Molecular mechanisms of thrombopoietin signaling


Kenneth Kaushansky, Department of Medicine, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0671, USA.
Tel.: +858 822 3345; fax: +858 822 3344.


Summary.  The molecular pathways that regulate thrombopoiesis are becoming increasingly understood. Upon binding to its receptor, the product of the c-Mpl proto-oncogene, thrombopoietin activates a number of secondary messengers that promote cell survival, proliferation and differentiation. Amongst the best studied are the signal transducers and activators of transcription, phosphoinositol-3-kinase, and the mitogen-activated protein kinases. Additional signals activated by these secondary mediators include mammalian target of rapamycin, β-catenin, hypoxia-inducible factor 1α and the homeobox proteins HOXB4 and HOXA9, and a number that are reduced, including glycogen synthase kinase 3α and the FOXO3 family of forkhead proteins. More recently, a number of signaling pathways have been identified that turn the thrombopoietin signal off, a step necessary to avoid uncontrolled myeloproliferation, and include the phosphatases PTEN, SHP1 and SHIP1, the suppressors of cytokine signaling, and down-modulation of surface expression of c-Mpl. This review will focus on these pathways in normal and neoplastic hematopoiesis.

An adequate supply of platelets is essential to repair the minute vascular damage that frequently occurs with daily life, and to initiate thrombus formation in the event of overt vascular injury. Accumulating evidence also indicates vital roles for platelets in wound repair, the innate immune response and metastatic tumor cell biology. The average platelet count in humans ranges from 150–400 × 109 L−1, although the level for any individual is maintained within fairly narrow limits from day to day. While 150–400 × 109 L−1 is considered ‘normal’, values derived from the mean ± 2 SD of a group of ‘healthy’ individuals, epidemiologic evidence indicates that individuals that display a platelet count in the highest quartile of the normal range have a 2-fold increased risk of adverse cardiovascular events, and in both experimental animal models of metastatic cancer and in patients with tumors higher platelet levels carry an unfavorable prognosis. Hence, the mechanisms that regulate the production of platelets are of keen interest for both health and disease.

The primary regulator of platelet production is thrombopoietin, an acidic glycoprotein produced in many organs but primarily in the liver, kidney and bone marrow. The biochemistry and structure–activity relationships of thrombopoietin have been carefully evaluated, as have the binding sites to its receptor, the product of the cellular proto-oncogene, c-Mpl. This review will focus on the molecular mechanisms through which the hormone acts to stimulate hematopoiesis.

The type I hematopoietic growth factor (HGF) receptor family, of which c-Mpl is a member, consists of more than 20 molecules that bear one or two cytokine receptor motifs (CRMs), an approximately 200 amino acid module containing four spatially conserved Cys residues, 14 β-sheets, and a juxtamembrane Trp-Ser-Xaa-Trp-Ser sequence [1]. In addition to the CRM(s), type I receptors contain a 20–25 residue transmembrane domain and a 70–500 amino acid intracellular domain containing short sequences that bind intracellular kinases and other signal-transducing molecules. The thrombopoietin receptor is expressed primarily in hematopoietic tissues, specifically in megakaryocytes, their precursors (e.g. hematopoietic stem cells) and their progeny (platelets). For the most part, c-Mpl is constitutively expressed in these tissues, although receptor display is modulated by thrombopoietin binding and receptor internalization. Upon binding cognate ligand, hematopoietic cytokine receptors such as c-Mpl are activated to transmit numerous biochemical signals. The molecular details that initiate this process are understood, based on studies of the erythropoietin and growth hormone receptors. These receptors exist in a homodimeric state in the absence of ligand, in a conformation that separates cytoplasmic domains (e.g. 73 Å in the erythropoietin receptor [2]). Upon ligand-binding, receptor conformation shifts, bringing the erythropoietin receptor cytoplasmic domains to within 39 Å of one another. Additional studies indicate that the membrane-proximal box1 and box2 cytoplasmic domains constitutively bind JAK family kinases, even in an inactive state. Upon ligand-binding, the closer juxtaposition of the two tethered kinases is thought to allow their cross-activation, initiating signal transduction.

