Platelets release pathogenic serotonin andreturn to circulation after immunecomplex-mediated sequestrationNathalie Cloutiera,1, Isabelle Allaeysa,1, Genevieve Marcouxa, Kellie R. Machlusb, Benoit Mailhotc, Anne Zuffereya,Tania Levesquea, Yann Beckera, Nicolas Tessandiera, Imene Melkia, Huiying Zhid, Guy Poiriere, Matthew T. Rondinaf,Joseph E. Italianob, Louis Flamanda, Steven E. McKenzieg, Francine Coteh, Bernhard Nieswandti, Waliul I. Khanj,k,Matthew J. Flickl, Peter J. Newmand, Steve Lacroixc, Paul R. Fortinm, and Eric Boilarda,n,2

aCentre de Recherche du Centre Hospitalier Universitaire de Québec-Université Laval, Département de microbiologie et immunologie, Faculté deMédecine, Université Laval, QC, Canada G1V 4G2; bDivision of Hematology, Department of Medicine, Brigham and Women’s Hospital and Harvard MedicalSchool, Boston, MA 02115; cCentre de Recherche du Centre Hospitalier Universitaire de Québec-Université Laval, Département de Médecine Moléculaire,Faculté de Médecine, Université Laval, QC, Canada G1V 4G2; dBlood Research Institute, Milwaukee, WI 53213; eCentre de Recherche du Centre HospitalierUniversitaire de Québec-Université Laval, Département d’Oncologie, Faculté de Médecine, Université Laval, QC, Canada G1V 4G2; fDepartment of InternalMedicine, University of Utah, Salt Lake City, UT 84112; gCardeza Foundation for Hematological Research, Thomas Jefferson University, Philadelphia,PA 19107; hInstitut Imagine, INSERM U1163, CNRS ERL8254, Université Paris Descartes, Hôpital Necker, 75006 Paris, France; iDepartment of ExperimentalBiomedicine, University Hospital and Rudolf Virchow Center, 97080 Wuerzburg, Germany; jFarncombe Family Digestive Health Research Institute, McMasterUniversity, Hamilton, ON, Canada L8S 4L8; kDepartment of Pathology & Molecular Medicine, McMaster University, Hamilton, ON, Canada L8S 4L8; lCancerand Blood Diseases Institute, Division of Experimental Hematology and Cancer Biology, Cincinnati Children’s Hospital, Cincinnati, OH 45229; mCentre derecherche du Centre Hospitalier Universitaire de Québec-Université Laval, Département de médecine, Faculté de médecine, Université Laval, QC, CanadaG1V 4G2; and nCanadian National Transplantation Research Program, Edmonton, AB, Canada T6G 2E1

Edited by Barry S. Coller, The Rockefeller University, New York, NY, and approved January 5, 2018 (received for review December 1, 2017)

There is a growing appreciation for the contribution of platelets toimmunity; however, our knowledge mostly relies on plateletfunctions associated with vascular injury and the prevention ofbleeding. Circulating immune complexes (ICs) contribute to bothchronic and acute inflammation in a multitude of clinical conditions.Herein, we scrutinized platelet responses to systemic ICs in theabsence of tissue and endothelial wall injury. Platelet activation bycirculating ICs through a mechanism requiring expression of plateletFcγ receptor IIA resulted in the induction of systemic shock. IC-drivenshock was dependent on release of serotonin from platelet-densegranules secondary to platelet outside-in signaling by αIIbβ3 and itsligand fibrinogen. While activated platelets sequestered in the lungsand leaky vasculature of the blood–brain barrier, platelets also se-questered in the absence of shock in mice lacking peripheral sero-tonin. Unexpectedly, platelets returned to the blood circulation withemptied granules and were thereby ineffective at promoting sub-sequent systemic shock, although they still underwent sequestra-tion. We propose that in response to circulating ICs, platelets are acrucial mediator of the inflammatory response highly relevant tosepsis, viremia, and anaphylaxis. In addition, platelets recirculateafter degranulation and sequestration, demonstrating that in adap-tive immunity implicating antibody responses, activated plateletsare longer lived than anticipated and may explain platelet countfluctuations in IC-driven diseases.

platelets | immune complexes | Fc receptor | serotonin | thrombocytopenia

Platelets are best known for their involvement in hemostasis.Their abundance in blood underlies their perfect positioning

for constant surveillance of the endothelium to ensure vascularintegrity. The role of platelets, however, is not restricted to thehemostatic response (1). Not only do platelets express a vastarray of mediators serving wound repair but they also possessimmune receptors and inflammatory molecules (1–3). Moreover,following tissue injury or infection, platelets are a central effectorin driving an inflammatory process for recruiting leukocytes,principally neutrophils, to the affected site (4–6). The plateletcontribution at an inflamed site also includes the prevention ofbleeding (7, 8), as platelets seal the breaches formed during theextravasation of neutrophils (9).Platelet activation occurs at the interface with the vascula-

ture, is elicited by injury or inflammation, and is restricted to the

endothelium or tissue. However, given the great number ofplatelets in blood and their extensive set of immune receptors,systemic immune triggers can also activate platelets while incirculation (i.e., in the absence of endothelial or tissue injury). Inrheumatic diseases, such as rheumatoid arthritis (RA), systemiclupus erythematosus (SLE), and antiphospholipid syndrome,there is a prevalence of pathogenic antibody–antigen scaffolds,called immune complexes (ICs), in circulation in patients (10).Anaphylactic reactions due to allergens (11, 12) and i.v. ad-ministration of drugs, such as in heparin-induced thrombocyto-penia (HIT) (13), also implicate ICs, thereby provoking acute

Significance

Immune complexes (ICs) form when antibodies encounter theirantigens. ICs are present in blood in multiple pathologicalconditions. Given the abundance of platelets in blood and thatthey express a receptor for ICs, called Fcγ receptor IIA (FcγRIIA),we examined the impact of ICs in blood in a mouse model. Wefound that circulating ICs induced systemic shock, characterizedby loss of consciousness, by activating platelet FcγRIIA. Shockwas mediated by the liberation of serotonin, a molecule betterknown for its role in the brain, from platelet granules. Duringshock, platelets were sequestered in the lungs and brain andreturned to the blood circulation after their degranulation.Platelets are thus crucial in response to ICs.

Author contributions: N.C., I.A., K.R.M., and E.B. designed research; N.C., I.A., G.M.,K.R.M., B.M., A.Z., T.L., Y.B., N.T., I.M., and H.Z. performed research; G.P., M.T.R., J.E.I.,L.F., S.E.M., F.C., B.N., W.I.K., M.J.F., P.J.N., S.L., and P.R.F. contributed new reagents/analytic tools; N.C., I.A., G.M., K.R.M., B.M., A.Z., and E.B. analyzed data; and N.C., I.A.,and E.B. wrote the paper.

Conflict of interest statement: J.E.I. has a financial interest in and is a founder of PlateletBioGenesis, a company that aims to produce donor-independent human platelets fromhuman-induced pluripotent stem cells at scale. J.E.I. is an inventor on this patent. Theinterests of J.E.I. were reviewed and are managed by the Brigham and Women’s Hospitaland Partners HealthCare in accordance with their conflict of interest policies.

This article is a PNAS Direct Submission.

Published under the PNAS license.1N.C. and I.A. contributed equally to this work.2To whom correspondence should be addressed. Email: eric.boilard@crchudequebec.ulaval.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1720553115/-/DCSupplemental.

E1550–E1559 | PNAS | Published online January 31, 2018 www.pnas.org/cgi/doi/10.1073/pnas.1720553115

Dow

nloa

ded

by g

uest

on

July

10,

202

1

inflammatory responses and a rapid drop in platelet count(thrombocytopenia). ICs can also form in blood during sepsisand viremia, when antibodies in the immune host recognizemicrobial antigens or their toxins.The pathological link between ICs and cellular responses is

mediated by members of the Fcγ receptor (FcγR) family. The low-affinity receptor FcγRIIA is the sole FcγR expressed by plateletsin humans, and is thereby the most abundantly expressed receptorin blood (12, 14, 15). Of note, it is the dominant receptor activatedby Zika virus if in the presence of sera from convalescent patientsinfected by dengue or West Nile virus, pointing to the formation ofICs in blood in patients with preexisting antiflavivirus immunity(16). Moreover, diverse strains of bacteria and the influenza virusH1N1 trigger platelet FcγRIIA activation after forming ICs in animmune host (17, 18). In addition, studies with human plateletsshow that ICs from patients with SLE also activate plateletsthrough FcγRIIA (19). However, mice do not express FcγRIIA,and murine platelets are completely devoid of any FcγRs (12).Thus, IC-mediated responses in mice strongly favor leukocytes,which express other FcγRs, while platelets play small or accessoryroles in mouse models implicating ICs.To more accurately model FcγRIIA reactions in mice, trans-

genic FcγRIIA (FcγRIIATGN) mice that display FcγRIIA ex-pression on platelets and certain leukocytes, including monocytesand neutrophils as in humans, were developed (12, 20). Notably,stimulation of FcγRIIA in transgenic mice induces a shock re-sponse that is reminiscent of anaphylaxis in humans, with a strongreliance on neutrophil activity and implicating release of platelet-activating factor (PAF) (14, 21–24).How ICs trigger systemic responses is not completely un-

derstood, as no study has definitively determined the contribu-tion of platelets to this response in vivo. Here, we employedFcγRIIATGN mice to elucidate the precise mechanisms by whichplatelets contribute to IC-mediated shock.

