Single-cell transcriptome analysis profiles the enlarged subsets of myeloid-biased HSPCs with preleukemic characters in disuse osteoporosis mice

Abstract

Background

Osteoporosis (OP) could lead to the alteration of bone marrow microenvironment and non-homeostasis of hematopoiesis, which could increase the incidence of hematologic malignancies. However, whether myeloid-biased hematopoiesis occurred and contributed to the leukemogenesis under the condition of OP remains unclear.

Results

This study successfully induced a mouse model for OP by hindlimb unloading, which shows increased myeloid cells and decreased B cells in the peripheral blood (PB). Furthermore, our study demonstrates that the myeloid-biased subset of HSPCs (hematopoietic stem and progenitor cells) with reduced differentiation and apoptosis, including multipotent progenitor 3 (MPP3) and granulocyte-monocyte progenitors (GMPs), were expanded in the OP mice. The expansion of myeloid-biased HSPCs contributes to the accumulation of HSPCs in the bone marrow and increased myeloid cells in the PB of OP mice. In the expanded pool of HSPCs, OP mice specifically enriched subsets were identified and profiled by single cell RNA-seq, including subHSCs from primitive HSCs, MPP3-1 from MPP3, GMP5 from GMPs, MkP2 from megakaryocyte progenitors and EryP1 from erythrocyte progenitors. Meanwhile, those OP-HU mice enriched subsets shared significantly up- and down-regulated genes enriched in chromatin modification and cell differentiation and apoptosis such as Bromodomain-containing protein 4 (Brd4), encoding an important chromatin remodeling protein, and Proteinase 3 (Prtn3). Moreover, the specific transcription factors corresponding to the expansion of subHSCs, MPP3-1, GMP5 and EryP1 in OP-HU mice were identified as Zfp951, Nfic, Maz and Ezh2. Finally, inhibition of BRD4 in vivo could partially restore the phenotype of OP-HU mice and the expression of genes regulating HSPC expansion, differentiation and apoptosis.

Conclusions

First of all, our study shows that OP could induce the unbalanced hematopoiesis and enhances the myeloid-biased hematopoiesis. Secondly, OP mice enriched subsets of HSPCs were identified and characterized with enhanced chromatin remodeling, reduced differentiation and resistance to apoptosis. Finally, this study demonstrate that Brd4 regulated gene programs endow the myeloid-biased subsets of HSPCs with tumor cell-like characters in OP mice, which may increase the incidence of the leukemic evolution. This study sheds light on the importance for the prevention of myeloid leukemogenesis in human with OP.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13578-025-01499-3.

Background

In adult mammals, hematopoietic stem cells (HSCs) primary reside in the bone marrow (BM) which provides the supportive microenvironment for the maintenance, proliferation and differentiation of HSCs [1–4]. The altered BM microenvironment contributes to the functional decline of HSCs, clonal hematopoiesis and initiation of hematologic malignancies during aging and other diseases [5–9].

Emerging evidence has illuminated links between bone health and blood homeostasis [10–14]. Osteoporosis (OP) is a bone disease characterized with low bone mineral density, which occurs during aging and other conditions because of unbalanced bone remodeling [15, 16]. OP could lead to the alteration of BM microenvironment and then hematopoiesis. It has been suggested that the number of myeloid cells or white blood cells were negatively associated with the bone mineral density in humans and mice [10, 13]. Furthermore, prospective cohort studies showed elderly people with OP had a higher risk of developing high neutrophils and myelodysplastic syndromes (MDS) [17], which implied predisposition to leukemia. Although several studies have investigated the mechanisms of OP-related hematopoietic homeostasis and impaired lymphopoiesis [14, 18–20], little is known about the alteration in myelopoiesis and their contribution to leukemia initiation under the OP condition.

Myeloid-biased hematopoiesis is a hallmark of HSC dysfunction which could increase the odds of developing hematologic malignancy [21, 22]. HSC-extrinsic and -intrinsic cues have been proved to promote myeloid bias, which in turn increases the risk of leukemogenesis. The biochemical and biomechanical alteration of BM microenvironment, caused by unbalanced bone remodeling, may lead to myeloid-biased hematopoiesis under the condition of OP [23]. During enhanced bone resorption the development of OP, reduced osteopontin (OPN) [24], increased release of latent transforming growth factor 1 (TGF-β1) [25], bone morphogenetic proteins (BMPs) [26, 27], macrophage colony-stimulating factor (M-CSF) and matrix metalloproteinases (MMPs) [28] in BM may lead to the expansion of myeloid-biased HSCs and output of myeloid lineage cells [29]. Therefore, OP is an important extrinsic cue that contributes to BM niche alteration resulting in HSC dysfunction. Therefore, the aim of this study is to investigate whether OP could induce the myeloid-biased hematopoiesis and activate the key gene programs which facilitate leukemogenesis of HSPCs under the condition of OP.

Here, we demonstrate that mechanical unloading-induced OP resulted in the expansion of myeloid-biased subsets of HSPCs with cellular and molecular characters of preleukemia [30], partially regulated by BRD4. To the best of our knowledge, this is the first time to depict the expansion of myeloid-biased subsets of HSPCs and reveal the gene programs promoting the development of myeloid malignancy under the condition of OP. This study sheds light on the importance for the prevention of myeloid leukemogenesis in patients with OP.

Materials and methods

Animal and Hindlimb unloading model

The wild-type C57BL/6 mice were purchased from SPF(BeiJing) Biotechnology CO.,LTD. The male mice (8-week-old) were used for hindlimb unloading for 28 days as described previously. In brief, mice were suspended in individual cage at about 30 degrees by wrapping the tail and hooking on a swivel pulley system, which makes the hindlimbs of mice were unable to touch the ground. For BRD4 inhibiting in vivo [31], 10 mg/kg (body weight) JQ-1 was administrated by intraperitoneally injection per day for 6 consecutive days from the 22nd day of hindlimb unloading for both HU and control mice. Mice without JQ-1 treatment were administrated with PBS containing 5% DMSO by intraperitoneally injection for negative control. All procedures for the use of mice were performed according to the guidelines approved by the Institutional Animal Experiment Administration Committee of the Air Force Medical University. Carbon dioxide (CO2) inhalation was used for mice euthanasia after the experiments according to the AVMA Guidelines. Animal anaesthesia was not used in this study because the experiments do not contain any surgical intervention or other procedures needing animal anaesthesia according to the guidelines. The work has been reported in line with the ARRIVE guidelines 2.0.

