Inhibition of thrombin-induced platelet aggregation using human single-chain Fv antibodies specific for TREM-like transcript-1

Summary TREM-like transcript-1 (TLT-1) is a novel platelet membrane receptor, which has been recently characterized in mice. TLT-1 is expressed exclusively in platelets and megakaryocytes, and its expression is dramatically upregulated upon platelet activation, suggesting that it plays a unique role in hemostasis and/or thrombosis. In this study we identified and characterized highly specific human monoclonal antibodies that bind to TLT-1 by screening a naïve library of single chain Fv fragments (scFvs) displayed on filamentous phage (Thomlinson I library). These scFvs detected plate-bound TLT-1, captured soluble TLT-1, and readily reacted with cell-bound TLT-1 on transfectants and primary human platelets. Most importantly, anti-TLT-1 scFvs inhibited thrombin-mediated human platelet aggregation. This inhibition was specific for thrombin-induced aggregation and was reversible with higher doses of agonist. These data are the first to demonstrate a biological role for TLT-1 and its potential as a therapeutic target. The human scFvs isolated in this study may represent novel anti-platelet therapeutic agents.

Ourunderstanding of TLT-1function has beeninpartlimited by the lack of suitable tools to dissect TLT-1'srole in platelet-mediated thrombosis and homeostasis. In this study we sought to assess TLT-1'spotential as atherapeutic target in the treatment of thromboticd iseases using monoclonal antibodiesd irected againstf unctional epitopes. Phage display of singlec hain Fv (scFv) offers anew waytoproduce monoclonal antibodieswith defined binding specificities (13,14). ScFv antibodies contain the variable regionso fh eavy and light chains connected by a linker peptide and represent the smallest units retaining the antigen-binding specificity of whole IgGs ( 15). Importantly, when these antibodyfragmentsare of humanorigin, adverse immune responses in humantherapycan be minimized (16,17).
In this study, we used anaïve phage-displayedlibrarytoisolate apanel of humanscFvs thatare highly specificfor TLT-1and used them to identify afunctional role for TLT-1inplateletaggregation. Specifically,w ef ound that targeting TLT-1w ith the scFvs resultedi nr educed thrombin-induced aggregation of humanplatelets. We found thatthe scFvs differentiallyaffected platelet aggregation and thatthis inhibition wasdose-dependent. The potential of these scFvs for the treatment of thrombosis and other platelet-related diseases is discussed.

Isolationand characterization of monoclonal phage antibodies specific forTLT-1
With the aimofdefining the roleofTLT-1 in platelets and functionalepitopes of this receptor,weisolated humanscFvs specific for TLT-1from anaïve libraryofphage-displayedscFvs.After depleting the libraryfor clonesthat bound to the Fc fragment of IgG,threerounds of panning against TLT-1-Fc fusion resulted in significant enrichment of phage thatb ound specificallyt o TLT-1, as assessed by ELISA, whereas no significant binding wasd etected with pooledp hage from anyo ft he rounds when TLT-1-Fc wasreplaced with the humanFcfragment alone (data not shown).Atotalof180 cloneswere randomlypickedfrom the phage pools from each round of selection,and 57 cloneswith the highestr eactivity againstT LT-1-Fca nd no reactivity against human Fc fragment alone were selected in aphage ELISA and subjected to further characterization. These clonesalso showed no reactivity to bovine ubiquitin or MAR-1, ar eceptor with homology to TREMs and TLT-1. Arepresentativephage ELISA for six selectedclonesisshown in Figure 2A. To verify thatthe clonesb ound to TLT-1i ni ts native conformation, phage were used in flow cytometric analysis of (HEK)293Tcells transfected with TLT-1. The vast majority of phage clones (96%)bound to TLT-1e xpressing cells,a sr epresented forc lone C10 (Fig. 2B, red histogram and Fig. 2C, upper panel). No binding wasd etected forthe naïvelibrary (Fig. 2B,b lack histogram)orw hen clonesw ere screened againstc ontrol cells that do not express TLT-1 ( Fig. 2C, bottom panel). The analysis of the deduced amino acidsequences of V H and V L revealed the presence of 24 differentclones(data not shown).The complementarydetermining regions(CDR)1and the framework regionsofall clones were identical (datanot shown).The CDR2 and CDR3 of V H and V L domains were the same length in allclonesand showedpartial sequencev ariability.The deduced amino acid sequencesofthe CDR2 and CDR3 of six selectedclonesare reported in Table1. The complete DNAs equences of these scFvs are availablea t GenBank (accession numbers DQ375449 to DQ375454).

