Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Tissue Metabolic Changes Drive Cytokine Responses to Mycobacterium tuberculosis

Tissue Metabolic Changes Drive Cytokine Responses to Mycobacterium tuberculosis Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Tissue Metabolic Changes Drive Cytokine Responses to Mycobacterium tuberculosis Ekta Lachmandas, Ana B Rios-Miguel, Valerie A C M Koeken, Eva van der Pasch, Vinod Kumar, Vasiliki Matzaraki, Yang Li, Marije Oosting, Leo A B Joosten, Richard A Notebaart, Mahdad Noursadeghi, Mihai G Netea, Reinout van Crevel, Gabriele Pollara Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 The Journal of Infectious Diseases BRIEF REPORT Our group has made use of the human tuberculin skin test Tissue Metabolic Changes Drive (TST) challenge model to faithfully reflect the inflammatory Cytokine Responses to Mycobacterium changes that occur at the site of tuberculosis disease, char - tuberculosis acterizing the tissue immunological pathways induced early after mycobacterial antigen exposure [ 7]. However, to date, no 1 1 1 Ekta Lachmandas, Ana B. Rios-Miguel, Valerie A. C. M. Koeken, Eva van der 1 1,2 2 2 1 Pasch, Vinod Kumar, Vasiliki Matzaraki, Yang Li, Marije Oosting, Leo A. studies have explored the metabolic changes and their fun-c 1 1,3 4 1 B. Joosten, Richard A. Notebaart, Mahdad Noursadeghi, Mihai G. Netea, tional consequences on downstream cytokine responses in 1 4 Reinout van Crevel, and Gabriele Pollara such a model. Quantitative cytokine production in response Department of Internal Medicine and Radboud Center for Infectious Diseases, to mycobacterial stimulation has been associated with genetic Radboud University Medical Center, Nijmegen, and University of Groningen, University Medical Center Groningen, The Netherlands; Laboratory of Food polymorphisms [8]. These cytokine quantitative trait loci Microbiology, Wageningen University and Research, The Netherlands; and Division (cQTLs) provide a functional insight into how genetics influ - July of Infection and Immunity, University College London, United Kingdom ences an inflammatory response, and in turn identify criti - Cellula m r etabolism can influence host immune responses to cal pathways that may be amenable to host-directed therapy. Mycobacterium tuberculosis. Using a systems biology approach, In this study, we use the TST model to test the hypothesis differential expression of 292 metabolic genes involved in glyco- l that differential tissue expression of genes involved in reg- ysis, glutathione, pyrimidine, and inositol phosphate pathways ulating metabolic pathways can directly influence cytokine was evident at the site of a human tuberculin skin test challenge production following M.  tuberculosis stimulation. In turn, in patients with active tuberculosis infection. For 28 metabolic our findings provide putative mechanistic links between the genes, we identified single nucleotide polymorphisms that were trans-acting for in vitro cytokine responses to M.  tuberculosis activity of cellular metabolic pathways and immune effector stimulation, including glutathione and pyrimidine metabolism functions. genes that alter production of Th1 and Th17 cytokines. Our METHODS findings identify novel therapeutic targets in host metabolism that may shape protective immunity to tuberculosis. Transcriptomic Data Analysis Keywords. tuberculosis; immune response; immunome- Transcriptomes were derived from TST and blood of patients tabolism; metabolism; cytokines; transcriptomics; microarrays; with active tuberculosis, and human M.  tuberculosis-in- functional genomics; human challenge model. fected and healthy lymph nodes from a separate cohort (Supplementary Table  S1 ). All patients were human immu - e ac Th tivity of many cellular metabolic pathways can impact nodeficiency virus (HIV) seronegative and none had diabetes the host immune response to infections [1]. Individual met - mellitus. abolic pathways have been implicated in antimycobacterial OA-CC-BY We used the KEGG pathway database (http://www.genome. responses: glutathione enhances interleukin-12 (IL-12) and jp/kegg/pathway.html) to derive 33 pathways associated with interferon-gamma (IFN-γ) secretion following Mycobacterium human metabolism, yielding a list of 1422 metabolic genes co -n tuberculosis stimulation [2]; tryptophan catabolism is involved taining no duplicate genes and no annotation to the original in M.  tuberculosis-induced production of IL-1β and IL-23 [3] pathways (Supplementary Table S2 and Table S3 ). Bioinformatic and control of M.  tuberculosis growth [4]; and a shift towards analyses were performed as previously described [ 7] (see aerobic glycolysis in M. tuberculosis-infected macrophages reg- Supplementary Methods ). ulates IL-1β production [ 5, 6]. However, as metabolic reactions are intrinsically interdependent, the challenge lies in determin - SNP Extraction and cQTL Mapping ing the relative roles of these and other pathways during in vivo Single nucleotide polymorphisms (SNPs) within metabolic M. tuberculosis infection. genes and cQTLs were identified as previously described [8] (see Supplementary Methods ). Received 10 July 2017; editorial decision 20 March 2018; accepted 2 April 2018; published online April 3, 2018. Metabolite Reporter Analysis Correspondence: G. Pollara, MBBS, MRCP, FRCPath, PhD, Division of Infection and Immunity, Reporter metabolite analysis [ 9] was performed in Matlab University College London, London WC1E 6BT, UK ([email protected]). using the RAVEN Toolbox (http://biomet-toolbox.chalmers. The Journal of Infectious Diseases 2018;21800:165–70 se/index.php?page=downtools-raven ) and the human genome- © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons scale metabolic reconstruction network HMR 2.00 provided in Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted Human Protein Atlas (http://www.metabolicatlas.org/down- reuse, distribution, and reproduction in any medium, provided the original work is properly cited. DOI: 10.1093/infdis/jiy173 loads/hmr ) (see Supplementary Methods ). BRIEF REPORT • JID 2018:218 (1 July) • 165 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Metabolite Depletion Experiments Next, we sought to test the hypothesis that metabolic changes Peripheral blood mononuclear cells (PBMC) were isolated from in the