Pronounced thrombocytosis in transgenic mice expressing reduced levels of Mpl in platelets and terminally differentiated megakaryocytes

Thrombopoietin (Tpo) and its receptor Mpl are the principal regulators of megakaryopoiesis.^1,2  Mice deficient in Tpo or Mpl continue to produce functional platelets, albeit at much lower levels,^3,4  suggesting that Mpl mainly controls quantitative aspects of thrombopoiesis. Tpo serum levels are controlled by the platelet mass through Mpl-mediated Tpo uptake and degradation.^5,6  Consequently, Mpl^−/− mice show increased Tpo levels.³  Although an important function of Mpl is to regulate platelet numbers, it is also expressed on hematopoietic stem cells (HSCs) and early progenitors.^7,8  Consistently, Mpl-deficient mice show markedly decreased numbers of hematopoietic progenitors, and competitive repopulation assays indicate that the numbers of murine HSCs is reduced by 7- to 8-fold.^7,8  In humans, loss-of-function mutations in Mpl lead to congenital amegakaryocytic thrombocytopenia, a disorder that frequently leads to bone marrow failure.^9-11  The reason for the more severe phenotype in humans remains unknown.

Mutant versions of Mpl can lead to uncontrolled proliferation and survival signals as exemplified by the retroviral fusion oncogene v-Mpl, which can immortalize hematopoietic progenitors.¹²  An autosomal-dominant point mutation in the transmembrane domain of Mpl (S505N) was identified as the cause of thrombocytosis in families with hereditary thrombocytosis.^13,14  Recently, point mutations in the cytoplasmic domain of Mpl (W515L, W515K) were identified in patients with primary myelofibrosis and essential thrombocythemia, and W515L was shown in mouse models to elicit myeloproliferative disease (MPD) with marked thrombocytosis.^15,16  These findings underline that Mpl can trigger potent growth signals, and stringent mechanisms of signal attenuation are essential for ensuring an appropriate, controlled cellular response.

To study the effects of Mpl variants carrying mutations in the cytoplasmic domain that are expected to alter signaling, we used a transgene rescue strategy. The cDNAs encoding Mpl variants were placed under the control of a previously characterized Mpl promoter,¹⁷  and transgenic mice were generated on the Mpl^−/− background.³  Unexpectedly, transgenic mice expressing the wild-type Mpl cDNA that was designed as a control displayed marked thrombocytosis, with 5-fold elevated platelets when examined on the Mpl^−/− background. Since the platelet numbers were normal when the transgene was examined on the heterozygous Mpl^−/+ background, we investigated the possibility that thrombocytosis was due to loss of an alternatively spliced Mpl mRNA. Expression of alternative cytokine receptor variants, which act as dominant-negative isoforms, can serve as a mechanism of modulating the responses to cytokines.^18-20  Mechanistically, dominant-negative receptor variants can form nonfunctional heterodimers with the full-length receptor or, when expressed as secreted “soluble” isoforms, compete with the receptor for ligand binding.²¹  We have described a third mechanism, in which overexpression of the Mpl isoform Mpl-tr triggers protein degradation of the full-length receptor.²²  Mpl-tr is the only splice variant found both in human and mouse. This variant results from splicing of exon 8 directly to exon 11, eliminating the juxtamembrane extracellular part and the transmembrane domain.^23,24  However, our genetic analysis excluded loss of Mpl-tr as the cause of thrombocytosis in our transgene rescue mice. Instead, decreased expression of Mpl protein on platelets appears to be the cause of thrombocytosis through a dysbalance in the autoregulatory mechanism controlling Tpo. This mouse model recapitulates disease manifestations frequently associated with human MPDs, including thrombocytosis and a reduction of Mpl protein expression in platelets.

We used a 2-kb HindIII-Apa I mouse Mpl promoter fragment,¹⁷  which was cloned in front of a mouse Mpl cDNA and an SV40 polyadenylation signal. The resulting 5-kb fragment, which contains no vector sequences, was used for oocyte microinjection. We generated 5 transgenic founder mice in the B6D2Fn background. This study was approved by the institutional review boards of all participating institutions. All mice used in this study were kept under specific pathogen–free conditions and in accordance to Swiss and German federal regulations.

