JARID2 inhibition reprograms human hematopoietic progenitor cells to enhance bone marrow transplantation

Key Points

• •The genetic inhibition of JARID2 enhances repopulating activity of human HSPCs in vivo.
• •JARID2 inhibition reinforces stem cell gene expression programs via STAT1 upregulation.

Visual Abstract

Abstract

Hematopoietic stem cell (HSC) transplantation is a common treatment for many blood disorders and can be a lifesaving therapy for patients with leukemias, lymphomas, and multiple myeloma. Umbilical cord blood (UCB) serves as a valuable source of hematopoietic stem and progenitor cells (HSPCs) for transplantation, particularly for patients lacking a matched donor. However, the limited number of repopulating cells in UCB units restricts its clinical utility. Our prior studies showed that the genetic deletion of the polycomb repressive complex 2 (PRC2) cofactor Jarid2 in mouse multipotent progenitors (MPPs) conveyed ectopic self-renewal capacity. Here, we hypothesized that the function of human HSPCs could be enhanced through JARID2 inhibition. In this study, we demonstrate that both constitutive and transient knockdown (KD) of JARID2 increases the number and enhances the functionality of human HSPCs both in vitro and in vivo. This phenotype was distinct from the inhibition of EZH2 in UCB cells, suggesting that the mechanism was independent of the PRC2 cofactor activity of JARID2. Mechanistically, JARID2 KD promotes a quiescent, long-term self-renewal gene expression program governed by STAT1 upregulation and characterized by an MHC class II immunophenotype. Analogous to mice, these mechanisms conferred HSC-like potential to human MPPs in vivo. Taken together, these findings highlight JARID2 inhibition as a novel and reversible approach to expand functional UCB-derived HSPCs ex vivo, potentially improving access to stem cell transplantation for a wider patient population.

Introduction

Bone marrow transplantation (BMT) is a potentially curative therapy for a variety of hematopoietic diseases.¹ The functional units of BMT are hematopoietic stem cells (HSCs) contained within the donor graft. The effectiveness of BMT is limited by 2 factors: toxicity (most typically, graft-versus-host disease) and donor availability.^2,3 The choice of an allogeneic stem cell donor is determined by donor-recipient histocompatibility, recipient disease type, recipient condition, and age.⁴ The current sources of stem cell donation from adults are matched siblings (preferred), matched unrelated donors, mismatched unrelated donors, or haploidentical donors. However, only ∼30% of patients have a matched sibling available, and the likelihood of finding a matched unrelated donor is low for patients from diverse ethnic backgrounds.⁵ These challenges have led to the rise of umbilical cord blood (UCB) as a donor source. UCB transplantation is now a routine treatment for both children and adults who do not have a matched sibling or an unrelated donor. Because hematopoietic cells from newborn babies are immunologically more naïve, there is less risk of graft-versus-host disease after UCB transplantation.6, 7, 8 UCB transplantation is especially useful for patients of racial and ethnic minorities, and UCB sources are readily available.⁹

Although UCB transplantation has multiple advantages, it is associated with delayed engraftment and immune reconstitution, thereby leading to greater risks of infection and graft failure. This is primarily because of the lower content of HSCs within individual cord blood units. Each UCB unit contains a total HSC dose that is 1 to 2 logs lower than the dose contained in adult harvests.¹⁰ Double-UCB transplantation was implemented to increase the HSC dose and thereby allow adult patients without sufficient-sized single-cord units to proceed to BMT. Double-UCB transplantation is associated with less graft failure in adult patients but may produce graft-versus-graft immune complications.^11,12 With the increasing use of UCB in BMT, strong efforts are being made to overcome these limitations.

There are intense research efforts aiming to increase UCB-derived HSC numbers or enhance their potency ex vivo. The major challenge is to maintain self-renewal and promote the proliferation of cells with repopulation potential. Upon extraction from their native BM environment, HSCs rapidly lose self-renewal capacity in culture. Various strategies have been implemented to maintain or expand primary human HSCs ex vivo. Notable examples include aryl hydrocarbon receptor antagonists (eg, StemRegenin-1),¹³ pyrimidoindole derivatives (eg, UM171),¹⁴ peroxisome proliferator-activated receptor-γ antagonists¹⁵ (eg, GW9662), and prostaglandin E2 derivatives,16, 17, 18 which have all produced mixed success in clinical trials.^19,20 Despite these challenges, the pursuit of this endeavor remains a worthy goal.

During hematopoietic development, epigenetic modifications are crucial in programming lineage determination and cellular identity.²¹ Because epigenetic modifications are reversible and dynamically remodeled during normal development, the manipulation of the epigenome could provide a strategy to bias HSC fate decisions and expand populations with reconstituting potential. Our laboratory has identified a critical role for the polycomb repressive complex 2 (PRC2) cofactor Jarid2 in regulating hematopoietic repopulating potential. The genetic deletion of Jarid2 in mice confers long-term self-renewal to multipotent progenitors (MPPs), the cell population immediately downstream of HSCs in the hematopoietic hierarchy, which possesses only transient repopulating capacity under normal circumstances.²² As a mechanism, it was found that Jarid2 recruits PRC2 to epigenetically silence transcriptional programs associated with HSC cell identity in MPPs through the deposition of the H3K27me3 repressive chromatin mark.²² Another study suggested that the genetic inhibition of JARID2 may also enhance human HSC function.²³ With these data in mind, we hypothesized that the inhibition of JARID2 could epigenetically reprogram human UCB–derived MPPs with HSC-like potential and enhance BMT.

