An acidic microenvironment in Tuberculosis increases extracellular matrix degradation by regulating macrophage inflammatory responses

Mycobacterium tuberculosis (M.tb) infection causes marked tissue inflammation leading to lung destruction and morbidity. The inflammatory extracellular microenvironment is acidic, however the effect of this acidosis on the immune response to M.tb is unknown. Using RNA-seq we show that acidosis produces system level transcriptional change in M.tb infected human macrophages regulating almost 4000 genes. Acidosis specifically upregulated extracellular matrix (ECM) degradation pathways with increased expression of Matrix metalloproteinases (MMPs) which mediate lung destruction in Tuberculosis. Macrophage MMP-1 and -3 secretion was increased by acidosis in a cellular model. Acidosis markedly suppresses several cytokines central to control of M.tb infection including TNF-α and IFN-γ. Murine studies demonstrated expression of known acidosis signaling G-protein coupled receptors OGR-1 and TDAG-8 in Tuberculosis which are shown to mediate the immune effects of decreased pH. Receptors were then demonstrated to be expressed in patients with TB lymphadenitis. Collectively, our findings show that an acidic microenvironment modulates immune function to reduce protective inflammatory responses and increase extracellular matrix degradation in Tuberculosis. Acidosis receptors are therefore potential targets for host directed therapy in patients.


Introduction
Mycobacterium tuberculosis (M.tb) is an intracellular pathogen of macrophages. It infects a third of the world's population resulting in 1.6 million deaths annually [1]. Tuberculosis deaths are rising due to the COVID pandemic [2]. In the majority of infected individuals who remain well, disease is prevented by activation of macrophage intracellular defense mechanisms including phagolysosome maturation, generation of reactive oxygen species, apoptosis and autophagy which contains M.tb [3,4]. Failure to contain infection results in clinical Tuberculosis (TB), a disease characterized by high levels of proinflammatory cytokines and extensive tissue inflammation. Inflammation leads to tissue destruction causing morbidity and facilitates mycobacterial growth and transmission [5]. Lung destruction in TB is immune-mediated, resulting from the activity of enzymes particularly Matrix Metalloproteinases (MMPs). MMPs degrade pulmonary extracellular matrix (ECM) and are derived from both M.tb-infected macrophages, and lung parenchymal cells stimulated by proinflammatory cytokines. MMPs-1,-3,-8, -9 and -10 are all increased in TB patients [6][7][8] with collagenase MMP-1 the key driver of collagen breakdown and tissue necrosis [9,10]. The balance of MMP and cytokine activity is therefore critical to disease outcomes in TB.
Inflammation disrupts tissue homeostasis; generating a distinct inflammatory microenvironment characterized by hypoxia [11], tissue acidosis [12] with elevated extracellular lactate concentrations [13] and fever range temperature [14]. An acidic microenvironment arises during the host response to infection due to lactic acid secretion by inflammatory cells, and inflamed tissues typically have an extracellular pH around 7.0 compared to pH 7.4 in healthy tissue. In TB patients, CSF [15], pleural [16], peritoneal [17] and lung lining fluid [18] are more acidic than healthy tissue. Extracellular acidosis is signaled by the Ovarian Cancer G-Protein Coupled Receptor (OGR) family of cell surface receptors, including OGR-1, T-cell death associated gene-8 (TDAG-8) and G-protein coupled receptor-4 (GPR-4) [19]. These receptors utilize conserved histidine residues to detect extracellular protons and are silent at pH 7.8 and maximally active at pH 6.8 [20]. TDAG-8 is predominantly expressed in lymphoid tissue [21] and TDAG-8 knock out mice have enhanced delayed type hypersensitivity responses [22]. OGR-1 expression is widespread and OGR1 deficient mice have attenuated pulmonary [23] and intestinal [24] inflammatory responses indicating that acidosis receptors may have immune function.
We therefore investigated for the first time how extracellular acidosis and lactate in the inflammatory microenvironment regulates primary human macrophage responses to M.tb. RNA-seq of uninfected macrophages demonstrated limited transcriptional changes with acidosis with activation of cellular stress response gene sets. M.tb-infected MDMs show marked transcriptional change in acidosis which enhances expression of gene sets regulating ECM degradation and glucose metabolism. Further investigations in a cellular model confirmed acidosis increases M.tb driven MMP-1 secretion and suppresses TNF-α. TDAG-8 is highly expressed in granulomas and on multinucleate giant cells in a murine model of TB infection and regulates MMP-1 activity in a human cellular model.We further demonstrate that TDAG-8 is highly expressed in patients with active TB. Extracellular lactate enhances MMP-1 secretion independently of acidosis. The acidic inflammatory microenvironment in TB is therefore immunomodulatory driving matrix degradation and inhibiting activity of key protective cytokines.

