In Mycobacterium abscessus, the Stringent Factor Rel Regulates Metabolism but Is Not the Only (p)ppGpp Synthase

ABSTRACT

The stringent response is a broadly conserved stress response system that exhibits functional variability across bacterial clades. Here, we characterize the role of the stringent factor Rel in the nontuberculous mycobacterial pathogen, Mycobacterium abscessus (Mab). We found that deletion of rel does not ablate (p)ppGpp synthesis and that rel does not provide a survival advantage in several stress conditions or in antibiotic treatment. Transcriptional data show that Rel_Mab is involved in regulating expression of anabolism and growth genes in the stationary phase. However, it does not activate transcription of stress response or antibiotic resistance genes and actually represses transcription of many antibiotic resistance genes. This work shows that there is an unannotated (p)ppGpp synthetase in Mab.

IMPORTANCE In this study, we examined the functional roles of the stringent factor Rel in Mycobacterium abscessus (Mab). In most species, stringent factors synthesize the alarmone (p)ppGpp, which globally alters transcription to promote growth arrest and survival under stress and in antibiotic treatment. Our work shows that in Mab, an emerging pathogen that is resistant to many antibiotics, the stringent factor Rel is not solely responsible for synthesizing (p)ppGpp. We find that Rel_Mab downregulates many metabolic genes under stress but does not upregulate stress response genes and does not promote antibiotic tolerance. This study implies that there is another critical but unannotated (p)ppGpp synthetase in Mab and suggests that Rel_Mab inhibitors are unlikely to sensitize Mab infections to antibiotic treatment.

INTRODUCTION

Bacteria must adjust their physiology to permit survival in fluctuating conditions. The stringent response is a conserved signaling system that promotes survival of many species in stress and antibiotics by altering the transcription of about a quarter of the genome (1–5). In this work, we profile the role of Rel, the sole annotated stringent factor, in the nontuberculous, rapidly growing Mycobacterium abscessus (Mab). Mab is an opportunistic pathogen that lives in the environment and causes skin and respiratory infections that are increasingly prevalent in cystic fibrosis patients (6). Mab infections are difficult to treat due to intrinsic resistance to many antibiotics (7) and high tolerance under stress to almost all antibiotics tested (8, 9). One proposed strategy to help treat such antibiotic-recalcitrant infections is to inhibit regulatory systems, like the stringent response, which promote antibiotic tolerance (10–13).

The conserved aspect of the stringent response is the synthesis, upon stress, of the hyperphosphorylated guanine (p)ppGpp. Once made, (p)ppGpp affects transcription in different ways (14–17) and also directly modulates replication (18, 19), nucleotide metabolism (20–22), ribosome maturation (23, 24), and translation (25–27).

There are a few different protein families that synthesize (p)ppGpp across bacterial clades. The most widely conserved are the Rel/Spo homolog or RSH proteins, which contain, from N terminus to C terminus, a (p)ppGpp hydrolase domain, a (p)ppGpp synthase domain, and regulatory TGS (threonyl-tRNA synthetase/GTPas/SpoT) and ACT (aspartate kinase/chorismate mutase/TyrA) domains (28). Some RSH proteins associate with the ribosome and sense amino acid starvation by detecting when ribosomes have stalled due to an uncharged tRNA being in the A site (29, 30). Some RSH proteins detect other types of stress or nutrient deprivation via other mechanisms (31–34). Many species have only one RSH-type protein, which is competent for both (p)ppGpp hydrolysis and synthesis (28). Many species encode small alarmose synthetase (SAS) and small alarmone hydrolase (SAH) proteins, which contain only (p)ppGpp synthetase and hydrolase domains, respectively (30, 35–38). Because SAS proteins do not have regulatory domains, they are typically controlled transcriptionally (30). There are a few other (p)ppGpp synthesizing domains that have been preliminarily studied. ToxSAS proteins are part of toxin-antitoxin systems and synthesize (p)ppGpp as a toxin; they are found in both phage and bacterial genomes (39). Streptomyces antibioticus has a polynucleotide phosphorylase (PNPase), which can also synthesize (p)ppGpp (40–43).

