A signaling network stimulated by β2 integrin promotes the polarization of lytic granules in cytotoxic cells

Abstract

Cytotoxic lymphocytes kill target cells through the polarized release of the contents of intracellular perforin-containing granules. In natural killer (NK) cells, the binding of β2 integrin to members of the intercellular adhesion molecule family is sufficient to promote not only the adhesion of NK cells to target cells but also the polarization of intracellular lytic granules toward the target. We used NK cells in an experimental system designed to enable us to study the polarization of lytic granules in the absence of their release through degranulation, as well as β2 integrin signaling independently of inside-out signals from other receptors. Through a proteomics approach, we identified a signaling network centered on an integrin-linked kinase (ILK)-Pyk2-paxillin core that was required for granule and microtubule-organizing center (MTOC) polarization. The conserved Cdc42-Par6 signaling pathway, which controls cell polarity, was also activated by ILK and was required for granule polarization toward the target cell. A subset of the signaling components required for polarization contributed also to the convergence of granules on the MTOC. These results delineate two connected signaling networks that are stimulated upon β2 integrin engagement and control the polarization of the MTOC and associated lytic granules toward the site of contact with target cells to mediate cellular cytotoxicity.

Fig. 1. Components of β₂ Integrin Outside-In Signaling Identified by Mass Spectrometry

(A) Anti-phosphotyrosine (pY) immunoblot of samples used for mass spectrometry analysis. Tyrosine phosphorylated proteins and associated proteins were pulled-down with mAb 4G10 (pY) or isotype control IgG2b mAb MOPC141 (C) from lysates of primary NK cells stimulated on plates for 20 min with BSA only (–), ICAM-1, and human IgG1, as indicated. Samples were eluted with sodium phenyl phosphate, run on SDS-PAGE, and immunoblotted with mAb 4G10. (B) Venn diagram of the number of proteins identified by mass spectrometry in the samples stimulated with ICAM-1 or human IgG1, after subtraction of proteins in the control samples. Selection criteria are described in Table S4 to S6. (C) Ingenuity Pathway Analysis (IPA, http://www.ingenuity.com) of the proteins identified in the sample stimulated with ICAM-1 (Table S4). The intensity of the red color indicates the score from mass spectrometry analysis. (D) Network analysis of the 23 proteins in the sample stimulated with ICAM-1 that had higher scores than the sample stimulated with human IgG1 (Table S6).

Fig. 2. ILK is Required for Granule Polarization Toward S2–ICAM-1 Cells

(A) Primary NK cells were stimulated with BSA only (–), ICAM-1, or human IgG1 for 5 and 20 min, as indicated. Cell lysates were immunoprecipitated with isotype control IgG2b agarose or with 4G10-agarose. Samples eluted with sodium phenyl phosphate (Top panel) and total cell lysates (Bottom panel) were immunoblotted with anti-ILK mAb. (B) ILK mRNA abundance 48 hours after silencing with siRNA, relative to ILK mRNA in cells treated with control (Ctrl) siRNA. Data shows 3 experiments with mean ± SD. (C) Conjugateformation of NK cells with S2 and S2–ICAM-1 cells after ILK silencing. Cells were either mock transfected or transfected with control (Ctrl) or ILK siRNA. Data shows 3 experiments with mean ± SD. (D) Representative images of MTOC and granule polarization toward S2 or S2–ICAM-1 cells. Cells were fixed, permeabilized and stained with mAb to perforin (IgG2b) and β-tubulin (IgG1) followed byisotype-specific Alexa Fluor 488-conjugated and Alexa Fluor 647-conjugatedsecondary antibodies. Scale bar = 5 μm. (E) Quantitative analysis of MTOC and granule polarization toward S2 and S2–ICAM-1 cells in NK cells treated as in (C). n represents individual NK–S2 cell contacts.

Fig. 3. Signaling Components Required for β₂ Integrin-Dependent Granule Polarization

(A) NK cells treated as in Fig. 2A were tested for the presence of γ-parvin, Leupaxin, Pyk2 and RhoGEF7 in tyrosine-phosphorylated protein complexes. Representative immunoblots of 3 to 5 experiments are shown. (B) mRNA abundance, conjugate formation with S2–ICAM-1 cells, and MTOC and granule polarization, as indicated, of NK cells after silencing of the indicated genes by siRNA. (^#Pyk2 protein was monitored by immunoblot.) Data are shown as percent of control siRNA. Conjugation with S2–ICAM-1 cells varied from 32% to 38%, and MTOC and granule polarization varied from 34% to 38% in the controls. Graphs show mean ± SEM from 3 experiments. *p<0.05, **p<0.01.