Animals were killed with CO₂, and blood was obtained by cardiac puncture and mixed with EDTA. Blood counts were determined on an Advia 120 Hematology Analyzer (Bayer, Leverkusen, Germany). ELISA was performed with the Quantikine Mouse Tpo Immunoassay kit (R&D Systems, Abingdon, United Kingdom). For histology, freshly dissected tissues were fixed in Optimal*Fix (AMTS, Lodi, CA). Fixed specimens were embedded in paraffin, sectioned, and stained in the Transgenic Pathology Laboratory (University of California, Davis, CA). Megakaryocyte frequency was determined by 2-color flow cytometry as described.²⁵  In brief, bone marrow cells from femurs and tibias were isolated in CATCH (129 nM NaCl, 8.614 nM Na₂HPO₄, 1.6 mM KH₂PO₄, 13.6 mM sodium citrate, 11.1 mM glucose, 1 mM adenosine, 2 mM theophilline, and 1% bovine serum albumin [BSA]; pH 7.0). Cells were filtered through a 100-μm nylon mesh, and megakaryocytes were stained using the monoclonal 4A5 rat anti–mouse platelet antibody and a fluoroisothiocyanate (FITC)–goat anti–rat IgG F(ab′)2 antibody (Tago, Burlingame, CA). Samples for electron microscopy were processed essentially as described.²⁶  In brief, spleen tissue was diced into pieces of roughly 1 mm², fixed for 1 hour in 3% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), treated with osmium tetroxide, washed, and embedded in Epon (Fluka Chemie, Buchs, Switzerland). Ultrathin sections were collected on grids and examined on a Philips Morgani 268D electron microscope (FEI, Hillsboro, OR) and images were analyzed with Photoshop (Adobe, San Jose, CA).

For platelet isolation, blood was drawn by cardiac puncture into a syringe containing 3.8% citrate. Platelet-rich plasma was obtained by centrifugation at 350g for 10 minutes, and pure platelets were sedimented by centrifugation at 1300g, washed once with phosphate-buffered saline (PBS), and analyzed for purity on an Advia 120 Hematology Analyzer. For megakaryocyte isolation, femurs were flushed with CATCH medium. A total of 4 mL of a single-cell suspension was then mixed with 3 mL Percoll/PBS (1.02 g/mL; GE Healthcare, Chalfont St Giles, United Kingdom), gently layered on top of 4 mL Percoll/PBS (1.05 g/mL) and centrifuged for 20 minutes at 400g. The interface was then collected and washed with CATCH buffer. Alternatively, megakaryocytes were isolated using anti-CD41 antibodies (BD Pharmingen, Franklin Lakes, NJ) and Dynabeads conjugated with anti–rat IgG (Invitrogen, Carlsbad, CA). The purity of the megakaryocyte preparation was assessed by acetylcholine esterase staining on cytospin slides.^27,28  For protein extraction, isolated platelets and megakaryocytes were lyzed by boiling in 2× SDS-PAGE loading buffer for 5 minutes at a concentration of 2× 10⁶ platelets/μL and 50 000 megakaryocytes/μL. Protein lysates were size-separated by SDS-PAGE and transferred to nitrocellulose. The membranes were probed with a biotinylated mouse monoclonal anti-Mpl antibody and reprobed with a biotinylated 4A5 rat anti–mouse glycoprotein V monoclonal antibody (clone 4A5).^8,29  Signals were detected using horseradish peroxidase–coupled streptavidin and the enhanced chemiluminescence (ECL) system (Amersham Biosciences, Piscataway, NJ).

