Cell-autonomous megakaryopoiesis associated with polyclonal hematopoiesis in triple-negative essential thrombocythemia

A subset of essential thrombocythemia (ET) cases are negative for disease-defining mutations on JAK2, MPL, and CALR and defined as triple negative (TN). The lack of recurrent mutations in TN-ET patients makes its pathogenesis ambiguous. Here, we screened 483 patients with suspected ET in a single institution, centrally reviewed bone marrow specimens, and identified 23 TN-ET patients. Analysis of clinical records revealed that TN-ET patients were mostly young female, without a history of thrombosis or progression to secondary myelofibrosis and leukemia. Sequencing analysis and human androgen receptor assays revealed that the majority of TN-ET patients exhibited polyclonal hematopoiesis, suggesting a possibility of reactive thrombocytosis in TN-ET. However, the serum levels of thrombopoietin (TPO) and interleukin-6 in TN-ET patients were not significantly different from those in ET patients with canonical mutations and healthy individuals. Rather, CD34-positive cells from TN-ET patients showed a capacity to form megakaryocytic colonies, even in the absence of TPO. No signs of thrombocytosis were observed before TN-ET development, denying the possibility of hereditary thrombocytosis in TN-ET. Overall, these findings indicate that TN-ET is a distinctive disease entity associated with polyclonal hematopoiesis and is paradoxically caused by hematopoietic stem cells harboring a capacity for cell-autonomous megakaryopoiesis.


Noncanonical JAK2 and MPL mutant exhibits wild-type-equivalent levels of STAT5 activation.
Because noncanonical mutations in JAK2 and MPL can be found in a subset of triple-negative myeloproliferative neoplasm (TN-MPN) patients 11,12 , we sequenced all exons of JAK2 and MPL for all the TN-ET patients (see Methods) and found 4 noncanonical JAK2 or MPL variants in 4 patients: JAK2 I724T (germline), MPL X636WX12 (germline), MPL S204F (somatic), and MPL A58V (unknown status due to a lack of germline control) ( Fig. 1B; Table S1). To examine the oncogenic properties of these variant gene products, we performed a STAT5 reporter assay (see Methods). We found that, unlike canonical mutants, such as JAK2 V617F and MPL W515L, all noncanonical variants displayed wild-type-equivalent levels of STAT5 activation in the absence of TPO (Fig. 3). In agreement with this result, aside from MPL S204F, the noncanonical variants observed are rare variants in the general population (Table S2). In addition, by whole-exome sequencing (WES) analysis of 11 patients with available genomic DNA (gDNA) samples from both peripheral blood and CD3-positive cells, we found that 2 patients harbored somatic mutations (details shown in Table S1). One such patient was found to harbor MPL S204F. Nevertheless, no mutation was common in these patients and has been implicated as a driver mutation in MPN.