Extracellular acidosis drives transcriptional change in human macrophages which is synergistically increased by M.tb stimulation
Firstly, uninfected human MDMs cultured at either neutral pH 7.4 or in acidosis at pH 7.0 were RNA-sequenced to investigate the effects of acidosis alone. RNA-seq was then performed on MDMs infected with live virulent M.tb at pH 7.4 to determine the effects of infection, and on MDMs infected and cultured at pH 7.0 to assess the combined effects of infection and acidosis. Human MDMs were generated by monocyte culture in M-CSF alone so MDMs were not polarised prior to stimulation [31]. General trends in the resulting data were investigated with Principal Component Analysis (PCA). Plotting of the first and second principal components shows samples cluster by both M.tb infection state and pH suggesting both stimuli trigger independent transcriptional responses ( Fig 1A). Three biological replicates cluster closely on analysis of the third principal component with one (Donor 2) having a somewhat different pattern of gene expression (S1A Fig). As experimental conditions, RNA integrity and sequencing quality control were similar across all samples, this donor was included in further analysis as this level of diversity likely reflects the heterogeneity of human responses to M.tb. Hierarchical clustering of the 50 most differentially expressed genes shows grouping of samples by M.tb infection and clustering of infected samples by pH, further demonstrating distinct transcription signatures for M.tb infection and acidosis (Fig 1B).
Differential gene expression analysis was performed between conditions as shown in Fig 1C  Full gene lists in S1 Dataset. For the first time, the effect of extracellular acidosis alone on human macrophages was dissected by comparing gene expression of unstimulated MDMs at pH 7.0 with unstimulated MDMs at pH 7.4 (analysis labelled Effect of Acidosis (uninfected)). Acidosis alone significantly upregulated expression of 178 genes and downregulated 37 genes. Comparison of Mtb infected with uninfected MDMs cultured at pH 7.4 (Analysis labelled Effect of Infection (pH7.4)) demonstrated infection drives greater transcriptional change than acidosis alone with upregulation of 1556 and downregulation of 1441 genes. Comparing M.tbinfected MDMs with control cells, both cultured at pH 7.0, shows the effect of M.tb infection at pH 7.0. Infection at pH 7.0 greatly increased transcriptional change with 2919 up-and 2616 down-regulated genes. Thus, 3102 genes (1611 upregulated, 1491 downregulated) are differentially expressed by Mtb infection at pH 7.0 but not by infection at pH 7.4 ( Fig 1D). Finally, the Total variance attributable to that PC indicated. (B) Clustered expression intensities standardised to z-score are shown for the 50 most differentially expressed genes (by p value) across all samples. Each sample is labelled with condition, pH and replicate number (C) Schematic of the conditions compared in differential expression overall effect of acidosis on infected MDMs was determined by directly comparing gene expression of infected MDMs at pH 7.0 with M.tb infected MDMs at pH 7.4 (Effect of Acidosis (Infected)). In this analysis 2019 and 1975 genes were up and downregulated respectively, representing DE of 30.7% of the genome expressed by control macrophages. The magnitude of fold change across these DE genes is significant with 437 genes differently expressed by >4 fold (Fig 1E). Smear plots showing the magnitude of differential expression of individual genes for each analysis are in S1C Fig. Therefore, although extracellular acidosis has some effect on MDM gene expression alone, acidosis has a marked synergistic interaction with M.tb to induce significant transcriptional change in macrophages.

Extracellular Acidosis downregulated inflammatory pathways in unstimulated macrophages
Acidosis alone in uninfected MDMs causes differential expression of 210 genes (178 up, 37 down) representing only 1.6% of total expressed genes. Genes for chemokines (CXCR4, CXCR7, CXCL5 AND CXCL15), signal transducer STAT4 and IL-1 receptor (IL1R1) were the only genes upregulated by acidosis with direct immune function. The biological effects of acidosis alone were determined by performing functional analysis using Gene Set Enrichment Analysis (GSEA) on the results of the Acidosis (uninfected) analysis. GSEA revealed no Reactome gene sets upregulated by acidosis alone. A total of 219 gene sets were downregulated, many highly significantly with the top 133 of 219 having FDR-q values <0.001. GSEA outputs using the Reactome Knowledgebase gene sets [32] were similar to Gene Ontology (GO) database results and complete lists of enriched gene sets in each analysis are shown in S2 Dataset. Classification by uppermost category of the Reactome event hierarchy revealed downregulated gene sets are involved in the cell cycle, gene expression, DNA repair and replication, specific disease states, signal transduction and metabolism (Fig 2A) reminiscent of the transcriptional change seen with other forms of cellular stress [33]. Twelve immune system gene sets were downregulated including NF-κB activation pathways, antigen processing and cross presentation, Endoplasmic reticulum-phagosome pathways and C-type lectin signalling pathways ( Fig 2B)  interactions were downregulated as were metabolic pathways including the Tricarboxylic Acid Cycle (TCA cycle) and the electron transport chain (Fig 2C).

