Signaling pathways associated with Lgr6 to regulate osteogenesis

Abstract

Fracture management largely relies on the bone’s inherent healing capabilities and, when necessary, surgical intervention. Currently, there are limited osteoinductive therapies to promote healing, making targeting skeletal stem/progenitor cells (SSPCs) a promising avenue for therapeutic development. A limiting factor for this approach is our incomplete understanding of the molecular mechanisms governing SSPCs’ behavior. We have recently identified that the Leucine-rich repeat-containing G-protein coupled receptor 6 (Lgr6) is expressed in sub-populations of SSPCs, and is required for maintaining bone volume during adulthood and for proper fracture healing. Lgr family members (Lgr4–6) are markers of stem cell niches and play a role in tissue regeneration primarily by binding R-Spondin (Rspo1–4). This interaction promotes canonical Wnt (cWnt) signaling by stabilizing Frizzled receptors. Interestingly, our findings here indicate that Lgr6 may also influence cWnt-independent pathways. Remarkably, Lgr6 expression was enhanced during Bmp-mediated osteogenesis of both human and murine cells. Using biochemical approaches, RNA sequencing, and bioinformatic analysis of published single-cell data, we found that elements of BMP signaling, including its target gene, pSMAD, and gene ontology pathways, are downregulated in the absence of Lgr6. Our findings uncover a molecular interdependency between the Bmp pathway and Lgr6, offering new insights into osteogenesis and potential targets for enhancing fracture healing.

1. Introduction

Skeletal stem and progenitor cells (SSPCs) are highly heterogeneous cells essential for skeletal development, remodeling, repair, and modulating the hematopoietic niche [1–3]. Diverse SSPC sub-populations have varying degrees of proliferation, self-renewal, and differentiation capacity, able to form adipogenic, chondrogenic, osteogenic, and stromal lineages. The osteogenic capacity of SSPCs is critical for maintaining bone volume, facilitating remodeling, and fracture healing. These processes involve the convergence of numerous signaling pathways that control SSPC proliferation and differentiation, whose interactions which are yet to be fully understood.

Bone morphogenetic proteins (Bmp) and Wingless Integrated (Wnt) signaling are key regulators of osteoblast differentiation and maturation [4,5]. Bmp enhances Wnt signaling in primary cultures via upregulating Wnt ligands and receptors. Generally, Bmp enhances the commitment of osteochondral progenitors while counteracting proliferating cues promoted by canonical Wnt (cWnt) signaling [6]. While recombinant human BMP2 is used in clinical practice as an ortho-biologic to enhance bone regeneration with specific indications and as an osteoinductive adjuvant in spinal fusion, its efficacy is limited, and its use can have severe side effects [7]. Modulation of the cWnt pathway using romosozumab, an anti-sclerostin antibody, prevents fractures in osteoporotic patients by promoting osteoblastic activity. However, this approach is not effective for the treatment of acute fractures [8]. Understanding the cWNT, BMP, and other molecular mechanisms that govern osteogenic processes is essential to developing new treatments that effectively promote bone repair and address bone loss.

Leucine-rich repeat-containing G protein-coupled receptors (LGR 4, 5, and 6) are a class of non-canonical GPRCs that have been shown to regulate and promote cWNT signaling through R-Spondin (RSPO1–4) mediated mechanisms in various stem cell niches throughout the body [9–11]. The actions of LGRs and RSPOs are associated with many physiological and pathological processes in bone. Rspo2 and Rspo3 are essential for regulating proper bone mass [12–14]. Recently, our group and others have reported that Lgr6-null mice develop normally without an overt skeletal phenotype; however, in adult mice, Lgr6 is required for maintaining bone volume [15,16]. In humans, genetic variants of the LGR6 loci are associated with osteoporosis [17,18]. In mesenchymal stem cells (MSC) derived from fractured bones of patients with hip fractures, LGR6 was the most upregulated gene (Fold change 11.0, corrected P-value = 0.0083) [19]. These data support our recent findings that Lgr6-null mice display impaired fracture healing and that Lgr6 likely plays an important role in human skeletal health and repair [15]. However, whether LGRs or RSPOs can independently regulate osteogenesis is largely unknown. Building on our previous studies, here we investigate the mechanism by which Lgr6 can modulate osteogenic processes. Supported by biochemical assays, bulk RNA-sequencing, and reanalysis of scRNA-sequencing data sets, our findings suggest that the RSPO-LGR-WNT axis are not inherently connected within the bone. Instead, we found that Lgr6 is more closely associated with BMP signaling during the osteogenic process, providing insights into the molecular mechanisms of Lgr6’s role in the skeleton.

2. Materials and methods

2.1. Animal welfare

This study was conducted in accordance with and following approval by the local IACUC at the University of Connecticut Health Center, which strictly adheres to NIH Guidelines for animal use. The animals were housed in a controlled environment with regulated temperature, humidity, and a 12-hour light cycle and were kept in ventilated cages.

