Transforming Growth Factor β1 Enhances Platelet Aggregation through a Non-transcriptional Effect on the Fibrinogen Receptor*

Upon activation, platelets store and release large amounts of the peptide transforming growth factor β1 (TGFβ1). The released TGFβ1 can then act on nearby vascular cells to mediate subsequent vessel repair. In addition, TGFβ1 may circulate to bone marrow and regulate megakaryocyte activity. It is not known what effect, if any, TGFβ1 has on platelets. Adult TGFβ1-deficient mice exhibit thrombocythemia and a mild bleeding disorder that is shown to result from faulty platelet aggregation. TGFβ1-deficient platelets are shown to contain functional receptors, and preincubation with recombinant TGFβ1 improves aggregation, demonstrating that TGFβ1 plays an active role in platelet aggregation. TGFβ1-deficient platelets fail to retain bound fibrinogen in response to aggregation agonists, but they possess normal levels of the αIIb/β3 fibrinogen receptor. Signaling from agonist receptors is normal because the platelets change shape, produce thromboxane A2, and present P-selectin in response to stimulation. Consequently, activation and maintenance of αIIb/β3 into a fibrinogen-binding conformation is impaired in the absence of TGFβ1. 4-Phorbol 12-myristate 13-acetate treatment and protein kinase C activity measurements suggest a defect downstream of protein kinase C in its activation cascade. Because platelets lack nuclei, these data demonstrate for the first time a non-transcriptionally mediated TGFβ1 signaling pathway that enhances the activation and maintenance of integrin function.

stored in large amounts as a latent peptide in the secretory ␣-granules of circulating platelets and is one of the molecules released during platelet activation (4). Active platelets can release enough TGF␤1 to raise the local concentration of TGF␤1 to as much as 40 ng/ml at the injury site and in the developing thrombus (5). TGF␤1 can influence vessel repair through regulation of endothelial cell function (6,7), smooth muscle cell differentiation (8,9), and vessel wall remodeling (10,11). It is not known whether TGF␤1 can also regulate platelet activity.
In nucleated cells, members of the Smad gene family are believed the primary mediators of intracellular signaling from the TGF␤ receptor types I and II (12,13). SMADs are able to bind DNA (14,15) and can regulate transcription of TGF␤1responsive genes (16,17). However, the type I receptor can interact with other molecules such as the ␣ subunit of farnesyltransferase (18). Additionally, TGF␤1 can stabilize protein levels in epithelial cells (19). These findings suggest that there are possible transcription-independent signaling pathways for TGF␤1 (19).
The primary defect in the TGF␤1-deficient mouse is a severe multifocal inflammatory disease resulting in death by the 3rd week after birth (20 -22). Genetic combination of Tgfb1 knockout and Scid alleles eliminates the inflammation and increases longevity by 3-5 months (23), thereby permitting investigation of adult phenotypes primary to the absence of TGF␤1 and not compromised by secondary effects resulting from inflammation. Scid Tgfb1 Ϫ/Ϫ mice have thrombocythemia and a mild bleeding disorder that is associated with faulty platelet aggregation resulting from a failure to sustain fibrinogen binding. Preincubation with active TGF␤1 improves platelet aggregation, indicating the presence of a TGF␤1 signaling pathway in platelets. Because TGF␤1 signaling in platelets is necessarily independent of transcription, these results demonstrate the existence of a non-transcriptional TGF␤1 signaling pathway that mediates platelet function by affecting the state of integrin activation.
Platelet and Megakaryocyte Counts-Blood samples were obtained by cardiac puncture of anesthetized mice using EDTA (10 mM) as an anticoagulant. Platelets in diluted whole blood were counted in duplicate with a thin hemocytometer under 100ϫ phase microscopy. Bone marrow was collected from bare femurs and tibiae of mice by flushing with a 10-ml empty syringe fitted with a 21-gauge needle rinsed with 10 mM EDTA. Drops of marrow were smeared on glass slides and stained with Wright and Giemsa stains. The average number of megakaryocytes per 10ϫ field covering the entire smear was determined.