M.tb infection upregulates multiple inflammatory pathways and MMP genes
We next assessed the effect of M.tb infection at pH 7.4. In the Infection (pH 7.4) analysis numerous cytokine and chemokine genes of known relevance in TB immunity were highly upregulated. The top 100 most upregulated genes were all upregulated >64 fold and included multiple genes for cytokines (IL-1β, IL-6, IL Whilst in controls cells extracellular acidosis downregulated immune pathways, acidosis increased expression of many pro-inflammatory genes in M.tb-infected MDMs. Direct comparison of M.tb infected MDMs at pH 7.0 with pH 7.4 (Acidosis (Infected)) demonstrates that acidosis enhances expression of MMPs key to TB immunopathology including MMP-1, MMP-3 and MMP-10 which were increased 5.6, 4.3 and 7 fold respectively (p<0.05). Genes for proinflammatory TNF-α, IL-1β and IFN-γ, critical activators of antimycobacterial processes in macrophages, and IL-12, a key inducer of protective Th-1 responses [4], were also further upregulated >1.5 fold (p<0.05) by acidosis as were multiple chemokine genes (Fig 3).
To assess the overall effect of acidosis on M.tb infected macrophages we performed functional analysis on the results of the Acidosis (Infected) analysis. This demonstrated 22 upregulated Reactome reactions and 9 upregulated GO terms (Fig 4A and 4B). Gene sets relating to ECM organisation were the most frequently and significantly upregulated. Reactome gene sets ECM proteoglycans, Collagen Degradation, Integrin cell surface interactions, Degradation of the extracellular matrix and Extracellular Matrix Organisation and were all highly significantly upregulated (p<0.001, FDR-q<0.05). Extracellular Matrix Organisation is one of only 4 GO biological processes enriched along with Regulation of glucose metabolism, adrenal gland development and cell communication (all p<0.01, FDR-q <0.05). Also enriched were GO cellular functions Proteinaceous Extracellular Matrix and GO molecular function term Extracellular Matrix Structural Constituent. A total of 168 Reactome gene sets were downregulated in the Acidosis (Infected) analysis with the majority involved in the cell cycle, DNA repair and replication, specific disease states and metabolism of proteins similar to that seen in MDMs

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis with acidosis alone. Only 3 of these downregulated gene sets were immune system sets; Cross presentation of soluble exogenous antigens, NF-κB inducing kinase non canonical NF-κB signalling and antiviral mechanism by IFN-stimulated genes ( Fig 4C).
Examining individual genes included in the enriched Extracellular Matrix Organisation gene set shows that genes for MMPs-1,-3 and -10 were the highest ranked genes in the set and the genes most upregulated by M.tb infection with acidosis ( Fig 4D). MMPs-1, -3 and 10 were similarly the most significantly acidosis upregulated genes amongst Collagen Degradation and ECM degradation genes. After MMP genes, the next most upregulated gene in these sets was protease ADAMTS4 which also degrades ECM components and is highly expressed at sites of TB infection [39]. Transcriptome profiling therefore demonstrates that acidosis enhances expression of ECM degradation pathways in M.tb infected MDMs by increasing expression of key MMPs of known importance to TB immunopathology.

Extracellular Acidosis increases secretion of MMP-1 and -3 and IL-1β but suppresses TNF-α
As the transcriptome data showed acidosis enhancement of ECM degradation pathways due to upregulation of MMP genes, we next investigated changes in protein secretion to confirm phenotypic change. Secretion of MMPs and TIMPs broadly followed the transcriptomic data. In M.tb infected MDMs, acidosis increased secretion of MMP-1, 2.3 fold (p<0.001), and its activator MMP-3 [40] 4.9 fold (p<0.0001) but there was no change in MMP-10 concentrations (Fig 5A-5C). Baseline secretion of MMPs-1, -3 and -10 was unchanged by acidosis. Increases in MMP-1 gene expression were confirmed by qPCR which showed 2.8 fold increase with acidosis (p<0.001) ( Fig 5D). No significant changes in secretion of TIMP-1 or -2 was detected, despite the observed modest increase in gene expression, confirming that acidosis worsens MMP/TIMP imbalance (Fig 5E and 5F). We next measured secretion of cytokines key to TB immunity. Acidosis in uninfected cells did not induce secretion of any measured cytokines (IL-1β, TNF-α, IFN-γ, IL-6, IL-8, IL-10). Secretion of IL-1β, an important macrophage activator in TB, was increased by acidosis ( Fig 5G). Contrasting the transcriptome data however, key regulators of antimycobacterial responses TNF-α, IFN-γ and IL-6 were markedly suppressed by acidosis (Fig 5H and 5I). Secretion of IL-8 and anti-inflammatory IL-10 was unchanged (Fig 5K and 5L). Acidosis effects on TNF-α and MMP-1 were unchanged if macrophages were cultured in AB human serum rather than FCS (S4).

