Role of Granulocyte-Macrophage Colony-Stimulating Factor Production by T Cells during Mycobacterium tuberculosis Infection

Abstract

Mice deficient for granulocyte-macrophage colony-stimulating factor (GM-CSF^-/-) are highly susceptible to infection with Mycobacterium tuberculosis, and clinical data have shown that anti-GM-CSF neutralizing antibodies can lead to increased susceptibility to tuberculosis in otherwise healthy people. GM-CSF activates human and murine macrophages to inhibit intracellular M. tuberculosis growth. We have previously shown that GM-CSF produced by iNKT cells inhibits growth of M. tuberculosis However, the more general role of T cell-derived GM-CSF during infection has not been defined and how GM-CSF activates macrophages to inhibit bacterial growth is unknown. Here we demonstrate that, in addition to nonconventional T cells, conventional T cells also produce GM-CSF during M. tuberculosis infection. Early during infection, nonconventional iNKT cells and γδ T cells are the main source of GM-CSF, a role subsequently assumed by conventional CD4⁺ T cells as the infection progresses. M. tuberculosis-specific T cells producing GM-CSF are also detected in the peripheral blood of infected people. Under conditions where nonhematopoietic production of GM-CSF is deficient, T cell production of GM-CSF is protective and required for control of M. tuberculosis infection. However, GM-CSF is not required for T cell-mediated protection in settings where GM-CSF is produced by other cell types. Finally, using an in vitro macrophage infection model, we demonstrate that GM-CSF inhibition of M. tuberculosis growth requires the expression of peroxisome proliferator-activated receptor gamma (PPARγ). Thus, we identified GM-CSF production as a novel T cell effector function. These findings suggest that a strategy augmenting T cell production of GM-CSF could enhance host resistance against M. tuberculosisIMPORTANCEMycobacterium tuberculosis is the bacterium that causes tuberculosis, the leading cause of death by any infection worldwide. T cells are critical components of the immune response to Mycobacterium tuberculosis While gamma interferon (IFN-γ) is a key effector function of T cells during infection, a failed phase IIb clinical trial and other studies have revealed that IFN-γ production alone is not sufficient to control M. tuberculosis In this study, we demonstrate that CD4⁺, CD8⁺, and nonconventional T cells produce GM-CSF during Mycobacterium tuberculosis infection in mice and in the peripheral blood of infected humans. Under conditions where other sources of GM-CSF are absent, T cell production of GM-CSF is protective and is required for control of infection. GM-CSF activation of macrophages to limit bacterial growth requires host expression of the transcription factor PPARγ. The identification of GM-CSF production as a T cell effector function may inform future host-directed therapy or vaccine designs.

Keywords: GM-CSF; Mycobacterium tuberculosis; T cells; cytokines; lung infection; macrophages.

FIG 1

GM-CSF production in the lung increases over the course of M. tuberculosis infection. GM-CSF was measured in a Bioplex immunoassay in lung homogenates at certain weeks post-aerosol infection with the Erdman strain in WT C57BL/6J mice. GM-CSF and IFN-γ protein levels were normalized per lung. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (compared to previous time point). #, P < 0.05 for GM-CSF versus IFN-γ.

FIG 2

GM-CSF production by either radioresistant or radiosensitive cells promotes mycobacterial control. (A and C) Reciprocal radiation chimeric mice (donor BM → recipient) were allowed to reconstitute for 8 weeks and then challenged with aerosolized M. tuberculosis, and lung (A) and spleen (C) CFU were determined after 4 weeks. Each group contained 4 to 5 mice, and data were combined from 2 independent experiments (results of the first experiment are shown as open symbols, and those from the second experiment are shown as closed symbols). A one-way ANOVA was used. *, P < 0.05. See Table S1 for statistical analysis of each experiment’s data, analyzed separately and also combined. (B and D) A third experiment was done to compare the how WT → KO and KO → KO BM chimeras controlled M. tuberculosis infection in the lungs (B) and spleen (D). The groups contained 7 and 5 mice, respectively. Bars represent means ± SEM. Analysis was performed using an unpaired t test. *, P < 0.05 (WT versus C57BL/6 mice and KO versus GM-CSF^−/− mice).

FIG 3

Multiple T cell subsets produce GM-CSF in the lung during M. tuberculosis infection. (A) Representative flow cytometry plots after gating of iNKT cells, γδ T cells, and CD4⁺ and CD8⁺ T cells. Controls for CD1d tetramer and GM-CSF and IFN-γ ICS staining are included. (B and C) Absolute cell numbers (B) and the relative frequency of GM-CSF⁺ T cells in C57BL/6J mice at various time points after aerosol infection with M. tuberculosis Erdman (C). (D and E) Absolute cell numbers (D) and relative frequencies of IFN-γ⁺ T cells (E). (F) Absolute numbers of GM-CSF⁺ IFN-γ⁺ T cells. Each point represents the result for an individual mouse. Lung cells were cultured in brefeldin A without additional stimulation, and then intracellular cytokine staining was performed. iNKT cells (TCR-β⁺ CD1d tetramer⁺), γδ T cells (CD3⁺ TCR-β⁻ TCR-γδ⁺), CD4⁺ (TCR-β⁻ CD4⁺ CD8⁻), and CD8⁺ (TCR-β⁻ CD4⁻ CD8⁺) were evaluated. Data were compiled from 2 or 3 independent experiments for each time point.

