Mycobacterium tuberculosis Lsr2 is a global transcriptional regulator required for adaptation to changing oxygen levels and virulence.

UNLABELLED: To survive a dynamic host environment, Mycobacterium tuberculosis must endure a series of challenges, from reactive oxygen and nitrogen stress to drastic shifts in oxygen availability. The mycobacterial Lsr2 protein has been implicated in reactive oxygen defense via direct protection of DNA. To examine the role of Lsr2 in pathogenesis and physiology of M. tuberculosis, we generated a strain deleted for lsr2. Analysis of the M. tuberculosis Δlsr2 strain demonstrated that Lsr2 is not required for DNA protection, as this strain was equally susceptible as the wild type to DNA-damaging agents. The lsr2 mutant did display severe growth defects under normoxic and hyperoxic conditions, but it was not required for growth under low-oxygen conditions. However, it was also required for adaptation to anaerobiosis. The defect in anaerobic adaptation led to a marked decrease in viability during anaerobiosis, as well as a lag in recovery from it. Gene expression profiling of the Δlsr2 mutant under aerobic and anaerobic conditions in conjunction with published DNA binding-site data indicates that Lsr2 is a global transcriptional regulator controlling adaptation to changing oxygen levels. The Δlsr2 strain was capable of establishing an early infection in the BALB/c mouse model; however, it was severely defective in persisting in the lungs and caused no discernible lung pathology. These findings demonstrate M. tuberculosis Lsr2 is a global transcriptional regulator required for control of genes involved in adaptation to extremes in oxygen availability and is required for persistent infection. IMPORTANCE: M. tuberculosis causes nearly two million deaths per year and infects nearly one-third of the world population. The success of this aerobic pathogen is due in part to its ability to successfully adapt to constantly changing oxygen availability throughout the infectious cycle, from the high oxygen tension during aerosol transmission to anaerobiosis within necrotic lesions. An understanding of how M. tuberculosis copes with these changes in oxygen tension is critical for its eventual eradication. Using a mutation in lsr2, we demonstrate that the Lsr2 protein present in all mycobacteria is a global transcriptional regulator in control of genes required for adaptation to changes in oxygen levels. M. tuberculosis lacking lsr2 was unable to adapt to both high and very low levels of oxygen and was defective in long-term anaerobic survival. Lsr2 was also required for disease pathology and for chronic infection in a mouse model of TB.

INTRODUCTION

Mycobacterium tuberculosis is one of the most successful human pathogens, killing nearly two million people per year, and is the agent of an estimated two billion latent infections worldwide (1). The ability of M. tuberculosis to survive and to establish both active and latent infections despite the fact that the host is immunocompetent implies that the bacteria are able to survive insults from the host immune system, such as reactive oxygen species (ROS) and reactive nitrogen species (RNS). M. tuberculosis must also survive vast changes in oxygen tension, from high atmospheric oxygen levels during aerosol transmission to the microaerobic or anoxic environments within necrotic granulomas (2). The mechanisms employed by M. tuberculosis to protect itself from these stresses and changing environments during infection are not well understood. Deciphering the mechanisms that M. tuberculosis utilizes to survive these diverse conditions could lead to more effective antibiotics and greater success in tuberculosis (TB) treatment.

The Lsr2 protein in mycobacterial species is similar to H-NS from Escherichia coli (3–6). H-NS works mainly to repress gene transcription by binding to AT-rich sequences in a sequence-independent fashion and has been implicated in virulence in Shigella flexneri and Vibrio cholerae (7). H-NS is able to activate gene expression as well, but examples of activation are less common (8). Lsr2 from M. tuberculosis and from Mycobacterium smegmatis has been shown in vitro to bind directly to DNA and protect it against H2O2 (9) and DNase I degradation (6). An M. smegmatis lsr2 deletion mutant was also transiently more susceptible to H2O2 stress (9).

To examine the role of Lsr2 in M. tuberculosis, we deleted lsr2 from the M. tuberculosis H37Rv strain. While lsr2 has been reported to be essential in M. tuberculosis (6, 10), we were able to generate a deletion mutation in lsr2. In contrast to the findings in M. smegmatis, we found the M. tuberculosis Lsr2 protein is not important for protection against DNA damage. The Lsr2 protein is, however, a global transcriptional regulator important for allowing M. tuberculosis to respond to changes in oxygen levels that would be experienced during both transmission and infection.

RESULTS

The Δlsr2 strain is sensitive to high-oxygen conditions.

Initial experiments with the Δlsr2 strain demonstrated it was severely defective in initiating growth on solid agar plates. The lowest dilution formed a lawn, but all higher dilutions displayed highly defective growth (see Fig. S2 in the supplemental material). Cultures of the Δlsr2 strain, which had similar optical density at 600 nm (OD600) and growth rate to the wild type under low-aeration conditions, displayed several logs less viability on Dubos-Tween-albumin (DTA) or 7H11 agar plates. These cultures displayed equal viability to the wild type when assayed using liquid culture to determine a most probable number (MPN). The liquid culture method was therefore used for all subsequent viability assessment. The Δlsr2 strain was defective for pellicle formation (see Fig. S3 in the supplemental material), as was previously shown in an M. smegmatis Δlsr2 mutant (3). We hypothesized these findings were due to the sensitivity of the mutant to the high-oxygen environment experienced by bacteria on the surface of an agar plate or at the surface of a bacterial liquid culture. Oxygen levels in air are 50 times higher than in liquid, even without factoring in the consumption of oxygen by bacteria in liquid culture. We therefore analyzed the growth of the four strains under hypoxic and hyperoxic conditions compared to growth under ambient air. Cultures were started at an OD600 of 0.1 and stirred either in an incubator containing ambient atmospheric oxygen or in an oxygen-controlled chamber containing a constant supply of oxygen at either 2% or 60%. The Δlsr2 strain grew at the same rate as the wild-type strain under 2% oxygen (Fig. 1A). However, under atmospheric oxygen, the mutant showed a growth defect compared to the wild type (Fig. 1B). This growth defect was even more pronounced when the cultures were grown in 60% oxygen (Fig. 1C). As expected, the doubling time of all strains was greatest when grown in 2% oxygen compared to an 18% oxygen environment. However, whereas the growth rate of the wild type, the Δlsr2 complemented strain, and the lsr2 overexpression strain did not change from 18% to 60% oxygen tension, the doubling time of the Δlsr2 strain increased significantly (Fig. 1D).

