Epigenetic Regulation of Matrix Metalloproteinase-1 and -3 Expression in Mycobacterium tuberculosis Infection

In pulmonary tuberculosis (TB), the inflammatory immune response against Mycobacterium tuberculosis (Mtb) is associated with tissue destruction and cavitation, which drives disease transmission, chronic lung disease, and mortality. Matrix metalloproteinase (MMP)-1 is a host enzyme critical for the development of cavitation. MMP expression has been shown to be epigenetically regulated in other inflammatory diseases, but the importance of such mechanisms in Mtb-associated induction of MMP-1 is unknown. We investigated the role of changes in histone acetylation in Mtb-induced MMP expression using inhibitors of histone deacetylases (HDACs) and histone acetyltransferases (HAT), HDAC siRNA, promoter-reporter constructs, and chromatin immunoprecipitation assays. Mtb infection decreased Class I HDAC gene expression by over 50% in primary human monocyte-derived macrophages but not in normal human bronchial epithelial cells (NHBEs). Non-selective inhibition of HDAC activity decreased MMP-1/-3 expression by Mtb-stimulated macrophages and NHBEs, while class I HDAC inhibition increased MMP-1 secretion by Mtb-stimulated NHBEs. MMP-3 expression, but not MMP-1, was downregulated by siRNA silencing of HDAC1. Inhibition of HAT activity also significantly decreased MMP-1/-3 secretion by Mtb-infected macrophages. The MMP-1 promoter region between −2,001 and −2,942 base pairs from the transcriptional start site was key in control of Mtb-driven MMP-1 gene expression. Histone H3 and H4 acetylation and RNA Pol II binding in the MMP-1 promoter region were increased in stimulated NHBEs. In summary, epigenetic modification of histone acetylation via HDAC and HAT activity has a key regulatory role in Mtb-dependent gene expression and secretion of MMP-1 and -3, enzymes which drive human immunopathology. Manipulation of epigenetic regulatory mechanisms may have potential as a host-directed therapy to improve outcomes in the era of rising TB drug resistance.

