Myeloid HIF‐1α regulates pulmonary inflammation during experimental Mycobacterium tuberculosis infection

Summary

The transcription factor hypoxia‐inducible factor‐1 alpha (HIF‐1α) is a key regulator of the response and function of myeloid cells in hypoxic and inflammatory microenvironments. To define the role of HIF‐1α in tuberculosis, the progression of aerosol Mycobacterium tuberculosis infection was analysed in mice deficient in HIF‐1α in the myeloid lineage (mHIF‐1α^−/−). We show that myeloid HIF‐1α is not required for the containment of the infection, as both wild‐type (WT) and mHIF‐1α^−/− mice mounted normal Th1 responses and maintained control of bacterial growth throughout infection. However, during chronic infection mHIF‐1α^−/− mice developed extensive lymphocytic inflammatory involvement of the interstitial lung tissue and died earlier than WT mice. These data support the hypothesis that HIF‐1α activity coordinates the response of myeloid cells during M. tuberculosis infection to prevent excessive leucocyte recruitment and immunopathological consequences to the host.

Expression of HIF-1α in myeloid cells is shown to be required to limit lung inflammation during experimental tuberculosis. Mice deficient in myeloid HIF-1α expression showed increased mortality associated with a hyper-inflammatory response in the lung but normal protective type 1 immune responses.

Abbreviations

CFU

colony‐forming units

dpi

days post‐infection

H&E

haematoxylin & eosin

HIF‐1α

hypoxia‐inducible factor‐1 alpha

IFN‐γ

interferon‐gamma

IL

interleukin

iNOS

inducible nitric oxide synthase

mHIF‐1α^−/−

myeloid‐deficient HIF‐1α

Mtb

Mycobacterium tuberculosis

TB

tuberculosis

Th1

T helper 1

TNF‐α

tumour necrosis factor‐alpha

WT

wild‐type

Introduction

The development of granulomas is the most distinctive pathological hallmark of the host response to Mycobacterium tuberculosis (Mtb) infections.1 Classically, granulomas have been associated with a protective response that contains the infection. However, some granulomas undergo central necrosis (caseous necrosis), a process that is also required for the pathogenesis and dissemination of Mtb.1 Necrotic granulomas are hypoxic; 2, 3 thus, one possible mechanism for the development of central granuloma necrosis is macrophage cell death caused by the reduced oxygen tension and the inability of these cells to adapt to hypoxia. In this regard, adaptation of immune cells to hypoxia requires a metabolic shift from oxidative phosphorylation toward enhanced glucose uptake, glycolysis and production of lactate, a bioenergetic signature that is controlled by the transcription factor hypoxia‐inducible factor (HIF)‐1α.5, 6 In support of the hypothesis that HIF‐1α activity prevents the earlier emergence of granuloma necrosis, a recent report using a mouse model of Mycobacterium avium infection showed that mice deficient in HIF‐1α in the myeloid compartment developed granuloma necrosis before wild‐type (WT) mice.8 While most tuberculosis (TB) mouse models do not develop necrotic granulomas upon aerosol Mtb infection, another recent report has shown that the activity of HIF‐1α was critical for the antimycobacterial response of macrophages.9 Indeed, the transcriptional response of Mtb‐infected macrophages to interferon‐gamma (IFN‐γ) was partially dependent on HIF‐1α, and mice deficient in this transcription factor in the myeloid compartment were acutely susceptible to Mtb infection.9 This phenotype was in part associated with a reduced production of nitric oxide, a central effector molecule of the macrophage antimycobacterial response9, 10 that also regulates the pathological inflammatory response to Mtb infection.11, 12 These data are intriguing because the acute susceptibility to Mtb infection is generally associated with the deregulation of critical immune pathways.13 Therefore, these data suggest that HIF‐1α may play a key role in coordinating protective immunity to TB.

