Subcellular targeting of oxidants during endothelial cell migration

Abstract

Endogenous oxidants participate in endothelial cell migration, suggesting that the enzymatic source of oxidants, like other proteins controlling cell migration, requires precise subcellular localization for spatial confinement of signaling effects. We found that the nicotinamide adenine dinucleotide phosphate reduced (NADPH) oxidase adaptor p47(phox) and its binding partner TRAF4 were sequestered within nascent, focal complexlike structures in the lamellae of motile endothelial cells. TRAF4 directly associated with the focal contact scaffold Hic-5, and the knockdown of either protein, disruption of the complex, or oxidant scavenging blocked cell migration. An active mutant of TRAF4 activated the NADPH oxidase downstream of the Rho GTPases and p21-activated kinase 1 (PAK1) and oxidatively modified the focal contact phosphatase PTP-PEST. The oxidase also functioned upstream of Rac1 activation, suggesting its participation in a positive feedback loop. Active TRAF4 initiated robust membrane ruffling through Rac1, PAK1, and the oxidase, whereas the knockdown of PTP-PEST increased ruffling independent of oxidase activation. Our data suggest that TRAF4 specifies a molecular address within focal complexes that is targeted for oxidative modification during cell migration.

Figure 1.

TRAF4 accumulates in focal complexes. HUVECs were transfected with the indicated plasmids (color of font indicates pseudocolor designation), plated on fibronectin-coated coverslips, and examined live without fixation using confocal (A and B) or TIRF (C) microscopy. (A) TRAF4-GFP translocated to the tips of leading edges (right), a pattern not seen with pEGFP (left). (B) HUVECs were cotransfected with TRAF4-GFP and DsRed-zyxin. Insets show appearance of TRAF4-GFP and DsRed-zyxin in discontinuous structures at the leading (right) edge of a large protrusion (arrows). Larger stress fiber–anchored focal adhesions contain DsRed-zyxin but no detectable TRAF4-GFP (top insets). (C) TIRF microscopy image of TRAF4-GFP and DsRed-zyxin showing ventral location of small TRAF4-GFP and DsRed-zyxin focal complexes at an advancing protrusion (top inset). Larger DsRed-zyxin focal adhesions lacked TRAF4-GFP (bottom inset). (D) Immunofluorescent images of fixed cells were obtained using TIRF microscopy showing colocalization of endogenous TRAF4 (labeled with AlexaFluor488) within small (closed head arrows) to medium (open head arrow)-sized vinculin aggregates (AlexaFluor555) appearing at the edge of a protrusion. Bars (A–C), 20 μm; (D) 10 μm.

Figure 2.

p47^phox associates with TRAF4 in focal complexes. (A) Confocal image of DsRed-p47 in small, discrete zyxin-GFP–containing complexes within the leading (right) lamellar edge. Stress fiber–associated focal adhesions (top inset) lack DsRed-p47. DsRed-p47 also localized to the periphery (arrow). (B) TIRF image shows ventral appearance of DsRed-p47 with nascent zyxin-GFP aggregates at the leading edge of a protrusion. (C) Immunofluorescent image of endogenous p47^phox (AlexaFluor488) showing colocalization with small vinculin (AlexaFluor555) accumulations. (D) Immunofluorescent image showing the appearance of endogenous TRAF4 (AlexaFluor488) and p47^phox (AlexaFluor633) in small dotlike structures at the edge of a protrusion (arrows). B–D, TIRF images. Bars (A and B), 20 μm; (C and D) 10 μm.

Figure 3.

TRAF4 associates with Hic-5. (A) In vitro–translated full-length Hic-5 was specifically pulled down by GST-TRAF4, not by GST or beads alone. Top panel is an autoradiogram of captured ³⁵S-methionine–Hic-5; bottom panel is Coomassie stain of GST fusion input. (B) Phoenix-293 cells were transfected with TRAF4 with or without Hic-5 and immunoprecipitated with antibodies against Hic-5 (left), TRAF4 (right), or irrelevant antibodies. Immunoblots were sequentially probed for Hic-5 and TRAF4. Hic-5 appears as a doublet straddling the Ig heavy chain (HC). (C) Lysate from untransfected human lung microvascular endothelial cells was immunoprecipitated with antibodies for TRAF4 or Pyk2 and immunoblotted as shown. Endogenous Hic-5 coprecipitated with TRAF4 and both TRAF4 and Hic-5 coprecipitated with Pyk2. (D) HUVEC were infected with adenoviruses harboring lacZ, p47^phox(wt), p47(W193R), or p47(S303D,S304D,S328D) (MOI = 100:1). 24 h later, lysate was immunoprecipitated for Pyk2 and immunoblotted for p47^phox and Pyk2. Bottom panel shows total p47^phox. (E) Partial colocalization of Hic5-GFP and DsRed-TRAF4 at focal structures within the membrane edge (arrows) is seen. (F) Immunofluorescent image showing the presence of endogenous Hic-5 (AlexaFluor633) in small- to medium-sized vinculin (AlexaFluor555) structures at the edge of a protrusion (arrows). (G) Immunofluorescent stain showing the simultaneous appearance of TRAF4 (AlexaFluor488), Hic-5 (AlexaFluor633), and vinculin (AlexaFluor555) in small (closed head arrow) to medium (open head arrow)-sized complexes at the periphery of a lamella. E–G are TIRF microscopy images. Bars (E and F), 10 μm; (G) 20 μm.

