Interaction of thrombopoietin with the platelet c-mpl receptor in plasma: binding, internalization, stability and pharmacokinetics


Dr David J. Kuter Hematology/Oncology Unit, Massachusetts General Hospital, Harvard Medical School, 100 Blossom Street, Boston, MA 02114, U.S.A. e-mail: kuter.david


Thrombopoietin (TPO) is the primary regulator of platelet production and acts through binding its receptor, c-mpl, found on megakaryocyte progenitor cells, megakaryocytes and platelets. Circulating levels of TPO are regulated primarily by the clearance of TPO after it binds to c-mpl receptors on circulating platelets. In this study the interaction of TPO with the platelet c-mpl receptor has been analysed under physiological conditions using radiochemical and pharmacokinetic approaches. 125I-rHuTPO was prepared using a novel method of gentle iodination that preserved its biological activity and used to demonstrate that platelets, but not endothelial cells, have a single class of binding sites (56 ± 17 binding sites/platelet) with high affinity (Kd = 163 ± 31 pM). Cross-linking experiments confirmed that TPO, but not erythropoietin (EPO), specifically associated with the 95 kD platelet c-mpl receptor. Upon addition of TPO to platelets, 80% of the TPO binding sites were internalized within an hour and were not recycled. TPO that was not bound by platelets was stable for up to 6 d in both platelet-poor and platelet-rich plasma. Using unlabelled recombinant human TPO (rHuTPO), standard pharmacokinetic analysis demonstrated that platelets have an average TPO clearance of 1.24 ± 0.38 ml/h/109 platelets and that TPO clearance was reduced by low temperature but not by a number of drugs or metabolic inhibitors. The maximal amount of TPO removed by platelets in vitro was identical to that predicted by the total number of TPO binding sites. These results provide a biochemical and pharmacokinetic basis for the clinical use of TPO and for understanding possible disorders of platelet production.

The production of platelets from multipotent haemopoietic progenitor cells and megakaryocytes in humans and animals is precisely regulated (Kuter, 1997). Recently, thrombopoietin (TPO) has been identified, cloned, and shown to be the primary regulator of platelet production (Kaushansky, 1995). The evidence for its role in megakaryocytopoiesis and thrombopoiesis comes from in vitro and in vivo studies using animal and human models. Mice lacking either TPO or the TPO receptor (c-mpl) reduced their platelet production to a basal level of about 15% of normal (de Sauvage et al, 1996; Gurney et al, 1994). On the other hand, overexpression of TPO in mice increased the platelet concentration 3–4-fold (Villeval et al, 1997) and administration of recombinant TPO to normal animals or humans significantly increased the platelet count in a dose-dependent manner (Basser et al, 1997; Farese et al, 1995; Goodnough et al, 1997; Harker et al, 1996). TPO has been shown to stimulate megakaryocyte proliferation, differentiation and maturation (Broudy & Kaushansky, 1995; Debili et al, 1995; Kuter et al, 1994; Teramura et al, 1997) as well as prevent the apoptosis and support the proliferation of multilineage haemopoietic progenitor cells (Itoh et al, 1996; Kato et al, 1996; Kaushansky, 1997; Ramsfjell et al, 1996; Shibuya et al, 1998).

The thrombopoietin receptor is the product of the protooncogene, c-mpl, and is a member of the cytokine receptor superfamily as characterized by its double haemopoietin receptor domain with two pairs of conserved cysteine residues and the WSXWS motif. Although the c-mpl receptor was identified first, its ligand (c-mpl ligand) was later identified as thrombopoietin (Bartley et al, 1994; de Sauvage et al, 1994). c-mpl is present primarily on platelets, megakaryocytes and some CD34+ progenitor cells (Columbyova et al, 1995; Methia et al, 1993; Takeshita et al, 1997), consistent with the observation that the megakaryocyte lineage and CD34+ progenitor cells are the major physiological cellular targets for TPO. When c-mpl expression was blocked by using antisense oligonucleotides, megakaryocyte proliferation and differentiation were markedly reduced (Methia et al, 1993).

Upon binding to its receptor, TPO promotes megakaryocyte viability, proliferation and differentiation via a number of down-stream intracellular biochemical reactions, including tyrosine phosphorylation of the megakaryocyte TPO receptor and other signal molecules (Drachman et al, 1995). Similar signal transduction events occur in the enucleate platelets but do not promote viability, proliferation or differentiation. Instead, they appear to make platelets more haemostatically reactive (Chen et al, 1995; Harker et al, 1996).

In addition, upon binding to its receptor, TPO is internalized and degraded. Although presumably also occurring in megakaryocytes, this property of the TPO receptor has been studied primarily in platelets where it plays a key role in regulating the amount of TPO in the circulation (Kuter, 1997; Kuter et al, 1994; Kuter & Rosenberg, 1994, 1995). When the platelet mass declines, TPO levels rise but there is no change in hepatic mRNA production (Stoffel et al, 1996). With persistent thrombocytopenia, the steady-state thrombopoietin levels remain elevated and are increased exponentially and proportionally to the linear decrease in the platelet mass (Kuter & Rosenberg, 1995; Nichol et al, 1995). In animals lacking the c-mpl receptor, TPO levels are elevated due to a failure of the platelets to clear the circulating TPO (Gurney et al, 1994), again demonstrating that TPO binding to its platelet receptor is the major mechanism of TPO metabolism.

