C3G promotes bone marrow adipocyte expansion and hematopoietic regeneration after myeloablation by enhancing megakaryocyte niche function

Abstract

C3G, a Rap1 GEF, promotes megakaryopoiesis and platelet function. Using transgenic and knock-out mouse models targeting C3G in megakaryocytes, we investigated whether C3G also affects the niche function of megakaryocytes during bone marrow (BM) recovery after myeloablation induced by 5-fluorouracil (5-FU), or total body irradiation (TBI) followed by bone marrow transplantation. C3G promoted megakaryocyte maturation and platelet production during recovery, along with increased white and red blood cell counts and enhanced survival of female mice after repeated doses of 5-FU. Additionally, megakaryocytes favored adipocyte differentiation through a C3G-mediated mechanism, likely involving Fgf1. Changes in the number or behavior of BM megakaryocytes and adipocytes influenced the hematopoietic stem cell pool, with C3G promoting its bias towards the myeloid-megakaryocytic lineage in both 5-FU- and TBI-ablated models. Therefore, C3G could be a potential target in therapies aimed at enhancing hematopoiesis in patients undergoing chemotherapy and/or BM transplantation.

Background

C3G (gene name RAPGEF1) is a Guanine nucleotide Exchange Factor (GEF) that primarily activates Rap1 GTPases [1]. It plays a pivotal role in platelet functions such as activation, aggregation [2, 3], exocytosis, spreading, and clot retraction [4, 5]. C3G also contributes to platelet-mediated angiogenesis in various pathological contexts [6]. On the other hand, the GTPase Activating Protein (GAP) Sipa1 is the main negative regulator of Rap1 GTPases in the hematopoietic system, acting as a tumor suppressor in hematological malignancies [7].

Beyond its involvement in platelet functions, C3G regulates megakaryopoiesis and thrombopoiesis under stress conditions, such as tumor cell implantation [8] and 5-fluorouracil (5-FU)-induced bone marrow (BM) depletion [6]. This positive role of C3G in differentiation has also been reported for myocytes, adipocytes, neuroblasts [9] and embryonic stem cells [10]. C3G also participates in the differentiation of oval cells into hepatocytes [11]. In addition, our previous findings indicated that C3G overexpression in megakaryocytes (MKs) enhances the generation of bone marrow adipocytes (BMAs) following melanoma cell implantation [8], which supports a role for MK C3G in adipocyte expansion in response to stressors.

MKs are well-known niche regulators of hematopoietic stem cell (HSC) quiescence and activity through interactions with endothelial cells in the vascular niche [12, 13]. In resting conditions, MKs secrete factors like TGFβ1, PF4 (also known as CXCL4) and IL-6 to maintain HSC quiescence and self-renewal [12, 14]. MKs also express thrombopoietin (TPO), which further contributes to HSC quiescence [15]. In fact, ablation of MKs favors HSC proliferation [12]. However, the MK population increases following 5-FU-induced myeloablation and promotes HSC expansion by producing FGF1 [16]. MKs originate from an oligopotent megakaryocytic/erythroid progenitor (MEP) [17], but they can also arise directly from HSC or multipotent progenitor (MPP) subsets. Indeed, TPO directly promotes MK generation from biased progenitors [18].

In addition to MKs, the BM niche comprises various other cell populations including osteoblasts, osteoclasts, adipocytes, mesenchymal stem cells, and endothelial cells, among others [19]. The BM niche enhances the ability of HSCs to engraft and repopulate in transplant settings or following chemotherapy-induced myeloablation [20, 21]. To mobilize HSCs after BM depletion, disruption of the CXCR4/SDF1 axis is necessary [22], while SDF1 (also known as CXCL12) production by BM mesenchymal stromal cells is crucial for successful cell engraftment after transplantation [14, 23].

During a BM disorder or failure, such as that caused by 5-FU or irradiation, the body can initiate extramedullary hematopoiesis (EMH) to overcome the inadequate production of blood cells. As a result, organs such as the spleen take on this task, which could be reflected in splenomegaly [24].

On the other hand, it has been proposed that mature BMAs act as negative modulators of HSCs by secreting TNFα and adiponectin, thereby preserving the HSC pool [25, 26]. In fact, ablation of BMAs positively influences engraftment after BM transplantation [25]. However, BMAs contribute to hematopoietic regeneration after injury by producing SCF [26, 27], and their abundance in the BM increases following irradiation or chemotherapy, concomitantly with HSC proliferation [27, 28]. It has been described that BMAs and MKs share a reciprocal relationship. Adipocyte delipidation enhances MK maturation through the CD36 transporter present on the MK membrane, a process likely regulated by an as-yet unidentified delipidation signal from MKs [29, 30]. On the other hand, factors that regulate adipocyte differentiation, such as FGF1 [31], IGF1 [32], PREF1 [33] and IL-6 [34], are released by MKs within the BM [14, 35]. However, no direct link between these observations has been established thus far.

In this study, we investigated the role of C3G in the niche function of MKs following BM depletion induced by chemotherapy or irradiation. Our findings suggest that MK C3G positively contributes to hematopoietic recovery after myeloablation, enhancing survival. Specifically, MK C3G promotes BMA maturation and expansion both, in vitro and in vivo, through the release of factors from MKs that facilitate, directly or indirectly, adipocyte differentiation. Additionally, MK C3G supports the myeloid branch of hematopoiesis during recovery, particularly influencing the megakaryocytic and erythroid lineages.

Methods

Mouse models

The MK-specific transgenic (tgC3G^Pf4) and conditional knock-out (Rapgef1^fl/fl;Pf4-Cre^+/−, hereinafter C3G^Pf4-KO) mouse models for C3G used in this work have been described and characterized previously [2,3,4,5,6, 8]. wtC3G^Pf4 and Rapgef1^fl/fl;Pf4-Cre^−/− (hereinafter C3G^Pf4-wt) are their respective control mice. C57BL/6 Ly5.1-expressing mice (Ly5.1^+/+) were purchased from Charles River Laboratories (France) for transplant experiments. The Sipa1-KO mouse model was described in [7, 36]. All mice used were 6- to 10-week-old.

5-FU-induced myeloablation

5-FU was freshly prepared in 0.9% NaCl with 10% DMSO at 37 ºC with vigorous shaking. Mice were injected intraperitoneally (i.p.) with a sublethal dose of 150 mg/kg 5-FU (Merck Millipore, #F6627) to induce myelosuppression. Their recovery was then assessed for 21 days.

To test the effect of C3G on sensitivity to noxious injury, mice were injected weekly with 150 mg/kg 5-FU and monitored for survival. According to ethical guidelines, mice were euthanized when they showed a 20% body-weight reduction or two or more moderate grimaces of pain [37].

Bone marrow extracts

BM from the femur and tibia was collected by centrifugation method [38, 39]. Briefly, a femur and tibia were cut open and placed in nested tubes. BM was collected by spinning out the bones onto 100 µL of PBS containing 10% FBS. A homogeneous cellular suspension was prepared by adding 900 µL of PBS with 10% FBS and gently pipetting up and down. This BM extract was used for further applications.

For experiments involving Sipa1-KO and wtSipa1 mice, BM extracts were obtained by crushing the femur and tibia with a mortar and pestle in PBS with 10% FBS and filtering the resulting cell solution through a 70-µm filter.

Evaluation of BM cellularity

Twenty microliters of a 1/10 dilution of the BM extracts (from one femur and one tibia) were mixed with 380 µL of Muse® Count & Viability reagent (Merck Millipore, #MCH100102). This solution was then analyzed in a Muse® Cell Analyzer (Merck Millipore) using the appropriate software module.

Analysis of hematological parameters

Whole blood was collected via submandibular vein puncture in EDTA-coated tubes (Sarstedt, #41.1395.005) and analyzed using an ADVIA 120 Hematology System (Siemens AG).

Determination of platelet number by flow cytometry

In addition to hemocytometry analysis, platelet counts were also determined using a BD Accuri™ C6+ cytometer, as previously described [3].

Immunophenotyping of HSPCs

Hematopoietic stem and progenitor cell (HSPC) populations were identified according to the scheme by [40]. Briefly, erythrocytes from BM extracts were lysed in ACK buffer (0.155 M NH₄Cl, 10 mM KHCO₃, 10 mM EDTA pH 7.4) for 5 min at 4 ºC. After washing and resuspension in the same buffer, cells were stained with the following antibodies: FITC-Lin (#133302, 1/100), PE-c-Kit (#105,808, 1/333), APC-Sca-1 (#108112, 1/100), PE/Cy7-CD16/32 (#101318, 1/133), PerCP/Cy5.5-CD34 (#128608, 1/50), BV510-CD127 (#135033, 1/80), all from BioLegend, as well as PE/CF594-CD135 (#562537, 1/50), from BD Biosciences. One million cells were acquired in a BD FACSAria III flow cytometer.

