Analyzing the Regulatory Role of the HigA Antitoxin within Mycobacterium tuberculosis

Abstract

Bacterial chromosomally encoded type II toxin-antitoxin (TA) loci may be involved in survival upon exposure to stress and have been linked to persistence and dormancy. Therefore, understanding the role of the numerous predicted TA loci within the human pathogen Mycobacterium tuberculosis has become a topic of great interest. Antitoxin proteins are known to autoregulate TA expression under normal growth conditions, but it is unknown whether they have a more global role in transcriptional regulation. This study focuses on analyzing the regulatory role of the M. tuberculosis HigA antitoxin. We first show that the M. tuberculosis higBA locus is functional within its native organism, as higB, higA, and Rv1957 were successfully deleted from the genome together while the deletion of higA alone was not possible. The effects of higB-Rv1957 deletion on M. tuberculosis global gene expression were investigated, and a number of potential HigA-regulated genes were identified. Transcriptional fusion and protein-DNA-binding assays were utilized to confirm the direct role of HigA in Rv1954A-Rv1957 repression, and the M. tuberculosis HigA DNA-binding motif was defined as ATATAGG(N₆)CCTATAT. As HigA failed to bind to the next-most-closely related motif within the M. tuberculosis genome, HigA may not directly regulate any other genes in addition to its own operon.

Toxin-antitoxin (TA) loci were first characterized as plasmid-borne genes involved in bacterial plasmid maintenance (18, 32), where in a daughter cell lacking the plasmid, the antitoxin is degraded more rapidly than the toxin, resulting in growth inhibition. TA loci are categorized into two broad types based on the nature of the antitoxin: type I locus antitoxins are antisense small RNAs which prevent toxin translation, whereas type II locus antitoxins are proteins which inactivate the toxin through protein-protein interactions (17, 21). Both types of TA loci have been identified within prokaryotic chromosomes (15, 16, 23, 33), indicating that TA loci have functions unrelated to plasmid maintenance (16, 27).

Nine families of typical type II TA loci have been identified: ccdAB, mazEF, vapBC, phd/doc, parDE, higBA, relBE, hipBA, and hicAB, all of which can be plasmid or chromosomally encoded (17, 23, 24, 28). The toxin and antitoxin genes are usually cotranscribed, and the antitoxin protein acts as a transcriptional regulator, alone or in complex with the toxin, to repress the expression of the TA operon under normal growth conditions (17). Chromosomally encoded type II TA loci were originally thought to be involved in programmed cell death, as toxin overexpression reduced bacterial viability (1). However, more recent evidence showed that these chromosomal toxins exert a bacteriostatic effect which can be reversed by the subsequent controlled expression of their cognate antitoxins (6, 9, 23, 36). It is now commonly thought that chromosomal type II TA loci are involved in growth arrest that allows the bacteria to withstand unfavorable environmental conditions (7, 9, 16, 17, 30, 36, 40). The majority of type II toxins appear to target DNA replication or protein synthesis, processes that are highly expensive for the cell in terms of energy consumption, so their reduction would be beneficial during stress (7, 9, 16, 17, 22, 36).

The observation that slow-growing pathogens appear to contain numerous TA loci (33) and the association of toxin activation with bacteriostasis and antibiotic-tolerant populations (24, 44) have linked type II chromosomal TA loci to persistence and dormancy. This new role of TA loci in bacterial physiology could be of immense importance in the life cycle of pathogens that exhibit a dormant phenotype in the host, including the causative agent of tuberculosis (TB), Mycobacterium tuberculosis (20). It is estimated that a third of the world's population is infected with M. tuberculosis asymptomatically (14). An intervention that reduces the progression to active disease among these two billion latently infected people would represent an important complement to existing TB control strategies (47). Therefore, determining the mechanisms involved during the establishment, maintenance, and reactivation of latent TB is an important research goal (4). Bioinformatic studies predict that the M. tuberculosis H37Rv genome contains nearly 100 type II TA loci (3, 23, 29, 33, 37), the roles of which are only beginning to be examined (8, 11, 20, 26, 30, 40, 44, 54, 55).

The host inhibition of growth (higBA) TA locus was originally identified on the Proteus vulgaris plasmid Rts1, and it is unusual because the antitoxin-encoding gene, higA, is located downstream of the toxin-encoding gene, higB (52). This gene arrangement was thought to be a unique trait of higBA family TA loci until the newly identified hicAB family was shown to share this feature (23). Two recent studies have independently demonstrated that the sole predicted higBA locus of M. tuberculosis represents a functional TA system, since the expression of Rv1955 toxin inhibited the growth of Escherichia coli (20) and Mycobacterium smegmatis (37) and this effect was silenced when Rv1956 antitoxin was coexpressed. Although the M. tuberculosis Rv1955 and Rv1956 genes have not formally been renamed, they are referred to as higB and higA, respectively, throughout this study.

The M. tuberculosis higBA locus is unusual because the two TA genes are cotranscribed with the newly identified upstream gene Rv1954A (located opposite the annotated Rv1954c gene) and the downstream gene Rv1957 (45), both of unknown function. The Rv1954A gene lies between the two identified transcriptional start sites of higB (45). Therefore, the more distal P2 promoter controls the expression of the whole operon (namely, Rv1954A-Rv1957), while the DNA damage-inducible P1 promoter controls the expression of higB-Rv1957 only. The M. tuberculosis HigA protein is predicted to contain a helix-turn-helix DNA-binding domain toward its amino terminus (10), consistent with the hypothesis that antitoxin proteins act as transcriptional regulators (17). This study focuses on analyzing the regulatory role of HigA within M. tuberculosis.