Total RNA was isolated with TriFast (PeqLab, Erlangen, Germany) and treated with Turbo DNAse (Ambion, Austin, TX). Reverse transcription was performed with Omniscript reverse transcriptase (QIAGEN, Hilden, Germany), and control reactions without reverse transcriptase were included. Quantitative real-time polymerase chain reaction (qPCR) was performed using SYBR Green PCR master mix on an ABI Prism 7000 (Applied Biosystems, Foster City, CA) with the following primers: CTGTATGCCTACCGAGGAGAGAAG and GTTCCAAAGGTGGGCACACT for total mouse Mpl; TATTGGCAGCAGCCCTGAA and TGGATGGTGTTGAGGATGGATA for endogenous mouse Mpl (reverse primer located in 3′ untranslated region [UTR]); CAGCCCTGACTCGAATCGAA and TGCAGGAATTGACGGTATCG for transgenic Mpl (partially located in the SV40 polyadenylation signal); GGGCCTTCCTTCGGGAT and AGCACAATTGGGCTCAGCTT for mouse Itga2b (CD41); and ATCCGCAAGCCTGTGACTGT and TCGGGCCAGGGTGTTTTT for mouse Rpl19.

The chromosomal integration site of the Yall transgene was determined by thermal asymmetric interlaced PCR.³⁰  The gene-specific primers were GGAGCCTGGTGTAATAGCTCAC, AATCCCAGAACTTGGGAGAGAG, and GACCAGCCTGTTCTACAAGCAA.

Clonal cultures of hematopoietic cells were performed as described.³¹  Colonies containing at least 3 acetylcholinesterase-positive cells were scored as colony-forming unit–megakaryocyte (CFU-MK).³²  For megakaryocyte cultures, femurs and tibias were flushed with CATCH medium, and cells were subjected to lineage depletion (Mouse Hematopoietic Cell Lineage Depletion Kit; R&D Systems), leading to a relative enrichment of megakaryocytes. RNA was prepared from cells immediately or after 2 or 4 days of culture in RPMI with 10% fetal calf serum (FCS), Tpo-conditioned media,⁶  and 10 μg/mL recombinant murine stem cell factor (BioSource, Camarillo, CA). For ploidy sorts, lineage-depleted bone marrow cells were stained in PBS with 0.1% BSA and 1 mM EDTA with a FITC-conjugated anti-CD41 antibody (Becton Dickinson) for 30 minutes at 4°C. After washing, cells were incubated for 2 hours at 37°C with 6.25 μg/mL Hoechst 33342 (Invitrogen) in RPMI without phenol red containing 0.1% BSA and 5 mM EDTA. CD41⁺ cells were sorted into 2 fractions (2n/4n/8n and 16n/32n/64n) on a FACSVantage cell sorter (Becton Dickinson).

We generated transgenic mice expressing the full-length mouse Mpl cDNA under the control of a 2-kb Mpl promoter fragment (Figure 1A), which we previously found to direct expression specifically to the megakaryocytic lineage.¹⁷  We obtained 5 founder mice and detected Mpl transgene expression in 3 of them (data not shown). These mice were viable, fertile, and showed no overt abnormalities. To eliminate the expression of the endogenous Mpl gene, we bred the founders into the Mpl knockout (Mpl^−/−) genetic background.³  All 3 Mpl transgenic lines that expressed the transgene showed marked thrombocytosis in an Mpl^−/− background with platelet counts 3- to 6-fold higher than the normal range, whereas a nonexpressing line showed thrombocytopenia similar to Mpl^−/− mice (Table 1).³  One line, designated Yall (for all tyrosines [Y] of the cytoplasmic domain), was chosen for detailed analysis.

The expression of the Yall transgene and endogenous Mpl in mouse organs was assessed by qPCR in transgenic mice heterozygous for Mpl knockout (Yall;Mpl^−/+, Figure 1B). The distinction between transgenic and endogenous Mpl mRNA was achieved by placing the reverse primer of each qPCR primer pair in the respective 3′ UTRs. The transgene mRNA was expressed in bone marrow and brain, whereas the endogenous Mpl mRNA was found in bone marrow and testes. Megakaryocytes purified by anti-CD41 magnetic beads showed the highest expression of transgenic and endogenous Mpl mRNA. Comparable levels of Mpl protein were detected by immunoblot using anti-Mpl antibodies in purified megakaryocytes from Yall;Mpl^−/− mice and wild-type controls, and—as expected—Mpl protein was absent in megakaryocytes from Mpl^−/− mice (Figure 1C).

