Platelets Contain Tissue Factor Pathway Inhibitor-2 Derived from Megakaryocytes and Inhibits Fibrinolysis*

Background: TFPI-2 inhibits plasma kallikrein, FXIa, and plasmin, but its concentration in normal plasma is insufficient to inhibit clotting or fibrinolysis. Results: Platelets contain TFPI-2 derived from megakaryocytes and binds to platelet FV/Va and circulating FV in late pregnancy when plasma TFPI-2 is ∼7 nm. Conclusion: Platelet-derived TFPI-2 regulates intrinsic coagulation and tPA-induced fibrinolysis. Significance: Platelet-derived TFPI-2 promotes clot stabilization while attenuating intrinsic clotting. Tissue factor pathway inhibitor-2 (TFPI-2) is a homologue of TFPI-1 and contains three Kunitz-type domains and a basic C terminus region. The N-terminal domain of TFPI-2 is the only inhibitory domain, and it inhibits plasma kallikrein, factor XIa, and plasmin. However, plasma TFPI-2 levels are negligible (≤20 pm) in the context of influencing clotting or fibrinolysis. Here, we report that platelets contain significant amounts of TFPI-2 derived from megakaryocytes. We employed RT-PCR, Western blotting, immunohistochemistry, and confocal microscopy to determine that platelets, MEG-01 megakaryoblastic cells, and bone marrow megakaryocytes contain TFPI-2. ELISA data reveal that TFPI-2 binds factor V (FV) and partially B-domain-deleted FV (FV-1033) with Kd ∼9 nm and binds FVa with Kd ∼100 nm. Steady state analysis of surface plasmon resonance data reveal that TFPI-2 and TFPI-1 bind FV-1033 with Kd ∼36–48 nm and bind FVa with Kd ∼252–456 nm. Further, TFPI-1 (but not TFPI-1161) competes with TFPI-2 in binding to FV. These data indicate that the C-terminal basic region of TFPI-2 is similar to that of TFPI-1 and plays a role in binding to the FV B-domain acidic region. Using pull-down assays and Western blots, we show that TFPI-2 is associated with platelet FV/FVa. TFPI-2 (∼7 nm) in plasma of women at the onset of labor is also, in part, associated with FV. Importantly, TFPI-2 in platelets and in plasma of pregnant women inhibits FXIa and tissue-type plasminogen activator-induced clot fibrinolysis. In conclusion, TFPI-2 in platelets from normal or pregnant subjects and in plasma from pregnant women binds FV/Va and regulates intrinsic coagulation and fibrinolysis.

In addition to regulating ECM turnover, TFPI-2 could potentially play an important role in the inhibition of pKLK and FXIa as well as tPA-induced clot fibrinolysis. Human plasma TFPI-2 levels are reported to be negligible (6,(31)(32)(33)(34)(35)(36); however, a mean value of 30 pM (range from 3 to 300 pM) in one study (37) and an average value of 0.5 nM (range not given) in an another study (38) have also been reported. Thus, plasma levels of TFPI-2 are below the limit of detection (Յ20 pM or 0.5 ng/ml) in the majority of the reports and are also observed to be low in the TFPI-2-transfected apoE-knock-out mice (39). These observations suggest that the plasma concentration of TFPI-2 is too low to impact hemostasis or fibrinolysis. Nevertheless, during hemostasis, platelets play a significant role in producing a localized platelet/fibrin plug to stop the blood loss; this is followed by containment, stabilization, and dissolution of the clot (40,41). Accordingly, if the platelets contain TFPI-2, its release at the injury site could supplement the known clotting (42,43) and fibrinolytic inhibitors (44) and provide support in limiting the clot size and its stabilization. To date, neither the expression of TFPI-2 by megakaryocytes nor its presence in platelets has been evaluated.
Interestingly, plasma TFPI-2 levels rise steadily during normal pregnancy and are highest at the onset of labor (31)(32)(33)(34)(35)(36). The plasma TFPI-2 levels in the third trimester of pregnancy range from the early reports of 1-3 nM (31)(32)(33)(34)(35) to up to 10 nM in recent reports (36,38). In one study, the mean plasma TFPI-2 level in pregnancy is reported to be 320 pM (31). However, this represents an average value from the beginning of pregnancy to an unspecified gestation period; thus, this value does not reflect the true level of TFPI-2 at the onset of labor. TFPI-2 is abundantly made by syncytiotrophoblasts during pregnancy (45); its role could be to supplement inhibition of fibrinolysis to control bleeding. Noticeably, TFPI-1 is associated with FV in normal plasma (46,47) and binds to purified FV (46) and FVa (47). Currently, association of TFPI-2 with purified FV or FV in plasma of pregnant women is not known; further, contribution of plasma TFPI-2 toward the inhibition of intrinsic coagulation and/or fibrinolysis has not been determined.
Human FV is a 330-kDa single-chain glycoprotein organized into domain structure A1-A2-B-A3-C1-C2 (42,48,49). The second and third A domains are separated by a large heavily glycosylated B-domain, consisting of 836 amino acids located between residues 709 and 1546 of FV. Prior to participation in the coagulation cascade, FV is activated to FVa by proteolysis at Arg-709, Arg-1018, and Arg-1545 with concomitant release of the B-domain in two fragments (amino acids 710 -1018 and 1019 -1545). Although FXa can activate FV, thrombin is the most robust activator of FV (48). Compared with thrombin, FXa preferentially cleaves at Arg-1018 and Arg-709, yielding an intermediate that has partial FVa activity (50); in this case, only residues 710 -1018 of the B-domain are released. Subsequently, FXa slowly cleaves at Arg-1545 to yield the fully active FVa molecule (50,51). Recent evidence indicates that the release of the B-domain from FV is not a prerequisite for development of FVa activity (52,53). The B-domain contains two highly conserved sequences, termed the basic region (BR, residues 963-1008) and the acidic region (AR, residues 1493-1537). The BR and the AR interact with each other and maintain FV in its inactive state (52,53). Removal of either the BR or the AR results in a constitutively active FVa molecule, and removal of both regions yields the fully active FVa molecule (52,53). Further, similar to the BR of FV, the C-terminal regions of TFPI-1 and TFPI-2 contain several Arg and Lys residues (1)(2)(3). This basic C-terminal region of TFPI-1 plays an important role in binding to FV/FVa (47, 54 -56). However, the potential role of the TFPI-2 C-terminal basic region in binding to FV/FVa is not known.
