Endothelial MAPKs Direct ICAM-1 Signaling to Divergent Inflammatory Functions

Abstract

Lymphocyte transendothelial migration (TEM) is critically dependent on intraendothelial signaling triggered by adhesion to ICAM-1. Here we show that endothelial MAPKs ERK, p38, and JNK mediate diapedesis-related and diapedesis-unrelated functions of ICAM-1 in cerebral and dermal microvascular endothelial cells (MVECs). All three MAPKs were activated by ICAM-1 engagement, either through lymphocyte adhesion or Ab-mediated clustering. MAPKs were involved in ICAM-1-dependent expression of TNF-α in cerebral and dermal MVECs, and CXCL8, CCL3, CCL4, VCAM-1, and cyclooxygenase 2 (COX-2) in cerebral MVECs. Endothelial JNK and to a much lesser degree p38 were the principal MAPKs involved in facilitating diapedesis of CD4⁺ lymphocytes across both types of MVECs, whereas ERK was additionally required for TEM across dermal MVECs. JNK activity was critical for ICAM-1-induced F-actin rearrangements. Furthermore, activation of endothelial ICAM-1/JNK led to phosphorylation of paxillin, its association with VE-cadherin, and internalization of the latter. Importantly ICAM-1-induced phosphorylation of paxillin was required for lymphocyte TEM and converged functionally with VE-cadherin phosphorylation. Taken together we conclude that during lymphocyte TEM, ICAM-1 signaling diverges into pathways regulating lymphocyte diapedesis, and other pathways modulating gene expression thereby contributing to the long-term inflammatory response of the endothelium.

FIGURE 1.

Endothelial MAPK activation in response to lymphocyte adhesion. (A) All three MAPKs were activated in GPNT ECs cocultured with Con A–activated, nonmigratory rat PLNCs. Shown are representative immunoblots of MAPKs phosphorylation alongside tubulin loading controls and normalized densitometric quantification of three independent experiments. (B and C) MAPK activation in GPNT in 30 min cocultures was reduced when PLNCs were preincubated with function-blocking anti–LFA-1 Abs but not an anti–VLA-4 blocking Ab. Shown are representative blots and densitometric quantification of three independent experiments. Control phosphorylation levels (in response to PLNC adhesion without adhesion molecule neutralization) were set to 100%. Data were compared with the corresponding time 0 controls and significant differences are indicated. In (C), white separation lines indicate where lanes from the same blots were joined. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 2.

Endothelial MAPK activation in response to ICAM-1 ligation and cross-linking. (A) GPNT were subjected to ICAM-1 cross-linking (XL), with secondary clustering, for the indicated length of time and MAPK phosphorylation analyzed. Representative results and quantification of kinase activation (normalized mean ± SEM) from three independent experiments are shown. (B) Representative analysis of MAPK phosphorylation following ICAM-1 ligation (without secondary clustering) for the indicated times and densitometric quantification of kinase activation of three independent experiments. MAPK levels were not significantly affected by ICAM-1 ligation or cross-linking (Supplemental Fig. 1A, 1B). (C) Postconfluent, serum-starved GPNT cells were either left untreated or incubated with 5 μg/ml anti–ICAM-1 (1A29) or isotype-matched control IgG for 10 min. (D–F) Primary rat brain MVEC (D) or hCMEC/D3 (E) or human dermal MVEC (F) were stimulated by ICAM-1 cross-linking (XL) or ICAM-1 ligation for the indicated times and MAPK phosphorylation analyzed as described in (B). Data were compared with the corresponding time 0 controls and significant differences are indicated. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 3.

Role of MAPK in ICAM-1–mediated gene expression. (A) Total RNA was isolated from untreated (NT) or 4 h ICAM-1 cross-linked (XL) GPNTs and analyzed by semiquantitative RT-PCR for message levels of VCAM-1, TNF-α, COX-2, CCL2, ICAM-1, and GAPDH. (B and C) Confluent hCMEC/D3 (B) or human dermal MVEC (hDMEC) (C) were either left untreated or subjected to ICAM-1 cross-linking. At the indicated times TNF-α concentration in the culture supernatant was determined by ELISA. Shown are mean levels ± SEM of TNF-α above those in control cells from three independent experiments. (D and E) hCMEC/D3 (D) or human dermal MVEC (hDMEC) (E) were left untreated (NT) or subjected to ICAM-1 cross-linking (XL) for 24 h. The concentration of CXCL8, CXCL10, CCL2, CCL3, and CCL4 in the supernatant was measured by multianalyte flow assay. Where indicated anti–TNF-α (1 μg/ml) was included during the stimulation period to determine if altered chemokine secretion was a consequence of TNF-α induction. Shown are mean concentrations ± SEM of chemokines in the culture supernatant as determined from three independent experiments. (F) Confluent GPNT ECs were either left untreated (NT) or treated with 200 U/ml TNF-α for 12 h or subjected to ICAM-1 cross-linking (XL) for 12 h prior to immunoblot analysis of VCAM-1 and tubulin. Shown is a representative blot and densitometric quantification of three independent experiments. Where indicated cross-linking was performed in the presence of 50 μM U0126, SP600125, or SB202190. White separation lines indicate where lanes from the same blots were joined. (G) GPNT ECs were transiently nucleofected with pVCAM-1-luc, reseeded and allowed to grow to confluence (48–72 h posttransfection), and then subjected to ICAM-1 cross-linking (XL) for 6 h or stimulation with IL-1β (IL1β). Where indicated cross-linking was performed in the presence of 50 μM U0126, SP600125, or SB202190. Mean ± SEM luciferase activity (relative to untreated cells) was determined from three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4.