ResultsPlatelets Are Critical During the Systemic Response. ICs injected intoFcγRIIATGN mice trigger systemic shock (22–24). Althoughthrombosis in the lungs and thrombocytopenia accompany sys-temic shock in these mice (22, 25–27), the platelet contributionto IC-induced shock has never been formally assessed. We usedheat-aggregated (HA) IgG as an IC surrogate (28, 29), as itpermits precise investigation of the role of FcγR activation withno bias owing to the nature of antigen specificities. HA-IgG(160 ± 7 nm in diameter) was injected i.v. into WT mice lack-ing FcγRIIA (FcγRIIAnull) and FcγRIIATGN mice. Significanttemperature loss (15 min) preceded by rapid significant systemicshock (3 min) in all of the FcγRIIATGN mice was confirmed, butit was never observed in the FcγRIIAnull mice (Fig. 1A andMovie S1). All of the FcγRIIATGN mice collapsed, underwentprofound immobility, and lost consciousness (Fig. S1A). Micerecovered from the shock and hypothermia within 180 min, afterwhich no residual signs were apparent (Fig. S1B). The occur-rence of shock was dependent on the concentration of ICspresent in blood, not induced by monomeric IgG, similarly in-duced when murine ICs were used, and more profound in maleFcγRIIATGN mice compared with females (Fig. S1 C–E).To formally assess the contribution of platelets to the sys-

temic shock, we depleted platelets (>98%) before injection ofICs. Thrombocytopenic mice were completely protected fromIC-mediated systemic responses (Fig. 1B and Fig. S1F), identi-fying a critical role for platelets in shock induced by circulatingICs. In platelets, efficient αIIbβ3 signaling relies on theFcγRIIA immunoreceptor tyrosine-based activation motif, and,accordingly, the expression of FcγRIIA by platelets enhancesfibrinogen-mediated platelet activation (30). Thus, we verifiedthe functional association of FcγRIIA and the αIIbβ3 receptorin vivo by comparing the response in FcγRIIATGN/β3−/− and

FcγRIIATGN/β3+/+ mice. The FcγRIIATGN/β3−/− mice were com-pletely resistant to IC challenge (Fig. 1C), further emphasizing therole of platelets and pointing to the requirement of αIIbβ3 in themechanism of IC-mediated shock.While αIIbβ3 may amplify FcγRIIA signaling through com-

mon kinases located in the platelet cytoplasm, it is predominantlyknown as the fibrinogen receptor, which may contribute uponbinding to platelet activation (31–33). Fibrinogen also mediatesplatelet–neutrophil interactions by bridging αIIbβ3 and macro-phage-1 antigen (Mac-1) (34). Hence, platelet–neutrophil ag-gregates formed in blood in the presence of ICs (Fig. S2A), aswell as neutrophils, were also necessary in the systemic responses(Fig. S2B), in agreement with prior studies (23, 28). To verify thespecific role of fibrinogen, we used mutant fibrinogen chi-mera mice. We lethally irradiated fibrinogenγΔ5 (FibγΔ5) andfibrinogenγ390-396A (Fibγ390-396A) mutant mice, which express mu-tant fibrinogen unable to bind αIIbβ3 or Mac-1 (34, 35), re-spectively, and engrafted the mice with bone marrow fromFcγRIIATGN mice. After injection with ICs, we found that fibrin-ogen binding to αIIbβ3 was necessary for the systemic response,whereas bridging platelets with leukocytes through fibrinogenbinding to Mac-1 was dispensable (Fig. 1D and E). Consistent withthese data, the blockade of Mac-1 as well as PSGL-1, the coun-terreceptor of platelet P-selectin on leukocytes (36), had no effecton shock (Fig. S2C).

Fig. 1. Platelets are critical during the systemic response. Systemic shock(Upper) (scores defined in Materials and Methods) and temperature (Lower)were measured at the indicated time points following IC injection in FcγRIIATGN

and FcγRIIAnull mice (A; n = 12); FcγRIIATGN mice preinjected with platelet-depleting antibody (PLT Ab) and isotypic control antibody (Ctrl Ab) (B; n =8); FcγRIIATGN/β3+/+ and FcγRIIATGN/β3−/− mice (C; n = 15); bone marrow chi-meric mice generated by transfer of FcγRIIATGN cells into WT (D and E; nativefibrinogen), FibγΔ5 (D), and Fibγ390-396A (E) irradiated mice (n = 6); FcγRIIATGN

mice preinjected with GPIb Fab (Xia.B2) or diluent (F; n = 6); FcγRIIATGN micepretreated with aurintricarboxylic acid (ATA) or diluent (G; n = 8); andFcγRIIATGN mice pretreated with alteplase (H; n = 4) or diluent. null, FcyRIIAnull;TGN, FcyRIIATGN. Data are mean ± SEM. **P < 0.005, ***P < 0.001, and****P < 0.0001; repeated-measures two-way ANOVA, statistical variation be-tween groups (A–H).

Cloutier et al. PNAS | Published online January 31, 2018 | E1551

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

10,

202

1

Like αIIbβ3, GPIb is critical in the prevention of bleeding andthrombosis, and the ablation of the gene coding for GPIb in miceleads to severe bleeding defects (37). As GPIb is localized withinlipid raft membrane microdomains in physical proximity withFcγRIIA (38), we used a GPIb-blocking Fab to probe the con-tribution of GPIb, and found that it is dispensable in shockmediated by ICs (Fig. 1F). Furthermore, the pharmacologicalblockade of von Willebrand factor (vWF) binding to GPIb(ATA; aurintricarboxylic acid) or administration of recombinanttissue plasminogen activator (tPA; alteplase) to promote the lysisof any existing thrombi also had no effect on shock (Fig. 1 G andH). These data further demonstrate that platelet activation bycirculatory ICs implicates pathways distinct from those pro-moting vaso-occlusion and vascular injury.

Identification of the Shock Mediator. Consistent with the rapidresponse, an increase in platelet factor 4 (PF4) and serotonin,components stored in α- and δ-granules, respectively, was de-tected in blood within 10 min (at the time of mouse collapse andbefore hypothermia) of the IC trigger in FcγRIIATGN mice (Fig.2A). We therefore hypothesized that a component released fromplatelet granules may be responsible for induction of systemic

shock. Intravital microscopy of the microvasculature using two-photon microscopy revealed significant vessel leakage andvasodilatation in response to IC injection in FcγRIIATGN mice(Fig. 2 B and C). Vessel leakage was systematically observedin response to IC injection in both FcγRIIATGN mice andFcγRIIAnull mice, but was absent in FcRγ−/− mice, pointing tothe role of other receptor(s) for ICs (Fig. S3A). It was alsomaintained in the absence of neutrophils, further suggesting thatvascular leakage is not the key pathological feature downstreamof FcγRIIA driving IC/platelet-mediated shock (Fig. S3A).In stark contrast, vasodilatation was a response observed ex-

clusively in FcγRIIATGN mice (Fig. 2C). Identified first as a serumagent mediating vascular tone, serotonin is a powerful vasocon-strictor when added to smooth muscle cells. However, it is a po-tent vasodilator on endothelial cells (39). In the case of systemicactivation in the absence of vascular injury, and thus an intactendothelium, we hypothesized that serotonin might be the keyplatelet component driving vasodilatation. This is supported by thefact that the majority of peripheral serotonin, which represents95% of the total body serotonin pool, is stored in plateletδ-granules (40, 41). Platelets do not synthesize serotonin; theyutilize their serotonin transporter (SERT) to capture circulatoryserotonin, with the latter being generated by enterochromaffincells from the digestive tract by the enzyme tryptophan hydroxy-lase 1 (Tph1) (42, 43). Consistent with our hypothesis, vasodila-tation was abrogated in FcγRIIATGN/Tph1−/− mice in response toICs (Fig. 2C). It was also maintained in absence of neutrophils,further confirming the direct role of peripheral serotonin in IC-induced vasodilatation of microvessels (Fig. S3B).Serotonin levels in plasma were back to normal 1 h postshock

(Fig. S3C). To directly verify its role in systemic shock, serotoninwas injected into FcγRIIAnull mice, which resulted in a shock re-sponse reminiscent of the responses observed in IC-injected mice(Fig. 2D). We then used a selective serotonin reuptake inhibitor(SSRI), a SERT blocker used as an antidepressant, to inhibit se-rotonin storage by platelets. Administration of the SSRI for 3 wkto deplete the serotonin content of δ-granules (41) revealed theimportant contribution of serotonin uptake in IC-mediated in-flammation: SSRI treatment nearly abolished the systemic shockresponse (Fig. 2E). Furthermore, FcγRIIATGN/Tph1−/− micelacking platelet serotonin (43) showed almost complete resistanceto shock (Fig. 2F), confirming the critical role of peripheral se-rotonin in IC-mediated shock. In addition, blockade of the5-hydroxytryptamine 2 receptor family, which is expressed in theperiphery by platelets, but also by monocytes and macrophages,dendritic cells, eosinophils, B and T lymphocytes, endothelial cells,fibroblasts, cells from the cardiovascular system, and neurons inthe peripheral nervous system (40), was effective at reducing shock(Fig. 2G). In support of the potential role of αIIbβ3 in de-granulation, serotonin was not released in FcγRIIATGN/β3−/− miceafter IC injection, in agreement with the absence of shock in thosemice (Fig. S3D). Key actors in the prevention of bleeding andthrombosis, the platelet-derived mediators ADP and thromboxaneA2, were dispensable for shock (Fig. S3 E and F). In addition,neutrophil extracellular traps, which can be present in thrombosis(44), were detected during shock according to quantifications ofcirculating nucleosomes (Fig. S3G); however, the injection ofDNase did not protect mice from shock (Fig. S3H). Together,these observations highlight the importance of platelet-derivedserotonin in the systemic response to ICs in vivo, a platelet re-sponse that is distinct from that traditionally observed in hemo-stasis and thrombosis.

