Male Lrp5^A214V mice maintain high bone mass during dietary calcium restriction by altering the vitamin D endocrine system

Abstract

Environmental factors and genetic variation individually impact bone. However, it is not clear how these factors interact to influence peak bone mass accrual. Here we tested whether genetically programmed high bone formation driven by missense mutations in the Lrp5 gene (Lrp5^A214V) altered the sensitivity of mice to an environment of inadequate dietary calcium (Ca) intake. Weanling male Lrp5^A214V mice and wildtype littermates (control) were fed AIN-93G diets with 0.125%, 0.25%, 0.5% (reference, basal), or 1% Ca from weaning until 12 weeks of age (ie, during bone growth). Urinary Ca, serum Ca, Ca regulatory hormones (PTH, 1,25 dihydroxyvitamin D₃ (1,25(OH)₂D₃)), bone parameters (μCT, ash), and renal/intestinal gene expression were analyzed. As expected, low dietary Ca intake negatively impacted bones and Lrp5^A214V mice had higher bone mass and ash content. Although bones of Lrp5^A214V mice have more matrix to mineralize, their bones were not more susceptible to low dietary Ca intake. In control mice, low dietary Ca intake exerted expected effects on serum Ca (decreased), PTH (increased), and 1,25(OH)₂D₃ (increased) as well as their downstream actions (ie, reducing urinary Ca, increasing markers of intestinal Ca absorption). In contrast, Lrp5^A214V mice had elevated serum Ca with a normal PTH response but a blunted 1,25(OH)₂D₃ response to low dietary Ca that was reflected in the renal 1,25(OH)₂D₃ producing/degrading enzymes, Cyp27b1 and Cyp24a1. Despite elevated serum Ca in Lrp5^A214V mice, urinary Ca was not elevated. Despite an abnormal serum 1,25(OH)₂D₃ response to low dietary Ca, intestinal markers of Ca absorption (Trpv6, S100g mRNA) were elevated in Lrp5^A214V mice and responded to low Ca intake. Collectively, our data indicate that the Lrp5^A214V mutation induces changes in Ca homeostasis that permit mice to retain more Ca and support their high bone mass phenotype.

Lay Summary

Optimizing peak bone mass (PBM) is critical for strong bones and osteoporosis prevention. Both genetics and dietary factors like calcium (Ca) contribute to PBM. The goal of this research study was to determine how dietary Ca intake and genetics interact with each other to impact bone mass. Lowering dietary Ca in control mice causes hormonal changes that increase intestinal Ca absorption and reduce urinary Ca loss to protect bone; but this process fails when dietary Ca becomes too low. However, mice with genetically programmed high bone mass could maintain high bone mass even when challenged with Ca deficient diets. This protection is because the high bone mass mice maintain higher serum Ca, have altered production and utilization of Ca-regulating hormones, and have increased molecular indicators of intestinal Ca absorption and kidney Ca retention. Our findings are important because they demonstrate how a genetic program that increases bone formation can drive improved efficiency of Ca utilization to accommodate the increased need for Ca deposition into bone. We believe that our preclinical study provides important proof-of-principle support for the concept of personalized recommendations for bone health management.

Graphical Abstract

Introduction

Osteoporosis is a metabolic bone disease characterized by low bone mass that increases bone fragility and fracture risk. It has a complex etiology that can be prevented by optimizing peak bone mass (PBM) during growth⁽¹⁾ and reducing the rate of adult bone loss.⁽²⁾ PBM accrual is impacted by both genetics^(3,4) and environmental factors like diet.⁽⁵⁾ For example, low calcium (Ca) intake is associated with low bone mineral density (BMD) in humans⁽⁶⁾ and mice^(7,8) while common variants in the Lrp5 gene reduce BMD in the femur and lumbar spine.⁽⁹⁾

There is growing evidence that genetics interact with dietary Ca intake to influence Ca and bone metabolism, that is, gene x environment (diet) (GxD) interactions exist. For example, while Caucasian girls do not increase Ca absorption in response to low dietary Ca intake, and thus have a high dietary Ca requirement (1300 mg/d⁽¹⁰⁾), Asian–American⁽¹¹⁾ and African–American girls⁽¹²⁾ adapt well to low Ca intake and have lower estimated dietary Ca requirements (<1000 mg/d). Consistent with these human studies, our previous work in 11 inbred mouse lines⁽⁷⁾ and BXD recombinant inbred mice^(13-15) showed that the impact of genetics on serum Ca- and phosphorus-regulating hormones, intestinal Ca absorption, and bone phenotypes can be modulated by dietary Ca intake levels. While we have identified genetic loci controlling these GxD interactions and we have identified potential candidate genes within these loci, the mechanism by which dietary Ca intake and genetic variation interact to influence PBM attainment is not clear.

