Transcription factor 3 is dysregulated in megakaryocytes in myelofibrosis

Abstract

Transcription factor 3 (TCF3) is a DNA transcription factor that modulates megakaryocyte development. Although abnormal TCF3 expression has been identified in a range of hematological malignancies, to date, it has not been investigated in myelofibrosis (MF). MF is a Philadelphia-negative myeloproliferative neoplasm (MPN) that can arise de novo or progress from essential thrombocythemia [ET] and polycythemia vera [PV] and where dysfunctional megakaryocytes have a role in driving the fibrotic progression. We aimed to examine whether TCF3 is dysregulated in megakaryocytes in MPN, and specifically in MF. We first assessed TCF3 protein expression in megakaryocytes using an immunohistochemical approach analyses and showed that TCF3 was reduced in MF compared with ET and PV. Further, the TCF3-negative megakaryocytes were primarily located near trabecular bone and had the typical “MF-like” morphology as described by the WHO. Genomic analysis of isolated megakaryocytes showed three mutations, all predicted to result in a loss of function, in patients with MF; none were seen in megakaryocytes isolated from ET or PV marrow samples. We then progressed to transcriptomic sequencing of platelets which showed loss of TCF3 in MF. These proteomic, genomic and transcriptomic analyses appear to indicate that TCF3 is downregulated in megakaryocytes in MF. This infers aberrations in megakaryopoiesis occur in this progressive phase of MPN. Further exploration of this pathway could provide insights into TCF3 and the evolution of fibrosis and potentially lead to new preventative therapeutic targets.

Plain Language Summary

What is the context?

• We investigated TCF3 (transcription factor 3), a gene that regulates megakaryocyte development, for genomic and proteomic changes in myelofibrosis.

• Myelofibrosis is the aggressive phase of a group of blood cancers called myeloproliferative neoplasms, and abnormalities in development and maturation of megakaryocytes is thought to drive the development of myelofibrosis.

What is new?

• We report detection of three novel TCF3 mutations in megakaryocytes and decreases in TCF3 protein and gene expression in primary megakaryocytes and platelets from patients with myelofibrosis.

• This is the first association between loss of TCF3 in megakaryocytes from patients and myelofibrosis.

What is the impact?

• TCF3 dysregulation may be a novel mechanism that is responsible for the development of myelofibrosis and better understanding of this pathway could identify new drug targets.

Keywords:

• Megakaryocyte
• myelofibrosis
• myeloproliferative neoplasms
• next-generation sequencing
• platelets
• TCF3

Introduction

Philadelphia-negative myeloproliferative neoplasms (MPN) are a group of clonal hematopoietic stem cell disorders characterized by the over-production of cells of the myeloid lineage. These comprise polycythemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF), categorized as pre-fibrotic or overt myelofibrosis (MF), as per the WHO guidelines.¹ The majority of MPNs arise from driver mutations in JAK2 (Janus kinase 2), CALR (calreticulin) or MPL (myeloproliferative leukaemia virus oncogene) genes, although 10–15% of ET and MF patients do not have a driver mutation and are classified as triple-negative.^2,3 The mechanism by which these driver mutations promote MPN is through constitutive activation of the signal-transduction pathways for hematopoiesis.² PV and ET patients generally have an indolent course, but up to 20% will transform to MF.^4–6 At this stage, the marrow has reduced capacity to produce blood cells due to the pathologic fibrosis. Treatment options are limited for MF patients (e.g. JAK2 inhibitors), and the median survival is less than six-years.^7,8

The accumulation of pathologic reticulin fibers corresponds with the local release of fibrogenic growth factors and pro-inflammatory cytokines.^9–11 However, the molecular drivers responsible for the fibrotic progression have yet to be fully characterized. Megakaryocytes, which are hyperplastic and the most morphologically abnormal cell in all MPN subtypes, are also implicated.¹⁰ In MF, these are most marked with tight clustering, paratrabecular localization with high nuclear:cytoplasmic ratios and pyknotic nuclear chromatin. They have dysregulated proliferative, apoptotic and epigenetic mechanisms and have acquired somatic mutations.^9,11–13 Platelets, the progeny of megakaryocytes, have also been reported to have abnormal function and transcriptomic profile in MF compared to the indolent phases of MPN and this is thought to be the consequence of genomic defects in the neoplastic megakaryocytes.^14–19