. The rapid emergence of widespread drug resistance necessitates new strategies to improve the efficacy of treatment in TB, both to decrease transmission and to improve patient outcomes. Ideally, such therapies will shorten the duration of therapy, which is currently a minimum of 6 months and may be years in drugresistant disease. Host-directed therapies are increasingly of interest in TB (2). The primary site of Mycobacterium tuberculosis (Mtb) infection is the lung, and pulmonary disease is characterized by granulomatous inflammation with destruction of lung parenchyma. The outcome of infection is very variable between hosts, and the factors determining this are not well understood, although host genetics and innate immune responses are important determinants of disease (3,4). Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases, which have key roles in tissue repair and in diseases characterized by inflammatory tissue destruction such as emphysema (5). MMPs are key mediators of inflammatory cell migration, and modulators of chemokine and cytokine signaling (6,7). MMP activity is strongly implicated in the immunopathogenesis of TB. Our group and others demonstrated the involvement of MMP-1, the major human collagenase, and its activator MMP-3 (stromelysin-1) in driving pathology in pulmonary TB (8)(9)(10). MMPs are secreted by Mtb-infected monocytes and macrophages, and also by uninfected stromal cells stimulated via intercellular networks (11). Epigenetic mechanisms are emerging as major regulators of MMP activity in non-infectious diseases (12,13), including chronic lung diseases such as asthma and COPD (14), but their role in MMP expression in TB is less established.
Epigenetic regulation encompasses all chromosomal modifications that alter gene expression without altering the nucleotide sequence of coding DNA (15,16). Eukaryotic DNA is packaged as chromatin around octamers of histone proteins, which contain globular domains and negatively charged tails. These are subject to extensive post-translational modification, including acetylation of highly conserved lysine residues. Acetylation of histones H3 and H4, carried out by Histone acetyltransferases (HATs), is associated with increased gene transcription (17). Conversely, acetyl groups are removed by the histone deacetylases (HDACs), which are divided into four classes. The Class I HDACs (1, 2, 3, and 8) are ubiquitously expressed, whereas Class II HDACs (such as HDAC 4 and 5) are selectively expressed in different tissues. HDAC activity is usually associated with silencing of gene expression (18,19). However, this is not uniformly the case and opposite regulation may occur. For example, HDAC inhibition reduced MMP-9 gene expression in cancer cell lines resulting in a less invasive phenotype (20).
The epigenetic mechanisms regulating inflammatory immune responses in TB are an emerging field. Altered miRNA expression in serum and sputum from TB patients compared to controls has been shown, and potential biomarkers for diagnosis have been identified (21). A growing body of evidence exists to support the importance of epigenetic mechanisms in other respiratory infections, for example, altered DNA methylation patterns in asthma patients versus healthy controls have been implicated in the pathogenesis of rhinovirus infection (22). Similarly, altered DNA (cytosine-5-)-methyltransferase-1 (DMT-1) expression in nasal epithelial cells from smokers was identified as a possible mechanism of increased susceptibility to influenza (23). In vitro studies of airway epithelial cells demonstrated increased HDAC2 expression and decreased histone acetylation in respiratory syncytial virus (RSV)-infected cells, while chemical HDAC inhibition restricted RSV replication (24). In the current study, we have investigated whether epigenetic modifications, specifically histone acetylation/deacetylation, regulated the characteristic TB-associated expression of MMP-1 and MMP-3 by monocytederived macrophages and normal human bronchial epithelial cells (NHBEs), thereby augmenting TB immunopathology. The role of histone acetylation in induction of MMP-1/-3 expression was specifically investigated, since this dynamic epigenetic mark is associated with transcriptional activation. We demonstrate that Mtb infection alters macrophage Class I HDAC expression and that MMP-1 expression induced by Mtb is sensitive to HDAC/ HAT inhibition. In addition, increased histone acetylation was seen at MMP-1 and -3 promoter regions compared with unstimulated cells, specifically in the region −2,001 to −2,942 bp of the MMP-1 promoter, which contains key inducible sites activated in Mtb-stimulated cells.

MaTerials anD MeThODs reagents and antibodies
Trichostatin A (TSA) was purchased from Sigma-Aldrich (Gillingham, UK), CBHA, HAT inhibitor II and Anacardic acid (AA) from Calbiochem (Millipore, Watford, UK), and MS-275 from Enzo Life Sciences (Exeter, UK). Primary rabbit anti-human acetyl-histone H3 and acetyl-histone H4 (Millipore) were used for chromatin immunoprecipitation. Primary mouse anti-human HDAC4 and anti-HDAC7 were used for Western blot and HRP-linked goat anti-rabbit IgG and goat anti-mouse (all from Cell Signalling, Hertfordshire, UK) were used as secondary antibodies. All other reagents were purchased from Sigma-Aldrich unless otherwise stated.

Mtb culture
Mycobacterium tuberculosis strain H37Rv was cultured from frozen stocks stored at −80°C in Middlebrook 7H9 broth (BD Biosciences, Oxford, UK) supplemented with 10% OADC enrichment medium (BD Biosciences), 0.2% glycerol, and 0.02% Tween 20 with agitation at 37°C. Growth was monitored by measuring optical density (OD) using a Biowave cell density meter (WPA, Cambridge, UK). Infection experiments were performed using cultures at mid-log growth (at OD 0.55-0.65) corresponding to 1-2 × 10 8 CFU/ml. Correlation with optical density was checked by performing colony counts in triplicate on Middlebrook 7H11 agar. Cells were infected at a multiplicity of infection (MOI) of 1 unless otherwise stated.

