Gfi1 (Growth factor independence 1) is a transcriptional repressor with an important role in the function and differentiation of hematopoietic stem cells (HSCs) in the direction of myeloid and lymphoid lineage (Phelan et al., 2010; van der Meer et al., 2010). It regulates the maturation of myeloid cells and is required for maintaining different signalling pathways such as Notch signalling (Horman et al., 2009; Velu et al., 2009; Phelan et al., 2013). Analysis of Gfi1 expression in Gfi1:GFP heterozygous knock-in mice revealed that Gfi1 is primarily expressed in HSCs, granulocyte-macrophage progenitors (GMPs) and common lymphoid progenitors (CLPs), while it is absent in common myeloid progenitors (CMPs) and megakaryocyte-erythroid progenitors (MEPs) (Zeng et al., 2004). Gfi1-deficient mice are characterized by an accumulation of monocytic cells and an absence of granulocytes, leading to severe neutropenia and monocytosis (Karsunky et al., 2002a; Hock et al., 2003). Recently it has been shown that reduced expression of GFI1 in blast cells of myeloproliferative neoplasm (MPN), chronic myeloid leukemia (CML) and acute myeloid leukemia (AML) patients are associated with an inferior prognosis and event-free survival (Hock et al., 2004; Khandanpour et al., 2012; Fraszczak et al., 2019). We have also shown that reduced expression of Gfi1 accelerated leukemia development in the murine model of AML (Kok et al., 2013; Hönes et al., 2016; Vadnais et al., 2018). These studies thus also underscore the contribution of Gfi1 in the expression of oncogenes, apoptotic pathways and metabolic functions in a dose-dependent manner. However, it has also been postulated that the loss of Gfi1 negatively influences the repopulation capacity of HSCs (Hock et al., 2004; Zeng et al., 2004).

We thus explored whether the low level and loss of Gfi1 indeed negatively affected the stem cell capacity of HSCs and how the seemingly contradictory findings between reduced self-renewal capacity and the dose-dependent role of Gfi1 function in myeloid pathogenesis could be reconciled. To reach this goal we made use of different mouse strains: Gfi1-knockout (KO) mice, (with a complete loss of Gfi1) and GFI1-knockdown (KD) mice, a mouse strain in which the human GFI1 was cloned into the murine Gfi1 gene locus together with a Neo cassette in the opposite direction of transcription, leading to lower GFI1 expression (10%–15% of wild-type levels) (Hönes et al., 2016). We now show that GFI1-KD and Gfi1-KO mice feature an elevated number of hematopoietic progenitors (LSK cells, Lin⁻Sca1⁺c-Kit⁺), but lower HSC (Lin⁻Sca1⁺c-Kit⁺CD150⁺CD48⁻) numbers. In contrast to Gfi1-KO HSCs, GFI1-KD HSCs can compete with Gfi1-WT HSCs in a competitive transplantation setting and contribute to multi-lineage differentiation. On the other hand, Gfi1-KO HSCs can only engraft but are not able to expand and lose their ability to contribute to multi-lineage differentiation.

We used previously described GFI1-KD mice (express 15%–20% of wildtype GFI1) (Hönes et al., 2016) and Gfi1-KO mice with a complete loss of Gfi1 expression (Karsunky et al., 2002a). GFI1-WT (KI/KI) mouse strain expresses the human GFI1 at the murine locus and the Gfi1-WT (+/+) mouse strain expresses murine Gfi1 were used as controls. We previously showed that GFI1-WT (KI/KI) mice containing the human GFI1 cDNA are functionally equivalent to murine Gfi1-WT (+/+) mice (Botezatu et al., 2016; Hönes et al., 2016; Hönes et al., 2017). Here we initially measured the percentage of LSKs, HSCs and progenitors in Gfi1-WT (+/+), heterozygous knock-in (GFI1-+/KI) and homozygous knock-in (GFI1-KI/KI) and Gfi1-KO mice from the total bone marrow (BM) cells. BM-derived LSKs (Lin⁻Sca1⁺c-Kit⁺) and HSCs (Lin⁻Sca1⁺c-Kit⁺CD150⁺CD48⁻) were gated as shown in Figure 1A; Supplementary Figure S1. Our data indicated that Gfi1-WT (+/+), GFI1 +/KI and GFI1-KI/KI mice do not differ in the number (Figures 1B–F) and percentage (Supplementary Figures S2A–G) of LSKs, HSCs and progenitor cells. Thus, Gfi1-WT (+/+) mice were used as a control in further experiments.