TCF3 (transcription factor 3) is a DNA transcription factor and a component of the pro-megakaryocytic transcriptional complex containing TAL1 (TAL bHLH transcription factor 1), GATA1 (GATA binding protein 1), LMO2, (LIM domain only 2) and LDB1 (LIM domain binding 1).^20–25 In particular, TCF3 binds with TAL1 to determine the megakaryocytic or erythroid fate of hematopoietic stem cells, in conjunction with GATA1. Specifically, TCF3-CDKN1A (cyclin dependent kinase inhibitor 1A; p21) expression increases during megakaryocytic differentiation from progenitor cells.²⁶ Further, lentiviral-mediated gene silencing of TCF3 in in vitro studies led to failure of megakaryocytic differentiation of CD34-positive cells.²⁶ Late-stage megakaryocyte development factors GFI1B (growth factor independent 1B transcriptional repressor) and AURKA (aurora A kinase) have also been reported to be upregulated and suppressed by TCF3, respectively.^27–30

Given the seminal role of TCF3 in megakaryocyte development, we hypothesized that dysregulation of this key molecule may underlie abnormalities in megakaryopoiesis in patients with MPN. Further, we postulate that this would be most significant in MF where there is the greatest perturbation of megakaryopoiesis. To assess this, we performed genomic analyses of TCF3 and its protein expression in megakaryocytes, as well as TCF3 gene expression in platelets.

Materials and methods

Patient recruitment

The project was approved by the Sir Charles Gairdner Osborne Park Health Care Group (2012-094, RGS-1867, RGS-1894, RGS-3807) and the University of Western Australia Human Research Ethics Committees (#RA/4/1/6566, #RA/4/1/9100, #RA/4/1/9009, #2022/ET000097) and in accordance with the Declaration of Helsinki. A total of 148 consenting patients with MPN (ET = 56, PV = 43, MF = 49) were recruited at diagnosis or follow-up, and 61 individuals of similar age and no history of MPN as controls were recruited. Patient characteristics are shown in Supplementary Table S1. Blood count, blood film, bone marrow examination and allele-specific polymerase chain reaction to determine the driver mutation were performed by PathWest Laboratory Medicine Western Australia (Nedlands, Australia). Bone marrow trephine biopsies were fixed, decalcified and processed into paraffin; hematoxylin and eosin as well as Gordon and Sweet reticulin staining were performed on 3 μm sections and reviewed by hematopathologists in accordance with the 2022 WHO Classification.¹

Immunohistochemistry

Tissue microarrays (TMAs) of 1.5 mm cores of each bone marrow biopsy (MPN and control marrows) were generated using the TMA Master tissue microarray (3DHISTECH, Budapest, Hungary). Sections were cut at 3 μm and then stained for TCF3 (Atlas Antibodies, Stockholm, Sweden). Immunohistochemical staining was performed on an automated BOND RX immunostainer (Leica Biosystems, Mount Waverley, Australia) after heat induced epitope retrieval with BOND Epitope Retrieval (ER2) Solution 2 (Leica Biosystems; pH 9), as per Malherbe et al.³¹ A diaminobenzidine (DAB) substrate BOND Polymer Refine Detection kit (Leica Biosystems) was applied to detect antigen binding followed by nuclear counterstaining with Mayer’s hematoxylin. Sections were independently assessed for megakaryocyte expression of TCF3 by two independent observers (R.J.C. and W.N.E.) using an Olympus BX53 light microscope at x400 magnification (Olympus Life Science Solutions, Macquarie Park, Australia) and imaged using a Pixera Pro600ES digital camera system and Viewfinder application (Pixera, San Jose, USA).

