Multimerin 1 supports platelet function in vivo and binds to specific GPAGPOGPX motifs in fibrillar collagens that enhance platelet adhesion

Abstract Background Multimerin 1 (human: MMRN1, mouse: Mmrn1) is a homopolymeric, adhesive, platelet and endothelial protein that binds to von Willebrand factor and enhances platelet adhesion to fibrillar collagen ex vivo. Objectives To examine the impact of Mmrn1 deficiency on platelet adhesive function, and the molecular motifs in fibrillar collagen that bind MMRN1 to enhance platelet adhesion. Methods Mmrn1‐deficient mice were generated and assessed for altered platelet adhesive function. Collagen Toolkit peptides, and other triple‐helical collagen peptides, were used to identify multimerin 1 binding motifs and their contribution to platelet adhesion. Results MMRN1 bound to conserved GPAGPOGPX sequences in collagens I, II, and III (including GPAGPOGPI, GPAGPOGPV, and GPAGPOGPQ) that enhanced activated human platelet adhesion to collagen synergistically with other triple‐helical collagen peptides (P < .05). Mmrn1−/− and Mmrn1+/− mice were viable and fertile, with complete and partial platelet Mmrn1 deficiency, respectively. Relative to wild‐type mice, Mmrn1−/− and Mmrn1+/− mice did not have overt bleeding, increased median bleeding times, or increased wound blood loss (P ≥ .07); however, they both showed significantly impaired platelet adhesion and thrombus formation in the ferric chloride injury model (P ≤ .0003). Mmrn1−/− platelets had impaired adhesion to GPAGPOGPX peptides and fibrillar collagen (P ≤ .03) and formed smaller aggregates than wild‐type platelets when captured onto collagen, triple‐helical collagen mimetic peptides, von Willebrand factor, or fibrinogen (P ≤ .008), despite preserved, low shear, and high shear aggregation responses. Conclusions Multimerin 1 supports platelet adhesion and thrombus formation and binds to highly conserved, GPAGPOGPX motifs in fibrillar collagens that synergistically enhance platelet adhesion.


| INTRODUC TI ON
Platelet adhesion is a critical step in hemostasis and thrombosis that allows platelets to localize and accumulate at sites of vessel injury or thrombus formation. Multimerin 1 (human: MMRN1, mouse: Mmrn1) is a large, soluble, homopolymeric adhesive glycoprotein that is synthesized and stored by megakaryocytes/platelets and endothelial cells for regulated release, but is undetectable in normal plasma. [1][2][3][4] When released, MMRN1 supports platelet adhesion through shear-dependent mechanisms involving activated α IIb β 3 and α V β 3 under static conditions and low shear flow conditions (≤150 s −1 ), and involving von Willebrand factor (human: VWF; mouse: Vwf) and GPIbα, but not β 3 integrins, under high shear flow (1500 s −1 ). 5,6 MMRN1 binds to VWF with high affinity through a two-site, two-step interaction with the VWF A1 and A3 domains that enhances platelet adhesion to immobilized VWF at high shear flow. 7 MMRN1 also enhances platelet adhesion to vascular fibrillar collagens I and III and Horm collagen (equine collagen I, ~95%, and III, ~5%) under high shear flow through mechanisms requiring VWF and GPIbα 6 and uncharacterized motifs in collagen that support MMRN1 binding. The impact of Mmrn1 deficiency on platelet function has not been fully characterized. Similar to MMRN1, the adhesive proteins VWF, fibronectin (FN), vitronectin (VN), and fibrin self-associate to form large homopolymers, [8][9][10][11][12][13][14] and bind to α IIb β 3 on platelets [15][16][17][18][19][20] to mediate aggregate formation. Additionally, thrombospondin-1 (TSP-1) helps to crosslink platelets by self-associating and binding α IIb β 3 -bound fibrinogen (FG). 21 These interactions create large macromolecular complexes that increase the likelihood of platelet-platelet collisions and the avidity of platelet-matrix or platelet-platelet interactions. 22 We postulated that the large MMRN1 homopolymers released by platelets and endothelial cells might similarly enhance platelet-matrix or plateletplatelet interactions, as Mmrn1 −/− Snca −/− mice (evaluated by the ferric chloride [FeCl 3 ] mesenteric vessel injury model) have impaired platelet adhesion and thrombus formation in vivo, and impaired platelet adhesion to collagen in vitro, that are corrected by exogenous MMRN1. 23 We generated Mmrn1-deficient (Mmrn1 −/− ) mice to: (a) determine the impact of selective Mmrn1 loss on platelet adhesion and thrombus formation in vivo; (b) identify additional adhesive ligands for MMRN1/ Mmrn1, and the impact of Mmrn1 deficiency on platelet adhesion and aggregation; and (c) investigate the motifs in collagen that support binding to MMRN1. We demonstrate that Mmrn1 contributes to platelet adhesion and thrombus formation in vivo and update the current model of platelet adhesion to collagen to include GPAGPOGPX, a conserved MMRN1/Mmrn1 binding motif that synergistically enhances platelet adhesion.