MATERIALS AND METHODS

Bacterial strains and media.

E. coli strain DH5α (Invitrogen) was used for plasmid construction, strain XL1-Blue (Stratagene) was used for site-directed mutagenesis (SDM), and strain Rosetta 2 (DE3) (Novagen) was used for protein expression. E. coli was grown at 37°C on Luria-Bertani (LB) agar or in LB broth with shaking at 250 rpm. M. tuberculosis H37Rv was used as the wild type and as the parental strain for deletion strains. M. tuberculosis was grown at 37°C on Difco Middlebrook 7H11 agar (Becton Dickinson) or in modified Dubos medium (Difco), both supplemented with 4% Dubos medium albumin (Difco) and 0.5% or 0.2% (wt/vol) glycerol, respectively. M. tuberculosis liquid cultures were grown in a roller incubator at 2 rpm. Where appropriate, E. coli culture medium was supplemented with 100 μg/ml ampicillin, 50 μg/ml kanamycin, 20 μg/ml gentamicin, 34 μg/ml chloramphenicol, 200 μg/ml 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal), and/or 10 mg/ml glucose. Likewise, M. tuberculosis culture medium was supplemented with 25 μg/ml kanamycin, 15 μg/ml gentamicin, 50 μg/ml X-Gal, and/or 20 mg/ml sucrose.

Plasmid construction.

The plasmids used in this study are listed and their construction described in Table 1. Details of deletion-targeting plasmid construction are in the next section. All primers and oligonucleotides used in this study are listed in Table S1 in the supplemental material. SDM was performed using a QuikChange SDM kit (Stratagene). All plasmids were verified by DNA sequencing.

TABLE 1.

Plasmids used in this study

Plasmid	Description and construction	Reference or source
pET28a	Replicating E. coli protein expression vector containing both N-terminal and C-terminal His tags (kanamycin resistant)	Novagen
pBackbone	Mycobacterial suicide vector (kanamycin resistant and ampicillin resistant)	19
pUC-GM	Plasmid containing the gentamicin resistance cassette (ampicillin resistant and gentamicin resistant)	43
pGoal17	Plasmid containing the sacB-lacZ cassette (ampicillin resistant)	35
pMV306	Integrating mycobacterial cloning vector (kanamycin resistant)	25
pKP186	pMV306 derivative that does not contain integrase	39
pBS-Int	Mycobacterial suicide vector containing integrase (ampicillin resistant), electroporated in conjunction with pKP186 derivatives	46
pEJ414	pMV306 derivative containing a promoterless E. colilacZ reporter gene	34
pASF5	pET28a containing higA giving C-terminal His-tagged HigA; the 452-bp PCR product from primers Rv1956 CHis F and Rv1956 CHis R was inserted into NcoI and XhoI	This study
pASF9a	higB-Rv1957 deletion plasmid; pBackbone containing higB-Rv1957 5′ and 3′ flanking regions, gentamicin resistance cassette, and sacB-lacZ cassette	This study
pASF10a	higA deletion plasmid; pBackbone containing higA 5′ and 3′ flanking regions and sacB-lacZ cassette	This study
pASF12b	pKP186 containing higB-Rv1957 and 622 bp upstream of the experimentally determined higB translational start site; the 2,066-bp PCR product from primers Rv1955-57 comp F2 and Rv1955-57 comp R was inserted into XbaI and BamHI	This study
pAFS13	pKP186 containing the first 624 bp of the pASF12 insert as an Rv1954A/Rv1954c control complement; the 636-bp PCR product from primers Rv1955-57 comp F and Rv1954c comp R was inserted into EcoRI	This study
pASF32	pEJ414 containing the higB P2 promoter; the two 62-bp oligonucleotides Rv1955 lacZ3 and Rv1955 lacZ4 were annealed and inserted into XbaI and HindIII	This study
pASF39	pASF12 derivative where higA was deleted in-frame; SDM primers Comp SDM4F and Comp SDM4R were used	This study
pASF46	pEJ414 containing the higB P1 promoter; the 166-bp PCR product from primers lacZP1F and lacZP1R was inserted into XbaI and HindIII	This study
pASF47	pEJ414 containing the higB P2 promoter, as in pASF32 but with a longer upstream flanking region; the 165-bp PCR product from primers lacZP2F and lacZP2R was inserted into XbaI and HindIII	This study
pASF56	pEJ414 containing the Rv1954c promoter; the 323-bp PCR product from primers Rv1954clacF and Rv1954clacR was inserted into PmlI	This study
pASF58	pEJ414 containing the mutated higB P2 promoter; the two 62-bp oligonucleotides mutP2 F and mutP2 R were annealed and inserted into XbaI and HindIII	This study

^aConstruction of deletion plasmids involves multiple cloning steps (see Materials and Methods).

^bRefer to reference 45 for higB translational start site mapping.

Creating M. tuberculosis deletion strains.

The 5′ and 3′ flanking regions of interest were amplified by PCR and were sequentially cloned into the suicide plasmid pBackbone, using restriction enzymes for which sites were incorporated into the primers. The 900-bp gentamicin resistance cassette from pUC-Gm was inserted into the XbaI site of the higB-Rv1957 deletion plasmid. Finally, the 6.4-kb sacB-lacZ cassette from pGoal17 was cloned into the unique PacI site of the plasmids. Deletion plasmids were electroporated into M. tuberculosis H37Rv, and screening and counter-selection processes were performed as outlined previously (35).