Murine macrophages predominantly express TDAG-8 which regulates MMP and TNF-α secretion by M.tb infected human macrophages
To investigate mechanisms regulating enhancement of MMP secretion, we next determined if acidosis sensing GPCRs were expressed at sites of M.tb infection in a murine model. Immunohistochemistry was performed on lung sections of mice infected with live virulent M.tb ( Fig 6A). Areas of granulomatous inflammation stained brightly for TDAG-8 particularly the infiltrating inflammatory cells (Fig 6Aii). Staining for OGR-1in inflamed areas of tissue was less marked but still evident compared to isotype control (Fig 6Aiii). Similarly, in primary human MDMs, TDAG-8 was the most abundantly expressed acidosis sensing GPCR by qPCR with lower and variable levels of OGR-1 expression. GPR4 expression was not detected

TDAG-8 is highly expressed in the lymph nodes of TB patients
To investigate the role of TDAG-8 in human immune responses to tuberculosis, we examined TDAG-8 expression in granuloma macrophages in patients with TB. Immunohistochemistry was performed on lymph node biopsies from 5 patients with culture proven TB lymphadenitis. All biopsies demonstrated necrotising granulomatous inflammation with multinucleate giant cell (MGC) formation characteristic of TB (Fig 7A). Compared to isotype control (Fig 7B),

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis TDAG-8 stained throughout the lymph node but staining was more marked in the necrotic centre of granulomas where M.tb-infected macrophages reside (Fig 7C). Cell specific staining was seen in infiltrating monocytes which stained brightly for TDAG-8 ( Fig 7D) as well as Langhans type MGCs (Fig 7E). OGR-1 staining was present in the granuloma centre associated with necrosis (Arrowed in Fig 7F) but much less marked than for TDAG-8 and not apparently cell associated. Multinucleate Giant Cells did not demonstrate OGR-1 staining (Fig 7F).

Extracellular lactate increases MMP-1 and TNF-α secretion by macrophages dependent upon Monocarboxylate transporter (MCT)-1
The microenvironment in inflammation is acidic specifically due to the build up of lactic acid and consequently the microenvironment in tuberculosis has high lactate concentration in addition to acidosis [44][45][46]. We therefore sought to determine if lactate has independent effects to acidosis on the pro-inflammatory response. Sodium L-Lactate (NaL) was used to manipulate media lactate concentration without altering pH. Lactate alone did not cause TNFα or MMP-1 secretion by uninfected MDMs. Lactate increases TNF-α secretion from M.tbinfected MDMs 1.9 fold at 50mM (p<0.0001) but no effect was observed at lower concentrations (5-30mM NaL) (Fig 8A). MMP-1 secretion increased in a dose dependent fashion in response to NaL; 50mM NaL resulted in a 5.6 fold increase in MMP-1 concentrations (p<0.0001) ( Fig 8B) and 25mM lactate increased MMP-1 gene expression 2.3 fold (Fig 8C). Neither acidosis or lactate had any effect on intracellular M.tb survival (Fig 8D). Finally, we investigated whether the effect of lactate required intracellular transport via MCTs which are known to be expressed in mouse lungs following M.tb infection [47]. Human MDMs expressed both MCT-1, a pH dependent importer of lactate [48] and MCT-4 a lactate exporter [48] (Fig 8E-8G). RNA-seq showed infection at pH 7.4 decreased expression of SLC16A1 (MCT-1) 1.5 fold (p = 0.02) and increased SLC16A3 (MCT-4) 1.6 fold (p = 0.004). We confirmed this with qPCR and further showed no expression change with lactate alone. Combined infection with lactate increased SLC16A3 (MCT-4) expression 2.5 fold (p<0.01) (Fig 8E and  8F) however this change was not reflected in western blots (Fig 8G). MCT-1 activity is critical to lactate regulation of MMP-1 secretion as blockade of MCT-1 with specific inhibitor AR-C122982 abrogated the effect of lactate showing that lactate acts intracellularly to increase MMP-1 secretion (Fig 8H).

Discussion
The extracellular inflammatory microenvironment in TB is acidic due to lactic acid accumulation as a consequence of cellular metabolism. Using an in vitro model, we found that this extracellular acidosis generates a unique transcriptional signature in both M.tb infected and uninfected macrophages. In uninfected macrophages, modest transcriptional change was induced by acidosis, increasing expression of 117 genes and decreasing 23 genes. This change is broadly inhibitory with downregulation of genes involved in the cell cycle, gene expression, DNA repair and replication, signal transduction and metabolism. Similar transcriptional responses are seen in model organisms such as yeast, HeLa cells and human lung fibroblasts in response to numerous cellular stressors suggesting that acidosis activates the conserved "Environmental Stress Response" [33]. Acidosis stimulates IL-1β secretion in unstimulated human monocytes [49] however we found no macrophage cytokine gene expression or secretion was induced by acidosis alone. Acidosis has been proposed as a "novel danger signal" [26] but our data does not support this. No immune system gene sets were upregulated by acidosis alone, whilst antigen processing, Dectin-1 and NF-κβ signalling gene sets were downregulated. This

