Lack of the Transcription Factor Hypoxia-Inducible Factor 1α (HIF-1α) in Macrophages Accelerates the Necrosis of Mycobacterium avium-Induced Granulomas

Abstract

The establishment of mycobacterial infection is characterized by the formation of granulomas, which are well-organized aggregates of immune cells, namely, infected macrophages. The granuloma's main function is to constrain and prevent dissemination of the mycobacteria while focusing the immune response to a limited area. In some cases these lesions can grow progressively into large granulomas which can undergo central necrosis, thereby leading to their caseation. Macrophages are the most abundant cells present in the granuloma and are known to adapt under hypoxic conditions in order to avoid cell death. Our laboratory has developed a granuloma necrosis model that mimics the human pathology of Mycobacterium tuberculosis, using C57BL/6 mice infected intravenously with a low dose of a highly virulent strain of Mycobacterium avium. In this work, a mouse strain deleted of the hypoxia inducible factor 1α (HIF-1α) under the Cre-lox system regulated by the lysozyme M gene promoter was used to determine the relevance of HIF-1α in the caseation of granulomas. The genetic ablation of HIF-1α in the myeloid lineage causes the earlier emergence of granuloma necrosis and clearly induces an impairment of the resistance against M. avium infection coincident with the emergence of necrosis. The data provide evidence that granulomas become hypoxic before undergoing necrosis through the analysis of vascularization and quantification of HIF-1α in a necrotizing mouse model. Our results show that interfering with macrophage adaptation to hypoxia, such as through HIF-1α inactivation, accelerates granuloma necrosis.

INTRODUCTION

The main pathological feature of human tuberculosis is the induction by Mycobacterium tuberculosis of granulomas which undergo central necrosis (caseous necrosis) (1). The mechanisms underlying the latter phenomenon are not clear, but it is generally accepted that caseation plays an essential role in the pathogenesis and dissemination of the infection. Caseous necrosis and cavitation during infection by M. tuberculosis is more prominent in rabbits and guinea pigs than in mice (2). Nevertheless, human-like pathology, as central granuloma necrosis and associated hypoxia, can be obtained in M. tuberculosis-infected I/StSnEgYCit, C3HeB/FeJ, or IL-13-overexpressing C57BL/6 mouse strains (3–5). Confluent necrotic granulomas also are observed in mice deficient in gamma interferon (IFN-γ) or the type 1 tumor necrosis factor (TNF) receptor (2, 6, 7). However, the fact that immunodeficient human hosts, namely, AIDS patients, tend to present forms of tuberculosis with granulomas lacking necrotic features suggests that necrosis depends on the immune response (8).

We have shown that mice infected with a small inoculum of M. avium ATCC 25291 establish a chronic and progressive infection characterized by the formation of discrete granulomas, which gradually increase in size and undergo central necrosis within a few months (9). This requires CD4⁺ T cells as well as an intact interleukin-12 (IL-12)/IFN-γ cytokine axis (9) but is independent of the generation of toxic mediators induced by IFN-γ in macrophages, such as the products of the inducible (type 2) nitric oxide synthase (iNOS) or the phagocyte NADPH oxidase (NOX2) (9). Granuloma necrosis also occurs in mice with genetic ablation of death receptors, such as type 1 TNF receptor or Fas (CD95) or the TNF superfamily members TNF or TNF-related apoptosis inducing ligand (TRAIL) (9–11). On the other hand, granuloma necrosis is not observed in certain inbred strains of mice, such as BALB/c or DBA/1. In the former strain, this was due to the increased production of the anti-inflammatory cytokine IL-10 (12). In the latter, it was related to the inability to form large granulomas due to the lack of the complement component C5 (12).

Granuloma necrosis does not occur when these lesions fail to grow to the size observed in C57BL/6 mice, such as in animals lacking T cells altogether or, more specifically, CD4⁺ T cells in animals with reduced or no IFN-γ production, as well as in the DBA/1 strain (9). Interestingly, infection of immunocompetent C57BL/6 mice with a high dose of the same M. avium strain did not induce necrotic granulomas. The explanation for this result lies in the fact that in this type of infection, severe lymphopenia is induced and granulomas do not get to be as large as those observed when T cells are spared, i.e., in low-dose infections (13).

