RUNX3 overexpression inhibits normal human erythroid development

RUNX proteins belong to a family of transcription factors essential for cellular proliferation, differentiation, and apoptosis with emerging data implicating RUNX3 in haematopoiesis and haematological malignancies. Here we show that RUNX3 plays an important regulatory role in normal human erythropoiesis. The impact of altering RUNX3 expression on erythropoiesis was determined by transducing human CD34+ cells with RUNX3 overexpression or shRNA knockdown vectors. Analysis of RUNX3 mRNA expression showed that RUNX3 levels decreased during erythropoiesis. Functionally, RUNX3 overexpression had a modest impact on early erythroid growth and development. However, in late-stage erythroid development, RUNX3 promoted growth and inhibited terminal differentiation with RUNX3 overexpressing cells exhibiting lower expression of glycophorin A, greater cell size and less differentiated morphology. These results suggest that suppression of RUNX3 is required for normal erythropoiesis. Overexpression of RUNX3 increased colony formation in liquid culture whilst, corresponding RUNX3 knockdown suppressed colony formation but otherwise had little impact. This study demonstrates that the downregulation of RUNX3 observed in normal human erythropoiesis is important in promoting the terminal stages of erythroid development and may further our understanding of the role of this transcription factor in haematological malignancies.

RUNX proteins belong to a family of transcription factors essential for cellular proliferation, differentiation, and apoptosis with emerging data implicating RUNX3 in haematopoiesis and haematological malignancies. Here we show that RUNX3 plays an important regulatory role in normal human erythropoiesis. The impact of altering RUNX3 expression on erythropoiesis was determined by transducing human CD34 + cells with RUNX3 overexpression or shRNA knockdown vectors. Analysis of RUNX3 mRNA expression showed that RUNX3 levels decreased during erythropoiesis. Functionally, RUNX3 overexpression had a modest impact on early erythroid growth and development. However, in late-stage erythroid development, RUNX3 promoted growth and inhibited terminal differentiation with RUNX3 overexpressing cells exhibiting lower expression of glycophorin A, greater cell size and less differentiated morphology. These results suggest that suppression of RUNX3 is required for normal erythropoiesis. Overexpression of RUNX3 increased colony formation in liquid culture whilst, corresponding RUNX3 knockdown suppressed colony formation but otherwise had little impact. This study demonstrates that the downregulation of RUNX3 observed in normal human erythropoiesis is important in promoting the terminal stages of erythroid development and may further our understanding of the role of this transcription factor in haematological malignancies.
Transcription factors play an important role in the establishment of haematopoietic lineages by regulating not only the survival and proliferation of haematopoietic stem and progenitor cells (HSPC), but also cell fate decisions and differentiation 1 . Their disruption can lead to changes in haematopoietic differentiation and the subsequent development of haematopoietic malignancies. RUNX proteins are a family of transcription factors (RUNX1, 2 and 3) that participate in important developmental processes: RUNX1 is essential for definitive haematopoiesis 2,3 ; RUNX2 is involved in skeletal development 4,5 ; and RUNX3 is essential for neurogenesis 6,7 , T cell development 8,9 and gastric epithelium growth 10 . Whilst there are several studies describing the central role of RUNX1 in haematopoiesis, little is known regarding the role of RUNX3 in human haematopoiesis.