Extracellular acidosis affects macrophage responses in Tuberculosis
contrasts with other well characterised endogenous "danger" signals such as hypoxia [50] and ROS [51] which increase cytokine and NF-κβ expression even in unstimulated cells.
M.tb infection, irrespective of pH, upregulated multiple cytokine and chemokine genes, cell surface activation markers, Interferon stimulated genes and glucose metabolism genes supporting the findings of others that these are core macrophage transcriptional responses to M. tb infection [52,53]. Activated macrophages undergo "glycolytic reprogramming"; rapidly metabolising glucose via aerobic glycolysis yielding excess pyruvate which is converted to lactic acid by lactate dehydrogenase (LDH). Simultaneously, the TCA cycle and oxidative

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis phosphorylation is suppressed [54]. This increase in aerobic glycolysis produces the lactic acidosis which acidifies the inflammatory microenvironment. Supporting a metabolic shift in M. tb infected human MDMs we found increased expression of multiple glucose uptake (SLC2A1, SLC2A3, SLC2A6), glycolysis (HK2, PFKFB3, ALDOC) and Lactate Dehydrogenase (LDHA, LDHB) genes whilst the "TCA cycle and respiratory electron transport" was the 5th most significantly downregulated gene set by infection. Amongst genes regulated by infection with acidosis but not at pH 7.4 the Reactome gene set "glycolysis" was enriched and expression of glycolysis genes (HK2, ALDOC, LDHB) was increased by acidosis. GO biological process term "regulation of glucose metabolism" was one of only 4 GO processes upregulated in the Acidosis (Infected) analysis. This suggests acidosis increases glycolysis potentially favouring further acidification of the microenvironment. Metabolic reprogramming requires MCT-4 mediated lactate export [55], and MCT-4 expression is increased in granulomas of TB patients [56]. Expression of MCT-4 in MDMs was increased by M.tb however significant changes in protein expression were not observed, possibly as baseline MCT expression in MDMs was already high.
In contrast to uninfected macrophages, extracellular acidosis acts in marked synergy with M.tb to cause profound proinflammatory transcriptional change in infected MDMs. Comparing macrophages infected at pH 7.0, as encountered in the inflammatory microenvironment, with infection at pH 7.4 revealed specific upregulation of gene sets controlling ECM Organisation, Collagen Degradation and Degradation of the Extracellular Matrix. Within these gene sets, genes for MMPs-1, -3 and -10 were the most upregulated by acidosis, and we confirmed secretion of MMP-1 and its' activator MMP-3 was increased with acidosis. An acidic microenvironment activates MMP-3 [57] and MMP-3 catalytic activity is greater at low pH [58]. Extracellular lactate above 10mM; which occurs in meningeal [45], pleural [46] and peritoneal TB [17], also increased MMP-1 secretion in the presence of infection. Lactate effects on MMP-1 secretion was abrogated by inhibition of lactate transporter MCT-1 suggesting that lactate acts intracellularly to modulate MMP-1 secretion. Additionally acidosis increased IL-1β gene expression and secretion by M.tb infected macrophages and IL-1β drives lung parenchyma MMP production in TB [59]. Extracellular acidosis and lactate thus independently enhance secretion of MMP-1 and its activators from M.tb infected macrophages without significant change in inhibiting TIMPs. MMP-1 is increasingly recognised as the key driver of tissue destruction in TB. MMP-1 is one of few enzymes capable of degrading fibrillary collagen [60], the principal structural component of the lung. Macrophage MMP-1 secretion is required for collagen degradation [9] and caseating necrosis [10] in TB and MMP-1 is highly expressed in TB cavities [61]. In patients, sputum MMP-1 and -3 levels correlate with disease severity [9]. Microenvironment pH and lactate enhancement of macrophage MMP-1 activity may therefore worsen morbidity in TB patients however this requires confirmation in in vivo models.
TNF-α is critical for macrophage M.tb killing and granuloma formation [62], and patients receiving anti-TNF-α therapy are at high risk of TB reactivation [63]. Acidosis markedly suppressed TNF-α secretion by infected MDMs despite increases in TNF expression. This is reported in LPS stimulated macrophages due to both cytosolic retention [64] and post transcriptional regulation [65] of TNF-α. Acidosis reduces protein synthesis [66], and we found translation and protein synthesis gene sets were significantly reduced by acidosis in MDMs. Increases in MMP and IL-1β secretion with acidosis however demonstrates that TNF-α suppression is a specific effect rather than a consequence of globally reduced protein synthesis. Microenvironment lactate is unlikely to balance acidosis effects on TNF-α as lactate only increased TNF-α secretion at concentrations greater than 30mM, above that described in TB [44][45][46]. Acidosis additionally suppressed M.tb infected MDM secretion of IFN-γ whilst secretion of IL-10 was unchanged. Low TNF-α, INF-γ levels favours M.tb survival [3]. Despite this, early intracellular M.tb infection burden, was unaffected by acidosis possibly as increased glycolytic rate and IL-1β stimulation increase macrophage intracellular killing [67]. This observation also suggests uptake of M.tb by MDMs was not significantly changed by acidosis.
The mechanism by which acidosis might affect macrophage function and gene expression in acidosis was unknown so we investigated the OGR-1 family of GPCRs comprising OGR-1, TDAG-8, GPR4 which signal extracellular acidosis within the extracellular pH range in TB [68]. In a murine model of TB, we found inflammatory lesions stained clearly for TDAG-8 and OGR-1 with numerous inflammatory cells staining brightly for TDAG-8. We confirmed human MDMs express both TDAG-8 and OGR-1, consistent with previous data [24], but TDAG-8 expression was significantly higher. In human monocytic cell lines acidosis, hypoxia and LPS all increase acidosis receptor expression [69] but expression in primary human MDMs was unchanged by acidosis or M.tb infection. Granulomas from TB patients stained strongly for TDAG-8, particularly MNGs which are of the macrophage lineage and a histological hallmark of TB disease. TDAG-8 agonists had similar effects on MMP-1 and TNF-α concentrations to those observed in acidosis whilst receptor knockdown abrogates acidosis TNF-α suppression. OGR-1 knock down also reduced acidosis TNF-α suppression. OGR-1 expression is lower than TDAG-8 expression in MDMs, and expression at sites of infection in patients was less clear. Lung OGR-1 expression is predominantly by epithelial and smooth muscle cells and whole lung expression is lower than for TDAG-8 [23]. Together these observations show that microenvironment acidosis, signalled via TDAG-8 expression in M.tb infected macrophages suppresses inflammatory, and enhances ECM turnover pathways. This effect on macrophages may explain the observations that TDAG-8 knock out mice have exaggerated delayed type hypersensitivity and reduced pathology in a collagen induced arthritis model [22]. These central findings are illustrated in Fig 9. In summary, we show that extracellular acidosis and lactate in the inflammatory microenvironment modulates M.tb infected macrophage responses favouring MMP-1 activity and tissue degradation pathways which can be predicted to worsen disease pathology. Acidosis, but not lactate, suppresses TNF-α secretion from infected MDMs. Acidosis effects on macrophage MMP and cytokine secretion are dependent upon acidosis receptor signalling of which TDAG-8 is the most abundantly expressed. Fully elucidating how acidosis and TDAG-8 regulates inflammatory responses in the complex environment of TB lesions in patients requires further study and may identify a new target for host directed therapy.