The physiological outputs of the stringent response vary across species, but there are conserved themes. First, the stringent response generally downregulates genes required for growth, such as ribosome and cell wall synthesis factors, and alters transcription of central metabolism to prioritize survival rather than construction of new cells (2, 3, 17, 44). Stringent inhibition of growth (44–50) indirectly protects against some stresses and antibiotics that interfere with growth factors. In many species, the stringent response upregulates stress response genes such as heat shock proteins, hibernation factors, and stress-specific transcription factors (3, 51, 52) and promotes survival in stress (13). The stringent response also helps many bacteria survive through antibiotic treatment by promoting antibiotic tolerance (12).

The stringent response has been studied in Mycobacterium tuberculosis (Mtb) and Mycobacterium smegmatis (Msmeg). Mtb has one major (p)ppGpp synthetase, Rel_Mtb, which makes (p)ppGpp when respiration is inhibited, in the stationary phase, and in total carbon and nitrogen starvation (46, 53). Rel_Mtb also promotes survival of Mtb during nutrient and oxygen starvation, stationary phase (46), and chronic infection of mice (1) and guinea pigs (54, 55). Importantly, Rel_Mtb also makes Mtb more tolerant to the first-line clinical antibiotic isoniazid during starvation and infection in mice (11). A PNPase enzyme in Mtb, Rv2783, has been shown to have weak (p)ppGpp synthetase activity (56), and the Mtb genome contains another gene with a predicted GTP pyrophosphokinase domain, Rv1366. However, the Δrel_Mtb strain does not synthesize measurable (p)ppGpp in starvation conditions (46), so these other potential synthetases are either inactive or not activated under the conditions tested.

Msmeg has an RSH protein, Rel_Msmeg, which can both synthesize and hydrolyze (p)ppGpp (57–60). Although the Δrel_Msmeg strain showed defects in biofilm formation and stationary-phase viability, it can still synthesize (p)ppGpp (57, 61). The secondary (p)ppGpp synthetase, RelZ_Msmeg, has an RNaseHII domain in addition to the conserved SAS ppGpp synthetase domain (59) and also synthesizes pGpp (62). Strains missing both synthetases have further biofilm and aggregation defects but can still synthesize some (p)ppGpp (62), indicating that there is a third, uncharacterized synthetase in Msmeg.

In this study, we examined the Δrel strain of Mab. Rel_Mab is an RSH protein. There are no other RSH genes and no SAS genes in the Mab genome. We found that the Δrel_Mab strain still makes (p)ppGpp. The Δrel_Mab strain does not exhibit survival defects in several stress conditions including antibiotic treatment but has a growth defect relative to wild type. We measured transcriptional changes in Δrel_Mab relative to wild type and found that it helps downregulate many metabolic pathways in the stationary phase.

RESULTS

In order to explore the role of rel_Mab, we built a strain of Mab ATCC 19977 with a deletion of the rel gene (MAB_2876), which has the canonical RSH gene structure, including a (p)ppGpp hydrolase domain, a (p)ppGpp synthetase domain, a TGS domain, and an ACT domain. We measured (p)ppGpp in both the wild-type Mab and Δrel_Mab strains, in logarithmic phase and carbon starvation (Fig. 1). We found that both the wild-type and Δrel_Mab strain produce (p)ppGpp in both the logarithmic phase and starvation, and the amount of ppGpp is increased in both strains in starvation.

FIG 1.

Thin-layer chromatography (TLC) of guanine nucleotides extracted from ³²P-labeled M. abscessus strains from logarithmic phase and starvation.

