hypoxia and acidic stress. Finally, we demonstrated for the first time that Rv0191 plays a role during oxidative stress.

Raw sequencing data supporting the conclusions of this article has been deposited as a collection at figshare: Sao Emani, Carine; Reiling, Norbert (2024). The efflux pumps Rv1877 and Rv0191 play differential roles in the protection of Mycobacterium tuberculosis against chemical stress. figshare. Collection. https://doi.org/10.6084/m9.figshare.c.7001352.v1.

CS: Conceptualization, Data curation, Investigation, Methodology, Validation, Writing – original draft, Writing – review & editing. NR: Funding acquisition, Project administration, Resources, Supervision, Visualization, Writing – review & editing.