Emerging data has supported an important role for RUNX3, the evolutionary founder of the mammalian RUNX family, in murine haematopoiesis 11 . RUNX3 is highly expressed in HSPC and its conditional knockout in aged mice causes a mild HSPC expansion and myeloid proliferation, partially phenocopying RUNX1 conditional knockout mice 12 . Indeed, conditional loss of RUNX1 in adult mice was previously shown to induce a transient expansion of haematopoietic stem cells followed by their subsequent exhaustion 13,14 . An interplay between RUNX1 and RUNX3 has been found in a RUNX1/RUNX3 double knockout model, with mice dying as a result of either bone marrow failure or a myeloproliferative disorder 15 . Furthermore, RUNX3 overexpression was recently shown to facilitate the development of a myelodysplastic syndrome in TET2-deficient mice, characterised by a disruption of cancer-related pathways and RUNX1-mediated haematopoiesis 16 . Interestingly, RUNX3 overexpression is considered an independent prognostic factor associated with worse event-free survival in childhood AML 17 . On the other hand, RUNX3 expression was found to be downregulated in prognostically favourable core binding factor (CBF) AML involving RUNX1-ETO and CBFβ-MYH11 fusion proteins 17 . Previous studies have shown that RUNX1-ETO expression as a single abnormality in human HSPC blocks erythroid differentiation and promotes self-renewal of HSPC 18,19 . A recent study showed that HSPC from elderly patients with unexplained anaemia present a greater reduction in RUNX3 expression and yield fewer erythroid colonies mRNA expression (log 2 transformed) within distinct human haematopoietic cell subsets. MPP-Multipotent progenitor; GMP-Granulocyte/monocyte progenitor; Ery-Erythroblast; Mega-Megakaryocytic cell CD34 − CD41 + . RNA-sequencing data obtained from the BLUEPRINT study 43 . Data indicate mean ± 1SD (n ≥ 3). Statistical analysis was performed using ANOVA with Tukey's multiple comparisons test, ***p < 0.001 vs HSC. (b) RUNX3 mRNA expression (log 2 transformed) in distinct haematopoietic cell subsets based on cell surface marker expression. HSC-Haematopoietic stem cell CD133 + CD34 dim ; HSPC-Haematopoietic stem progenitor cell CD38 − CD34 + ; CMP-Common myeloid progenitor; MEP-Megakaryocyte/erythroid progenitor; Ery 1-Erythroid CD34 + CD71 + GPA − ; Ery 2-Erythroid CD34 − CD71 + GPA − ; Ery 3-Erythroid CD34 − CD71 + GPA + ; Ery 4-Erythroid CD34 − CD71 low GPA + ; Ery 5-Erythroid CD34 − CD71 − GPA + . Data obtained from GSE24759 using 204197_s_at probeset 42 . Data indicate mean ± 1SD (n ≥ 4). Statistical analysis was performed using one-way ANOVA with Tukey's multiple comparisons test, *p < 0.05; **p < 0.01 vs HSC.  Fig. S1).
Cultures were subsequently enriched for transduced erythroid cells (DsRed + CD13 low ) by FACS to aid the analysis of the retrovirally transduced erythroid population (CD13 low CD36 high ) (Supplemental Fig. S2). Overexpression of RUNX3 protein was validated by western blot, showing a 4.4-fold increase in RUNX3 nuclear levels compared to control cells ( Fig. 2a and Supplemental Fig. S3). Erythroid differentiation can be divided into an early developmental stage which occurs independently of EPO and a late developmental stage strictly dependent on this cytokine 21 . The growth and differentiation of the erythroid committed population (CD13 low CD36 high ) was first assessed by culturing cells in the absence of EPO. We found that while the growth of control cultures continued to day 13, the growth of RUNX3 cultures ceased by day 10 displaying a four fold reduction in proliferative capacity by day 13 compared to controls (Fig. 2b).
Phenotypic changes associated with early erythropoiesis are characterized by an increase of CD36 expression with a simultaneous loss of CD34 22 . RUNX3 overexpression delayed upregulation of CD36 (Fig. 2c) though no significant impact on CD34 expression was observed (data not shown). Together, these data suggest that overexpression of RUNX3 in human HSPC suppresses the growth and early development of erythroid progenitors in the absence of EPO.