Flow cytometry for BM cells and peripheral blood

BM cells were obtained by crushing or flushing femurs and tibia in staining media composed of PBS containing 2% heat-inactivated FBS as described previously [6]. Blood cells were removed with RBC Lysis buffer (eBioscience™, 00–4300-54). Unfractionated BM cells or c-kit enriched BM cells were stained with rat anti-mouse lineage antibodies (B220, BioLegend, 103,202; CD3, BioLegend, 100,202; CD4, eBioscience, 16–0041-85; CD8, BioLegend, 100,702; CD5, BioLegend, 100,602; Gr-1, eBioscience, 14–5931-85; Mac-1, eBioscience, 16–0112-85; Ter-119, eBioscience, 16–5921-85), then goat anti-rat-FITC (Abcam, ab6840) and blocked with rat IgG (Abcam, ab182931). For HSC analysis and sorting, cells were further incubated with c-Kit-APC-Cy7 (BioLegend, 105,826), Sca-1-PB (BioLegend, 108,120), CD48-A647 (BioLegend,103,416), CD150-PE (BioLegend, 115,904), and Flk2-Bio (eBioscience, 13–1351-85) followed by Streptavidin-PE-Cy7 (eBioscience, 25–4317-82). For analysis of HSC apoptosis [32], cells were further incubated with c-Kit-APC-Cy7 (BioLegend, 105,826), Sca-1-PB (BioLegend, 108,120), CD48-A647 (BioLegend,103,416), CD150-PE-Cy7 (BioLegend, 115,913), and Annexin V-PE (4A Biotech, FXP027-50). For progenitors, cells were further incubated with c-Kit-APC-Cy7 (BioLegend, 105,826), Sca-1-PB (BioLegend, 108,120), CD34-PE (BioLegend, 152,203) and CD16/32-PE-Cy7 (BioLegend, 101,317). For BM neutrophils, BM cells were incubated with TruStain FcX™ (BioLegend, 101,320) for Fc receptor blocking and then with Gr-1-PE-Cy7 (Biolegend, 108,415) and CD11b-APC (BioLegend, 101,211). For megakaryocyte, BM cells were stained with CD41-PE (BioLegend, 133,905). For peripheral blood, blood samples were treated with RBC as described in protocol and then stained with CD45R/B220-FITC (BioLegend, 103,205), CD3e-PE (eBioscience, 12–0031-81) and CD11b-APC (BioLegend, 101,211). 7AAD was used for excluding dead cells. Cell sorting were performed on BD FACS Aria III. Cell analysis were performed on Beckman Navious.

Peripheral blood cell counting

Peripheral blood was obtained from the tail of mice and counted using an Auto Hematology Analyzer BC-5000 (MINDRAY) [33].

Micro-CT analysis

Femurs were dissected and fixed in 4% paraformaldehyde for 24 h, following scanned and analyzed using high-resolution micro-CT (Quantum GX2, PerkinElmer, Massachusetts, USA) following the manual [18]. The scanner was at a voltage of 50 kV, a current of 85 uA and a resolution of 10 um per pixel. The images were reconstructed and analyzed for the trabecular bone and cortical bone and the region of interest (ROI) was defined as the area of 5% proximal to the growth plate in the distal femurs, and BV/TV, Tb. Th, Tb. N, Tb. Sp were analyzed.

Colony-forming unit assay

As described previously [34], Colony-forming unit assay for mouse HSCs was performed followed the technical manual. In brief, freshly isolated BM cells (2 × 10⁵ cells/ml) were resuspended in IMDM with 2% FBS. Then, 0.4 mL of diluted cells were added to 4 ml complete methylcellulose medium (MethoCult GF M3434; Stem Cell Technologies). The welly mixed MethoCult medium containing BM cells were dispensed into three 35 mm culture dishes using a 3-cm a sterile 3 mL luer lock syringe with a sterile 16 gauge Blunt- End Needle and were incubated at 37 °C, in 5% CO₂ with≧95% humidity for 12 days. At the end of culture, CFU colonies (CFU-GM, CFU-E, CFU-GEMM) with more than thirty cells were enumerated under an inverted microscope (IX73, Olympus).

Bulk RNA-seq of bone marrow cells

Whole bone marrow cells were harvested for bulk RNA-seq. Total RNA was isolated using TRNzol (DP424, TIANGEN) according to the manufacturer’s instructions and assessed by the Bioanalyzer2100system (Agilent Technologies, CA, USA). RNA libraries for RNA-seq were prepared using NEBNext® Ultra™ RNA Library Prep Kit for Illumina® following manufacturer’s protocols. In brief, mRNA was further purified from total RNA and fragmentated for the synthesis of first strand cDNA, second strand cDNA, sequencing adaptor ligation, PCR amplification and purification for sequencing library preparation described previously [35] and library were sequenced by the Illumina NovaSeq 6000 (Novogene Biotech Co., Ltd, Beijing). After sequencing, clean data were qualified and mapped to the reference genome. The expression of each gene was calculated and differential expression genes were analyzed using DESeq2Rpackage (1.20.0). Gene Ontology (GO) enrichment was performed by the cluster Profiler R package (3.8.1).

Histochemistry

As described [18], femurs were dissected and fixed overnight with 4% paraformaldehyde at 4℃, Samples were decalcified at room temperature using OSTEOSOFT® (Sigma, 1,017,281,000) for 72 h and then embedded in paraffin. Sections with the thickness of 6-μm were prepared and stained with hematoxylin and eosin for tissue histology.