In this work, we show that HIF‐1α activity in the myeloid lineage is not required for the long‐term containment of Mtb H37Rv growth. We also found that the genetic ablation of this transcription factor was associated with pathological pulmonary inflammation and reduced survival during chronic infection. These data show that HIF‐1α activity is not a critical component of the antimycobacterial response, but its activity is required to restrain pathological lung inflammation and long‐term survival of the host.

Materials and methods

Mice and Mtb aerosol infections

Mice deficient in HIF‐1α in the myeloid compartment (Hif‐1α^fl/fl × LysM‐cre^+/+, mHIF‐1α^−/−) were obtained by crossing single transgenic mice from the Jackson Laboratory (Bar Harbor). C57BL/6 mice were originally from Charles River Laboratory (Barcelona, Spain). All mice used in this study were age‐ and sex‐matched, and between the ages of 8 and 12 weeks. Aerosol infections with Mtb strains H37Rv (originally from the Trudeau Institute) or HN878 were carried out using the Glas‐Col airborne infection system, as previously described.14 Bacterial loads were determined in the homogenate of the target organs that were plated onto Middlebrook 7H11 agar (BD Biosciences) and incubated at 37º for 3 weeks. All procedures were carried out in accordance with the European Directive 86/609/EEC, and were approved by the Subcomissão de Ética para as Ciências da Vida e da Saúde (SECVS 074/2016) and the Portuguese National Authority Direcção Geral de Veterinária (014072).

Leucocyte isolation for flow cytometry

Aseptically excised lungs were sectioned and incubated at 37º for 30 min with collagenase D (0.7 mg/ml; Sigma, St. Louis, MO). Lungs were then disrupted into a single cell suspension by passage through a 70‐μm nylon cell strainer (BD Biosciences, San José, CA). After centrifugation, the cell‐free suspensions were aliquoted and frozen at −80º until the concentration of cytokines was determined using ELISA kits (Thermo Fisher Scientific, Waltham, MA), following the manufacturer’s instructions. Lung single cell suspensions were then treated with erythrocyte lysis buffer (0·87% of NH₄Cl). In order to remove cell debris and non‐haematopoietic cell interference, lung cells were further subjected to a 40%:80% percoll (GE Healthcare, Chicago, IL). The resulting cell suspension was washed twice and counted. For flow cytometry analysis, lung cells were stained with fluorochrome‐conjugated antibodies for 30 min on ice. For intracellular cytokine detection, cells were cultured in 5 µg/ml of ESAT‐6_1‐20 peptide for 1·5 hr before 10 μg/ml of Brefeldin A (Sigma) was added to the culture for 3·5 more hours. After surface staining, cells were fixed with 2% paraformaldehyde (PFA) for 20 min, permeabilized and stained for 30 min on ice. Antibodies specific for Siglec‐F (E50‐2440), Ly6C (HK1.4), CD11b (M1/70), CD45 (30‐F11), CD11c (N418), Ly6G (1A8), CD3 (145‐2C11), CD44 (IM7), CD4 (GK1.5), interleukin (IL)‐17 (TC11‐18H10.1) and IFN‐γ (XMG1.2) were from BD Biosciences, Biolegend or eBioscience. Data were acquired on a LSRII flow cytometer (BD Biosciences) with diva Software and analysed using flowjo software (TreeStar, Ashland, OR). The total number of cells for each population was determined based on the percentage of cells determined by flow cytometry and the total number of cells per lung.

Histology and immunohistochemistry

The apical lobe of each lung was inflated with 4% PFA, embedded in paraffin, and sectioned in 2−3‐μm slices before being stained with haematoxylin & eosin (H&E) or Picrosirius Red. For immunohistochemistry, tissue sections were deparaffinized with xylene and rehydrated in sequential steps from absolute ethanol to distilled water. Antigens were then retrieved using Antigen Retrieval Solution (DAKO) in a 96º pre‐warmed water bath for 30 min, and the endogenous peroxidases were inhibited with 3% hydrogen peroxide in methanol for 10 min. Samples were incubated overnight at 4º with rabbit anti‐mouse inducible nitric oxide synthase (iNOS; sc‐650; Santa Cruz Biotechnology, Dallas, TX). The day after slides were incubated with a biotinylated horse anti‐rabbit secondary antibody (BA‐1100; Vector Laboratories, Burlingame, CA) for 1 hr, followed by horseradish peroxidase‐conjugated streptavidin (S2438; Sigma) for 30 min. The staining was revealed with DAB (DAKO) for 10 min following haematoxylin. Images were captured using an Olympus BX61 microscope and recorded with a digital camera (DP70) using the cell^p software. Image analysis and stain density were performed using Fiji (imagej) software, as previously described.15, 16