In many species, rel orthologs promote survival during stress conditions, such as stationary phase, acid stress, starvation, or oxidative stress (5, 35, 46, 49, 63, 64). To evaluate the physiological role of rel in Mab, we assayed survival of wild type, Δrel_Mab, and complemented strains upon and after transfer to either carbon starvation (Fig. 2A), salt stress (Fig. 2B), oxidative stress (Fig. 2C), or acidic media (Fig. 2D). Treatment with these stressors did not induce measurable differences in growth or survival of Δrel_Mab relative to wild type and the complemented strain. Thus, Rel_Mab does not regulate responses to these stresses under the conditions tested or at least not enough to affect growth or survival. We also found that Rel_Mab does not promote survival in the stationary phase (Fig. 3A). Robust survival in the Δrel_Mab in all stress conditions is likely aided by the continued synthesis of (p)ppGpp under stress (Fig. 1). Growth curves showed that Δrel_Mab had a significant growth defect relative to wild type and complemented strains (Fig. 3B).

FIG 2.

Contribution of rel_Mab to survival in various stresses. (A) CFU of wild-type M. abscessus strain ATCC 19977 (WT), the Δrel_Mab strain, and the complemented Δrel_Mab L5::rel_Mab strain in Hartman’s du Bont medium with no glycerol and tyloxapol as a detergent. (B) CFU in 7H9 Middlebrook medium with a pH of 4. (C) CFU in Lennox LB with 1 M NaCl. (D) CFU in Hartmans du Bont medium with 5, 25, or 60 mM tert-butyl peroxide after 24 h. Relative CFU is calculated by taking the ratio between each CFU value and the initial CFU value at time zero. All data points are averages from three biological replicates. Error bars represent standard deviation. There are no significant differences in any of these data by a two-tailed Student’s t test.

FIG 3.

Growth and stationary-phase survival of Δrel_Mab. (A) CFU in the stationary phase in 7H9 medium. The 2-day time point is 48 h after diluting logarithmic phase cultures to optical density at 600 nm (OD₆₀₀) = 0.05. (B) CFU during logarithmic phase growth in 7H9 medium. The graph is set at a log₂ scale. Doubling times ± 95% confidence interval are as follows: wild type (WT), 2.15 h ± 0.08; Δrel_Mab strain, 4.67 h ± 4.26; Δrel_Mab L5::rel_Mab, 2.84 ± 1.85. The P values are for the wild type compared to Δrel_Mab. are as follows: for t = 4, P = 0.0415; for t = 7, P = 0.00015; for t = 10, P = 0.006; and for t = 12, P = 0.00029. The P values between wild type and the complemented strain were not significant. The P values between strains in the stationary phase were not significant (data not shown). Asterisks represent significance as measured by the two-tailed Student’s t test: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; NS, P > 0.05.

Because the stringent response is a major activator of antibiotic tolerance and persistence in many species (5, 65, 66), we sought to assess how Rel contributes to antibiotic tolerance in Mab. First, we treated Mab cultures in logarithmic phase with amikacin, clarithromycin, and cefoxitin, which are commonly prescribed to treat Mab infections (67). We found that clarithromycin and cefoxitin alone were ineffective against all of the strains (Fig. 4A). However, amikacin treatment resulted in 10- to 100-fold decreases in the viability of wild-type and complemented strains but had no effect on Δrel_Mab.

FIG 4.

Contribution of rel_Mab to survival in antibiotic treatment. Relative CFU of strains treated with either 200 μg/mL of clarithromycin (Clar), 150 μg/mL of amikacin (Amk), or 80 μg/mL of cefoxitin (Fox) for either (A) 48 h in logarithmic phase or (B) 72 h in the stationary phase. Relative CFU were calculated by taking the ratio between the CFU value after treatment and the initial CFU value at time zero. The bars represent the mean of six to nine biological replicates; the individual values are shown by the dots. Error bars represent standard deviation. The logarithmic phase P values were as follows: (Amk150) WT versus Δrel_Mab strain, 0.005; WT versus Δrel_Mab L5::rel_Mab strain, 0.01; Δrel_Mab versus Δrel_Mab L5::rel_Mab strain, 0.004. The stationary phase P values were as follows: (AMK150) WT versus Δrel_Mab strain, 0.00017; WT versus Δrel_Mab L5::rel_Mab strain, 0.0217; Δrel_Mab versus Δrel_Mab L5::rel_Mab strain, 0.0017. Asterisks represent significance as measured by the two-tailed Student’s t test: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; NS, P > 0.05.