In this report, we used RT-PCR, Western blotting, immunohistochemistry, and confocal microscopy to assess the expression of TFPI-2 by a megakaryoblastic cell line (MEG-01), platelets, and bone marrow megakaryocytes. In further studies, we examined binding of TFPI-2 to plasma-derived FV (pdFV), partially B-domain-deleted FV (FV-1033) 5 (57), and recombinant FVa (rFVa); these data were then compared with the data obtained using TFPI-1. We also examined whether TFPI-2 binds to FV in the plasma of pregnant women and whether or not platelets contain TFPI-2 that is associated with FV/Va. Finally, the anticoagulant and antifibrinolytic role of TFPI-2 in platelets and in plasma from women at the onset of labor was assessed.
Protease Inhibition Assay-FXIa (1 nM), pKLK (1 nM), or plasmin (3 nM) was incubated with various concentrations of TFPI-2 or KD1-WT for 1 h at room temperature in a 96-well microtitration plate (total volume, 100 l) as outlined (11,12). The buffer used was 50 mM Tris-HCl, 100 mM NaCl, pH 7.4 (TBS, pH 7.4) containing 0.1 mg/ml BSA (TBS/BSA) and 2 mM Ca 2ϩ . Briefly, 5 l of synthetic substrate, S-2251 for plasmin and S-2366 for FXIa and pKLK, was then added (final concentration, 1 K m ), and residual amidolytic activity was measured in a Molecular Devices V max kinetic microplate reader. The inhibition constants (K i *) were determined using the non-linear regression data analysis program Grafit. Data were analyzed with an equation for a tightly binding inhibitor (Equation 1), where v i and v 0 are the inhibited and uninhibited rates, respectively, and [I] 0 and [E] 0 are the total concentrations of inhibitor and enzyme, respectively (61,62).
K i values were obtained by correcting for the effect of synthetic substrate concentration, as outlined by Beith (61), using Equation 2, where [S] is substrate concentration, and K m is specific for each enzyme.
Blood Collection and Platelet Preparation-Protocols for human studies were approved by the UCLA institutional review board, and informed consent was obtained from all participants. Platelets were isolated as described (63). Fifty ml of blood was collected from normal subjects into acid citrate dextrose anticoagulant (12 mM trisodium citrate⅐2H 2 O, 10 mM citric acid monohydrate, and 15 mM dextrose) and centrifuged at 200 ϫ g at room temperature for 20 min. Apyrase (final concentration, 0.25 unit/ml) was added to platelet-rich plasma (PRP) and incubated for 10 min at 37°C prior to centrifugation at 800 ϫ g for 20 min at room temperature. Platelet-poor plasma (PPP) was centrifuged again at 800 ϫ g to remove traces of platelets. The platelet pellet was resuspended in 10 ml of 15 mM HEPES-buffered Tyrode's solution (0.13 M NaCl, 5 mM KCl, and 6 mM dextrose, pH 6.5) containing 2 mg/ml BSA and 0.25 unit/ml apyrase; the suspension was incubated for 10 min at 37°C and centrifuged at 800 ϫ g for 10 min. The platelets were resuspended in 10 ml of the above Tyrode's solution, washed, and suspended in 3 ml of the above buffer without apyrase and counted on a Beckman Coulter counter. Platelets were activated with 10 M TRAP for 10 min at 37°C, at which point benzamidine was added to yield 1 mM concentration. The suspension was centrifuged at 12,000 ϫ g for 30 min for Western blot and plasmin ligand blot analysis.
Ten ml of citrated blood was obtained from each healthy pregnant woman (5 subjects in total) at the onset of labor. FV (ϳ81-123%) and platelet counts (2.3-3.1 ϫ 10 5 /l) were in the normal range. Fibrinogen levels were 421 Ϯ 92 mg/dl in late pregnancy plasma, which represents an ϳ200% increase as compared with normal healthy subjects. PPP and platelets were isolated as above for use in plasmin ligand blot and plasma clot fibrinolysis assays. TFPI-2 in normal PPP, in pregnant woman PPP, and in adult platelets was measured by the procedure outlined by Iino et al. (6).
Flow Cytometry-To examine whether platelets in the cord blood also contain TFPI-2, the blood was subjected to Ficoll-Paque gradient centrifugation, and the red blood cell-depleted fraction was used to sort for the CD41a-positive fraction containing platelets and to remove leukocytes (CD45) and the contaminating red blood cells (CD235a). After incubation with the appropriate antibodies, cells were washed with phosphate-buffered saline (PBS) containing 1% BSA and analyzed using a BD FACSAria or LSRII cytometer (BD Biosciences). FACS files were exported and analyzed using FACSDiva software (BD Biosciences). The isolated platelets were smeared on a glass slide, fixed with cold methanol, air-dried, and blocked with 2% normal horse serum before incubating with the rabbit anti-TFPI-2 antibody (16 g/ml in TBS/BSA) overnight at 4°C. The bound rabbit anti-TFPI-2 was detected using Alexa Fluor 488 (green) anti-rabbit IgG at a 1:100 dilution in TBS/BSA.
Expression of CD41 and TFPI-2 in MEG-01 Cells Using RT-PCR and Western Blotting-MEG-01 human megakaryocyte leukemia cells obtained from American Type Culture Collection were maintained in RPMI 1640 medium with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin at 37°C in 5% CO 2 . For induced expression of CD41 and TFPI-2, the MEG-01 cells were cultured in the presence of 10 ng/ml PMA for 8 days (64,65). Total RNA was prepared from ϳ10 6 cells using the RNeasy kit (Qiagen) with DNase I treatment as per the manufacturer's protocol. One g of RNA in 20 l was reverse transcribed into cDNA using random primers and the SuperScript III first strand synthesis kit (Invitrogen) containing Moloney murine leukemia virus reverse transcriptase. The cDNA mixture was diluted 1:10, and 1 l was used for amplification of a 440-bp region of human TFPI-2 using the following primers: forward, 5Ј-GTCGATTCTGCTGCTTTTCC-3Ј; reverse, 5Ј-ATGGAATTTTCTTTGGTGCG-3Ј (7). The CD41 expression was analyzed using the following primers: forward, 5Ј-AGTGCCCCTCGCTGCTCTTTGACC-3Ј; reverse, 5Ј-AGT-TTTCGGTCTGCCCGGCTCTC-3Ј (64). Glyceraldehyde-3phosphate dehydrogenase (GAPDH) served as a housekeeping gene, using the following primers: forward, 5Ј-ATCCCATCA-CCATCTTCCAG-3Ј; reverse, 5Ј-CCATCACGCCACAGTT-TCC-3Ј. For TFPI-2 and GAPDH, PCR products were amplified (30 cycles) using 60°C annealing temperature and 72°C elongation temperature each for 1 min. For CD41, PCR products were amplified using 58°C annealing temperature and 72°C elongation temperature, each for 30 s.