And the Platelet Count? Circulating platelets were also monitoredthroughout the systemic response and beyond. We observed thatFcγRIIATGN mice rapidly underwent profound thrombocytope-nia (∼10% of total normal platelet count) in response to ICs,whereas FcγRIIAnull mice presented with only very minimal or

Fig. 2. Serotonin is the mediator of shock. (A) PF4 (Left) and serotonin(Right) concentrations were measured in platelet-free plasma prepared fromblood of FcγRIIATGN mice collected 10 min after IC injection (n = 5). Thebaseline (measured in nonchallenged FcγRIIATGN mice) concentration is in-dicated using a dotted line. Dil, diluent. (B and C) Intravital microscopy re-veals profound changes to the mouse ear vasculature in response to ICs.(B) Vascular leakage, evidenced by the presence of Evans Blue (magenta)outside blood vessels in the subendothelial matrix rich in collagen (blue), isrepresented using FcγRIIATGN. (Scale bar: 75 μm.) (C) Vasodilatation wasmeasured prior to leakage (t = 8 min) in FcγRIIATGN, FcγRIIAnull, or FcγRIIATGN/Tph1−/− mice (n > 5 vessels per field in three mice per group). (D–G) Systemicshock (Upper) and temperature (Lower) were measured at the indicatedtime points following injection of serotonin (Sero) or its Dil in FcγRIIAnull

mice (D; n = 6), after IC injection in FcγRIIATGN mice treated or not treatedwith the SSRI fluoxetine (E; n = 10), after IC injection in FcγRIIATGN/Tph1+/+

and FcγRIIATGN/Tph1−/− mice (F; n = 9), and after injection of ICs in FcγRIIATGN

mice pretreated with the 5-hydroxytryptamine receptor 2 blocker ketanserinor Dil (G; n = 5). null, FcyRIIAnull; TGN, FcyRIIATGN. Data are mean ± SEM.*P < 0.05, **P < 0.005, ***P < 0.001, and ****P < 0.0001, using an unpairedt test (A); one-way ANOVA (C); and repeated-measures two-way ANOVA,statistical variation between groups (D–G).

E1552 | www.pnas.org/cgi/doi/10.1073/pnas.1720553115 Cloutier et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

1

no changes in platelet counts (Fig. 3A). In fact, thrombocyto-penia occurred rapidly (<10 min), and the platelet count grad-ually increased, reaching ∼20% by 60 min, and it was 50%resolved within 24 h (Fig. 3A). The occurrence of thrombocyto-penia was critically dependent on the expression of FcγRIIA andwas similarly induced in males and females (Fig. S4A). Onlyplatelets were affected, as the levels of RBCs, monocytes, andneutrophils were unchanged in response to ICs (Fig. S4 B–D).Thrombocytopenia was as profound in native fibrinogen mice as inFibγΔ5 and Fibγ390-396A mice (Fig. 3B), suggesting that fibrinogenbinding was dispensable, and that thrombocytopenia only mod-estly involved β3, potentially through binding to its other ligands.Conversely, thrombocytopenia was unaltered by the inhibition ofGPIb and vWF interactions, or the blockade of thromboxane A2synthesis and ADP (Fig. 3B). Thrombin–antithrombin complexesand D-dimers, which are evidence of coagulation activation andthrombus degradation, respectively, were not significantly elevated24 h after shock (Fig. S4 E and F), and consistent with this, thedestruction of potential thrombi by the injection of alteplase didnot impact platelet count (Fig. 3B). These data further disso-ciate the platelet response in coagulation and thrombosis fromIC-induced thrombocytopenia.Platelet conversion to microparticles also did not explain the

profound thrombocytopenia, as microparticle levels only in-creased 60 min after the IC trigger (Fig. S4G). The decrease inplatelet number could also not be attributed to platelet in-teraction with leukocytes, as neutrophil depletion, or blockade ofMac-1 and PSGL-1, had no effect on thrombocytopenia (Fig. S4H and I). Given that FcγRIIATGN/Tph1−/− mice also underwentprofound thrombocytopenia (Fig. 3B), and that exogenous se-rotonin did not induce thrombocytopenia (Fig. S4J), this suggeststhat serotonin serves as the effector of IC-mediated shock but isnot the driver of the resulting thrombocytopenia. Thus, althoughthrombocytopenia occurs concurrently with shock, it is notcausative in the mechanism of inducing the shock.The unanticipated rapid recovery in platelet count prompted a

detailed analysis of the platelets present in blood 24 h after ICchallenge. We found that platelets were not activated or apo-ptotic, as evidenced by the lack of phosphatidylserine (PS) andP-selectin on their surface (Fig. 3C). However, the platelet granulecontent was markedly reduced (Fig. 3D), as ∼30% of the circu-lating platelets contained no detectable serotonin or PF4 (Fig.3E), suggesting significant degranulation. Electron microscopyfurther confirmed that circulating platelets were frequently de-void of any granules (Fig. 3F). Notably, platelets still expressedsurface FcγRIIA 24 h postshock (Fig. S5 A and B), and stillunderwent thrombocytopenia if challenged a second time withICs (Fig. S5C). This is in contrast to shock, where FcγRIIATGN

mice were resistant to shock induced by further challenges (Fig.3G), and neither serotonin nor PF4 was induced in the blood ofthese mice (Fig. 3H). Thus, we hypothesized that platelets cir-culating 24 h postshock were, in fact, platelets that had alreadydegranulated and had undergone temporary sequestration. Toverify this, we performed fluorescent labeling of FcγRIIATGN

platelets, which we adoptively transferred into FcγRIIATGN

mice. As expected, 87 ± 12% of fluorescent platelets, which werenegative for PS (Fig. S5D), rapidly became undetectable fromthe blood circulation after an IC trigger. Of particular impor-tance, 42 ± 16% of the fluorescently labeled FcγRIIA platelets,devoid of any surface PS, were identified in blood 24 h aftershock (Fig. 3I), confirming a temporary sequestration of plateletsfollowing FcγRIIA activation and their return to the blood cir-culation after degranulation.

Localization of Sequestration Sites. We next aimed to determinethe sites of platelet sequestration in response to FcγRIIA acti-vation. An intravital imaging system (IVIS) provided evidence offluorescently labeled platelets in the lungs (Fig. 4A), but not in

Fig. 3. Thrombocytopenia in response to ICs. (A and B) Blood samples werecollected at the indicated time points following IC injection, and CD41+

platelets in whole blood were counted using a flow cytometer. (A) Plateletcount was determined in FcγRIIATGN and FcγRIIAnull mice (n = 11). (B) Per-centages of circulating platelets 10 min after IC injection compared withinitial platelet count were calculated in FcγRIIAnull mice (black) and the in-dicated three different groups of FcγRIIATGN mice. Mice were pretreatedwith acetylsalicylic acid (ASA) (n = 4), apyrase (n = 6), aurintricarboxylic acid(ATA) (n = 3), alteplase (n = 5), or diluent (n = 14) (all shown in blue).FcγRIIATGN mice pretreated with GPIb Fab or control (n = 4), FcγRIIATGN/β3+/+

or FcγRIIATGN/β3−/− mice (n = 10); FcγRIIATGN/Tph1+/+ mice; or FcγRIIATGN/Tph1−/− mice (n = 4) are represented (all shown in purple). Bone marrowchimeric mice generated by transfer of FcγRIIATGN cells into WT (native fi-brinogen), FibγΔ5, and Fibγ390-396A irradiated (IRRAD) mice are represented(n = 3 per group) (shown in green). (C and D) Mice were injected with diluentor IC, and blood was collected 24 h later. (C) Exposition of PS and P-selectin(P-sel) was assessed on CD41+ platelets using flow cytometry in FcγRIIATGN

mice (n = 7). (D) Platelet content of PF4 (n = 6) and serotonin (n = 5) wasmeasured by ELISA in 106 platelets retrieved 24 h after shock. Baseline(measured in nonchallenged FcγRIIATGN mice) contents are indicated usingdotted lines. (E) PF4 and serotonin-negative cells were counted using im-munofluorescence microscopy (n = 3 different mice). Representative imageof Z-stack projections using confocal microscopy. Platelets that returned tocirculation 24 h after IC injection were used for quantification. Tubulin (red)was used as a platelet marker. PF4 (green) and serotonin (green) were ob-served in less than 60% of platelets. Empty platelets (arrowheads) andplatelets (arrows) are represented. (Scale bars: 2 μm.) As a negative control,serotonin labeling was performed on Tph1−/− platelets (Fig. S9). (F) Electronmicroscopy of FcγRIIATGN platelets before IC injection (Upper) and 24 h afterIC injection (Lower). Empty platelets (arrowheads) and platelets (arrows) arerepresented. (Scale bars: 0.8 μm.) (G) Systemic shock was measured inFcγRIIATGN mice injected with ICs at t = 0 and rechallenged at t = 24 h (n = 3).(H) Plasma levels of PF4 (n = 4) and serotonin (n = 5) were determined inmice rechallenged 24 h after the first challenge with ICs. Results werecompared with the level after the first challenge (dotted lines). (I) Plateletswere purified from FcγRIIATGN mouse blood, fluorescently labeled, andadoptively transferred i.v. in FcγRIIATGN mice. One hour later, mice wereinjected with ICs, and fluorescent platelets in circulation were determined atthe indicated time points following injection of ICs (n = 10). Dil, diluent; null,FcyRIIAnull; TGN; FcyRIIATGN. Data are mean ± SEM. **P < 0.005, ***P <0.001, and ****P < 0.0001, using an unpaired t test (A, D, and H); one-wayANOVA (B, C, E, and I); and repeated-measures two-way ANOVA, statisticalvariation between groups (G).