tissue environment influence M.  tuberculosis-induced 9 healthy Dutch adult volunteers and stimulated witM.  h tuber- cytokine production. We used natural genetic variation to culosis lysate in the presence or absence of pharmacological identify putative cis-acting SNPs for metabolic gene expres- manipulators of the glutathione pathway (see Supplementary sion and tested whether they might be trans-acting SNPs for Methods). cytokine responses. We identified metabolic genes differen - tially expressed in the TST and assessed their impact on cyt-o RESULTS kine secretion following M. tuberculosis lysate stimulation in a We have previously demonstrated that the transcriptional cohort of 500 healthy individuals (500FG cohort) 8 []. First, we response at the site of TST is characterized by upregulation used genotypes extracted from the 1000 Genomes Project to of 1725 genes that closely reflect changes seen in dissected identify 16 061 SNPs from the 109 metabolic genes that com- human tuberculosis granuloma relative to healthy lung t -is prised the 10 most enriched metabolic pathways in the TST sue [7]. We now show that the TST signature is also enriched (Figure 1C). We then assessed which of these SNPs were associ- within M. tuberculosis-infected relative to healthy lymph nodes ated with variable cytokine secretion, generating 2376 putative (Figure 1A), confirming that the TST transcriptional responses cQTLs. To reduce multiple testing false positives, we focused mirror pathology seen at established sites of human tuberculo - on SNPs in the mRNA coding region of the gene of interest that sis disease. influence the same gene’s transcription (ie, metabolic gene SNPs Out of 1422 metabolic genes, 292 were differentially that were cis-expression QTLs [eQTLs]). expressed at the site of TST versus saline injection in patients This analysis generated 47 cQTL SNPs from 28 metabolic with active tuberculosis disease (Supplementary Figure  S1 and genes (Figure  2A, 2B, Supplementary Table  S6 , and Table  S7). Supplementary Table S4 ). Expression of these genes at tissue level Based on these genes, the most over-represented metabolic showed a remarkable lack of concordance with the blood tra -n pathways included the glutathione, glycolysis, inositol ph -os scriptome of the same patients, as only 9 metabolic genes were phate metabolism, and pyrimidine pathways ( Supplementary differentially expressed in the blood ( Figure 1B, Supplementary Figure  S4). Many amino acid metabolism pathways were Figure S2, and Supplementary Table S5 ). This demonstrates that also observed, and ALDH3A2, ALDH3A1, LDHA, and IL4I1 the TST challenge model provides greater molecular resolution were the most frequent constituent genes from these path- to identify host differential metabolic gene expression to myco- ways (Supplementary Figure  S4). Glutathione and pyrimidine bacterial infection than the blood compartment [ 7]. metabolism predominantly influenced the secretion of IFN-γ Given the interconnected relationship between multiple and IL-17. Seven of 15 (47%) cQTLs that regulated IFN-γ secre - metabolic pathways, we generated a network depicting the 10 tion and 4 of 7 (57%) cQTLs that regulated IL-17 secretion most enriched KEGG annotated metabolic pathways based on were derived from genes assigned to glutathione or pyrimidine the 292 differentially expressed metabolic genes (Figure 1C). As metabolic pathways, whereas no cQTLs from these pathways expected, the TST induced changes in multiple interlinked m- et influenced cytokine secretion by macrophages ( Figure  2B). In abolic pathways. These included not only previously described contrast, genes involved in glycolysis, amino acid, and inositol pathways such as glycolysis and glutathione metabolism, but phosphate metabolism acted as cQTLs more ubiquitously, influ - also others such as inositol phosphate metabolism and specific encing the secretion of both T-cell and myeloid cell-derived amino acid metabolic pathways not previously associated with cytokines (IL-1β, IL-6, IL-22, and tumor necrosis factor- alpha tuberculosis [1, 2]. Because differentially expressed genes within [TNF-α]) (Figure 2B). the same metabolic pathway were both up- or downregulated We validated some of the relationships between metabolic (Supplementary Figure S3), we evaluated which metabolites in changes and cytokine secretion in vitro using buthionine sulf - these pathways were most ae ff cted using reporter metabolite oximine (BSO), an inhibitor of glutathione synthesis, and dieth - analysis. This identified metabolites in the human metabolic ylmaleate (DEM), which depletes cells of glutathion 11 e [ ]. Both network around which the most transcriptional changes occur resulted in increased IL-6 and decreased IL-17 secretion by [10]. e Th model predicted changes in several key metabolites M.  tuberculosis-stimulated PBMC ( Figure  2C), consistent with between TST and saline, including 1,3-bisphospho- d-glycer - the cQTL data demonstrating that variable expression of glu - ate (glycolysis), l-formylkynurenine (tryptophan metabolism), tathione genes impacted the secretion of both these cytokines 1-phosphatidyl-1d-myo-inositol-3,4-bisphosphate, and glu - (Figure 2B). tathione (Figure  1D). Therefore, the model predicts that gene Finally, to explore the physiological relevance of the 28 m - et expression changes in the TST alter the concentration of several abolic genes that act as cQTLs, we showed that their expression bioactive metabolites in multiple pathways, and that this has the in M.  tuberculosis-infected lymph nodes strongly correlated potential to impact the nature of the host immune response to with expression in the TST (Figure  2D), indicating that these M. tuberculosis. metabolic genes have the potential to exert cQTL activity at 166 • JID 2018:218 (1 July) • BRIEF REPORT Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Figure 1. Metabolic pathways enriched in tuberculin skin test (TST) responses relative to saline injection. A, Expression of TST signature in Mycobacterium tuberculosis (Mtb)-infected lymph nodes (LN) relative to healthy LN. Each dot represents 1 sample. Horizontal lines represent median value expression. * P < .0001 by Mann-Whitney test. B, Pairwise comparison of 292 genes differentially expressed in TST relative to saline injection