Differentiation of mesenchymal stem/stromal cells into CD45⁺ macrophage-like cells: expanding insights into MSC plasticity

Summary

Mesenchymal stem/stromal cells (MSCs) are notable for their differentiation potential and immunomodulatory properties. They share several features with macrophages as well as fibrocytes. This study aimed to investigate the relationship between MSCs, fibrocytes, and macrophages, focusing on the hypothesis of a possible direct lineage. Particularly, the focus lies on a freshly isolated murine MSC population known as PαS cells, which were transplanted into irradiated recipients, resulting in cells expressing a macrophage-like phenotype with CD45, CD68, and CSF1R (also known as CD115) expression. Further analysis showed that PαS cells cultured in macrophage differentiation media acquired characteristics of M2 macrophages (CD45, CD68, and CD206) through interaction with CSF1R. These results were confirmed in several mouse strains and human MSCs. Single-cell RNA sequencing revealed macrophage-like populations among the PαS cells, demonstrating the phenotypic plasticity of non-cultured MSCs, giving into MSC potential roles in immune regulation and tissue repair.

Graphical abstract

Highlights

• •Freshly isolated MSCs transitioned into CD45⁺ CSF1R⁺ macrophage-like cells
• •M-CSF stimulation induced MSCs to express CD206 with macrophage-like morphology
• •RNA-seq revealed MSC subpopulations with hematopoietic markers after M-CSF stimulation
• •CSF1R activation drives MSC-to-macrophage differentiation in both mouse and human MSCs

Cell biology; Immunology

Introduction

Mesenchymal stem/stromal cells (MSCs) are widely used in medical applications owing to their immunomodulatory properties, differentiation potential, and low immunogenicity.^1,2 As markers for MSCs, it has been reported that hematopoietic lineage markers such as CD45 and Ter119 are negative, while Sca-1, PDGFRα, CD73, CD90, and CD105 are commonly recognized as specific markers.^3,4,5 Previously reported MSC populations with various combinations of these markers are believed to represent largely the same populations as they can differentiate into various cells, including adipocytes, chondrocytes, osteoblasts, and myofibroblasts.^6,7

MSCs and fibrocytes are involved in wound healing, tissue repair, and fibrotic diseases.^{8,9,10,11,12,13} As hematopoietic derivatives, fibrocytes express markers such as CD11b, CD14, CD34, and, most importantly, CD45¹⁴ and migrate to injury sites where they transition to a mesenchymal phenotype and contribute to tissue fibrosis.¹⁵ Fibrocytes and macrophages have several similarities despite a distinct phenotype. Both have been associated with fibrosis, are hematopoietic-derived, and share several surface markers, leading some to suggest a lineage connection.^16,17

The interactions between immune cells and MSCs have been extensively studied. Their immunomodulatory potential makes them promising candidates for treating immune-mediated diseases.¹⁸ Numerous studies have shown that these cells exert immunomodulatory functions mainly via interactions with immune cells such as T, B, and natural killer cells.^19,20,21 Their low immunogenicity and ease of availability make them excellent candidates for stem cell therapy.^1,2,22

However, other studies have suggested that MSCs can aggravate the disease. Our study demonstrated that fresh (not cultured) MSCs contribute to the development of graft-vs-host disease.²³ Similarly, human bone marrow-derived MSCs upregulate major histocompatibility complex (MHC) class II expression when exposed to lower levels of IFNγ, acting as immunostimulants.²⁴ However, this effect was not observed at higher IFNγ levels. Another study showed that human MSCs can be polarized via toll-like receptor (TLR) signaling. MSCs also display a proinflammatory phenotype in an underactivated immune system. This polarization is mediated by TLR4 signaling, whereas the anti-inflammatory polarization is mediated by TLR3 signaling.²⁵ Owing to their similarities to monocytes, Waterman et al. suggested the terms MSC1 and MSC2, paralleling M1/M2 macrophages.

The MSC bipolar characteristics resemble those of macrophages, which become pro- or anti-inflammatory in response to their environment.^26,27 Pro-inflammatory MSCs also recruit lymphocytes and activate the T cell secretion of macrophage inflammatory protein-1, CCL5, CXCL9, and CXCL10.^28,29,30 Anti-inflammatory MSCs suppress immune responses primarily when exposed to IFN-γ, TNF-α, and IL-1β,^31,32,33 with IFN-γ specifically, inducing PD-L1 and PD-L2 inhibitors.^19,34 This characteristic is shared with bone marrow-derived macrophages, as IFN-γ can induce PD-L1 expression in tumor-associated macrophages, which promotes M2 polarization.^35,36

In this study, based on the similarities among MSCs and macrophages, we conducted a detailed analysis of the relationships between these cell types. After purifying the MSC fraction (CD45⁻ TER119⁻ Sca-1⁺ PDGFRα⁺), these cells expressed macrophage markers in vivo. Furthermore, we stimulated a macrophage phenotype change by exposing MSCs to M-CSF, suggesting that these cells may be directly related through differentiation. Using single-cell RNA sequencing (scRNA-seq) revealed the presence of a population within the MSC fraction that differentiates into macrophages. These findings may reflect the previously unrecognized in vivo heterogeneity of MSCs.

Results

Freshly isolated mouse mesenchymal stem/stromal cells to transition into the hematopoietic lineage

For in vivo analysis, CD45⁻ TER119⁻ Sca-1⁺ PDGFRα⁺ (PαS) cells were isolated from transgenic GFP C57BL/6 mice and transplanted into irradiated recipient C57BL/6 mice (Figure 1A). After 3 weeks, the lacrimal glands, lungs, and spleen were examined. These organs are involved in immune responses and associated with various diseases, including autoimmune disorders such as graft vs. host disease and Sjögren’s syndrome, which may be candidates for MSC therapy.^37,38 Although the transplanted cells (GFP⁺ PαS cells) did not express Cd45 (denoted as CD45) (Figure 1A), the presence of GFP⁺ CD45⁺ CSF1R⁺ CXCR4^+/− cells was observed in all examined tissues, following the transplantation of GFP⁺ PαS cells (Figures 1B and 1C) (Table 1). No discernible or consistent patterns of cell distribution were observed within the tissues. In addition, the number of GFP⁺ cells varied significantly among the samples; however, they were generally scarce, except in the spleen.

Figure 1.