Cloutier et al. PNAS | Published online January 31, 2018 | E1553

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

10,

202

1

other locations, in agreement with previous studies (22, 25–27).Platelet thrombi populated with neutrophils were also evident inthe lungs (Fig. 4B), although no pulmonary edema was observed(Fig. 4C). Thrombi contribution to platelet sequestration andshock was not significant: ablation of the β3 gene, or blockade ofGPIb using Fab, significantly reduced thrombus formation inFcγRIIATGN mice (Fig. 4D) with only a modest or no impact onthrombocytopenia, respectively (Fig. 3B). Furthermore, seroto-nin had no effect on thrombus formation, as the number ofthrombi in the lungs remained unchanged in FcγRIIATGN/Tph1−/− mice (Fig. 4D). These data identify the lungs as theapparent major site for platelet sequestration, and highlight thatalthough thrombi are present in the lungs, these thrombi play norole in the thrombocytopenia and shock mediated by ICs.

However, we suspected that other anatomical sites of plateletsequestration might exist, as the IVIS approach may not permitan optimal distinction between immobilized platelet aggregatesand circulating platelets in the microvasculature, where smallplatelet aggregates are to be expected. In addition, if the entireplatelet population was sequestered in the lungs, it was puzzlingthat none of the mice died from the IC trigger. Using a quanti-tative approach to measure fluorescently labeled CD41+ plate-lets, we estimated that the lungs, in fact, contained only 16% ofthe total platelet load (Fig. 4E), confirming that other sites ofplatelet sequestration likely existed.The presence of platelets outside the vasculature was excluded

first, as no platelets were detected in the thoracic lymph (Fig.S6A). Furthermore, assessment using whole-mouse imaging ofyellow fluorescent protein (YFP) in FcγRIIATGN/CD41-YFPmice, which constitutively express YFP in CD41-expressingcells, efficiently identified thrombi in the lungs and megakaryo-cytes in the bone marrow, but did not reveal platelets outsideblood vessels or in other tissues (Fig. S6B). Therefore, wespeculated that tracking ICs in mice might be a more efficientmeans of leading us to platelet sequestration sites. We found thatICs not only localized to the lungs but were also observed in thebrain vasculature (Fig. 4F). Thus, the brain microvasculature wasexamined using two-photon microscopy in live FcγRIIATGN/CD41-YFP mice, with the inclusion of FcγRIIAnull/CD41-YFPmice for comparison. We observed profound leakage of the brainvasculature in both FcγRIIAnull and FcγRIIATGN mice wheninjected with ICs (Fig. 4G and Movies S2 and S3). Of impor-tance, in the presence of ICs, small platelet aggregates readilyformed in the leaky brain microvasculature, but only if FcγRIIAwas expressed by platelets (Fig. 4G and Movie S2). Thrombi werenot detected in the microvasculature of the kidney, liver, or spleenof FcγRIIATGN mice injected with diluent or ICs (Fig. S6C), andtwo-photon microscopy of the femurs in live FcγRIIATGN/CD41-YFP mice did not reveal any platelet aggregation in sinusoids inthe bone marrow (Fig. S6D). These data suggest that in thepresence of ICs, platelets sequester in certain microvascularbeds, notably in the lungs and brain, and that this event occursindependent of thrombosis and leakage.

Role of Platelet FcγRIIA and Serotonin in Acute InflammatoryResponses. During microbial invasion, foreign antigens are rec-ognized by host antibodies and form ICs. Hence, incubation ofhuman platelets in the presence of influenza virus or variousstrains of bacteria leads to platelet activation, which strictly re-quires the presence of plasma and FcγRIIA (17, 18). While theseobservations suggest that motifs on pathogens contribute to theformation of ICs, whether they can trigger platelet activation invivo has not been established.Antibodies against lipopolysaccharide (LPS), a common gram-

negative pathogen-associated molecular pattern (PAMP), weredetected in the blood of healthy volunteers similar to patientswith septic shock due to confirmed gram-negative bacteria in-fection (Fig. 5A). In contrast to humans, mice housed in a facilitywith high standards of cleanliness present no detectable anti-LPSantibodies (nonimmune mice) (Fig. 5B). Therefore, we immu-nized mice with small quantities of bacterial PAMP (LPS), which hadno perceptible effect on FcγRIIAnull or FcγRIIATGN mice (Fig. 5C).FcγRIIAnull and FcγRIIATGN mice developed equivalent anti-LPS antibody levels within 3 wk (LPS-immune mice) (Fig. 5B),and, importantly, a subsequent injection of LPS only inducedserotonin release and shock in FcγRIIATGN mice (Fig. 5 C andD). The shock induced by LPS-containing ICs was dependenton the presence of platelets, and was greatly diminished in theabsence of peripheral serotonin or β3 gene expression (Fig. 5E–G). Moreover, plasma serotonin levels were significantlyreduced in FcγRIIATGN/β3−/− mice after LPS injection (Fig. 5D).Thrombocytopenia was dramatically increased in the presence of

Fig. 4. Localization of sites of sequestration. (A) Mouse platelets fromwhole blood of FcγRIIATGN mice were labeled in vivo before IC injection. Thekidneys (Kid), spleen (Spl), liver (Liv), heart (He), lungs (Lu), and brain (Br)were harvested 10 min after IC injection, and fluorescence was measured ineach organ using an IVIS to determine platelet localization (n = 3). Resultswere compared with FcγRIIAnull mice injected with diluent, and are pre-sented as the percentage of fluorescence in diluent-injected FcγRIIAnull mice.(B) Lungs were collected 10 min after IC injection. Thrombi (star) andneutrophils (arrows) were examined by microscopy after hematoxylin andeosin coloration. (Magnification: 400×.) (C) FcγRIIATGN mice were injectedwith diluent (Dil) or ICs, and lungs were collected after 10 min. The lung wet-to-dry ratio was evaluated. (D) Number of lung thrombi per square millimeterwas quantified in FcγRIIAnull (n = 4), FcγRIIATGN (n = 12), FcγRIIATGN/β3−/− (n =7), and FcγRIIATGN/Tph1−/− (n = 3) mice, as well as in FcγRIIATGN mice pretreatedwith GP1b Fab antibody (n = 4). (E) Mouse platelets from FcγRIIAnull andFcγRIIATGN mice were labeled in vivo. Lungs were collected 10 min after the ICtrigger, and the number of fluorescent platelets was estimated (n = 4) using astandard curve designed with known numbers of fluorescent platelets spikedinto nonfluorescent control lung homogenates. Percentages were obtained bycomparison with platelet count obtained before the experiment and arepresented as the percentage of total platelet number in the whole-mousebody. (F) ICs were labeled with DyLight-647 anti-Human IgG (fluo ICs)before injection in mice. The Kid, Spl, Liv, He, Lu, and Br were harvested 10minlater, and fluorescence was measured in each organ using an IVIS to determineIC localization (n = 3). Results are presented as the percentage of fluo de-termined in FcγRIIAnull mice. (G) Two-photon intravital microscopy in brainmicrovasculature of FcγRIIATGN/CD41-YFP mice injected with ICs. Thrombi wereobserved 5 min after IC injection (arrows), whereas brain vasculature leakageoccurred after 10 min. (Scale bar: 50 μm.) null, FcyRIIAnull; TGN; FcyRIIATGN.Data are mean ± SEM. *P < 0.05, **P < 0.005, ***P < 0.001, and ****P <0.0001 using one-way ANOVA (A and D) and an unpaired t test (C, E, and F).

E1554 | www.pnas.org/cgi/doi/10.1073/pnas.1720553115 Cloutier et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

1

FcγRIIA expression (Fig. 5H), as in the passive model of the ICtrigger, and was also transient, as platelets with significantly re-duced PF4 and serotonin content were identified in blood whenthey reappeared (Fig. 5I).We further verified whether the same conclusions hold in vi-

remia and in a well-described model of active systemic anaphylaxis(23, 28, 29). The i.v. injection of herpes simplex virus-1 (HSV-1),capable of blood dissemination in immunized humans (45, 46), inHSV-1–immunized mice induced shock reminiscent of the re-sponses seen in IC-injected mice and dependent on FcγRIIA(Fig. S7A). Moreover, BSA injection in BSA-immunized miceinitiated systemic anaphylactic responses, and a proportion ofthe FcγRIIATGN mice, but never the FcγRIIAnull mice, died.Modest shock response and moderate thrombocytopenia werealso observed in FcγRIIAnull mice (Fig. S7B), possibly due to thehigh antigenicity of BSA and the potent immunization protocolimplicating adjuvant, consistent with other studies implicatingneutrophils and FcγRIIA-independent responses in these experi-

mental conditions (23, 28, 29). Of importance is that in all models,platelet FcγRIIA and serotonin were involved in shock, whileFcγRIIA was implicated in thrombocytopenia (Fig. S7 C–E).Thus, as in our passive model of IC-mediated immune re-

action, active immunization with gram-negative PAMPs, virus, orprotein antigen dominantly implicates platelet FcγRIIA and se-rotonin, and presents with transient platelet sequestration. Themechanisms unveiled in this study are illustrated in Fig. 6.