We next examined the frequency of hematopoietic progenitor cells in GFI1-WT (+/+, +/KI and KI/KI), GFI1-KD and Gfi1-KO mice. The total cell number and the percentage of LSK cells were significantly increased in GFI1-KD and Gfi1-KO mice compared to GFI1-WT mice (Figure 1B; Supplementary Figure S2A). In contrast, the number and percentage of HSCs were significantly reduced in GFI1-KD and Gfi1-KO mice compared to GFI1-WT mice (Figure 1C; Supplementary Figure S2B). We then assessed the percentage of LSKs and HSCs in old and young mice to examine if this effect was irrespective of the age of the mice. We did not observe a significant increase in LSKs but a significant reduction of HSCs in GFI1-KD and Gfi1-KO mice in both old and young groups (Supplementary Figures S2C, D). Our data thus recapitulate previous findings of our and other groups regarding an expansion of LSKs upon loss of GFI1 (Hock et al., 2003; Khandanpour et al., 2011). Our data further expands our knowledge concerning a dose-dependent effect of GFI1 expression on HSCs resulting in a significant decrease in HSCs number. In addition to LSKs and HSCs, we also quantified progenitor cells. Reduced expression of GFI1 (GFI1-KD) or loss of GFI1 (Gfi1-KO) significantly increased the number and percentage of GMPs (Figure 1D; Supplementary Figure S2E), but no changes in the number and percentage of CMPs or MEPs were observed (Figures 1E, F, Supplementary Figures S2F, G).

Next, we evaluated the proliferation status of LSKs, HSCs and progenitor cells in GFI1-WT, -KD and -KO mice by measuring the cell cycle and the apoptosis rate. Reduced expression or loss of GFI1 significantly enhanced proliferation as demonstrated by the percentage of cells in the proliferative phases of G1 and S-G2-M (Figures 2A, B; Supplementary Figures S3A–C). We then measured the rate of apoptosis by flow cytometry. The low level or loss of GFI1 resulted in a significantly enhanced apoptosis rate compared to GFI1-WT in LSKs, HSCs and progenitor cell populations. (Figure 2C; Supplementary Figure S3D). We subsequently monitored the clonogenic capacity of different HSCs via colony-forming units (CFU) assay. Initial plating (round 1) of HSCs, followed by replating (round 2) of HSCs from GFI1-KD and Gfi1-KO mice resulted in a significantly higher colony number underscoring their proliferative nature of the cells compared to HSCs from GFI1- KI mice (Figures 2D, E). The addition of the standard chemotherapeutic agent such as 5-Fluorouracil (5-FU) demonstrated that the GFI1-KD and Gfi1-KO HSCs were more sensitive to 5-FU treatment compared to GFI1-KI cells (Figures 2D, E).

