IGF-1 from bone marrow Adipoq-lineage cells stimulates endocortical bone formation in mature female mice

Abstract

Insulin-like growth factor 1 (IGF-1) is an anabolic signal promoting growth, differentiation and function of both embryonic and postnatal tissues. Both endocrine and paracrine functions of IGF-1 have been documented to regulate bone growth and BM hematopoiesis. Local production of IGF-1 from various cell types may contribute differently to the overall bioactivity of IGF-1 in bone, but relevant sources and mechanisms are yet to be fully elucidated. Here, we report that the Adipoq⁺ stromal cells are a notable source of IGF-1 in the BM of postnatal mice. Deletion of IGF-1 with Adipoq-Cre diminished endocortical bone formation and cortical bone mass in mature female mice. On the other hand, the trabecular bone parameters or hematopoietic properties were not affected in mutant mice of either sex. The study uncovers a local source of IGF-1 in the BM microenvironment that contributes to bone anabolic regulation in a site-specific manner.

Introduction

Insulin-like growth factor 1 (IGF-1), is among the most well characterized growth-promoting signals for both embryonic and postnatal tissues. In mouse KO studies, IGF-1 deletion caused severe intrauterine growth retardation and perinatal mortality.^1,2 Postnatally, IGF-1 functions both by mediating the function of growth hormone (GH) and through a GH-independent mechanism.³ IGF-1 is abundantly produced by the liver and secreted to the circulation, but also generated in local tissues like bone.⁴ In the serum, the vast majority of IGF-1 is bound to members of the IGF binding protein family (IGFBP 1-6), and the acid labile subunit (ALS).⁵ The association with ALS is believed to both prolong the half-life of IGFs and restrict its passage from the circulation to the extravascular compartment, thereby modulating its biological activities.⁶ Evidence indicates that both circulating and locally produced IGF-1 contribute to growth regulation. In support of the endocrine function of hepatic IGF-1, liver-specific IGF-1 ablation combined with deletion of ALS resulted in >85% reduction in the circulating IGF-1 level and significant growth retardation and bone loss in postnatal mice.⁷ On the other hand, liver-specific deletion of IGF-1 alone did not impair linear growth and only modestly reduced cortical bone growth even though the serum IGF-1 level was reduced by 75%.^8–10 The studies highlight a low threshold level of circulating IGF-1 necessary for supporting most of its endocrine function.

Genetic studies have identified important paracrine or autocrine functions for locally produced IGF-1 in promoting bone growth. Deletion of IGF-1 with either Col1a2-Cre in mesenchymal cells including osteoblasts, or Col2a1-Cre that targeted both chondrocytes and osteoblasts, suppressed skeletal growth and bone accrual without affecting serum IGF-1 levels.^11,12 Deletion of IGF-1 in late osteoblasts and osteocytes with DMP1-Cre impaired both the normal growth of cortical bone and the anabolic response to mechanical loading.^13,14 Conversely, IGF-1 overexpression from the osteocalcin promoter increased BMD and bone formation rate.¹⁵ Consistent with the ligand studies, deletion of the IGF-1 receptor (IGFR1) in osteoblasts with OC-Cre appeared to reduce bone formation by impeding mineralization, whereas deletion in preosteoblasts with Osx-Cre suppressed bone formation via impaired osteoblast differentiation.^16,17 More recently, IGF-1 deletion in a subset of BM stromal cells (BMSC) with LepR-Cre reduced both trabecular and cortical bone mass due to impaired bone formation.¹⁸ Collectively, the studies establish that locally produced IGF-1 serves as an important anabolic signal for bone growth. Furthermore, the various local sources from different cell types likely contribute differently to the overall IGF-1 bioactivity in bone.

Local IGF-1 levels have been implicated in the regulation of hematopoiesis that occurs in the BM microenvironment. A decline in BM IGF-1 levels in middle-aged mice (~12 mo) has been shown to initiate hematopoietic stem cell (HSC) aging and induce myeloid-biased hematopoiesis.¹⁹ However, as previous scRNA-seq studies have uncovered a high degree of heterogeneity among BMSCs, it is important to delineate the cellular sources of IGF-1 responsible for osteogenic or hematopoietic activities in normal or pathological conditions.^20,21

Here, we study the physiological relevance of IGF-1 derived from the Adipoq-lineage BMSCs. By deleting IGF-1 with Adipoq-Cre, we report cortical-specific bone loss without overt effects on hematopoietic stem and progenitor cells.