Southern blotting.

An amount of 5 μg of genomic DNA was digested with PvuII, separated on an agarose gel, and treated according to standard protocols (42). DNA was transferred to a Hybond-N+ membrane (GE Healthcare) and cross-linked by UV irradiation. Southern blotting was performed using an ECL direct nucleic acid labeling and detection system (GE Healthcare). The 217-bp PCR product from primers Rv1955-57 probe F and Rv1955-57 probe R was used as the probe.

RNA preparation and qRT-PCR analysis.

M. tuberculosis cultures were grown to mid-exponential phase (optical density at 600 nm [OD₆₀₀] of 0.8), and RNA was prepared using a FastRNA pro blue kit (Qbiogene). Contaminating DNA was removed using a TURBO DNA-free kit (Ambion), and 1 μg of RNA was converted to cDNA using SuperScript II reverse transcriptase (RT) (Invitrogen) with 250 ng of random primers (Invitrogen). Quantitative RT-PCR (qRT-PCR) was carried out on a 7500 fast real-time PCR system (Applied Biosystems) using fast SYBR green master mix (Applied Biosystems). RNA without RT (RT−) was analyzed alongside cDNA (RT+). Standard curves were performed for each gene analyzed, and the quantities of cDNA within the samples were calculated from cycle threshold values. Data were averaged, adjusted for chromosomal DNA contamination (RT+ minus RT−), and normalized to corresponding sigA values.

Transcriptional start site mapping.

5′ RACE (rapid amplification of cDNA ends) (Invitrogen) was performed to map the transcriptional start site of Rv1954c according to the manufacturer's guidelines, using primers Rv1954c GSP1, GSP2, and GSP3. A single PCR product was obtained, and this was sequenced directly using the Rv1954c GSP3 primer. As 5′ RACE uses a polycytosine tail, and the complementing strand was sequenced, sites cannot be precisely mapped where transcription may begin at a guanine residue.

Microarray analysis.

Whole-genome M. tuberculosis microarray slides (version 2) were purchased from the Bacterial Microarray Group at St. George's (BμG@S), University of London. The array design is available in BμG@Sbase (accession no. A-BUGS-23; http://bugs.sgul.ac.uk/A-BUGS-23) and ArrayExpress (accession no. A-BUGS-23). Each microarray slide processed was of test cDNA competitively hybridized against single-stranded H37Rv genomic DNA, obtained from Colorado State University. Three biological replicates of each strain were each arrayed in duplicate. After DNase treatment, RNA was concentrated by ethanol precipitation. Seven micrograms of RNA was converted to cDNA and fluorescently labeled with Cy5 as described previously (12), with 3 μg of random primers. Single-stranded Cy3-labeled genomic DNA was created in a final volume of 50 μl consisting of 1× Klenow buffer (Promega), 1 μg DNA, 3 μg random primers, 100 μM each dATP/dGTP/dTTP, 40 μM dCTP, 30 μM Cy3-dCTP (GE Healthcare), and 5 U Klenow fragment (Promega). Random primers and DNA were incubated at 95°C for 5 min and immediately placed on ice, after which the remaining components were added and the reaction mixture was further incubated in the dark at 37°C for 1.5 h. Cy5-labeled cDNA and Cy3-labeled DNA were combined and purified using a MinElute PCR purification kit (Qiagen), eluting in 30.2 μl elution buffer. Microarray slide prehybridization, hybridization, and washing were performed as described previously (12). As each set of RNA was competitively hybridized against genomic DNA rather than with each other, dye-swap studies were not necessary (49).

Slides were scanned using a GenePix 4000B Scanner (Axon Instruments) with GenePix Pro 6.1 software (Axon Instruments). The image data were quantified using Bluefuse software (BlueGnome Ltd.) and were normalized and interpreted using GeneSpring GX 7.3.1 software (Agilent Technologies). Data were normalized per spot; the measured Cy5 intensity (cDNA signal) was divided by its control channel Cy3 intensity (genomic DNA signal). Data were then normalized by slide; values for each spot were divided by the median value of all spots. Log transformed data from different strains were compared to wild-type H37Rv data using analysis of variance and Student's t test (≥2-fold, P ≤ 0.05) with a Benjamini and Hochberg multiple testing correction. Fully annotated microarray data have been deposited in BμG@Sbase (accession no. E-BUGS-107; http://bugs.sgul.ac.uk/E-BUGS-107) and ArrayExpress (accession no. E-BUGS-107).

β-Gal assays.

M. tuberculosis cultures were grown to mid-exponential phase (OD₆₀₀ of 0.8). Cell-free protein extracts were prepared and filtered, and total protein and β-galactosidase (β-Gal) assays were performed as described previously (13). Specific β-Gal activity (in units/mg protein) was calculated using the formula defined by Miller (31).

HigA protein expression and purification.