Once JAK kinases are active, their targets include (i) tyrosine residues within the receptor itself, creating phosphotyrosine (P-Y) docking sites for signaling molecules that contain Shc homology (SH)2 or phosphotyrosine-binding (PTB) motifs [1], (ii) molecules bound to the docking sites that promote cell survival and proliferation, including the signal transducers and activators of transcription (STATs) [3], phosphoinositol-3-kinase (PI3K) [4]; and the mitogen-activated protein kinases (MAPKs), and (iii) and molecules bound to the receptor docking sites that limit cell signaling, including phosphatases such as SHP1 [5], SHIP1 [6] and PTEN [7] and suppressors of cytokine signaling (SOCS) [8]. Moreover, ligand engagement leads to receptor internalization and either recycling to the cell surface, or destruction [9], which also affect signaling. These processes are vital for the proper regulation of hematopoiesis, as alterations in many of these mediators result in pathologic marrow failure or myeloproliferation.

Signaling mediators that promote cell survival, cell cycle progression and differentiation

The signaling pathways that promote cell survival, proliferation and differentiation in megakaryopoiesis have been widely studied, at least the ones common to a number of other microenvironmental signals (cytokines, integrins, mechanical stimuli of cells). Activation of the MAPK pathway is required for maturation of megakaryocytic progenitor cells and the generation of highly polyploid cells [10]. One MAPK-dependent pathway that contributes to megakaryocyte differentiation is its activation of RUNX1 (also termed AML1) by post-translational modification [11], which then induces expression of p19 INK4 that leads to endomitotic arrest and megakaryocyte maturation [12]. Phosphorylation of AKT by PI3K controls cell cycle progression and survival of these cells [13] through silencing of FOXO family of transcription factor [14], which left unchecked would stimulate expression of the cell cycle inhibitor p27 and the proapoptotic molecule fas ligand. The STAT family of transcription factors is also indispensable for the normal development of megakaryocytes, as Stat5a- and Stat5b-deficient mice show impaired platelet production [15]. Moreover, transgenic mice with megakaryocytic lineage-specific overexpression of a dominant negative form of STAT3 display reduced platelet recovery following 5-FU induced myelosuppression [16]. One target of HGF-induced STAT5 that mediates these effects on cell survival is the anti-apoptotic molecule BclXL [17]. The STATs are also affected by additional Mpl-related adaptor molecules. For example, overexpression of LNK, a known inhibitor of cytokine signaling, inhibits thrombopoietin (TPO)-induced STAT5 and MAPK activation in 32D-mpl cells, introduction of LNK into bone marrow Lin- cells reduced TPO-dependent growth and ploidy of megakaryocytes, and increased megakaryocyte ploidy was found in LNK-deficient mice [18,19].

A number of studies have also demonstrated an important role for TPO in hematopoietic stem cell (HSC) biology. For example, TPO supports cell survival in cultures of highly purified murine HSCs [20]. The molecular signals that support this function include induction of HoxB4 expression in a p38 MAPK- and USF1/2-dependent fashion [21], nuclear translocation of HoxA9 expression secondary to induction of its heterodimeric partner, MEIS1 [22], and hypoxia-inducible factor (HIF)1α-induced autocrine production of vascular endothelial growth factor [23].

Molecular signals that extinguish TPO-signaling

Once a cell is stimulated by a HGF, the induced signals must be extinguished, lest uncontrolled proliferation ensue. At least three mechanisms down-modulate the sensitivity of stimulated cells to further HGF signaling: (i) induction of SOCS proteins; (ii) activation of phosphatases that remove P-Tyr sites, and (iii) a number of adaptor proteins that negatively regulate signaling. Removal of the activated HGFR from the cell surface by endocytosis will be discussed below. SOCS proteins are induced by STAT-mediated transcription, and once translated bind to P-Y residues in the HGF receptor and activated JAK kinases, precluding binding of additional signaling molecules and triggering proteolytic destruction [8]. Hematopoietic cells contain a number of phosphatases that eliminate P-Y from receptors and signaling adaptors (e.g. SHP1); their physiologic importance is illustrated by disorders of macrophage activation or profound erythropoiesis. Recent evidence suggests that the dual function phosphatase, PTEN, is also important for hematopoiesis. Moreover, TPO has been shown to induce the expression of SOCS1 and SOCS3 [24], implicating these STAT – induced proteins in the regulation of TPO-signaling. Finally, several proteins that bind to Mpl, either directly or indirectly, initiate signals that dampen the proliferative signals that emanate from c-Mpl. In addition to LNK, mentioned above, the Src family kinase Lyn down-modulates TPO-induced proliferation [25], likely acting downstream of the focal adhesion kinase, FAK [26].

