Thrombopoietin induces production of nucleated thrombocytes from liver cells in Xenopus laevis

The development of mammalian megakaryocytes (MKs) and platelets, which are thought to be absent in non-mammals, is primarily regulated by the thrombopoietin (TPO)/Mpl system. Although non-mammals possess nucleated thrombocytes instead of platelets, the features of nucleated thrombocyte progenitors remain to be clarified. Here, we provide the general features of TPO using Xenopus laevis TPO (xlTPO). Hepatic and splenic cells were cultured in liquid suspension with recombinant xlTPO. These cells differentiated into large, round, polyploid CD41-expressing cells and were classified as X. laevis MKs, comparable to mammalian MKs. The subsequent culture of MKs after removal of xlTPO produced mature, spindle-shaped thrombocytes that were activated by thrombin, thereby altering their morphology. XlTPO induced MKs in cultured hepatic cells for at least three weeks; however, this was not observed in splenic cells; this result demonstrates the origin of early haematopoietic progenitors in the liver rather than the spleen. Additionally, xlTPO enhanced viability of peripheral thrombocytes, indicating the xlTPO-Mpl pathway stimulates anti-apoptotic in peripheral thrombocytes. The development of thrombocytes from MKs via the TPO-Mpl system in X. laevis plays a crucial role in their development from MKs, comparable to mammalian thrombopoiesis. Thus, our results offer insight into the cellular evolution of platelets/MKs in vertebrates. (200/200).

In addition to platelet production, the expansion of HSCs is also a function of the TPO/Mpl system. Therefore, the question to be addressed is whether TPO function in pluripotent progenitor cells is conserved among vertebrates. The development of polyploidy in MKs is unclear 24 , although they can be artificially generated from embryonic stem cells, induced pluripotent stem cells, and fibroblasts [25][26][27] . The aim of this study was to characterize TPO/ Mpl function in the development of mature/immature nucleated thrombocytes in non-mammalian vertebrates.
The structures and expression of mammalian TPO and Mpl are well understood. Mammalian TPO mRNA is expressed in hepatocytes, osteoblasts, kidneys, and the spleen 28,29 . Native human TPO circulates and maintains thrombopoiesis 30 , with constitutive hepatic expression of TPO and total Mpl mass, comprising MKs and platelets 31 . Posttranslational processing, including proteolytic cleavage, cleaves the C-terminal carbohydrate-rich domain of TPO, modulating its activity 32 . In addition, TPO mRNA expression is increased in the bone marrow and spleen of thrombocytopenic animals 33 . It is not clear if those particular mechanism are common to all vertebrates, and direct purification of TPO protein that provides physicochemical knowledge about native mammalian TPOs, has not been achieved in non-mammalian vertebrates 34,35 . Non-mammalian TPO genes have been identified and cloned from zebrafish and chicken 18,36 . Chicken TPO stimulates erythro-and thromboblastosis, and zebrafish TPO stimulates thrombopoiesis 18,36 . In the embryo of the African clawed frog, Xenopus laevis, rat TPO promotes expansion of haematopoietic progenitors, thus demonstrating cross-species TPO stimulation of haematopoieisis 37 .
X. laevis is one of the most popular animal models in embryology and physiology. We have directed our efforts to establishing a new animal model for the study of haematopoieisis [38][39][40][41][42][43][44] and have investigated the physiological haematopoieisis response under a variety of environmental stress such as changes in temperature 41,45 . We recently reported that thrombocytic progenitors are localized in the liver and spleen of X. laevis and have a greater DNA content than do peripheral erythrocytes and thrombocytes 43 . Here, we describe the identification, cloning, and expression of biologically active X. laevis TPO (xlTPO) and Mpl (xlMpl). XlTPO induces the development of hepatic and splenic MKs as progenitors of nucleated thrombocytes in X. laevis and regulates the fate of peripheral Mpl-expressing thrombocytes via anti-apoptotic signalling. To our knowledge, this is the first report of the development of nucleated thrombocytes from MKs induced by the TPO/Mpl system.