cell culture
Monocytes were adhesion-purified from leukocyte residues from healthy blood donors (NHS Blood Transfusion Service) and differentiated into macrophages for 4 days in RPMI 1640 (Life Technologies, Paisley, UK) supplemented with 10% fetal bovine serum (FBS) and 100 ng/ml M-CSF (R&D Systems, Abingdon, UK). After a further 24 h without M-CSF, the medium was changed for M-SFM (Life Technologies) and cells were infected with Mtb H37Rv strain. Cells were pretreated with chemical inhibitors for 2 h prior to infection where relevant.
Primary NHBEs (Lonza, Wokingham, UK) were cultured according to the supplier's instructions in supplemented bronchial epithelial growth medium (BEGM). Medium was replaced every 3 days. Cells were subcultured at 80% confluence and used at passage 4 or 5.
conditioned Medium from Mycobacterium tuberculosis-infected Monocytes (coMTb) Peripheral blood monocytes isolated as above from healthy blood donors were infected with H37Rv at a MOI of 1 in RPMI without FBS for 24 h. The culture medium was then collected and sterilized by passage through a 0.2 µM Anopore syringe filter (Whatman, Brentford, UK). Paired samples of conditioned medium from uninfected monocytes (CoMCont) from the same donor were used as controls.

MMP elisas
Supernatants were collected at 72 h poststimulus, sterile filtered, and MMPs were quantified using the Duoset MMP-1 and MMP-3 ELISA kits (R&D Systems) according to manufacturer's instructions. Lower limits of sensitivity for the Duoset kits are: 156 pg/ ml for MMP-1 and 31.2 pg/ml for MMP-3. Samples were run with appropriate controls and at dilutions calculated to give readings within the linear range of detection as indicated by the manufacturer.

luminex Multiplex immunoassay
Quantification of MMP-1, -3, -7, and -9 concentrations was performed using the Fluorokine MultiAnalyte Profiling MMP Base Kit (R&D Systems) and the Luminex platform Bio-Plex 200 (Bio-Rad, Hemel Hempstead, UK) dual laser analyzer. Standard curves were generated using Bio-Plex Manager version 5.0. Lower limits of sensitivity for the Fluorokine Luminex were: 1.1 pg/ml for MMP-1, 7.3 pg/ml for MMP-3, 6.6 pg/ml for MMP-7, and 13.7 pg/ml for MMP-9. All samples were run with appropriate controls and were within the linear range of detection as indicated by the manufacturer.

Western Blotting
Cells were washed in PBS and scraped and homogenized in lysis buffer (62.5 mM Tris, 2% SDS, 10% glycerol, 50 mM DDT with bromophenol blue) prior to storage at −80°C. Samples were denatured by heating to 90°C for 5 min and loaded on 4-12% NuPAGE Bis-Tris mini gels (Life Technologies) and run at 200 V for 70-90 min. Proteins were transferred to a nitrocellulose membrane (GE Healthcare) at 4°C, 30 V, for 1 h before blocking with 5% skimmed milk protein/TBS/0.1% Tween 20 buffer and incubating with diluted primary antibody. Membranes were incubated with diluted HRP-linked secondary antibody and developed using ECL Prime developing kit (GE Healthcare, Hatfield, UK).
A549 cells were transfected when 60% confluent with FuGene 6 (Roche, Lewes, UK), and 0.8 µg plasmid DNA or control plasmid DNA. Sixteen hours after transfection, the cells were stimulated according to the experimental conditions. Twenty-four hours later, cells were washed once in PBS and lysed in passive lysis buffer (Promega). Luciferase assays were performed using the Promega Dual-Luciferase Reporter Assay kit (Promega) using an L-Max 2 luminometer (Molecular Devices, Sunnydale, CA, USA).