2.2. Mice

Mice with knock-in targeted insertion of the EGFP-Ires-CreERT2 into the transcriptional start of Lgr6 locus (Lgr6^{EGFP-IRES-CreERT2}) [20] were bred into a mixed 129/SV background. Heterozygous breeding of these knock-in mice into homozygotes yields Lgr6^{EGFP-IRES-CreERT2}/Lgr6^{EGFP-IRES-CreERT2} (Lgr6-null mice), Lgr6^{EGFP-IRES-CreERT2}/⁺ (Lgr6/+ mice) and control littermates; functional endogenous Lgr6 expression is blocked from both alleles in Lgr6-null mice. Mice were genotyped at weaning and before tissue harvest using primers described in Table 1. Our previous studies have found that the skeletal phenotypes of the Lgr6/+ and Control mice were indistinguishable; therefore, in this study, we have compared cell cultures derived from Lgr6-null and control mice.

Table 1.

Primer sequences.

Gene	Forward	Reverse
Gapdh	5′-AGGTCGGTGTGAACGGATTTG-3′	5′-TGAGACCATGTAGTTGAGGTCA-3′
Alpl	5′-TGTGTGGGGTGAAGGCCAAT-3′	5′-TCGTGGTGGTCACAATGCCC-3′
Axin2	5′-ATGTCCTGTCTGCCAGCGTTC-3′	5′-CAAGCACTAGCCAGTGGGTCAA-3′
Bglap	5′-CACCATGAGAGCCCTCACACTC-3′	5′-CCTGCTTGGACACAAAGGCTGC-3′
Bmp2	5′-TGGAAGTGGCCCATTTAGAG-3′	5′-TGACGCTTTTCTCGTTTGTG-3′
Bsp	5′-AAAGTGAAGGAAAGCGACGA-3′	5′-GTTCCTTCTGCACCTGCTTC-3′
Col1a1	5′-TGAACGTGACCAAAAACCAA-3′	5′-GCAGAAAAGGCAGCATTAGG-3′
Dlx5	5′-GCCCCTACCACCAGTACG-3′	5′-TCACCATCCTCACCTCTGG-3′
Dmp1	5′-AGTGAGGAGGACAGCCCTGAA-3′	5′-GAGGCTCTCGTTGGACTTCAC-3′
Id1	5′-CTCTACGACATGAACGGCTGT-3′	5′-TGCTCACCTTGCGGTTCTG-3′
Lef1	5′-TGGCATCCCTCATCCAGCTATTGT-3′	5′-TGAGGCTTCACGTGCATTAGGTCA-3′
Lgr6	5′-ATCATGCTGTCCGCTGACTG-3′	5′-ACTGAGGTCTAGGTAAGCCGT-3′
Noggin	5′-CACTATCTACACATCCGCCCAG-3′	5′-AGCGTCTCGTTCAGATCCTTCT-3′
Ocn	5′-CACCATGAGAGCCCTCACACTC-3′	5′-CCTGCTTGGACACAAAGGCTGC-3′
Runx2	5′-CCACCACTCACTACCACACG-3′	5′-CACTCTGGCTTTGGGAAGAG-3′
Sp7	5′-GATGGCGTCCTCTCTGCTTG-3′	5′-GCCATAGTGAGCTTCTTCCTCAA-3′
hGAPDH	5′TGGTATCGTGGAAGGACTCATGAC-3′	5′-ATGCCAGTGAGCTTCCCGTTCAGC-3′
hLGR4	5′-CCAAGCGCTGGATATCAGTATG-3′	5′-CCCGCCAATTGTAGCTCTTC-3′
hLGR5	5′-GTCTTCACCTCCTACCTAGACCT-3′	5′-GTTAGCATCCAGACGCAGGG-3′
hLGR6	5′-TAACCCAATCCAGTTTGTGGGAA-3′	5′-GCAGCTCCTCAATTTGATTGTGA-3′
Genotyping primers
Lgr6 WT	5′-CTCGCCCGTCTGAGCG-3′	5′-GCAGGCACCACTGAGAGC-3′
Lgr6 mut	5′-GCCCACCGACGGCGCAGCCC-3′	5′-GCTGAACTTGTGGCCGTTTA-3′

2.3. Cell cultures

Primary bone-marrow-derived human MSCs (from healthy donors aged 22 to 29 years) were obtained as individual frozen vials from the Institute of Regenerative Medicine, Texas A&M University. These cells were pre-characterized for cell surface expression (CD166+ CD90+ CD105+ CD36− CD34− CD10− CD11b− CD45−) and tri-lineage differentiation (osteoblastic, adipogenic and chondrogenic) potential. Cells were cultured at a density of 3000 cells/cm² using alpha-MEM supplemented with 16.5 % FBS (Atlas Biologicals, CO) in standard culture conditions for propagation and used within passage 7. Osteoblast differentiation was achieved using BMP stimulation by adding 200 ng/ml of recombinant human BMP2 (R&D Systems, MN) to the cell monolayer using osteopermissive media (serum-free alpha-MEM additionally supplemented with 25 μg/ml l-Ascorbic Acid Phosphate Magnesium Salt n-Hydrate (AA2P) (FUJIFILM Wako Chemicals USA Corp., VA), 5 mM β-glycerophosphate disodium salt hydrate (BGP) (Sigma, MO) and 1× Insulin-transferrin-selenous acid premix (ITS) (Corning, NY))[21,22]. Stimulated cells were harvested after 3 days for total RNA extraction. To detect extracellular calcium, cells received osteopermissive media every 3–4 days for 10 days and were stained using 1 % Alizarin Red S using standard procedure.