Materials and methods

Mice

All mouse work was approved by the Children’s Hospital of Philadelphia Animal Care and Use Committee (IACUC approval number IAC#24-001296). Mice were housed at 22 °C with a 12-h light cycle (6 am to 6 pm) and free access to food and water. Both male and female mice were analyzed in the study. Adipoq-Cre (Strain #: 028020) and IGF-1^f/f (Strain #: 016831) mouse lines were obtained from The Jackson Laboratory.^22,23

scRNA-seq analyses of BMSCs

The scRNA-seq data was generated previously with FACS-purified endosteal stromal cells from tibias and femurs of 8-wk-old male C57BL/6J mice²⁴ (Data accession GSE232738). scRNA-seq was performed with 10x Genomics. Downstream analyses were performed with Seurat 4.1.1 according to standard procedures.

RT-qPCR in BMSC or bone

IGF1 and ADIPOQ mRNA levels were measured by RT-qPCR in BMSC directly following isolation without culture. To isolate BMSC from the mice, the epiphyses were removed with scissors from femurs or tibias. The marrow content was flushed out with MEM-α media containing 10% FBS from both ends of the bone until the diaphysis appeared white. The cells were centrifuged and resuspended in MEM-α media containing 10% FBS and washed with PBS. Following this, the cells were mixed with CD45 microbeads (Miltenyi Biotec, 130-052-301), washed, centrifuged, and resuspended before being passed though LD columns (Miltenyi Biotec, 130-042-901) on a MidiMACS separator (Miltenyi Biotec, 130-042-302) attached to a multistand. The flow-through CD45-negative cells were subjected to RNA extraction according to the RNeasy Micro Kit (Qiagen). cDNA was made using the High-Capacity RNA-to-cDNA kit (Applied Biosystems). qPCR was performed with SYBR Green on QuantStudio3 (Applied Biosystems) with gene-specific primers for IGF-1, ADIPOQ, or HPRT (Table 1). Relative expression levels of IGF-1 or ADIPOQ were normalized to those of HPRT and calculated with the 2^−∆∆CT method.

Table 1.

Sequence of qPCR primers.

Primer name	Sequence
IGF-1 forward	5′-CTG GTG GAT GCT CTT CAG TTC G-3′
IGF-1 reverse	5′-TGC TTT TGT ACG CTT CAG TGG G-3′
TNFRSF11B (OPG) forward	5′-ACCCAGAAACTGGTCATCAGC-3′
TNFRSF11B (OPG) reverse	5′-CTGCAATACACACACTCATCACT-3′
TNFSF11 (RANKL) forward	5′-CAGCATCGCTCTGTTCCTGTA-3′
TNFSF11 (RANKL) reverse	5′-CTGCGTTTTCATGGAGTCTCA-3′
SOST forward	5′-AGCCTTCAGGAATGATGCCAC-3′
SOST reverse	5′-CTTTGGCGTCATAGGGATGGT-3′
ACTB forward	5′-GTGACGTTGACATCCGTAAAGA-3′
ACTB reverse	5′-GCCGGACTCATCGTACTCC-3′
HPRT forward	5′-TCA GTC AAC GGG GGA CAT AAA-3′
HPRT reverse	5′-GGG GCT GTA CTG CTT AAC CAG-3′

For RNA extraction from cortical bone, the growth plates were cut off from the femurs and the bone shafts were centrifuged briefly to remove the BM and cleaned of soft tissues. The bone shafts were then chopped up with scissors and homogenized in Precellys Evolution with Cryolys Evolution system (Bertin Technologies) using tubes CKMix50-R in TRIzol Reagent. Settings used were as follows: 6000 rpm, 10 s burst, 30 s rest, 3× repeat, 4 °C. Samples were then incubated on ice for 5 min and centrifuged at 12 000 × g for 10 min at 4 °C. Supernatant was pipetted into fresh tube, 200 μL chloroform added and shaken vigorously for 15 s, followed by 3 min rest. Samples were centrifuged at 21 000 × g for 15 min at 4 °C. Upper aqueous phase containing RNA was transferred to fresh tube. RNeasy Mini Kit (Qiagen, 74104) was used to isolate total RNA according to manufacturer’s protocol. RT-qPCR was performed as above using gene-specific primers for RANKL, OPG, or SOST. Beta-Actin (ACTB) was used as an internal control.