For Western blots (6,58), the PMA-stimulated and unstimulated MEG-01 cells (2 ϫ 10 7 ) were lysed in PBS containing 50 mM octyl-␤-D-glucopyranoside, 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml aprotinin, 1 mM sodium orthovanadate, 10 mM EDTA, and 1% Triton X-100 as described (7). After removing the cell debris, protein in each cell lysate was determined using a Bradford protein assay kit; 10 g was used for 12% SDS-PAGE under nonreducing conditions. The ECM proteins from the cell-free culture plate (150 mm) were eluted with 2 ml of 1% SDS in PBS; 15 l was used for 12% SDS-PAGE under nonreducing conditions. Proteins were transferred to Bio-Rad PVDF membrane, and Western blot analyses were carried out as detailed for binding of FV to TFPI-2.
Immunofluorescence Staining of MEG-01 Cells-One hundred l of washed MEG-01 cells (10 5 cells) in cold 1% BSA plus PBS were used to prepare the smears, using 500 rpm for 5 min (Shandon Cytospin 3). The smears were air-dried, permeabilized with ice-cold methanol, and blocked with 2% horse serum for 20 min prior to incubating with rabbit anti-TFPI-2 (16 g/ml) and mouse anti-CD41 antibody (10 g/ml) for 2 h at room temperature. After three washes with TBS, 0.05% Tween 20, the slides were incubated with Alexa Fluor 488 (green) antirabbit IgG (1:100 dilution), and Alexa Fluor 594 (red) antimouse IgG (1:100 dilution) for 30 min in the dark. The slides were washed with TBS plus 0.05% Tween 20, rinsed with deionized water, mounted, and coverslipped using Fluormount-G. Confocal images were obtained using an LSM 700 spinning disk confocal Carl Zeiss microscope with ϫ63 oil immersion objectives.
Detection of TFPI-2 in Bone Marrow-derived Megakaryocytes-Formalin-fixed paraffin-embedded sections of human fetal bone marrow tissue were used for these studies. The antigen was retrieved by immersing the tissue sections at 60°C for 30 min in citrate buffer, pH 6.0. The endogenous peroxidase activity was blocked by incubating the sections with 3% H 2 O 2 for 10 min. Slides were then incubated with 2% normal horse serum to prevent nonspecific binding. After incubation with rabbit anti-TFPI-2 IgG (16 g/ml) overnight at 4°C, the sections were washed (3 times) for 3 min each with PBS plus 0.05% Tween 20 and rinsed with PBS. The sections were then incubated with secondary antibody (goat anti-rabbit IgG-HRP, 1:500 dilution) for 30 min. 3,3Ј-diaminobenzidine tetrahydrochloride-H 2 O 2 substrate was added, and the slides were incubated at room temperature for 3 min. Specimens were counterstained lightly with hematoxylin. Slides were dehydrated and FIGURE 1. Purification and inhibitory constants (K i ) of TFPI-2 and Western blot analysis of TFPI-1. A, SDS-PAGE analysis of TFPI-2 and KD1-WT. KD1-WT and TFPI-2 were expressed, purified, and refolded as described (11,58). Lane 1, molecular weight markers; lane 2, TFPI-2 (M r ϳ26,000); lane 3, KD1-WT (M r ϳ8000). SDS-PAGE was performed using the Laemmli buffer system (68). Acrylamide concentration was 12%, and ␤-mercaptoethanol was 5% (v/v). B, equilibrium inhibition constants (K i ) of TFPI-2 with plasmin, FXIa, and pKLK. The enzyme activity (mean of three experiments) is expressed as the percentage of fractional activity (inhibited/uninhibited) at increasing TFPI-2 concentrations: percentage of activity of 3 nM plasmin (F), percentage of activity of 1 nM FXIa (E), and percentage of activity of 1 nM pKLK (OE). The inhibition constants (K i ) were determined as outlined under "Experimental Procedures." C, Western blot analysis of TFPI-1 and TFPI-1 161 . Reduced samples (10 ng each) were electrophoresed on 12% SDS-PAGE and transferred to nitrocellulose membrane. The primary antibody used was mouse N-terminal TFPI-1 peptide antibody (left) or the rabbit C-terminal TFPI-1 peptide antibody (right). The secondary antibody used was HRP-conjugated rabbit anti-mouse IgG (left) or donkey anti-rabbit IgG (right). The protocol used was as outlined by the manufacturer for the mouse monoclonal antibody, and by Broze and co-workers (47,59) for the rabbit polyclonal antibody. The blots were developed with the SuperSignal Femto chemiluminescent substrate for 5 min and scanned using Alpha Innotech FluorChem FC2 digital imaging system. In both left and right panels, lane 1, full-length TFPI-1; lane 2, TFPI-1 161 that lacks the third Kunitz domain and the C-terminal basic segment.
TFPI-2 Binding to pdFV, FV-1033, and rFVa-The binding of TFPI-2 or KD1-WT to pdFV, FV-1033, and rFVa was studied using a solid phase protein-binding ELISA technique (66). The Immulon 4HBX microtiter 96-well plates were coated overnight at 4°C with 100 l of TFPI-2 or KD1-WT at 10 g/ml in 50 mM NaHCO 3, pH 9.5. The control (background) wells were coated with BSA. The unbound protein was removed by washing the wells three times for 3 min with a washing buffer (25 mM HEPES, 100 mM NaCl, 2 mM CaCl 2 , pH 7.4, containing 0.1% Tween 20) and blocked for 2 h at room temperature with 300 l of blocking buffer (25 mM HEPES, 100 mM NaCl, 2 mM Ca 2ϩ , pH 7.4, containing 5% nonfat milk). The wells were washed three times with the washing buffer. The pdFV, FV-1033, or rFVa was added to the wells with different concentrations in blocking buffer (100 l) for 4 h at room temperature. The wells were rinsed (3 times for 5 min each) with the washing buffer and incubated with sheep anti-FV (primary antibody) at 20 g/ml for 2 h. After washing (3 times for 5 min each), the wells were incubated with donkey anti-sheep IgG-HRP (secondary antibody) at 1:2000 dilution for 1 h. After washing, the bound pdFV or FV-1033 or rFVa was detected using SIGMAFAST TM o-phenylenediamine dihydrochloride as substrate for 30 min. The results presented are the average of three determinations.
