GREM2 inactivation increases trabecular bone mass in mice

Osteoporosis is a common skeletal disease affecting millions of individuals world-wide, with an increased risk of fracture, and a decreased quality of life. Despite its well-known consequences, the etiology of osteoporosis and optimal treatment methods are not fully understood. Human genetic studies have identified genetic variants within the FMN2/GREM2 locus to be associated with trabecular volumetric bone mineral density (vBMD) and vertebral and forearm fractures, but not with cortical bone parameters. GREM2 is a bone morphogenetic protein (BMP) antagonist. In this study, we employed Grem2-deficient mice to investigate whether GREM2 serves as the plausible causal gene for the fracture signal at the FMN2/GREM2 locus. We observed that Grem2 is moderately expressed in bone tissue and particularly in osteoblasts. Complete Grem2 gene deletion impacted mouse survival and body growth. Partial Grem2 inactivation in Grem2+/− female mice led to increased trabecular BMD of femur and increased trabecular bone mass in tibia due to increased trabecular thickness, with an unchanged cortical thickness, as compared with wildtype littermates. Furthermore, Grem2 inactivation stimulated osteoblast differentiation, as evidenced by higher alkaline phosphatase (Alp), osteocalcin (Bglap), and osterix (Sp7) mRNA expression after BMP-2 stimulation in calvarial osteoblasts and osteoblasts from the long bones of Grem2−/− mice compared to wildtype littermates. These findings suggest that GREM2 is a possible target for novel osteoporotic treatments, to increase trabecular bone mass and prevent osteoporotic fractures.


Assessment of bone parameters
Dual X-ray absorptiometry Total body bone mineral density (BMD) was measured using the Lunar PIXImus densitometer (Wipro GE Healthcare).

High-resolution microcomputed tomography (µCT)
High-resolution microcomputed tomography (µCT) analyses were performed on proximal tibia and vertebra L5 using the 1172 µCT (Bruker MicroCT), as previously described 14 .The trabecular bone parameters were analyzed in the metaphyseal region of tibia, starting approximately 650 µm from the proximal growth plate, continuing for roughly 134 µm in distal direction.Cortical thickness was measured in the diaphyseal region of the tibia, starting approximately 5.2 mm from the growth plate, continuing for roughly 134 µm in distal direction.For trabecular parameters from vertebra L5, trabecular bone caudal of the pedicles were analyzed starting at a distance of approximately 4.5 µm caudal of the lower end of the pedicles, continuing in a longitudinal direction for 230 µm in the caudal direction.The data was analyzed using the CTAn software (Bruker MicroCT).

Peripheral quantitative computed tomography (pQCT)
Computed tomography (pQCT) analyses were performed using the XCT Research M (v.4.5B,Norland Stratec) at a resolution of 70 µm 15 .In order to measure trabecular volumetric BMD, the scans were positioned in the metaphysis of femur at a distance corresponding to 3.4% of the total femur length, proximal from the distal growth plate.The trabecular region was defined as the inner 45% of the total cross-sectional area 16 .

Bone histomorphometry
For dynamic histomorphometry, the mice were labeled intraperitoneally using the fluorochromes calcein (C0875, Merck GmbH) and alizarin (A3882, Merck GmbH) 8 and 1 day prior to termination, respectively.After dissection, tibia was fixated in formalin, dehydrated in ethanol, and imbedded in methyl methacrylate.Dynamic trabecular bone parameters were measured in unstained 8-µm-thick sections, and static trabecular parameters were determined in 4-µm-thick plastic sections stained in Masson-Goldner's Trichrome.All parameters were analyzed using the OsteoMeasure7 histomorphometry system (OsteoMetrics), according to the guidelines of the American Society for Bone and Mineral Research 17 .

