Intestinal Vitamin D Receptor Is Dispensable for Maintaining Adult Bone Mass in Mice With Adequate Calcium Intake

Abstract

1,25-Dihydroxyvitamin D₃ (1,25(OH)₂D₃)-mediated intestinal calcium (Ca) absorption supplies Ca for proper bone mineralization during growth. We tested whether vitamin D receptor (VDR)-mediated 1,25(OH)₂D₃ signaling is critical for adult Ca absorption and bone by using mice with inducible Vdr gene knockout in the whole intestine (villin-CreER^T2+/− × Vdr^f/f, WIK) or in the large intestine (Cdx2-CreER^T2+/− ×Vdr^f/f, LIK). At 4-month-old, Vdr alleles were recombined (0.05 mg tamoxifen/g BW, intraperitoneally [i.p.], 5 days) and mice were fed diets with either 0.5% (adequate) or 0.2% (low) Ca. Ca absorption was examined after 2 weeks while serum 1,25(OH)₂D₃, bone mass, and bone microarchitecture were examined after 16 weeks. Intestinal and renal gene expression was measured at both time points (n = 12/genotype/diet/time point). On the 0.5% Ca diet, all phenotypes in WIK and LIK mice were similar to the controls. Control mice adapted to the 0.2% low-Ca diet by increasing renal Cyp27b1 mRNA (3-fold), serum 1,25(OH)₂D₃ level (1.9-fold), and Ca absorption in the duodenum (Dd, + 131%) and proximal colon (PCo, + 28.9%), which prevented bone loss. In WIK mice, low-Ca diet increased serum 1,25(OH)₂D₃ (4.4-fold) but Ca absorption remained unaltered in the Dd and PCo. Consequently, significant bone loss occurred in WIK mice (e.g., cortical thickness, Ct.Th, −33.7%). LIK mice adapted to the low-Ca diet in the Dd but not the PCo, and the effect on bone phenotypes was milder (e.g., Ct.Th, −13.1%). Our data suggest intestinal VDR in adult mice prevents bone loss under low Ca intake but is dispensable under adequate calcium intake.

Osteoporosis is a global public health concern characterized by low bone mass and bone structural deterioration, causing bone fragility and increased risk for fracture at multiple bone sites (1). Increasing peak bone mass (PBM) and reducing adult bone loss are key strategies for osteoporosis prevention (2). Intestinal calcium (Ca) absorption has been positively associated with higher PBM during growth (3, 4) and reduced bone loss in adults (5). Higher intestinal Ca absorption can also lower the hip fracture risk (−2.5-fold) for older women with inadequate dietary intake (6).

Intestinal Ca absorption is the sum of 2 processes, an energy-dependent saturable route that is subject to physiological and nutritional regulation and a nonsaturable route that is a linear function of the luminal Ca concentration (7). The major regulator of saturable intestinal Ca absorption is 1,25-dihydroxyvitamin D₃ (1,25(OH)₂D₃) signaling. Under low dietary Ca intake, renal 1,25(OH)₂D₃ production and serum 1,25(OH)₂D₃ levels are elevated and this increases intestinal Ca absorption efficiency during growth in animal models (8, 9) and humans (10). The action of 1,25(OH)₂D₃ on intestinal Ca absorption is mediated via the vitamin D receptor (VDR) by regulating the expression of genes that control Ca fluxes through the enterocytes (8, 11, 12). When the Vdr gene is deleted from the mouse, duodenal Ca absorption is decreased by 47% leading to the hypocalcemia, hyperparathyroidism, osteomalacia, and growth arrest (13). We have previously shown that intestine-specific transgenic expression of the VDR in Vdr knockout mice (KO/TG) restored intestinal Ca absorption and reversed the abnormal Ca metabolism observed in Vdr KO mice (13). These findings are direct proof that the regulation of intestinal Ca absorption is the single most important role for 1,25(OH)₂D₃ and VDR signaling during growth.

In addition to the proximal small intestine, accumulating evidence suggests that the lower bowel also plays an important role in Ca metabolism and bone health. For example, Ca absorption was significantly lower in patients with Crohn disease whose colon was removed than in patients with a functional colon (14). Similarly, Ca homeostasis was disrupted and bone microstructure was compromised when the large intestine of 8-week-old rats was surgically removed (15, 16). In addition, kinetic studies have revealed that 1,25(OH)₂D₃ regulates Ca absorption in the cecum and colon of rats (17‐19). Furthermore, we recently showed that when a Vdr transgene was expressed in the distal intestine of Vdr KO mice, it rescued the abnormalities in Ca metabolism and bone phenotypes normally seen in these mice (20).