TFPI-2 and TFPI-1 Binding to FV-1033 and rFVa Using Surface Plasmon Resonance (SPR)-Binding of FV-1033 and rFVa to TFPI-2 and TFPI-1 was also studied using a Biacore T100 flow biosensor (Biacore, Uppsala, Sweden). TFPI-2 or TFPI-1 was immobilized on a carboxymethyl-dextran flow cell (CM5 sensor chips, GE Healthcare) using amine-coupling chemistry. Flow cell surfaces were activated with a mixture of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-hydroxysulfosuccinimide for 5 min at 10 l/min, after which TFPI-2 or TFPI-1 in 10 mM sodium acetate, pH 5.5, was injected across the flow cell. Unreacted sites were blocked with 1 M ethanolamine. Each analyte (FV-1033 and rFVa) was perfused (10 l/min) through flow cells in HBS-P buffer (20 mM HEPES, 100 mM NaCl, 2 mM Ca 2ϩ , pH 7.4, 0.01% P20, 0.1% PEG 8000). Fourmin association and 10-min dissociation times were used. Flow cells were regenerated with HBS-P containing 1 M NaCl followed by equilibration with HBS-P buffer. Data were corrected for nonspecific binding by subtracting the signal obtained with the analyte infused through a flow cell without the coupled protein. Dissociation constants (K d ) were obtained using the equilibrium response (RU eq ) for each concentration of FV-1033 or rFVa. Data were fitted to the steady-state 1:1 interaction FIGURE 2. Flow cytometry and fluorescence microscopy to detect TFPI-2 in cord blood platelets. A, dot plot analysis of red blood cell-depleted cord blood cells using the fluorescent labeled mouse isotype IgG. B, gating of red blood cell-depleted cord blood cells stained for allophycocyanin anti-CD45.
C, dot plot analysis of the CD45 negative cells stained for FITC anti-CD235A (glycophorin A) and for the R-phycoerythrin anti-CD41a. D, detection of TFPI-2 in platelets by fluorescence microscopy. The anti-CD41a-positive cells sorted using flow cytometry were smeared on a glass slide and stained with rabbit anti-TFPI-2 IgG. The rabbit anti-TFPI-2 IgG was detected using Alexa Fluor 488 (green) goat anti-rabbit IgG as described under "Experimental Procedures." Origin and Biologic Significance of Platelet TFPI-2 NOVEMBER 7, 2014 • VOLUME 289 • NUMBER 45 equation, RU eq ϭ (RU max ϫ C)/(K d ϩ C), where C represents the concentration of FV-1033 or rFVa, and RU max is the maximum binding capacity. The dissociation constants (K d ) were obtained using the non-linear regression data analysis program Grafit.
TFPI-2 Binding to pdFV, FV-1033, and rFVa using Pull-down Assay and Western Blots-One hundred l of pdFV, FV-1033, or rFVa (each at 30 nM) was incubated with 100 l of purified TFPI-2 (150 nM) in TBS/BSA, pH 7.4, 2 mM calcium, 1 mM benzamidine for 2 h at room temperature. The complex was then incubated with either 100 l of AHV-5101-Seph resin or with anti-TFPI-2 SK9 antibody resin overnight at 4°C in the same buffer. The samples were centrifuged, and the resin was washed four times with 0.5 ml of the incubation buffer. The bound complexes were eluted with 40 l of SDS sample buffer, 6 M urea (50°C for 30 min) and run on 10% SDS-PAGE under non-reducing conditions. The proteins were transferred to Bio-Rad PVDF membrane and probed overnight with rabbit anti-TFPI-2 antibody (6, 58) at 16 g/ml in TBS plus 1% nonfat milk. The secondary antibody used was goat anti-rabbit IgG-HRP at 1:15,000 dilution in TBS plus 1% milk for 2 h. The blots were developed with the SuperSignal Femto chemiluminescent substrate for 5 min and scanned using Alpha Innotech FluorChem FC2 digital imaging system. The blots were stripped with Thermo Scientific Restore Western blot stripping buffer and then probed with sheep anti-FV antibody (20 g/ml in TBS plus 1% milk, overnight). The secondary antibody used was donkey anti-sheep IgG-HRP (1:2000 dilution in TBS plus 1% milk for 2 h), and the blots were developed and scanned similarly.
Association of TFPI-2 with FV in Plasma of Pregnant Women and in Normal Platelets-One ml of plasma from each pregnant woman and their pooled plasma as well as 1 ml of normal pooled plasma were incubated (in the absence of Ca 2ϩ ) with either anti-FV AHV-5101-Seph resin or with anti-TFPI-2 SK9 antibody resin as described above. Similarly, the TRAP-activated platelet supernatants from normal individuals and pregnant women were incubated with anti-FV AHV-5101-Seph resin or with anti-TFPI-2 SK9 antibody resin and processed for Western blots as outlined above.
Plasma Clot Fibrinolysis-With slight modification, the procedure used was that described previously (11,12,67). FXIa or rTF was used to initiate intrinsic or extrinsic clotting and fibrin formation in plasma. Two hundred twenty-five l of PPP, PRP containing 3 ϫ 10 8 platelets/ml, or PPP from pregnant women at the onset of labor were mixed with 12.5 l of PL vesicles and 12.5 l of buffer or rabbit anti-TFPI-2 IgG. In experiments where PPP was used, 240 l of the above mixture was added to The arrows indicate differently glycosylated TFPI-2 species. C, confocal microscopy of unstimulated MEG-01 cells depicting the expression of TFPI-2 (green) and CD41 (red). D, confocal microscopy of PMA-stimulated MEG-01 cells depicting increased expression of TFPI-2 (green) and CD41 (red). The merged images in C and D reveal that CD41 is membrane-associated, and TFPI-2 is intracellular. CaCl 2 . In the 250-l final volume, concentration of FXIa was 400 pM (or 10 pM rTF), PL was 100 M, and tPA was 0.5 g/ml.