FIG 1 

Growth of the Δlsr2 strain under hypoxic, normoxic, and hyperoxic oxygen tension. Cultures of H37Rv (squares and white bars), the Δlsr2 strain (open triangles and horizontally striped bars), the Δlsr2 complemented strain (diamonds and diagonally striped bars), or the lsr2 overexpression strain (closed triangles and checkered bars) were grown in DTA medium in either a chamber containing 2% oxygen (A), under atmospheric oxygen (~18% oxygen) (B), or a chamber containing 60% oxygen (C). (D) The doubling time was calculated from between day 1 and day 3 for each strain.

The Δlsr2 strain is not more sensitive to H2O2 or mitomycin C.

In order to determine if the M. tuberculosis Δlsr2 strain was more sensitive to DNA-damaging agents, we exposed all strains to 5 mM H2O2 or to 50 nM mitomycin C during mid-logarithmic growth (Fig. 2A) or to 3 mM H2O2 and 50 nM mitomycin C during anaerobic stasis at day 14 in the rapid anaerobic dormancy (RAD) model (Fig. 2B). As the Δlsr2 strain could not grow well on solid agar, the liquid infinite dilution method was used to obtain an MPN measurement of viability. In contrast to previous data from M. smegmatis, the M. tuberculosis Δlsr2 strain displayed no statistically significant difference in survival in response to H2O2 from the wild-type strain during aerobic growth or in the RAD model (Fig. 2A and B). To ensure that the slowed rate of the Δlsr2 strain was not responsible for the lack of susceptibility to H2O2, we exposed all strains to either 3, 4, or 5 mM H2O2 under 2% oxygen, a condition during which all strains grew the same. We also exposed the cultures to 500 nM mitomycin C to determine if the Δlsr2 strain was more sensitive to DNA damage under hypoxic conditions. Under hypoxic conditions, the Δlsr2 strain was no more sensitive to all concentrations of H2O2 tested or to either concentration of mitomycin C (see Fig. S4 in the supplemental material). These data indicate that Lsr2 does not play a role in protection against DNA damage from oxidative stress and other DNA-damaging agents as indicated in M. smegmatis. However, overexpression of lsr2 did confer increased resistance to H2O2 stress compared to the other strains under anaerobic conditions (Fig. 2B), indicating that an increase in lsr2 expression may have a slight protective effect when oxygen is lacking, but overexpression did not increase resistance to H2O2 under aerobic or hypoxic conditions.

FIG 2 

Survival of the Δlsr2 strain after addition of H2O2 or mitomycin C during aerobic growth or in an anaerobic model. Cultures of H37Rv (white bars), the Δlsr2 strain (horizontally striped bars), the Δlsr2 complemented strain (diagonally striped bars), or the lsr2 overexpression strain (checkered bars) were exposed to water, 5 mM H2O2, or 50 nM mitomycin C (MitC) during aerobic growth (A), and the most probable number (MPN) was determined after 48 h. (B) Cultures were exposed to either water, 3 mM H2O2, or 50 nM mitomycin C at day 14 in a RAD model, and the MPN was determined after 5 days. Panels C and D are the calculated percentages of survival from panels A and B, respectively.

Lsr2 is critical for adaptation to anaerobiosis.

As the Δlsr2 strain had clear defects in adaptation to changes in oxygen levels, its ability to survive anaerobiosis was investigated. A modified RAD model was used to examine the adaptation and fitness of all strains to anaerobic conditions (11). The modified RAD model starts at a relatively high culture density with little available oxygen in order to prevent a high-oxygen growth phase that would be detrimental to the Δlsr2 strain. OD600 and MPN measurements were taken throughout the course of adaptation to and during maintenance of anaerobic conditions. As demonstrated in Fig. 3A and B, the OD600 of the wild-type and Δlsr2 complemented cultures increased in the first 24 h. After the bacilli consumed available oxygen and adapted to the anaerobic environment, the OD600 slowly decreased. In the Δlsr2 strain, however, the OD600 decreased immediately and remained significantly lower than those of all other strains throughout the hypoxic and anaerobic experiments. Interestingly, the lsr2 overexpression strain initially increased in OD600 more than that of either the wild type or the Δlsr2 complemented strain and remained at a higher OD600 throughout anaerobic stasis. MPN measurements were taken to assess viability. Figure 3C demonstrates that there was no significant difference in survival rates among all four strains up to day 14 in the anaerobic model. By day 21, however, the Δlsr2 strain demonstrated a marked decrease in survival compared to all other strains. Although the lsr2 overexpression strain displayed a higher OD600 throughout the experiment, this strain did not demonstrate a measurable increase in growth or survival.

FIG 3 

Growth and survival of the Δlsr2 strain in an anaerobic model. Cultures of H37Rv (squares and white bars), the Δlsr2 strain (open triangles and horizontally striped bars), the Δlsr2 complemented strain (diamonds and diagonally striped bars), or the lsr2 overexpression strain (closed triangles and checkered bars) were cultured in the RAD anaerobic model. (A) OD600 measurements were taken during the course of the RAD model. (B) OD600 measurements during adaptation from aerobic to anaerobic phase of the model. (C) Viability MPN analysis at 0, 7, 14, 21, 28, 35, and 42 days in the RAD model.

Lsr2 is critical for recovery from anaerobiosis.

To determine if Lsr2 is also important for recovery from long-term anaerobiosis, anaerobic tubes of each strain were opened and exposed to aerobic conditions. Recovery from anaerobiosis was monitored at days 7 and 14 of the RAD model, both time points at which the Δlsr2 strain did not demonstrate a significantly decreased viability (Fig. 3C). OD600 measurements were determined once every 24 h to assess growth. After 7 days in the RAD model, the recovery growth rates of all four strains were the same as during aerobic growth (Fig. 4A compared with Fig. 1B). The wild-type, Δlsr2 complemented, and lsr2 overexpression strains were able to readily reactivate from anaerobiosis at day 14, albeit with a slight lag. The Δlsr2 strain, however, had a significant lag in reactivation by day 14 in the RAD model (Fig. 4B). The doubling time of the mutant recovering from 14 days in the RAD model was pronounced by a statistically significant lag compared to the wild type (Table 1).