Flow cytometry of BMSC

Targeting efficiency was assessed by tdTomato expression in central marrow vs endosteal BMSC from Adipoq-Cre; Ai9 mice at 2-3 mo of age. After the epiphyses of tibias and femurs were excised with scissors and discarded, the central marrow fraction was collected by flushing with MEM-α media containing 10% FBS from both ends of the bone. To collect the endosteal fraction, the remaining bone was cut into small pieces and digested with 1 mg/mL dispase II (Roche, 4942078001) and 1 mg/mL STEMxyme1 (Worthington, LS004106) at 37 °C for 30 min. The central marrow and endosteal fractions were cultured separately with MEM-α and 10% FBS for 7 d with media changed at day 4. The CD45⁻ BMSC were then purified from each culture with magnetic cell separation (MACS) as described above. Single cell suspensions were subjected to flow cytometry with CytoFLEX LX (Beckman Coulter) for tdTomato expression. Cells not expressing tdTomato were used as a gating control.

Micro-CT analyses

Micro-CT (μCT) analysis was performed with μCT45 (Scanco Medical AG, Switzerland) according to guidelines of the American Society of Bone and Mineral Research. The femurs were scanned with the X-ray source of 55 kVp, 145 μA, 8 W, and 0.5 mm aluminum filter. The nominal voxel size was 4.5 μm and integration time was 400 ms. For quantifying trabecular bone parameters, two consecutive regions (ROI 1 and ROI 2) of 400 CT slices (1.8 mm) each were collected starting from 100 slices (0.45 mm) below the distal growth plate of the femur using a lower threshold of 350 and a Gaussian noise filter (sigma = 1.2, support = 2.0). For quantifying cortical bone parameters, 70 slices were analyzed using a lower threshold of 380 and a Gaussian noise filter (sigma = 1.2, support = 2.0). Males and females are quantified separately with genotypes blinded during analyses.

Dynamic histomorphometry

For double labeling of bone forming surfaces, calcein (5 mg/mL, pH 7.2, and 5 μL/g body weight) (Sigma, C0875) was administered intraperitoneally 7 d prior to harvest, followed by Alizarin Red S (15 mg/mL, pH 7.2, and 5 μL/g body weight) (Sigma, A5533) 2 d prior to harvest. Following harvest, femurs or tibias were fixed in 4% paraformaldehyde in PBS for 3 h at room temperature (RT) before being cryoprotected in 30% sucrose in PBS for 3 d at 4 °C. The bones were then embedded in Tissue-Plus O.C.T. compound, cryosectioned at 10 μm using a cryostat (Leica CM1950) and adhered to cryofilm type II membrane (Section Lab, Co., Ltd.). Sections were washed in PBS and mounted with ProLong Gold Antifade Mountant (Thermo Fisher, P36930) before images were captured with ZEISS Axio Scan.Z1.

BIOQUANT 2025 image analysis software was used for bone histomorphometry. Trabecular bone parameters were collected in ROI 1 starting at 450 μm below the growth plate with a fixed length of 1800 μm. The width of ROI 1 varied to encompass the entire trabecular bone width which differed across mice. The cortical region began adjacent to the end of ROI 1 at 2500 μm from the growth plate and extended for 3000 μm. Measurements included bone surface (mm), mineralizing surface (mm), bone formation rate (%), inter-label width (μm), mineral apposition rate (MAR) (μm/day), and bone formation rate normalized by bone surface (μm/day). Genotypes of the mice were blinded during analyses.

Serum PINP or CTX-1 ELISA

Whole blood was collected from mice via cardiac puncture using a 25 μm needle, and placed into an SST Microcontainer (BD) on ice. Tubes were centrifuged for 15 min at 4000 × g at 4 °C to isolate serum. Serum samples were stored at −80 °C until use. PINP and CTX-1 were measured in diluted serum with Rat/Mouse PINP EIA and RatLaps CTX-I EIA, respectively, by following the manufacturer’s instructions (Immuno Diagnostic Systems). The readings were acquired with a Cytation 5 image reader (Biotek).

