Jarid2 regulates hematopoietic stem cell function by acting with polycomb repressive complex 2

Polycomb repressive complexes (PRCs) are major epigenetic regulators that control multiple aspects of stem cell fate.¹  PRC2 consists of 3 core polycomb group proteins: Eed, Suz12, and the histone methyltransferase Ezh2 or Ezh1, which catalyze histone H3 lysine 27 dimethylation and trimethylation,^2-4  the latter of which is enriched at transcriptionally silent loci.⁵  The generic histone chaperone proteins Rbbp4 and Rbbp7 are also often considered core PRC2 components.^2,3 

The majority of studies regarding the molecular mechanism of PRC2 targeting have been performed in embryonic stem cells (ESCs), in which PRC2 represses a number of key developmental regulators to safeguard pluripotency.^6,7  Although core PRC2 components lack DNA binding ability, several accessory factors in ESCs, including Jarid2 and the mammalian orthologs of the Drosophila polycomb-like (Pcl) protein—Phf1, Mtf2, and Phf19—are important for PRC2 recruitment to target genes and for modulating its histone methyltransferase activity.

Jarid2 is a catalytically inactive jumonji family histone demethylase that is essential for PRC2 recruitment in ESCs.^8-12  Jarid2 has AT-rich interaction domain DNA binding and zinc finger domains that demonstrate low-affinity binding to DNA with a preference for CpG-rich regions, although this alone cannot explain the specificity of its genomic distribution.^9,10  Jarid2 additionally exhibits nucleosome and long noncoding RNA binding capabilities that promote PRC2 assembly, association with chromatin, and stimulation of methyltransferase activity.^13-15 

The Pcl proteins are also enriched at some PRC2 targets in ESCs, but they predominantly form distinct complexes to PRC2-Jarid2.^16-21  Pcl proteins bind the active H3K36me3 mark via their Tudor domain, thereby recruiting PRC2 to transcriptionally active chromatin.^18-21  Although these ESC studies have formed the basis for the paradigms of PRC2 accessory factor function, the extent to which they hold true in other cell types, particularly other rare adult stem cell populations, is unknown.

Hematopoietic stem cells (HSCs) are a well-characterized, clinically relevant stem cell population. HSCs generate the full array of mature blood cell types in a tightly regulated process that balances self-renewal and differentiation; however, alterations to PRC2 disrupt this delicate balance. Although somewhat controversial, Ezh2 appears to be important in highly proliferative fetal HSCs, yet appears to be dispensable in their adult counterparts.^22,23  By contrast, Ezh1 is critical for adult HSCs²⁴ ; Ezh1 knockout results in bone marrow (BM) failure due to Cdkn2a-induced senescence and reduced homing capacity.²⁴ Eed knockout leads to adult HSC exhaustion through the disruption of self-renewal, differentiation, and apoptosis.²²  Therefore, HSCs represent a relevant and interesting population in which to study PRC2 accessory factor function.

Although complete loss of PRC2 core components compromises hematopoietic stem and progenitor cell (HSPC) function and viability, in a seemingly contradictory manner, heterozygous deletion or depletion by short hairpin RNA (shRNA) –mediated knockdown leads to enhanced progenitor proliferation and contribution in competitive transplantation assays.^22,25-27  Therefore, we have surveyed the function of known PRC2 accessory factors in HSPCs by using shRNA-mediated knockdown and competitive reconstitution assays to determine which factors behave similarly to Suz12 knockdown and demonstrate enhanced contribution to all hematopoietic lineages.

We report that similar to Suz12 knockdown, Jarid2/JARID2 knockdown leads to enhanced capacity for transplantation in fetal and adult HSPCs in mouse cells and in human cells cultured in vitro or transplanted into recipient mice. Gene expression profiling of knockdown HSPCs revealed that Jarid2 represses a subset of Suz12 targets in this population, suggesting that Jarid2 plays a dose-sensitive role as a PRC2 accessory factor in HSPCs.