Next, we investigated how reduced levels or loss of GFI1 affected the ability of HSCs to engraft post-transplantation and measured their competence to differentiate into the various cell lineages. To this end, we performed competitive transplantation assays as illustrated (Figure 3A; Figure 4A). We isolated HSCs and BM cells from CD45.1-expressing Gfi1-WT and CD45.2-expressing GFI1-KD or -KO mice and mixed 200 HSCs and 500,000 BM from each of the CD45.1+ and CD45.2 + mice and transplanted into lethally irradiated CD45.2 + C57BL/6 or CD45.1 + Gfi1-WT recipient mice. Usage of both CD45.1+ and CD45.2 + recipient mice ruled out a possible background bias. Post transplantation, every 4 weeks we analysed the PB for CD45.1 and CD45.2 expression on different myeloid cells (monocytes- CD11b+ and granulocytes-Gr1+), T cells (CD8⁺ and CD4⁺), B cells (B220+) and erythroid cells (Ter119+). After 16 weeks of transplantation, we euthanized the mice and analyzed them for mature and progenitor cells and HSCs of the PB and BM. We found that in the competitive setting involving GFI1-KD mice, the number of CD45.2 + mature cells analysed in the PB increased over time and nearly reached 100% in 16 weeks after transplantation (Figures 3B, C; Supplementary Figures S4A, B). The presence of short-lived myeloid cells derived from CD45.2 + GFI1-KD cells 4 weeks after transplantation into CD45.2 + host mice rules out a potential carry-over of these cells as a result of the transplantation (Figure 3B). 8 weeks after transplantation, about 90% of all monocytes in PB were CD45.2+ in both transplantation settings, hence originating from CD45.2 + GFI1-KD HSCs in the competitive transplantation assay (Figures 3B, C). Erythroid cells, B-cells and long-lived cells like T-cells are derived from CD45.2 + GFI1-KD HSCs and reached 90% after 16 weeks in both the CD45.2 and CD45.1 recipient mice (Supplementary Figures S4A, B). It has previously been shown that long-term reconstitution can be determined 12–16 weeks after transplantation (Morrison and Weissman, 1994; Akala et al., 2008). Hence, to determine the reconstitution capacity in the competitive setting, the mice were sacrificed after 16 weeks of transplantation and measured the percentage of HSCs, progenitors (CMPs, GMPs and MEPs) and differentiated cells from the total BM. The BM analysis showed that the major proportion of HSCs, progenitors and differentiated cells were derived from CD45.2 cells (Figures 3D, E). This indicates GFI1-KD HSCs possess reconstitution capacity, indeed long-term HSC activity and that GFI1 has a dose-dependent effect in contribution to PB cell formation when transplanted in a competitive setting with Gfi1-WT HSC.

In contrast to the results from the GFI1-KD transplantations, CD45.2-derived cells from Gfi1-KO mice were not present in the Gfi1-KO competitive transplantation setting (Figures 4A–E; Supplementary Figures S4A, B). About 4 weeks after transplantation the percentage of CD45.2-derived monocytes and granulocytes was less than 11% in both the CD45.2 and CD45.1 recipient mice (Figures 4B, C). Moreover, T cells, B-cells and erythroid cells were contributed from CD45.1-expressing HSCs even 16 weeks after transplantation (Supplementary Figures S5A, B). This finding is in line with previous publications that Gfi1-KO cells do not contribute to mature cells (Hock et al., 2004; Zeng et al., 2004). However, analysis of total BM revealed that the majority of HSCs are CD45.2, whereas the progenitor populations and differentiated cells were derived from CD45.1 HSCs (Figures 4D, E).

To measure self-renewal and clonogenic capacity in-vivo in a competitive setup, secondary transplantation was performed after euthanizing and analysing the BM from the primary transplanted mice. 10 Mio total BM cells from the primary transplanted mice (GFI1-WT/KD and GFI1-WT/KO groups) were collected and transplanted into lethally irradiated CD45.1 recipient mice (Supplementary Figure S6A). The PB was analysed for the expression of CD45.1 and CD45.2 on mature cells. After 8 weeks, mice transplanted with GFI1-WT/KD cells expressed CD45.2 receptors, hence the majority of the blood cells were derived from GFI1-KD mice (Supplementary Figure S6B), similar to the results from the primary transplanted mice. On the other hand, the mature cells from the mice transplanted with GFI1-WT/KO group expressed CD45.1 receptors, indicating their origin from GFI1-WT mice (Supplementary Figure S6C).