The effects of RUNX3 overexpression on the EPO dependent phase of erythroid development were subsequently analysed. In the presence of EPO, erythroid progenitors re-enter cell cycle and upregulate glycophorin www.nature.com/scientificreports/ A (GPA). Subsequently, they undergo maturation-associated growth arrest accompanied by reduction of cell size and of CD36 expression 18,[23][24][25] . In the presence of EPO, RUNX3-overexpressing erythroid cells showed enhanced proliferation compared to controls (7.0-fold by day 20; Fig. 3a). Developmentally, both cultures showed a decrease in CD36 expression as cells matured ( Fig. 3b and Supplemental Fig. S4). However, RUNX3 overexpression significantly suppressed GPA expression (1.8-fold on day 20 Fig. 3c and Supplemental Fig. S5), implying that terminal differentiation was inhibited. In support of this, RUNX3 erythroid cells showed a consistently higher forward scatter (FSC) compared to control (Fig. 4a,b), suggesting that RUNX3 cells were significantly larger than control cells. Furthermore, morphological analysis demonstrated that while control cells were predominantly in a late stage of erythroid differentiation (OrthoE; orthochromatic erythroblast), RUNX3 overexpression reduced the number of cells with orthochromatic erythroblast morphology (Fig. 4c,d). Taken together, these data suggest that RUNX3 overexpression suppresses terminal erythroid differentiation. We next determined the effects of RUNX3 overexpression on erythroid colony forming capacity and selfrenewal under clonal conditions. RUNX3 overexpression resulted in a significant reduction in colony forming ability by 1.7-fold compared with control (Fig. 5a). To gauge the impact of RUNX3 overexpression on self-renewal potential we carried out serial replating of colony forming cells. RUNX3 overexpressing cells were able to form www.nature.com/scientificreports/ 2.7-fold more erythroid colonies than controls upon replating (Fig. 5b). These results indicate that while expression of RUNX3 impairs erythroid colony formation, these cells have a higher self-renewal potential consistent with the inhibition of differentiation observed above.
Knockdown of RUNX3 expression impairs the colony forming efficiency of erythroid cells. We next sought to examine the effects of reducing endogenous levels of RUNX3 on human erythroid development. Lentiviral vectors encoding different RUNX3 shRNA were employed for this study (Supplemental Fig. S1).  www.nature.com/scientificreports/ Upon addition of EPO to the culture medium, no overall significance was observed in erythroid growth for RUNX3 KD cells, apart from shRNA 2 which could be attributable to off-target effects (Fig. 6b). Developmentally, reduced RUNX3 expression induced a transient delay in GPA upregulation (Fig. 6c); however, cell size and morphology were unaffected by RUNX3 KD (Supplemental Fig. S6C,D). While RUNX3 KD impaired erythroid colony forming efficiency by 15-41% (Fig. 6d), self-renewal potential as scored by colony replating was not significantly affected, possibly due to selection bias for colonies with poorer RUNX3 KD in the replating round (Supplemental Fig. S7A). Taken together, reduced expression of RUNX3 impairs the colony forming ability of erythroid cells compared to controls but fails to induce similar effects on erythroid growth and development in bulk liquid culture. This suggests that RUNX3 KD impaired survival under clonal conditions, an observation supported by the fact that cluster formation was similarly impacted (Supplemental Fig. S7B). Overall, and in contrast to the effect observed in the overexpression studies, KD of RUNX3 had only minor consequences on erythroid development, with the caveat that we were unable to generate high efficiency RUNX3 KD in these cells. Figure 5. RUNX3 overexpression inhibits erythroid colony formation and increases colony formation in replating assays. (a) Erythroid colony forming efficiency of sorted DsRed + CD13 low control and RUNX3expressing cultures following 7 days of growth in erythroid supporting growth medium containing IL-3, IL-6, SCF and EPO. Data indicate mean ± 1SD (n ≥ 3). No differences were observed in cluster formation between control and RUNX3 cultures (data not shown). (b) Self-renewal potential, assessed by a single replating round of control and RUNX3 cultures in the same conditions as previously. Following the initial 7 days of growth, erythroid colonies were counted, harvested and cells were replated in erythroid supporting growth medium containing IL-3, IL-6, SCF and EPO. Replating #1 data indicate mean ± 1SD (n = 3). Significant difference of RUNX3-expressing cells from controls was analysed by paired t-test, *p < 0.05. Replating #2 data was obtained from a single experiment.

