GATA2 mutated allele specific expression is associated with a hyporesponsive state of HSC in GATA2 deficiency syndrome

Introduction

GATA2 is one of the master regulatory transcription factors (TF) of hematopoiesis, characterized by two zinc finger domains (N-ZF and C-ZF), which bind DNA and interact with protein partners to regulate its function [1]. During embryogenesis, GATA2 plays a crucial role in the generation and maintenance of the hematopoietic stem and progenitor cells (HSPCs) [2, 3]. Its deregulation leads to a decrease in the number of HSCs and a loss of their self-renewal capacity in young mice [4, 5].

Over the past decade, germline GATA2 mutations have been extensively identified in patients with a wide range of clinical presentations, including hematological malignancies, immunological defects and vascular disorders [6,7,8,9,10,11,12]. The clinical spectrum of germline GATA2 mutations is broad [13]. Overall, frameshift, nonsense and enhancer +9.5 mutations suggest a haploinsufficiency mechanism of pathogenesis, while recurrent missense mutations may have aberrant activity [14, 15]. Moreover, patients with missense mutations are more prone to develop leukemic transformation, suggesting that these mutations may confer additional features [16]. However, their impacts remain poorly studied in vivo due to the very limited number of available models [17, 18]. To investigate the impact of the Gata2 R396Q mutation on hematopoiesis, we established a new knock-in mouse model.

Located at the apex of the hematopoietic hierarchy, long-term HSCs (LT-HSCs) have the ability to self-renew and differentiate to maintain hematopoietic homeostasis throughout the life of an organism [19]. We report here that Gata2^R396Q/+ mice, in contrast to the Gata2^+/- mice, have an increased number of LT-HSCs from embryogenesis to old age. Challenging assays unveiled functional impairments elucidated at the molecular level by enrichment in pathways associated with exhaustion and hyporesponsiveness.

Moreover, it is now admitted that the HSC population is phenotypically and functionally heterogeneous [20]. Previously, Gata2 has been identified as a gene exhibiting allelic-specific expression (ASE) in non-physiological contexts such as acute myeloid leukemia (AML) [21] or GATA2 deficiency [22]. In this study, we demonstrate that a distinct subset of LT-HSCs predominantly express the mutated allele of Gata2, highlighting a link between the heterogeneity of HSC function and the allele-specific expression (ASE) of the GATA2 gene. Thus, our findings provide insights into the influence of a Gata2 missense mutation on hematopoietic tissue and facilitate a deeper understanding of the pathophysiology in patients with Gata2 germline mutations.

Methods

Mice

C57BL/6 N and CD45.1 B6.SJL mice were sourced from Janvier (Orleans, France) and Jackson Laboratory (Bar Harbor, ME, USA), respectively. Gata2^+/- mice, provided by Prof. Stuart Orkin (Harvard Stem Cell Institute, USA), were backcrossed to C57BL/6 N. Gata2^R396Q/+ mice were generated in-house by the CIPHE animal facility (Marseille, France). All experiments complied with French law and were approved by the local ethics committee for animal experimentation (Protocol numbers: 19-16U1037-CB-01 and 20-16U1037-CB-02).

Genotyping

DNA for genotyping PCR was extracted from mouse ear tissues (or tails for embryos) using the “Extract-N-Amp PCR” kit (Sigma #XNAT2-1KT). For Gata2^R396Q/+ mice, primers were Gata2-forward (5’ AGGCTGTGCAGGCATTTA 3’) and Gata2-reverse (5’ CTGCCAAACCACCCTTGAT 3’). PCR conditions: 94 °C for 3 min; 31 cycles of 94 °C for 30 sec, 58 °C for 40 sec, 72 °C for 30 sec; and 72 °C for 7 min. DNA bands were visualized on a TapeStation (Agilent). Gata2^+/- mice were genotyped by amplifying a 650 bp band using Neospecific PCR with Neo504 and Neo505R primers.

Clonogenic Assay

Human GATA2 WT, GATA2 R204X, and GATA2 R396Q cDNAs were subcloned into a MIG plasmid (Addgene #52107) [23] and tagged with 5’ 3xHA to produce retroviruses. Murine lineage-negative cells were transduced with retroviral supernatant supplemented with polybrene, mIL-3, hIL-6, and mSCF (Peprotech). Clonogenic assays were conducted per the manufacturer’s instructions (Stem cell technologies).