Figure 1 depicts possible gene-by-diet (GxD) interaction models and their skeletal outcomes. Figure 1A shows a model where gene and diet have independent effects on the accrual of PBM. Alternatively, Figure 1B shows a model where the benefit of genetic variation may only be recognized in the presence of the proper dietary environment. In this case, diet does not define bone acquisition, but limits whether an individual can meet his or her genetic potential for high bone mass. A final model is shown in Figure 1C. In this case, genetically programmed high bone mass might protect an individual from the consequences of low dietary Ca intake (Figure 1C) by rapidly removing Ca from the blood and increasing serum PTH and 1,25 dihydroxyvitamin D₃ (1,25(OH)₂D₃) levels to promote intestinal Ca absorption and renal Ca reabsorption, for example, similar to what normally happens during low dietary Ca intake.⁽¹⁶⁾ We tested these alternatives using growing male mice with genetically programed high bone mass and challenging them with diets containing different levels of Ca. Genetically programmed high bone mass was modeled using knock-in mice (ie, Lrp5^A214V) that have a loss-of-function A214V mutation in the LRP5 gene that causes high bone mass in humans.⁽¹⁷⁾ This mutation renders the LRP5 receptor resistant to antagonists (eg, sclerostin) leading to hyperactivation of osteogenic Wnt signaling.⁽¹⁸⁾ Our data indicate that genetically driven high bone mass reprograms Ca metabolism to allow mice to maintain high PBM even in the face of deficient dietary Ca intake.

Figure 1.

Models of gene x environment interactions. (A) Independent action of genetics and dietary calcium (Ca) on the phenotype (eg, bone mass), (B, C) models depicting a response to diet that is conditional upon the genotype. In (B), the benefit of genotype depends upon adequate dietary Ca. In (C), genetics makes the individual insensitive to changes in dietary Ca. D = diet, G = genotype, HBM = high bone mass genotype.

Materials and methods

Experimental design

To model a genetic propensity for developing high bone mass through increased bone formation, we used mice carrying a high bone mass-inducing, human missense A214V mutation knocked into exon 3 of the Lrp5 gene (Lrp5^A214V/A214V).⁽¹⁹⁾ These mice were provided to us by Dr. Alexander G Robling (Indiana University School of Medicine). Lrp5^A214V/A214V mice were on a mixed 129S1/SvIMJ and C57BL/6J genetic background. After crossing Lrp5^A214V/A214V mutant mice with C57BL/6J mice, the heterozygous Lrp5 knock-in mice on the mixed genetic background (Lrp5^A214V/wt) were used as breeders to obtain homozygous male Lrp5^A214V/A214V (Lrp5^A214V) and Lrp5^wt/wt (control) mice for experiments. To control for the mixed genetic background, we used littermates as matched controls for the Lrp5^A214V mice in our experiments. Animals were genotyped for the Lrp5 wild type and knock-in gene A214V alleles using the following conditions: Primers (10 nM): (F) AGTACTGGCTGGCACAGA, (R) CAGGCTGCCCTTGCAGAT (WT = 250 bp, homozygote knock-in = 400 bp); 94°C 3 min; 34 cycles: 94°C 30 sec/60°C 30 sec/72°C 1 min; 72°C 10 min; GoTaq polymerase (Promega) and 10 nM dNTP. Bands for genotypes were resolved on 2% agarose gels.

