Interindividual Genetic Differences Drive Discordance Between Serum Calcidiol and Calcitriol Concentrations in Females

Abstract

Vitamin D insufficiency (VDI) is primarily determined by serum levels of calcidiol, which serves as a biomarker for the less abundant but most potent bioactive metabolite, calcitriol. However, population studies often show discordance between calcidiol and calcitriol. Here, a genetically diverse population of 7 inbred mouse strains was used to investigate the role of interindividual genetic differences in driving calcidiol-to-calcitriol discordance under vitamin D sufficient (VDS) vs depleted (VDD) conditions. We found high interstrain variability in calcitriol that was discordant with calcidiol under VDS and VDD conditions. However, under VDS conditions, stratification by calcitriol level revealed that strains with serum calcitriol >60 pM (HighC) exhibited the expected positive calcidiol-to-calcitriol association, whereas strains with low calcitriol (<60 pM, LowC) did not. Thus, discordance under VDS was driven by genetically divergent strains with LowC. Discordance under VDD was not associated with LowC. LowC was not caused by increased calcitriol degradation or by transcriptional dysregulation of canonical vitamin D metabolism enzymes. Instead, LowC strains exhibited low renal expression of Lrp2 (megalin), the primary transporter required for renal calcitriol production. LowC strains also exhibited reduced renal expression of the vitamin D receptor (Vdr) and several target genes, demonstrating impaired vitamin D signaling. These findings reveal novel, naturally occurring genetic determinants of VDI that function by disrupting calcitriol production and signaling in a manner that cannot be predicted by calcidiol levels. Cross-species conservation of this phenomenon would have important implications for clinical management of VDI and related disease risks across genetically diverse populations.

Vitamin D (VitD) functions as a steroid hormone estimated to regulate transcription of more than 5000 genes (1). In addition to its well-characterized roles regulating calcium homeostasis and skeletal health (2), VitD also has important roles in cardiometabolic, immune, neurobehavioral, and reproductive health (3-7). VitD is especially important during pregnancy to support maternal and fetal health. VitD insufficiency (VDI) during pregnancy increases the risk of pregnancy complications, including preeclampsia (8), gestational diabetes (9, 10), cesarean delivery (11), preterm birth (12), and small for gestational age (13). Emerging studies from our laboratory and that of others implicate maternal VDI in offspring metabolic and neurobehavioral health, with some effects persisting into adulthood (14-17). Given these wide-ranging effects and the high global prevalence of VDI among pregnant women (18, 19), it is important to fully understand the factors disrupting VitD status and its functions.

In humans, VitD is obtained by sunlight-induced synthesis in the skin or by diet and has 2 vitamer forms, D3 (synthesized primarily by animals) and D2 (synthesized primarily by fungi) (20). D3 is typically the most abundant form found in human serum. Figure 1 summarizes the canonical VitD3 metabolism pathway, which is conserved between humans and rodents. VitD3 activation starts in the liver with the production of 25(OH)D3 (calcidiol) and finishes in the kidney with the production of 1,25(OH)₂D3 (calcitriol) (20). Calcitriol is the most potent ligand of the VitD receptor (VDR), which regulates gene transcription by binding to VitD response elements (VDRE) in the genome (20). However, because of the short half-life of calcitriol (4-6 hours) and low abundance in serum, VitD status is primarily monitored using the 1000× more abundant but less active precursor metabolite 25(OH)D3 (calcidiol) (35). Regulatory agencies differ in their recommendations for the threshold of serum calcidiol required to maintain VitD sufficiency in humans, but most recommend maintaining serum calcidiol ≥50 nmol/L (36).

Figure 1.

Canonical vitamin D metabolism pathway. Figure shows the primary genes/proteins regulating production and activities of the most abundant VitD metabolites, which are conserved between mouse and humans. VitD obtained through consumption or synthesis in the skin (21) is metabolized primarily in the liver and kidney (20). VitD binding protein (DBP, encoded by Gc) is primarily expressed in the liver and binds to VitD3 and its metabolites for stable transport in the bloodstream to target tissues (22). Cytochrome P450 Family 2 Subfamily R Member 1 (Cyp2r1) is primarily expressed in the liver and catalyzes hydroxylation of carbon 25 on VitD to form 25(OH)D3 (23, 24). Megalin (encoded by Lrp2) is primarily expressed in the kidney and mediates transport of 25(OH)D3 into the kidney via endocytosis (25). Cytochrome P450 Family 27 Subfamily B Member 1 (Cyp27b1) is primarily expressed in liver (although extrarenal expression is well documented (26)) and catalyzes hydroxylation of carbon 1 on 25(OH)D3 in the kidney to generate the most potent active metabolite, 1,25(OH)₂D3 (27, 28). Cytochrome P450 Family 24 Subfamily A Member 1 (Cyp24a1) is primarily expressed in kidney and catalyzes hydroxylation of carbon 24 on 25(OH)D3 and 1,25(OH)₂D3 to generate catabolic products 24,25(OH)₂D and 1,24,25(OH)₃D (29, 30). The VitD receptor (Vdr) is primarily expressed in the kidney and binds to 1,25(OH)₂D3 to enable translocation into the nucleus and genomic binding to target genes for transcriptional regulation including positive self-regulation (31-34).

The primary factors proposed to drive VDI are low dietary intake and darker skin pigmentation that blocks skin synthesis of VitD (37). However, conditions affecting tissues important for VitD absorption (eg, intestines (38)) and activation (eg, liver, kidney (39)) can also cause VDI. Recent work also strongly implicates the role of genetic regulation of VitD metabolism via modulation of the enzymes that catalyze VitD activation (CYP2R1, CYP27B1) and catabolism (CYP24A1), or mediate transport (GC) (40, 41). However, the distinct contributions of naturally occurring genetic differences, environmental causes, and organ dysfunction in driving VDI within populations are poorly defined. Furthermore, most genome-wide association discovery studies have targeted genetic determinants of calcidiol (40-46), whereas our knowledge of genetic factors independently driving variability in calcitriol remains limited (47). A 2015 mouse study from the Fleet laboratory, which shows genetic factors associated with serum calcitriol concentrations map to different loci than those associated with calcidiol, strongly supports the presence of distinct regulatory mechanisms (48). Human population data showing discordance between serum calcidiol and calcitriol concentrations (48-51) warrants further study to understand the mechanisms controlling the relationship between these 2 metabolites and the implications for disease risk. This is especially important for managing VDI among genetically diverse populations that exhibit high variability in calcidiol or calcitriol.