Immunofluorescence

BAF/3 cells were transduced with viral supernatants in cytokine-supplemented RPMI by spinoculation and sorted by GFP expression. Cells were fixed, permeabilized, and stained for immunofluorescence analysis using an AXIO microscope and ZEN software (Zeiss). Whole E12.5 fetal livers were fixed, embedded in agarose, sectioned using a vibratome, blocked, permeabilized, and incubated with Endomucin and CD150 antibodies. Sections were stained, mounted, and examined using confocal microscopy.

Flow Cytometry, Cell Sorting, and Data Analyses

Bone marrow (BM) single-cell suspensions were prepared in IMDM with 2% FBS. Immunostaining used antibodies from Pharmingen (BD Biosciences), Invitrogen (Thermo Fisher Scientific), and Miltenyi Biotec. Flow cytometry utilized BD antibodies. Flushed BM cells from Gata2^+/+ or Gata2^R396Q/+ mice were enriched in the lineage-negative cell fraction using EasySep Mouse Hematopoietic Progenitor Cell Isolation Kit (Stem Cell Technology, MethoCult GF M3434). Cell sorting of LSK or LT-HSC, ST-HSC, and MPP3-4 was performed using a BD FACSAria™ III Cell Sorter with DIVA software. Flow cytometry data were analyzed using FlowJo Version 10.8 and normalized using the FlowAI package (V2.1) [24]. Arc sin transformation was used for TriMap dimensional reduction analysis [25].

Transcriptomic Analysis

Total RNA from 60,000 purified LSK Gata2^+/+ and Gata2^R396Q/+ mice was isolated using the Rneasy Plus Mini Kit (Qiagen, #74134). RNA-seq libraries were prepared with the TruSeq Stranded mRNA Low Sample kit, starting with 300 ng of total RNA. Cluster generation and sequencing were conducted on the Illumina NextSeq 500/550 High Output Kit v2.5 with a read length of 2 × 75 nucleotides. Specific LSK subpopulations were purified by cell sorting, RNA extracted, and Smart-seq v4 libraries prepared. Paired-end sequencing was performed on an Illumina NextSeq 500, and data was processed using fastp, RNA-STAR, and featurecount.

ATAC-Seq

Sorted LSK populations were prepared using the Active-Motif ATAC-Seq kit (#53150). Libraries were assessed on a Tapestation with a DNA High Sensitivity kit and sequenced on a NextSeq 500/550 with an average of 125 million reads per sample. FastQC was used to assess sequence quality. Foreign sequences were removed and trimmed with Sickle. Sequences were mapped to the murine genome with Bowtie2, and various cleaning and filtering steps were performed. GC bias was diagnosed and corrected using deepTools. The pre-processed output was used for peak calling with MACS2 and DiffTF analysis was conducted using the diffTF pipeline.

Functional in Vivo Assays

For BM reconstitution assays, myeloablation was induced by two doses of 5-Fluorouracil (Accord) administered intraperitoneally at days 0 and 11. Acute inflammatory stress was induced using lipopolysaccharide (Sigma #L2630) injected intraperitoneally at 5 mg/kg in 8-week-old mice. Mice were euthanized after 16 or 96 h, and bone marrow cells were characterized by flow cytometry. Transplantations of LSK or LT-HSC cells were performed on myeloablated mice, and cell chimerism quantified by flow cytometry.

WGS and RNA-Seq

Data were generated by France medicine genomic platform (detailed in supplementary methods)

Statistics

Statistical differences were assessed using a two-tailed unpaired Student’s t-test or Mann-Whitney test as appropriate. Error bars denote the standard deviation (SD) or standard error of the mean (SEM) as indicated. Survival variances were evaluated using the log-rank test. A p-value below 0.05 was considered statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). All analyses were conducted using GraphPad Prism software, version 7.