Results
Identification and cloning of the X. laevis TPO and Mpl genes. We identified more than 60 putative orthologous of TPO by reference to public databases. Until now, there have been no comparative studies of molecular structure and function in other organisms ( Supplementary Fig. S1). Among X. laevis hepatic and splenic T12 + / CD41 + thrombocytic cells demonstrated in our previous study 43 , only hepatic large T12-positive cells are morphologically similar to MK in mammals (Fig. 1A). Therefore, we first hypothesized that thrombocyte progenitor in X. laevis originated from large cells, and attempted to clone TPO and Mpl in X. laevis. To identify the TPO sequence in X. laevis, we identified neighbouring genes in the human genome database. Polymerase RNA II DNA-directed polypeptide H (POLR2H) and chordin (CHORDIN) were then identified in the expressed sequence tag (EST) sequence of the X. tropicalis TPO locus ( Supplementary Fig. S2A). cDNA encoding full-length X. laevis TPO was obtained by RT-PCR amplification from the adult X. laevis liver and spleen. The X. laevis, mammalian, zebrafish, and chicken TPOs were subsequently aligned (Fig. 1B). X. laevis shares an overall sequence identity of 87% with X. tropicalis TPO (Fig. 1B). The full-length xlTPO encodes a 153-amino acid-long protein; the non-mammalian TPO lacks the C-terminal domain. Mature xlTPO protein contains four cysteine residues (Cys-4, Cys-24, Cys-78, and Cys-97) that are conserved in human TPO (hTPO) at positions Cys-7, Cys-29, Cys-85, and Cys-151 ( Supplementary  Fig. S3). Zebrafish, chicken and X. laevis TPO lack a C-terminal half domain; however, the first to the fourth Cys residues of xlTPO are essential for biological activity in humans and rodents 30 . Therefore, we identified low but consistent sequence identities with the human (23%), mouse (23%), rat (23%), chicken (24%), and zebrafish (18%) homologs (Fig. 1B). The Mpl-neighbouring genes in the human genome database were identified as TIE-1 and CDC-20 (cell division cycle 20 homologue) ( Supplementary Fig. S2B). The extracellular domain of Mpl contains highly conserved tyrosine and cysteine residues, WSXWS motifs, and Box1 and Box2 motifs necessary for intercellular signalling ( Fig. 1D and Supplementary Fig. S4). The extracellular domain of X. laevis Mpl, which mediates ligand binding, shares homology with the human (22%), rat (24%), mouse (24%), chicken (30%), and zebrafish (22%) Mpl sequences (Fig. 1C). The low similarities of the X. laevis TPO and Mpl suggest substantial differences in the biological functions of the X. laevis TPO/Mpl system. mRNA expression of xlTPO and xlMpl in the liver, spleen, kidney, and lung was confirmed by RT-PCR analysis; xlMpl mRNA was detected in the spleen, kidney, intestine, and liver, and in thrombocytes in the peripheral blood (Fig. 1E).
Biological activity of xlTPO. Cells of the kidney, liver, and spleen were cultured in serum containing semisolid media with xlTPO. After 2 days, non-haemoglobinized colonies emerged from the hepatic and splenic cells and reached a peak on day 4; kidney cells did not yield colonies ( Fig. 2A). XlTPO (0.1-10 ng/mL) produced a concentration-dependent effect on thrombocytic cell proliferation in the spleen and liver (Fig. 2B). The colonies were T12-positive, indicating they were thrombocytic colonies, and they did not appear by erythropoietin (EPO) stimulation (Fig. 2C). To validate whether xlTPO stimulates proliferation with xlMpl, we produced a fusion protein containing the extracellular domain of xlMpl and the murine IgG2a Fc region. Thrombocytic colony formation was inhibited by xlMpl-Fc in xlTPO-containing medium, whereas hepatic cells gave rise to thrombocytic colonies in the presence of xlTPO and IgG2a as the control (Fig. 2D). Thus, xlTPO stimulates thrombocytic cells via Mpl signalling. In addition, the effects of xlTPO activity in human Mpl were tested in cell proliferation assays using an hTPO-dependent cell line, UT-7/TPO. Despite