Bleeding Time-Bleeding times were determined by a previously reported method (24). Briefly, the base of each tail was treated with a depilatory, cleaned with alcohol, and air-dried. A puncture was made along the lateral side of the tail using a sterile lancet with a blade depth of 2.5 mm and width of 1.5 mm (Microlance; Becton and Dickinson). The average bleeding time was measured by constant observation under a dissecting microscope while blotting with filter paper every 15 s so that the bleeding stop could be accurately observed. Measurements from three separate punctures were made for each mouse.
Platelet Aggregation-Blood samples were collected with a butterfly needle (25 g ϫ 3/8, 3.5-inch tubing, Abbott Laboratories) and a 3-ml syringe containing 3.8% sodium citrate through a clean abdominal vena cava puncture. Typically, we obtained 0.6 -0.8 ml of blood from Scid Tgfb1 ϩ/ϩ mice and 0.5-0.6 of blood from Scid Tgfb1 Ϫ/Ϫ mice. A blood: citrate ratio of approximately 6:1 was used in the collections by using 100 l and 80 l of citrate for Scid Tgfb1 ϩ/ϩ and Scid Tgfb1 Ϫ/Ϫ mice, respectively. Platelet-rich plasma (PRP) was prepared by centrifugation of blood at 150 ϫ g for 15 min. The remaining portion was then centrifuged 10 min at 2000 rpm to recover platelet-poor plasma (PPP). Previous experience indicated that samples showing signs of red blood cell hemolysis did not perform well and thus were discarded. PRP recovered from 3-5 mice were pooled and the platelet concentration adjusted to 3 ϫ 10 8 platelets/ml with PPP. Aggregation experiments were performed with 225 l of PRP at 37°C with constant stirring in an optical aggregometer (Chrono-log Corp.). Aggregation was initiated by adding 25 l of 10ϫ agonist. Experiments involving thrombin included 0.25 mM glycyl-prolyl-arginyl-proline peptide (GPRP, Sigma), to prevent fibrin clot formation (25). For experiments involving preincubation with TGF␤1, 250 l of PRP was incubated at room temperature for at least 1 h with active recombinant human TGF␤1 (R&D Systems) prior to performing the aggregation experiment.
Fibrinogen Binding and Flow Cytometry-Detection of bound fibrinogen was carried out essentially as described previously (25). After incubation in the dark at room temperature for 30 min, 0.5 ml of 0.2% formalin was added. To follow the release of bound fibrinogen by platelets, the anti-fibrinogen antibody was added at 0, 5, 10, and 15 min after stimulation by ADP to separate aliquots of PRP diluted in HEPES buffer. The experiment continued as described above. For detection of P-selectin, 1 g of a FITC-conjugated monoclonal rat antibody to mouse P-selectin (PharMingen) was used in place of the anti-fibrinogen antibody and was present at the time of agonist addition. Samples stimulated with thrombin contained 0.25 mM GPRP. Analysis was performed on a FACSTAR instrument (Becton Dickinson) using the FACSTAR Lysis II software. Specificity of the anti-fibrinogen antibody for fibrinogen was determined in two ways. Wild type platelets were activated in the presence of a FITC-conjugated rat anti-mouse T lymphocyte antibody (Thy1.2, PharMingen) to control for nonspecific binding of IgG by platelets. In the other control experiments, an excess of non-immune rabbit IgG was included with the anti-fibrinogen antibody during platelet activation. Activated wild type mouse platelets gave a slightly higher background signal with the anti-Thy1 antibody than resting platelets incubated with the anti-fibrinogen antibody. The excess IgG did not affect the positive fluorescent peak typically obtained with activated wild type platelets (data not shown). PAC-1, an antibody that specifically recognizes the activated form of human ␣ IIb /␤ 3 (26) did not interact with the mouse platelets (from either Tgfb1 ϩ/ϩ or Tgfb1 Ϫ/Ϫ mice) used in our experiments.
Measurement of Thromboxane Production-PRP containing 2 ϫ 10 7 platelets was incubated in the presence or absence of 10 M ADP at 37°C for 10 min without stirring. Samples were immediately centri-fuged and the supernatants transferred to fresh tubes. An aliquot (25 l) from each supernatant and plasma collected during preparation of the PRP were assayed for thromboxane B 2 levels using a colorimetric enzyme immunoassay (Biotrak, Amersham Pharmacia Biotech) according to manufacturer's instructions.