Mouse models provide a valuable tool for studying genetic determinants of VDI because there is strong conservation in VitD biology between mice and humans (52, 53). Of equal importance, both genotype and environment can be controlled in mouse models to isolate genetic factors from environmental factors driving VDI and the associated diseases. The current study aimed to investigate the role of naturally occurring interindividual genetic differences in driving low calcitriol and to characterize the relationship between calcidiol and calcitriol in a genetically diverse population. We leveraged the high interstrain genetic diversity present among mouse strains from the Collaborative Cross (CC) recombinant inbred population. Population effects were studied under VitD-sufficient (VDS) or VitD-depleted (VDD) conditions to assess differences in VDI caused by genetic vs environmental factors. We assessed effects in reproductive-aged adult females building on our previous research showing interindividual differences in maternal VitD status (54-56). Our findings shed new light on the complex relationship between calcidiol and calcitriol that underlies VDI, including new evidence that noncanonical genetic mechanisms play a key role.

Materials and Methods

Mice, Housing, and Dietary Treatments

Animal handling was performed in accordance with the Guide for the Care and Use of Laboratory Animals under the corresponding animal use protocol at the University of North Carolina at Chapel Hill. Mice from 7 inbred CC strains (CC001/Unc [CC001], CC006/TauUNC [CC006], CC011/Unc [CC011], CC017/Unc [CC017], CC026/GeniUNC [CC026], CC032/GeniUnc [CC032], and CC051/TauUNC [CC051]) were obtained from the UNC Systems Genetics Core Facility (Chapel Hill, NC). CC strains are genetically divergent recombinant inbred mice that carry different combinations of alleles derived from 8 genetically divergent founder strains: A/J, C57BL/6J (B6), 129S1Sv/ImJ (129), NOD/ShiLtJ (NOD), NZO/H1LtJ (NZO), CAST/EiJ (CAST), PWK/PhJ (PWK), and WSB/EiJ (WSB) (57). The last CC strain genotyping that is relevant to this study occurred in 2011 through 2012, 5 years before the selected experimental CC strains were born in 2017 through 2018. At this time, the selected strains were at least 91% homozygous (91.4%-95.4%). Mice were housed in the UNC Division of Comparative Medicine vivarium under standard conditions (microisolator cages, ad libitum water and food access, controlled 22-24 °C temperatures, 12-hour light-dark cycle, and unfiltered fluorescent light with ultraviolet B wavelength ∼280-315 nm, ∼8.39E⁻⁷ W/cm²). Mice were cohoused with siblings and fed standard chow (2.4 IU/g VitD3, 8604, Envigo Teklad, Madison, WI) from birth to adulthood. At 4 to 5 months old, cohoused females were split into different treatment groups to minimize litter/cage effects and placed on either purified VDS or nutrient-matched VDD diet containing “VitD-free” tested casein (VDS: 2.2 IU VitD3/g, 0.47% calcium, 0.3% phosphorus, TD.89124; VDD: 0 IU VitD3/g, TD.89123, Envigo Teklad, TD.89123). Mice were treated with purified diets for 6 weeks (except CC051, on diet for ∼19 weeks). At the end of the diet treatment, mice were euthanized by CO₂ inhalation followed by cervical dislocation for tissue and serum collection. Samples were randomized and run blinded for all assays.

Food Intake and Body Composition

Food intake and bodyweight were measured weekly. Average daily food intake was calculated for the first 5 weeks of treatment using the formula: [Total food intake/cage/# treatment days/# of mice]. In week 6 of treatment, fat and lean mass were measured via EchoMRI (3-in-1 Body Composition Analyzer, EchoMRI, Houston, TX) by the UNC Nutrition Obesity Research Center Animal Metabolism Phenotyping Core. Free water mass was subtracted from body weight for fat and lean mass calculations.

Quantification of VitD Metabolites

At the end of the diet treatment, whole blood was collected via cardiac puncture (after CO₂ inhalation and cervical dislocation), and serum was isolated, snap-frozen in liquid nitrogen, and stored at −80 °C until use. Protein bound and unbound VitD metabolites [25(OH)D3, 1,25(OH)₂D3, 24,25(OH)₂D3, 1,24,25(OH)₃D3)] were quantified via liquid chromatography-tandem mass spectrometry (LC/MS/MS) at the UC Davis Lipid Analytical Core Facility (Davis, California), as previously described (58). Measurements below the limit of quantification were imputed to the values of the limit of detection. Activation ratios were determined using the formula: [1,25(OH)₂D3 nM/25(OH)D3 nM]. Catabolism ratios were determined using the formulas: [1,24,25(OH)₃D3/1,25(OH)₂D3 nM] and [24,25(OH)₂D3/25(OH)D3 nM].

Founder Strain and Single Nucleotide Polymorphism Assessment

Gene locations and sizes were extracted from the UCSC genome browser (59) (GRCm38/mm10). Single nucleotide polymorphism (SNP) data were extracted from the Mouse Phenome Database (60) (GRCm38/mm10) for available genotypes. Loci with missing data were excluded. Nonsynonymous variants predicted to affect protein function were identified using the SIFT tool (61) in Ensembl (62) (GRCm39/mm39). Founder haplotypes at VitD metabolism and transport genes were obtained using the CC viewer tool (63).

Tissue Collection and Gene Expression

Liver and kidney (renal capsule removed) tissues were snap-frozen in liquid nitrogen and stored at −80 °C until use. Frozen tissues were manually pulverized into a homogenous mixture for RNA extraction by Trizol reagent (Invitrogen), and gene expression was measured by quantitative real-time PCR, as previously described (54). Expression was measured in technical triplicate for Cyp2r1 and Gc in liver and for Cyp27b1, Cyp24a1, Cyp27a1, Vdr, Lrp2, S100g, mKlotho, Npt2a/Slc34a1, and Npt2c/Slc34a3 in the kidney. Relative expression was normalized to reference genes that were not independently altered by VDD, Arpp0 (Acidic ribosomal phosphoprotein P0) for liver and B2M (Beta-2-Microglobulin) for kidney, and expression was calculated using Pfaffl's method (64). Primer sequences and references are provided in Table S1 (65).