low similarity in the primary structures of TPO and MPL (Fig. 1B,C), xlTPO stimulated proliferation of UT-7/TPO cells ( Supplementary Fig. S5A,B), and JAK2 was tyrosine-phosphorylated ( Supplementary Fig. S5C), similar to the effects of hTPO. Subpopulation of splenic and hepatic T12-positive cells by xlTPO stimulation. In X. laevis, PHZ induced acute haemolysis, erythrocyte counts reached a nadir within 8 days, and the number of immature haematopoietic progenitors increased in the liver and spleen 39 . Anaemic serum samples were obtained 4 days after PHZ administration to obtain the maximal colony-forming activity. Hepatic or splenic cells from PHZ-induced anaemic X. laevis were cultured in a semisolid culture system in the presence of serum collected at day 4. Thereafter, T12 and o-dianisidine co-positive colonies emerged ( Supplementary Fig. S6A), suggesting the presence of bipotent thrombocyte/erythrocyte progenitors in the liver and spleen of X. laevis, as in chicken 46 . We also observed   Characterization of MKs in X. laevis. MKs in X. laevis expressed T12 and CD41 and appeared when hepatic or splenic cells were stimulated by recombinant xlTPO (Fig. 4A,B). After 8 days of culture, T12-positive large cells were collected by cytometry, and the cytosolic structures were observed by transmission electron microscopy. The vacuolar system in T12-positive large cells closely resembled the surface-connecting system of MKs and platelets in mammals, and the granules in T12-positive large cells in X. laevis also resemble the dense granules in MKs and platelets in mammals (Fig. 4C). RT-PCR showed the T12-positive large cells express Mpl, CD41, and Fli-1 but not acetylcholine esterase (AChE), EPOR, or myeloperoxidase (MPO) (Fig. 4D). The T12-positive large cells at 8 days of culture were stained with Hoechst 33342 for flow cytometry, which showed the fluorescence intensities of the T12-positive large cells were 2-to 4-fold higher than those of mature thrombocytes (Fig. 4E). On the basis Differentiation of mature thrombocytes. Cultured MKs were enriched by Percoll density-gradient centrifugation (Fig. 5A) and cultured in the presence or absence of xlTPO. After 2 days, spindle-shaped cells appeared in MKs cultured without xlTPO (Fig. 5B). These cells were T12-positive, suggesting they have features of peripheral thrombocytes (Fig. 5C). The peripheral thrombocyte fraction was collected by density-gradient centrifugation and cultured in dα -MEM containing 20% FCS. After 4 days, the thrombocytes had acquired a barbell morphology similar to cultured thrombocytic cells (Fig. 5D and Supplementary Fig. S8A). To determine whether these cells were functional thrombocytes, a thrombocyte activation assay was performed. After stimulation with thrombin, spindle-shaped thrombocytic cells were counted; upon thrombin activation, their morphology changed (Fig. 5E). These results were consistent with the functions of peripheral thrombocytes.
TPO-Mpl signalling in peripheral thrombocytes. The Mpl gene was also expressed in peripheral thrombocytes. To examine TPO/Mpl signalling in X. laevis, peripheral thrombocytes were collected and cultured in the presence of xlTPO. XlTPO enhanced thrombocyte viability by 50% relative to untreated thrombocytes at day 10 (Fig. 6A). The irregular shape of the non-stimulated thrombocytes indicated the cells were dead  . 6B). We then tested apoptosis induction by analysing karyorrhexis. Unlike thrombocytes treated with xlTPO, untreated thrombocytes exhibited increased rates of apoptosis (Fig. 6C). XlTPO (0.1-10 ng/mL) also produced a concentration-dependent effect on antiapoptosis of peripheral thrombocytes after 8 days of culture ( Supplementary  Fig. S8B). In addition, xlTPO activated STAT5 phosphorylation (Fig. 6D), indicating the role of xlTPO as an anti-apoptotic regulator in peripheral thrombocytes. The proposed mechanism of xlTPO function in the liver, spleen, and peripheral blood is summarized in Fig. 7.