Affinity Cross-linking of 125 I-TGF␤1 to Receptors-Identification of platelet surface TGF␤1-binding proteins was performed by affinity cross-linking of 125 I-TGF␤1 with disuccinimidyl suberate (27). Platelets were collected by venipuncture and washed twice with cold binding buffer (128 mM NaCl, 5 mM KCl, 1.2 mM CaCl 2 , 5 mM MgSO 4 , 50 mM HEPES, and 5 mg/ml bovine serum albumin at pH 7.5) containing 1 M prostacyclin to prevent activation. Washed platelets (3 ϫ 10 7 ) were incubated at 4°C for 2.5 h with 150 pM 125 I-TGF␤1 in 0.5 ml of binding buffer. All incubations were performed in siliconized microcentrifuge tubes to minimize binding of the 125 I-TGF␤1 to tube surfaces. Following three washes in cold binding buffer, platelets were incubated in 0.5 ml of binding buffer with 0.25 mM disuccinimidyl suberate at 4°C for 15 min. The cross-linking reaction was stopped by washing the platelets twice in cold stop buffer (0.25 M sucrose, 1 mM EDTA, 10 mM Tris, pH 7.4) containing protease inhibitors. Each platelet pellet was solubilized for 40 min at 4°C in 20 l of detergent buffer (1% Triton X-100, 1 mM EDTA, 125 mM NaCl, 10 mM Tris, pH 7.4) containing protease inhibitors. Samples were centrifuged to remove the insoluble material and the supernatants boiled for 5 min in an equal volume of SDS loading buffer under reducing (50 mM dithiothreitol) or non-reducing (no dithiothreitol) conditions. Samples were separated on a 7% polyacrylamide gel electrophoresis gel, dried, and exposed to x-ray film (X-Omat, Eastman Kodak Co.) for 1-2 days. Confluent cultures of human iliac vein endothelial cells (HIVE) in 100-mm dishes were collected without trypsin and affinity-labeled as above. Nonspecific binding was determined by incubating HIVE with 7.5 nM cold TGF␤1 (a 50-fold excess) for 30 min before the addition of 125 I-TGF␤1.
Protein Kinase C Activity-The extent of phosphorylation of a protein kinase C (PKC)-specific peptide substrate by platelet lysates was measured using a kit supplied by Amersham Pharmacia Biotech. All procedures were performed on ice and according to the manufacturer's instructions. Equal numbers of platelets (3 ϫ 10 7 ) in PRP were centrifuged, washed with calcium-free Tyrode's buffer, and sonicated in lysis buffer. 25 l of lysate was added to a mixture containing reaction buffer, artificial membranes, 10 M 4-phorbol 12-myristate 13-acetate (PMA), peptide substrate, and [␥-32 P]ATP. After 15 min, the reaction was stopped with acid and the peptide substrate was captured on binding paper, washed in acid, and counted in a scintillation counter. Controls included reactions lacking platelet lysates or the peptide substrate to determine background counts and endogenous PKC substrate levels, respectively.

RESULTS
Scid Tgfb1 Ϫ/Ϫ Mice Exhibit a Mild Bleeding Disorder and Elevated Circulating Platelet Numbers-Scid Tgfb1 Ϫ/Ϫ mice exhibit outward signs of a mild bleeding disorder. Brains of 5/6 Scid Tgfb1 Ϫ/Ϫ neonates exhibited multiple petechia indicative of microhemorrhages, while none of the Scid Tgfb1 ϩ/ϩ neonates examined (0/4) had this condition. Adult Scid Tgfb1 Ϫ/Ϫ mice routinely have "flushed" small bowels that show no signs of inflammation by histology. Furthermore, these mice tend to bleed more readily from incisions. Scid Tgfb1 Ϫ/Ϫ mice have bleeding times nearly twice as long as wild type litter-mates (Table I). The numbers of peripheral blood platelets in both Scid and non-Scid Tgfb1 Ϫ/Ϫ mice are elevated approximately 3-fold over those from Scid or non-Scid Tgfb1 ϩ/ϩ and Scid or non-Scid Tgfb1 ϩ/Ϫ controls. Megakaryocyte counts from bone marrow are also significantly (p Ͻ 0.05) elevated in all Scid or non-Scid Tgfb1 Ϫ/Ϫ mice.
