A novel role for endoplasmic reticulum protein 46 (ERp46) in platelet function and arterial thrombosis in mice

Abstract

Although several members of protein disulfide isomerase (PDI) family support thrombosis, other PDI family members with the CXYC motif remain uninvestigated. ERp46 has 3 CGHC redox-active sites and a radically different molecular architecture than other PDIs. Expression of ERp46 on the platelet surface increased with thrombin stimulation. An anti-ERp46 antibody inhibited platelet aggregation, adenosine triphosphate (ATP) release, and αIIbβ3 activation. ERp46 protein potentiated αIIbβ3 activation, platelet aggregation, and ATP release, whereas inactive ERp46 inhibited these processes. ERp46 knockout mice had prolonged tail-bleeding times and decreased platelet accumulation in thrombosis models that was rescued by infusion of ERp46. ERp46-deficient platelets had decreased αIIbβ3 activation, platelet aggregation, ATP release, and P-selectin expression. The defects were reversed by wild-type ERp46 and partially reversed by ERp46 containing any of the 3 active sites. Platelet aggregation stimulated by an αIIbβ3-activating peptide was inhibited by the anti-ERp46 antibody and was decreased in ERp46-deficient platelets. ERp46 bound tightly to αIIbβ3 by surface plasmon resonance but poorly to platelets lacking αIIbβ3 and physically associated with αIIbβ3 upon platelet activation. ERp46 mediated clot retraction and platelet spreading. ERp46 more strongly reduced disulfide bonds in the β3 subunit than other PDIs and in contrast to PDI, generated thiols in β3 independently of fibrinogen. ERp46 cleaved the Cys473-Cys503 disulfide bond in β3, implicating a target for ERp46. Finally, ERp46-deficient platelets have decreased thiols in β3, implying that ERp46 cleaves disulfide bonds in platelets. In conclusion, ERp46 is critical for platelet function and thrombosis and facilitates αIIbβ3 activation by targeting disulfide bonds.

Graphical abstract

Figure 1.

Functional role of ERp46 in human platelets. Expression of ERp46 on the surface of nonactivated platelets (A) and thrombin (1 U/mL)-activated platelets (B). Mean fluorescent intensity (MFI) ± standard error of the mean (SEM), n = 5, ****P < .0001, Student t test. Normal rabbit IgG (nI IgG, 30 μg/mL, red) and anti-ERp46 antibody (30 μg/mL, blue) were both labeled with Alexa-488. (C) SFLLRN (100 μM)-induced platelet activation/aggregation releases ERp46 into the supernatant. Shown is the fold increase of ERp46 protein in the supernatant with platelet activation. ERp46 was immunoprecipitated from the supernatant and analyzed by immunoblotting. (D-E) Platelet activation increases thiol labeling in surface ERp46. After activation by collagen (10 μg/mL) (D) or thrombin (1 U/mL) (E), platelets were labeled with MPB or SSB as described in “Methods.” After immunoblotting with the anti-ERp46 antibody, the band intensities were used to calculate the ratio of MPB/SSB in ERp46 on the surface of activated platelets relative to ERp46 on the surface of nonactivated platelets. (F) The polyclonal anti-ERp46 antibody inhibits platelet aggregation and ATP release. Representative aggregation and ATP release tracings (left) and combined results (right) showing inhibition by the anti-ERp46 antibody of human platelets activated with thrombin (0.05 U/mL); mean ± SEM, n = 3, *P < .05, ***P < .001, analysis of variance (ANOVA). Washed human platelets were preincubated with normal rabbit IgG and polyclonal anti-ERp46 antibody at the indicated concentration. Aggregation and ATP secretion were monitored in the lumi-aggregometer. (G-I) Anti-ERp46 antibody inhibits αIIbβ3 activation measured by PAC1 binding in platelets when added before (G) but not after (H) thrombin (0.025 U/mL) activation. MFI ± SEM, n = 5, ****P < .0001, ANOVA. In these experiments, the anti-ERp46 antibody (30 μg/mL) was incubated with the human platelets for 10 minutes before activation (G) or after activation (H). The red and blue histograms represent thrombin-activated platelets. (I) Addition of ERp46 (2 μM) (blue) potentiates thrombin (0.25 U/mL) activation of αIIbβ3 measured by PAC1 binding on platelets but does not itself induce αIIbβ3 activation. PBS control (red). Left panels, representative histograms; right panels, bar graph of combined results; mean ± SEM, n = 5, ****P < .0001, ANOVA. Non-act, nonactivated platelets.

Figure 2.