Results

Disruption of early HSC development by GATA2
^R396Q mutation

Reported nonsense and frameshift coding mutations occur upstream of the second zinc finger (C-ZF), resulting in haploinsufficiency [14]. GATA2 germline missense mutations cluster on the second C-ZF and are associated with a higher risk of leukemic transformation [16, 26]. Among them, arginine 396 is a mutational hotspot. The GATA2^R396Q mutant maintains its nuclear localization, as wild-type GATA2, contrasting with the cytoplasmic localization observed with the truncated variant GATA2^R204* lacking the nuclear localization signal (Supplementary Fig. 1A-B).

Previous report indicates that GATA2^R396Q exhibits notably reduced DNA-binding and transactivation capabilities [26], suggesting an altered functional phenotype. To further analyze its functional consequences, we generated a Gata2^R396Q/+ knock-in mouse (Supplementary Fig. 1C-D). Breeding Gata2^R396Q/+ and Gata2^+/- mice resulted in the expected genotypes at E12.5 when the yolk sac hematopoiesis ends [27] (Supplementary Fig. 1E). In contrast, Gata2^R396Q/- embryos exhibited a translucent fetal liver (Fig. 1A) with an absence of CD150⁺ hematopoietic stem cells (Fig. 1B). No Gata2^R396Q/- pups were present at birth (Supplementary Fig. 1F), similarly to the phenotype observed in Gata2^-/- mice [3], demonstrating that Gata2^R396Q is ineffective in rescuing the wild-type Gata2 allele.

**Fig. 1: Impaired emergence and differentiation of hematopoietic cells due to recurrent Gata2^R396Q mutation.**

At E14.5, Gata2^R396Q/+ embryos exhibited an increase in fetal liver LSK cell numbers compared to WT embryos, primarily due to a significant increase in Long-Term (LT)-HSCs and Multipotent Progenitors 2 (MPP2) cells while Short-Term (ST)-HSC cells decreased (Fig. 1C, Supplementary Fig. 2A). This substantial impact of the Gata2^R396Q mutant on hematopoiesis persisted 2 months after birth, with an additional increase of MPP3 cells (Fig. 1D, Supplementary Fig. 2B). The rise of the number of Gata2^R396Q/+ MPP2 cells was associated with increased quiescence, suggesting that their increase is unlikely to be related to cell cycle dysregulation (Supplementary Fig. 2C). Notably, the effect of the Gata2^R396Q mutant was significantly different from the loss of one Gata2 allele, as evidenced by the Gata2^+/- mouse showing a reduction in both LT-HSCs and ST-HSCs (Supplementary Fig. 2D-E).

In WT mice, Gata2 expression was prominent in LT-HSCs and MPP2, consistent with its role in self-renewal and influencing erythroid and megakaryocyte differentiation, whereas Gata2^R396Q/+ mice showed comparatively higher Gata2 levels in all compartments (Fig. 1F, Supplementary Fig. 2G). A specific Gata2^high CD150^High LT-HSC population is present in the mutant mice, the population also Kit and Sca-1 low (Fig. 1E, F and Supplementary Fig. 2B, F) and absent from the Gata2^+/- mice (Supplementary Fig. 2D-F).

We examined the protein expression of Gata2 and one of Gata2’s canonical target genes, Gata1, in the Lin- Kit+ Sca-1- (LK) compartment as Gata1 is poorly expressed in LSK. While Gata2 protein is mainly increased in the GMP subset, Gata1 expression is reduced only in MEP cells of Gata2^R396Q/+ mice compared to WT mice (Fig. 1G, Supplementary Fig. 2H-I), suggesting that Gata2^R396Q may affect the expression of Gata2 target genes independently of the overall GATA2 protein load in a specific cell subset.

Gata2^R396Q affects the mature hematopoietic compartment

We analyzed the more mature hematopoietic compartment in 2-month-old Gata2^R396Q/+ mice. LK progenitors in the bone marrow were decreased due to a reduced CMP and MEP cell compartment compared to WT and Gata2^+/- mice (Fig. 2A, Supplementary Fig. 3A-D). We observe a decreased hemoglobin and platelet counts in the blood of Gata2^R396Q/+ mice (Fig. 2B) likely due to an increased apoptosis rate in mature mutant cells (Fig. 2C, D), despite a similar absolute number of mature cells in the bone marrow including myeloid, B, and T cells (Fig. 2E). Such abnormalities were not detected in Gata2^+/- mice (Supplementary Fig. 3D-E), supporting the notion of an ectopic function associated with this recurrent missense mutation.

