G protein–coupled receptors activate p38 MAPK via a non-canonical TAB1–TAB2– and TAB1–TAB3–dependent pathway in endothelial cells

Endothelial dysfunction is induced by inflammatory mediators including multiple G protein–coupled receptor (GPCR) agonists. However, the GPCR signaling pathways that promote endothelial dysfunction are incompletely understood. We previously showed that thrombin promotes endothelial barrier disruption through autophosphorylation and activation of p38 mitogen-activated protein kinase (MAPK) via a non-canonical transforming growth factor-β–activated protein kinase-1–binding protein-1 (TAB1) and TAB2–dependent pathway rather than the canonical three-tiered kinase cascade. Here, we sought to determine whether other GPCR agonists stimulate p38 MAPK activation via this non-canonical pathway in human endothelial cells derived from different vascular beds. Using primary human umbilical vein endothelial cells (HUVECs), HUVEC-derived EA.hy926 cells, and human dermal microvascular endothelial cells (HDMECs), we found that both non-canonical and canonical p38 activation pathways components are expressed in these various endothelial cell types, including TAB3, a structurally-related TAB2 homolog. Moreover, multiple GPCRs agonists, including thrombin, histamine, prostaglandin E2, and ADP, stimulated robust p38 autophosphorylation, whereas phosphorylation of the upstream MAPKs MAP kinase kinase 3 (MKK3) and MKK6, was virtually undetectable, indicating that non-canonical p38 activation may exist for other GPCRs. Indeed, in EA.hy926 cells, thrombin- and histamine-stimulated p38 activation depended on TAB1–TAB2, whereas in primary HUVECs, both TAB1–TAB2 and TAB1–TAB3 were required for p38 activation. In HDMECs, thrombin-induced p38 activation depended on TAB1–TAB3, but histamine-induced p38 activation required TAB1–TAB2. Moreover, thrombin- and histamine-stimulated interleukin-6 production required both TAB1–TAB2 and TAB1–TAB3 in HUVEC. We conclude that multiple GPCR agonists utilize non-canonical TAB1–TAB2 and TAB1–TAB3–dependent p38 activation to promote endothelial inflammatory responses.

