Neutrophil cathepsin G proteolysis of protease-activated receptor 4 generates a novel, functional tethered ligand

Human platelets express 2 G-protein coupled, protease-activated receptors (PAR1 and PAR4).¹  Although many serine proteases activate platelets through PARs,²  thrombin is the best studied. PARs are activated by proteolysis of their amino termini.¹  The resulting new N-terminus functions as a tethered ligand that binds to an extracellular loop of the PAR,^3,4  resulting in platelet activation. Compared with PAR1, PAR4 has been less studied, but recent findings emphasize its importance and potential as a therapeutic target.²  PAR1 has a higher affinity for thrombin, and Ca²⁺ transients rise and fall sharply after activation. In contrast, PAR4 acts as an “amplifier,” inducing a gradual but sustained rise in Ca²⁺ that accounts for greater intracellular Ca²⁺ flux, thrombin generation, and fibrin formation under shear stress.^5,6 

Elevated blood neutrophil levels correlate with ischemic arterial events,⁷  and platelet-leukocyte aggregates are elevated during a stroke.⁸  Activated neutrophils release the serine protease cathepsin G (CatG) at sites of injury and inflammation.⁹  CatG is a potent human PAR4-dependent platelet agonist,^10-13  but it is unknown where CatG cleaves PAR4 and how this cleavage might activate the receptor.

A more detailed methods section is provided in the supplemental Data.

Platelets and neutrophils were isolated as previously described.¹⁴  Platelet isolation from dogs, rats and mice was approved by the IACUC. Platelet activation and aggregation were performed as previously described.¹⁵ 

PAR4-dependent calcium flux was measured using Fura2-QBT in platelets and HEK293T/17 cells with wild-type (WT) and mutant PAR4.¹⁶ 

PAR4 synthetic peptides were incubated with protease, snapped frozen, and sent for mass spectrometry analysis.

Of the major serine proteases released from neutrophils, CatG was the only protease to induce platelet aggregation and αIIbβ3 activation (supplemental Figure 1A-D). Supplemental Figure 1A shows the effect of neutrophil activation on platelet activation is mediated in large part by CatG. Using PAR-specific inhibitors, we observed CatG-dependent platelet activation was mediated by PAR4 and not PAR1 (Figure 1A; supplemental Figure 2). Because platelet-neutrophil interactions are critical in stroke pathophysiology and a common PAR4 Ala120Thr variant has been associated with human stroke risk,^17,18  the effect of this variant on CatG-induced platelet activation was studied. Compared with platelets from homozygous Ala¹²⁰ individuals, CatG stimulation of platelets homozygous for the racially divergent and hyperreactive PAR4 Thr¹²⁰ variant demonstrated modest but significantly greater αIIbβ3 activation and P-selectin expression (Figure 1B; supplemental Figure 3). Thus, neutrophil-released CatG may contribute to the increased stroke risk associated with the PAR4Thr120 variant.

Thrombin cleavage of PAR4 at Arg⁴⁷-Gly⁴⁸ is blocked by the monoclonal antibody RC3.⁶  RC3 had little effect on CatG-induced platelet aggregation despite largely abolishing thrombin-induced aggregation (Figure 1C-D; supplemental Figure 4) (note thrombin activation of PAR1 was blocked in Figure 1C-D, as determined by supplemental Figure 4). These results suggest CatG cleaves PAR4 at a site different than thrombin. It is not clear why our data apparently differ from previous observations using a polyclonal antibody to PAR4^6,13  since the antibodies were raised against the same sequence, but perhaps the polyclonal antibody also blocked the CatG cleavage site while monoclonal RC3 does not.