Understanding the mechanism by which the platelet TPO receptor binds and clears thrombopoietin from the circulation is crucial for the rational clinical use of TPO and has been the focus of our studies. Although high-affinity TPO binding to the c-mpl receptor on the platelet has been described (Broudy et al, 1997; Fielder et al, 1997; Li et al, 1996) in animals and in a small number of human subjects, there is a discrepancy in terms of the Kd and TPO binding site density. Furthermore, there is little information available about how the TPO/receptor complexes formed upon binding are subsequently internalized and metabolized in human platelets. Finally, there has been no pharmacokinetic analysis of the interaction of TPO with its receptor on the platelet. In this report we extensively analysed the biochemical and pharmacokinetic properties of TPO binding to human platelets under physiological conditions.



Chemicals of special note were purchased as follows: anagrelide from Roberts Pharmaceutical Co. (Eatontown, N.J.); Na125I (643.8 MBq/μg) from DuPont Corp. (Wilmington, Del.); vincristine, cytochalasin B, sodium azide, Iodogen (1,3,4,6-tetrachloro-3a,6a-diphenylglycouril), and agarose conjugated with Protein A/Protein G from Sigma Chemical Co. (St Louis, Mo.); disuccinimidyl suberate (DSS) from Pierce Co. (Rockford, Ill.); goat antibody against human TPO from R&D Systems (Minneapolis, Minn.). Recombinant human thrombopoietin (rHuTPO), pegylated recombinant human megakaryocyte growth and development factor (PEG-rHuMGDF), and the c-mpl/32D cell line were gifts from Amgen Inc. (Thousand Oaks, Calif.). The c-mpl/32D cell line expresses the full-length human c-mpl receptor and is dependent for growth upon the presence of either Il-3 or TPO (Bartley et al, 1994) and was maintained in Dulbecco's Modified Eagle's Medium supplemented with 5% fetal calf serum and 10 ng/ml human IL-3.

Iodination of rHuTPO

The procedure for iodination was modified from a method previously described (Li & Roberts, 1994). In brief, 5 μg of rHuTPO were iodinated with 18.5 MBq of 125I in 50 μl of 0.5 M Na-phosphate buffer (pH 7.4) in the presence of 1 μg of Iodogen. The reaction was allowed to proceed at 25°C for 4 min and stopped by addition of 200 μl of ice-cold PBS. The radioactive protein was separated from free 125I by chromatography on a column of Sephadex G-50 equilibrated with PBS containing 0.1% gelatin.

Bioassay of rHuTPO

The c-mpl/32D cell line was used to determine the biological activity of rHuTPO. c-mpl/32D cells (2 × 104) were incubated with a series of concentrations of rHuTPO in Dulbecco's Modified Eagle's Medium containing 5% fetal calf serum for 48 h at 37°C. The growth of the cells was measured by the MTT [3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide, Sigma Chemical Co., St Louis, Mo.] method as described previously (Mosmann, 1983).

Preparation of platelet-rich plasma (PRP) and platelet-poor plasma (PPP)

Peripheral blood was drawn into a tube containing 15% acid-citrate-dextrose anticoagulant. PRP was prepared by centrifuging the sample at 300 g for 10 min at room temperature. For some experiments, PRP was obtained from MGH Blood Transfusion Service from whole blood obtained from normal human donors after centrifugation in a preparative blood centrifuge at 400 g and the PRP siphoned off. The platelet number and purity were determined by using the Unopette collection system (Becton Dickinson, New Jersey) and phase contrast microscopy. Less than 0.0001% of the cells were white blood cells. PPP was obtained by centrifugation of PRP for 15 min at 2000 g and then filtered through a 0.45 μm filter to remove platelet microparticles.

125I-rHuTPO binding to platelets in human plasma

For the competitive binding assays, 3 ng of 125I-rHuTPO were incubated with 200 μl of PRP (300–400 × 109 platelets/l) in the absence or in the presence of different amounts (0.2 ng/ml to 2 μg/ml) of unlabelled rHuTPO at 25°C for 1 h. For the rHuTPO binding saturation assay, various amounts of 125I-rHuTPO [from 1 to 28 ng/ml (12–340 pM)] were incubated with 250 μl PRP in the absence (total binding) or the presence (non-specific binding) of 500 ng of unlabelled rHuTPO at 25°C for 1 h. The platelets were pelleted at 13 000 g for 3 min, washed once with 400 μl of Hank's Balanced Salt Solution without Ca++ or Mg++ (HBSS), and the radioactivity associated with the washed pellets counted in a gamma counter. Specific binding was calculated by subtracting the non-specific binding from the total binding.

Measurement of TPO receptor internalization into platelets

The internalization of the TPO receptor into platelets was determined by measuring the amount of 125I-rHuTPO bound to platelets which could not be displaced by an excess of unlabelled TPO. Platelets in 500 μl of plasma were pre-treated with 125I-rHuTPO or unlabelled rHuTPO for various times, then washed with HBSS by gentle centrifugation at 800 g for 2 min in the microcentrifuge, and the pellet resuspended with homologous PPP containing unlabelled rHuTPO or 125I-rHuTPO, respectively. After further incubation for the indicated time, the platelets were pelleted by centrifugation and washed once with HBSS followed by counting platelet-associated radioactivity.