Figure 4.

TRAF4–Hic-5 interactions affect endothelial cell migration. (A) HUVEC were transfected with siRNA against either TRAF4 or Hic-5, and migration across filters in response to a VEGF gradient was measured. Immunoblots show protein levels 48 h after transfection. Control siRNA was against luciferase. 100 μM MnTMPyP was present only during the 16-h migration period. Knockdown of Hic-5 or TRAF4 or antioxidant treatment decreased VEGF-induced migration of HUVEC. hpf, high-powered field. (B) Phoenix-293 cells were cotransfected with full-length TRAF4 and Flag-tagged Hic-5 truncations as indicated. Immunoblots of lysates before TRAF4 immunoprecipitation are shown as input. Cartoon below diagrams Hic-5 truncations showing NH₂-terminal Flag tag, three LD motifs (shaded), and four COOH-terminal LIM domains (open). Only full-length Flag–Hic-5 containing LIM 4 coprecipitated with TRAF4 (first lane). (C) Phoenix-293 cells were cotransfected with full-length TRAF4 and full-length Flag–Hic-5 (FL), Flag–Hic-5(329–444) (LIM3,4), or Flag–Hic-5(388–444) (LIM4). Lysates before TRAF4 immunoprecipitation were immunoblotted for Hic-5 and reprobed for TRAF4 to show input. Top panel shows coprecipitation of full-length Flag–Hic-5 and Flag–Hic-5(329–444) but not Flag–Hic-5(388–444) with TRAF4. (D) Phoenix-293 cells were cotransfected with full-length Flag–Hic-5 and either Flag-TRAF4(1–260) (TRAF4-N) or Flag-TRAF4(261–470) (TRAF4-C). Lysate was immunoblotted with anti-Flag (bottom) to show Hic-5 and TRAF4 inputs simultaneously. After immunoprecipitation with anti–Hic-5, blots were probed with anti-Flag to demonstrate the recovery of Flag–TRAF4-C but not Flag–TRAF4-N. Ig heavy and light chains (HC and LC) are shown. (E) HUVEC were cotransfected with pEGFP and empty vector (pCIN), TRAF4-C, or Hic-5(LIM3,4). GFP-expressing cells migrating across 8-μm pore filters in response to a VEGF gradient is shown. Migration in response to VEGF was blocked by the coexpression of either TRAF4-C or Hic-5(LIM3,4). (A and E) *, P < 0.05 compared with no VEGF control; ^†, P < 0.05 compared with VEGF control. (F) Discrete DsRed-p47 dotlike structures characteristically appeared at the edges of lamellar protrusions (arrows). Bar, 20 μm. Cotransfection of HUVEC with TRAF4-C or knockdown of endogenous TRAF4 or Hic-5 decreased the number of cells (*, P < 0.05) with such DsRed-p47 structures. (G) HUVEC were cotransfected as in D and plated on fibronectin-coated etched coverslips. Phase-contrast and epifluorescent (green) images were obtained immediately after wounding and at 16 and 24 h after wounding. (H) Histogram represents mean speed of GFP-expressing cells migrating into the wound. Entry of GFP-expressing cells into the wound was decreased by the coexpression of TRAF4-C or Hic-5(LIM3,4). Non–GFP-expressing cells migrated into wounds at comparable speeds. Error bars represent SEM. *, P < 0.05 compared with vector alone.

Figure 5.