Compared to the well-defined role of 1,25(OH)₂D₃ and VDR signaling in intestinal Ca absorption and bone health during growth, their significance in adult intestinal Ca absorption and adult bone health has been a matter of debate. As individuals mature and PBM has been attained, the need for bone Ca deposition drops substantially (21). Coincident with these changes, intestinal Ca absorption efficiency is reduced during adulthood both in humans and in animal models (22‐26) and the ability of individuals to adapt to low Ca intake is also reduced (27). Interestingly, adult patients with hereditary vitamin D–resistant rickets (HVDRR) can maintain normal serum Ca levels and maintain bone mass without the high dose of Ca supplementation normally needed to protect bone health in these patients during childhood (28, 29). Collectively, the evidence suggests that 1,25(OH)₂D₃ and VDR may not be important for intestinal Ca absorption and bone health in the adult. However, this research question has not been formally tested. Hence, the goal of our study was to determine the critical role of intestinal VDR on Ca and bone metabolism under both adequate Ca intake and in response to dietary Ca restriction.

Material and Methods

Animals

All of the animal experiments were approved by the Purdue University Animal Care and Use Committee. Mice were housed in groups of 3 to 5 mice and exposed to a 12-hour light/12-hour dark cycle. Food and water were given ad libitum.

Generation of Mice With Whole-Intestine– or Large-Intestine–Specific Vdr Gene Deletion

Heterozygous mice containing loxP sites flanking exon 4 of the Vdr gene (Vdr^f/w) were a gift from Dr Gardner (University of California San Francisco) (30). Mice with inducible Vdr gene deletion in the entire intestinal epithelium (whole-intestine KO, WIK) were made by crossing VDR^f/f mice and villin promoter-Cre-ER^T2 mice (31) to create Villin-CreER^T2+/−×Vdr^f/f mice. Mice with inducible Vdr deletion in distal ileum, cecum, and proximal colon (large-intestine KO, LIK) were generated by crossing Vdr^f/f mice with CDX2 promoter-Cre-ER^T2+/− mice (32) (Jackson Laboratory; stock No. 022390) to create Cdx2-CreER^T2+/−×Vdr^f/f mice. Littermates with the VDR floxed alleles but no Cre-ER^T2 transgene were used as controls. Mouse genotypes were confirmed using genomic DNA isolated from tail clipping with the DNeasy Blood & Tissue Kits (QIAGEN). Vdr genotyping was performed using primers and conditions described elsewhere (30). Cre genotyping of heterozygous mice was conducted as we have previously described (33).

Confirmation of Vdr Deletion in Whole-Intestine Knockout and Large-Intestine Knockout Mice

WIK, LIK, and control mice (n = 8/genotype, balanced by sex) were fed a modified AIN93G diet (0.5% Ca [wgt/wgt × 100], 0.4% P, 200 IU VD₃/kg diet, Research Diets Inc) from weaning. At age 4 months all mice received tamoxifen (0.05 mg/g body weight, intraperitoneal [i.p.] injection for 5 days) to induce recombination of the floxed VDR alleles. Two weeks after tamoxifen treatment, we collected kidney and mucosal scrapings from the duodenum (Dd), distal ileum, cecum, and proximal colon (PCo) for DNA isolation and Vdr genotyping. Assessment of Vdr gene recombination was performed using polymerase chain reaction (PCR) primers and conditions described elsewhere (30). We also assessed the effect of Vdr gene deletion in WIK and LIK mice on intestinal Vdr messenger RNA (mRNA) level using mucosal scrapings from mouse Dd and PCo of mice 16 weeks after tamoxifen treatment. Real-time (RT)-PCR was conducted using mouse Vdr primers (Integrated DNA Technologies): sense 5'-TCAACGCTATGACCTGTGAAG-3' and antisense 5'-CCGGTTGTCCTTGGTGATG-3'.

Experimental Design

WIK, LIK, and control mice (n = 48/genotype, balanced by sex) were fed a modified AIN93G diet (0.5% Ca [wgt/wgt ×100], 0.4% P, 200 IU VD₃/kg diet, Research Diets Inc) from weaning. At age 4 months, mice have reached a plateau for bone Ca accrual and they are considered growth stable, mature adults (21). At this point, all mice received tamoxifen (0.05 mg/g body weight, i.p. injection for 5 days) to induce recombination of the floxed VDR alleles. Mice were then randomly assigned to a modified AIN93M diet (0.4% P and 200 IU VD₃) with either 0.5% (adequate) or 0.2% (low) Ca (n = 24/genotype/diet). After 2 weeks on the experimental diets, we used 12 mice/genotype/diet (6 male, 6 female) to assess the short-term effect of diet on: Ca absorption efficiency in the Dd and the PCo using in situ ligated loops procedure and Cyp27b1 gene expression in kidney. After 16 weeks on the experimental diets, we used the remaining 12 mice/genotype/diet to evaluate the long-term effect of diet on serum 1,25(OH)₂D₃ level, Cyp27b1 gene expression in the kidney, Trpv6 and S100g gene expression in the Dd and PCo (the same region where Ca absorption was assessed in the short-term study); bone mineral content (BMC) and density (BMD), and bone microarchitecture.