Real‐time reverse transcriptase‐polymerase chain reaction

Total RNA was extracted using triple TRIzol (Thermo Fisher Scientific) according to the manufacturer’s instructions. cDNA was generated from 1 mg RNA using the GRS cDNA Synthesis Master Mix (Grisp) following the manufacturer’s instructions. The resultant cDNA template was used to quantify the expression of target genes by real‐time polymerasechain reaction (PCR; Bio‐Rad CFX96 Real‐Time System with C1000 Thermal Cycler) and normalized to Ubiquitin mRNA levels using the ΔCt method. Target gene mRNA expression was quantified using SYBR green (Thermo Scientific) and specific oligonucleotides (Invitrogen, Carlsbad, CA).

Statistical analysis

Statistical differences between experimental groups were determined using unpaired Student’s t‐test with a Welch correction. Differences were considered significant for P ≤ 0·05.

Results

Wild‐type and mutant mice were infected aerogenically with M. tuberculosis H37Rv [500 colony‐forming units (CFU)] and their survival was followed for 200 days. mHIF‐1α^−/− mice began to show signs of illness and decreased survival beginning at 140 days post‐infection (dpi), whereas WT mice remained healthy throughout the experiment (Fig. 1a). We next infected mice to determine whether this susceptibility to infection was associated with a higher bacterial burden. At 25 and 40 dpi we observed that both WT and mHIF‐1α^−/− mice exhibit similar bacterial loads (Fig. 1b). We thus decided to study the bacterial growth and the immune response in mHIF‐1α^−/− mice at later time‐points during the chronic phase. In a first experiment (Exp. 1), we found that both WT and mHIF‐1α^−/− mice maintained similar bacterial burdens in the lungs at 60 and 120 dpi (Fig. 1c). We performed two additional experiments (Exps 2 and 3, each representing the data from two independent infections) and assessed bacterial burdens as well as lung pathology. As shown in Fig. 1(c), lung bacterial burdens were again similar in the two mouse strains, except for day 120 in experiment 3 where mutant mice were slightly more susceptible than B6 animals. Bacterial burdens in the liver and spleen were also not different between the two strains (data not shown). Histological analysis of the lungs revealed no statistically significant differences in inflammatory foci between the two strains at days 25 and 40 (Fig. S1), but an increase in the inflamed area was evident at 60 and 120 dpi in Mtb‐infected mHIF‐1α^−/− mice as compared with controls (Fig. 2a). These data were further confirmed upon measuring the areas of inflammation that showed a clear and reproducible increase in cellular infiltrates in mHIF‐1α^−/− mice when compared with WT mice at 60 and 120 dpi (Fig. 2b). We next tested a hypervirulent clinical strain of M. tuberculosis in a high‐dose (500 CFU) aerosol infection. mHIF‐1α^−/− mice showed increased mortality during the chronic stage of infection with strain HN878 (Fig. 3a). However, bacterial loads in mHIF‐1α^−/− mice at a time‐point preceding the onset of mortality (i.e. day 60) were only very slightly increased in the lung as compared with WT mice, and no differences in splenic loads were observed between the two strains (Fig. 3b). mHIF‐1α^−/− mice also showed exacerbated lung pathology in response to infection by strain HN878 as compared with WT animals (Fig. 3c).

Figure 1.