Animal studies were approved by the Walter and Eliza Hall Institute Animal Ethics Committee (AEC 2011.027). Gene-specific shRNA sequences (supplemental Table 1, available on the Blood Web site) were designed by using the Designer of Small Interfering RNA Web site (http://biodev.cea.fr/DSIR/DSIR.html)²⁸  and subcloned into the LMS-green fluorescent protein (GFP) or LMP-blue fluorescent protein (BFP) vectors (adapted from Dickins et al²⁹ ) with selectable markers EGFP or EBFP/puromycin, respectively. Retrovirus production and E14.5 fetal liver (FL) cell transduction and transplantation were performed as described.^26,27  Lethally irradiated (11 Gy) secondary transplant recipients were injected with 1 to 3 million BM cells isolated from FL primary recipients, 2 or 3 recipients per donor.

Adult BM was harvested from the iliac, femur, and tibia of CD45.1⁺ C57BL/6 donor mice. Cells were stained with rat monoclonal antibodies against lineage markers (Ter119, B220, CD19, Mac1, Gr1, CD2, CD3, CD8) and then incubated with BioMag goat anti-rat immunoglobulin G beads (Qiagen) for magnetic depletion. Lineage-depleted cells were stained with fluorochrome-conjugated anti-rat immunoglobulin G, c-Kit, and Sca1 and sorted on the FACSAria or BD InFlux cell sorter (BD Biosciences). HSPCs were incubated for 16 hours in StemSpan Serum-Free Expansion Medium (STEMCELL Technologies) with cytokines (50 ng/mL stem cell factor and 10 ng/mL each of thrombopoietin, interleukin-6, and Flt3), placed on viral-coated dishes, and incubated for 24 hours before intravenous injection into lethally irradiated CD45.2⁺ recipients with buffer whole BM cells.

Xenograft transplantation experiments were performed at Lund University and were approved by the Lund/Malmö Ethical Committee (M39-13). Approval for the use of the human cell samples from healthy donors was given by the Ethical Committee at Lund University Hospital (2010/696). Human CD34⁺ cord blood cell isolation, transduction, in vitro expansion, and transplantation were performed as described.³⁰  Human FL cells were obtained from Novogenix Laboratories LLC (Los Angeles, CA). Human BM cells were obtained from healthy donors age 20 to 25 years after informed consent. BM aspirations were performed from the posterior superior iliac spine at the Lund University Hospital. BM-derived CD34⁺ cells were acquired as for cord blood. Human CD34⁺ cells were cultured in StemSpan Serum-Free Expansion Medium with 100 ng/mL each of recombinant human thrombopoietin, stem cell factor, and Flt3L (Peprotech). Cells (200 000 per well) were transduced overnight in 96-well viral-loaded plates. Cells were re-plated in fresh media and maintained with twice-weekly half volume media changes, and wells were split as required. For transplantation, CD34⁺ cells were further enriched for HSCs by flow sorting for CD34⁺38^low90⁺45RA^– cells. After transduction and following a 36-hour culture as above, transduced GFP⁺ cells were sorted and intravenously injected (2000 cells per mouse) into the tail vein of 8- to 12-week-old sublethally irradiated (3 Gy) NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/SzJ (NSG) recipients. For secondary transplantations, a half femur equivalent of BM from primary recipients was injected into new mice. Contribution of human cells was measured by flow cytometry for human CD45 (huCD45) in the peripheral blood at 8 weeks and in the BM at 16 to 18 weeks posttransplant.

Donor⁺ lineage^– c-Kit⁺Sca1⁺GFP⁺/BFP⁺ HSPCs were isolated from BM of recipient mice with 30 000 to 100 000 HSPCs pooled from 6 to 10 recipients per sample. We generated 5 nonsilencing (NS) shRNA, 4 shRNA-Suz12, 4 shRNA-Jarid2.1, and 3 shRNA-Jarid2.2 transduced HSPC samples. Sequencing libraries were prepared by using the TruSeq messenger RNA Sample Prep Kit (Illumina), and 100 bp single-end read sequencing was performed on an HiSeq2000 or HiSeq2500 ultra-high-throughput sequencing system (Illumina).

The reads were aligned to mm10 by using TopHat version 2.0.8 with the b2-very-sensitive preset, then summarized over exons by using featureCounts from the Rsubread package. Differential analysis was performed by using Voom³¹  together with Limma³²  on genes with a count per million (cpm) larger than 1 in at least 2 samples. The data are deposited in Gene Expression Omnibus (GEO; GSE60808). More details on methods and reagents are provided in the supplemental Data.