Male Lrp5^A214V and control mice were randomly assigned to purified AIN93G diets containing 0.4 % P and 1000 IU vitamin D/kg and either 0.125%, 0.25%, 0.5%, or 1% Ca (Research Diets, New Brunswick, NJ) from weaning until 12 weeks of age (n = 8–11 per group). The 0.5% level of Ca is the rodent requirement for calcium⁽²⁰⁾ and so this is the reference control diet. We chose to use 12 weeks of age to study PBM because others have shown that trabecular bone in distal femur peaks at 2 months of age and cortical bone in the femur midshaft peaks around 3 months of age.^(21,22) Dietary Ca levels were chosen to meet the dietary requirement for rodents (0.5% Ca) or to modulate an adaptive physiological response.⁽¹⁶⁾ Mice were grouped-housed (4/cage) on TEK-fresh bedding (Harlan Laboratories, Indianapolis, IN) in conventional shoebox cages at the Purdue University Animal Facilities. Food and water were fed ad libitum and mice were maintained in an UVB light-free environment on a 12 h-light/dark cycle under standard conditions of temperature and humidity. Mice were monitored for their health daily and were fasted overnight prior to the termination of the experiment. Mice were anesthetized with ketamine (22 mg/mL) and xylazine (33 mg/mL) (0.1 mL/20 g body weight, IP injection) after which they were bled by cardiac puncture to obtain serum samples for hormone analysis. Urine samples for Ca analysis were collected directly from the bladder. Duodenal scrapings from the first 2 cm segments and minced kidneys were collected into TriReagent (Molecular Research Center, Inc., Cincinnati, OH) and frozen in liquid nitrogen. Femora from mice were harvested and prepared for imaging analysis as previously described.⁽²³⁾ Femur length (FL) was recorded using a digital caliper (Mitutoyo America Corporation, Aurora, IL). A summary of the experimental design and study outcomes is presented in Supplementary Figure S1. The animal protocol was approved by the Institutional Animal Care and Use Committee at Purdue University.

Gene expression

RNA was isolated using TriReagent according to the manufacturer’s directions. The isolated RNA was reverse transcribed into cDNA as we have previously described.⁽²⁴⁾ Reverse transcription polymerase chain reaction (RT-PCR) was conducted using the MyiQ RT-PCR system containing SYBR green (Bio-Rad, Hercules, CA) and the primers and PCR conditions as previously reported for RPLP0, TRPV5, CaBPD_28k, Trpv6, S100g_, Cyp27b1, and Cyp24a1.^(25,26) We used the delta–delta-Ct method to calculate the relative mRNA level of samples and we normalized our data based on the expression of the housekeeping gene RPLP0.

Serum and urinary analysis

Serum and urinary Ca and urinary creatinine concentrations were measured using the QuantiChrom™ Ca and creatinine assay kits, respectively (BioAssay Systems, Hayward, CA). Serum 1,25(OH)₂D₃ was measured by radioimmunoassay (ImmunoDiagnostic Systems, Fountain Hills, AZ) and intact PTH was measured by an enzyme-linked immunosorbent assay (Immutopics Inc, San Clemente, CA).

Bone imaging

Femurs were fixed and scanned by micro-computed tomography (μCT 40, Scanco Medical AG, Bassersdorf, Switzerland) using our published methods.⁽¹⁵⁾ ISQ files obtained from the Scanco imager were analyzed by CTAn (version 1.21.1, Bruker). Briefly, the region of interest (ROI) for trabecular bone in the distal femur was set to scan 94 slices (1.5 mm) and 1 mm of trabecular bone was analyzed starting from the first slice containing no evidence of the metaphyseal growth plate or primary spongiosa. For cortical bone, the ROI was set at 50% of the length of the bone (midshaft) where 0.24 mm was scanned and analyzed.

Ash weight and Ca content

Femurs were weighed and then dried in an oven at 42°C for 48 h to determine the dry weight. Dried femora were then ashed in the furnace at 300°C for 24 h followed by 600°C for 5 days. Ash weight was recorded and expressed as a percent of dry weight (% ash). The ash was dissolved in 70% trace element free nitric acid. The sample was diluted with deionized water and further with 0.5% lanthanum chloride for analysis of Ca content using atomic absorption spectrometry (AAnalyst 300, Perkin Elmer, Waltham, MA).

Statistical analysis

The sample size for each treatment group was calculated using variance estimates from our published data and α = 0.05 and 1-β = 0.8. This analysis indicated that with n = 8 mice per group, we could detect a 30% difference between groups for bone μCT phenotypes and a 50% difference in serum hormone levels and gene expression. All experimental results represent the data collected from individual mice. All the samples for analysis were processed by block randomization; investigators assessing the experimental outcomes were blinded to the genotype and dietary treatment.