Granulomas are believed to be poorly vascularized structures; thus, there may be difficulty in providing nutrients and oxygen to the cells, in particular to those at the core of the lesion in bigger lesions. Therefore, it is fair to hypothesize that nutrient deprivation and/or hypoxia underlie the genesis of granuloma necrosis. Given that macrophages are the most abundant cells at the center of the granuloma, these might be the cells that die. However, macrophages are known to adapt to hypoxic conditions, namely by using the glycolytic pathway to generate ATP (14). Central to these metabolic adaptations is the transcription factor hypoxia-inducible factor 1α (HIF-1α).

HIF-1α is directly involved in angiogenesis (15), erythropoiesis (16), cell growth and differentiation (17), survival, and apoptosis (18). Under normoxic conditions, HIF-1α has a very short half-life, being suppressed by hydroxylation of two prolyl residues (Pro-402 and Pro-564). This modification allows the interaction with the von Hippel-Lindau tumor suppressor protein (pVHL), the recognition component of an E3 ubiquitin ligase complex that targets HIF-1α for ubiquitination and proteasomal degradation. Under hypoxic conditions, prolyl hydroxylation is suppressed and HIF-1α escapes from proteasomal degradation, becoming stable , where it dimerizes with HIF-1β. The HIF complex formed becomes transcriptionally active, binding hypoxia response elements (HREs) (19, 20). Classical HIF-1α target genes include the vascular endothelial growth factor (21), erythropoietin (16), glucose transporters (22), and transferrin (23). The influence of HIF-1α in the innate immune system is relatively well established, and recently its role in adaptive immune responses has been recognized (24, 25).

Mice lacking HIF-1α in cells from the myeloid lineage show impairment of myeloid cell functions (25), whereas HIF-1α activation increased their activity under hypoxic conditions (26). HIF-1α is strongly stimulated upon exposure to bacteria and has the ability to regulate the production of nitric oxide (NO) and TNF-α by phagocytes (27). HIF-1α also plays a role in T cells (28), B cells (29), neutrophils (30), or dendritic cells (DCs) (31, 32) and regulates the differentiation of Th17 cells and regulatory T cells (33, 34) and the production of proinflammatory cytokines in response to T cell receptor activation (28). Nonhypoxic stimuli, such as glucose concentrations (35), lipopolysaccharide (LPS) (36), interferons (37), TNF-α (38), and nuclear factor κB (NF-κB) (39), are additional modulators of HIF-1α transcription; therefore, they are able to influence the expression of hypoxic genes.

Here, we found that the genetic ablation of HIF-1α in the myeloid lineage causes the earlier emergence of granuloma necrosis during low-dose infection by virulent M. avium. These data support the view that hypoxia is one of the causes for granuloma necrosis and that interfering with the adaptation of macrophages to this condition exacerbates this type of pathology.

MATERIALS AND METHODS

Mice, bacteria, and infection.

C57BL/6 wild-type (WT) mice were bred in our facilities from a breeding pair purchased from Harlan Iberica (Barcelona, Spain). C57BL/6 mice with a myeloid deficiency in HIF-1α (HIF-1α^−/−) were obtained in our facilities after back-crossing the C57BL/6.129-Hif1α^tm3RSjo/J and the C57BL/6.129P2-Lyz2^tm1(cre)Ifo/J strains from the Jackson Laboratory (Bar Harbor, ME, USA). The genotyping was performed according to Jackson Laboratory protocols (Bar Harbor, ME, USA). All mice were kept in our animal facilities in HEPA filter-bearing cages under 12-h light/dark cycles and fed autoclaved chow and water ad libitum. Experimental mice were age and sex matched and used between the ages of 8 and 12 weeks. The highly virulent Mycobacterium avium strain 25291 (ATCC 25291 SmT) was obtained from the American Type Culture Collection (Manassas, VA, USA). Bacterial inocula were prepared as described previously (40). Mice were infected with 10² CFU of M. avium strain 25291 through a lateral tail vein. The bacterial load in the organs was determined as previously described (40). All animal experiments were performed in accordance with the recommendations of the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (ETS 123) and 86/609/EEC Directive and Portuguese rules (DL 129/92). The animal experimental protocol was approved by the national authority, Direção Geral de Veterinária.