Discussion
Mammalian platelets are derived from the cytoplasm of MKs through endocytosis, primarily induced by TPO, which mediates MK maturation and stem cell expansion. In this study, we explored the origin of nucleated thrombocytes in the circulation and the functional role of the X. laevis TPO/Mpl system in the development of nucleated thrombocytes. The N-terminal region of hTPO is homologous to human EPO, with which it shares 23% sequence identity. The xlTPO sequence also overlaps with X. laevis EPO (xlEPO) by 30%, although xlTPO stimulates the production of thrombocytes, i.e., thrombopoiesis, and xlEPO stimulates erythropoiesis 39 . Therefore, the function of EPO is distinct from that of TPO, at least in amphibians. XlTPO stimulates phosphorylation of cellular proteins in human UT-7/TPO cells ( Supplementary Fig. S5C), demonstrating a certain tertiary structure in the TPO-Mpl binding is partly shared, despite differences in the primary structure (Fig. 1B,C and Supplementary Fig. S1). The conservation of Cys residues also suggests receptor recognition is conserved. This finding is crucial to understanding the evolution of TPO signalling.
The C-terminal region of hTPO contains multiple N-glycosylation sites that mediate secretion of the protein 47 , and circulating native hTPO comprises this full-length form 48 . In contrast, xlTPO lacks the C-terminal domain of mammalian TPOs but does possess potential N-glycosylation sites in the N-terminus. In X. laevis, xlTPO is produced in the liver where thrombocytes develop, in an analogous way to X. laevis erythropoiesis 39,40 . In contrast, xlTPO mRNA is highly expressed in the lung where there are no Mpl-expressing cells reside. Therefore, xlTPO likely circulates in the blood and is stabilized by the sugar chain, which influences the lifespan of circulating thrombocytes (Fig. 6). Human TPO primes platelet aggregation induced by shear stress and various agonists 49 , and blood levels are regulated by the platelet mass through TPO binding to Mpl on the platelet surface under constitutive hepatic expression of TPO mRNA 31 . In this study, xlTPO stimulated phosphorylation of STAT5 and anti-apoptosis in peripheral thrombocytes in vitro (Fig. 6C,D). Phosphorylation of STAT5 induced the expression of the anti-apoptotic factors bcl-xL and pim-1 50 . Therefore, circulating xlTPO might suppress apoptosis in peripheral thrombocytes through bcl-xL signalling and might regulate the fate of peripheral thrombocytes.
Colony formation in the presence of xlTPO and xlEPO revealed thrombocyte and erythrocyte progenitors in the liver and spleen of X. laevis (Fig. 2) 40 . We have previously reported that peripheral thrombocytes and T12-positive cells in the liver have 2-and 4-fold greater DNA content than do erythrocytes, whereas polyploidization is the same in splenic T12-positive cells and peripheral thrombocytes 43 . While hepatic cells could be cultured for more than 2 weeks in the presence of xlTPO stimulation, splenic cells could not (Supplementary Fig. S7A). In addition, cellular distribution of 10-20 μ m-T12 positive cell in the spleen was higher than the liver ( Supplementary  Fig. S7B), suggesting that thrombocyte progenitors are present at different developmental stages in the liver and spleen. Functional thrombocytes are produced from CD41-and T12-positive MKs, primarily in the liver, whereas morphologically mature thrombocytes localize in the spleen 43 . Although thrombocyte progenitors reside in the spleen and liver, thrombocyte progenitors are more immature in the liver than in the spleen. Thus, the X. laevis liver niche regulates MKs maturation, whereas the spleen niche is superior at a later stage, when it comprises peripheral thrombocytes. In adult mammals, haematopoietic progenitors reside in the microenvironment of the bone marrow composed of osteoblast, endothelial, and stromal cells that regulate stem cell quiescence and differentiation 51 . The spleen and liver niches in X. laevis comprise soft tissue lacking in osteogenesis, and both organs are potential destruction sites of thrombocytes; the difference may reveal underlying factors of terminal differentiation.
