Recruitment of the adaptor protein Nck to PECAM-1 couples oxidative stress to canonical NF-κB signaling and inflammation

See allHide authors and affiliations

Science Signaling  24 Feb 2015:
Vol. 8, Issue 365, pp. ra20
DOI: 10.1126/scisignal.2005648


Oxidative stress stimulates nuclear factor κB (NF-κB) activation and NF-κB–dependent proinflammatory gene expression in endothelial cells during several pathological conditions, including ischemia/reperfusion injury. We found that the Nck family of adaptor proteins linked tyrosine kinase signaling to oxidative stress–induced activation of NF-κB through the classic IκB kinase–dependent pathway. Depletion of Nck prevented oxidative stress induced by exogenous hydrogen peroxide or hypoxia/reoxygenation injury from activating NF-κB in endothelial cells, increasing the abundance of the proinflammatory molecules ICAM-1 (intracellular adhesion molecule–1) and VCAM-1 (vascular cell adhesion molecule–1) and recruiting leukocytes. Nck depletion also attenuated endothelial cell expression of genes encoding proinflammatory factors but not those encoding antioxidants. Nck promoted oxidative stress–induced activation of NF-κB by coupling the tyrosine phosphorylation of PECAM-1 (platelet endothelial cell adhesion molecule–1) to the activation of p21-activated kinase, which mediates oxidative stress–induced NF-κB signaling. Consistent with this mechanism, treatment of mice subjected to ischemia/reperfusion injury in the cremaster muscle with a Nck inhibitory peptide blocked leukocyte adhesion and emigration and the accompanying vascular leak. Together, these data identify Nck as an important mediator of oxidative stress–induced inflammation and a potential therapeutic target for ischemia/reperfusion injury.


Oxidative stress contributes to inflammation in various cardiovascular pathologies, including ischemia/reperfusion injury, diabetic complications, and atherosclerosis (1, 2). In endothelial cells, oxidative stress promotes increased endothelial permeability and expression of mRNAs encoding proinflammatory adhesion molecules [for example, intercellular adhesion molecule–1 (ICAM-1), vascular adhesion molecule–1 (VCAM-1)] that mediate leukocyte homing (3, 4). The nuclear factor κB (NF-κB) family of redox-sensitive transcription factors classically mediate proinflammatory gene expression (5). The best-characterized NF-κB isoform consists of a p65 subunit (hereafter referred to as NF-κB) either as a homodimer or as a heterodimer with a p50 subunit (5). Typically, proinflammatory stimuli activate the IκB kinase (IKK) complex to stimulate serine phosphorylation, ubiquitination, and degradation of inhibitory IκB proteins, thereby allowing nuclear localization of NF-κB. IKK also phosphorylates NF-κB on Ser536 in the transactivation domain, enhancing its transcriptional activity (6).

Oxidative stress may activate NF-κB through both IKK-dependent and IKK-independent mechanisms (3, 4). Tyrosine phosphorylation mediates various oxidative stress–induced signaling responses because oxidation of critical cysteine residues in the catalytic domain of tyrosine phosphatases inactivates the phosphatase domain, thereby enhancing tyrosine phosphorylation (7). The tyrosine kinase inhibitor herbimycin A blunts NF-κB activation after hypoxia/reoxygenation or addition of exogenous hydrogen peroxide (H2O2) (8, 9); however, the role of tyrosine phosphorylation in oxidative stress–induced NF-κB activation remains unclear. Early work found that H2O2 stimulates IKK-independent NF-κB activation through direct IκB tyrosine phosphorylation in T cells (10, 11). Although high amounts of oxidative stress (300 to 500 μM H2O2) can induce tyrosine phosphorylation of IκB in endothelial cells (9, 12, 13), moderate oxidative stress activates canonical IKK-dependent NF-κB activation in certain cell types (3). Consistent with the latter model, endothelial responses to lipopolysaccharide, angiotensin II, and hemodynamic shear stress all require oxidative stress for canonical IKK-dependent activation of NF-κB, albeit through largely unknown mechanisms (1416).

The Nck family of Src homology 2 (SH2) and SH3 domain–containing adaptor proteins [Nck1 and Nck2 (Nck1/2)] classically couple tyrosine kinase signaling to cytoskeletal remodeling responses during cell migration (17, 18). Nck1 and Nck2 share 68% amino acid identity, are present in nearly all cell types, and show both distinct and conserved downstream signaling partners (18). Nck1/2 recruitment to tyrosine-phosphorylated proteins at the plasma membrane drives the activation of the serine/threonine kinase p21-activated kinase (PAK) (19), and we have demonstrated a critical role for PAK in oxidative stress–dependent canonical NF-κB activation by shear stress (20). In the current work, we tested the hypothesis that Nck critically coupled oxidative stress–induced tyrosine phosphorylation to activation of PAK and NF-κB pathways to drive proinflammatory responses.