First, we confirmed expression of all candidate PRC2 accessory factors (Rbbp4, Rbbp7, Phf1, Mtf2, Phf19, Jarid2) in FL-derived lineage^– Sca1⁺c-Kit⁺ HSPCs (Figure 1A). The rarity of long-term HSCs precludes the use of biochemical techniques to identify PRC2 binding partners, so we tested the function of these factors in an in vivo assay of stem cell function. FL cells were transduced with retrovirus containing validated shRNAs (supplemental Figure 3A) to deplete the expression of the candidate PRC2 accessory factors within these donor cells and their progeny. We analyzed the repopulating capacity of transduced cells in competition with nontransduced cells in reconstitution assays and compared this to mice receiving cells transduced with a negative control (NS). At 10 weeks posttransplant, the contribution of transduced (GFP⁺ or BFP⁺) cells to the B- and T-lymphocyte and myeloid lineages in peripheral blood was analyzed, reasoning that an effect in multiple lineages was likely indicative of an effect in the HSPC population.

To account for differences in transduction efficiency, the percentage of transduced cells within each lineage at 10 weeks posttransplant (output) was compared with the percentage of transduced FL cells at the time of transplant (input). The output:input ratio for each test shRNA was compared with the output:input ratio for the negative control recipients generated from the same original donor FL cells, allowing for comparison between different cohorts. This is expressed as the “relative contribution,” where “1” means no difference to negative controls (Figure 1B).

Suz12 knockdown led to enhanced contribution of transduced cells to both lymphoid and myeloid lineages (Figure 1C), as expected.^26,27  Of the 6 factors assessed, only Jarid2 knockdown induced a similar phenotype (Figure 1C), suggesting that Jarid2 may act with PRC2 in hematopoiesis. We assessed the contribution of Rbbp4-, Rbbp7-, Phf1-, Mtf2-, or Phf19-depleted donor cells to hematopoietic cell populations in the spleen, thymus, and BM at 16 weeks posttransplant (data not shown), including the HSPCs (supplemental Figure 4). No consistent change in relative contribution to HSPCs by both validated shRNAs for any of these 5 accessory factors was found; thus we focused on Jarid2. The efficacy of all hairpins was confirmed by reverse transcriptase quantitative polymerase chain reaction on splenic lymphocytes (supplemental Figure 3A), and Jarid2 and Suz12 knockdown was confirmed by western blot on transduced thymocytes isolated at 16 weeks posttransplant (supplemental Figure 3B).

In addition to PRC2, Jarid2 interacts with several H3K9 methyltransferases: Setdb1, Ehmt1, and Ehmt2.^33,34  To assess whether Jarid2 may function with these proteins in hematopoiesis, validated knockdown constructs against each of these (supplemental Figure 3C) were used in competitive transplantation assays (Figure 1B). In contrast to Suz12 or Jarid2 knockdown, H3K9 methyltransferase knockdown cells displayed a reduced relative contribution to all cell lineages analyzed in the peripheral blood at 10 weeks posttransplant (Figure 1D). These data suggest that Jarid2 is not predominantly functioning in concert with H3K9 methyltransferases and indicate previously uncharacterized roles for these enzymes in hematopoiesis.

We analyzed the relative contribution of Suz12 or Jarid2 knockdown FL cells to hematopoietic cell populations within the thymus, spleen, and BM at 6 and 16 weeks posttransplant. We observed increased relative contribution of both Suz12 and Jarid2 knockdown cells to the HSPC compartment from 6 weeks posttransplant (Figure 2A). Despite this increase, total peripheral blood counts were normal at 6 and 10 weeks posttransplant, there was no significant difference in the total cell number of the thymus, spleen, or BM between recipients of different constructs at 16 weeks posttransplant, nor did any recipients develop leukemia or lymphoma during these experiments (data not shown).

At 16 weeks posttransplant, a time point that evaluates the contribution of long-term repopulating HSCs,³⁵  the relative contribution of Suz12 or Jarid2 knockdown cells was enhanced for all cell types examined, including the HSPC compartment (Figure 2B and supplemental Figures 1 and 5A). Although the phenotype was similar between Suz12 and Jarid2 knockdown FL cell recipients, Jarid2 knockdown resulted in a marked increase in relative contribution to the T-cell lineage (Figure 2B), evident from the early double-negative stage of thymocyte development and maintained in mature splenic T cells (supplemental Figure 1).