In the second set of experiments where PRP was used, 240 l of the mixture was added to 10 l of FXIa (or rTF) and tPA in TBS/BSA containing 250 M TRAP and 250 mM CaCl 2 . In this case, the 250-l final volume contained 400 pM FXIa (or 10 pM rTF), 100 M PL, 0.5 g/ml tPA, and 10 M TRAP. The effect of TFPI-2 present in platelets or in pregnant woman plasma was evaluated by including rabbit anti-TFPI-2 IgG at 300 g/ml in the final 250-l mixture. In control experiments, non-immune rabbit IgG or anti-TFPI-2 IgG at 300 g/ml was added to PPP (that lacks TFPI-2) in order to rule out their contributions to the plasma clot lysis. In additional experiments, PPP was supplemented with 7 nM TFPI-2 (comparable with that in pregnant woman PPP) to assess its effect on clotting and/or fibrinolysis. Clot formation and lysis were monitored at 37°C with a microplate reader (SPECTRAmax 190, Molecular Devices) by measuring the optical density at 405 nm (A 405 ) as described (11,12).
Plasmin Ligand Blot to Assess TFPI-2 in Platelets-Thirty l of TRAP-activated platelet suspension (3 ϫ 10 8 platelets) was centrifuged, and supernatant and the pellet were subjected to 12% SDS-PAGE under nonreducing conditions. Western blot analysis was carried out as above except that plasmin (20 g/ml) was used as the ligand. The bound plasmin was detected by the rabbit anti-plasmin antibody (1:100 dilution) followed by incubation with goat anti-rabbit IgG-HRP at 1:15,000 dilution.
Flow Cytometry and Fluorescence Microscopy to Detect TFPI-2 in Cord Blood Platelets-Next, we examined whether platelets in the cord blood also contain TFPI-2. We used flow cytometry to obtain platelets free of other blood cells. The platelets were stained with rabbit anti-TFPI-2 antibody followed by Alexa Fluor 488 anti-rabbit IgG. The data presented in Fig. 2 clearly reveal that the TFPI-2 antigen is present in the cord blood platelets.
Expression of TFPI-2 and CD41 by MEG-01 Cells and Bone Megakaryocytes-Here, we examined the expression of TFPI-2 and CD41 by the megakaryoblastic MEG-01 cells using RT-PCR (Fig. 3A), Western blots (Fig. 3B) and immunofluorescence (Fig. 3, C and D). The unstimulated MEG-01 cells expressed low levels of TFPI-2 and CD41, whereas PMAstimulated MEG-01 cells expressed increased levels of TFPI-2 and CD41. In addition to the cell lysates, the ECM of

Origin and Biologic Significance of Platelet TFPI-2
NOVEMBER 7, 2014 • VOLUME 289 • NUMBER 45 PMA-stimulated MEG-01 cells also contained significant amounts of TFPI-2 (Fig. 3B). The presence of TFPI-2 in the ECM of MEG-01 cells is consistent with the earlier data obtained using ECM from different cell types (6,23,27).
In subsequent experiments, we assessed TFPI-2 expression in megakaryocytes using fetal bone marrow tissue. The immunostaining data using non-immune rabbit IgG (Fig. 4A) and the rabbit anti-TFPI-2 IgG revealed that the fetal bone megakaryocytes express TFPI-2 (Fig. 4B). Cumulatively, the antigen levels in adult platelets and the data presented in Figs. 2-4 support a conclusion that megakaryocytes are the source of platelet TFPI-2.
Binding of TFPI-2 and TFPI-1 to pdFV, FV-1033, and rFVa Using ELISA-Plasma concentration of FV is ϳ20 nM, and platelets contain ϳ20% of the total FV present in whole blood (48). Previous work on the association of FV with TFPI-1 raises the possibility that TFPI-2 in platelets and in the plasma of pregnant women might also be associated with FV. We examined the binding of TFPI-2 and KD1-WT to FV and FVa using a solid phase protein-protein interaction assay presented in Fig.  5, A and B. TFPI-2 bound to pdFV, FV-1033 and rFVa with K d values of 7 Ϯ 2, 9 Ϯ 3, and 95 Ϯ 11 nM, respectively (Table 1). However, KD1-WT did not bind to pdFV or FV-1033. Moreover, prior incubation (1 h) of 10 nM pdFV or FV-1033 with 100 nM TFPI-1 abrogated Ն90% of their binding to the surfacebound TFPI-2 in the ELISA. However, incubation of pdFV or FV-1033 with TFPI-1 161 under similar conditions had no effect on their binding to TFPI-2. These data indicate that TFPI-2 and TFPI-1 compete for binding to FV. Moreover, KD1-WT of TFPI-2 or residues 1-161 of TFPI-1 do not significantly contribute to this interaction.
Binding of TFPI-2 and TFPI-1 to FV-1033 and rFVa Using SPR-TFPI-1 has been shown previously to bind to pdFV with ϳ15 nM K d using SPR (46). Here, we employed SPR to study the binding of FV-1033 and rFVa to immobilized TFPI-2 or TFPI-1. These data are presented in Fig. 5, C-F. The sensograms indicate that the interactions reached equilibrium during the injection of each analyte concentration; therefore, the dissociation constants for TFPI-2 and TFPI-1 with FV-1033 or rFVa could be determined using the relationship between the maximum binding response (RU max ) and the analyte concentration (FV-1033 or rFVa) using the steady-state 1:1 interaction equation (see "Experimental Procedures"). TFPI-2 binds to FV-1033 with K d ϭ 36 Ϯ 4 nM and to rFVa with K d ϭ 252 Ϯ 16 nM. Similarly, TFPI-1 binds to FV-1033 with K d ϭ 48 Ϯ 6 nM and to rFVa with K d ϭ
In SPR experiments, TFPI-2 or TFPI-1 was coupled to the CM5 chip. The dissociation constants were obtained using the steady state equilibrium model. 456 Ϯ 18 nM. Thus, both TFPI-2 and TFPI-1 bind to FV-1033 with higher affinity than rFVa. These data are summarized in Table 1. TFPI-2 Binding to pdFV, FV-1033, and rFVa Using Pull-down Assays and Western Blots-The purpose of these experiments was to determine whether TFPI-2 in the plasma of pregnant women and in platelets from normal healthy subjects is associated with FV. For this, we first developed an in vitro pull-down assay method followed by Western blots to study this interac-tion using purified proteins. The data presented in Fig. 6, A (using anti-FV AHV-5101-Seph resin) and B (using anti-TFPI-2 SK9 antibody resin), confirm that the procedure can be used to study the interaction of TFPI-2 with FV/Va. The data presented in Fig. 6, C and D, reveal that plasma obtained from women during the onset of labor contained TFPI-2 that, in part, was bound to FV; in contrast, TFPI-2 was not detected in normal plasma. Further, normal platelets also contained TFPI-2 associated with FV/Va (Fig. 6, C and D).