Oxidative stress activates Nck-dependent canonical NF-κB signaling

To determine how oxidative stress regulates endothelial NF-κB activation, we first examined the dose response and time course for H2O2-induced NF-κB activation in human aortic endothelial cells (HAECs). Treatment with a low dose of H2O2 (100 μM) was sufficient to induce the phosphorylation of Ser536 in NF-κB (Fig. 1A), indicative of IKK-dependent NF-κB activation, which was maximal by 5 to 15 min and was sustained for at least 60 min (Fig. 1B). In contrast, this dose of H2O2 did not increase tyrosine phosphorylation of IκB, as shown by Western blotting and immunoprecipitation with the anti-phosphotyrosine antibody 4G10 (21, 22) (fig. S1, A and B). Consistent with these phosphorylation patterns, low-dose H2O2 treatment resulted in an enhanced IKK activation within 15 min (Fig. 1, C and D), suggesting that low amounts of oxidative stress promote canonical NF-κB activation in endothelial cells. To determine whether Nck facilitates oxidative stress–induced NF-κB activation, we tested H2O2-induced NF-κB activation after small interfering RNA (siRNA)–mediated knockdown of Nck1 and Nck2 (~75% knockdown; fig. S2A). Treatment with Nck siRNA completely blunted oxidative stress–induced IKK kinase activity (Fig. 1, C and D). Consistent with Nck-dependent NF-κB activation, Nck siRNA also blunted H2O2-induced nuclear accumulation (Fig. 1E and fig. S2B) and phosphorylation (Fig. 1F) of NF-κB.

Fig. 1 Oxidative stress requires Nck for canonical NF-κB activation in endothelial cells.

(A) Phosphorylation (Phos or P) of NF-κB in HAECs treated with increasing doses of H2O2 was determined by Western blotting. Phospho-NF-κB was normalized to total NF-κB and conveyed as fold change (Δ) compared to untreated conditions. (B) Phosphorylation of NF-κB in HAECs treated with H2O2 for the indicated times was determined as in (A). (C and D) IKK activity (act) was determined in HAECs transfected with Nck siRNA and treated with H2O2. IP, immunoprecipitate; B+L, beads plus lysate. (E) HAECs were treated as in (C) and nuclear translocation of NF-κB activation was determined by immunofluorescence staining (fig. S2B). At least 100 cells per condition were scored for the presence or absence of nuclear NF-κB staining for each experiment. (F) Phosphorylation of NF-κB in HAECs treated as in (C) was determined by Western blotting. (G) Phosphorylation of NF-κB in HAECs treated with Nck siRNA and exposed to hypoxia followed by reoxygenation (Hyp/Reox) was determined by Western blotting. Norm, normoxia; n = 4 to 5 independent experiments in all panels. *P < 0.05, **P < 0.01, ***P < 0.001.

Reperfusion of ischemic tissues stimulates oxidative stress–dependent NF-κB activation (3, 4). We next sought to determine whether Nck mediated inflammation in the hypoxia/reoxygenation model of endogenous oxidative stress. HAECs were exposed to short-term hypoxia (5% O2, 1 hour) followed by reoxygenation (21% O2, 2 hours), conditions that increase ICAM-1 and VCAM-1 abundance in an oxidative stress–dependent manner (23). Whereas reoxygenation enhanced NF-κB activation, Nck siRNA completely inhibited NF-κB activation in response to hypoxia/reoxygenation injury (Fig. 1G).

NF-κB activation drives proinflammatory gene expression (3, 4), and Nck siRNA effectively blocked increases in ICAM-1 and VCAM-1 abundance induced by H2O2 but not by the proinflammatory cytokine tumor necrosis factor–α (TNFα) (Fig. 2, A to C). Consistent with these results, the addition of a membrane-permeable peptide containing the Nck-binding sequence of PAK (Nck-blocking peptide), but not a control peptide lacking critical proline residues (24), suppressed H2O2-induced increases in VCAM-1 abundance (fig. S2C). Furthermore, Nck siRNA similarly blunted increases in both VCAM-1 (Fig. 2, D and E) and ICAM-1 (Fig. 2, D and F) abundance in cells exposed to hypoxia/reoxygenation injury. Multiple proinflammatory stimuli promote leukocyte–endothelial cell interactions through oxidative stress–induced increases in VCAM-1 and ICAM-1 abundance (3, 4). Whereas treating endothelial cells with either H2O2 (Fig. 2G) or hypoxia/reoxygenation injury (Fig. 2H) enhanced THP-1 monocyte adhesion, Nck depletion blunted adhesion of these cells after either stimulus (Fig. 2, G and H, and fig. S3, A and B). Together, these data demonstrate that the Nck adaptor proteins couple oxidative stress to activation of the NF-κB pathway and proinflammatory gene expression in endothelial cells.

Fig. 2 Depleting Nck expression blunts oxidative stress–induced endothelial activation and leukocyte recruitment.

(A to C) The abundance of VCAM-1 (A and B) or ICAM-1 (A and C) in HAECs transfected with Nck1/2 siRNA and treated with H2O2 or TNFα was determined by Western blotting. Exp, expression; NT, no treatment (D to F) The abundance of VCAM-1 (D and E) or ICAM-1 (D and F) in HAECs transfected with Nck siRNA and exposed to hypoxia/reoxygenation was determined by Western blotting. (G and H) HAECs transfected with Nck1/2 siRNA were exposed to H2O2 (G) or hypoxia/reoxygenation injury (H). The adhesion of Cell Tracker Green–labeled THP-1 monocytes was determined in static adhesion assays (fig. S3) and conveyed as percent of monocytes adhering under each condition. Adh, adhesion. (I and J) HAECs transfected with Nck1/2 siRNA were exposed to H2O2 (I) or hypoxia/reoxygenation (J). Changes in the expression of genes encoding proinflammatory and antioxidant factors were determined by quantitative real-time polymerase chain reaction (qRT-PCR). Results show the fold change in gene expression compared to untreated conditions. Ind, induced; n = 5 independent experiments in all panels. *P < 0.05, **P < 0.01, ***P < 0.001.