Discussion
RUNX3 (located at 1p36, a chromosomal region often deleted in several types of cancer) has a major role in the development of gastro-intestinal tract, neurogenesis and thymopoiesis 17,26 . Whilst this transcription factor has also been shown to be crucial during haematopoiesis in non-human models, there remains a paucity of studies regarding its role in normal human haematopoiesis. This study investigated the expression of RUNX3 and its role on human erythroid development using a normal human primary cell haematopoietic model. In human cord blood derived haematopoietic cells, RUNX3 mRNA expression levels gradually decreased as cells differentiate into mature erythroid cells. The increase in mRNA observed at a later stage of maturation is unlikely to have functional relevance as erythroid cells expel their nucleus as part of their terminal differentiation, and RUNX3 function and localisation is nuclear. Human RNA-seq data demonstrated a similar trend, with erythroid cells having the lowest expression of RUNX3 compared to HSPC and compared to cell types from HSPC were initially cultured with IL-3, IL-6 and SCF for 10 days followed by addition of EPO at day 10. Data indicate mean ± 1SD (n ≥ 3). Significant differences were analysed by one-way ANOVA using Tukey's multiple comparisons test, **p < 0.01. (c) Summary data showing GPA expression (MFI) in GFP + CD13 low CD36 high cells over time. Data indicate mean ± 1SD (n ≥ 3). (d) Erythroid colony forming efficiency of shRNA control and RUNX3 KD cultures after 7 days of growth in liquid culture. Data indicate mean ± 1SD (n ≥ 4). Significant differences were analysed by one-way ANOVA using Tukey's multiple comparisons test, *p < 0.05, **p < 0.01. www.nature.com/scientificreports/ other lineages. We next examined the consequences of RUNX3 overexpression as a single abnormality on erythroid development. In early erythropoiesis, erythroid committed progenitors require SCF but not EPO for their proliferation 27 . Phenotypically, EPO independent erythroblast maturation is characterised by downregulation of CD34 on their cell surface while the thrombospondin receptor (CD36) is upregulated when cells commit to the erythroid lineage. We found that RUNX3 overexpression imposed a reduction in erythroid growth which was accompanied by a delayed upregulation of CD36 compared to controls suggesting a suppression of early human erythroid development by RUNX3. EPO is absolutely required for the survival and proliferation of late erythroid progenitor cells and for their terminal differentiation 27 . In the presence of EPO, the erythroid marker GPA is upregulated, concomitant with a gradual downregulation of CD36 25 and also a decrease in cell size 27 . During the EPO dependent phase of growth we found that RUNX3 overexpression promoted proliferation and impaired differentiation, evidenced by reduced upregulation of GPA and increased cell size compared to controls. Assessment of morphology supported the flow cytometric analysis of impaired differentiation. Previous studies have implicated RUNX3 in haematopoietic development using non-human models 12,15,28 . Recently, RUNX3 was identified as a key determinant of erythroid-myeloid lineage balance in the bone marrow and its expression was involved in the development of ageing-associated anaemias in humans 20 . Consistent with an inhibition of differentiation we also found evidence that RUNX3 overexpression increased colony formation. Interestingly, RUNX3 has been implicated in iron metabolism of the liver through regulation of BMP and TGF-β signallng 29 . In addition, a new role for TGF-β ligands in erythropoiesis has been discovered 30 where the SMAD2/3 pathway is activated leading to increased cell proliferation of early (EPO-dependent) erythroid cells. Erythropoiesis and iron metabolism are intrinsically linked and RUNX3 maybe mediating this process. Further, our data parallels RUNX1-ETO-induced disruption of erythroid development and increased self-renewal of human HSPC 18 . RUNX1-ETO is known for retaining the Runt domain region of RUNX1 present in all RUNX proteins 31 , and therefore overexpression of RUNX3 and RUNX1-ETO could target similar processes in these cells. In the haematopoietic system, RUNX1 expression is lost during erythroid development similarly to that of RUNX3 32,33 , and RUNX3-mediated repression of RUNX1 has been previously reported in different haematopoietic cells 16,34,35 . Considering the dysregulation of RUNX1 target genes by RUNX1-ETO 36 , overexpressing RUNX3 in HSPC could have similar effects leading to the inhibition of normal human erythropoiesis.