FIG 4 

Recovery from anaerobiosis. H37Rv (squares), the Δlsr2 strain (open triangles), the Δlsr2 complemented strain (diamonds), or the lsr2 overexpression strain (closed triangles) was added to RAD model tubes. After 7 days (A) or 14 days (B), tubes were opened as indicated by the arrows, all but 5 ml of culture was removed to ensure proper aeration, and the OD600 was obtained every 24 h.

TABLE 1 

Doubling time between day 1 and day 3 during recovery from anaerobiosis

Condition	Doubling time for:
	H37Rv (wild type)	Δlsr2 strain	Δlsr2 complemented strain	lsr2 overexpression strain
Aerobic growth	21.1 ± 0.6	29.7 ± 3.4	21.2 ± 0.3	22.8 ± 0.4
RAD model
Day 7	30.6 ± 2.3	61.2 ± 17.1	30.6 ± 1.9	32.3 ± 2.2
Day 14	41.8 ± 5.8	144.3 ± 6.4[a]	103.7 ± 101.1	41.3 ± 1.7

Statistically significant difference between H37Rv and the Δlsr2 strain with a P value of <0.05.

Lsr2 modulates a shift away from aerobic respiration.

To investigate further the defect in adaptation to anaerobic conditions, we added methylene blue to RAD model tubes to analyze the oxygen consumption of all strains while adapting to anaerobic conditions. Methylene blue decolorization correlates with oxygen depletion in the RAD model (11). OD600 measurements were obtained at regular intervals until all oxygen had been consumed, as measured by complete methylene blue decolorization. As is shown in Fig. 5, the Δlsr2 mutant consumed oxygen much more rapidly within the first 48 h than all other strains. The strain overexpressing lsr2 consumed oxygen more slowly than all other strains. The rate of oxygen consumption directly correlated with the OD600 measurements of the strains during adaption to anaerobiosis.

FIG 5 

Oxygen consumption during adaptation to anaerobiosis as measured by methylene blue decolorization. H37Rv (squares), the Δlsr2 strain (open triangles), the Δlsr2 complemented strain (diamonds), and the lsr2 overexpression strain (closed triangles) were added to a RAD model. Methylene blue (MB) was added (final concentration, 9 µg/ml) at time zero, and the OD600 measurement was taken at regular intervals until methylene blue decolorization was complete.

Punctate DNA staining is unchanged in the Δlsr2 strain.

During anaerobiosis, a small proportion of M. tuberculosis bacilli displayed a punctate 4′,6-diamidino-2-phenylindole (DAPI) staining pattern, which appears to indicate DNA condensation. If Lsr2 was involved in condensation of DNA during anaerobiosis, the difference in DNA nucleoid structure should be visualized using the DNA-specific stain DAPI. Both the wild type and the Δlsr2 strain were stained following anaerobic stasis and analyzed via microscopy. There were no visual differences between the wild type and the Δlsr2 strain in the compaction of DNA during anaerobiosis (see Fig. S5 in the supplemental material).

Lsr2 is a global transcriptional regulator.

To determine the cause of heightened susceptibility of the Δlsr2 strain to an ambient oxygen environment or to an anaerobic environment, we performed whole-genome expression profiling by microarray analysis to compare wild-type gene expression to expression in the Δlsr2 strain during aerobic growth or after 24 or 48 h in a RAD model. A large set of genes was differentially regulated in the Δlsr2 strain (see Table S1 in the supplemental material). Most of these genes were also identified by chromatin immunoprecipitation with microarray technology (ChIP-chip) analysis of M. tuberculosis as being directly bound by Lsr2 (see Fig. S6 and Table S1 in the supplemental material) (5). The majority of genes directly controlled by Lsr2 from the ChIP-chip data were upregulated in the microarray experiments. These data indicate that Lsr2 is more prominent in gene repression than in gene activation. Many genes differentially regulated during aerobic growth were also differentially regulated in the RAD model. Some genes, however, were regulated by Lsr2 in the RAD model, which were not differentially regulated during aerobic growth (see Table S1). In order to understand why the Δlsr2 mutant was more susceptible to H2O2 in M. smegmatis but not in M. tuberculosis, we analyzed the transcriptional response of the M. tuberculosis Δlsr2 strain compared to that of the wild type after the addition of H2O2 during aerobic growth and at 48 h in the anaerobic model (see Table S1). Microarray analysis revealed that the genes responsible for responding to oxidative stress (e.g., katG, trxC, trxB2, recA, and radA [12]) were induced to similar levels in both the Δlsr2 and wild-type strains after H2O2 stress.

Lsr2 is critical for persistent infection.

Lsr2 was important for survival during the changes in oxygen availability that are also experienced by bacilli during infection. Therefore, we tested H37Rv, the Δlsr2 strain, and the Δlsr2 complemented strain for in vivo growth and pathogenesis in BALB/c mice after aerosol infection (Fig. 6). Equal genome equivalents were detected for all three strains up to 2 weeks into infection. However, genome equivalents fell in the lungs of the mice with Δlsr2 strain infections by 4 weeks of infection (Fig. 6A). There were equal genome equivalents detected in the spleens of mice infected with all three strains at each time point (Fig. 6B). The numbers of genome equivalents detected for the wild type and the Δlsr2 complemented strain were comparable to CFU counts obtained by plating these 2 strains on 7H11 agar plates (Fig. 6C). Histopathology analysis was performed on lungs of infected BALB/c mice, and disease severity was scored. As shown in Fig. 6D, the Δlsr2 strain displayed no visible disease pathology in lung tissue at any time point, whereas the wild-type strain and the Δlsr2 complemented strain displayed significant disease by 4 weeks of infection. Histology analysis was performed on the lungs of 5 mice sacrificed at either 2 or 4 weeks post-aerosol infection (Fig. 7). Whereas inflammatory lesions were visible in mice infected with the wild type and the Δlsr2 complemented strain (Fig. 7A, B, E, and F), there were no signs of inflammation or any inflammatory lesions present within lungs of mice infected with the Δlsr2 strain (Fig. 7C and D).