Identification and distribution of GFP⁺ PαS-derived cells in vivo

(A) Visual summary of the experimental process of in vivo experiments. GFP⁺CD45⁻TER119⁻Sca-1⁺PDGFRα⁺ (PαS) cells were isolated from transgenic GFP mice and transplanted intravenously into recipient mice. The inset graph shows CD45 mRNA expression, confirming the absence of hematopoietic markers in PαS cells compared to bone marrow (BM) cells (n.d., not detected). After transplantation, various organs, including the lacrimal gland, lung, spleen, and BM, were analyzed.

(B) Immunofluorescence staining of recipient organs (lacrimal gland, lung, and spleen) shows localization of GFP⁺ cells. Immunohistological staining of PαS-derived cells (GFP, red) and CD45 (gray) of lacrimal gland, lung, and spleen tissue, and counterstained with DAPI (blue). Scale bars, 200 μm.

(C) Immunohistological staining of PαS-derived cells (GFP, green), CSF1R (red), and CXCR4 (gray) of lacrimal gland, lung, and spleen tissue, and counterstained with DAPI (blue). Scale bars, 200 μm.

(D) Flow cytometric analysis of blood and bone marrow after transplantation. Gating strategy of detection of PαS (GFP⁺) cells derived from cells. CD45⁺ (green box), GFP⁺ (red box) were investigated for the expression of CXCR4 (y axis) and CSF1R (x axis). Gates were set based on FMOs.

(E) Overall measurement of PαS cells derived CD45⁺ cells from (D) (Blood, n = 12, BM, n = 8, ∗p < 0.05, Student’s t tests).

(F) CD45⁺, CXCR4^±, and CSF1R^± cells derived from PαS cells in the blood (n = 12) from (D).

(G) CD45⁺, CXCR4^±, and CSF1R^± cells derived from PαS cells in the BM (n = 9) from (D).

Table 1.

Antibody list for tissue sections and cell culture

Name	Clone	Reporter	Company	Dilution
Anti-mouse CD45	30-F11	PE	BioLegend	1:500
Anti-mouse CD45	30-F11	APC	BioLegend	1:500
Goat anti-mouse CD45	Polyclonal	None	R&D Systems	1:500
Goat anti-GFPuv	Polyclonal	None	R&D Systems	1:500
Anti-mouse CD68	FA-11	APC	BioLegend	1:500
Anti-mouse CXCR4 (flow cytometry)	1276F12	APC	BioLegend	1:500
Anti-mouse CXCR4 (IHC)	UMB2	None	Abcam	1:500
Anti-mouse CD115	AFS98	None	Thermo Fisher	1:500
Donkey anti-Goat IgG	Polyclonal	Alexa Fluor® 555	Invitrogen	1:400
Goat Anti-Rabbit IgG	Polyclonal	Alexa Fluor® 647	Abcam	1:400
Anti-human CD45	HI30	PE	BioLegend	1:500
Anti-human CD206	15–2	A488	BioLegend	1:500
Anti-human CD68	Y1/82A	APC	BioLegend	1:500

To validate these observations, bone marrow and blood samples were collected and analyzed for the fibrocyte markers CD45 and CXCR4, as well as the macrophage marker CSF1R, which is not expressed by fibrocytes.³⁹ Flow cytometry revealed that many GFP⁺ cells in the blood and bone marrow were positive for CD45, CXCR4, and CSF1R (Figure 1D). Notably, there were significantly more CD45⁺ GFP⁺ cells in the bone marrow than in the blood (Figure 1E). Further subset analysis revealed that GFP⁺ CD45⁺ CXCR4⁺ CSF1R⁺ cells were dominant, consisting of 87% of GFP⁺ CD45⁺ cells in the blood (Figure 1F) and 96% in the bone marrow (Figure 1G). The second most common subset was GFP⁺ CD45⁺ CXCR4⁻ CSF1R⁺ in both cases. In addition, blood-derived GFP⁺ cells formed a well-defined, distinct population of cells in the dot plot compared to the bone marrow. These findings were confirmed in a second mouse strain (B10D2), which revealed consistent findings (Figure S1).

Mesenchymal stem/stromal cells express macrophage markers upon stimulation with defined factors

CSF1R plays a crucial role in macrophage development, maintenance, and function. It supports survival, proliferation, self-renewal, and M2-like polarization.^40,41 The unusual presence of CD45⁺ CSF1R⁺ mMSCs derived cells in vivo raised the question of whether freshly isolated mMSCs also express CSF1R. To further investigate the role of CSF1R, fresh PαS cells were cultured in medium containing M-CSF, also known as CSF1, its corresponding ligand.⁴²

After isolation, the cells were plated on round cover glasses in a 12-well plate. The cells were cultured for 7 days in four different media: Dulbecco modified Eagle’s medium (DMEM), Roswell Park Memorial Institute medium (RPMI), RPMI media supplemented with M-CSF (RM), M-CSF and IL-10 (RMI), and GM-CSF (RGM) (Figure 2A) (Table 2). These supplemented RPMI media are commonly used for bone marrow macrophage differentiation.^43,44,45,46 First, to assess CD45 expression, cells were stained with monoclonal and polyclonal antibodies against CD45 and the macrophage marker CD68. After 7 days, no CD45 expression was observed in DMEM (Figure 2B). Similarly, cells in RPMI medium did not express CD45 or CD68 (Figure 2C). However, in RM and RMI media, a subset of MSC (PαS) expressed CD45, of which most cells co-expressed CD68 (Figures 2D and 2E). Moreover, their cell morphology was clearly distinct from that of initial PαS cells, often with pseudopodia and sometimes with an elongated, spindle, or stellate morphology. Regularly, these cells formed uniform colonies composed solely of CD45⁺ cells; however, they also grew in mixed colonies alongside CD45⁻ cells (Figures 2G and 2H). When cultured in RGM, there was no change in morphology observed, and no investigated markers were expressed (Figure 2F).

Figure 2.

Induction of CD45-positive cells from PαS cells under different culture conditions

(A) Visual summary of the methodology of in the in vitro cell culture. PαS cells from the bone Marrow were isolated and cultured in different media: DMEM, RPMI, RM (RPMI supplemented with M-CSF), RMI (RM supplemented with M-CSF and IL-10), and RGM (RM supplemented with GM-CSF).

(B–F) Immunofluorescence staining of PaS-derived cells after culture in each condition. PαS cells after 7 days of culture in the media described in (A). Cells were stained with a mono- (green) and polyclonal (red) CD45 antibody, as well as CD68 (grey). Scale bars, 200 μm.

(G and H) Representative high-magnification images show the distinct morphologies of PaS-derived macrophage-like cells in RM (G) and RMI (H) conditions. Scale bars, 200 μm.

Table 2.