Cross-linking of 125I-rHuTPO to platelet receptors

20 ng of 125I-rHuTPO were incubated with 500 μl of PRP (400 × 109 platelets/l) for 45 min at 25°C and then pelleted by centrifugation. The platelet pellet was washed once with 500 μl of PBS and resuspended in 500 μl of PBS. Cross-linking was performed by the addition of 1 mM DSS at room temperature for 15 min. The reaction was stopped by adding 1 volume of ice-cold 2 × RIPA buffer (40 mM Tris-HCl, pH 7.4, 2% Na-deoxycholate, 1% NP-40, 300 mM NaCl, 2 mM NaF, 2 mM Na-orthovanadate, 1 mM EDTA, 1 mM EGTA, 1 mM PMSF, 20% glycerol). The cells were lysed by agitation at 4°C for 20 min. Insoluble protein was removed by centrifugation at 13 000 g for 10 min. The 125I-rHuTPO/c-mpl complex was concentrated by immunoprecipitation and analysed by autoradiography as follows.

TPO immunoprecipitation and electrophoresis

Platelet-poor plasma or RIPA buffer containing 125I-rHuTPO was incubated with 5 μg of goat antibodies against rHuTPO (R&D Systems, Minneapolis, Minn.) at 4°C for 14–16 h. The 125I-rHuTPO/Ab complex was collected by addition of 25 μl of agarose-conjugated protein A/protein G at 4°C for 4 h. The beads were washed with 1 × RIPA buffer three times and the immunoadsorbed complexes were solubilized by addition of 20 μl of sample buffer (50 mM Tris-HCl, pH 6.8, 1% 2-mercaptoethanol, 2% SDS, 0.1% bromphenol blue and 10% glycerol) and boiled for 2 min. The 125I-rHuTPO was analysed by SDS-PAGE after which the gel was dried and subjected to autoradiography.


TPO concentration was measured using a ‘sandwich’ ELISA assay (Nichol, 1998) kindly provided by Janet Nichol (Amgen Inc., Thousand Oaks, Calif.). Its sensitivity was 17 pg/ml and it was linear up to 1000 pg/ml (Vonderheide et al, 1998). Samples were assayed in triplicate at two different protein concentrations and an average value was calculated. In our hands the average coefficient of variation of this assay was 12%.

Pharmacokinetic analysis of the interaction of TPO with platelets in vitro

1500 pg of rHuTPO or an equivalent volume of PBS were added to 1 ml aliquots of PRP or PPP and incubated at 37°C for up to 5 h. At intervals, samples were centrifuged at 2000 g for 5 min, the supernatant removed and assayed for its content of TPO by the TPO ELISA assay. In some experiments, variable amounts (ranged from 1/1000 to 1000 times the ED50) of vincristine, cytochalasin B, sodium azide, anagrelide, or infusible platelet membrane (IPM, Cypress Bioscience Inc., Redmond, Wash.) were added to the incubation mixture to assess their in vitro effect on TPO clearance by platelets.

Statistical methods

in vitro pharmacokinetic parameters for the interaction of TPO with platelets were determined with the WinNonlin program (Scientific Consulting Inc.) using a single bolus, one-compartment model (Gabrielsson & Weiner, 1994). Scatchard parameters (Bmax., Kd) for TPO binding to platelets were determined using Ligand 4.0 (Biosoft, Cambridge, U.K.).


Iodination of recombinant human TPO (rHuTPO)

One of the problems in studying TPO binding to its receptor is the difficulty in producing high-quality radiolabelled ligand that retains enough radiospecific activity without sacrificing the ligand's biological specific activity. Conventional iodination of rHuTPO protein with chloramine-T (McConahey & Dixon, 1980) results in rapid loss of biological function (data not shown), whereas the standard Bolter-Hunter method (Langone, 1980) yields a protein with very low radiospecific activity (<74 MBq/mg protein). With Iodogen, rHuTPO becomes totally inactivated after exposure to 10 μg Iodogen for just 1 min (data not shown) in the presence or absence of 125I. To minimize the presumed oxidative damage, we developed a milder Iodogen-based iodination method using one-tenth the amount of Iodogen as previously suggested (Li & Roberts, 1994). Under these conditions, the specific radioactivity of 125I-rHuTPO typically ranged from 444 to 629 MBq/mg, whereas >60% of its initial biological activity on c-mpl/32D cells was retained, as shown in Fig 11A. As demonstrated by SDS-PAGE, the iodinated rHuTPO was homogenous with a relative molecular weight of 95 kD (Fig 1B) and showed minimal signs of degradation.

Figure 1.

. Evaluation of 125I-rHuTPO. (A) The biological activity of 125I-rHuTPO was compared with unlabelled rHuTPO using c-mpl/32D cells as described in Materials and Methods. At increasing TPO concentrations (horizontal axis) there was increased viability of the c-mpl/32D cells as measured by the metabolism of MTT (vertical axis). (B) Analysis of fractions of 125I-rHuTPO eluting at the peak of radioactivity after iodination and Sephadex G-25 chromatography. Fractions were examined on a 10% SDS-PAGE gel followed by autoradiography. Fraction 7 (Fxn 7) was routinely used in all subsequent experiments. The arrow indicates the position of rHuTPO.