Polyclonal hematopoiesis in TN-ET.
Because of the considerably low frequency of somatic mutations in the TN-ET patients (Table S1), we suspected that the majority of the TN-ET patients had reactive thrombocytosis with bone marrow (BM) characteristics resembling the MPN. To this end, we performed a human androgen receptor (HUMARA) assay to examine the clonality of hematopoietic cells in the TN-ET patients by measuring the degree of skewed methylation on paternal and maternal X-chromosomes 14 . Seventeen of 18 female TN-ET patients had available gDNA from granulocytes (n = 12) or mononuclear cells (MNCs) (n = 5); we assessed their samples, and only one (9.1%) of 11 patients who exhibited judgeable results showed a clonal pattern, while 10 patients (90.9%) showed a polyclonal pattern (Fig. 4B). In contrast, all the patients (n = 3) harboring canonical driver mutations showed a clonal pattern as previously described 15 , and the patients (n = 3) with reactive thrombocytosis showed a polyclonal pattern (data not shown). The frequency (9.1%) of clonal hematopoiesis in TN-ET was within the range expected from a previous study where a higher frequency (14 of 31, 45.2%) of clonal hematopoiesis was observed in elderly females 14 . Nevertheless, despite the histology defining neoplastic features in the BM, most of the TN-ET patients in our cohort exhibited a polyclonal pattern according to the HUMARA assay.
The levels of serum cytokines were comparable among the TN-ET and ET patients harboring canonical mutations. To search for the cause of thrombocytosis associated with polyclonal hematopoiesis, we examined the levels of serum cytokines, such as TPO [16][17][18][19] and interleukin-6 (IL-6) 16,18,19 , that have been shown to promote platelet production. By analyzing samples from 16 patients with TN-ET, 17 patients with mutated ET, and 8 healthy individuals, we found that the levels of TPO and IL-6 were increased in the patients with TN-ET and mutated ET compared to the healthy controls, but there was no significant difference among the groups, implying that the increases in the levels of TPO and IL-6 do not or only partly promote thrombocytosis in TN-ET (Fig. 4C,D). www.nature.com/scientificreports/ . Twelve TN-ET, 7 mutated ET, and 7 control samples, including normal (n = 6) and reactive (n = 1) samples, were evaluated. Even in the absence of TPO, Mk colonies formed in TN-ET and mutated ET cell cultures but much less in control cell cultures (Fig. 5A). The capacity to form TPO-independent Mk colonies was determined by the ratio of the number of colonies formed in the absence and presence of TPO and compared between the groups. As shown in Fig. 5B, TN-ET and mutated ET exhibited equivalent capacities to form TPO-independent Mk colonies, and the capacity of these cells was significantly increased compared to

Discussion
In the present study, we performed an in-depth analysis of the clinical and biological features of TN-ET in a single institution. Statistical analysis of clinical records revealed that young females were predominant in the TN-ET patient population with no incidence of thrombosis and no disease progression to fibrosis and leukemia (Table 1; Fig. 2), strongly suggesting that TN-ET is a disease entity that is biologically distinct from mutated ET. Sequencing analysis failed to identify somatic mutations in the majority of the TN-ET patients, suggesting reactive thrombocytosis associated with polyclonal hematopoiesis in TN-ET. In contrast, no significant elevation of cytokines related to platelet production was observed (Fig. 4). Biological analyses revealed that hematopoietic stem/progenitor cells in TN-ET presented a capacity to promote megakaryopoiesis in the absence of TPO (Fig. 5).
Collectively, these findings indicate that TN-ET, which may still be a heterogeneous population, is a distinctive disease entity associated with polyclonal hematopoiesis but is caused in a cell-autonomous manner. Compared to previous studies with large cohorts, in our cohort, the TN-ET patients (median age of 36.0 years) were younger (ranging between 47 and 59 years) 7,9,10 . Despite such differences that may reflect racial and geographical differences, we observed biological features such as polyclonal hematopoiesis and a low frequency of somatic mutations that resembled those observed among other cohorts. However, it is notable that no thrombotic events were observed in our cohort, while 10-20% of the TN-ET patients in other studies 2,7,9,10 exhibited thrombosis. While the thrombosis risk determined by the IPSET-thrombosis model 20 was comparable between www.nature.com/scientificreports/ TN-ET and CALR or MPL mutant ET (Table 1), the lack of thrombotic events in TN-ET might indicate the specific biological feature of TN-ET in our cohort. Consistent with a previous study 11,12,21 , noncanonical JAK2 and MPL mutations were detected by WES or next-generation sequencing (NGS) in a fraction of the TN-ET patients in our cohort. The STAT5 reporter assay revealed that noncanonical mutants of JAK2 and MPL did not show activity similar to that of the canonical mutant proteins but rather exhibited the same level of activity as the wild-type proteins (Fig. 3). Although the noncanonical mutant MPL S204P has been shown to possess a weak gain-of-function property 12 , the present and previous 11 studies implied that MPL S204F may not be the case. Nevertheless, the low frequency of noncanonical mutations in JAK2 and MPL in TN-ET does not fully explain the pathogenesis of TN-ET.
Polyclonal hematopoiesis was observed in nearly all the TN-ET patients examined, and these patients lack acquired mutations (Table S1). While this suggests the possibility of hereditary thrombocytosis, the platelet count before the diagnosis of all the patients (n = 10) with available blood count data showed no sign of thrombocytosis before diagnosis. Paradoxically, hematopoietic stem/progenitor cells in TN-ET exhibited an oncogenic property in megakaryopoiesis, which may explain the observation of indistinguishable morphology of BM in TN-ET with mutated ET. Based on these facts, we proposed that an environmental cue triggers a persistent epigenetic change in megakaryocytic progenitor cells to confer a capacity to promote megakaryopoiesis in the absence of TPO and results in thrombocytosis. Aberrations in epigenomic regulation not associated with genetic alteration have been shown to promote uncontrolled cell proliferation 22 . Such epigenomic changes may be reversible, and in fact, we observed spontaneous regression of thrombocytosis in one TN-ET patient diagnosed with thrombocytosis www.nature.com/scientificreports/ 1.5 years previously. Further analyses are required to better understand the pathogenesis of TN-ET, which should lead to the establishment of an appropriate treatment strategy against TN-ET.
In conclusion, we performed in-depth analysis of TN-ET and found that most of the TN-ET patients exhibited polyclonal hematopoiesis with no acquired mutation and no sign of hereditary thrombocytosis. Despite the possibility of reactive thrombocytosis, BM specimens exhibited the features of ET, and hematopoietic stem cells from TN-ET patients showed a capacity for cell-autonomous megakaryopoiesis, the hallmark of ET 23 . Based on these data, we propose that TN-ET, which may still be a heterogeneous population, is a biologically distinctive disease entity; thus, a different treatment strategy may need to be considered from that for mutated ET.