Media composition for cell culture

Base medium	Abbreviation	FBS	P/S	GlutaMAX (Gibco)	M-CSF (BioLegend)	IL-10 (BioLegend)	GM-CSF (BioLegend)
DMEM	DMEM	10%	1%	–	–	–	–
RPMI	RPMI	10%	1%	2 mM	–	–	–
RPMI	RM	10%	1%	2 mM	50 ng/mL	–	–
RPMI	RMI	10%	1%	2 mM	50 ng/mL	25 ng/mL	–
RPMI	RGM	10%	1%	2 mM	–	–	50 ng/mL

Next, PαS cells were stained for CD206 to assess macrophage polarization toward the M2 phenotype (Figure 3). Most (approximately 99%) cells that expressed CD45 and CD68 were also positive for CD206. To confirm these results, CD73⁺ mMSC were isolated and cultured in the same manner. The cells changed their phenotype identically as seen in PαS (Figure S2A). These results were replicated with CD73⁺mMSCs, where they expressed CD45 and CD68 upon exposure to M-CSF (RM) or M-CSF and IL-10 (RMI) (Figure S2B). In line with PαS cells, the morphology was different compared to CD45⁻ cells, and the CD45⁺ CD68⁺ population also expressed CD206. As a reference for macrophage morphology, CD45⁺ CD68⁺ cells were isolated from the bone marrow and cultured in DMEM and RPMI (Figure S2C). In both media, a morphology similar to that of differentiation medium-cultured PαS and CD73⁺mMSCs was observed.

Figure 3.

Characterization of macrophage marker expression in PαS-derived cells

Immunofluorescence staining of PαS-derived cells cultured in different media: DMEM, RPMI, RM, and RMI. Cells were stained for the macrophage markers CD206 (green), CD45 (red), and CD68 (gray). Nuclei were counterstained with DAPI (blue). Scale bars, 200 μm.

To test if hematopoietic stem/progenitor cells (HSCs) can grow in these media, HSCs were isolated and cultured using the same protocol. However, after 1 week, no cells were observed (Figure s3A). These results suggest the presence of a subpopulation of MSCs that can differentiate into macrophages in response to stimulation with M-CSF and IL-10. To validate the purity of the isolated PαS, TaqMan-based gene expression analysis and colony-forming unit (CFU) assays were performed (Figures 1A and S3B). No CD45 gene expression was found in the isolated PαS. While they were able to form CD45⁺ colonies when exposed to M-CSF, they did not expand in MethoCult medium, in contrast to control bone marrow cells. Finally, PαS cells were isolated, cultured for 2 passages, frozen, replated in RM, and cultured for one week. In contrast to freshly isolated PαS, these cells did not express any of the critical markers, CD45, CD68, or CD206. Additionally, they also did not express CSF1R (Figure S3C).

Single-cell RNA-Seq supports the transition of PαS cells into macrophage and fibrocyte-like lineages

To further investigate these results, single-cell RNA-Seq (scRNA-seq) was performed with freshly isolated PαS cells, which were cultured for 1 week in DMEM or RM. First, the t-distributed stochastic neighbor embedding was investigated for the positive isolation markers PDGFRα (denoted as PDGFRα) and Ly6a (Sca-1) Ptprc, Cd45 (PTPRC) (Figure 4A). Although both samples displayed a similar distribution, only the RM-cultured sample clearly contained Cd45-expressing cells.

Figure 4.

Single-cell transcriptomic analysis of PαS-derived cells cultured with M-CSF

(A) tSNE plots showing representative gene expression patterns of PαS cells cultured in DMEM or M-CSF-containing medium. Expression of PDGFRα, LY6A (Sca-1), and PTPRC (CD45) is shown. Arrows indicate CD45-positive populations.

(B) Integrated tSNE of PαS cells cultured in DMEM and M-CSF conditions, identifying ten distinct clusters (clusters 1–10). Notably, Cluster 6a has been manually separated from Cluster 6.

(C) Bubble heatmap of all clusters from (B) with the expression of macrophage relevant markers, as well as MSC markers (PDGFRα and Sca-1).

(D) Expression of classical macrophage markers CD45, CD68, ITGAM (CD11b), CSF1R (CD115), CD34, and COL1A1 (collagen 1).

(E) Classification of clusters based on marker gene signatures.

(F) Expression of macrophage-relevant genes based on (E) classification displayed as bubble heatmap.

Next, the cells were clustered into 10 clusters using the Louvain Clustering of BBrowser (Figure 4B). However, owing to the distinct location and different levels of key marker expression, cluster 6 was manually subdivided into clusters 6 and cluster 6a (Figures 4B and 4C). The isolation markers Sca-1 (denoted as LY6A), PDGFRα (PDGFRA), Cd73 (NT5E), Cd45 (PTPRC), as well as Cd68 (CD68), Cd11b (ITGAM), Csf1r (CSF1R), Cd34 (CD34), and Collagen 1 (COL1A1) were blotted in a bubble heatmap (Figure 4C). Clusters 4, 5, and 6a-10 all highly expressed Sca-1 and PDGFRα, whereas clusters 1, 2, and 3 expressed low levels of Sca-1 and PDGFRα, while cluster 6 did not express PDGFRα and expressed low levels of Sca-1. Cd73 was only notably expressed in clusters 3 and 4. Cluster 3 was the only cluster with high Cd34 expression. While devoid in the other clusters, clusters 6 and 6a both expressed high levels of Csf1r, Cd11b, Cd68, and Cd45. Notably, Collagen 1 expression was high in all clusters except for cluster 6, which was approximately half of that in the others (Figure 4C).

Based on these markers and co-expression within clusters, cells were divided into MSCs (Figure 4E, blue, clusters 5, 7–10) based on Sca-1 and PDGFRα expression; adipocyte progenitors (yellow, cluster 4) based on Cd73 expression; fibroblast progenitors (green, cluster 3) based on Cd73 and Cd34 expression; macrophages (red, cluster 6) based on Cd45, Cd68, Cd11b, and Csf1 expression; and fibrocyte-like macrophages (violet, cluster 6a) based on the low expression of macrophage markers and high expression of collagen I. Both clusters (macrophages and fibrocyte-like macrophages) clearly express macrophage-relevant genes such as Cd14 (CD14), Cd48 (CD48), Cd53 (CD53), Stab1 (STAB1), Ccl9 (CCL9), and Cd206 (MRC1), and lose the characteristics of MSCs (Figure 4F). These data strongly suggest that freshly isolated PαS cells contain heterogeneous subpopulations, including macrophages and fibrocyte-like macrophages, that emerge in response to M-CSF stimulation.