**Fig. 2: Impact of the mutant on hematopoietic differentiated compartment.**

Hyporesponsiveness and inflammatory molecular signatures are a hallmark of Gata2^R396Q

To elucidate the molecular mechanisms underlying the impact of Gata2^R396Q on hematopoiesis, we performed comprehensive analyses of gene expression and chromatin accessibility using purified Gata2^R396Q/+ and Gata2^+/+ LSK cells from 2-month-old mice (Fig. 3A). We assessed TF accessibility [28] and observed a significant reduction in accessibility for Meis [29] and Irf [30] associated motifs along with a decreased expression, suggesting an impact on hematopoietic commitment and interferon response, respectively (Fig. 3B, Supplementary Fig. 4A). Conversely, motifs linked to TF associated with cell quiescence (e.g., Egr) or BMP signaling (e.g., Smad) showed substantial increase in activity and expression (Fig. 3C, Supplementary Fig. 4B). Interestingly, GATA motifs showed only a marginal decrease in accessibility in the mutant cells (Fig. 3B).

**Fig. 3: Hyporesponsiveness and exhaustion signatures in mutated LSK cells.**

Gata2 is a component of a heptad transcription complex critical for defining HSPC identity [31] and among these 7 factors, only Erg and Fli1 showed increased activity without concomitant changes in expression (Supplementary Fig. 4C-D), suggesting a potential disruption of the complex. Subsequent analyses focusing on regions of differential accessibility, identified both enrichments and depletions of Gata motifs across the genome in mutated LSK cells, explaining the absence of widespread repression (Fig. 3D, E). The accessibility of regions containing ETS motifs was predominantly located at promoters (Fig. 3D, E), highlighting the significant role of these motifs in regulating gene transcription (Fig. 3D, E). However, regions containing both Gata2 and ETS motifs were less accessible at key hematopoietic genes such as Meis, Cd34, and Gata2 itself (Supplementary Fig. 4D). In contrast, regions specifically associated with ETS motifs alone, particularly at enhancers and promoters, tended to be more accessible. This increased accessibility correlated with higher expression levels of genes related to hematopoiesis, such as Slamf1 (encoding CD150), Mmp8, and Smad1 (Supplementary Fig. 4D).

The Over-Representation Analysis (ORA) uncovered concurrent changes in cellular processes at both chromatin accessibility and expression levels in Gata2^R396Q/+ LSK cells. These cells exhibited enrichment in pathways associated with innate inflammatory response, migration, and extracellular matrix degradation (Fig. 3F, G, Supplementary Fig. 5A-B). Futhermore, the observed depletion of interferon-gamma response pathways (Fig. 3F, G, Supplementary Fig. 5A) was correlated with an observed reduction in Irf expression and accessibility (Supplementary Fig. 4A), suggesting a diminished capacity to respond to stimuli. In conclusion, our analysis of the transcriptional landscape and network of mutated LSK cells revealed specific signatures of inflammation, exhaustion and hyporesponsiveness.

Gata2^R396Q expression is associated with a loss of HSCs identity in the mutant mice

We investigated expression profiles by RNA sequencing in sorted LT-, ST-HSC, and MPP3-4 compartments. Notably, Gata2^R396Q expression displayed elevated allelic expression levels compared to the wild type in LT- and MPP3-4 (58% and 67% respectively; Fig. 4A), unveiling distinct allele-specific expression (ASE) patterns. Differential gene expression observed in sorted LT-HSCs and MPP3-4 compared to ST-HSCs is consistent with ASE (Fig. 4A, B). Notably, Gata1 and Gfi1b, genes activated by Gata2, showed significantly lower expression in LT-HSCs and MPP3-4, while Cebpa, a gene repressed by Gata2 [32], exhibited increased expression (Fig. 4C). Additionally, key genes for HSPC functions (Epor, Esam, Aqp9, Gpc3) have a lower expression in LT-HSCs and MPP3-4 (Supplementary Fig. 5C). GSEA analysis revealed notable dysregulation of pathways associated with protein synthesis, myeloid differentiation, and TNFα response in all subpopulations, indicating an influence of Gata2^R396Q expression level on these processes. Conversely, repression of heme metabolism and interferon-gamma responses showed variability depending on ASE status, suggesting direct involvement of mutant functions. (Fig. 4D).