Similar to the increase in platelets (Figure 2A), the percentage of megakaryocytes in the bone marrow of Yall;Mpl^−/− mice was elevated as determined by flow cytometry (Figure 2B). To assess whether enhanced polyploidization could also be a contributing factor for the thrombocytosis, we determined ploidy of bone marrow megakaryocytes by flow cytometry (Figure 2C). Yall;Mpl^−/− mice showed a slight shift toward higher ploidy, whereas a shift toward lower ploidy was seen in Mpl^−/− mice. We also observed a marked increase in the number of megakaryocytic precursors (CFU-MK) in Yall;Mpl^−/− mice, as determined by colony assays (Figure 2D). Thus, Mpl transgene expression in Mpl^−/− mice resulted in an expansion of the megakaryocytic lineage at all stages of differentiation. In addition, we found that the frequencies of other hematopoietic progenitor types were also elevated in Yall;Mpl^−/− mice (Figure 2E). This finding is in agreement with the role of Mpl in the maintenance of hematopoietic progenitors.^7,8,33 

Histopathologic analysis revealed a marked increase in megakaryocyte numbers in the bone marrow and spleen of Yall;Mpl^−/− mice (Figure 3A). Furthermore, spleen sections of Yall;Mpl^−/− but not wild-type mice contained aggregates of small particles (Figure 3A brown color). Such aggregates were also visible at a lower frequency in bone marrow. To identify the nature of the particles, electron microscopic analysis of spleen sections was performed. We detected areas containing densely packed platelets in Yall;Mpl^−/− mice, but not in wild-type controls (Figure 3B). Platelets were identified by size, morphology, and the presence of dense granules.^34,35 

We have recently shown that a splice variant of Mpl, called Mpl-tr, acts as a dominant-negative regulator of Mpl function in vitro.^22,36  Because the Mpl knockout allele carries a neomycin cassette inserted in Mpl exon 3, the Mpl^−/− mice cannot express functional Mpl-tr mRNA, which contains exon 3. We therefore hypothesized that the increased megakaryopoiesis and thrombopoiesis could be caused by the absence of the Mpl-tr transcript in Yall;Mpl^−/− mice. Consistently, the Yall transgene on the Mpl^−/+ background displayed platelet numbers similar to their nontransgenic littermates, indicating that a single copy of the wild-type Mpl gene is dominant over the transgene in respect to platelet production (Figure 4A). Similarly, comparable numbers of CFU-MKs were found in the bone marrow from transgenic and nontransgenic Mpl^−/+ mice (data not shown). The Yall transgene on the Mpl^+/+ background showed normal platelet numbers, and the heterozygous Mpl^−/+ mice showed slightly higher platelet counts than Mpl^+/+ mice independent of the presence or absence of the transgene (Figure 4A). To more directly test the hypothesis that the absence of Mpl-tr is the cause of thrombocytosis in Yall;Mpl^−/− mice, we took advantage of an Mpl knockin allele named Mpl^Δ60.³⁷  This allele was generated by homologous recombination replacing exons 11 and 12, which encode the cytoplasmic signaling domain of Mpl, with a cDNA fragment encoding a truncated cytoplasmic domain that lacks the last 60 amino acids (Figure 4B). Mpl^Δ60/Δ60 mice have normal platelet counts, demonstrating that the last 60 C-terminal amino acids of Mpl are not essential for steady-state thrombopoiesis in vivo.³⁷  Mpl-tr cannot be produced by the Mpl^Δ60 allele, because the splice acceptor for the generation of Mpl-tr is absent (Figure 4B). We confirmed the absence of Mpl-tr in Mpl^Δ60/Δ60 mice by 3′ rapid amplification of cDNA ends (RACE; data not shown). We then crossed Yall;Mpl^−/− mice with Mpl^Δ60/Δ60 mice to obtain Yall;Mpl^−/Δ60 mice, expecting that these mice would also show thrombocytosis. However, platelet levels in these mice were normalized to levels comparable with Yall;Mpl^−/+ or Yall;Mpl^+/+ mice (Figure 4A). Thus, the presence of the Mpl^Δ60 allele was dominant over the transgene in respect to platelet production despite the fact that Mpl-tr cannot be expressed. These results exclude the hypothesis that the absence of Mpl-tr is the primary cause of thrombocytosis in Yall;Mpl^−/− mice.