125I-rHuTPO binds to its receptor on platelets

To assess the biochemical properties of thrombopoietin binding to platelets under physiological conditions, all of the following binding experiments were carried out in platelet-rich plasma at a platelet concentration at 300–400 × 109/l. The extent of specific binding of 125I-rHuTPO to platelets was dependent on the concentration of unlabelled TPO as increasing the concentration of unlabelled rHuTPO, but not recombinant human erythropoietin (rHuEPO), could abolish 125I-rHuTPO binding to platelets (Fig 2A). To quantify the binding affinity and binding site density for rHuTPO on platelets, Scatchard analysis was applied to TPO saturation binding studies. The amount of specific binding of 125I-rHuTPO to platelets increased as the concentration of the radioactive ligand was raised and approached a plateau at a concentration of approximately 200 pM (data not shown). Scatchard analysis indicated a single class of high-affinity binding sites for TPO on the platelets. The platelet binding data (Fig 2B) for eight normal individuals showed that each fit a model for a single class of binding sites for TPO with an average binding dissociation constant (Kd) of 163 ± 31 pM (range 123–218 pM, n = 8). Each platelet contained an average of 56 ± 17 (range 39–81, n = 8) TPO binding sites.

Figure 2.

. Binding of 125I-rHuTPO to the surface of the platelet. (A) Competitive binding of rHuTPO to human platelets. 3 ng of 125I-rHuTPO were incubated with 400 μl of PRP (platelet concentration = 300 × 109/l) at 25°C for 1 h in the presence of various concentrations of either rHuTPO or recombinant human erythropoietin (rHuEPO). After washing the platelets with HBSS, the platelet associated radioactivity was counted. Each point represents an average of duplicates. (B) Analysis of TPO binding affinity (Kd) and binding site density (Bmax) for platelets from eight normal individuals. The means ±SD are also plotted.

Formation of a 125I-rHuTPO/c-mpl complex on platelets

TPO binding to its receptor and the formation of a ligand/receptor complex is the initial step for TPO to exert its biological activity. To assess directly the formation of this TPO/c-mpl complex, chemical cross-linking experiments were performed as shown in Fig 3. After chemical cross-linking with disuccinimidyl suberate (DSS), a new radioactive protein band with a molecular weight of 190 kD (lane 1) appeared. Subtracting the 95 kD molecular weight of 125I-rHuTPO, the molecular weight of its receptor may be estimated at about 95 kD, which is the size of full-length c-mpl (Wendling & Vainchenker, 1995). Addition of an excess of unlabelled rHuTPO (lane 3) or omission of the chemical cross-linker DSS (lane 4), but not the addition of unlabelled rHuEPO (lane 2), could abolish this band (Fig 3A), indicating that the band represents a specific TPO/c-mpl complex. Further confirmation of these results comes from the demonstration that increasing the concentration of unlabelled rHuTPO could reduce the amount of radioactive TPO/c-mpl receptor complex in a dose-dependent manner (Fig 3B).

Figure 3.

. Electrophoretic analysis of rHuTPO/c-mpl complexes (arrow) formed upon chemical cross-linking with DSS. (A) 20 ng of 125I-rHuTPO were incubated with 5 × 108 platelets in 1 ml PRP for 45 min at 25°C in the absence (lane 1) or presence of 1 μg of either rHuEPO (lane 2) or rHuTPO (lane 3). The cross-linking reaction was performed in the presence (lanes 1–3) or absence (lane 4) of 1 mM DSS. The rHuTPO/c-mpl complexes were immunoprecipitated with antibody against TPO and analysed by 6% SDS-PAGE and autoradiography as described in Materials and Methods. (B) 20 ng of 125I-rHuTPO were bound to 5 × 108 platelets in 1 ml of plasma in the presence of 0 (lane 1), 1 (lane 2), 10 (lane 3), 100 (lane 4) and 1000 (lane 5) ng/ml of unlabelled rHuTPO followed by cross-linking and electrophoresis as described above.

Internalization of rHuTPO bound to platelets

The binding of a cytokine to its receptor on the cell surface usually initiates a dynamic irreversible process of internalization of the bound ligand, resulting in a reduction of the total ligand binding capacity of the cell. To test if a similar phenomenon occurs upon TPO binding, platelets were pretreated with unlabelled rHuTPO and their subsequent ability to bind 125I-rHuTPO compared to platelets without pre-treatment. As shown in Fig 44A, pretreatment of platelets with unlabelled rHuTPO reduced the subsequent capacity of the platelets to bind 125I-rHuTPO by 70%.

Figure 4.