Methods

Human cord blood samples

Mononuclear cells were isolated using Ficoll gradient centrifugation. CD34⁺ cells were then enriched by magnetic selection (Miltenyi Biotec, catalog no. 130-100-453) and cultured in SFEM II medium (Stemcell Technologies, catalog no. 09605) supplemented with penicillin-streptomycin (50 U/mL), human stem cell factor (100 ng/mL), human thrombopoietin (100 ng/mL), and human FLT3L (100 ng/mL).

Animals

NOD-scid IL2Rγ^null (NSG) mice were obtained from The Jackson Laboratory (stock no. 005557). Mice aged 8 to 12 weeks were used as recipients and were transplanted via retro-orbital injection after 200 rads of gamma irradiation (Cesium-137).

CRISPR/Cas9 gene knockout

Guide RNAs (gRNAs) targeting human JARID2 were synthesized (Synthego) and included 2 sequences targeting exon 3: CTTGGTAATGACCAGTCTAA (JARID2#1) and GTTGCTAGTGGAGGACACTT (JARID2#2). A gRNA targeting the AAVS1 locus (GGGGCCACTAGGGACAGGAT) was used as a negative control. CD34⁺ cells were nucleofected with Cas9 ribonucleoprotein complexes (Integrated DNA Technologies, cat. no. 1074181) assembled with gRNAs. Twenty-four hours after nucleofection, ∼250 000 cells were transplanted into sublethally irradiated (200 rads) NSG mice.

Hematopoietic stem and progenitor cell (HSPC) culture and expansion assays

CD34⁺ cells were cultured in SFEM II medium (Stemcell Technologies, catalog no. 09605) supplemented with penicillin-streptomycin (50 U/mL), human stem cell factor (100 ng/mL), human thrombopoietin (100 ng/mL), and human FLT3L (100 ng/mL). For doxycycline-inducible short hairpin RNA (shRNA) lentiviral vectors, doxycycline (200 ng/mL) was added to induce shRNA expression for 8 days, after which the entire cell culture was transplanted into NSG mice. For mice receiving doxycycline chow, 1250 ppm doxycycline-containing chow was used.

Lentiviral vectors for shRNAs

We used an optimized microRNA lentiviral backbone to express shRNAs (Addgene plasmid, catalog no. 111170).²⁴ shRNA sequences: shJARID2#1 - CCAGCATTAATCTTTCTTTTA, shJARID2#2 - AAGCATTAATCTTTCTTTTAA, shJARID2#3 - CGCAGTTATTTTTGAATGTGA, and shJARID2#4 - ATGAACTGTTTTGGAATATTT; shEZH2#1 - CCAGGATGGTACTTTCATTGA, shEZH2#2 - CCAGCAGAAGAACTAAAGGAA, shEZH2#3 - ATCGGTGTCAAACACCAATAA, shEZH2#4 - AACCAATAAAGATGAAGCCAA; shSTAT1#1 - CAAGGAAGTAGTTCACAAAAT, shSTAT1#2 - CCCAGATGTCTATGATCATTT, shSTAT1#3 - CGAGCACAGTGATGTTAGACA, shSTAT1#4 - CCAGAAAGAGCTTGACAGTAA. An shRNA targeting Renilla luciferase (CAGGAATTATAATGCTTATCT) was used as a control.

Human HSPC transplantation

CD34⁺ cells were transduced with lentiviral vectors carrying shRNAs, using LentiBOOST (1:100; Sirion Biotech, catalog no. SB-P-LV-101-12) and protamine sulfate (1:1000).²⁵ Peripheral blood (PB) engraftment was evaluated to monitor human cell engraftment and differentiation. Twenty weeks after transplantation, BM was harvested (femur, tibia, and fibula) for flow cytometric analysis of HSPCs.

Statistical analysis

Statistical analyses were performed using the Student t test, Mann-Whitney U test, 1-way analysis of variance, and 2-way analysis of variance, as appropriate. P values are stated in which comparisons reach statistical significance. All data are presented as the mean ± standard error of the mean.

Anonymous cord blood specimens were obtained from Barnes-Jewish Hospital with written consent, in accordance with a protocol approved by the Washington University Human Studies Committee and following the Declaration of Helsinki. All animal procedures were approved by the Institutional Animal Care and Use Committee at Washington University.

Results

Long-term deletion of JARID2 produces durable hematopoiesis without transformation

JARID2 is a tumor suppressor in chronic myeloid neoplasms.26, 27, 28 Consequently, any potential perturbation of JARID2 in HSPCs could carry the risk of neoplastic transformation. To assess the safety of long-term JARID2 deletion, we performed CRISPR/Cas9 targeting of JARID2 in UCB CD34⁺ cells. A total of 2.5 × 10⁵ edited cells were transplanted into NSG mice and maintained for 1 year. Targeting efficiencies were similar for 2 independent gRNAs targeting JARID2 and the AAVS1 negative control gRNA (supplemental Figure 1A). PB engraftment of human CD45⁺ (hCD45⁺) cells reached saturation by 8 weeks after transplantation and remained stable at high levels throughout (Figure 1A). The percentage of edited cells for the 2 gRNAs targeting JARID2 was higher than those in the AAVS1 control (Figure 1B). One year after transplantation, there were no significant differences in the PB lineage differentiation spectrum between the groups (Figure 1C) in terms of proportions of myeloid cells (CD33⁺), B cells (CD19⁺), and T cells (CD3⁺). Similarly, lineage output was maintained in the BM (supplemental Figure 1B) in the long-term absence of JARID2 (Figure 1D), and complete blood count analysis revealed no significant differences between the groups (Figure 1E; supplemental Figure 1C). Histological analysis showed no pathological changes in the BM or spleens of recipient mice (Figure 1F). In summary, long-term JARID2 deficiency in human HSPCs cells did not induce hematopoietic pathologies in vivo, suggesting that manipulating JARID2 could be a safe translational approach. This is consistent with the observation that JARID2 is not mutated in human clonal hematopoiesis or is deleted as a founding event in human blood cancers,²⁹ suggesting that JARID2 loss of function cannot operate as an oncogenic event in isolation.