M-CSF/CSF1R pathway induces the differentiation of mesenchymal stem/stromal cells into macrophages

Given the critical role of CSF1R in macrophage biology, it is likely to exert similar functions in mMSCs. To investigate the potential role of CSF1R activation, the receptor was blocked using various approaches to determine whether the differentiation is mediated through CSF1R activation. For this purpose, freshly isolated PαS cells were cultured identically as described above, with the addition of a CSF1R blocking antibody, and/or small molecule inhibitor GW2580 (Figure 5A). In all three conditions - antibody treatment, small molecule inhibition, and their combination, no macrophage transition was observed, in contrast to the inhibitor-free M-CSF-containing medium (Figure 5A).

Figure 5.

Functional validation of M-CSF-induced macrophage differentiation from PαS and human MSCs

(A) Visual summary for blocking M-CSF-induced macrophage differentiation. PαS cells isolated from the bone marrow and cultured in M-CSF-containing medium with or without CSF1R-blocking antibody or small molecule inhibitor (GW2580). Representative immunofluorescence images show CD206 (green), CD45 (red), and CD68 (gray) expression. Scale bars, 200 μm.

(B) Experiment process for M-CSF injection. M-CSF was injected into the tail vein 7 days and again 3 days before collecting PαS cells, and the PαS cell yield was analyzed by flow cytometry.

(C) CD45⁻ TER119-cells after M-CSF tail vein injection (PBS, n = 5, MCSF, n = 5) (ns, not significant, Student’s t tests).

(D) PαS cells in the bone in response to M-CSF injection (PBS, n = 5, MCSF, n = 5) (∗p < 0.05, Student’s t tests).

(E) CSF1R expressing PαS cells in response to M-CSF injection (PBS, n = 5, MCSF, n = 5) (ns, not significant, Student’s t tests).

(F) Experimental setup for human cell analysis. CD90⁺ CD271⁺ human MSCs (huMSCs) and human bone marrow cells (huBMCs) were cultured under DMEM or M-CSF-containing RPMI (RM) conditions.

(G) Expression of CSF1R in freshly isolated huMSCs from the BM (CD271⁺ CD90⁺ cells).

(H) Immunofluorescence staining of huBMCs and huMSCs after culture in DMEM or RM. Cells were isolated from the bone marrow and stained for CD206 (green), CD45 (red), and CD68 (gray). All results were confirmed in three independent experiments. Scale bars, 200 μm.

As CSF1R stimulation has been shown to influence macrophage proliferation, it may also influence CSF1R⁺ MSC proliferation.⁴⁷ To further investigate the potential of M-CSF, 10 μg M-CSF was injected 7 days and again 3 days before collecting PαS cells (Figure 5B). Flow cytometry analysis showed a clear trend of decreased CD45⁻ population in the M-CSF injected group (p = 0.057) (Figure 5C), with a significant decrease of the PαS cells within the CD45⁻ population (Figure 5D). This is accompanied by an increase of CSF1R⁺ PαS cells (p = 0.12) (Figure 5E). Finally, the reproducibility in human cells was investigated. Cryopreserved human whole bone marrow was thawed, and MSCs were isolated as CD271⁺ CD90⁺ cells (hMSCs)⁴⁸ and cultured under the same conditions as mMSCs (Figure 5F). After 1 week, hMSCs demonstrated similar morphology and marker expression as PαS cells, where a subset of cells expressed CD45, CD68, and CD206 (Figure 5G). In addition, around 20–25% of freshly isolated hMSCs showed the expression of CSF1R (Figure 5H). These data suggest that MSC-to-macrophage differentiation is driven by CSF1R activation through M-CSF, and that M-CSF influences both differentiation and the expansion of CSF1R⁺ MSCs in both mice and humans.

Discussion

In this study, we determined the relationships between MSCs, fibrocytes, and macrophages. First, GFP⁺ PαS mice from transgenic C57BL/6 mice were isolated and transplanted to irradiated C57BL/6 recipient mice. After transplantation, GFP⁺ cells became not only positive for CXCR4 and CSF1R, but also expressed CD45, a marker that is exclusively part of the hematopoietic lineage. This result was reproducible with CD73⁺mMSCs using both, B10.D2 and C57BL/6 mice. Notably, GFP⁺ CD45⁺ CXCR4⁺ CSF1R⁺ cells were found in the blood and bone marrow, with the second most common type being GFP⁺ CD45⁺ CXCR4⁻ CSF1R⁺, both expressing the macrophage marker CSF1R. However, a strong preference for bone marrow was observed, suggesting that the bone marrow harbored the transplanted cells. The more distinct population detected in the blood by FACS suggests that cells enter circulation after maturation in the bone marrow. These results suggest that fresh mMSCs home to the bone marrow and differentiate into macrophage-like cells in vivo. As CSF1R is an essential receptor for macrophage biology, it may be so for mMSCs as well, and indeed a small fraction of freshly isolated PαS cells naturally expressed CSF1R, indicating that this receptor might be responsible for macrophage differentiation.

To further corroborate these findings in vitro, mMSCs were cultured in two different macrophage differentiation media consisting of either M-CSF, the ligand for CSF1R, with the optional supplement IL-10, or GM-CSF, the counterpart to M-CSF, essential for M1 differentiation.⁴⁹ While DMEM or RPMI medium did not show changes in mMSCs, RPMI with M-CSF, as well as RPMI with M-CSF and IL-10 supplementation, changed the phenotype of the cultured mMSCs. In addition, these cells were CD45⁺ CD68⁺ CD206⁺, with the latter being a common marker of M2 macrophages. However, when cultured in GM-CSF, no transition was observed in PαS cells. In addition, cultured PαS did not show any expression of CD45, CD68, CD206, or CSF1R, indicating that the cells lose their ability to express these markers during culture, likely through the loss of CSF1R or loss of the subpopulation that expresses this marker.

As a reference, CD45⁺ CD68⁺ macrophages were isolated from the bone marrow and cultured in DMEM and RPMI medium. The morphology was nearly identical to that of PαS, and CD45⁺ CD68⁺ cells had pseudopodia. To explore the possible contamination by mHSPCs, Lin⁺, c-Kit ⁺, and Sca1⁺ cells were isolated and cultured under identical conditions as PαS cells. No cells were observed growing in any media. In contrast to MSCs, HSPCs do not adhere to glass and were most likely washed out.^50,51 In vivo, host macrophages might become GFP⁺ by phagocytosing transplanted GFP⁺ PαS cells. The in vitro experiment demonstrated that the in vivo findings of MSCs becoming macrophage-like cells were reproducible, and since the seeded cells consisted solely of mMSCs it is unlikely that in vivo findings are credited to phagocytosis by host macrophages.