Liver mononuclear cell isolation.

Liver samples were collected in Dulbecco's modified Eagle's tissue culture medium (DMEM; Life Technologies, Paisley, United Kingdom) containing 10% fetal bovine serum (FBS; Life Technologies), 5% glutamine, and 1% penicillin-streptomycin (P/S) and processed individually. A single-cell suspension was obtained by passing the liver through a 70-μm cell strainer (BD Biosciences, San Jose, CA, USA). Cells were washed with phosphate-buffered saline (PBS) until the supernatant became clear, and then they were resuspended in DMEM and subjected to density centrifugation using Histopaque 1083 (Sigma-Aldrich, St. Louis, MO, USA). After centrifugation at 1,000 × g for 25 min at room temperature (RT) without acceleration or brake, liver mononuclear cells were recovered, washed, and resuspended in DMEM, and the number of viable cells was counted by trypan blue exclusion using a hemocytometer.

HIF-1α quantification.

Liver mononuclear cells were fractionated into cytoplasmic and nuclear protein extracts using a cell fractionation kit according to the manufacturer's instructions (BioVision, Mountain View, CA, USA). Total HIF-1α was quantified using the cytoplasmic extracts by a two-site enzyme-linked immunosorbent assay (ELISA) according to the manufacturer's instructions (R&D Systems, Minneapolis, MN, USA). Positive controls were provided by the manufacturer. All samples were assayed in duplicate. The values were normalized to the total liver mononuclear cell number isolated from each mouse.

Flow cytometry.

For the immunofluorescence staining, 10⁶ cells were labeled with distinct combinations of the following antibodies (Abs): fluorescein isothiocyanate (FITC)-conjugated anti-CD19, phycoerythrin (PE)-conjugated anti-CD3, V450-conjugated anti-CD4, V500-conjugated anti-CD8, brilliant violet 510 (BV 510)-conjugated anti-CD11b, BV 421-conjugated anti-CD11c, allophycocyanin (APC)-conjugated anti-Ly6G, and APC with cyanin-7 (APC/Cy7)-conjugated anti-F4/80. All of the Abs were obtained from BioLegend (San Diego, CA, USA) except for V450-CD4 and V500-CD8, which were obtained from BD Biosciences (San Jose, CA, USA). The acquisition of cells was performed using a FACSCanto II flow cytometer using BD FACSDiva software (BD Biosciences). Data were analyzed using FlowJo software (Tree Star, Ashland, OR). Cells were selected on the basis of forward scatter/side scatter (FSC/SSC), and the singlets were gated according to size versus width. For the lymphocyte analysis we used CD19 as a marker of B cells and CD3, CD4, and CD8 to label T cells. CD11b⁺ Ly6G^high were defined as neutrophils. CD11b⁺ Ly6G⁻ F4/80⁺ and CD11b⁺/Ly6G⁻/CD11c^high were defined as macrophages and DCs, respectively. To determine the cell number, the number of gated events for each cell population was multiplied by the total cell number (counted using Kova chambers) and divided by the total number of events selected by FSC/SSC parameters.

Histology and morphometric analysis of the granulomatous area.

Portions of the livers were fixed in buffered formaldehyde and embedded in paraffin. Sections (4 μm) were stained with hematoxylin and eosin (H&E). Slides were photographed using an Olympus CX31 light microscope equipped with a DP-25 camera (Imaging Software CellB̂; Olympus, Center Valley, PA, USA). One liver section per animal with random fields selected in a blind way was analyzed. The determination of the granulomatous infiltration of necrotic and nonnecrotic areas was done in a total tissue area ranging from 6 × 10⁶ to 9 × 10⁶ μm², corresponding to six fields analyzed in each section. To determine the liver area covered by granulomas, the NIH ImageJ software program was used. The percentage of granuloma area was calculated for each mouse by dividing the sum of granulomatous areas by the total area of the liver section analyzed. The number of cellular infiltrates was the sum of all granulomatous/cell infiltration areas.