Megakaryocyte isolation, whole genome amplification and next-generation sequencing

To perform DNA mutation analyses, megakaryocytes were first isolated from approximately 1 mL of bone marrow aspirate collected into an EDTA anti-coagulated vacutainer (Greiner Bio-One, Frickenhausen, Germany) as per Guo et al.¹² Megakaryocyte DNA was isolated, amplified and sequenced as per Guo et al.¹² In brief, a human albumin density gradient centrifugation was used to enrich megakaryocytes from aspirated bone marrows. The megakaryocyte suspension was washed, cytocentrifuged onto a slide and stained with Mayer’s hematoxylin. Megakaryocytes were captured by laser microdissection using an Arcturus XT Laser Capture Microdissection System (Thermo Fisher Scientific, Scoresby, Australia) and stored at −80°C, as previously described.¹² Megakaryocyte-depleted bone marrow cells were collected from the gradient pellet and stored at −80°C until DNA was extracted using a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany).

Laser captured megakaryocytes were thawed at room temperature and lyzed as per protocol.³² DNA from megakaryocytes and megakaryocyte-depleted bone marrow cells underwent whole-genome amplification using the Illustra GenomiPhi V2 DNA Amplification Kit (GE Health Care, Chicago, USA).³² Amplified DNA was purified using Agencourt AMPure XP beads (Beckman Coulter, Brea, USA) and assessed for quantity and quality using a High Sensitivity DNA Kit (Agilent Technologies, Santa Clara, USA) on a 2100 Bioanalyzer (Agilent Technologies).

20 ng of DNA were used to prepare libraries using an Ion AmpliSeq custom panel (Thermo Fisher Scientific) and an Ion AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific), as per Guo et al.¹² Libraries were quantified on a 2100 Bioanalyzer using a High Sensitivity DNA Kit. Equimolar concentrations were pooled and underwent template preparation on an Ion Chef System (Thermo Fisher Scientific) using an Ion PI Hi-Q Chef Kit (Thermo Fisher Scientific) and sequenced on an Ion Proton Sequencer (Thermo Fisher Scientific) using an Ion PI Hi-Q Sequencing Kit and Ion PI v3 Chip (Thermo Fisher Scientific).

Torrent Suite Software v4.4 (Thermo Fisher Scientific) was used to process barcoded reads, perform base calling, align reads to the reference genome (human genome build 19), generate run metrics, analyze coverage and identify variants. The variants were annotated and filtered using the Ion Reporter Software v4.4 (Thermo Fisher Scientific), as per Guo et al.¹²

TCF3 DNA variant validation

Amplicons of regions containing the TCF3 variants were amplified from samples with sufficient DNA (10 ng) using Platinum II PCR MasterMix (Thermo Fisher Scientific) and custom primers (Thermo Fisher Scientific), detailed in Supplementary Table S2. Amplicons were purified using Agencourt AMPure XP beads (Beckman Coulter) and quantitated using a High Sensitivity DNA Kit on a 2100 Bioanalyzer. Purified amplicons were submitted along with sequencing primers (Integrated DNA Technologies, Singapore Science Park II, Singapore), listed in Supplementary Table S2, to the Australian Genome Research Facility (Perth, Australia) for Sanger Sequencing. Chromatograms were qualitatively assessed using FinchTV v1.4.0 (Geospiza Inc., Seattle, USA).

Platelet gene expression

To perform platelet transcriptomic expression analysis, blood was collected in EDTA anti-coagulated vacutainers (Greiner Bio-One) and processed within 72 hours. Platelets were isolated and assessed for purity and quantified as previously described.³³ RNA was extracted from platelets using a miRNeasy Mini Kit (QIAGEN) and quantified on a 2100 Bioanalyzer using an Agilent RNA 6000 Pico Kit (Agilent Technologies). Libraries were prepared using 5 ng of total RNA and an Ion AmpliSeq Transcriptome Human Gene Expression Kit (Thermo Fisher Scientific) as per Collinson et al.³³ The libraries were quantified on a 2100 Bioanalyzer using a High Sensitivity DNA Kit. Equimolar concentrations of libraries were pooled and loaded onto Ion PI v3 Chips using an Ion Chef Instrument. The chips were sequenced using an Ion PI Hi-Q Sequencing Kit on an Ion Proton Sequencer.