Real-time quantitative PCR
Total RNA from cortical (shafts from flushed tibia) and trabecular-rich (vertebral body) bone, brain cortex, muscle gastrocnemius, brown fat, retroperitoneal fat, and gonadal fat was prepared using TRIzol Reagent (15596018, Thermo Fisher Scientific) followed by the RNeasy Mini Kit (74116, Qiagen).Total RNA from ovary, hypothalamus, lung, liver, aorta, kidney, spleen, uterus, and heart was prepared using the RNeasy Mini Kit (74116, Qiagen).The RNA was reversed transcribed into cDNA (4368814, Applied Biosystems) and real-time PCR analyses were performed using the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific).Grem2 expression was www.nature.com/scientificreports/measured using the following custom-made primer and probe sets located within the deleted region of exon 2: forward primer: 5′-TGG CTG TGC TGG TAA AGG TA-3′, reverse primer: 5′-TTG ATC TGG TGA TGC CAC CT-3′, and probe: FAM-5′-CGC CCG CAG GCC GGT TCT TC-3′.The following Assay-on-Demand primer and probe sets were used: Alp, Mm00475834_m1; Bglap (encoding osteocalcin), Mm03413826_mH; and Sp7 (encoding osterix), Mm04209856_m1.For Grem2 expression analyses in liver and gonadal fat of Grem2 +/− heterozygote and wildtype littermates, Mm00501909_m1 Assay-on-Demand primer and probe set was used.Relative gene expression was calculated by the 2 -∆∆Ct method using the expression of the 18S ribosomal subunit (4310893E, Thermo Fisher Scientific) as internal standard.

Primary osteoblast cell cultures from long bones
Primary osteoblast cell cultures from tibia from 12-week-old female Grem2 +/+ and Grem2 −/− mice were isolated by outgrowth of cells from collagenase-treated diaphyseal cortical bone, as previously described 19 .Primary bone cells from tibia were incubated in complete osteogenic αMEM, and the experiments were conducted as described for primary calvarial bone cells above.To study osteoblast differentiation, cells were treated with or without BMP-2 (100 ng/ml, 355-BEC, R&D Systems).For mineralization analyses, the cells were cultured for 10 days and then washed in PBS, fixed in 2.5% glutaric aldehyde in 70% ethanol, washed three times in 70% ethanol and air dried.Mineralized nodules were visualized by staining with 1% alizarin red in 50% ethanol for 5 min, followed by three washes in 50% ethanol and air dried.Whole well stitched images were taken using a Jenoptik Gryphax camera on a Nikon Eclipse 80i microscope.The percentage of the well surface covered by mineralized matrix was determined using the Bioquant Osteo software (version 2023 v23.5.60).

Primary osteoclast cell cultures
Bone marrow cells from wildtype mice were cultured in suspension culture discs (430591, Corning) in complete αMEM with 30 ng/ml macrophage-stimulating factor (M-CSF, 416-ML-050, R&D Systems), for 2 days.The adherent bone marrow macrophages (BMMs) were used as osteoclast progenitors 20,21 .The cells were detached and spot-seeded in 24-well plates (20,000 cells per well).The cells were cultured in complete αMEM with 30 ng/ ml M-CSF with or without the addition of 4 ng/ml receptor activator of nuclear factor κ-B ligand (RANKL, 462-TEC, R&D Systems) to induce osteoclast differentiation.After 3 days, the medium was changed, and 4 days after seeding, the cells were harvested for RNA preparation by lysis in RLT buffer with β-mercaptoethanol.

Statistics
Results are presented as mean ± standard error of the mean.Genotype ratios were tested against a Mendelian ratio (1:2:1) with a one sample chi-square goodness of fit test (SPSS, version 29.0.0.0).One-way ANOVA followed by Dunnett's multiple comparisons test (GraphPad Prism, version 10.0.3) was used to assess the effect of genotype on body weight between Grem2 −/− versus Grem2 +/+ and Grem2 +/− versus Grem2 +/+ mice.To calculate the difference in Grem2 expression over time in wildtype calvarial osteoblast, one-way ANOVA followed by Dunnett's multiple comparisons test (GraphPad Prism) was used, using the expression of day 1 as the reference.To evaluate the effect of genotype (Grem2 +/− and Grem2 +/+ ), age (10 or 12 weeks), and the interaction thereof, two-way ANOVA was used (GraphPad Prism).For analyses of primary osteoblast cell cultures, Student's t test was used to evaluate the difference in gene expression between two groups (Grem2 +/+ mice vs. Grem2 +/− mice, Grem2 +/+ cells vs. Grem2 −/− cells, or Ctrl vs. BMP-2).Two-way ANOVA (GraphPad Prism) was used to assess overall interaction between BMP-2 effect in Grem2 +/+ cells vs. BMP-2 effect in Grem2 −/− cells.Power analysis was performed to determine group sizes.The analysis suggested that the use of eleven Grem2 +/+ and eleven Grem2 +/− mice would give 80% power to detect a biological significant effect (where P < 0.05) with a 1.26 s.d.change in trabecular bone mass.We therefore aimed to use at least eleven mice per group in the in vivo studies.