FIG 6 

Survival of the Δlsr2 strain after aerosol infection in a BALB/c mouse model. Six- to 8-week-old BALB/c mice were infected via aerosol infection with either H37Rv (squares), the Δlsr2 strain (open triangles), or the Δlsr2 complemented strain (diamonds). Lungs (A) or spleens (B) were harvested at 1, 14, and 28 days, and genome equivalents (GE) were determined using qPCR with primer/probe sets for both Rv1738 and Rv2626. Genome equivalents were averaged together for both primer/probe sets. (C) CFU counts were obtained from lung and spleen samples for the wild type (white bars) and the Δlsr2 complemented strain (diagonally striped bars). (D) Disease scores were assessed for each mouse, and averages are shown for the wild type (Rv [white bars]), the Δlsr2 strain (LKO), or the Δlsr2 complemented strain (LCO [diagonally striped bars]). Five mice were used for each time point. An asterisk denotes differences that are statistically significant.

FIG 7 

Histology of infected BALB/c mouse lungs using hematoxylin and eosin stain. BALB/c mice were infected by the low-dose aerosol method with the H37Rv strain of M. tuberculosis (A and B), the Δlsr2 strain (C and D), or the Δlsr2 complemented strain (E and F). Low-magnification photomicrographs (A, C, and E) or higher-magnification photomicrographs (B, D, and E) are shown.

Survival of the Δlsr2 strain after long-term exposure to nitric oxide.

The oxidative burst within a mouse during M. tuberculosis infection occurs at around 2 weeks, and the nitrosative burst occurs at around 4 weeks (13). As the Δlsr2 strain did not display a defect until after 2 weeks of infection, we postulated the decrease in fitness of the mutant at 4 weeks could be due to sensitivity to RNS present during the nitrosative burst produced at later time points within the lung. We added six doses of 100 µM diethylenetriamine NONOate (DETA-NO), once every 6 h, to aerobic cultures and measured both recovery via OD600 and survival via MPN. The time to half-maximum value was calculated for each strain during recovery from long-term DETA-NO exposure and compared to that of the wild type. As demonstrated in Fig. S7A and S7B in the supplemental material, the Δlsr2 strain displayed delayed recovery after a prolonged exposure to nitric oxide compared with all other strains. None of these strains showed any decrease in viability after the addition of long-term DETA-NO exposure, as determined by MPN (data not shown) as we previously described for wild-type M. tuberculosis (12).

DISCUSSION

Lsr2 has been extensively studied in M. smegmatis, but Lsr2 in M. tuberculosis was believed to be essential (6, 10). We were successful in creating an lsr2 M. tuberculosis deletion mutant. We have demonstrated that the Lsr2 protein of M. tuberculosis is important both for survival in a high-oxygen environment and for adaptation to an anaerobic environment. The defect in growth at ambient oxygen levels likely explains the original designation of Lsr2 essentiality. The inability to grow under high oxygen tension could be due to increased susceptibility to ROS generated by the Fenton reaction during aerobic respiration (14), as an M. smegmatis lsr2 deletion strain was more sensitive to H2O2 stress (9). However, the M. tuberculosis Δlsr2 strain was not more sensitive to exogenous H2O2, either aerobically or anaerobically. The aerobic growth defect in the Δlsr2 strain is not primarily due to a defect in H2O2 defense, but perhaps the lsr2 mutant is defective in defense against other ROS. However, the genes most highly expressed after addition of H2O2 are not induced in the lsr2 mutant, and the types of genes controlled by Lsr2 indicate that the lack of Lsr2 results in deregulation genes important for aerobic growth.

Lsr2 from M. smegmatis has also been implicated in the direct protection of DNA. A study by Colangeli et al. demonstrated recombinant Lsr2 binds directly to pUC19 plasmid DNA in vitro and protects it against DNA-damaging agents. The authors speculated that Lsr2 could play a role similar to Dps in E. coli, condensing DNA during the stationary phase and protecting it against insults such as ROS (9). However, M. tuberculosis Lsr2 in vivo does not appear to have a direct role in protecting DNA from damage via condensation. This conclusion is based on several lines of evidence. First, the M. tuberculosis Δlsr2 strain is not more susceptible to DNA-damaging agents such as H2O2 or mitomycin C. Second, Lsr2 from M. tuberculosis binds and regulates a large number of genes throughout the genome in an AT-rich, sequence-independent manner, similar to the H-NS protein from E. coli (see Table S1 in the supplemental material and references 5 and 15). Third, DAPI staining of anaerobic M. tuberculosis (see Fig. S5 in the supplemental material) shows similar staining patterns, indicating that Lsr2 is not solely responsible for global DNA compaction, although it is likely that Lsr2 binding results in localized changes in tertiary structure. In addition, overexpression of Lsr2 in M. tuberculosis was not associated with global gene repression, indicating that it is not condensing DNA nonspecifically (6). The disparity between the results of Colangeli et al. and those of this study could be due to the fact that φX174 DNA was used to assess the role of Lsr2 as a protector of DNA in vitro. φX174 DNA has a high AT content (55%) (16) compared to M. tuberculosis (45%) (17), and Lsr2 readily binds to high-AT-content DNA. Thus, in vitro conditions and AT-rich DNA may have promoted nonspecific Lsr2 binding.

Whole-genome expression profiling demonstrated many genes were upregulated in the Δlsr2 strain, including several PE or PPE family proteins and genes that were horizontally acquired, such as Rv2339. These genes contain a higher AT content, to which Lsr2 readily binds (5). However, while many genes were differentially regulated in the Δlsr2 strain, genes highly responsive to H2O2 stress, such as katG, trxC, trxB2, recA, and radA, were not among them, further indicating that the defect in growth of the Δlsr2 strain is not due to sensitivity to ROS. Some genes that are normally upregulated by H2O2 stress were upregulated in the Δlsr2 strain in the absence of exogenous H2O2, such as the operon comprising Rv1461 to Rv1466 (Rv1461-Rv1466), involved in iron-sulfur cluster repair, sigB (an alternate stress sigma factor), Rv2729, and Rv3018 (12). Some of these genes, such as Rv1460 to Rv1463, are also upregulated under low-iron conditions (18). The mmpL4 and mmpS4 genes (Rv0450 and Rv0451, respectively), were upregulated in the Δlsr2 strain and are normally repressed in the presence of high iron concentrations by the iron-responsive regulatory protein IdeR (18). The gene encoding the iron storage protein bacterioferritin (bfrB) was upregulated in the Δlsr2 strain and is directly bound by Lsr2. The upregulation of an iron storage protein could cause a decrease in available intracellular iron and consequently result in slower growth. The Δlsr2 strain is not likely more sensitive to the H2O2 produced during growth under aerobic conditions but may instead display slowed growth due to actual or perceived low-iron levels. Several genes (such as Rv1067c, sigB, 3288c, Rv3424c, Rv3879c, and Rv3903c) that were expressed more highly in the Δlsr2 strain also respond to cell envelope disruption via addition of vancomycin (19). The Rv1501-Rv1507c operon involved in cell wall biosynthesis was upregulated in the Δlsr2 strain (20). The Δlsr2 mutant could be experiencing cell envelope stress due to differential regulation of genes responsible for cell wall biosynthesis. Cell envelope stress and low internal iron availability due to differential gene regulation in the mutant appear to be the main stresses that are being experienced by the Δlsr2 strain and could result in slowed to no growth under highly oxygenated conditions. Lsr2 is an important regulatory protein part of a larger regulatory network (21), the disruption of which clearly causes a severe growth defect during standard aerobic conditions.