The Torrent Suite Software v4.4 was used to process barcoded reads, perform base calling, align reads to the reference transcriptome (human genome build 19), generate run metrics, and produce gene expression counts. Differential gene expression analyses and count normalization were performed using the DESeq2 package (R/Bioconductor; http://www.bioconductor.org/).^17,34 Significant differential expression was defined as having at least a ±two-fold change and adjusted p (padj) <0.05 (using the Benjamini-Hochberg adjustment).

TCF3 pathway analysis

Pathway analysis was performed using Ingenuity Pathway Analysis (IPA; https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis; Ingenuity Systems, Redwood City, USA) based on differentially expressed genes in platelets between MF and the indolent phase of ET and PV. The causal networks analysis was used to identify upstream regulators.³⁵ Network activation and inhibition were stratified to a z-score below −2 and above 2. Networks fulfilling the “bias” criteria were excluded. Minimum network-bias-corrected p-value of 10⁻⁴ was implemented as per Krämer et al.³⁵

Statistical analysis

The indolent phases of MPN (ET and PV) were stratified as one group (“ET/PV”) and compared with MF and controls. Statistical significance of age was assessed by one-way ANOVA follow by post-hoc Holm–Sidak test. Statistical significance of sex was assessed by Kruskal–Wallis one-way ANOVA followed by post-hoc Dunn tests. The percent of megakaryocytes expressing TCF3 in all MPN and control bone marrow trephines were analyzed using a Mann–Whitney U test. Control, ET/PV and MF patients were analyzed using a Kruskal–Wallis one-way ANOVA followed by Benjamini–Hochberg adjustment. Statistical significance was set at a p-value of <0.05.

Results

Megakaryocyte expression of TCF3

Immunohistochemical staining for TCF3 gave the expected positivity in myeloid and erythroid progenitor cells and lymphocytes in all bone marrow samples (Figure 1A–E).³⁶ In the marrow controls, 0–57.1% (mean = 13.3%) of megakaryocytes were TCF3-positive, showing weak nuclear expression. In the MPN samples, 0–83.7% (mean = 27.7%) of megakaryocytes were positive in the 68 ET/PV cases and 0–69.9% (mean = 21.2%) in the bone marrow cores of MF (n = 34). This gave an overall 1.92-fold (padj = 0.016) increase of TCF3-positive megakaryocytes in MPN marrows compared with controls. In MF, there was a 0.77-fold reduction in megakaryocytes expressing TCF3 compared with the ET/PV group (padj = 0.030) (Figure 1F).

Figure 1. Quantitative and qualitative assessment of TCF3 expression in megakaryocytes of MPN (ET/PV = 68, MF = 34) and control bone marrow trephines (n = 71). (A–E) Immunohistochemical staining showing TCF3 in bone marrow trephines of normal marrow and MPN subtypes. Black arrows indicate examples of TCF3-positive and white arrows TCF3-negative megakaryocytes. Positive erythroid, granulocyte and lymphoid cells are present in all cases. A) control bone marrow (x600); B) ET (x600); C) PV (x600); D) MF with distorted TCF3-negative megakaryocytes adjacent to trabecular bone (x400); E) MF that arose secondary to preceding ET. The interstitial megakaryocytes are positive and have a morphological appearance as seen in B). The TCF3-negative megakaryocytes are angulated with pyknotic nuclear chromatin, a feature of MF (x400). (DAB substrate; hematoxylin counterstain.) F) percentage of megakaryocytes expressing TCF3 in control, ET/PV and MF patients (Kruskal–Wallis one-way ANOVA analysis followed by Benjamini–Hochberg adjustment). Data shown as the mean ±95% confidence interval. *padj < 0.05, **padj < 0.01, ***padj < 0.001, ****padj < 0.0001.