Single-cell RNA sequencing and analysis

Bone marrow cells from femurs and tibias of 4 mice were harvested as mentioned above and combined per group. Lin⁻ c-kit⁺ BM cells were enriched according to methods described previously [33, 36]. Briefly, bone marrow cells were isolated femurs and tibias as mentioned above. After being labeled with lineage microbeads (Miltenyi Biotech, 130–090-858), Cells were passed through a LS MACS column on a MidiMACS magnetic separator and the effluent was collected in the same tube (This fraction represents the enriched lineage negative cells, lin⁻ BM cells). Following, lin⁻ BM cells were incubated with CD117 MicroBeads (Miltenyi Biotech, 130–091-224) and passed through a MS MACS column on a MiniMACS magnetic separator. The magnetically labeled cells were flushed out and collected immediately after washing step by firmly pushing the plunger into the column (This fraction represents the enriched lin⁻ c-kit⁺ BM cells). Lin⁻ c-kit⁺ enriched HSPCs (90,000 cells) were dispelled into single cell for further single-cell RNA sequencing using high density microwell chip (containing 250,0000 wells/chip for capturing 20,000–50,000 single cells, Singleron Biotech Co., Ltd, Nanjing, China) for each sample. Raw reads were processed to generate gene expression profiles using CeleScope v1.12.0 (Singleron Biotechnologies) with default parameters. Scanpy v1.8.1 was used for quality control, dimensionality reduction and clustering under Python 3.7 [37]. For each sample dataset, we filtered expression matrix by the following criteria: 1) cells with gene count less than 200 or with top 2% gene count were excluded; 2) cells with top 2% UMI count were excluded; 3) cells with mitochondrial content > 10% were excluded; 4) genes expressed in less than 5 cells were excluded. Transcriptional Lin⁻ (CD3⁻ CD4⁻ CD5⁻ CD8⁻ B220⁻ Ly6G⁻ CD11b⁻) c-kit⁺ cells were used for further analysis. Cell doublets were estimated based on the expression pattern of canonical cell markers. Any clusters enriched with multiple cell type-specific markers were excluded for downstream analysis. Batch effect between samples was removed by Harmony v1.0 using the top 20 principal components from PCA [38]. After filtering, 48,974 cells were retained for the downstream analyses, with on average 2695.868 genes and 6453.753 UMIs per cell. Cells were clustered using Louvain algorithm and setting resolution parameter at 0.6. Cell clusters were visualized by using Uniform Manifold Approximation and Projection (UMAP). Cell types were annotated using multivariate approach that extracts gene signatures for each individual cell and perform cell identity recognition through hypergeometric tests (HGT). HGT were performed on these gene sets against bone marrow reference from SynEcoSysTM database, which contains all cell-type’s featured genes in the specific organ/tissue. Identity of each cell was determined as the cell-type has the minimal HGT p value. Typical marker genes for the different hematopoietic populations were used for confirming the cell type annotation. Cell differentiation trajectory of monocyte subtypes was reconstructed with the Monocle2 v 2.10.0 [39]. For constructing the trajectory, top 2000 highly variable genes were selected by Seurat (v3.1.2) FindVairableFeatures, and dimension-reduction was performed by DDRTree. The trajectory was visualized by plot cell trajectory function in Monocle2. Transcription factor network was constructed by pySCENIC (v0.11.0) [40] using scRNA expression matrix and transcription factors in AnimalTFDB. First, GRNBoost2 predicted a regulatory network based on the co-expression of regulators and targets. CisTarget was then applied to exclude indirect targets and to search transcription factor binding motifs. After that, AUCell was used for regulon activity quantification for every cell. Cluster-specific TF regulons were identified according to Regulon Specificity Scores (RSS) and the activity of these TF regulons were visualized in heatmaps.

RNA isolation and quantitative PCR analysis

LSK HSCs were sorted by flowcytometry for each sample as described above. Total RNA was purified from 10,000 LSK HSCs using M5 HiPer Supermicroscale RNA Mini Kit (Mei5 Biotechnology, MF789-plus-01). Hifair® Advance Fast 1st Strand cDNA Synthesis Kit (Yeasen, 11150ES60) was used for reverse transcription of low amount mRNA. The expression of genes was determined using 2X M5 HiPer Realtime PCR mix containing SYBR Green (Mei5 Biotechnology, MF015-01) by quantitative PCR. The sequence of primers used in this paper was listed in Table S3.

ELISA

Femurs and tibias were flushed with 200 μL phosphate-buffered saline (PBS) for BM fluid collection. BM fluid was further spun at 500 g for 5 min to get clear supernatant. Supernatants were spun at 12,000 g for 10 min to for further clarifying. Samples were subsequently aliquoted and stored at- 80 °C until use. Amounts of TGFβ in BM fluid were determined by QuantiCyto® Mouse TGF-β1 ELISA kit (Neo-bioscience, #EMC107b) according to the manufacturer’s protocol.

Statistical analysis

Statistical analysis was made using Prism software 8 (Prism, GraphPad Softw are Inc., San Diego, CA). Unpaired two-tailed t test was used to compare data for two groups. One way ANOVA test or two-way ANOVA test was used to compare data for more than two groups or multiparameter analysis [41]. Data are expressed as the mean ± SEM. p < 0.05 was considered statistically significant (* p < 0.05, ** p < 0.01, *** p < 0.001).

Results

Osteoporosis increases myeloid cells in the peripheral blood of mice

Disuse osteoporosis occurred during aging, bed rest and spaceflight, which is characterized with bone loss resulted from unbalanced bone remodeling induced by mechanical unloading [42]. To investigate whether disuse osteoporosis could affect the myelopoiesis in the bone marrow, mice were treated by a well-known method (hindlimb unloading HU) to induce disuse osteoporosis [20] (Fig. 1A). Compared with control mice, the femurs of HU induced OP mice showed increased adipocytes in the bone marrow (Fig. 1B) accompanied with enlarged medullary cavity space (Fig. 1C) and decreased bone mass (Fig. 1D), trabecular bone volume per tissue volume (Tb.BV/TV, Fig. 1E), trabecular number (Tb.N, Fig. 1F) and higher value of trabecular separation (Tb.Sp, Fig. 1G).

Fig. 1.