Fig. 4. A Conserved Signaling Pathway for Cell Polarity is Used by β₂ Integrin to Induce Granule Polarization

(A) Potential pathway for β₂ integrin-dependent granule polarization in NK cells. Molecules in blue have been identified by mass spectrometry. Cdc42, known as a master regulator of microtubule-dependent cell polarity, can be activated by RhoGEF7 and Pyk2. Cdc42 controls cell polarity through both CLIP-170 and APC. Interactions in the diagram indicate direct binding (circles), activation (arrows), indirect activation (dashed arrow), or inhibition (cross bar), according to Ingenuity Pathway Analysis. (B) mRNA abundance after silencing of the indicated proteins. (C) Conjugation with S2–ICAM-1 cells and MTOC and granule polarization after silencing of the indicated proteins. Conjugation varied from 30% to 36%, and MTOC and granule polarization varied from 33% to 37% in the controls. Graphs show mean ± SEM from 3 experiments. **p<0.01. (D) Primary NK cells stimulated with BSA only (–), ICAM-1, or human IgG1 for the indicated times. Cell lysates were immunoblotted with antibodies to Ser⁹ phosphorylated GSK3β, total GSK3β, and actin. (E) ILK abundance monitored by immunoblot after silencing with siRNA. (F) After stimulation with BSA only (–) or ICAM-1 for 5 min, lysates were immunoblotted as in (D). (G) Intensity of GSK3β Ser⁹ phosphorylation, relative to phosphorylation in siRNA controls. Squares, triangles, and diamonds represent data from, from 3 independent experiments. **p<0.01.

Fig. 5. Convergence and Retention of Granules at the MTOC Require Leupaxin, Pyk2, and CLIP-170

(A) Staining of perforin and β-tubulin in KHYG-1 cells after ILK or Leupaxin silencing. The contours in the merged panel of the control indicate a 7.5 μm circle centered over the MTOC (a), and the whole cell (b). (B) Fraction of perforin located near the MTOC relative to perforin in the whole cell after ILK or Leupaxin silencing. (C) mRNA abundance and fraction of cells with perforin near the MTOC after silencing of the indicated genes. (^#Pyk2 abundance was monitored by immunoblot.) The fraction of NK cells in which perforin was near the MTOC in the controls (control siRNA) varied from 71% to 78%, and is set to 100% for each experiment. Graphs show mean ± SEM from 3 experiments. **p<0.01. (D-F) KHYG-1 cells analyzed after silencing of the indicated genes, exactly as described for A-C. The fraction of NK cells in which perforin was near the MTOC in the controls (control siRNA) varied from 71% to 78%, and is set to 100% for each experiment. Graphs show mean ± SEM from 3 experiments. *p<0.05, **p<0.01.

Fig. 6. β₂ Integrin Signaling Controls MTOC Polarization and Granule Convergence to the MTOC

(A) Distance of the MTOC in NK cells from the site of contact with S2 or S2–ICAM-1 cells, relative to the cell diameter, after silencing of the indicated genes. Each dot represents a measurement from a single cell. The graph shows the median from measurements collected in 3 experiments. N. s., not significant; *p<0.05; **p<0.01. (B) Polarization of the MTOC toward S2 and S2–ICAM-1 cells and granule convergence to the MTOC, as indicated, were scored in the absence or presence of an anti-S2 rabbit serum. A minimum of 34 conjugates were analyzed.

Fig. 7. Proximity of β₂ Integrin with ILK at the Site of Contact with S2–ICAM-1 Cells

(A) PLAs were performed with primary NK cells mixed with S2 or S2–ICAM-1 cells for 20 min, fixed, permeabilized, and incubated with rabbit polyclonal Abs to the CD18 cytoplasmic tail or to WASp, together with mouse mAbs to talin, ILK, or Cdc42, as indicated on the left. In the last two panels, S2 and S2–ICAM-1 cells were pre-loaded with rabbit anti-S2 serum. (B) Quantitative analysis of PLAs. Each dot represents an NK cell. A minimum of 19 contacts with S2 cells and 22 contacts with S2–ICAM-1 cells were analyzed. The combination of S2 cells and Abs used are indicated below the graph. The fraction of PLA signals that were polarized toward S2 cells is indicated at the bottom of each column. (C) PLAs for CD18 with ILK in NK cells treated with 1 μM Syk inhibitor II, or with Syk siRNA, as indicated. A minimum of 19 contacts with S2 cells and 26contacts with S2–ICAM-1 cells were analyzed. **p<0.01.