A 1.5-liter culture of E. coli Rosetta 2 (DE3) containing pASF5 was grown to early-exponential phase (OD₆₀₀ of 0.5). The expression of higA-His was induced using 0.5 mM isopropyl-β-d-thiogalactopyranoside, and the culture was incubated at 37°C for 4.5 h. Cells were harvested, resuspended in 15 ml binding buffer (20 mM sodium phosphate, 500 mM sodium chloride, 2.5 mM β-mercaptoethanol, and 5 mM imidazole [pH 7.4]) containing 1 mg/ml lysozyme (Sigma) and 25 U/ml Benzonase nuclease (Novagen), and incubated on ice for 30 min. Cells were lysed by sonication for a total of 3 min. After centrifugation, the insoluble protein fraction containing HigA-His was resuspended in 15 ml binding buffer containing 8 M urea to denature and solubilize the proteins. The tagged protein was purified by nickel affinity chromatography under denatured conditions using two adjoining 1-ml HiTrap chelating HP columns (GE Healthcare). Purification was performed at a flow rate of 1 ml/min on an AKTA Prime machine (GE Healthcare). The column was washed with 5% elution buffer (binding buffer containing 500 mM imidazole and 8 M urea) and eluted in 100% elution buffer. To refold HigA-His, the urea was slowly removed by dialysis using a 3- to 12-ml, 2,000-molecular-weight-cutoff Slide-A-Lyzer dialysis cassette (Thermo Fisher Scientific). Stepwise dialysis was performed in 3 liters binding buffer containing 6 M, 4 M, 2 M, 1 M, and 0 M urea for 2 h each. Imidazole was removed during the first dialysis step. The liquid recovered from the dialysis cassette was filtered (to remove precipitate) and concentrated from 12 ml to ∼1 ml. The concentration of final pure refolded HigA-His protein was estimated to be ∼10 μg/ml (or ∼560 μM) when compared to protein standards on a Coomassie blue-stained SDS-PAGE gel.

Protein-DNA binding assays.

XbaI and HindIII sticky ends of annealed oligonucleotide fragments were filled in and labeled with [α-³²P]dCTP in a final volume of 50 μl, consisting of 1× Klenow buffer, 1 μg DNA, 200 μM each dATP/dGTP/dTTP, 2 μl Easytide [α-³²P]dCTP (0.37 MBq μl⁻¹; PerkinElmer) and 10 U Klenow fragment (Promega), at 37°C for 1.5 h. The labeled DNA probe was purified using a QIAquick nucleotide removal kit (Qiagen) and was diluted in distilled water (dH₂O) to ∼1 ng/μl. Standard protein-DNA binding reactions were performed in a final volume of 10 μl consisting of 1× PfuUltra buffer (Stratagene), 0.01 U poly(dI-dC) (Sigma), 1 μl ³²P-labeled DNA probe (equal to 1 ng or 23 fmol), and 1 μl pure refolded HigA-His (equal to 10 ng, giving a final concentration of 56 μM). When less protein was added, 1 μl of an appropriate dilution in dH₂O was used. For competition or supershift assays, the appropriate amount of unlabeled competitor DNA or rabbit anti-HigA antibody (raised and purified by Cambridge Research Biochemicals Ltd. against a synthesized peptide equivalent to residues 112 to 129 of the HigA protein) was added to the reaction mixture. Reaction mixtures were incubated at room temperature for 30 min. Protein-DNA complexes were resolved from free DNA probe on 6% nondenaturing PAGE gels in 0.5× Tris-borate-EDTA buffer at 25 mA at 4°C. Gels were dried under heat and vacuum, and radioactive bands were visualized by autoradiography.

RESULTS

The M. tuberculosis higBA TA locus is functional within its native organism.

This study set out to create two M. tuberculosis deletion strains; a higB-Rv1957 deletion strain in which the three genes were removed and replaced with a gentamicin resistance cassette and an in-frame higA deletion strain. These deletion strains were designed before the identification of Rv1954A. Transcriptional start site mapping (using 5′ RACE) was performed on Rv1954c to aid in the design of the higB-Rv1957 deletion strain. A single Rv1954c transcriptional start site was identified, located 93 to 96 bp downstream of the annotated translational start site. This result indicates that a single promoter controls the expression of Rv1954c and that Rv1954c may be shorter than predicted (Fig. 1 A). In conjunction with the published translational start site mapping of higB (45), the coding regions of Rv1954c and higB do not appear to overlap and the Rv1954c promoter does not appear to be located within the higB coding region (Fig. 1A), facilitating the design of the higB-Rv1957 deletion strain.

FIG. 1.

The M. tuberculosis higBA locus and the creation of the higB-Rv1957 deletion strain. (A) Schematic of the M. tuberculosis higBA locus (not to scale). The open reading frames are indicated by thick arrows, the positions of the higB and Rv1954c promoters are shown by bent arrows, and the region deleted within the M. tuberculosis higB-Rv1957 deletion strain is indicated by the double-headed arrow. The relative locations of the PvuII sites and the probe used for Southern blot analysis are marked. (B) Southern blot analysis of three independently generated M. tuberculosis higB-Rv1957 deletion strains (1 to 3) alongside wild-type H37Rv (WT). The PvuII site within HigA of the wild-type strain is absent within the three deletion strains, and so a larger-sized product was detected. Deletion strain 1 was used for all future experiments. (C) qRT-PCR of higB-Rv1957 was performed on RNA extracts of wild-type M. tuberculosis H37Rv, the higB-Rv1957 deletion strain, and the full and control complementation strains. The data presented are averages of results for three biological replicates, each assayed in triplicate, and error bars represent standard deviations. higB-Rv1957 transcription was undetectable within the higB-Rv1957 deletion strain and the control complementation strain (i.e., RT+ values were statistically equal to RT− values; P > 0.05, Student's t test). There were no significant differences in normalized higB-Rv1957 transcription levels between the full complementation strain and the wild-type strain (P > 0.05, Student's t test).