Molecular mechanisms that regulate c-Mpl receptor display

A great deal is now understood of what governs c-Mpl receptor display on the cell surface, both its trafficking to the cell surface following biosynthesis, and its removal from the cell surface following TPO-binding, processes vital for normal regulation of hematopoiesis.

Using a BaF3 cell model and immunoblotting we found that c-Mpl displays two distinct electrophoretic mobilities; a more slowly migrating band that represents mature, N-glycosylated receptor and a smaller band representing an immature, hypoglycosylated form. Using three distinct experimental approaches, we found that only N-glycosylated c-Mpl is expressed on the cell surface and participates in signaling. For example, cells labeled with a cell-impermeable form of biotin display biotin only on the more slowly migrating, higher Mr form of c-Mpl. There are four sites of potential N-linked glycosylation of c-Mpl, N117, N178, N298 and N358. By expressing a series of N to Q mutants of the receptor, we determined which N residues bear carbohydrate modification; we found that N117Q displayed normal electrophoretic mobility and cell surface expression, but mutation at any combination of the other three sites altered the electrophoretic mobility and cell surface expression of the receptor.

Alterations in glycosylation of c-Mpl appears to play an important role in two distinct disorders of hematopoiesis, polycythemia vera, where c-Mpl is underglycosylated and surface display of the receptor is reduced [27], and in the most common form of congenital amegakaryocytic thrombocytopenia, c-Mpl R102P, which causes reduced glycosylation, obviating access of the receptor to the hormone [28].

Once expressed on the cell surface and exposed to TPO, c-Mpl is rapidly removed from the cell membrane, in a clathrin-mediated process [29]. The molecular components of clathrin coated pit removal of surface proteins are clathrin triskeletons and adaptor protein (AP)2 complexes, the latter a heterotetramer composed of 100–115 kDa α and β adaptins, a 50 kDa μ2 and a 17 kDa δ2 subunit [30]. Target proteins for clathrin-mediated endocytosis bear recognition sequences NPXY or YXXθ (θ = bulky hydrophobic), LL, or an acidic cluster; of note, human c-Mpl bears two YRRL sequences (Y521RRL and Y591RRL), highly conserved in the murine receptor. Previously, we showed that an Y591F mutant displays enhanced TPO-signaling [31]. Moreover, reduced receptor internalization and enhanced signaling could explain why truncation of c-Mpl beyond S574 [32] or L582 [31] signals so well, despite elimination of the P-Y residues that activate STATs and MAPKs. We found that cell surface clearance of c-Mpl is greatly diminished when Y591 is mutated to F, an effect associated with intense and prolonged signaling, and that Y591 is part of the clathrin/AP2 complex recognition site. Once internalized, pathways that either promote recycling or degradation of the receptor may regulate the duration of signaling. In that regard, we have found that Y521 is responsible for the trafficking of internalized c-Mpl to the lysosome, as its mutation to F allows enhanced recycling of the internalized receptor to the cell surface [29]. c-Mpl contains two intracellular lysines that are potential targets for ubiquitination (K553, K573) and hence might mediate its degradation through the proteasome. In addition, acquired mutations of the c-Cbl ubiquitin ligase lead to impaired ubiquitination and over-expression of FLT3 in myeloid leukemias [33]; using in vitro signaling systems with mutant forms of the c-Mpl receptor we have now shown that c-Cbl is an E3 ligase for the ubiquitinated c-Mpl receptor and is responsible for receptor signaling intensity.


Our current understanding of c-Mpl- and TPO-signaling is evolving; we know that JAK2, STAT3 and STAT5, PI3K and Akt and p38, p42/p44 MAPKs are involved in the primary signals that emanate from c-Mpl, and that HOXB4, HOXA9, RUNX-1, HIF-1α are secondarily affected, and enhance cell survival, proliferation and differentiation. We also know that a number of phosphatases, SOCS proteins and receptor internalization down modulate TPO-signaling once the hormone binds to the receptor. While this pattern of TPO-signaling is rather complex, it was built almost entirely on a candidate gene approach to identifying the signaling components of c-Mpl. One might argue that an unbiased approach to signal mediator identification could yield important new pathways, studies of which are currently underway in our laboratory.

Disclosure of Conflict of Interests

The author receives funding from the National Institutes of Health and is on the scientific advisory board of Ligand Pharmaceuticals.