Under constant hypoxia, the Δlsr2 strain had the same growth rate as the wild-type strain but displayed a growth defect during the microaerobic phase and a survival defect during the anaerobic phase in an anaerobic model. This discrepancy is likely because the oxygen concentration in the RAD model rapidly drops below the dissolved oxygen level in the 2% hypoxia experiment. Methylene blue starts to decolorize immediately in the RAD model but does not decolorize in the cultures under a 2% oxygen atmosphere. The difference in growth rates for the Δlsr2 strain between early in the anaerobic model and the 2% oxygen atmosphere experiment indicates Lsr2 is required either for growth at the very low oxygen tension before anaerobiosis or for adaption to the rapidly declining oxygen tension that occurs in the aerobic to anaerobic model. The difference in growth rates as seen by OD600 was not reflected in survival as observed by MPN, but this is most likely due to the fact that this method of detecting viability is not as sensitive as plating for CFU, so small differences in viability will not be observed. Additionally, M. tuberculosis undergoes morphological changes during adaptation to anaerobiosis resulting in an increase in OD600 without a corresponding increase in CFU (22–24). Thus, the small decrease in OD600 observed in the Δlsr2 strain that is not reflected in the MPN of these cultures is likely due to changes in cell morphology of the mutant rather than to changes in viability. Lsr2 was also required for recovery after extended anaerobiosis. The anaerobic recovery and survival defects may be linked to uncontrolled rapid consumption of a dwindling oxygen supply during the shift to anaerobiosis. The ability of M. tuberculosis to slow the consumption of oxygen during adaptation to anaerobic conditions appears to be important for its survival. In addition, the rapid oxygen consumption of the Δlsr2 strain led to a shorter period of oxygen availability and less time for growth prior to the onset of anaerobiosis, during which obligate aerobes such as M. tuberculosis are unable to grow. The rapid oxygen consumption during microaerobic conditions is similar to that observed for a mutant with mutation of another key M. tuberculosis regulator, DosR, which is required for oxygen adaptation and anaerobic survival (11). A dosR mutant also consumes oxygen at a much higher rate than the wild type during microaerobic conditions and has a delayed recovery from anaerobic dormancy (11). A reciprocal trend was observed in the strain overexpressing lsr2, which had a slower oxygen consumption rate than the wild type and achieved a higher OD600 in the anaerobic model. As oxygen is required for M. tuberculosis growth, these data suggest that conservation of oxygen by the bacilli in this model allowed for more growth due to the longer time interval in the presence of oxygen (Fig. 5 compared with Fig. 3). As M. tuberculosis faces both microaerobic conditions and changing oxygen levels within the necrotic granuloma, Lsr2 should be important for survival within this environment.

The differential gene transcription profile displayed in this study correlated well with ChIP-chip data found by Gordon et al. (5), demonstrating that control of transcription by Lsr2 is mostly direct. A higher proportion of genes identified in the ChIP-chip analysis were also found by microarray analysis to be upregulated rather than downregulated in the Δlsr2 strain, indicating Lsr2 plays a greater role in gene repression than in gene activation. The Lsr2 protein in mycobacterial species has a similar function to that of the H-NS protein from E. coli (3–6, 21). Like Lsr2, H-NS is a small, basic protein (8) that functions mainly to repress gene transcription (7). Like H-NS, Lsr2 is able to activate as well as repress genes by nonspecific binding to AT-rich regions of DNA, but the role in activation of genes by H-NS is less extensive and not well understood (8). Not all genes differentially regulated in the Δlsr2 strain are directly controlled by Lsr2. Regulators such as sigB and whiB3 were differentially regulated in the Δlsr2 strain, are either directly or indirectly controlled by Lsr2 (21), and are likely responsible for some of the indirect gene regulation in the mutant. The combination of direct binding studies and expression profiling demonstrates that Lsr2 directly regulates a multitude of genes important for growth at high and very low oxygen tension and survival during anaerobic stasis.

To determine if Lsr2 is required for pathogenesis, BALB/c mice were infected via aerosol infection with wild type, the Δlsr2 strain, and the Δlsr2 complemented strain from static growing cultures. The aerosol method of infection used in this study did not decrease the initial viability of the mutant, as genome equivalents detected on days 1 and 14 were similar for all three strains. After 4 weeks of infection, there was a significant decrease in genome equivalents detected in lungs of mice infected with the Δlsr2 strain. No notable disease pathology was observed in terms of lesion formation in mice infected with the Δlsr2 strain after 4 weeks, whereas in the mice infected with the complement and the wild-type strain, the pathology was clearly visible. These findings may be explained by the dynamics of the oxidative and nitrosative bursts within a murine M. tuberculosis infection. The oxidative burst within a mouse predominates within the first 2 weeks, whereas RNS exposure starts around 4 weeks into infection (13). The equal bacterial numbers in the lungs of all strains after 2 weeks in the mouse were therefore not surprising, as the Δlsr2 strain was not more susceptible to H2O2 stress. Granulomas formed within a mouse do not become anaerobic (25); thus, the difference in viabilities at 4 weeks is likely not attributed to total lack of oxygen. We therefore speculate that this decrease in viability may be due to the presence of a nitrosative stress within the mouse lungs. Nitric oxide, like anaerobiosis, is a potent inhibitor of respiration (26). M. tuberculosis responds very similarly to nitric oxide and anaerobiosis and displays a similar pattern of growth arrest and slow recovery (11, 27). In vitro experiments demonstrated that the Δlsr2 strain displayed a prolonged lag in recovery when exposed to nitric oxide, similar to the delay in recovery observed from anaerobiosis. We speculate that a defect in recovery from a period in which aerobic respiration has been inhibited by a combination of limited oxygen and nitric oxide exposure may give the wild type an advantage over the Δlsr2 strain in the nonnecrotic BALB/c mouse model. The viability of the mutant was equal to that of the wild type at all time points in the spleen, indicating the splenic lesions were less restrictive with respect to respiration. As the bacterial counts were the same in the lungs for all strains at 2 weeks, the number of bacilli initially seeding the spleen would have been the same and likely accounts for the successful colonization of the spleen even though persistence in the lung was defective.