**Fig. 4: Molecular profiling reveals dysregulated pathways and gene expression patterns in *Gata2*^R396Q/+ hematopoietic stem and progenitor cell subpopulations.**

To go deeper, we used molecular signatures associated with the hallmark of HSC properties. Both The MoIO [33] signature, linked with superior HSC function, and the NoMO [33] signature which is associated with less quiescent and functionally inferior HSCs, are overrepresented in wild-type LT-HSCs compared to mutated cells (Fig. 4E). Notably, specific genes within the MolO signature like Neo1, Smtnl1, Cd82 and Vwf exhibited reduced expression in mutated LT-HSCs whereas Ly6a and Procr genes encoding Sca-1 and Epcr respectively showed increased expression (Fig. 4F). Similarly, most of the NoMO signature genes, including Muc13, Itga2b, Gata1 or Aqp9, displayed decreased expression in mutated LT-HSCs while Nkg7 expression increased (Fig. 4C, G). Notably, the expression of Dlk1, a gene associated with increased myeloid potential, quiescent state, and self-renewal capacity [34], was significantly decreased in mutated LT-HSCs (Supplementary Fig. 5D). Furthermore, mutated LT-HSCs expressed genes not typically found in hematopoietic cells, such as Igf1, Sfrp2, S1pr3 or Nsg2 (Fig. 4H and Supplementary Fig. 5E). These findings indicate that mutated LT-HSCs undergo a loss of HSC features, encompassing both dormant and active HSC properties and acquire inappropriate markers.

Gata2^R396Q mutation impairs HSCs development and function, mostly at LT-HSC stage

To assess the impact of these molecular features associated with Gata2^R396Q expression, we performed clonogenic assays from 2-month-old mice. Gata2^R396Q/+ cells exhibited a significant decrease in CFU-GEMM colonies (Fig. 5A). When subjected to serial passages, these cells displayed an increase in colony number (Supplementary Fig. 6A). Morphological and phenotypic analyses revealed a predominance of granular precursors and mature granulocytes in Gata2^R396Q/+ cells whereas mast cells were more prevalent in Gata2^+/+ condition (Supplementary Fig. 6B-C). Gata2^+/- LSK cells have a reduced number of CFU-GM without increased clonogenic activity. As in Gata2^+/+, mast cells were the most prominent cells observed in the Gata2^+/- condition (Supplementary Fig. 6D-F).

**Fig. 5: Functional characterization of hematopoietic stem cells in *Gata2*^R396Q/+ mice under challenging conditions.**

To evaluate bone marrow repopulation capacity, myeloablation was induced with 5-FU, revealing that Gata2^R396Q/+ mice displayed reduced hematopoietic reconstitution ability, resulting in a significantly higher mortality rate compared to Gata2^+/+ mice (Fig. 5B). In a syngeneic transplantation experiment, Gata2^R396Q/+ LSK cells exhibited reduce repopulation capacity compared to their Gata2^+/+ counterparts after 2 months, with all subpopulations showing reduced expansion. This engraftment deficit worsened after 6 months, indicating a severe impairment of Gata2^R396Q/+ LT-HSCs (Fig. 5C-F). Secondary transplantation of total bone marrow from the primary transplanted mice confirmed the loss of long-term engraftment capacity of Gata2^R396Q/+ cells (Fig. 5E). Overall, transplantation assays uncovered a long-term exhaustion of Gata2^R396Q/+ cells, as evidenced by a decrease in self-renewal capacity.

Furthermore, to assess the functional consequences of the Gata2^R396Q mutation on hematopoiesis under challenge conditions, we determined the hematopoietic response after a single dose of lipopolysaccharide (LPS), mimicking an acute bacterial infection (Fig. 5G). LSK cells exhibited an increased proliferation in response to inflammatory stress [35] (Supplementary Fig. 7A) triggering emergency myelopoiesis [36]. After16 hours, we observed an increase in MPP2 and MPP3 numbers in Gata2^R396Q/+ mice but to a lesser extent than in Gata2^+/+ LSK cells (Supplementary Fig. 7B-C).