Nck siRNA does not affect expression of antioxidant genes after oxidative stress

In addition to inflammation, oxidative stress drives the expression of components in protective antioxidative response systems to alleviate the pathological consequences of future oxidative stress (25). The redox-sensitive transcription factor nuclear factor (erythroid-derived 2)–related factor 2 (Nrf2) promotes the expression of multiple antioxidant genes, such as NAD(P)H dehydrogenase quinone 1 (NQO1), heme oxygenase-1 (HMOX1), peroxyredoxin-1 (PRDX1), and the glutamate-cysteine ligase catalytic subunit (GCLC) (25, 26). Because oxidative stress activates the Keap1 (Kelch-like ECH-associated protein 1)–Nrf2 signaling pathway directly through oxidative modification of critical cysteine residues in Keap1 (26), the Nck adaptors should not contribute to Nrf2 activation. Therefore, we next tested whether Nck depletion differentially affects the inflammatory and adaptive responses to oxidative stress. H2O2 enhanced endothelial cell mRNA expression of both proinflammatory (VCAM-1 and ICAM-1) and antioxidant genes (NQO1, HMOX1, PRDX1, and GCLC) (Fig. 2I). However, only proinflammatory genes showed significantly reduced expression after siRNA-mediated Nck depletion (Fig. 2I). Similarly, Nck siRNA blunted inflammatory gene expression but not antioxidant gene expression after hypoxia/reoxygenation injury (Fig. 2J). Together, these data show that the adaptive response to oxidative stress remains intact after Nck depletion.

Oxidative stress activates Nck signaling through tyrosine phosphorylation of platelet-endothelial cell adhesion molecule–1

Lack of IκB tyrosine phosphorylation at low amounts of H2O2 did not indicate a reduced capacity for oxidative stress–induced tyrosine phosphorylation. Western blotting with the phosphotyrosine-specific antibody 4G10 demonstrated that low-dose H2O2 treatment induced a twofold increase in total cellular tyrosine phosphorylation content by 15 min that continued to increase for 60 min (Fig. 3A and fig. S4A). This increase in tyrosine phosphorylation was accompanied by an increase in Nck coimmunoprecipitation with tyrosine-phosphorylated proteins (Fig. 3B). Proximity ligation assays, which are used to assay protein-protein interactions, similarly showed a significant increase in Nck interaction with tyrosine-phosphorylated proteins after H2O2 treatment (Fig. 3C). This enhancement of Nck-phosphotyrosine interactions correlated with the enhanced recruitment of Nck to the cell membrane fraction (Fig. 3D).

Fig. 3 Nck couples oxidative stress–induced tyrosine phosphorylation to activation of PAK2.

(A) Total cellular tyrosine phosphorylation was determined in cells treated with H2O2 for the indicated times. Representative blots are shown. IB, immunoblot. (B) Nck coimmunoprecipitation (co-IP) with tyrosine-phosphorylated proteins was determined in cells treated with H2O2. (C) HAECs were treated as in (B) and Nck/phosphotyrosine interactions were analyzed in situ by proximity ligation assays (PLA). Average proximal ligations per cell (X¯) are shown. (D) Cells were treated as in (B), and Nck recruitment to the membrane fraction was determined by Western blotting and normalized to the integral membrane protein α5 integrin (int). (E) HAECs were treated as in (B) and PAK2 coimmunoprecipitation with Nck1 and Nck2 was analyzed. (F) Activation of PAK2 was determined by Western blotting in cells pretreated with the Nck-blocking peptide before treatment with H2O2. (G) Cells were treated as in (F), and PAK2 recruitment to the membrane fraction was determined by Western blotting and normalization to the integral membrane protein α5 integrin. n = 5 independent experiments for each panel. *P < 0.05, **P < 0.01, ***P < 0.001.

We have shown that PAK signaling enhances oxidative stress–dependent canonical NF-κB activation in endothelial cells exposed to shear stress (20). Because Nck can promote PAK membrane translocation and activation (19, 27), we next characterized whether H2O2 activates PAK in a Nck-dependent fashion. Treatment with exogenous H2O2 enhanced the interaction between PAK2, the primary isoform in endothelial cells (27), and both Nck1 and Nck2, as determined by coimmunoprecipitation (Fig. 3E), suggesting that these isoforms may be functionally redundant. Nck siRNA inhibited the increase in the phosphorylation of PAK2 induced by both exogenous H2O2 (fig. S4, B and C) and hypoxia/reoxygenation injury (fig. S4D), consistent with a critical role for Nck in oxidative stress–induced proinflammatory signaling. Similarly, treatment with the Nck-blocking peptide completely suppressed both the phosphorylation (Fig. 3F) and translocation of PAK2 to the membrane fraction (Fig. 3G) after oxidative stress.