| ME THODS
The study was conducted in accordance with the recently revised collagen peptides (P < .05). Mmrn1 −/− and Mmrn1 +/− mice were viable and fertile, with complete and partial platelet Mmrn1 deficiency, respectively. Relative to wild-type mice, Mmrn1 −/− and Mmrn1 +/− mice did not have overt bleeding, increased median bleeding times, or increased wound blood loss (P ≥ .07); however, they both showed significantly impaired platelet adhesion and thrombus formation in the ferric chloride injury model (P ≤ .0003). Mmrn1 −/− platelets had impaired adhesion to GPAGPOGPX peptides and fibrillar collagen (P ≤ .03) and formed smaller aggregates than wild-type platelets when captured onto collagen, triple-helical collagen mimetic peptides, von Willebrand factor, or fibrinogen (P ≤ .008), despite preserved, low shear, and high shear aggregation responses.
Conclusions: Multimerin 1 supports platelet adhesion and thrombus formation and binds to highly conserved, GPAGPOGPX motifs in fibrillar collagens that synergistically enhance platelet adhesion.

K E Y W O R D S
blood platelets, fibrillar collagens, multimerin, platelet adhesiveness, von Willebrand factor Essentials • Multimerin 1 (Mmrn1) is a homopolymeric protein that supports platelet adhesion in vitro.
• Mmrn1-deficient mice and collagen peptides were used to assess Mmrn1 contributions to platelet function.
• Mmrn1 loss impaired platelet adhesion in vivo and to GPAGPOGPX motifs in collagens in vitro.
• Mmrn1 contributes to platelet function and binds to adhesive GPAGPOGPX motifs in collagen.
Mmrn1-deficient mice were generated as outlined in Appendix S1 in supporting information. Experiments were done with wild-type, Mmrn1 −/− , and Mmrn1 +/− mice obtained from crosses of Mmrn1 +/− mice that were regenerated every three to five generations by additional crosses, as outlined in Appendix S1.

| Evaluation of mouse bleeding after tail transection
Bleeding times (BT) and blood loss following tail transection 1.5 mm from the distal tip were evaluated as described. 24 BT were recorded as 900 seconds if bleeding had not stopped by then.

| Blood collection
Murine blood was obtained by terminal exsanguination of anesthetized mice after carotid artery cannulation. 25 Human blood was collected from general population controls with written informed consent.

| Intravital microscopy
Platelet adhesion and thrombus formation in mesenteric arterioles treated with 250 mmol/L FeCl 3 were evaluated as described, 23 to assess: (a) fluorescent platelet deposition on the vessel wall per minute, between 3 and 5 minutes following injury, (b) time to form the first 20 µm diameter thrombus, and (c) vessel occlusion time.