To determine whether the function of Mpl protein in platelets as a negative regulator of Tpo serum levels was disrupted in Yall;Mpl^−/− mice, we compared the expression of Mpl in platelets and megakaryocytes from Yall;Mpl^−/−, Mpl^Δ60/Δ60, and wild-type mice (Figure 5A). Mpl mRNA expression in platelets was strongly reduced in Yall;Mpl^−/− mice compared with wild-type and Mpl^Δ60/Δ60 mice (Figure 5A top panel), which translated into marked differences in protein levels as determined by Western blot analysis (Figure 5A bottom panel). In contrast, no differences in Mpl mRNA or protein expression were observed in purified megakaryocytes from the different genotypes. Thus, Mpl mRNA and protein encoded by the Yall transgene was dramatically reduced in platelets, but not in megakaryocytes, whereas Mpl expression driven by the endogenous Mpl promoter in wild-type and Mpl^Δ60 mice was comparable in both platelets and megakaryocytes. Based on these findings, a possible cause for the observed thrombocytosis in Yall;Mpl^−/− mice could be the decreased expression of Mpl in platelets and thus a reduced capacity of platelets to absorb circulating Tpo protein.

In Mpl^−/− mice, which lack the Tpo receptor and have low platelet counts, Tpo levels are substantially increased.³  Platelets in Yall;Mpl^−/− mice still express Mpl protein, albeit at much lower levels than wild-type mice. We therefore asked whether the reduction in Mpl protein in platelets leads to increased Tpo levels and determined Tpo by enzyme-linked immunosorbent assay (ELISA) in both serum and plasma. Tpo in serum and plasma from Yall;Mpl^−/− mice was not elevated, but rather slightly diminished compared with the Tpo levels detected in Yall;Mpl^−/+ control mice (Figure 5B). As shown previously, Mpl^−/− mice displayed elevated Tpo levels. Thus, reduction in Mpl protein expression in platelets did not result in elevated Tpo serum levels in Yall;Mpl^−/− mice. Because megakaryocytes have been shown to also have an effect on the circulating Tpo levels,^38,39  it seems plausible that the increase in megakaryocyte mass in bone marrow may compensate for the decrease in Mpl protein expression in platelets in Yall;Mpl^−/− mice. Thrombocytosis could also be the consequence of increased half-life of platelets in Yall;Mpl^−/− mice. To address this possibility, we marked blood cells in wild-type and Yall;Mpl^−/− mice with biotin and determined the fraction of labeled platelets over time by flow cytometry.⁴⁰  No significant difference in half-life was observed (Figure 5C).

To explain the reduction of Yall transgene expression in platelets, we considered 2 possibilities. First, the transgenic mRNA could be less stable than the endogenous Mpl mRNA. Since platelets do not contain nuclei and thus no longer transcribe mRNA, a faster decay of the transgene message could result in strongly reduced Mpl expression levels in platelets. Second, the 2-kb Mpl promoter used in the Yall transgenic mice could be transcriptionally less active in late stages of megakaryocyte maturation than the endogenous Mpl promoter.

The stability of many mRNAs is regulated by sequence elements in the 3′ UTR. The Yall transgene was constructed with a polyadenylation signal derived from the SV40 virus that is different from the polyadenylation signal of endogenous Mpl mRNA. However, the Mpl^Δ60 knockin allele, which results in normal Mpl expression level in platelets, also features an SV40 polyadenylation signal. Since different variants of the SV40 polyadenylation signal were used in the Yall and the Mpl^Δ60 constructs, we investigated whether this difference affects the half-life of the respective mRNAs. We constructed 2 Mpl expression plasmids differing only in their polyadenylation signal, one corresponding to Yall and the other to Mpl^Δ60. Stable transfectants were obtained in murine hematopoietic BaF3 cells, and the half-life of Mpl mRNA was measured after blocking transcription with actinomycin D. We observed no difference in the mRNA half-life of the 2 constructs (data not shown), suggesting that differences in mRNA stability are not responsible for the loss of transgene expression in platelets of Yall;Mpl^−/− mice.