Immunofluorescence and quantification of BM adipocytes

Femurs were fixed in 4% paraformaldehyde in PBS for 3 h at RT and then cryoprotected in 30% sucrose in PBS for 3 d at 4 °C. They were embedded in Tissue-Plus O.C.T. compound, cryosectioned at 10 μm using a cryostat (Leica CM1950) and adhered to cryofilm type II membrane (Section Lab, Co., Ltd.).

For immunostaining, sections were washed in PBS for 15 min, blocked with Antibody Diluent with BSA for 30 min at RT, and incubated with rabbit anti-perilipin monoclonal antibody (1:200, Cell Signaling Technology, #9349) overnight at 4 °C, followed by Alexa Fluor 647-conjugated F(ab’)2-goat anti-rabbit IgG (1:500, Thermo Fisher A21246) for 1 h at RT. Sections were mounted with ProLong Gold Antifade Mountant (Thermo Fisher, P36930) and imaged with Leica TCS SP8 Confocal Microscope. Perilipin+ BM adipocytes were quantified within an approximately 7 mm² area encompassing the primary spongiosa and trabecular bone (3 mm in height) on three sections per femur.

Analysis of BM hematopoietic stem and progenitor cells

To isolate the BM cells, both femurs were dissected from mice at 2 or 8 wk of age. The femurs were cut at the metaphysis on both ends, and BM was isolated by centrifugation using a quick pulse at 10 000 × g. BM was resuspended in PBS with 2% FBS. After cells were filtered through a 70 μm filter, red cell lysis was performed to obtain BM mononuclear cells. For flow cytometric analyses of hematopoietic stem and progenitor cells (HSPCs) cell surface markers, BM mononuclear cells were incubated with a cocktail of anti-Ter119, B220, TCRβ, CD48, Gr-1, CD11b, c-kit, Sca-1, CD135, CD150, CD48 (BioLegend), and analyzed on a Cytek Aurora spectral flow cytometer with the FlowJo 10.10.0 software. Hematopoietic stem and progenitor cell populations were gated and quantified as previously described.²⁵ For colony forming assays, approximately 8000 BM mononuclear cells were plated in fully supplemented methylcellulose (MethoCult GF M3434, Stemcell Technologies). Colonies were quantified after 7 d of culture based on morphology using phase contrast microscopy. All experiments included mice of both sexes.

Results

IGF-1 expression is enriched in CAR cells

To elucidate the molecular features of BMSC, we have further analyzed the single-cell RNA-sequencing (scRNA-seq) dataset that we have previously generated.²⁴ Consistent with previous analyses, the predominant majority of BMSC (clusters 0, 1, and 3-7) represents Cxcl12 abundant reticular (CAR) cells that are also enriched in Adipoq and LepR expression (Figure 1A and B).^20,21,26 Importantly, we discovered that IGF-1 was prominently expressed by most CAR cells (Figure 1A and B). In particular, the CAR cells in clusters 0, 1, 3, 4, and 7 exhibited higher IGF-1 levels than osteoblasts (cluster 2) and the related cells (clusters 9, 11, and 13) (Figure 1A and B). As IGF-1 is known to function as a paracrine factor regulating bone formation and hematopoiesis, we decided to pursue the potential function of IGF-1 derived from the BM CAR cells.

Figure 1.

Enrichment of Igf1 mRNA among CAR cells in 8-wk-old mice. (A) UMAP plot of BM mesenchymal cell clusters by scRNA-seq. (B) Violin plots of select marker genes. CAR, Cxcl12-abundant reticular cells; Osteo, osteoblast; Chondro, chondrocyte; Sm Musc, smooth muscle. (C and D) feature plots of Igf1 (C) and Adipoq (D) showing co-expression in CAR cells.

