HIV disease progression correlates with the generation of dysfunctional naive CD8(low) T cells

Abstract

HIV infection can result in depletion of total CD4(+) T cells and naive CD8(+) T cells, and in the generation of dysfunctional effector CD8(+) T cells. In this study, we show that naive CD8(+) T cells in subjects with progressive HIV disease express low levels of CD8α and CD8β chains. Such naive CD8(low) T cells display broad signaling defects across the T-cell receptor complex, and their appearance correlates with generalized up-regulation of major histocompatibility complex class I (MHC-I) antigens on peripheral blood mononuclear cells (PBMCs). To explore a causal link between increased MHC-I up-regulation and the generation of naive CD8(low) T cells, we used the humanized SCID-hu Thy/Liv mouse model to show that HIV infection of the thymus and interferon α (IFNα) treatment alone result in MHC-I up-regulation and in the generation of dysfunctional CD3(high)CD8(+)CD4(-) single-positive 8 (SP8) thymocytes with low expression of CD8. We suggest that dysfunctional naive CD8(low) T cells are generated as a result of IFNα-mediated up-regulation of MHC-I on stromal cells in the thymus and antigen-presenting cells in the periphery, and that dysfunction in this naive compartment contributes to the immunodeficiency of HIV disease. This study is registered at www.clinicaltrials.gov as NCT00187512.

Figures

Figure 1

Naive CD8low T cells are prevalent in untreated HIV-infected human progressors. Cryopreserved PBMCs of HIV-negative and HIV-infected subjects in various stages of disease progression, viral control, and treatment were analyzed for cell-surface expression of the CD8α and CD8β chains in the naive pool of CD8+ T cells. 3 separate studies (study A, B, and C) were performed with samples from subjects of SCOPE and Options cohorts (see Table 1 and “Subjects” and supplemental Methods for more details and definitions of these groups). (A) FACS plot representing gating strategy, CD4 and CD8 expression in CD3+ T cells, and CD45RA and CD27 expression in CD8+ T cells (left to right panels). Naive (CD45RAhighCD27high) CD8+ T cells are shown from a subject with progressive disease (“Prog”), from a virologically suppressed subject (“HAART”) and from a HIV-negative subject (Neg) from study A (as indicated by the blue rectangle) with measurement of CD8α expression (MdFI). (B) CD8α expression (MdFI) in naive CD8+ T cells in all subjects from study A. (C) CD8α (top) and CD8β (bottom) expression in naive CD8+ T cell in all subjects from study B. (D) Grouping strategy of subjects in study C, based on viral load (“VL”) and immune activation (“IA” measured by CD38 expression) at approximately 12 months after infection. (E) Expression of CD8α (higher) and CD8β (lower) in naive CD8+ T cell over time, as measured in subjects at 3 and 12 months after infection (“acute” and “chronic,” respectively) from group 1 to 3 in study C (left to right). P values were calculated using the Mann-Whitney test for group analysis as indicated by the horizontal bar (B) or compared with HIV-Neg group when not indicated (C). The paired t test was used for longitudinal follow-up of the same subject (E). When the P value is not indicated on the graph, the symbols * and ** indicate that P < .05 or P < .005 for each test, respectively.

Figure 2

Naive CD8low T cells from untreated HIV-infected subjects are impaired with respect to calcium flux response and phosphorylation of p38MAPK after TCR stimulation. Calcium flux and the level of phosphorylation of p38MAPK were measured by flow cytometry in naive (CD45RAhighCD27high) CD8+ T cells from PBMCs of study A (described in Figure 1B). Calcium flux was measured after TCR/CD8 cross-linking, and phosphorylation of p38MAPK was measured by the fold induction of p38MAPK phosphorylation from PHA-stimulated compared with mock-stimulated samples (see “Phenotypic and functional analysis in HIV-infected subject PBMCs” and supplemental Methods and Schweneker et al). (A-C) Calcium flux responses over time were determined by the percentage of responding naive CD8+ T cells over 75% of the baseline level (prior TCR stimulation). Calcium flux responses are shown in 3 HAART and 3 Prog. CD4 < 350 (A) and for all subjects (B), and were correlated with the fold-induction of phopho-p38MAPK (C left) and shown in a representative example (C right). (D-E) Finally, both calcium flux (D) and phospho-p38MAPK (E) responses were correlated with the CD8α expression in naive (CD45RAhighCD27high) CD8+ T cells. P values are indicated for Spearman rank correlation test. R2 is also indicated (from the Pearson coefficient of correlation).

Figure 3

CD8low CD8 naive T cells from untreated HIV-infected subjects are functionally impaired with respect to IL-2 secretion after TCR stimulation. PBMCs from subjects in study D, including 16 Prog. CD < 350 and 10 HAART subjects, were stimulated or not by the superantigen SEB to determine IFNγ and IL-2 intracellular cytokine response. Cells were stimulated overnight and analyzed for cytokine secretion in combination with surface staining for CD3, CD4, CD8, CD45RA, CD27, CD28, and CCR7. (A) Ethidium monoazide (EMA) labeling was also used to discriminate live from dead cells. (B) The frequency of IL-2 but not IFNγ-responding cells was determined in naive CD8+ T cells (EMA−CD3+CD8+CD45RA+CD27+CCR7+CD28+). P values were calculated using Mann-Whitney test for group analysis as indicated by the horizontal bar.