In M-CSF-supplemented culture conditions, the number of CD45⁺ cells varied from a few single cells to large colonies where CD45⁺ cells coexisted next to CD45⁻ cells. Quantification is difficult when combined with varying amounts of plated MSCs. However, scRNA-seq data suggested that the percentage of CD45⁺ cells was approximately 5.85%. scRNA-seq was performed to confirm these results. A small population of CD45⁺ and CD68⁺ cells was identified, which was also positive for CD11b, CD206, and CD48, commonly expressed on various hematopoietic cell surfaces. Therefore, these cells were identified as macrophages. Notably, another small population exhibited a profile similar to that of macrophages; however, the expression levels of the aforementioned key markers were generally lower. Instead, the isolation marker PDGFRα, especially Sca-1 and collagen 1, was strongly expressed, indicating that these cells might be in between the MSC and macrophage phenotype and resemble fibrocytes.

MSCs naturally share several markers with fibrocytes, such as CXCR4 and SMA,^52,53 and M2 macrophages can produce collagen (adopt a fibrocyte feature) within atherosclerotic plaques,⁵⁴ further highlighting the similarities between these cells. The expression of monocyte markers, such as CD11b, CD48, and CD68, signifies a macrophage phenotype.^55,56 CD206 is a key marker for M2 macrophages, known for anti-inflammatory functions through TGF-β and IL-10, similar to MSCs.^28,57 M2 macrophages contribute to wound healing in a manner similar to that of MSCs and recruit fibroblasts, a cell type “phenotypically indistinguishable from MSCs.”^58,59 In contrast, MSCs can induce an M2-like phenotype in macrophage co-cultures, supporting a close relationship between these cells.⁶⁰ This study also revealed a notable decrease in PDGFRα expression, accompanied by an increase in macrophage markers, among the populations. While the exact role of PDGFRα in macrophage function and behavior remains underexplored, its expression may not be significant in macrophages.⁶¹

Previous reports showed that CD49⁺ MSCs with CFU-F capability exhibited a CD34⁺ CD45^med/low phenotype after isolation, which was substantially downregulated during culture.⁵² Another recent study found “distinct CD45⁻ B-lymphoid and erythroid progenitor populations, whose activity was enhanced by BM stromal cells” within CD45⁻ CD31⁻ Ter119⁻ cells. Notably, these triple-negative cells replaced over 95% of CD45⁺ cells in lethally irradiated mice 4 months after transplantation.⁶² MSCs are traditionally seen as invisible immunoregulators; however, they can occur in the immune system through MHC II upregulation, suggesting a more direct role.⁶³ These studies have shown that macrophages, fibrocytes, and MSCs are closely associated. However, increasing evidence suggests that this relationship is not merely collaborative.

The data presented in this study showed that mMSCs can change phenotype from CD45⁻ to CD45⁺ in vivo in multiple strains as well as in vitro. Furthermore, this study showed that CSF1R is not only important for macrophages but is also the driving factor for the differentiation of mMSCs, as blocking the receptor with antibody and/or small molecules resulted in a halt of the differentiation. This highlights the essential role of the M-CSF-CSF1R interaction in this process. Additionally, M-CSF injection decreased the percentage of CD45⁻ cells (p = 0.057), decreased the percentage of PαS cells within the CD45⁻population significantly and also increased the percentage of CSF1R⁺ PαS cells (p = 0.125) in an in vivo model. Other studies showed that M-CSF injection increases macrophages in vivo.⁴⁷ Taken together, these findings indicate that M-CSF injection might cause an increase in CSF1R⁺ PαS cells, which then shift to CD45⁺. However, this topic exceeds the focus of this study and needs to be thoroughly explored in the future.

Finally, this study showed that CD271⁺ CD90⁺ hMSCs showed the same reaction to M-CSF as demonstrated by PαS cells and CD73⁺mMSCs, indicating that this behavior is generally applicable, as demonstrated in multiple strains and species. However, the number of isolated cells from human bone marrow was much lower compared to the murine model, which may explain the scarcity and lack of large CD45⁺ cell populations. Further refinement of culture protocol may enhance the size of CD45⁺ cell populations. In addition, we previously reported how MSCs in the bone marrow may contribute to the pathogenesis of graft-vs-host disease (GVHD) in patients following bone marrow transplantation.²³ We postulated that non-professional antigen-presenting cells (APCs) may have been involved, but our current data indicate that macrophages derived from donor MSCs may play a role. The translational implications of our findings need to be investigated in the future.

While it can be argued that MSC populations may have been contaminated with hematopoietic cells that subsequently became CD45⁺ cells, contamination was unlikely since the data were consistently replicated (N = 8), with no CD45 expression in the unsupplemented medium. Furthermore, HSCs generally do not attach to uncoated glass, a feature commonly utilized to isolate MSCs from bone marrow. In addition, qPCR showed no CD45 expression in the cells immediately after isolation. While Methocult only affect colony forming progenitors, not HSC directly, the combination of RNA data and culture control data in combination underlines the purity of the isolated cells.

To date, no study has investigated whether MSCs change their phenotype before performing their immunomodulatory roles. This study demonstrated that, in addition to CD45, macrophage markers CD11b and CD68, and to some degree fibrocyte markers such as CXCR4 and collagen I, were co-expressed in mMSCs both in vivo and in vitro, which was confirmed by single-cell RNA-seq. While it is established that MSCs can become fibroblasts, this study suggests that freshly isolated MSCs can develop a fibrocyte-like and macrophage-like phenotype. Currently, it is unclear whether there are two steps on the same path or two individual paths. This macrophage-like state is more specific to anti-inflammatory CD206⁺ M2 macrophages, which share many immunomodulatory features through the same mechanisms as MSCs, as demonstrated previously in numerous studies.

In conclusion, this study revealed that freshly isolated bone marrow MSCs are more representative of in vivo MSCs than conventionally cultured MSCs. We believe that our data convincingly show that MSCs have the potential to transition to what has hitherto been considered an HSC line. This highlights additional MSC potential and indicates the need to re-evaluate the current understanding of MSCs.

