Ex vivo rectal explant model reveals potential opposing roles of Natural Killer cells and Marginal Zone-like B cells in HIV-1 infection

Abstract

Our understanding of innate immune responses in human rectal mucosal tissues (RM) and their contributions to promoting or restricting HIV transmission is limited. We defined the RM composition of innate and innate-like cell subsets, including plasmacytoid dendritic cells; CD1c + myeloid DCs; neutrophils; macrophages; natural killer cells (NK); Marginal Zone-like B cells (MZB); γδ T cells; and mucosal-associated invariant T cells in RM from 69 HIV-negative men by flow cytometry. Associations between these cell subsets and HIV-1 replication in ex vivo RM explant challenge experiments revealed an inverse correlation between RM-NK and p24 production, in contrast to a positive association between RM-MZB and HIV replication. Comparison of RM and blood-derived MZB and NK illustrated qualitative and quantitative differences between tissue compartments. Additionally, 22 soluble molecules were measured in a subset of explant cultures (n = 26). Higher production of IL-17A, IFN-γ, IL-10, IP-10, GM-CSF, sFasL, Granzyme A, Granzyme B, Granulysin, and Perforin following infection positively correlated with HIV replication. These data show novel associations between MZB and NK cells and p24 production in RM and underscore the importance of inflammatory cytokines in mucosal HIV infection, demonstrating the likely critical role these innate immune responses play in early mucosal HIV replication in humans.

Figure 1

Quantification of innate and innate-like cellular subsets within the rectal mucosa. (A) Representative gating strategy for rectal mucosal, (B) innate and (C) innate-like subsets of CD45 + cells. (D) Median (red bar) percentage of innate and innate-like cell subsets as a percentage of CD45 + cells within rectal mucosal biopsies (n = 85 γδ T cells and MAIT, n = 69 all other subsets). pDCs were the least abundant cell subset analyzed (Kruskal–Wallis, p < 0.0001, all comparisons except neutrophils, p = 0.0002). Both MZB (Kruskal–Wallis, p < 0.0001, all comparisons) and NK (Kruskal–Wallis, p < 0.0001, all comparisons, except γδ T cells, p = 0.009) were the most abundant innate cell type quantified.

Figure 2

Correlations between NK and MZB cell percentages within the rectal mucosa, and ex vivo HIV-1 replication. Rectal biopsies were exposed to HIV-1 BaL, placed on collagen rafts, and culture supernatants were collected and probed for HIV replication via p24 ELISA over the course of 18 days. (A) Longitudinal p24 values for all study participants (n = 86), normalized to biopsy weight. P24 accumulation was above assay limit of detection, but minimal in some individuals (triangles, reds) and substantial in others (squares, blues). (B) The median log of the cumulative area under the curve (AUC) of p24 accumulation was calculated and correlated with innate cellular subsets, as a percentage of CD45 + cells. Spearman r and p values indicated. (C) A positive association emerged between MZB cells and logAUC p24, (D) while NK percentages resulted in a negative correlation with p24.

Figure 3

Quantitative differences between RM residing MZB and NK cells, and those circulating in blood. In contrast to the RM residing subsets (Fig. 2), there was no correlation with MZB (A) or NK (B) percentages within blood and p24 production in the rectal explant. (C) Within the RM, there are more B cells, and proportionally more MZB cells, than found within the blood (2way ANOVA, p < 0.0001). (D) There are fewer NK within the RM vs. blood, however substantially more demonstrate the CD16-CD56 + phenotype (2way ANOVA, p < 0.0001).

Figure 4

HSNE analysis of blood and RM residing MZB cells. Analysis was performed with matched blood and RM MZB cells, defined as CD45 + , CD3−, CD20 + , HLA-DR + , CD1c + (no more than 5000 events/participant/compartment). (A) HSNE analysis illustrated a polarization of blood (red) and RM (blue) MZB cells. (B) Heatmap analysis illustrates higher CD20 and CD1c expression on MZB from blood, while increased HLA-DR expression on RM residing MZB. C) MFI comparisons on the same cells demonstrate statistical differences in expression of these markers on blood vs. RM MZB (Wilcoxon test, p < 0.0001, all markers).

Figure 5

HSNE analysis of blood and RM residing NK cells. Matched blood and RM NK cells were compared in this instance, defined as CD45 + , Lin− , CD16± , CD56± (no more than 5000 events/participant/compartment). (A) This HSNE analysis resulted in discrete populations of blood (red) and RM (blue) NK cells. (B) Heatmap analysis illustrates a near absence of CD16 on RM residing MZB. C) MFI comparisons on the same cells demonstrate statistical differences in expression of both CD16 and CD56 on blood vs. RM MZB (Wilcoxon test, p < 0.0001, all markers).

Figure 6

Associations between cytokine and effector molecule production and p24 production within the rectal explant model of infection. Cytokine and effector molecule production was quantified, longitudinally, in a subset of rectal explant supernatants (n = 26). Spearman analysis revealed correlations between log p24 AUC (x axis) and IL-17A, IFN-γ, IL-10, IP-10, GM-CSF, sFasL, GZA, GZB, Granulysin, and Perforin AUC (y axis). Spearman p and r values indicated. Data points where molecules were not detectable (zero values) are not visible on graph due to log scale, however all values were included in correlation analysis.