Binding of the anti-TLT-1scFvs to platelets
To test the binding activity of the selected scFvs to TLT-1inthe context of the plateletsurface, flowcytometric analysis wasperformedu sing thrombin-activatedh uman platelets and purified scFvs (Fig. 4A).Expression of TLT-1onthe surfaceofthe activatedplatelets wasdetected using the anti-TLT-1pAb (Fig.4B, bottom panels). Resting platelets and an irrelevant scFv,isolated from the same library,w ere used as negative controls. Thes ix anti-TLT-1scFvs reacted to adifferent extent with activatedplatelets (Fig. 4A). Strongtomoderate reactivity with activatedplatelets wasfound with clonesC10, A8, D1 and A1, while scFv B8 and A7 only weaklyreacted with TLT-1onplatelets. No binding of the negative control scFv wasdetected to activatedplatelets ( Fig. 4A and B, rightupper panel). All scFvs demonstrated minimalbindingonresting platelets, consistent with the lowlevel of TLT-1onthe surfaceofthese cells.Arepresentative histogram of binding to resting platelets compared to activatedp latelets is shown for scFvC10 in Figure 4B (upperpanel). Basedonthese results, the scFv that showedthe highest binding activity to activatedplatelets compared to controls wasC10. Moreover,the scFv C10 performed better than the commercially availablepAb anti-TLT-1 (Fig. 4B). Therefore, we used scFv C10 to performanimmunoprecipitation of TLT-1from whole cell lysates of resting or thrombin activatedplatelets. ThescFvC10 efficiently immunoprecipitatedT LT-1 from both celll ysates, as shown in the Westernb lot analysis of the immunoprecipitates probedwith anti-hTLT-1 pAb (Fig. 4C).The twomajor bands of 35 and 25 kDadetected in the Westernblot analysis of plateletlysate were both present in scFv C10 immunoprecipitates, while no bands were detected in the control immunoprecipitate.
Takent ogether,t hese data confirmt hat we have identified scFvs highlyspecific for TLT-1and that some of these antibodies readilyreact with plateletTLT-1. Furthermore, this approach resulted in the isolation of scFvs with variousdegrees of reactivity with TLT-1asdemonstrated in Figure 4D.
The majority of TLT-1isfound within the α -granulesofresting platelets suggesting thatthe scFvs must actprimarily after initials ignaling and degranulation. Consistent with this notion, platelets stimulatedwith thrombin in the presence of scFvs exhibit rapid shape change consistent with primarysignaling (Fig.  5B,D -F). However, we felt it important to test whether scFvs might function, in part,byblocking thrombin-induced degranulation. Therefore, we stimulatedplatelets in the presence of scFv C10, stoppedthe reaction using FACS buffer,and assessed their expression of CD62P by flow cytometry. Figure 6A shows arepresentativea ggregation tracing denoting where cells were removedf or analysis of CD62P (Fig. 6A, arrow). Thesee xperiments demonstrated that thrombin-inducedC D62P expression wasnot decreasedbyscFvC10 pretreatment even though aggregation wasb locked ( Fig. 6B). Because thrombin-induced CD62Pi sc alcium-dependent, these data also demonstrate that scFv pretreatment doesnot affect proximal thrombin signaling, including calcium mobilization, and degranulation.