Immunohistochemistry.

Paraffin sections were dewaxed, rehydrated, and permeabilized in PBS containing 0.1% Triton X-100 and 0.1% Tween. After antigen retrieval with 10 mM sodium citrate buffer for 30 min at 96°C, endogenous peroxidase activity was blocked with 0.3% hydrogen peroxidase in methanol for 35 min at RT, followed by blocking the nonspecific Ab binding with normal horse serum (Vector Laboratories, Burlingame, CA, USA). Sections were incubated overnight at 4°C with rat IgG monoclonal anti-mouse F4/80 (clone Cl:A3-1; Abcam) or rat IgG monoclonal anti-mouse endomucin (clone V.7C/.1; Abcam). After washing with PBS, sections were incubated with goat anti-rat IgG and horseradish peroxidase-conjugated Ab (GE Healthcare). Development was performed with a 3,3-diaminobenzidine (DAB) labeling system (Vector Laboratories). Sections then were counterstained with Gill's hematoxylin, dehydrated, and mounted in DPX (Sigma-Aldrich).

Immunofluorescence.

Paraffin sections were dewaxed, rehydrated, and permeabilized in PBS containing 0.1% Triton X-100 and 0.1% Tween. After antigen retrieval with 10 mM sodium citrate buffer for 30 min at 96°C, nonspecific Ab binding was blocked with 5% bovine serum albumin (BSA) in PBS, 0.1% Triton X-100, 0.1% Tween. Sections were incubated for 3 h at RT with rabbit IgG monoclonal anti-human/mouse active caspase 3 (R&D Systems). After washing with PBS, sections were incubated with goat anti-rabbit IgG and Alexa Fluor 488-conjugated Ab (Life Technologies). Sections then were counterstained and mounted with 4′,6-diamidino-2-phenylindole (DAPI)-Vectashield and were observed using a Zeiss Imager Z1 microscope.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay.

Paraffin sections were dewaxed, rehydrated, permeabilized in 0.1% Triton X-100, 0.1% sodium citrate for 8 min at RT, and washed twice with PBS. Nicked DNA was labeled by incubating the cells with the label solution containing fluorescein-2′-deoxyuridine 5′-triphosphate (dUTP) and terminal deoxynucleotidyl transferase (TdT) (Roche Diagnostics, Penzberg, Germany) for 1 h at 37°C in a humidified atmosphere in the dark. Negative and positive controls were performed by incubating the slides without TdT and with recombinant DNase I (Roche) in 50 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 1 mg/ml BSA for 10 min at RT, respectively. Sections were counterstained and mounted with DAPI-Vectashield and were observed using a Zeiss Imager Z1 microscope.

In vitro stimulation of spleen cells and IFN-γ quantification.

Spleen cell suspensions were processed individually as described previously (10). Cells were cultivated at a density of 2 × 10⁵ cells/well in a U-bottom 96-well microtiter plate and incubated in triplicate in DMEM with 10% fetal calf serum (FCS) with no stimulus or were stimulated with mycobacterial envelope proteins (4 μg/ml). Supernatants from the cultures were collected after 72 h of culture at 37°C in a 7% CO₂ incubator, and the IFN-γ produced was quantified by the ELISA method using anti-IFN-γ-specific affinity-purified MAb (R4-6A2 as capture Ab and biotinylated AN-18 as detecting Ab). A standard curve was generated with known amounts of recombinant murine IFN-γ (Genzyme, Cambridge, CA, USA). The optical densities (OD) were recorded at 450 nm.

Infection of bone marrow macrophages and treatments.

Macrophages from WT and HIF-1α^−/− mice were obtained as described previously (41). Recombinant murine IFN-γ (Gibco), 100 U per culture well, and recombinant murine TNF-α (Genzyme), 50 U per culture well, were added daily to the cultures, starting immediately after infection and until day 7.

Statistics.

