SIV-induced terminally differentiated adaptive NK cells in lymph nodes associated with enhanced MHC-E restricted activity

Natural killer (NK) cells play a critical understudied role during HIV infection in tissues. In a natural host of SIV, the African green monkey (AGM), NK cells mediate a strong control of SIVagm infection in secondary lymphoid tissues. We demonstrate that SIVagm infection induces the expansion of terminally differentiated NKG2alow NK cells in secondary lymphoid organs displaying an adaptive transcriptional profile and increased MHC-E-restricted cytotoxicity in response to SIV Env peptides while expressing little IFN-γ. Such NK cell differentiation was lacking in SIVmac-infected macaques. Adaptive NK cells displayed no increased NKG2C expression. This study reveals a previously unknown profile of NK cell adaptation to a viral infection, thus accelerating strategies toward NK-cell directed therapies and viral control in tissues.


Supplementary Figure 5. Gene expression profiles in LN and blood NK cells from chronically infected AGMs.
Heatmaps of the transcript signatures for (a) NK cell surface markers, (b) transcription factors, (c) surface receptors. In each panel, NK cell subsets are organized as in Fig.2. Each row represents a variable gene among clusters, and each column represents the median of 3 monkeys. (d) Mean fluorescent intensity (MFI) for a given marker on NK cells during the chronic phase (day 150 p.i.) of SIVagm infection in AGM LN. Each monkey is represented by a triangle. The black bar represents the median, the error bar shows the interquartile range. For groups comparisons two-sidedWilcoxon signed-rank test with Bonferroni correction were used (n=3). P values of less or equal to 0.05 were considered statistically significant. Asterix indicate significant change. Exact P values are provided on the graphs.

Supplementary Figure 6. Transcriptome analysis of NK cells in blood and LN from chronically SIVagm-infected AGMs. (a)
The pie chart shows the enriched Gene Ontology (GO) terms for the up-regulated genes shared by all NK cell subsets in LN when compared to blood. The 'cellular component', 'biological process', 'molecular function' and 'immune system process' GO terms were selected for this analysis. The twosided hypergeometric test was used in the statistical inference. The term value corrected with the Bonferroni step down method was applied for pvalue correlation. The adjusted p-value threshold was set to 0.001. The list of the genes found for each GO term is given in the Source Data file. Each pie slice represents the percentage of genes found in a given pathway among total up-regulated genes in LN. (b) and (c) Heatmaps showing all the genes differentially expressed in blood and LN cell subsets. P values of less or equal to 0.05 were considered statistically significant. In each panel the name of the proteins corresponding to the transcripts are indicated. (d) The pie chart shows the enriched GO terms for the downregulated genes shared by all LN NK cells when compared to blood ones. The analysis was the same as described in (a). The list of genes found in each pathway is given in the Source Data file. Each pie slice represents the percentage of genes found in a given pathway among the total genes downregulated in LN NK cells. (e) and (f) Heatmaps showing all the genes found to be significantly downregulated in LN NK cells in at least one comparison between blood and LN cell subsets. P values of less or equal to 0.05 were considered statistically significant. In each panel the name of the proteins corresponding to the transcripts are indicated. Three chronically infected monkeys were analysed. Source data have been deposited in the Gene Expression Omnibus database; the accession number is GSE140600. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140600. All the pathways obtained during the GO enrichment analysis are provided in the Source Data file. Figure 7. MHC-E expression on CD4 + T cells in LN during pathogenic and nonpathogenic SIV infection. (a-b) Frequencies of MHC-E + cells among (a) memory and (b) naive CD4 + T cells in LN from MAC (orange) and AGM (blue) before and during SIV infection. Each symbol represents an individual monkey. The black bar represents the median and error bars the interquartile range. Comparisons were done with a two-sided Wilcoxon signed-rank test with Bonferroni correction (n=3). P ≦0.05 was considered statistically significant and marked by asterixs. (c-d) Longitudinal analyses on MHC-E + cell frequencies among distinct memory CD4 + T cell subpopulations in LN from infected MAC (orange) and AGM (blue). Each symbol represents an individual monkey. The black bar represents the median and the error bar the interquartile range. Comparisons were performed for time points where samples from 6 animals have been analyzed (days 0, 9 and 150 p.i.), using a two-sided Wilcoxon signed-rank test with Bonferroni correction (n=3). P ≦ 0.05 was considered statistically significant and indicated by asterixs.

Supplementary Figure 9. MHC-E dependent cytotoxic activity of NK cells before and after SIV infection. (a) Competition assay for
binding to MHC-E. K562-HLA-E*0101 cells were loaded with biotin-A2 VL9 peptides, biotin-HSP60 peptides, biotin-SIVmac ENV LS peptides and biotin-SIVagm ENV LS peptides. An increasing concentration of the indicated unlabeled competitor peptides was introduced into the culture. MHC-E-bound biotinylated peptide was measured at the surface of K562 cells. Data represent the median of three independent experiments. The sequence of the region coding for the SIVmac251 peptide is identical to that of SIVmac239. (b) MHC-E dependent cytotoxic activity of NK cells from blood before and during chronic SIV infection. Uninfected cynomolgus MAC are indicated in orange squares, uninfected rhesus MAC in red circles and uninfected AGM in blue triangles. SIV infected animals are indicated in black. Rhesus MAC corresponded to 4 uninfected, 6 controller (SIC) and 4 viremic (VIC) animals. For cynomolgus MAC and AGM, comparisons were done with a two-sided Wilcoxon signed-rank test with Bonferroni correction (n=3). P values of less or equal to 0.05 were considered statistically significant. Asterix indicate significant change when compare to the base line. Exact P value are provide on the graphs. For . Rhesus MAC, significant difference between the two conditions was determined using two-tail Mann-Whitney test . P values of less or equal to 0.05 were considered statistically significant and indicated by asterixs. (c) MHC-E dependent cytotoxic activity of NK cells from spleen before and during chronic SIV infection. Peptides derived from ENV LS of SIVagm.sab92018, SIVagm.sab1, SIVmac239/251, SIVgsn and SIVagm.tan were analyzed. NK cell activities for uninfected cynomolgus MAC are indicated in orange squares and uninfected AGM in blue triangles. NK cell activities for SIV infected animals are depicted in black. The SIVmac251 peptide is identical to the corresponding SIVmac239 region. Each symbol represents an individual monkey. The black bar represents the median and the error bar the interquartile range. Significant difference between healthy and SIV-infected monkeys were determined using twotail Mann-Whitney test. P values of less or equal to 0.05 were considered statistically significant and indicated by asterixs. This is a T cell epitope prediction tool, which takes into account an amino acid sequence and determines each subsequence's ability to bind to a specific MHC class I molecule 94,95 . Here the sequence corresponds to the leader sequence. The multiple possible subsequences of it are represented by the black dots in the graph. Each of the subsequence and its predicted capacity for binding to HLA-E is provided in the Source Data file. The location of each dot represents the predicted capacity of binding to HLA-E. For each peptide, percentile ranks for binding to HLA-E*0101 are indicated on the y-axis, and percentile ranks for binding to HLA-E*0103 are shown on the x-axis. A low percentile rank value indicates high affinity. The position of the known VL9 and HSP60 control peptides and of the selected SIV peptides are shown in red. The sequences for the selected SIVmac239 and SIVmac251 peptide are identical to each other.

Supplementary table 1: Correlation between NK cell subsets in LN and viral load in LN.
The data on the NK cell subsets correspond to those shown in Figure 3b. The rho and p values of the correlations are shown.

SIVmac251
GAG quantification of plasma viral load of SIVmac251 Backward