We also assessed the localization of megakaryocytes within the marrow, their cytological characteristics and TCF3 immunohistochemical staining in the MPN samples. The most strongly stained megakaryocytes in the MPN were large, present in the interstitium of the marrow and had hyperlobated nuclei; these were frequent in ET/PV samples (Figures 1B, C). In MF, the TCF3-negative megakaryocytes were generally adjacent to trabecular bone and had the most abnormal morphology, being angulated with pyknotic nuclear chromatin (Figure 1D). Of interest was that in some post-ET/PV cases of MF there were both hyperlobated TCF3-positive megakaryocytes (as seen in ET/PV) and pyknotic-angulated TCF3-negative megakaryocytes (Figure 1E). These appearances were seen in marrow from patients with secondary MF that had transformed from preceding ET or PV.

TCF3 DNA mutations in Megakaryocytes and Megakaryocyte-depleted bone marrow

Having determined that there was reduced megakaryocytic TCF3 expression in MF, we proceeded to examine if TCF3 was mutated in megakaryocytes. To do this, we assessed mutations in both megakaryocytes and megakaryocyte-depleted bone marrow cells from each patient (n = 39). The samples were successfully sequenced with an average on target percentage of 97.6% and mean depth of 927 x. In 3 of 10 patients with MF, TCF3 mutations were identified in DNA extracted from megakaryocytes. The mutations, TCF3^Q500G, TCF3^L88V and TCF3^A602V were detected at frequencies of 21.9%, 9.6% and 9.1%, respectively, and were not detected in the non-megakaryocytic compartment. These mutations were validated using Sanger Sequencing. All three mutations were predicted to be disease-causing based on the SIFT and PolyPhen2 scores.^37,38 TCF3^Q500G and TCF3^A612V were also predicted to be disease-causing by MutationTaster2. TCF3^Q500G was predicted to alter protein function and splicing.^37,38 This predicted loss was also seen at the protein level, as per Figure 1D and 1E. No TCF3 mutations were detected in megakaryocytes extracted from ET/PV patients (n = 29). There were, however, two TCF3 frameshift-mutations (TCF3^V357fs and TCF3^G410fs), at variant allelic frequencies of 21.7% and 10.1%, in the megakaryocyte-depleted bone marrow cells in two of these patients with both predicted to be pathogenic (MutationTaster2).^37,38

Platelet gene expression and pathway analysis

Since megakaryocyte mutations were detected, and there was reduced TCF3 expression in megakaryocytes, we then explored whether there was altered gene expression. As the method of enriching and isolating primary megakaryocytes includes the use of chemicals and reagents that cause degradation of RNA, gene expression of primary megakaryocytes was unable to be performed. Instead, we performed transcriptomic next-generation sequencing of platelets isolated from blood samples.

The samples were successfully sequenced with an average of 10.5 × 10⁶ mapped reads at 95.6% valid reads, sufficient for this approach.^17,33,39 TCF3 gene expression was upregulated in platelets from MF patients (n = 26) compared to both ET/PV (n = 48) and controls (n = 10) by 4.91- (padj <0.0001) and 14.36-fold, (padj < 0.0001), respectively. There were no significant differences between ET/PV and controls (0.89-fold, padj = 0.833; Figure 2A). To further understand this unexpected upregulation of TCF3 gene expression in MF, we assessed genes that are direct downstream targets of TCF3 (i.e. CDKN1A, GFI1B and AURKA) and regulate late-megakaryocyte development.^{21,25,29,40,41} CDKN1A and GFI1B, which are positively regulated by TCF3, were significantly downregulated in MF platelets compared to ET/PV (0.58-fold, and 0.60-fold respectively; both padj < 0.0001). This was also seen when compared with the controls (CDKNIA; 0.66-fold, padj = 0.006, GFI1B; 0.49-fold, padj < 0.0001) (Figures 2B, 2C). In contrast, AURKA, which is suppressed by TCF3, was significantly upregulated in MF compared with ET/PV (19.2-fold, padj < 0.0001) and controls (34.4-fold, padj < 0.0001) (Figure 2D).