The life cycle of M. tuberculosis requires the pathogen to be transmitted via small aerosolized droplet nuclei that must penetrate deep into the lungs of a new host, where it is taken up within resident macrophages and neutrophils. These cells react to the growing bacteria by forming a hypoxic/anaerobic granuloma and applying antimicrobial agents, such as nitric oxide, that contain and control bacterial growth. Changes in oxygen availability during this infectious cycle are dramatic and numerous. During aerosol transmission, the pathogen is exposed to high oxygen tension in ambient air. The moment that the bacilli enter the lungs of a new host, oxygen levels rapidly drop, and they continue to decrease during granuloma formation. In vitro, Lsr2 is essential for growth on agar plates where the bacilli are exposed to air at an oxygen concentration of about 260 mg/liter, analogous to oxygen levels during aerosol transmission. Lsr2 is also essential for adaptation to microaerobic and anaerobic conditions, such as those found within necrotic lesions. Lsr2 is important for recovery from nonrespiring states after prolonged anaerobiosis or after exposure to nitric oxide, both of which are present during human infection. From the data presented herein, Lsr2 is a global transcriptional regulator similar to H-NS that directly controls gene expression in response to changes in oxygen availability. Lsr2 is important for persistent infection and likely plays a role in several stages of infection in which oxygen levels and the corresponding metabolic systems are in constant flux.

MATERIALS AND METHODS

Culture conditions, strains, and plasmids.

Liquid cultures of M. tuberculosis strain H37Rv, the lsr2Δ174-268 strain (here referred to as the Δlsr2 strain) which was deleted for the C-terminal portion of the protein responsible for binding to DNA (5), the Δlsr2::pMV306-lsr2 strain (Δlsr2 complemented strain), and a strain of H37Rv containing the expression vector pMV261 (28) containing the gene for lsr2 (here referred to as the lsr2 overexpression strain) were maintained in Dubos-Tween-albumin broth (DTA) (Difco Dubos broth base [Becton Dickinson], 0.5% bovine serum albumin [BSA] fraction V, 0.75% glucose, 0.17% NaCl, and 0.05% Tween 80) as semisettled cultures at 37°C in flasks with daily agitation to ensure slight aeration. These low-aeration cultures were used to start all experiments listed below, as all strains grew at the same rate under these hypoxic conditions as they did in defined 2% oxygen (Fig. 1A). The lsr2 overexpression strain was maintained in the presence of 20 mg/ml kanamycin in order to maintain the pMV261 plasmid. Cultures were maintained at an OD600 of less than 0.5. All aerobic survival experiments after exposure to H2O2, mitomycin C, or diethylenetriamine NONOate (DETA-NO) were carried out at a starting OD600 of 0.1 in 10-ml volumes in glass tubes (20 mm by 125 mm) containing stir bars (12 mm by 4.5 mm). Growth of H37Rv, the Δlsr2 strain, the Δlsr2 complemented strain, and the lsr2 overexpression strain in ambient air (~18% oxygen), 60%, and 2% oxygen concentrations was conducted in an oxygen-controlled chamber (Coy Laboratories). Ten-milliliter volumes of culture with a starting OD600 of 0.05 were maintained in 20 mm by 125-mm glass tubes containing 12- by 4.5-mm stir bars, and OD600 measurements were obtained daily. The modified rapid anaerobic dormancy (RAD) model was performed as follows (11). Sixteen-milliliter cultures at an OD600 of 0.15 were started in glass tubes (16 by 125 mm) containing stir bars (12 mm × 4.5 mm). Minimal headspace remained in the tubes (~1 ml). The tubes were sealed with caps containing butyl rubber septa and stirred at 60% of maximum speed using a rotary magnetic tumble stirrer from V & P Scientific (San Diego, CA). Methylene blue (final concentration, 9 µg/ml) was added to a set of tubes in order to assess oxygen depletion by all strains, and OD600 was obtained at regular intervals over a period of 4 days, or until the methylene blue became completely decolorized. Recovery from anaerobiosis was assessed by removing the lids from days 7 and 14 in the RAD model, with all but 5 ml of culture from the glass tubes removed to ensure proper aeration. Tubes were covered with loose-fitting caps, and the OD600 was measured every 24 h. Each experiment was performed using three biological replicates.

Construction of the Δlsr2 strain and Δlsr2 complemented strains.

The Δlsr2 strain was constructed as previously described (29–31). Briefly, flanking regions comprising upstream and downstream regions of the lsr2 gene were amplified using the following primers: upstream forward primer TTTTTTTTCCATAGATTGGAGACCCGTCAGCACCGCAGT, upstream reverse primer TTTTTTTTCCATCTTTTGGACGCCGCCACCCATTGCTTC, downstream forward primer TTTTTTTTGCATAGATTGCGGCTCGTCGTAACGGGCACA, and downstream reverse primer TTTTTTTTGCATAGATTGCCTGCATGACCCGCTCGATTT. Flanking sequences for upstream and downstream regions of lsr2 were amplified by PCR and cloned into the pAES0004S plasmid containing a hygromycin resistance cassette, which was confirmed by sequence analysis. The vector was then ligated into the phAE159 phasmid, which was electroporated into M. smegmatis, and the resulting phage was amplified to obtain a high-titer stock. The high-titer phage was used to infect M. tuberculosis, which was plated onto DTA plates containing hygromycin. Plates were incubated for 3 months instead of the typical 1 to 2 months in sealed bags, which eventually led to the growth of a small number of lsr2 mutant colonies. Colonies were picked, and PCR analysis was used to confirm the presence of the hygromycin-gene flanking region and absence of the wild-type gene. Sequence analysis was performed on these PCR products from the wild type, the Δlsr2 strain, and the phasmid construct in order to further confirm the absence of the wild-type lsr2 gene in the Δlsr2 strain (see Fig. S1B in the supplemental material). The Δlsr2 complemented strain was constructed using the integrative vector pMV306. The lsr2 gene was amplified along with 200 bases of upstream DNA to include the native promoter. The Δlsr2 strain was electroporated with the complementing plasmid, and colonies were picked, grown in the presence of antibiotic, and confirmed for presence of the vector using PCR analysis. For the lsr2-overexpressing strain, the lsr2 gene was amplified and inserted downstream of the hsp60 promoter within the exogenous pMV261 plasmid. The overexpression plasmid was introduced into H37Rv by electroporation, and the presence of the vector was confirmed by PCR.