Myeloid‐deficient hypoxia‐inducible factor‐1 alpha (mHIF‐1α^−/−) mice are more susceptible to chronic Mycobacterium tuberculosis (Mtb) infection. B6 and mHIF‐1α^−/− mice were challenged with different doses of the H37Rv Mtb strain via the aerosol route. (a) Survival of B6 and mHIF‐1α^−/− mice challenged with 500 CFU of Mtb infection. Data represent 10 animals per group from two independent experiments each with five mice per group. (b) Lung bacterial burdens of mice challenged with Mtb (75 CFU) at 25 and 40 days post‐infection (dpi). (c) Mice challenged with Mtb via the 150 CFU in experiment 1, 75 CFU in experiment 2, and 500 CFU in experiment 3. At 60 and 120 dpi, bacterial burdens were assessed in the lungs. Data represent the mean CFU ± SD from four−nine mice per group. Individual plots from experiments 2 and 3 are from two independent assays each with four−five mice per group. Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·05; ns, not significant).

Figure 2.

Hypoxia‐inducible factor 1‐alpha (HIF‐1α) activity in the myeloid compartment regulates pulmonary inflammation during chronic Mycobacterium tuberculosis (Mtb) infection. B6 and myeloid‐deficient (m)HIF‐1α^−/− mice were challenged with Mtb (H37Rv) infection via the aerosol route (experiments 2 and 3). (a) Representative haematoxylin & eosin (H&E) lung sections at 60 and 120 days post‐infection (dpi). (b) Percentage of the infiltrated area in the lungs at 60 and 120 dpi for three experiments. Individual data points represent one lung section. Data represent the mean ± SD from six−10 mice per group. Individual plots from experiments 2 and 3 are from two independent assays each with four−five mice per group. Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·05; **P < 0·01).

Figure 3.

Myeloid‐deficient hypoxia‐inducible factor‐1 alpha (mHIF‐1α^−/−) mice are more susceptible to chronic infection with hypervirulent Mycobacterium tuberculosis (Mtb). B6 and mHIF‐1α^−/− mice were challenged with 500 CFU of Mtb HN878 strain via the aerosol route. (a) Survival curve. (b) Lung and spleen bacterial burdens at 60 days post‐infection (dpi). (c) Representative haematoxylin & eosin (H&E) lung sections. (d) Percentage of the infiltrated area. Individual data points represent one lung section. Data represent the mean ± SD from five mice per group. Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·5; **P < 0·01; ns, not significant).

Considering the role of HIF‐1α^−/− in the transcriptional control of the macrophages responses,9, 10 we next investigated if the susceptibility of mHIF‐1α^−/− mice was caused by an altered expression of protective molecules in the lungs. To this end, we measured the expression of IFN‐γ and antimycobacterial genes induced by IFN‐γ, including that for iNOS (encoded by Nos2)17 and that for the immunity‐related GTPase family M member 1 gene (Irgm1).18 We chose to study the response to strain H37Rv as bacterial loads between the two mouse strains were comparable thus providing similar antigenic stimulation. As shown in Fig. 4(a), there was a similar pattern of iNOS expression in the lung tissue of both WT and mHIF‐1α^−/− mice. This was confirmed by quantitative densitometric analysis of the staining (Fig. 4b), which showed similar intensities of staining in both strains except for a minor decrease in mutant mice at day 60 in experiment 3. Expression of IFN‐γ‐regulated genes such as Nos2 and Irgm1 as well as those encoding for the chemokines CXCL9 to 11 was similar in the two strains (Table 1). While expression of Nos2 is indicative that lung phagocytes express toxic nitrogen radicals, the concomitant expression of Arginase 1 (Arg1) can limit the production of nitric oxide by iNOS.19, 20 Arginase expression can be induced by IL‐10 and IL‐6,20, 22 and promotes exacerbated lung inflammation and TB disease severity.23 Therefore, we measured the expression of these genes as well as other type 2‐induced ones (Fizz1 and Ym1) but found a similar pattern of expression in both WT and mHIF‐1α^−/− mice (Table 1). These data demonstrate that genetic ablation of HIF‐1α in the myeloid lineage does not impair the expression of the type 1 immune response or the activation of macrophages being therefore not required for the containment of Mtb H37Rv growth. However, mutant mice displayed increased pulmonary inflammation and reduced survival during chronic infection.