Only non-mammalian HSCs have been identified in ginbuna crucian carp and zebrafish 52,53 . In mice, TPO regulates HSC quiescence and interaction with the osteoblastic niche 54 , and HSCs are absent in Mpl-deficient mice 55 . A murine Mpl derivative lacking the distal 60 amino acids revealed the essential nature of the membrane-proximal region for maintenance of HSC activity 56 ; xlMpl also possesses this membrane-proximal region (Fig. 1D). hTPO stimulates in vitro proliferation of progenitor cells for more than 2 weeks; xlTPO also stimulates proliferation of X. laevis hepatic cells for more than 2 weeks, suggesting that the presence of immature and earlier haematopoietic progenitors in X. laevis needs to be addressed in future studies. Recent studies have found that heterogeneous HSCs can directly differentiate into megakaryocyte progenitors 57 . Although we could not evaluate the HSCs in X. laevis, it is important to identify the main pathway to produce the thrombocytes for development.
Our study demonstrated that nucleated thrombocyte progenitors in X. laevis resemble platelet progenitors in mammals. Both have multilobed nuclei and produce haemostatic cells activated by thrombin, suggesting the platelets evolved from nucleated spindle-shaped elliptic thrombocytes. In X. laevis, blood capillaries do not develop as in mammals 58 and a proplatelet formation system is unnecessary. The number of peripheral thrombocytes in X. laevis (30 g) is approximately 3 × 10 7 cells; moreover, the ratio of peripheral thrombocytes to total thrombocytes is lower in X. laevis than in mammalian species 43 . The DNA content of mammalian MKs increases from 2 N to 256 N, and 4000-8000 platelets may be produced from a single MK 59 . In this study, the DNA content of X. laevis MKs increased from 4N (allotetraploid) to 32N, and a single MK divided at least twice and produced 1-4 thrombocytes.
The X. laevis model will provide new insights in the field of comparative haematology with the added support of the X. laevis genome database [Horb M et al. Xenbase. 2014 Xenopus Community White Paper. Improvement of Xenopus antibody resources. 2014. Available at: http://www.xenbase.org/community/xenopuswhitepaper.do. Accessed June 10, 2015]. Our data showed the function of xlTPO in thrombocyte production in the liver and spleen, and as an inducer of anti-apoptosis in peripheral thrombocytes. We demonstrated the derivation of thrombocytes from MKs induced by TPO/Mpl signalling in X. laevis and conclude that the process of platelet production is partially conserved in the mechanism of non-mammalian thrombocyte production.

Materials and Methods
All animal experiments were performed in accordance with the approved protocols and guidelines of the Steering Committee for Animal Experimentation at Waseda University.
Animals. Wild-type male X. laevis (10-30 g) were purchased from Aquatic Animal Supply (Misato, Saitama, Japan) and housed in a 20-L aquarium on a 12-h light/dark cycle at 22 °C.
Isolation of the X. laevis TPO and Mpl genes. To identify the xlTPO and xlMpl genes, we searched neighbouring genes at the TPO locus (NCBI gene ID 4352) and Mpl locus (NCBI gene ID 7066) in the human genome database. POLR2H and CHORDIN were identified at the TPO locus, and tyrosine kinase with immunoglobulin and epidermal growth factor homology domains TIE-1 and CDC-20 were identified at the Mpl locus. These were then searched against the NCBI Homologene or BLAST database for X. tropicalis. For xlTPO, we found an EST sequence in the X. tropicalis TPO locus. The EST sequence was obtained from GenBank (accession number: DR881950). For xlMpl, we searched for conserved Mpl sequences by TBLASTN of the X. tropicalis Mpl locus with the TIE-1 and CDC-20 genes. The program GENESCAN (http://genes.mit.edu/GENSCAN.html) was used to predict the exons and coding sequences. X. laevis RNA preparation and DNA sequencing. Total RNA was extracted from various adult X. laevis tissues using TRIZOL reagent (Invitrogen). The purified total RNA was reverse-transcribed into cDNA using ReverTraAce (Toyobo). To generate the putative xlTPO cDNA fragment, RT-PCR of total RNA from spleen and liver cells was performed using primers specific for X. tropicalis TPO (xtTPO)_Fw2 (5′ -TCGATTCGCTCATATTCTGC-3′ ) and xtTPO_Re2 (5′ -TGGCAAGGTACAGTGTAGTCCA-3′ ) and Ex Taq polymerase or pfu polymerase (Takara). Primers were complementary to the 5′ and 3′ edges of the EST sequence obtained from the X. tropicalis genome database. The same primer pair was used to analyse TPO expression in various X. laevis tissues. The obtained cDNA fragment was cloned into pGEM-T Easy (Promega) and sequenced on an Applied Biosystems 3130 Genetic Analyzer with dye-terminator chemistry.