ICAM-1–induced p38-mediated message stabilization. (A) Levels of COX-2, TNF-α and GAPDH transcripts were determined by semiquantitative RT-PCR analysis in response to ICAM-1 cross-linking (XL) for 4 h. Where indicated cross-linking was performed in the presence of 50 μM U0126, SP600125, or SB202190. (B and C) Confluent GPNT cells were serum starved and either left untreated (NT, filled squares) or ICAM-1 cross-linked (XL, filled triangles) in the absence or presence of 50 μM SB202190 (XL + SB, open triangles). Ten micrograms per milliliter actinomycin D was added to block transcriptional activity and total RNA was isolated after 0, 2.5, and 5 h. Subsequently transcript levels of TNF-α and COX-2 were determined by quantitative RT-PCR. The amount of each transcript was quantified by densitometry, normalized, plotted, and analyzed by linear regression and ANOVA. The values are mean ± SEM of seven (TNF-α) and five (COX-2) independent experiments. **p < 0.01, ***p < 0.001.

FIGURE 5.

Endothelial JNK regulates lymphocyte TEM. (A) GPNT monolayers were pretreated or not with actinomycin D (Act D, 5 μg/ml) and subsequent TEM of Th1 lymphocytes (PAS, see also Supplemental Fig. 3A) measured after 30 min. Whereas Act D did not affect 30 min TEM rates, it inhibited all subsequent TEM events (measured up until 4 h). (B) TEM assay as in (A) with the exception that GPNT monolayers were left untreated (NT) or treated with 50 μM U0126 (U0), SP600125 (SP) or SB202190 (SB) for 1 h prior to a 30 min TEM assay. Due to the high washout rate of U0126 from GPNT cells (see Supplemental Fig. 2B), TEM experiments were also conducted with U0126 present throughout. However, even under these conditions TEM was not inhibited (data not shown). (C) TEM assay as in (B) except that primary rat brain MVEC were either left untreated (NT) or pretreated with 50 μM SP600125 for 1 h prior to addition of Ag-specific T lymphocytes. (D) TEM assay as in (B) with the exception that GPNT were either left untreated (NT) or treated with 1 μM L-JNKi for 1 h prior to the addition of T lymphocytes. (E) TEM assay as in (B) except that GPNT cells were transfected with wild-type (WT) or dominant-negative (DN) JNK1, JNK2, or MKK7 48 h before TEM and adhesion were analyzed. (F and G) TEM assay as in (B) except that TEM of human CD4⁺ cells across hCMEC/D3 (F) or human dermal MVEC (G) was measured following EC pretreatment with 50 μM U0126 (U0), SP600125 (SP), or SB202190 (SB) for 1 h. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 6.

Role of Src, Rho GTPase, and PKC in ICAM-1–mediated MAPK activation and lymphocyte TEM. Postconfluent, serum-starved GPNT cells were either left untreated (NT) or pretreated with 10 μM PP2 for 30 min (A), 10 μg/ml C3 transferase for 12 h (B), or 20 μM Gö6983 (Gö) for 30 min (C). In (C) white separation lines indicate where lanes from the same blots were joined. Where indicated EC monolayers were subjected to ICAM-1 cross-linking (XL) for 10 min. MAPK phosphorylation was then analyzed and quantified as described for Fig. 1. Results similar to those shown with PP2 were also found with 10 μM SU6656 (data not shown). MAPK levels were not significantly affected by any of the pretreatments (Supplemental Fig. 1C). (D) GPNT EC monolayers were pretreated with 10 μM PP2 for 1 h, 10 μg/ml C3 transferase for 16 h, or 20 μM Gö6983 (Gö) for 1 h prior to analysis of lymphocyte TEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 7.