We used SAS Enterprise Guide 8.3 (SAS Institute Inc., Cary, NC) for the statistical analyses. While several outliers were identified as values whose z-score was in the extreme 2.5% tails of the line/diet group or whole population distribution, sample values were removed only when there is an additional reason, for example, sample values were outside an ELISA/EIA standard curve, evidence of technical error, broken bone. The exact number of samples used for each analysis is reported in Supplementary Table S1. For bone phenotypes, the covariate effect of body weight and/or FL was determined by Pearson’s correlation coefficients and, when significant, removed by linear regression.⁽²⁷⁾ Most cortical and trabecular bone parameters were affected by FL and analysis of covariance (ANCOVA) was performed on these parameters. Trabecular separation (Tb.Sp), femur dry bone weight, and femur ash weight were not correlated to FL and analysis of variance (ANOVA) was used on these parameters. A normal distribution of the residuals was determined by assessing Q-Q plots, Predicted vs Residual plots, and with the Shapiro–Wilk test. Equality of variance was tested by assessing Predicted vs Residual plots and by Brown–Forsythe for ANOVA in raw data and ANCOVA in residual data. The following transformations were made: BV/TV (log10), PTH (natural log), Urinary Ca/Creatinine ratio (square root), renal Cyp24a1 mRNA and duodenal S100g (quad root), renal Cyp27b1 mRNA, renal TRPV5 mRNA, (natural log), and renal S100g (square root) (Supplementary Table S1). Two-way ANOVA or ANCOVA was used to test for significant main effects (genotype, diet) and a GxD interaction; the P-values for main effects and the interaction are reported within each figure and in Supplementary Table S1. When significant diet or GxD interactions were observed, the Tukey–Kramer post-hoc test was used to determine differences among the groups. To test for associations between parameters, Pearson’s correlations were performed in Origin 2022, (Northampton, MA, USA). Main effect, interaction, and pairwise comparisons were considered significant when P < 0.05.

Data were plotted using Origin 2022 (Northampton, MA, USA) as dot plots superimposed with box plots with the central box spanning the 25^th and 75^th percentiles, the central line representing the median, and the whiskers representing the 5^th and 95^th percentiles. Supplementary Table S1 contains the mean ± SEM for all phenotypes as well as summary statistical tables from ANOVA/ANCOVA.

Results

PBM accrual in Lrp5^A214V mice was elevated even in the presence of dietary Ca restriction

All mice were healthy throughout the study and their body weight was not affected by the dietary treatments. Lrp5^A214V mice were smaller but their femora were longer than controls (Supplementary Table S1). Consistent with other reports,^(19,28) Lrp5^A214V mice had a high bone mass phenotype, reflected by statistically significant genotype effects for all of the bone parameters we measured, that is, for trabecular bone: higher BV/TV, Tb.N, Tb.Th, and lower Tb.Sp than control mice; for midshaft cortical bone, higher Ct.Th, Ct.Ar, Tt.Ar, and Ct.Ar/Tt.Ar than control mice. Bone mineral levels were also higher in Lrp5^A214V mice as were dry bone weight, ash weight, and percent ash (Figure 2, Supplementary Table S1).

Figure 2.

Analysis femur from 12-week-old wild-type (control) and homozygous Lrp5^A214V mutant mice fed different levels of calcium from weaning. (A.1) cortical thickness (Ct.Th) (A.2) cortical bone area (Ct.Ar) (B.1) trabecular bone volume fraction (BV/TV); (B.2) trabecular separation (Tb.Sp); (C.1.) femur ash weight (C.2.) femur ash %. 2-way ANOVA (Tb.Sp., ash weight, % ash) and ANCOVA (Ct.Th, Ct.Ar, BV/TV) results for main effects and interactions are depicted in the tables inset within each graph. n = 7–10 mice per diet and genotype group. D = diet, G = genotype, G*D = gene-by-diet.

Regardless of genotype, low dietary Ca intake had a significant negative effect on all trabecular bone parameters, Ct.Ar, Ct.Th, femur dry weight, ash weight, and percent ash (Figure 2, Supplementary Table S1). The low Ca diet did not affect Tt.Ar., Ct.Ar/Tt.Ar, or FL. Despite the negative impact of low Ca intake in both genotype groups, the Lrp5^A214V mice fed the deficient, 0.125% Ca diet had improved phenotypes for most bone endpoints compared with control mice fed the 1% Ca diet, suggesting the Lrp5 mutation protects bone mass even when Ca is lacking (Supplementary Table S1).

We did not observe a statistically significant GxD interaction for any of the bone phenotypes (Supplementary Table S1, Figure 2). As such, our data do not support the hypothesis the potential of the high bone mass genotypes are significantly dependent upon the level of dietary Ca intake.