We used the signaling lymphocytic activation molecule cell surface markers CD48 and CD150 to delineate the long-term HSCs (LT-HSCs), short-term HSCs (ST-HSCs), and multipotent progenitors (MPPs) within the HSPC population (Figure 2C).³⁶  Within all these subpopulations, both Suz12 and Jarid2 knockdown led to an elevated relative contribution, most markedly with Suz12 knockdown (Figure 2D and supplemental Figure 6).

To confirm that transduced HSCs were capable of maintaining hematopoiesis through serial transplantation, whole BM from recipients of transduced FL cells was injected into lethally irradiated secondary recipients (Figure 2E). Relative contribution was calculated by using the percentage of transduced cells in the HSPC compartment of the primary donor as input.

Jarid2-depleted donor cells contributed to secondary recipients but no longer demonstrated an augmented repopulating capacity in competition with nontransduced cells, except within the T-cell lineage (Figure 2E and supplemental Figure 5B). As expected,²⁶  the relative contribution of Suz12 knockdown cells to recipient hematopoiesis increased with serial competitive transplantation, particularly in the myeloid and progenitor compartments (Figure 2E). As the proportion of Suz12-depleted cells within the secondary transplant recipients increased, peripheral blood platelet count also increased (Figure 2F). This is the phenotype by which a Suz12 N-ethyl-N-nitrosourea–induced point mutation was first identified on a sensitized genetic background.²⁶  Target gene knockdown was maintained in secondary recipients at 16 weeks posttransplant (supplemental Figure 3D).

These data suggest that although Jarid2 has a role in controlling HSPC behavior as a PRC2 accessory factor, the extent of its role may be more specialized compared with that of core PRC2 component Suz12. Furthermore, it appears that Jarid2 has an additional function within the T-cell lineage, which is interesting in light of the JARID2, SUZ12, EED, and EZH2 mutations found in acute T lymphoblastic leukemia.^37-39 

Fetal HSCs are known to functionally and transcriptionally differ from adult HSCs.⁴⁰  Previous studies imply they may also have a distinct dependency on PRC2: fetal HSCs on PRC2-Ezh2 and adult HSCs on PRC2-Ezh1.^22,24  Jarid2 is associated with PRC2-Ezh2 rather than PRC2-Ezh1 in ESCs¹¹  and is required only for PRC2-Ezh2 nucleosome binding.¹³  However, Jarid2 has been shown to stimulate the methyltransferase activity of both complexes.¹³  On the basis of these reports plus our secondary transplant data in which Suz12 and Jarid2 knockdown donor cells behaved differently, we examined whether Jarid2 knockdown affects adult HSPCs.

We confirmed Jarid2 expression in fetal and adult LT-HSCs, ST-HSCs, and MPPs (Figure 3A). Jarid2 expression was comparable between fetal and adult LT-HSCs. Transduced adult BM HSPCs were used to reconstitute lethally irradiated recipients and the relative contribution of Suz12 or Jarid2 knockdown cells to hematopoiesis was analyzed (Figure 3B). Both Suz12 and Jarid2 knockdown resulted in enhanced contribution to all cell types examined at early time points in the peripheral blood (Figure 3C) and at 16 weeks posttransplant in the thymus, spleen, and BM (Figure 3D). Again, the most pronounced effect of Jarid2 knockdown was within the T-cell lineage (Figure 3D). Target gene knockdown was confirmed in transduced BM cells isolated at 16 weeks posttransplant (supplemental Figure 3E). These data suggest that, like Suz12, Jarid2 plays a pivotal role in restraining both fetal and adult HSPC repopulation capacity. This may be the result of an increased proliferative burden induced by in vitro culture and/or reconstitution that leads to an increased dependence of PRC2-Ezh2 and thus Jarid2 in adult HSPCs.