Adipoq-Cre deletes IGF-1 in BMSC

The strong overlap between IGF-1 and Adipoq expression prompted us to employ Adipoq-Cre to delete IGF-1 in CAR cells. We first determined the efficacy of Adipoq-Cre in targeting BMSC in Adipoq-Cre;Ai9 mice. BM stromal cells isolated from the central marrow vs endosteal niche were analyzed separately by flow cytometry. The results showed that ~88% central marrow BMSC and ~70% of endosteal BMSC expressed tdTomato, indicating adequate targeting of both populations by Adipoq-Cre (Figure 2A-C). We next examined directly the deletion efficiency of IGF-1 in the BMSC of Adipoq-Cre;IGF-1^f/f mice (CKO). BM stromal cells were enriched from flushed marrow through exclusion of the CD45⁺ cells with MACS beads, and then analyzed by RT-qPCR. The results showed notable reduction of IGF-1 mRNA in both male and female mice, indicating effective deletion of IGF-1 in BMSC by Adipoq-Cre (Figure 2D). In contrast, Adipoq mRNA levels in BMSC were not altered by the deletion (Figure 2E). Thus, Adipoq-Cre provides a useful tool for restricting IGF-1 production by BMSC.

Figure 2.

Targeting of BMSC by Adipoq-Cre. (A and B) Representative flow cytometry graphs showing tdTomato detection in central marrow (A) vs endosteal fraction (B) of BMSC from a 9-wk-old Adipoq-Cre;Ai9 female mouse. (C) Quantification of targeting efficiency in central marrow vs endosteal BMSC in Adipoq-Cre;Ai9 female mice at 9-12 wk of age. (D-G) Relative levels of IGF-1 (D and E) or ADIPOQ mRNA (F and G) in BMSC of 8-wk-old females (D and F) or males (E and G). CTRL, Igf1^f/f; CKO, Adipoq-Cre;Igf1^f/f; F, females; M, males. Statistics: unpaired t-test, *p < .05, **p < .01, each dot representing a single mouse.

IGF-1 deletion causes age-dependent defect in endocortical bone formation in female mice

We next examined the potential effects of IGF-1 deletion on bone parameters by μCT. We evaluated the cortical bone at the mid-diaphysis and the trabecular bone in upper (ROI 1) vs lower (ROI 2) regions. To our surprise, at 8 wk of age, no abnormalities in any of the cortical or trabecular bone parameters were observed in Adipoq-Cre;IGF-1^f/f (CKO) mice of either sex (Figure S1 and S2). As a previous study reported that deletion of IGF-1 by LepR-Cre reduced bone mass in 12-wk-old mice, we suspected that BMSC-derived IGF-1 might influence bone mass in an age-specific manner.¹⁸ To test this notion, we analyzed the bones of female CKO vs control (CTRL) mice at 12 wk of age. Here, like in the younger mice, all trabecular bone parameters were normal in the CKO mice (Figure 3A and B). However, notable defects were detected in the cortical bone, as indicated by significant reductions in bone area (BA), bone area fraction (BA/TA), and cortical thickness (C.Th), without changes in the total cross-sectional area (TA) (Figure 3C and D). Thus, whereas much of the BMSC-derived IGF-1 appears to be dispensable for trabecular bone mass or organization, it is necessary to support a normal cortical bone thickness during the third month of postnatal life when bone continues to grow in mice.

Figure 3.

Effects of BMSC-derived IGF-1 on bone mass in 12-wk-old female mice. (A-E) Representative 3-D reconstruction image (A) and quantification (B-E) of trabecular bone by μCT. (F-J) Representative image (F) and quantification (G-J) of cortical bone by μCT. CTRL, Igf1^f/f; CKO, Adipoq-Cre;Igf1^f/f. Statistics: unpaired t-test, *p < .05, **p < .01, ns: non-significant, p > .05, each dot representing a single mouse.

We next assessed the cellular basis for the reduced cortical bone thickness in the 12-wk-old female CKO mice. Dynamic histomorphometry showed that bone formation rate (BFR/BS) at the endosteal bone surface was significantly reduced owing to suppressed MAR even though mineralizing surface areas (MS/BS) were not altered (Figure 4A and B), indicating suppression of osteoblast activity without affecting the relative osteoblast numbers. In contrast, the trabecular bone formation parameters were normal in the CKO mice (Figure 4C and D). Moreover, BS/BV, an indicator of local bone resorption activity, was not altered in either cortical or trabecular bone (Figure 4B and D). Similarly, serum biochemistry did not reveal any significant changes in the overall bone formation (P1NP) or bone resorption (CTX-I) activity in the CKO vs CTRL mice (Figure 5A and B). Finally, as osteocytes are known to secrete several factors to influence osteoblasts or osteoclasts, we examined the mRNA levels of RANKL, OPG, and SOST in the cortical bone but observed no difference between the CTRL and CKO genotypes (Figure 5C-E). Thus, the data supports the notion that IGF-1 deletion in BMSC locally impaired osteoblast activity at the endosteal bone surface, causing the cortical bone defect.