Although multiple tyrosine-phosphorylated proteins may recruit Nck to the membrane to initiate proinflammatory signaling, Nck coimmunoprecipitations showed a prominent phosphotyrosine band at ~130 kD (Fig. 4, A and B). Therefore, we next sought to identify this protein on the basis of size, abundance, and known tyrosine phosphorylation. We identified the 130-kD protein as platelet-endothelial cell adhesion molecule–1(PECAM-1) because PECAM-1 siRNA caused the disappearance of this phosphotyrosine band from Nck immunoprecipitates (Fig. 4, A and B). PECAM-1 showed increased tyrosine phosphorylation after H2O2 treatment (Fig. 4, C and D), and H2O2 promoted the coimmunoprecipitation of PECAM-1 and Nck in endothelial cells (Fig. 4, C and E, and fig. S5A). Immunocytochemistry indicated that PECAM-1, phosphotyrosine, and Nck1/2 colocalized after H2O2 treatment (fig. S5C). Consistent with PECAM-1–driven Nck recruitment, proximity ligation assays demonstrated enhanced interactions between PECAM-1 and Nck1/2 after H2O2 treatment (Fig. 4, F and G). Oxidative stress similarly stimulated the interaction of PECAM-1 with PAK2 (fig. S5B). Consistent with our model for Nck recruitment to tyrosine-phosphorylated PECAM-1, endothelial PECAM-1 showed increased precipitation with the recombinant glutathione S-transferase (GST)–Nck1 SH2 domain after H2O2 treatment, whereas SH2 domains containing a point mutation reducing its affinity for phosphotyrosine (R308K) demonstrated diminished PECAM-1 binding under both basal and H2O2-stimulated conditions (Fig. 5A). Similarly, wild-type PECAM-1 expressed in human embryonic kidney (HEK) 293 cells showed enhanced precipitation with GST-Nck1 SH2 domains after H2O2 treatment, whereas these SH2 domains failed to precipitate a phosphorylation-deficient PECAM-1 (YYFF, mutation to Phe at Tyr663 and Tyr686; Fig. 5B). The PECAM-1 YYFF mutant also showed reduced coimmunoprecipitation with endogenous Nck1/2, suggesting that PECAM-1 phosphorylation mediates its interaction with the Nck adaptor proteins (Fig. 5C).

Fig. 4 Oxidative stress–induced PECAM-1 phosphorylation stimulates interactions with NcK.

(A and B) Nck was immunoprecipitated from HAECs treated with H2O2, and Nck-interacting tyrosine-phosphorylated proteins were identified by Western blotting with the 4G10 antibody. Enhanced interaction with a 130-kD tyrosine-phosphorylated protein was blunted by PECAM-1 siRNA. Representative Western blots are shown. (C to E) PECAM-1 was immunoprecipitated from HAECs treated with H2O2. Changes in PECAM-1 phosphorylation (C and D) and Nck coimmunoprecipitation (C and E) were analyzed by Western blotting and normalized to total PECAM-1 in the immunoprecipitates. Representative Western blots are shown. (F) HAECs were treated as in (A) and PECAM-1–Nck interactions were analyzed in situ by PLA. Scale bars, 50 μm. (G) Average proximal ligations per cell are shown. n = 5 independent experiments for (A) to (E) and n = 4 independent experiments for (F). *P < 0.05, **P < 0.01.

Fig. 5 PECAM-1 phosphorylation is required for its interaction with the Nck SH2 domain.

(A) Pull-down of PECAM-1 with GST-tagged Nck1 SH2 domains after H2O2 treatment in HAECs was determined by Western blotting and normalized to PECAM-1 abundance in the cell lysates. The GST-tagged R308K mutant SH2 domain served as a negative control for phosphotyrosine-dependent interactions. Representative Western blots are shown. (B and C) HEK293 cells expressing wild-type PECAM-1 or the phosphorylation-deficient PECAM-1 YYFF mutant were lysed, and precipitation with either the GST-Nck1 SH2 domain (B) or endogenous Nck1/2 (C) was determined by Western blotting. Precipitated PECAM-1 was normalized to either total PECAM-1 in the lysates (A) or total Nck1/2 abundance in the immunoprecipitates (B). Representative Western blots are shown. n = 5 independent experiments for each panel. **P < 0.01, ***P < 0.001.

To assess the functional importance of PECAM-1–Nck interactions in the response to oxidative stress, we analyzed oxidative stress–induced proinflammatory responses after PECAM-1 knockdown using siRNA. PECAM-1 knockdown significantly blunted both NF-κB activation (Fig. 6, A and B) and increases in VCAM-1 abundance (Fig. 6, A and C), suggesting that oxidative stress causes PECAM-1 phosphorylation, resulting in Nck recruitment and Nck-dependent proinflammatory signaling through PAK2 and NF-κB (Fig. 6D).

Fig. 6 PECAM-1 is required for oxidative stress–induced proinflammatory responses.

(A to C) HAECs transfected with PECAM-1 siRNA were treated with H2O2 for the indicated times and analyzed for phosphorylated NF-κB (A and B) and VCAM-1 abundance (A and C) by Western blotting. Representative Western blots are shown. n = 5 independent experiments. ***P < 0.001, ****P < 0.0001. (D) Model of oxidative stress–induced activation of NF-κB through PECAM-1–dependent recruitment of Nck. I/R, ischemia/reperfusion; ROS, reactive oxygen species.