Limitations of the study

One of the biggest technical challenges was the scalability and conductor-dependent yield of MSC isolation. Processing the bone marrow of 10 mice takes approximately 9–11 h per experiment, and increasing the number of mice was not feasible with the current sorting equipment. Although female mice were used in this study, male mice were included as a source of donor cells due to limits in the availability of inbred mice. This is unlikely to have influenced the overall outcome. The yield MSCs varies depending on various factors, making comparisons of experimental data that require a consistent yield problematic. While this study identified CSF1R as a key player for the transition to CD45⁺ mMSCs, we were not able to establish whether the transition from MSCs to macrophage-like or fibrocyte-like cells occurs in a single step or through multiple intermediate stages. Further analysis will require a method to scale up the cells, as CSF1R⁺ PαS cells represent only 0.5% of a population that is only 0.1% of the whole bone marrow. Finally, although in vitro experiments were designed to corroborate in vivo findings, the artificial culture conditions may not fully recapitulate the complexity of the in vivo environment. Future studies will require additional genetic lineage tracing experiments using Cre/loxP mice.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Shigeto Shimmura (shigeto.shimmura@fujita-hu.ac.jp).

Materials availability

No new reagents were generated in this study.

Data and code availability

• •Data: The RNASeq data are available at the following link: https://www.ebi.ac.uk/biostudies/ArrayExpress/studies/E-MTAB-16511?key=486e1e84-1d42-43c2-8aab-b95f704aff1c (BioStudies: E-MTAB-16511).
• •Code: This article does not report original code.
• •Other: This article does not report any additional resources.

Acknowledgments

This work was supported by a grant from the Japan Society for the Promotion of Science (KAKENHI) 23K24504. We thank Hiroko Taniguchi for her technical assistance in the laboratory. We also acknowledge Mari Fujiwara at Keio University School of Medicine, Central Equipment Management Division, for her assistance with FACS cell isolation and for providing access to the FACS sorting machine. This work was partially supported by JST, CREST, Japan (Grant Number JPMJCR2124), and the Japan Agency for Medical Research and Development (AMED) (Grant Number 23bm1223011h0001). Illustrations used in Figures 1, 2, 5, and Graphical Abstract were created using BioRender.

Author contributions

Conceptualization: R.M.R., Y.M., Y.O., and S.S. Methodology: R.M.R., Y.M., and Y.O. Validation: R.M.R. Formal analysis: R.M.R. Investigation: R.M.R., Y.M., and Y.O. Data curation: R.M.R. Writing – original draft: R.M.R. Writing – review and editing: R.M.R., Y.M., and S.S. Visualization: R.M.R. Resources: Y.M., S.M., and S.S. Supervision: Y.O. and S.S. Project administration: R.M.R., Y.M., and S.S. Funding acquisition: S.S.

Declaration of interests

The authors do not have competing interests.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the author(s) used ChatGPT in order to optimize grammar and formulation. After using this tool or service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Experimental models: organisms/strains
C57BL/6NCrSlc Mouse	Sankyo Labo Service Corporation	N/A
C57BL/6-Tg(CAG-EGFP) Mouse	Sankyo Labo Service Corporation	N/A
B10.D2/nSnSlc	Sankyo Labo Service Corporation	N/A
Antibodies
PE anti-mouse CD45 Antibody	Biolegend	Cat#103106; RRID: AB_312971
APC anti-mouse CD45 Antibody	Biolegend	Cat#103112; RRID: AB_312977
Goat anti-mouse CD45	R&D Systems	Cat# AF114; RRID: AB_442146
Goat anti-GFPuv Antibody	R&D Systems	Cat# AF4240; RRID: AB_884445
APC anti-mouse CD68 Antibody	Biolegend	Cat# 137007; RRID: AB_10575299
APC anti-mouse CXCR4 Antibody (flow cytometry)	Biolegend	Cat# 146508; RRID: AB_2562785
Rabbit anti-CXCR4 Antibody (IHC)	Abcam	Cat# ab124824; RRID: AB_10975635
Rat anti-mouse CD115 Antibody	Thermo Fisher	Cat# 14-1152-82; RRID: AB_467428
Alexa Fluor® 555 Donkey anti-Goat IgG	Invitrogen	Cat# A-21432; RRID: AB_141788
Alexa Fluor® 647 Goat Anti-Rabbit IgG	Abcam	Cat# ab150079; RRID: AB_2714032
PE/Cyanine7 anti-mouse CD45	Biolegend	Cat# 103114; RRID: AB_312979
PE/Cyanine7 anti-mouse CD31	Biolegend	Cat# 102524; RRID: AB_2572182
PE/Cyanine7 anti-mouse TER-119/Erythroid Cells	Biolegend	Cat# 116222; RRID: AB_2281408
PE anti-mouse Ly-6A/E (Sca-1)	Biolegend	Cat# 108108; RRID: AB_313345
APC anti-mouse CD140a	Biolegend	Cat# 135908; RRID: AB_2043970
APC anti-mouse CD73	Biolegend	Cat# 127210; RRID: AB_11218786
PE anti-human CD45	Biolegend	Cat# 304008; RRID: AB_314396
Alexa Fluor(R) 488 anti-human CD206 (MMR)	Biolegend	Cat# 321113; RRID: AB_571874
APC anti-human CD68	Biolegend	Cat# 333809; RRID: AB_10567107
Ultra-LEAF Purified anti-mouse CD115	Biolegend	Cat# 135541; RRID: AB_2832485
Biological samples
Human Bone Marrow Mononuclear Cells	LONZA	2M-125C
Chemicals, peptides, and recombinant proteins
DNase	Qiagen	Cat# 79254
Collagenase	Wako	Cat# 032-22364, CAS RN® : 9001-12-1
TaqMan™ Fast Advanced Master Mix	Thermo Fisher	Cat# 4444557
Recombinant Mouse M-CSF	Biolegend	Cat# 576406
Recombinant Human M-CSF	Biolegend	Cat# 574804
Recombinant Mouse GM-CSF	Biolegend	Cat# 576304
Recombinant IL-10	Biolegend	Cat# 575804
Glutamax 100	Gibco	Cat# 35050061
Tissue-Tek® O.C.T. Compound	Sakura Finetek Japan	Cat# 4583
Formalin	Nacalai	Cat# 09154-85
fetal bovine serum	JRH Biosciences	Cat# 12107C
Normal Goat Serum	Thermo Fisher	Cat# 50197Z
Normal Donkey Serum	Sigma Aldrich	Cat# S30-100ML
Triton X-100	Nacalai	Cat# 35501-02
Water deionized & sterilized	Nacalai	Cat# 06442-95
HBSS	Nacalai	Cat# 17461-05
DMEM/Ham’s F-12	Nacalai	Cat# 08460-95
RPMI	Nacalai	Cat# 30264-85
GW2580	Selleck Chemicals LLC	Cat# S8042
Critical commercial assays
NucleoSpin® RNA XS	Takara	Cat# 740902.50
MethoCult™ GF M3434	Stemcell Technologies	Cat# #03444
Deposited data
Single-cell RNA sequence data	This paper	E-MTAB-16511; https://www.ebi.ac.uk/biostudies/ArrayExpress/studies/E-MTAB-16511?key=486e1e84-1d42-43c2-8aab-b95f704aff1c
Oligonucleotides
TaqMan™ Assay FAM-MGB PTPRC (CD45)	Thermo Fisher	Assay ID: Mm01293577_m1
TaqMan™ Gene Expression Assay, VIC primer-limited	Thermo Fisher	Assay ID: Mm99999915_g1
Software and algorithms
BBrowser	Bioturing	https://bioturing.com/
Kaluza	Beckman Coulter	Version 2.3
FIJI/ImageJ	N/A	Version 1.54p
Graphpad Prism	GraphPad	Version 10.5.0