Ethics statement
Patients were recruited from several hospital sites in Lima, Peru. Written informed consent was obtained from each patient and study protocol was approved by Local and National Research Ethics Committee in Peru. All animals were handled in accordance with UK national Home Office regulations and all procedures were approved locally by Imperial College, London.

M. tuberculosis culture
M. tuberculosis H37Rv, obtained from The National Collection of Type Cultures (UKHSA, UK), was cultured in Middlebrook 7H9 media (BD Diagnostics) supplemented with 10% ADC enrichment medium (BD Diagnostics), 0.2% glycerol and 0.02% Tween 80. Clumping was prevented by circular agitation (100rpm) and brief sonication prior to use. M.tb was grown to mid log phase and used at a multiplicity of infection (MOI) of 1 as described previously [70].

Macrophage and monocyte culture
Peripheral Blood Mononuclear Cells (PBMC) were isolated from single donor leukocyte cones purchased from the National Blood Transfusion Service (Colindale, UK) by density gradient centrifugation and adhesion purification. Briefly, leukocytes were diluted 1:2 with Hanks-Balanced Salt Solution (HBSS) (Gibco), layered on Ficoll-Paque (GE Healthcare) and centrifuged at 1500 rpm for 30 mins. The PBMC layer was aspirated, washed and monocytes were plated at 250,000 monocytes/cm 2 in RPMI 1640 supplemented with 2mM glutamine and 10μg/ml ampicillin (Gibco). After 1 hour, non-adherent cells were removed by washing with HBSS. Monocyte Derived Macrophages (MDMs) were generated by culturing monocytes for 4 days in RPMI 1640/2mM glutamine/10μg/ml ampicillin/10% heat inactivated Fetal Calf Serum (FCS) (Biowest) and 100ng/ml of Macrophage Colony Stimulating Factor (M-CSF) followed by 24 hour culture in RPMI/10%FCS without growth factors. Experiments were performed using Macrophage Serum Free Media (M-SFM) (Gibco). Purity of macrophages was >94%.

Adjusting pH and Lactate concentration of cell culture media
The pH of cell culture media was adjusted with predetermined volumes of 1M HCL or 1M NaOH. Media was incubated at 37˚C and 5% CO 2 for 24 hours to reach equilibrium. pH was

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis measured with a 3 point calibrated pH meter accuracy +/-0.01. 1M Sodium L-Lactate (Sigma-Aldrich) was used in lactate experiments and did not affect pH. Cell viability at 24 hrs was unaffected by the pH range used (pH6. 8-7.4) and >95% in all experiments. Viability was assessed by measuring supernatant LDH activity using LDH Cytotoxicity Detection Kit (Roche) according to manufacturers instructions and exclusion of fluorescent DNA binding dye Fixable Viability Dye eFlour450 (eBiosciences) by flow cytometry using a BD FACSCalibur (BD Biosciences).