. Internalization of TPO binding sites on human platelets. (A) Pretreatment of platelets with unlabelled rHuTPO decreases 125I-rHuTPO binding to platelets. 500 μl of PRP (platelet concentration = 400 × 109/l) were treated with buffer [total binding (TB) and non-specific binding (NSB)] or 1 nM unlabelled rHuTPO (pretreatment) at 25°C for 60 min. After washing twice with HBSS, the platelets were resuspended in 500 μl of homologous filtered PPP containing 200 pM125I-rHuTPO and incubated at 25°C for another 90 min in the absence [total binding (TB) and pretreatment] or in the presence [non-specific binding (NSB)] of 1 nM unlabelled rHuTPO, washed once with HBSS, and the platelet-associated radioactivity (cpm ± SD) measured. (B) Extent of platelet TPO receptor internalization after 60 min incubation with rHuTPO. 500 μl of PRP (400 × 109 platelet/l) were incubated with 3 ng of 125I-rHuTPO for 60 min (◆). Then 1 μg of unlabelled rHuTPO was added (arrow) to a portion of the sample and the incubation continued for another 120 min (▪). At the indicated times platelet-associated radioactivity (±SD) was counted. The amount of non-specific binding (1 μg of unlabelled rHuTPO added at the start of the 125I-rHuTPO labelling) in this experiment is also presented (▴). (C) Time course of TPO receptor internalization into platelets. 500 μl of PRP (350 × 109 platelet/l) were incubated with 6 ng of 125I-rHuTPO at 25°C in the absence (total binding) or in the presence (non-specific binding) of 1 μg of unlabelled rHuTPO. At the indicated times the platelets were washed with HBSS and a portion of each sample removed for measurement of the platelet-associated radioactivity. Platelet total specific binding (total binding − non-specific binding) was then calculated. The rest of each sample was resuspended in 500 μl of homologous filtered plasma containing 1.5 μg of unlabelled rHuTPO for another 90 min incubation followed by measurement of platelet-associated radioactivity. Specific non-displaceable binding (total binding − non-specific binding) was then calculated. At each time point the extent of internalization of TPO binding sites was calculated as the ratio of specific nondisplaceable binding divided by total specific binding. Each point represents the average of duplicates (±SD). For (A), (B) and (C) each figure is representative of three independent experiments.

This reduction in the platelet binding capacity for 125I-rHuTPO may be attributed either to reversible receptor occupancy or to irreversible receptor internalization. To distinguish between these two mechanisms, platelets were incubated with 60 pM125I-rHuTPO for 60 min, at which time excess unlabelled TPO (20 nM) was added to the platelets and platelet-associated radioactivity was measured at different times as shown in Fig 44B. Addition of unlabelled TPO decreased the 125I-rHuTPO platelet-associated radioactivity within 30 min with a maximal displacement of 30% reached after about 1 h. However, 60–80% of the platelet-associated TPO was no longer able to be displaced by an excess of unlabelled TPO after 60 min (Fig 4B). This indicates that the majority of TPO binding sites are internalized after 1 h.

To quantify further the time course for internalization of the TPO/c-mpl complex, the percentage 125I-rHuTPO associated with the platelet but not displaceable by unlabelled rHuTPO was measured after 2–120 min of incubation (Fig 4C). The percentage of internalized TPO binding sites increased rapidly with time reaching a maximum within 30 min at which time 70–80% of all the TPO binding sites were internalized. Subsequently, the extent of internalized TPO binding sites remained constant at about 80% even after incubation times as long as 120 min. These results suggest that once TPO binds to its receptor, the receptor–ligand complex is rapidly internalized leaving no more than 20% of the total binding sites available on the platelet surface.

The TPO receptor does not recycle after internalization

To assess the possibility that the TPO receptor is recycled for further TPO binding and internalization, 125I-rHuTPO was incubated with platelets for up to 24 h at 25°C. As shown in Fig 5, platelet uptake of 125I-rHuTPO was maximum after 1 h; subsequent incubation of 125I-rHuTPO with platelets for up to 24 h did not increase specific TPO uptake by platelets, suggesting that the TPO receptor is used only once to transport TPO into platelets.

Figure 5.

. Platelet TPO receptors are not recycled. 500 μl of PRP (320 × 109 platelet/l) were incubated with 3 ng of 125I-rHuTPO in the absence (total uptake) or in the presence (non-specific uptake) of 1 μg unlabelled rHuTPO. At the indicated times the platelet-associated radioactivity was measured. The specific platelet TPO uptake was obtained by subtracting the non-specific uptake from the total uptake. Each bar represents the average of duplicates (±SD).

125I-rHuTPO does not bind to human umbilical vein endothelial cells (HUVEC)

c-mpl mRNA was previously reported to be present in human endothelial cells when measured by RT-PCR (Cardier & Dempsey, 1998; Methia et al, 1993). We therefore sought to determine whether 125I-rHuTPO could bind to HUVEC cells. The total binding of 125I-rHuTPO to HUVEC cells was not affected by increasing the concentration of unlabelled rHuTPO and demonstrated only a low level of non-specific binding (data not shown), suggesting that HUVEC cells do not express significant amounts of functional c-mpl protein on the cell surface.

Unbound 125I-rHuTPO is stable in platelet-rich plasma

In order to assess the stability of the rHuTPO protein under physiological conditions, we measured the change in the molecular weight of 125I-rHuTPO during its incubation in platelet-poor plasma (PPP). Under these conditions 125I-rHuTPO undergoes no significant proteolysis in PPP even after a 6 d incubation at 25°C (data not shown). To determine whether the presence of platelets would increase TPO proteolysis, 125I-rHuTPO was incubated with platelet-rich plasma (PRP) under identical conditions. The concentration and molecular weight of unbound 125I-rHuTPO did not change during a 6 d incubation and there was no significant degradation of the protein (data not shown). However, the relatively small amount of TPO bound to platelets underwent partial degradation (data not shown), as reported by others (Fielder et al, 1997).