Vascular endothelial dysfunction is caused by inflammatory mediators, many of which signal through G protein-coupled receptors (GPCRs) 4 (1,2). A single monolayer of endothelial cells lines the lumen of blood vessels and forms a semipermeable barrier that is controlled by the integrity of cell-cell junctions (3). Endothelial dysfunction results in production of inflammatory mediators and disruption of cell-cell junctions, vascular leakage, and tissue edema that contributes to organ failure (4,5). The activation of RhoA and phosphorylation of myosin light chain (MLC) by MLC kinase are known mediators of thrombin-induced endothelial barrier disruption (6), however, we and others have shown that thrombin-stimulated p38 mitogen-activated protein kinase (MAPK) activation also controls endothelial barrier permeability through a pathway that is independent MLC phosphorylation and presumably RhoA signaling (7,8). Thus, GPCRs utilize multiple signaling pathways to promote endothelial dysfunction.
The p38 MAPK is a critical mediator of inflammation (9). However, despite enormous efforts, pharmacological inhibitors of p38 that generally target the ATP-binding site have failed to advance in clinical trials due to adverse side effects such as skin and liver toxicity. Thus, a better mechanistic understanding of the activators of p38-driven inflammatory responses is critical for discovery of new strategies to manipulate p38 signaling in various disease settings. All four p38 isoforms (␣, ␤, ␥, and ␦) are activated by the canonical three-tiered kinase cascade mediated by upstream MAP2Ks and MAP3Ks (10,11). Interestingly, however, authenticated non-canonical pathways for p38 activation also exist. The direct binding of transforming growth factor-␤-activated kinase-1-binding protein-1 (TAB1) to p38␣ induces autophosphorylation and autoactivation and bypasses the requirement for upstream MAP2Ks, mitogen-activated protein kinase kinase-3 (MKK3) and MKK6 (12,13). The physiological relevance of the TAB1-p38␣ pathway has been demonstrated in vivo and shown to function in inflammation, cardiotoxicity, and myocardial ischemia (14 -16). A different non-canonical pathway for p38 activation is mediated by ZAP-70 binding, which results in p38␣ and -␤ autophosphorylation and activation in immune T cells (17). Although it is presumed that GPCRs activate p38 through the three-tiered kinase cascade there is limited supportive evidence (18). In fact, several studies have shown that GPCRs stimulate p38 MAPK activation through diverse G s , G q , and G 13 signaling pathways (18), however, rarely has the function of MAPK2Ks been directly examined (19).
In previous studies, we showed that activation of proteaseactivated receptor-1 (PAR1), a GPCR for the coagulant protease thrombin, in endothelial cells promotes p38␣ activation via a TAB1-dependent pathway and is independent of upstream MAP2Ks, MKK3, and MKK6 (8). We also showed that ubiquitination of activated PAR1 drives recruitment of TAB2, an adaptor protein that binds TAB1 (20) and contains a Npl4 zinc finger (NZF) domain that binds K63-linked ubiquitin (21). The ubiquitin binding capacity of TAB2 and p38␣ binding determinants for TAB1 are both required for thrombin-stimulated p38 signaling (8). TAB3 is a structurally related homolog of TAB2 that can also bind ubiquitin and mediate inflammatory signaling (22,23). Ubiquitin-driven p38 signaling induced by thrombin-activated PAR1 further promotes endothelial barrier permeability in vitro and p38 activity is required for PAR1stimulated vascular leakage in vivo (8). Thus, PAR1 stimulates p38 inflammatory signaling via a non-canonical TAB1-TAB2dependent pathway in endothelial cells, however, it is not known if this pathway is broadly applicable to other GPCRs expressed in endothelial cell types derived from different vascular beds.
In this study, we sought to determine whether non-canonical TAB1-dependent p38 activation is induced by other GPCRs in a panel of extensively studied endothelial cell models including human endothelial cells of venous macrovascular origin, human endothelial vein umbilical cells (HUVECs), and HUVEC-derived EA.hy926 cells, and human dermal microvascular endothelial cells (HDMECs). We found that critical components of the canonical and non-canonical p38 -activation pathways are expressed in these endothelial cell types, and multiple GPCRs agonists including thrombin, histamine, prostaglandin E 2 (PGE 2 ), and ADP, stimulated non-canonical p38 autophosphorylation and activation. In addition, whereas all GPCR agonists stimulated robust p38 activation, each displayed a unique requirement for either TAB1-TAB2 or TAB1-TAB3 for p38 activation in distinct endothelial cells types. Thrombin and histamine also stimulated production of the inflammatory mediator interleukin-6 (IL-6) via a TAB1-dependent pathway, suggesting that noncanonical activation of p38 inflammatory signaling is important for multiple GPCR agonists.

Non-canonical activation of p38 by GPCRs
were easily detected in HUVEC-derived EA.hy926 cells (Fig.  1A, lane 1), primary HDMECs (Fig. 1A, lane 2), and primary HUVECs (Fig. 1A, lane 3) by immunoblot analysis and realtime quantitative (q) PCR (Fig. 1, B and C). However, the various cell lines showed differences in expression of the individual components relative to each other. Notably, EA.hy926 cells displayed a significantly higher amount of TAB1 compared with HDMEC, TAB2 compared with HUVEC, and TAB3 and MKK6 compared with both HDMEC and HUVEC based on both immunoblot and quantitative PCR (qPCR) analysis ( Fig. 1, A-C). In contrast, there was no significant difference in expression of TAB1 in EA.hy926 cells versus HUVEC, TAB2 in EA.hy926 versus HDMEC, whereas MKK3 and p38 exhibited comparable expression in all endothelial cell lines detected by immunoblotting (Fig. 1, A and B). Interestingly, however, qPCR analysis of individual mRNA transcripts encoding the p38-activation pathway components revealed a significant higher abundance of TAB1, TAB2, TAB3, MKK3, and MKK6 mRNA transcripts in EA.hy926 cells compared with HDMECs and HUVECs, whereas the levels of p38 MAPK mRNA transcripts were comparable across all endothelial cell types similar to that observed with p38 protein (Fig. 1, B and C). These findings likely reflect higher transcriptional activity in EA.hy926 cells versus HDMECs and HUVECs as well as possibly differences in protein stability of the individual components, which has been clearly demonstrated for TAB1 (8), a critical mediator of noncanonical p38 activation.