Figure 1.

Long-term deletion of JARID2 produces durable hematopoiesis without transformation. (A) Longitudinal analysis of human cell chimerism in the PB of NSG mice transplanted with UCB CD34⁺ cells targeted with gRNAs for AAVS1 (n = 4), JARID2#1 (n = 5), or JARID2#2 (n = 5). (B) Longitudinal analysis of modified alleles from CRISPR edits in hCD45⁺ PB cells in transplanted NSG mice. (C) Lineage composition of hCD45⁺ PB cells 1 year after transplantation (B cells = CD19⁺; T-cells = CD3⁺; myeloid cells = CD33⁺). (D) Western blot of hCD34⁺ cells isolated from recipient NSG BM 1 year after transplantation. (E) CBC analysis of WBC, RBC, Hb, HCT, and PLT of recipient mice 1 year after transplantation. The dashed lines represent normal range. (F) Representative histology of PB smear, BM cytospins, and spleen sections harvested 1 year after transplantation of CRISPR-edited hCD34⁺ cells into NSG mice. CBC, complete blood count; HCT, hematocrit; Hb, hemoglobin; PLT, platelet; RBC, red blood cell; WBC, white blood cell.

JARID2 inhibition enhances functional output of human HSPCs in vivo

To facilitate the genetic inhibition of JARID2 in human HSPCs, lentiviral vectors expressing shRNAs and a green fluorescent protein (GFP) reporter were constructed.²⁴ The 2 most effective shRNAs targeting JARID2 (supplemental Figure 2A) relative to the negative control Renilla shRNA (shRen) were chosen for experimentation. UCB CD34⁺ cells were transduced with lentiviruses and cultured for 8 days ex vivo. Transduction with JARID2 shRNAs produced more overall cellular output (supplemental Figure 2B), increased numbers of CD34⁺ cells (supplemental Figure 2C), and increased colony-forming unit (CFU) activity (supplemental Figure 2D). To benchmark JARID2 inhibition against current agents used to expand human HSPCs, we compared the treatment of UCB CD34⁺ cells from the same donors with SR1^13,19 and/or UM171^14,30,31 vs JARID2 shRNA. The inhibition of JARID2 enhanced human HSPC function comparably to either SR1 or UM171 as single agents (supplemental Figure 2E). Combining JARID2 inhibition with SR1 and UM171 synergistically enhanced human HSPC function (supplemental Figure 2F-G). Genetic inhibition by JARID2 knockdown (JARID2-KD) in vivo significantly improved PB engraftment of UCB CD34⁺ cells relative to the shRen control group (Figure 2A) without altering lineage differentiation bias (Figure 2B) or blood counts (supplemental Figure 2H). At 20 weeks after transplantation, JARID2 inhibition was sustained in targeted CD34⁺ cells (supplemental Figure 2I), and the engraftment of JARID2-KD cells was significantly increased in the BM (Figure 2C). Moreover, flow cytometric analysis (supplemental Figure 2J) revealed increased numbers of GFP⁺ HSPCs, including HSCs (Lineage⁻ CD34⁺ CD38⁻ CD90⁺ CD45RA⁻), MPPs (Lineage⁻ CD34⁺ CD38⁻ CD90⁻ CD45RA⁻), and multilymphoid progenitors (MLPs; Lineage⁻ CD34⁺ CD38⁻ CD90⁻ CD45RA⁺), resulting from JARID2-KD (Figure 2D). To assess long-term HSC self-renewal, 1.0 × 10⁶ CD34⁺ cells from primary recipients were transplanted into secondary NSG mice. JARID2-KD cells maintained higher PB engraftment in secondary recipients (Figure 2E) with normal differentiation (Figure 2F). JARID2-KD cells showed increased engraftment in the BM (Figure 2G) and increased numbers of HSPCs (Figure 2H). Notably, shJARID2#3 produced more pronounced functional enhancement than shJARID2#1, which was likely because of dose-dependent JARID2 inhibition. Cumulatively, these data show that JARID2 inhibition enhances lineage reconstitution and the self-renewal capacity of human HSPCs.

Figure 2.

JARID2 inhibition enhances the functional output of human HSPCs in vivo. (A) hCD45⁺ cell engraftment in the PB of NSG mice transplanted with UCB CD34⁺ cells transduced with indicated constitutive lentiviral shRNAs. (B) Lineage composition of GFP⁺ hCD45⁺ PB cells 16 weeks after transplantation. (C) Engraftment of GFP⁺ hCD45⁺ cells in the BM of recipient mice 20 weeks after transplantation. (D) Quantification of GFP⁺ human HSPC populations in the BM of recipient mice 20 weeks after transplantation. (E) PB engraftment of CD34⁺ cells transduced with indicated constitutive lentiviral shRNAs isolated from primary recipients in secondary NSG mice. (F) Lineage composition of GFP⁺ PB cells 16 weeks after secondary transplantation. (G) Engraftment of GFP⁺ human cells in the BM of secondary recipient mice 20 weeks after transplantation. (H) Quantification of GFP⁺ human HSPC populations in the BM of secondary recipient mice 20 weeks after transplantation.