Western Blot Analyses

Frozen pulverized kidney tissue was placed in RIPA Lysis Buffer (Sigma #R0278) and homogenized for 1 hour at room temperature on a vortexer and for ten 5-second cycles in a Diagenode Sonicator. Lysates were then cleared by centrifugation. Protein was quantified using a BCA assay (Pierce #PI23227), and for each sample, 10 µg of protein lysate was mixed with 4× Laemli Loading Buffer (BioRad #1610747) and β-mercaptoethanol, denatured for 5 minutes at 95 °C, and loaded onto a precast 4% to 15% polyacrylamide gel (BioRad #4561085) beside a reference marker (BioRad #161-0375). The gel was run at 4 °C for 15 minutes at a constant 50 V, followed by 1 hour and 15 minutes at 75 V, then transferred for 2 hours at 80 V onto a low fluorescence PVDF membrane (BioRad #1620260). The membrane was blocked in blocking buffer (LICOR #927-5000) for 1 hour at room temperature, then incubated with primary antibodies for VDR (D2K6W, Cell Signaling, Cat #12550, Antibody Registry ID: AB_2637002) and B-actin (Cell Signaling, Cat #4970S, Antibody Registry ID: AB_2223172) simultaneously overnight at 4 °C. Membranes were then probed with a fluorescent secondary antibody for IgG (LICOR, Cat #92632211, Antibody Registry ID: AB_621843) for 1 hour at room temperature. Bands were imaged and quantified by densitometry using a LICOR Odyssey Clx with Image Studio 6.0 software.

Calcium Measurements

At the end of the diet treatment, whole blood was collected via submandibular puncture and serum isolated by incubating at room temperature for 30 minutes, followed by centrifugation at 4 °C for 10 minutes at 2000× g, followed by flash freezing in liquid nitrogen and storage at −80 °C until use. Serum concentrations of calcium were measured using the Calcium Colorimetric Assay Kit (Abcam, #ab102505).

Statistical Analysis

Unless otherwise noted, all statistical analyses were performed using JMP Pro software version 16.1.0 (SAS, NC) using the following methods. To test for main diet or main strain effects, linear regression analyses were performed using the model (y∼diet + strain) with additional variables added where indicated. The model (y∼diet + strain + diet*strain) was used to test for interactive strain × diet effects. For outcomes with significant strain effects, Tukey's (honest significance difference) post hoc analyses were used to identify which strains differed, and differences are noted with different letters. Error bars represent the standard error of the mean (SEM) on all bar graphs. Associations were determined by linear regression, and a 95% CI was determined using the “Line of Fit” command in JMP (Polynomial, linear). P values <.05 are considered statistically significant. Association plots show measurements for each mouse along the line of fit for the population tested with a shaded 95% confidence interval. R-squared (r²) is provided for significant associations to indicate the amount of variance explained.

Results

Interindividual Genetic Differences Drive Low Serum Calcitriol That Is Not Predicted by Serum Calcidiol Levels

To assess the impact of genetic background on VitD status in reproductive-aged females, we measured strain effects in an adult female population of 7 genetically divergent CC inbred mouse strains treated with a purified VDS diet (2.2 IU/g VitD3) (Fig. 2A). To measure strain differences in response to acute dietary VitD depletion, we used a cohort of females matched to the VDS group for age, litter, and timing, treated with a purified VDD diet (0 IU/g VitD3) (Fig. 2A).

Figure 2.

Strain differences in food intake and body composition under VitD-sufficient (VDS) and VitD-depleted (VDD) conditions. (A) Treatment scheme: 4- to 5-month-old female mice were experimentally treated with purified VitD sufficient (VDS) or depleted (VDD) diet (VitD content shown in parentheses) for 6 weeks (^#CC051 mice treated for ∼19 weeks). (B) Average daily food intake. (C) Average bodyweight at start of diet treatment (top panel), in week 6 of treatment (middle panel), and difference between diet start and week 6 (bottom panel). (D) Average percent fat mass (top panel), percent lean mass (middle panel), and fat/lean mass (g) ratio (bottom panel) in week 6 of treatment. P values shown for strain (adjusted for diet), diet (adjusted for strain), and strain × diet interactive effects. Sample sizes: Total population (n = 58); N per strain from left to right (7, 7, 6, 6, 6, 12, 14); n per diet (VDS = 30, VDD = 28).

Strains had similar food intake levels except for CC017, which ate the least (mean = 2.3 g/day) and CC011, which ate the most (mean = 3.3 g/day) (Fig. 2B). Strain differences in body weight and body composition, before and after diet treatments, mostly reflected differences in food intake, with CC017 also exhibiting the lowest body weight and leanest body composition (fat/lean mass ratio) compared to other strains (Fig. 2C and 2D). Over the course of the diet treatment, strains differed significantly in their change in body weight, with CC017 and CC032 exhibiting little change, whereas CC001 and CC051 exhibited the greatest increase in body weight (Fig. 2C). VDD treatment (alone or in interaction with strain) did not significantly alter food intake, body weight, or body composition measures in the population (Fig. 2B-2D).

The influence of genetic background on the availability of calcitriol (1,25(OH)₂D3) and its relationship with calcidiol (25(OH)D3) was assessed by measuring serum metabolite concentrations at the end of the diet treatment using LC/MS/MS, as previously described (58). Under VDS conditions, calcidiol concentrations differed by genetic background up to ∼2-fold (Fig. 3A). This strain effect remained significant after adjustment for strain differences in body weight and food intake (P = 2.6 ×10⁻³). VDD substantially reduced calcidiol concentrations across all strains (Fig. 3A). However, strains differed significantly in the extent of calcidiol retention with VDD, ranging on average from ∼16% (CC011) to ∼1% (CC032) (Fig. 3A and Fig. S1A (65)).

Figure 3.

Impact of interindividual genetic differences on serum calcitriol concentrations and its concordance with calcidiol. (A) Serum calcidiol and (B) Serum calcitriol concentrations under VDS and VDD conditions. Dots represent measures from individual mice with mean and SEM overlaid. (C) Association plot showing predictive relationship between serum calcidiol and calcitriol under VDS conditions only for all strains (left panel) or stratified by calcitriol phenotype (right panel). (D) Association plot showing predictive relationship between serum calcidiol and calcitriol under VDD conditions only for all strains (left panel) or stratified by calcitriol phenotype (right panel). (E) Association plot showing predictive relationship between serum calcidiol and calcitriol within the full genetically diverse population of all strains and diet groups (VDS and VDD) combined (left panel) or stratified by calcitriol phenotype (right panel). (A and B), P values shown for strain (after stratified by diet, VDS-grp and VDD-grp), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠b = ab). For (C-E), P values, line of fit, and shaded 95% CI are shown for each population tested. For significant associations (P < .05), r² indicates the amount of variance in calcitriol explained by variance in calcidiol. Sample sizes: Total population (n = 42) (HighC = 24, LowC = 18); n per strain (6); n per diet (21).