Antibiotic tolerance increases in the stationary phase in most bacterial species relative to logarithmic phase (5, 68). We repeated the antibiotic survival experiments on cultures in stasis and found that Rel_Mab does not affect tolerance to clarithromycin or cefoxitin (Fig. 4B). Similar to amikacin treatment in growth, Rel_Mab increased susceptibility relative to wild type and the complemented strains in stasis.

A major function of the stringent response in other bacteria is to regulate transcription (4). To determine the effects of Rel_Mab on transcription, we compared the transcriptome of wild type and Δrel_Mab in both the logarithmic and stationary phases using RNA-Seq. We found that Rel_Mab represses many more genes than it activates (see File S1 in the supplemental material).

In logarithmic phase (optical density [OD] = 0.5), when Δrel_Mab grows more slowly relative to wild type (Fig. 3B), we found 150 genes repressed by rel_Mab in the wild-type strain at least 3-fold, and only 7 genes were activated. The only annotated upregulated genes are an efflux pump (MAB_0677) and a major facilitator superfamily (MFS) transporter (MAB_0069). We found several mce family genes that were repressed by rel_Mab (Table S2). Mce proteins are typically lipid transporters, but they also play roles in host cell entry and immune modulation (69). We also found two antibiotic resistance genes that are repressed by rel_Mab in logarithmic phase (Table 1).

TABLE 1.

Antibiotic resistance genes (under whiB7 regulon)^a

Mab gene annotation	FC-ΔrelMab vs WT Log	FDR-corrected P value ΔrelMab vs WT Log	FC- ΔrelMab vs WT Stat	FDR-corrected P value ΔrelMab vs WT Stat	FC-WT Stat/WT Log	FDR-corrected P value WT Log vs WT Stat	Mab GO molecular function
MAB_0163c	+2	1.1E−04	+40	5.1E−12	+2	1.7E−311	Aminoglycoside phosphotransferase
MAB_0185c	+1	0.3	+5.4	6.4E−03	−5.4	0	Arabinosyl transferaseb
MAB_0186c	+1	0.8	+9.7	1.8E−04	−7.4	0	Arabinosyl transferaseb
MAB_1341	+1	0.45	+34	2.3E−11	−1.3	5.5E−13	Decarboxylaseb
MAB_1342	+1.4	0.03	+14	0	−1	0	Acyl coenzyme A synthetaseb
MAB_1395	+2.7	1.3E−04	+48	5.2E−10	+1	7.2E−29	Transporterb
MAB_1396	+2.5	3.7E−06	+36	0	−1	0	Multidrug MFS transporter
MAB_1846	−1.3	0.44	+28	4.2E−06	−1.4	2.9E−265	ABC transporterb
MAB_2273	+2.3	8.8E−12	+101	1.96E−08	+2.2	0	MFS transporterb
MAB_2297	+1.5	0.02	+99.2	7.1E−08	−2.1	0	Methyltransferase-erm41b
MAB_2310	+1.3	0.5	+5.7	0.03	+3.6	1.5E−50	Multidrug transporter
MAB_2355c	+2.2	1.5E−08	+17	0	+2.8	0	ABC transporterb
MAB_2396	+2.1	8.2E−03	+18	7.3E−09	+1.5	4.5E−122	Probably acetyltransferaseb
MAB_2640c	+1.2	0.122	+5	5.4E−03	−2.4	0	Mmr (multidrug transport integral membrane protein)
MAB_2736c	+1	0.6	+13	1.99E−10	−3.7	0	ABC transporter
MAB_2780c	+1.7	0.01	+27	1.14E−07	+3.3	0	MFS transporterb
MAB_2807	−1	0.7	+5	4.4E−03	−4.7	0	MFS transporter
MAB_2875	+5.4	9.7E−06	+38	0	+1.2	1.6E−171	β-Lactamase
MAB_2989	+2	5.1E−04	+6.8	2.9E−05	+2.93	1.1E−222	Chloramphenicol acetyltransferase
MAB_3042c	+2.7	1.9E−12	+24	0	+1.82	0	GTpase-Hflxb
MAB_3467c	+6	2.5E−03	+21	0	+92	0	Heat shock proteinb
MAB_3508c	+1.8	0.3	+31	0.08	+14	0.38	WhiB7
MAB_3762	+2	2.4E−09	+11	2.9E−10	+7.09	4.9E−143	Membrane proteinb
MAB_3869c	−1.3	0.158	+6.7	5.4E−03	−1.67	0	DNA-directed RNA polymeraseb
MAB_4294	+1.8	3.1E−03	+28	0	+1.88	0	Aminotransferaseb
MAB_4395	+2.4	0	+8	8.8E−07	+1.1	0	Aminoglycoside-2′-N-acetyltransferase
MAB_4837	+4.6	0	+26	0	+1.84	1.9E−312	Aminoglycoside phosphotransferase