Nck inhibitors blunt oxidative stress–induced leukocyte–endothelial cell interactions during ischemia/reperfusion injury in vivo

To test the role of Nck in oxidative stress–driven inflammation in vivo, we used intravital microscopy to visualize leukocyte recruitment in the cremaster muscle model of ischemia/reperfusion injury. Because endothelial PAK signaling also regulates vascular permeability in multiple systems (2830), fluorescein isothiocyanate–labeled bovine serum albumin (FITC-BSA) was also used to measure endothelial barrier integrity. Ischemia/reperfusion injury induced both leukocyte adhesion (Fig. 7A) and emigration (Fig. 7B) over the course of the reperfusion period as compared to sham-operated controls. Mice receiving the Nck-blocking peptide but not the control peptide showed significantly attenuated ischemia/reperfusion-induced leukocyte adhesion and emigration to amounts comparable to those seen in sham-operated controls (Fig. 7, A and B). Similar to other models (2830), treatment with the Nck-blocking peptide also significantly blunted vascular permeability after ischemia/reperfusion injury (Fig. 7, C and D).

Fig. 7 The Nck-blocking peptide blunts inflammation and permeability during ischemia/reperfusion injury in vivo.

(A to C) Mice were pretreated with vehicle (veh), control peptide (ctrl pep), or the Nck-blocking peptide (Nck pep) and subjected to sham or ischemia/reperfusion (I/R) injury. Intravital microscopy was performed on cremasteric postcapillary venules. (A) Leukocyte adhesion within the venules (conveyed as the number of leukocytes adhering to the vessel wall per square millimeter), (B) leukocyte emigration out of the vessel (number of emigrated cells per square millimeter of interstitium), and (C) leakage of FITC-albumin (as a measure of vessel permeability) were measured before ischemia (pre) and throughout reperfusion. (D) Representative images of FITC-albumin leakage at 60-min reperfusion are shown. n = 4 to 8 mice per group. Statistical comparisons (Bonferroni, P ≤ 0.0033): *, comparing I/R-veh and I/R-ctrl pep to pretreatment values (P < 0.001); #, comparing I/R-veh and I/R-ctrl pep to all other groups (P < 0.001); &, comparing I/R-veh group to pretreatment (P < 0.001); ‡, comparing I/R-ctrl pep to pretreatment (P < 0.001); §, comparing I/R-veh to I/R-Nck pep (P < 0.01); and §, comparing I/R-ctrl pep to I/R-Nck pep (P < 0.01).

Together, these data suggest that blunting Nck signaling either by reduced expression or by competitive inhibition abrogates oxidative stress–induced inflammation after ischemia/reperfusion injury.


Oxidative stress in ischemia/reperfusion injury can induce both direct and indirect tissue damage by stimulating local inflammation. We characterized a role for the Nck adaptor proteins in critically coupling oxidative stress–induced tyrosine phosphorylation to endothelial NF-κB activation and proinflammatory gene expression. Nck inhibition with either siRNA or a membrane-permeable peptide significantly blunted oxidative stress–induced proinflammatory signaling through PAK2 and NF-κB, proinflammatory adhesion molecule expression (ICAM-1, VCAM-1), and leukocyte binding. Nck regulated proinflammatory responses to both exogenous oxidative stress (H2O2) and endogenous oxidative stress (hypoxia/reoxygenation injury), whereas Nck did not affect antioxidant gene expression or TNFα-mediated cell adhesion molecule expression, suggesting that Nck selectively couples oxidative stress to proinflammatory responses. Nck promoted these proinflammatory effects by coupling the tyrosine phosphorylation of PECAM-1 to the activation of PAK2, a Ser/Thr kinase that mediates oxidative stress–induced NF-κB activation. Consistent with the results obtained in these in vitro systems, treatment with a Nck inhibitory peptide completely ablates leukocyte recruitment and vascular permeability after ischemia/reperfusion injury in the cremaster muscle. Together, these data reveal that the Nck family of adaptor proteins couples oxidative stress to proinflammatory responses in endothelial cells and may serve as a therapeutic target to limit oxidative stress–induced inflammation.

Multiple studies demonstrate an important role for Nck in mediating inflammatory responses, primarily through Nck’s role in actin dynamics. Interaction between Nck and the T cell receptor (TCR) stimulates PAK-dependent actin polymerization, which is important for the formation of the immunological synapse (31, 32), and mice deficient in both Nck1 and Nck2 show reduced thymic selection and reduced TCR sensitivity (33, 34). Similarly, Nck modulates actin remodeling during phagocytosis in neutrophils and macrophages (35). Work from our group shows an important role for Nck in mediating endothelial activation (27); however, the mechanisms of action appear to be independent of actin remodeling responses. Nck is required for PAK2 membrane recruitment in endothelial cells exposed to hemodynamic shear stress, and inhibiting PAK2 signaling blunts shear stress–induced activation of NF-κB (20, 27). PAK2 drives NF-κB activation through an NF-κB–inducing kinase and an IKK-dependent pathway (20), and endothelial cells expressing a PAK2 construct showing enhanced Nck interactions demonstrate enhanced basal NF-κB signaling (27). We demonstrated that oxidative stress drives PAK2 activation by stimulating Nck-dependent membrane targeting. Although membrane targeting can promote PAK activation (19), activation of PAK typically requires interaction with the small guanosine triphosphatases Rac and Cdc42 (36). Whereas a role for Nck in oxidative stress–induced Rac or Cdc42 signaling has yet to be described, Nck can directly interact with the Rac guanine nucleotide exchange factor ELMO1 (engulfment and cell motility 1) (37), suggesting that Nck could localize membrane-targeted PAK to sites of increased Rac activation. The interplay between these pathways should provide an interesting direction for future research.