Under VDS conditions, serum calcitriol concentrations also differed by genetic background up to ∼4-fold (Fig. 3B). This strain effect remained significant after adjustment for body weight and food intake (P = 5.7×10⁻⁵). Using a mean serum calcitriol threshold of 60 pM (threshold at which VDD reduced bone mineral density (66)), we categorized strains as low calcitriol (LowC < 60 pM; CC026, CC006, and CC032) or high calcitriol (HighC > 60 pM; CC001, CC011, CC017, and CC051) (Fig. 3B). VDD substantially reduced serum calcitriol concentrations across all strains, further exacerbating the LowC phenotype (Fig. 3B). There was a significant strain-dependent diet effect on calcitriol concentrations but no significant strain differences in calcitriol retention after VDD, which ranged on average from ∼36% (CC017) to ∼5% (CC032) (Fig. 3B and Fig. S1B (65)).

Under VDS or VDD conditions, there was no concordance between calcidiol and calcitriol concentrations when all strains were combined (Fig. 3C and 3D). However, after stratification by calcitriol phenotype, VDS-fed HighC strains exhibited the expected strong positive association between calcidiol and calcitriol, whereas VDS-fed LowC strains exhibited a striking negative association between the metabolites (Fig. 3C). Under VDD conditions, neither HighC nor LowC strains showed concordance between the metabolites (Fig. 3D). Assessment of the full population (VDS and VDD combined) showed concordance between the VDD-induced reductions in calcitriol and calcidiol and revealed a stronger relationship between metabolites among HighC strains, where variability in calcidiol was associated with ∼89% of the variability in calcitriol, compared to only ∼42% among LowC strains (Fig. 3E).

LowC Phenotype Is Linked to a Reduced Activation Ratio That Is Not Driven by Aberrant Transcriptional Dysregulation of Canonical VitD Activation Enzymes

To test whether LowC may be driven by disruption of VitD activation, we assessed differences in the VitD activation ratio, a measure of the amount of active calcitriol circulating relative to the amount of 25(OH)D3 available in circulation for activation. Under VDS conditions, strains exhibited significantly different VitD activation ratios (Fig. 4A), with LowC strains exhibiting significantly lower ratios than HighC strains (Fig. 4B). VDD drove a >10-fold increase in the activation ratio (P = 8.4 ×10⁻³), resulting in a similar activation ratio between LowC and HighC strains (Fig. 4A and 4B).

Figure 4.

Genetic strain differences in the VitD activation potential. (A) Activation ratio: Proportion of circulating active calcitriol metabolite concentrations relative to amount of available calcidiol precursor under VDS or VDD conditions graphed by strain or by (B) HighC vs LowC mice. (Activation ratio for 1 VDD-CC026 sample removed due to value >150× higher than other biological replicates.) Dots represent measures from individual mice with mean and SEM (bar graph) or median and 95% CI (box and whiskers plot) overlaid. P values shown in (A) for strain (stratified by diet), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠ b = ab). Sample sizes: Total population (n = 41) (HighC = 24, LowC = 17); n per strain (5-6); n per diet (20-21).

To determine whether LowC may be explained by transcriptional dysregulation of the canonical VitD activation enzymes, Cyp2r1 or Cyp27b1, we assessed whether the genetic makeup of the genes or their expression levels are associated with serum metabolite concentrations. Cyp2r1 catalyzes the production of 25(OH)D3 in the liver (Fig. 1) and is expected to be a positive regulator of 25(OH)D3 (and indirectly calcitriol). Among the 7 CC strains tested, Cyp2r1 contained 80 synonymous SNPs in cis (Table 1) and haplotypes derived from 6 founder strains (Table 2). Neither founder strain nor sequence similarity at Cyp2r1 segregated completely with the LowC phenotype (Table 2). Liver Cyp2r1 expression differed significantly among the strains, with a ∼10-fold difference between the lowest (CC011, HighC) and highest (CC032, LowC) (Fig. 5A). LowC mice had significantly higher Cyp2r1 expression levels on average compared to HighC mice (Fig. 5B) despite similar 25(OH)D3 concentrations (Fig. 5C). VDD significantly reduced Cyp2r1 transcript levels with no strain × diet interactive effect (Fig. 5A). Under VDS conditions, when all strains were combined, there was no association between Cyp2r1 and calcidiol or calcitriol (Fig. 5D). However, stratification by calcitriol phenotype revealed that HighC strains exhibited a significant positive association where variability in Cyp2r1 expression was associated with ∼60% of the variability in calcidiol and calcitriol, whereas LowC strains showed no association (Fig. 5D). Cyp2r1 expression was negatively associated with the activation ratio, but only when HighC and LowC strains were combined, and only associated with ∼22% of the variation in the activation ratio (Fig. 5D). VDD-fed mice showed no associations between Cyp2r1 and serum metabolite concentrations (Fig. S2A (65)).

Table 1.

Genetic variation at vitamin D metabolism pathway genes among the 7 CC mouse strains tested

	Gene	Description	Role in vitamin D metabolism	Gene location (GRCm38/mm10)	Size (bp)	SNPs (no.)	Nonsynonymous variants (no. predicted deleterious)	UTR/splice variants (no.)
Activation	Cyp2r1	25-hydroxylation	Production of 25(OH)D	Chr7: 114549682-114562890	13,209	80	0 (0)	2
	Cyp27b1	1-hydroxylation	Activation of 25(OH)D to generate 1,25(OH)2D	Chr10: 127048250-127052969	4720	47	7 (1)	15
Catabolism	Cyp24a1	24-hydroxylation	Catabolism of 25(OH)D and 1,25(OH)2D	Chr2: 170482708-170497145	14,438	202	1 (0)	26
Transport	Gc	Vitamin D binding protein (DBP)	Vitamin D transport in circulation	Chr5: 89417522-89457898	40,377	545	5 (2)	12
	Lrp2	Megalin	25(OH)D transport into kidney	chr2: 69424340-69586065	161,726	1363	18 (1)	8

Table 2.