MC3T3-E1 cells were grown at a 5000 cells/cm² density in alpha-MEM supplemented with 10 % FBS (Atlas Biologicals, CO) in standard culture conditions for propagation and used within passage 10. Osteoblast differentiation was achieved using 100 ng/ml of recombinant mouse Bmp2 (R&D Systems, MN) to sub-confluent cell monolayer using osteopermissive media (alpha-MEM containing 5 % FBS and additionally supplemented with 25 μg/ml AA2P and 5 mM BGP). Stimulated cells were harvested after 3 days for total RNA extraction. Cells received fresh osteopermissive media every 3–4 days until stained for extracellular calcium deposition using 1 % Alizarin Red S.

As previously reported, calvarial cultures were derived from 5- to 7-day-old pups [23]. Cells were allowed to reach confluency and replated for osteogenic differentiation assay using methods described for MC3T3-E1 cells. Bone marrow stromal cell (BMSC) cultures were prepared as described previously [24].

2.4. CFU assay

Bone marrow-derived cells from individual mice were plated in duplicates at a density of 1 × 10⁶ cells in 60 mm culture plates and cultured in 5 % carbon dioxide without cell passage for ten days. Fixed cells were stained for ALP activity using an ALP detection kit according to the manufacturer’s instructions (Sigma-Aldrich) or with crystal violet. Colonies were imaged using a Leica microscope and were evaluated using converted greyscale 8-bit images in ImageJ.

2.5. Osteogenic differentiation

For osteogenic differentiation, isolated bone marrow-derived cells acquired from individual mice were seeded (1 × 10⁶ cells/well) in a 6-well plate. Cells were treated with differentiation media (50 μg/mL ascorbic acid, 4 mM β-glycerophosphate) after culturing for 2–3 days to reach 80–90 % confluency. Media was changed every other day for 10 days. Cells were fixed in cold methanol and stained with Alizarin Red S (ScienCell) to visualize in vitro mineralization. In a parallel experiment, another set of control and Lgr6-null cultures derived from individual mice was used for bulk mRNA sequencing analysis.

2.6. Bulk mRNA-sequencing and analysis

RNA was isolated from cell cultures using a TRIzol (Invitrogen) extraction method. Total RNA was quantified, and purity ratios were determined for each sample using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). To further assess RNA quality, total RNA was analyzed on the Agilent TapeStation 4200 (Agilent Technologies). High-quality RNA (RIN > 9.0) was isolated from osteogenic cultures, and mRNA sequencing, including library preparation, was performed at the UConn Genomics facility using the Illumina Stranded mRNA Ligation Sample Preparation kit following the manufacturer’s protocol (Illumina). Libraries were validated for length and adapter dimer removal using the Agilent TapeStation 4200 D1000 High Sensitivity assay (Agilent Technologies), then quantified and normalized using the dsDNA High Sensitivity Assay for Qubit 3.0 (Life Technologies). All samples were pooled into one sequencing pool, equally normalized, and run as one sample pool across the Illumina NovaSeq 6000 using version 1.5 chemistry. A target read depth of 50 M reads was achieved per sample with paired-end 100 bp reads.

After trimming and quality control, reads were mapped to the Mus musculus genome (GRCm39) with HISAT/2.1.1 (https://doi.org/10.1038/s41587-019-0201-4). Counts were generated against the features with HTSeq-count (htseq/0.13.5) [25]. Differential expression of genes between conditions was generated using DESeq2 [26]. Genes showing <10 counts across the compared samples were dropped from the analysis. P-values were adjusted for comparison using the false discovery rate (FDR). The criteria for DEGs were >1.0 log2 fold change/<−1.0 log2 fold change (absolute fold change 2) and adjusted P-values < 0.05. KEGG annotations (https://www.genome.jp/kegg/pathway.html) were used for functional annotation. KEGG terms with FDR (q < 0.05) were significantly enriched within the gene set. Gene Set Enrichment Analysis (GSEA) was performed using multiple annotated gene set databases available from the Molecular Signatures Database v7.7.5.1 (MsigDB) (gsea-msigdb.org). For the present study, we used KEGG gene sets and GO Biological Processes and a list of ranked genes based on a score calculated as −log10 of adjusted P-value multiplied by the sign of fold-change. Enrichment of gene sets was performed using the default GSEA using 1000 permutations, with gene sets containing between 30 and 500 genes being reported [27]. Selected enriched gene sets are shown from all enriched genes found in the Supplementary.

2.7. RT-qPCR

Total RNA was extracted from cultured cells, and RNA integrity was checked using BioAnalyzer RNA NanoChips (Agilent Technologies). Only samples with a RIN value of >8.0 were used. Reverse transcription (RT) to synthesize cDNA was done using the Superscript III cDNA synthesis kit (Invitrogen). RT-qPCR reactions were performed on a Bio-Rad real-time PCR system using SYBR Green (BioRad) PCR reagents. For each gene of interest, qPCR was performed in both technical and biological triplicates. Data were normalized to Gapdh. Primer sequences are listed in Table 1.