Scientific Reports
To assess the importance of RUNX3 expression during human erythroid development, its endogenous levels of expression were reduced using targeted shRNA. While this had little impact on development, RUNX3 KD efficiency was at best ~ 50% for any of the shRNA constructs employed in this study, hence we are unable to conclude that RUNX3 does not have a non-redundant role in human erythroid development. Functional redundancy between RUNX1 and RUNX3 could however rescue RUNX3 KD cells, as RUNX3 expression was previously shown to overlap that of RUNX1 in the haematopoietic system 37 where combined RUNX1/RUNX3 knockout blocked murine erythropoiesis at early stages of development 15 . Under clonal conditions, KD of RUNX3 significantly inhibited erythroid colony formation indicative of a role in maintaining survival of erythroid progenitor cells. In support of this, conditional RUNX3 knockout in aged mice was previously shown to significantly reduce the erythroid compartment 12 and a similar inhibition of erythroid colony formation was recently shown in RUNX3 KD human HSPC 20 . This study suggested that RUNX3 has an important role in the maintenance of bone marrow lineage balance as RUNX3 KD selectively reduced the megakaryocyte-erythroid compartment along with myeloid skewing similar to ageing 20 . Phenotypic data obtained here contrast with recently published data by Balogh et al. suggesting that RUNX3 KD inhibits erythroid differentiation of human HSPC based on inhibition of GPA expression at a single timepoint after 3 days of culture 20 . We observed only a transient delay in GPA expression with all other differentiation endpoints being insignificantly altered. The reasons for contrasting results could be explained using different shRNA constructs targeting distinct regions of the RUNX3 sequence, as well as different experimental designs.
In summary, RUNX3 expression decreases with erythroid maturation in human HSPC, and its ectopic expression leads to an impairment of normal erythroid differentiation and increased self-renewal. Taken together, this study demonstrates that the downregulation of RUNX3 observed in normal human erythropoiesis is important in promoting the terminal stages of erythroid development and may further our understanding of the role of this transcription factor in haematological disorders.

Plasmids and generation of retro-and lentivirus.
A retroviral vector co-expressing RUNX3 and Discosoma sp. red fluorescent protein (DsRed) was generated by directional cloning of RUNX3 (NM_001031680.2) into BamH1/EcoR1 sites of a PINCO vector modified to express DsRed 38 . PINCO DsRed vector lacking RUNX3 cDNA was used as control. Short hairpin RNA (shRNA) vectors co-expressing green fluorescent protein (GFP) were purchased from VectorBuilder (Guangzhou, China) (Supplemental Materials and Methods). RUNX3 shRNA vectors were selected using the Genetic Perturbation Platform (https:// porta ls. broad insti tute. org/ gpp/ public/) according to their specificity score and match to RUNX3 CDS. Retro and lentivirus were subsequently generated by transient transfection of Phoenix or HEK293 packaging cells, respectively, using Lipofectamine 3000 (Fisher Scientific, Loughborough, UK) according to manufacturer's instructions.
Generation of control and RUNX3 expressing/knockdown human erythroid progenitor cells. Normal human CD34 + HSPC were isolated, cultured and transduced with unconcentrated retro/lentivirus as previously described (Supplemental Materials and Methods) 18,39 . For overexpression, cultures were transduced through two separate rounds (days) of infection. Lentivirus transduction underwent one round of infection to limit toxicity 40 . For each assay described, 3 or more independent cord blood samples were used. To aid the analysis of transduced viable erythroid committed cells, day 3 cultures were stained with CD13allophycocyanin (APC) and further enriched for DsRed + CD13 low cells by FACS using a BD FACSAria III (BD Biosciences, Wokingham, UK), as previously described; cells were gated on the above parameters including FSC/ SSC and doublet discrimination 18 . Sorted cells were subsequently used in colony assays (see "Colony assay") or grown in bulk-liquid culture containing iron saturated human transferrin for growth and differentiation assessment by flow cytometry (see "Phenotypic, differentiation and morphology analysis").