compared to the expression difference between the blood of patients with active tuberculosis (TB) and healthy volunteers (HV). C, The top 10 most statistically enriched KEGG metabolic pathways in the TST relative to saline injection are represented in a network plot, in which the edges indicate associations between genes (light gray nodes) and named pathways (dark gray nodes), and the node size is proportional to the respective pathway −log10 P value enrichment statistic. D, Reporter metabolites predicted to be differentially expressed in TST compared to saline injection. Metabolites selected for known association with metabolic pathways and ranked by increasing P value. Abbreviations: r , Spearman rank correlation coefficient; CoV, covariance. BRIEF REPORT • JID 2018:218 (1 July) • 167 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Figure 2. Identification of cytokine quantitative trait loci (cQTLs) within metabolic genes differentially expressed in a tuberculin skin test (TST). A, Representative box plots of association between single nucleotide polymorphisms (SNP) genotypes and Mycobacterium tuberculosis (Mtb)-induced cytokine levels. Length of the box is the interquar- tile range and whiskers indicate the range of 1.5 × the length of the box from either end of the box. P values were obtained using linear regression analysis of cytokine on genotype data. B, Heatmap of all 47 cQTL SNPs and their relationship to cytokine secretion following peripheral blood mononuclear cell (PBMC) or macrophage (Mfs) stimula- tion with M. tuberculosis lysate. C, Interleukin-6 (IL-6) and IL-17 secretion from M. tuberculosis lysate-stimulated PBMC in the presence or absence of buthionine sulfoximine (BSO) or diethylmaleate (DEM). Horizontal lines represent median value expression. * P < .01 Wilcoxon signed-rank test. D, Gene expression of 28 metabolic genes with cQTL SNPs in tuberculin skin test (TST) relative to saline injection compared to the expression in M. tuberculosis-infected lymph nodes (TBLN) relative to healthy lymph nodes (LN). Abbreviations: IFN-γ, interferon-gamma; r , Spearman rank correlation coefficient; RPMI, RPMI 1640 medium; TNF- α, tumor necrosis factor-alpha. 168 • JID 2018:218 (1 July) • BRIEF REPORT Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 the site of human tuberculosis disease, thus shaping the local genes predominantly impact the secretion of T-cell–derived inflammatory response to M. tuberculosis. cytokines IFN-γ and IL-17. As such, therapeutic manipulation of pyrimidine and glutathione pathways may influence T helper Discussion (T 1) and T 17 polarization, and thus the balance between pro - H H tection and pathology in M. tuberculosis-infected tissues [14]. In Individual metabolic pathways have been shown to ae ff ct the contrast, genes involved in amino acid and inositol phosphate host response to M. tuberculosis [2, 3, 5], but their relative con- metabolism impact a wider array of cytokines, including those tribution in a multicellular tissue infection setting has not been secreted from myeloid cells, which may in part relate to the cen - investigated. We took a systems approach using the human in tral role of the metabolite phosphatidylinositol as the backbone vivo TST challenge model, revealing gene expression changes in of signal transduction components inositol triphosphate (IP3) multiple metabolic pathways that in turn predict enrichment of and protein kinase B (Akt) [6]. several bioactive metabolites. Genetic polymorphisms in these Our study has some limitations. Firstly, we restricted our differentially expressed metabolic genes, as well as pharmaco - cQTL analyses to SNPs that were also cis-eQTLs to limit mul - logical inhibition, were found to controM.  l tuberculosis-induced tiple testing errors, thus likely missing other functionally r -el cytokine production. Finally, expression of these same genes evant cQTLs that may act in trans via other genes. Equally, at the site of tuberculosis disease closely correlated with that identification of SNPs that were eQTLs was limited to databases observed in the TST, underlying the functional relevance of our that probed blood and tissues not infected with M.  tuberculo- findings to the pathogenesis of human tuberculosis infection. sis, possibly missing other functionally relevant SNPs in the e infl Th ammatory response to TST challenge reflects the context of M.  tuberculosis infection. Furthermore, the effector immunopathological changes in human tuberculosis disease cytokines studied for cQTL analysis comprise only one facet [7]. Interestingly, in terms of metabolic gene expression, the TST of the host immune response to M.  tuberculosis infection and site also closely mirrors M.  tuberculosis-infected lymph nodes, were restricted to the manually selected panel available in the unlike the peripheral blood of tuberculosis patients, which 500FG database, introducing bias into the breadth of immu- showed quantitatively smaller changes. Although this may nological effector functions exerted by metabolic gene SNPs. reflect the TST response occurring aer ac ft ute antigenic chal - Finally, although reporter metabolite analysis predicted co -m lenge in contrast to steady-state assessment in blood, in these putationally that the gene expression changes impacted on the same patients the blood transcriptome shows significant devia - concentration of several key metabolites, we were not able to tion from health [7, 12], illustrating that tissue may be the more assess this directly. The study was conceived retrospectively, and appropriate context when studying changes in metabolism. thus tissue samples had already been used wholly for transcr-ip We observed differential gene expression in the TST of mul- tomic analysis. The relationship between tissue gene expression tiple metabolic pathways, including glycolysis, glutathione, changes and metabolite concentrations will need to be verified pyrimidine, and inositol phosphate metabolism. The relevance experimentally in future studies. of changes in the metabolic environment aer ft M.  tuberculo- In summary, this study made a comprehensive assessment of sis infection was supported by the fact that we identified sev - the human tissue metabolic transcriptional response to in vivo eral SNPs in these genes that both act as cis-eQTLs and