In experiments involving BM isolated from transplanted mice, the PE/Cy7-CD16/32 antibody was replaced by PE/Cy7-CD45.2 (#109830, 1/100) antibody, and Pacific Blue™-CD45.1 (#110722, 1/100) (both from BioLegend) was added to the cocktail to distinguish donor and host cells.

Immunophenotyping of mesenchymal stem cells and adipocyte-committed progenitors

BM CD45⁻CD31⁻Sca-1⁺ cells were identified as mesenchymal stem cells (MSCs) if they were positive for the CD24 marker; or as adipocyte-committed progenitors (APCs), if they were negative for CD24, according to [28]. BM cells were erythrocyte-lysed, washed, and resuspended in PBS with 3% FBS. Cells were stained with FITC-CD31 (#102405, 1/100), FITC-CD45 (#103,107, 1/100), PE-CD24 (#101807, /100) and APC-Sca-1 (#108112, 1/75) from BioLegend, for 30 min on ice in the dark. Samples were acquired using a BD Accuri™ C6+ cytometer. At least 5 × 10⁵ events were recorded.

Immunophenotyping of mature lineages

Whole blood was collected via submandibular vein puncture in EDTA-coated tubes (Sarstedt, #41.1395.005). After lysing erythrocytes, cells were resuspended in 50 µL of PBS with 3% FBS and stained with the following antibody combinations (from BioLegend): PE-CD11b (Mac-1, #101207, 1/100) and APC-Ly-6G/Ly-6C (Gr-1, #108412, 1/100) for the identification of neutrophils, FITC-CD4 (#100509, 1/100), PE-CD8a (#100707, 1/100) and APC-CD3 (#100236, 1/100) for detecting T lymphocytes, and FITC-B220 (#103205, 1/100) for detecting B lymphocytes.

In experiments involving the analysis of peripheral blood (PB) from transplanted mice, PerCP/Cy5.5-CD45.1 (#110728, 1/100) and APC-CD45.2 (#109814, 1/100) antibodies from BioLegend were added to the cocktails to distinguish donor and host cells. In this setting, FITC-Ly-6G/Ly-6C (Gr-1, #108405, 1/100) antibody was used instead of APC-Ly-6G/Ly-6C, and the APC-CD3 antibody was omitted.

In all cases (except for transplant settings), cells were resuspended in PBS containing 3% FBS and 1 μg/mL propidium iodide (PI) before analysis to exclude dead cells. Samples were acquired using a BD Accuri™ C6+ cytometer. At least 1.5 × 10⁴ events were recorded. All cytometry data were analyzed using FlowJo v10.7.1 software.

Bromodeoxyuridine (BrdU) incorporation

BrdU (MedChemExpress, #HY-15910) was prepared in PBS at a final concentration of 10 mg/mL. Mice were i.p. injected with a single dose of 100 mg/kg. Sixteen hours later, the mice were sacrificed, and bone marrow harvested. After lysing erythrocytes, cells were stained with lineage markers FITC-Lin (BioLegend, #133302, 1/100) for 20 min on ice in the dark and fixed with 4% paraformaldehyde (PFA, Sigma-Aldrich, #P6148) in PBS for 20 min at RT. Cells were then permeabilized by adding 300 µL of 0.5% Triton-X100 (Sigma-Aldrich, #X100) in PBS for 15 min at RT and treated with 30 µg of DNase I (Roche, #10104159001) in 100 µL PBS for 1 h at 37 °C. After washing, the cell pellet was resuspended in 50 µL PBS with 1% FBS and stained with PE/Cy7-BrdU antibody (BioLegend, #364117, 1/100) for 20 min at RT in the dark. Cells were analyzed using a BD FACSAria III flow cytometer, and data were analyzed with FlowJo v10.7.1 software.

Maturity of MKs

BM cells were stained with PE-CD61 (eBiosciences, #12–0611-81, 1/100), a MK marker, and APC-CD42 (BioLegend, #148505, 1/100), a marker of mature MKs [41]. At least 2 × 10⁴ double positive events were acquired using a BD Accuri™ C6+ cytometer. Analysis was carried out using FlowJo v10.7.1 software.

Hematoxylin–eosin (H&E) staining for MK and BMA counting

Femora were harvested and fixed in 4% formaldehyde for 24 h at 4 °C, following decalcification in Osteosoft (Sigma-Aldrich, #1.01728). Femora were paraffin-embedded, and sectioning was performed using a HistoCore Autocut microtome (Leica). Tissue sections were stained with H&E (Casa Álvarez, Spain, #10–3002) at the CIC Pathology Unit. Images were acquired with an Olympus BX51TF microscope coupled to an Olympus DP74 camera.

MKs were manually counted, as they are easily recognized as large pinkish cells with multilobulated dark purple nuclei. A semi-automatic approach was used for BMAs, which can be identified as large, rounded empty-appearing spaces with a thin bright cytoplasmic boundary in pink and peripheral nuclei. First, images were converted to 8 bits and a threshold was automatically established using “Triangle” algorithm. Second, open, close, and fill holes binary operations were applied. Finally, object recognition was checked, and BMAs not considered by the algorithm were manually identified. The numbers of MKs and BMAs were quantified relative to the bone marrow area.

Lineage-negative enrichment of BM via MACS

BM cell extracts from tgC3G^Pf4, C3G^Pf4-KO mice and their wild-type littermates were enriched for progenitors using magnetic-activated cell sorting (MACS). Briefly, BM cells were resuspended at a concentration of 200 × 10⁶ cells/mL in PBS with 10% FBS and the following purified antibodies from BioLegend added: anti-Gr-1 (#108401, 1/400), anti-CD11b (#101201, 1/400), anti-CD19 (#115501, 1/200), anti-CD3 (#100201, 1/200), anti-Ter119 (#116201, 1/200). Following incubation, washed Sheep anti-Rat IgG Dynabeads™ (Thermo Fisher Scientific #11035) were added, following manufacturer’s instructions, to capture mature cells. The cell suspension was then separated using the DynaMag™-2 magnet (Thermo Fisher Scientific, #12321D). Supernatants containing progenitor-enriched BM cells were recovered, centrifuged, and resuspended at 2 × 10⁶ cells/mL in DMEM with 20% FBS and 50 ng/mL TPO. Finally, 1 × 10⁶ cells were seeded in the upper chamber of a 0.4 μm cell culture insert to initiate the co-culture with 3T3-L1 cells (see next).

Adipokine evaluation of BM supernatants

Seven days after 5-FU administration, BM supernatants from at least three mice per genotype were pooled and 300 μL of sample was assayed using the Proteome Profiler Mouse Adipokine Kit (R&D Systems, #ARY013) according to manufacturer’s instructions. Blots were developed on Super RX-N X-ray films (Fujifilm) and scanned using an Epson V700-Photo scanner. Densitometric analysis was carried out using ImageJ/FiJi analysis software.

Co-culture of MKs and 3T3-L1 pre-adipocytes

To study the contribution of MKs to adipocyte differentiation and/or expansion, we established an indirect co-culture system based on [29]. Briefly, 1 × 10⁵ 3T3-L1 cells were seeded in 400 μL of maintenance medium (MM: DMEM + 10% calf serum (CS) + 1% penicillin/streptomycin) in a 24-well plate coated with 5 μg/cm² collagen type I (Corning, #354,236). After 24 h, MM was replaced with differentiation medium I (DM-I: MM containing 500 μM isobutylmethylxanthine (IBMX), 1 μM dexamethasone (DEX), 5 μM troglitazone (TROG) and 10 μg/mL insulin), for two days to initiate commitment. Subsequently, cells were cultured in differentiation medium II (DM-II: MM + 10 μg/mL insulin) until day 4. From day 4, two conditions were tested: (i) cells were maintained in DM-II or (ii) in MM (i.e., without insulin), with half of the culture medium replaced every 2 days until day 8.