JARID2 inhibition enhances HSC function without altering lineage specification or gene expression programs

We performed an immunophenotypic analysis of primary transplant recipients to comprehensively characterize the spectrum of cells produced after JARID2-KD. Representative t-distributed stochastic neighbor embedding plots (Figure 3A; supplemental Figure 3A) and heat maps (supplemental Figure 3B) reflecting cell distributions demonstrated comparable cell populations between JARID2-KD and the shRen control groups. In a parallel experiment, transduced UCB CD34⁺ cells were transplanted into NOD.Cg-Prkdc^scid Il2rg^tm1Wjl mice to better assess myeloid output (supplemental Figure 3C). No differences in myeloid differentiation potential were observed between the groups (Figure 3B). hCD34⁺ cells were isolated from the BM of primary transplant recipients and subject to single-cell RNA sequencing. Cell identities were assigned based on selected signature genes (supplemental Figure 3D), and unsupervised uniform manifold approximation and projection clustering identified distinct hematopoietic lineages (supplemental Figure 3E). Efficient KD of JARID2 was observed across all shJARID2 populations (Figure 3C). The quantification of cell numbers within each population (Figure 3D) using Azimuth reference data set within Seurat showed no significant differences in lineage distribution between groups (Figure 3E).^32,33 These data suggest that JARID2 inhibition enhances the functional capacity of human HSPCs without dramatically altering transcriptional networks or lineage specification programs.

Figure 3.

JARID2 inhibition enhances HSC function without altering lineage specification or gene expression programs. (A) t-SNE plots of lineage distribution of hCD45⁺ cells in the BM of NSG recipient mice 20 weeks after transplantation by flow cytometry. (B) Frequency of maturing myeloid cells in NSGS mice transplanted with UCB CD34⁺ cells transduced with indicated constitutive lentiviral shRNAs 20 weeks after transplantation. (C) Dot plot from single-cell RNA sequencing (scRNA-seq) data depicting JARID2 expression across indicated cell populations showing KD efficiency. (D) Uniform manifold approximation and projection (UMAP) of scRNA-seq data for hCD34⁺ cells isolated 20 weeks after transplantation. Cell populations are identified by marker genes. (E) Number of cells of indicated genotypes per UMAP cluster based on Azimuth cell-type annotations. CLP, common lymphoid progenitor; cDC, conventional dendritic cell; EMP, erythroid-myeloid progenitor; GMP, granulocyte-monocyte progenitor; LMPP, lymphoid-primed MPP; MEP, megakaryocyte-erythroid progenitor; NK, natural killer; NSGS, NOD.Cg-Prkdc^scid Il2rg^tm1Wjl; pDC, plasmacytoid dendritic cell; PMN, polymorphonuclear neutrophil; PreB, pre–B cell; ProB, pro–B cell; SSC, side scatter; t-SNE, t-distributed stochastic neighbor embedding.

Transient JARID2 inhibition imparts long-term but reversible functional benefits to human HSPCs

We aimed to more closely simulate clinical protocols in which UCB CD34⁺ cells would be transiently exposed to JARID2 inhibition ex vivo before transplantation, after which the exposure would be ceased. To facilitate this, doxycycline-inducible lentiviruses were engineered to constitutively express a reverse tetracycline-controlled transactivator protein from a phosphoglycerate kinase promoter with separate coexpression of shRNA of interest and blue fluorescent protein (BFP) driven by a Tet-On 3G promoter in the presence of doxycycline (supplemental Figure 4A). Dose titration identified 200 ng/mL doxycycline as optimal, achieving effective KD with minimal cytotoxicity. Vector functionality was validated by western blot, confirming efficient JARID2 suppression upon doxycycline exposure that was reversed when doxycycline was withdrawn (supplemental Figure 4B). The kinetics of vector activity loss after doxycycline withdrawal was confirmed in both the HEL cell line (supplemental Figure 4C) and UCB CD34⁺ cells (supplemental Figure 4D). After transduction and exposure to 8 days of doxycycline ex vivo, inducible JARID2 inhibition increased the total cell yield and CD34⁺ cell number from a fixed input of UCB CD34⁺ cells (supplemental Figure 4E), similar to constitutive shRNAs.