Discussion
Our previous work had determinedthat TLT-1isanabundantly expressed,alpha-granule specific, platelet receptor in the mouse. Thedatawepresenthere confirmsthat TLT-1isplatelet-specific  in humans as well. Despite the substantiall evelso fTLT-1 expressed in human and murine platelets and its regulation during plateletactivation, until nowthe receptor has not beenlinked to anyspecific function. Therefore, we sought to develop amethod to affect TLT-1f unction in vitro witht he goal of carrying any findings through to the clinic. The approachwechose wastouse aphage-displayedhuman antibody librarytoderive scFvs specificfor TLT-1. Arepertoireofmanydifferent scFvs can be displayed on the surface of filamentousb acteriophage, allowing phage with aspecific antigen-binding activity to be selectedby panning on the target antigen (13).This approachhas several advantagesc ompared to the traditional hybridoma technology; i) monoclonal antibodies canbeisolated faster and withoutthe needfor animal immunization (23);ii) the use of anaïve library (derivedf rom non-immunized donors)a llows the selection of antibodiesa gainst self-antigena nd weaklyi mmunogenic proteins (17,24); iii)scFvs can be efficiently and economically producedinbacteriaorinotherexpression systems (25,26).
Using these monoclonal scFvs we were able to demonstrate for the first time arole for TLT-1inplateletaggregation. Based on our aggregation traces it seemst hat TLT-1-mediated inhibition comesafter thrombin-orU46619-induced shape change and during aperiod thatwould be describedasthe onset of primaryaggregation. This finding is consistent with the inabilityof the scFv to activate plateletaggregation (datanot shown).Anti-TLT-1scFvinhibition would,therefore,not be expected to take effect until after initiala ctivation hadr eleased TLT-1f rom the α -granules. Interestingly, twoofthe identified clones(C10 and D1)showedfairlystrong recognition of TLT-1byflowcytometric analysis (Fig. 4D) butvariable levels of inhibition on platelet aggregation (Fig. 5F).This variance mayreflect the differences in binding characteristics exemplifiedb yt he flowc ytometry data.Analternativepossibilityisthat these scFvs bind different epitopes with the region bound by C10 being moreimportant for activation. To understand these distinctions scFv epitope mapping and/or direct measurement of their affinity for TLT-1would be required.
TLT-1isanITIM containing receptordemonstrated to bind SHP-1 and/or SHP-2 (12). Classicalmodels of ITIM-mediated inhibition would suggest that TLT-1might inhibit an ITAM containing,phosphotyrosine-based receptor system such as the collagenreceptor GPVI, whichsignals via FcRγ .Inaddition, most models of inhibitorys ignaling requirec o-clustering of the inhibitorya nd target receptors to promote inhibitorys ignaling. ScFvs are small, monovalent molecules,m aking co-clustering unlikely. We cannot rule out, however, thatathigh enoughconcentrations,the scFv engagesenough TLT-1molecules to send an inhibition signal. Amore likelyinterpretation of our data is the scFv blocking TLT-1-ligand interactions during aggregation. Ouruse of washedplatelets here would implythat the TLT-1ligand is eitheronplatelets or at leaststored in platelets at sufficient concentrations to becomeb iologically activeu pon plateletd egranulation. This type of co-activator modelfor TLT-1function would be consistent with data showing that co-engagement of TLT-1with the Fc receptor of RBL cells enhances calcium mobilization in an ITIM-dependentfashion (12).These authorssug-gest this co-activator activity wasdue to ITIM-mediated recruitment of SHP-2 to the receptor complex.
Regardless of their biochemical mechanism of action, anti-TLT-1s cFv, liket he anti-integrin α IIb β 3 scFvs describedb y Schwartz et al. (31), have several features thatmakethem promising candidates for the development of anti-thrombotic drugs. scFvs are of completeh uman origin avoiding major immunologicalimcompatabilities. In addition, their small sizeshould result in favorable pharmacological propertiessuch as faster elimination with consequent reductions of bleeding complications and betterpenetration of thrombi. Lastly, should it be required, scFv are easytoengineertoincrease affinity or construct derivativeswhile retaining antigen specificity.
In summary,wedemonstrate the isolation, characterization, and utility of scFv monoclonal antibodiesthat recognize TLT-1. Our application of these antibodies in in vitro assays of human plateletf unction has, for the first time,d efinedap latelet-specificfunction for TLT-1inthe regulation of aggregation. Regardless of the mechanisms thatr esponsible for inhibition, these findings open the possibilityfor the use of these and/or related TLT-1specific agents for therapeutic intervention in avarietyof human disordersassociatedwith thrombosis and coagulation.