Although PECAM-1 has been ascribed both proinflammatory and anti-inflammatory functions (38, 39), inhibiting PECAM-1 reduces ischemia/reperfusion injury in rodent models (40). PECAM-1 is a sensor for oxidative stress, showing enhanced phosphorylation and SHP-2 recruitment after a rapid increase in cellular H2O2 production (41). Increased PECAM-1 tyrosine phosphorylation is correlated with Nck coimmunoprecipitation and Nck translocation to the plasma membrane, and PECAM-1 depletion by siRNA prevented oxidative stress–induced NF-κB activation and increases in VCAM-1 abundance. The amino acid sequences surrounding both Tyr663 and Tyr686 in the PECAM-1 cytoplasmic tail show similarities to Nck SH2 domain–binding sequences (42), and mutation of these two Tyr residues to Phe blunted the interaction between PECAM-1 and the Nck SH2 domain (Fig. 5). Nck can interact with several tyrosine-phosphorylated proteins, including PECAM-1, and can localize to multiple cellular sites, including cell-cell junctions and cell-matrix adhesions. PECAM-1 appeared to be critical for the proinflammatory signaling downstream of Nck recruitment, and we speculate that PECAM-1 may preferentially couple Nck to other important signaling partners or may alter the local substrates for PAK2. Consistent with this proinflammatory role, PECAM-1–deficient mice showed reduced leukocyte transmigration after interleukin-1β stimulation (43, 44). Given Nck’s ability to stimulate cytoskeletal remodeling, PECAM-1–mediated interactions between leukocytes and endothelial cells may use Nck-dependent signaling to enhance cytoskeletal remodeling during leukocyte diapedesis.

Modular adaptor proteins, such as Nck, facilitate highly specific protein-protein interactions to couple environmental stimuli to distinct intracellular signals and effector responses (45). We show that H2O2 treatment significantly enhances the interaction between Nck and PAK2 (Fig. 4E), an interaction that involves the second SH3 domain of Nck. This domain can interact in an intramolecular fashion with a (K/R)x(K/R)RxxS sequence in the linker region between the first and second SH3 domain, an interaction that lowers the affinity of the second SH3 domain for its binding partners, indicative of an inactive confirmation (46). Whereas direct oxidation of Nck has not been described, Nck can be phosphorylated on serine, threonine, and tyrosine residues (47), and the presence of a Ser residue in the linker sequence suggests that inducible phosphorylation at this site could enhance Nck’s affinity for binding partners. Platelet-derived growth factor stimulates the phosphorylation of Nck on both Ser and Tyr residues in this linker region between the first and second SH3 domain (48), although the function of these phosphorylation events is unknown. The effect of oxidative stress on Nck posttranslational modification and inducible recruitment of signaling intermediates should provide an interesting direction for future research.

Although cell culture and animal models implicate oxidative stress in cardiovascular inflammation, treating cardiovascular disease with exogenous antioxidants has not revealed therapeutic benefit in human trials (49). Whereas multiple factors may account for this deficiency, discrete protein-protein interactions and signaling processes that couple oxidative stress to inflammatory gene expression may provide a more stable therapeutic target and may provide a greater therapeutic benefit by leaving the oxidative stress–induced activation of negative feedback pathways intact. Because Nck inhibition does not blunt Nrf2-driven antioxidant genes, Nck-based therapeutics may prevent inflammation without reducing the preconditioning effect to subsequent ischemia/reperfusion injury. Furthermore, blocking PAK-Nck interactions reduces endothelial cell permeability to multiple stimuli (28, 30), suggesting that targeting Nck will provide an additional benefit compared to blocking NF-κB directly by reducing tissue edema. However, long-term Nck inhibition may also be detrimental because Nck inhibitors blunt endothelial cell migration and angiogenic remodeling (24, 50). Therefore, acute, reversible Nck inhibitors may provide the most therapeutic benefit for targeting acute inflammation while allowing preconditioning without blocking later tissue remodeling responses in chronically ischemic tissue.


Cell culture

HAECs were cultured in MCDB131 containing 10% fetal bovine serum (FBS), heparin (60 μg/ml), bovine brain extract (24 μg/ml), 2 mM l-glutamine, and antibiotics (Gibco) and used between passages 6 and 10. Nck siRNA (SMARTPool, Dharmacon) transfections were performed using Lipofectamine2000 (Invitrogen) per the manufacturer’s instructions. HAEC were treated with various doses of H2O2 or hypoxia under low-serum conditions (0.5% FBS). After initial dose requirements were determined, subsequent experiments used 100 μM H2O2. For hypoxia/reoxygenation experiments, HAECs were exposed to 5% O2 in a hypoxia chamber (Coy Laboratory Products) for 60 min before reoxygenation at 21% O2 for up to 2 hours. HEK293 cells, grown in Dulbecco’s modified Eagle’s medium containing 10% FBS and antibiotics, were transfected with wild-type PECAM-1 or PECAM-1 containing Tyr-to-Phe mutations in Tyr663 and Tyr686 (YYFF mutation; gift of D. Newman, Medical College of Wisconsin) (41). Membrane-permeable TAT-tagged peptides derived from the Nck-binding sequence in PAK1 (Nck peptide) were used at 20 μg/ml (24).