2.8. Western blotting

To analyze protein expression, cultured cells were lysed in modified RIPA buffer, and the clarified total cell lysate (30 μg) was electrophoresed on 10 % SDS-PAGE gels as described previously [23]. Western blots were probed with the indicated antibodies. The blot was stripped and reprobed with anti-Gapdh antibodies or pERK, pAKT, and pSMAD1 to verify equal protein loading. Band intensities were measured using BioRad Image Lab v6.0.

2.9. Single-cell RNA-sequencing analysis

Single-cell RNA sequencing (scRNA-seq) data from the Gene Expression Omnibus (GEO) was downloaded for analysis from GSE138689 [28], GSE136970 [29], and GSE195940 [30], which included sample barcodes, features, and matrix files. For all analyses, processing was done using RStudio 4.2.2 with the Seurat 4.3.0 package [31]. The GSE138689 data sets were derived from CD45-Ter119-Tie2-Lepr-Cre+ cells that were isolated from bone marrow and digested bone fragments from intact bones (GSM4116165) or bones seven days after fracture where femurs were dissected and crushed for enzymatic dissection (GSM4116169) [28]. For the GSE136970 data set, FACS-sorted Cxcl12-GFP+ cells from two P28 mice (GSM4064136 and GSM4064137) were used [29]. GSE195940 data set used confluent cultured periosteal cells derived from intact and 3-day post-fractured tibiae [30].

Quality control, removal of low-quality cells, and cluster analysis were performed similarly to what was described in the original manuscript unless otherwise noted. For GSE15940, cells with >10 % mitochondrial transcripts and <100 or >8000 features were removed. For GSE138689, cells with >15 % mitochondrial transcripts, <500 features, and >2 % hemoglobin (UMIs with Hbb-UMI) were removed. For GSE136970, cells with >15 % mitochondrial transcripts and <1000 features were removed. Cluster identification and markers used were based on those described by [14] for GSE136970. Filtered samples were normalized and integrated using Seurat’s updated ‘v2’ variance stabilizing of scRNA-seq data (SCTransform, v2) [32,33]. Clustering and visualizing of genes and samples were also done using Seurat.

3. Results

Recent findings from our group and others indicate that Lgr6 is dynamically expressed during the osteogenic differentiation of SSPCs derived from various mesenchymal tissues, including calvaria, periosteum, and bone marrow [15,16,34]. Our previous studies have shown that SSPCs from the periosteum and bone marrow of 3- to 5-month-old Lgr6-null mice exhibit severely impaired osteogenic differentiation and mineralization as evaluated by ALP (alkaline phosphatase) staining and Alizarin Red S staining. Concomitantly, compared to controls, the expression of osteogenic markers, including Runx2, Alpl, Ocn, and Col1a1, decreased. Additionally, we found a modest decrease in the expression of downstream targets of Wnt signaling Axin2 and Lef1 only upon prolonged exposure (72 h) of BMSCs to Wnt3a stimulation [15]. To further confirm the modest effect on cWnt signaling in the absence of Lgr6, we show that BMSCs stimulated with Wnt3a for 24 or 48 h show comparable expression of Axin2 between genotypes (Supplementary Fig. 1A and data not shown). Additionally, western blotting analysis revealed similar levels of active (non-phosphorylated) β-catenin and total β-catenin after Wnt3a stimulation of control and Lgr6-null cells (Supplementary Fig. 1B). These findings suggest that signaling pathways other than cWnt are impaired in Lgr6-null cells undergoing osteoblast differentiation.

Recent reports have shown that stimulation of MC3T3-E1 cells or wild-type BMSCs with BMP increases Lgr4 and Lgr6 expression [35,36]. Furthermore, in another study, stimulation with Wnt3a decreases Lgr6 expression in periosteal cultures derived from Prxx1^Cre/⁺; Bmp2^fl/fl mice [37]. Based on these findings, we investigated the relationship between Bmp signaling and Lgr6 in osteogenic cell types, including primary human MSCs, MC3T3-E1 cells, and primary calvarial cultures derived from 1 to 5-day-old pups (Fig. 1A–C). Our studies showed that compared to Lgr4 and Lgr5, Lgr6 expression was robustly enhanced in cultures treated with BMP2. Bmp response was confirmed by assessing mineralization in different culture systems (Fig. 1A–C) and the expression of Id1 and Noggin in calvarial cultures (Fig. 1D). Next, we examined the effect of BMP2 on Lgr6-null cells and found that expression of BMP-responsive genes Id1 and Noggin was attenuated in Lgr6-null calvarial cultures compared to controls (Fig. 1E). Additionally, western blotting analysis showed reduced phosphorylation of SMAD 1/3/5, the canonical component of the Bmp2 signaling pathway, in Lgr6-null cultures compared to controls. However, AKT and ERK phosphorylation, non-canonical components of Bmp2 signaling, were comparable between the genotypes (Fig. 1F).

Fig. 1.