Colony assay. Colony assays in liquid medium were performed as described previously 18 . Erythroid colony assays were performed on DsRed + CD13 low sorted HSPC on day 3 of culture by limiting dilution in 96-U plates (0.3 cells/well) in Iscove's Modified Dulbecco's Medium (IMDM; Fisher Scientific, Loughborough, UK) supplemented with IL-3, IL-6, SCF and EPO (BioLegend, London, UK) at 5 ng/mL or 2 U/mL for EPO and incubated at 37 °C with 5% CO 2 . Following 7 days of growth, individual erythroid colonies (> 50 cells) and clusters (> 5, < 50 cells) were counted and scored. BFU-e and CFU-e were not discriminated in these counts. To assess their self-renewal potential, colonies were harvested, replated at higher density (1 cell/well), and cultured for an additional week.
Phenotypic, differentiation and morphological analysis. To assess erythroid cell growth and differentiation in bulk liquid culture, sorted HSPC were maintained in IMDM medium supplemented with IL-3, IL-6, and SCF at 5 ng/mL during the initial 10 days (EPO independent phase of development). On day 10 of culture, EPO at 2 U/mL was added to the growth medium, and the EPO dependent phase of development was monitored for additional 12 days. Transduced cultures were analysed by flow cytometry using a BD FACSCanto II at different time points using a panel of cell surface markers (Supplemental Materials and Methods and Supplemental Fig. S4) as previously described 18 . Morphology was assessed on day 20 of culture as previously described (Supplemental Materials and Methods) 18 .

Validation of RUNX3 expression by western blot and qRT-PCR. Cytosolic and nuclear proteins
were extracted using the Biovision Nuclear/Cytosol Fractionation Kit (Cambridge Bioscience, Cambridge, UK) following manufacturer's instructions. Western blotting was performed as previously described (Supplemental Material and Methods) 41 and RUNX3 protein expression was detected using a primary rabbit monoclonal antibody (D6E2, Cell Signaling Technologies, London, UK). Total RNA was extracted from GFP + sorted HSPC using the RNeasy Plus Mini Kit (Qiagen, Manchester, UK). RNA concentration and purity were assessed using a NanoDrop (Fisher Scientific UK Ltd, Loughborough, UK). RUNX3 mRNA expression was determined using a TaqMan gene expression assay (Hs00231709_m1, Fisher Scientific UK Ltd, Loughborough, UK). GAPDH was used as reference gene (Hs02786624_g1). Gene expression was assessed using QuantStudio 5 Real-Time PCR System (Fisher Scientific UK Ltd, Loughborough, UK). Gene expression data was analysed using QuantStudio Design and Analysis software v1.5.1 by Thermo Fisher Scientific.
Statistical and data analysis. Statistical analysis was performed using a paired sample t-test, or oneway ANOVA. Minitab 18 software (Minitab LLC, State College, Pennsylvania, USA) was used for all statistical analyses.
Gene expression data was obtained from GSE24759 42 . RNA-sequencing data from different human haematopoietic cells was obtained from the BLUEPRINT epigenome programme 43 . Ethics declaration. Human neonatal cord blood was obtained from the Maternity Unit of the University Hospital of Wales (Cardiff) in accordance with the 1964 Declaration of Helsinki. All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all subjects or, if subjects are under 18, from a parent and/or legal guardian. Use of cord blood was approved by South East Wales Local Research Ethics Committee 06/WSE03/6).

Data availability
Gene expression array data analysed in this study is available in the Gene Expression Omnibus (GEO) repository with the accession number GSE24759. Additional human RNA-seq data analysed in this study was obtained from the BLUEPRINT epigenome programme. All other data generated or analysed during this study are included in this published article (and its Supplementary Information files).