To evaluate the potential contribution of MKs to this process, either wtC3G^Pf4/tgC3G^Pf4 or C3G^Pf4-wt/C3G^Pf4-KO lineage-negative (Lin⁻) cells (1 x 10⁶) isolated via MACS were seeded on day 4 in the upper chamber of a 0.4 µm Boyden chamber/Transwell insert (Costar®, #3470) in DMEM with 20% FBS and 50 ng/mL TPO to promote the differentiation of progenitor cells towards MKs. One Transwell insert without progenitor cells was used to account for effects derived from TPO addition. On days 6 and 8, BODIPY™ staining was performed to evaluate lipid content in differentiated 3T3-L1 cells using image analysis. Afterward, cells were lysed in NZYol reagent (NZYTech # MB18501) for RNA isolation.

For inhibition assays, TPO-stimulated Lin⁻ cells were treated with 70 µM suramin (MedChemExpress, HY-B0879A) to unselectively block growth factors secreted by HSPCs/MKs. Alternatively, 3T3-L1 cells were treated with 0.5 µM infigratinib (NVP-BGJ398), a selective FGFR1-3 inhibitor [42].

BODIPY™ staining of lipid droplets

On days 6 and 8, culture medium was removed, and 1 μg/mL BODIPY™ 493/503-containing DMEM without phenol red (Gibco™, #21041025) was added to the wells and cells were incubated for 20 min. After staining, the BODIPY™-containing medium was discarded and fresh DMEM without phenol red was added for image acquisition in an Incucyte® SX5 Instrument (Sartorius). Thirty-six pictures per well were captured using the AI Scan module and a 20X objective. Raw phase and raw green images were exported and analyzed using CellProfiler v4.2.6 software. The configuration used was based on the method described in [43] with some modifications: (i) the diameter of objects was set at 1–80 px, (ii) a threshold correction factor of 1.2 was applied, and (iii) the method to identify individual objects within clumped objects was changed to “Propagate”. For images taken on day 8, a threshold correction factor of 1.6 was used instead.

BM transplantation

BM cells from at least three mice were pooled and counted using a Beckman Coulter Z2 hemocytometer. The cell concentration was adjusted to 2 × 10⁶ cells/200 μL in sterile PBS. The cell solution was kept on ice until injection.

Female tgC3G^Pf4, C3G^Pf4-KO mice and their corresponding wild-type littermates were irradiated with two doses of 5.5 Gy separated by 4 h. After 2–3-h recovery period, 2 × 10⁶ pooled whole BM cells from Ly5.1/CD45.1 donors were injected via the lateral tail vein. The number of neutrophils, T lymphocytes, and B lymphocytes in PB was assessed monthly until the 9th month, when mice were sacrificed to evaluate potential changes in BM and spleen.

RT-qPCR analysis

cDNA was synthesized from MK and 3T3-L1 cell RNA using the NZY First-Strand cDNA Synthesis kit (NZYTech, #MB12501). MKs were freshly isolated from steady-state BM by FACS as CD41/CD42 double positive cells and immediately added to the lysis buffer. Gene expression was assessed by qPCR using SYBR Green master mix (NZYTech, #MB22202) and primers listed in Additional file 1, Suppl Table 1. Primers were designed using the Primer-BLAST Tool. In the analysis of the expression of genes in Sipa1-KO MKs, TaqMan probes were used (Suppl Table 2). Data were analyzed using the 2^(−ΔΔCt) method [44]. β-actin (Actb) or hypoxanthine guanine phosphoribosyl transferase (Hprt) genes were used as housekeeping genes, as indicated.

Statistical analysis

Unless otherwise specified, all the results were reported as mean ± standard error of the mean (SEM) from at least three independent experiments, each including multiple replicates. Statistical analyses were performed using GraphPad Prism v8.2.1. Data normality was assessed with Shapiro–Wilk tests. For comparisons between two experimental groups, a t-test was used. To evaluate differences among three or more groups, one-way or two-way ANOVA was applied, depending on whether one or two factors were analyzed. ANOVA tests were followed by Fisher’s Least Significant Difference (LSD) post-hoc test. The choice of statistical test was based on the normality and homoskedasticity of the data. For Kaplan–Meier survival analysis, a log-rank test was performed. p values less than or equal to 0.05 were considered statistically significant.

Results

C3G boosts MK maturation following myeloablation, leading to hematopoiesis recovery

Previous data from our group have confirmed that C3G promotes megakaryocytic differentiation both in vitro and ex vivo in response to TPO stimulation [8]. Additionally, injection of TPO or 5-FU in MK-specific C3G-overexpressing mouse models increases platelet count, whereas C3G ablation results in a decrease in platelet levels [6, 8].

To further investigate the role of C3G in MK physiology under biological stress conditions, we induced myelosuppression in tgC3G^Pf4 and C3G^Pf4-KO mice, along with their respective wild-type siblings through i.p. administration of a sublethal dose of 5-FU. Recovery was monitored at several time points (Fig. 1A). As previously reported using flow cytometry [6], hemocytometric analysis revealed a significant increase in platelet rebound in tgC3G^Pf4 mice (Fig. 1B), while deletion of C3G delayed platelet count recovery (Fig. 1C). Similarly, white blood cell (WBC) counts recovered faster in tgC3G^Pf4 mice, surpassing resting levels (Fig. 1D), accompanied by elevated red blood cell (RBC) and hemoglobin levels (Fig. 1E, F). In contrast, WBC counts remained lower in C3G^Pf4-KO mice, compared to C3G^Pf4-wt mice, during recovery (Fig. 1G), with no effect on RBC counts or hemoglobin levels (Fig. 1H, I). This suggests that C3G overexpression in MKs may help mitigate the decline in blood cell populations, other than platelets, following chemotherapy.

Fig. 1

MK C3G promotes hematopoietic recovery following 5-FU treatment by enhancing megakaryocyte (MK) maturation (A) Scheme of 5-FU-induced myeloablation. Blood drops indicate the days of blood extraction. Femurs indicate the days of femur and BM harvest. Platelet count in (B) tgC3G^Pf4 and (C) C3G^Pf4-KO mice and their controls. Evaluation of (D, G) WBC, (E, H) RBC, and (F, I) hemoglobin in (D–F) tgC3G^Pf4 and (G–I) C3G^Pf4-KO and their controls. Data (mean ± SEM) correspond to two independent experiments. wtC3G^Pf4, n = 4; tgC3G^Pf4, n = 5; C3G^Pf4-wt, n = 5; C3G^Pf4-KO, n = 5. (J, L) Graphics showing the mean ± SEM of MK count in the first 5 mm of femoral BM from the growth plate, at the indicated days after 5-FU, in (J) wtC3G^Pf4 and tgC3G^Pf4 and in (L) C3G^Pf4-wt and C3G^Pf4-KO. (K, M) Representative photomicrographs of BM sections stained with H&E. Arrowheads indicate MKs. The insets show a 2.5X magnification of the regions highlighted in the images. Scale bar: 0.25 mm. (N, O) Normalized counts (mean ± SEM) of mature MKs (CD42⁺CD61⁺ cells) by flow cytometry in BM extracts from (N) tgC3G^Pf4, and (O) C3G^Pf4-KO mice and their controls at the indicated days after 5-FU. (P, Q) Violin plots show the median size of MKs in the inner part of the diaphysis. (R, S) MK (mean ± SEM) quantification along (R) wtC3G^Pf4 and tgC3G^Pf4 femora 7 days after 5-FU injection, and (S) C3G^Pf4-wt and C3G^Pf4-KO femora 10 days after 5-FU injection. Two to three independent experiments were performed per day of analysis. wtC3G^Pf4, n = 4–6; tgC3G^Pf4, n = 4–8; C3G^Pf4-wt, n = 4–6; C3G^Pf4-KO, n = 4–6. p-values in (B–I, J, L, R, S) were calculated using two-way ANOVA with uncorrected Fisher’s LSD post-hoc test. Unpaired t-test was used in (N–Q). *p ≤ 0.05. **p ≤ 0.01. ***p ≤ 0.001

Comparative analysis of hematoxylin–eosin-stained femur sections revealed that the number of MKs in the BM of tgC3G^Pf4 mice 7 days after 5-FU administration was significantly lower than in wtC3G^Pf4 mice, though the numbers equalized at later recovery stages (Fig. 1J, K). Conversely, C3G^Pf4-KO mice showed a significant increase in MK numbers in the BM compared to wild-type controls (Fig. 1L, M), with differences becoming significant 10 days post-treatment. To determine whether MK maturity correlated with platelet production, we examined the surface levels of CD42 and CD61 in BM extracts using flow cytometry. As expected, MKs from tgC3G^Pf4 mice were more mature than those from C3G^Pf4-KO mice, compared with their respective control littermates (Fig. 1N, O). This difference in maturity was also evident in the bone marrow sections, where MKs in tgC3G^Pf4 femora were notably larger than those in wtC3G^Pf4 marrow (Fig. 1P), with no differences between C3G^Pf4-KO and C3G^Pf4-wt mice (Fig. 1Q). This suggests that C3G may regulate MK differentiation, maturation and platelet production upon 5-FU treatment, with this process being delayed in the absence of C3G. Consistent with other reports [26, 45], the distribution of MKs along the femoral marrow was not random, with a progressive enrichment from the proximal to the distal region, and the most notable differences between genotypes occurring in the latter region (Fig. 1R, S).