Bmp signaling is diminished in Lgr6-null mesenchymal cells: (A) Representative Alizarin Red stained wells show extracellular calcium deposition after BMP2 stimulation of hMSCs for ten days. The graph depicts LGR4, LGR5, and LGR6 expression levels after BMP2 stimulation for three days in the osteopermissive media. Each colored dot represents cells from a unique hMSC donor. (B and C) Alizarin Red S staining of MC3T3-E1 and murine calvarial osteoblasts stimulated with BMP2 in osteopermissive media for 12 and 16 days, respectively. Expression of Lgr4, Lgr5, and Lgr6 was determined after BMP2 stimulation for three days in osteopermissive media for both cell types. (D) Temporal expression of Id1, Noggin, and Lgr6 in response to Bmp2 (50 ng/ml and 100 ng/ml) following 24 h serum starvation of WT calvarial cultures was determined by RT-qPCR. (E) Control and Lgr6-null calvarial cultures derived from 5-day-old pups after one passage were serum starved for 24 h and then treated with Bmp2 (50 ng/ml), Dkk1 (50 ng/ml), or a combination of both for 24 h. The expression of Id1 and Noggin was examined using RT-qPCR. (F) Calvarial cultures were grown to confluence, and serum starved for 24 h, followed by stimulation with Bmp2 (50 ng/ml) for 5 and 20 min. Protein expression of phospho-SMAD1, phospho-AKT, and phospho-ERK was determined by western blotting. Blots were stripped and reprobed with anti-SMAD1, anti-AKT, or anti-ERK antibodies to determine the phosphorylated/total protein ratio. (G) Cultures from Lgr6-null and control mice were grown to confluence and treated with 10 % FBS or osteogenic media (OM; 50 μg/ml ascorbic acid and 8 mM β-glycerophosphate) or treated with Bmp2 (50 ng/ml), or Rspo2 (50 ng/ml), or combination of both factors (50 ng/ml each). All growth factors were added to media containing no serum. Media was changed every day for five days. Cells were stained using a commercial kit for ALP activity to visualize osteogenesis. Cells from four-five independent donors were used for the experiment using hMSCs. A representative of four independent experiments is shown for MC3T3-E1 cells. A representative experiment of three separate experiments is shown for all murine calvarial culture experiments. Data were analyzed by Student’s t-test; **p < 0.01, ***p < 0.001; ****p < 0.0001 vs untreated. White bars indicate controls and red bars represent Lgr6-null samples. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

We previously showed that ALP expression and activity are significantly diminished in Lgr6-null cultures derived from bone marrow and periosteum [15]. Therefore, we investigated whether adding recombinant BMP2 or Rspo2, or a combination of the two, could restore ALP activity in Lgr6-null cultures (Fig. 1G). Control cells exposed to BMP2 had robust induction of ALP staining. However, continuous treatment with BMP2, Rspo2, or a combination of Rspo2 and Bmp2 for five days did not improve deficient ALP staining in Lgr6-null cultures. The lack of ALP staining in control cultures upon Rspo2 treatment aligns with our previous findings [12]. We have reported that despite the enhanced mineralization of cultures in response to Rspo2 plus Bmp2, Rspo2 alone is insufficient to induce ALP [12,38]. Overall, the absence of Lgr6 leads to attenuated Bmp2-responsive gene expression, reduced canonical SMAD phosphorylation, and impaired ALP activity, all of which highlight the importance of Lgr6 in osteogenic processes. However, adding exogenous Bmp2, or Bmp2 + Rspo2, does not rescue deficient ALP staining in Lgr6-null cultures, suggesting that Lgr6 modulates Bmp2-mediated osteoblast differentiation.

3.1. Bulk RNA-sequencing of Lgr6-null bone marrow-derived cells shows differential expression of osteogenic genes

We have previously shown that compared to age- and sex-matched controls, osteogenic differentiation is compromised in calvarial cultures derived from 5-day-old pups and in periosteal and BMSC cultures derived from 5-month-old Lgr6 null mice (data not shown; [15]). To determine whether the effect of Lgr6 deficiency persists with age, we used 18-month-old male control and Lgr6-null mice for this experiment. In agreement with previous results from 3 and 5-month-old mice, BMSCs isolated from aged Lgr6-null mice had a 62 % reduction in colony-forming potential, 85 % decrease in ALP+ colonies, and significant reductions in mineralization and the expression of osteogenic markers (Fig. 2A–C). These data confirm that Lgr6 expression is required for regulating osteogenesis. To gain insight into underlying mechanisms, specifically whether reduced Bmp signaling is responsible for decreased osteogenesis in Lgr6-null cells, we performed bulk RNA-sequencing (bulk RNA-seq) on control and Lgr6-null BMSCs grown in an osteogenic medium for ten days.

Fig. 2.