CC mouse strain founder haplotypes and sequence similarity at vitamin D regulatory genes

	Strain	Cyp2r1	Cyp27b1	Cyp24a1	Gc	Lrp2
	Number of founder haplotypes	6	5	4	4	4
LowC strains	CC006	WSB (34)	a CAST (40)	129 (0)	NZO (58)	A/J (876)
	CC032	129 (32)	A/J (0)	CAST (193)	a CAST (508)	NZO (296)
	CC026	B6 (0)	WSB (3)	CAST (199)	NZO (58)	b WSB (804)
HighC strains	CC001	NZO (33)	NOD (0)	129 (1)	129 (60)	A/J (896)
	CC051	A/J (33)	129 (0)	129 (0)	NZO (58)	B6 (8)
	CC011	WSB (34)	WSB (11)	A/J (2)	B6 (0)	A/J (899)
	CC017	PWK (78)	a CAST (40)	WSB (4)	129 (54)	A/J (901)

Gene sequence similarity shown in parentheses (number of single nucleotide polymorphisms between strain and C57BL/6J reference sequence).

^a Haplotypes with single nucleotide polymorphisms predicted to affect protein function.

^b Stop gained mutation.

Figure 5.

Genetic strain differences in transcriptional regulation of canonical VitD activation genes and their relationship with calcidiol and calcitriol. (A) Liver expression of Cyp2r1 (relative to Arpp0) graphed by strain or (B) calcitriol phenotype. (C) Serum calcidiol concentrations for HighC vs LowC strains. (D) Association plots showing predictive relationship between liver Cyp2r1 expression and serum calcidiol (top), calcitriol (middle), or the activation ratio (bottom) under VDS conditions only for all strains (left) or stratified by calcitriol phenotype (right). (E) Kidney expression of Cyp27b1 (relative to B2M) graphed by strain or (F) calcitriol phenotype. (G) Serum calcitriol concentrations for HighC vs LowC strains. (H) Association plots showing predictive relationship between liver Cyp27b1 expression and serum calcidiol (top), calcitriol (middle), or the activation ratio (bottom) under VDS conditions only for all strains (left) or stratified by calcitriol phenotype (right). (Activation ratio for 1 VDD-CC026 outlier sample removed due to value >150× higher than other biological replicates.) A and E, P values shown for strain (adjusted for diet), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠ b = ab). (D and H) P values, line of fit, and shaded 95% CI are shown for each population tested. For significant associations (P < .05), r² indicates the amount of variance in metabolite concentrations explained by variance in gene expression levels. Sample sizes: Total population (n = 41-42) (HighC = 24, LowC = 17-18); n per strain (5-6); n per diet (20-21).

Cyp27b1 catalyzes the activation of 25(OH)D3 in the kidney to generate calcitriol (Fig. 1) and is expected to be a positive regulator of calcitriol but a negative regulator of 25(OH)D3. Among the 7 CC strains tested, Cyp27b1 contained 47 SNPs in cis, including 7 nonsynonymous variants (Table 1), and haplotypes derived from 5 founder strains (Table 2). One variant (rs248879305), derived from the CAST founder strain and carried by both CC006 (LowC) and CC017 (HighC), is predicted to affect protein function assessed by SIFT (61) (Table 1). Neither founder strain nor sequence similarity at Cyp27b1 segregated completely with the LowC phenotype (Table 2). Kidney Cyp27b1 expression differed significantly among the strains, with an ∼11-fold difference between the lowest (CC001, HighC) and the highest (CC011, HighC) (Fig. 5E). However, LowC mice had similar levels of Cyp27b1 expression (Fig. 5F) on average despite substantially lower calcitriol (Fig. 5G). VDD did not significantly alter Cyp27b1 expression levels, and there was no strain × diet interactive effect (Fig. 5E). Under VDS conditions, when strains were either combined or stratified by calcitriol phenotype, variability in Cyp27b1 expression was negatively associated with ∼30% to 40% of the variability in calcidiol (Fig. 5H). Cyp27b1 was not associated with calcitriol in any population but was positively associated with the activation ratio, explaining ∼26% of the variation (when HighC and LowC strains were combined) and ∼45% of the variation among LowC strains only (Fig. 5H). VDD-fed mice showed no associations between Cyp27b1 and serum metabolite concentrations (Fig. S2B (65)).

We also investigated the role of Cyp27a1, an alternative VitD 25-hydroxylase proposed to contribute to the generation of calcidiol (66). Cyp27a1 expression levels in the liver were similar between LowC and HighC strains and were not significantly altered by diet, making it unlikely to play a causal role in the LowC phenotype (Fig. S3A and S3B (65)).

LowC Phenotype Is Not Driven by Increased Catabolism

Catabolism of VitD occurs in the kidney, catalyzed by Cyp24a1 via hydroxylation of 25(OH)D3 or 1,25(OH)₂D3 to form 24,25(OH)₂D3 or 1,24,25(OH)₃D3, respectively (Fig. 1). To maintain homeostasis, excess calcitriol triggers upregulation of Cyp24a1, whereas calcitriol insufficiency may decrease Cyp24a1 expression to increase calcitriol retention (67). To assess whether the LowC phenotype is related to either scenario, we assessed whether the Cyp24a1 genetic makeup or expression levels are associated with serum metabolite concentrations among the 7 CC strains.

Cyp24a1 contained 202 SNPs in cis, including 1 nonsynonymous variant that was not predicted to affect protein function (Table 1) and haplotypes derived from 4 founders (Table 2). Neither founder strain nor sequence similarity at Cyp24a1 segregated completely with the LowC phenotype (Table 2). Kidney Cyp24a1 expression was similar among the strains except for CC011, which had ∼3- to 6-fold higher levels (Fig. 6A). Cyp24a1 expression was not affected by diet and was similar between LowC and HighC mice under VDS or VDD conditions (Fig. 6B).

Figure 6.

Genetic strain differences in Cyp24a1 expression. (A) Kidney expression of Cyp24a1 (relative to B2M) graphed by strain or (B) calcitriol phenotype. P values shown in (A) for strain (adjusted for diet), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠ b = ab). Sample sizes: Total population (n = 42) (HighC = 24, LowC = 18); n per strain (6); n per diet (21).