We next tested whether JARID2 has a role in human HSCs. JARID2 expression was detected in HSPC populations isolated from human cord blood, which contains a mixture of fetal and adult cell types (Figure 4A and supplemental Figure 2A). Two validated shRNA constructs (Figure 4B) were used to transduce cord blood–derived HSPC-enriched CD34⁺ cells that normally demonstrate limited survival in vitro.⁴¹  We observed a dose-dependent outgrowth of JARID2 knockdown GFP⁺CD34⁺ cells relative to untransduced GFP^– cells, which was not observed in the negative control (shRNA-Scramble; Figure 4C and supplemental Figure 2B). A similar phenotype was observed with transduced FL, cord blood, and adult BM CD34⁺ cells after 14 and 21 days in culture (Figure 4D).

To test the effect of JARID2 knockdown in vivo, transduced cord blood–derived cells enriched for human HSC markers (CD34⁺CD38^lowCD90⁺CD45RA^–) were transplanted into sublethally irradiated NSG recipient mice. Primary recipients receiving shRNA-JARID2–transduced cells showed an increased percentage of human CD45⁺ cells in the peripheral blood and BM compared with the negative control (Figure 4E and supplemental Figure 2C), with a normal distribution of lymphoid and myeloid cells (supplemental Figure 2D). Upon secondary transplantation, shRNA-JARID2 transduced human CD45⁺ cells were still detectable in the peripheral blood and BM of secondary recipients at 18 weeks posttransplant, whereas negative control human CD45⁺ cells could not be detected in the BM at this time point (Figure 4F). These data suggest that JARID2 functions in both fetal and adult human HSPCs. The continued presence of detectable JARID2 knockdown human cells in the BM of secondary transplant recipients likely indicates that JARID2 affects long-term HSCs.

We next investigated the molecular effects of Jarid2 knockdown in murine HSPCs. The gene expression profiles of Suz12 and Jarid2 knockdown HSPCs isolated from primary recipients of transduced FL cells at 16 weeks posttransplant were compared with the equivalent negative controls. Suz12 and Jarid2 transcripts were reduced to half that of control samples (Figure 5A-B); however, depletion of Jarid2 or Suz12 did not alter transcript levels of PRC2 core components (Figure 5A), other accessory factors (Figure 5B), or other jumonji domain proteins (data not shown).

In ESCs, Suz12 and Jarid2 share common targets.^8,10,12  In HSPCs, 529 genes were differentially expressed between Suz12 knockdown and negative controls (false discovery rate [FDR] <0.05; supplemental Table 4), 78% of which were upregulated (supplemental Figure 7A-B). Jarid2 knockdown (shRNA-Jarid2.1 and shRNA-Jarid2.2 grouped) had a more muted effect, with 76 genes attaining genome-wide significance (FDR <0.05; supplemental Table 5 and supplemental Figure 7C-D). Although 14 genes were significantly upregulated in both comparisons, a simple overlap test is biased as a result of each knockdown being compared with the same control samples and because highly expressed genes have increased power to detect differential expression.⁴²  Therefore, to study the correlation between the effect of the knockdowns in an unbiased manner, we first split the control samples between the two comparisons. Second, we selected genes with evidence of differential expression in both comparisons (P < .01; Fisher’s method of combining P values). Finally, we compared the log2 fold-change between the Jarid2 and Suz12 comparisons. Genes upregulated in the Suz12 knockdown were significantly upregulated in the Jarid2 knockdown compared with genes that were downregulated in the Suz12 knockdown (P < 10⁻⁵; Mann-Whitney test; Figure 5C). This suggests that there are common targets of Jarid2 and Suz12.

We contemplated the biological significance of the differentially expressed genes. Both Lin28b and Hmga2, genes important for the high self-renewal of fetal compared with adult HSCs,⁴³  were significantly upregulated in Suz12 knockdown HSPCs (supplemental Table 4). Gene set enrichment analysis demonstrated that both Jarid2 and Suz12 knockdown produce a significant upregulation of genes expressed in fetal HSCs⁴⁴  (Figure 5D), suggesting that PRC2-Jarid2 is responsible for repressing the fetal HSC expression signature. We broadened our analysis to 15 additional hematopoietic gene sets and found a significant enrichment for genes upregulated in HSCs compared with other hematopoietic cells in both Suz12 and Jarid2 knockdown HSPCs (supplemental Table 6). Furthermore, an LT-HSC–specific gene signature⁴⁵  was significantly and strongly upregulated in Suz12 knockdown HSPCs and modestly upregulated upon Jarid2 knockdown (FDR q = 0.061; Figure 5E). These effects had strong signals (supplemental Table 6) and were driven by alteration in expression of different groups of genes (data not shown), shown by leading edge analysis.