Myr-TRAF4 activates Rho family GTPases and PAK1. (A) HUVECs were transfected with empty vector, native TRAF4 (wt), or Myr-TRAF4. After 24 h, cells were lysed, and GTP-loaded Rho proteins were captured with GST-CRIB or GST–Rho-binding domain. Immunoblots for Cdc42, Rac1, and RhoA are shown in lysates and after capture. (B and C) HUVECs were transfected with the indicated vectors, infected with Ad-lacZ, Ad-RhoA(N19), or Ad-Rac1(N17) (MOI = 100:1), and RhoA, Rac1, and Cdc42 activity were assessed. Myr-TRAF4 activated Cdc42 and Rac1 upstream of RhoA. (D) HUVECs were transfected with the indicated plasmids and assessed for PAK1 activity by immunoprecipitation kinase. Phosphorylation of myelin basic protein (top) and immunoblot for captured PAK1 (bottom) are shown. Myr-TRAF4 but not wtTRAF4 activated PAK1.

Figure 6.

Myr-TRAF4 activates the NADPH oxidase through PAK1. (A) Phoenix-293 cells overexpressing wt-p47^phox were transfected with the indicated plasmids. Whole cell lysates were immunoblotted for p47^phox (pS328) and reprobed for total p47^phox, demonstrating phosphorylation of p47(S328) by Myr-TRAF4. (B) HUVECs were cotransfected with DsRed and one of the indicated plasmids and were infected with Ad-lacZ or Ad-p67(V204A). DCF fluorescence of adherent DsRed-expressing cells was quantified by image analysis as described in Oxidant production. Myr-TRAF4–induced oxidant production was blocked by p67(V204A). (C) HUVECs were cotransfected with DsRed and the indicated plasmids, and oxidant production of DsRed-expressing cells was measured without manipulation. The PAK inhibitory domain (PID(wt)) but not its nonbinding mutant (PID(LF)) decreased Myr-TRAF4–induced oxidant production. (B and C) *, P < 0.05 compared with control; ^†, P < 0.05 compared with Myr-TRAF4 alone. Error bars represent SEM.

Figure 7.

Myr-TRAF4 oxidatively modifies the focal contact phosphatase PTP-PEST through the NADPH oxidase. (A) Phoenix-293 cells were transfected with either empty vector or Myr-TRAF4. In top panels, cells were cotransfected with Flag–PTP-PEST. p67(V204A) was adenovirally transduced. Lysates were labeled with 5-IAF, and the indicated PTPs were immunoprecipitated and immunoblotted for fluorescein to detect 5-IAF labeling and for the respective PTPs to assess capture. Myr-TRAF4 decreased 5-IAF labeling of PTP-PEST, indicating oxidative modification of the active site cysteine. 5-IAF labeling of MKP-1 and SHP-2 was unaffected. Error bars represent SEM. (B) Immunofluorescent images were obtained with TIRF microscopy showing colocalization of p47^phox, TRAF4, and Hic-5 (AlexaFluor488 or 633) with phosphotyrosinylated proteins (AlexaFluor555) within peripheral dots at the edges of protrusions. In bottom panels, endogenous TRAF or Hic-5 were knocked down in HUVECs as indicated. Bars, 10 μm. (C) HUVECs were transfected with the indicated plasmids and infected with the indicated Ad viruses. 200 μM MnTMPyP was present for 3 h before lysis. Rac1 GTP loading was assessed by binding to GST-CRIB. Both MnTMPyP and p67(V204A) blocked Myr-TRAF4–induced Rac1 activation.

Figure 8.

Myr-TRAF4 induces marked membrane ruffling through Rac1, PAK1, and the NADPH oxidase. (A) HUVECs were cotransfected with either empty vector or Myr-TRAF4 and either actin-GFP or DsRed-p47. Myr-TRAF4–induced ruffling is seen on multiple edges of cotransfected cells (middle). Ruffles contained DsRed-p47 (right). (B and C) HUVECs were cotransfected with actin-GFP and either empty vector (first bar) or Myr-TRAF4 and additionally with one of the indicated plasmids. PAK1(K298A) and p67(V204A) were expressed via adenoviral transduction. 100 μM MnTMPyP was added for 1 h. Inhibition of PAK1, the NADPH oxidase, Pyk2, or Src decreased the percentage of cells ruffling in response to Myr-TRAF4. *, P < 0.05 compared with vector control; ^†, P < 0.05 compared with Myr-TRAF4 alone. (D) HUVECs were transfected with siRNA against PTP-PEST. Immunoblot shows protein levels after 48 h. PTP-PEST knockdown increased the percentage of ruffling cells. *, P < 0.05 compared with control siRNA. Expression of p67(V204A) had no effect on the ruffling rate in PTP-PEST knockdown cells. Examples are shown below. Error bars represent SEM. Bars, 20 μm.