Bone marrow adipocytes (BMA) expand parallel to C3G expression in MKs after 5-FU

We previously described a significant increase in BMA density following melanoma cell implantation in tgC3G^Pf4 mice [8], suggesting that C3G expression in MKs might influence the response of BM stromal cells to an insult. Based on this, we first explored whether MK C3G affects the adipocyte progenitor cell (APC) population during recovery from 5-FU treatment. Seven days after treatment initiation, APCs were significantly less abundant in the BM of tgC3G^Pf4 mice (Fig. 2A), whereas C3G^Pf4-KO mice exhibited the opposite trend (Fig. 2B). These changes in the progenitor population resulted in a higher number of BMAs in the femora of tgC3G^Pf4 mice (1.6-fold increase) by day 10 compared to their wtC3G^Pf4 littermates, likely due to a role of MK C3G in the differentiation of APCs into BMAs (Fig. 2C, D and Suppl Fig. 1A). Supporting this, the number of BMAs in the BM of C3G^Pf4-KO mice was lower than that in C3G^Pf4-wt mice (Fig. 2E, F and Suppl Fig. 1B). This suggests a defective differentiation capability of APCs upon C3G deletion in MKs, resulting in the observed accumulation of adipocyte progenitors (Fig. 2B). Furthermore, BMAs from tgC3G^Pf4 mice were on average 57.64 ± 14.95 μm² larger than those from wtC3G^Pf4 mice (Fig. 2C, G). In contrast, no differences in adipocyte size were detected in the C3G^Pf4-KO genotype (Fig. 2E, H), indicating that the observed variations in total marrow adipose tissue (MAT) were due to MK C3G-induced MAT hyperplasia rather than hypertrophy. No differences were observed in either the number of adipocytes or their size before and after recovery (21 days) from 5-FU treatment (Fig. 2C, E, and Suppl Fig. 1C–F).

Fig. 2

MK C3G promotes BMA expansion following 5-FU-induced BM depletion (A, B) Mean ± SEM of APC (CD45⁻CD31⁻Sca-1⁺CD24⁻) percentage within the MSC compartment in BM from (A) tgC3G^Pf4 and (B) C3G^Pf4-KO mice and their controls. (C, E) H&E-stained BM sections of femurs at the indicated days post-5-FU treatment. The insets show a three-fold magnification from the 10-day images. Scale bar: 0.25 mm. (D, F) BMA number/total marrow area normalized to controls. (G, H) Violin plots depict the median area of the counted adipocytes. Three to five independent experiments were performed per day of analysis. Statistics were calculated using two-way ANOVA with uncorrected Fisher’s LSD post-hoc test (A, B) and unpaired t-test (D, F, G, H). wtC3G^Pf4, n = 4–7; tgC3G^Pf4, n = 3–8; C3G^Pf4-wt, n = 6–11; C3G^Pf4-KO, n = 5–11. (I) Adipokine content in BM supernatants from either tgC3G^Pf4 or C3G^Pf4-KO mice compared to their controls 7 days after 5-FU administration. Quantification was performed in short or long exposures of the Proteome Profiler membranes as indicated in Suppl Fig. 2. (J) In vitro setting for the analysis of MK influence on adipocyte differentiation. (K, S) Representative images of 3T3-L1 cells co-cultured without (3T3-L1) or with HSPCs/MKs of the indicated genotypes (K) in the absence of insulin and (S) under the effect of inhibitors. Scale bar: 0.2 mm. Violin plots showing (L, T) BODIPY™ mean fluorescence intensity, (M, U) number and (N) mean size of each lipid droplet. In each assay, 24 to 36 fields were analyzed per condition. Three independent experiments were performed in (K–N). Statistical analysis was performed using one-way ANOVA with uncorrected Fisher’s LSD post-hoc test. Unless directly stated on the graph, significance was indicated as follows: * vs Ctrl sample; # vs the corresponding wild-type. One symbol p ≤ 0.05, two symbols p ≤ 0.01, three symbols p ≤ 0.001. a.u.: arbitrary units; px: pixels. ns: non-significant. (O–R) Mean ± SEM of the log₂(Fold change) of mRNA expression of (O) Fgf1, (P) Igf1, (Q) Pref1, and (R) Il6 in CD41⁺CD42⁺ MKs from the BM of tgC3G^Pf4, C3G^Pf4-KO mice and their wild-types. *p ≤ 0.05

These changes in adipogenesis observed during mid-recovery from 5-FU suggest that factors within the BM microenvironment of both tgC3G^Pf4 and C3G^Pf4-KO mice differ from those of their corresponding wild-type counterparts. To address this, we evaluated adipokine levels of pooled BM supernatant samples using a specific Proteome Profiler array. The results shown in Fig. 2I and Suppl Fig. 2 indicate a differential release of factors between tgC3G^Pf4 and C3G^Pf4-KO BM, compared to their respective controls. It is worth noting the lower levels of IL-6 in tgC3G^Pf4 BM, which plays a negative role in adipogenesis by downregulating PPARγ and C/EBPα transcription factors [46]. In contrast, IL-6 levels were higher in C3G^Pf4-KO BM compared to C3G^Pf4-wt. The same pattern was found for RAGE and RANTES (CCL5), both playing a role in inflammation and insulin resistance associated with obesity [47, 48]. Although only IL-6 has been described to be secreted by MKs [49], these results support the observed increase in BMA in tgC3G^Pf4 BM, and suggest that C3G may promote the differential release of MK factors that favor adipogenesis.

C3G from MKs promotes the differentiation of 3T3-L1 pre-adipocytes

The aforementioned results suggest that MKs may influence BMA differentiation. Previous studies have shown that MKs promote the delipidation of adipocytes, which, in turn, enhance MK differentiation and maturation [29]. However, to our knowledge this is the first report suggesting a potential role for MKs in BMA differentiation.

To determine whether the changes observed in the BMA population in vivo were driven by signals from MKs that directly affect adipocytes, and whose release is modulated by C3G, we established an in vitro assay. This consisted of the indirect co-culture of committed pre-adipocyte 3T3-L1 cells with MKs, using a Transwell system. Freshly MACS-isolated Lin⁻ progenitor cells, cultured in the presence of TPO to induce megakaryocytic differentiation, were seeded in the upper chamber of the Transwell containing in the lower chamber a 4-day culture of pre-differentiated 3T3-L1 cells (Fig. 2J). This setup was maintained for an additional 4 days, and adipocyte differentiation was assessed on days 6 and 8. Notably, no differences were found between 3T3-L1 cells grown alone (control condition) or in the presence of TPO in the upper chamber of the Transwell system, ruling out any potential effect of TPO (data not shown), consistent with findings by Valet et al. [29]. However, contrary to Valet´s findings, our results suggest that the indirect co-culture of 3T3-L1 cells with HSPCs/MKs enhanced lipid synthesis in the 3T3-L1 cells, particularly during the first 2 days (Fig. 2K, L and Suppl Fig. 3A, B), regardless of the absence (Fig. 2K, L) or presence (Suppl Fig. 3A, B) of insulin in the culture medium. Notably, Lin⁻ progenitor cells in the upper chamber differentiated similarly across all four genotypes analyzed throughout the assay (data not shown). In cultures without insulin (MM medium), adipocyte differentiation increased in all conditions, as evidenced by a higher number of lipid droplets (Fig. 2M) and an increase in their size (Fig. 2N). Furthermore, 3T3-L1 cells co-cultured with HSPCs/MKs from tgC3G^Pf4 mice exhibited higher lipid droplet content compared to those co-cultured with wtC3G^Pf4 cells (Fig. 2M). This effect was also observed under DM-II conditions (Suppl Fig. 3C), closely reflecting the in vivo results (Fig. 2C, D, G).