Lgr6-null BMSCs have decreased osteoblastogenesis and mineralization: (A) BMSCs harvested from individual 18-month-old male Control and Lgr6-null mice. Graphs show CFU and CFU-ALP+ colonies. (B) BMSC cultures from individual mice were grown to confluence for three days. Cultures were treated with osteogenic medium (50 μg/ml ascorbic acid and 8 mM β-glycerophosphate) for ten days. The osteogenic medium was changed every other day. Calcium deposits were assessed by staining with Alizarin Red S. (C) RT-qPCR analysis of indicated osteogenic markers. N = 4 mice/genotype. Data analyzed by Student’s t-tests *p < 0.05; **p < 0.001 ***p < 0.001; ****p < 0.0001 vs control. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The principal component (PC) analysis of our bulk RNA-seq data showed PC1 accounted for 81 % of the variation and PC2 accounted for 7 %, which showed close grouping between the genotypes (Fig. 3A). Of the 19,461 genes identified, 4154 were statistically different (Adj.P < 0.05) (Fig. 3B). Of these, 764 genes were upregulated, and 492 genes were downregulated (Log2FC+/− 1) in Lgr6-null cells. Gene-set enrichment analysis (GSEA) was performed on differentially expressed genes using the KEGG Pathway Database [39]. Our analysis showed that TGFβ superfamily signaling was downregulated in Lgr6-null cultures (Fig. 3C). Consistent with decreased osteogenesis and mineralization in Lgr6-null cultures (Fig. 2), the expression of mineralization enzymes (Alpl, Phospho1, Phex) and osteoblast markers (Sp7, Bglap, and Dmp1) was decreased in Lgr6-null osteogenic cultures compared to controls (Fig. 3D).

Fig. 3.

Differential gene expression comparing control and Lgr6-null BMSCs osteogenic cultures. (A) Principal Component Analysis. (B) Volcano plot showing differentially expressed genes. (C) KEGG analysis. (D) Heat map for osteogenic and mineralization marker genes comparing Lgr6-null and control. N 4 male mice/genotype.

3.2. Signaling pathways associated with Lgr6

Lgrs are considered high-affinity receptors for Rspo, and in turn, Lgr-Rspo signaling is known to enhance cWnt signaling. Intriguingly, KEGG pathway analysis revealed that cWnt pathway genes were not changed in Lgr6-null cultures (nom p < 0.331) (Fig. 4A), and the Wnt Signaling Pathway was the 30th most down-regulated pathway in the Lgr6-null cultures. Lef1 and Axin2 expression were used as surrogates for Wnt activity, and Rspo 2 and Rspo 3 expression levels were also comparable between the genotypes (Fig. 4B). These data agree with the minimal changes in Axin2 expression and phosphorylation of β-catenin (Supplementary Fig. 1).

Fig. 4.

cWnt signaling is unperturbed in Lgr6-null cells: (A). Enrichment plot. (B) RT-qPCR analysis of Wnt target genes Lef1 and Axin2 and Wnt signaling modulators Rspo2 and Rspo3. N = 4 mice/genotype. Student’s t-tests were used to determine statistical significance.

The TGFβ superfamily includes TGFβs, activins, and BMPs, structurally related signaling molecules grouped in the KEGG Pathway Dataset as a single pathway. Our analysis showed the most downregulated KEGG term was the TGFβ signaling pathway (BMP-signaling pathway; p < 0.001) (Fig. 5A). A heat map comparing control and Lgr6-null samples shows that expression of Bmp ligands (Bmp2, Bmp3, and Bmp7), Bmp receptors (Bmpr1a, Bmpr2), transcriptional targets (Id1–3), and inhibitors (Noggin and Gremlin) were downregulated in Lgr6-null samples (Fig. 5B). RT-qPCR confirmed decreased Bmp2, Bmpr1a, Id1, and Noggin expression in BMSCs-derived Lgr6-null cultures obtained from 18-month-old mice (Fig. 5C).

Fig. 5.

Genes regulating Bmp signaling are downregulated in Lgr6-null osteoblasts cultures derived from mice: (A) Enrichment plot. (B) Heat map comparing Lgr6null and control. (C) RT-qPCR analysis confirmed decreased expression of genes in the Bmp pathway in Lgr6-null samples. N = 4 male-mice/genotype. Data analyzed by Student’s t-tests *p < 0.05; **p < 0.001 ***p < 0.001; ****p < 0.0001 vs control.

3.3. Differential expression of LGRs and RSPOs in Cxcl12+ and Lepr+ cells derived from intact and injured bones

We next sought to understand better cell types that express Lgr6 and its homologs and R-spondins under homeostasis and injury. We first examined bone marrow stromal cells that are described as being Cxcl12+. We reanalyzed a dataset obtained from FACS-sorted Cxcl12+ stromal cells derived from intact bone (GSE136970; [29]). Using the same data set, as reported by Nilsson et al. [13,40], we confirmed that Rspo3 is most robustly expressed in osteogenic and adipogenic Cxcl12+ cells (Fig. 6A and B). For the Lgr family, Lgr4 was expressed at a low level, Lgr5 was expressed in only a few cells, and Lgr6 expression was not detected in any cell cluster (Fig. 6B).

Fig. 6.

scRNA-seq of Cxcl12+ cells from intact bones. (A) Feature plots of FACS-sorted Cxcl12+ cells derived from bone marrow. (B) The expression pattern of Lgr4, Lgr5, Lgr6, Rspo2, and Rspo3 was determined. Rspo1 and Rspo4 were not expressed at a detectable level.