Experimental model and study participant details

Mice

The animals (C57BL/6NCrSlc, C57BL/6-Tg (CAG-EGFP), B10.D2/nSnSlc) were procured from Sankyo Labo Service Corporation (Tokyo, Japan). B10.D2/nSnSlc were bred with C57BL/6-Tg(CAG-EGFP) mice and inbred for over 10 generations to create B10.D2 EGFP mice. All mice were 8 weeks old at the time of bone marrow acquisition. Mice obtained through Sankyo Labo Service Corporation were female, inbred transgenic EGFP mice used for donors were mixed sex due to limits of availability. Animal studies were conducted in compliance with protocols approved by the Institutional Animal Care and Use Committee of Keio University (#A2022-178 and #A2022-122). These procedures strictly adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all experimental methods were performed in accordance with Keio University’s Institutional Guidelines on Animal Experimentation.

MSC isolation

Mouse bone marrow was obtained as previously described with minor modifications.²³ Briefly, tibia, femur, and coxal bones were extracted and washed thrice with 10 mL HBSS (Nacalai Tesque, Kyoto, Japan) containing 10 mM HEPES (Nacalai Tesque, Kyoto, Japan), 2% fetal bovine serum (JRH Biosciences, USA), and 1% penicillin/streptomycin (Nacalai Tesque, Kyoto, Japan). Next, the bones were crushed and cut using scissors to obtain a paste-like consistency. The bones of 5–7 animals (roughly divided in half) were incubated with 20 mL 0.2% (g/v) collagenase (Wako, Osaka, Japan) containing DMEM/Ham’s F-12 (Nacalai Tesque, Kyoto, Japan) in a MACSmix™ Tube Rotator (Miltenyi Biotech, Bergisch Gladbach, Germany) at 37 °C for 1 h, with the addition of 10 μL DNase (Qiagen, Venlo, Netherlands) 15 min before the end (after 45 min).

After collecting the medium and adding the supernatant from washing the bones with HBSS used above, the cells were centrifuged at 300 ×g for 5 min at room temperature, and blood cells were lysed using 1 mL water per tube (Nacalai, Kyoto, Japan). Bone marrow cells were then transferred to a 1.5 mL Eppendorf tube for antibody staining. Per 1x10⁷ cells, 1 μL of [PE-Sca1 (clone: D7), PE-Cy7-CD45 (clone: 30-F11), PE-Cy7-TER119 (clone: TER-119), and 2 μL of APC-PDGFRα (clone: APA5)] or [PE-Cy7-CD31 (clone: MEC13.3), PE-Cy7-CD45 (clone: 30-F11), PE-Cy7-TER119 (clone: TER-119), and APC-CD73 (clone: TY/11.8)] (all Biolegend, San-Diego, CA, USA) were added to cells in 100 μL HBSS wash buffer. After 30 min on ice in the dark, cells were washed and isolated using a BC MoFlo XDP (Beckman Coulter, Brea, CA, USA). MSCs were isolated with two established protocols: as CD45^-, TER119^-, PGFR⁺, Sca-1⁺ (PαS) or CD31^-, CD45^-, TER119^-, and CD73⁺ (CD73⁺mMSC).

Purity verification with TaqMan

To validate the purity, cells were isolated as described above for PαS cells, and RNA was isolated using NucleoSpin® RNA XS (Takara, Shiga, Japan) according to the manufacturer’s protocol. Next, RNA was analyzed using TaqMan™ Fast Advanced Master Mix for qPCR according to the manufacturer’s protocol. The following primers were used: TaqMan™ Assay FAM-MGB PTPRC (CD45) (20X) (Assay ID: Mm01293577_m1) (Thermo Fisher, Waltham, MA, USA) and TaqMan™ Gene Expression Assay, VIC primer-limited (Assay ID: Mm99999915_g1). CD45⁺ cells from the bone marrow, cultured MSC, and water (no RNA) were used as references and controls.

Bone marrow transplantation

Recipient mice (8-week-old) were irradiated at 7 Gy. Subsequently, 2 × 10⁴ GFP⁺ PαS cells isolated from GFP transgenic mice together with 1×10⁵ GFP^-, CD45⁺, and TER119⁺ cells from wild-type mice were transplanted through the tail vein in 200 μL DMEM F12 (Nacalai, Kyoto, Japan).

Flow cytometry

Blood and bone marrow samples were collected on the day of the acquisition. Before staining, the red blood cells were lysed using water (Nacalai Tesque, Kyoto, Japan). Bone marrow cells were cultured overnight on 6 cm FNC-coated (Athena Enzyme Systems, Baltimore, MD, USA) plastic dishes in DMEM/Ham’s F-12 (Nacalai, Kyoto, Japan). The following day, the supernatant was aspirated, and only the adherent cells were stained. Initially, cells were washed with FACS buffer (R&D Systems, Minneapolis, MN, USA) thrice and centrifuged at 300 ×g and 4 °C for 5 min. Subsequently, the cells were stained with primary antibodies against APC-CXCR4 (1276F12), PE-Cy7-CD45 (30-F11), and BV421-CD115 (AFS99) (BioLegend, San Diego, CA, USA) for 30 min on ice. The cells were washed with the staining buffer and analyzed using BC CytoFLEX S (Beckman Coulter, Brea, CA, USA). A gating strategy was developed using Fluorescence Minus One (FMO) controls and wild-type cells for GFP^- signaling. If the event count for GFP⁺ cells was <50 for blood or <100 for bone marrow, the data were excluded.