A two-step strategy (35) was utilized to create deletion strains in this study. Strains containing the higB-Rv1957 deletion were readily isolated; 11 of the 12 white, gentamicin-resistant, sucrose-resistant, kanamycin-sensitive colonies analyzed by PCR were as expected for the higB-Rv1957 deletion (data not shown), and three of these were further confirmed by Southern blotting (Fig. 1B). However, we were unable to obtain an M. tuberculosis in-frame higA deletion strain.

As the creation of a M. tuberculosis in-frame higA deletion strain was unsuccessful, attempts were made to produce an equivalent strain by complementation. An in-frame higA deletion was made within the full complementation plasmid pASF12 by SDM, to give pASF39. Numerous colonies were obtained when pASF12 was integrated into the M. tuberculosis higB-Rv1957 deletion strain, but attempts to independently integrate pASF39 yielded no viable colonies. These observations are consistent with the hypothesis that HigA is an antitoxin and is required to inhibit the toxic effect of HigB. The M. tuberculosis higBA TA locus therefore appears to be functional within its native organism. However, deleting higB-Rv1957 together did not affect the in vitro growth of M. tuberculosis (data not shown).

The presence of pASF12 restored higB-Rv1957 transcription levels within the M. tuberculosis higB-Rv1957 deletion strain back to wild-type H37Rv levels, demonstrating successful complementation (Fig. 1C). As the pASF12 plasmid contains the higB P2 promoter, it also contains Rv1954A and/or Rv1954c. A control complementation plasmid, encoding Rv1954A and/or Rv1954c only (pASF13), was therefore also independently integrated into the M. tuberculosis higB-Rv1957 deletion strain to enable the effects of Rv1954A and/or Rv1954c to be distinguished from those of higB-Rv1957. The presence of pASF13 within the deletion strain did not affect higB-Rv1957 transcription levels (Fig. 1C).

Global gene expression profiling identifies potential HigA-regulated genes.

To investigate whether HigA acts as a global transcriptional regulator, the effects of higB-Rv1957 deletion on global gene expression were assessed. The M. tuberculosis higB-Rv1957 deletion strain and its two complementation strains were analyzed alongside wild-type H37Rv using cDNA microarrays. The relative transcription levels of 61 M. tuberculosis genes differed significantly, by at least 2-fold, upon higB-Rv1957 deletion compared to their expression in the wild type (see Table S2 in supplemental material), although these genes did not include higB-Rv1957 themselves. This discrepancy is most likely due to the low higB-Rv1957 transcription levels within wild-type M. tuberculosis H37Rv.

If a certain gene is regulated by HigA, the effect on transcription upon higB-Rv1957 deletion would be restored to wild-type levels within the full complementation strain and would remain as for the higB-Rv1957 deletion strain within the control complementation strain. Of the 61 genes affected by higB-Rv1957 deletion, only seven genes showed this pattern of expression (Table 2). The microarray PCR product probe for the annotated Rv1954c gene was also internal to the newly identified Rv1954A gene. Rv1954c and/or Rv1954A was the most differentially expressed gene within the M. tuberculosis higB-Rv1957 deletion strain, being significantly upregulated, by 55.7-fold (Table 2). Although the Rv1954c and/or Rv1954A transcription level within the full complementation strain was still statistically different from the wild-type-strain levels, complementation was considered to have been successful because the fold change dropped from 55.7-fold to 1.9-fold (Table 2). Rv1954c and/or Rv1954A was upregulated by 90.1-fold within the control complementation strain (Table 2); this 1.6-fold increase in expression compared to the expression level in the higB-Rv1957 deletion strain can be explained by the fact that the control complement contains an additional copy of Rv1954c and/or Rv1954A but does not contain higB-Rv1957. These changes in Rv1954c and/or Rv1954A expression were confirmed by qRT-PCR (Fig. 2) and strongly imply that HigA acts as a repressor of the Rv1954c promoter, the higB P2 promoter, or both of these promoters.

TABLE 2.

Microarray data for genes that may be regulated by HigA

Locus tag	Gene name	Microarray data (compared to results for H37Rv) for each M. tuberculosis strain
		higB-Rv1957 deletion		Full complement		Control complement
		Fold regulated	P value	Fold regulated	P value	Fold regulated	P value
Rv0860	fadB	2.3↓	0.001	1.6↓	0.021	2.0↓	0.002
Rv1953	Rv1953	2.1↑	3.E-05	1.4↑	0.002	2.7↑	0.001
Rv1954c	Rv1954c	55.7↑	4.E-07	1.9↑	0.048	90.1↑	2.E-09
Rv2919c	glnB	2.1↓	0.003	1.3↓	0.520	2.1↓	0.001
Rv3173c	Rv3173c	2.3↓	0.023	1.8↓	0.264	3.7↓	0.002
Rv3290c	lat	2.7↓	0.024	1.8↓	0.252	3.3↓	0.008
Rv3662c	Rv3662c	3.9↓	1.E-04	1.6↓	0.111	2.9↓	3.E-04

FIG. 2.