Western blotting was performed as previously described (20, 28). Primary antibodies included anti-PAK2, phospho-NF-κB (p65 subunit; Ser536), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ICAM-1 (Cell Signaling Technology), phosphotyrosine (4G10), phospho-PAK1/2/3 (Ser141; Invitrogen), Nck1/2, p65, VCAM-1, α5 integrin, and extracellular signal–regulated kinase 1/2 (Santa Cruz Biotechnology). Nck1- and Nck2-specific antibodies were provided by L. Larose (McGill University) (51). Densitometry was performed using ImageJ software.


Cells were fixed in phosphate-buffered serine (PBS) containing 4% formaldehyde, permeabilized in 0.1% Triton X-100, and blocked in 1% BSA containing 10% goat serum. Primary antibodies (incubated overnight) included rabbit anti–NF-κB p65 (Santa Cruz Biotechnology), mouse anti-phosphotyrosine (4G10; Millipore), rabbit anti-Nck1/2 (Millipore), mouse anti-Nck (Abcam), goat anti–PECAM-1 (Santa Cruz Biotechnology), and mouse anti–PECAM-1 (Cell Signaling Technology). Staining was visualized with Alexa-conjugated secondary antibodies (Invitrogen) and viewed on a Nikon Eclipse Ti inverted fluorescence microscope. Images were taken by using a Photometrics Coolsnap120 ES2 camera and analyzed by the NIS Elements BR 3.00, SP5 imaging software. More than 100 cells were analyzed for nuclear NF-κB staining for each condition of individual experiments. Proximity ligation assays were performed according to the manufacturer’s instructions; cells were counterstained with 4′,6-diamidino-2-phenylindole and phalloidin and quantified using NIS Elements software.


Cells were lysed for immunoprecipitation as previously described (27). Lysates were precleared for 15 min with γ-bind G beads and centrifuged. Cleared lysates were incubated for 3 hours with rabbit anti-Nck (Millipore), mouse anti-phosphotyrosine (4G10; Millipore), mouse anti–PECAM-1 (Cell Signaling Technology), or rabbit anti-IKKβ. Lysates were incubated with γ-bind G sepharose beads for 2 hours, and the beads were washed three times in lysis buffer. Immunoprecipitated proteins were dissociated from the beads by the addition of 2× sample buffer followed by boiling at 95°C for 2 min. For IKK activity assays, IKKβ immunoprecipitates were washed two times in kinase buffer [25 mM tris, 5 mM β-glycerophosphate, 2 mM dithiothreitol, 0.1 mM sodium orthovanadate (Na3VO4), 10 mM magnesium chloride; pH 7.5] and incubated for 30 min at 30°C with 1 μg of GST-IκBα in kinase buffer containing 200 μM adenosine 5′-triphosphate. Kinase activity was measured by Western blotting for phosphorylated IκBα and normalized to IKKβ abundance in the immunoprecipitates. SH2 pull-down assays used the GST-tagged Nck1 SH2 domain or the R308K mutant SH2 domain with reduced affinity for phosphotyrosine (gift of L. Larose, McGill University) (52). For this experiment, lysates were incubated for 2 hours with glutathione Sepharose beads preconjugated to GST-SH2 domain or GST-SH2 R308K, and subsequent interactions were analyzed by Western blotting as previously described.

Cytosol and membrane fractionation

Cells plated on 60-mm dishes were washed in ice-cold PBS and lysed in 200 μl of cytosol buffer [20 mM tris (pH 7.5), 2 mM 2-mercaptoethanol, 5 mM EGTA, 2 mM EDTA, 2 mM Na3VO4, and 1× protease inhibitor cocktail (Research Products International)]. Cell lysates were collected and centrifuged for 30 min at 15,000 rpm at 4°C. The supernatant was then collected as the cytosolic fraction. The remaining pellet was resuspended in 200 μl of cytosol buffer and centrifuged again for 30 min at 15,000 rpm at 4°C to completely remove any residual cytosolic proteins. The remaining pellet was resuspended in membrane buffer [50 mM tris (pH 8.0), 150 mM sodium chloride, 1% NP-40, 10 mM sodium fluoride, 2 mM Na3VO4, and 1× protease inhibitor] and spun for 30 min at 15,000 rpm at 4°C. The supernatant was collected as the membrane fraction and analyzed by Western blot for PAK2 and Nck. The membrane protein α5 integrin and the cytosol protein GAPDH were used to test isolation efficiency.

Monocyte adhesion assay

Human THP-1 monocytes labeled with Cell Tracker Green (Invitrogen) according to the manufacturer’s protocol were resuspended in Hanks’ balanced salt solution containing calcium and magnesium. Monocytes (5.0 × 106) were added to confluent HAEC cultures and allowed to attach for 15 min at 37°C. The unbound monocytes in the supernatant and two washes were collected. Bound monocytes were visualized by epifluorescence microscopy, and both bound and unbound monocytes were lysed in 200 mM NaOH. Fluorescence in the individual lysates was measured in a FLUOStar fluorospectrometer, and monocyte adhesion was calculated as percent change compared with untreated cells by normalizing to the unbound fraction.