Figure 2. Normalized counts of A) TCF3 and direct targets B) CDKN1A, C) GFI1B and D) AURKA in control (n = 10), ET/PV (n = 48) and MF (n = 26) patients. Normalized counts were analyzed using a one-way ANOVA followed by Benjamini–Hochberg adjustment. Data shown as mean ± 95% confidence interval. *padj < 0.05, **padj < 0.01, ***padj < 0.001, ****padj < 0.0001.

Pathway analysis was performed to provide insight on changes that may be occurring in the parental megakaryocytes based on platelet gene expression changes. Overall, a total of 2,846 genes were differentially expressed in platelets from patients with MF compared to ET/PV and used for pathway analysis. Six networks were identified as master upstream regulators using the causal networks analysis tool within IPA (Suplementary Table S3).³⁵ Of these six, the TCF3-regulated network was the only one with an overlap with the megakaryocyte mutation data. TCF3 was predicted to be inhibited (z = −2.941) and statistically significant (Fisher’s exact test: p = 3.17 × 10⁻¹⁷).³⁵ The TCF3 network includes 26 other genes, all of which have previously been associated with MF (Figure 3).^{17,18,42–48}

Figure 3. Genes involved in the downstream effects of TCF3 inhibition. Network activation and inhibition was stratified to a z-score below −2 and above 2. Networks fulfilling the “bias” criteria were excluded. Minimum network-bias-corrected p-value of 10⁻⁴. Genes in orange are predicted to be activated and in blue are predicted to be inhibited. Red indicates genes that have increased transcript expression in platelets from patients with ET/PV (n = 48) and MF (n = 26) patients.

Discussion

This analysis of megakaryocytes and platelets in MF has shown that cell-specific mutations and genetic dysregulation are associated with reduced TCF3 expression and aberrant megakaryopoiesis. This is linked with the abnormal megakaryocyte localization and morphology within the bone marrow. We identified three novel TCF3 mutations in megakaryocytes, all of which are predicted to result in loss of function of TCF3. This is supported by gene expression analysis of platelets which revealed a transcriptomic profile that is consistent with loss of TCF3 function in parental megakaryocytes in MF.

The TCF3 mutations detected in megakaryocytes were novel and have not been reported in MPN or other diseases previously. According to the Genome Aggregation Database (gnomAD; https://gnomad.broadinstitute.org/) and Protein Data Bank in Europe-Knowledge Base (PDBe-KB; https://www.ebi.ac.uk/pdbe/pdbe-kb/), TCF3^A612V is in proximity to splice region c.1823 and located in a functional helix-loop-helix DNA-binding domain.^49–51 TCF3^L88V and TCF3^Q500Gare also located near confirmed frameshift mutations (p.Ser91HisfsTer29, p.His497_Arg498del, p.His505GInfsTer100).⁵¹ This suggests that mutations in these regions may interrupt the ability of TCF3 to bind to the megakaryocytic DNA transcriptional complex to form the mature complex required for its regular function. It was interesting to note that two TCF3 frameshift mutations were detected in the megakaryocyte-depleted bone marrow cells of two ET patients. Neither patient showed morphological evidence of bone marrow fibrosis. This suggests that loss of TCF3, specifically in the megakaryocyte lineage, may underpin dysregulated megakaryopoiesis and contribute to fibrotic progression in MPN.

Although megakaryocyte-restricted mutations in the exons of TCF3 were not detected in all patients with MF, it remains possible that mutations in non-coding regulatory regions (e.g. promoter sites) or changes at the epigenetic level could also contribute to loss of TCF3. For example, epigenetic regulators such as EZH2, SUZ12 and DNMT3A, have been shown to suppress TCF3 through targeted methylation of lysine-27 of histone-3, which may impact TCF3 function.^52–56 Further insight into which promoter regions are critical for regulating TCF3 expression and profiling epigenetic changes are needed to fully characterize alternate mechanisms of TCF3 loss in megakaryocytes in MF.

