Invariant Natural Killer T-cell Dynamics in Human Immunodeficiency Virus-associated Tuberculosis

Abstract

Background: Tuberculosis (TB) is the leading cause of mortality and morbidity in people living with human immunodeficiency virus (HIV) infection (PLWH). PLWH with TB disease are at risk of the paradoxical TB-associated immune reconstitution inflammatory syndrome (TB-IRIS) when they commence antiretroviral therapy. However, the pathophysiology is incompletely understood and specific therapy is lacking. We investigated the hypothesis that invariant natural killer T (iNKT) cells contribute to innate immune dysfunction associated with TB-IRIS.

Methods: In a cross-sectional study of 101 PLWH and HIV-uninfected South African patients with active TB and controls, iNKT cells were enumerated using α-galactosylceramide-loaded CD1d tetramers and subsequently functionally characterized by flow cytometry. In a second study of 49 people with HIV type 1 (HIV-1) and active TB commencing antiretroviral therapy, iNKT cells in TB-IRIS patients and non-IRIS controls were compared longitudinally.

Results: Circulating iNKT cells were reduced in HIV-1 infection, most significantly the CD4+ subset, which was inversely associated with HIV-1 viral load. iNKT cells in HIV-associated TB had increased surface CD107a expression, indicating cytotoxic degranulation. Relatively increased iNKT cell frequency in patients with HIV-1 infection and active TB was associated with development of TB-IRIS following antiretroviral therapy initiation. iNKT cells in TB-IRIS were CD4+CD8- subset depleted and degranulated around the time of TB-IRIS onset.

Conclusions: Reduced iNKT cell CD4+ subsets as a result of HIV-1 infection may skew iNKT cell functionality toward cytotoxicity. Increased CD4- cytotoxic iNKT cells may contribute to immunopathology in TB-IRIS.

Keywords: HIV; innate; invariant natural killer T cell; paradoxical immune reconstitution inflammatory syndrome; tuberculosis.

Figure 1.

Reduced invariant natural killer T (iNKT) cells in human immunodeficiency virus type 1 (HIV-1) infection and active tuberculosis (TB). iNKT cells were enumerated by flow cytometry using ⍺-galcer–loaded CD1d tetramers. Each sample was stained in parallel with a control tetramer (without ⍺-galcer) to identify nonspecific tetramer binding for subtraction. A, The gating strategy demonstrates an iNKT cell frequency of 0.77%, with no control tetramer binding, equivalent to 7700 cells per million CD3⁺CD19^– live lymphocytes. B, Decreased iNKT cell frequency was found in people living with HIV (PLWH) with active TB (HIV⁺TB⁺) and without active TB (HIV⁺TB^–), compared to HIV-uninfected patients without active TB (HIV^–TB^–). Similarly, in PLWH with and without active TB, decreased iNKT cell numbers (cells/mL peripheral blood) (C) were found compared to HIV^–TB^– patients. Additionally, in HIV-uninfected patients with active TB (HIV^–TB⁺), iNKT cell numbers were reduced compared to HIV^–TB^– patients. Analysis was by Kruskal-Wallis test with Dunn multiple comparisons test to calculate adjusted P values: *P < .05; **P < .01; ***P < .001. In (B) and (C), zero values were replaced by 1 for representation on a log scale. Abbreviations: α-galcer, α-galactosylceramide; FSC-A, forward scatter area; FSC-H, foward scatter height; SSC-A, side scatter area; VIVID, Violet LIVE/DEAD Fixable stain.

Figure 2.

CD4⁺ invariant natural killer T (iNKT) cell subset depletion in human immunodeficiency virus type 1 (HIV-1)–associated tuberculosis (TB). People living with HIV (PLWH), most significantly those with active TB, had depleted CD4⁺ iNKT cells as measured by percentage of total iNKT cell count (A) and frequency per million CD3⁺CD19^– live lymphocytes (B). In PLWH, peripheral blood CD4 T-cell count positively correlated with CD4⁺ iNKT cell percentage (C). HIV-1 viral load negatively correlated with CD4⁺ iNKT cell percentage (D). In HIV-uninfected patients without active TB, iNKT cells were mostly either CD4⁺CD8^– or double negative (CD4^–CD8^–), whereas in PLWH, CD4⁺CD8^– iNKT cells were depleted and double-negative iNKT cells were the predominant subset (E and F). Analysis was by Kruskal-Wallis test with Dunn multiple comparisons test to calculate multiplicity-adjusted P values: *P < .05; **P < .01; ***P < .001 or by Spearman correlation (C and D).

Figure 3.

Invariant natural killer T (iNKT) cell cytotoxicity in human immunodeficiency virus type 1 (HIV-1)–associated tuberculosis (TB). In HIV-associated TB, there were increased percentages of CD107a⁺ iNKT cells, suggestive of cytotoxic degranulation (A). In people living with HIV (PLWH) with clinical features of extrapulmonary TB (EPTB), there were increased CD107a⁺ iNKT cell percentages compared to PLWH with pulmonary TB (B). Analysis was by Kruskal-Wallis test with Dunn multiple comparisons test to calculate multiplicity-adjusted P values or by Mann-Whitney U test in (B): * P < .05.

Figure 4.

Invariant natural killer T (iNKT) cells are elevated in patients with tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) and are CD4⁺CD8^– subset depleted. iNKT cells were enumerated longitudinally by flow cytometry using ⍺-galcer–loaded CD1d tetramers, in a cohort of 46 patients with active tuberculosis (TB) and human immunodeficiency virus type 1 (HIV-1) infection. Samples were collected around the time of TB diagnosis (TB0), at antiretroviral therapy initiation (ARV0, a median of 17.5 days after TB treatment initiation), and at 2 weeks (ARV2) and 4 weeks (ARV4) post–ART initiation; not all patients contributed data to the first timepoint as patients who had taken >4 doses of TB treatment at enrollment contributed data from ARV0, resulting in fewer data points at TB0. TB-IRIS presentation was typically at ARV2. Increased iNKT cell frequency (A) was observed in TB-IRIS patients compared to non-IRIS controls. CD4⁺CD8^– iNKT cell percentages were reduced in TB-IRIS patients (B). Statistical analysis was by multivariable negative binomial modeling to examine associations of iNKT cell frequency and number with TB-IRIS status and by multivariate linear regression modeling to estimate difference in CD4/CD8 cell subset percentages between TB-IRIS and non-IRIS patients, including data from all timepoints to derive P values. In (A), zero values were replaced by 1 for representation on a log scale.

Figure 5.

Invariant natural killer T (iNKT) cell cytotoxicity associated with tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS). iNKT cells were characterized longitudinally by flow cytometric analysis for surface markers CD161 and CD107a, in patients with TB-IRIS and without IRIS. Between antiretroviral therapy initiation (ARV0) and 2 weeks post–ART initiation (ARV2), there was a reduction in CD161⁺ iNKT cell percentage in TB-IRIS compared with non-IRIS patients (A), whereas CD107a⁺ iNKT cell percentage increased in TB-IRIS patients between ARV0 and ARV2, compared to non-IRIS patients (B). CD107a⁺ iNKT cell frequency (cells per million CD3⁺CD19^– live lymphocytes) was increased in TB-IRIS patients compared to non-IRIS controls at ARV2, the most common time of TB-IRIS presentation (C). Mann-Whitney U test for TB-IRIS vs non-IRIS: *P < .05; **P < .01.