Pharmacokinetic analysis of the interaction of TPO with platelets

Although the preceeding studies clearly demonstrated that 125I-rHuTPO binds to platelets with high affinity and is internalized, we next sought to quantify this interaction using unlabelled rHuTPO and standard pharmacokinetic methods. Platelets were incubated with 1500 pg/ml rHuTPO, a concentration 10-fold greater than normal but one which was found in other experiments to be the steady state level in patients with low platelet counts (i.e. aplastic anaemia). In PRP, there was a rapid decline in the amount of unbound TPO over the 5 h incubation (Fig 6A) with a decline to an average of 20% of the starting amount (from 1172 ± 116 pg/ml at T = 0 to 231 ± 86 pgml at 5 h, n = 7). In contrast, the addition of rHuTPO to PPP prepared from the same normal subjects was followed by a modest decline in the concentration of TPO to 77% of the starting amount (from 1172 ± 116 pg/ml at T = 0 to 896 ± 114 pg/ml after 5 h, n = 7). For PRP the majority of the decline in TPO concentration occurred in the first 3 h and there was a small subsequent decrease over another 45 h of incubation. The small reduction in TPO concentration in PPP is apparently not due to proteolysis (vide supra) but may be due to non-specific binding to other proteins or the surface of the tube. However, the large decrease in TPO concentration in PRP was attributable to the platelets and over 5 h resulted in the removal of 65% of the rHuTPO (757 ± 241 pg, n = 7).

Figure 6.

. Pharmacokinetic analysis of the interaction of TPO with platelets in vitro. (A) As described in Materials and Methods, PRP (○, platelet count = 477 × 109/l) and PPP (▪) were incubated with 1500 pg/ml of rHuTPO and the concentration (±SD) of unbound TPO measured at the indicated times. (B) Linear regression analysis of the data in (A) using a single bolus, one-compartment model with C(t) = D/V × exp (−kt), where C is the TPO concentration, D is the total dose of TPO introduced into the compartment, V is the volume of distribution, and k is the rate constant for elimination of the TPO from the compartment. (r = 0.99).

The decline in TPO concentration in vitro closely fit (r = 0.99) a simple first order rate of elimination pharmacokinetic model (Fig 6B) for all platelet samples tested. Using this first-order model, standard pharmacokinetic parameters, including the clearance rate, could be obtained. Analysis of platelets from seven normal individuals (Table I) showed that an average human platelets have a TPO clearance (ClPlt) of 0.45 ± 0.15 ml/h. Since TPO clearance in this model was linearly related to the platelet count (data not shown), the average ClPlt for normal human platelets was determined to be 1.24 ± 0.38 ml/h/109 platelets.

Table 1. Table I. Clearance of rHuTPO by normal human platelets.* * Platelet-rich plasma (PRP) and platelet-poor plasma (PPP) were obtained from seven normal individuals and the in vitro clearance of rHuTPO by PRP (ClPRP) and by PPP (ClPPP) determined as described in Materials and Methods. Platelet-specific clearance (ClPlt) was determined using the formula: ClPlt = ClPRP − ClPPP. Since ClPlt is directly proportional to the platelet concentration, all values were normalized for 109 platelets.† Prior to the addition of rHuTPO, endogenous TPO concentration was determined in the PPP by ELISA as described in Materials and Methods.Thumbnail image of

Comparison of pharmacokinetic data with Scatchard binding data

Since the platelet TPO receptor does not appear to recycle, the maximum binding capacity of platelets may be estimated. For 109 human platelets each of which binds 56 ± 17 TPO molecules, the predicted binding capacity is 8370 ± 2541 pg. This predicted maximum platelet binding capacity was then compared with the actual maximum amount of TPO cleared in vitro. An approximation of the actual maximum TPO clearance was determined by adding increasing amounts of TPO to a fixed amount of platelets and measuring the amount of TPO cleared in 5 h. As shown in Fig 7, with increasing TPO concentration, the total amount of TPO cleared reached a plateau value of approximately 2500 pg per 0.206 × 109 platelets, or approximately 12 500 pg per 109 platelets. This experiment has been performed with platelets from three different donors and the ratio of actual maximal platelet TPO clearance to the predicted binding capacity was an average of 0.97.

Figure 7.

. Relationship between the plasma TPO concentration and the total amount of TPO cleared by platelets. PRP (platelet count = 200 × 109/l) and PPP were incubated at 37°C in the presence of the indicated concentrations of rHuTPO and the amount of unbound TPO after 5 h was measured by ELISA. The total amount of TPO cleared over 5 h (±SD) was calculated by subtracting the TPO concentration in the PRP from that in the PPP and is plotted versus the concentration of rHuTPO added to the samples at the beginning of the incubation.