Results were expressed as means ± standard deviations (SD). Statistical significance was calculated by using the unpaired Student t test for data presented in Fig. 3 and 4 and the one-way analysis of variance (ANOVA) test with a Tukey's posttest for data presented in Fig. 5. P < 0.05 was considered statistically significant.

FIG 3.

HIF-1α accumulates and its absence increases the susceptibility to M. avium 25291 infection. (A) Analysis of total HIF-1α protein in liver mononuclear cells during M. avium infection. Groups of noninfected mice (day 0) were included at every time point of infection studied. Each group of noninfected mice was comprised of three to four animals, and each group of infected mice was comprised of four to six animals. Data are expressed as mean protein levels ± SD calculated from the total liver mononuclear cells for individual mice. P < 0.01 (**) and P < 0.001 (***) for comparisons of values from infected and noninfected animals. (B and C) Representative photographs of spleens (B) and livers (C) from the infected mice at day 96 postinfection. (D) Representative kinetics of M. avium 25291 infection of the liver and spleen from WT and HIF-1α^−/− mice. Data represent the mean CFU and SD from five mice per group of one of three experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (statistically significant differences between WT and HIF-1α^−/− mice).

FIG 4.

Myeloid ablation of HIF-1α influences granuloma progression during M. avium infection. (A, C, and E) Representative lesions in liver sections from infected WT mice. (B, D, and F) Representative lesions in liver sections from infected HIF-1α^−/− mice. Liver sections are from 30 days (A and B), 60 days (C and D), and 104 (E and F) days postinfection. (G) Percentage of infiltrated area for WT and HIF-1α^−/− mice. ***, P < 0.001 for total percent infiltrated area of WT versus that of HIF-1α^−/− mice. +, P < 0.05 for comparisons of necrotic to nonnecrotic areas of HIF-1α^−/− mice. (H) Number of cellular infiltrates for WT and HIF-1α^−/− mice. Data are expressed as mean areas ± SD from mice analyzed individually in each group. ***, P < 0.001 for comparisons of WT and HIF-1α^−/− mice.

FIG 5.

Evidence of apoptotic cell death within the granulomas of HIF-1α^−/− mice. A representative lesion from an HIF-1α^−/− mouse infected for 104 days is shown. Tissue sections of liver tissue were analyzed by performing the immunodetection of cleaved caspase-3 and the TUNEL assay (green fluorescence). Nuclei were stained blue by DAPI. Panels C and D are magnifications of the insets shown in panels A and B, respectively.

RESULTS

Evolution of the granuloma induced by M. avium.

WT mice were infected with a low dose of M. avium, and liver histological sections were analyzed at different time points of infection. Liver was selected, as it is the organ where most of the inoculum is implanted following intravenous infection and where the delineation of the granuloma from the surrounding tissue is easier. To assess the cellular constitution of the granuloma, we stained macrophages by immunohistochemistry (IHC) using F4/80-specific antibodies (42). As shown in Fig. 1, granulomas from day 30 to day 120 postinfection were mostly constituted by macrophages forming granulomas of progressively larger sizes. Caseating granulomas (Fig. 1D) had the necrotic center surrounded by F4/80-positive cells.

FIG 1.

Immunohistochemical staining of F4/80-positive cells in liver sections from WT mice infected with M. avium 25291 for 30 days (A), 60 days (B), 90 days (C), and 120 days (D).

It has been described that caseating granulomas show signs of hypoxia in the necrotic areas (43), raising the possibility that hypoxia is common in the granulomatous environment. In fact, granulomas are considered to be poorly vascularized structures, and the limited blood supply could cause a reduction in nutrient and oxygen supply at the core of the granuloma (43). Alternatively, hypoxia could simply reflect the destruction of the integrity of the tissue and be a consequence rather than a cause of the granuloma necrosis. Thus, we analyzed vascularization during granuloma development by the immunohistochemical detection of endomucin (44). As shown in Fig. 2, small granulomas show an extensive vascular bed and preservation of cellular integrity (Fig. 2A), but as they increase in size, signs of cellular decay with accumulation of neutrophils can be seen in areas with a smaller density of blood vessels (Fig. 2B). At advanced stages, granulomas show large areas of necrosis and no evidence of vascularization except at the periphery of the granuloma, where cells still are intact (Fig. 2C and D). These data indicate that macrophages at the center of granulomas with poor vascularization can die due to reduced access to nutrients and/or oxygen.