Results of qRT-PCR of selected genes to assess the effects of higB-Rv1957 deletion and full and control complementation on transcription. qRT-PCR of selected genes was performed on RNA extracts of wild-type M. tuberculosis H37Rv, the higB-Rv1957 deletion strain, and the full and control complementation strains. The data presented are averages of the results for three biological replicates, each assayed in triplicate, and error bars represent standard deviations. It is indicated when the gene's normalized transcription level within a selected strain was significantly different from the equivalent wild-type-strain level (*, P < 0.05; Student's t test).

Of the six other genes where the pattern of expression was as expected for HigA-dependent regulation, only Rv1953 was upregulated upon higB-Rv1957 deletion (Table 2). However, qRT-PCR analysis indicated that Rv1953 expression was not restored to wild-type levels within either of the complementation strains (Fig. 2), so Rv1953 does not appear to be regulated by HigA. Conversely, the microarray data for fadB, Rv3173c, and Rv3662c were confirmed by qRT-PCR (Fig. 2). Therefore, in addition to repressing Rv1954c and/or higBA locus expression, HigA may be involved in the activation of fadA-fadB, Rv3173c, and dppA-Rv3662 expression.

HigA binds to a perfect palindromic DNA motif and directly represses expression from the higB P2 promoter.

Promoter-lacZ transcriptional fusions were initially used to investigate the role of M. tuberculosis HigA in the regulation of the higBA locus and Rv1954c. The activity of the higB P2 promoter (from pASF47) was significantly higher within the M. tuberculosis higB-Rv1957 deletion strain than within wild-type H37Rv, while the higB P1 promoter (from pASF46) and the Rv1954c promoter (from pASF56) were unaffected by the absence of higB-Rv1957 (Fig. 3). The higB P2 promoter is therefore directly or indirectly repressed by HigA, while the higB P1 promoter and the Rv1954c promoter do not appear to be regulated by HigA.

FIG. 3.

The results of β-Gal assays indicate that the higB P2 promoter is repressed by HigA. β-Gal assays were performed on cell-free protein extracts of wild-type M. tuberculosis H37Rv and the higB-Rv1957 deletion strain containing the promoter-lacZ fusion plasmids pASF46 (higB P1 only), pASF47 (higB P2 only), and pASF56 (Rv1954c promoter only) alongside the empty vector control (pEJ414). The data presented are averages of the results for at least two biological replicates, each assayed in duplicate, and error bars represent standard deviations. The β-Gal activity level for the higB-Rv1957 deletion strain containing pASF47 was significantly higher than the equivalent wild-type-strain level (*, P < 0.05; Student's t test).

Carboxyl-terminal His-tagged HigA protein (HigA-His) was purified by nickel affinity chromatography under denaturing conditions for use within protein-DNA binding assays. Although a large quantity of pure denatured recombinant HigA protein was obtained (Fig. 4, lane 5), the majority of this precipitated out of solution upon dialysis to remove the denaturant and refold the protein (Fig. 4, lane 6). However, a small quantity of pure recombinant HigA protein remained soluble after dialysis (Fig. 4, lanes 7 and 8). Mass spectrometry verified the identity and integrity of the protein (S. Howell, personal communication), and circular dichroism indicated that the protein had secondary structure (S. Martin, personal communication).

FIG. 4.

Expression and purification of recombinant HigA protein. Coomassie blue-stained SDS-PAGE showing the results at various stages of carboxyl-terminal His-tagged HigA protein expression (from pASF5), purification (using nickel affinity chromatography), refolding by dialysis, and concentration. Lanes: 1, soluble protein fraction; 2, solubilized denatured insoluble protein fraction; 3, purification flowthrough fraction; 4, purification wash fraction; 5, purification elution fraction; 6, precipitated protein after dialysis; 7, filtered soluble protein after dialysis; 8, filtered soluble protein concentrated 12-fold after dialysis. Right- and leftmost lanes contain protein ladders (Bio-Rad).

Protein-DNA binding assays were performed between recombinant HigA and a 68-bp, ³²P-labeled higB P2 promoter DNA fragment (formed by annealing the complementary oligonucleotides Rv1955 lacZ3 and Rv1955 lacZ4). HigA bound to the higB P2 promoter in a concentration-dependent manner (Fig. 5A). This interaction was shown to be sequence specific, since the formation of the protein-probe complex was inhibited by unlabeled specific competitor DNA (the higB P2 promoter fragment) (Fig. 5B) but was not inhibited by unlabeled nonspecific competitor DNA (the higB P1 promoter fragment, formed by annealing the complementary oligonucleotides Rv1955 lacZ1 and Rv1955 lacZ2) (Fig. 5C). The latter also demonstrates that HigA does not bind to the higB P1 promoter. To confirm that HigA was responsible for the retardation of the higB P2 probe, a “supershift” assay was performed where anti-HigA antibody was added to the reaction mixture, and an antibody-protein-probe complex was observed (Fig. 5D).

FIG. 5.

Recombinant HigA protein directly and specifically binds to the higB P2 promoter. Protein-DNA binding assays were performed between pure refolded HigA-His protein and the 68-bp ³²P-labeled higB P2 promoter probe (23 fmol) using increasing amounts of protein (A), unlabeled higB P2 promoter fragment (identical to the probe) as specific competitor (B), unlabeled higB P1 promoter fragment as nonspecific competitor (C), and anti-HigA antibody (D). The concentrations of recombinant protein used were 3.5, 7, 14, and 28 μM for lanes 2 to 5, respectively, 0 μM for lanes 1 and 17, and 56 μM for all other lanes. Competitor DNA fragments were added in 5-, 10-, 100-, and 200-fold molar excess of the labeled probe (lanes 8 to 11 and 13 to 16). Anti-HigA antibody was used at final dilutions of 1 in 5,000, 500, 50, and 5 for lanes 19 to 22, respectively.