Based on our recent findings that Lgr6 is critical for mediating the healing response to fracture injury, we next reanalyzed the scRNA-seq data set (GSE138689) that used LepR+ cells derived from fractured tibia seven days after injury [28]. LepR+ cells largely overlap Cxcl12+ cells and are an essential source of SSPCs in the bone marrow and for intramembranous bone healing [28]. Similar to our Cxcl12+ cell analysis, only Rspo3 was expressed in LepR+ cells, and there was a modest increase in response to injury; Rspo2 was expressed at a low level. No expression of Rspo1 or Rspo4 was detected (Fig. 7). Minimal expression of Lgr4 and Lgr5 was found in intact or injured cell clusters; no appreciable Lgr6 expression was observed in any of the BMSC clusters. We confirmed our findings by reanalyzing a published bulk RNA-seq dataset from LepR+ cells from the intact bone (GSE173371; [49]), which showed that Lgr6 was expressed at either very low levels or not at all in skeletal LepR+ cells depending on the sample replicate. Our analysis of scRNAseq data shows that Lgr6 is not co-expressed in Cxcl12+ bone marrow stromal cells derived from either intact or injured bones, and Rspo3 likely exerts its effects via either Lgr4 or other receptors. Furthermore, our scRNA-seq analysis of LepR+ cells showed that Bmp2 was minimally expressed in non-injured cells, and its expression did not increase in response to injury (Fig. 7).

Fig. 7.

scRNA-seq of LepR+ cells of fractured and intact bones. (A) Integrated data sets of scRNA-seq of LepR+ cells of fractured and intact bone were visualized using UMAP-based clustering; Nine distinct cell clusters (Cluster 0–8) were identified. (B) Violin plots show Lgr4, Lgr5 Lgr6, Rspo2, Rspo3 and Bmp2 expression pattern. Rspo1 and Rspo4 were not detected in the datasets.

3.4. Lgr6 expression increase in periosteal SSPCs derived from fractured bones

We identified Lgr6 as one of the highly expressed genes in periosteal cultures derived from a mouse model with enhanced PI3K activity [15]; this model exhibits a robust periosteal reaction in response to fracture injury. Furthermore, in Lgr6-null mice, fracture healing is delayed due to a deficient periosteal response characterized by decreased periosteal cell proliferation and ALP activity at the injury site [15]. Therefore, to gain mechanistic insights into the Lgr6 function, we re-analyzed the published scRNA-seq dataset (GSE195940) reported by the Colnot group [30]. This single-cell data set used ex vivo expanded confluent periosteal cell cultures derived from intact and fractured tibiae (three day post-fracture). We identified the same 3 clusters of immune cells, 9 clusters of skeletal stem/progenitor cells (SSPCs), fibroblast, and fibrochondro progenitors (FCPs) as was originally described (Fig. 8A). The integrated dataset contained a mixture of hematopoietic CD45+ (Ptprc+) and mesenchymal Prrx1+ cells (Fig. 8B).

Fig. 8.

Lgr6 and Bmp2 expression is enriched in periosteal-derived SSPC clusters. (A) Integrated data sets of scRNA-seq of fractured and intact periosteal cells were visualized using UMAP-based clustering, and 13 distinct cell clusters (Cluster 0–12) were identified. (B) The integrated dataset contained hematopoietic CD45+ (Ptprc) and mesenchymal (Prrx1+) cells. (C) UMAPs showing indicated gene expression in periosteal cells derived from uninjured (intact) and injured (fractured) bones.

We found robust Lgr6 expression in SSPC clusters only in cells derived from fractured injury samples; minimal expression was detected in non-injured SSPCs or in the Fibro Chondro Progenitor (FCP) cluster, which is predicted to have a cell trajectory toward the chondrogenic state (Fig. 8C). This Lgr6 distribution aligned closely with our published work using the Lgr6/+; Ai9 reporter mouse in which Lgr6+ cells and their progeny (marked as being Tomato+) were only present in injured periosteum five days post fracture and in newly formed bone callus on day 14 after injury. We did not detect Tomato+ cells in the cartilaginous callus [15]. Of the other Lgr family, Lgr4 expression was minimal in uninjured and injured samples; Lgr5 expression was below the detection limit. Interestingly, Rspo2 and Rspo3, major ligands of Lgr6, had minimal expression in all clusters. Julien et al. also found that osteochondral differentiation of periosteal SSPCs was mediated by BMP signaling [30]. The scRNA-seq re-analysis showed that the injury-induced cell specific Bmp2 expression was similar to the Lgr6 expression pattern (Fig. 8).

4. Discussion

In this study, we investigated the role of Lgr6 in modulating osteogenesis. Classically, BMP boosts the commitment of osteochondral progenitors while mitigating the proliferative signals facilitated by cWnt. Our studies suggest that the osteogenic phenotype of Lgr6-null is likely due to decreased canonical BMP signaling. We found that in the absence of Lgr6, canonical Bmp signaling in mesenchymal cells is reduced. RT-qPCR analysis of Lgr6-null cultures derived from aged BMSCs showed decreased expression of Bmpr1a and Bmpr2, suggesting that the decreased receptor expression could be one of the factors contributing to the diminished Bmp response in Lgr6-null cultures. In support of our studies, microarray data comparing the effect of Wnt3 on mesenchymal cells derived from control (Bmp2^fl/fl) and Prxx1Cre; Bmp2^fl/fl mice showed an attenuated Lgr6 expression but no changes in Lgr4 levels [37]. Other reports have shown that Bmp9 also stimulates the expression of Lgrs in MC3T3-E1 cells and primary osteogenic cultures [35,36].