Calcium Absorption Test

We assessed the capacity to absorb Ca in the Dd and PCo within the same mouse using an in situ ligated loop procedure with modifications to the procedure we have previously described (34). The Ca-absorption test was initiated in the proximal colon by making a small opening at 3 cm distal from the cecocolic junction, through which we removed the fecal remnants in the lumen by gently messaging the loop. Then we tied 2 ligatures at 0.5 and 2.5 cm distal from the cecocolic junction, respectively. ⁴⁵Ca buffer (transport buffer containing 2 mM Ca and 0.001 mCi/mL ⁴⁵Ca, Perkin Elmer) was then injected into the lumen of the ligated segment of proximal colon. At this concentration of Ca, active Ca transport predominates in the intestine (35, 36). While the proximal colon absorption test was ongoing, the same procedure was repeated at the Dd (2-cm Dd segment starting 0.5 cm after the pyloric sphincter). The only exception is that we did not make a cut to remove lumen contents because the Dd did not contain digesta after fasting overnight. At the end of the 10-minute absorption period, each loop was removed and the loop length measured. Then each segment was placed into a 20-mL glass scintillation vial, digested, and the radioactivity in the digesta was measured as we have described elsewhere (34). Efficiency of Ca absorption during the 10-minute incubation period was calculated by measuring the disappearance of ⁴⁵Ca from the loop, that is, [1 – (amount of Ca remaining in the loop/amount of Ca injected into the loop)] × 100. We took the following precautions to ensure the quality of our Ca absorption test. First, mice were kept warm and intestinal loops were kept moist throughout the absorption period. Second, we carefully observed and recorded the appearance and coloring of each loop (e.g., disruption of the circulation leads to a purple loop). When a loop had an abnormal appearance, the data from that loop were not used in the final analysis. Finally, we assessed radioactivity leakage from each loop with a Geiger counter after completion of the absorption period and removal of the loop. When leakage was detected in the body cavity after removal of the loop, the data from that mouse were not used in the final analysis.

Gene Expression

RNA from intestine mucosal scrapings and kidney were isolated using the Zymo Direct-zol RNA kit (Zymo Research) according to the manufacturer's directions. RNA was reverse transcribed to complementary DNA using a reaction cocktail and PCR conditions described elsewhere (34). Afterward, quantitative real-time PCR (qPCR) was conducted using CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Primers (Integrated DNA Technologies) and qPCR conditions that we have previously described (34). Fold change of gene expression for Cyp27b1, Trpv6, and S100g were determined using the 2^–ΔΔCT method with normalization relative to Rplp0 (37).

Serum 1,25(OH)₂D₃ Analysis

Serum 1,25(OH)₂D₃ was measured by enzyme immunoassay following the manufacturer's instructions (Immuno Diagnostic Systems, catalog No. AC-62F1).

Bone Analysis

On harvest, the right leg femur was harvested and thoroughly cleaned to remove muscle and connective tissue. Femur samples were then fixed in 10% neutral buffer formalin and stored at 4 °C for 7 days. Afterward, femur samples were rinsed with and stored in 70% ethanol at 4 °C until scanning. Scanning was performed using a PIXImus densitometer (Lunar; GE-Healthcare) to yield BMC (g) and BMD (g/cm²). We also scanned femur using microcomputed tomography (µCT 40, Scanco Medical) at the midshaft and distal metaphysis to yield cortical and trabecular bone microarchitecture parameters, correspondingly (38). µCT settings were reported elsewhere (39, 40) with the following exceptions. Images were obtained using a cubic voxel size of 8 µm. For all samples, the regions of interest (ROIs) were binarized using a global threshold (310.6 mg HA/cm³ for trabecular bone and 505.7 mg HA/cm³ for cortical bone). Cortical bone ROI in the femur midshaft was at 50% of the length of the bone where 60 slices were scanned and reconstructed. We reported total area (Tt.Ar, mm²), cortical bone area (Ct.Ar, mm²), cortical area fraction (Ct.Ar/Tt.Ar, %), and cortical bone thickness (Ct.Th, mm). Trabecular bone ROI in the distal femur was defined as 112 slices starting from the first slice containing no evidence of distal growth plate. We manually contoured trabecular bone every 10 slices with the reference contour drawn 2 to 3 pixels away from the endocortical surface. Intermediate slices were interpolated with Scanco contouring algorithm to generate volume of interest (38). We reported bone volume fraction (BV/TV, %), trabecular number (Tb.N, 1/mm), trabecular thickness (Tb.Th, mm), and trabecular separation (Tb.Sp, mm).