Alternatively, expression of Mpl mRNA in our Yall transgene could be reduced at late but not early stages of megakaryocytic differentiation (eg, due to a lack of regulatory elements in the 2-kb promoter construct or due to effects related to transgene insertion). The Yall transgene was mapped to mouse chromosome 9, band D, between the genes Grinl1a, encoding a glutamate receptor, and Lipc, encoding hepatic lipase c, neither of which have any known function in megakaryopoiesis. Since the Mpl^Δ60 allele has been generated by homologous recombination into the endogenous Mpl locus, transcriptional regulation of Mpl^Δ60 is expected to be the same as for endogenous Mpl. To compare the expression of transgenic and endogenous Mpl mRNA during megakaryocyte maturation, we cultured megakaryocyte-enriched bone marrow from Yall;Mpl^−/+ mice in the presence of Tpo and determined the mRNAs of Yall and endogenous Mpl on days 0, 2, and 4 of culture by qPCR with primers that are specific for the respective mRNAs (Figure 6A).

Values were normalized to the mRNA of the megakaryocyte-specific marker CD41. We found that expression of endogenous Mpl slightly decreased after 2 and 4 days in culture, whereas the levels of Yall transgene mRNA dropped dramatically during the same period of time (Figure 6A). To show that this decrease correlates with megakaryocyte maturation, we sorted bone marrow–derived megakaryocytes according to increasing ploidy using a fluorescence-activated cell sorter (FACS). Megakaryocytes with a DNA content of 2n, 4n, and 8n were pooled (R1) and compared with a pool of megakaryocytes with a DNA content of 16n or higher (R2). The expression levels of the Yall transgene were decreased in megakaryocytes with higher ploidy, whereas the mRNA of endogenous Mpl was essentially equal (Figure 6B). We propose that such dysregulated Mpl expression during late megakaryopoiesis is causing thrombocytosis in Yall;Mpl^−/− mice through decreased Tpo clearance by platelets in peripheral blood.

We generated mice expressing a wild-type Mpl cDNA transgene in the Mpl^−/− background and surprisingly observed a marked thrombocytosis with approximately 5-fold higher platelet counts than normal. Initially, these experiments were designed to study Mpl transgenes with tyrosine-to-phenylalanine mutations in the intracellular domain, and the transgenic mice with the wild-type Mpl cDNA served as the positive control. At the same time, this transgenic model also allowed us to assess the in vivo role of alternatively spliced Mpl isoforms that are disrupted in the knockout allele and missing from the transgene. Here, we described in detail 1 of 3 transgenic strains that express the wild-type Mpl cDNA. Another team independently generated very similar Mpl transgenic mice in which the same 2-kb Mpl promoter was used, but was crossed with a different Mpl knockout strain.⁷  Interestingly, these mice also displayed thrombocytosis, indicating that the phenotype is independent of the Mpl knockout strain used (B. Lannutti and N. Josephson, written personal communication, March 2008).

We initially pursued the hypothesis that the reason for expansion of the megakaryocytic lineage in Yall;Mpl^−/− mice could be the absence of Mpl-tr, an Mpl splice form which we have previously characterized as a negative regulator of Mpl in cell lines.²²  However, we found that the Mpl^Δ60 knockin allele, which cannot generate Mpl-tr, was nevertheless able to suppress thrombocytosis in Yall;Mpl^−/Δ60 mice (Figure 4), which proves that loss of Mpl-tr cannot be the cause of thrombocytosis in Yall;Mpl^−/− mice. In humans, 2 alternate mRNA Mpl species are known in addition to Mpl-tr. The Mpl-K variant is due to a read-through beyond the exon 10 splice donor site.⁴¹  The resulting K-form of Mpl diverges from the native sequence after the ninth cytoplasmic amino acid and terminates after 57 codons within intron 10. A second isoform, Mpl-del, arises as a consequence of alternative splicing between exons 8 and 9 and encodes a protein of unknown function with an in-frame deletion of 24 amino acids.⁴²  There is no evidence for a negative regulatory role of Mpl-K or Mpl-del proteins, and these human Mpl isoforms have not been found in the mouse.