Figure 4.

Dynamic histomorphometry for bone formation in 12-wk-old female mice. (A-E) Representative image (A) and quantification (B-E) of endosteal bone labeling. (F-J) Representative image (F) and quantification (G-J) of trabecular bone labeling. MS, mineralizing surface; BS, bone surface; BV, bone volume; MAR, mineral apposition rate; BFR, bone formation rate; CTRL, Igf1^f/f; CKO, Adipoq-Cre;Igf1^f/f. Statistics: unpaired t-test, ***p < .001, ****p < .0001, ns: non-significant, p > .05, each dot representing a single mouse.

Figure 5.

Quantification of serum bone markers, osteocyte factors and marrow adiposity. (A, B) Serum levels of bone formation (A) and resorption (B) markers. (C-E) Relative mRNA levels for osteocyte-derived factors as indicated. (F) Representative images for immunofluorescence staining of perilipin 1 (green) on sections of distal femur. Tissue morphology shown by differential interference contrast (DIC) imaging. (G) Quantification of adipocyte numbers in distal femur. CTRL, Igf1^f/f; CKO, Adipoq-Cre;Igf1^f/f. Statistics: unpaired t-test, ns: non-significant, p > .05, each dot representing a single mouse.

IGF-1 is well established to promote adipocyte differentiation from preadipocytes.²⁷ We therefore examined the potential effect of IGF-1 deletion on BM adiposity in the CKO mice. Immunofluorescence staining with a perilipin 1 antibody revealed normal adipocyte numbers in the long bones of the mutant mice (Figure 5F and G). Thus, IGF-1 originated from the Adipoq-lineage BMSC is likely dispensable for BM adipogenesis under normal conditions.

IGF-1 from Adipoq-lineage BMSC is dispensable for hematopoiesis

We next determined whether loss of IGF-1 in the BM microenvironment impaired HSPCs. For this, we isolated BM cells from the femurs of both males and females at either 2 or 8 wk of age, and performed flow cytometry and methylcellulose colony forming assays with the BM mononuclear cells. Flow cytometry indicated that the fractions of total Lin⁻Sca1⁺c-Kit⁺ (LSK) cells, long-term or short-term HSCs, and various multipotent progenitors (MPP2-4) were all normal among the BM mononuclear cells in the CKO male or female mice of either age (Figure 6A and B). Multipotent progenitor designations were based on a previous publication.²⁵ Moreover, the colony forming assays revealed no effect on the multipotential myeloid progenitors (CFU-GEMM), erythroid burst-forming units (BFU-E), or granulocyte-macrophage progenitors (CFU-G/M/GM) by IGF-1 deletion (Figure 6C). Thus, IGF-1 production by Adipoq-lineage BMSC appears to have no discernible effect on HSPC population abundance or colony-forming potential in young postnatal mice.

Figure 6.

Quantification of BM HSPC populations and colony formation potential. (A-L) Analysis of HSPC populations by flow cytometry in mice at 2 (A-F) or 8 (G-L) weeks of age. HSPC populations expressed as absolute frequencies among BM mononuclear cells. (M) Methylcellulose colony forming assays of BM mononuclear cells derived from 2-wk-old mice. Colonies identified by morphology and counted at day 7 (n = 4 per genotype). CTRL, Igf1^f/f; CKO, AdipoqCre⁺ Igf1^f/f. LSK, Lineage⁻Sca1⁺c-Kit⁺ hematopoietic stem and progenitor cells; LT-HSC, long-term hematopoietic stem cell; ST-HSC, short-term hematopoietic stem cell; MPP2, multipotent progenitor population 2 (megakaryocyte and erythroid bias); MMP3, multipotent progenitor population 3 (granulocyte and macrophage bias); MPP4, multipotent progenitor population 4 (lymphoid bias); CFU-GEMM, colony forming unit—granulocyte, erythrocyte, megakaryocyte, monocyte; BFU-E, burst forming unit—erythrocyte; CFU-G/M/GM, colony forming unit—granulocyte, macrophage. Data represented as mean ± SEM. Statistics: unpaired t-tests; n.s. (non-significant): p > .05, each dot representing a single mouse in (A-L).