Conversely, in 3T3-L1 cultures without insulin, a trend toward reduced lipid droplet size was observed when C3G^Pf4-KO HSPCs/MKs were in the upper chamber of the Transwell system (Fig. 2N). Interestingly, when insulin was present, C3G deficiency in HSPCs/MKs promoted lipid droplet formation in 3T3-L1 cells (Suppl Fig. 3C), though these droplets were significantly smaller than those in 3T3-L1 cells maintained with factors from C3G^Pf4-wt HSPCs/MKs (Suppl Fig. 3D).

Differentiation of 3T3-L1 cells under the influence of HSPCs/MKs was further enhanced after 4 days of co-culture (day 8 of differentiation), even in the absence of insulin, regardless of genotype (Suppl Fig. 3E–H). This mirrored the effects seen after 2 days of co-culture (day 6 of differentiation) in the presence of insulin (Suppl Fig. 3A–D).

Interestingly, all cells cultured under cues from HSPCs/MKs exhibited fewer lipid droplets compared to control cells (Suppl Fig. 3G), likely due to fusion into larger droplets, as suggested by the increase in droplet size across all conditions (Suppl Fig. 3H). Thus, the variations in BMA area observed in femur sections (Fig. 2G, H) were recapitulated in this in vitro system. Unfortunately, quantitative analysis of 4-day co-cultures with insulin was not feasible due to cell overgrowth and detachment. However, qualitative observations suggested increased proliferation of 3T3-L1 cells influenced by factors secreted by tgC3G^Pf4 HSPC/MKs. Notably, although factors from C3G^Pf4-KO TPO-stimulated progenitors promoted greater lipid accumulation in 3T3-L1 cells in the absence of insulin after 4 days of co-culture (day 8) (Suppl Fig. 3F), these co-cultures displayed slower growth (and less cell detachment) in DM-II medium, compared to cells influenced by factors from C3G^Pf4-wt HSPC/MKs (Suppl Fig. 3I).

C3G modulates the expression of MK genes involved in BMA differentiation, eventually enhancing the expression of adipocyte differentiation genes in 3T3-L1 cells

Since our results suggested that factors secreted by MKs could modulate adipocyte differentiation, we next explored the expression of candidate genes, such as Fgf1, Igf1, Pref1 and Il6 in freshly isolated CD41⁺/CD42⁺ MKs from the different genotypes. The expression of all these genes, with the exception of Il6, was reduced in MKs lacking C3G (Fig. 2O–R), potentially explaining the smaller lipid droplets observed when C3G^Pf4-KO HSPCs/MKs were present in the environment (Fig. 2N and Suppl Fig. 3D, H). Although not statistically significant, there was a general trend toward increased expression of Fgf1, Igf1 and Pref1 in tgC3G^Pf4 MKs compared to wtC3G^Pf4 MKs (Fig. 2O–Q). In addition, a significant positive correlation was found between the expression of C3G (Rapgef1) and Fgf1 (Suppl Fig. 4A, B). This suggests that C3G may directly regulate Fgf1 expression in MKs, which may contribute to the expansion of the adipocyte population in the BM. Similar correlations were observed for Igf1 (Suppl Fig. 4D) and Pref1 (Suppl Fig. 4F), in the C3G^Pf4-KO model, but not in the transgenic model (Suppl Fig. 4C, E). In contrast, a negative correlation between Rapgef1 and Il6 expression was observed (Suppl Fig. 4G, H), along with a significant reduction in Il6 mRNA levels in tgC3G^Pf4 MKs (Fig. 2R). These results align with the decreased levels of IL-6 cytokine detected in 5-FU-treated tgC3G^Pf4 BM (Fig. 2I).

To further evaluate the potential contribution of FGF1 to the role of MK C3G in adipogenesis, we treated HSPCs/MKs from wtC3G^Pf4 and tgC3G^Pf4 mice with suramin, able to block the action of factors released by TPO-differentiated Lin⁻ cells. As shown in Fig. 2T, U and Suppl Fig. 3J, suramin completely inhibited the increase in lipid synthesis induced by both wtC3G^Pf4 and tgC3G^Pf4 MKs, suggesting the involvement of factors released by MKs. Next, we specifically examined the impact of FGF1 by inhibiting FGF receptor family with infigratinib on 3T3-L1 cells and found that it abrogated the enhanced lipogenesis induced by transgenic C3G expression (Fig. 2T, U and Suppl Fig. 3J). These results strongly support the notion that C3G regulates MK secretion of factors directly involved in adipogenesis, with FGF1 likely playing a key role.

Several transcription factors (e.g. C/EBPα, β γ and PPARγ and other regulatory signals (e.g. Pref-1 and Adiponectin) are known to orchestrate the terminal differentiation of pre-adipocytes into mature adipocytes [50] (Suppl Fig. 5A). We examined whether the expression of any of these regulators was affected in our 2-day co-cultures (day 6 of differentiation). As shown in Suppl Fig. 5B–G, insulin induced the expression of all tested genes in control 3T3-L1 cells. Interestingly, MKs from all genotypes also induced the expression of these genes at levels comparable to insulin. Although not statistically significant, HSPCs/MKs from tgC3G^Pf4 mice promoted the expression of the terminal differentiation genes Cebpa (Suppl Fig. 5E), Pparg (Suppl Fig. 5F), and Adipoq (Suppl Fig. 5G), at least in the insulin condition, indicating enhanced adipocyte differentiation. This expression pattern aligns with the increased number (Fig. 2M and Suppl Fig. 3C) and signal intensity (Fig. 2L) of lipid droplets, their larger size (Suppl Fig. 3D, H), and the larger size of adipocytes (Fig. 2D) influenced by cues from tgC3G^Pf4 MKs.

Interestingly, Pref1 gene expression was notably elevated in 3T3-L1 cells co-cultured with HSPCs/MKs, even in the absence of insulin (MM medium), with levels comparable to those seen in the insulin-containing (DM-II) condition (Suppl Fig. 5D). Among the early genes assessed, Pref1 is the only one encoding a protein that negatively regulates adipocyte differentiation when secreted [51]. Its persistent expression at this late stage (days 6 of differentiation) is, therefore, intriguing, although it could explain the reduced BODIPY™ intensity observed in the tgC3G^Pf4 + insulin condition (Suppl Fig. 3B) and the lower number of lipid droplets in MK-co-cultured 3T3-L1 cells compared to controls on day 8 (Suppl Fig. 3G). However, it is important to note that only the secreted form of Pref-1 appears to inhibit adipocyte differentiation [52].

Taking together, these data indicate that MK C3G promotes BMA expansion and maturation, which could contribute to HSPC mobilization following myeloablation.

C3G overexpression in MKs promotes myeloid lineage expansion following 5-FU treatment, while C3G deletion favors CLP accumulation

Our data indicate that C3G levels in MKs influence the amounts of MKs and BMAs in the BM. Given that both cell types function as BM niche cells, we hypothesize that changes in MK C3G expression may affect the HSPC pool during recovery from 5-FU treatment. Although no significant differences in BM cellularity were observed between C3G-modified mice and their corresponding wild-type controls (Fig. 3A, B), we found significant changes in the relative proportions of specific subpopulations during the mid-recovery phase (days 7 and 10) (Fig. 3 and Suppl Figs. 6, 7). Specifically, LKS cells (but not LKS⁻) were reduced on day 7 in tgC3G^Pf4 mice (Fig. 3C, D), although none of its subpopulations (i.e., LT-HSCs, ST-HSCs, MPPs, and CLPs, (see Abbreviations Section)) were proportionally altered (Fig. 3E–H). However, there was a significant reduction in CMP and MEP populations (Fig. 3I, K), accompanied by a slight increase in the GMP population (Fig. 3J). These results, along with the reduced number of MKs (Fig. 1J), the exacerbated platelet rebound (Fig. 1B), and the increased RBCs (Fig. 1E, F), suggest that C3G-overexpressing MKs induce the mobilization of HSPCs and promote the myeloid branch of hematopoiesis, particularly the MK-erythrocyte lineage.