To assess catabolism effects independent of Cyp24a1, we measured serum concentrations of the catabolic products (24,25(OH)₂D3 and 1,24,25(OH)₃D3) using LC/MS/MS, as previously described (58). Similar to its precursor calcidiol, 24,25(OH)₂D3 differed by individual strain under VDS and VDD conditions (Fig. 7A) but did not differ between LowC vs HighC mice (Fig. 7B). Also similar to its precursor calcitriol, 1,24,25(OH)₃D3 differed by strain under VDS and VDD conditions (Fig. 7C), and LowC mice exhibited lower concentrations than HighC (Fig. 7D). VDD drove a reduction in 24,25(OH)₂D3 and 1,24,25(OH)3D3 concentrations that was seemingly dependent on baseline (VDS) concentrations (Fig. 7A and 7D). Catabolism ratios (measure of circulating catabolic product relative to the available precursor metabolite) for 25(OH)D3/24,25(OH)₂D3 differed by strain, but were not associated with the LowC phenotype and were not significantly altered by VDD (Fig. 7E and 7F). In contrast, the catabolism ratios for 1,25(OH)₂D3/1,24,25(OH)₃D3 were similar among the strains and also unaffected by LowC but were significantly increased by VDD (Fig. 7E and 7F). These findings demonstrate the LowC phenotype is not caused by increased catabolism of VitD. Furthermore, interstrain variability in catabolic products seemed to mostly reflect the availability of the precursor metabolite and not differences in Cyp24a1 expression.

Figure 7.

Genetic strain differences in VitD catabolism products. (A) Serum concentrations of 24,25(OH)₂D3 under VDS or VDD conditions graphed by strain or (B) calcitriol phenotype. (C) Serum concentrations of 1,24,25(OH)₃D3 under VDS or VDD conditions graphed by strain or (D) calcitriol phenotype. (E) Catabolism ratio: Proportion of circulating catabolic product relative to amount of available precursor under VDS or VDD conditions graphed by strain or (F) calcitriol phenotype. (Catabolism ratio for 1 VDD-CC026 outlier sample removed due to value >40× higher than other biological replicates.) (A, C, and E) P values shown for strain (after stratified by diet, VDS-grp and VDD-grp), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠ b = ab). Sample sizes: Total population (n = 42) (HighC = 24, LowC = 18); n per strain (6); n per diet (21).

LowC Strains Exhibit Low Expression of the Kidney Transporter Megalin (Lrp2)

VitD activation to produce calcitriol requires transport of calcidiol from the liver into the kidney, which is mediated by the VitD-binding protein (DBP), which is encoded by Gc, and megalin, which is encoded by Lrp2 (Fig. 1). Disruption of transport could result in LowC. Therefore, we assessed the role of differences in genetic makeup or dysregulation in expression levels of Gc and Lrp2 among the 7 CC strains.

DBP is encoded by Gc and is responsible for systemic transport of VitD metabolites. We identified 545 SNPs in cis at Gc, including 5 nonsynonymous variants of which 2 variants (rs51039323 and rs47789395), derived from CAST and carried only by CC032 (LowC), were predicted to affect protein function (Table 1). Strain haplotypes at Gc were derived from 4 founders (Table 2), but neither founder strain nor sequence similarity at Gc segregated completely with the LowC phenotype (Table 2). Liver Gc expression differed among the strains, mostly driven by high expression in CC051 (Fig. 8A). However, Gc expression was similar between LowC and HighC mice and was not affected by VDD (Fig. 8B).

Figure 8.

Genetic strain differences in VitD transport genes and concordance with LowC phenotype. (A) Liver expression of Gc (relative to Arpp0) graphed by strain or (B) calcitriol phenotype. (C) Kidney expression of Lrp2 (relative to B2M) under VDS or VDD conditions graphed by strain or (D) calcitriol phenotype. (A and C) P values shown for strain (adjusted for diet), diet (adjusted for strain), and strain × diet effects. Letters denote strains that differ by Tukey post hoc (a ≠ b = ab). Sample sizes: Total population (n = 41-42) (HighC = 24, LowC = 17-18); n per strain (5-6); n per diet (20-21).

Megalin is encoded by Lrp2 and mediates transport of calcidiol into the kidney via endocytosis (25) (Fig. 1). We identified 1363 SNPs in cis at Lrp2, including 18 nonsynonymous variants, of which 1 variant (rs3540406789), derived from WSB and carried only by CC026 (LowC), was predicted to generate a premature stop codon (stop gained) (Table 1). Strain haplotypes at Lrp2 were derived from 4 founders (Table 2), but neither founder strain nor sequence similarity at Lrp2 segregated completely with the LowC phenotype (Table 2). Renal Lrp2 expression differed significantly among the strains (Fig. 8C), with LowC mice exhibiting lower levels compared to HighC (Fig. 8D). In contrast, low calcitriol driven by VDD did not significantly reduce Lrp2 expression (Fig. 8C), implicating a causal role in driving the LowC phenotype. Further assessment of cubilin (Cubn), the endocytic receptor that partners with megalin for transport of calcidiol into the kidney (68-70), found no statistically significant differences among strains or diets (Fig. S4A and S4B (65)). However, renal expression patterns among the strains were similar between Lrp2 and Cubn under VDS conditions (Fig. S4A and S4C (65)).

LowC Strains Exhibit Impaired VitD Signaling

Calcitriol is the most potent ligand for VDR, the primary nuclear receptor that binds and transports calcitriol into the nucleus and facilitates VitD signaling via genomic binding to VitD response elements (Fig. 1). The Vdr gene locus contains a VDRE that allows it to be positively autoregulated by calcitriol-bound VDR (71). To assess whether the genetically determined LowC phenotype disrupts VitD signaling, we measured the impact on kidney Vdr expression and several classical calcitriol-bound VDR targets that are typically upregulated by calcitriol, including calbindin D9k/VitD-dependent calcium-binding protein (S100g), Klotho (Kl), Solute carrier family 34 member 1 (Npt2a/Slc34a1), and Solute carrier family 34 member 3 (Npt2c/Slc34a3).

Under VDS conditions, strains exhibited significant differences in renal Vdr transcript levels that were primarily driven by lower levels among LowC strains compared to HighC strains (Fig. 9A and 9B). VDD reduced Vdr transcript levels across all strains except CC032, which seemingly exhibited no response to VDD (Fig. 9A). HighC strains exhibited the greatest VDD-induced reductions in Vdr transcript levels, resulting in levels similar to those of VDS-treated LowC strains (Fig. 9A). Similar to Vdr transcript levels, strain differences in renal VDR protein levels were primarily driven by lower levels among LowC strains compared to HighC (Fig. 9C, 9D and Fig. S5 (65)). In contrast, VDD had no effect on VDR protein levels for most strains, with the exception of CC006 (LowC) and CC051 (HighC), which exhibited substantial reductions (Fig. 9C). These findings show that genetically determined LowC drives a reduction in Vdr transcript levels that is similar to the reduction caused by dietary VitD depletion and is sufficient to impact VDR protein levels.