Statistical Analysis

The statistical analysis was conducted in SAS Enterprise Guide v8.2 (SAS Institute Inc). Data were examined for outliers through the outlier test (z score < or > 2 SD). Ca absorption data were further excluded based on these criteria: a) intestine loop appeared purple; b) radioactivity leakage. After removing outliers, data were analyzed for normal distribution and for the presence of potential covariates (ie, body weight, intestine segment length, femur length). Data that were not normally distributed were transformed as follows: Vdr mRNA (16-week Dd and PCo) (natural log); serum 1,25(OH)₂D₃ level ($\sqrt{y}$); Ca absorption efficiency (2-week Dd), Cyp27b1 mRNA (16-week kidney), S100g mRNA (16-week Dd and PCo), and BV/TV (log 10); Ct.Th and Ct.Ar/Tt.Ar ($y^2$); Ct.Ar ($y^3)$. Main effects (diet, genotype, sex) and their interactions were determined by analysis of variance (ANOVA), or when covariates affected a parameter, analysis of covariance (ANCOVA). When the F statistic in the 2-way ANOVA/ANCOVA test was significant, post hoc pairwise comparisons were conducted with the Tukey-Kramer test. Differences were considered statistically significant when P was less than .05.

Results

Confirmation of Intestine-Specific Vdr Gene Deletion

As expected, tamoxifen treatment caused complete Vdr gene deletion along the whole intestine in WIK mice (Fig. 1A). In comparison, LIK mice had limited Vdr deletion in the distal ileum, complete deletion in the cecum, and partial deletion in the PCo (Fig. 1B). Tamoxifen treatment did not alter the Vdr gene in the kidney of either model. Consistent with the genotyping results, Dd Vdr mRNA level was not altered in the Dd of LIK mice (Fig. 2A) but PCo Vdr mRNA level was reduced by 50% (P < .0001) (Fig. 2B). Vdr mRNA was completely eliminated in the Dd and PCo of WIK mice (P < .0001) (see Fig. 2).

Figure 1.

Generation of intestine-specific Vdr gene knockout mice. Ethidium bromide–stained Vdr allele genotyping gels. A, Four-month-old whole-intestine VDR knockout (WIK) and B, large-intestine VDR knockout (LIK) mice received tamoxifen-treatment (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Two weeks later mucosal scrapings from cecum, proximal colon (P. Colon), distal ileum (D. Ileum), and duodenum and kidney were collected for DNA extraction. Vdr genotyping yields a 1.5-kb band corresponding to the Vdr floxed allele and a 1.0-kb band corresponding to the recombined/deleted allele.

Figure 2.

Vdr gene expression in the mouse A, duodenum and B, proximal colon. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. Then, 16 weeks later, RNA was isolated from the duodenum (Dd) and proximal colon (PCo) and examined for Vdr messenger RNA (mRNA) level by real-time polymerase chain reaction. AU, arbitrary units. Analysis of variance was conducted on natural log-transformed data without covariates. Data derived from each genotype and diet group are presented as box plots (n = 10-12/group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test. Bars with different letter superscripts are statistically significantly different from each other with P less than .0001.

Effect of Vdr Gene Deletion in Adult Mice After 2 Weeks of Dietary Calcium Intervention

After 2 weeks on the adequate-Ca diet, WIK and LIK mice had similar levels of kidney Cyp27b1 gene expression as the Ctrl mice (Fig. 3). Feeding the low-Ca diet caused a significant increase in kidney Cyp27b1 gene expression in the Ctrl (+ 3.2-fold), WIK (+ 2.2-fold), and LIK (+ 4.6-fold) mice. Ca absorption efficiency in the Dd or PCo of the WIK and LIK mice was similar to Ctrl mice when they were fed the adequate-Ca diet (Fig. 4). In Ctrl mice, the low-Ca diet induced adaptive increase in Ca absorption efficiency both in the Dd (+ 131%; P = .011) and PCo (+ 29%; P = .049). In contrast, Ca absorption efficiency was not increased by feeding the low-Ca diet in either the Dd or PCo of WIK mice, while in LIK mice the low-Ca diet increased Ca absorption in the Dd (+66%; P = .083) but not in the PCo (+5%, ns; see Fig. 4).

Figure 3.

Renal Cyp27b1 gene expression in response to short-term low dietary Ca intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 2 weeks, RNA was isolated from the kidney (Kd) and examined for Cyp27b1 messenger RNA (mRNA) by real-time polymerase chain reaction. AU, arbitrary units. Analysis of variance was conducted on nontransformed data without covariates. Data derived from each genotype and diet group are presented as box plots (n = 10-12/group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated.

Figure 4.