Figure 4.

Expression of inducible nitric oxide synthase (iNOS) in lungs from wild‐type (WT) and myeloid‐deficient hypoxia‐inducible factor‐1 alpha (mHIF‐1α^−/−) mice at 60 and 120 days post‐infection (dpi) with Mycobacterium tuberculosis (Mtb) H37Rv (experiments 2 and 3). (a) Representative images of the immunohistochemical analysis of iNOS in lung tissues of Mtb‐infected mice. (b) Optical density quantification of iNOS staining. Data represent the mean ± SD of eight−10 mice per group. Individual plots from experiments 2 and 3 are from two independent assays each with four−five mice per group. Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·5; ns, not significant).

Table 1.

Gene expression at 60 and 120 dpi (Mtb H37Rv) in the lungs of WT and mHIF‐1α^−/− mice

Gene		60 dpi			120 dpi
		WT	mHIF−/−	(P‐value)	WT	mHIF−/−	(P‐value)
Nos2	Exp. 2	0·67 ± 1·1	0·53 ± 0·6	(0·739)	0·93 ± 0·7	0·65 ± 0·6	(0·411)
	Exp. 3	0·68 ± 0·9	0·47 ± 0·6	(0·598)	0·31 ± 0·2	0·58 ± 0·5	(0·216)
Irgm1	Exp. 2	5·06 ± 8·0	5·07 ± 6·6	(0·997)	20·44 ± 20·9	11·62 ± 11·4	(0·325)
	Exp. 3	6·59 ± 10·9	4·31 ± 5·2	(0·604)	5·25 ± 5·1	11·45 ± 5·8	(0·045)*
Arg1	Exp. 2	3·08 ± 5·2	1·36 ± 2·1	(0·375)	5·21 ± 5·6	8·83 ± 11·0	(0·452)
	Exp. 3	1·49 ± 2·6	0·70 ± 1·2	(0·451)	3·10 ± 4·2	4·60 ± 4·5	(0·491)
Fizz1	Exp. 2	1695·68 ± 489·2	1429·08 ± 799·2	(0·408)	3153·05 ± 2300·1	2801·21 ± 2026·2	(0·758)
	Exp. 3	1592·08 ± 793·1	2376·44 ± 1198·9	(0·131)	1955·10 ± 2103·2	2921·43 ± 2333·1	(0·385)
Ym1	Exp. 2	0·36 ± 0·5	0·80 ± 1·3	(0·360)	1·46 ± 1·7	1·51 ± 1·5	(0·944)
	Exp. 3	0·53 ± 0·6	0·28 ± 0·3	(0·269)	1·33 ± 1·8	0·67 ± 0·7	(0·335)
Il10	Exp. 2	0·86 ± 0·5	0·72 ± 0·3	(0·513)	0·64 ± 0·4	0·51 ± 0·3	(0·524)
	Exp. 3	0·36 ± 0·3	0·60 ± 0·5	(0·226)	0·51 ± 0·5	0·38 ± 0·2	(0·477)
IL6	Exp. 2	58·43 ± 41·1	50·28 ± 52·2	(0·731)	54·73 ± 51·0	32·18 ± 18·8	(0·274)
	Exp. 3	50·73 ± 39·1	66·36 ± 110·2	(0·699)	33·75 ± 22·9	22·64 ± 7·8	(0·201)
Il23a	Exp. 2	2·13 ± 3·1	4·04 ± 6·5	(0·440)	18·46 ± 22·7	16·88 ± 20·7	(0·890)
	Exp. 3	9·57 ± 13·6	5·26 ± 6·1	(0·458)	20·34 ± 40·7	17·14 ± 16·7	(0·841)
Ifng	Exp. 2	3·17 ± 2·8	2·34 ± 1·5	(0·472)	5·21 ± 5·1	2·96 ± 2·4	(0·285)
	Exp. 3	2·96 ± 3·4	5·38 ± 5·3	(0·278)	2·93 ± 2·6	3·36 ± 4·3	(0·811)
Cxcl1	Exp. 2	11·17 ± 15·5	10·65 ± 14·4	(0·942)	44·43 ± 60·2	34·70 ± 38·5	(0·707)
	Exp. 3	19·36 ± 27·9	12·56 ± 14·1	(0·547)	27·69 ± 30·1	34·78 ± 21·9	(0·599)
Cxcl2	Exp. 2	28·30 ± 22·8	30·22 ± 43·8	(0·910)	31·44 ± 26·9	22·87 ± 28·6	(0·560)
	Exp. 3	15·10 ± 10·9	24·74 ± 29·8	(0·385)	20·55 ± 19·1	17·42 ± 24·2	(0·774)
Cxcl5	Exp. 2	12·23 ± 10·2	19·53 ± 24·0	(0·480)	38·64 ± 26·1	41·68 ± 25·4	(0·823)
	Exp. 3	18·34 ± 27·1	4·05 ± 2·6	(0·181)	19·60 ± 22·5	30·31 ± 23·7	(0·377)
Cxcl9	Exp. 2	189·50 ± 76·4	87·30 ± 67·10	(0·013)*	216·78 ± 76·1	168·21 ± 85·1	(0·237)
	Exp. 3	78·83 ± 90·7	90·97 ± 91·2	(0·781)	105·20 ± 48·7	194·12 ± 140·1	(0·125)
Cxcl10	Exp. 2	305·66 ± 216·2	201·99 ± 150·0	(0·249)	180·27 ± 310·4	91·85 ± 115·8	(0·473)
	Exp. 3	180·93 ± 158·5	325·51 ± 134·8	(0·064)	333·65 ± 416·7	189·76 ± 249·0	(0·427)
Cxcl11	Exp. 2	3·38 ± 1·9	4·65 ± 4·7	(0·445)	6·63 ± 5·2	2·95 ± 1·3	(0·089)
	Exp. 3	3·76 ± 1·9	4·61 ± 2·7	(0·463)	2·42 ± 1·7	3·25 ± 2·9	(0·503)