A perfect palindromic DNA motif [ATATAGG(N₆)CCTATAT] was identified within the 68-bp higB P2 promoter region (Fig. 6A). To investigate whether this motif was required for HigA binding, complementary oligonucleotides (mutP2 F and mutP2 R) were designed in which every other base of the palindromic motif was mutated (Fig. 6A). Binding of recombinant HigA to the ³²P-labeled mutated higB P2 promoter DNA fragment could not be detected (Fig. 6B). Furthermore, when the mutated DNA fragment was used as an unlabeled competitor within binding assays between HigA and the ³²P-labeled wild-type higB P2 promoter DNA fragment, the formation of the protein-probe complex was not inhibited (Fig. 6C). Specific HigA binding of the 68-bp higB P2 promoter DNA fragment was therefore abolished upon mutation of the identified perfect palindromic motif.

FIG. 6.

Identification of the HigA DNA-binding motif and the effect of mutating this motif on HigA binding and higB P2 promoter activity. (A) The wild-type and mutated sequences of the 68-bp higB P2 promoter fragment. The experimentally determined P2 promoter transcriptional start site (45) and putative −10 and −35 sequences are in boldface and boxed, respectively. The identified perfect palindromic DNA motif is underlined, with mutated bases highlighted in gray. (B) DNA-binding assays were performed between pure refolded HigA-His protein (0 and 56 μM for lanes marked − and +, respectively) and the ³²P-labeled mutated higB P2 promoter probe (23 fmol). (C) Protein-DNA-binding assays were performed between pure refolded HigA-His protein (56 μM) and the ³²P-labeled wild-type higB P2 promoter probe (23 fmol) using increasing amounts of unlabeled mutated higB P2 promoter fragment as competitor (5-, 10-, 100-, and 200-fold molar excess of the labeled probe for lanes 2 to 5, respectively). (D) β-Gal assays were performed on cell-free protein extracts of wild-type M. tuberculosis H37Rv and the higB-Rv1957 deletion strain containing the promoter-lacZ fusion plasmids pASF32 (wild-type higB P2 promoter) and pASF58 (mutated higB P2 promoter) alongside the empty vector control (pEJ414). The data presented are averages of the results for three biological replicates, each assayed in duplicate, and error bars represent standard deviations. The β-Gal activity level for the higB-Rv1957 deletion strain containing pASF32 was significantly higher than the equivalent wild-type-strain level (*, P < 0.05; Student's t test). The β-Gal activity level for the higB-Rv1957 deletion strain containing pASF58 was not significantly different from the equivalent wild-type-strain level (P > 0.05; Student's t test).

To determine whether HigA binding to the higB P2 promoter has a biological effect on promoter activity, wild-type (oligonucleotides Rv1955 lacZ3 and Rv1955 lacZ4) and mutated (oligonucleotides mutP2 F and mutP2 R) versions of the higB P2 promoter were transcriptionally fused to the lacZ reporter gene. Upon mutation of the identified perfect palindromic HigA-binding motif (within pASF58), higB P2 promoter activity within wild-type H37Rv was elevated to levels similar to those seen for the wild-type motif (within pASF32) within the higB-Rv1957 deletion strain (Fig. 6D). HigA therefore appears to directly repress expression from the higB P2 promoter by binding to the perfect palindromic DNA motif, and this repression was alleviated upon mutation of this DNA-binding motif.

HigA is unlikely to directly control the expression of any other M. tuberculosis genes.

A DNA pattern search did not identify the perfect palindromic HigA DNA-binding motif anywhere else within the M. tuberculosis H37Rv genome (TubercuList database). The next-most-similar DNA motif identified contained two mismatches to the HigA DNA-binding motif and is located within the 152-bp intergenic region between the divergent genes whiB5 and Rv0023 (Fig. 7A). No binding of recombinant HigA to a 68-bp ³²P-labeled whiB5-Rv0023 DNA fragment (formed by annealing the complementary oligonucleotides whiB5-Rv0023 F and whiB5-Rv0023 R) could be detected (Fig. 7B). Furthermore, when the whiB5-Rv0023 DNA fragment was used as an unlabeled competitor within binding assays between HigA and the ³²P-labeled wild-type higB P2 promoter DNA fragment, the formation of the protein-probe complex was not inhibited (Fig. 7C). Therefore, HigA does not appear to bind to the DNA motif between the whiB5 and Rv0023 genes. While it remains possible that a site with more mismatches could bind HigA if some positions are more critical than others, the lack of a recognizable motif upstream of the few genes identified as potentially regulated by HigA (by microarray analysis) supports the notion that HigA is unlikely to directly control the expression of any other M. tuberculosis genes in addition to its own operon.

FIG. 7.

HigA does not bind to the next-most-closely related motif within the genome. (A) Sequence of the 68-bp fragment containing the DNA motif (underlined) between whiB5 and Rv0023, with the two bases mismatched to the HigA DNA-binding motif highlighted in gray. (B) Protein-DNA binding assays were performed between pure refolded HigA-His protein (0 and 56 μM for lanes marked − and +, respectively) and the ³²P-labeled whiB5-Rv0023 probe (23 fmol). (C) Protein-DNA-binding assays were performed between pure refolded HigA-His protein (56 μM) and the ³²P-labeled wild-type higB P2 promoter probe (23 fmol) using increasing amounts of unlabeled whiB5-Rv0023 fragment as competitor (5-, 10-, 100-, and 200-fold molar excess of the labeled probe for lanes 2 to 5, respectively).