Effect of small molecules on platelet TPO clearance

It has previously been suggested (Kuter, 1996a, 1997) that several small molecules or other substances might alter the clearance of TPO by platelets. By comparing the total amount of TPO removed over 5 h of incubation, the in vitro pharmacokinetic model provided a simple way to screen for substances that might affect platelet TPO clearance. To demonstrate the feasibility of this approach, the effect of temperature on platelet TPO clearance was first studied. As shown in Fig 8, platelet TPO clearance at 4°C was only 2% of that at 37°C. However, vincristine (Jackson & Edwards, 1977), anagrelide (Anagrelide Study Group, 1992), cytochalasin B, and sodium azide had no effect on TPO clearance (Table II). Finally, when TPO was incubated with infusible platelet membrane (Scigliano et al, 1997) at a dose equal to that which is physiologically active in haemostasis, there was absolutely no clearance of TPO (data not shown), indicating the absence of c-mpl receptors in this novel investigational platelet derivative.

Figure 8.

. The effect of temperature on the clearance of TPO by platelets. As described in Fig 7, PRP (platelet concentration = 301 × 109/l) and PPP were incubated with 1500 pg/ml of rHuTPO at the indicated temperatures and the amount of unbound TPO measured after 5 h by ELISA. The total amount of TPO cleared over 5 h was calculated by subtracting the TPO concentration in the PRP from that in the PPP. The total amount of TPO (±SD) cleared at each temperature was then expressed as a percentage of the total amount of TPO cleared at 37°C (100% = 570 ± 21 pg).

Table 2. Table II. The effect of various small molecules on TPO clearance by platelets. * As described in Materials and Methods, 1500 pg of rHuTPO were added to 1 ml of PPP or PRP (platelet count = 316 × 109/l) and incubated for 5 h at 37°C with the indicated amount of vincristine, anagrelide, cytochalasin B, or sodium azide. The specimens were then centrifuged to remove the platelets and the supernatants assayed for its content of TPO. The amount of TPO (±SD) removed over the 5 h incubation was then calculated.Thumbnail image of


There are two general consequences of the binding of TPO to its receptor c-mpl. The first is the initiation of specific signalling pathways which affect cellular viability, proliferation and differentiation of not only megakaryocytes but also bone marrow precursors of other lineages and possibly the pluripotential stem cell. Activation of these signalling pathways in the enucleate platelets is associated with increased platelet reactivity but the biological significance of this effect is unclear. The second consequence of TPO binding to its receptor is that TPO is internalized and removed from the circulation. It is this mechanism which has been the subject of our studies. As initially proposed (Kuter et al, 1994), circulating TPO levels appear to be regulated primarily by the clearance of TPO after binding to c-mpl receptors on platelets and possibly megakaryocytes (Chang et al, 1996; Shivdasani et al, 1997), and not by changes in TPO production (Stoffel et al, 1996). Our studies have sought to define the critical interaction of TPO with its platelet receptor using both radiochemical and standard pharmacokinetic methods.

TPO is relatively sensitive to iodination since standard iodination conditions of 18.5 MBq 125I in the presence of 10 μg Iodogen resulted in no detectable biological activity of the radiolabelled protein despite radiochemical specific activities as high as 1.48–1.85 GBq/mg. With all of the iodinatable tyrosine residues located in the C-terminal domain of the protein, the inactivation of TPO during iodination is unlikely to be due to the binding of iodine to the protein's receptor binding domains which are located in the N-terminal region. Indeed, exposure of TPO to 10 μg Iodogen alone for just 1 min at room temperature reduced the biological activity by >99%, presumably because it oxidized some critical amino acid residue(s) in the protein. But by decreasing the amount of Iodogen to 1 μg and reducing the labelling time to 4 min, the oxidative damage was minimized and effective TPO labelling obtained with retention of most of the biological activity. This gentle method may be applicable to other sensitive proteins.

Human platelets have a single class of TPO binding sites with high affinity of around 163 pM and each platelet expresses approximately 56 binding sites (Fig 2). Since the endogenous plasma TPO concentration of approximately 1 pM is much lower than the TPO receptor binding affinity, the few occupied TPO receptors on platelets from normal individuals should not significantly affect the Scatchard analysis results. This low receptor occupancy also probably accounts for why platelets, at normal physiological concentrations of TPO, do not have significant potentiation of their activation whereas platelets exposed to 10–100 pM TPO have their threshold for agonist stimulation reduced by approximately 50% (Chen et al, 1995; Harker et al, 1996).

Our narrow range of values for the affinity and number of TPO binding sites on normal platelets is in contrast to the more variable results of Fiedler et al (1997) who analysed platelets from eight normal donors and reported a Kd of ~350 pM (range 125–814 pM) and ~100 (range 24–224) binding sites per platelet. Our results are comparable to those of Broudy et al (1997) who found that platelets from five donors had a Kd of 190 pM and 30 ± 9 (range 20–40) binding sites per platelet. Possible explanations for this discrepancy include differences in the quality of the 125I-TPO and the age of the platelets used.

Although TPO and EPO share considerable sequence homology, especially in the amino terminal 153 amino acids where they are 50% similar (Barley et al, 1994; de Sauvage et al, 1994), only TPO binds to the c-mpl receptor. This has been demonstrated by showing that unlabelled TPO but not unlabelled EPO displaces 125I-rHuTPO bound to the platelet in both receptor competition experiments (Fig 2) and cross-linking experiments (Fig 3). In the latter, we have shown that a receptor–ligand complex of 190 kD is formed upon cross-linking, identical to the predicted size of the complex formed by the 95 kD TPO with the 95 kD TPO receptor, c-mpl.