Mice with osteoporosis (OP) induced by hindlimb unloading (HU) exhibited typical phenotype of OP and perturbation in peripheral blood cells. A Cartoon shows mice were treated by HU for inducing OP for 4 weeks (right). B Histology of femurs from OP-HU mice with increased adipocytes (Bottom) compared to control mice (Top, Con), scale bars, 200 μm. C Representative images of Micro-CT show the enlargement of medullary cavity space of femurs from OP-HU mice (right). D–G Quantification of bone mineral density (BMD, p = 0.0185) (D), trabecular bone volume per tissue volume (Tb.BV/TV, p = 0.0032) E, trabecular number (Tb.N, p = 0.0357) F, and trabecular separation (Tb.Sp, p = 0.0234) G in femurs of control and OP-HU mice respectively. H–J Counts of white blood cell (WBC, p = 0.0136) (H), red blood cells (RBC, p = 0.1838) I and platelets (J, p = 0.0357) in the peripheral blood were compared between control and OP-HU groups. K–L Representative FACS plots of gating strategy for blood cells K and cell number of myeloid cells (p = 0.0134), B cells (p = 0.0018) and T cells (p = 0.1404) in the peripheral blood were compared between control and OP-HU groups L. Con, represents control group, OP-HU, represents osteoporosis group induced by hindlimb unloading. Unpaired t-test was used for the statistics (* p < 0.05, ** p < 0.01, *** p < 0.001)

In the peripheral blood (PB) of OP-HU mice, White blood cells (WBC) count decreased significantly (Fig. 1H), while the count of red blood cells (RBC) and platelets reduced slightly (Fig. 1I-J). Furthermore, B cells were largely reduced in PB of OP-HU mice (Fig. 1K-L), which is consistent with the reports that lymphopoiesis was impaired in mice under the condition of HU [18–20, 43]. On the contrary, both the number and frequency of myeloid cells (CD11b⁺) increased significantly in the PB of OP-HU mice (Fig. 1K-L, Fig. S1A), indicating enhanced myelopoiesis. Together, these findings indicate that myelopoiesis was enhanced in OP-HU mice while lymphopoiesis was reduced.

Myeloid-biased HSPC pools expanded with impaired differentiation and apoptosis in the bone marrow of OP-HU mice

To investigate whether the enhanced myelopoiesis was occurred in the bone marrow of OP-HU mice, the number and frequency of Lin⁻/Sca1⁺/c-Kit⁺ (LSK) cells and its subpopulations were determined respectively. Consistent with previous studies, LSK cells increased significantly in OP-HU mice [43] (Fig. 2A and B, Fig. S1B). Furthermore, in myeloid-biased progenitor compartment, multipotent progenitor 2 (MPP2, Lin⁻/c-kit⁺/Sca-1⁺/CD135⁻/CD150⁺/CD48⁺) and MPP3 (Lin⁻/c-kit⁺/Sca-1⁺/CD135⁻/CD150⁻/CD48⁺) increased significantly (Fig. 2A and C, Fig. S1C), suggesting the presence of myeloid skewing. In addition, the lymphoid-primed MPP4 increased significantly (Fig. 2 A and C, Fig. S1 C), which should be due to the decreased lymphopoiesis of common lymphoid progenitors (CLP) from MPP4 under the mechanical unloading condition reported by recent studies [18, 19]. Moreover, the increase of granulocyte–macrophage progenitors (GMP, Lin⁻/c-kit⁺/Sca-1⁻/CD34⁺/ FcγR^hi) were also observed, which indicated the occurrence of myeloid skewing in the more differentiated progenitors (Fig. 2D and E, Fig. S1D). However, the descendant of GMP, mature neutrophils (CD11b⁺Gr1^hi) and immature neutrophils (CD11b⁺Gr1^low), did not increased as much as the progenitors (Fig. 2D and F, Fig. S1E), suggesting impaired terminal differentiation of myelopoiesis. No extramedullary hematopoiesis was found in the spleens of OP-HU mice (Fig. S1F and G).

Fig. 2.

Myeloid-biased HSPCs expanded with reduced ability of differentiation and apoptosis in bone marrow of OP-HU mice. A Representative FACS plots of gating strategy for Lin⁻/Sca1⁺/c-Kit⁺ (LSK) cells and the subpopulation from bone marrow (top), and the representative FACS plots for determining the frequency of all cell populations in OP-HU and control mice (bottom). B–C The alteration of cell number of LSK cells (B, p = 0.0213), and cell number of LT-HSC (p = 0.2398), ST-HSC (p = 0.3033), MPP2 (p = 0.0493), MPP3 (p = 0.0494), and MPP4 (p = 0.0021) fractions in the bone marrow (BM) of control or OP-HU mice (C) are shown. D Representative FACS plots of gating strategy for common myeloid progenitors (CMP), granulocyte–macrophage progenitors (GMP), megakaryocytic-erythroid progenitors (MEP) (upper panel), mature and immature neutrophils (bottom panel) in mice. E–F Cell number of CMP (p = 0.3365), GMP (p = 0.0236), MEP (p = 0.7791) (E), and mature neutrophils (p = 0.1246), immature neutrophils (p = 0.7251) (F) in the bone marrow of control or OP-HU mice are shown. G Representative image of CFU colony formed by culturing BM cells of control mice with MethoCult GF M3434 (scale bars, 500 μm). H Colony number of GM (p = 0.0175), E (p = 0.0098), GEMM (p = 0.0213) were counted under an inverted microscope and shown as control and OP-HU group respectively. I Representative FACS plots of gating strategy for analysis the apoptosis of HSPCs after hindlimb unloading. J–L The percentage of apoptotic cells in LSK (J, p = 0.0106), MPP3/4 (K, p = 0.0134) and LK.⁺ myeloid progenitors (L, p = 0.0021) in the bone marrow of control or OP-HU mice were shown. Unpaired t-test was used for the statistics (* p < 0.05, ** p < 0.01, *** p < 0.001)

Following, we performed clonogenic methylcellulose assays to test whether the differentiation of HSPCs was impaired in OP-HU mice. The number of colony-forming unit (CFU-GM, CFU-E, CFU-GEMM) decreased significantly (Fig. 2G and H), suggesting the impaired ability of stem cell maintenance or differentiation into mature myeloid cells. Our data are consistent with a previous study which showed the impaired ability of reconstitution of BM from OP-HU mice by transplantation [43]. Moreover, the apoptosis of LSK cells, MPP3/4 and myeloid cell progenitors (Lin⁻/c-kit⁺/Sca-1⁻) decreased significantly in OP-HU group (Fig. 2I-L), which could result in the expansion of HSPC pools. Taken together, these findings demonstrate that in OP-HU mice, MPP3 and GMPs undergo myeloid-biased hematopoiesis with impaired differentiation and resistance to apoptosis, indicating a potential for leukemogenesis.