^aFDR, false discovery rate; Log, logarithmic; Mab, Mycobacterium abscessus; Stat, stationary phase; WT, wild type.

^bUnder the whiB7 regulon.

Even though there was no apparent difference in survival between the wild-type and Δrel_Mab strains in the stationary phase, we observed significant differences in transcription. We isolated RNA from stationary phase cultures that were shaken for 48 h after being diluted to an optical density of 0.05. We found hundreds of genes that were repressed by rel_Mab in the wild-type strain in the stationary phase, but none that were activated by rel_Mab 3-fold or more. We found many genes in the WhiB7 regulon (Table 1) that are repressed in the stationary phase, although they are mostly unaffected in logarithmic phase. WhiB7 is a transcription factor that activates many antibiotic resistance genes and promotes resistance to many classes of antibiotics in Mab (70). It is notable that these antibiotic resistance genes are repressed by rel_Mab in stasis, which would imply that wild-type Mab would be more susceptible to antibiotics under this condition, which is what we see in amikacin treatment. In the case of clarithromycin and cefoxitin, increased tolerance through downregulation of target expression may counterbalance the repression of the antibiotic resistance genes, resulting in no differences in susceptibility in our assays (Fig. 4).

We also found several cell wall biosynthetic genes that are downregulated by rel_Mab in the stationary phase (Table S2). Downregulation of growth factors is typical in stringent responses across many bacterial species (2, 3, 47, 66, 71). We see that rel_Mab downregulates many central metabolism genes in the stationary phase (Fig. 5; Table S3), which shows that when Rel is present, it contributes to the decrease in central carbon metabolism during stress. However, it is notable that not all the genes in a given pathway are downregulated equally. We hypothesize that this uneven regulation of certain pathways may allow certain metabolites to accumulate and be redirected to other pathways. In a metabolic pathway in which most of the enzymes are downregulated, we expect that the product of a single gene that is not downregulated will accumulate. We note that many of the metabolites we expect to accumulate converge on the NAD synthesis pathway. In addition, none of the genes in the NAD synthesis pathway are downregulated by rel_Mab, which implies that continued metabolism of NAD, which is a critical cofactor in many pathways, may be important in the stationary phase. From our preliminary analysis, it is clear that Rel_Mab helps regulate growth and central metabolism and affects expression of antibiotic resistance genes; however, it does not seem to upregulate specific stress responses under the conditions tested.

FIG 5.

Repression of central metabolic genes by Rel_Mab during stationary phase. Genes in red are downregulated by Rel_Mab at least 3-fold in the stationary phase, P < 0.05. Genes in black are not significantly regulated by Rel_Mab in the stationary phase. See Table S4 for data. TCA, tricarboxylic acid.