FIG 2.

Immunohistochemical staining of endomucin-positive cells in liver sections from WT mice infected with M. avium 25291 for 90 days (A) and 120 days (B to D).

Evidence for the development of hypoxia in the tissues of infected mice.

Necrotic areas in M. tuberculosis or M. avium granulomas are hypoxic (43, 45, 46). Our observations showed that small nonnecrotic granulomas were well vascularized structures, but as granulomas enlarged the central core became less vascularized. This could be responsible for a decrease in oxygen supply and lead to a hypoxic environment. Since HIF-1α is an important regulator of hypoxia (19, 20), we evaluated HIF-1α protein content in liver mononuclear cells during M. avium infection. We found a significant increase of total HIF-1α protein levels in cells from infected animals compared to those of the noninfected animals beginning at day 60 postinfection (Fig. 3A). These results indicate HIF-1α accumulation at the onset of granuloma necrosis (60 and 90 days postinfection).

Consequences of a myeloid HIF-1α deficiency.

The stabilization of HIF-1α and the consequent activation of a targeted transcriptional program allow cells to adapt to hypoxia (20). In the absence of HIF-1α, it is expected that cells fair less well in hypoxic environments and die. In a preliminary assessment of the impact of myeloid deficiency in HIF-1α on the course of M. avium infection in vivo, we infected HIF-1α^−/− and WT mice with a low dose (100 CFU) of M. avium 25291. Mice were sacrificed at different time points postinfection and their organs analyzed. Infected mutant mice exhibited increased organ sizes and macroscopic tubercles at 96 days of infection (Fig. 3B and C). The bacterial burdens in the spleen and liver also were increased in mutant compared to WT mice (Fig. 3D). Importantly, the exacerbation of bacterial growth occurred late in infection with no differences in the replication of M. avium during the first month of infection in the spleen and in the first 2 months of infection in the liver.

Impact of myeloid HIF-1α deficiency on granuloma integrity and macrophage apoptosis/necrosis.

We studied the evolution of the granulomas in both strains by analyzing H&E-stained liver sections. At day 30 postinfection, lesions of both HIF-1α^−/− and WT mice were very small and incipient (Fig. 4A and D). After day 60 postinfection, differences between the lesions in either strain were evident, showing initial necrosis of granulomas in HIF-1α^−/− mice and smaller granulomas still lacking signs of necrosis in WT animals (Fig. 4B and E). At day 104 of infection, extensive granuloma necrosis was found in all HIF-1α^−/− mice analyzed and in none of the WT animals analyzed. We quantified the type of pathology in the liver by distinguishing necrotic from nonnecrotic areas within the infiltrates. Morphometric analysis determining the infiltrated necrotic or nonnecrotic areas confirmed the importance of HIF-1α in the evolution of the granuloma. No differences in liver inflammatory infiltrates were found at day 30 postinfection between the two mouse strains. At day 60 postinfection, a slight albeit not statistically significant increase in inflammatory area was observed in HIF-1α^−/− mice compared with that of WT mice. On day 104 postinfection, the organs of HIF-1α^−/− mice showed a marked and statistically significant increase in inflammatory areas (Fig. 4G) and number of lesions (Fig. 4H) compared to levels for WT mice. More important, a large fraction of the inflammatory lesions in HIF-1α^−/− mice was constituted by necrotic material, whereas no necrosis was found yet in WT mice. The development of necrosis in liver granulomas was coincident with the increase in bacterial numbers (Fig. 3D). These results show that the lack of HIF-1α in myeloid cells anticipates the emergence of necrosis, augments the extent of inflammation, and increases the susceptibility to infection. In order to determine the molecular mechanism underlying granuloma necrosis in the absence of HIF-1α, we performed the TUNEL assay and detected caspase-3 activation by immunofluorescence in liver sections of organs from HIF-1α-deficient animals presenting necrotic lesions. On day 104 postinfection, the granulomas of HIF-1α^−/− mice showed cleaved caspase-3-positive cells and marked internucleosomal DNA fragmentation at the center of the lesions, demonstrating that cell demise within a granuloma occurs by apoptosis followed by secondary necrosis (Fig. 5). Granuloma evolution, vascularization, and bacterial load also were evaluated in the lungs. The histological observation of the organ at the longest time points showed small lesions without signs of necrosis, comparable granuloma vascularization, and no differences in bacterial burden in either strain of mouse. Further morphometric analysis indicated no differences in infiltrated area or in the number of lung inflammatory infiltrates from HIF-1α^−/− and WT mice (see Fig. S1 in the supplemental material).