Values were determined by qPCR and are normalized to Ubq expression. Data represent the mean ± SD of experiments 2 and 3 each with at least eight mice per group.

Abbreviation: dpi, days post‐infection; mHIF^−/−, myeloid‐deficient hypoxia‐inducible factor; WT, wild‐type.

^*Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (^* P < 0·5).

To examine the basis for the increased pulmonary inflammation of mHIF‐1α^−/− mice, we compared the dynamics of cell infiltration in the lungs of Mtb‐infected WT and mHIF‐1α^−/− mice at 60 dpi. The marked fibrosis of the lungs at day 120 (Fig. S2) precluded an efficient isolation of leucocytes and their precise quantification at that time‐point. In a steady‐state, both WT and mHIF‐1α^−/− mice exhibited similar numbers of the assessed cellular populations (Table S1, and Fig. S3 for gating strategies). At 60 dpi, in accordance with the increased inflammatory foci of mHIF‐1α^−/− mice, we found significantly higher numbers of CD45⁺ cells in the lungs of these mice with a twofold increase in experiment 2 and a 2·8‐fold increase in experiment 3 (Table 2). Although some myeloid populations were significantly increased in mHIF‐1α^−/− mice, the bulk of the increase was due to the T‐cell compartment with a threefold increase in experiment 2 and a 3·2‐fold increase in experiment 3, with an excess of 60% of the lung leucocytes being T‐cells in mutant animals. To determine the frequency and number of cytokine‐producing cells, lung single cell suspensions were restimulated with the Mtb immunodominant epitope ESAT‐6_1‐20.14 We found a similar frequency of CD4⁺ T‐cells producing IFN‐γ, tumour necrosis factor (TNF) or both cytokines which, due to the increased numbers of T‐cells in mHIF‐1α^−/− mice, translated into higher numbers of these cell subsets in mHIF‐1α^−/− mice and increased IFN‐γ protein in lung tissue (Fig. 5). On the other hand, both the frequency and numbers of IL‐17‐producing CD4⁺ T‐cells were elevated in mHIF‐1α^−/− mice (Fig. 5). Although the latter correlated with a slightly higher accumulation of neutrophils in mutant mice (Table 2), there was no difference in the expression of IL23a and the neutrophil chemotactic factors Cxcl1, 2 and 5 (Table 1). However, because the PCR data provide information of gene expression on a per cell basis, the fact that the total number of leucocytes roughly doubled in mutant mice supports an overall increase in protein levels for these cytokines and chemokines.