Additionally, 96 h after LPS injection, similar patterns were still observed in all cell populations. This persistence suggests that the Gata2^R396Q mutation may affect the onset of the response rather than its magnitude (Fig. 5H and Supplementary Fig. 7C). In Gata2^R396Q/+ LT-HSCs, we observed two phenotypically distinct populations based on Gata2 and CD150 protein levels (Fig. 1E). Under acute inflammatory stress with LPS, Gata2^R396Q/+ mice unveiled an aberrant LT-HSCs with higher CD150 and Sca-1 marker intensities and lower Kit intensity (Fig. 5H, red arrow, Supplementary Fig. 7D). Ki67 labeling revealed that CD150^high aberrant LT-HSCs cycled less than CD150^low and WT LT-HSCs under steady-state conditions, and the CD150^high LT-HSC population exhibited hyporesponsive cycling in response to LPS stimulation (Fig. 5H, I, Supplementary Fig. 7E). Sixteen hours after LPS treatment, the majority of Gata2^+/+ LT-HSCs expressed CD41 (Itga2b-encoded platelet marker) compared to a fraction of Gata2^R396Q/+ LT-HSCs (Fig. 5H, grey arrow).

Consequently, the Gata2^R396Q mutation results in a hematopoietic functional impairment with reduced responsiveness and exhaustion following both short- and long-term inflammatory stress.

Functional decline in Gata2^R396Q LT-HSCs induces a specific inflammaging

Given the presence of the CD150^high LT-HSC population and the exhaustion and hyporesponsiveness functionality observed in LSK cells from young mutant mice, we examined one-year-old mice, representing a midpoint in their lifespan. Gata2^R396Q/+ mice exhibited higher LSK cell numbers, along with increased LT-HSC, MPP2, and MPP3 cells, while MPP1 and ST-HSC were significantly reduced (Fig. 6A, Supplementary Fig. 8A). The increase in LT-HSC numbers was more pronounced at one year than at 2 months (Fig. 6B). Repeated measurements of blood parameters revealed a progressive anemia in Gata2^R396Q/+ mice while the platelet count increased rapidly, suggesting a differentiation bias in favor of platelet over erythroid development (Fig. 6C). Because of the cellular phenotypic features resembling a premature aging phenotype, we compared the significant dysregulations observed in mutant LT-HSCs with those found in old versus young WT LT-HSCs [37], revealing similar deregulated pathways, such as an increased enrichment of oxidative phosphorylation and proteasome-related pathways and a decrease of Kit and Cxcr4 pathways (Fig. 6D).

**Fig. 6: Functional decline of HSPC in older *Gata2*^R396Q/+ mice.**

While increased interferon signaling and inflammation are typically associated with aging [30], these markers were not elevated in young Gata2^R396Q/+ HSPCs, suggesting an unconventional premature aging phenotype (Fig. 6E). Physiological hematopoietic aging is associated with changes in LT-HSCs, including increased expression of the CD150 marker, presence of megakaryocytic surface markers such as CD41 or CD61 [38, 39], and upregulation of Gata2 (Supplementary Fig. 8B) and its target gene Gata1 [40], correlating with a platelet bias. We identified an aberrant CD150^high/CD41^–/CD61^low/- LT-HSCs in middle-aged Gata2^R396Q/+ mice, which progressively expanded over time and represented the majority of the LT-HSCs population (Fig. 6F, G). This population exhibited a reduced functionality as evidenced by decreased clonogenic capacity and multipotency with a notable decrease in CFU-GEMM and CFU-Mk colony numbers in clonogenic assay, correlating with the lack of CD41 protein expression (Supplementary Fig. 8C-D).