Endothelial JNK mediates F-actin rearrangements and paxillin phosphorylation in the regulation of lymphocyte TEM. (A) GPNT cells were pretreated without or with 50 μM SP600125 for 1 h before 30 min ICAM-1 ligation. Cell lysates were analyzed by immunoblots using anti–phospho-Y118 paxillin and anti-paxillin Abs. (B) Densitometric quantification of paxillin phosphorylation from three independent experiments as shown in (A). (C) Expression of phosphorylation-deficient Y31F/Y118F (Y31/118F) significantly inhibited lymphocyte TEM (measured as described in Fig. 5E). (D) Pretreatment of GPNT with the FAK inhibitors PF573228 (PF, 10 μM) or FAK inhibitor 14 (C-14, 50 μM) for 1 h led to inhibition of TEM (measured as described in Fig. 5B). (E) Postconfluent, serum-starved GPNT ECs were subjected to ICAM-1 ligation for 30 min, and then fixed and stained for phospho-Y118 paxillin (green) and total paxillin (red). Shown are representative confocal micrographs. (F) Postconfluent GPNT monolayers were left untreated (upper panels), or PLNCs (∼5 PLNCs per EC) were added (lower panels) and allowed to adhere for 30 min. Cultures were then vigorously washed to remove all loosely attached PLNCs, fixed and stained for phospho-paxillin (red) and DNA (blue), and analyzed by confocal microscopy. Numbers indicate three individual adherent T cells. Arrowheads indicate phospho-paxillin in cell-cell contact strands. Scale bars, 10 μm. (G) ICAM-1 was ligated in postconfluent, serum-starved GPNT cells for the indicated times before the cells were lysed and VE-cad immunoprecipitated. Representative immunoblots show the level of paxillin and VE-cad found in immunoprecipitates. Densitometric analysis (mean ± SEM) from five such experiments is shown on the right. (H) As in (G) with the exception that postconfluent GPNT were pretreated with 50 μM SP600125 for 1 h and subjected to ICAM-1 ligation for 15 min before analysis; quantification was from four independent experiments. (I) TEM assay as described in Fig. 5B except that GPNTs were pretreated with 50 μM SP600125 (SP), 1 mM L-NAME, or both as indicated. (J) TEM assay as described in Fig. 5E except that GPNTs were cotransfected with combinations of plasmids encoding wild-type or phosphorylation-deficient Y31F/Y118F (Y31/118F) paxillin and wild-type or phosphorylation-deficient Y731F mouse VE-cad as indicated. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 8.

ICAM-1 mediates VE-cad internalization in a JNK-dependent manner. (A) ICAM-1 was ligated in postconfluent, serum-starved GPNT. At indicated times, cells were transferred to ice to stop endocytosis, treated with trypsin, and lysed, and the level of trypsin-resistant VE-cad determined by immunoblot analysis. Shown are representative immunoblots and densitometric quantification of trypsin-resistant (i.e., internalized) VE-cad in comparison with cellular tubulin content. A 10th of nontreated control cell extract was loaded to reveal total VE-cad content (no trypsin/10). Where indicated, cells were pretreated with 50 μM SP600125. (B) VE-cad endocytosis was visualized by internalization of a FITC-labeled anti–VE-cad Ab (green) in postconfluent, serum-starved hCMEC/D3 cells. Cells were left untreated or treated with anti–ICAM-1 (5 μg/ml) for the indicated times. Where indicated extracellular Ab was removed by acid wash (AW). Subsequently cells were fixed, their nuclei counterstained (blue), and analyzed by confocal microscopy. Scale bar, 10 μm. (C) Cryo-immuno–EM of VE-cad distribution in control (Ctrl) and anti–ICAM-1 stimulated (5 min) hCMEC/D3 cultures. Shown are interendothelial junction areas with the two abutting plasma membranes. Arrowheads point out gold labeled VE-cad, which in control cells was found predominantly associated with the plasmalemmal membrane (within 20 nm, i.e., the distance expected by the primary and the secondary bridging Ab) (23). (D) Distances measured from cell-cell junction for VE-cad gold particles as determined from three independent preparations as shown in (C). (E) As in (A) with the exception that postconfluent, serum-starved GPNT were cocultured with PLNCs (∼5 PLNCs per EC) for 15 min. (F) hCMEC/D3 were cocultured with human CD4⁺ lymphocytes (green) for 30 min. VE-cad (magenta) distribution and endocytosis was visualized as described in (B) by Ab labeling and acid wash before fixation and confocal microscopy. Scale bar, 10 μm. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 9.

Proposed signaling networks in brain MVEC downstream of ICAM-1. Circled P indicates protein phosphorylation.