Lrp5^A214V mice had an altered vitamin D metabolic response to low Ca diet

Serum Ca levels were significantly higher in Lrp5^A214V mice and they fell significantly in response to the low dietary Ca in both groups (Figure 3A). Both genotype groups exhibited a similar correlation between serum Ca and PTH (Control: r = −0.42, P = 0.04; Lrp5^A214V: r = −0.46, P = 0.024; Supplementary Figure S2). Despite the higher serum Ca levels, neither fasting serum PTH nor 1,25(OH)₂D₃ levels were suppressed in Lrp5^A214V mice (Figure 3B, C). In contrast, we detected a significant genotype-by-diet interaction affecting serum 1,25(OH)₂D₃, but not serum PTH. While Ca restriction increased serum 1,25(OH)₂D₃ levels 3-fold in control mice, this rise was absent in Lrp5^A214V mice, indicating that 1,25(OH)₂D₃ synthesis in response to elevated PTH levels resulting from low dietary Ca intake was abnormal in these mice (Figure 3C).

Figure 3.

Serum calcium (Ca) and Ca regulating hormone levels in wild-type (control) and homozygous Lrp5^A214V mutant mice fed diets with different Ca levels from weaning until 12 weeks of age. (A) Serum Ca, (B) PTH, and (C) 1,25(OH)₂D₃. 2-way ANOVA of main effects and interactions are depicted in the tables inset within each graph. n = 6–10 mice per diet and genotype group. In panel (C), groups with common letter superscripts are not significantly different from one another (Tukey–Kramer test, P < 0.05). D = diet, G = genotype, G*D = gene-by-diet.

Lrp5^A214V mice had altered molecular markers of renal 1,25(OH)₂D₃ production and degradation

Similar to what has been shown previously,⁽²⁹⁾ in control mice renal Cyp27b1 mRNA levels progressively increased upon feeding low Ca diets and renal Cyp24a1 mRNA levels increased only on the highest Ca diet (Figure 4, Supplementary Table S1). Renal Cyp27b1 gene expression was higher in Lrp5^A214V than control mice (Supplementary Table S1). In addition, renal Cyp27b1 mRNA fell in Lrp5^A214V mice only when Ca was fed at the highest level (Figure 4A). For renal Cyp24a1 gene expression there were significant main effects for genotype and diet Ca (Figure 4B, Supplementary Table S1). Serum PTH level was positively correlated with Cyp27b1 mRNA (r = 0.62, P = 0.001) and negatively associated with Cyp24a1 mRNA (r = −0.57, P = 0.003) in control mice (Figure 4C, D). In contrast, no significant relationships were found between the serum PTH and either renal Cyp24a1 or Cyp27b1 mRNA levels in Lrp5^A214V mice. This suggests that PTH-mediated actions are changed in Lrp5^A214V mice, which in turn impacts the regulation of serum 1,25(OH)₂D₃ by low dietary Ca intake.

Figure 4.

Renal vitamin D metabolizing enzyme gene expression in wild-type (control) and homozygous Lrp5^A214V mutant mice fed diets with different Ca content from weaning until 12 weeks of age. (A) renal Cyp27b1 mRNA and (B) Cyp24a1 mRNA expressed as arbitrary units (AU). 2-way ANOVA of main effects and interactions are depicted in the tables inset within each graph. n = 7–9 mice per diet and genotype group. (C) the natural log of Cyp27b1 mRNA or (D) quad root (qr) of Cyp24a1 mRNA were correlated with the natural log (ln) of serum PTH (independent variable). P = P-value; r = Pearson’s r. Blue squares depict data from individual control mice and red circles depict data from individual Lrp5^A214V mice. D = diet, G = genotype, G*D = gene-by-diet.

Lrp5^A214V mice have elevated molecular markers of renal Ca reabsorption and intestinal Ca absorption

Hormonal adaptation to low dietary Ca intake induces molecular events that increase intestinal Ca absorption and reduce renal Ca excretion. As expected, the urinary Ca/creatinine ratio was reduced in control mice on low Ca diets (Figure 5A) and this was accompanied by increases in renal S100g mRNA levels (Figure 5B), an intracellular Ca transport protein whose gene expression is increased by 1,25(OH)₂D₃.⁽¹⁶⁾ Despite having elevated serum Ca levels that are normally correlated with urinary Ca excretion,⁽³⁰⁾ urinary Ca/creatinine levels were not elevated in Lrp5^A214V mice, and they had a similar response to changes in dietary Ca as control mice (Figure 5A). Renal S100g gene expression was higher in Lrp5^A214V mice (+40% vs Control) but Trpv5 and Calb1 mRNA levels were not (Supplementary Table S1). In addition, there was a GxD interaction affecting S100g mRNA, that is, renal S100g mRNA was suppressed by 0.5% dietary Ca intake in control mice but was not reduced in Lrp5^A214V mice until they were fed the 1% Ca diet (Figure 5B). Consistent with a role for S100g in renal Ca reabsorption, we found that renal S100g mRNA expression was inversely correlated with urinary Ca excretion in both control (r = −0.45, P < 0.012) and Lrp5^A214V (r = −0.62, P < 0.002) mice (Supplementary Figure S3). This indicates that Ca reabsorption machinery was working as expected in both genotype groups.