Given that JARID2 is a PRC2 cofactor,^34,35 we also generated inducible EZH2-targeting shRNAs (supplemental Figure 4F) to determine if similar effects would be recapitulated by PRC2 inhibition. UCB CD34⁺ cells per well were transduced with inducible vectors. After 48-hour exposure to doxycycline, 4.0 × 10⁴ transduced (BFP⁺) CD34⁺ cells were purified and cultured for 8 days with doxycycline. The entire progeny of these cultures was transplanted into NSG mice. BFP⁺ hCD45⁺ cells became undetectable in the PB of recipient mice by 4 weeks after transplantation, suggesting that shRNA expression was not induced in the absence of doxycycline (supplemental Figure 4G). Even with transient inhibition ex vivo, JARID2-KD enhanced the PB engraftment of UCB CD34⁺ cells (Figure 4A) with normal lineage differentiation (supplemental Figure 4H). This effect was not observed from the shEZH2 groups (Figure 4A). BM engraftment was more variable but significantly lower in the shEZH2 groups (Figure 4B). Human HSPC numbers were consistently increased in the group that received the more potent JARID2 shRNA (Figure 4C). To evaluate long-term HSC self-renewal, 4.0 × 10⁵ CD34⁺ cells were isolated from the BM of primary recipients and transplanted into secondary NSG mice. Western blot of hCD34⁺ cells from the BM of secondary recipients confirmed the restoration of JARID2 protein levels after doxycycline withdrawal (supplemental Figure 4I). In secondary transplant, the engraftment of human cells in the PB, BM, and CD34⁺ HSPC compartments was comparable among all groups (supplemental Figure 4J). These data suggest that transient JARID2 inhibition can enhance initial reconstitution, but this phenotype is lost when JARID2 is restored. Consequently, any reprogramming resulting from JARID2 inhibition is not permanent.

Figure 4.

Transient JARID2 inhibition imparts long-term but reversible functional benefits to human HSPCs. (A) PB engraftment of hCD45⁺ cells in NSG mice from the transplant of UCB CD34⁺ cells transduced with the indicated inducible lentiviral shRNAs induced for 8 days ex vivo before transplantation. (B) Engraftment of hCD45⁺ cells in the BM of recipient mice 20 weeks after transplantation. (C) Quantification of human HSPC populations in the BM of recipient mice 20 weeks after transplantation. (D) PB engraftment of hCD45⁺ cells in NSG mice from the transplantation of UCB CD34⁺ transduced with the indicated inducible shRNAs after induction with doxycycline chow in vivo at 14 weeks after transplantation (normalized to shRen). (E) Engraftment of BFP⁺ hCD45⁺ cells in the BM of recipient mice 26 weeks after transplantation (12 weeks after doxycycline induction). (F) Quantification of BFP⁺ HSPC populations in the BM of recipient NSG mice 26 weeks after transplantation. Dox, doxycycline.

We leveraged this inducible system to determine if JARID2 inhibition could directly reprogram human HSPCs in vivo. UCB CD34⁺ cells were transduced with inducible lentiviral shRNAs in the absence of doxycycline and transplanted into NSG mice. In the absence of induction, PB engraftment of human cells was comparable among all groups. Fourteen weeks after transplantation, the expression of shRNAs was induced through the administration of doxycycline in rodent chow. Immediately after induction, shJARID2 groups displayed a significant increase in PB engraftment (Figure 4D). After 12 weeks of doxycycline induction in vivo, there was increased engraftment of human cells in the BM (Figure 4E) and increased numbers of BFP⁺ HSPCs (Figure 4F) from the JARID2-KD groups. This experiment shows that human HSPCs can be directly reprogrammed in vivo by JARID2 inhibition to enhance functional output.

Cell culture changes the phenotypic identity of functional human repopulating cells

Our prior data in mouse models showed that the genetic inhibition of Jarid2 conveyed long-term repopulating potential to MPPs.²² It is possible that a similar mechanism operates in human cells, and the enhanced function of UCB CD34⁺ cells after JARID2-KD is attributable to the differential effects on specific HSPC subpopulations. To examine this, UCB CD34⁺ cells were transduced with lentiviral vectors constitutively expressing shRNAs targeting JARID2 or Renilla. After ex vivo culture, 1250 HSCs (Lineage⁻ CD34⁺ CD38⁻ CD90⁺ CD45RA⁻) or 1500 MPPs (Lineage⁻ CD34⁺ CD38⁻ CD90⁻ CD45RA⁻) were purified based on canonical human cell surface markers and transplanted into NSG mice. At 12 weeks after transplantation, PB engraftment of HSCs receiving JARID2-KD exhibited higher engraftment than the shRen control group (supplemental Figure 5A). However, MPPs from all groups failed to engraft (supplemental Figure 5B), and overall engraftment levels were markedly lower than that observed in prior experiments. This suggested that the immunophenotype used to identify native human HSPC populations may be influenced by ex vivo culture and/or lentiviral transduction.

CFU assays were performed to define the phenotype of functional HSPCs after ex vivo culture. Although a marker of human HSCs in vivo,36, 37, 38 CD49f was not a reliable marker of functional cells after ex vivo culture because all CD49f⁺ cells failed to generate colonies (supplemental Figure 5C). Rather, the CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ phenotype enriched for most robust CFU activity (supplemental Figure 5C-D). Eight days after transduction, JARID2-KD increased the number of CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ cells (supplemental Figure 5E) and enhanced CFU activity from this population (supplemental Figure 5F). Cellular indexing of transcriptomes and epitopes by sequencing was performed to integrate single-cell transcriptomics with immunophenotype after ex vivo culture (supplemental Figure 5G). Because CD49f⁺ cells lacked CFU potential after culture, this population (supplemental Figure 5H) was excluded from the final uniform manifold approximation and projection (Figure 5A). HSCs were identified by the expression of HSC-defining genes (supplemental Figure 5I). In addition to a cluster of cells expressing human HSC genes such as HLF, the expression of other HSC-specific genes, such as MECOM and MEIS1, spread to another cluster of cells after JARID2-KD (Figure 5B). This emergent population was also marked by cell surface expression of EPCR (CD201) and CD90 (Figure 5C), with the expression of CD71 demarcating this population from canonical HSCs (Figure 5C). CD71 has been noted as a marker of activated HSCs,³⁹ and given the ectopic expression of some HSC-specific genes in this population and its expansion after JARID2-KD, we hypothesized that this may represent the cell population that is functionally augmented after JARID2 inhibition.