Bone marrow microenvironment of OP-HU mice supports the expansion of HSC pool and myeloid-biased hematopoiesis

In order to investigate the way by which the bone marrow environment of OP-HU mice supports the expansion of HSCs and myeloid-biased hematopoiesis, bulk RNA-seq of was performed with whole bone marrow cells of mice after hindlimb unloading (Fig. 3A, Fig. S2A). Transcriptome showed genes for leukocyte differentiation were downregulated (Fig. 3B). Among these genes, Pf4 was the only one encoding secreted protein mainly expressed in megakaryocytes (MKs) and has been proved to limit the myeloid output of HSCs [44] (Fig. 3C). Similar to PF4, lack of MKs secreted TGFβ1 has been shown to resulting in the expansion of HSCs [45]. Although Tgfb1 was slightly downregulated at the transcriptional level, the level of TGFβ1 protein in the fluid of BM decreased significantly (Fig. 3D). A possible reason for the dropping of total TGFβ1 protein is that OP reduced the storage of latent TGFβ1 in the bone matrix. Further study showed the declined frequency of MKs in the BM of OP-HU mice (Fig. 3E and F) indicating the reduced source of PF4 and TGFβ1. In addition, the slightly reduced Ccl3 [46] or increased Ccl5 [47] also contributed to reduced maturation of myeloid cells or myeloid skewing of HSCs (Fig. S2B). Therefore, these data show that the concentration of several cytokines varied in the BM environment of OP-HU mice, which in turn could support the expansion of HSC pool and myeloid-biased hematopoiesis.

Fig. 3.

The alteration of bone marrow (BM) cells in OP-HU mice. A Heatmap shows the transcriptional change of BM cells from OP-HU and control mice with 3 replicates per group. B Gene ontology (GO) shows the downregulated gene term enriched in the BM cells of OP-HU mice based on bulk RNA-seq. C Expression of Pf4 (Cxcl4, p = 0.0002), TGF-β1 (p = 0.1210) as shown by the fragments per kilobase of transcript per million mapped reads (FPKM) based on bulk RNA-seq. D The concentration of total TGF-β1 in BM fluid of control and OP-HU mice was determined by ELISA (p = 0.0065). E Representative FACS plots of gating strategy for megakaryocytes in BM of mice. F Frequency of megakaryocytes declined significantly in OP-HU mice compared to control (p = 0.0040). Unpaired t-test was used for the statistics (* p < 0.05, ** p < 0.01, *** p < 0.001)

Single cell RNA-seq identifies the specifically expanded subsets of HSPCs in OP-HU mice

To understand the molecular mechanisms driving the myeloid-biased hematopoiesis and the potential of leukemia evolution (resistant to differentiation and apoptosis) in HSPCs of OP-HU mice, single cell RNA sequencing was performed to analyze the transcriptional changes of Lin⁻/c-kit⁺ HSPCs from the femur and tibia of OP-HU and control mice. In total, we profiled 48,974 HSPCs (control 25,396 cells vs OP-HU 23578 cells) which formed 8 clusters including HSCs, GMP, megakaryocyte progenitors (MkP), erythroid progenitors (EryP) and more differentiated myeloid progenitors (Fig. 4A). HSCs stood at the center surrounding with GMP, MkP, EryP and other myeloid cells. The annotation of different cell types was performed based on the differentially expressed genes and lineage markers [2, 6, 48] (Fig. 4B, Fig. S3A–E).

Fig. 4.

The alteration of landscape and transcriptional programs in HSPCs from BM of OP-HU mice. A Transcriptomic landscape of lin⁻ c-kit⁺ HSPCs from BM of control (middle, n = 4) or OP-HU mice (right, n = 4) on U-MAP. All represents the combination of lin⁻ c-kit⁺ HSPCs from control and OP-HU mice. B Dot plot shows top differently expressed genes for identifying different cell types within lin⁻ c-kit⁺ HSPCs. C U-MAP shows the alteration of subset distribution within HSCs, GMPs, Eryp and Mkp from control to OP-HU mice. D–G Bar plot shows the percentage of subsets changing in HSCs (D), GMPs (E), Ery (F) and MkP (G) between control and OP-HU mice respectively. H–K GSEA enrichment shows epigenetic changes of HSCs (H), GMPs (I), Ery (J) and MkP (K) in OP-HU mice compared to control. L Heatmap shows the alteration of gene expression related to epigenetic remodeling (Brd4, Kmt2e, Dek, Npm1), HSC maintaining (Malat1, Meis1, Hoxa9), differentiation (Prtn3, Cebpa) and apoptosis (Bak1, Noxa) between control and OP-HU mice. M Comparison of the expression of CD52 and CD48 among subsets within MPP3 for isolating CD52^high CD48^highMPP3-1 cells based on single cell RNA-seq data

In order to identify subsets of the four main cell types specifically enriched in OP-HU mice, cells were further divided into subsets for HSCs (subHSCs, MPP2, MPP3 (MPP3-1 to MPP3-3), MPP4 (MPP4-1 and MPP4-2)), GMP (GMP1 to GMP), EryP (EryP1 to EryP4) and MkP (MkP1, MkP2) based on the differentially expressed genes ((Fig. 4C, Fig. S3F-I). HSCs, GMP, EryP and MkP comprised clusters specifically enriched in OP-HU mice including subHSCs, MPP3-1, MPP4-1 from HSCs, GMP5 from GMP, EryP1 from EryP and from MkP (Fig. 4C). The increased percentage of subHSCs, MPP3-1, MPP4-1, GMP5, EryP1and MkP2 in OP-HU mice were shown (Fig. 4D-G). Because MPP3 mainly contributes to the downstream myelopoiesis, we tried to explore the cell surface markers to distinct the OP mice specific MPP3-1 with other subsets in MPP3 population based on our data. Among the different expressed genes between MPP3-1 and the other two subsets, CD52 and CD48 was highly expressed in MPP3-1 (Fig. 4 M–N). Therefore, the phenotype of Lin⁻ Sca-1⁺ c-Kit + CD150⁻ CD48 ^highCD52^high from LSK HSCs could be isolated for future study.