GPCR agonists induce significant p38 phosphorylation but limited induction of MKK3/MKK6 phosphorylation in human endothelial cells
To assess the capacity of endothelial GPCRs to activate p38 MAPK signaling via the non-canonical versus the canonical MKK3/MKK6 pathway, we first examined the extent of four different agonists that signal through GPCRs to induce phosphorylation of p38 and MKK3/MKK6 in HUVEC-derived EA.hy926 cells. PAR1 is the major GPCR for thrombin in HUVEC and HUVEC-derived EA.hy926 cells (25,26), whereas histamine signals through H1 and H2 GPCRs in HUVEC and EA.hy926 cells (27,28). ADP signaling is primarily mediated by the P2Y1 GPCR in HUVEC (29), whereas PGE 2 responses are mediated primarily by the EP4 prostanoid GPCR in endothelial cells (30,31).
Thrombin induced a robust ϳ5-fold increase in p38 phosphorylation at 2.5 and 5 min ( Fig. 2A, lanes 1-3, and B), with no significant change in the phosphorylation status of MKK3 or MKK6 (Fig. 2, A, lanes 1-3, C and D), consistent with our previously published studies (8). Histamine also caused a significant ϳ5-fold increase in p38 phosphorylation detected at 5 and 7.5 min following agonist incubation ( Fig. 2A, lanes 4 -6, and B). Although histamine failed to induce a marked change in MKK3 phosphorylation at either time point (Fig. 2, A, lanes 4 -6, and C), a significant ϳ1.5-fold increase in MKK6 phosphorylation was detected after 5 min of agonist stimulation (Fig. 2, A, lanes 4 -6, and D). Cells stimulated with PGE 2 also showed an ϳ2-3-fold increase in p38 phosphorylation at 5 and 10 min (Fig. 2, A, lanes 7-9, and B), and a modest but significant increase in MKK3 and MKK6 phosphorylation at 10 min (Fig. 2, A, lanes 7-9, C and D). Incubation with ADP also caused a significant but modest ϳ1.5-fold increase in p38 phosphorylation (Fig. 2, A, lanes 10 -12, and B), with no significant change in MKK3 phosphorylation, whereas a significant decrease in MKK6 phosphorylation was detected (Fig. 2, A, lanes 10 -12, C and D). These data demonstrate that multiple GPCR agonists consistently stimulate robust p38 phosphorylation in HUVEC-

Non-canonical activation of p38 by GPCRs
derived EA.hy926 cells with modest and varying differences in MKK3 and MKK6 phosphorylation.
We next determined if GPCR agonists induced a similar trend in p38, MKK3, and MKK6 phosphorylation in primary HUVECs. Thrombin stimulated a significant ϳ7-fold increase in phosphorylation of p38 with no change in the phosphorylation status of MKK3 or MKK6 (Fig. 3, A, lanes 1-3, B-D), like that observed in HUVEC-derived EA.hy926 cells. Similar to thrombin, histamine also caused a significant ϳ6-fold increase in p38 phosphorylation without affecting the MKK3 or MKK6 phosphorylation ( Fig. 3, A, lanes 7-9, B-D). Cells stimulated with ADP and PGE 2 displayed a modest but significant ϳ2-fold increase in p38 phosphorylation, and no change in MKK3 or MKK6 phosphorylation (Fig. 3, A, lanes 4 -6 and 10 -12, B-D). Taken together these data suggest that whereas multiple GPCR agonists stimulate activation of p38 MAPK, they fail to consistently induce MKK3 or MKK6 phosphorylation in primary HUVEC and HUVEC-derived EA.hy926 cells, suggesting that p38 activation may occur through non-canonical pathways in different endothelial cell types.
To ensure that phosphorylation of MKK3 and MKK6 is responsive to stimuli in HUVEC and HUVEC-derived EA.hy926 cells, cells were treated with vascular endothelial growth factor (VEGF). VEGF has been previously shown to activate p38 MAPK through an MKK3-and MKK6 -dependent pathway in endothelial cells (32). A significant ϳ10-fold increase in p38 phosphorylation was induced by VEGF at 10 min, which was diminished after 15 min of stimulation in EA.hy926 cells (Fig. 4, A and B). VEGF-induced p38 phosphorylation coincided with a significant ϳ1.5-fold increase in phosphorylation of MKK3 and MKK6 in EA.hy926 cells detected at 10 min (Fig. 4, A, C, and D). HUVECs stimulated with VEGF resulted in a similar significant ϳ12-fold increase in p38 phosphorylation and ϳ1.5-fold increase in MKK3 and MKK6 phosphorylation after 10 min of stimulation (Fig. 4, E-H). These findings indicate that both MKK3 and MKK6 phosphorylation occurs in response to growth factor stimulation in HUVECs and HUVEC-derived EA.hy926 cells. Thus, the failure of certain GPCR agonists to promote MKK3 or MKK6 phosphorylation consistently in various endothelial cell types suggest that p38 activation may not depend exclusively on activation of the upstream MAP2Ks, MKK3 and MKK6.