Fig. 3

MK C3G regulates the mobilization of progenitors following 5-FU treatment (A, B) BM cellularity is depicted as the mean ± SEM of cells from one femur and one tibia of (A) wtC3G^Pf4 and tgC3G^Pf4 or (B) C3G^Pf4-wt and C3G^Pf4-KO mice throughout the recovery after 5-FU injection. Box plots depict the median and range of (C, L) LKS, (D, M) LKS⁻, (E, N) LT-HSC, (F, O) ST-HSC, (G, P) MPP, (H, Q) CLP, (I, R) CMP, (J, S) GMP, (K, T) MEP and (Y, Z) Lin⁻ subpopulations of HSPCs in the BM of tgC3G^Pf4 or C3G^Pf4-KO mice (as indicated) compared to their corresponding wild-type littermates 7 days after 5-FU administration. The values correspond to the proportion within each corresponding parent gate and are normalized to the corresponding wild-types. (U, V) Percentage of Lin⁻ cells in BM (mean ± SEM) from (U) wtC3G^Pf4 and tgC3G^Pf4 or (V) C3G^Pf4-wt and C3G^Pf4-KO mice during the 21-day assessment period. (W, X) Box plots depict the median value and range of the BrdU⁺ Lin⁻ cells from (W) tgC3G^Pf4 mice and (X) C3G^Pf4-KO mice normalized against their controls, 7 days after the injection of 5-FU. Two independent experiments were performed. Statistics were performed using two-way ANOVA with uncorrected Fisher’s LSD post-hoc test in (A, B, U, V), and using t-test in (C–T, W–Z). *p ≤ 0.05. **p ≤ 0.01. ***p ≤ 0.001. Three independent experiments were performed

In contrast, C3G deletion in MKs did not lead to any changes in the LKS or LKS⁻ populations (Fig. 3L, M) nor in MK-biased progenitors (Fig. 3R, T). However, there was a decrease in the LT-HSC compartment (Fig. 3N), while the ST-HSC (Fig. 3O), MPP (Fig. 3P), and CLP (Fig. 3Q) populations increased proportionally 7 days after 5-FU injection, with no changes in GMPs (Fig. 3S). This, along with the reduced number of WBCs (Fig. 1G), suggests that C3G deletion in MKs promotes HSPC bias toward the lymphoid lineage but impairs their terminal differentiation. No alterations in hematopoiesis due to changes in MK C3G expression were observed in untreated mice (Suppl Fig. 6).

Moreover, while the Lin⁻ progenitor cell compartment remained significantly increased on day 10 in tgC3G^Pf4 mice compared to wtC3G^Pf4 (Fig. 3U and Suppl Fig. 7A), C3G^Pf4-KO mice showed a slight decline in this population by day 10 relative to their controls (Fig. 3V and Suppl Fig. 7K). In agreement, the proliferative capacity of C3G^Pf4-KO Lin⁻ cells was reduced on day 7 compared to those from C3G^Pf4-wt mice (Fig. 3X), which may explain the population decrease observed 3 days later. Conversely, the proliferative capacity of tgC3G^Pf4 Lin⁻ cells was increased, albeit not significantly (Fig. 3W), consistent with the higher number of immature cells noted on day 10 (Suppl Fig. 7A). These tendencies were already noticed on day 7 (Fig. 3Y, Z). In both scenarios, the Lin⁻ population returned to baseline levels by day 21 (Suppl Fig. 8A–K), along with the other populations (Suppl Fig. 8).

Collectively, these data suggest that MK C3G promotes the proliferation and mobilization of HSPCs, favoring the myeloid branch of hematopoiesis after 5-FU administration, which leads to an increased presence of platelets and erythrocytes in the blood. In contrast, C3G ablation promotes the lymphoid branch.

Altered C3G levels in MKs lead to changes in the levels of mature neutrophils and T-lymphocytes

Next, we explored whether the changes in hematopoietic progenitor populations influenced by MK C3G expression, following 5-FU treatment, extended to mature differentiated cells. Notably, both C3G overexpression (Fig. 4A) and deletion (Fig. 4B) led to an increased percentage of neutrophils under steady-state conditions, which persisted on day 3 in tgC3G^Pf4 mice. An increase in neutrophil counts in PB has previously been reported in tgC3G^Pf4 mice [2]. During recovery from neutrophil depletion, C3G^Pf4-KO mice exhibited an increase in neutrophils, while tgC3G^Pf4 mice showed a decrease, in line with a role for C3G in promoting megakaryopoiesis. No significant changes were observed in the percentage of T lymphocytes in the PB of tgC3G^Pf4 mice (Fig. 4C). However, between days 7 and 14, the proportion of CD4⁺ and CD8⁺ cells shifted in favor of the latter (Fig. 4D, E). Consistently, a bias toward the CD4⁺ phenotype was observed within the CD3⁺ lymphocyte population of C3G^Pf4-KO mice compared to their C3G^Pf4-wt counterparts (Fig. 4G, H) with no significant changes in CD3⁺ cell expansion (Fig. 4F). No changes in B cell number were observed in tgC3G^Pf4 mice (Fig. 4I); however, a slight increase was noted in C3G^Pf4-KO mice (Fig. 4J).

Fig. 4

MK C3G regulates the levels of mature neutrophils and T-lymphocytes after 5-FU-induced myeloablation XY plots display the mean ± SEM of the percentage of (A, B) neutrophils, (C–H) T lymphocytes and (I, J) B lymphocytes in (A, C–E, I) wtC3G^Pf4 and tgC3G^Pf4 or (B, F–H, J) C3G^Pf4-wt and C3G^Pf4-KO PB during the 21-day recovery period after 5-FU administration. Percentages shown in (D, E, G, H) are those within the CD3⁺ T lymphocyte gate. Five independent experiments were performed. wtC3G^Pf4, n = 11; tgC3G^Pf4, n = 13; C3G^Pf4-wt, n = 14; C3G^Pf4-KO, n = 14. (K, L) Interleaved scatter plots display the mean ± SEM of spleen weight of either (K) tgC3G^Pf4 or (L) C3G^Pf4 KO mice, as well as their corresponding wild-type siblings at the indicated time points following 5-FU administration. Three to five independent experiments were performed per day of analysis. Statistical analysis was performed using two-way ANOVA with uncorrected Fisher’s LSD post-hoc test. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001

These findings suggest that MK C3G alters blood cell population percentages, likely by promoting a shift towards the MK-platelet and erythrocyte lineages following BM depletion, consistent with the results shown in Figs. 1 and 3.

C3G deletion prevents 5-FU associated splenomegaly

5-FU administration induces EMH, leading to splenomegaly [24]. To assess EMH in our models, spleens were harvested at three time points: before treatment (day 0), at mid-recovery (day 7), and after recovery (day 21). As expected, 5-FU administration resulted in splenomegaly on day 21 in both control and tgC3G^Pf4 mice (Fig. 4K), reflecting the spleen supportive role in hematopoietic recovery following myeloablation. However, C3G^Pf4-KO mice did not develop splenomegaly compared to untreated mice (Fig. 4L). This suggests that MKs are involved in 5-FU-induced EMH, and that C3G is critical for this MK function.

C3G exerts a protective role in recovery following repeated cycles of 5-FU, but only in female mice

The data in Fig. 1B, D, E, together with the increased precursor mobilization shown in Fig. 3C, I, K, suggest that C3G overexpression in MKs accelerates the recovery of blood cell populations following 5-FU-induced myeloablation, while C3G^Pf4-KO mice exhibited a delayed recovery (Figs. 1C–G and 3O–Q). To investigate this further, mice were subjected to consecutive weekly cycles of 5-FU and their survival was analyzed (Fig. 5A). Altering C3G levels in MKs significantly impacted survival, but only in female mice (Fig. 5B–G). Specifically, female tgC3G^Pf4 mice survived an average of 3.5 days longer than their wild-type littermates (Fig. 5C), whereas C3G deletion resulted in a 1-day reduction in survival (Fig. 5F). These findings are consistent with the enhanced proliferative capacity of progenitor cells in tgC3G^Pf4 mice and the reduced capacity in C3G^Pf4-KO mice (Fig. 3W, X), further supporting a positive role for MK C3G in hematopoietic recovery following myeloablation.