real-time Pcr
After 24 h incubation in the specified experimental settings, cells were lysed in TRI-reagent and total RNA extracted with the PureLink RNA mini kit (Life Technologies) with on-column DNase treatment. RNA concentrations and purity were evaluated using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE, USA). A total of 1 µg of sample RNA was reverse transcribed using the Quantitect RT Kit (Qiagen, Crawley, UK) according to the manufacturer's instructions. Real-time PCR was performed using Brilliant II qPCR mastermix (Agilent, UK) on a Stratagene Mx3000p platform (Stratagene, La Jolla, USA). The thermal profile was 10 min at 95°C, followed by 40-45 cycles of 30 s at 95°C and 1 min at 60°C. The cycle threshold (Ct) at which amplification entered the exponential phase was determined for each well and analyte. 18S ribosomal RNA, β-actin, and GAPDH RNA were used as reference genes. The following primers and probes were used to analyze target and reference genes.

Transfection of epithelial cells with sirna
ON-TARGETplus SMARTpool siRNA oligonucleotides and transfection reagents were purchased from Dharmacon (Thermo Scientific). NHBEs were cultured in complete medium and transfected at 60-70% confluence. siRNA and Lipofectamine 2000 were diluted in Optimem and used at final concentrations of 30 nM and 25 µg/ml, respectively. After 4 h, each well was washed with PBS and fresh BEGM was added. The cells were then rested overnight prior to stimulation with the experimental conditions. For analysis of mRNA expression, cells were lysed and total RNA was extracted 24 h after stimulation. For analysis of protein expression, cell culture supernatants were collected for ELISA and cells were washed in PBS and lysed in Western lysis buffer 48 h poststimulation.

chromatin immunoprecipitation assay (chiP)
Magna-ChIP kits and antibodies were purchased from Millipore (Watford, UK) and used according to manufacturer's instructions. The rabbit polyclonal anti-acetyl-Histone H4 antibody was raised against tetra acetylated H4 and recognizes intermediately acetylated H4, but not acetylation on lysine 16. The rabbit polyclonal anti-acetyl-Histone 3 antibody was raised against acetylated N-terminus of H3.
These assays were not performed in technical triplicates due to the number of cells required per condition.
statistics Unless otherwise stated, results shown are from experiments performed in triplicate and representative of at least two independent experiments. Comparisons between two groups were made using the Student's t-test (two-tailed with significance set as p < 0.05). For comparison of three or more groups, one-way ANOVA was used with Tukey's correction for multiple comparisons. Unless otherwise stated in figure legends, graphs show mean values for triplicate samples and error bars are the SD.