These are the first studies to demonstrate that the TPO receptor is internalized after binding TPO. This aspect of the interaction of TPO with its receptor has been analysed in two ways. First, in vitro pharmacokinetic clearance studies showed that over half of unlabelled rHuTPO was removed within 1 h. Second, using 125I-rHuTPO it has been shown that 80% of TPO specifically bound to the platelet surface could be internalized within 60 min, whereas 20% failed to be internalized even after 2 h of incubation. One possible explanation for the failure to internalize all of the binding sites is that, at the concentrations of TPO employed, there is a high ratio of ligand to receptor which may prevent dimerization and subsequent internalization. Another possible explanation is that different forms of the c-mpl receptor, the P and K forms (Skoda et al, 1993; Vigon et al, 1992; Wendling & Vainchenker, 1995), may be present on platelets. The K form lacks half of the cytoplasmic domain found in the P form and fails to initiate intracellular signalling pathways. As assessed by RT-PCR, approximately 80% of platelet c-mpl RNA was in the P form and 20% in the K form. If the K form protein is expressed, even though it may not transduce a cytoplasmic signal, it may account for the residual, non-internalized binding sites.

Following binding and internalization, TPO is degraded in the platelet in a time-dependent way as described by others (Fielder et al, 1997; Kato et al, 1997; Stefanich et al, 1997). The receptor, however, does not appear to be recycled to the surface to bind more TPO because in studies (Fig 5) looking at the uptake of 125I-rHuTPO by platelets over time, total TPO uptake reached a plateau after 1–3 h of incubation. This concept was further confirmed by TPO pharmacokinetic studies which also demonstrated nearly maximal clearance of rHuTPO after 3 h.

Receptor binding and internalization of TPO is the only apparent mechanism by which platelets remove and degrade TPO. TPO which is not bound to the platelet appears to be very stable and is not degraded by either plasma or platelet membrane-associated proteases even after 6 d of incubation. Indeed, Stefanich et al (1997) have demonstrated that recombinant mouse TPO was not degraded in the plasma of mice after a single i.v. injection.

Other circulating blood cells such as red blood cells and white blood cells do not bind or metabolize TPO (Kuter et al, 1994; Kuter & Rosenberg, 1995). Since endothelial cells were reported to synthesize c-mpl (Cardier & Dempsey, 1998), we explored the possibility that the endothelial cells which line the blood vessels contribute to TPO metabolism. No specific binding to HUVEC cells was observed using 125I-rHuTPO and no clearance was noted by pharmacokinetic analysis. Although these results suggest that the blood vessel wall may not be important in the clearance of TPO, given the significant heterogeneity of endothelial cells in the microvasculature, in vivo studies will be required to resolve this issue.

To confirm the results using 125I-rHuTPO and allow further quantitation of the interaction of TPO with its receptor, standard pharmacokinetic studies in platelet-rich plasma were done in vitro. The advantage of this approach is the use of a recombinant molecule unmodified by iodination and at clinically relevant doses. A single compartment model perfectly fit the in vitro clearance curves and allowed the measurement of the clearance of TPO by normal platelets (ClPlt). This was a metabolically active process that was temperature sensitive. Furthermore, it was a saturable process with a finite total binding capacity for TPO of approximately 12.5 ng/109 platelets. Since platelets probably neither synthesize new TPO receptor nor recycle TPO receptor, this total platelet binding capacity for TPO could be compared with that predicted by the radiolabelled TPO binding studies; the two values were virtually identical.

The presence of a pool of platelet receptors for TPO [with an estimated total binding capacity of 12.5 μg (0.18 μg/kg) for a 70 kg man with a platelet count of 200 × 109/l] is consistent with a number of observations in recent human clinical trials. First, it supports the finding that the minimal clinically effective dose of TPO is 0.3 μg/kg (Basser et al, 1997) since doses <0.18 μg/kg would simply saturate the average platelet total binding capacity. Second, it explains why with constant daily dosing of TPO, steady-state TPO levels rise daily (Roskos et al, 1997) as platelet receptors become increasingly saturated.

Since clearance of TPO is the primary determinant of circulating TPO levels, pharmacological methods to reduce platelet TPO clearance might be clinically employed to elevate TPO levels and stimulate platelet production in lieu of exogenous cytokine administration (Kuter, 1996a, b, 1997). A number of drugs with known effects on platelet production (Anagrelide Study Group, 1992; Jackson & Edwards, 1977) as well as metabolic inhibitors were therefore tested in vitro for their effect on TPO clearance. Although none were found to alter TPO clearance, this approach may ultimately allow the detection of drugs that might be used clinically to alter platelet TPO clearance.

Our studies have measured the interaction of TPO with its platelet receptor under physiological conditions and defined the binding affinity, time course of internalization, stability, and clearance rate of TPO by normal platelets. These results may help predict the adequate clinical dosing of TPO as well as provide the means to assess alterations in TPO clearance caused by diseases or drugs acting on the platelet.


This work was supported in part by NIH grant HL54838 (D.J.K.) and in part by a grant from Amgen Inc. (Thousand Oaks, Calif.). We are grateful to Drs John Romo and Chris Stowell of the Massachusetts General Hospital Blood Transfusion Service for the provision of the blood products used in this study. Dr Lorin K. Roskos (Amgen, Thousand Oaks, Calif.) provided valuable help with the pharmacokinetic analysis.