Fig. 5

Overexpression of C3G in MKs enhances the survival of female mice following weekly 5-FU administration (A) Scheme illustrating the 5-FU administration schedule for the 5-FU challenge. Kaplan-Meyer curves depict the overall survival of mice after repeated weekly administration of 150 mg/kg 5-FU in (B–D) wtC3G^Pf4 and tgC3G^Pf4 mice, and (E–G) C3G^Pf4-wt and C3G^Pf4-KO mice. Results are shown for (B, E) all the mice tested, (C, F) only females and (D, G) only males. The first injection was performed on day 0. Each additional dose is represented as a dotted line in the plot. The difference in median survival and p-value according to the Log-rank (Mantel-Cox) test are shown in the squared box. Five to six independent experiments were performed. wtC3G^Pf4, n = 11 (4 females, 7 males); tgC3G^Pf4, n = 14 (5 females, 9 males); C3G^Pf4-wt, n = 17 (10 females, 7 males); C3G^Pf4-KO, n = 19 (6 females, 13 males). (H–K) Bar plots show the mean ± SEM of the log₂(Fold change) of mRNA expression analyzed by RT-qPCR of (H) Tpo, (I) Pf4/Cxcl4, (J) Sdf1/Cxcl12, and (K) Tgfb1 in CD41⁺CD42⁺ MKs isolated from the BM of tgC3G^Pf4 and C3G^Pf4-KO mice, and their corresponding wild-types. Values are relative to β-actin levels and normalized to the corresponding wild-type control. *p ≤ 0.05

C3G modulates the expression of genes affecting HSCs in MKs

To further investigate the mechanism by which C3G levels in MKs influence HSPC fate, we examined whether C3G affects the expression of key MK genes known to modulate HSPCs, specifically Tpo, Pf4, Sdf1, and Tgfb1 (Fig. 5H–K and Suppl Fig. 4I–P), in addition to the previously analyzed Il6 (Fig. 2R and Suppl Fig. 4G, H). In C3G^Pf4-KO MKs Tpo and Tgfb1 expression was reduced (Fig. 5H, K), consistent with the diminished thrombopoiesis observed in C3G^Pf4-KO mice (Fig. 1C) and in previous studies [3, 53]. Pf4, which is associated with a reduced number of HSCs and increased quiescence [54], also exhibited a trend toward lower expression in C3G-deficient MKs (Fig. 5I). Similarly, SDF1, a chemokine involved in HSPC quiescence and MK migration [55], followed the same pattern (Fig. 5J). This indicates that C3G might play a role in promoting MK-induced HSC quiescence in the absence of external stimuli. This observation is further supported by the elevated Sdf1 mRNA expression in MKs lacking the Rap1 GAP Sipa-1 (Suppl Fig. 9A), which also regulates the BM niche [36]. Il6 expression was also increased in Sipa1-KO MKs (Suppl Fig. 9B), consistent with the involvement of both IL-6 and the Rap1 signaling pathway in myeloid differentiation [56, 57]. In contrast, C3G overexpression led to decreased Il6 mRNA levels in MKs (Fig. 2R), which supports the better recovery of these mice, since IL-6 is known to play an inhibitory role in the early stages of megakaryopoiesis [58]. A correlation between C3G expression and Tpo expression was observed in C3G^Pf4-KO mice (Suppl Fig. 4J), but not in tgC3G^Pf4 mice (Suppl Fig. 4I). Similarly, Rapgef1 expression correlated with Pf4 expression in C3G^Pf4-KO (Suppl Fig. 4L). No correlation was detected in the rest of the conditions (Suppl Fig. 4K, M, N, O, P).

C3G influences the niche function of MKs in a transplant setting

Next, we investigated whether the effects of MK C3G on the HSC population could be observed in different contexts. To this end, we assessed the impact of MK C3G on the recovery of mice subjected to TBI followed by whole bone marrow (WBM) transplantation (Fig. 6A). WBM cells from Ly5.1 mice repopulated PB populations similarly across all host genotypes. However, the percentages were slightly lower in tgC3G^Pf4 mice compared to wtC3G^Pf4 controls (Fig. 6B), and higher in C3G^Pf4-KO mice (Fig. 6C), particularly at 1 month of age. Indeed, with the exception of B cells, CD45.1⁺ cells generally contributed equal or less to BM repopulation in tgC3G^Pf4 mice compared to controls (Fig. 6D–G), consistent with the lower percentage of CD45.1⁺ cells in tgC3G^Pf4 BM (Fig. 7A). The opposite trend was observed in C3G^Pf4-KO mice compared to their controls for neutrophils, but not for T cells (Fig. 6H–K). Since total BM cellularity nine months after transplantation was similar in wtC3G^Pf4 and tgC3G^Pf4 mice (Fig. 7B), cells other than the transplanted ones are contributing to BM repopulation in tgC3G^Pf4 mice. One potential explanation is that the BM of tgC3G^Pf4 mice was more resistant to myeloablation than that of wtC3G^Pf4 mice, which could enhance repopulation, consistent with the results in Fig. 5C. This hypothesis is supported by the lower BM cellularity observed in C3G^Pf4-KO mice (Fig. 7D), alongside a similar contribution of CD45.1⁺ cells compared to C3G^Pf4-wt mice (Fig. 7C), indicating that host (CD45.2⁺) BM cells from C3G^Pf4-KO mice may be less resistant to TBI. In addition, the reduced proportions of CD41⁺, ST-HSC and MPP populations in tgC3G^Pf4 BM (Fig. 7E, H, I), with no changes in CD41⁺ and MPP populations in C3G^Pf4-KO mice (Fig. 7F, L), indicates that MK C3G may promote enhanced mobilization of myeloid/megakaryocytic precursors, consistent with findings from the 5-FU model. Moreover, C3G^Pf4-KO animals exhibited an accumulation of the most immature precursors in the BM (Fig. 7J, K) and increased neutrophil counts in PB (Fig. 6H). This suggests impaired MK differentiation, further supporting the hypothesis.

Fig. 6

MK C3G modulates neutrophil and lymphocyte reconstitution after TBI. (A) Transplantation setup. Two million BM CD45.1⁺ cells from a 7- to 10-week old Ly5.1 mouse were transplanted into lethally irradiated wtC3G^Pf4, tgC3G^Pf4, C3G^Pf4-wt and C3G^Pf4-KO, recipient mice (CD45.2). PB of the recipients was monitored monthly after transplantation and the BM analyzed at the endpoint of the experiment (9 months, see Fig. 7). XY plots display the mean ± SEM of the assessment of the proportion of donor (CD45.1⁺) cells evaluated during a 9-month period after transplantation. The percentages of (B, C) BM cells, (D, H) neutrophils, (E, I) CD4⁺ T lymphocytes, (F, J) CD8⁺ T lymphocytes, and (G, K) B lymphocytes in the PB were analyzed following the transplantation of 2 × 10⁶ WBM cells from Ly5.1 mice into lethally irradiated recipients: either (B–D–G) wtC3G^Pf4 and tgC3G^Pf4 or (C, H–K) C3G^Pf4-wt and C3G^Pf4-KO mice. Statistical analysis was performed using two-way ANOVA with uncorrected Fisher’s LSD post-hoc test. *p ≤ 0.05. **p ≤ 0.01. Three independent experiments were performed. wtC3G^Pf4, n = 7; tgC3G^Pf4, n = 9; C3G^Pf4-wt, n = 5; C3G^Pf4-KO, n = 6

Fig. 7

C3G influences the niche function of MKs in a transplant setting. Box plots depict the median and range of (A, C) the percentage of donor (CD45.1⁺) BM cells, (B, D) total marrow cellularity, (E, F) CD41-expressing BM cells, (G, J) LT-HSC, (H, K) ST-HSC, (I, L) MPP, (M, N) Lin⁻, (O, P) LKS and (Q, R) LKS⁻ subpopulations of HSPCs in the BM of (A, B, E, G, H, I, M, O, Q) tgC3G^Pf4 mice or (C, D, F, J, K, L, N, P, R) C3G^Pf4-KO mice compared to their corresponding wild-type littermates 9 months after transplantation with a 2 × 10⁶-WBM-cell graft. The proportion values in (G–R) were normalized to wtC3G^Pf4 or C3G^Pf4-KO, as appropriate. Statistical analyses were performed using t-test. *p ≤ 0.05. Three independent experiments were performed

No changes in LT-HSC population were observed in tgC3G^Pf4 mice (Fig. 7G). Furthermore, the proportions of Lin⁻ (Fig. 7M, N), LKS (Fig. 7O, P) and LKS⁻ (Fig. 7Q, R) populations were not affected by C3G overexpression or deletion in MKs. Finally, no alterations in spleen size were noted in any of the recipient mice nine months after transplantation (Suppl Fig. 9C, D).

These results indicate that MK C3G enhances the mobilization of precursor cells biased toward the myeloid/megakaryocytic branch in the bone marrow following transplantation, leading to decreased percentages of lymphocyte and neutrophil populations in PB. This phenotype mirrors that observed in the 5-FU model.