also mycobacterial antigenic stimulation. A  number of known and impact M.  tuberculosis-induced cytokine secretion by PBMC novel metabolic pathways were differentially expressed, and and macrophages [8], cells that are enriched at the site of TST genetic variation in identified genes ae ff cted cytokine responses [7]. For functional validation, we selected one of the pat-h to M. tuberculosis. Therefore, our systems approach provided a ways identified through gene expression and cQTL analysis, novel list of putative metabolic gene cQTLs that are amenable the glutathione pathway, whereby pharmacological inhibition to further experimental validation, and revealed a new layer of affected M. tuberculosis-induced production of IL-6 and IL-17. complexity to the host antimycobacterial response, supporting Importantly, the 500FG cohort data demonstrated that var-i the use of host-directed strategies that target cellular meta -b able expression of genes involved in glutathione metabolism olism, such as the regulation of glycolysis by metformin [ 15]. (GSTM2, ANPEP, LAP3, and GSTA3) impacted the secretion of Furthermore, our approach combining transcriptomics and these same cytokines. Furthermore, we have previously demon - functional genomics illustrates a pipeline that can be used to strated that inhibiting glycolysis using 2-deoxyglucose impaired identify novel and clinically relevant pathways in the context of IL-22 secretion by M.  tuberculosis-stimulated PBMC, in turn other infectious diseases. validating the observation from the 500FG data that SNPs in the glycolysis gene ADPGK are cQTLs for IL-22 secretion [6]. Supplementary Data Our analyses also identified new roles for pyrimidine metabolism, which has recently been associated with regu - Supplementary materials are available a e Th t Journal of Infectious lating inflammasome activity and cytokine secretion [ 13]. Diseases online. Consisting of data provided by the authors to We demonstrate that pyrimidine and glutathione metabolism benefit the reader, the posted materials are not copyedited and BRIEF REPORT • JID 2018:218 (1 July) • 169 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 are the sole responsibility of the authors, so questions or com - glycolysis in human alveolar macrophages that is required ments should be addressed to the corresponding author. for control of intracellular bacillary replication. J Immunol 2016; 196:2444–9. Notes 6. Lachmandas E, Beigier-Bompadre M, Cheng SC, et  al. Financial support. This work was supported by the Rewiring cellular metabolism via the AKT/mTOR pathway Wellcome Trust Research Fellowship to G.  P. (grant number contributes to host defence against Mycobacterium tuber- WT101766/Z/13/Z); Wellcome Trust Investigator Award to culosis in human and murine cells. Eur J Immunol 2016; M.  N. (grant number 207511/Z/17/Z); the National Institute 46:2574–86. for Health Research Biomedical Research Centre at University 7. Bell LC, Pollara G, Pascoe M, et al. In vivo molecular -dis College London Hospital; by a European Research Counci l section of the effects of HIV-1 in active tuberculosis. PLoS Consolidator Grant to M.  G. N.  (grant number 310372); and Pathog 2016; 12:e1005469. a Spinoza grant of the Netherlands Organization for Scientific 8. Li Y, Oosting M, Smeekens SP, et al. A functional genomics Research (to M.  G.). N.  R.  v.  C.  and E.  L.  were supported by approach to understand variation in cytokine production in the European Union’s Seventh Framework Programme project humans. Cell 2016; 167:1099–1110.e14. TANDEM (grant number HEALTH-F3-2012–305279). 9. Patil KR, Nielsen J. Uncovering transcriptional regulation Potential conifl cts of interest. All authors: No reported con - of metabolism by using metabolic network topology. Proc flicts.. All authors have submitted the ICMJE Form for Disclosure Natl Acad Sci U S A 2005; 102:2685–9. of Potential Conflicts of Interest. Conflicts that the editors consider 10. Agren R, Liu L, Shoaie S, Vongsangnak W, Nookaew I, relevant to the content of the manuscript have been disclosed. Nielsen J. The RAVEN toolbox and its use for generating a genome-scale metabolic model foP r enicillium chrysogenum. References Maranas CD, editor. PLoS Comput Biol 2013; 9:e1002980. 11. Mitchell JB, Russo A, Biaglow JE, McPherson S. Cellular 1. Lachmandas E, Boutens L, Ratter JM, et al. Microbial stim - glutathione depletion by diethyl maleate or buthionine su -lf ulation of different Toll-like receptor signalling pathways oximine: no effect of glutathione depletion on the oxygen induces diverse metabolic programmes in human mono - enhancement ratio. Radiat Res 1983; 96:422–8. cytes. Nat Microbiol 2016; 2:16246. 12. Roe JK, Thomas N, Gil E, et al. Blood transcriptomic diag - 2. Guerra C, Morris D, Sipin A, et al. Glutathione and ada- p nosis of pulmonary and extrapulmonary tuberculosis. JCI tive immune responses against Mycobacterium tuberculosis Insight 2016; 1:e87238. infection in healthy and HIV infected individuals. PLoS 13. Furman D, Chang J, Lartigue L, et al. Expression of specific One 2011; 6:e28378. inflammasome gene modules stratifies older individuals 3. Memari B, Bouttier M, Dimitrov V, et  al. Engagement of into two extreme clinical and immunological states. Nat the aryl hydrocarbon receptor in Mycobacterium tubercu- Med 2017; 23:174–84. losis-infected macrophages has pleiotropic effects on innate 14. Mourik BC, Lubberts E, Steenwinkel JEM de, Ottenho ff immune signaling. J Immunol 2015; 195:4479–91. THM, Leenen PJM. Interactions between Type 1 interfer- 4. Zhang YJ, Reddy MC, Ioerger TR, et  al. Tryptophan bio - ons and the Th17 response in tuberculosis: lessons learned synthesis protects mycobacteria from CD4 T-cell-mediated from autoimmune diseases. Front Immunol 2017; 8:294. killing. Cel2013 l ; 155:1296–308. 15. Singhal A, Jie L, Kumar P, et al. Metformin as adjunct ant-i 5. Gleeson LE, Sheedy FJ, Palsson-McDermott EM, et  al. tuberculosis therapy. Sci Transl Med 2014; 6:263ra159. Cutting edge: Mycobacterium tuberculosis induces aerobic 170 • JID 2018:218 (1 July) • BRIEF REPORT http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journal of Infectious Diseases Oxford University Press

Tissue Metabolic Changes Drive Cytokine Responses to Mycobacterium tuberculosis

Download PDF
7 pages

Loading next page...