DiscussionIn this study, we examined platelet activity in IC-inducedsystemic inflammation in the absence of vascular insult. Weconfirmed a central role for platelet activation and identified se-rotonin as a critical platelet component mediating mechanismsof shock.Albeit artificial, models utilizing ferric chloride, laser-induced

injury, or injection of cytokines or PAMP in tissue/organs (e.g.,cremaster muscle, liver, lungs) (4–6, 9, 47) have provided crucialinsights to key pathways implicated in inflammation. Herein, weused HA-IgG as a surrogate model of ICs and revealed criticalcomponents in the in vivo response to systemic ICs, which werefurther confirmed using active immunization with endotoxin,virus, or a commonly used antigen in the study of anaphylaxis.Our observations were only possible using FcγRIIATGN mice,where we confirmed the functional association of FcγRIIA andαIIbβ3 in platelet activation, and suggest that targeting thesereceptors may have clinical benefits in severe conditions in-volving ICs. Conversely, other molecules (i.e., P-selectin, GPIb,ADP, thromboxane) that classically play a dominant role inplatelet activation in thrombosis, or favor platelet and neutro-phil interactions, were dispensable. The contribution of neu-trophils to the systemic response to ICs is, however, not excluded,as neutrophils have also been previously implicated, possiblythrough the release of PAF (12, 23, 28). Hence, although se-rotonin release was maintained in the absence of neutrophils,shock was dramatically reduced (Figs. S2B and S8A). We thuspropose that the release of serotonin precedes the neutrophilcontribution to shock, consistent with the reported role of se-rotonin in neutrophil activation (48). As vasodilation was alsopresent in the absence of neutrophils (Fig. S3B), these datasuggest that platelets, through serotonin, orchestrate neutrophilactivation and endothelial cell functions, and that these eventscan occur independently (Fig. 6).Vascular leakage is another feature systematically observed

when ICs were present in blood. Intriguingly, leakage and va-sodilatation were not connected, as leakage took place in-dependent of FcγRIIA but vasodilation critically requiredFcγRIIA and serotonin. As such, vessel permeability likely in-volved smaller postcapillaries vessels and more subtle changes inendothelial cell interactions, and was also insufficient to inducesignificant edema in the lungs, although platelets and neutrophilsaccumulated in great number in the lungs. In addition, leakagecritically involved IC-mediated signaling, given that it was totallyabsent in FcRγ−/− mice (Fig. S3A). While neutrophils expressother IgG receptors (other than FcγRIIA), they were dispens-able in the process (Fig. S3A). Therefore, leakage could be at-tributed to mast cells or basophils, for instance, which alsoexpress an array of receptors, such as FcγRI and FcγRIII, ca-pable of responding to ICs and can mediate permeability, po-tentially through histamine release (49).Whereas serotonin is mostly known for its role in mood, anxiety,

psychosis, or memory in the central nervous system, more than95% of total body serotonin is present in the periphery (39). Themajority of peripheral serotonin is stored in platelets, and ourobservations further revealed that platelets from female micecontain less serotonin than those from males (Fig. S8B), whichmay explain, in part, the aggravated phenotype observed in malemice compared with females following IC challenge. Moreover,

Fig. 5. Role of the FcγRIIA/serotonin axis in acute inflammation. Endogenousantibodies directed against LPS (anti-LPS) were measured by ELISA in plasma ofhealthy volunteers (HV) and patients with ongoing septic shock (SS) (A; n = 11)and in FcγRIIAnull and FcγRIIATGN mice that were immunized with LPS (LPS-immune) or diluent (nonimmune) (B; n = 4). OD, optical density. (C) Systemicshock and temperature were evaluated in nonimmune (n = 4) or LPS-immunemice (n = 6) expressing or not expressing FcγRIIA. (D) Serotonin levels in plasmawere determined in LPS-immune FcγRIIAnull, FcγRIIATGN, and FcγRIIATGN/β3−/−

mice immediately after LPS injection (n = 5) and compared with nonimmuneFcγRIIATGN mice (baseline level, indicated as a dotted line). Systemic shock andtemperature were measured after LPS injection in LPS-immune FcγRIIATGN micepreinjected with platelet-depleting (PLT Ab) or control (Ctrl Ab) antibodies(E; n = 5) and in LPS-immune FcγRIIATGN/Tph1+/+ mice and FcγRIIATGN/Tph1−/−

mice (F; n = 6). (G) Systemic shock was measured after LPS injection in LPS-immune FcγRIIATGN/β3+/+ mice and FcγRIIATGN/β3−/− mice (n = 4). (H) CD41+

platelets in whole blood were counted using flow cytometry at the indicatedtime points after injection of LPS in LPS-immune FcγRIIAnull and FcγRIIATGN mice(n = 5). (I) Platelet content in PF4 (n = 5; Left) and serotonin (n = 6; Right) weredetermined 48 h after LPS injection in LPS-immune FcγRIIAnull and FcγRIIATGN

mice and compared with nonimmune FcγRIIATGN mouse levels (dotted lines). Dil,diluent; null, FcyRIIAnull; TGN, FcyRIIATGN. Data are mean ± SEM. **P < 0.005,***P < 0.001, and ****P < 0.0001, using an unpaired t test (A and I); repeated-measures two-way ANOVA, statistical variation between groups (C and E–G);and one-way ANOVA (B, D, and H).

Cloutier et al. PNAS | Published online January 31, 2018 | E1555

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

10,

202

1

although platelets in females and males expressed similar levels ofFcγRIIA, female FcγRIIATGN mice presented significantly lowerplatelet counts than males, which could also partially explain thereduced shock in females in comparison to male mice (Fig. S8 C–E).Other immune components, such as complement C5a, are alsomore abundant in males than in females (50). Whether theserotonin reservoir in platelets explains a fundamental gender-related dichotomy in susceptibility to inflammatory responsesto ICs remains to be established.The function of serotonin in platelets is not clear; studies

suggest that it is important for the serotonylation of proteinsnecessary in platelet aggregation (51). However, SSRIs are typ-ically used by patients during the perioperative period and micelacking peripheral serotonin present only mild bleeding defects(43). Thus, the present study sheds light on a major role ofplatelet serotonin in response to a systemic stimulus, which oc-curs independent of other molecules typically implicated in theprevention of bleeding. The advantages for an organism to re-

lease bulk serotonin in response to systemic ICs are unclear. Wecan speculate that in response to a microbial invasion in animmune host, it might be preferable to reduce blood flow toprevent dissemination of the pathogen to vital organs and tofacilitate its capture by phagocytes. As serotonin also mediatesorgan regeneration (52), its liberation may be pivotal to regrowthfollowing insults caused by pathogen invasion.Thrombocytopenia, which coincides with thrombi formation in

the lungs, has been reported to occur in FcγRIIATGN mice, andrequires the expression of guanine nucleotide exchange factorCalDAG-GEF1 and 12-lipoxygenase (22, 25–27, 53). We showedhere that the formation of thrombi or occlusion of blood ves-sels was not the primary cause of shock; thrombi formed nor-mally in FcγRIIATGN/Tph1−/− mice, whereas shock was nearlytotally abrogated in these mice. Furthermore, reduction of throm-bosis by blockade of GPIb was without effect on shock and throm-bocytopenia, thereby revealing that thrombocytopenia resultedmainly from platelet sequestration, not thrombosis. Whereas platelets

Fig. 6. Platelets release pathogenic serotonin and return to blood circulation after IC-mediated degranulation and sequestration. Sequential events (numbered1–6) were observed when circulating ICs encountered platelets. Platelets are abundant in blood and in humans (not in WT mice); they express FcγRIIA, a low-affinity receptor for IgG. ICs activate FcγRIIA present on platelets (1), which changes αIIbβ3 to its active conformation (2). (3) Active αIIbβ3 binds its extracellularligand fibrinogen, which mediates outside-in signaling and granule release. In the absence of αIIbβ3, there is no granule release. It is suggested (dotted line) thatserotonin engages neutrophils (4), and it is further hypothesized (dotted line) that serotonin induces vasodilatation through its action on endothelial cells (5).(6) Multiple manifestations are observed when ICs form in blood. Shock, characterized by loss of consciousness, immobility, shallow respiration, and hypothermia,strictly implicates platelets, αIIbβ3 binding to fibrinogen, and serotonin release. In the absence of neutrophils, serotonin is released but shock is abolished. It issurmised that neutrophils produce PAF in response to serotonin, which may contribute to shock downstream of serotonin release. Mediators of shock are in-dicated in the figure. Thrombocytopenia is due, at least in part, to platelet sequestration in certain vascular beds, notably in the lung and brain microvasculature.Sequestration implicates FcγRIIA but, in contrast to shock, occurs independent of serotonin and neutrophils, and only partially implicates β3. Thrombocytopenia isonly transient, and platelets return to blood circulation with emptied granules. Microparticle release is observed in blood before return of platelets. Roles ofdifferent molecules in thrombocytopenia are indicated in the figure. Vasodilatation is implicating platelet-derived serotonin and is occurring independent of thepresence of neutrophils. Thrombosis was characterized in lungs following IC administration. It implicates FcγRIIA and β3, and, in contrast to shock, it involves GPIband not serotonin. These data support the notion that shock and thrombosis are independent events. Vascular leakage occurs independent of platelets andFcγRIIA. Molecules implicated in leakage are presented in the figure. NETosis, neutrophil extracellular traps; P-sel, P-selectin.