Discussion

In this study, we provide further evidence for the role of C3G in megakaryopoiesis, within the context of 5-FU–induced bone marrow myeloablation, and TBI followed by WBM transplantation, using genetically modified mouse models targeting C3G in MKs. In addition, our data suggest that MK C3G contributes to hematopoietic regeneration and that MKs are involved in BM adipogenesis, with C3G potentially playing a key role in this MK niche function.

Our group has previously demonstrated the significant role of C3G in platelet homeostasis [2, 3, 5], as well as in other platelet-mediated functions, including angiogenesis, tumor growth, metastasis, and the regulation of platelet mass via c-Cbl-mediated TPO endocytosis and degradation [4, 6]. Additionally, we have identified a role for C3G in both megakaryopoiesis and thrombopoiesis after challenging the hematopoietic system with TPO or chemotherapy [6, 8], with our data also suggesting that MK C3G may contribute to bone marrow adipogenesis after melanoma cell implantation [8].

In the 5-FU model, MK C3G facilitated MK maturation and platelet production during recovery, which correlated with a transient decrease in the MK population in the BM, particularly in the diaphysis. Interestingly, MK C3G also enhanced the recovery of WBCs and RBCs, which was associated with increased proliferation of Lin⁻ cells and mobilization of LKS cells. These findings suggest a potential role for C3G in the niche function of MKs, known to promote HSC proliferation and differentiation after injury [12]. The platelet rebound induced by 5-FU is the result of an indirect effect on BM niche cells, rather than a direct impact on HSCs [59]. Therefore, the involvement of MK C3G in this process, as observed here and previously [6], further supports a key role for C3G in MK function as niche cells within the BM. Furthermore, C3G expression in MKs positively correlated with Fgf1 expression, a factor known to promote HSC expansion during 5-FU-induced BM depletion [16].

Despite increased WBC counts in tgC3G^Pf4 mice, we observed a trend toward lower proportions of neutrophils and CD4⁺ T lymphocytes during recovery, with the opposite trend seen in C3G^Pf4-KO mice. This pattern aligns with the observed CMP mobilization and its bias favoring MEP over GMP differentiation. The C3G-driven bias toward megakaryopoiesis over other myeloid lineages is further supported by the decreased Il6 expression detected in tgC3G^Pf4 MKs. While IL-6 promotes myeloid lineage differentiation [60], it has a negative impact on the early stages of megakaryopoiesis [58]. This appears to be in contrast with the Sipa1-KO results (Suppl Fig. 9B), which show an increase in Il6 expression. However, it is important to consider that Sipa1-KO is a systemic knock-out [7], which could explain the observed differences.

On the other hand, C3G ablation in MKs led to an accumulation of progenitors primarily within the lymphoid branch. C3G deletion favored CLP production, but appeared to block their progression to mature lymphocytes, suggesting a role for MK C3G in lymphocyte maturation. Consistent with this, previous studies have reported a role for C3G-Rap1 signaling in T-cell homeostasis and function [61, 62]. Our data also indicate an essential role for MK C3G in EMH following 5-FU treatment, as evidenced by the absence of splenomegaly in C3G^Pf4-KO animals.

In parallel, MK C3G altered the adipokine content of the BM (Fig. 2I), promoting an enrichment in mature BMAs, accompanied by a reduction in APCs. This MK C3G-mediated increase in BMAs following a stress stimulus aligns with previous findings [8] and suggests a potential role for MK C3G in regulating adipogenesis. Our in vitro results further corroborate this, demonstrating that MKs promote the adipocytic differentiation of 3T3-L1 cells, with C3G likely enhancing this process by modulating the expression of MK-derived factors involved in adipogenesis. Specifically, C3G positively modulated the expression of Fgf1, Igf1, Pref1, and negatively that of Il6 in MKs, all of which are relevant regulators of adipogenesis [31,32,33,34]. Interestingly, the reduced ability of tgC3G^Pf4 HSPCs/MKs to promote adipogenesis following selective FGFR inhibition (Fig. 2S–U, Suppl Fig. 3J), postulates FGF1 as a key factor in MK C3G-mediated adipocyte expansion in the BM after myeloablation. Although the secreted form of Pref-1 can act as a negative regulator of adipogenesis [52], the effect of membrane-bound Pref-1 remains unclear. Furthermore, a hierarchical relationship exists among these factors, such that the presence of IGF1 supersedes Pref-1 [63]. This hierarchy, along with the complexity of Pref-1 effect, may explain how C3G overexpression promotes adipogenesis despite elevated Pref1 mRNA levels. On the other hand, there was neither a synergistic nor additive effect between MK cues and insulin on the expression of adipocyte late differentiation genes. However, factors from C3G-overexpressing MKs appeared to enhance the insulin response, consistent with C3G role in regulating platelet secretion [5].

Other studies have established a relationship between MKs and BMAs. Specifically, adipocytes enhance megakaryopoiesis by promoting MK polyploidization and proplatelet formation through fatty acid transfer from adipocytes to MKs via CD36 translocase. MKs may then create a positive feedback loop by inducing delipidation of adipocytes, likely mediated by MK releasate [29, 30]. However, to our knowledge, this is the first report identifying a role for MKs in BM adipogenesis, highlighting the significant contribution of C3G.

The effect of MK C3G on HSC proliferation and differentiation after injury may also be indirectly linked to its influence on BMAs, which are known to promote hematopoiesis in response to stress stimuli [26, 27]. Yet, in the absence of stimuli, C3G appears to promote MK-induced HSC quiescence by modulating Tpo, Pf4, and Sdf1 expression. Notably, the expression of Tpo in MKs correlates with that of Rapgef1, further supporting the involvement of C3G in megakaryopoiesis and thrombopoiesis. As previously noted, our group has demonstrated that the regulation of platelet secretion is likely the most critical function of C3G in platelet biology. C3G regulates actin remodeling and trans-SNARE complex formation through mechanisms involving PKCδ, RalA-Rap1, and Rac1 pathways [5]. This, combined with the observation that MKs influence BMAs without requiring direct contact, strongly supports the hypothesis that C3G acts as a key regulator of MK secretion.

The positive effect of C3G on BM recovery after depletion is further evidenced by the increased survival observed in tgC3G^Pf4 female mice and the decreased survival seen in C3G^Pf4-KO female mice. Some studies suggest that certain signaling pathways involving C3G could influence metabolic conditions differently in males and females due to broader hormonal and genetic interactions. For instance, polymorphisms in the Rapgef1 gene have been associated with diabetes [64, 65], a condition globally more prevalent in women than in men [66]. Additionally, Rapgef1 polymorphisms have been linked to persistent lactation and mastitis in dairy cows [67], and the Rapgef1 promoter has been found demethylated in gynecological cancers [68]. However, direct research confirming gender-specific effects of C3G remains limited.

In the transplantation model, although tgC3G^Pf4 mice showed a slightly reduced contribution of CD45.1⁺ cells to hematopoiesis nine months post-transplantation, our results indicate increased mobilization of hematopoietic precursors due to MK C3G overexpression, particularly of CD41⁺ cells. As with the 5-FU model, this finding suggests a C3G-mediated differentiation of MK-primed progenitors. Consistently, C3G ablation in MKs led to reduced progenitor mobilization. The reduced number of CD45.1⁺ cells in the BM of tgC3G^Pf4 mice may be attributed to several factors: (i) poorer engraftment influenced by the MK C3G niche, and/or (ii) greater resistance of the tgC3G^Pf4 niche to irradiation, mediated by MK C3G, resulting in a lower proportion of CD45.1⁺ cells. In the first scenario, one possible explanation is an increased BMA population due to MK C3G, which is known to negatively impact BM engraftment post-transplant [25]. However, as previously discussed, BMA abundance may also have a positive effect on HSC proliferation following irradiation or chemotherapy [27, 28], making both scenarios plausible.

Conclusions

In conclusion, our findings highlight a novel role for C3G as a key factor in the niche function of MKs during BM recovery following chemotherapy or transplantation. Moreover, we reveal that MKs are significant regulators of BM adipogenesis during this process, with C3G playing a crucial role by promoting the release of FGF1 and potentially other factors from MKs that alter the adipokine content of the BM. Hence, the enhanced hematopoietic regeneration observed in the tgC3G^Pf4 model is likely due to increased secretion by mature MKs and, potentially, the indirect effects of elevated BM adipocytes. These insights not only deepen our understanding of BM recovery mechanisms but also underscore the potential for targeting C3G in therapeutic strategies aimed at enhancing hematopoietic reconstitution and improving outcomes in chemotherapy-treated and transplanted patients.