Platelet ERp46 is required for hemostasis, thrombosis, and platelet accumulation into a growing thrombus. (A-B) Characterization of Pf4-Cre/ERp46^fl/fl mice. (A) PCR products of tail DNA from Cre⁻ ERp46^fl/fl mice and Pf4-Cre/ERp46^fl/fl mice. The bands represent PCR product from the ERp46-floxed allele (upper panel, 431-bp) and the Cre gene (lower panel, 420-bp), respectively. (B) Western blots of platelet lysates using a polyclonal rabbit anti-ERp46 antibody and antibodies against PDI, ERp57, ERp72, and ERp5. Shown are the actin loading controls. Left panel representative blot; right panel, quantitative analysis of protein level by densitometry of band density of PDIs relative to the ERp46^fl/fl wild-type (WT) control, which was set at 100%; mean ± SEM, ****P < .0001, n = 4, Student t test. (C) Tail bleeding times; mean ± SEM, n = 16 for each group, ****P < .001, Student t test. (D-E) Incorporation of platelets into a growing thrombus in ERp46^fl/fl mice and Pf4-Cre/ERp46^fl/fl mice was detected by Alexa 488 anti-CD41 using FeCl₃-induced mesenteric arterial injury. Mean artery diameters were 140.2 ± 3.35 μm (SEM) in ERp46^fl/fl mice, 136.0 ± 3.73 μm in Pf4-Cre/ERp46^fl/fl mice (P = not significant [ns]), 136.3 ± 3.24 μm in Pf4-Cre/ERp46^fl/fl mice plus rERp46 (P = ns), 132.6 ± 5.21 μm in Pf4-Cre/ERp46^fl/fl mice plus rERp46(aaaaaa), P = ns, ANOVA. (D) Images at 7, 11, and 15 minutes. Dotted lines mark the vessel wall. Scale bar, 200 μm. Composite of fluorescence intensity (FI) per area analyzed (FI/μm²) in ERp46^fl/fl (21 thrombi from 8 mice), Pf4-Cre/46^fl/fl (20 thrombi from 8 mice), Pf4-Cre/46^fl/fl plus rERp46 (24 thrombi from 8 mice), and Pf4-Cre/ERp46^fl/fl plus rERp46(aaaaaa) (25 thrombi from 9 mice); mean ± SEM, *P < .05, ****P < .0001, ANOVA. (F-H) Cremaster laser injury in arterioles of Pf4-Cre/ERp46^fl/fl mice and their Cre⁻ ERp46^fl/fl littermate control mice. Platelets at the site of injury were detected using anti-CD41 F(ab)₂ fragments conjugated to Alexa Fluor 647. (F) Representative fluorescence images from widefield intravital microscopy for platelet accumulation (red) at the indicated time points after injury. (G) The mean ± SEM integrated FIs of anti-CD41 F(ab)₂ fragments over 240 seconds from ERp46^fl/fl (37 thrombi from 4 mice) and Pf4-Cre/ERp46^fl/fl (26 thrombi from 4 mice). (H) The areas under the FI curves over 240 seconds were analyzed with a Mann-Whitney U test; **P < .01. (I) Time to occlusion of FeCl₃-induced carotid artery thrombosis in Pf4-Cre/ERp46^fl/fl mice compared with ERp46^fl/fl littermate controls; ***P < .001, Student t test. ns, not significant.

Figure 3.