Grem2 is moderately expressed in bone tissue
To study the expression pattern of Grem2, we compared the Grem2 expression in 16 tissues from mice (female, aged 12 weeks) and observed high expression in ovary, certain brain structures, such as hypothalamus and brain cortex, and lung, and moderate expression in cortical bone and vertebral body, rich in trabecular bone (Fig. 1a).Analyses of isolated calvarial cells, containing a large proportion of osteoblasts, indicated that cultured primary osteoblasts express Grem2, but the expression decreases with time in these cultures (Fig. 1b).No expression of Grem2 was present in cultured bone marrow macrophages (BMM) or RANKL-differentiated osteoclasts (OCL)  22 , publicly available at Single Cell Portal (https:// singl ecell.broad insti tute.org/ single_ cell/ study/ SCP13 37).Freshly isolated cells and cultured cells form separate clusters as indicated in the figure 22 .
Vol.:(0123456789) www.nature.com/scientificreports/(Fig. 1a).To further elucidate expression of Grem2 in bone cells, single cell RNA-sequencing (scRNAseq) datasets of freshly isolated primary mouse calvarial cells and cultured calvarial cells were used 22 .Analysis of the datasets demonstrated expression of Grem2 both in freshly isolated primary osteoblasts and in cultured osteoblasts from the calvaria (Fig. 1c, d).Grem2 was modestly expressed in osteoblasts in scRNAseq datasets of mouse bone marrow cells (Supplementary Fig. 1) 23 .FMN2 is expresssed in brain tissues but no expression of FMN2 was observed in freshly isolated primary mouse calvarial cells or in cultured calvarial cells (Supplementary Fig. 2).

Complete Grem2 gene deletion affects mouse survival and growth
To obtain Grem2-deficient mice, we bred Grem2 +/− female mice with Grem2 +/− male mice.We observed that Grem2 −/− mice were not born according to Mendel's law of inheritance with the expected Mendelian ratio (1:2:1).

Increased trabecular bone mass in female mice with partial Grem2 inactivation
As the Grem2 −/− mice born were severely growth restricted, it was not feasible to study their skeletal phenotype.Therefore, skeletal analyses on adult mice were performed only on the Grem2 +/− heterozygote and wildtype littermates.Grem2 mRNA expression analysis demonstrated a reduction in Grem2 expression in the trabecular-rich vertebral body (−42.8 ± 8.2%, P = 0.0023) in the heterozygote Grem2 +/− mice, compared with wildtype Grem2 +/+ littermates (Fig. 2a).A similar decrease in Grem2 expression was seen in liver and gonadal fat (Supplementary Fig. 4).