Cell culture

Round cover glasses (18 mm; Matsunami, Osaka, Japan) were placed in a 12-well plate. MSCs from wild-type mice were isolated as previously described. Next, 1.0–1.5 × 10⁴ MSCs were seeded in individual wells, with each well receiving 1 mL of the medium listed in Table 2. The base media used were DMEM/Ham’s F-12 (Nacalai, Kyoto, Japan) or RPMI 1640 (Nacalai, Kyoto, Japan). After plating, the cells were cultured undisturbed for 2 days. From day 3, 500 μL were exchanged for fresh medium daily until day 7. On the last day, cells were fixed with 4% paraformaldehyde, as described above. For blocking experiments either 5 μL/mL of Ultra-LEAF Purified anti-mouse CD115 (CSF-1R) blocking antibody (clone: AFS98, Biolegend, San-Diego, CA, USA), 5 μg/mL GW2580 (Selleck Chemicals LLC, Cologne, Germnay) or a combination of both.

As control PαS were isolated, cultured for 2 passages, frozen according to standard protocol, replated and exposed to MCSF containing medium (RM) for one week. Then, cells were analyzed with SonyMA800 for their expression of PE-CD45 (30-F11), APC-CD68 (FA-11), Alexa488-CD206, and BV421-CD115 (AFS99).

HSPC isolation and culture

The femur and tibia were harvested and crushed using a pestle, as described above. Next, the whole bone marrow was stained with PECy7-Lineage [CD3ε (clone: 145-2C11), CD45R (clone: RA3-6B2), Gr-1 (clone: RB6-8C5), CD11b (clone: M1/70), Ter119 (clone: TER-119)], APC-cKit (clone: 2B8), FITC-Sca1 (clone: D7), and PE-CD45 (clone 30-F11) (all Biolegend, San-Diego, CA, USA). Cells were stained as described above and sorted into Lin⁺, c-Kit ⁺, and Sca1⁺ cells. After isolation, the cells were directly plated in a 12-well plate with 18 mm round cover glasses (Matsunami, Osaka, Japan) in DMEM, RPMI, RM, RMI, and RGM.

CFU test (CFU-F, CFU-GM, CFU-G and CFU-M)

To further test the purity of the isolated PαS, 5000 freshly isolated cells were cultured in MethoCult™ GF M3434 (Stemcell Technologies, Vancouver, Canada) for one week. As control, 5000 bone marrow cells were cultured in the same medium, and 5000 PαS were cultured in RPMI media supplemented with M-CSF (RM) (see Table 2) on the same 6 well plate.

Human MSC culture

BONE MARROW MNC (POIETICS, Lonza, Basel, Switzerland) (Human male bone marrow cells, 2.5x10⁷ cells/vial) were suspended in ice-cold HBSS at 1–5×10⁷ cells/mL and stained for 30 min on ice with a monoclonal antibody. Three vials (7.5x10⁷ cells) were divided into control and study groups. The following antibodies were used: LNGFR (CD271)-PE (clone ME20.4-1. H4; Miltenyi Biotech, Bergisch Gladbach, Germany), and THY-1(CD90)-APC (clone 5E10; BioLegend, San Diego, CA, USA).⁴⁸ Propidium iodide (PI: 2 μg/mL) was used to eliminate dead cells during flow cytometric analysis. Flow cytometric analysis and sorting were performed using a BD FACSAria (BD Biosciences, Franklin Lakes, NJ, USA).

Method details

Immunohistochemistry

The tissues were fixed in a 10% buffered formalin solution for 3 h. Following fixation, samples were briefly rinsed and then immersed in OCT compound (Sakura Finetek Japan, Co., Ltd., Japan) before being frozen at -80 °C. Subsequently, frozen samples were sectioned using a cryostat. Cell culture samples were each washed thrice with PBS (Nacalai, Kyoto, Japan) for 5 min and fixated with 4% formalin (Nacalai, Kyoto, Japan) for 15 min at room temperature. After three PBS washings, fixated cells were either stored in PBS at 4 °C or stained immediately.

Samples were blocked with Normal Goat Serum (Thermo Fisher, Waltham, MA, USA) or 10% Normal Donkey Serum (Sigma Aldrich, St. Louis, MO, USA) with 1% Triton X-100 (Nacalai, Kyoto, Japan) for 60 min at room temperature. GFP-containing samples were protected from light. Next, the samples were incubated overnight at 4 °C with the primary antibody (Table 1). The following day, the sections were washed three times with PBS and stained with Alexa Fluor 555- or 647-labeled secondary antibody for 30 min at room temperature. After washing three times, the cells were mounted with DAPI and anti-fading reagent-containing mounting medium (Abcam, Cambridge, UK). Images were obtained using a Leica LSM 980 microscope (Leica, Wetzlar, Germany).

Single-cell RNA-seq and data analysis

PαS were cultured in DMEM or RM for 1 week, as described above. After culture, cells were washed twice with PBS and dissociated into single-cell. Approximately 10,000 viable cells per sample were loaded onto the Chromium Controller (10x Genomics, Pleasanton, CA, USA) for single-cell encapsulation and barcoding. Single-cell RNA-seq libraries were prepared using the Chromium Single Cell 3′ Reagent Kits v3.1 (Dual Index, 10x Genomics) according to the manufacturer’s instructions (v3.1 Chemistry Dual Index, 10x Genomics, Pleasanton, CA, USA). Sequencing was performed on an Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA). Raw sequencing data were processed using Cell Ranger software. The filtered feature-barcode matrices were imported into BioTuring BBrowserX⁶⁴ for downstream analysis. Dimensionality reduction was performed using Principal Component Analysis (PCA), followed by k-nearest neighbor graph construction and Louvain clustering (resolution = 0.5). Visualization was conducted using t-distributed stochastic neighbor embedding (t-SNE) via Vinci software (BioTuring).

Quantification and statistical analysis

All statistical analyses were performed using GraphPad Prism 9. Statistical tests used throughout the study are indicated in the relevant figure legends and results sections. The primary statistical test used was the unpaired t-test. Displayed data represents means ± SEM. Significance was defined as follows: p < 0.05 (∗), p < 0.01 (∗∗), and p < 0.001 (∗∗∗). Each experiment was conducted independently at least three times. Randomization and stratification were not applicable due to the nature of the experimental design. In flow cytometry data was excluded if the event count for the target population was below 50 (blood) or below 100 (bone marrow).

Published: February 5, 2026