 
/lp/ou_press/tissue-metabolic-changes-drive-cytokine-responses-to-mycobacterium-7FVxecomig

References (15)

Publisher
Oxford University Press
Copyright
Copyright © 2022 Infectious Diseases Society of America
ISSN
0022-1899
eISSN
1537-6613
DOI
10.1093/infdis/jiy173
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Tissue Metabolic Changes Drive Cytokine Responses to Mycobacterium tuberculosis Ekta Lachmandas, Ana B Rios-Miguel, Valerie A C M Koeken, Eva van der Pasch, Vinod Kumar, Vasiliki Matzaraki, Yang Li, Marije Oosting, Leo A B Joosten, Richard A Notebaart, Mahdad Noursadeghi, Mihai G Netea, Reinout van Crevel, Gabriele Pollara Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 The Journal of Infectious Diseases BRIEF REPORT Our group has made use of the human tuberculin skin test Tissue Metabolic Changes Drive (TST) challenge model to faithfully reflect the inflammatory Cytokine Responses to Mycobacterium changes that occur at the site of tuberculosis disease, char - tuberculosis acterizing the tissue immunological pathways induced early after mycobacterial antigen exposure [ 7]. However, to date, no 1 1 1 Ekta Lachmandas, Ana B. Rios-Miguel, Valerie A. C. M. Koeken, Eva van der 1 1,2 2 2 1 Pasch, Vinod Kumar, Vasiliki Matzaraki, Yang Li, Marije Oosting, Leo A. studies have explored the metabolic changes and their fun-c 1 1,3 4 1 B. Joosten, Richard A. Notebaart, Mahdad Noursadeghi, Mihai G. Netea, tional consequences on downstream cytokine responses in 1 4 Reinout van Crevel, and Gabriele Pollara such a model. Quantitative cytokine production in response Department of Internal Medicine and Radboud Center for Infectious Diseases, to mycobacterial stimulation has been associated with genetic Radboud University Medical Center, Nijmegen, and University of Groningen, University Medical Center Groningen, The Netherlands; Laboratory of Food polymorphisms [8]. These cytokine quantitative trait loci Microbiology, Wageningen University and Research, The Netherlands; and Division (cQTLs) provide a functional insight into how genetics influ - July of Infection and Immunity, University College London, United Kingdom ences an inflammatory response, and in turn identify criti - Cellula m r etabolism can influence host immune responses to cal pathways that may be amenable to host-directed therapy. Mycobacterium tuberculosis. Using a systems biology approach, In this study, we use the TST model to test the hypothesis differential expression of 292 metabolic genes involved in glyco- l that differential tissue expression of genes involved in reg- ysis, glutathione, pyrimidine, and inositol phosphate pathways ulating metabolic pathways can directly influence cytokine was evident at the site of a human tuberculin skin test challenge production following M.  tuberculosis stimulation. In turn, in patients with active tuberculosis infection. For 28 metabolic our findings provide putative mechanistic links between the genes, we identified single nucleotide polymorphisms that were trans-acting for in vitro cytokine responses to M.  tuberculosis activity of cellular metabolic pathways and immune effector stimulation, including glutathione and pyrimidine metabolism functions. genes that alter production of Th1 and Th17 cytokines. Our METHODS findings identify novel therapeutic targets in host metabolism that may shape protective immunity to tuberculosis. Transcriptomic Data Analysis Keywords. tuberculosis; immune response; immunome- Transcriptomes were derived from TST and blood of patients tabolism; metabolism; cytokines; transcriptomics; microarrays; with active tuberculosis, and human M.  tuberculosis-in- functional genomics; human challenge model. fected and healthy lymph nodes from a separate cohort (Supplementary Table  S1 ). All patients were human immu - e ac Th tivity of many cellular metabolic pathways can impact nodeficiency virus (HIV) seronegative and none had diabetes the host immune response to infections [1]. Individual met - mellitus. abolic pathways have been implicated in antimycobacterial OA-CC-BY We used the KEGG pathway database (http://www.genome. responses: glutathione enhances interleukin-12 (IL-12) and jp/kegg/pathway.html) to derive 33 pathways associated with interferon-gamma (IFN-γ) secretion following Mycobacterium human metabolism, yielding a list of 1422 metabolic genes co -n tuberculosis stimulation [2]; tryptophan catabolism is involved taining no duplicate genes and no annotation to the original in M.  tuberculosis-induced production of IL-1β and IL-23 [3] pathways (Supplementary Table S2 and Table S3 ). Bioinformatic and control of M.  tuberculosis growth [4]; and a shift towards analyses were performed as previously described [ 7] (see aerobic glycolysis in M. tuberculosis-infected macrophages reg- Supplementary Methods ). ulates IL-1β production [ 5, 6]. However, as metabolic reactions are intrinsically interdependent, the challenge lies in determin - SNP Extraction and cQTL Mapping ing the relative roles of these and other pathways during in vivo Single nucleotide polymorphisms (SNPs) within metabolic M. tuberculosis infection. genes and cQTLs were identified as previously described [8] (see Supplementary Methods ). Received 10 July 2017; editorial decision 20 March 2018; accepted 2 April 2018; published online April 3, 2018. Metabolite Reporter Analysis Correspondence: G. Pollara, MBBS, MRCP, FRCPath, PhD, Division of Infection and Immunity, Reporter metabolite analysis [ 9] was performed in Matlab University College London, London WC1E 6BT, UK ([email protected]). using the RAVEN Toolbox (http://biomet-toolbox.chalmers. The Journal of Infectious Diseases 2018;21800:165–70 se/index.php?page=downtools-raven ) and the human genome- © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons scale metabolic reconstruction network HMR 2.00 provided in Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted Human Protein Atlas (http://www.metabolicatlas.org/down- reuse, distribution, and reproduction in any medium, provided the original work is properly cited. DOI: 10.1093/infdis/jiy173 loads/hmr ) (see Supplementary Methods ). BRIEF REPORT • JID 2018:218 (1 July) • 165 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Metabolite Depletion Experiments Next, we sought to test the hypothesis that metabolic changes Peripheral blood mononuclear cells (PBMC) were isolated from in the tissue environment influence M.  tuberculosis-induced 9 healthy Dutch adult volunteers and stimulated witM.  h tuber- cytokine production. We used natural genetic variation to culosis lysate in the presence or absence of pharmacological identify putative cis-acting SNPs for metabolic gene expres- manipulators of the glutathione pathway (see Supplementary sion and tested whether they might be trans-acting SNPs for Methods). cytokine responses. We identified metabolic genes differen - tially expressed in the TST and assessed their impact on cyt-o RESULTS kine secretion following M. tuberculosis lysate stimulation in a We have previously demonstrated that the transcriptional cohort of 500 healthy individuals (500FG cohort) 8 []. First, we response at the site of TST is characterized by upregulation used genotypes extracted from the 1000 Genomes Project to of 1725 genes that closely reflect changes seen in dissected identify 16 061 SNPs from the 109 metabolic genes that com- human tuberculosis granuloma relative to healthy lung t -is prised the 10 most enriched metabolic pathways in the TST sue [7]. We now show that the TST signature is also enriched (Figure 1C). We then assessed which of these SNPs were associ- within M. tuberculosis-infected relative to healthy lymph nodes ated with variable cytokine secretion, generating 2376 putative (Figure 1A), confirming that the TST transcriptional responses cQTLs. To reduce multiple testing false positives, we focused mirror pathology seen at established sites of human tuberculo - on SNPs in the mRNA coding region of the gene of interest that sis disease. influence the same gene’s transcription (ie, metabolic gene SNPs Out of 1422 metabolic genes, 292 were differentially that were cis-expression QTLs [eQTLs]). expressed at the site of TST versus saline injection in patients This analysis generated 47 cQTL SNPs from 28 metabolic with active tuberculosis disease (Supplementary Figure  S1 and genes (Figure  2A, 2B, Supplementary Table  S6 , and Table  S7). Supplementary Table S4 ). Expression of these genes at tissue level Based on these genes, the most over-represented metabolic showed a remarkable lack of concordance with the blood tra -n pathways included the glutathione, glycolysis, inositol ph -os scriptome of the same patients, as only 9 metabolic genes were phate metabolism, and pyrimidine pathways ( Supplementary differentially expressed in the blood ( Figure 1B, Supplementary Figure  S4). Many amino acid metabolism pathways were Figure S2, and Supplementary Table S5 ). This demonstrates that also observed, and ALDH3A2, ALDH3A1, LDHA, and IL4I1 the TST challenge model provides greater molecular resolution were the most frequent constituent genes from these path- to identify host differential metabolic gene expression to myco- ways (Supplementary Figure  S4). Glutathione and pyrimidine bacterial infection than the blood compartment [ 7]. metabolism predominantly influenced the secretion of IFN-γ Given the interconnected relationship between multiple and IL-17. Seven of 15 (47%) cQTLs that regulated IFN-γ secre - metabolic pathways, we generated a network depicting the 10 tion and 4 of 7 (57%) cQTLs that regulated IL-17 secretion most enriched KEGG annotated metabolic pathways based on were derived from genes assigned to glutathione or pyrimidine the 292 differentially expressed metabolic genes (Figure 1C). As metabolic pathways, whereas no cQTLs from these pathways expected, the TST induced changes in multiple interlinked m- et influenced cytokine secretion by macrophages ( Figure  2B). In abolic pathways. These included not only previously described contrast, genes involved in glycolysis, amino acid, and inositol pathways such as glycolysis and glutathione metabolism, but phosphate metabolism acted as cQTLs more ubiquitously, influ - also others such as inositol phosphate metabolism and specific encing the secretion of both T-cell and myeloid cell-derived amino acid metabolic pathways not previously associated with cytokines (IL-1β, IL-6, IL-22, and tumor necrosis factor- alpha tuberculosis [1, 2]. Because differentially expressed genes within [TNF-α]) (Figure 2B). the same metabolic pathway were both up- or downregulated We validated some of the relationships between metabolic (Supplementary Figure S3), we evaluated which metabolites in changes and cytokine secretion in vitro using buthionine sulf - these pathways were most ae ff cted using reporter metabolite oximine (BSO), an inhibitor of glutathione synthesis, and dieth - analysis. This identified metabolites in the human metabolic ylmaleate (DEM), which depletes cells of glutathion 11 e [ ]. Both network around which the most transcriptional changes occur resulted in increased IL-6 and decreased IL-17 secretion by [10]. e Th model predicted changes in several key metabolites M.  tuberculosis-stimulated PBMC ( Figure  2C), consistent with between TST and saline, including 1,3-bisphospho- d-glycer - the cQTL data demonstrating that variable expression of glu - ate (glycolysis), l-formylkynurenine (tryptophan metabolism), tathione genes impacted the secretion of both these cytokines 1-phosphatidyl-1d-myo-inositol-3,4-bisphosphate, and glu - (Figure 2B). tathione (Figure  1D). Therefore, the model predicts that gene Finally, to explore the physiological relevance of the 28 m - et expression changes in the TST alter the concentration of several abolic genes that act as cQTLs, we showed that their expression bioactive metabolites in multiple pathways, and that this has the in M.  tuberculosis-infected lymph nodes strongly correlated potential to impact the nature of the host immune response to with expression in the TST (Figure  2D), indicating that these M. tuberculosis. metabolic genes have the potential to exert cQTL activity at 166 • JID 2018:218 (1 July) • BRIEF REPORT Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Figure 1. Metabolic pathways enriched in tuberculin skin test (TST) responses relative to saline injection. A, Expression of TST signature in Mycobacterium tuberculosis (Mtb)-infected lymph nodes (LN) relative to healthy LN. Each dot represents 1 sample. Horizontal lines represent median value expression. * P < .0001 by Mann-Whitney test. B, Pairwise comparison of 292 genes differentially expressed in TST relative to saline injection