E1556 | www.pnas.org/cgi/doi/10.1073/pnas.1720553115 Cloutier et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

1

were observed in the lungs and brain, principally in the smallerand intertwined vessels, platelets may also be hiding in other yet-to-be-discovered vascular beds despite our careful investigations.In thrombotic thrombocytopenic purpura, profound thrombocy-topenia is explained by the failure of ADAMTS13 to performproteolysis of vWF attached to the endothelium (54), a mecha-nism distinct from what is observed in IC-induced thrombocyto-penia, as the blockade of vWF and GPIb had no effects onthrombocytopenia in our study. As in-depth whole-mouse imaginguniquely identified megakaryocytes and thrombi, which comprisemore stable and adherent platelets, it suggests that the majority ofthe sequestered platelets were dislodged by the perfusion pro-cedure. Therefore, we propose that platelets might be bridgedtogether by ICs, and that platelet–IC scaffolds may be looselytrapped in the microvasculature.Degranulated platelets recirculate, a finding of potential sig-

nificance for elucidating mechanisms underlying thrombocyto-penia. The absence of PS at the surface suggests that they arenot procoagulant platelets, known as balloon- or zombie-likeplatelets (55). How platelets return to the circulation after se-questration is unclear, but it is reasonable to speculate thatdisengagement of FcγRIIA after its desensitization by yet un-known mechanisms (e.g., unidentified immunoreceptor tyrosine-based inhibitory motif-containing receptors or phosphatases) orIC internalization by platelets (56) might release platelets fromplatelet–IC scaffolds and permit their liberation from the mi-crovasculature. FcγRIIA expression was maintained, however,on recirculating platelets, suggesting that it might be recycledif the internalization of ICs is implicated. Of interest is thatplatelets at thrombi surfaces visualized in vascular injury modelsappear loosely packed and lightly activated (57, 58), and mightalso return to the circulation. As platelets activated in vitro withthrombin can also circulate after degranulation if transfused (59,60), our study reveals that thrombotic and immunological trig-gers can induce degranulation independent of platelet elimina-tion. These models contrast with the general belief that platelets“have only one life,” and may not recirculate after undergoingactivation in vivo.The insertion of human activating (FcγRIIA/IIIA/IIIB) and

inhibitory (FcγRIIB) FcγR into the equivalent murine locusconfirmed the predominance of FcγRIIA in systemic shock in thehumanized mouse model (28), suggesting that our findings maywell be translatable to humans as platelets from transgenic miceand humans express equivalent levels of FcγRIIA (18). Whilethese approaches cannot fully recapitulate all of the subtletiesof IC-driven inflammation in humans, it is very likely that themechanisms revealed in our study may, at least in part, take placein disease states such as rheumatic disease, HIT, sepsis, viremia,anaphylaxis, and adverse reactions following i.v. IgG therapy.PAMPs trigger innate immune responses through activation of

pattern-recognition receptors, but the recognition of PAMPs byantibodies in adaptive immunity can modulate different re-sponses (61). We found that LPS, a Toll-like receptor 4 (TLR-4)trigger, dominantly implicates platelets and serotonin when in-volving the adaptive immune response in mice that had beenpreexposed to LPS. All adult volunteers we examined also dis-played antibodies directed against the endotoxin. Interestingly,humans lacking TLR signaling molecules are extremely suscep-tible to infections in infancy and childhood, and thereafter de-velop significant resistance, consistent with the prevalent role ofadaptive immunity in adults (62). Sepsis susceptibility is associ-ated with FcγRIIA polymorphism (63) and is accompanied byelevated serotonin and endothelial hyperpermeability (64),which could be attributed to platelets. Furthermore, thrombo-cytopenia measured in patients with sepsis is a strong predictorof mortality (65, 66), and our data suggest that platelets may besequestered through FcγRIIA activation in some patients. It isthus important to note that most mouse models of sepsis or viral

infection may be suboptimal, as they overlook the contribution ofFcγRIIA in a nonimmune host. Observations from trials duringwhich LPS was injected into volunteers revealed that humansresponded more promptly to small doses of LPS than mice (67),pointing to the involvement of ICs and FcγRIIA in humans, apathway absent in mice.Circulating platelets presenting reduced granule content are

reported in different contexts, such as cardiovascular diseases,type-2 diabetes mellitus, preeclampsia, autoimmune diseases,and sepsis (68–72). In patients with severe sepsis, ADP and se-rotonin contents in platelets were reduced by 40% and 50%,respectively (71). However, very little is known about the impactof in vivo platelet degranulation on platelet life span. In chronicconditions implicating ICs, such as rheumatic diseases, ICsconstantly trigger platelet FcγRIIA, consistent with the frequentshifts in platelet count. Hence, microthrombi are recognized inthese diseases, including in the lungs, and the platelet content ofserotonin is reportedly reduced in patients with SLE and RA by25% and 27%, respectively (72, 73), suggesting that plateletsmight indeed circulate in their degranulated form in variouscontexts implicating ICs (70). Neuropsychiatric SLE is the leastunderstood, yet the most prevalent, manifestation of SLE (74).Our identification of localized platelet activation in the brainmicrovasculature and leakage of the blood–brain barrier maythus have implications in the important and poorly understoodneurological manifestations in rheumatic diseases.In summary, our study reveals platelet contributions to in-

flammation in reactions involving ICs. It appears that the FcγRIIAsignaling and serotonin release are unique in regard to theirmajor role in inflammation and minor roles in the prevention ofbleeding, suggesting that interference in this process might be apromising avenue for further research.

Materials and MethodsMice. C57BL/6J (FcγRIIAnull mice), FcγRIIATGN hemizygous mice (12, 20), andFcRγ−/− mice were obtained from The Jackson Laboratory. FcγRIIATGN hemi-zygous mice described to express human FcγRIIA on platelets, megakaryo-cytes, monocytes, macrophages, neutrophils, eosinophils, basophils, mastcells, and dendritic cells (12, 20) were backcrossed to C57BL/6J more than10 times. The β3−/− mice (30), Tph1−/− mice (43), and CD41-YFP mice (75)were crossed with FcγRIIATGN mice to obtain FcγRIIATGN/β3−/− (30), FcγRIIATGN/Tph1−/−, and FcγRIIATGN/CD41-YFP mice. Chimeric mice were generated bytransfer of bonemarrow cells of FcγRIIATGNmice into FcγRIIAnull (WT), Fibγ390-396A,and FibγΔ5 irradiated mice (34, 35, 76). Guidelines of the Canadian Councilon Animal Care were followed in a protocol approved by the Animal Wel-fare Committee at Laval University (2013-106-3).

HA-IgG as an IC Model. Human IgG (IC) (MP-biological, Sigma–Aldrich andInnovative Research) were i.v. injected (600 μg and 750 μg) in males andfemales, respectively. In some experiments, human IgG was dissolved butnonaggregated (monomeric) or mouse IgG was aggregated and injected inmice (29).

Measure of Shock. Temperature was measured using a rectal probe ther-mometer at indicated time points. Signs of apparent shock were assessed asdescribed previously (24, 77) using a score from 0 to 3 (24, 77). A score of3 represents completely immobilized and unconscious mice (mice collapseand do not react to sound or touch), a score of 2 represents mice with im-paired mobility and irregular respiration, a score of 1 corresponds to micewith slow motions and shallow respiration, and a score of 0 describesnormal mice. In anaphylaxis experiments implicating BSA injection in BSA-immunized FcγRIIATGN mice, death was sometimes observed. In those ex-ceptional cases, shock was scored as 4. Scores measured in experimentalgroups were averaged and are presented as a function of time (minutes).Scores for individual mice for key experiments are provided in SI Materialsand Methods (Fig. S1 A and F).

LPS Immunization Model. Mice were immunized with i.v. injection of 1 mg/kgLPS (Escherichia coli 0111:B4; Sigma–Aldrich) at days 0, 14, 28, and 42. At thefirst and the last immunizations, shock and temperature were monitored for1 h. Mouse blood was drawn by cardiac puncture 10 min after the fourth

Cloutier et al. PNAS | Published online January 31, 2018 | E1557

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

10,

202

1

injection and used for platelet count and ELISA. Lungs were also collected asdescribed below.

Flow Cytometry. Flow cytometry was performed using a BD FACSCanto IIinstrument with forward scatter coupled to a photomultiplier tube “smallparticles option” flow cytometer (BD Biosciences). Platelets, platelets inter-acting with neutrophils, and platelet microparticles were analyzed.