Multiple GPCR agonists induce p38 autophosphorylation
Activation of p38 MAPK occurs through direct phosphorylation mediated by upstream MAP2Ks, MKK3, and MKK6, or through direct binding of TAB1 to p38␣ resulting in autophosphorylation and activation. To examine if GPCR-induced p38 phosphorylation is mediated by upstream MAP2Ks or through p38 autophosphorylation, cells were pretreated with the p38 inhibitor SB203580. SB203580 specifically blocks the catalytic activity of p38␣ and -␤ isoforms inhibiting autophosphorylation without effecting phosphorylation of p38 mediated by upstream MAP2Ks (12). HUVEC-derived EA.hy926 cells were pretreated with SB203580 and then incubated with thrombin for various times. In control DMSO vehicle-treated cells, thrombin-induced a significant ϳ2.5-fold increase in p38 activation (Fig. 5A, lanes 1-3), which was significantly inhibited in endothelial cells pretreated with SB203580 (Fig. 5A, lanes 4 -6). These data are consistent with our previously published results (8). Similarly, pretreatment of endothelial EA.hy926 cells with SB203580 resulted in significant inhibition of histamine-, PGE 2 -, and ADP-induced p38 activation measured after various times of agonist incubation compared with control treated Non-canonical activation of p38 by GPCRs cells (Fig. 5, B-D, lanes 1-3 versus 4 -6). These findings suggest that p38 activation induced by multiple GPCR agonists occurs through autophosphorylation in EA.hy926 cells.
Primary HDMEC were next used to assess the impact of SB203580 on GPCR agonists on p38 autophosphorylation. However, unlike HUVEC, the subtypes of GPCRs that mediate responses to thrombin, histamine, ADP, and PGE 2 in HDMECs are less characterized but likely to be similar (34 -36). In contrast to HUVEC-derived EA.hy926 and primary HUVEC, pretreatment of HDMEC with SB203580 caused a significant decrease in basal p38 phosphorylation compared with control cells (Fig. 7, A-D, lanes 1 and 4), suggesting that detectable basal p38 phosphorylation in HDMEC is mediated predominantly by autophosphorylation. Despite a reduction in basal p38 phosphorylation, pretreatment with the SB203580 p38 inhibitor virtually abolished thrombin, histamine-, PGE 2 -, and ADP-induced p38 phosphorylation compared with control treated HDMECs (Fig. 7, A-D, lanes 1-3 versus 4 -6). Collec-

Non-canonical activation of p38 by GPCRs
tively, these data suggest that several GPCR agonists induce p38 activation via autophosphorylation independent of upstream MKK3 or MKK6 phosphorylation in multiple endothelial cell lines.