Immune response in the absence of myeloid HIF-1α.

Given the involvement of HIF-1α in the adaptive immune response, we assessed cellular responses in the liver as well as in the spleen. The number of macrophages, DCs, and neutrophils in the spleen remained low until day 60 of infection in both strains but increased by day 104 of infection with significantly higher numbers of these cells in mutant mice than in the WT animals (Fig. 6A). Similar findings were obtained for macrophages and DCs in the liver (Fig. 6B). However, neutrophils could not be quantified given the isolation procedure employed, which does not allow the recovery of this cell type. The analysis of spleen lymphocyte populations revealed no significant differences at day 60 postinfection, as previously shown (9). However, at the late time point of infection analyzed, a significant increase of CD4⁺ and CD8⁺ cell populations was observed in HIF-1α^−/− mice compared to levels in WT mice (Fig. 6A). A significant increase in the number of splenic CD19⁺ cells was found in both infected mouse strains compared to levels in the respective noninfected strains (Fig. 6A). The analysis of liver lymphocyte populations indicated no differences for CD4⁺ or CD8⁺ lymphocytes at the time points analyzed (Fig. 6B). A significant increase in CD19⁺ cells in the liver was observed in infected WT mice compared to the level for HIF-1α^−/− mice, suggesting a higher accumulation of B cells in the liver in the presence of HIF-1α (Fig. 6B).

FIG 6.

M. avium-infected HIF-1α^−/− mice present increased numbers of mononuclear cells at late time points of infection. Cellular composition of spleens (A) and livers (B) from WT and HIF-1α^−/− mice. Groups of noninfected mice (day 0) were included at every time point of infection studied. Each group was comprised of 5 animals. All data are expressed as means ± SD from one of three experiments. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***) for comparisons of infected WT and HIF-1α^−/− mice. P < 0.05 (+), P < 0.01 (++), and P < 0.001 (+++) for comparisons of infected versus noninfected (day 0) animals.

Spleen cells from mutant mice infected for 99 days secreted amounts of IFN-γ similar to those of cells from wild-type animals (26.9 ± 3.8 pg/ml versus 27.3 ± 1.6 pg/ml, respectively) in response to M. avium antigen. Thus, T cells appear to respond similarly to M. avium in either strain.

Given that the physiology of the macrophage itself might be affected by the absence of HIF-1α, we performed in vitro infections of bone marrow-derived macrophages that were activated or not activated by a combination of IFN-γ plus TNF-α. Bacterial proliferation in nonactivated macrophages was similar in mutant and WT cells (1.08 versus 1.20 log₁₀ increase in CFU in 7 days, respectively). Activation of macrophages with cytokines led to a similar reduction in M. avium growth in either mutant or WT cells (0.70 versus 0.83 log₁₀ increase in CFU in 7 days, respectively). Nevertheless, macrophages from mutant mice appeared to control the mycobacteria as well as WT cells, at least under normoxic conditions (data not shown).