Figure 5.

Cell culture changes the phenotypic identity of functional human repopulating cells. (A) UMAP of CITE-seq data from GFP⁺ CD34⁺ cells after 8 days of culture (CD49f⁺ cells excluded). (B) Feature plots from CITE-seq data showing the expression of HSC marker genes HLF, MEIS1, and MECOM in cells of indicated genotypes. The purple circle denotes emergent population with HSC-like transcriptional profile in the shJARID2 group. (C) Feature plots from CITE-seq data showing the expression of adt for surface proteins CD90, CD201 (EPCR), and CD71. (D) PB engraftment of hCD45⁺ cells in NSG recipient mice transplanted with 1000 cells of indicated HSC (GFP⁺ CD34⁺ CD90⁺ EPCR⁺ CD49f⁻) phenotype (CD71⁺/CD71⁻) transduced with the indicated lentiviral shRNAs. (E) Engraftment of GFP⁺ human cells in the BM and HSPC populations of recipient mice 20 weeks after transplantation. adt, antibody-derived tag; CITE-seq, cellular indexing of transcriptomes and epitopes by sequencing; CLP, common lymphoid progenitor; cDC, conventional dendritic cell; EMP, erythroid-myeloid progenitor; EPCR, endothelial protein C receptor; Eryth, erythroid; GMP, granulocyte-monocyte progenitor; LMPP, lymphoid-primed MPP; MEP, megakaryocyte-erythroid progenitor; Mk, megakaryocyte; MkP, megakaryocyte progenitor; pDC, plasmacytoid dendritic cell; PreB, pre–B cell; ProB, pro–B cell.

To analyze the functional potential of this emergent population, UCB CD34⁺ cells were transduced with constitutive lentiviral shRNA vectors, and after 8 days of culture, equal numbers of CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ cells were purified from each group with additional separation based on CD71 expression (supplemental Figure 5J) and transplanted into NSG mice. Only the CD71⁻ cell fraction displayed long-term engraftment potential, which was enhanced by JARID2-KD (Figure 5D). JARID2-KD in CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ CD71⁻ cells resulted in higher human cell chimerism in the BM of recipient mice and increased numbers of HSPCs (Figure 5E). Similar results were observed from serial CFU replating (supplemental Figure 5K). Although JARID2-KD expanded a population of cells with HSC transcriptional and phenotypic characteristics in culture, ultimately, this CD71⁺ cell fraction possessed no reconstitution capacity in vivo. Our data reveal that functional repopulating activity was contained within the CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ CD71⁻ cell fraction after culture manipulation.

JARID2 KD produces subtle molecular effects in HSPCs that are distinct from PRC2 inhibition

The epigenetic profiles of UCB CD34⁺ cells after JARID2 and EZH2 KD were compared. Western blot analysis after ex vivo culture confirmed that EZH2 KD led to a significant depletion of H3K27me3 in HSPCs (Figure 6A). In contrast, JARID2 KD did not significantly alter global H3K27me3 levels (Figure 6A), suggesting a distinct role from the PRC2 complex in human HSPCs. To assess specific loci, cleavage under targets and release using nuclease and cleavage under targets and tagmentation were performed for a variety of epigenetic modifications on purified GFP⁺ CD34⁺ EPCR⁺ CD90⁺ CD49f⁻ CD71⁻ cells after 8 days of culture. Experiments were performed on 3 independent UCB samples and compared to the freshly isolated (naïve) UCB CD34⁺ cells as a reference control. Consistent with western blot results, H3K27me3 peaks were maintained in JARID2-KD samples but markedly reduced after EZH2-KD (Figure 6B-C). Pathway analysis of H3K27me3-enriched regions (supplemental Table 1) did not identify significantly enriched terms (supplemental Figure 6A). Transcriptional profiling by bulk RNA sequencing of the same cell populations revealed a similar effect, with very few differentially expressed genes between comparisons (supplemental Table 2), none of which seemed to explain the observed functional results (supplemental Figure 6B). Of note, EZH1 expression was reduced in EZH2-KD cells (supplemental Table 2), suggesting that any phenotypic effects were not compensated by EZH1 upregulation. Quantitative assessment of the relationship between gene expression (RNA sequencing) and H3K27me3 (cleavage under targets and tagmentation) by Spearman correlation analysis revealed no clear correlation between gene expression changes and H3K27me3 enrichment (Figure 6D). Reciprocally, differentially expressed genes showed minimal change or no H3K27me3 occupancy at these loci (supplemental Figure 6C). These findings suggest that JARID2 inhibition does not regulate target genes through an H3K27me3-dependent mechanism, as has been implicated by other studies.⁴⁰

Figure 6.

