Altered erythropoiesis via JAK2 and ASXL1 mutations in myeloproliferative neoplasms

Abstract

Myeloproliferative neoplasms (MPN) are driven by hyperactivation of JAK-STAT signaling but can demonstrate skewed hematopoiesis upon acquisition of additional somatic mutations. Here, using primary MPN samples and engineered embryonic stem cells, we demonstrate that mutations in JAK2 induce a significant increase in erythroid colony formation, whereas mutations in additional sex combs-like 1 (ASXL1) led to erythroid colony defect. RNA-sequencing revealed upregulation of protein arginine methyltransferase 6 (PRMT6) induced by mutant ASXL1. Furthermore, genetic perturbation of PRMT6 exacerbated MPN disease burden including leukemic engraftment and splenomegaly in patient-derived xenograft models, highlighting a novel tumor-suppressive function of PRMT6. However, augmented erythroid potential and bone marrow human CD71+ cells following PRMT6 knockdown were reserved only to primary MPN samples harboring ASXL1 mutations. Lastly, treatment of CD34+ hematopoietic/stem progenitor cells with PRMT6 inhibitor, EPZ020411 induced expression of genes involved in heme metabolism, hemoglobin, and erythropoiesis. These findings highlight interactions between JAK2 and ASXL1 mutations and a unique erythroid regulatory network in the context of mutant ASXL1.

Graphical Abstract

Introduction

JAK2 V617F is the most frequent driver mutation found in myeloproliferative neoplasms (MPNs), in approximately 50–60% of myelofibrosis (MF)¹. Mutations in additional sex combs-like 1 (ASXL1), a polycomb chromatin-binding epigenetic regulator, often co-occur with JAK2 V617F and are associated with decreased survival and increased risk of transformation to secondary acute myeloid leukemia^2,3. How mutant ASXL1 contributes to the MPN disease phenotype and confers poor prognosis is not fully understood. Controversy remains as to whether ASXL1 mutations found in patients confer a loss of function, gain of function, and/or dominant negative phenotype^4–6. Additionally, Asxl1 mutant knock-in mouse models present with a relatively modest phenotype, following a very long latency period^7–9. A human model system has the potential to provide a useful tool to understand how these mutations affect ASXL1 function. In this study, we sought to investigate the impact of ASXL1 and JAK2 V617F mutations alone and in combination within distinct hematopoietic compartments and disease progression.

Methods

Ethics statement

Primary samples were obtained according to a protocol approved by the Washington University Human Studies Committee (WU no. 01–1014). In vivo procedures were conducted in accordance with the Institutional Animal Care and Use Committee of Washington University (no. 20–0463).

Generation and differentiation of pluripotent stem cells

JAK2, ASXL1, and double knock-in mutants in the H1 ES cell line (WA01, WiCell) were generated via CRISPR-mediated gene editing at the GEiC at Washington University in St. Louis. Cells were electroporated with the gRNA/Cas9 ribonuclease protein (RNP) complex (200 pmol gRNA + 80 pmol SpCas9). Mutated cells were confirmed with junction PCRs following nucleofection and then single-cell sorted. Targeted deep sequencing analysis using primer sets specific to the target regions of interest were utilized to screen the single cell clones. Hematopoietic differentiation was performed as previously described¹⁰. In brief, cells underwent stepwise differentiation and CD34+ hematopoietic stem/progenitor cells were then plated in methylcellulose.

Synthetic gRNA (IDT):

JAK2(V617F): 5’- AATTATGGAGTATGTGTCTG −3’

ASXL1(p920T): 5’- CCAACCTGGGGCTCAACAGA −3’

ASXL1(dupG): 5’- CGGCCACCACTGCCATCGGA −3’

Colony formation assays

Lineage-negative CD34⁺ cells were sorted and plated in triplicate in MethoCult H4034 (STEMCELL Technologies) in the presence or absence of drug treatment and enumerated after 2 weeks.

Flow cytometry

Cells were suspended in live cell buffer, stained and data was collected on an LSR Fortessa (BD Biosciences). Analysis was performed using FlowJo version 10.0 (BDBiosciences). Utilized antibodies:

Anti-human CD34-PE/Cy7 (BD #348791), anti-human CD43-FITC (BD #555475), anti-human CD73-PE (BD #550257), anti-human CD184-APC (BD #555976), anti-human CD34-APC (BD #340441), anti-human CD45-V450 (BD #642275), anti-human CD45-APC (BioLegend #368512), anti-human CD71-APCcy7 (BioLegend #334110), and anti-mouse CD45-BV605 (BioLegend #103139).

qRT-PCR

Quantitative reverse transcription polymerase chain reaction (RT-qPCR) was performed on CD34+ cells following PRMT6 knockdown with PRMT6 expression normalized to ACTB.