Figure 5.

Renal calcium (Ca) handling phenotypes in wild-type (control) and homozygous Lrp5^A214V mutant mice fed diets with different Ca content from weaning until 12 weeks of age. (A) Urinary Ca/creatinine ratio and (B) renal S100g mRNA levels expressed as arbitrary units (AU). 2-way ANOVA of main effects and interactions are depicted in the tables inset within each graph. n = 5–11 mice per diet and genotype group. In panel (B), groups with common letter superscripts are not significantly different from one another (Tukey–Kramer test, P < 0.05). D = diet, G = genotype, G*D = gene-by-diet.

In control mice, duodenal expression of the Ca transport proteins Trpv6 and S100g was increased by low dietary Ca feeding (Figure 6, Supplementary Table S1), consistent with the increase in serum 1,25(OH)₂D₃ seen as dietary Ca levels fell (Figure 3). However, despite the observation that serum 1,25(OH)₂D₃ levels were not altered by changes in dietary Ca intake in Lrp5^A214V mice (Figure 3), the expression levels of Trpv6 and S100g were upregulated by dietary Ca restriction in Lrp5^A214V mice (Figure 6). In addition, Trpv6 and S100g mRNA levels in Lrp5^A214V mice were significantly higher compared with control mice (Figure 6A, 6B), suggesting that Lrp5^A214V mice were absorbing more Ca from the intestine. We found that in control mice, the circulating levels of 1,25(OH)₂D₃ were positively and significantly correlated with Trpv6 mRNA (r = 0.45, P = 0.03, Supplementary Figure S4) In contrast, this relationship was not observed in the Lrp5^A214V mice (Trpv6, r = −0.014, P = 0.95). This suggests that the regulation of intestinal Trpv6 gene expression is uncoupled from serum 1,25(OH)₂D₃ levels in Lrp5^A214V mice.

Figure 6.

Molecular markers for intestinal calcium (Ca) absorption in wild-type (control) and homozygous Lrp5^A214V mutant mice fed diets with different Ca content from weaning until 12 weeks of age. (A) Duodenal Trpv6 and (B) S100g mRNA levels. 2-way ANOVA of main effects and interactions are depicted in the tables inset within each graph. n = 7–10 mice per diet and genotype group. D = diet, G = genotype, G*D = gene-by-diet.

Discussion

The most striking finding of this study is that despite the critical need for adequate dietary Ca to support mineralization of newly synthesized matrix,^(31-33) Lrp5^A214V mice with genetically programmed high bone formation maintain their phenotype even when dietary Ca was severely limiting. Very low Ca intake (0.125% Ca diet or 25% of the rodent Ca requirement) reduced bone and ash weight relative to 1% Ca diet, yet femur and ash weight were still dramatically higher in Lrp5^A214V mice fed the low Ca diet. This indicates that Lrp5^A214V mice are tremendously efficient as utilizing Ca for bone growth and mineralization.

In the classical model, physiologic adaptation to low diet Ca intake occurs to maintain serum Ca by increasing the production of hormones that increase the efficiency of intestinal Ca absorption (1,25(OH)₂D₃) and renal Ca reabsorption (PTH, 1,25(OH)₂D₃).⁽³⁴⁾ However, the adaptive increases in Ca absorption and reabsorption efficiency have limits, so extended periods of low Ca intake induces osteoclastic bone resorption, and the maintenance of serum Ca occurs at the expense of bone.^(34,35) Our observation that restricting dietary Ca intake regulates Ca and bone metabolism in control mice is consistent with this classical model. Our a priori hypothesis, built upon the classical model, was that very low Ca diets would limit the genetic potential of Lrp5^A214V mice because Ca intake would be insufficient to maintain the serum Ca levels necessary for the mineralization of bone (see the model in Figure 1B). However, our study revealed a model like Figure 1A where the impact of genetics and diet on bone are independent. This suggests that the genetics driving bone formation in Lrp5^A214V mice alters whole body Ca metabolism to permit more bone mineralization and PBM accrual. This raises the question “How can the Lrp5^A214V-driven genetic programming for high bone mass overcome inadequate Ca intake?” Our data suggest this is because Lrp5^A214V mice maintain serum Ca levels that are modestly, but not pathologically, elevated compared with control mice (10.01 ± 0.10 vs 9.59 ± 0.09 mg/dL, respectively, across all diets). This is observed even in the face of very low dietary Ca intake, where the Lrp5^A214V mice have serum Ca levels (9.74 ± 0.14 mg/dL) that are comparable with those seen in control mice fed the 1% Ca diet (9.79 ± 0.08 mg/dL) (Figure 3A).