compared to the expression difference between the blood of patients with active tuberculosis (TB) and healthy volunteers (HV). C, The top 10 most statistically enriched KEGG metabolic pathways in the TST relative to saline injection are represented in a network plot, in which the edges indicate associations between genes (light gray nodes) and named pathways (dark gray nodes), and the node size is proportional to the respective pathway −log10 P value enrichment statistic. D, Reporter metabolites predicted to be differentially expressed in TST compared to saline injection. Metabolites selected for known association with metabolic pathways and ranked by increasing P value. Abbreviations: r , Spearman rank correlation coefficient; CoV, covariance. BRIEF REPORT • JID 2018:218 (1 July) • 167 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 Figure 2. Identification of cytokine quantitative trait loci (cQTLs) within metabolic genes differentially expressed in a tuberculin skin test (TST). A, Representative box plots of association between single nucleotide polymorphisms (SNP) genotypes and Mycobacterium tuberculosis (Mtb)-induced cytokine levels. Length of the box is the interquar- tile range and whiskers indicate the range of 1.5 × the length of the box from either end of the box. P values were obtained using linear regression analysis of cytokine on genotype data. B, Heatmap of all 47 cQTL SNPs and their relationship to cytokine secretion following peripheral blood mononuclear cell (PBMC) or macrophage (Mfs) stimula- tion with M. tuberculosis lysate. C, Interleukin-6 (IL-6) and IL-17 secretion from M. tuberculosis lysate-stimulated PBMC in the presence or absence of buthionine sulfoximine (BSO) or diethylmaleate (DEM). Horizontal lines represent median value expression. * P < .01 Wilcoxon signed-rank test. D, Gene expression of 28 metabolic genes with cQTL SNPs in tuberculin skin test (TST) relative to saline injection compared to the expression in M. tuberculosis-infected lymph nodes (TBLN) relative to healthy lymph nodes (LN). Abbreviations: IFN-γ, interferon-gamma; r , Spearman rank correlation coefficient; RPMI, RPMI 1640 medium; TNF- α, tumor necrosis factor-alpha. 168 • JID 2018:218 (1 July) • BRIEF REPORT Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 the site of human tuberculosis disease, thus shaping the local genes predominantly impact the secretion of T-cell–derived inflammatory response to M. tuberculosis. cytokines IFN-γ and IL-17. As such, therapeutic manipulation of pyrimidine and glutathione pathways may influence T helper Discussion (T 1) and T 17 polarization, and thus the balance between pro - H H tection and pathology in M. tuberculosis-infected tissues [14]. In Individual metabolic pathways have been shown to ae ff ct the contrast, genes involved in amino acid and inositol phosphate host response to M. tuberculosis [2, 3, 5], but their relative con- metabolism impact a wider array of cytokines, including those tribution in a multicellular tissue infection setting has not been secreted from myeloid cells, which may in part relate to the cen - investigated. We took a systems approach using the human in tral role of the metabolite phosphatidylinositol as the backbone vivo TST challenge model, revealing gene expression changes in of signal transduction components inositol triphosphate (IP3) multiple metabolic pathways that in turn predict enrichment of and protein kinase B (Akt) [6]. several bioactive metabolites. Genetic polymorphisms in these Our study has some limitations. Firstly, we restricted our differentially expressed metabolic genes, as well as pharmaco - cQTL analyses to SNPs that were also cis-eQTLs to limit mul - logical inhibition, were found to controM.  l tuberculosis-induced tiple testing errors, thus likely missing other functionally r -el cytokine production. Finally, expression of these same genes evant cQTLs that may act in trans via other genes. Equally, at the site of tuberculosis disease closely correlated with that identification of SNPs that were eQTLs was limited to databases observed in the TST, underlying the functional relevance of our that probed blood and tissues not infected with M.  tuberculo- findings to the pathogenesis of human tuberculosis infection. sis, possibly missing other functionally relevant SNPs in the e infl Th ammatory response to TST challenge reflects the context of M.  tuberculosis infection. Furthermore, the effector immunopathological changes in human tuberculosis disease cytokines studied for cQTL analysis comprise only one facet [7]. Interestingly, in terms of metabolic gene expression, the TST of the host immune response to M.  tuberculosis infection and site also closely mirrors M.  tuberculosis-infected lymph nodes, were restricted to the manually selected panel available in the unlike the peripheral blood of tuberculosis patients, which 500FG database, introducing bias into the breadth of immu- showed quantitatively smaller changes. Although this may nological effector functions exerted by metabolic gene SNPs. reflect the TST response occurring aer ac ft ute antigenic chal - Finally, although reporter metabolite analysis predicted co -m lenge in contrast to steady-state assessment in blood, in these putationally that the gene expression changes impacted on the same patients the blood transcriptome shows significant devia - concentration of several key metabolites, we were not able to tion from health [7, 12], illustrating that tissue may be the more assess this directly. The study was conceived retrospectively, and appropriate context when studying changes in metabolism. thus tissue samples had already been used wholly for transcr-ip We observed differential gene expression in the TST of mul- tomic analysis. The relationship between tissue gene expression tiple metabolic pathways, including glycolysis, glutathione, changes and metabolite concentrations will need to be verified pyrimidine, and inositol phosphate metabolism. The relevance experimentally in future studies. of changes in the metabolic environment aer ft M.  tuberculo- In summary, this study made a comprehensive assessment of sis infection was supported by the fact that we identified sev - the human tissue metabolic transcriptional response to in vivo eral SNPs in these genes that both act as cis-eQTLs and also mycobacterial