Histology. In some experiments, organs were collected at the end of theexperiment. Intratracheal instillation with 1 mL of 4% paraformaldehyde inlungs was performed before collection. Brain, kidneys, spleen, heart, liver,and lungs were collected and then fixed in 4% paraformaldehyde for 24 h(lungs) or 72 h (other organs); they were then washed and stored in PBS at4 °C before histology. After fixation, paraffin-embedded organs were cutinto 5-μm sections and stained with hematoxylin and eosin. Thrombi wereobserved on five different spots at 400× resolution, they were counted usinglight microscopy (BX51; Olympus) by a blinded investigator, and the num-bers of lung thrombi per square millimeter were calculated. Neutrophilswere identified in lungs of IC-injected FcγRIIATGN mice.

Two-Photon Intravital Microscopy. For in vivo imaging of the mouse brain,FcγRIIATGN/CD41-YFP mice (8–12 wk old) were anesthetized with 1–2% iso-flurane (vol/vol) and a cranial window was made to expose the vasculatureof the sensorimotor cortex. Animals were imaged 2 wk after the surgery.Briefly, for the imaging session, the head of the mouse was restrained usinga custom-built cranial stereotaxic apparatus (David Kopf Instruments) andplaced under the microscope. For the ear imaging, mice were anesthetizedand the hair recovering the ears was gently removed using Nair, a com-mercial depilatory lotion. One ear was then gently flattened and fixed ona Plexiglas bloc using MSI-EpiDermGlu (Medisav Services). Gelseal (GEHealthcare) was applied around the tissues to form a watertight rim, and theimaging cavities were filled with sterile HBSS without Ca2+/Mg2+ (ThermoFisher Scientific). Blood vessels were stained by an i.v. injection of 0.05%Evans Blue (Sigma–Aldrich) diluted in 0.9% sterile saline. Body temperaturewas kept at 37 °C during all procedures with a heating pad. Four to sixdifferent vessels per mouse were analyzed at 1 and 8 min postinjection.

Patients with Septic Shock. Adult (age ≥ 21 y) patients with septic shock andconfirmed, gram-negative bacteremia were included. Each patient or an au-thorized family member provided written informed consent. Septic shock wasdefined as the presence of sepsis (requiring evidence of systemic infection andtwo or more of the following: temperature >38 °C or <36 °C; heartrate >90 beats per minute; respiratory rate >20 breaths per minute or partialpressure of carbon dioxide in arterial blood <32 mmHg; WBC count >12,000/mL,

<4,000/mL, or >10% bands) and the need for vasopressors to maintain asystolic blood pressure >90 mmHg or within 40 mmHg of baseline despiteadequate fluid resuscitation (5, 6). For identification of gram-negative bac-teremia, blood samples were obtained from patients upon intensive careunit (ICU) admission as part of their routine clinical care. Blood samplesunderwent gram staining and culturing in a clinical pathology laboratory.Gram-negative pathogens were identified from the gram stain and/or culturesby the clinical laboratory. Healthy, fasting adult (age ≥ 21 y) control subjectsprovided informed consent. Following informed consent, demographic data,physiological parameters, and laboratory data were recorded. Plasma was har-vested by centrifugation on whole blood collected in sterile ACD vacutainertubes. Plasma was frozen at −80 °C until used for assays. In patients with septicshock, plasma was obtained within 48 h of ICU admission.

Study Approval. Informed consent was obtained from all human subjects inthe study. The study was approved by the institutional review board atUniversity of Utah.

Statistical Analysis. Results are presented as mean ± SEM. The statisticalsignificance for comparisons between groups was determined using one-way ANOVA, two-way repeated- measures ANOVA, an unpaired Student’st test, or a Mann–Whitney U test. The D’Agostino–Pearson test was used as anormality test. All statistical analysis was done using Prism software package6 (GraphPad Software).

ACKNOWLEDGMENTS. We thank Isabelle Dubuc for expert technicalassistance with virus preparation. This work was supported by a foundationgrant from the Canadian Institutes of Health Research (CIHR) (to E.B.); it wasalso supported, in part, by NIH/National Heart, Lung, and Blood Institute Grant5F32HL118865 (to K.R.M.), Grant HL130054 (to P.J.N.), and Grant R01Hl68130(to J.E.I.); by Grant 1K01DK111515 from the National Institutes of Diabetesand Digestive and Kidney Diseases (to K.R.M.); by a grant from the CanadianFunds for Innovation (to G.P.); by Grant HL112311 (to M.T.R.); by GrantHL126547 (to M.T.R.); by National Institute on Aging Grant AG048022 (toM.T.R.), by a CIHR foundation grant (to W.I.K.); by Grant P01HL110860(to S.E.M.), by a CIHR operating grant (to S.L.); and by CIHR Grant MOP93575 (to L.F.). E.B. is the recipient of an investigator award from the CIHR.K.R.M. and J.E.I. are American Society of Hematology Fellow Scholars. G.M. hasan award from the Canadian Blood Services, N.T. and I.M. are recipients offellowships from The Arthritis Society (TAS), A.Z. is a recipient of fellowshipsfrom the TAS and CIHR, and B.M. has a fellowship from the Fondation duCentre Hospitalier Universitaire de Quebec. This material is also the result ofwork supported with resources and the use of facilities at the George E.Wahlen Veterans Affairs Medical Center (M.T.R.). The contents do not representthe views of the US Department of Veterans Affairs or the US Government.

1. Semple JW, Italiano JE, Jr, Freedman J (2011) Platelets and the immune continuum.Nat Rev Immunol 11:264–274.

2. Morrell CN, Aggrey AA, Chapman LM, Modjeski KL (2014) Emerging roles for plateletsas immune and inflammatory cells. Blood 123:2759–2767.

3. Kapur R, Zufferey A, Boilard E, Semple JW (2015) Nouvelle cuisine: Platelets servedwith inflammation. J Immunol 194:5579–5587.

4. Kolaczkowska E, Kubes P (2013) Neutrophil recruitment and function in health andinflammation. Nat Rev Immunol 13:159–175.

5. Sreeramkumar V, et al. (2014) Neutrophils scan for activated platelets to initiate in-flammation. Science 346:1234–1238.

6. Imhof BA, et al. (2016) CCN1/CYR61-mediated meticulous patrolling by Ly6Clowmonocytes fuels vascular inflammation. Proc Natl Acad Sci USA 113:E4847–E4856.

7. Goerge T, et al. (2008) Inflammation induces hemorrhage in thrombocytopenia.Blood 111:4958–4964.

8. Boulaftali Y, et al. (2013) Platelet ITAM signaling is critical for vascular integrity ininflammation. J Clin Invest 123:908–916.

9. Gros A, et al. (2015) Single platelets seal neutrophil-induced vascular breaches viaGPVI during immune-complex-mediated inflammation in mice. Blood 126:1017–1026.

10. Suurmond J, Zou YR, Kim SJ, Diamond B (2015) Therapeutics to block autoantibodyinitiation and propagation in systemic lupus erythematosus and rheumatoid arthritis.Sci Transl Med 7:280ps5.

11. Jönsson F, et al. (2011) Mouse and human neutrophils induce anaphylaxis. J Clin Invest121:1484–1496.

12. Bruhns P, Jönsson F (2015) Mouse and human FcR effector functions. Immunol Rev268:25–51.

13. Cai Z, et al. (2015) Atomic description of the immune complex involved in heparin-induced thrombocytopenia. Nat Commun 6:8277.

14. Pincetic A, et al. (2014) Type I and type II Fc receptors regulate innate and adaptiveimmunity. Nat Immunol 15:707–716.

15. Gardiner EE, Andrews RK (2014) Structure and function of platelet receptors initiatingblood clotting. Adv Exp Med Biol 844:263–275.

16. Bardina SV, et al. (2017) Enhancement of Zika virus pathogenesis by preexistingantiflavivirus immunity. Science 356:175–180.

17. Arman M, et al. (2014) Amplification of bacteria-induced platelet activation is trig-gered by FcγRIIA, integrin αIIbβ3, and platelet factor 4. Blood 123:3166–3174.

18. Boilard E, et al. (2014) Influenza virus H1N1 activates platelets through FcγRIIA sig-naling and thrombin generation. Blood 123:2854–2863.

19. Duffau P, et al. (2010) Platelet CD154 potentiates interferon-alpha secretion byplasmacytoid dendritic cells in systemic lupus erythematosus. Sci Transl Med 2:47ra63.

20. McKenzie SE, et al. (1999) The role of the human Fc receptor Fc gamma RIIA in theimmune clearance of platelets: A transgenic mouse model. J Immunol 162:4311–4318.

21. Ishii S, et al. (1998) Impaired anaphylactic responses with intact sensitivity to endo-toxin in mice lacking a platelet-activating factor receptor. J Exp Med 187:1779–1788.

22. Reilly MP, et al. (2001) Heparin-induced thrombocytopenia/thrombosis in a transgenicmouse model requires human platelet factor 4 and platelet activation throughFcγRIIA. Blood 98:2442–2447.

23. Jönsson F, et al. (2012) Human FcγRIIA induces anaphylactic and allergic reactions.Blood 119:2533–2544.

24. Meyer T, et al. (2015) CD32a antibodies induce thrombocytopenia and type II hy-persensitivity reactions in FCGR2A mice. Blood 126:2230–2238.

25. Stolla M, et al. (2011) CalDAG-GEFI deficiency protects mice in a novel model ofFcγ RIIA-mediated thrombosis and thrombocytopenia. Blood 118:1113–1120.

26. Amirkhosravi A, et al. (2014) CalDAG-GEFI deficiency protects mice from FcγRIIa-mediated thrombotic thrombocytopenia induced by CD40L and beta2GPI immunecomplexes. J Thromb Haemost 12:2113–2119.

