Differential Integrin Adhesome Expression Defines Human NK Cell Residency and Developmental Stage

Abstract

NK cells are innate immune cells that reside within tissue and circulate in peripheral blood. They interact with a variety of microenvironments, yet how NK cells engage with these varied microenvironments is not well documented. The adhesome represents a molecular network of defined and predicted integrin-mediated signaling interactions. In this study, we define the integrin adhesome expression profile of NK cells from human tonsil, peripheral blood, and those derived from human hematopoietic precursors through stromal cell coculture systems. We report that the site of cell isolation and NK cell developmental stage dictate differences in expression of adhesome associated genes and proteins. Furthermore, we define differences in cortical actin content associated with differential expression of actin regulating proteins, suggesting that differences in adhesome expression are associated with differences in cortical actin homeostasis. These data provide understanding of the diversity of human NK cell populations and how they engage with their microenvironment.

Figure 1.. Tonsil and peripheral blood NK cell subsets have unique profiles of adhesome gene expression that reflect distinct developmental trajectories.

Bulk RNA sequencing was performed on sorted primary human NK cell developmental intermediates from 3 healthy peripheral blood and tonsil donors as described in Methods. A) PCA of gene expression of NK developmental stages 3, 4A, 4B and 5 isolated from tonsil, and stages 4B, 5 and 6 from peripheral blood. Ton, tonsil; PB, peripheral blood; number represents NK developmental stage. Each data point represents a single technical replicate (n=3) pooled from 12 biological donors. B) 18,475 genes ordered and plotted by PC1 loading weights. 229 adhesome genes are highlighted; red points represent 56 adhesome genes with PC1 weights greater than 0.01 or less than −0.01; turquoise points are adhesome genes with PC1 weight values between −0.01 and 0.01. Labeled points represent relevant adhesome and NK cell associated genes. C) PCA of peripheral blood and tonsil NK subset expression of 229 genes following filtering of whole transcriptome on integrin adhesome genes (Supp. Table I).

Figure 2.. Human NK cell adhesome gene expression is determined by developmental stage and tissue residency. Adhesome gene expression data of tonsil and PB NK cells was subjected to K means clustering (K=5) and expression of genes within each cluster were averaged and plotted.

A) Average expression of K means clusters of tonsils (grey) and PB (blue) NK cell subsets. Standard deviation is depicted by dashed lines. B) K means heatmap of adhesome gene expression of tonsil and PB NK cells subsets. C) Average gene expression of select representative adhesome genes from K means clusters. Ton, tonsil; PB, peripheral blood. Data derived from the means of 3 technical replicates pooled from 12 tonsil or peripheral blood donors.

Figure 3.. Leukocyte-specific integrin expression reflects the tissue specificity of human NK cell developmental subsets.

Expression (MFI) of integrin subunits on NK cell developmental subsets was measured as described in Methods. Data shown are from 4–7 peripheral blood or tonsil healthy donors. A-C) MFI (left) and representative histograms (right) of leukocyte associated integrins CD11a, CD11b, and CD11c. D) Frequency (percent) of CD103⁺ NK cells in PB and tonsil and representative histograms of CD103 expression showing the frequency of tonsil CD103⁺ NK cells. E) Relative density (MFI) of CD103 on tonsil and peripheral blood NK cells. Ordinary one-way ANOVA with multiple comparisons was used to compare MFI of tonsil CD103⁺ or PB CD103^– NK cell subsets. Multiple comparisons are reported in Supp. Table II. F) Relative density (MFI) of CD18 on PB and tonsil NK cells and representative histograms (right). Error bars indicate mean±SD. MFIs from each tonsil or PB NK cell subset were compared by ordinary one-way ANOVA with multiple comparisons. P-values from the multiple comparison test are reported in Supp. Table II. FMO, fluorescence minus one.

Figure 4:. Integrin β1 and associated subunit expression is dictated by NK cell developmental stage and tissue residency.

Expression of β1 associated integrins CD49a, CD49d, CD49e, CD29 and integrin β7 from flow cytometry of dissociated tonsil mononuclear cells or peripheral blood mononuclear cells. Data shown are from 4–10 peripheral blood or tonsil healthy donors and technical replicates. A) Frequency (percent) of tonsil and PB CD49a⁺ NK cells and representative histogram with percent of CD49a⁺ tonsil NK cells. B) Relative density (MFI) of CD49a on CD49a⁺ NK cells (open circles) and CD49a⁻ NK cells (filled circles). Tonsil CD49a⁺ or PB CD49a^– NK cell subsets were compared to one another by ordinary one-way ANOVA with multiple comparisons. Multiple comparisons are reported in Supp. Table II. C-E) MFI and representative histograms of β1 associated integrins CD49d, CD49e, and CD29. F) Frequency (percent) of tonsil and PB NK cells with detectable integrin β7 and representative histogram with percent of integrin β7⁺ tonsil NK cells. G) Relative density (MFI) of integrin β7 on NK cells. Error bars indicate mean±SD. MFIs from each condition were compared by ordinary one-way ANOVA with multiple comparisons. P-values from the multiple comparison test are reported in Supp. Table II. FMO, fluorescence minus one.