DISCUSSION

Mtb, Msmeg, and Mab all have a stringent response, but they have evolved use of different sets of genes to synthesize the same alarmone. The Mab genome encodes only one annotated RSH gene, rel_Mab (MAB_2876), and no homologs of SAS or SAH genes. rel_Mab is homologous to RSH genes in Mtb and Msmeg. We found that the Δrel_Mab strain could still produce (p)ppGpp, consistent with the presence of an unannotated (p)ppGpp synthase. Mab does not have a homolog of the SAS-RNaseHII relZ, which has been characterized in Msmeg (57, 59, 62). Despite this, Msmeg and Mab are similar in having significant (p)ppGpp synthesis in the absence of the RSH gene (Fig. 1) (53) and in exhibiting greater survival with some antibiotics (Fig. 4) (61, 68). Mab does have a homolog of the PNPase, which has been shown to be capable of (p)ppGpp synthesis in Mtb (MAB_3106c) (56). MAB_3106c is transcriptionally upgregulated 4-fold in the Δrel strain compared to wild-type in the stationary phase, indicating that the presence of Rel may decrease the need for transcription of this gene. In the wild-type strain, this gene is also upregulated ∼7-fold in the stationary phase compared to the logarithmic phase, indicating that this gene is used in stress. We therefore propose MAB_3106c as a possible additional (p)ppGpp synthase.

Our results show that the stringent factor Rel_Mab does not promote survival during in vitro stress, although it does promote growth during logarithmic phase (Fig. 2 and 3A). Our transcriptomics analyses corroborate these results, as they indicate that Rel_Mab does not upregulate stress response genes but does alter transcription of growth metabolism genes (see Table S2 and File S1 in the supplemental material). It is possible that increases in (p)ppGpp stimulate transcription of stress response genes in Mab, as is seen in other species (3, 52, 72, 73); our experiments did not test this because in our assays, Δrel_Mab synthesizes amounts of (p)ppGpp comparable to those of wild type (Fig. 1). The unannotated ppGpp synthase may contribute to stress response signaling. Our transcriptional data show that Rel_Mab is involved in downregulating metabolism for growth arrest (Fig. 5; Fig. S2), even though (p)ppGpp levels are still present in the Δrel_Mab strain. There is precedent for metabolic genes being regulated separately from stress response genes in the stringent response. In Escherichia coli, some metabolic genes are repressed at a lower (p)ppGpp level than is required to induce a regulon of stress-response genes (74). (p)ppGpp likely does not directly bind RNAP in mycobacteria as it does in E. coli (75), and it is currently unknown how it exerts its effects on transcription. Further knowledge of the functioning of the mycobacterial stringent response will be required before we can speculate about the mechanism by which (p)ppGpp could differentially affect transcription of metabolic and stress response genes in Mab.

Mab is notorious for having resistance to many clinical antibiotics and expressing many antibiotic resistance genes (70, 76, 77), which is why it is problematic to treat infections in cystic fibrosis patients (78). We observed in our transcriptional data that Rel_Mab downregulated numerous antibiotic resistance genes in the stationary phase (Table 1). In other species, RSH proteins promote antibiotic tolerance (5, 11, 66) and sometimes also increase expression of antibiotic resistance genes (79, 80). Studies are ongoing to find drugs that would inhibit (p)ppGpp synthesis by RSH proteins (11, 13), as such drugs are expected to increase susceptibility to clinically available antibiotics. Our results indicate that Rel_Mab inhibitors, should they become available, are unlikely to help treat Mab infections. Instead, efforts should focus on identifying the other (p)ppGpp synthetase(s) in Mab so that it can be explored as a drug target.

MATERIALS AND METHODS

Construction of strains.