Allele-specific expression drives functional defects in LT-HSCs and disease progression in GATA2 R396Q model

Based on our identification of an ASE mechanism within HSPC subpopulations, we investigated into the prospect of different allelic imbalances in these two subpopulations. We sorted CD150^high and CD150^low LT-HSCs from 2- and 12-month-old Gata2^R396Q/+ mice and measured the variant allelic frequency (VAF) of the Gata2 mutation at the RNA level. Our results revealed that Gata2^R396Q VAF_RNA is higher in CD150^high compared to CD150^low LT-HSCs (73% vs 53% at 2 months and 79% vs 58% at 12 months, respectively), providing evidence for a persistent ASE mechanism in the aberrant CD150^high LT-HSCs (Fig. 6H).

In LT-HSC subpopulation transplantation assay, we observed a negative relationship between the level of donor chimerism and Gata2 VAF_RNA. Transplantation of CD150^high LT-HSC subpopulation resulted in a severe repopulation defect in recipient bone marrow one-month post-injection (Fig. 6I), confirming the hypofunctionality of the CD150^high LT-HSCs. On the other hand, transplantation of CD150^low underscored a milder defect in BM repopulation capacity compared to Gata2^+/+ LT-HSCs.

Interestingly, these functional defects were also detected in old mutant mice at steady state. Phenotypic characterization of 18-month-old Gata2^R396Q/+ mice revealed a significantly reduced hemoglobin levels (6 g/dL compared with 10 g/dL for the Gata2^+/+ mouse of the same age), enlarged spleen with extramedullary erythropoiesis and a substantial decrease in bone marrow cells (Supplementary Fig. 8E-F). The LSK compartment is predominantly composed of CD150^high LT-HSC cells (Supplementary Fig. 8G) suggesting bone marrow hypoplasia could be due to the exhaustion of the most functional LT-HSCs (CD150^low). Our RNA-Seq analysis of a patient with the germline missense GATA2^R362G (c.1084 C > G) mutation [16, 41], confirmed the presence of ASE with higher relative expression of the mutated allele at leukemic transformation (Fig. 6J). These findings highlight the clinical relevance of the ASE mechanism in human GATA2 deficiency syndrome and its potential implications for disease development and progression.

Discussion

Our work comprehensively illustrates the impact of the GATA2^R396Q germline mutation on HSPCs, generating a Gata2^R396Q/+ knock-in mouse model. This model shows increased LT-HSC numbers, diminished self-renewal potential, and impaired responsiveness to inflammatory stimuli.

While the missense GATA2^R396Q mutant maintains nuclear localization unlike truncated mutants [15, 42], it doesn’t rescue the wild-type allele, indicating a global loss-of-function mutation.

Interestingly, three GATA2 missense mutations—Gata2^R396Q, Gata2^R398W, and Gata2^L359V (our study; [17, 18])—have been introduced into mouse models, each with distinct molecular impact. Structurally, leucine 359 interacts with DNA in the major groove, while arginine 361 and R396 are essential for recognizing the WGATAR motif and interacting with thymine and adenine residues. R362 and R398 enhance DNA binding by stabilizing interactions. The L359V mutation swaps one hydrophobic residue for another, whereas R396Q and R398W reduce positive charge, affecting DNA binding differently. Homology modeling suggests arginine mutations disrupt DNA interactions but leave the second zinc finger structure intact [26, 43].

Despite reduced DNA-binding activity [26, 42], Gata2^R396Q/- embryos survive longer than Gata2^-/- embryos [3], suggesting potential ectopic functions. Unlike models with lower Gata2 expression, such as Gata2^+/- or Gata2^+9.5+/- models [5, 44], our Gata2^R396Q/+ model, like Gata2^R398W/+, does not exhibit a decrease in overall Gata2 levels within the LSK compartment [17]. Instead, we observed a notable increase in global Gata2 protein levels, particularly in LT-HSCs, MPP2, and MPP3, with no alterations in ST-HSCs. These changes mirror the abnormal LSK subpopulation distribution from embryonic stages throughout their lifespan. This distribution pattern is distinct from Gata2^+/- and Gata2^R398W/+ models in which a reduction or no difference in LT-HSC proportions is observed, respectively [5, 17, 45]. Furthermore, GATA2^R396Q mutant affects the more mature HSPC compartment by reducing the CMP and MEP compartments inducing anemia over time while platelet number increase rapidly, indicating differentiation bias in favor of platelet over erythroid development.