PRMT6 Forward: 5’- TCTGGTTCCAGGTGACCTTC-3’

PRMT6 Reverse: 5’- AGGTAGAGGAGCGCCTGTTT-3’.

ACTB Forward: 5’-GCATGGAGTCCTGTGGCAACCACG-3’

ACTB Reverse: 5’-GGTGTAACGCAACTAAGTCATAG-3’.

RNA-sequencing and gene-set analyses

Genetically engineered ES cells underwent stepwise differentiation and CD34+ hematopoietic stem/progenitor cells were collected and RNA was isolated for RNA-seq. Sample preparation, library construction, sequencing, and data analysis was performed as previously described¹¹. Similar RNA isolation and sequencing protocols were utilized for CD34+ primary MPN cells treated with EPZ020411 for 6 hours. Enrichment analysis was performed using https://www.gsea-msigdb.org/gsea/msigdb.

PRMT6 RNA-seq and mutational data of CD34+ cells from the WashU Cohort were accessed from source publication¹².

Patient-derived xenograft (PDX) models

PDX models were established as previously described^13–15. In brief, CD34⁺ cells were isolated by bead purification and transduced with two shRNAs targeting PRMT6 (TRCN0000034686, TRCN0000034688) or scrambled control. 100,000 transduced cells were then transplanted into sub-lethally irradiated NSGS mice. Two weeks post-transplant and until 12 weeks, PB was collected, and engraftment was evaluated by flow cytometry. At endpoint, BM was collected and analyzed by flow cytometry and spleens were weighed and normalized to mouse body weight. Mouse tibias and spleens were extracted at endpoint and stained with hematoxylin and eosin (H&E).

Statistical analysis and figure schematics

Statistical analysis was performed using GraphPad Prism version 9 (La Jolla, CA). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Figure schematics were created with BioRender.com.

Results and discussion

MPN patients and mouse models harboring ASXL1 mutations or loss-of-function have been described to have low red blood cell (RBC) and hemoglobin (Hb)^3,16–19. To evaluate potential causes, we first performed colony formation assays (CFA) on sorted CD34+ hematopoietic stem/progenitor cells (HSPCs) from MPN patients carrying JAK2 mutations alone (wildtype ASXL1) or dual JAK2/ASXL1 mutations, and from healthy bone marrow (NBM) donors. There was a significant increase of erythroid colonies indicated by burst forming uniterythroid colonies (BFU-E) from JAK2-mutant patients whereas JAK2/ASXL1 mutants demonstrated an erythroid colony defect (Fig. 1A, B). Due to 1) the relative scarcity of MPN patients harboring ASXL1 mutations devoid of JAK2 mutations, and 2) possible contribution by other non-identified mutations, we then utilized CRISPR-Cas9 to generate isogeneic models using human embryonic stem cells (hESCs; H1 line). We knocked-in the JAK2 V617F mutation, two MPN patient-specific ASXL1 exon 12 hotspot mutations (P920Tfs*4 and G646Wfs*12), or their combination of JAK2 and ASXL1 (double knock-in; DKI) and differentiated stem cells into CD34+ CD45+ progenitors over a two-week period (Fig. 1C, D), which were then seeded for CFAs. Consistent with our patient CD34+ CFAs, compared to wildtype (WT), JAK2-mutant cells generated more erythroid colonies while single ASXL1-mutants showed the opposite phenotype (Fig. 1E). DKI cells showed restoration of erythroid colony potential to WT levels thereby demonstrating that mutations in JAK2 and ASXL1 inflict opposing regulation of erythropoiesis.

Figure 1. Erythropoiesis defect in ASXL1-mutant MPN.