Intestinal calcium (Ca) absorption efficiency in response to short-term low dietary Ca intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 2 weeks, Ca absorption efficiency (%) was simultaneously assessed at A, duodenum (Dd) and B, proximal colon (PCo) in the same mouse using the in situ ligated loops (2-cm segments, 2 mM Ca, 10 minutes). For Dd, analysis of variance was conducted on log10-transformed data without covariates. For PCo, analysis of covariance was conducted on nontransformed data with body weight and sex as covariates. Data derived from each genotype and diet group are presented as box plots (n = 10-12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated (ns = not significant).

Effect of Vdr Gene Deletion in Adult Mice After 16 Weeks of Dietary Calcium Intervention

As we reported for the short-term dietary Ca intervention, WIK and LIK mice on the adequate-Ca diet had similar levels of kidney Cyp27b1 gene expression and serum 1,25(OH)₂D₃ as the Ctrl mice after 16 weeks (Fig. 5). The low-Ca diet increased renal Cyp27b1 mRNA (+ 2.8-fold) and serum 1,25(OH)₂D₃ level (+ 1.9-fold) in Ctrl mice. Similarly, LIK mice adapted to the low-Ca diet by increasing their kidney Cyp27b1 gene expression (+ 4.5-fold; P = .0005) and serum 1,25(OH)₂D₃ level (+ 2.5-fold; P = .005). However, in WIK mice, long-term feeding of the low-Ca diet caused a more dramatic increase in renal Cyp27b1 mRNA (+ 12.2-fold; P < .0001) and serum 1,25(OH)₂D₃ (+ 4.4-fold; P < .0001) levels, suggesting intestinal resistance to 1,25(OH)₂D₃ action due to the lack of intestinal VDR.

Figure 5.

Renal Cyp27b1 gene expression and serum 1,25-dihydroxyvitamin D₃ (1,25(OH)₂D₃) level in response to long-term low dietary calcium (Ca) intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 16 weeks, A, renal Cyp27b1 messenger RNA (mRNA) levels were assessed by real-time polymerase chain reaction and B, serum 1,25(OH)₂D₃ level by enzyme immunoassay. AU, arbitrary units. For Cyp27b1 mRNA, analysis of covariance (ANCOVA) was conducted on log10-transformed data without covariates. For PCo, ANCOVA was conducted on square-root–transformed data without covariates. Data derived from each genotype and diet group are presented as box plots (n = 10-12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated. The value for 0.2% Ca-fed WIK mice was higher than all other groups (P < .0001 for Kd Cyp27b1, P < .005 for serum 1,25(OH)₂D₃).

On the adequate-Ca diet there was no difference in S100g mRNA level across different genotype groups in either the Dd or PCo (Fig. 6). When fed the low-Ca diet, Ctrl mice had increased S100g mRNA levels both in the Dd (+ 3.5-fold) and PCo (+ 2.6-fold). In contrast, the low-Ca diet feeding did not increase S100g mRNA in either Dd or PCo of WIK mice (see Fig. 6). In LIK mice, the low-Ca diet significantly increased duodenal S100g mRNA (+ 3.7-fold) but did not affect PCo S100g mRNA (Fig. 6). Similar data were also observed for Trpv6 mRNA expression (data not shown).

Figure 6.

Intestinal gene expression in response to long-term low dietary calcium (Ca) intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/1 g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 16 weeks, RNA was isolated from the A and C, duodenum (Dd) and B and D, proximal colon (PCo) and examined for A and B, S100g and C and D, Trpv6 messenger RNA (mRNA) by real-time polymerase chain reaction. AU, arbitrary units. Log10- (Dd, PCo S100g, Dd Trpv6) or Cube-root transformed (PCo Trpv6) data were analyzed by analysis of variance (Dd S100g, PCo S100g, Trvp6) or analysis of covariance (PCo S100g, body weight covariate). Data derived from each genotype and diet group are presented as box plots (n = 10-12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated (ns = not significant).

All bone phenotypes were similar across Ctrl, WIK, and LIK mice fed the adequate-Ca diet. Long-term feeding of the low-Ca diet did not alter their BMD or BMC (Fig. 7), nor did it cause any loss of cortical (Fig. 8) or trabecular bone (Fig. 9) in Ctrl mice. These data indicate that physiologic adaptation to inadequate Ca intake was sufficient to prevent adult bone loss in mice with normal intestinal VDR levels. In WIK mice, the long-term low-Ca diet significantly reduced BMC and BMD (−28.2% and −23.6%; P < .0001 for both) (see Fig. 7), several cortical bone µCT parameters (Ct.Ar/Tt.Ar, Ct.Ar and Ct.Th by −27.3, −28.7%, and −33.7%, respectively; P < .0001 for all) (see Fig. 8), and 3 of the trabecular bone µCT parameters (BV/TV, Tb.N, and Tb.Th, −50.9%, −16.3%, and −19.2%, respectively; P < .05 for all) (see Fig. 9). In comparison, the low-Ca diet had a milder, but still statistically significant, bone phenotype on LIK mice than WIK: loss of BMC and BMD (−12% for both), loss of cortical bone in Ct.Ar (−10.2%), Ct.Th (−13.1%), and Ct.Ar/Tt.Ar. (−11.1%) (P < .05 for all) (see Fig. 8).