In addition to the elevated serum Ca levels in Lrp5^A214V mice, several other aspects of Ca metabolism did not function in these mice as predicted in the classical model. This includes the observation that high serum Ca levels did not suppress serum PTH and that diet-induced changes in serum PTH did not increase serum 1,25(OH)₂D₃ levels. Our data suggest that one point of dysregulation in the adaptive system may be in the PTH mediated expression of the renal vitamin D metabolizing enzymes. Research by Meyer et al.⁽³⁶⁾ shows that PTH signaling modulates CREB-mediated transcription to stimulate renal Cyp27b1 gene expression and inhibit renal Cyp24a1 gene expression. We observed that in Lrp5^A214V mice the expression of Cyp27b1 mRNA is higher and Cyp24a1 mRNA is lower than would be expected by the serum PTH levels (Figure 4, Supplementary Table S1). This suggests that PTH mediated regulation of these genes is enhanced in Lrp5^A214V mice. However, neither Cyp27b1 nor Cyp24a1 mRNA responded to the increasing PTH levels induced by the low Ca diets in Lrp5^A214V mice (Figure 4C, D). The loss of a response to low Ca intake suggests the PTH mediated regulation of CREB transcriptional activity may be saturated in Lrp5^A214V mice. Additional studies are necessary to clarify how PTH-regulated, CREB-mediated gene regulation is modified in Lrp5^A214V mice.

Our data also suggest that traditional vitamin D target tissues, the small intestine and kidney, may be more responsive to 1,25(OH)₂D₃ in the Lrp5^A214V mouse and that this improves calcium utilization and supports increased bone mass. We noted that the mRNA level of vitamin D-target genes in the duodenum (Trpv6, S100g, Figure 4A-B) were significantly higher in Lrp5^A214V mice than in the control mice. In addition, they were induced by low Ca intake even though serum 1,25(OH)₂D₃ levels were not elevated nor regulated by dietary Ca in Lrp5^A214V mice. We and others have reported that Trpv6 and S100g mRNA levels are positively correlated to Ca absorption efficiency.^(7,16) Thus, the enhanced expression of these genes suggests that Ca absorption is higher in Lrp5^A214V compared with control mice.

Our data also suggest that renal Ca handling is altered in Lrp5^A214V mice. Urinary Ca is normally elevated when intestinal Ca absorption is high⁽³⁷⁾ or when serum Ca levels are elevated⁽³⁸⁾ but this was not observed in Lrp5^A214V mice (Figure 5). This suggests that renal Ca reabsorption may be higher in Lrp5^A214V mice. Consistent with this, we found that Lrp5^A214V mice have elevated renal mRNA levels for S100g, a vitamin D-responsive gene⁽¹⁶⁾ whose cellular levels reflect transcellular Ca fluxes.⁽²⁴⁾ Collectively, elevated intestinal Ca absorption coupled with increased renal Ca reabsorption in Lrp5^A214V mice could explain why serum Ca remains high in these mice, even during dietary Ca restriction.

The mechanism accounting for the impact of Lrp5^A214V on whole body Ca metabolism is not clear. Wnt signaling is an important regulator of intestinal stem cell function that supports cell proliferation⁽³⁹⁾ but there is very little data to suggest that Lrp5 regulates Ca metabolism through effects on intestinal cell function. Thus, while deletion of both Lrp5 and Lrp6 from intestinal epithelial cells reduced proliferation and caused premature differentiation of crypt-base stem cells,⁽⁴⁰⁾ mice with single Lrp5 or Lrp6 gene deletions were phenotypically normal⁽⁴⁰⁾ and intestinal epithelial cell-specific Lrp5^A214V or Lrp5^G171V gene expression did not affect trabecular bone mass.⁽¹⁹⁾ Gene variants in the Lrp5 gene have been associated with autosomal dominant polycystic kidney disease (ADPKD).⁽⁴¹⁾ Cells from ADPKD cysts have low resting cytosolic Ca, indicating a possible association of Lrp5 signaling mutations in the kidney and intracellular Ca signaling that might influence whole body Ca homeostasis.⁽⁴²⁾ However, a role for renal Lrp5 in Ca handling by the kidney has not yet been reported.