Quantitative real-time PCR

Cells were lysed in TRIzol reagent (Invitrogen), and mRNA was extracted per manufacturer’s instructions. After complementary DNA (cDNA) synthesis (iScript cDNA Synthesis Kit, Bio-Rad), qRT-PCR was performed in a Bio-Rad iCycler using SYBR Green Master Mix (Bio-Rad). All primers used for qRT-PCR (table S1) were validated by melt curve analysis, gel electrophoresis of the PCR product, and sequencing of the PCR product. Expression of target genes was normalized to the housekeeping gene β2-microglobulin and conveyed as fold change compared to the untreated condition using the 2–ΔΔCt method.

Intravital microscopy

All animal protocols were approved by the Louisiana State University (LSU) Health Sciences Center Shreveport Animal Care and Use Committee and followed the National Institute of Health (NIH) guidelines for the care and use of laboratory animals. The cremaster muscle was prepared for light and fluorescence intravital microscopy as described previously (53). Briefly, 8-week-old male C57Bl/6J mice were anesthetized with ketamine hydrochloride (150 mg/kg) and xylazine (7.5 mg/kg) intraperitoneally, and the jugular vein was canulated. At this time, the mice received saline (vehicle), control peptide (100 μg, ~ 4 mg/kg), or PAK-Nck peptide (100 μg, ~ 4 mg/kg) through the canula. The cremaster muscle was then isolated, spread across a viewing pedestal, and superfused with warm bicarbonate buffered saline at 1 ml/min. The tissue was transilluminated, and red blood cell velocity (VRBC) was measured using an optical Doppler velocimeter. Postcapillary venules with a diameter between 20 and 40 μm were monitored, and after a 20- to 30-min stabilization period, a section of the least inflamed venule with a wall shear rate ≥500 s−1 was chosen for the study. Background fluorescence was recorded. The mouse then received FITC-albumin (25 mg/kg) in 0.1 ml of saline through the jugular vein canula. This was allowed to circulate before a 1-min recording of the venule was taken using light microscopy (to assess leukocyte recruitment), followed by a recording of the fluorescence image (for measurement of albumin leakage). Thereafter, a vascular clamp was used on the whole pedicle to induce ischemia for 30 min. The clamp was then released, and recordings of the venule using light and fluorescence microscopy were taken at 10-min intervals up to 60 min of reperfusion. Leukocyte adhesion and emigration were determined by off-line analysis. Leukocytes were considered adherent if they stopped for at least 30 s (conveyed as number of cells per square millimeter of vessel wall), and emigrated leukocytes were leukocytes identified in the interstitium (conveyed as number of cells per square millimeter of tissue). The index of vascular albumin leakage (permeability index) was determined from the ratio (interstitial intensity − background)/(venular intensity – background) as previously reported (54) and measured using ImageJ 1.46 software from the NIH.

Statistical analysis

Data were tested for normality (Kolmogorov-Smirnov test) and significance using GraphPad Prism software. Data that passed the normality assumption were analyzed using Student’s t test and one-way or two-way analysis of variance with Bonferroni posttests where indicated. Data that failed the normality assumption were analyzed using the nonparametric Mann-Whitney U test with post hoc analysis. Data are shown as mean values ± SEM. Differences were considered statistically significant at a value of P < 0.05.


Fig. S1. H2O2 does not stimulate Tyr phosphorylation of IκBα at concentrations that activate NF-κB in endothelial cells.

Fig. S2. Nck siRNA reduces H2O2-induced NF-κB activation and increase in VCAM-1 abundance.

Fig. S3. Micrographs of monocyte adhesion to oxidative stress–activated endothelial cells.

Fig. S4. Oxidative stress–induced tyrosine phosphorylation activates Nck-dependent PAK2 signaling.

Fig. S5. H2O2 stimulates Nck and PAK recruitment to PECAM-1.

Table S1. Quantitative real-time PCR primers.


Acknowledgments: We would like to thank L. Larose (McGill University) for providing the Nck1 and Nck2 antibodies and the GST-Nck1 SH2 constructs, D. Newman (Medical College of Wisconsin) for providing the PECAM-1 constructs, and S. Jain (LSU Health Sciences Center Shreveport) for providing the THP-1 monocytes. Funding: This work was supported by the NIH (R01 HL098435 to A.W.O., R01 HL113303 to C.G.K., and P20-GM103433 from the National Institute of General Medical Sciences to K.Y.S.), by the Louisiana Board of Regents Superior Toxicology Fellowship [LEQSF (2008-13)-FG-20 to A.Y.J.], and by an American Heart Association Postdoctoral Fellowship (12POST12030375 to J.C.) and Predoctoral Fellowship (14PRE18660003 to A.Y.J.). Author contributions: Experiments were performed by J.C., I.L.L., A.Y.J., and A.W.O., and data analysis was carried out by J.C., A.Y.J., B.T., K.Y.S., and A.W.O. Consultation on experimental design and setup was provided by I.L.L., C.G.K., and K.Y.S., and J.C., A.Y., C.G.K., K.Y.S., and A.W.O. contributed to the preparation of the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article