RNA-seq
At 24 hours for each condition 2.5x10 6 macrophages were lysed in Tri-Reagent RNA isolation buffer (Sigma-Aldrich) and purified by spin column purification with in column DNAase step using DirectZol Miniprep kit as per manufactures instructions (Zymo Research). Only samples yielding >2.2μg of RNA with RNA Integrity Number > 7.0 were used to generate a stranded, polyadenylated mRNA focused library using a Tru-seq Stranded mRNA Library Prep Kit (Illumina). Sequencing was performed on the Illumina HiSeq 4000 platform (Edinburgh Genomics) using a single flow cell lane to generate >30 million 2x75bp paired end reads per sample. Reads were aligned to a reference Homo Sapiens genome (Ensembl v 37.71) using STAR version 2.5 and the resulting BAM files sorted with Picard-tools (version 1.115) using default parameters. Read counts were generated using htseq-count version 0.6.0 and python version 2.7.3 with parameters "-m union-s reverse-I gene id-t exon" using ensemble gene annotations. Genes with less than 20 counts mapping across all samples were excluded from analysis. Differential expression analysis was performed using Bioconductor package EdgeR (version 3.10.5) with R version 3.2.1. The Quasi likelihood F-test approach was used due to the low replicate number. Genes with an individual False Discovery Rate (FDR) of <0.01 were considered differentially expressed. A fold change cut off was not used. Functional analysis was performed using Gene Subset Enrichment Analysis (GSEA) from the Broad Institute (version 2.2.1). GSEA was performed on a pre ranked list of all genes in each condition ranked either by fold change, or p value (p) using metric -1*(log(1/p)). Displayed results show gene sets enriched in both fold change and significance analyses. Analysis parameters were: permutations 1000, Maximum gene set size 500, Minimum gene set size 15, weighted analysis. Gene sets from both the Gene Ontology and Reactome databases were used with similar results. GSEA results displayed in S2 Dataset shows Size (number of genes in the gene set after excluding genes not expressed in the dataset), Nominal p value (Statistical significance of the enrichment score), False Positive Rate q value (estimated probability that the normalized enrichment score represents a false positive finding), Rank at Max (position in the ranked list at which the maximum enrichment score occurs) and Leading Edge data. GO terms enriched in the analysis are displayed using REVIGO [71] where bubble shading indicates significance (darker more significant) and edges link related GO terms. The raw sequencing data has been deposited in the European Nucleotide Archive (ENA) under accession number PRJEB20229 and is publicly available.

Intracellular M. tuberculosis survival
Infected MDMs were lysed in sterile water after 24 hours, serially diluted and plated onto 7H11 agar (Middlebrook). Individual colony forming units were counted after incubation at 37˚C for 2-3 weeks.

Enzyme Linked Immunosorbent Assay (ELISA) and Luminex Muliplex Assay
Levels of MMPs, TIMPs and cytokines were measured by ELISA assay (R&D duoset, R&D systems) according to the manufacturer's instructions. Where stated in results, levels of MMPs

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis and cytokines in supernatants were also measured by multiplex microsphere based immunoassay on the Bio-Rad Bio-Plex 200 system (Bio-Rad Laboratories). For MMPs the Human Luminex Performance Assay base Kit (R&D systems) was used in conjunction with selected human MMP primary and secondary magnetic bead sets (R&D systems) according to manufacturer's instructions. Cytokine levels were measured using Human Cytokine Multiplex Assay (Millipore).

Western blot
Samples were washed in phosphate buffered saline and lysed in SDS lysis buffer (2%SDS, Tris 62.5mM, glycerol 10%, DDT 50mM) heat denatured at 70˚C for 10mins and centrifuged at 10000rpm for 1min before loading onto a NuPAGE 4-12% Bis Tris Gel (Life Technologies). Proteins were separated by electrophoresis using the SureLock MiniCell system (Life Technologies) and MES running buffer (Life technologies) then transferred to nitrocellulose membrane (GE Healthcare) using an X-Cell Blot Module and Transfer buffer (Life Technologies). Membranes were blocked for 1 hour (5% non-fat dry milk, 0.1% Tween-20 in Tris buffered saline) and incubated overnight with primary antibody diluted in blocking buffer at 4˚C. Membranes were then incubated with secondary HRP-conjugated antibody, diluted in blocking buffer for 1 hour at room temperature. Blots were developed with Bio-Rad Clarity (Bio-Rad) for 1-2 mins and chemiluminescence imaged using the Chemidoc MP Imaging system (Bio-Rad, UK). Image Lab v5.2.1 (Bio-Rad, UK) was used for post acquisition image processing. Antibodies used MCT-1 (Santa Cruz, sc-365501), MCT-4 (Santa Cruz, sc-50329), β-Actin (Cell Signalling, 8H10D10).