Physiologic Function of TFPI-2 in Platelets and Pregnant
Woman PPP-Biologic function of TFPI-2 in normal platelets and in the plasma of pregnant women at the onset of labor was evaluated using intrinsic or extrinsic clotting and tPA-induced fibrinolysis. Initially, the effect of varying concentrations of FXIa and tPA using normal PPP was assessed. These data are shown in Fig. 7A. At 400 pM FXIa, 0.5 g/ml tPA, clot formation started at ϳ4 min, and complete lysis occurred at ϳ26 min (curve 1); in the absence of tPA, no lysis was evident during the 2-h time course of the experiment (curve 2); at 300 pM FXIa, 0.38 g/ml tPA, the clot formation started at ϳ5 min, and complete lysis occurred at ϳ34 min (curve 3); at 200 pM FXIa, 0.25 g/ml tPA, the clot formation started at ϳ6 min, and complete lysis occurred at ϳ50 min (curve 4); and at 100 pM FXIa, 0.12 g/ml tPA, the clot formation started at ϳ7 min, and complete lysis occurred at ϳ100 min (curve 5). When PPP was replaced by PRP (3 ϫ 10 8 platelets/ml) at 400 pM FXIa, 0.5 g/ml tPA, the clot formation started at ϳ6 min, and complete lysis occurred at ϳ120 min (Fig. 7B, curve ϪAb). The addition of rabbit anti-TFPI-2 IgG to the PRP reduced the clot formation to ϳ5 min and lysis to ϳ42 min (Fig. 7B, curve ϩAb). Thus, by inference, TFPI-2 released from platelets inhibits FXIa by ϳ25% (in the first 6 min) and tPA-induced lysis by ϳ50%.
In additional experiments, when normal PPP was replaced by pregnant woman PPP at 400 pM FXIa, 0.5 g/ml tPA, the clot formation started at ϳ5 min, and complete lysis occurred at ϳ80 min (Fig. 7C, curve ϪAb). The addition of rabbit anti-TFPI-2 IgG to the pregnant woman PPP reduced the clot formation to ϳ4 min and lysis to ϳ40 min (Fig. 7C, curve ϩAb). Thus, TFPI-2 present in the plasma of pregnant women at the onset of labor inhibits FXIa by ϳ25% (during the first 5 min) and tPA-induced lysis by ϳ40%.
To validate the results of Fig. 7, B and C, we examined the effect of non-immune IgG and anti-TFPI-2 IgG on the clot fibrinolysis of normal PPP; these data presented in Fig. 7D reveal no effect, confirming that the anti-TFPI-2 IgG is TFPI-2-specific. Further, when normal PPP was supplemented with 7 nM TFPI-2, it increased the clot lysis time from ϳ25 min to ϳ40 min. Moreover, the addition of anti-TFPI-2 IgG reduced the lysis time to ϳ25 min, similar to PPP (Fig. 7D); in contrast, non-immune IgG had no effect.
Next, we assessed the effect of varying concentrations of rTF and tPA while using normal PPP (Fig. 7E). At 10 pM rTF, 0.5 g/ml tPA, the clot formation started at ϳ2 min, and complete lysis occurred at ϳ24 min (curve 1); in the absence of tPA, no lysis was evident during the 2-h time course of the experiment (curve 2); at 7.5 pM rTF, 0.38 g/ml tPA, the clot formation started at ϳ3 min, and complete lysis occurred at ϳ33 min (curve 3); at 5 pM rTF, 0.25 g/ml tPA, the clot formation started at ϳ4 min, and complete lysis occurred at ϳ55 min (curve 4); and at 2.5 pM rTF, 0.12 g/ml tPA, the clot formation started at ϳ5 min, and complete lysis occurred at Ͼ120 min (curve 5). When PPP was replaced by PRP (3 ϫ 10 8 platelets/ml) at 10 pM rTF, 0.5 g/ml tPA, the clot formation started at ϳ3 min, and complete lysis occurred at ϳ80 min (Fig. 7F, curve ϪAb). The addition of rabbit anti-TFPI-2 IgG to the PRP did not affect the clot formation time (ϳ3 min) but reduced the lysis time to ϳ40 min (Fig. 7F, curve ϩAb). Thus, TFPI-2 in PRP inhibits tPA-induced lysis by ϳ40% but does not affect extrinsic coagulation.
In further experiments, when normal PPP was replaced by PPP from pregnant women at 10 pM rTF, 0.5 g/ml tPA, the clot formation started at ϳ2 min, and complete lysis occurred at ϳ40 min (Fig. 7G, curve ϪAb). The addition of rabbit anti-TFPI-2 IgG to the pregnant woman PPP did not affect the clot formation time but reduced the lysis time to ϳ30 min (Fig. 7G, curve ϩAb). Thus, TFPI-2 present in the plasma of pregnant women at the onset of labor inhibits tPA-induced lysis by ϳ30% without affecting extrinsic coagulation. Similar to the results obtained in Fig. 7D, non-immune IgG or anti-TFPI-2 IgG had no effect on the clot fibrinolysis of normal PPP when the clotting was initiated with rTF ( Fig. 7H). In this system, when normal PPP was supplemented with 7 nM TFPI-2, it increased the lysis time from ϳ24 min to ϳ36 min. In addition, anti-TFPI-2 IgG reduced the lysis time to a value similar that of PPP (Fig.  7H), whereas non-immune IgG had no effect.