We examined whether genes upregulated in Suz12 or Jarid2 knockdown HSPCs were H3K27me3 marked by using published chromatin immunoprecipitation (ChIP) followed by next generation sequencing (ChIP-seq) data generated from wild-type adult BM HSPCs⁴⁶  and our own ChIP-seq data from E14.5 FL HSPCs (Figure 6A-B). Genes repressed by Suz12 and Jarid2 (positive log2-fold change, P < .01) showed significant H3K27me3 enrichment in adult (Figure 6A-B) and fetal (Figure 6A) HSPCs. These genes are likely direct PRC2 targets that are sensitive to reduced Suz12 and/or Jarid2 expression.

To investigate whether Jarid2 and Suz12 directly interact in hematopoietic progenitors, we performed co-immunoprecipitation assays in an ESC-derived hematopoietic progenitor cell line, HPC7, that can partially sustain hematopoiesis when transplanted into lethally irradiated recipient mice.⁴⁷  Immunoprecipitation of Suz12 pulled down Jarid2 (supplemental Figure 8A), an interaction that was validated by reverse immunoprecipitation (supplemental Figure 8B). This confirms that Jarid2 binds PRC2 in a relevant hematopoietic progenitor population. Furthermore, we found that Suz12 knockdown depleted Jarid2 in transduced thymocytes isolated 16 weeks posttransplant (supplemental Figure 3B). Because Suz12 is required for complex stability,⁴⁸  this suggests that Jarid2 is also part of PRC2 in thymocytes.

Here, we investigated the role of PRC2 accessory factors in HSPCs by using sensitive, dose-dependent competitive reconstitution assays. Of the factors examined, only knockdown of Jarid2 altered the reconstitution capacity of transduced cells. In both FL and BM HSPC primary transplant recipients, depleting Jarid2 mirrored the phenotype observed upon knockdown of PRC2 core component Suz12; both knockdowns increased contribution to the HSC compartment compared with the negative control. Furthermore, JARID2 knockdown in human cord blood, FL, and adult BM CD34⁺ cells enhanced their capacity to be cultured in vitro and augmented the contribution of HSC-enriched cord blood cells to transplant recipient mice. These data indicate that Jarid2/JARID2 is a newly characterized regulator of HSPC function in mice and humans.

The function of Jarid2 in HSC activity has previously been examined in an RNA interference screen targeting jumonji domain–containing methyltransferases.⁴⁹  Although two unvalidated Jarid2 targeting hairpins were scored in their assay,⁴⁹  which is similar to ours, Jarid2 was not selected for further study. Mice carrying Jarid2 (Jmj) gene trap alleles have also been used to investigate hematopoiesis; however, transplantation assays were qualitative so the results do not contradict our observation of enhanced contribution by Jarid2 knockdown cells.⁵⁰ 

Although Jarid2 was the only PRC2 accessory factor to influence hematopoiesis in our assay, this does not exclude a role for the other accessory factors. Other factors may require depletion below a threshold level for a functional outcome. Factors such as histone chaperone proteins Rbbp4 and Rbbp7 or the Pcl proteins may be redundant in HPSCs. Indeed, compensation by another accessory factor might also account for the differences between Jarid2 and Suz12 knockdown in HSCs upon secondary transplantation. Although we did not observe increased messenger RNA expression of other accessory factors in Jarid2 knockdown HSPCs, there may be increased protein stability upon long-term Jarid2 knockdown or other as yet unidentified compensatory factors that play a role.