Sequence analysis, alignment. Protein sequences for the mammalian TPO and Mpl were retrieved from the NCBI database. The signal peptide of xlTPO was predicted by SignalP and compared to the reported amino acid sequences of other mammalian species. Alignments of the X. laevis and mammalian TPO and Mpl sequences were generated with ClustalX software. Morphology and staining. Cells were collected and stained with MGG and immunostained with T12 and CD41 as previously described 38,43 . Western blotting. Thrombocytes were collected by gradient density centrifugation, washed twice with 0.8 × Dulbecco's modified phosphate-buffered saline with EDTA to remove Mg 2+ and Ca 2+ ions. Total thrombocyte protein was extracted in M-PER lysis buffer (ThermoFisher) containing protease and phosphatase inhibitors (Roche Applied Science). SDS polyacrylamide gel-electrophoresis (PAGE) was performed as previously described 43 . Lysates were prepared and 3 μ g of total lysate were quantified and separated by SDS-PAGE; gels were blotted onto polyvinylidene difluoride membranes (Millipore) and stained polyclonal antibodies to STAT-5 (1:200 dilution, Santa Cruz) and phosphorylated STAT-5 (1:1000 dilution, BD Biosciences). Antibody binding was detected by incubation with a horseradish peroxidase-labelled secondary antibody, followed by chemiluminescence detection (ECL-Plus; Amersham Pharmacia Biotech).
In vitro colony assay and liquid cell culture. Haematopoietic cells were isolated from the liver and spleen of X. laevis. These cells were filtered (BD Biosciences Clontech) and washed four times with dα -MEM (Invitrogen). The cells were cultured at 0.8 × 10 5 cells/mL in 35-mm plastic Petri dishes (Corning) in 2 mL dα -MEM containing 20% heat-inactivated FCS, 0.8% methylcellulose (Shin-Etsu Chemical), and 100 μ g/mL streptomycin (Invitrogen) and 100 μ g/mL kanamycin (Invitrogen). Escherichia coli BL21(DE3) was transformed with plasmid pET19b (Takara), which encodes a full-length cDNA copy of xlTPO encoding the enterokinase inserted between NdeI and BamHI sites, and xlTPO was expressed by this plasmid. Stimulators were added where appropriate, and the cells were incubated in 5% CO 2 at 23 °C. Colony formation was assessed on alternate days from day 2 to day 18. Colonies were picked with a 20-μ L pipet tip, suspended in dα -MEM containing 10% FCS, and centrifuged (400 rpm). The samples were stained with various dyes in triplicate experiments. Error bars in figures depict standard errors. Differences were considered significant at the 95% confidence level (P < 0.05). We produced a soluble xlMpl-Fc fusion protein (xlMpl-Fc) containing the extracellular domain of xlMpl (aa 11-246) and the murine IgG2a Fc region in HEK293 cells. To verify xlTPO/xlMpl signalling, 1 ng/mL xlTPO was incubated with 500 ng/ mL xlMpl-Fc for 30 min and added to hepatic cells in semisolid culture to assess neutralization; mouse IgG2a (Dako) was incubated with xlTPO as the control. Hepatic or splenic cells were cultured as the liquid suspension in dα -MEM containing 20% FCS and xlTPO (5 ng/mL). Cells were counted and suspended at 5.0 × 10 5 cells/mL, then cultured with 500 μ L dα -MEM containing 10% FCS supplemented with 100 μ g/mL streptomycin and 100 μ g/ mL kanamycin for 20 days at 23 °C.

Statistical analysis.
Results are presented as means ± SE. Statistical analysis included analysis of variance and Student's t-test. A P-value of 0.05 or less was considered statistically significant.