TAB1-TAB2 or TAB1-TAB3 are required for p38 activation induced by multiple GPCR agonists
TAB1 directly binds to p38␣ inducing autophosphorylation and activation (12). TAB2 associates with TAB1 (20), and TAB3 is a structurally related homolog of TAB2 that can also bind TAB1 (22). TAB1, TAB2, and TAB3 are all expressed in endothelial cells (Fig. 1), however, the relative contribution of TAB1-TAB2 versus TAB1-TAB3 to GPCR-induced p38 activation has not been previously determined. We showed that thrombin-stimulated p38 activation requires TAB1 and TAB2 in EA.hy926 cells but whether thrombin or other GPCRs agonists utilize TAB1 and TAB2 or TAB3 for activation of p38 in other endothelial cell types is not known. As expected, co-depletion of TAB1 and TAB2 by siRNA in EA.hy926 cells caused a significant ϳ40% inhibition of thrombin-induced p38 activation compared with nonspecific siRNA-transfected control cells (Fig. 8A, lanes 1 and 4 versus 2 and 5), whereas co-depletion of TAB1 and TAB3 by siRNA failed to significantly affect thrombin-induced p38 activation compared with nonspecific siRNA control (Fig. 8A, lanes 1 and 4 versus 3 and 6). Next, we examined the role of TAB1-TAB2 versus TAB1-TAB3 in histamine-induced p38 activation in EA.hy926 cells. Co-depletion of TAB1 and TAB2 by siRNA caused a modest but significant ϳ35% reduction in p38 activation stimulated by histamine (Fig.  8B, lanes 1 and 4 versus 2 and 5). Although siRNA-targeted depletion of TAB1 and TAB3 caused an apparent significant reduction in p38 activation, the data were variable and not significant (Fig. 8B, lanes 1 and 4 versus 3 and 6). These data suggest that multiple GPCR agonists utilize TAB1 and TAB2 to promote non-canonical p38 activation in EA.hy926 cells.
To determine whether thrombin and histamine induced p38 activation via the non-canonical TAB1-dependent pathway in primary endothelial cells we examined HUVECs and HDMECs. In TAB1-and TAB2-deficient HUVEC, thrombin-induced p38 activation was reduced significantly by ϳ70% compared with

Non-canonical activation of p38 by GPCRs
nonspecific siRNA-transfected control cells (Fig. 9A, lanes 1  and 4 versus 2 and 5). In contrast to EA.hy926 cells, co-depletion of TAB1 and TAB3 by siRNA in HUVEC also resulted in a significant ϳ55% reduction in thrombin-induced p38 activation compared with control cells (Fig. 9A, lanes 1 and 4 versus 3  and 6). Similar to thrombin, histamine-induced p38 activation was diminished significantly in both TAB1-TAB2 and TAB1-TAB3 co-depleted cells, ϳ50 and ϳ60% respectively, compared with control cells (Fig. 9B, lanes 1 and 4 versus 2 and 5 and 3 and  6), suggesting that TAB3 also makes important contributions to p38 activation induced by GPCRs in endothelial cells.

Non-canonical TAB1-dependent p38 activation is required for IL-6 production induced by GPCR agonists in endothelial cells
IL-6 is a potent multifunctional cytokine secreted by endothelial cells following stimulation by various GPCRs agonists including thrombin and histamine (37,38). However, the role and relevance of GPCR-stimulated TAB1-dependent p38 activation in IL-6 production has not been determined and was examined by measuring IL-6 production in HUVECs transfected with siRNAs targeting TAB1 and TAB2 or TAB1 and TAB3. Thrombin induced a ϳ2-fold increase in IL-6 production that was significantly reduced in endothelial cells co-depleted of TAB1-TAB3 (Fig. 11A), consistent with thrombininduced TAB1-TAB3-dependent activation of p38 in HUVEC (Fig. 9A). Despite the apparent decrease in IL-6 production stimulated by thrombin in HUVECs co-depleted of TAB1-TAB2, the results were variable and not significant (Fig. 11A). In contrast, co-depletion of either TAB1-TAB2 or TAB1-TAB3 resulted in a marked and significant decrease in IL-6 production stimulated by histamine compared with nonspecific siRNA-transfected control cells (Fig. 11A). These findings are consistent with a role for both TAB1-TAB2 and TAB1-TAB3 in histamine-stimulated p38 activation in HUVEC (Fig.  9B). Together, these findings indicate that GPCR agonists induce p38 proinflammatory signaling in various endothelial cell types via a non-canonical TAB1-dependent pathway.