Figure 4

In HIV-infected subjects, generalized up-regulation of cell-surface MHC-I in PBMCs is associated with disease progression, chronic immune activation, and naive CD8low T cells. MHC-I cell-surface expression was measured by the MFI of MHC-I in PBMCs from subjects in studies A, B, and C, as shown here in CD4+ T cells. (A) MHC-I cell-surface expression in study A (left) and study B (right); of note, differences in MHC-I MFI measurements between study A and B are because of the use of different flow cytometers with different instrument settings. (B) MHC-I cell-surface expression in CD4+ T cells from study C at 3 and 12 months infection (“acute” and “chronic,” respectively). (C) Correlation between MHC-I cell-surface expression and CD8α expression in naive CD8+ T cells in study A (left) and study B (right). P values were calculated using the Mann-Whitney test for group analysis as indicated by the horizontal bar (A). Two-tailed paired and unpaired t tests were used for longitudinal follow-up of the same subject and for group analysis, respectively (B). When the P value is not indicated on the graph, the symbols * and *** indicate that P < .05 or P < .0005 for each test, respectively. P values are indicated for Spearman rank correlation for correlation test. R2 from the Pearson coefficient of correlation is also indicated.

Figure 5

HIV infection generates dysfunctional CD8low SP CD8 thymocytes in the SCID-hu Thy/Liv mouse model. Saline, or 1000 50% tissue culture infective dose (TCID50) of X4-tropic (NL4-3), R5-tropic (Ba-L), or dual-tropic X4-R5 (primary isolate JD) HIV were inoculated into individual cohorts of 30-40 SCID-hu Thy/Liv mice each. In most cohorts, several groups were also treated with antiretroviral drugs (eg, 3TC). Viral load (HIV RNA content per million total cells) as well as phenotypic and functional parameters were measured by flow cytometry in human thymocytes from each implant 3 to 7 weeks after inoculation. (A) FACS plot of SP8 thymocytes in Mock- and HIV-infected human thymus. (B) Correlation between CD8α expression in SP8 thymocytes and MHC-I expression in double-positive (DP) thymocytes from individuals within the same cohort of HLA-A2+ SCID-hu Thy/Liv mice, including animals inoculated with medium (Mock), HIV NL4-3, Ba-L, or JD, and compared with IFNα treatment alone. (C-E) Calcium flux response after TCR/CD8 cross-linking in SP8 thymocytes from a cohort of SCID-hu Thy/Liv mice, 21 days postinoculation with NL4-3: representative calcium flux kinetic from 2 mock-infected (blue line) or 2 NL4-3-infected mice (red line). (C) Calcium flux peak responses in all animals (D) and correlation with CD8α expression in SP8 thymocytes (E). P values were calculated using the Mann-Whitney test for group analysis and by Spearman rank correlation for correlation test. R2 from the Pearson coefficient of correlation is also indicated.

Figure 6

CD8low SP8 thymocytes have impaired proliferative capacities after TCR stimulation. Thymi from SCID-hu Thy/Liv mice were processed 21 days after direct intrathymic inoculation of saline (mock) or HIV JD, as described in Figure 5. MHC-I MFI and CD8 MdFI were determined at day 0 on total fresh thymocytes. In parallel, CD3 and CD28 proliferative responses were assessed on purified SP8 thymocytes after 5 days by CFSE dilution. (A) CFSE proliferative responses (flow plots, higher), histogram representation (red curve, lower), and modeling (blue curve, lower) in SP8 thymocytes from representative examples of mock- or HIV-infected SCID-hu Thy/Liv mice, indicating the percentage of divided, proliferative index, and division index after proliferation, as well as CD8 expression at the starting day (Day 0). (B) Percent of divided SP8 thymocytes (modeling) in HIV-infected SCID-hu Thy/Liv mice (n = 8) compared with mock- and virally suppressed 3TC/HIV-infected animals (n = 9, P = .001) and (C) correlation to CD8 MdFI at day 0 (P = .0001). P values were calculated using the Mann-Whitney test for group analysis and by Spearman rank correlation for correlation test. R2 from the Pearson coefficient of correlation is also indicated.

Figure 7

Generation of CD8low SP8 thymocytes correlates with reduced frequency of Melan-A–specific thymocytes. SP8 thymocytes from HLA-A2+ SCID-hu Thy/Liv mice (see Figure 5B) were enriched by CD4 depletion on magnetic beads and stained with MelanA pentamers or control CMV tetramers. (A) Representative examples in each group. (B-D) Frequency of MelanA+ SP8 thymocytes for all thymi (B), correlation with MHC-I up-regulation on DP thymocytes (C), and correlation with CD8α expression in MelanA+ SP8 thymocytes (D). P values were calculated using the Mann-Whitney test for group analysis and by Spearman rank correlation for correlation test. R2 from the Pearson coefficient of correlation is also indicated.