Infinite dilution to determine viability.

As the Δlsr2 strain was unable to grow on agar plates as well as the wild type, the viabilities of all strains were determined using the infinite dilution method to obtain the most probable number (MPN) of viability, as described previously (11). Briefly, 50 µl of culture was added to 450 µl of DTA medium in 24-well plates, and 10-fold serial dilutions were performed up to a 10−10 dilution. Plates were sealed with parafilm wax and placed in plastic bags in a 37°C incubator for 8 to 10 weeks. Viability was determined using the highest dilution to show growth. All experiments were plated in duplicate, and three biological replicates were performed for each experiment.

H2O2 and mitomycin C challenge.

Either 5 mM H2O2, 50 nM mitomycin C, or water was added to cultures of aerobically growing H37Rv, Δlsr2, Δlsr2 complemented, and lsr2 overexpression strains at a starting OD600 of 0.1. Viability was determined at both 0 and 48 h by the infinite dilution method to determine MPN. Alternatively, RAD model cultures of all strains were started, and 3 mM H2O2, 50 nM mitomycin C, or an equal volume of water was added after 14 days in the RAD model, after which samples were collected at days 0 and day 5 after addition of stress to determine viability as measured by MPN.

Microscopic analysis.

Staining of dormant cultures of the H37Rv and Δlsr2 strains was performed as previously described (11). Briefly, cultures were resuspended in 0.025 M HEPES with 0.02% Tween 80 (pH 7.75). The DNA-specific stain 4′,6-diamidino-2-phenylindole (DAPI) was added at a final concentration of 300 nM for 20 min at 37°C. Samples were washed once in HEPES-Tween, resuspended in a small volume of phosphate-buffered saline (PBS), viewed on a Nikon Eclipse TE200-U inverted microscope, and analyzed using MetaVue software.

Microarray analysis.

Oligonucleotide microarrays were printed at the University of Colorado Anschutz Medical Campus using oligonucleotides purchased from Operon and were processed and prehybridized as previously described (26). RNA isolation and cDNA preparation, labeling and hybridization were performed as previously described (32). Microarrays were scanned using a GenePix 4000b scanner (Axon Instruments). The intensities of the two dyes at each spot were quantified using GenePix Pro 6.0 (Molecular Devices). Microarray-determined ratios were calculated from three biological replicates. The ArrayStar program (DNASTAR) was used to determine statistically significant regulated genes (33). Genes included in Table S1 in the supplemental material exhibited at least a 2-fold induction or repression ratio under at least one condition and had a significance analysis of microarrays (SAM) corresponding false discovery q value of zero.

BALB/c mouse infections.

Six- to 8-week-old BALB/c mice (Jackson Laboratories) were rested for 2 weeks before aerosol infection. The mice were then aerosol infected with around 4,000 bacilli (per mouse) of H37Rv, the Δlsr2 strain, or the Δlsr2 complemented strain by the Glas-Col inhalation exposure system as previously described (34). Lungs were harvested from three BALB/c mice that were sacrificed at 1 day postinfection, or lungs and spleens were harvested from five mice that were sacrificed at 2 and 4 weeks postinfection. Samples were homogenized and were assessed for bacterial load via determination of genome equivalents. The right caudal lung was collected and preserved in 4% paraformaldehyde for histological analysis.

DNA isolation.

To determine genome equivalents in mouse tissues, tissue samples were homogenized and suspended into 4.5 ml of DTA broth. Cellular material from 1 ml of homogenate was pelleted by centrifugation for 10 min at 13,200 rpm. This pellet was resuspended with 1 ml Trizol reagent (Invitrogen). Cells were disrupted using 0.1-[Merops: inline graphic #1]mm-diameter zirconia/silica beads (BioSpec Products) with 6- to 30-s pulses in a bead beater and were placed on ice between bead-beating steps. Cellular debris was separated from the cell lysate by centrifugation for 1 min at 13,200 rpm, and lysates were transferred to a new tube. Three hundred microliters of chloroform was added to each sample, and tubes were mixed for 15 s and then were allowed to sit 2 min at room temperature with occasional mixing. Centrifugation was performed at 13,200 rpm for 10 min, and the aqueous phase of each sample was removed. Five hundred microliters of filtered back extraction buffer (4 M guanidine thiocyanate, 50 mM sodium citrate, 1 M Tris base) was added to the organic phase of each sample, and samples were incubated at room temperature for 10 min. Separation of aqueous and organic phases was attained by centrifugation at 12,000 rpm for 30 min. The aqueous phase of each sample was then placed in a new tube, 10 µl of glycol blue (Life Technologies) and 400 µl of isopropanol were added to each sample, and samples were incubated at room temperature for 10 min. DNA was pelleted by centrifugation at 12,000 rpm for 15 min at 4° C, and the supernatant was removed. The DNA pellet was washed twice with 70% ethanol. The ethanol was removed, the pellet was briefly allowed to dry at room temperature, and the DNA pellets were resuspended in 20 µl of nuclease-free water.

Determination of genome equivalents.

Five microliters of purified DNA from mouse tissue was used per quantitative PCR (qPCR). The primer sets for two M. tuberculosis genes, Rv1738 and Rv2626, were used to determine genome equivalents (35). The primer and probe sets were as follows: Rv2626c forward, CCGCGACATTGTGATCAAAG, reverse, GCTCTGAGATGACCGGAACAC, and probe, CGAACGCAAGCATCCAGGAGATGC; and Rv1738 forward, CACTGGACCGTCGACATAT CG, reverse, CGGTCGGCCGGATTG, and probe, CCAACGCAGCCGTGCCTTCG. Five microliters of DNA purified from mouse tissue, 0.6 µl water, 1.6 µl (each) forward and reverse primers, and 1.2 µl corresponding probe were mixed with 10 µl of master mix (Roche), and reaction mixtures were added to a 96-well plate (Roche). Quantitative PCR was performed on a LightCycler 480 (Roche). All samples were compared to a standard curve, generated using a 10-fold dilution series of M. tuberculosis genomic DNA. The equation of the plotted trendline from the standard curve was obtained and used to calculate the number of copies of genomic DNA present within each experimental sample. The values herein represent the average between the data obtained from the two primer/probe sets.