Discussion
Vascular endothelial dysfunction is induced by various inflammatory mediators, and many signal through activation of GPCRs. In previous work, we showed that thrombin induced

Non-canonical activation of p38 by GPCRs
endothelial barrier disruption, a hallmark of inflammation, through non-canonical TAB1-dependent p38 autophosphorylation and activation (8). Ubiquitination of PAR1 was also shown to be critical for initiating the recruitment of TAB2, TAB1, and p38 activation in HUVEC-derived EA.hy926 cells. The P2Y 1 receptor was further shown to activate p38 signaling through a ubiquitin-dependent non-canonical TAB1-dependent pathway in HeLa cells (8). However, it is not known if other GPCR agonists promote non-canonical p38 activation and inflammatory signaling in endothelial cells derived from different vascular beds. Here, we report that multiple GPCR agonists induce p38 autophosphorylation and activation via a TAB1-TAB2 and TAB1-TAB3 dependent pathway rather than the three-tiered kinase cascade to promote inflammatory responses in three distinct endothelial cell types.
MAPKs exist as distinct signaling cascades, comprised of three evolutionarily conserved, sequential acting kinases including a MAPK, MAP2K, and MAP3K. MAP3Ks are typically activated by phosphorylation mediated by upstream MAP2Ks. In contrast to canonical MAPK cascades, the p38␣ isoform can also be autoactivated through its interaction with TAB1 or through phosphorylation facilitated by the tyrosine kinase Zap70 (12, 17, 39). Although MKK3 and MKK6 are the major upstream MAP2Ks for p38␣ activation in response to cytokines or stress (40,41), we found that multiple GPCR agonists failed to induce consistent and robust phosphorylation of MKK3 or MKK6 in distinct endothelial cell types compared with VEGF. We also previously showed the siRNA targeted depletion of MKK3 and MKK6 failed to inhibit thrombin-induced p38 activation (8), suggesting that upstream MAP2Ks are not the major pathway for GPCR-stimulated p38␣ activation. In addition, the p38␣and ␤-selective inhibitor SB203580 virtually abolished p38 phosphorylation induced by multiple GPCR agonists in endothelial cells, whereas hyperosmolar NaCl-induced p38 phosphorylation remained intact. These data indicate that the catalytic activity of p38 is important of activation induced by GPCRs. Finally, co-depletion of either TAB1-TAB2 or TAB1-TAB3 caused a significant loss of GPCR-induced p38 phosphorylation in primary endothelial cells including HUVECs and HDMECs, as well as HUVEC-derived EA.hy926 cells. Although activation of p38␣ by the threetiered kinase cascade is the presumed major pathway for many inflammatory mediators, our data provide compelling evidence that TAB1-mediated p38 autoactivation is the predominate pathway utilized by mammalian GPCRs in human cultured endothelial cells and reveal a new paradigm by which GPCRs stimulate p38 MAPK inflammatory signaling.

Non-canonical activation of p38 by GPCRs
We previously showed that agonist-induced ubiquitination of PAR1, the GPCR for thrombin, is required for the recruitment of TAB2 and p38 signaling in HUVEC-derived EA.hy926 cells (8). TAB2 is an adaptor protein that binds ubiquitin via its NZF domain, and associates with TAB1 (20, 21). TAB3 is a structurally-related homolog of TAB2, contains an NZF domain, and can also bind ubiquitin (22,23). In addition to PAR1, the P2Y 1 receptor, a GPCR for ADP, also requires agonist-induced receptor ubiquitination and TAB1-TAB2 to stimulate p38 activation in HeLa cells (8). Here, we now show that ADP, histamine, and PGE 2 utilize a TAB1-TAB2 and TAB1-TAB3-dependent pathway for p38 activation. However, it is not known if the endothelial expressed P2Y 1 , H 1 and H 2 histamine receptors, or the endothelial PGE 2 receptors (EP 2 , EP 3 , or EP 4 ) are modified with ubiquitin following agonist stimulation and if receptor ubiquitination is required for either TAB1-TAB2 or TAB1-TAB3-dependent p38 activation. It also remains to be determined if TAB3 is recruited to activated GPCRs via an ubiquitin-dependent process similar to TAB2 and associates with TAB1 to promote p38 activation and necessitates further exploration.
The PAR1 ubiquitin-driven TAB1-mediated mechanism for p38 MAPK activation illustrates a new function for ubiquitin in the regulation of GPCR biology. Our findings now indicate that ubiquitin-driven signaling is likely relevant for at least four different GPCRs from distinct receptor families expressed in endothelial cells. The covalent attachment of ubiquitin to GPCRs is mediated by the sequential actions of E1, E2, and E3 ubiquitin-conjugating enzymes. The neural precursor cell expressed, developmentally down-regulated-4 (NEDD4) family