Analysis of the expression pattern of transgenic and endogenous Mpl revealed a likely cause for the phenotype of Yall;Mpl^−/− mice: terminally differentiated megakaryocytes and platelets showed strongly reduced transgene mRNA expression, while in unfractionated megakaryocytes, the expression of transgenic Mpl mRNA and protein were not grossly different from endogenous Mpl of wild-type mice (Figures 5,6). Importantly, Yall-transgenic mice on a heterozygous Mpl^−/+ background displayed almost normal platelet counts (Figure 2), and at the same time showed Mpl protein expression in platelets at levels similar to nontransgenic Mpl^−/+ control mice (Figure 5). These findings are compatible with a model in which decreased expression of Mpl protein in late megakaryocytes and platelets concurrent with persistent Mpl expression on early megakaryocytes and progenitors is the cause for thrombocytosis in Yall;Mpl^−/− mice (Figure 7). Interestingly, heterozygous Mpl^−/+ mice consistently showed a slightly higher platelet count than wild-type littermates^3,7  (Figure 4). According to our model, the subtle phenotype of heterozygous Mpl^−/+ mice may be explained by a moderate decrease in Mpl expression in platelets. A trend toward a higher production of platelets in vitro has been described in Mpl^−/− embryonic stem (ES) cells that were rescued with a full-length Mpl cDNA driven by the hEF-1alpha promoter.⁴³  Since these experiments were conducted in vitro, we suspect that the mechanism of slightly increased platelet production in the ES cell experiments is different from the mechanism in our transgenic mice.

It has been shown that Tpo serum levels are controlled by the platelet mass through Mpl-mediated Tpo uptake and degradation.^5,6  Consequently, Mpl^−/− mice showed increased Tpo levels³  (Figure 5B). In Yall;Mpl^−/− mice, we found that Tpo levels were normal or slightly reduced (Figure 5). This could be explained by 2 factors: first, platelets make up for decreased Mpl expression by their increased numbers; second, the mass of megakaryocytes is increased and contributes to decreasing the Tpo levels. Megakaryocyte mass was previously shown to be an additional factor in the regulation of Tpo levels, as knockout mice lacking the transcription factor NF-E2, which are profoundly thrombocytopenic but have an increased mass of immature megakaryocytes, display normal Tpo levels.^38,39  Similarly, in patients with platelet destructive disorders where megakaryocyte mass is increased, Tpo levels were found to be low.⁴⁴  To reconcile the normal Tpo levels with our model, we suggest that the loss of the negative regulatory feedback normally exerted by platelets leads to an increased production of megakaryocytes and platelets in Yall;Mpl^−/− mice. As these platelets lack the capability to provide a negative feedback, this expansion is expected to continue until the total amount of Mpl protein, contributed predominantly by an increase in megakaryocyte numbers, is able to consume Tpo at a normal rate. A new equilibrium is reached at a higher total number of megakaryocytes and platelets, keeping serum Tpo at normal levels. The production of platelets from mature megakaryocytes has been shown to be Tpo independent.⁴⁵ 

To explain why the Yall transgene is expressed at much lower levels than endogenous Mpl in late megakaryocytes and platelets in Yall;Mpl^−/− mice, we considered several possible explanations. First, the mRNA of the Yall transgene could be less stable than the mRNA of endogenous Mpl, possibly due to the difference in the 3′ UTR. However, expression of Mpl constructs bearing different 3′ UTRs in cell lines did not support this hypothesis. Second, transcription of the Yall transgene might decrease during terminal megakaryocytic differentiation. This hypothesis is supported by our findings that Yall expression was reduced in ploidy-sorted mature megakaryocytes and also decreased during maturation of megakaryocytes cultured in vitro (Figure 6). A possible explanation for decreased Mpl expression in late megakaryocytes could be that the 2-kb Mpl promoter fragment used in the Yall construct lacks some elements required for expression in late megakaryocytes. A putative chromatin-dependent enhancer was described in Mpl intron 6.⁴⁶  Alternatively, the integration site of the transgene could exert a silencing effect at later stages of megakaryocyte differentiation. Since several independent transgenic strains with the wild-type Mpl cDNA showed thrombocytosis when examined on the Mpl^−/− background, this possibility appears less likely. Gene silencing of multicopy transgenes is a well-known phenomenon,⁴⁷  but there is currently no evidence that such an effect could be selective for late stages of megakaryopoiesis. Finally, a decreased efficiency of the transgenic Mpl mRNA translation in mature megakaryocytes could in addition also contribute to decreasing the Mpl protein levels. This possibility is supported by the observation that the total Mpl mRNA was slightly higher in unfractionated megakaryocytes from Yall;Mpl^−/− mice than from wild-type mice, while the Mpl protein levels were approximately the same (Figure 5).