resUlTs class i hDacs are suppressed by Mtb infection of Macrophages
Experiments were designed to investigate epigenetic regulation of TB immunopathology, first investigating whether Mtb infection of macrophages altered Class I HDAC expression. Expression of Class I HDACs was repressed by Mtb infection, with a 68% decrease in HDAC1, 69% in HDAC2, 76% in HDAC3, and 58% in HDAC8 compared with uninfected controls (all p < 0.05) (Figures 1A-D). Next, we examined monocyte networkdependent stimulation of NHBEs by CoMTb. In contrast to macrophage stimulation by Mtb, CoMTb stimulation of NHBEs did not significantly alter accumulation of any class I HDAC mRNAs (Figures 1E-H). Expression of Class II HDACs 4 and 5 was also examined by RT-PCR. HDAC4 mRNA accumulation was increased by 48% in Mtb-infected macrophages, while no difference was observed for HDAC5 (Figures 1I,J). Increase in HDAC4 protein level was confirmed by western blot ( Figure  S1 in Supplementary Material). Class II HDAC expression in CoMTb-stimulated NHBEs was similar to control conditions (Figures 1K,L). Thus, exposure to live Mtb selectively suppressed macrophage class I HDAC expression.   (Figures 2A,B). In NHBEs, TSA significantly decreased baseline MMP-1 secretion by over 50% (p < 0.05; Figure 2C). CoMTbstimulated MMP-1 secretion was decreased by TSA treatment, although this did not reach statistical significance (Figure 2C). In contrast, CoMTb-stimulated MMP-3 secretion was significantly reduced in a dose-dependent manner by 25% with 1 ng/ml TSA (from 2.4 to 1.8 ng/ml) and by 72% with 10 ng/ml TSA (685 pg/ml) ( Figure 2D). Further experiments were performed using an alternative chemical HDACi, CBHA (m-Carboxycinnamic Acid bis-Hydroxamide), to confirm that the observed effects were due to HDAC inhibition and not non-specific. Pretreatment of macrophages with 4 µM CBHA before Mtb infection decreased MMP-1 secretion by 77% (from 8.89 to 2.05 ng/ml) and MMP-3 secretion to undetectable levels (Figures 2E,F). CoMTb-driven MMP-1 secretion was reduced by CBHA treatment in a dosedependent manner. 5 µM CBHA reduced MMP-1 secretion by 41% (from 797 to 473 pg/ml; p < 0.05; Figure 2G), while 1 µM CBHA was sufficient to completely inhibit CoMTb-induced MMP-3 secretion (p < 0.0001; Figure 2H). These results indicate that HDACs may differ between cell types, each with specific MMP regulatory pathways. The HDAC class I-selective inhibitor MS-275 (1 µM) inhibited Mtb-driven macrophage MMP-1 secretion by 92% and reduced MMP-3 concentrations by 94.4% (Figures 2I,J). This supports the earlier finding and implies a key regulatory role for class I HDACs in macrophages MMP-1/-3 expression. In contrast, MS-275 enhanced MMP-1 secretion by both unstimulated (p < 0.05) and CoMTb-stimulated NHBEs (p < 0.0001; Figure 2K). MS-275 at a low concentration of 1 µM increased MMP-3 secretion 4.6-fold compared to CoMTb alone (p < 0.05), but this was not observed at the higher concentration of 10 µM MS-275 (Figure 2L), possibly due to sublethal cell toxicity, which could not be detected by cell viability experiments. The effects of these HDACi on MMP secretion were selective, since neither TSA nor 1 µM MS-275 significantly altered MMP-7 secretion by Mtb-infected macrophages (Figures 2M,N).

Macrophage-Derived MMP-1 and -3 gene expression and secretion during Mtb infection are Blocked by haT inhibition
Next, the role of HAT activity was investigated using the inhibitor HATi II. 10 µM HATi II significantly decreased MMP-1 secretion from Mtb-infected macrophages by 56% (from 5,029 to 2,187 pg/ml) and mRNA accumulation by 62% (Figures 4A,B). MMP-3 secretion was decreased from 1,653 to 190 pg/ml (p < 0.0001, Figure 4C). There was a non-significant trend to decreased MMP-3 mRNA accumulation with HATi II treatment ( Figure 4D). HATi II treatment did not affect MMP-7 secretion in response to Mtb infection ( Figure S3 in Supplementary Material).
A second HAT inhibitor, AA (10 µM), reduced Mtb-infected macrophage MMP-1 secretion by 74% (from 5,029 to 1,302 pg/ml; p < 0.0001) (Figure 4E). Similarly, a significant decrease in MMP-1 secretion was also detected in stimulated NHBE cells pretreated with AA ( Figure S4 in Supplementary Material). Secretion of MMP-3 in Mtb-infected macrophages was also inhibited by 10 µM AA (p < 0.01; Figure 4F). The AA compound is closely related to salicylic acid and has been reported to have some antimicrobial activity, including against Mtb (26,27). We therefore investigated whether these results might be secondary to an effect on Mtb growth, but this was not altered in broth cultures containing AA at concentrations between 2 and 25 µM ( Figure 4G). These experiments support the hypothesis that HAT activity is required for inducible expression of MMP-1 and MMP-3 in macrophages and NHBEs stimulated with Mtb.