Figure 9.

LowC strains exhibit low renal VitD receptor (Vdr) expression. (A) Kidney transcript levels of Vdr (relative to B2M) graphed by strain or (B) calcitriol phenotype. (C) Kidney protein levels of Vdr (relative to B-actin) graphed by strain or (D) calcitriol phenotype. (A and C) P values shown for strain (stratified by diet), diet (adjusted for strain), and strain × diet effects. Sample sizes: Total population (n = 41-42) (HighC = 23-24, LowC = 18); n per strain (5-6); n per diet (20-21).

Expression of all 4 VDR-mediated VitD target genes followed a pattern similar to Vdr transcript levels, with LowC strains mostly exhibiting reduced expression levels compared to HighC (Fig. 10A-10H). In contrast with Vdr transcript levels, dietary VDD did not affect renal S100g and Klotho levels (Fig. 10A and 10B) and significantly upregulated Npt2a and Npt2c expression (Fig. 10C and 10D). Effects of VDD were similar for both HighC and LowC mice. These findings strongly implicate that LowC disrupts signaling to VDR genes, likely via reduction in VDR availability. Given that several of these genes regulate calcium homeostasis, we also measured serum calcium concentrations in a subset of strains. There was no effect of VDD diet on calcium concentrations, but there was a significant strain effect primarily driven by lower calcium in CC006 (Fig. S6 (65)).

Figure 10.

LowC strains exhibit reduced expression of VDR-mediated calcitriol target genes. (A-D) Kidney expression of s100g, Klotho, Npt2a, and Npt2c (relative to B2M) graphed by strain or (E-H) by calcitriol phenotype. (A-D) P values shown for strain (stratified by diet), diet (adjusted for strain), and strain × diet effects. Sample sizes: Total population (n = 42) (HighC = 24, LowC = 18); n per strain (6); n per diet (21).

Discussion

Here, we used a genetically diverse mouse population to show for the first time that even when there is sufficient dietary VitD intake, naturally occurring interindividual genetic differences can drive distinct genetically determined calcitriol phenotypes: low serum calcitriol that cannot be predicted using the calcidiol biomarker (LowC group) and low serum calcitriol that can be predicted by calcidiol (HighC group). The presence of LowC individuals in the population disrupts the expected calcidiol-to-calcitriol concordance. Despite detecting high interstrain variability in expression of the canonical VitD metabolism pathway genes and high variability in downstream metabolites, LowC was not caused by transcriptional upregulation of VitD catabolism, transcriptional downregulation of VitD activation, nor systemic transportation. Instead, LowC strains have lower renal Lrp2 (megalin) expression, implicating disruption of calcidiol transport into the kidney, leading to reduced calcitriol production. LowC mice also exhibit evidence of impaired VDR-mediated VitD signaling, with markedly reduced expression of VDR and several primary target genes. Taken together, these findings highlight the presence of yet undefined genetic regulators of VitD metabolism and illuminate the need for more research into the effectiveness of calcidiol as a biomarker of VitD sufficiency and its relationship with calcitriol in the context of genetically diverse populations.

Calcitriol is not routinely screened clinically or experimentally because early studies showing its short half-life led scientists to conclude that calcitriol is unstable and would not have much clinical relevance (35). We used a new highly sensitive LC/MS/MS method that uses click chemistry to precisely measure calcitriol at low concentrations (limit of quantification = 2 pM) in small volumes (100 µL) (58). Our data from a genetically diverse mouse population showed that serum calcitriol concentrations were highly stable (reproducible) among genetically identical mice on VDS diet but were highly variable (up to 4-fold) among the different strains/genetic lineages. As expected, VDD substantially reduced calcitriol concentrations across all strains, but concentrations remained highly reproducible among genetically identical mice, and strains with low calcitriol under VDS conditions exhibited the lowest calcitriol under VDD. Thus, we identified genetic effects on calcitriol concentrations that were distinct from diet effects. We also identified a significant strain × diet interactive effect on calcitriol concentrations, implicating strains differed in how they responded to VDD. However, we had limited capacity to further characterize interactive effects because of low sample sizes. Profiling the degradation of VitD under VDS conditions revealed that all genetic lineages had higher overall degradation of calcitriol to 1,24,25(OH)₃D3 (max 1:4) compared to calcidiol to 24,25(OH)₂D3 (max 1:1). However, all strains exhibited similar calcitriol catabolism ratios under VDS and VDD conditions, ruling out differences in calcitriol stability as an explanation for the variability in calcitriol. These findings demonstrate that despite low abundance in serum, calcitriol measurements are relatively stable and reproducible. Thus, when accurately measured, variability in calcitriol concentrations likely reflects biologically relevant interindividual genetic differences or differences in environmental VitD exposure.

Global public health efforts to maintain healthy VitD status rely primarily on screening calcidiol, which has been shown to be a strong predictor of VitD exposure (72). However, human studies showing discordance between serum calcidiol and calcitriol concentrations (48-51) suggest that having “sufficient” calcidiol may not always predict VDS. Here, we used strains from the genetically diverse CC mouse genetic reference population to reveal calcidiol-to-calcitriol discordance that is not driven by differences in VitD exposure (diet) but rather driven primarily by genetic differences. We showed that stratifying the population by calcitriol level revealed HighC genetic lineages with significant variability in calcitriol concentrations that were positively correlated with calcidiol. In contrast, LowC genetic lineages exhibited similar levels of variability in calcitriol but exhibited a striking negative correlation. This finding demonstrates that the relationship between calcidiol and calcitriol is heavily dependent on the genetic makeup of the population and that populations with genetically determined LowC are more likely to show discordance. These data also show that calcitriol concentrations are regulated by calcidiol-dependent and calcidiol-independent genetic mechanisms, with the latter responsible for driving discordance in the population.