Discussion

We have investigated the potential paracrine function of IGF-1 derived from BMSC in the regulation of bone growth and hematopoiesis. By deleting IGF-1 with Adipoq-Cre which targets most BMSC, the study uncovered a site-specific contribution of the local IGF-1 to bone formation at the endosteum of long bones, resulting in thinner cortices by 12 wk of age. On the other hand, the trabecular bone parameters or the HSPC frequencies were not affected by the deletion. Together with previous studies, the current findings support the view that multiple sources of IGF-1 in the BM environment likely exert distinct niche functions towards osteogenic or hematopoietic cells.

As Adipoq-Cre also targets mature adipocytes, it is important to note that a previous study reported no changes to adipose depots, whole-body metabolism or circulating IGF-1 levels when IGF-1 was deleted with Adipoq-Cre in mice under normal feeding conditions.²⁸ Thus, the endocortical bone phenotype observed in the mutant mice here is most consistent with the paracrine function of IGF-1 locally produced by BMSC.

We observed no HSPC phenotype in the CKO mice. This is consistent with a previous study where deletion of IGF-1 in BMSC with LepR-Cre did not impair hematopoiesis.¹⁸ In contrast, IGF-1 deletion with Nestin-Cre^ER was reported to cause HSC aging and myeloid-biased hematopoiesis,¹⁹ or impaired bone formation.²⁹ However, a similar Nestin-Cre^ERT2 line was found to target mostly endothelial cells (a known source of IGF-1) instead of stromal cells in the BM of postnatal developing mice.³⁰ Thus, it remains to be further determined whether the phenotypes reported earlier were strictly dependent on IGF-1 produced by BMSC.

The bone phenotype here seems to be at odds with a previous report that deletion of IGF-1 with LepR-Cre significantly reduced both trabecular and cortical bone mass due to impaired bone formation.¹⁸ The LepR-Cre-mediated deletion also increased BM fat which was not observed in the current study. Although LepR-Cre differs from Adipoq-Cre with additional activity in the periosteum, they both target BMSC with high efficiency. It is therefore surprising that deletion of IGF-1 by each Cre resulted in seemingly different bone phenotypes. However, some technical differences should be noted. Whereas the previous study focused on male mice at a single time point (12 wk of age), we analyzed both sexes at 8 wk and only the females at 12 wk of age. Whether the age and/or sex differences could explain some of the phenotypic discrepancy is not clear at present. Future work is necessary to analyze both sexes in both models at the same age for direct comparison. A more interesting alternative is that the 2 Cre drivers may target subsets of BMSC with different efficiencies. If so, the different phenotypes would highlight distinct contributions of IGF-1 from the various BMSC sources to trabecular vs endocortical bone formation. Future studies are necessary to examine potential functional diversity across subpopulations of BMSC.

Author contributions

Joshua C. Bertels (Data curation, Formal analysis, Investigation, Methodology, Writing—original draft), Jasmin Koehnken Sawall (Investigation, Methodology, Writing—review & editing), Brian Dulmovits (Formal analysis, Investigation, Methodology), Xiaobin Liu (Investigation, Methodology), Ashley Phan (Formal analysis), Xing Ji (Investigation), Fangfang Song (Formal analysis), Christopher Thom (Funding acquisition, Methodology, Writing—review & editing), and Fanxin Long (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Supervision, Writing—review & editing)

Funding

The work is partially supported by NIH grants R01 AG077911 (F.L.), R01 DK125498 (F.L.), and NHLBI K99 HL156052 (C.S.T.).

Conflicts of interest

The authors declare no conflict of interest.

Data availability

The data underlying this article are available in the article and in its online supplementary material. The scRNA-seq data is publicly available (Data accession GSE232738).