Figure 7.

Bone mineral content (BMC) and density (BMD) in response to long-term low dietary calcium (Ca) intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 16 weeks, the right leg femur was scanned using a PIXImus densitometer for A, BMC and B, BMD. Analysis of covariance was conducted on nontransformed data with covariates (BMD, femur length; BMC, femur length, body weight). Data derived from each genotype and diet group are presented as box plots (n = 10-12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated (ns = not significant). The value for 0.2% Ca-fed WIK mice was lower than all other groups (P < .0001 for BMC, P < .0005 for BMD).

Figure 8.

Cortical bone microarchitecture in response to long-term low dietary calcium (Ca) intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 16 weeks, the right leg femur was scanned at the midshaft using microcomputed tomography for A, total cross-sectional area (Tt.Ar); B, cortical bone area (Ct.Ar); C, cortical area fraction (Ct.Ar/Tt.Ar); and D, cortical thickness (Ct.Th). Analysis of covariance was conducted on nontransformed (Tt.Ar) or transformed (Ct.Th, Ct.Ar, Ct.Ar/Tt.Ar) data with femur length, and body weight (Ct.Th, Tt.Ar, Ct.Ar) or sex (Tt.Ar, Ct.Ar/Tt.Ar) as covariates. Data derived from each genotype and diet group are presented as box plots (n = 12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated (ns = not significant). The values of Ct.Th and Ct.Ar for 0.2% Ca-fed WIK mice were lower than all other groups (P < .05).

Figure 9.

Trabecular bone microarchitecture in response to long-term low dietary calcium (Ca) intake. Four-month-old control (Ctrl), whole-intestine Vdr knockout (WIK), and large-intestine Vdr knockout (LIK) mice received tamoxifen (0.05 mg/g BW, intraperitoneally, 5 days) to induce recombination of the floxed Vdr alleles. Afterward, mice were randomly assigned to an AIN93M diet with either 0.5% or 0.2% Ca. After 16 weeks, the right leg femur was scanned at the distal metaphysis using microcomputed tomography for A, bone volume fraction (BV/TV); B, trabecular number (Tb.N); C, trabecular thickness (Tb.Th); and D, trabecular separation (Tb.Sp). Analysis of covariance was conducted on nontransformed (Tb.N, Tb.Sp, Tb.Th) or transformed (BV/TV) data with femur length (Tb.Th), and body weight (all parameters) or sex (BV/TV, Tb.N, Tb.Sp) as covariates. Data derived from each genotype and diet group are presented as box plots (n = 12 per group) with individual data points represented by ▴ = males, ● = females, and the median represented by a line. Multiple comparisons were made by the Tukey-Kramer test with P values for the effect diet within each genotype group as indicated (ns = not significant).

Discussion

Using growing animal models, it has been well established that consuming a low-Ca diet stimulates a physiologic adaptation of the 1,25(OH)₂D₃–parathyroid hormone (PTH) axis to increase the efficiency of intestinal Ca absorption (8, 41). In humans, there is a positive correlation between serum 1,25(OH)₂D₃ levels and fractional Ca absorption both in children (42) and adults (23). However, there is conflicting evidence on whether elevated serum 1,25(OH)₂D₃ levels are necessary for the upregulation of Ca absorption efficiency that results from feeding a low-Ca diet to adult humans, with some studies showing no increases in serum 1,25(OH)₂D₃ (43, 44) and others showing increases that were dependent on race (ie, higher response to low-Ca diet in Black vs White women) (45). This may reflect the effect of aging, which is accompanied by a loss of the adaptive increases in renal 1α hydroxylase activity after dietary Ca deprivation (27, 46, 47) that may be due to the loss of insulin-like growth factor-1 following growth (48). Nonetheless, these studies highlight the need to determine whether basal Ca absorption, or the increased Ca absorption efficiency that results from dietary Ca restriction, is dependent on 1,25(OH)₂D₃ signaling through the intestinal VDR in growth-stable adults.

Our data clearly show that when adult mice have adequate dietary Ca intake, intestinal VDR is not required for normal Ca metabolism and bone health, and they rely predominantly on a VDR-independent Ca absorption mechanism to meet their Ca requirement. Our finding is consistent with clinical observations in patients with HVDRR. Normally, children with HVDRR require treatment with high doses of oral and/or intravenous Ca to attain normal PBM and prevent hypocalcemia; this bypasses the saturable Ca absorption route that depends on a functional VDR in the intestine (28). However, once they are into adulthood, these individuals experience disease remission characterized by normalization of serum Ca, Ca absorption efficiency comparable to age-matched controls, and healing of rickets without the need for high-Ca treatment (28, 29, 49‐51).