Scid Tgfb1 Ϫ/Ϫ Platelets Do Not Aggregate Properly in Vitro-To test for a deficiency in platelet aggregation, PRP collected from mice via venipuncture was assayed for aggregation competence. When stimulated with the agonists ADP, collagen, or the TXA 2 analog I-BOP (28), Scid Tgfb1 Ϫ/Ϫ platelets fail to properly aggregate as compared with Scid Tgfb1 ϩ/ϩ control platelets (Fig. 1). Similar results are obtained with ADP using platelets recovered from Rag2 Ϫ/Ϫ Tgfb1 Ϫ/Ϫ mice, which also have no inflammatory disease (data not shown). Scid Tgfb1 Ϫ/Ϫ platelets show a range of aggregation responses to 10 M ADP from approximately 0 -80% of controls. In contrast to ADP-mediated responses, Scid Tgfb1 Ϫ/Ϫ platelets consistently fail to aggregate properly following stimulation with either 20 g/ml collagen or 1 M I-BOP. TGF␤1-deficient platelets that show an 80% aggregation response to ADP fail to aggregate when stimulated with I-BOP (data not shown), indicating an agonist-specific aggregation response that is not due to variability in platelet collection. Unlike these other agonists, thrombin induces maximal, yet delayed, aggregation in platelets from Scid Tgfb1 Ϫ/Ϫ and Scid Tgfb1 ϩ/ϩ mice (Fig. 1).
Delayed Aggregation following PKC Activation of Scid Tgfb1 Ϫ/Ϫ Platelets-Platelets from Scid Tgfb1 Ϫ/Ϫ mice exhibit a rapid shape change in response to agonists as detected by the drop in light transmission during in vitro aggregation to ADP, I-BOP, and collagen (Fig. 1). In addition, TGF␤1-deficient platelets produce normal levels of TXA 2 and mobilize P-selectin to the platelet surface in response to ADP (Table II). P-selectin levels of non-stimulated platelets from both Scid Tgfb1 Ϫ/Ϫ and Tgfb1 ϩ/ϩ mice were similar to background levels, indicating that our collection protocol does not prestimulate the platelets (Table II). Stimulation of Scid Tgfb1 Ϫ/Ϫ platelets with PMA, which activates PKC independently of receptor-mediated upstream events, also results in delayed aggregation (Fig. 3A). However, when measured directly, total PKC activity in Scid Tgfb1 Ϫ/Ϫ platelets is not different from control platelets (Fig.  3A), indicating that there is no reduction in PKC enzymes.
FIG. 1. Aggregation of platelets from Scid Tgfb1 ؉/؉ and Tgfb1 ؊/؊ mice in response to various agonists. A, PRP is stirred at 37°C in a photo-aggregometer and the resulting aggregation following agonist addition is measured as the percentage of light transmission through a platelet-poor plasma blank. The agonist was added at "0" min and the aggregation response followed until the traces reached a plateau. Changes in platelet shape immediately following agonist addition appear as a drop in the percent of transmitted light (arrow in traces for ADP, collagen, and I-BOP). Each aggregation is performed with PRP pooled from three to five mice and is representative of at least four separate experiments. The traces shown for ADP labeled 1-3 are of three separate aggregation experiments using pooled platelets from different groups of TGF␤1-deficient mice and show the possible range of responses obtained with this agonist. The concentration of ADP, I-BOP, collagen, and thrombin used in the aggregations was previously determined to produce a maximal response in wild type mouse platelets. For aggregation with thrombin, 0.25 mM GPRP peptide, which inhibits fibrinogen clot formation, was included. B, distribution of aggregation responses to ADP (10 M) by platelets from Scid Tgfb1 ϩ/ϩ (q) and Scid Tgfb1 Ϫ/Ϫ (E) mice. Each data point represents an aggregation experiment involving platelets pooled from three to five mice. Each aggregation experiment with Tgfb1 Ϫ/Ϫ platelets was accompanied by a separate experiment with wild type mouse platelets using the same reagents. Paired Scid Tgfb1 ϩ/ϩ and Scid Tgfb1 Ϫ/Ϫ aggregation experiments are plotted together at the same x axis position. In response to ADP, TGF␤1-deficient platelets from 13/16 (81%) experiments aggregated to 20% of maximum or less.