In HSPCs, Jarid2 appears to function as an integral component of PRC2. In addition to the similar phenotypic outcomes of Suz12 or Jarid2 knockdown in primary transplants, we observed a direct interaction between Suz12 and Jarid2 in a hematopoietic progenitor cell line and a significant overlap in genes regulated by Suz12 and Jarid2 in HSPCs that are also H3K27me3 enriched. Both Jarid2 and Suz12 knockdown resulted in significant upregulation of a fetal HSC expression signature,⁴⁴  a set of genes in which some account for the high self-renewal of fetal compared with adult HSCs.⁴³  Moreover, Suz12 knockdown resulted in significant alterations in the expression of genes involved in LT-HSCs,⁴⁵  whereas Jarid2 depletion led to modest changes in these genes. These findings raise the intriguing possibility that although Suz12 and Jarid2 depletion do not alter the proportion of LT-HSCs within the HSPC compartment (as determined by cell surface markers), they may increase the self-renewal capacity of LT-HSCs. This may occur in part via de-repression of the fetal HSC program without overly compromising differentiation, consistent with what occurs in B-cell progenitors depleted of Suz12.⁵¹ 

Although there was overlap in Jarid2- and Suz12-regulated genes, there were also Suz12-regulated genes not affected by Jarid2 knockdown. This suggests that Jarid2 may be present in only a proportion of PRC2 complexes within HSCs and therefore does not affect all aspects of PRC2 function observed with Suz12 knockdown. Alternatively, a reduced level of an accessory factor may have less impact on complex stability or absolute activity than depletion of a core component. Certainly, the contribution of Suz12 knockdown cells to the HSPC compartment, particularly the LT-HSC population, was greater than that observed for Jarid2 knockdown, despite a significant enhancement compared with the negative control. The more muted effect of Jarid2 depletion on HSPC behavior and gene expression in primary recipients may account for the differing phenotypic outcome between Suz12 and Jarid2 knockdown in secondary recipients; a small enrichment of Jarid2-depleted cells may be too small to be observed above the high variance inherent in competitive transplantation. Alternatively, Jarid2 knockdown may even result in mild exhaustion of the HSPCs in secondary recipients, again with a small effect size making it indistinguishably different from controls.

Our study has identified Jarid2/JARID2 as a regulator of HSC function and highlights JARID2 as a potential therapeutic target for HSC manipulation. Our data suggest that Jarid2 performs this role via interaction with PRC2, which provides insight into the mechanistic basis of PRC2 targeting in HSPCs and broadens our understanding of the circumstances in which Jarid2 behaves as a PRC2 component. Interestingly, depletion of PRC2 has different cellular and gene expression outcomes compared with deletion of the PRC2 core component Eed.²²  It is worth considering these differing phenotypes in the context of inhibitors that target EZH2,⁵²  in which activity is likely to be reduced but not eliminated. Additionally, accessory factors and nonenzymatic components of these complexes are emerging as promising therapeutic targets,^53,54  so these differing phenotypes may be informative in regard to how such drugs, when administered systemically, may affect HSC behavior. In the future, it will be interesting to determine whether deletion of Jarid2 in HSCs has a similar effect on depletion.

The authors thank the Walter and Eliza Hall Institute Fluorescence-Activated Cell Sorter Facility and Bioservices (particularly Keti Stoev), Samir Taoudi, and Douglas Hilton for helpful discussions, and The Dyson Bequest and The DHB Foundation for philanthropic funding (M.E.B.).

This work was supported by grants from the Australian National Health and Medical Research Council (NHMRC; APP1027398) (M.E.B., A.O., and I.J.M.) and Program 1016647 (W.S.A.). J.L. is supported by grants from the Swedish Research Council, the Swedish Cancer Foundation and the Swedish Pediatric Cancer Foundation. M.E.B. is a Queen Elizabeth II Fellow of the Australian Research Council (DP1096092). I.J.M., A.O. (1051481), and W.S.A. (1058344) have fellowships from the NHMRC. This work was made possible through Victorian State Government Operational Infrastructure Support and the Australian NHMRC Research Institute Infrastructure Support Scheme.

Contribution: S.A.K., R.G., C.F., K.C., W.S.A., M.D., I.J.M., A.O., J.L., and M.E.B. designed the experiments; S.A.K., R.G., A.K., O.G., S.L., J.L., K.B., L.J.G., D.L.M., and M.E.B. performed the experiments; C.F. and A.O. performed the bioinformatic analysis; S.A.K. and M.E.B. wrote the manuscript with input from R.G., C.F., and J.L.; and all authors contributed to discussions.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Marnie E. Blewitt, The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, VIC 3052, Australia; e-mail: blewitt@wehi.edu.au.