Histological analysis.

The right caudal lobe was harvested from each mouse at 2 and 4 weeks into infection and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS). Tissue sections were embedded in paraffin and cut to 5 µm, mounted on glass slides, deparaffinized, and stained with hematoxylin and eosin. Slides were examined by a veterinary pathologist (R. J. Basaraba) blinded to the treatment groups. Microscopic changes, including an estimate of lung involvement, number of lesions, lesion type, the extent of lesion necrosis, and the relative distribution of lesions (alveolitis, peribronchiolitis, and perivasculitis), were scored for each animal and totaled. Once treatment groups were revealed, the mean scores were calculated, the section that scored closest to the different treatment groups was digitally scanned, and photomicrographs illustrating relevant features were generated (Fig. 7).

Nitric oxide challenge.

Recovery and survival of all strains after long-term exposure to DETA-NO were assessed as previously described (11). Briefly, 10 ml of aerobically grown cultures at a starting OD600 of 0.1 was stirred as described above. A 100 nM concentration of DETA-NO was added to cultures six times, once every 6 h, and OD600 measurements were monitored. A sample of culture from each tube was also taken at 0 and 48 h to determine survival via MPN using the infinite dilution method. The time to half-maximum OD600 was determined as previously described (36). Briefly, half-maximum growth was defined as (maximum OD600 − starting OD600) divided by 2. Linear interpolation using the two OD600 measurements flanking the half-maximum OD600 was used to calculate the time to half-maximum OD600 for each strain. The time to half-maximum OD600 value for each strain was divided by the value calculated for the H37Rv strain from the same treatment group to determine a percentage of the wild type for each strain.

Statistical analysis.

A two-way analysis of variance (ANOVA) test was used to determine statistical significance for survival and growth data. A Student’s t test was used to determine statistical significance for all mouse data and DETA-NO survival data. In all cases, values of <0.05 were considered statistically significant.

Microarray data accession numbers.

Microarray data are available at the NCBI Gene Expression Omnibus website (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE57948.

Sequence analysis of the Δlsr2 strain compared with the wild type and phasmid construct. PCR analysis was performed on phAE159, the phasmid construct used to create the Δlsr2 strain (lane 1), wild-type H37Rv (lane 2), the Δlsr2 strain (lane 3), and the Δlsr2 complemented strain (lane 4), using primers for the lsr2 gene, and products were visualized on a 1% agarose gel with ethidium bromide (A). PCR products marked with an asterisk were subsequently excised for sequence analysis, and the sequences are shown (B). The “scar” sequences, the residual nucleotides remaining from the phasmid gene deletion method, are highlighted in yellow. a, H37Rv and Δlsr2 complemented strain sequences are identical. Download

Figure S1, TIF file, 1.8 MB

Growth on DTA solid agar plates. H37Rv, the Δlsr2 strain, the Δlsr2 complemented strain, and the lsr2 overexpression strains were subjected to 10-fold dilutions, and 5 µl of each dilution was spotted on DTA or 7H11 agar plates. DTA plates are shown. Download

Figure S2, TIF file, 0.7 MB

Pellicle formation in long-term settled culture. H37Rv, the Δlsr2 strain, and the Δlsr2 complemented strain were grown in a settled culture for several weeks until pellicle formation occurred in the wild type and the Δlsr2 complemented strain. The Δlsr2 strain was severely deficient in pellicle formation. Download

Figure S3, TIF file, 1.7 MB

Survival of the Δlsr2 strain after the addition of DNA-damaging agents during hypoxia. Cultures of H37Rv (white bars), the Δlsr2 strain (horizontally striped bars), the Δlsr2 complemented strain (diagonally striped bars), or the lsr2 overexpression strain (checkered bars) were exposed to water (B), 500 nM mitomycin C (MitC), 3 mM H2O2 (3H), 4 mM H2O2 (4H), or 5 mM H2O2 (5H) and were placed in a chamber containing 2% oxygen (Coy Laboratories). The MPN was determined at time zero and after 48 h. Two biological replicates were sampled, each in duplicate. Download

Figure S4, TIF file, 0.6 MB

DAPI staining to visualize macroscopic DNA staining patterns. H37Rv and the Δlsr2 strain were exposed to anaerobic dormancy. Bacilli were stained using the DNA-specific stain DAPI and visualized by microscopy. The compact DNA staining in the wild type was similar to that observed in the Δlsr2 strain (bacilli in white circles). DIC, differential inference contrast. Download

Figure S5, TIF file, 1.9 MB

Correlation between microarray results and previously published ChIP-chip data. Microarray analysis was performed on wild-type strain H37Rv versus the Δlsr2 strain under aerobic and microaerobic conditions. Genes either upregulated (A and C) or downregulated (B and D) during aerobic (A and B) or microaerobic (C and D) conditions in our microarray analysis were compared with genes directly bound by Lsr2 as determined by ChIP-chip analysis (5). The numbers in each section are as follows: the top number indicates if the gene identified by microarray analysis was 0, 1, 2, 3, 4, or >4 genes away from a gene found in the ChIP-chip assay, and the bottom number indicates the proportion of genes that fell into the above category. Download

Figure S6, TIF file, 0.9 MB

Survival after long-term nitric oxide exposure. Cultures of H37Rv (white bars), the Δlsr2 strain (horizontally striped bars), the Δlsr2 complemented strain (vertically striped bars), and the lsr2 overexpression strain (checkered bars) were exposed to six doses of 100 µM DETA-NO, once every 6 h, and the OD600 was monitored over time. (A) The time to half-maximum OD600 was determined for each strain either for the blank or for the DETA-NO-treated samples, and the half-maximum OD600 was determined. (B) The time to half-maximum OD600 of each strain was calculated relative to that for H37Rv. An asterisk denotes statistical significance. Download

Figure S7, TIF file, 1.5 MB

Differentially regulated genes in Δlsr2 relative to H37Rv.

Table S1, XLSX file, 0.1 MB.