A) CD34+ colony assay from unique MPN donors carrying JAK2 (n = 7) or JAK2/ASXL1 (n= 8) mutations, and from healthy bone marrow donors (NBM; n = 5). B) BFU-E summary data from (A). Statistics were assessed by two-tailed Student’s t-test. C) Bi-axial flow cytometry plots showing differentiation of different engineered ES lines. On day 8 of differentiation CD34+CD43-CD184-CD73- hemogenic endothelium were sorted from each genotype. D) CD34+CD45+ definitive progenitors were sorted on day 15 to utilize for CFU and RNA sequencing assays. E) Colony assay showing BFU-E proportions in engineered ES lines, including the ASXL1 P920Tfs*4 and G646Wfs*12 mutations. WT (n = 10), JAK2 (n = 9), ASXL1 P920Tfs*4 (n = 5), ASXL1 P920Tfs*4 DKI (n = 5), ASXL1 G646Wfs*12 (n = 3), and G646Wfs*12 DKI (n = 3). Statistics were assessed by two-tailed Student’s t-test. F) Total colonies identified as BFU-E, CFU-M, or CFU-GM in differentiation experiments in engineered ES lines with ASXL1 mutations from E). 5 paired control and ASXL1 P920Tfs*4 experiments. 3 paired WT control and ASXL1 ASXL1 G646Wfs*12 experiments. G)Total colonies identified as BFU-E, CFU-M, or CFU-GM in representative differentiation experiments in engineered ES lines with JAK2, ASXL1, or DKI mutations. H) Gene set enrichment analysis was performed following RNA-sequencing of WT, JAK2, ASXL1 P920Tfs*4, and DKI at day 31 of differentiation. Top 500 genes upregulated and downregulated assessed by relative mRNA expression change in ASXL1 P920Tfs*4 compared to wildtype were analyzed. I) Expression heatmap of individual downregulated genes identified in the Hallmark “Heme metabolism” dataset. Expression changes were normalized to the wildtype. J) Expression heatmap of protein arginine N-methyltransferase family genes. Expression changes were normalized to the wildtype. K) PRMT6 mRNA expression following RNA-seq analysis of CD34+ cells from 24 MF patients from the WashU Cohort. ASXL1 WT (n = 17) ASXL1-mutant (n = 7). # denotes patients who did not have next-generation gene panel analysis at the time of sample collection but were confirmed to have an ASXL1 mutation at a later sample timepoint. Statistics were assessed by two-tailed Student’s t-test.

The ASXL1 mutations showed some alteration of myeloid macrophage and granulocyte-macrophage (CFU-M and CFU-GM) colonies as well, whereas JAK2 mutants showed a mild increase in total numbers (Fig. 1F, G).

We then performed RNA-seq on differentiated CD34+ HSPCs. Enrichment analysis using the Hallmark gene set revealed a profound upregulated pathway in epithelial to mesenchymal transition (EMT) in ASXL1-mutants compared to WT (Fig. 1H). Heme metabolism was a key downregulated pathway in ASXL1-mutants compared to WT, including erythroid transcription factor KLF1 (Fig. 1H, I). Although JAK2-mutants demonstrated a relative upregulation of these heme metabolism genes, expression patterns were similar between ASXL1-mutants and DKI. We also investigated other effectors with potential to regulate erythropoiesis, such as members of protein arginine methyltransferase (PRMT) family. Of these effectors, differential expression patterns between the genotypes yielded PRMT6 to be upregulated in ASXL1-mutants, downregulated in JAK2-mutants, and in-between for DKI (Fig. 1J). We then verified upregulated expression of PRMT6 mRNA by RNA-seq of CD34+ cells from the spectrum of MPN patients from the WashU Cohort¹² in ASXL1 mutant MF samples compared to wildtype ASXL1 (Fig. 1K). These data support a role of mutant ASXL1 in the regulation of PRMT6.

A previous report demonstrated that PRMT6 represses erythroid genes and colony formation of CD34+ HSPCs²⁰. To further assess the role of PRMT6 in the MPN setting in the context of JAK2 and ASXL1 mutations, we knocked-down PRMT6 utilizing two independent shRNAs in CD34+ MPN cells from two JAK2/ASXL1-mutant (692345, 652054) and two JAK2-mutant (551599, 829376) patients (Fig. 2A, B). PRMT6 knockdown (KD) conferred greater erythroid colony formation in both JAK2/ASXL1-mutants but not in the JAK2-mutants devoid of ASXL1 mutations (Fig. 2C). In parallel, we established patient-derived xenograft (PDX) models by transplanting PRMT6-KD CD34+ cells into NSGS mice to assess the role of PRMT6 in MPN pathogenesis. We observed that PRMT6 inhibition augmented leukemic engraftment in both the peripheral blood (PB; Fig. 2D) and bone marrow (BM; Fig. 2E), increased spleen weights (Fig. 2F), and resulted in pathogenic bone marrow osteosclerosis (Fig. 2G), regardless of mutational status, suggesting a conserved disease suppressive role of PRMT6 in MPN. However, an increase of BM hCD71+ cells was observed solely in samples carrying an ASXL1 mutation (Fig. 2H), suggesting ASXL1-dependent erythroid gene expression is at least partially dependent on PRMT6, albeit potential contributions by other mutations cannot be ruled out. Recent studies have demonstrated a role for Prmt6 in maintaining leukemic stem cell maintenance in secondary and tertiary transplants, in addition to cell cycle regulation^21,22. Our data here suggests a possible alternative role of PRMT6 in our MPN setting across primary patient samples.