Figure 5.. Integrin conformation is determined by tissue specificity and developmental stage.

Flow cytometric analysis of dissociated tonsil mononuclear cells or peripheral blood mononuclear cells was used to measure expression (MFI) of activated integrin subunits. Data shown is of 5 tonsil and 4 peripheral blood healthy donors. Monoclonal antibodies that detect the open (activated) conformation of CD29 (CD49a-f/CD29) or CD18 (CD11a-c/CD18) were used to probe developmental subsets from tonsil or peripheral blood as indicated. Frequency (A and B) or mean fluorescence intensity (C and D) of open conformation integrins on NK cells were quantified relative to FMO samples. Representative MFI data shown are from 4 peripheral blood and 4 tonsil donors analyzed on the same day; cells were gated on HUTS4⁺ or mAb24⁺ before calculating MFI. Frequency of integrin activation are pooled data from 11–13 PB and 9 tonsil donors run on separate days as indicated by the shape of the data point. Normally distributed frequencies of cells with activated CD29 or CD18 were compared by ordinary one-way ANOVA with multiple comparisons, while non-parametric data was analyzed by Kruskal-Wallis Test with multiple comparisons. MFIs from each condition were compared by ordinary one-way ANOVA with multiple comparisons, P-values reported in Supp. Table II.

Figure 6.. Expression of leukocyte-specific integrins by in vitro derived human NK cells.

NK cells were differentiated in vitro with co-culture on EL08.1D2 or OP9 stromal cells and their phenotype was analyzed at week 4 by flow cytometry. Percent of NK cells expressing CD11a, CD11b, CD11c, CD103, and CD18 (left panel) and their respective density (right panel), by MFI, on in vitro derived NK cells from EL08.1D2 (A, C, E, G, I) and OP9 (B, D, F, H, J) conditions. n= 4–5 biological repeats per stromal cell condition. Statistical significance was tested by ordinary one-way ANOVA with multiple comparisons. P-values from multiple comparisons are reported in Supp. Fig. 2C, D.

Figure 7.. In vitro derived human NK cell subsets exhibit stage-specific expression of β1- and β7-associated integrins.

Analysis of CD49a, CD49d, CD49e, CD29 and integrin β7 expression frequency (left panel) and MFI (right panel) on in vitro stage 3, 4, and 5 NK cells from EL08.1D2 (A, C, E, G, I) or OP9 (B, D, F, H, J) conditions. n= 4–5 biological replicates per stromal cell condition. Statistical significance was tested by ordinary one-way ANOVA with multiple comparisons. P-values from multiple comparisons are reported in Supp. Fig. 2C, D.

Figure 8.. Integrin activation profile and cortical actin density of in vitro derived human NK cells.

Percent positive NK cells (left) or MFI (right) of open conformation CD18 or CD29 on in vitro derived NK cell subsets measured by flow cytometry from EL08.1D2 (A, C) or OP9 conditions (B, D). Statistical significance was tested by ordinary one-way ANOVA with multiple comparisons. Number above graph indicates P value. P-values from multiple comparisons are reported in Supp. Fig. 2C, D. E) Representative histogram of activated LFA-1 expression (MFI) on stage 3, stage 4 and stage 5 of in vitro derived (left) or peripheral blood (right) NK cells. F) Intensity of phalloidin detected by intracellular staining of in vitro derived NK cells from EL08.1D2 (open circles) and OP9 (open triangles) at week 4 measured by flow cytometry. n= 2–7 biological and technical replicates per stromal cell condition. **P<0.05 ordinary one-way ANOVA with multiple comparisons. Error bars indicate mean±SD. P-values from multiple comparisons are reported in Supp. Fig. 2C, D.

Figure 9.. Terminally mature primary NK developmental subsets have increased density of cortical actin.

Primary NK cells were isolated from tonsil and subsets were isolated as described in Methods. A) ACTB (top) or ACTG1 (bottom) gene expression (FPKM) from RNA-Seq data of tonsil and peripheral blood NK cell developmental intermediates. n=3 biological replicates, mean±SD. ns, not significant by Kruskal-Wallis. B) Representative flow cytometric analyses of phalloidin content in peripheral blood NK cells. C) Quantification of actin intensity in NK cell developmental subsets from tonsil or peripheral blood. n=5 biological replicates, mean±SD. **P<0.05 by Kruskal-Wallis Test with multiple comparisons. D) Structured illumination microscopy images of primary NK cells freshly isolated, fixed, permeabilized and immunostained with phalloidin (magenta) and anti-CD56 antibody (cyan). Scale bar = 5 μm. E) CD56 integrated density. n = 11 (CD56^bright), 16 (CD56^dim); ****P<0.0001 by Mann-Whitney test. F) Actin integrated density. n = 11 (CD56^bright), 16 (CD56^dim); ***P<0.001 by Mann-Whitney test. G) Cell volume of stage 4 and stage 5 cells. n=10 per condition; ns, not significant by Mann-Whitney test. H) Normalized estimated gene expression (FPKM) of selected genes ARPC2 and PFN1 related to actin remodeling and nucleation. P = 0.01 by Kruskal-Wallis test.