Kristin Holl, Nicolas Chatain, Susanne Krapp, Anja Scheufen, Nathalie Brock, Steffen Koschmieder, Daniel Moreno-Andrés
Abstract
Myeloproliferative neoplasms (MPNs) encompass a diverse group of hematologic disorders driven by mutations in JAK2, CALR, or MPL. The prevailing working model explaining how these driver mutations induce different disease phenotypes is based on the decisive influence of the cellular microenvironment and the acquisition of additional mutations. Here, we report increased levels of chromatin segregation errors in hematopoietic cells stably expressing CALRdel52 or JAK2V617F mutations. Our investigations employing murine 32D MPL and human erythroleukemic TF-1MPL cells demonstrate a link between CALRdel52 or JAK2V617F expression and a compromised spindle assembly checkpoint (SAC), a phenomenon contributing to error-prone mitosis. This defective SAC is associated with imbalances in the recruitment of SAC factors to mitotic kinetochores upon CALRdel52 or JAK2V617F expression. We show that JAK2 mutant CD34 + MPN patient-derived cells exhibit reduced expression of the master mitotic regulators PLK1, aurora kinase B and PP2A catalytic subunit. Furthermore, the expression profile of mitotic regulators in CD34 + patient-derived cells allows to faithfully distinguish patients from healthy controls, as well as to differentiate primary and secondary myelofibrosis from essential thrombocythemia and polycythemia vera. Altogether, our data suggest alterations in mitotic regulation as a potential driver in the pathogenesis in MPN.
Introduction
Philadelphia chromosome-negative myeloproliferative neoplasms (Ph-neg. MPNs) are a heterogeneous group of clonal hematopoietic disorders clinically subdivided into polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF) [1]. The mutations in the genes of the Janus kinase 2 (JAK2), calreticulin (CALR), or the thrombopoietin receptor (TPOR/MPL) are driver mutations of these diseases [2, 3]. Their occurrence and variant allele frequency, together with specific bystander mutations, determine the clinical features, disease severity, and whether these disease evolve with dismal prognosis and decreased survival[3–6], towards secondary myelofibrosis (SMF), secondary acute myeloid leukemia and/or secondary solid tumors [7, 8]. The main pathogenic molecular signaling event of Ph-neg. MPNs is the constitutive activation of JAK2- STAT-dependent signaling pathways by mutations in CALR, JAK2 or the MPL receptor [2, 3]. Yet, noncanonical mechanisms of mutant JAK2[9], and CALR[10] have been recently linked to aspects of the disease pathology. However, the molecular mechanisms of the phase transition towards acute disease states are poorly defined [2, 5].
In contrast to other myeloid neoplasms such as primary acute myeloid leukemia (AML) [11, 12], myelodysplastic syndromes [13, 14] or chronic myeloid leukemia [15, 16], the cytology and molecular status of mitosis in Ph-neg. MPNs has not been studied in detail. However, karyotype abnormalities likely caused by chromatin segregation defects due to defective mitosis are present in up to 5% of ET, 20% of
PV, and 57% of PMF cases at the time of diagnosis [4, 17, 18] and accumulate over time, especially at blast-phase transformation[19–22] and frequently are associated with unfavorable prognosis and decreased survival[23–25]. Therefore, mitotic defects induced by driver Ph-neg. MPN mutations could play a role in the pathological mechanisms and contribute to the phase transition. Mitosis is tightly regulated by the crosstalk between the kinases Aurora B, CDK1 (cyclic dependent kinase 1)-Cyclin B1, and Polo-Like Kinase 1 (PLK1), and the protein phosphatase PP2A, as well as by the spindle assembly checkpoint (SAC) [26, 27]. The latter constitutes a protein network recruited to chromosome kinetochores to ensure proper chromosome-spindle attachments and accurate chromatin segregation. It includes several evolutionarily conserved proteins, like BubR1, Aurora B, MAD1, MAD2, MPS1, CDC20 and kinesin motor proteins, such as CENP-E, which are required for precise SAC function [26, 28]. Precise maintenance of the molecular equilibrium in gene expression and accurate subcellular positioning of these mitotic regulators play a critical role in preserving chromosome integrity and ensuring the stability of the karyotype [16, 28–30]. Consequently, defects in mitotic regulation promote chromosome instability (CIN), acquisition and evolution of heterogeneous karyotypes, inflammation, and epigenetic dysregulation. All these pathological mechanisms are linked to the malignant transformation in many solid cancers [31–33]. Similarly, hematological malignancies [16] such as AML [11, 12] and myelodysplastic syndromes [13, 14], show defects or dysregulation in crucial mitotic factors linked to CIN and heterogeneous karyotypes. Here, we have analyzed the mitotic cytology in murine and human cells stably expressing CALRdel52 or JAK2V617F and found error-prone mitosis. The examination of the molecular status of key mitotic regulators suggests defective SAC function. Also, CD34 + Ph-neg. MPN patient cells display differential expression profiles of a subset of important mitotic regulators, including the SAC components BUB1, MAD2L1, INCENP, CDC20, CDK1, PLK1 and Aurora A/B.
