Dietary fructose inhibits lactation-induced adaptations in rat 1,25-(OH)₂D₃ synthesis and calcium transport

Abstract

We recently showed that excessive fructose consumption, already associated with numerous metabolic abnormalities, reduces rates of intestinal Ca²⁺ transport. Using a rat lactation model with increased Ca²⁺ requirements, we tested the hypothesis that mechanisms underlying these inhibitory effects of fructose involve reductions in renal synthesis of 1,25-(OH)₂D₃. Pregnant and virgin (control) rats were fed isocaloric fructose or, as controls, glucose, and starch diets from d 2 of gestation to the end of lactation. Compared to virgins, lactating dams fed glucose or starch had higher rates of intestinal transcellular Ca²⁺ transport, elevated intestinal and renal expression of Ca²⁺ channels, Ca²⁺-binding proteins, and CaATPases, as well as increased levels of 25-(OH)D₃ and 1,25-(OH)₂D₃. Fructose consumption prevented almost all of these lactation-induced increases, and reduced vitamin D receptor binding to promoter regions of Ca²⁺ channels and binding proteins. Changes in 1,25-(OH)₂D₃ level were tightly correlated with alterations in expression of 1α-hydroxylase but not with levels of parathyroid hormone and of 24-hydroxylase. Bone mineral density, content, and mechanical strength each decreased with lactation, but then fructose exacerbated these effects. When Ca²⁺ requirements increase during lactation or similar physiologically challenging conditions, excessive fructose consumption may perturb Ca²⁺ homeostasis because of fructose-induced reductions in synthesis of 1,25-(OH)₂D₃.—Douard, V., Suzuki, T., Sabbagh, Y., Lee, J., Shapses, S., Lin, S., Ferraris, R. P. Dietary fructose inhibits lactation-induced adaptations in rat 1,25-(OH)₂D₃ synthesis and calcium transport.

Pregnancy and lactation are unusually challenging periods for the maintenance of Ca²⁺ and vitamin D homeostasis. Ca²⁺ transfer to the fetus through the placenta during the final trimester of gestation and subsequently to the neonate through the mammary glands during lactation can reach 250 and 500 mg/d, respectively, in humans, as well as 50 and 200 mg/d, respectively, in rats (1). These high rates of Ca²⁺ flux between mother and fetus or neonate are feasible only if there is a greater intake of dietary Ca²⁺ and if intestinal (1) and renal (2) Ca²⁺ transport systems are operating efficiently. The primary method by which mammals adapt to long-term physiological and nutritional challenges requiring more Ca²⁺ is by chronic increases in intestinal Ca²⁺ absorption (3), achieved by increasing levels of the active form of vitamin D, 1,25(OH)₂D₃ (2), whose genomic effects are mediated by the interaction of the activated nuclear vitamin D receptor (VDR) with vitamin D target genes (4). While the kidney can make acute adjustments in renal Ca²⁺ excretion to compensate for rapid changes in levels of blood Ca²⁺, a secondary method of adaptation to lactation involves modest chronic increases in renal Ca²⁺ reabsorption (2).

In humans, defects in the development of the newborn skeleton are associated with maternal vitamin D deficiency (5). Preeclampsia and hypertensive disorders during pregnancy have been associated with lower plasma levels of ionized Ca²⁺ and vitamin D (6). Incidents of hypocalcemia and deaths increase in pregnant rodents fed low-Ca²⁺ diets, in vitamin D-deficient pregnant rats, or in VDR-null mice (7, 8). Moreover, Ca²⁺ plays a crucial role not only in maintaining a healthy skeleton but also in regulating muscle contraction, blood clotting, nerve activity, cell signaling, and enzyme function (9). In turn, Ca²⁺ homeostasis and skeletal integrity are regulated by vitamin D, which also has numerous extraskeletal effects related to hypertension, dyslipidemia, diabetes mellitus, and coronary artery disease (9). Thus, to sustain conditions optimal for maternal and infant health, perturbations in intestinal and renal Ca²⁺ transport and vitamin D synthesis must be avoided during pregnancy and lactation.