Reduced expression of Mpl protein in platelets was described as a feature in many patients with MPD,^48,49  and could be detected in patients with all 3 disease entities (polycythemia vera, essential thrombocythemia [ET], and primary myelofibrosis).^50-52  Loss of Mpl protein was also observed in megakaryocytes from patients with MPD,^49,53,54  but it is currently unclear whether this specifically affects late megakaryocytes, as we observed in our mice. Recently, a correlation between decreased expression of Mpl and mutant JAK2-V617F was noted.⁵⁵  In 2 studies, platelets from patients with ET with decreased Mpl protein showed a concomitant reduction of Mpl mRNA resembling our Yall mouse model,^48,56  whereas in another study, the platelet Mpl mRNA was normal in patients with MPD with decreased platelet Mpl protein⁵¹  (our unpublished data). Furthermore, decreased expression of Mpl protein was also found in patients with hereditary thrombocythemia due to a mutation in the THPO gene,⁵¹  and in some cases of reactive thrombocytosis.⁵⁰  These data suggest that decreased platelet Mpl expression can contribute to thrombocytosis in MPD, but different mechanisms may be effective that need to be further investigated.

A single nucleotide substitution (G1238T) in the Mpl gene that results in a change from lysine to asparagine at amino acid 39 of the Mpl protein (MPL^K39N) was found in 7% of African Americans.⁵⁷  Termed MPL^Baltimore, the defect is associated with mild thrombocytosis in heterozygous individuals and more marked thrombocytosis (> 800 × 10⁹/L) in the homozygous state.^57,58  These patients also had reduced platelet protein Mpl expression,⁵⁷  suggesting that a decrease of Mpl protein expression in platelets can contribute to the development of thromobocytosis in human patients. In contrast, mutations in the transmembrane domain (Mpl^S505N) or cytoplasmic juxtamembrane domain (Mpl^W515K/L) cause thrombocytosis through a different mechanism (ie, constitutive activation of Mpl signaling).^15,16,59-62  It will be interesting to develop mouse models for these mutations. Our results indicate that such mouse models will need to be examined in the presence of the wild-type Mpl allele, as we demonstrate here that dysregulated expression of Mpl even in the absence of any mutations can be sufficient to cause marked thrombocytosis when examined on an Mpl^−/− background.

In summary, our findings support the concept that expression of Mpl protein in platelets and megakaryocytes is an important negative feedback factor in the regulation of thrombopoiesis, and that decreased Mpl expression can cause a phenotype resembling essential thrombocythemia.

We thank Verena Jäggin for FACS sorting, Dr Kenneth Kaushansky and Dr Norma Fox for providing serum and plasma from Tpo^−/− mice, and Dr Francois Lanza and Dr Lucia Kubovcakova for helpful comments on the manuscript.

This work was supported in part by grants 31-53677.98, 3100-066949.01, and 310000-108006/1 from the Swiss National Science Foundation to R.C.S.

Contribution: R.T. and J.C. designed and performed research, analyzed data, and wrote the paper; S.Z., A.W., H.H.-S., C.B., J.S., S.K., F.J.d.S., and C.W.J. performed research and analyzed data; and R.C.S. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: F.J.d.S. is an employee and stockholder of Genentech. The remaining authors declare no competing financial interests.

Correspondence: Radek C. Skoda, Department of Research, Experimental Hematology, University Hospital Basel, Hebelstrasse 20, 4031 Basel, Switzerland; e-mail: radek.skoda@unibas.ch.