The canonical VitD metabolism pathway (Fig. 1) has served as the nexus for integrating genetic and environmental influences on VitD status (20). Our study examined whether the LowC phenotype could be caused by genetically induced transcriptional dysregulation of canonical VitD metabolism genes and whether these effects are comparable to those driven by dietary VitD depletion (20, 67, 73, 74). We detected significant strain-specific variability in expression of the primary activation genes (Cyp2r1, Cyp27b1), catabolism gene (Cyp24a1), and transport genes (Gc and Lrp2). However, LowC was not associated with expression differences in renal Cyp27b1 and Cyp24a1. This finding is consistent with previous findings from the Fleet laboratory in which they used 2 genetically diverse inbred mouse populations (panel of 11 inbred strains and panel of 51 BxD recombinant inbred lines) to demonstrate high variability in serum calcidiol and calcitriol concentrations that was discordant and not correlated with renal Cyp27b1 or Cyp24a1 (48). VitD binding protein isoforms derived from genetic variants in the GC gene are associated with VitD status in humans (75). However, in mice, DBP isoforms and their roles in VitD status have not been characterized. Here, we found that LowC was not associated with hepatic Gc transcript levels and did not segregate with variants in Gc among the 7 strains tested, implicating that it is unlikely that DBP isoforms play an independent causal role in the LowC phenotype. On the other hand, we showed that Cyp2r1 and Lrp2 differed significantly between LowC and HighC strains. Surprisingly, LowC strains exhibited elevated hepatic expression of Cyp2r1, implicating a compensatory response to LowC and not a causal effect. This is supported by our finding that VDD reduced hepatic Cyp2r1 in what was likely a response to reduced substrate (calcidiol). LowC mice also had reduced renal expression of Lrp2, which encodes the endocytic receptor megalin that partners with cubilin (CUBN) (68) to regulate transport of calcidiol into the kidney for activation (25, 69, 70). Cubn exhibited a similar strain-specific expression pattern as Lrp2, but it was not significant. This finding, combined with our finding that LowC strains exhibit an inverse calcidiol-to-calcidiol relationship, strongly implicates that LowC is caused by disrupted transport of calcidiol into the kidney for calcitriol production. This presents a unique opportunity to further elucidate the processes regulating renal transport of calcidiol and its critical role in calcitriol production.

Calcitriol performs most of the known biologically relevant functions of VitD (1). However, the molecular and physiological impacts of calcitriol insufficiency have primarily only been studied within the context of calcidiol-dependent effects or in the context of chronic kidney disease (76, 77). The LowC strains defined here provide a new genetically determined context for low calcitriol, which can be used to characterize the direct impacts of low calcitriol that are calcidiol independent. Our findings present strong evidence of functional effects of LowC implicating impaired VitD signaling. LowC strains exhibited reduced transcript and protein levels of Vdr, implicating a reduction in transcriptional activation of Vdr driven by calcitriol insufficiency (71). This is supported by our finding that LowC strains exhibit reduced expression of well-established VitD target genes (s100g, Klotho, Npt2a, and Npt2c) that are normally positively regulated by the binding of the calcitriol-bound VDR protein complex to VDREs in the gene promoters (32, 78). The most well-studied physiological role for calcitriol is regulating calcium homeostasis (27) and its role in bone health. Although LowC strains did not all exhibit reduced serum calcium concentrations, there was a significant difference in calcium concentrations between the LowC strain with the lowest calcitriol (CC006) and the HighC strain with the highest (CC001). Although previous studies using CC strains have demonstrated significant variability in bone health parameters that were linked to genetic variability among the strains (79, 80), bone health phenotypes have not been assessed for the specific CC strains tested here. Nonetheless, it is important to note that LowC strains exhibit serum calcitriol concentrations similar to those induced by acute dietary VitD depletion and are below the threshold (<60 pM) shown to reduce bone mineral density in a C56BL/6J strain dietary depletion model (66). However, additional studies are required to determine whether LowC in these CC models impacts bone health outcomes. These novel findings strongly implicate that genetic lineages with LowC have impaired VitD signaling and warrants further investigation into molecular mechanisms driving these effects and the physiological impact of these molecular changes.

Although informative, our study design had several limitations, including limited capacity to fully define the strain-specific effects of diet on VitD metabolites and relationships between genotype and phenotype. These analyses require substantially increased sample sizes and strain numbers in addition to expanded assessment of genetic differences at intergenic regulatory regions, such as enhancers, which play critical roles in regulating complex polygenic phenotypes. Furthermore, the short duration of VDD (6 weeks) limited our study to acute responses and likely did not capture the full range of adaptive responses, whereas the focus on only VitD3 metabolites leaves the effects on D2 metabolites unknown. Finally, because of our research focus on maternal VitD status, this study characterized effects in females only. Further work is necessary to understand the effects in males and investigate the presence of sex-specific effects.

Conclusions

These findings highlight significant gaps in our understanding of the complex genetic architecture of VitD insufficiency and establish the CC genetic reference population as a valuable model system for further elucidating factors regulating the availability of this important nutrient. Studying VitD metabolism within the context of a genetically diverse mouse population where environmental and genetic contributors are controllable allows for distinguishing effects driven by genotype vs environment that are not easily separated in human populations. Our findings provide definitive evidence that genetic factors in the population are significant contributors to the discordance between calcidiol and calcitriol and can substantially impact the ability of serum calcidiol concentrations to predict calcitriol sufficiency. Importantly, the causative genetic determinants are likely acting outside of the current canonical VitD metabolism enzymes and may instead be driven by impaired renal transport of calcidiol. This work lays the foundation for identifying causal genetic factors that drive LowC in the mouse and subsequently translating this knowledge across species (including, but not limited to, humans) to gain a better understanding of the biological and clinical relevance.

Abbreviations

CC

Collaborative Cross

DBP

vitamin D-binding protein

HighC

high calcitriol

LC/MS/MS

liquid chromatography tandem mass spectrometry

LowC

low calcitriol

SNP

single nucleotide polymorphism

VitD

vitamin D

VDD

vitamin D depleted

VDR

vitamin D receptor

VDRE

vitamin D response element

VDS

vitamin D sufficient

Funding

This work was supported by the National Institutes of Health—National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK, P30DK056350—P&F award to F.I., P30DK056350 to V.S.V., R00-DK129705 to R.A.); and National Institute of Environmental Health Sciences (NIEHS, P30ES010126—P&F award to F.I., R35ES030443 and P42ES004699 to B.H.). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This study was also supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES Finance Code 001 to C.V.C.).

Disclosures

B.D.H. is a cofounder of EicOsis Human Health. The remaining authors have nothing to disclose.

Data Availability

The analyzed data from this study are presented in the Materials and Methods and Results sections of this manuscript. Raw data and metadata are available on request from the corresponding author. Supplementary Figures and Tables are available under the collection “Supplementary Data for Hutchins et al (2025) “Interindividual genetic differences drive discordance between serum calcidiol and calcitriol concentrations” (65).