Table 2.

Analysis of different cell populations in the lungs of WT and mHIF‐1α^−/− mice at 60 dpi (Mtb H37Rv)

Cell subset (× 103)		WT	mHIF−/−	(P‐value)
CD45	Exp. 2	1965·77 ± 644·5	3907·41 ± 1277·8	(0·001)**
	Exp. 3	1559·70 ± 354·6	4290·09 ± 1711·1	(0·001)**
Neut.	Exp. 2	145·98 ± 74·4	232·06 ± 42·7	(0·010)*
	Exp. 3	174·49 ± 139·1	475·34 ± 446·1	(0·084)
CD11b DCs	Exp. 2	39·61 ± 13·3	90·46 ± 28·2	(0·0004)***
	Exp. 3	39·98 ± 16·9	95·68 ± 36·1	(0·014)**
CD103 DCs	Exp. 2	14·83 ± 5·3	34·11 ± 16·1	(0·007)**
	Exp. 3	12·77 ± 8·0	23·11 ± 8·8	(0·019)*
Alv MΦ	Exp. 2	0·92 ± 0·4	1·00 ± 0·8	(0·796)
	Exp. 3	0·57 ± 0·5	0·91 ± 0·9	(0·310)
rMono	Exp. 2	14·16 ± 7·8	15·70 ± 7·2	(0·667)
	Exp. 3	7·12 ± 3·7	13·87 ± 7·2	(0·028)*
iMono	Exp. 2	25·41 ± 6·8	20·04 ± 7·8	(0·139)
	Exp. 3	14·53 ± 8·0	21·00 ± 7·7	(0·100)
T‐cells	Exp. 2	855·77 ± 242·5	2640·10 ± 762·1	(<0·0001)****
	Exp. 3	818·56 ± 181·7	2597·72 ± 1040·1	(0·0008)***

The values were obtained by flow cytometric analysis. Data represent the mean ± SD of experiments 2 and 3 each with nine mice per group.

Abbreviation: mHIF^−/−, myeloid‐deficient hypoxia‐inducible factor; WT, wild‐type.

Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·5; **P < 0·01; ***P < 0·01; ****P < 0·001).

Figure 5.

Myeloid hypoxia‐inducible factor‐1 alpha (HIF‐1α) deficiency prompts enhanced CD4⁺ T‐cell responses in myeloid‐deficient (m)HIF‐1α^−/− mice challenged with Mycobacterium tuberculosis (Mtb) H37Rv (experiments 2 and 3). (a) Lung CD4⁺ T‐cell numbers determined by flow cytometry at 60 days post‐infection (dpi). (b−d) Lung cells were restimulated with ESAT‐6_1‐20 to assess the polyfunctionality of CD4⁺ T‐cells. (b) Representative flow cytometry data of interferon‐gamma (IFN‐γ)‐, tumour necrosis factor‐alpha (TNF‐α)‐ and interleukin (IL)‐17‐expressing CD4⁺ T‐cells. (c) Frequencies and (d) numbers of IFN‐γ⁺ single producers (SP), IFN‐γ⁺TNF‐α⁺ double producers (DP), TNF‐α⁺ SP and IL‐17 SP CD4⁺ T‐cells. (e) IFN‐γ concentration in the supernatants of lung single cell suspensions at 60 dpi determined by ELISA. Data represent the mean ± SD of nine−10 mice per group. Individual plots from experiments 2 and 3 are from two independent assays each with four−five mice per group. Statistical significance was calculated by using unpaired Student’s t‐test with a Welch correction (*P < 0·05; **P < 0·01; ***P < 0·01; ****P < 0·0001; ns, not significant).