The observed increase of TCF3 expression in platelets from patients with MF was unexpected. This was also observed for other transcripts, including GATA1, LMO2, and LDB1 (all co-factors of TCF3). In particular, GATA1 is upregulated in platelets in MF compared to ET/PV, which is in contrast to reported downregulation in megakaryocytes in MF.^57–60 Megakaryocyte RNA packaging into nascent platelets was thought to be a strictly linear relationship.^61–64 However, more recent studies show it is a more selective process, meaning the loss of TCF3 in megakaryocytes could be due to the over-packaging of TCF3 RNA into platelets in MF.^61,63,65,66 Alternatively, it is possible that a mutation in TCF3 could result in a nonfunctional protein being made without impacting on gene expression.^67,68 Further research into how RNA is packaged into platelets by megakaryocytes is required to understand the functional implication of TCF3 upregulation in platelets. Although TCF3 expression is increased in MF platelets, the expression of downstream targets CDKN1A, GFI1B and AURKA were significantly dysregulated in a manner that was consistent with loss of TCF3. This was supported by pathway analysis which identified TCF3 to be a master upstream regulator and predicted to be inhibited. Taken together, this suggests that there is a loss of TCF3 in the parental megakaryocytes, which is in agreement with the mutations that were detected.

TCF3 protein expression was specifically reduced in megakaryocytes with morphological and topographical abnormalities that are characteristic of MF.¹ This included paratrabecular localization, angular appearances and pyknotic nuclear chromatin.^11,69 This is concordant with previous studies using primary human CD34-positive cells and haploinsufficient Tcf3^+/- mice, which showed that loss of Tcf3 resulted in similar abnormalities in megakaryocytes.^21,26 Both models supported the notion that decreased TCF3 expression leads to impaired megakaryocytic maturation in vivo and is associated with the characteristic megakaryocytic morphological features of MF.^13,26,70,71

Although TCF3 is a known regulator of megakaryopoiesis, the stage of differentiation at which this is most prominent remains unknown. However, the data from the present study demonstrates that there is protein expression in normal mature megakaryocytes (Figure 1). TCF3 levels are higher than the controls likely due to “residual” ET/PV TCF3-positive megakaryocytes. TCF3-negative megakaryocytes in ET/PV were heterogeneous in morphology, being either physiologically normal (i.e. pertaining to controls) or pleomorphic. TCF3-negative megakaryocytes in MF patients suggest they may be in a pre-apoptotic state, as immature and necrotic megakaryocytes are typically surrounded by fibrotic areas of the marrow.^72,73

Additional insights into biological changes that may be occurring in megakaryocytes and platelets in MF were obtained from the pathway analysis.^11,13 For example, TCF3 downregulation is shown to be causal to AURKA over-expression in platelets and upregulation of TGFβ1 (Figure 3). TCF3 downregulation likely perturbs megakaryocyte maturation in MF patients through the upregulation of AURKA. AURKA protein expression has been reported to be increased in megakaryocytes of myelofibrotic murine models.¹³ Drugs targeting AURKA ameliorate fibrosis and restore megakaryocyte polyploidization.^11,13 Aberrant megakaryocytic expression of inflammatory cytokines such as TGFβ in GATA1_low mice appear to promote fibrosis. TGFβ is implicated in driving bone marrow fibrosis through AURKA over-expression, which is also upregulated in platelets from MF patients and a component of the TCF3 network.^13,42,74 Treatment with AURKA-inhibitor Alisertib (MLN8237) clinically has shown potent anti-fibrotic properties, restored polyploidization and reduced TGFβ cytokine levels.^11,13

Loss of TCF3 in megakaryocytes has previously been reported in a study using primary cell culture models.²⁶ Expression of TCF3, as well as CDKN1A, was shown to be decreased in megakaryocytes following Anagrelide treatment.²⁶ In another study, GFI1B and TCF3 co-factor TAL1, were also reported to be decreased in CD34+ cells after treatment with Anagrelide.⁷⁵ Interestingly, a retrospective analysis of survival in ET patients treated with Anagrelide reported decreased MF-free survival.⁷⁶ Together, these studies suggest that impaired megakaryopoiesis and TCF3 function are associated with Anagrelide treatment, providing a potential link between loss of TCF3 and fibrotic progression. However, the effect of Anagrelide on TCF3 downregulation has been disputed by other reports, which suggest that the effect is concentration-dependent, by showing no evidence of TCF3 mRNA reduction at 1 μM and 50 μM.^58,77 Megakaryocytes treated with Anagrelide over time at a physiological concentration would address whether Anagrelide-treated megakaryocytes have decreased TCF3 expression.