To investigate the low and loss of GFI1 induced molecular and global gene expression changes, we performed RNA-seq analysis in HSCs derived from GFI1-KD or Gfi1-KO mice and compared it to GFI1-KI mice. The RNA-seq data analysis demonstrated there are indeed distinct differences in GFI1-KD/-KO/-KI mice. Genes involving significant alterations in expressions are depicted in Supplementary Figures S7A, B. Further analysis revealed that 89 and 95 genes are differentially expressed in GFI1-KI vs. -KD and Gfi1-WT vs. KO settings, respectively. To better illustrate the differences at the pathway levels, the gene set enrichment analysis (GSEA) was performed against the combined Reactome (RT) and hallmark (HM) genesets and the significantly altered pathways in GFI1-KD and Gfi1-KO in comparison with GFI1-KI mice and Gfi1-WT mice respectively was identified. Pathway analysis identified that the cell cycle and related pathways such as G2M checkpoints, and DNA damage checkpoints were significantly upregulated in GFI1-KD HSCs (Figure 5). On the contrary, these pathways were downregulated in Gfi1-KO HSCs (Supplementary Figure S8). Interestingly it was also previously shown that GFI1 plays a role in overriding the G1 cell-cycle checkpoint in T-cells and contributing to T-cell lymphomagenesis (Karsunky et al., 2002b). Our observations as cell cycle regulation, thus shed light on different molecular alterations induced by varied GFI1 expression.

GFI1 influences the prognosis of AML, CML and MPN patients in a dose-dependent manner, and reduced levels of GFI1 accelerated disease development in murine models of human leukemia (Kok et al., 2013; Hönes et al., 2016; Volpe et al., 2017; Vadnais et al., 2018). AML, CML and MPN arise as a result of the transformation of HSCs or early hematopoietic progenitors that still possess some degree of stem cell capacity (Goardon et al., 2011). It has been described that complete loss of Gfi1 leads to an impaired function of HSCs (Hock et al., 2003; Zeng et al., 2004) and an increase in LSK frequency, possibly due to increased proliferation (Hock et al., 2003; Khandanpour et al., 2011). It has also been previously postulated that Gfi1-KO cells fail to compete with Gfi1-WT cells in a competitive setting, demonstrated by a lack of reconstitution capacity of the myeloid and lymphoid compartments in the PB (Zeng et al., 2004). The question arising from these results is whether the lack of mature cells in a competitive transplantation setting is a) due to a failure of Gfi1-KO HSCs to engraft, b) due to a lack of survival of Gfi1-KO HSCs in competition with Gfi1-WT HSCs or c) due to a loss of Gfi1-KO HSC capacity to contribute to PB cell production. We were hence interested to determine whether reduced levels of GFI1 or loss of Gfi1 influence a) the number of HSCs and b) their ability to engraft and reconstitute the PB and BM of recipient mice in a competitive transplantation setting.

We expand our knowledge about the distinct role of GFI1 in various cellular populations by showing that reduced levels of GFI1 and loss of Gfi1 lead to the significant expansion of LSK cells, but to a highly significant reduction of HSCs. On a functional level, reduced expression or loss of GFI1 expression did not dramatically impede the ability of GFI1-KD or Gfi1-KO HSCs to engraft or to expand after transplantation (up to 80% of HSCs were of GFI1-KD origin and up to 60% of HSCs were of Gfi1-KO origin). Yet, loss of GFI1 negatively influenced the maturation of HSCs into the lymphoid and myeloid lineages when Gfi1 was absent but enhanced the ability to give rise to progenitor cells and mature cells if GFI1 level were reduced. Lymphoid cells are generally long-lived, if they were generated from the transplanted HSCs early after transplantation, these cells would still be detectable at later time points, even though HSCs do not self-renew and differentiate anymore. Our data are in line with other reports highlighting the role of Gfi1 in lineage choice, similar to that of other transcription factors such as IRF8 and PU.1 (Spooner et al., 2009; Olsson et al., 2016). We thus hypothesized that knockdown and loss of GFI1 might regulate the choice of symmetric and asymmetric division of HSCs which could explain the leukemia-propagating function of reduced GFI1 levels in CML and AML (Hönes et al., 2016). The differential effect on the self-renewal and differentiation status in the in-vivo environment is further evidenced by upregulated cell-cycle related pathways in GFI1-KD HSCs while the opposite effect in Gfi1-KO HSCs.

However, further work is required to dissect this and get a deeper insight into which of the different epigenetic and gene-expression changes are induced by different GFI1 levels and how this influences the function of HSCs. Nevertheless, our work lays the foundation to dissect this in more detail in future studies.

In conclusion, HSCs with reduced expression of GFI1 could self-renewal and reconstitute, but HSCs with a complete loss of GFI1 only have the self-renew but lose reconstitution ability.