| Preparation of recombinant human multimerin 1
Recombinant (r) human MMRN1 was affinity purified from media of stably transfected human embryonic kidney (HEK)-293 cells and assessed for concentration and purity by ELISA, western blotting, and silver staining as described. 4,5 2.6 | Protein binding assays rMMRN1 binding to immobilized human FG (±pre-treatment with 0.2 U/mL thrombin to induce fibrin formation), FN, or triple-helical collagen peptides was evaluated based on methods described, 6 using Immulon 2 HB flat-bottom plates (Thermo Fisher Scientific) coated (overnight, 4°C) with 1 µg/well protein or peptide.

| Assays of platelet adhesion under shear
Endpoint and real-time analyses of mouse platelet adhesion were evaluated as described, 7
Whole blood from Mmrn1 −/− and wild-type mice was collected in 1:10 (v/v) 3.2% sodium citrate, and supplemented with 5 mmol/L CaCl 2 and 93 µmol/L PPACK (to promote aggregation and prevent clotting, respectively) immediately before loading samples. Blood was sheared by the rotating cone in a 0.3 mm gap for 1 minute at room temperature before fixation with 1:4 (v/v) 0.625% paraformaldehyde (0.5% final).

| Quantification of platelet adhesion and sizes of adherent platelet aggregates
Percent surface area covered by platelets was estimated using ImageJ (National Institutes of Health), and sizes of captured platelet aggregates on surfaces were estimated using representative regions (in-focus) at the center of microcapillary channels, using the following objectives and areas: Horm collagen: 40× objective, ~60 mm 2 square, 1020 × 1020 pixels; rVwf, fibrin(ogen), fibronectin, and collagen peptides: 20× objective, ~20 mm 2 squares, 515 × 515 pixels.
Grayscale images were converted to binary using the Threshold tool before separating features using the Watershed tool and counting using the Analyze Particles tool. As platelet aggregate sizes on Horm collagen varied greatly (~1--40 000 µm 2 ), data were binned and evaluated by frequency plots. Platelet aggregates captured onto rVwf, fibrin(ogen), fibronectin, and collagen peptides (sizes: ~1--5000 µm 2 ) were evaluated using mean feature sizes/image.

| Statistical analyses
Two-tailed independent or paired Student's t-tests were used to evaluate data with normal distributions. Mann-Whitney U-tests were used to evaluate data with non-normal distributions. One-way or repeated measures analysis of variance were used to evaluate data with more than two groups (α < 0.05 considered significant). Bonferroni correction was used for post hoc multiple comparisons where k ≤ 6 or the Holm-Sidak method where k > 6, with α < 0.05 considered significant for each family of comparisons. One-tailed z test of population proportions was used to evaluate proportional data. Unless stated otherwise, data are reported as mean ± standard error of the mean (SEM).

| Mmrn1 loss selectively impairs platelet adhesion
In high shear flow experiments (1500 s −1 ), with whole blood, both whole blood, activated Mmrn1 −/− platelets showed a minor reduction in adhesion to Fg (P = .03), accompanied by a reduction in size of captured platelet aggregates (P = .007; Figure 4B), but normal adhesion to fibrin and Fn ( Figure S8A-D in supporting information).