MMP-1 and MMP-3 Promoter-reporter analysis in TB
Plasmid promoter-reporter constructs of the MMP-1 and MMP-3 promoter regions were transfected into A549 respiratory epithelial cells to investigate the effect of Mtb-stimulation on promoter activity. A schematic representation of the relevant region of the human MMP-1 promoter region is shown in Figure 5A. CoMTb treatment increased promoter activity of the WT construct by more than threefold compared to controls (p = 0.02; Figure 5B). CoMTb-mediated promoter activation was significantly enhanced by 98 and 71% in the 3,830 and 2,942 bp constructs, respectively, compared to CoMTb-stimulated WT (p < 0.01). Further truncation of the promoter resulted in loss of CoMTbdriven promoter activity. The MMP-3 promoter, examined using similar methodology, showed a progressive reduction in both basal and CoMTb-stimulated promoter activity with truncation (F) HDAC2 mRNA normalized to the reference gene ACTB and accumulation was silenced by HDAC2 siRNA. HDAC2 silencing increased MMP-1 mRNA accumulation (g) and did not affect MMP-3 mRNA accumulation (h) in CoMTb-stimulated NHBEs. mRNA of target genes was normalized to mRNA of the reference gene ACTB. Bars represent mean ± SD and analysis was performed using one-way ANOVA with Tukey's posttest. *p < 0.05; **p < 0.01; ****p < 0.0001; ns, non-significant; ACTB, beta-actin; HDAC, histone deacetylase siRNA; NT, non-target siRNA.  (g) H37Rv was cultured in the presence of AA 2-25 µM and bacterial growth was accessed by optical density (OD) measurements. Bars represent mean ± SD and analysis was performed using one-way ANOVA with Tukey's posttest. *p < 0.05; **p < 0.01; ****p < 0.0001; ns, non-significant; ACTB, beta-actin; AA, anacardic acid. (c) MMP-3 promoter activity is suppressed by truncations upstream −1,612 bp from the transcriptional start site. Bars represent mean ± SD and analysis was performed using one-way ANOVA with Tukey's posttest. *p < 0.05; **p < 0.01; ***p < 0.001. AU, arbitrary unit; WT, wild-type. To further investigate whether epigenetic modifications of the MMP-1 promoter controlled the response to CoMTb, we examined the histone acetylation status of the MMP-1 promoter region by chromatin immunoprecipitation. Preliminary experiments using the respiratory epithelial A549 cell line suggested marked increases in histone H4 acetylation with CoMTb treatment at 1 and 2 h poststimulus (data not shown). In primary NHBEs, RNA Polymerase II binding to the MMP-1 promoter was increased between 10-and 15-fold in cells after 2 h of CoMTb treatment (Figure 6A). Histone H3 acetylation was increased 2 h poststimulation and was approximately threefold greater than under control conditions in the proximal promoter, and fivefold higher than control when measured 2 kbp upstream of the MMP-1 transcriptional start site ( Figure 6B). Detection of acetylated histone H4 after 2 h of CoMTb stimulation was also increased across the MMP-1 promoter region. Histone H4 acetylation was 10-fold higher at −2 kbp and −500 bp, and 6-fold higher than unstimulated controls at the transcriptional start site ( Figure 6C).