antigenic stimulation. A  number of known and impact M.  tuberculosis-induced cytokine secretion by PBMC novel metabolic pathways were differentially expressed, and and macrophages [8], cells that are enriched at the site of TST genetic variation in identified genes ae ff cted cytokine responses [7]. For functional validation, we selected one of the pat-h to M. tuberculosis. Therefore, our systems approach provided a ways identified through gene expression and cQTL analysis, novel list of putative metabolic gene cQTLs that are amenable the glutathione pathway, whereby pharmacological inhibition to further experimental validation, and revealed a new layer of affected M. tuberculosis-induced production of IL-6 and IL-17. complexity to the host antimycobacterial response, supporting Importantly, the 500FG cohort data demonstrated that var-i the use of host-directed strategies that target cellular meta -b able expression of genes involved in glutathione metabolism olism, such as the regulation of glycolysis by metformin [ 15]. (GSTM2, ANPEP, LAP3, and GSTA3) impacted the secretion of Furthermore, our approach combining transcriptomics and these same cytokines. Furthermore, we have previously demon - functional genomics illustrates a pipeline that can be used to strated that inhibiting glycolysis using 2-deoxyglucose impaired identify novel and clinically relevant pathways in the context of IL-22 secretion by M.  tuberculosis-stimulated PBMC, in turn other infectious diseases. validating the observation from the 500FG data that SNPs in the glycolysis gene ADPGK are cQTLs for IL-22 secretion [6]. Supplementary Data Our analyses also identified new roles for pyrimidine metabolism, which has recently been associated with regu - Supplementary materials are available a e Th t Journal of Infectious lating inflammasome activity and cytokine secretion [ 13]. Diseases online. Consisting of data provided by the authors to We demonstrate that pyrimidine and glutathione metabolism benefit the reader, the posted materials are not copyedited and BRIEF REPORT • JID 2018:218 (1 July) • 169 Downloaded from https://academic.oup.com/jid/article/218/1/165/4959339 by DeepDyve user on 13 July 2022 are the sole responsibility of the authors, so questions or com - glycolysis in human alveolar macrophages that is required ments should be addressed to the corresponding author. for control of intracellular bacillary replication. J Immunol 2016; 196:2444–9. Notes 6. Lachmandas E, Beigier-Bompadre M, Cheng SC, et  al. Financial support. This work was supported by the Rewiring cellular metabolism via the AKT/mTOR pathway Wellcome Trust Research Fellowship to G.  P. (grant number contributes to host defence against Mycobacterium tuber- WT101766/Z/13/Z); Wellcome Trust Investigator Award to culosis in human and murine cells. Eur J Immunol 2016; M.  N. (grant number 207511/Z/17/Z); the National Institute 46:2574–86. for Health Research Biomedical Research Centre at University 7. Bell LC, Pollara G, Pascoe M, et al. In vivo molecular -dis College London Hospital; by a European Research Counci l section of the effects of HIV-1 in active tuberculosis. PLoS Consolidator Grant to M.  G. N.  (grant number 310372); and Pathog 2016; 12:e1005469. a Spinoza grant of the Netherlands Organization for Scientific 8. Li Y, Oosting M, Smeekens SP, et al. A functional genomics Research (to M.  G.). N.  R.  v.  C.  and E.  L.  were supported by approach to understand variation in cytokine production in the European Union’s Seventh Framework Programme project humans. Cell 2016; 167:1099–1110.e14. TANDEM (grant number HEALTH-F3-2012–305279). 9. Patil KR, Nielsen J. Uncovering transcriptional regulation Potential conifl cts of interest. All authors: No reported con - of metabolism by using metabolic network topology. Proc flicts.. All authors have submitted the ICMJE Form for Disclosure Natl Acad Sci U S A 2005; 102:2685–9. of Potential Conflicts of Interest. Conflicts that the editors consider 10. Agren R, Liu L, Shoaie S, Vongsangnak W, Nookaew I, relevant to the content of the manuscript have been disclosed. Nielsen J. The RAVEN toolbox and its use for generating a genome-scale metabolic model foP r enicillium chrysogenum. References Maranas CD, editor. PLoS Comput Biol 2013; 9:e1002980. 11. Mitchell JB, Russo A, Biaglow JE, McPherson S. Cellular 1. Lachmandas E, Boutens L, Ratter JM, et al. Microbial stim - glutathione depletion by diethyl maleate or buthionine su -lf ulation of different Toll-like receptor signalling pathways oximine: no effect of glutathione depletion on the oxygen induces diverse metabolic programmes in human mono - enhancement ratio. Radiat Res 1983; 96:422–8. cytes. Nat Microbiol 2016; 2:16246. 12. Roe JK, Thomas N, Gil E, et al. Blood transcriptomic diag - 2. Guerra C, Morris D, Sipin A, et al. Glutathione and ada- p nosis of pulmonary and extrapulmonary tuberculosis. JCI tive immune responses against Mycobacterium tuberculosis Insight 2016; 1:e87238. infection in healthy and HIV infected individuals. PLoS 13. Furman D, Chang J, Lartigue L, et al. Expression of specific One 2011; 6:e28378. inflammasome gene modules stratifies older individuals 3. Memari B, Bouttier M, Dimitrov V, et  al. Engagement of into two extreme clinical and immunological states. Nat the aryl hydrocarbon receptor in Mycobacterium tubercu- Med 2017; 23:174–84. losis-infected macrophages has pleiotropic effects on innate 14. Mourik BC, Lubberts E, Steenwinkel JEM de, Ottenho ff immune signaling. J Immunol 2015; 195:4479–91. THM, Leenen PJM. Interactions between Type 1 interfer- 4. Zhang YJ, Reddy MC, Ioerger TR, et  al. Tryptophan bio - ons and the Th17 response in tuberculosis: lessons learned synthesis protects mycobacteria from CD4 T-cell-mediated from autoimmune diseases. Front Immunol 2017; 8:294. killing. Cel2013 l ; 155:1296–308. 15. Singhal A, Jie L, Kumar P, et al. Metformin as adjunct ant-i 5. Gleeson LE, Sheedy FJ, Palsson-McDermott EM, et  al. tuberculosis therapy. Sci Transl Med 2014; 6:263ra159. Cutting edge: Mycobacterium tuberculosis induces aerobic 170 • JID 2018:218 (1 July) • BRIEF REPORT

Journal

The Journal of Infectious Diseases – Oxford University Press

Published: Jun 5, 2018

Keywords: cytokine; metabolism; immune response; genes; mycobacterium tuberculosis; glutathione

There are no references for this article.