| MMRN1 binds to GPAGPOGPX motifs in fibrillar collagens, which enhance platelet adhesion and show high specificity for Mmrn1
Collagen Peptide Toolkits were used to determine the locations and sequences in fibrillar vessel wall collagens that bind MMRN1. rMMRN1 bound two Collagen Toolkit peptides, II-9 and III-38, that share a GPAGPOGPX sequence where X is valine (II-9) or glutamine (III-38; Figure 5A,B). Tests with truncated derivatives indicated that MMRN1 bound to GPAGPOGPX but not to other regions of III-38 ( Figure 5C). In silico searches indicated that the GPAGPOGPX motif is unique to fibrillar collagens, and that the GPAGPOGPV locus (helix residues: 151--159) of peptide II-9 is poorly conserved in collagens I and III. Searches for motifs conserved with the MMRN1-binding sequence in peptide III-38 (GPAGPOGPQ, helix residues: 682--690), and searches for variants of GPAGPOGPX elsewhere in collagen I (which consists of one α 2 and two α 1 chains) identified GPAGPOGPI at helix residues 667 to 675 in D-period 3 of the α 1 (I) triple helix, and GPAGSOGFQ in D-period 2 at residues 457 to 465 that aligns with a conserved GPAGPOGFQ sequence in α 2 (I). GPAGPOGPI in collagen I overlaps in sequence alignments with GPAGPOGPQ in collagen III, which has a 9-residue extension at its N-terminus, offsetting helix numbering. Testing of these collagen I sequences as homotrimeric triple-helical peptides, in parallel with GPP (negative control) indicated that GPAGPOGPI (P = .0002), but not the variants GPAGPOGFQ (P = .12) or GPAGSOGFQ (P = .52), supported MMRN1 binding ( Figure 5D).
As the GPAGPOGPX motifs that bind MMRN1 in human collagens I, II, and III are fully conserved in murine collagens, these motifs were further tested using peptide-coated wells and mouse platelets.
When GFOGER was present, GPAGPOGPX peptides in combination with III-23 further enhanced platelet adhesion ( Figure 6B,C). The enhancing effect of GPAGPOGPX on platelet adhesion to GFOGER was increased by CRP activation to induce MMRN1 release (P ≤ .007, Figure 6D), either by adding CRP to the sample or to coating peptides (P > .24, Figure S9 in supporting information).
High shear flow experiments (1500 s −1 ) with CRP-activated mouse platelets provided further evidence that GPAGPOGPX is highly specific for Mmrn1 as co-presentation of GFOGER with GPAGPOGPQ increased the size of captured aggregates formed by wild-type (P < .0001) but not Mmrn1 −/− (P = .07) platelets ( Figure 7A,B). Mmrn1 −/− platelets were also noted to form smaller aggregates than wild-type platelets on surfaces coated with GFOGER and GPP only (P < .001, Figure 7A,B), suggesting that Mmrn1 stabilizes platelet-platelet interactions through additional mechanisms.  While we found that Mmrn1 is not required for platelet adhesion to Fg, fibrin, or Fn, its absence did reduce platelet adhesion, and the F I G U R E 5 Multimerin 1(MMRN1) binding to collagen, evaluated using Collagen Toolkit and other triple-helical peptides, and static platelet adhesion assays. Panels (A)-(B) show data from representative experiments to assess MMRN1 binding to Collagen Toolkit II (A) and Toolkit III (B) peptides, the control peptide GPP, and GFOGER. The sequence of the peptides showing the most binding is indicated in text. C, MMRN1 binding to Toolkit peptides II-9 and III-38 is compared to MMRN1 binding of shortened, derivatives of these peptides and GPP (negative control). D, MMRN1 binding to GPAGPOGPV and homotrimeric variants of the GPAGPOGPX sequence in collagen α1(I) and α2(I). E, Evaluation of the specificity of MMRN1-binding peptide GGSGPAGPOGPQGVK by static platelet adhesion assays using collagen-related peptide-activated Mmrn1 +/+ (average for n = 4 mice) or Mmrn1 −/− (average for n = 3 mice) platelets. F, Similar static platelet adhesion assay evaluation of the MMRN1-specificity of GPAGPOGPI (average for n = 3 mice/group). In panels (A)-(C), absorbance values are corrected for non-specific binding to bovine serum albumin (BSA). Bars and whiskers in Panels (C) and (D) represent the average and standard deviation of three identical experiments, each performed in triplicate. Symbols in Panels (E) and (F), respectively, show data for Mmrn1 +/+ (solid) and Mmrn1 −/− (open) mice, and absorbance values are corrected for non-specific adhesion by subtracting adhesion to GPP at each concentration tested size of platelet aggregates captured onto Fg-but not fibrin-coated surfaces under low shear flow. The latter discrepancies could reflect differences in the ultrastructure of surface-adsorbed Fg versus fibrin, and/or differences in Fg-and fibrin-α IIb β 3 binding interactions. 46,47 It is also possible that the high-affinity conformation of α IIb β 3 or activation-induced receptor clustering is required for platelets to bind Mmrn1, as activation enhances human platelet adhesion to MMRN1. 5, 6 We suggest that Mmrn1 works synergistically with other proteins, including Fg and Vwf, to enhance platelet adhesion onto a growing platelet aggregate. The binding site(s) for MMRN1 on FG are unknown, but Mmrn1 does not appear to inhibit the ability of platelets to adhere to immobilized Fg, which occurs predominantly via the Fg γ-chain binding to α IIb β 3 . 48 Further, we did not detect an in- shear-stretched Vwf multimers to form larger, heteropolymers that enhance platelet tethering and adhesion. As Mmrn1 loss reduced the size of platelet aggregates that adhered to collagen, rVwf, Fg, and GFOGER peptides, we suggest that Mmrn1 also functions to stabilize platelet-platelet interactions in wounds and injured vessels. Figure 8A summarizes the proposed role of Mmrn1 in platelet adhesion.
Additionally, Mmrn1-GPAGPOGPX binding could be one of the VWFindependent mechanisms that mediates platelet adhesion and aggregate formation on collagen under low shear flow conditions. 62 The in vivo defects in platelet adhesion and platelet-rich thrombus formation of Mmrn1 −/− and Mmrn1 +/− mice may reflect other defects as collagen exposure in FeCl 3 injured vessels appears minimal, 40