Moreover, Gene set enrichment analysis showed that epigenetic remodeling of chromatin was shared by HSCs, GMPs and MkP (Fig. 4 H, I, K), except non-coding RNA processing was enriched in EryP (Fig. 4 J). This result indicates that the alteration of epigenetics occurred in HSPCs of OP-HU mice. Furtherly, OP-HU mice specifically enriched subsets also shared significantly upregulated genes related to chromatin modification and downregulated genes related to differentiation and apoptosis (Fig. 4 L, Table S1–S2), indicating a systematic change from HSCs to more differentiated progenitors. Although the specific cluster in MkP did not increased as much as other specific populations, several important genes involved in megakaryopoiesis (Itgb3, Vwf, Gp1ba, Pecam1) were also downregulated in OP-HU mice (Fig. S3J), suggesting a suppression of megakaryocyte lineage production. Therefore, our data identifies the specifically enriched cell population from HSCs to myeloid progenitors with shared transcriptional alteration in OP-HU mice at single cell resolution.

The expanded specific subsets of HSPCs in OP-HU mice were characterized with altered transcriptional programs contributed to myeloid-biased hematopoiesis and leukemogenesis

To further analyze whether those subsets enriched in OP-HU mice were corresponding to the myeloid-biased hematopoiesis, the molecular characters of each subpopulation was profiled. Firstly, subHSCs showed the status of quiescence, while MPP3-1 and MPP4-1 were more active in mitochondrion activity and cell division respectively among all HSC subsets (Fig. S4A–C). This result confirmed the accuracy of cell subset identification.

Moreover, subHSCs, MPP3-1 and MPP4-1 shared upregulated genes regulating the epigenetic modification (Brd4, Kmt2e, Dek, Npm1) [49–52], stem cell maintaining and renewal (Malat1, Meis1, Hoxa9) [53–56], as well as downregulated genes for terminal differentiation of myeloid cells (Prtn3, Mpo, Cebpa) [57–60] (Fig. 4L, Fig. S4D, Table S1–S2). This indicates the molecular changes of the MPP3-1 and MPP4-1 should be originated from subHSCs, which led to the myeloid-biased hematopoiesis and inhibition of terminal differentiation.

Furthermore, genes promoting apoptosis (Bak1, Pmaip1/Noxa) [61, 62] were also downregulated in both MPP3-1 and MPP4-1 rather than subHSCs (Fig. 4L, Fig. S4D, Table S1–S2). This implies HSPCs became resistance to apoptosis from the stages of MPPs in OP-HU mice. Together, these results demonstrate the epigenetics of HSCs in OP-HU mice altered from primitive stage to downstream progenitors, which may lead to the impairment of terminal differentiation and apoptosis.

To examine whether those altered expression of genes could be found in myeloid progenitors downstream of MPPs, OP-HU mice enriched GMP5 and EryP1 were further analyzed. GMP5 and MPP3-1were immediately adjacent on the U-MAP (Fig. S4F) and shared many upregulated genes for epigenetic modification (Brd4, Kmt2e, Dek, Npm1), stemness maintaining (Malat1, Meis1,) and mitochondria activity (Slc25a5/Ndufa4/Uqcr10) and downregulated genes for terminal differentiation of myeloid cells (Prtn3) and promoting apoptosis (Bak1) (Fig. 4L, Table S1–S2). This indicates the altered molecular characters of GMP5 were inherited from MPP3-1. In addition, EryP1 showed a proliferation and undifferentiation status with upregulated genes for epigenetic modification (Dek, Prmt5, Ezh2) [63, 64] and ribosome biogenesis (Npm1, Ncl, Lyar) [65, 66], and downregulated genes for myeloid or erythrocyte differentiation (Prtn3, Gfi1b, Hmgb2, Gata1) [67–69] (Fig. 4L, Table S1-S2). Therefore, GMP5 and EryP1 had similar epigenetic changes to their precursors with reduced ability of differentiation, which may account for the expansion of GMP and megakaryocytic-erythroid progenitors (MEP) (Fig. 2D and E).

Lastly, the pseudotime trajectories showed OP mice specifically enlarged subsets of HSCs, GMPs and EryP distribution (Fig. 5A–C), which demonstrate the cell status shift of subsets between OP-HU mice and control group. SCIENIC analysis was performed to explore cell type specific regulons potentially driving the expansion of OP mice enriched subsets. Zfp951 (regulating 13 genes), Nfic (regulating 7 genes), Maz (regulating 13 genes) and Ezh2 (regulating 748 genes) were identified as potential regulons driving the expansion of subHSCs, MPP3-1, GMP5 and EryP1 cell populations respectively (Fig. 5D). While the function of Zfp951 in hematopoiesis is unknown, Nfic, Maz and Ezh2 are known regulator for murine HSPCs survival and erythropoiesis. Therefore, the cell type specific regulons should be responsible for the alteration of cell development trajectories.

Fig. 5.

The alteration of cell subset development trajectories and key regulons regulating the myeloid-biased hematopoiesis in OP-HU mice. A–C The pseudotime trajectories show alterative developing routine of OP-HU mice specially enriched subsets subHSCs and MPP3-1 (A), GMP5 (B) and EryP1 (C) respectively. D The top five regulons regulating the cell status of OP mice enriched subHSCs and MPP3-1, GMP5 and EryP1 were analyzed by pySCENIC based on single cell RNA-seq data

Taken together, these data show that epigenetic remodeling take placed from HSCs to myeloid progenitors, which could promote the myeloid-biased hematopoiesis and the leukemic evolution programs via regulating gene expression involved in cell expansion, differentiation and apoptosis.

Inhibition of BRD4 partially reverses the tumor cell-like characters of HSCs in OP-HU mice

According to the single cell RNA-seq data, Brd4 was upregulated in HSCs, GMP of OP-HU mice and the specific enriched subsets, MPP3-1, MPP4-1 and GMP5 (Fig. S4D–E). BRD4 is an important epigenetic reader which recognizes the acetylated histone for transcription activation and elongation [70] and is required for normal adult hematopoiesis and overexpressed in myeloid malignancies. BRD4 also functions as a histone acetyltransferase for chromatin decompaction [71]. However, it is still unknown whether inhibition of BRD4 could reverse the tumor-like characters of HSPCs in OP-HU mice.

To test whether the inhibition of BRD4 could reverse the tumor-like characters of HSCs induced by OP, both control and OP-HU mice were treated with JQ1(BET inhibitor, 10 mg/kg/day) for 6 consecutive days starting at 22nd day of hindlimb unloading (Fig. 6A). JQ-1 treatment significantly reduced the frequency of MPP3/4 in OP-HU mice, indicating that BRD4 plays crucial role in the enlargement of HSC pool (Fig. 6B). In addition, the frequency of LSK cells and MPP3/4 increased significantly in JQ-1 treat ed control mice, which maybe a compensation to the increased apoptosis of HSCs Fig. 6B and C).