Although absolute concentration of TFPI-2 in the PRP was only ϳ40% of that present in the pregnant woman PPP, the FXIa or plasmin inhibition was more pronounced in the PRP system (Fig. 7, B versus C and F versus G). A possible explanation for this observation might be that the TFPI-2 in PRP versus the pregnant woman PPP is more effectively localized at the FIGURE 7. Effect of TFPI-2 in normal PRP and in the pregnant woman PPP on clotting and fibrinolysis. Clot formation was initiated with FXIa or rTF, and fibrinolysis was induced by the simultaneous addition of tPA. Fibrin formation and lysis were monitored by measuring the optical density at 405 nm as outlined under "Experimental Procedures." A, effect of varying FXIa and tPA on clotting and fibrinolysis using normal PPP. Curve 1, 400 pM FXIa, 0.5 g/ml tPA; curve 2, 400 pM FXIa, no tPA; curve 3, 300 pM FXIa, 0.38 g/ml tPA; curve 4, 200 pM FXIa, 0.25 g/ml tPA; curve 5, 100 pM FXIa, 0.12 g/ml tPA. B, effect of TFPI-2 in PRP/TRAP on intrinsic clotting and fibrinolysis. C, effect of TFPI-2 in pregnant woman PPP on intrinsic clotting and fibrinolysis. In B and C, FXIa was 400 pM and tPA was 0.5 g/ml, and the curve labeled ϪAb contained non-immune rabbit IgG, and the curve labeled ϩAb contained rabbit anti-TFPI-2 IgG. Control curves 1 and 2 in B and C represent repeats of curves 1 and 2 in A. Further, anti-TFPI-2 IgG or non-immune rabbit IgG had no effect on clotting or fibrinolysis (see below). D, effect of TFPI-2 added to PPP on intrinsic clotting and fibrinolysis. Control curves 1 (green) and 2 (blue) are duplicates of curves 1 and 2 in A. Additional components included in the green curve are non-immune IgG (dark green squares), anti-TFPI-2 IgG (brown squares), TFPI-2 (cyan squares), and TFPI-2 plus anti-TFPI-2 IgG (red open triangles) or plus non-immune IgG (magenta closed triangles). TFPI-2 concentration was 7 nM, and non-immune or anti-TFPI-2 IgG was 300 g/ml. E, effect of varying rTF and tPA on clotting and fibrinolysis using normal PPP. Curve 1, 10 pM rTF, 0.5 g/ml tPA; curve 2, 10 pM rTF, no tPA; curve 3, 7.5 pM rTF, 0.38 g/ml tPA; curve 4, 5 pM rTF, 0.25 g/ml tPA; curve 5, 2.5 pM rTF, 0.12 g/ml tPA. F, effect of TFPI-2 in PRP/TRAP on extrinsic clotting and fibrinolysis. G, effect of TFPI-2 in pregnant woman PPP on extrinsic clotting and fibrinolysis. In both F and G, rTF was 10 pM, and tPA was 0.5 g/ml, and the curve labeled ϪAb contained non-immune rabbit IgG, and the curve labeled ϩAb contained rabbit anti-TFPI-2 IgG. Origin and Biologic Significance of Platelet TFPI-2 NOVEMBER 7, 2014 • VOLUME 289 • NUMBER 45 platelet aggregation/clot site. Further detailed prospective experiments are needed to address this issue.
Because a major effect of the activated platelets in PRP appears to be the inhibition of tPA-induced fibrinolysis (Fig. 7, B and F), we examined whether TFPI-2, in part, stays associated with the platelet membrane or is fully released into the medium. To assess this, we activated the isolated platelets from normal subjects and from pregnant woman blood with TRAP and performed plasmin ligand blot analysis. The data for normal platelets are shown in Fig. 7F (inset). Similar results were obtained with platelets from pregnant subjects (data not shown). Clearly, a majority of the TFPI-2 is released into the medium upon activation of platelets with TRAP.

DISCUSSION
The role of TFPI-2 in regulating ECM turnover and tumor invasion has been well documented and reviewed (8,9,27,30). However, whether or not TFPI-2 plays a role in regulating blood coagulation and/or fibrinolysis remains uncertain. An important reason for this prevailing perspective is the finding (6, 31-36) (present study) that levels of TFPI-2 are undetectable (Յ20 pM) in normal plasma. Here, we show that normal platelets, which constitute an integral part of the hemostatic system, contain TFPI-2. TRAP activation of 3 ϫ 10 8 platelets (average number present per ml of blood) released ϳ85 ng of TFPI-2, most of which would be localized at the platelet aggregation site. Further, because most of the TFPI-2 is released into the medium (Fig. 7F, inset), it may be stored in the platelet ␣-granules. However, additional work will be required to corroborate this view because TFPI-1, a homologue of TFPI-2, has not been localized to ␣-granules or lysosomes (69).
We suggest that platelet TFPI-2 is derived from megakaryocytes. Three lines of evidence support this possibility. First, TFPI-2 concentration in blood is too low for manifestation of endocytosis by platelets. Second, MEG-01 cells express both TFPI-2 and CD41, as revealed by RT-PCR, Western blotting, and immunofluorescence (Fig. 3). Moreover, PMA up-regulates the expression of CD41 and production of platelets in MEG-01 cells (64,65). Therefore, PMA-treated MEG-01 cells might also express proportionately increased TFPI-2 for adequate amounts to exist in platelets. Our data presented in Fig. 3 corroborate this conclusion. Finally, the fetal bone marrow tissue stained with rabbit anti-TFPI-2 IgG show the presence of TFPI-2 in megakaryocytes (Fig. 4). Although in our study we used fetal bone marrow tissue, we postulate that adult megakaryocytes would also express TFPI-2. This conclusion is supported by the observation that platelets in the adult (Figs. 6, C and D, and 7F (inset)) and in the cord blood (Fig. 2D) contain TFPI-2.
In agreement with recent observations (36,38), our data indicate that plasma levels of TFPI-2 in pregnancy during the onset of labor are ϳ7 nM. Based upon previous work with TFPI-1 (46, 47, 54 -56), we evaluated the binding of TFPI-2 to purified FV and FVa as well as to FV/Va in platelets and FV in pregnant woman plasma. In ELISA experiments, TFPI-2 bound to pdFV and FV-1033 with ϳ10 nM K d and to rFVa with ϳ100 nM K d (Fig. 5 and Table 1). However, in SPR experiments, TFPI-2 bound to FV-1033 with ϳ36 nM K d and to rFVa with ϳ252 nM K d . The difference in K d values using the two techniques could reflect the experimental approaches employed in each methodology. Further, in SPR experiments, TFPI-1 bound to FV-1033 with ϳ48 nM K d and to rFVa with ϳ456 nM K d (Fig. 5 and Table  1). In previous studies, pdFV bound to TFPI-1 with ϳ15 nM K d (46). 6 Because pdFV and FV-1033 both contain all essential features for binding to TFPI-1 (57), the moderate variation in K d previously observed with pdFV (46) and currently obtained with FV-1033 could reflect slight differences in the experimental conditions used in the two laboratories. Importantly, pulldown and Western blot analysis data presented in Fig. 6 demonstrate association of TFPI-2 with FV in plasma of pregnant women and with FV/Va AR in platelets.