Results
CALRdel52 and JAK2V617F 32D MPL cells have a stresssensitive and error-prone mitosis To investigate chromatin segregation and the duration of mitosis, we performed long term live-cell imaging of murine 32D MPL cells (Fig. 1A) for a duration of 20 h followed by image analysis. In comparison with control 32D MPL (EV) cells, 32D MPL cells transduced with CALRdel52 or JAK2V617F showed a slight and non-significant increase in the numbers of chromatin bridges and lagging chromosomes (Fig. 1B). In contrast, the percentage of telophase micronuclei is significantly increased in the JAK2V617F mutant cell (p < 0.008, Fisher´s exact test). The occurrence of all three kinds of chromatin segregation errors further increases significantly in comparison to EV when DNA damage is induced with the chemotherapeutic agent doxorubicin [34] (p < 0.05, Fisher´s exact test), or SAC malfunction with the antimitotic drug NMS-P715 [35] (p < 0.02, Fisher´s exact test), which inhibits the checkpoint kinase MPS1 (Fig. 1B). The average mitotic timing in untreated cells or upon treatment with doxorubicin or MPS1 inhibitor is similar between mutants and control (EV) transfected cells (Supplementary Fig. S1). As expected, treatment the SAC inhibitor NMS-P715 reduced the mitotic timing with respect to untreated samples (Supplementary Fig. S1). These data suggest that mitosis in CALRdel52 and JAK2V617F mutant 32D MPL cells is stress sensitive. A weakened SAC contributes to error-prone mitosis in murine 32D
MPL CALRdel52 and JAK2V617F cells JAK2V617F [9, 36] and CALRdel52 [37] mutations have been linked to increased ROS production. In
addition, JAK2V617F is also linked to replication stress[38], and to lower p53 levels, a factor which is critical for the DNA damage response[39]. Replication stress and DNA damage signalling pathways, together with mitotic dysregulation, are well-known sources of karyotype aberrations such as aneuploidy, CIN, and genomic instability [40]. Therefore, we investigated whether the observed error-prone mitosis in CALRdel52 and JAK2V617F cells after doxorubicin treatment could be due to an altered response to DNA damage or replication stress.
First, we tested whether CALRdel52 or JAK2V617F transduced 32D MPL cells show increased levels of double-strand breaks during entry into mitosis as compared to control (EV) cells by immunofluorescence staining of γ-H2AX (H2AX S139ph) a well-described marker for DNA damage [41]. Visual inspection of γH2AX foci in untreated prometaphase EV and mutant cells revealed similar low levels of DNA damage. As
expected, the number of γ-H2AX foci increased to a similar extent in control (EV) and mutant cells after doxorubicin treatment (Supplementary Fig. S2A). In agreement with the literature [39], p53 basal levels in untreated JAK2V617F cells were much lower than in control EV cells or CALRdel52 (Supplementary Fig. S2B, Supplementary Figure S8). After doxorubicin treatment, p53 levels increased more than four-fold in all the cell lines. These results indicate that DNA damage is similarly induced by doxorubicin in control (EV) and mutant cells during mitotic entry and all cells showed a comparable functional stabilization of p53 after genotoxic stress. To test whether defects in SAC could contribute to the observed increase of chromatin segregation errors in CALRdel52 and JAK2V617F mutant cells, we challenged them with the spindle poison nocodazole. Cells with a weakened or defective SAC escape faster from the nocodazole induced mitotic arrest [29, 42]. Nocodazole treatment induced mitotic arrest in all three cell lines, but compared to control (EV) cells (257 ± 45 min), CALRdel52 (210 ± 36 min, p < 0.02 one-way ANOVA) or JAK2V617F (186 ± 35 min, p < 0.0001) transduced 32D MPL cells showed significantly shorter mitotic arrest and faster mitotic exit (Fig. 2A, B). The outcomes of mitotic arrest induced with microtubule inhibitors are diverse among cancer and normal cell lines due to the different pathways they induce [43, 44]. The current ‘competing networks-threshold’ model proposes that cell fate determination of – either cell death or extended survival – hinges on which of the two thresholds is reached first: either the activation of pro-apoptotic caspases or the degradation of cyclin B1 leading to mitotic slippage [43, 45]. To discriminate between these two options, we directly analyzed the fate of individual cells under mitotic arrest. The fraction of cells with mitotic death after mitotic arrest upon nocodazole treatment is very low (< 10%) and without significant differences between control (EV), CALRdel52, and JAK2V617F cells (two-way ANOVA with Dunnet post test) (Fig. 2C). In all conditions the arrested cells mostly escaped mitotic arrest by mitotic slippage without significant differences between the different cell types (two-way ANOVA with Dunnet post test) (Fig. 2C).
It has been proposed that murine cells are naturally more resistant than human cells to mitotic poisons due to the presence of clearance systems [46]. Therefore, we determined whether cyclin B1 accumulation was different between control (EV) and CALRdel52 or JAK2V617F cells after 3 hours of nocodazole treatment, which is the lower average time the cells (JAK2V617F cells, Fig. 2A) spend in mitotic arrest
before they undergo slippage. Cyclin B1 accumulated similarly in all three cell lines (Fig. 3A). Once accumulated, cyclin B1 was degraded faster in cells expressing CALRdel52 and JAK2V617F mutations (Fig. 3B; Supplementary Fig. S3; Supplementary Fig. S6) suggesting that a weakened SAC contributes to the error-prone mitosis in these cells.