The degree of rescue by TGF␤1 is independent of the extent of aggregation by TGF␤1-deficient platelets prior to addition of TGF␤1. Those platelets that initially fail to respond to ADP show the same absolute degree of response when preincubated with TGF␤1 as do platelets from Scid Tgfb1 Ϫ/Ϫ mice that exhibit 80% responsiveness to ADP alone (Fig. 3B). Co-incubation with a neutralizing antibody against TGF␤1 prevents the effect of added TGF␤1 on aggregation (data not shown). In order for the added TGF␤1 to facilitate aggregation, it is necessary to preincubate mutant platelets with at least 20 ng/ml TGF␤1 for 1 h or longer before inducing aggregation. Shorter incubation times or lower concentrations result in no improvements to aggregation. Furthermore, the addition of TGF␤1 to wild type platelets has no effect on aggregation responses to ADP, nor does TGF␤1 alone induce aggregation in either Scid Tgfb1 Ϫ/Ϫ or Scid Tgfb1 ϩ/ϩ platelets (data not shown).

Murine Platelets Possess Type I and II TGF␤1 Receptors-
The effect of exogenous TGF␤1 on Scid Tgfb1 Ϫ/Ϫ platelets demonstrates that TGF␤1 signals in platelets. To address whether the receptors are present to mediate TGF␤1 signaling, we examined mouse platelets for 125 I-TGF␤1 binding. Platelets from both Scid Tgfb1 ϩ/ϩ and Scid Tgfb1 Ϫ/Ϫ mice bind 125 I-TGF␤1 mainly through types I and II receptors (Fig. 4). Human platelets also have the type II receptor (data not shown). In contrast to human iliac vein endothelial cells, murine platelets do not possess endoglin. Furthermore, murine platelets lack the type III receptor. DISCUSSION Mice lacking TGF␤1 produce increased numbers of bone marrow megakaryocytes and circulating platelets. Previous studies have indicated that TGF␤1 inhibits bone marrow-derived megakaryocyte growth, colony formation, and platelet production (29 -32). The results observed in both the Scid and non-Scid Tgfb1 Ϫ/Ϫ mice are consistent with this role for TGF␤1 as a negative regulator of megakaryocyte growth and hence platelet production. Additionally, these mice exhibit a mild bleeding disorder characterized by a 2-fold longer bleeding time, cerebral petechia in neonates, and a "flushed" appearing a Thromboxane B 2 (TXB 2 ) (pg/l), a stable metabolite of TXA 2 , in plasma or supernatants of 1.6 ϫ 10 7 platelets (in PRP) activated for 10 min at 37°C with 10 M ADP was detected by enzyme-linked immunosorbent assay. Measurements of plasma and platelet-derived TXB 2 were made from the same three mice for each genotype.