Primers 1233 to 1238 (see Table S3 in the supplemental material) were used to amplify a 502-bp segment upstream of Mab rel, which included the start codon, a 448-bp segment downstream of Mab rel that included the stop codon, and a 788-bp ZeoR cassette. All three segments were stitched together by PCR to form the Δrel::zeoR double-stranded recombineering knockout construct. Δrel_Mab was generated through double-stranded recombineering, as previously described (81) (Fig. S1C). Colonies from the transformation of the Δrel::zeoR construct were PCR-checked by using primers 761 and 1424, primers 1235 and 1236, and primers 762 and 1425 (Fig. S1A). To make the complemented strain, the rel_Mab gene was amplified through PCR using primers 1329 to 1330 and inserted into pKK216 (82) with NdeI and HindIII. This new plasmid, pCB1248, was transformed into the Δrel_Mab mutant strain in order to create the complementation strain, Δrel_Mab L5::rel_Mab, in which rel_Mab expression is driven by a constitutive promoter (BN17, Fig. S1B).

Media and culture conditions.

All M. abscessus ATCC 19977 wild-type and mutant cultures were started in 7H9 (Becton, Dickinson, Franklin Lakes, NJ) medium with 5 g/liter bovine serum albumin, 2 g/liter dextrose, 0.85 g/liter NaCl, 0.003 g/liter catalase, 0.2% glycerol, and 0.05% Tween 80 and shaken overnight at 37°C until cultures entered the logarithmic phase. For starvation and other specific assays, Hartmans de Bont (HdB) minimal medium was made as described previously (83). The cultures were inoculated to an OD₆₀₀ of 0.05, unless otherwise stated. All CFU time points were plated on Luria-Bertani (LB) agar and placed in a 37°C incubator for 4 days.

Mab (p)ppGpp extraction and detection.

Mab strains were grown until logarithmic phase (OD₆₀₀ = 0.5) in homemade 7H9 medium. To maximize ³²P labeling, 7H9 medium with 1/25 of normal phosphate levels was used (46). Logarithmic phase Mab cells were pelleted and resuspended in 1 mL of low phosphate 7H9, followed by the addition of ³²P-labeled orthophosphoric acid to a final concentration of 100 μCi/mL. ³²P-labeled cells were then incubated at 37°C and shaken at 200 rpm for 4 h. Following incubation, the cells were pelleted and resuspended with 100 μl of TBST (Tris-buffered saline, pH 8, with 0.05% tyloxapol) and treated with 1 mg/mL of lysozyme on ice for 20 min. Equal volume 2 M formic acid was then added to each sample, followed by a 5-min centrifugation at maximum speed. Supernatant was then collected for each sample and stored in −20°C. For starvation cultures, a second set of logarithmic phase Mab strains (OD₆₀₀ = 0.5) were washed and resuspended in TBST. ³²P-labeled orthophosphoric acid was added, and (p)ppGpp extraction was performed similarly as logarithmic phase cultures. Following extraction, 20 μl of extract from all cultures were spotted on polyethyleneimine (PEI)-cellulose thin-layer chromatography (TLC) plates and developed in 1.5 M potassium monophosphate buffer (pH 3.4). The plates were then air dried and placed on a phosphor screen (Molecular Dynamics) overnight. The phosphor screens were scanned with a Storm 860 scanner (Amersham Biosciences), and the images were analyzed with ImageQuant software (Molecular Dynamics).

Δrel_Mab stress assays.

For all stress assays, strains were prepared and grown into logarithmic phase. Unless otherwise stated, cultures for stress assays were done in 24-well plates and shaken at 130 rpm at 37°C. For carbon starvation, the strains were inoculated in 30 mL inkwells in HdB minimal medium with no glycerol and with tyloxapol as a detergent. For acid stress, the strains were inoculated in 7H9 medium (pH 4). For osmotic stress, the strains were inoculated in LB broth medium with 1 M salt (ACS sodium chloride, VWR Chemicals BDH). For oxidative stress, all strains were inoculated in complete HdB minimal medium, which does not contain catalase, and strains were exposed to different concentrations of tert-butyl hydroperoxide (Alfa Aesar). CFU time points were taken upon inoculation and at 1, 3, and 24 h postinoculation.