Together, these data show that HIF‐1α deficiency in the myeloid compartment prompts the establishment of a pathological microenvironment with increased, but differential, accumulation of IFN‐γ‐ and IL‐17‐producing CD4⁺ T‐cells in the lungs of Mtb‐infected mice.

Discussion

Recent studies suggest that HIF‐1α is a central regulator of the macrophage response to mycobacterial infections.8, 9, 10 One particular study showed that mice deficient in HIF‐1α in the myeloid lineage (mHIF‐1α^−/−) were acutely susceptible to aerosol Mtb Erdman challenge.9 Contrary to these findings, in this study we showed that mHIF‐1α^−/− mice were not acutely susceptible to Mtb H37Rv strain or to the hypervirulent HN878 strain. Instead, our data showed that mHIF‐1α^−/− mice were capable of mounting normal Th1 responses, exhibited unimpaired responses to IFN‐γ and contained Mtb H37Rv growth. Additionally, mutant mice were only slightly more permissive to HN878 growth than WT mice. However, mHIF‐1α^−/− mice developed extensive lung inflammatory foci during chronic infection and died earlier than WT mice. These findings correlated both with inoculum dose and strain virulence. These data support the hypothesis that HIF‐1α activity in myeloid cells is crucial to regulate pathological pulmonary inflammation during chronic Mtb infection. The discrepancies between our data and the aforementioned study do not appear to be due to differences in the virulence and/or inflammatory profile of the Mtb strain used. Although Mtb H37Rv and Erdman strains differ in their in vivo grow rate and degree of inflammation induced,25 we had no acute mortality during infections with strain HN878. We did find an increase in mortality to H37Rv that was dose‐dependent, and an earlier mortality rate to high doses of infection with the hypervirulent HN878 strain as compared with H37Rv. In all cases, the increase in susceptibility occurred at a chronic stage of infection and correlated with extensive lung pathology.

Inflammation in mHIF‐1α^−/− mice during the chronic infection with H37Rv was characterized by increased accumulation of both IFN‐γ‐ and IL‐17‐producing CD4⁺ T‐cells as compared with infected B6 mice. While strong IFN‐γ responses can be pathological,26 we have previously shown that IL‐17 production during chronic infection promotes pathological inflammation without impacting control of the bacterial burden.27, 28 Consistent with this, we only found a slight exacerbation of the infection at the latest time‐point in one of the experiments. Whether this was caused by the increased inflammation is not known.

In summary, we propose that HIF‐1α expression in the myeloid cells induces an anti‐inflammatory response that limits an exaggerated influx of leucocytes and therefore the consequent development of more severe hypoxia. Further research is required to clarify the mechanisms by which HIF‐1α coordinates the response of myeloid cells to regulate pulmonary inflammation. Defining these mechanisms may have important implications in the design of new therapies for treatment of TB.

Disclosures

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Supporting information

Figure S1 . B6 and mHIF‐1α^−/− mice challenged with Mtb H37Rv strain (75CFU) via 2 the aerosol route. (a) Representative H&E lung sections at 25 and 40 days post‐infection. (b) 3 Percentage of the infiltrated area in the lungs at 25 and 40 days post‐infection. Individual data 4 points represent one lung section. Data represent the mean ± SD from 5 mice per group. Statistical 5 significance was calculated by using unpaired Student’s t‐test with a Welch correction (ns, not 6 significant).

Figure S2 . Representative Picro Sirius staining of lung sections from WT and 2 mHIF‐1α^−/− mice infected (H37Rv) for 120 days.

Figure S3 . Gating strategy used to identify the (a) different 1 lung myeloid cell 2 subsets and (b) lung CD3⁺ and CD4⁺ T‐cells.

Table S1 . Analysis of different cell populations in the lungs of uninfected WT and mHIF‐1α^−/− mice.