Our findings on intestine-specific Vdr gene deletion in adult mice extend what we have previously reported in growing mice with global Vdr gene deletion (13, 20, 26) or large-intestine–specific Vdr gene deletion using Cdx2-Cre transgenic mice (52) and what others have shown with whole-intestine–specific Vdr gene deletion using Villin-Cre transgenic mice (53). In these earlier studies, Vdr gene deletion was present from birth and mice were studied when they were rapidly growing mice and required large amounts of Ca for bone mineralization. As such, global Vdr KO mice fed diets with normal Ca levels had low Ca absorption leading to hypocalcemia, high serum levels of PTH and 1,25(OH)₂ D, reduced growth, and severe osteomalacia (13, 20, 26). Similar but less severe effects were seen in whole-intestine Vdr KO mice (53) (ie, no growth arrest but low Ca absorption, elevated serum PTH and 1,25(OH)₂ D, and large losses in bone mass and microarchitecture). Bone was not affected in growing large-intestine–specific Vdr KO mice because of compensatory upregulation of Ca absorption in the small intestine (52). Duodenal Ca absorption in the 4-month-old control mice fed an adequate Ca diet (14.2 ± 3.08%) was similar to what we previously observed in 90-day-old mice (18.0 ± 0.9%), and this is significantly lower than duodenal Ca absorption in rapidly growing mice (55.4 ± 3.8% for 60-day-old) (26). However, whereas lifelong, global Vdr deletion and whole-intestine–specific Vdr deletion significantly reduced Ca absorption in growing mice fed Ca-adequate diets, inducible, whole intestine-epithelial cell deletion of Vdr in growth stable 4-month-old adult mice did not. Thus, in contrast to the effects in growing mice with Vdr gene deletion, adult WIK mice on Ca-adequate diets had normal serum 1,25(OH)₂D₃ level, renal Cyp27b1 expression, Ca absorption, intestinal S100g mRNA levels, and bone mass and microarchitecture. This suggests two things. First, the multitissue effect of global Vdr gene deletion beginning in utero is more severe than intestine-specific deletion of Vdr in adults who have experienced healthy growth (e.g., even on a high-Ca rescue diet Vdr KO mice are smaller than wild-type mice). In addition, our data indicate a minimal role for 1,25(OH)₂D₃/VDR signaling in adult intestinal Ca absorption and bone health when the dietary Ca intake requirement is met.

A large difference in Ca and bone metabolism between adulthood and growth is that adults have a reduced Ca requirement because bone Ca deposition rates fall significantly after the rapid growth period (21, 54‐56). Thus, in adulthood, humans and animals may rely more on a 1,25(OH)₂D₃-independent, nonsaturable (presumably paracellular) Ca absorption pathway. However, paracellular transport is estimated to occur at a rate of just 13% per hour of the luminal load (∼ 2.2% in our 10-minute absorption period) (57), and we found that the efficiency of Ca absorption in adult mice fed the adequate-Ca diet was significantly higher than this. Although the VDR-independent intestinal Ca absorption mechanism in adult mice remains uncertain, some evidence suggests that other hormones may be contributors. For example, estrogen induced intestinal Ca transport and Trpv6 mRNA levels in Vdr KO and vitamin D–deficient animals (58‐60)

In contrast to the 1,25(OH)₂D₃ independence of adult Ca absorption when intake is adequate, we show that intestinal VDR is critical for the adaptive upregulation of intestinal Ca absorption that occurs following dietary Ca restriction. In control mice, renal 1,25(OH)₂D₃ synthesis and serum 1,25(OH)₂D₃ level were elevated by the low-Ca diet and this led to an increase in intestinal S100g gene expression and Ca absorption both in the Dd and PCo. This adaptation was sufficient to prevent significant loss of bone mass and microarchitecture in the control mice. In contrast, the loss of functional VDR in the whole intestine of WIK mice eliminated 1,25(OH)₂D₃/VDR-mediated intestinal adaptive responses. Consequently, the long-term feeding of a low-Ca diet to WIK mice caused significant harm to bone mass (−23.6% BMD), cortical bone (eg, −27.3% Ct.Ar/Tt.Ar), and trabecular bone (eg, −51% in BV/TV) (see Figs. 7-9). This is likely due to the bone-resorptive effects of elevated serum 1,25(OH)₂D₃ levels (and presumably PTH, the driver of renal Cyp27b1 gene transcription) in WIK mice.