A recent report suggests that Rspo2/3 are Bmp receptor antagonists in Xenopus’s early development [41]. In this model, Rspo2 specifically interacts with Bmpr1a and tethers it to ubiquitin ligase Znrf3 in an Lgr-independent manner to trigger endocytosis and degradation of Bmpr1a, thereby regulating osteogenesis. We show that Rspo2/3 expression is comparable between Lgr6-null and control cells; however, future studies could investigate the relationship between Bmp receptors, Rspos, and Lgr6. Previously, in Lgr6-null cells, we found a modest decrease in Axin2 expression after prolonged exposure to Wnt3a. Here, we found that in response to Wnt3a stimulation, β-catenin levels only modestly changed in Lgr6-null cells, and no changes were found in Axin2 expression. The activity/levels of cWnt signaling components depend on osteogenic differentiation. Since Lgr6-null cells have impaired osteogenic differentiation, it is difficult to interpret if the results are due to alterations in cWnt signaling or osteogenic differentiation potential affected by a mechanism unrelated to Wnt.

In the scRNA-seq dataset comparing periosteal cultures derived from intact and fractured bones developed by the Colnot group, expression of Bmp2 is remarkably increased in periosteal cultures derived from injured bones, while Bmpr1a expression was comparable between cultures derived from intact and injured bones. Lgr6 expression mirrored Bmp2 expression with significant increases in fractured tibiae (Fig. 8), suggesting that Lgr6 is likely expressed in a specific subpopulation of periosteal cells activated in response to injury. These data support our observations that Lgr6 expression augments BMP signaling (Fig. 1).

Analysis of scRNA-sequencing of bone marrow-derived Cxcl12+, Lepr+, and Prrx1+ periosteal cells support the premise that Rspos may activate cWnt signaling via Lgr-independent mechanisms or act in a paracrine manner, affecting different cell types or supporting the hematopoietic niche. Cxcl12+ and Lepr+ cells differentiate into osteoblasts and support hematopoiesis by secreting various factors to act on neighboring cells. Within the Lepr+ and Cxcl12+ FACS sorted cells, Rspo3 is highly expressed, in congruence with other studies analyzing the same data [13]. However, cells that expressed Rspo did not express Lgrs at an appreciable level, suggesting that Rspo proteins may interact with different receptors in addition to Lgrs. Rspos have binding affinity for all three Lgr family members, and it remains unclear if a distinct Rspo/Lgr axis regulates osteoblast differentiation, as disruption of Rspo2 in the osteoblast lineage results in decreased bone mass [12]. Intriguingly, Rspo3 deletion in osteoprogenitors has divergent effects on regulating bone mass: in long bones, Rspo3 deficiency results in increased trabecular bone mass, while deficiency in vertebrae results in low trabecular bone mass [13,40,42]. Recent evidence indicates that heparan sulfate proteoglycans (HSPGs) act as alternative co-receptors for R-spondins, and this interaction can modulate cWnt signaling independent of Lgr6 [43–45]. Rspo2/3 are Bmp receptor antagonists in Xenopus’s early development, which may be another mechanism by which they regulate osteogenesis [41]. Moreover, Lgr6-null mice have a pronounced impaired fracture phenotype. Still, it appears that deficiency of Rspo2 does not have an appreciable role in regulating osteochondral progenitor cell expansion in response to fracture healing [46].

Lgr6 can also interact with Maresin 1 (MaR1), enhancing G protein-mediated signaling and cAMP activity [16]. MaR1-Lgr6 interaction has been identified in the context of immune resolution of inflammation in human and mouse phagocytes [47], but it is not reported in other cell types. However, MaR1 has been associated with promoting an anabolic bone phenotype [48], suggesting that Lgr receptors may signal through other pathways in addition to Wnt and Bmp pathways to regulate osteogenesis.

In summary, our study highlights an important relationship between Bmp signaling and Lgr6, utilizing multiple cell types derived from humans and mice of various ages in different bone compartments, and suggests a potential mechanism for the diminished osteogenesis observed in Lgr6-null cultures. In ex vivo cultures from Lgr6-null mice, we observed impaired osteogenesis and downregulated Bmp signaling. In contrast, cWnt signaling remained unchanged, indicating that Lgr6’s role in osteogenesis may only be partially mediated through the Lgr6-Rspo-axis that regulates cWnt signaling while significantly impacting Bmp signaling. In our working model (Supplementary Fig. 5), we propose that BMP treatment enhances Lgr6 expression and in the absence of Lgr6, mineralization is adversely affected in vitro cultures. How increased Lgr6 modulates BMP signaling to promote mineralization, whether it directly interacts with the BMP receptors or other membrane associated protein(s) to promote BMP signaling is not known. Therefore, understanding the signaling networks that govern osteogenesis through Lgr6’s influence could aid in developing novel and more effective therapies for orthopedic applications.

Supplementary Material

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bone.2024.117207.

Data availability

The scRNA-seq datasets analyzed for this study can be found in the Gene Expression Omnibus (GEO) GSE195940, GSE136970, and GSE138689. Bulk RNA-seq data generated for this study can be accessed via GSE235758.