To examine the possible cleavage fragments generated by CatG, 2 portions of the PAR4 extracellular N-terminus were synthesized and analyzed by LC-MS/MS: PAR4-B containing the thrombin cleavage site, and PAR4-C, downstream of the thrombin cleavage site (Figure 1E). As expected, PAR4-B was cleaved by thrombin between amino acids Arg⁴⁷ and Gly⁴⁸, while little cleavage was observed with CatG (supplemental Figure 5). In contrast, thrombin failed to cleave PAR4-C, while CatG cleavage generated several fragments, including DSDTLELPSS (Figure 1G; and data not shown). Identification of the DSDTLELPSS fragment (amino-terminal sequence of PAR4-C) indicated CatG cleavage between Ser⁶⁷ (S67)-Arg⁶⁸ (R68). As a control, in the absence of thrombin or CatG, PAR4-B and PAR4-C remained intact in these analyses (Figure 1F; supplemental Figure 5).

To specifically examine if CatG induces PAR4 activation and signaling by cleaving Ser⁶⁷-Arg⁶⁸, we assessed calcium flux using various PAR4 mutants. HEK cells transfected with WT PAR4 demonstrated increased calcium flux when treated with CatG compared with mock transfection (Figure 1H). CatG activation of HEK cells expressing either PAR4 mutated Ser⁶⁷ to Glu (S67E) or Arg⁶⁸ to Glu (R68E) resulted in significantly less calcium flux than WT (P < .001) (Figure 1H). Notably, the PAR4 S67E and R68E mutants induced minimal calcium flux above baseline, such that we cannot exclude other minor CatG cleavage sites indicated by our mass spectrometry data (Figure 1H; supplemental Figure 5; and data not shown) could induce receptor activation, as has been shown with elastase-cleavage of PAR2.¹⁹  Since neighboring amino acids can be critical in protease cleavage,²⁰  a Ser⁶⁶ (S66E) mutant was generated that showed a slight decrease in CatG-induced calcium flux compared with WT and a significantly greater calcium flux than mock transfection (Figure 1H). None of the PAR4 mutants were significantly different than WT when stimulated by thrombin in the presence of a PAR1 inhibitor (Figure 1I). Taken together, these data support CatG, but not thrombin, cleavage of PAR4 between Ser⁶⁷-Arg⁶⁸.

A synthetic peptide RALLLGWVPTR representing the CatG-generated PAR4 tethered ligand was synthesized to assess its potential functionality. RALLLGWVPTR induced calcium flux in HEK cells transfected with WT PAR4 compared with mock, indicating the tethered ligand induces signaling through PAR4 (Figure 2A). Platelets treated with RALLLGWVPTR demonstrated increased integrin activation, α granule release, and platelet aggregation (Figure 2B-E). In addition, RALLLGWVPTR induced a significant and sustained calcium flux in platelets similar to other peptide-tethered ligands²¹  (Figure 2F). Peak calcium flux was similar between CatG and RALLLGWVPTR, and consistent with previous literature, CatG-induced platelet calcium flux decreased over time¹³  (supplemental Figure 6).

LC-MS/MS analysis also indicated a minor CatG cleavage between Arg⁶⁸ and Ala⁶⁹ under the conditions of our experiment, so we also synthesized an ALLLGWVPTR 10mer. However, platelets exposed to ALLLGWVPTR displayed no discernable platelet aggregation or calcium flux (Figure 2D-E; supplemental Figure 6) independent of peptide concentration (supplemental Figures 6 and 7). Notably, 1 mM RALLLGWVPTR produced maximal platelet aggregation, while 6 mM GYPGQV (the thrombin-generated PAR4 tethered ligand) was necessary to induce maximal platelet aggregation consistent with previous literature, demonstrating high concentrations of GYPGQV are needed to induce receptor activation (Figure 2D-E; supplemental Figure 7). The CatG-generated tethered ligand appears to be a more potent activator of PAR4 than the thrombin-generated tethered ligand, but the CatG enzyme is much less potent than the thrombin enzyme. Perhaps thrombin has greater affinity for the PAR4 extracellular domain than does CatG.