ERp46 is critical for aggregation of mouse platelets and interacts with αIIbβ3 (A) ERp46-deficient platelets have defective thrombin (0.02 U/mL-induced activation of αIIbβ3 detected by the JON/A activation-dependent antibody). (B) P-selectin expression is decreased in thrombin-stimulated ERp46-null platelets. (A-B) left panels, representative histogram; right panels, combined results; mean ± SEM, n = 5 for each group, ***P < .001, Student t test. (C-E) Representative aggregation and ATP release tracings (left panels) and combined results (right) showing the defects in ERp46-deficient platelets using (C) thrombin, (D) CRP, or aggregation for (E) ADP; mean ± SEM, n = 4 (thrombin), n = 6 (CRP), n = 5 (ADP), **P < .01, ***P < .001, ****P < .0001, Student t test. Aggregation and ATP secretion were monitored in the lumi-aggregometer. (F) Anti-ERp46 inhibits CHAMP-induced aggregation of human platelets; mean ± SEM, n = 4, ****P < .0001, ANOVA. Normal rabbit IgG (90 μg/mL) was used as a control. (G) CHAMP-induced aggregation of ERp46-null mouse platelets is decreased; mean ± SEM, n = 5, ****P < .0001, Student t test. CHAMP was added to final concentration of 3 μM and aggregation was performed in the presence of indomethacin (100 µM) and Apyrase (4 U/mL). (H) ERp46 interaction with αIIbβ3 by surface plasmon resonance. Recombinant full-length human αIIbβ3 (25 μg/mL) was immobilized on the surface of a CM5 chip. Different concentrations of WT ERp46 without or with MnCl₂ (1 mM) were infused over the chip in the running buffer. The equilibrium dissociation constant was calculated based on the Kon and Koff values with Biacore T200 evaluation software. (I) ERp46 interacts with β3 integrins on mouse platelets. Binding of Alexa Fluor 488-conjugated ERp46 to Mn²⁺-treated WT and β3-null mouse platelets. Representative histogram (left panels); cumulative data (right panels); mean ± SEM, n = 5 for each group, ***P < .001, ****P < .0001, ANOVA. Washed mouse platelets (3 ×10⁸/mL) were preincubated with Alexa Fluor 488 ERp46 (30 μg/mL) for 10 minutes at room temperature and then treated with Mn²⁺ (12 mM) for 5 minutes at room temperature. Surface binding of Alexa Fluor 488 ERp46 was detected by flow cytometry. (J) Stimulation-dependent association of ERp46 with integrin β3. Platelets (1 × 10⁹/mL) were stimulated in the presence of EGTA, apyrase, and indomethacin at varying concentrations of convulxin (30, 60, or 120 ng/mL) for 90 seconds. Following sample lysis, proteins were precipitated with mouse anti-β3 antibody SZ21 and protein G–resin. Immunoblotting with goat anti-ERp46 antibody and polyclonal rabbit anti-β3 antibody (Abcam) showed interacting proteins.

Figure 4.

The a°, a, and a′ active sites of ERp46 each contribute to aggregation and ATP release of mouse and human platelets. (A) Schematic diagram of the ERp46 variants. (B) Characterization of variant ERp46 proteins in the Di-E-GSSG assay. ERp46(cc-cc-cc), WT ERp46; ERp46(cc-cc-aa) had the a′ CGHC motif inactivated; ERp46(cc-aa aa) had the a and a′ active sites inactivated; ERp46(aa-aa-aa) had all 3 sites inactivated. (C-E) Correction of the aggregation and secretion defects of ERp46-null platelets (2 × 10⁸ platelets/mL) by ERp46 variants (100 nM). (F) Effect of preincubating human platelets (2 × 10⁸ platelets/mL) with ERp46 variants (1 μM). Submaximal aggregation (baseline) was stimulated with thrombin (0.015 U/mL). The ERp46 variants were added 5 minutes prior to the addition of thrombin. Representative aggregation and ATP release tracings (left panels) and combined results (right); mean ± SEM, n = 3, mouse platelets; n = 5, human platelets, *P < .05, **P < .01, ***P < .001, ****P < .0001, ANOVA. ns, not significant.

Figure 5.

ERp46 but not other PDIs recover aggregation and ATP release of ERp46-null platelets. (A) ERp46-null platelets were incubated with 100 nM PDI, ERp57, ERp72, or ERp46. (B) PDI-null platelets were incubated with 100 nM ERp46 or PDI. (C) ERp57-null platelets were incubated with 100 nM ERp46 or ERp57. Platelet aggregation was stimulated with thrombin (0.015 U/mL); representative aggregation and ATP release tracings (left panel) and combined results (right); mean ± SEM, n = 3 (A); n = 5 (B); n = 3 (C); **P < .01, ***P < .001, ****P < .0001, ANOVA.

Figure 6.