Discussion
The FMN2/GREM2 locus was previously recognized as a bone-related locus through GWAS 5 .This study aimed to determine whether GREM2 functions as the plausible causal gene underlying the fracture signal within the FMN2/GREM2 locus.Utilizing a globally Grem2-deficient mouse model in our study, we scrutinized the model's phenotype, revealing evidence that establishes GREM2 as the plausible causative gene for the fracture signal at the FMN2/GREM2 locus.Primary bone cell cultures derived from Grem2 −/− mice demonstrated a greater increase in Alp, Bglap, and Sp7 expression following BMP-2 stimulation compared to wildtype cells, indicating an increase in osteoblast differentiation as a result of the GREM2 inactivation.By using mRNA expression analyses and publicly available databases containing analyses from single cell RNA sequencing data, we demonstrated that Grem2 is expressed in several tissues and cell types including osteoblasts but not osteoclasts.These results are consistent with a previous report indicating expression of Grem2 in pre-osteoblasts of embryonic day 18.5 mouse calvaria 12 , and our previous study, demonstrating GREM2 expression in human osteoblasts 5 .The expression of Grem2 in calvarial bone cells decreased over time in osteogenic media.However, since the time course gene expression analyses were made in total RNA preparations it cannot be concluded if the expression per osteoblast is changed or if the relative number of osteoblasts in the cultures decrease compared to other cells in the primary cell cultures.Several studies have shown that the percentage of macrophages increase during in vitro culture of calvarial bone cells 22,24 .
The low number of born Grem2 −/− mice implies that GREM2 directly affects crucial developmental processes.This observation is in line with a previous indication from the International Mouse Phenotyping Consortium (IMPC), suggesting preweaning lethality of Grem2 −/− mice 25 .A previous study in zebrafish 26 demonstrated that loss of grem2 leads to overexpression of pitx2 and downregulation of lefty2, two crucial regulators of asymmetric cardiac development 27,28 .Moreover, grem2 knockdown in zebrafish increases the levels of activated phosphorylated Smad and BMP signaling within cardiac cells, interfering with both ventricular and atrial differentiation, indicating that GREM2 is required for pharyngeal mesoderm patterning, proper cardiac differentiation, and atrial chamber formation during cardiac development 26 .Thus, the low number of Grem2 knockout mice born may be attributed to the pivotal role of GREM2 in cardiac development.Furthermore, the few Grem2 −/− mice born in the present study were growth restricted, which might also be a consequence of impaired cardiac development.Given the growth restriction in Grem2 −/− mice, further skeletal analyses were focused on Grem2 +/− heterozygote mice and their wildtype littermates.
The Grem2 +/− mice appeared healthy, with no differences in body weight or long bone length compared to wildtype controls.Detailed skeletal analyses using DXA and CT scans revealed intriguing findings.In female mice, no differences were observed in total body areal bone mineral density (aBMD), but Grem2 +/− mice displayed an increase in trabecular BMD of femur and trabecular BV/TV of the tibia, attributed to an elevated trabecular number, whereas there was no change in cortical thickness.
As neither static nor dynamic histomorphometric analyses provided insights into the significance of GREM2 for the number of osteoclasts or osteblasts, or its impact on the bone formation process, possibly due to the establishment of a new equilibrium within the bone tissue in mice having a reduced expression of GREM2 from birth, additional in vitro studies were conducted.GREM2 is a BMP antagonist and we showed that stimulating calvarial osteoblasts and osteoblasts derived from the long bones with BMP-2, resulted in extensively enhanced expression   of Grem2.These findings confirm and extend previous findings from both pre-osteoblasts from embryonic mouse calvaria and human bone marrow-derived mesenchymal stem cells among others, demonstrating an upregulation of Grem2 expression by BMP-2 stimulation 12,[31][32][33] .These findings emphasize that expression of Grem2 is involved in a negative feedback mechanism.Importantly, the increase in expression of the osteogenic markers Alp, Bglap, and Sp7 was more pronounced in BMP-2 stimulated osteoblasts, from either calvaria or long bones, derived from Grem2-inactivated mice, further emphasizing the functional role of GREM2 as a BMP-2 antagonist.These results are in line with results from previous reports demonstrating that silencing Grem2 expression in cell cultures increases expression of Alp after BMP-2 stimulation, whereas overexpression of Grem2 in cell cultures decreases expression of Alp 12,32 .Mineralization studies on osteoblasts derived from long bones further confirmed the effect of Grem2 on osteoblast differentiation, as cells from mice lacking Grem2 had a higher percentage of mineralized surface after stimulation with BMP-2, when compared to cells from wildtype mice.Together with previous studies, our results demonstrate that suppression of Grem2 promotes osteoblast differentiation.The   mRNA expression levels of Bglap in primary calvarial bone cells (e) and primary cells from long bones (f), after being cultured in osteogenic media and stimulated with or without BMP-2 for 7 days.(g,h) mRNA expression levels of Sp7 in primary calvarial bone cells (g) and primary cells from long bones (h), after being cultured in osteogenic media and stimulated with or without BMP-2 for 7 days.(i) Alizarin red staining of mineralized nodules in osteoblast cultures of long bones from Grem2 +/+ (left) and Grem2 −/− (right) mice, after being cultured in osteogenic media and stimulated with BMP-2 for 10 days.Scale bar: 500 µm.(j) Quantification of mineralized nodules stained with alizarin red in osteoblast cultures of long bones from Grem2 +/+ and Grem2 −/− mice, after being cultured in osteogenic media and stimulated with or without BMP-2 for 10 days.Individual values are presented in all graphs with the mean presented as horizontal lines and ± standard error of the mean presented as vertical lines.Statistical analyses were performed using two-sided Student's t test when comparing two groups (Grem2 +/+ vs. Grem2 −/− , or Ctrl vs. BMP-2 treatment), and a mixed model two-way ANOVA was used to evaluate the overall interaction effect between BMP-2 treatment in Grem2 +/+ cells vs. BMP-2 treatment in Grem2 −/− cells.
single cell RNA sequencing data demonstrated no expression of FMN2 in either freshly isolated primary mouse calvarial cells or cultured calvarial cells.Furthermore, a recent report on large-scale whole exome sequencing data in the UK Biobank dataset revealed that missense variants in the GREM2 gene, but not in the FMN2 gene, are strongly associated with estimated BMD in the heel, supporting the notion that GREM2 and not FMN2 is crucial for bone mass regulation 29 .
In conclusion, we functionally show that a partial deletion of Grem2 in mice leads to an increase in trabecular bone mass of long bones, with unchanged cortical bone.By using two different primary osteoblast cell cultures, we report mechanistic data showing that Grem2-inactivated cells display enhanced response to BMP-2 stimulation on osteoblast differentiation, as evident by an augmented increase in Alp, Bglap, and Sp7 expression.We believe that GREM2, being a BMP antagonist, is a promising new target for osteoporosis treatment.Further studies are warranted to explore the impact of GREM2 on bone mass and strength in neonatal and aged mice, and utilizing cell-specific and inducible knockout mouse models will provide a more comprehensive understanding of GREM2's involvement in bone biology.