TFPI-1 and TFPI-2 compete with each other in binding to FV. In contrast, TFPI-1 161 fails to compete, indicating that it does not bind to FV (Fig. 5A). These data confirm previous observations (47,54,56) and further suggest that, similar to TFPI-1, the C-terminal basic region of TFPI-2 plays an essential role in its interaction with FV. Our data are also consistent with the earlier observations that TFPI-1 binds to FVa (47) but with 10 -20-fold reduced affinity (56). Whether or not the acidic residues ( 659 DDDEDS 664 and/or 688 EDEESDADYD 697 ) in FVa (70) are involved in this interaction is not known. However, concentrations of full-length TFPI-1 (71) and TFPI-2 in platelets or plasma are quite low to appreciably bind to FVa, suggesting this interaction is not biologically significant.
Five basic amino acids (KRKRK) in the C-terminal region of TFPI-1 have been implicated in binding to the B-domain AR segment of FV AR , a molecule that lacks the BR region (56). Because TFPI-1 and TFPI-2 compete with each other in binding to FV, and the C-terminal of TFPI-2 also has five conserved basic amino acids (KKKKK; Table 2), we predict that these residues bind to the FV-AR. Based on the ϳ20 nM K d , only ϳ50% of ϳ0.5 nM TFPI-1 in plasma would be associated with FV. Similarly, based on the ϳ30 nM K d , only ϳ40% or less of the ϳ7 nM TFPI-2 in pregnant women would be associated with FV. In contrast, based upon the ϳ0.1 nM K d for TFPI-1 for FVa AR present in platelets (56), a significant portion of TFPI-1 and by analogy TFPI-2 is expected to be associated with FVa AR . 7 TFPI-2 in platelets and in the plasma of pregnant women affects the hemostatic process. The primary effect observed in our assay system was on fibrinolysis (Fig. 7), whether the clotting was initiated with FXIa or with rTF. We observed ϳ40 -50% inhibition of fibrinolysis specifically attributable to TFPI-2 whether platelets or plasma from pregnant women was used. When PRP was used (Fig. 7, B and F), the remaining antifibrinolytic activity could stem from the release of plasminogen activator inhibitor 1 and ␣ 2 -antiplasmin from platelets (44,72,73). 6 Our efforts to obtain the K d value for binding of pdFV to TFPI-2 or TFPI-1 using SPR were unsuccessful. The background binding to the CM5 chip in the absence of coupled TFPI-2 or TFPI-1 was extremely high, which precluded obtaining reliable data. 7 We attempted purification of TFPI-2 from platelet lysates using an anhydroplasmin affinity column. After extensive washing with 50 mM Tris, 1 M NaCl, 2 mM CaCl 2 , pH 7.5, the bound TFPI-2 was eluted with 10 mM HCl, 500 mM KCl, pH 2.1, in tubes containing 100 l of 1 M Tris, pH 8.0, to a final volume of ϳ500 l. However, Western blots of the HCl-eluted fractions revealed the presence of both TFPI-2 and FV/Va. These data indicate that TFPI-2 has very high affinity for platelet FV/Va.
Similarly, comparison of the antifibrinolytic activity of PPP from pregnant women (Fig. 7, C and G) and normal PPP supplemented with TFPI-2 (Fig. 7, D and H) indicates that the additional antifibrinolytic activity in pregnant women possibly stems from the increased levels of plasminogen activator inhibitors 1 and 2 and thrombin-activatable fibrinolysis inhibitor (74 -76). As expected, rTF was not inhibited by TFPI-2 in PRP (Fig. 7F) or PPP (Fig. 7G) from pregnant women; the increase in clot formation time in the case of PRP (Fig. 7F) could be due to TFPI-1 released from the platelets (71). Approximately 25% inhibition of FXIa could be attributed to TFPI-2 in PRP (Fig. 7B) or PPP (Fig. 7C) from pregnant women. The additional inhibition observed in the case of PRP (Fig. 7B) could stem from protease nexin-2 released from platelets (43,77,78). Although only ϳ25% inhibition of FXIa was observed during the first 5 min of the experiment (Fig. 7), inhibition of FXIa was expected to continue during the course of the experiment. Although we did not initiate intrinsic clotting through the activation of factor XII, we expected that TFPI-2 would efficiently inhibit the generated pKLK and FXIa in our experimental system. Thus, both intrinsic clotting and fibrinolysis are regulated by platelet TFPI-2 and by the TFPI-2 present in plasma of pregnant women.
Pregnant women are at risk for life-threatening hemorrhage during labor when the placenta is separated from the uterus, exposing large maternal arteries. These profound hemostatic challenges are overcome by local and systemic enhancement of coagulation and impairment of fibrinolysis. Plasma levels of several coagulation factors and fibrinolysis inhibitors are increased severalfold in pregnancy (74 -76). In this context, TFPI-2 synthesized by the syncytiotrophoblasts (45) would provide additional antifibrinolytic activity during pregnancy/ delivery to control bleeding.
In conclusion, platelet TFPI-2 plays an important role in the hemostatic process and adds to the list of other platelet factors, including polyphosphate (79), FV (80), protease nexin 1 and 2 (77,78,81), TFPI-1 (56), and plasminogen activator inhibitor-1 (72), that act as coagulant, anticoagulant, and antifibrinolytic agents. The effect of TFPI-2 on fibrinolysis is large and agrees with the K i for plasmin being lower than for FXIa and pKLK. The data suggest a protective role of TFPI-2 against postpartum bleeding.

TABLE 2 The conserved residue Lys-131 in Kunitz domain 3 and basic residues in the C-terminal region of TFPI-2 across mammalian species
The TFP-2 residues predicted to bind to FV-AR region are in boldface type. Corresponding residues in TFPI-1 predicted to participate in binding to the FV-AR region are 254 KTKRKRKK 261 (56).