b The percentage of platelets positive for surface P-selectin staining before (resting) or following activation (activated) by ADP was measured by flow cytometry using a FITC-conjugate antibody directed against mouse P-selectin.  3. Effect of TGF␤1 on aggregation of TGF␤1-deficient  platelets. A, a representative trace of PMA-induced platelet aggregation. Pooled PRP from Scid Tgfb1 ϩ/ϩ (ϩ/ϩ) or Scid Tgfb1 Ϫ/Ϫ (Ϫ/Ϫ) mice was stimulated by 10 M PMA with stirring in an aggregometer. PMA was added at time 0 and the change in light transmission followed for 25 min. Inset, total PKC activity in lysates from equal numbers of platelets from Scid Tgfb1 ϩ/ϩ and Scid Tgfb1 Ϫ/Ϫ mice. Values are mean Ϯ S.E. of (n) mice indicating the picomoles of P i /min/3 ϫ 10 7 platelets. Results are corrected for the labeling of endogenous PKC substrates. B, improved aggregation of platelets in the presence of TGF␤1. Representative aggregation tracings from two separate experiments (1 and 2) similar to those described in Fig. 1. PRP pooled from four TGF␤1-deficient mice were prepared for aggregation with PPP as described in Fig. 1 and incubated at room temperature with (ϩTGF␤1) or without (ϪTGF␤1) recombinant human TGF␤1 (20 ng/ml) for 1 h before activation with 10 M ADP in the aggregometer. Incubated samples were used directly for aggregation so that recombinant TGF␤1 was present during the aggregation. An aggregation trace of platelets in plasma from Scid Tgfb1 ϩ/ϩ mice (ϩ/ϩ) incubated at room temperature for 1 h is included as a reference.
small bowel in adults. Associated with the elevated platelet numbers is an aggregation deficiency in response to ADP, collagen and I-BOP, a TXA 2 analog. Thrombin, a potent platelet agonist, induces slightly delayed but full aggregation. The platelet aggregation defect is sufficient to explain the bleeding disorder. However, contributing factors such as vascular fragility or endothelial cell dysfunction cannot be discounted. Essential thrombocythemia, a human myeloproliferative disorder, is characterized by clonal expansion of megakaryocytes, persistent elevated platelet counts, and abnormal platelet function (33,34). The similarities between TGF␤1-deficient mice and essential thrombocythemia patients suggest that defects in the TGF␤1 signaling pathway or a pathway regulated by TGF␤1 may be altered in essential thrombocythemia patients.
Platelet aggregation in response to ADP is poor but variable. The cause of this variability is not clear. Platelets from Scid Tgfb1 Ϫ/Ϫ mice that responded well to ADP failed to aggregate in response to the thromboxane A 2 analog, I-BOP, suggesting that the variability is specific to ADP. The difference in responses to ADP and I-BOP also indicates that, unlike human platelets, murine platelets do not depend on TXA 2 as an intermediate in ADP-stimulated aggregation. Recently, murine platelets were shown to be insensitive to cyclooxygenase inhibitors (35), consistent with our observations. From this, we conclude that ADP activates mouse platelets through a direct, TXA 2 -independent pathway and that ADP is a stronger platelet agonist in mouse than in human. Thus, the variability in Scid Tgfb1 Ϫ/Ϫ platelet response to ADP may reflect the strong stimulation by ADP that can partially override the defect.
Even though TGF␤1-deficient platelets do not aggregate properly in response to ADP, they do change shape and generate TXA 2 . Furthermore, TGF␤1-deficient platelets exhibit normal agonist-induced ␣-granule secretion. However, platelets from Scid Tgfb1 Ϫ/Ϫ mice exhibit impaired fibrinogen binding, which can fully account for the defective aggregation. These observations suggest that this defect occurs late in the activation cascade, specific to ␣ IIb /␤ 3 activation. Similar to Scid Tgfb1 Ϫ/Ϫ mouse platelets, normal human platelets treated with phosphoinositol 3-kinase (PI3-K) inhibitors change shape, produce thromboxane, and mobilize P-selectin in response to agonists. However, PI3-K-inhibited platelets are unable to induce and maintain ␣ IIb /␤ 3 in an active form (36,37). The similarity strongly suggests the aggregation defect observed in Scid Tgfb1 Ϫ/Ϫ platelets reflects impaired PI3-K-mediated ␣ IIb /␤ 3 activation.