Patients.
A total of 483 patients in the Hematology Department, Juntendo University Hospital who were suspected to have ET were analyzed and defined based on the WHO 2016 criteria 24 . Those who were suspected to have familial MPNs were excluded from this study. Preparation of gDNA and analysis of JAK2 V617F, MPL exon 10 and CALR exon 9 mutations were performed as described previously 6,[25][26][27] . Histological analysis of BM was performed as described previously 28 . TN-ET was defined by the bone marrow morphology and by evidence of persistent thrombocytosis (> 450 × 10 9 /L for more than six months for all except one who was only followed up for two months) with no potential cause of reactive thrombocytosis. This study was conducted in accordance with the Helsinki Declaration of 1975 and approved by the ethics committee of the School of Medicine, Juntendo www.nature.com/scientificreports/ University (IRB#2013020). Written informed consent was obtained prior to the use of samples and the collection of clinical records.
STAT5 reporter assay. A STAT5 reporter assay was performed as described previously 31 . cDNA of JAK2 and MPL variants were subcloned into pcDNA3.1 plasmids, which were transfected and expressed in HEK293T cells. MPL was coexpressed with wild-type JAK2.
Statistics. Blood cell counts and biochemical parameters at the first visit prior to the treatment were analyzed. The durations for the development of fibrosis and leukemia were defined from the date of the first visit to that of the diagnosis of grade 2 or 3 fibrosis and to that of the detection of > 20% blasts in the peripheral blood or BM, respectively. Risk stratification according to the chromosome karyotype was defined according to a previous study 13 Fig. 1D), chi-square test/chi-square test with Bonferroni correction (Table 1), logrank test (Fig. 2), and Student's t test (Fig. 3) were used for statistical analysis. P values < 0.05 were considered to indicate statistical significance.
Determination of the serum concentrations of TPO and IL-6. Serum samples were prepared by centrifugation at 1000 g for 15 min at room temperature and stored in a − 80 °C freezer until examination. To determine the levels of human TPO and IL-6, ELISAs were performed with the Human Thrombopoietin Quantikine ELISA Kit (R&D Systems) and the IL-6 Human Instant ELISA Kit (Invitrogen), respectively, according to the manufacturer's instructions.