DISCUSSION

Much has already been learnt about the cellular and molecular mechanisms underlying the structuring of the mycobacterial granuloma, but the mechanisms specifically leading to granuloma necrosis still are poorly understood. We have shown in a mouse model of M. avium infection that this process is highly dependent on CD4⁺ T cells, IFN-γ, and IL-12 and partly on IL-6 and CD40 (9, 47). Recently, it has been described that necrotic granulomas developed in M. tuberculosis-infected guinea pigs, rabbits, and nonhuman primates (45), as well as in M. avium (strain TMC724)-infected WT mice, are hypoxic (43). In this work, we tested the hypothesis that hypoxia underlies the origin of granuloma necrosis. We have used WT mice infected with M. avium 25291 to study the vascularization of the granuloma. The results obtained showed that small- and medium-sized granulomas were well vascularized, but in advanced lesions blood vessels were found only at the periphery of the granulomas, as already published (48). Therefore, it is possible that the centers of large granulomas become hypoxic due to this reduction in blood supply, leading to the death of the cells further away from the blood vessels, i.e., macrophages within the deep core of the granuloma. However, it is known that macrophages can adapt to low-oxygen tensions found in inflammatory sites (49, 50). This adaptation, as in most cells, involves the regulation of metabolic pathways controlled by the HIF system. To clarify a possible role of hypoxia in the development of necrotic granulomas in mycobacterial infections, we studied the influence of HIF-1α in the development of necrosis of M. avium-induced granulomas. Total HIF-1α protein increased in liver mononuclear cells at 60 days of infection, a time point where reduced vascularization of the granulomas was observed. Mice defective in myeloid HIF-1α developed earlier granuloma necrosis than WT animals, suggesting that HIF-1α is required for cell survival within the granuloma and that those cells at the core of these lesions are subjected to severe hypoxia. The cell death involved caspase-3 activation and internucleosomal DNA nicking, suggesting that necrosis is secondary to apoptosis. The accelerated granuloma necrosis preceded an increase of the bacterial growth, suggesting that mycobacterial control is not affected by the lack of HIF-1α as long as the cells remain viable. This interpretation is supported by the fact that (i) HIF-1α-deficient mice are as resistant to a low-virulence M. avium strain as the WT mice (unpublished observations); (ii) no phenotype was observed in HIF-1α-deficient mice infected with a high dose of M. avium 25291, where no necrosis of granulomas ever occurs (unpublished data); and (iii) no major defects in the adaptive immune response were found, except for the increase in the accumulation of phagocytes, likely a consequence of the tissue damage and consequent exacerbation of the inflammation. Therefore, we propose that the increase in bacterial loads at the late time points of infection is the result of the death of the host cells in granulomas undergoing necrosis. Although factors other than hypoxia may regulate the expression of this transcription factor either at the transcriptional or posttranslational levels, such as IFN-γ and LPS (51), we found similar expression of IFN-γ (52, 53) and only minor changes in T cell populations. HIF-1α is involved in the regulation of transcription in activated macrophages, namely, those of iNOS (51). Although NO does not participate in the IFN-γ-induced control of M. avium growth, HIF-1α could be required for the expression of additional antimicrobial systems. In fact, even under normoxia, HIF-1α is induced by bacterial infection and regulated by TNF-α and IFN-γ (15, 27, 38). However, we detected no macrophage defect in controlling M. avium proliferation in vitro even upon cytokine activation of the cells. This study failed to provide evidence of necrotic granulomas in the lung at the time points analyzed. This likely was due to the small size of the lesions in an organ with high oxygen tension. The formation of necrotic granulomas in the lung was achieved by Ehlers and colleagues by using a massive inoculum of a similar M. avium strain (10⁵ CFU) following an aerosol inhalation and a long-term infection (19 weeks) (54). Further work is being undertaken, studying longer time points of infection with bigger inocula, in order to assess the role of granuloma volume and confluence and consequent hypoxia in the development of lung pathology.

Our data are consistent with the hypothesis that growing granulomas become progressively less vascularized, more hypoxic, and eventually reach a state where the macrophages at the inner core of the lesion no longer survive the lack of oxygen and die. Such a sequence of events was accelerated here by depriving macrophages of the major adaptive system against hypoxia. As a consequence, cells died recruiting more phagocytes and bacteria found the appropriate ground for replication. The latter may not necessarily happen in an intact host where the oxygen tension reached within the necrotic granulomas would be expected to be much lower and likely restrictive of mycobacterial growth. We propose that interfering with adaptation to hypoxia or reducing the growth of granulomas will provide ways to reduce pathology and prevent dissemination in mycobacterial infections.