DiscUssiOn
Upregulation of MMP-1 in TB is a critical event for the development of cavitation (9). Previous studies from our group and other have demonstrated that Mtb-stimulation upregulates MMP-1 expression, which is enzymatically active and able to degrade type I collagen, the main component of the lung's ECM [Al (10,28,29)]. Expression of other MMPs is also induced in pulmonary TB (28,30), and MMPs are implicated in central nervous system immunopathology (31,32). In this study, we demonstrated that expression of MMP-1 and -3 in response to Mtb is controlled by epigenetic changes in histone acetylation. HDACs are canonically considered to be negative regulators of gene expression, and we showed a change in the profile of HDAC expression following Mtb infection of macrophages. Downregulation of Class I HDAC gene expression was observed with a concomitant and selective upregulation of HDAC4 but not HDAC5. In contrast, in respiratory epithelial cells, Class I and Class II HDAC expression were unchanged after CoMTb stimulation, implying that this is a cell type-specific effect. TLR signaling is known to induce changes in HDAC expression and activity (33), and our findings are consistent with published evidence of selective changes in macrophage HDAC expression in response to specific inflammatory stimuli (34). Our data show that Mtb infection of macrophages induces a change in the cellular HDAC profile. We went on to investigate the likely effects of this change by performing experiments in which MMP responses were examined under conditions of both general and specific HDAC blockade. Non-selective HDAC inhibition using TSA and the bipolar hybrid CBHA suppressed Mtb-driven MMP-1 and -3 secretion and mRNA accumulation in primary epithelial cells and macrophages. Such a paradoxical effect on MMP expression has been previously reported in an in vitro model of arthritis, where chondrocyte collagenase activity was inhibited by TSA (35). In a mouse model of arthritis, TSA also inhibited MMP-1, -3, and -13 expression (36). Similarly, LPS-induced MMP expression by murine bone marrow-derived macrophages was inhibited by TSA (37). The Class I selective HDACi MS-275 had contrasting effects: increasing basal and CoMTb-stimulated MMP-1 and -3 secretion in epithelial cells, while in Mtb-infected macrophages, it decreased MMP-1/-3 secretion. Similar inhibition of cytokine-induced MMP-1 was seen in MS-275 treated human chondrocytes (38). The effects of MS-275 may be due to selective inhibition of HDAC1 at lower concentrations, and additional effects on the activity of HDAC2 and 3 at higher concentrations (39). Individual HDAC enzymes are likely to play different roles in regulation of MMP expression, and since these chemicals affect function of multiple HDACs, it is impossible to dissect the relative contribution of each HDAC by a chemical inhibition approach alone.
The increased secretion of MMP-1 and -3 from NHBEs observed with Class I HDAC inhibition, as compared to nonselective HDAC inhibition, is consistent with the premise that Class I HDACs are key negative regulators of MMP expression. HDAC1 was shown to be recruited to the MMP-9 promoter site, reducing histone H3 acetylation and NF-κB recruitment, leading to repression of MMP-9 expression in fibrosarcoma cells (40).
The enhancement of CoMTb-stimulated MMP expression observed with MS-275 led us to hypothesize that silencing the expression of HDAC1 might similarly increase inducible MMP expression. The catalytic activity of HDAC 1 and 2 is reliant on their incorporation as heterodimers into multiprotein assemblies (41) and therefore we expected that silencing either HDAC 1 or HDAC 2 would affect MMP expression. However, in spite of a high efficiency of HDAC1 silencing, MMP-1 expression was unaffected, and MMP-3 secretion was decreased in HDAC1-silenced cells. In contrast, upregulation of MMP-1 in CoMTb-stimulated conditions was further enhanced with HDAC2 inhibition, suggesting that MS-275 could be having its effect via inhibition of HDAC2 rather than HDAC1. The differing results for HDAC1 and HDAC2 silencing are unexpected given the close homology between these proteins, and consequently further dissection of their relative contributions to control of MMP expression is needed. In other experimental systems, different HDACs have been implicated in MMP regulation and there may be cell and stimulus specificity in host responses. In synovial fibroblasts from arthritis patients, HDAC1 siRNA enhanced TNFα-induced MMP-1 expression (42), whereas HDAC4 was identified as a negative regulator of MMP-1 expression (43).