Despite higher LT-HSC counts mutant mice, repopulation and serial transplantation assays unveiled impaired LSK functionality. Chromatin accessibility and gene expression profiles in young mice showed reduced accessibility for motifs linked to commitment and interferon response, compromised infection responses to mycobacteria for example in patients [43]. In addition, GATA2 deficient patients bone marrow with spectrum 1 showed a higher proportion of MEP and a lower proportion of GMP than aplastic anaemia patients [41, 46]. Under chronic inflammation, this differentiation bias is due to exhaustion of CMP and GMP populations, rather than MEP proliferation.

Gata2 is a constituent of a crucial heptad of TF defining stemness [31]. Our findings demonstrate that the Gata2^R396Q mutation selectively impairs DNA binding at the heptamer binding site, but does not affect other TF expression. Furthermore, we observe increased accessibility specifically at Fli1/Erg motifs, resulting in higher expression of BMP pathway genes like the Smad family. This supports previous research showing a crosstalk between Smad signaling and Gata2 [47, 48] and may explain Gata2 levels similar to WT cells [32]. There is also a correlation between CD150 and Gata2 expression levels in WT HSCs [49]. In mutated HSPCs, Fli1 sites at Slamf1 (encoding CD150) transcriptional regulatory elements show heightened accessibility. This suggests a compensatory response to the diminished Gata2 activity.

Hence, we identify a mutant specific CD150^high-expressing LT-HSCs population in young Gata2^R396Q/+ mice compared to WT mice, showing similarities to certain features of an aging. However, some pathways were not elevated in young Gata2^R396Q/+ HSPCs compared to aged wild-type HSPCs, indicating an unconventional premature aging phenotype [37].

Mutated LT-HSCs exhibited a notable reduction in the expression of Neo1, a component of the MolO signature. In HSPC from Neo1-mutant mice, the NoMO signature was enriched along with increased Gata1 and Itga2b expressions [50]. In Gata2^R396Q/+ mouse model, the physiological upregulation of Gata1was hindered leading to the absence of enrichment of the NoMO signature in mutated LT-HSCs. Additionally, mutated LT-HSCs In addition, mutated LT-HSC express genes from cells in the microenvironment. This is in line with findings from De Pater’s group [51], which demonstrated impaired suppression of endothelial identity in Gata2^+/- HSPCs during embryonic maturation due to reduced Gfi1b activity, as observed in Gata2^R396Q/+ LT-HSCs. These observations suggest a loss of molecular stemness identity in mutated HSCs.

LPS challenge uncovered functional diversity within the LT-HSC compartment, notably linked to CD150 expression levels. While CD150^low mutant LT-HSCs behaved similarly to WT cells, CD150^high cells displayed pronounced hyporesponsiveness. The CD150^high LT-HSC population exhibited the highest expression levels of the mutant allele, suggesting an allele-specific expression mechanism at play. This ASE mechanism has previously being reported in the context of GATA2 [21]. Interestingly, Al Seraihi and colleagues have reported that differential allele expression can influence the clinical phenotypes in a GATA2 deficient patient [22]. Our findings showed that ASE is present in mice as early as 2 months old and might persist even in the context of a hematologic progression in patients. This raises questions about potential secondary events that could favor the selection of cells demonstrating ASE. Furthermore despite the existence of some genotype-phenotype correlations [41], questions remain regarding the wide range of symptoms observed even within a same family. The investigation of ASE mechanism could provide valuable insights into understanding these divergences.

In conclusion, our study provides compelling evidence for the substantial impact of a missense Gata2 mutation on the HSPC compartment during hematopoietic development. The alterations at phenotypic, functional, and molecular levels highlight the crucial role of Gata2 in maintaining HSPC homeostasis. The disruption of critical molecular pathways in mutated HSPCs hampers their ability to respond appropriately to environmental signals and may contribute to their exhaustion. These insights significantly advance our understanding of initiation of bone marrow failure in GATA2 deficiency. Therefore, our findings hold promising implications for improving clinical follow-up for patients with GATA2-related conditions, providing valuable insights into the mechanisms involved in maintaining HSPC homeostasis.