Fig. 6.

Inhibition of BRD4 partially restored the BM phenotype of OP-HU mice. A Experiment design of the JQ1 injection for control-JQ1 and OP-HU-JQ1 mice. Untreated mice and OP-HU mice without JQ1 are included as baseline control. B Frequency of LT-HSC (Con-DMSO vs. Con-JQ1, p = 0.0444; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.0907), ST-HSC (Con-DMSO vs. Con-JQ1, p = 0.0239; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.5265), MPP2 (Con-DMSO vs. Con-JQ1, p = 0.0680; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.34394), MPP3/MPP4 (Con-DMSO vs. Con-JQ1, p = 0.0089; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.0079) fractions in the bone marrow (BM) of control, control-JQ1, OP-HU, OP-HU-JQ-1 mice were analyzed by flowcytometry and shown in bar plot (Unpaired t-test was used for the statistical analysis). C Frequency of LSK cells in control and OP-HU mice treated with or without JQ1were were analyzed by flowcytometry and shown (Con-DMSO vs. Con-JQ1, p = 0.0003; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.2127, One-way ANOVA). D The changes of CFU ability of BM cells from OP-HU and control mice treated with or without JQ1 were determined CFU assay (Con-DMSO vs. Con-JQ1: GM, E and GEMM, p = 0.0224, p < 0.0001, p = 0.0359 respectively; OP-HU-DMSO vs. OP-HU-JQ1: GM, E and GEMM, p < 0.7608, p = 0.7754, p = 0.9986 respectively; Con-DMSO vs. OP-HU-DMSO: GM, E and GEMM, p = 0.0092, p < 0.0001, p = 0.0359 respectively, One-way ANOVA was used). E The percentage of apoptosis in LSK cells from OP-HU and control mice were analyzed by flowcytometry (Con-DMSO vs. Con-JQ1: GM, E and GEMM) after JQ1 treatment in vivo (Unpaired t-test was used). F Quantitative PCR shows the change of genes regulating HSC maintenance, differentiation, and apoptosis after JQ1 treatment. HoxA9 (Con-DMSO vs. Con-JQ1, p = 0.1330; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.0001), Meis1 (Con-DMSO vs. Con-JQ1, p < 0.0001; OP-HU-DMSO vs. OP-HU-JQ1, p < 0.0001), Runx2 (Con-DMSO vs. Con-JQ1, p < 0.0001; OP-HU-DMSO vs. OP-HU-JQ1, p < 0.0001), Cebpα (Con-DMSO vs. Con-JQ1, p = 0.0067; OP-HU-DMSO vs. OP-HU-JQ1, p < 0.0001), Bcl2 (Con-DMSO vs. Con-JQ1, p = 0.0012; OP-HU-DMSO vs. OP-HU-JQ1, p = 0.0059), Noxa (Con-DMSO vs. Con-JQ1, p < 0.0001; OP-HU-DMSO vs. OP-HU-JQ1, p < 0.0002), Two-way ANOVA was used. (* p < 0.05, ** p < 0.01, *** p < 0.001)

Furthermore, CFU results revealed a slightly decrease in CFU-G, CFU-E, CFU-GEMM in OP-HU mice with JQ-1 treatment compared to untreated OP-HU mice (Fig. 6D). OP-HU mice treated with JQ-1 showed comparable rate of apoptosis with untreated control mice, indicating partially recovering of the apoptosis by inhibiting of BRD4 (Fig. 5E). Together, inhibition of BRD4 could almost restored the apoptosis, partially restored the differentiation of HSPCs.

Finally, we test whether the upregulation of Brd4 also regulate the gene programs related to preleukemia characters of HSPCs in OP-HU mice. we examined the expression of BET target genes (Hoxa9, Meis1, Runx2, Cebpa, Bcl2, Noxa) in our single cell-seq data (Fig. S4D). In the OP-HU mice specific enriched subsets of HSCs, genes promoting HSC expansion (Hoxa9, Meis1) and inhibiting differentiation (Runx2) and apoptosis (Bcl2) were upregulated, while genes promoting HSC differentiation (Cebpa) and apoptosis (Noxa) were downregulated. Those genes were sensitive to JQ1 in AML and other cancer cells [70, 72, 73] and were proved to regulate the normal hematopoiesis and malignancy development [53, 58, 74–77].

Moreover, inhibition of BRD4 in vivo by JQ1 could partially restore the expression of Hoxa9, Mesi1, Cebpa and Noxa or nearly restore the expression of Runx2 and Bcl2 in HSCs, which further confirmed the role of Brd4 in regulating HSC function and leukemia development (Fig. 6F). Together, these results reveal that upregulation of Brd4 play important roles in the apoptosis and differentiation of HSPCs in OP-HU mice.

Discussion

Although people with OP have increased risk for developing high neutrophils and MDS, the cellular and molecular alterations of myeloid-biased HSPCs induce by OP remain unclear. Here, we provide evidence that myeloid-biased HSCs with impaired ability of differentiation and resistance to apoptosis were expanded in the HSPC pool of OP-HU mice. The specifically enriched cell populations (from HSCs to downstream myeloid progenitors) in OP-HU mice were identified by single RNA-seq. Those specifically enriched subsets share similar transcriptional alterations related to epigenetic remodeling. Among the upregulated genes related to epigenetic remodeling, Brd4 should be one of the key players in this process, because inhibition of BRD4 could partially restored the preleukemia characters of HSPCs during hindlimb unloading.

The majority of people who suffered from OP including postmenopausal women, older men (> 50 years of age), astronauts and others [12, 78]. Given the reduced mechanical loading occurred in various types of OP and could lead to unbalanced bone remolding, hindlimb unloading model of mice were used to investigate the effect of OP related BM niche alteration on myelopoiesis in this study.

Compared to a previous study which used two-week hindlimb unloading and did not find any effect of OP on myeloid-biased HPSCs except for reduced frequency of CLPs [19], four-week hindlimb unloading was used in our study. This indicated the effect of hindlimb unloading on HSPCs is time dependent. Therefore, our results provide the evidence for the expansion of myeloid-biased HSC pool in BM of OP-HU mice.