JARID2 KD produces subtle molecular effects in HSPCs that are distinct from PRC2 inhibition. (A) Western blot analysis of UCB CD34⁺ cells transduced with indicated shRNAs after 8 days of ex vivo culture. (B) H3K27me3 peak counts from CUT&TAG analysis of GFP⁺ CD49f⁻ CD90⁺ CD34⁺ EPCR⁺ CD71⁻ cells transduced with the indicated shRNAs after 8 days of ex vivo culture. (C) Heat maps of H3K27me3 signal intensity centered at transcription start sites (±5 kb) for genes identified in all 4 experimental conditions. Data represent the combined results from 3 biological replicates per condition. Heat maps are sorted by decreasing H3K27me3 signal intensity in naïve UCB CD34⁺ reference sample. (D) Spearman correlation analysis between DEGs identified from scRNA-seq HSC populations and H3K27me3 signal intensity at these loci. (E) UMAP of scRNA-seq data for hCD34⁺ cells sorted from NSG mice transplanted with cord blood cells transduced with constitutive lentiviral shRNAs 20 weeks after transplantation. Cell populations are identified by marker genes. shRen, shJARID2#1, and shJARID2#3 samples are combined; the HSC population is highlighted. (F) Dot plot highlighting the differential expression of a curated HSC signature gene set (Aguade-Gorgorio et al⁴¹) among shRNA genotypes within the HSC population. (G) HSC module scores of individual cells within the UMAP HSC cluster calculated for each shRNA genotype using the Seurat package based on the curated HSC signature gene list. CLP, common lymphoid progenitor; CUT&TAG, cleavage under targets and tagmentation; DEG, differentially expressed gene; GMP, granulocyte-monocyte progenitor; LMPP, lymphoid-primed MPP; MEP, megakaryocyte-erythroid progenitor; pDC, plasmacytoid dendritic cell.

Because cell culture conditions modified the properties of HSCs, to uncover the true effects of JARID2-KD, we revisited single-cell RNA sequencing data generated from GFP⁺ CD34⁺ cells isolated from primary NSG recipient mice, reasoning that this would provide sufficient time to re-establish native cell surface markers. Within the annotated HSC cell cluster (Figure 6E), JARID2-KD increased the expression of well-annotated, HSC-defining genes,^41,42 such as HLF, MLLT3, MYCT1, and MEIS1 (Figure 6F). We incorporated a curated list of HSC-defining genes to compute an HSC module score.⁴¹ This HSC module score was significantly elevated in shJARID2 HSCs (Figure 6G). Although the HSC module peaked in the HSC cluster, elevated scores were also detected in other JARID2-KD HSPC populations (supplemental Figure 6D), suggesting that JARID2 inhibition may reprogram some of these populations with HSC-like potential. To test this, HSCs, MPPs, and MLPs were isolated from the BM of secondary recipients of constitutive shRNA-transduced UCB CD34⁺ cells and subject to functional analysis via CFU assay. JARID2-KD significantly enhanced serial replating of both HSC and MPP cell fractions, but not MLPs (supplemental Figure 6E-F). Consequently, in human HSPCs, JARID2 inhibition not only enhanced the function of HSCs but also conveyed HSC-like potential to MPPs, which lack a self-renewal capacity under normal conditions, analogous to our prior results from murine models.²²

JARID2 inhibition upregulates STAT1 to enhance the function of human HSPCs

Differential gene expression analysis was compared specifically within the HSC population. The most significantly upregulated genes in JARID2-KD HSCs were largely related to the major histocompatibility complex class II (MHCII; Figure 7A). Recent studies have suggested a role for MHCII in HSC regulation.^43,44 Flow cytometry confirmed a higher expression of MHCII proteins on the surface of JARID2-KD HSCs (Figure 7B). Because MHCII has been suggested to mark long-term HSCs in mice,⁴³ to assess if increased MHCII marked more functional human cells after JARID2-KD, HSCs from recipient mice transplanted with UCB CD34⁺ cells transduced with constitutive shRNAs were sorted based on MHCII expression and subject to CFU assay. Serial replating showed that JARID2-KD enhanced the functionality of both MHCII⁻ and MHCII^high HSCs, but most of the CFU potential was contained in the MHCII^high fraction, which was enhanced after JARID2 inhibition (Figure 7C). MHCII^high HSCs show reduced cycling and are resistant to stress-induced proliferation.^43,44 Cell-cycle analysis of posttransplant transduced HSCs (supplemental Figure 7A) confirmed that JARID2-KD HSCs contained a significantly higher proportion of cells in the quiescent G₀ phase (Figure 7B). Because deeper quiescence is a hallmark of more functional HSCs, this presents 1 potential mechanism through which JARID2 inhibition can enhance the long-term reconstitution of human HSCs.

Figure 7.

JARID2 inhibition upregulates STAT1 to enhance the function of human HSPCs. (A) Volcano plot showing differentially expressed genes between shJARID2 and shRen cells within the HSC population from scRNA-seq data. (B) Representative flow cytometry plots showing the proportion of MHCII^high (red boxes) HSCs from each shRNA genotype from the BM of recipient mice 20 weeks after transplantation. (C) Serial replating showing the CFU efficiency of indicated HSPC genotypes separated by MHCII levels isolated from the BM of recipient mice 20 weeks after transplantation. (D) Western blot showing STAT1 protein levels in hCD34⁺ cells isolated from primary recipient mice of UCB CD34⁺ cells transduced with indicated lentiviral shRNAs 20 weeks after transplantation. (E) CFU analysis of indicated HSC genotypes isolated from primary recipient mice transduced with indicated BFP⁺ lentiviral shRNAs. (F) MACS2-called STAT1 peak counts (relative to IgG) in freshly isolated (naïve) UCB CD34⁺ cells and those transduced with indicated shRNAs after 8 days of culture. (G) Genome browser tracks showing STAT1 binding at indicated loci in freshly isolated (naïve) UCB CD34⁺ cells and those transduced with indicated shRNAs after 8 days of culture. IgG, immunoglobulin G; ncRNA, non-coding RNA; miRNA, microRNA; SSC, side scatter; UTR, untranslated region.