Technical challenges associated with the analysis of primary megakaryocytes is well established and a major challenge for the field. In comparison with other hematopoietic cells, megakaryocytes are rare (<1% of cells) and much larger (20–100 μm in diameter), and thus are unable to be isolated using standard flow cytometry-based techniques. This is further complicated by difficulties in aspirating fresh marrow, especially with established fibrosis. In situ hybridization, to assess megakaryocyte gene expression in biopsy specimens, is challenging to perform on bone marrow sections due to the formalin fixation and acid decalcification which reduces RNA integrity. Although past studies of megakaryocyte gene expression have utilized cell lines and megakaryocytes derived from CD34+ cells in vitro, assessment of fully mature primary megakaryocytes has been inconsistent.^78–80 These technical limitations made direct analysis difficult and resulted in the need to analyze platelets as a surrogate to primary megakaryocyte RNA.

The genomic and proteomic data presented here show that there is significant dysregulation in TCF3 regulation in megakaryocytes in MPN, which is most marked in MF. This work demonstrates that there are defects in TCF3 in megakaryocytes in MF and these correlate with the classical morphological abnormalities. Whether TCF3 is directly involved in the fibrotic progression remains uncertain. However, TCF3 is part of a network of genes, some of which have already been implicated in MF in clinical studies. This suggests that TCF3 may be a key driver of dysregulated megakaryopoiesis in MF. These findings may open new therapeutic avenues that target genes in the TCF3 network and reduce or prevent bone marrow fibrosis in MPN.

Author contributions

R.J.C., B.B.G. and M.D.L. designed and performed the research. R.J.C. and B.B.G. analyzed the data. B.M., L.W. and D.B. isolated platelets. B.B.G, B.M., L.W. and K.A.F. isolated megakaryocytes. ZY.N., H.C., R.H., C.S.G., J.A.J.M., and M.F.L. recruited patients and analyzed the clinical data. H.C., R.J.C., W.N.E. and ZY.N. reviewed and analyzed the bone marrow trephines. R.J.C. and W.N.E reviewed the immunohistochemistry-stained sections. R.J.C. performed the pathway analysis. All authors contributed to drafting the paper and had final approval of the submitted and published versions.

Acknowledgments

We acknowledge Alan Morling, Brett Bettridge and Aaron Osborn from PathWest, Royal Perth Haematology, Sir Charles Gairdner Hospital Haematology and Rockingham General Haematology for their assistance with sample collection.

Disclosure statement

The authors declare that they have no competing interests.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Funding

The work was supported by grants from the Cancer Council Western Australia (APP1065493 and CCWA ECCRoY 2016), the University of Western Australia (RA/1/1200/883), the MPN Research Foundation and the Ruby Red Foundation (RA/1/3246/2 and RA/1/3472/2), Ann Helene Toakley Charitable Endowment (IPAP2018/1802, IPAP2019/0291) and Raine Medical Research Foundation (RPG043-19). R.J.C. is supported by Research Training Program Scholarship provided by the University of Western Australia. B.B.G. is supported by the Gunn Family National Career Development Fellowship for Women in Haematology from Snowdome Foundation and Maddie Riewoldt’s Vision. M.D.L. was an ISAC Marylou Ingram Scholar. ZY.N is supported by a Fellowship of the Western Australia Cancer and Palliative Care Network Health from the Western Australia Health Department. H.C. was supported by a Fellowship of the Western Australia Cancer and Palliative Care Network Health from the Western Australia Health Department. J.A.J.M. received financial support from the Royal College of Pathologists of Australasia.