Δrel_Mab growth curve and stationary-phase survival.

Logarithmic-phase cultures of all strains were inoculated to an OD of 0.05 in 30-mL inkwells in 7H9 medium. Cultures were then placed in shaking incubator at 37°C and 130 rpm. CFU time points were then taken throughout a 12-h period. For stationary-phase survival, a second set of cultures were grown into stationary phase up to 48 h. Initial CFU time point was taken at 48 h after dilution of logarithmic phase samples to an OD₆₀₀ of 0.05, with subsequent time points taken at 5, 6, 7, 8, 9, and 10 days.

Antibiotic assays.

For the logarithmic phase experiments, the strains were kept in logarithmic phase in 7H9 for ∼24 h and diluted to an OD₆₀₀ of 0.05 and treated with either 150 μg/mL of amikacin, 200 μg/mL clarythromycin, or 80 μg/mL of cefoxitin. These antibiotic concentrations were chosen because they were shown in a previous study to kill Mab in certain conditions (9). CFU were measured upon treatment (t = 0) and 48 h after treatment (t = 48). For the stationary phase, logarithmic-phase cells at an OD₆₀₀ of 0.05 were shaken for 48 h and then treated as above. CFUs were measured upon treatment and 72 h after treatment.

RNA isolation, library preparation, and data analysis.

RNA from three biological replicates of each strain and condition was isolated as previously described (84) with some modifications. After growth for ∼24 h in either logarithmic or stationary phase, the cells were transferred to 15-mL conical tubes and centrifuged at 4°C for 3 min at 1,467 × g. The cell pellets were immediately resuspended in 750 μL of TRIzol (Invitrogen) and lysed by bead beating. RNA was purified according to protocol with the Zymogen Direct-zol RNA Miniprep Plus (catalog no. 2070). RNA was processed for Illumina sequencing using the TRuSeq Total RNA Library Prep from Illumina, with bacterial rRNA removal probes provided separately by Illumina. Sequencing was performed using Illumina NovaSeq at the North Texas Genome Center at the University of Texas in Arlington.

Between 50 and 300 million pair end reads per library were mapped to the M. abscessus subsp. abscessus strain ATCC 19977 published genome using CLC Genomic Workbench software (Qiagen). To minimize the skewing effect that certain PCR jackpots had on the data, we adjusted the number of reads mapped from each library so that the number of median reads per gene was equivalent within an experiment. In the logarithmic phase samples, the number of median reads per gene was ∼600. In the stationary-phase samples, the median reads per gene was ∼100. After normalization, the reads per kilobase million (RPKM) values were determined for each gene, and the weighted proportion fold change of RPKM between wild type and Δrel_Mab for each condition were calculated by CLC Workbench. The Baggerley’s test was used to generate a false discovery rate-corrected P value. We used an arbitrary cutoff 3-fold change with a false discovery rate corrected P ≤ 0.05 to identify significantly differentially regulated genes between wild type and Δrel_Mab under each condition. Because the median reads per gene for logarithmic phase samples was six times higher than for stationary-phase samples, we linearly scaled the fold change values when comparing logarithmic to stationary-phase data to normalize for this difference in read depth.

Real-time PCR.

For RNA-seq validation, real-time PCR (RT-PCR) was performed with the Kapa Biosystems Sybr Fast one-step qRT-PCR kit. RNA was extracted in triplicate from each Mab strain as described above. Primers for reverse-transcription (RT) were designed for each gene of interest by using Primer3 (Table S3). Each 20-μl RT reaction mixture contained 10 μl of one-step SYBR green Master Mix, 1 μl of each primer, 4.5 μl of nuclease-free water, 0.5 μl of qScript One-step RT, and 3 μl of RNA. RT-PCR was done by using Bio-Rad CFX Connect real-time system. The relative target levels (fold change) were calculated using the ΔΔC_T method (85), with normalization of RNA targets to TetR transcription factor gene (MAB_1638).