DISCUSSION

This study was designed to investigate the transcriptional regulatory properties rather than the antitoxin properties of M. tuberculosis HigA. Nevertheless, we demonstrate that the M. tuberculosis higBA locus is functional within its native organism, as attempts to delete higA alone, either directly or by partial complementation of the higB-Rv1957 deletion strain, resulted in apparent cell death. Although chromosomal TA loci are thought to be bacteriostatic, not bactericidal, prolonged overexpression of toxin can result in a “point of no return” situation leading to cell death (2), and this presumably occurs here in the absence of the antitoxin. However, deleting higB-Rv1957 together did not affect the in vitro growth of M. tuberculosis, and so the higA antitoxin is only essential in the presence of the higB toxin. A similar situation was observed when various deletions of the higBA1 locus within Vibrio cholerae were attempted (6). This phenomenon does not exclude the possibility that the M. tuberculosis higBA locus may play an essential role in vivo.

Analysis of the effects of higB-Rv1957 deletion on M. tuberculosis global gene expression found that the most pronounced effect was on Rv1954c and/or the newly identified Rv1954A. Promoter activity studies demonstrated that the higB P2 promoter was modulated by HigA while the higB P1 promoter and the Rv1954c promoter were not. This was further confirmed by protein-DNA binding assays using recombinant HigA protein. Upon the mutation of a perfect palindromic DNA motif identified within the higB P2 promoter region, HigA binding was abolished, and this conferred a derepressed P2 promoter phenotype within wild-type M. tuberculosis H37Rv. The two halves of this palindrome are located around a putative −35 promoter motif (Fig. 6). HigA therefore appears to directly repress expression from the higB P2 promoter by binding to this palindromic DNA motif, thereby preventing RNA polymerase from binding to the promoter.

It is believed that the higB P2 promoter controls Rv1954A-Rv1957 expression (45), and so HigA appears to repress the whole higBA operon. This would further explain the M. tuberculosis cell death which occurred when the attempt to delete higA alone was made, as the complete removal of the HigA antitoxin would completely derepress the higB P2 promoter and allow unnaturally high levels of the higB toxin to be expressed. It is possible that the higB P2 promoter primarily expresses Rv1954A, with only partial read-through into higB-Rv1957. Even if this is the case, the dramatic increase in expression from P2 in the absence of HigA would be expected to lead to increased transcription of these downstream genes.

The M. tuberculosis HigA palindromic DNA-binding motif was defined as ATATAGG(N₆)CCTATAT. HigA failed to bind to the next-most-similar motif within M. tuberculosis, which is located within the intergenic region between the divergent genes whiB5 and Rv0023 and contains two mismatches to the identified HigA DNA-binding motif. This illustrates that the two nucleotides in question are important for HigA binding. In agreement with this binding data, neither whiB5 nor Rv0023 was differentially expressed upon higB-Rv1957 deletion. It remains possible that HigA might bind to other, less-well-conserved motifs if certain positions are more critical than others; this could be assessed in vitro by extensive binding analyses using many probes of various sequences or in vivo by global chromatin immunoprecipitation (ChIP) assays (ChIP and microarray analysis [ChIP-chip] or ChIP and high-throughput sequencing [ChIP-seq]), but such analyses are beyond the scope of this study. A small number of genes were identified whose expression patterns in the M. tuberculosis higB-Rv1957 deletion and the two complementation strains were as expected for HigA-dependent regulation (namely, fadB, Rv3173c, and Rv3662c), but no recognizable motifs resembling the confirmed HigA binding site could be identified in their upstream regions, suggesting that their expression pattern is more likely to be an indirect effect of higB-Rv1957 deletion and/or high Rv1954A expression. Taken together, these observations suggest that it is unlikely that M. tuberculosis HigA directly regulates any other genes in addition to its own operon.

No other chromosomally encoded HigA antitoxin DNA-binding motifs have been experimentally determined, but the HigA DNA-binding region of the Rst1 plasmid-borne higBA TA locus consists of two palindromic motifs with a consensus sequence of GTATTACAC(N₃)GTGTAATAC (51). There does not appear to be any correlation between this motif and the M. tuberculosis HigA DNA-binding motif other than that they are both AT-rich palindromes. The Rst1 plasmid-borne HigB toxin appears to corepress higBA expression (50, 51), while the chromosomally encoded HigB1 toxin of V. cholerae does not appear to function as a corepressor (6). It is yet to be investigated whether the M. tuberculosis HigB toxin acts as a corepressor of Rv1954A-Rv1957 expression. In addition, the role of the HigA-independent, DNA damage-inducible P1 promoter also warrants further investigation.

The results of gene expression analyses indicate that the M. tuberculosis higBA locus is upregulated when the bacterium is exposed to heat shock (48), nutrient starvation (5), DNA damage (38), and hypoxia (37, 41), supporting the hypothesis that TA loci are involved in survival under stress. Hypoxia is known to be a major factor in inducing dormancy (or nonreplicating persistence) of M. tuberculosis (53), and so it is possible that the M. tuberculosis higBA locus is important in pathogenesis, perhaps playing a role in inducing dormancy and allowing long-term survival of the bacteria within the host.