Figure 1 .
Figure 1.Grem2 is expressed in osteoblasts but not in osteoclasts in mice.(a) Grem2 mRNA expression pattern in various tissues of 12-week-old female, wildtype mice (n = 6), bone marrow macrophages (BMM), and osteoclasts (OCL).Data presented as % of the expression in trabecular-rich vertebral body and shown as individual values with the mean presented as horizontal lines and ± standard error of the mean as vertical lines.N.D. not detectable.(b) Grem2 mRNA expression in primary calvarial osteoblasts from wildtype mice decreases over time in culture.Data presented as % of the expression at day 1 and shown as individual values with the mean presented as horizontal lines and ± standard error of the mean as vertical lines.Difference in expression over time was analyzed using one-way ANOVA followed by Dunnett's multiple comparisons test with day 1 as the reference group.n = 4 per timepoint.The experiment was repeated three times.(c,d) Cluster analysis (c), and feature plots of Grem2 expression (d) in freshly isolated primary mouse calvarial cells and cultured calvarial cells published by Ayturk et al.22 , publicly available at Single Cell Portal (https:// singl ecell.broad insti tute.org/ single_ cell/ study/ SCP13 37).Freshly isolated cells and cultured cells form separate clusters as indicated in the figure22 .

Figure 3 .
Figure 3. Grem2 inactivation stimulates osteoblast differentiation in primary bone cells.(a,b) mRNA expression levels of Grem2 in primary calvarial bone cells (a) and primary cells from long bones (b), after being cultured in osteogenic media and stimulated with or without bone morphogenetic protein-2 (BMP-2) for 7 days.(c,d)mRNA expression levels of Alp in primary calvarial bone cells (c) and primary cells from long bones (d), after being cultured in osteogenic media and stimulated with or without BMP-2 for 7 days.(e,f) mRNA expression levels of Bglap in primary calvarial bone cells (e) and primary cells from long bones (f), after being cultured in osteogenic media and stimulated with or without BMP-2 for 7 days.(g,h) mRNA expression levels of Sp7 in primary calvarial bone cells (g) and primary cells from long bones (h), after being cultured in osteogenic media and stimulated with or without BMP-2 for 7 days.(i) Alizarin red staining of mineralized nodules in osteoblast cultures of long bones from Grem2 +/+ (left) and Grem2 −/− (right) mice, after being cultured in osteogenic media and stimulated with BMP-2 for 10 days.Scale bar: 500 µm.(j) Quantification of mineralized nodules stained with alizarin red in osteoblast cultures of long bones from Grem2 +/+ and Grem2 −/− mice, after being cultured in osteogenic media and stimulated with or without BMP-2 for 10 days.Individual values are presented in all graphs with the mean presented as horizontal lines and ± standard error of the mean presented as vertical lines.Statistical analyses were performed using two-sided Student's t test when comparing two groups (Grem2 +/+ vs. Grem2 −/− , or Ctrl vs. BMP-2 treatment), and a mixed model two-way ANOVA was used to evaluate the overall interaction effect between BMP-2 treatment in Grem2 +/+ cells vs. BMP-2 treatment in Grem2 −/− cells.

Table 2 .
Histomorphometry analyses of tibia in 12-week-old female mice.Bone histomorphometry analyses of trabecular bone in tibia in 12-week-old female Grem2 +/+ (n = 10) and Grem2 +/− (n = 11) mice.A two-tailed Student's t test was used to evaluate the difference between the groups.

Table 3 .
Bone parameters of 10-week-old male mice.Whole-body DXA analyses and µCT analyses of tibia in male 10-week-old mice.All values are given as mean ± standard error of the mean.Grem2 +/+ mice, n = 12; Grem2 +/− , n = 10.A two-tailed Student's t test was used to evaluate the difference between Grem2 +/+ and Grem2 +/− mice.