RWF is Scientific Advisor and AB is the Lead Peptide Scientist at
CambCol Laboratories Ltd. The other authors declare that they have no real or perceived conflicts of interest to disclose.

AUTH O R CO NTR I B UTI O N S
A. Leatherdale and D. Parker designed and conducted experiments, analyzed and interpreted the findings, and led the manuscript F I G U R E 8 Proposed model of multimerin 1 (Mmrn1) functions in platelet adhesion and an updated model of the motifs in types I and III vessel wall fibrillar collagens that support platelet adhesion or activation. A, Proposed role of Mmrn1 in platelet adhesion. Following Mmrn1 release from platelets and endothelial cell storage granules, α IIb β 3 , α v β 3 , and other unidentified receptors mediate Mmrn1 binding to platelets. Mmrn1 binding to platelets increases the size and stability of platelet aggregates that are captured onto the macromolecular protein complexes that promote platelet-matrix and platelet-platelet interactions. Shear-exposed and matrix bound von Willebrand factor (Vwf), fibrin(ogen), and fibronectin in these macromolecular protein complexes provide multiple binding sites for Mmrn1 attachment. Additionally, at sites of injury that expose blood to fibrillar collagen, Mmrn1 binds to GPAGPOGPX motifs to synergistically increase platelet adhesion to collagen beyond the adhesion supported by Vwf-and α 2 β 1 -dependent mechanisms. B, Scale representation of the spatial arrangement of the functional sequences in the triple-helical (COL) domain of human type I (top) and type III (bottom) collagen that support platelet adhesion or activation. *The α1-α1-α2 heterotrimer consisting of GPRGQAGVMGFO and GPRGEOGNIGFO in type I collagen has been verified to support VWF binding using heterotrimeric triple-helical peptides. 72 GXOGER sequences that bind α 2 β 1 , the VWF-binding sequence GPRGQOGVMGFO, and the GPVI-binding sequence GAOGLRGAGPOGPEGGKGAAGPOGPO have been previously described elsewhere [Colour figure can be viewed at wileyonlinelibrary.com]