A recent study showed that STAT1 is essential in the maintenance of MHCII^high HSCs in mice.⁴⁴ In addition to MCHII components, STAT1 was one of the most significantly upregulated genes in JARID2-KD HSCs (Figure 7A). Western blot analysis confirmed the upregulation of STAT1 protein in JARID2-KD CD34⁺ cells (Figure 7D). We hypothesized that if the mechanism of augmented HSPC function resulting from JARID2 inhibition was through STAT1 upregulation, then the inhibition of STAT1 in JARID2-KD cells may ameliorate their functional enhancement. shRNAs targeting STAT1 were cloned into a BFP-expressing lentiviral vector (supplemental Figure 7C). GFP⁺ CD34⁺ cells from primary transplant recipients of constitutive shRNA lentiviruses were transduced with BFP⁺ STAT1 shRNA lentiviruses and subject to functional assessment by CFU assay. Results showed that STAT1 inhibition was able to rescue the increased CFU activity of JARID2-KD HSPCs (Figure 7E). Although there were no obvious epigenetic changes at the STAT1 locus in JARID2-KD HSCs that would explain the increased expression (supplemental Figure 7D-E), we performed cleavage under targets and release using nuclease to analyze STAT1 binding to the genome after the genetic inhibition of JARID2. After 8 days of shRNA expression, there were more STAT1 binding peaks in the genome of JARID2-KD CD34⁺ cells than in control cells (Figure 7F). Moreover, significantly increased STAT1 binding was observed at many of the upregulated target genes in JARID2-KD HSCs, such as CD74 and STAT1 itself (Figure 7G). Collectively, these findings suggest that JARID2 inhibition enhances human HSPC functionality, at least in part, by upregulating STAT1 and MHCII and promotes a quiescent gene expression signature characteristic of long-term self-renewing HSCs.

Discussion

The ability to propagate functional human HSCs ex vivo would have tremendous benefit for BMT and open new avenues for gene therapy/gene editing applications. The approach we undertook was to use acute epigenetic manipulation during ex vivo expansion culture to prevent the repression of self-renewal genetic networks. Because epigenetic modifications are reversible and dynamically remodeled during normal development, we reasoned that the manipulation of the epigenome could provide a novel approach to biasing HSC fate decisions toward self-renewal and expand functional HSCs. Our data show that JARID2 inhibition enhances the output of human HSPCs by promoting a quiescent gene expression signature characteristic of long-term self-renewing HSCs, at least partly driven by the upregulation of STAT1 and marked by increased MHCII expression. The inhibition of JARID2 also appears to reprogram human MPPs in vivo to endow this population with some measure of self-renewal potential. This finding is analogous to murine studies²² and extends previous observations regarding the role of JARID2 in human hematopoiesis.²³ Importantly, for potential translational approaches, transient JARID2 inhibition through inducible shRNA expression was sufficient to reprogram human HSPCs with enhanced functional potential. This effect was reversible, showing that transient JARID2 inhibition does not permanently reprogram human HSPCs, which could alleviate concerns about future malignant potential. Although the experimental tools used here are not ideal for clinical applications, they serve as proof of concept that the development of small-molecule inhibitors of JARID2 or PROTAC-based degradation systems may have translational potential.

Because we did not observe the major redistribution of H3K27me3 in human HSCs after JARID2 inhibition, the observed mechanism of functional enhancement resulting from JARID2 inhibition in human HSPCs appears to be at least partly attributed to the upregulation of STAT1. KD of JARID2 in human HSCs led to increased STAT1 expression and increased binding of STAT1 at upregulated target genes, such as CD74, B2M, and STAT1 itself. The absence of H3K27me3 at these loci may provide a permissive, but not sufficient, chromatin context for transcriptional activation in the setting of JARID2 inhibition. This aligns with recent data from mouse models showing that HSCs from STAT1-deficient mice are functionally impaired in competitive transplantation assays, and STAT1 is required to maintain the protective transcriptional programs in homeostatic HSCs, including the inhibition of cell cycling.⁴⁴ In mice, STAT1 is also required to maintain MHCII^high HSCs because this population is lost upon the genetic deletion of STAT1.⁴⁴ Our data suggest that this appears to be a mechanism that augments the HSC function that is conserved between mouse and human. A caveat of this study is that the exact identity of the human HSPCs that obtain a functional benefit from JARID2 inhibition could not be definitively identified because the nature of ex vivo culture conditions alters the immunophenotypic profile of repopulating cells. Nevertheless, results from these studies position JARID2 inhibition as a novel method to enhance human HSPC function from limiting donor stem cell sources that could be implemented to improve clinical BMT applications.

Conflict-of-interest disclosure: T.M.P. has served as a consultant for Pillar Patient Advocates, Silence Therapeutics, PharmaEssentia, and the MPN Research Foundation. A.L.Y. has served as a consultant for BioGenerator; and is a cofounder, chief executive officer, and shareholder of Pairidex Inc. G.A.C. has served as a consultant for and received research funding from Incyte, Ajax Therapeutics, Atavistik Bio, and ReNAgade Therapeutics Management; and is a cofounder, a scientific advisory board member, and a shareholder of Pairidex Inc. The remaining authors declare no competing financial interests.