PAR4 signaling can be induced by the thrombin-tethered ligand with as few as 6 amino acids (GYPGQV), but a 6mer from the CatG tethered ligand (RALLLG) was not able to induce platelet aggregation (Figure 2G-H). We also synthesized an intermediate length peptide, RALLLGWV 8mer, but were unable to test its activating potential due to its inability to go into solution despite numerous attempts with different solvents. Thus, we cannot exclude the possibility that the minimal tethered ligand length required for PAR4 activation in vivo may be less than 11 residues. Having said that, perhaps the hydrophobic portion of the C-terminus of RALLLGWVPTR stabilizes the tethered ligand on the platelet plasma membrane.

Conservation of amino acid sequence across species can be supportive and informative of sequence function. Supplemental Table 1 shows the sequence alignments of PAR4 N-terminal from 12 animals, 4 of which are shown in Figure 2I. Human thrombin induced platelet aggregation in humans, dogs, mice, and rats (Figure 2J), whereas human CatG only induced platelet aggregation in humans and dogs (Figure 2K). Similar to CatG, the RA-11mer induced aggregation only in humans and dogs (Figure 2L). Dog platelets were not as responsive to human CatG, perhaps due to inefficient binding of human CatG to dog PAR4 or the substitution of an alanine for a proline in the dog PAR4 sequence. Furthermore, the human RA-11mer could be less effective binding to the dog PAR4 extracellular domain that mediates the signaling response to the dog tethered ligand. This genetic data lends support to the mutagenesis and functional peptide data regarding the importance of Arg⁶⁸ in mediating CatG-mediated PAR4 cleavage and activation. Perhaps the positive charge of Arg68 is important for the activity of the tethered ligand (mouse Gln is uncharged while rat Glu is negatively charged).

There are a relatively small number of serine protease-generated functional PAR tethered ligands. Identifying the potentially diverse array of PAR tethered ligands may be important if they activate different cellular signaling pathways or vary by pathophysiologic conditions. In certain clinical settings, such as those with extensive platelet-neutrophil interactions or leukocytosis in malignancy, CatG may dampen or eliminate thrombin signaling through PAR1 and PAR4, while the CatG tethered ligand could induce persistent PAR4 signaling.²²  PAR4 is also expressed in nonplatelet tissues, where inflammation alters function and expression levels, and CatG could affect activation.^23,24  Lastly, in hemorrhagic conditions, the RALLLGWVPTR peptide could be therapeutic for rescuing or enhancing platelet reactivity.

The authors wish to acknowledge Mark Cody for his technical assistance with experiments; Orvelin Roman, Ravi Ranjan, and Candace Reno for their assistance obtaining dog and rat platelets; Cynthia David from the University of Arizona, Analytical and Biological Mass Spectrometry Core Facility; James Cox from the University of Utah Mass Spectrometry Core for assistance with peptide analysis; and Guy Zimmerman for his thoughtful insight on experiments. The authors are grateful for figure preparation expertise from Diana Lim.

This work was supported by grants from the National Institutes of Health (R01HD093826 [C.C.Y.], R01HL102482 [P.F.B.], and K01AG059892 [R.A.C.]), the American Heart Foundation (21POST830138 [F.D.]), and an American Society of Hematology Bridge Grant (DSS-10060581 [P.F.B.]). Research reported in this publication was supported by the Flow Cytometry Core at the University of Utah and the National Center for Research Resources of the National Institutes of Health under Award Number 1S10RR026802-01.

Contribution: M.L.S., P.F.B., and R.A.C. conceived and designed the experiments; M.L.S., I.B., F.D., J.W.R., J.A., K.P., M.T.N., C.C.Y., J.R.H., P.F.B., and R.A.C. performed experiments; M.L.S., I.B., K.P., P.F.B., and R.A.C. analyzed results and made the figures; M.L.S., P.F.B., and R.A.C. wrote the manuscript; and all authors reviewed and critically edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Robert A. Campbell, University of Utah Health Sciences Center Eccles Institute of Human Genetics, 15 North 2030 East, Building 533, Suite 4225, Salt Lake City, UT 84112; e-mail: rcampbell@u2m2.utah.edu.