Figure 2. PRMT6 perturbation confers erythropoiesis potential in ASXL1-mutant MPN.

A) Mutations in patient samples utilized. B) qRT-PCR of PRMT6 mRNA following knockdown by shRNA or control vector. C) Proportion of erythroid colonies in CD34+ colony formation assays following PRMT6 knockdown. Cells were plated in triplicate. Statistics were assessed by two-tailed Student’s t-test. D) Leukemic engraftment in the peripheral blood (PB) in PDX models (n = 5 mice per group). Statistics assessed by two-way ANOVA. E) Leukemic engraftment in the bone marrow (BM) at endpoint (n = 5 mice per group, except n = 4 for 652054 shPRMT6 #2). Statistics were assessed by Mann-Whitney u test. F) Normalized spleen weights at endpoint (n = 5 mice per group, except n = 4 for 652054 shPRMT6 #2). Statistics were assessed by Mann-Whitney u test. G) H&E histology of bone marrow at endpoint. Scale = 75 μM. H) Human CD71+ (hCD71+) cells in the BM at endpoint (n = 5 mice per group, except n = 4 for 652054 shPRMT6 #2). Statistics were assessed by Mann-Whitney u test. I) Proportion of erythroid colonies in CD34+ colony formation assays. Cells were grown in 0.5 μM EPZ020411 for 14 days. Seeded samples were from unique donors/patients: NBM (n = 2), JAK2 (n = 3), ASXL1 (n = 3). J) Expression heatmap of pertinent genes from RNA-seq analysis of CD34+ cells treated with 0.5 μM EPZ020411 for 6 hours comparing EPZ treatment to control within each sample. Treated samples were from unique donors/patients: NBM (n = 1), JAK2 (n = 3), ASXL1 (n = 3).

Lastly, we pharmacologically perturbed PRMT6 using the selective small molecule inhibitor, EPZ020411 (EPZ). In CD34+ CFAs, we observed trends that EPZ treatment did not affect erythropoietic potential in NBM cells, suppressed colonies from JAK2-mutant patients, and either did not affect or rather, increased colonies from JAK2/ASXL1-mutants (Fig. 2I). Further RNA-seq on EPZ-treated CD34+ HSPCs ex vivo revealed upregulation of genes involved in heme metabolism, hemoglobin, and notably in erythropoiesis including HIF1A and EDRF1 (Fig. 2J), further highlighting the unique erythroid responses in the context of mutant ASXL1.

In summary, through utilizing primary MPN samples and isogenic cell models, we demonstrate that mutations in JAK2 and ASXL1 play counteractive roles in expansion of the erythroid lineage. While mutant ASXL1 impaired colony production overall with substantial myeloid skewing, cells expressing both mutations presented with an intermediate phenotype. We observed that mutation of ASXL1 upregulated PRMT6, and that PRMT6 inhibition conferred greater leukemic potential and disease exacerbation in MPN, with specific increase of erythropoiesis in the context of mutant ASXL1. Additional profiling of alterations in global activation and/or repression of gene expression, and chromatin accessibility by mutant ASXL1 will provide novel, functional insight into myeloid pathogenesis. Lastly, further investigation combining a PRMT6 inhibitor with current JAK2 inhibitors may prove to be beneficial in balancing elimination of myeloproliferative disease and accompanying therapy-induced anemia.

Highlights.

• JAK2 and ASXL1 mutations inflict opposing regulation of erythropoiesis.

• Mutant ASXL1 upregulates PRMT6.

• PRMT6 inhibition specifically augments disease burden in ASXL1-mutant MPN PDX models.

• PRMT6 inhibitor, EPZ020411, upregulate genes involved in heme metabolism and erythropoiesis.