Our results show that murine platelets bind TGF␤1 through a competent TGF␤ type I/II receptor system. Binding of TGF␤1 to both receptor types requires receptor heterodimerization, thereby forming the active signaling complex (38). In murine platelets binding of 125 I-TGF␤1 to both types I and II receptors indicates that heterodimerization is occurring. Unlike endothelial cells, which share a common lineage with platelets, murine platelets do not bind TGF␤1 through endoglin (Fig. 4). Furthermore, platelets do not posses the type III binding protein. Thus, any TGF␤1-derived signals would be transduced from primarily the classic type I/II receptor system. Because platelets lack a nucleus, TGF␤1 must be regulating platelet activity independent of transcription. In other cell types the type II TGF␤ receptor can interact with the ␣ subunit of farnesyltransferase (18,39), suggesting that TGF␤1 may be affecting protein prenylation states in platelets. In addition, TGF␤1 can regulate protein translation and stability (19,40) and may play a similar role in platelets. The length of preincubation required for TGF␤1 to facilitate platelet function is consistent with either of these two mechanisms. Preliminary studies using puromycin suggest that TGF␤1 does not signal through protein translation (data not shown). Further studies involving TGF␤1-deficient mouse platelets should prove useful in identifying nonnuclear mediators of TGF␤ signaling and may reveal additional insights into platelet activation mechanisms.
Preincubating platelets from Scid Tgfb1 Ϫ/Ϫ mice with active TGF␤1 facilitates aggregation. However, TGF␤1 alone does not induce aggregation. Thus TGF␤1 augments rather than initiates aggregation. The time of TGF␤1 preincubation (Ͼ1 h) needed to affect platelet activity and the relatively short time of ADP aggregation (5 min) suggest that TGF␤1 is not acting during platelet aggregation. Instead, it may precondition the platelet to ensure a vigorous response in conditions requiring rapid and complete formation of a hemostatic plug. In this regard, the long preincubation period may mimic the continual exposure to low levels of TGF␤1 during the life of the platelet in the wild type mouse. This may also explain why platelets from control mice are unresponsive to exogenous TGF␤1. Plasma contains low levels of circulating, active TGF␤1 bound to carriers such as ␣-macroglobulin (41,42). The presence of the TGF␤1 carriers in the plasma of Scid Tgfb1 Ϫ/Ϫ mice may account for why a relatively large amount of TGF␤1 (20 ng/ml) was needed to improve aggregation responses of Scid Tgfb1 Ϫ/Ϫ platelets. Presumably, these carrier molecules are depleted of TGF␤1 in the Scid Tgfb1 Ϫ/Ϫ mice and may sequester a significant portion of the exogenously added active TGF␤1 during our experiments.
In conclusion, the absence of TGF␤1 in mice leads to compromised platelet aggregation resulting in a mild bleeding disorder. The inability of TGF␤1-deficient platelets to aggregate properly is due to faulty fibrinogen binding involving downstream targets of PKC (e.g. PI3-K). The compromise in platelet activity within the Scid Tgfb1 Ϫ/Ϫ mouse demonstrates that TGF␤1 is important in hemostatic activities other than putative vascular cell regulation. Because platelets lack a nucleus, FIG. 4. Autoradiograph of 125 I-TGF␤1 cross-linked platelets detecting TGF␤1-binding proteins on the platelet surface. Equal numbers of platelets from Scid Tgfb1 ϩ/ϩ mice (ϩ/ϩ) and Scid Tgfb1 Ϫ/Ϫ mice (Ϫ/Ϫ) were washed twice in the presence of 1 M prostacyclin and labeled by chemical cross-linking to 125 I-TGF␤1. Platelet lysates were separated on 7% polyacrylamide gel electrophoresis gel under reducing (R) or nonreducing (NR) conditions and the position of receptor-bound 125 I-TGF␤1 was detected by exposure to film. Under reducing conditions, the 185 kDa endoglin molecule migrates to a position near the type II receptor while under non-reducing conditions, endoglin migrates as a broad smear at the top of the gel. We detected no differences in TGF␤1 receptor levels between platelets of Scid Tgfb1 ϩ/ϩ and Scid Tgfb1 Ϫ/Ϫ mice in three separate experiments. Experiments with HIVE, which possess endoglin but not type III receptor, were included for comparison. The result from preincubation of HIVE for 30 min with a 50-fold excess of cold TGF␤1 is shown in the rightmost lane under HIVE. The numbers to the right of the autoradiograph represent the positions of molecular markers in kDa. TRI, TGF␤ receptor type I; TRII, TGF␤ receptor type II. such activities would be independent of nuclear-mediated responses such as gene transcription.