RNA Extraction, cDNA Synthesis and Real Time PCR
RNA was extracted as previously described. One microgram of RNA was then reverse transcribed to cDNA using the QuantiTect Reverse Transcription Kit (Qiagen). Resulting cDNA was then used in an RT-PCR reaction using specific primer and probe mixes with Brilliant II mastermix (Agilent) on the Stratagene Mx3000p platform (Agilent). The following thermal profile was used: 10min at 95˚C followed by 40 cycles at 95˚C for 30 sec then 60˚C for 1 min. For MMPs genes cycle thresholds were compared to a plasmid standard curve of known MMP and housekeeping gene concentrations. For other genes relative gene expression was calculated using the Pfaffl comparative method. The following custom oligonucleotides (MMP-1 forward primer 5'-AAGATGAAAGGTGGACCAACAATT -3'; Reverse primer 5' -CCAAGAGAAT GGCCGAGTTC -3'; Probe 5'-FAM CAGAGAGTACAACTTACATCGTGTTGCGGCTC-T AMRA -3', Sigma) and proprietary mixes were used (Thermofisher TaqMan reagents; OGR-1 Hs00268858_s1, TDAG-8 Hs00269247_s1, SCL16A1 Hs01560299_m1, SLC16 A3 Hs00358829_m1, 18S Hs03003631_g1, ACTB Hs01060665_g1)

Macrophage Transfection with siRNA
MDMs were generated as previously described and seeded at 105,000 cells/cm 2 . Transfection with DharmaFect 3 transfection reagent (Dharmacon) was performed as per manufactures instructions. For each gene a mix of 4 specific "On-target plus Smart-pool" siRNAs was compared with a similar pool of 4 non-targeting control siRNA (Dharmacon). Transfection was performed in antibiotic/serum free RPMI 1640 supplemented with 2mM glutamine in 12 well plates. 5μl/well of DharmaFect 3 was suspended in the appropriate concentration of serum free media. The siRNA was suspended in siRNA buffer (Dharmacon) and diluted in media to a final concentration of 25nM. The transfection reagent and siRNA were combined by gentle agitation at 20˚C for 20mins, cells transfected for 8 hours then rested for 12hrs in RPMI 1640

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis (2mM glutamine, 10μg/ml ampicillin, 10% heat FCS). Transfection efficiency as measured by uptake of florescence (FAM) labelled siRNA was > 95%. The described protocol was optimised for cell seeding density, transfection reagent concentration and transfection duration and yielded target gene expression compared to control of 30% for TDAG-8 and 49% for OGR-1.

Immunohistochemistry of TB Tissue
Lymph node biopsies were taken from HIV negative, adult patients with culture proven TB lymphadenitis. All biopsies had the typical histological features of TB lymphadenitis as confirmed by 3 independent, blinded, histopathologists. Additionally, paraffin fixed lung tissue blocks were obtained from Balb/C mice infected with 10 4 live virulent M. tuberculosis H37Rv by intranasal installation. Paraffin fixed patient biopsies were deparaffinised by heating (60˚C for 8mins) and staining performed on the Ventana Benchmark Ultra platform (Ventana Inc). Antigen retrieval was performed by incubation with Ventana Protease 3 cocktail (Ventana Inc). Primary antibody (1:300) was incubated for 32 mins at 32˚C. Endogenous peroxidase activity was blocked with Ventana Optiview Peroxidase Inhibitor (3%) and visualization performed with Ventana Optiview DAB Detection Kit and counterstained with Haematoxylin II/ Ventana Bluing Reagent (Ventana Inc). For animal studies cut sections of paraffin embedded mouse lung were deparaffinised and heat induced antigen retrieval performed. Primary antibody (1:50) was incubated at 4˚C overnight in blocking buffer (4% goat serum/1%BSA/PBS). Slides were washed with PBS and incubated for 45mins at room temperature with secondary antibody (1:200) in blocking buffer followed by detection using NovaRed HRP Detection Kit (Vector Labs). Antibodies used; Rabbit anti-human TDAG-8 IgG (ab188907, Abcam), Rabbit anti-human OGR-1 IgG (ab72500, Abcam), biotinylated anti-rabbit IgG1 (BA-1000, Vector labs)

Statistical analysis
Statistical analysis of sequencing data was performed as described in detail in the RNA-seq methodology section. Statistics on cell culture experiments (Figs 4-8) was performed using Prism Graphpad Software v7. Unless otherwise stated in the text all experiments were performed in triplicate and displayed results are representative of a minimum of 3 independent experiments with separate donors. Significance was assessed by 1-way ANOVA with Tukey's test except for experiments with repeated conditions when a 2 way ANOVA with Bonferoni correction was used. Results were considered statistically significant if p<0.05. Graphs display the mean +/-SD unless otherwise stated. On graphs statistical significance between conditions is marked with *. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

PLOS PATHOGENS
Extracellular acidosis affects macrophage responses in Tuberculosis