Another factor that might contribute to our observations is FGF23, a hormone that is classically linked to the control of phosphate metabolism and can suppress the production of both PTH and 1,25(OH)₂D₃.⁽⁴³⁾ FGF23 can regulate Ca metabolism,⁽⁴⁴⁾ increase renal Ca reabsorption,⁽⁴⁵⁾ and suppress bone mineralization through inhibitory effects on bone alkaline phosphatase expression.⁽⁴⁶⁾ Despite these emerging roles for FGF23 in Ca metabolism, our observation of reduced renal 1,25(OH)₂D₃ production with normal PTH levels and elevated mineralization in Lrp5^A214V mice are inconsistent with both elevated FGF23 (which would lower 1,25(OH)₂D₃ and PTH and inhibit mineralization) or reduced FGF23 (which would increase 1,25(OH)₂D₃ and PTH and promote mineralization). As a result of these inconsistencies, we did not measure FGF23 in this study.

The primary strength of our study is that we could carefully control the environment of mice with the Lrp5^A214V mutation that causes a high bone mass phenotype in humans.⁽²⁸⁾ This allowed us to directly test how Ca intake regulates genetically programmed high bone mass. The use of animal models also allows us to measure endpoints that could not be measured in humans, for example, tissue mRNA levels, bone microarchitecture, and bone mineralization. Similar studies are not possible in human subjects for practical and ethical reasons. Despite the strength of our study design and our outcomes, there were several weaknesses in our study. First, we did not measure whether higher vitamin D receptor (VDR) levels in intestine and kidney of Lrp5^A214V mice could account for the increased tissue sensitivity to 1,25(OH)₂D₃ we observed_. However, while VDR gene deletion impairs vitamin D signaling in the intestine and kidney and disrupts Ca/bone metabolism,^(26,47) we recently reported that elevated VDR does not make the intestine more responsive to 1,25(OH)₂D₃.⁽⁴⁸⁾ Second, we only conducted a complete analysis on male mice because of our concerns regarding low trabecular bone mass in female mice⁽²²⁾ as well a sex differences in vitamin D action we previously reported,⁽⁴⁹⁾ However, in a pilot study we found that the bones of female Lrp5^A214V mice are similarly protected from the consequences of feeding a very low Ca diet (data not shown). Nonetheless, since others have reported sex differences for the effects of human Lrp5 mutations on bone,^(19,50) sex differences should be addressed in future studies. Third, we only examined one type of genetically programmed high bone mass so future studies using mouse models where high bone mass is driven by mutations affecting osteoblast/osteocyte activity (eg, other LRP receptors, SMAD9, ANKH) or suppressing osteoclast function (eg, OSTM1, PLEKHM1) will be needed to test the generalizability of our findings.⁽⁵¹⁾ Fourth, limitations on blood volume prevented us from measuring serum ionized calcium and bone biomarkers of formation or resorption. However, the total calcium levels we measured are generally well-correlated to serum ionized calcium.⁽⁵²⁾ Also, the Lrp5^A214V mutation is known to promote osteoblast-mediated bone formation⁽¹⁹⁾ and low Ca diets increase bone resorption without harming bone formation in growing rodents.⁽³⁵⁾ Thus, serum bone formation/resorption markers would not alter interpretation of our data.

This is the first study to evaluate the physiologic responses of Lrp5^A214V mice with genetically programmed high bone mass to dietary Ca restriction, that is, GxD interactions. We found that although Lrp5^A214V mice were sensitive to dietary changes, their genetic programming allowed them to develop very high bone mass even under the physiologic stress of very low Ca intake. This was due to unexpected and interesting alterations in vitamin D and calcium metabolism in Lrp5^A214Vmice that suggest the increased bone mass may provide a physiologic signal for improved efficiency of calcium utilization. These findings add to our understanding of how genetic variants controlling bone metabolism and different dietary environments interact to modulate the susceptibility of individuals to bone loss.

Author contributions

Serra Ucer Ozgurel (Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing), Perla C Reyes Fernandez (Data curation, Formal analysis, Writing—original draft, Writing—review and editing), Krittikan Chanpaisaeng (Formal analysis, Writing—review and editing), and James C Fleet (Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing—original draft, Writing—review and editing)

Funding

This work was supported by National Institutes of Health (grant nos. ES019103 and DK118036), and UT start-up funds to J.C.F.

Conflicts of interest

None declared.

Data availability

The data that support the findings of this study are available from the corresponding author upon request.