Our study found the number of myeloid-biased MPP2 (CD135⁻/CD150⁺/CD48⁺ LSK) and MPP3(CD135⁻/CD150⁺/CD48⁻ LSK) increased significantly in BM of OP-HU mice. In contrast to another study [43], they divided LSK HSCs into three subsets and found the number of subsets significantly increased in BM of mice with OP. However, the myeloid-biased HSCs could not be distinguished from LSK cells based on the combination of CD34 and CD135. This study only showed the impaired differentiation, apoptosis of LSK HSCs and transcriptional change of Flt3- LSK cells (also named CD135⁻ LSK cells, containing MPP2, MPP3, HS^CST and HSC^LT in our paper) in bulk. In contrast, our study focuses on the myeloid-biased HSCs, explores the altered genes programs at single cell level and provides an explanation for OP patients with higher risk for developing high neutrophils and MDS.

In order to get the whole picture of transcriptome information for all abundant and rare population in the subpopulation of HSPCs at the same time, we successfully acquired the transcriptome data of more than 20, 000 LK⁺HSPCs (including approximately 1% primitive HSCs) at single cell resolution for each group. By doing this, we could analyze the changes of stemness and lineage differentiation for HSPCs at different stages, thereby depicting the whole picture of HSPC status in OP-HU mice. Our study demonstrated the systematic molecular changes from primitive HSCs to more differentiated progenitors included enhanced chromatin modification, reduced differentiation and apoptosis in OP-HU mice. Moreover, the enhanced chromatin modification was found in OP-HU mice enriched subsets including subHSCs, MPP3-1, GMP5 and so on. In addition, the pseudotime trajectories and SCIENIC analysis of cell subsets further reveals the key regulons regulating shift of cell status in OP mice. These data indicate that the enhanced epigenetic modification might drive the expansion, resistance to differentiation and anti-apoptosis of myeloid-biased HSCs in OP-HU mice.

It has been suggested that Brd4 is necessary for normal adult hematopoiesis [49] and myeloid malignancies [70, 79, 80]. Our data showed that the upregulation of Brd4 in HSPCs resulted in the upregulation of genes related to HSC expansion, and downregulation of genes related to HSC differentiation and apoptosis. Therefore, the upregulation of Brd4 is not only important for the proliferation of leukemia cells but also for the myeloid-biased hematopoiesis. Together, our study expands the understanding of OP induced myelopoiesis and leukemogenesis by presenting a detailed transcriptional landscape of HSPCs at single-cell resolution. Future studies are needed to investigate the role of Brd4 in clonal hematopoiesis and how the signals from BM niche induced by OP were integrated and transmitted into HSCs, and lead to epigenetic remodeling.

Conclusions

Overall, our findings demonstrate disuse osteoporosis, occurred during aging and other conditions, could lead to the enlargement of myeloid-biased compartment of HSPCs and initiate the gene programs involved in leukemogenesis. Osteoporosis induced the epigenetic remodeling of HSPCs which endows the myeloid-biased subsets with tumor cell-like characters by reducing differentiation and apoptosis. Moreover, our study demonstrates the upregulation of epigenetic reader BRD4 promoting myeloid-biased hematopoiesis and tumor cell-like state of HSPCs in OP-HU mice. In a word, this study not only demonstrates the way that osteoporosis contributes to the myeloid-biased hematopoiesis and leukemogenesis, but also shows that BRD4 is a potential target for preventing the initiation of leukemogenesis induced by osteoporosis. These discoveries give a promising future for keeping fit of HSCs during osteoporosis.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

OP

Osteoporosis

MPP3

Multipotent progenitor 3

GMPs

Granulocyte-monocyte progenitors

Brd4

Bromodomain-containing protein 4

HSCs

Hematopoietic stem cells

BM

Bone marrow

MDS

Myelodysplastic syndromes

OPN

Osteopontin

TGF-β1

Transforming growth factor 1

BMPs

Bone morphogenetic proteins

M-CSF

Macrophage colony-stimulating factor

MMPs

Matrix metalloproteinases

OP-HU

Osteoporosis induced by hindlimb unloading

CO2

Carbon dioxide

ROI

Region of interest

GO

Gene ontology

UMAP

Uniform Manifold Approximation and Projection

HGT

Hypergeometric tests

HU

Hindlimb unloading

MKs

Megakaryocytes

MkP

Megakaryocyte progenitors

EryP

Erythroid progenitors

MEP

Megakaryocytic-erythroid progenitors

BMD

Bone mineral density

Tb.BV/TV

Trabecular bone volume per tissue volume

Tb.N

Trabecular number

Tb.Sp

Trabecular separation

WBC

White blood cell

RBC

Red blood cells

PB

Peripheral blood

LSK

Lin⁻/Sca1⁺/c-Kit⁺

CMP

Common myeloid progenitors

MEP

Megakaryocytic-erythroid progenitors

Author contributions

Chen Zhang, Yueru Ji, and Xiaotong Gao contributed equally to this work. Chen Zhang performed conceptualization, investigation, experiments, data analysis and wrote the manuscript. Yueru Ji contributed to investigation and data analysis. Xiaotong Gao contributed to FACS analysis. Zhuo Wan contributed to data analysis. Fangna Gu contributed to mouse maintenance. Dian Lou contributed to preparing the manuscript. Han Liang contributed to cell sorting by flowcytometry. Li Liu., Weiwei Qin contributed to conceptualization supervision, project administration.

Funding

This work was supported by grants from Yinfeng Plan of Tangdu hospital (2021YFJH009), Natural Science Basic Research Plan in Shaanxi Province of China (Program No. 2018JQ1033, Program No. 2018SF153), and from the National Natural Science Foundation of China (82200171).

Data availability

The bulk RNA-seq data and single-cell RNA-seq data that support the findings of this study have been deposited in the Gene Expression Omnibus under accession code GSE281210 and GSE281970. This paper does not report original code. Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

Declarations

Ethics approval and consent to participate

All procedures were approved by the Animal Care Committee of the Air Force Medical University. The title of the approved project was the study of epigenetic mechanisms in dysregulated hematopoiesis mediated by Brd4 under the condition of disuse osteoporosis. Approval number:20240177. Date of approval: 03/01/2024. This study complied with the ARRIVE guidelines statement during the experiment. There is no human sample in this article and informed consent is not applicable.

Consent for publication

All authors confirm their consent for publication.

Competing interests

The authors declare no conflicts of interest.