Non-canonical activation of p38 by GPCRs
of nine E3 ligases are primarily responsible for GPCR ubiquitination, currently reported for about 40 different GPCRs (42). However, in most cases it remains unclear how NEDD4 E3 ligase activity is regulated to facilitate GPCR ubiquitination. In recent work, we showed that thrombin activation of PAR1 stimulates c-Src-mediated tyrosine phosphorylation and activation of NEDD4-2 to promote p38 signaling and endothelial barrier disruption (43). Moreover, the P2Y 1 also required c-Src and NEDD4-2 tyrosine phosphorylation for p38 activation in endothelial cells (43). However, it remains to be determined whether other endothelial GPCR agonists utilize a conserved c-Src-dependent mechanism to control NEDD4 E3 ligase activity similarly to promote p38 MAPK-induced inflammatory responses in endothelial cells.
In summary, our findings indicate that multiple endothelial GPCR agonists stimulate p38 activation through a TAB1-dependent mechanism rather than the three-tiered kinase cascade in various endothelial cell types. However, the dependence on TAB1-TAB2 versus TAB1-TAB3 varies depending on the endothelial cell type and GPCR agonist. The underlying basis for the dependence on TAB1-TAB2 versus TAB1-TAB3 is not known and requires further investigation. In addition, the mechanisms by which GPCR-stimulated TAB1-induced p38 MAPK signaling controls various inflammatory responses including induction of cytokine production and endothelial barrier disruption remains poorly understood and warrants further exploration. Finally, in vivo studies have documented the relevance of TAB1-p38␣ activation in various disease settings including myocardial ischemia (13,16,44), indicating that TAB1-dependent p38 activation will be likely important for endothelial dysfunction in vivo. A recent paper further showed that disruption of TAB1-p38␣ interaction in vivo reduced myocardial ischemic injury (14), suggesting that the development of small molecules targeting the TAB1-p38␣ interface may provide a specific therapeutic intervention with limited side effects, not otherwise achievable with global p38 inhibition.

Cell lines
Human umbilical vein endothelial cell-derived EA.hy926 cells were grown and cultured as previously described (24). Primary HUVECs and HDMECs were purchased from Lonza (Basel, Switzerland), grown in human endothelial cell media or human microvascular endothelial cell media per the manufacturer's instructions, expanded and frozen at passage 2. Primary cells were used from these pools for all experiments up to passage 6.

Immunoblotting
EA.hy926 endothelial cells were seeded into 24-well plates, transfected, and grown as described above. Primary HUVECs and HDMECs were seeded into wells pre-coated with 10 g/cm of rat tail collagen, Type IV. In some experiments, cells were pretreated with 1 M SB203580, a p38␣ and ␤ selective inhibitor, for 30 min at 37°C prior to agonist stimulation. Cells were then lysed, and protein concentrations were determined using a BCA protein assay (ThermoFisher Scientific), diluted in 1ϫ Laemmli sample buffer containing 100 mM DTT and sonicated at 10% amplitude. Equivalent amounts of cell lysates were loaded and resolved by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and incubated with specific antibodies as indicated. The range of protein concentrations loaded for the individual experiments varied between experiments and ranged from 6 to 10 g. Membranes were developed by chemiluminescence and quantified by densitometry using ImageJ software.

Data analysis
Data were analyzed using Prism 7.0 software (GraphPad Software, La Jolla, CA) and statistical significance determined using Student's t test and one-way analysis of variance, as indicted.