Consistent with the hypothesis that epigenetic modifications regulate MMP secretion in TB, expression of MMP-1 and -3 were both suppressed by HAT inhibition with HATi II. Experiments using a structurally unrelated HAT inhibitor, AA, generated consistent data. Similarly, in human dermal fibroblasts exposed to UV light, AA inhibited MMP-1 expression, as did siRNA silencing of p300 expression. Increased HAT activity and histone H3 acetylation and decreased HDAC activity preceded changes in MMP-1 gene expression (44). The finding that both HAT and HDAC inhibition decreased MMP-1 and -3 expression, while apparently contradictory, may reflect the complex interdependence of these processes. It is well recognized that many non-histone substrates of these enzymes exist (45,46) and indeed phylogenetic studies have indicated that bacterial HDAC homologs pre-date the existence of histones (47). Lysine acetylation of many non-histone proteins has been shown to be enhanced by HDAC inhibition, for example, with MS-275 (46).
Many transcription factors contain such lysine acetylation sites, including cAMP response element-binding protein (CREB), whose activity is increased in the presence of TSA (48). CREB can be acetylated at three sites, enhancing its transcription factor activity, and HDAC8 is known to act on the CREB acetylation sites (49). The NF-κB family of transcription factors is also subject to posttranslational modification including acetylation as well as phosphorylation (50). Myocyte enhancer factor-2 is deacetylated by HDAC3, which also acts on the HATs PCAF and p300/CBP (51). In addition to transcription factors and HATs, the HDACs themselves contain lysine acetylation sites, as do a number of structural and regulatory proteins (52). Therefore, it is difficult to dissect out the relative contribution of inhibition of histone acetylation/deacetylation compared to effects on these other substrates when considering the effects of chemical inhibitors, and our findings indicate a complex interplay of signaling pathways occurs during infection.
The promoter-reporter analysis showed that inhibitory elem ents located 4,372-2,942 bp upstream from the MMP-1 transcriptional start site decrease promoter activation, since deletion of this region enhanced CoMTb-driven promoter activity. The area between −2,942 and −2,001 bp contains several critical elements required for the induction of gene expression, including a putative NF-κB binding site. There is also an AP-1 site at −1,950 bp just proximal to the −2,001 bp truncation that might be functionally disrupted by this truncation. There is substantial evidence that MMP-1 expression is regulated by both NF-κB and AP-1 family transcription factors (8,53,54), and our data support a central role. Similarly, multiple transcription binding sites may be important in MMP-3 promoter function. Our chromatin immunoprecipitation studies demonstrated that CoMTb stimulation leads to increased H3 and H4 acetylation at the MMP-1 promoter region. Histone acetylation was an early event after CoMTb stimulation occurring concurrently with binding of RNA Pol II to the MMP-1 promoter.
In summary, MMP-1 and 3 expression in TB is regulated by HDAC and HAT activity. MMP-1 upregulation, as a result of epigenetic control, has the potential to drive tissue damage in the lung, thereby facilitating spread of infection and development of pathology. Chemical inhibition suggests that HDAC and HAT activity is necessary for inducible expression of MMP-1 and -3 in Mtb-infected macrophages, but that different mechanisms operate in NHBEs, where class I HDACs appear to act as a brake on collagenase expression. This is a selective effect, as MMP-7, which is constitutively expressed in MDMs and upregulated by Mtb infection, was unaffected by HDAC and HAT inhibition.
The minimal inhibition of MMP responses seen with siRNA targeting individual class I HDACs implies that there may be some redundancy of function. Increased histone acetylation in the MMP-1 promoter region follows Mtb stimulation, favoring RNA Pol II binding, and results in upregulation of MMP gene transcription and enzyme secretion. Tissue breakdown mediated by MMP activity is a key event in TB immunopathology and manipulation of host epigenetic changes have potential applications as host-directed therapy in the era of rising drug resistance in TB.
aUThOr cOnTriBUTiOns JF conceived the project. RM, PE, and JF designed the experiments and analyzed the data. RM, SB, and FS performed the experiments and generated the data. SB, RM, PE, and JF wrote the manuscript, which was reviewed and final version approved by all authors.
FUnDing RM was a Wellcome Trust Clinical Research Fellow. SB was supported by the Portuguese Foundation for Science and Technology (FCT). SB and JF were supported by the Rosetrees Trust and Breathing Matters charities. JF acknowledges support of the Biomedical Research Centre at Imperial College.