27. Arman M, Krauel K (2015) Human platelet IgG Fc receptor FcγRIIA in immunity andthrombosis. J Thromb Haemost 13:893–908.

28. Gillis CM, et al. (2016) Mechanisms of anaphylaxis in human low-affinity IgG receptorlocus knock-in mice. J Allergy Clin Immunol 139:1253–1265.e14.

29. Bruhns P, et al. (2009) Specificity and affinity of human Fcgamma receptors and theirpolymorphic variants for human IgG subclasses. Blood 113:3716–3725.

30. Zhi H, et al. (2013) Cooperative integrin/ITAM signaling in platelets enhancesthrombus formation in vitro and in vivo. Blood 121:1858–1867.

31. Nieswandt B, Varga-Szabo D, Elvers M (2009) Integrins in platelet activation. J ThrombHaemost 7(Suppl 1):206–209.

E1558 | www.pnas.org/cgi/doi/10.1073/pnas.1720553115 Cloutier et al.

Dow

nloa

ded

by g

uest

on

July

10,

202

1

32. Ren Q, Ye S, Whiteheart SW (2008) The platelet release reaction: Just when youthought platelet secretion was simple. Curr Opin Hematol 15:537–541.

33. Li Z, Delaney MK, O’Brien KA, Du X (2010) Signaling during platelet adhesion andactivation. Arterioscler Thromb Vasc Biol 30:2341–2349.

34. Flick MJ, et al. (2007) Fibrin(ogen) exacerbates inflammatory joint disease through amechanism linked to the integrin alphaMbeta2 binding motif. J Clin Invest 117:3224–3235.

35. Flick MJ, et al. (2004) Leukocyte engagement of fibrin(ogen) via the integrin receptoralphaMbeta2/Mac-1 is critical for host inflammatory response in vivo. J Clin Invest 113:1596–1606.

36. Ehlers R, et al. (2003) Targeting platelet-leukocyte interactions: Identification of theintegrin Mac-1 binding site for the platelet counter receptor glycoprotein Ibalpha.J Exp Med 198:1077–1088.

37. Kato K, et al. (2004) Genetic deletion of mouse platelet glycoprotein Ibbeta producesa Bernard-Soulier phenotype with increased alpha-granule size. Blood 104:2339–2344.

38. Shrimpton CN, et al. (2002) Localization of the adhesion receptor glycoprotein Ib-IX-Vcomplex to lipid rafts is required for platelet adhesion and activation. J Exp Med 196:1057–1066.

39. Vanhoutte PM (2016) Regenerated endothelium and its senescent response to ag-gregating platelets. Circ J 80:783–790.

40. Adnot S, Houssaini A, Abid S, Marcos E, Amsellem V (2013) Serotonin transporter andserotonin receptors. Handb Exp Pharmacol 218:365–380.

41. Cloutier N, et al. (2012) Platelets can enhance vascular permeability. Blood 120:1334–1343.

42. Zhang X, Beaulieu JM, Sotnikova TD, Gainetdinov RR, Caron MG (2004) Tryptophanhydroxylase-2 controls brain serotonin synthesis. Science 305:217.

43. Côté F, et al. (2003) Disruption of the nonneuronal tph1 gene demonstrates theimportance of peripheral serotonin in cardiac function. Proc Natl Acad Sci USA 100:13525–13530.

44. Fuchs TA, et al. (2010) Extracellular DNA traps promote thrombosis. Proc Natl Acad SciUSA 107:15880–15885.

45. Berrington WR, et al. (2009) Clinical correlates of herpes simplex virus viremia amonghospitalized adults. Clin Infect Dis 49:1295–1301.

46. Zahariadis G, Jerome KR, Corey L (2003) Herpes simplex virus-associated sepsis in apreviously infected immunocompetent adult. Ann Intern Med 139:153–154.

47. McDonald B, et al. (2010) Intravascular danger signals guide neutrophils to sites ofsterile inflammation. Science 330:362–366.

48. Duerschmied D, et al. (2013) Platelet serotonin promotes the recruitment of neu-trophils to sites of acute inflammation in mice. Blood 121:1008–1015.

49. Binstadt BA, et al. (2006) Particularities of the vasculature can promote the organspecificity of autoimmune attack. Nat Immunol 7:284–292.

50. Strait RT, et al. (2011) MHC class I-specific antibody binding to nonhematopoietic cellsdrives complement activation to induce transfusion-related acute lung injury in mice.J Exp Med 208:2525–2544.

51. Walther DJ, et al. (2003) Serotonylation of small GTPases is a signal transductionpathway that triggers platelet alpha-granule release. Cell 115:851–862.

52. Lesurtel M, et al. (2006) Platelet-derived serotonin mediates liver regeneration.Science 312:104–107.

53. Yeung J, et al. (2014) Platelet 12-LOX is essential for FcγRIIa-mediated platelet acti-vation. Blood 124:2271–2279.

54. Dong JF, et al. (2002) ADAMTS-13 rapidly cleaves newly secreted ultralarge vonWillebrand factor multimers on the endothelial surface under flowing conditions.Blood 100:4033–4039.

55. Podoplelova NA, et al. (2016) Coagulation factors bound to procoagulant plateletsconcentrate in cap structures to promote clotting. Blood 128:1745–1755.

56. White JG, Krumwiede MD, Cocking-Johnson DJ, Escolar G (1996) Uptake of vWF-anti-vWF complexes by platelets in suspension. Arterioscler Thromb Vasc Biol 16:868–877.

57. Stalker TJ, et al. (2013) Hierarchical organization in the hemostatic response and itsrelationship to the platelet-signaling network. Blood 121:1875–1885.

58. Welsh JD, et al. (2016) A systems approach to hemostasis: 4. How hemostatic thrombilimit the loss of plasma-borne molecules from the microvasculature. Blood 127:1598–1605.

59. Michelson AD, et al. (1996) In vivo tracking of platelets: Circulating degranulatedplatelets rapidly lose surface P-selectin but continue to circulate and function. ProcNatl Acad Sci USA 93:11877–11882.

60. Berger G, Hartwell DW, Wagner DD (1998) P-Selectin and platelet clearance. Blood92:4446–4452.

61. Kawai T, Akira S (2010) The role of pattern-recognition receptors in innate immunity:Update on toll-like receptors. Nat Immunol 11:373–384.

62. Picard C, et al. (2010) Clinical features and outcome of patients with IRAK-4 andMyD88 deficiency. Medicine (Baltimore) 89:403–425.

63. Beppler J, et al. (2016) Fc gamma receptor IIA (CD32A) R131 polymorphism as amarker of genetic susceptibility to sepsis. Inflammation 39:518–525.

64. Li Y, et al. (2016) Sepsis-induced elevation in plasma serotonin facilitates endothelialhyperpermeability. Sci Rep 6:22747.

65. Claushuis TA, et al.; Molecular Diagnosis and Risk Stratification of Sepsis Consortium(2016) Thrombocytopenia is associated with a dysregulated host response in criticallyill sepsis patients. Blood 127:3062–3072.

66. Greinacher A, Selleng S (2016) How I evaluate and treat thrombocytopenia in theintensive care unit patient. Blood 128:3032–3042.

67. Copeland S, Warren HS, Lowry SF, Calvano SE, Remick D; Inflammation and the HostResponse to Injury Investigators (2005) Acute inflammatory response to endotoxin inmice and humans. Clin Diagn Lab Immunol 12:60–67.

68. Willoughby S, Holmes A, Loscalzo J (2002) Platelets and cardiovascular disease. Eur JCardiovasc Nurs 1:273–288.

69. Fateh-Moghadam S, et al. (2005) Platelet degranulation is associated with progressionof intima-media thickness of the common carotid artery in patients with diabetesmellitus type 2. Arterioscler Thromb Vasc Biol 25:1299–1303.

70. Janes SL, Goodall AH (1994) Flow cytometric detection of circulating activatedplatelets and platelet hyper-responsiveness in pre-eclampsia and pregnancy. Clin Sci(Lond) 86:731–739.

71. Protti A, et al. (2015) Platelet mitochondrial dysfunction in critically ill patients:Comparison between sepsis and cardiogenic shock. Crit Care 19:39.

72. Zeller J, Weissbarth E, Baruth B, Mielke H, Deicher H (1983) Serotonin content ofplatelets in inflammatory rheumatic diseases. Correlation with clinical activity.Arthritis Rheum 26:532–540.

73. Lood C, et al. (2015) Type I interferon-mediated skewing of the serotonin synthesis isassociated with severe disease in systemic lupus erythematosus. PLoS One 10:e0125109.

74. Hanly JG (2014) Diagnosis and management of neuropsychiatric SLE. Nat RevRheumatol 10:338–347.

75. Zhang J, et al. (2007) CD41-YFP mice allow in vivo labeling of megakaryocytic cells andreveal a subset of platelets hyperreactive to thrombin stimulation. Exp Hematol 35:490–499.

76. Flick MJ, et al. (2011) The development of inflammatory joint disease is attenuated inmice expressing the anticoagulant prothrombin mutant W215A/E217A. Blood 117:6326–6337.

77. Robles-Carrillo L, et al. (2010) Anti-CD40L immune complexes potently activateplatelets in vitro and cause thrombosis in FCGR2A transgenic mice. J Immunol 185:1577–1583.

Cloutier et al. PNAS | Published online January 31, 2018 | E1559

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

July

10,

202

1