FIG 4

T cell-derived GM-CSF controls M. tuberculosis growth. (A) Experimental strategy for transfer of donor WT or GM-CSF^−/− T cells into RAG^−/− recipient mice, followed by M. tuberculosis aerosol infection. (B) CFU in lungs 4 weeks postinfection after WT or GM-CSF^−/− T cells were transferred into RAG^−/− recipients. Data are representative of five independent experiments. (C) Percentage of CD4⁺ T cells producing IFN-γ or GM-CSF after 5 h of ESAT6_1-20 peptide stimulation or anti-CD3/anti-CD28 stimulation with brefeldin A and IL-2 at 4 weeks postinfection. (D) Experimental strategy for adoptive transfer experiments using sublethally irradiated GM-CSF^−/− mice as recipients, who were then infected with M. tuberculosis via aerosol. (E) CFU in lungs 4 weeks postinfection after WT or GM-CSF^−/− CD4⁺ T cells were transferred into GM-CSF^−/− recipients. Data were compiled from 4 independent infections with a total of n = 16 to 19 mice per condition. (F) Frequency of ESAT6_3-17 tetramer-positive CD4⁺ T cells from donor WT and GM-CSF^−/− infected mice 4 weeks after sublethal irradiation, adoptive transfer, and aerosol infection. (G) Experimental strategy for adoptive cotransfer of CD4⁺ T cells from donor WT and GM-CSF^−/− Thy1.1 recipients 4 weeks after infection (at a 1:1 ratio) with M. tuberculosis infection via aerosol. (H) Origin of CD4⁺ T cells based on congenic markers 4 weeks after infection. (I and J) Frequency of ESAT6_3-17-positive (I) or Ag85b_241-256 tetramer-positive (J) CD4⁺ T cells among total host (e.g., endogenous), WT, or KO CD4⁺ T cells. Statistical testing was performed by using a paired 1-way ANOVA. *, P < 0.05; **, P < 0.01; ****, P < 0.0001. Error bars indicate SEM.

FIG 5

GM-CSF production by human T cells. (A) Representative flow cytometric analysis results for GM-CSF and IFN-γ production by unstimulated human peripheral blood CD4⁺ and CD4⁻ T cells, or after stimulation with M. tuberculosis lysate (Mtb lysate), or after treatment with PMA plus ionomycin (P + I). (B and C) The frequency of CD4⁺ (B) or CD4⁻ (C) T cells producing GM-CSF only, both GM-CSF and IFN-γ, or IFN-γ only (left), or the total GM-CSF or total IFN-γ (right), from healthy controls (n = 9) or active TB patients (n = 9) stimulated with M. tuberculosis lysate. Red circles, HC; blue circles, TB patients. Lines indicate means ± standard deviations. Statistical testing was performed with a two-way ANOVA. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

FIG 6

GM-CSF and IFN-γ have an additive antimicrobial effect. Mycobacterial growth inhibition was performed using H37Rv-infected C57BL/6 macrophages, which were infected overnight as described elsewhere (23). CFU were measured on day 1 (macrophages [mφ] alone, baseline) and on day 5 postinfection. Recombinant GM-CSF and/or IFN-γ was added on day 1 postinfection. Error bars indicate means ± SEM. **, P < 0.01. One-way ANOVA with Dunnett’s posttest was used to compare combination treatment with individual cytokine treatment at each concentration. Data are representative of two independent experiments.

FIG 7

GM-CSF antimicrobial activity requires PPARγ expression in macrophages. (A) Mycobacterial growth inhibition was performed using H37Rv-infected macrophages from C57BL/6 mice (e.g., WT) (left) or PPARγ^fl/fl; LysM-cre mice (right). Baseline CFU were measured on day 1 in the absence of GM-CSF. Recombinant cytokines, including GM-CSF (0.01 to 10 ng/ml), were added on day 1, and bacterial growth was determined on day 5. Recombinant IFN-γ (10 U/ml) and TNF (10 ng/ml) were used as positive controls. (B) The percent CFU reduction compiled from 4 independent experiments. Error bars indicate means ± SEM. **, P < 0.01; ***, P < 0.001. One-way ANOVA with Dunnett’s posttest was used to compare day 5 untreated macrophages for the data in panel A; multiple Student’s t tests were used for the data in panel B. Data are representative of four independent experiments with three replicates each. mφ, macrophages.