ERp46 generates thiols in the β3 subunit of αIIbβ3. (A) Reductase activity of ERp46 compared with ERp57, PDI, ERp5, ERp72, and inactivated ERp46 [ERp46(aaaaaa)] (30 nM each) in the Di-E-GSSG assay. (B-C) ERp46 generates thiols in β3 more effectively than PDI, ERp57, ERp72, or ERp5. ERp46 or other PDIs (1 μM) were incubated with αIIbβ3 (0.5 μM) (Abcam) for 20 minutes at 37°C. IodoTMT (400 μM) was then added for 1 hour at 37°C and labeling of thiols performed in 1% SDS with 5 mM ethylenediaminetetraacetic acid . These studies were performed in the absence of added GSH. (D-E) The active site cysteines of ERp46 generate thiols in αIIbβ3. Shown is the labeling of β3 with WT ERp46 or inactivated ERp46 [ERp46(aaaaaa)]. ERp46 or ERp46(aaaaaa) (1 μM) was incubated with αIIbβ3 (0.5 μM) and GSH (100 μM), and labeling was performed with iodoTMT (500 μM). (F-G) Effect of fibrinogen (Fbgn) on thiol generation by ERp46 and PDI in αIIbβ3. ERp46 or PDI (1 μM) were incubated with αIIbβ3 (0.5 μM) with GSH (100 μM). In some samples, fibrinogen (0.5 μM) was added. Thiols were labeled with iodoTMT (500 μM). Blotting for iodoTMT was performed using the anti-TMT antibody. (H-I) Effect of fibrinogen on thiol generation by ERp46 and PDI in αIIbβ3 in platelets. ERp46 or PDI (1 μM) were incubated with human platelets (4 × 10⁸ platelets/mL) without GSH. In some samples, fibrinogen (1 μM) was added. Thiols were labeled with MPB and blotting was performed as described in supplemental Methods. The intensity of each band was calculated using the Image J program, and the ratio of iodoTMT (TMT) or MPB to β3 protein was compared with the untreated sample. Shown are the blots for β3 (H-96 or B-7 for D), PDI and ERp46 (rabbit antibodies), and the β chain of fibrinogen; mean ± SEM, n = 4 (B-C), n = 3 (D-E), n = 3 (F-G), n = 7 (H-I); *P < .05, **P < .01, ***P < .001, ****P < .0001, ANOVA. (J-K) Addition of ERp46 followed by PDI (ERp46/PDI) maximally generates thiols in β3. αIIbβ3 was treated with 1 μM PDI or ERp46 for 40 minutes, except when added sequentially they were added for 20 minutes each. Addition of PDI followed by ERp46 (PDI/ERp46) gave similar results to ERp46 alone or PDI plus ERp46 added simultaneously. Labeling was with iodoTMT (400 μM) in the absence of GSH; mean ± SEM, n = 4; ****P < .0001, ANOVA. (L-M) Addition of ERp46 followed by PDI (ERp46/PDI) provides maximal enhancement of platelet aggregation. 1.5 μM PDI or ERp46 were added for 10 minutes to platelets, except when added sequentially they were added for 5 minutes each. Aggregation was induced with thrombin (0.012 U/mL). The aggregation tracings for ERp46 alone, PDI plus ERp46 added simultaneously, or PDI followed by ERp46 (PDI/ERp46) are overlapping; mean ± SEM, n = 5; *P < .05, ****P < .0001, ANOVA. ns, not significant.

Figure 7.

ERp46 generates thiols in the EGF-2 domain of the β3 subunit of the αIIbβ3 integrin. (A) The MS/MS spectrum of a 5+ charged ion (m/z 618.72) corresponding to the peptide GECLCGQCVCHSSDFGK from integrin β3 with all 4 cysteines alkylated by iodoTMT reagent. The y- and b-ion series in (A) enabled the confident identification of the peptide sequence. During the MS/MS stage of acquisition to derive fragment ions and sequence information, a unique reporter ion mass is also generated. These reporter ions are in the low mass region of the MS/MS spectrum. The intensity of iodoTMT tag (126, 127, 130, and 131) in (B) indicate the relative quantitation of redox changes. IodoTMT-126 or iodoTMT-130 represent the initial abundance of thiols in the peptide, whereas iodoTMT-127 and iodoTMT-131 represent the thiols in the peptide labeled after reduction of disulfide bonds in each sample with DTT/TCEP. (C) Representative MS3 spectrum of the first Cys (Cys501) in the peptide showing the increase in iodoTMT-130 induced by ERp46 relative to iodoTMT-126 in the sample without ERp46. IodoTMT-127 and iodoTMT-131 represent the thiols labeled after reduction of disulfide bonds in the samples. (D) ERp46 generates thiols in Cys503, Cys501, Cys508, and Cys521 of the EGF-2 domain as measured by MS3; mean ± SEM, n = 3; *P < .05, **P < .01, t test. These cysteines form the Cys473-Cys503, Cys486-Cys501, and Cys508-Cys521 disulfide bonds. (E-G) Thiols are decreased in the β3 subunit of nonactivated (NA) and activated (Act) ERp46-null platelets but are unchanged in PDI-null platelets. Platelets from Pf4-Cre/ERp46^fl/fl mice, Pf4-Cre/PDI^fl/fl mice, or Cre⁻ littermates were activated with collagen (10 μg/mL) for 10 minutes without stirring followed by MPB labeling. After platelet lysis, the MPB was pulled down with streptavidin-agarose beads. The samples were analyzed by blotting with rabbit anti-β3 and anti-ERp46 or anti-PDI antibodies and the band densities calculated by densitometry using Odyssey V3.0. (E and G) Representative blots (F and H) cumulative data for nonactivated (NA) and activated platelets; mean streptavidin SEM, n = 4 (E-F), n = 6 (G-H), ***P < .001, ****P < .0001, Student t test.