Our data also clarify the physiologic significance of 1,25(OH)₂D₃-mediated Ca absorption in the proximal colon. Dhawan et al (20) previously demonstrated that transgenic expression of VDR restricted to the distal intestine was sufficient to recover the abnormal Ca metabolism and bone phenotypes of Vdr KO mice, indicating the potential of the distal intestine to contribute to Ca homeostasis. However, we previously showed that despite a significant reduction of Trpv6 and S100g mRNA levels (indicators of Ca absorption) in the PCo of growing mice with large-intestine specific VDR deletion, bone was not strongly affected in these mice, even when dietary Ca intake was low (52). This suggests that during growth, Ca absorption in other intestinal segments can compensate for the loss of absorptive capacity in the large intestine. Our current data show that when VDR is deleted from the lower bowel of adult mice and they are fed a low-Ca diet, there was a significant decline in femoral BMC and BMD (−12%; see Fig. 7) and in several cortical bone microarchitecture phenotypes (eg, −13.1% Ct.Th; see Fig. 8). While VDR deletion in the whole intestine resulted in more bone loss (eg, −33.7% Ct.Th) than that in the large intestine, the effect of large-intestine Vdr gene deletion was still substantial and accounted for approximately 30% of the Ct.Th loss seen from the whole-intestine Vdr gene deletion. This critical role of the large intestine in adult bone health is likely due to the prolonged residence time of dietary Ca in the lower bowel (61).

The strengths of the present study include the in vivo measurement of Ca absorption efficiency both in the Dd and the PCo on the same mouse simultaneously; careful evaluation of the quality of our data using several criteria (eg, health of the intestine loops, statistical outlier tests); and the assessment of the short-term as well as the long-term effect of Vdr deletion in the whole intestine and the large intestine among adult mice. We also included male and female mice, making our findings translational to both sexes. Last, we addressed our research goals by conducting our experiment in skeletally mature adult mice (21). There are also some weaknesses. For example, our LIK mice model has only partial Vdr gene deletion in the proximal colon (see Fig. 1). This is consistent with the mosaic expression pattern of Cre in the Cdx2P-CreER^T2 mouse, in which Cre-mediated deletion occurs in some, but not all cells of the PCo (Supplementary Fig. S1 in Feng et al (32)). As a result, although this leads to significant bone loss (e.g., Ct. Th, −13.1%; P = .0012), we may have underestimated the physiologic effect of Vdr gene deletion in the PCo. Additionally, we could not measure serum 1,25(OH)₂D₃ levels after the 2-week intervention because blood draw interferes with our primary outcome of interest, intestinal Ca absorption. Therefore, we chose to use renal Cyp27b1 mRNA level as the surrogate of serum 1,25(OH)₂D₃ level. Finally, we also noticed higher than expected variability in our Ca absorption data, which could be reduced by including more mice per genotype × diet group.

In summary, our study reveals important facets of 1,25(OH)₂D₃ and VDR signaling in adult Ca absorption and adult bone health. First, we clearly show that 1,25(OH)₂D₃ and VDR play a minimal role in Ca absorption and bone health of adult mice as long as their dietary Ca intake is adequate. This suggests that VDR-independent Ca-absorption mechanisms are dominant in these conditions. Second, our work shows that intestinal VDR is a critical mediator of the adaptive responses of Ca absorption that protects against adult bone loss when dietary Ca intake is low. Third, our data demonstrate that Ca absorption in the PCo has a physiologic role in protecting adult bone when dietary Ca intake is low. Finally, by defining the role of vitamin D signaling in the adult intestine, and by clarifying the physiologic importance of Ca absorption in the adult intestine, our present work lays the foundation for future research on approaches to enhance adult Ca absorption, including targeting the PCo as a means to improve Ca homeostasis and attenuate bone loss in individuals at high risk for bone loss due to conditions such as bariatric surgery or short bowel syndrome (62).

Abbreviations

µCT

microcomputed tomography

1,25(OH)₂D₃

1,25-dihydroxyvitamin D₃

BMC

bone mineral content

BMD

bone mineral density

BV/TV

bone volume fraction

Ca

calcium

Ct.Ar

cortical bone area

Ct.Ar/Tt.Ar

cortical area fraction

Ct.Th

cortical bone thickness

Dd

duodenum

IP

intraperitoneal

HVDRR

hereditary vitamin D–resistant rickets

KO

knockout

LIK

Cdx2-CreER^T2+/− ×Vdr^f/f

mRNA

messenger RNA

PBM

peak bone mass

PCo

proximal colon

PCR

polymerase chain reaction

PTH

parathyroid hormone

qPCR

quantitative polymerase chain reaction

ROI

region of interest

RT

real-time

Tb.N

trabecular number

Tb.Sp

trabecular separation

Tb.Th

trabecular thickness

Tt.Ar

total area

VDR

vitamin D receptor

WIK

villin-CreER^T2+/− × Vdr^f/f

Funding

This work was supported by the National Institutes of Health (grant No. R01DK112365 to J.C.F. and S.C.).

Disclosures

The authors have nothing to disclose and no conflicts of interest.

Data Availability

Some or all data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.