Ca²⁺ enters the apical membrane through specialized Ca²⁺ channels, transient receptor potential vanilloid 5 and 6 (TRPV5 and TRPV6). Ca²⁺ then diffuses through the cytosol while bound to Ca²⁺-binding proteins (calbindins, CaBP9k, and CaBP28k). Finally, exit across the basolateral membrane is mediated by a Na⁺/Ca²⁺ exchanger (NCX1) along with the plasma membrane Ca²⁺-ATPase (PMCA1). PMCA1 and NCX1 are found in the kidney and intestine. TRPV6 and CaBP9k are mainly involved in intestinal Ca²⁺ transport, while TRPV5 and CaBP28k are primarily responsible for renal Ca²⁺ reabsorption. Failure to adaptively increase rates of intestinal and renal Ca²⁺ absorption, as well as levels of 1,25(OH)₂D₃ during pregnancy and lactation, will perturb Ca²⁺ homeostasis, so that Ca²⁺ is resorbed from maternal bones to meet fetal demand (1). Thus, optimal serum Ca²⁺ concentrations are maintained by a balance of intestinal absorption, renal excretion, and bone remodeling, and this balance is tightly controlled primarily by parathyroid hormone (PTH), calcitonin, and 1,25-(OH)₂D₃.

Recently, we found that in nephrectomized rats displaying renal insufficiency, intestinal luminal fructose markedly inhibited active intestinal absorption of Ca²⁺, as well as expression levels of CaBP9k and TRPV6 (10). However, the nephrectomized rats were also hyperphosphatemic and hypercalcemic. These findings indicate that excessive consumption of fructose may adversely affect intestinal Ca²⁺ transport when renal function is compromised[b]. However, it is not clear whether this fructose-induced perturbation in Ca²⁺ homeostasis occurs during pregnancy and lactation when the demand for Ca²⁺ and 1,25-(OH)₂D₃ is high.

In humans, 1,25-(OH)₂D₃ synthesis and intestinal Ca²⁺ transport peak during pregnancy (1). Because these adaptive increases occur during lactation in rats, the lactating rat is an appropriate model to evaluate whether a high fructose intake perturbs active Ca²⁺ transport and whether the mechanism is a fructose-induced reduction in 1,25-(OH)₂D₃ synthesis. In this study, we show that chronic consumption of a high-fructose diet from gestation through lactation prevents adaptive increases in expression and activity of intestinal and renal Ca²⁺-transporting and -binding proteins, as well as exacerbates the effects of lactation on Ca²⁺ homeostasis and bone quality. We then show that fructose specifically prevents lactation-induced increases in TRPV5 and 6, as well as CaBP9k and 28k expression in the duodenum and the kidney, respectively. Fructose does not alter PTH levels but inhibits the lactation-induced increase in renal expression of 1α-hydroxylase, a key enzyme in the 1,25-(OH)₂D₃ metabolism, thereby preventing increases in blood levels of 1,25-(OH)₂D_3. This failure to increase 1,25-(OH)₂D₃ levels results in reduced VDR binding to the promoter regions of TRPV6 and CaBP9k.

MATERIALS AND METHODS

Animals

All of the procedures in this study were approved by the Institutional Animal Care and Use Committee, University of Medicine and Dentistry of New Jersey (UMDNJ)–New Jersey Medical School. Studies were conducted on timed-pregnant primiparous Sprague-Dawley rats (∼2 mo old, ∼200 g, and 2 d of gestation at the beginning of the experiment) and virgin rats of similar age and weight (Charles River Laboratories, Wilmington, MA, USA). Rats were kept under standard conditions: 12-h light-dark cycle and 22–24°C.

Experimental design

Pregnant and virgin females were each randomly assigned to 3 groups fed a 63% glucose, a 63% fructose, or a 63% starch diet modified from a standard, published American Institute of Nutrition (AIN)-93G formula containing normal Ca²⁺ and P_i levels. These isocaloric diets are designed to meet the nutrient requirements of gestation and lactation in rodents. Animals were fed the high-carbohydrate diets ad libitum for 6 wk (from d 2 of gestation until the end of lactation, which corresponded to d 21 after birth). The day after parturition, litters were normalized to 8 pups/dam. Blood sampling from the tail vein was done on d 21 of lactation (prior to removal of pups). Same-age virgin controls were also sampled on the same days (Supplemental Fig. S1).

In vitro intestinal transport measurements

All of the intestine segments were everted quickly after isolation and prepared as everted sacs or sleeves to determine nutrient transport rates at 37°C with 95% O₂:5% CO₂, as described previously (10).

Ca²⁺ uptake

The everted gut sacs were made by using the first 4 cm of proximal duodenum where active transcellular transport of Ca²⁺ is localized (11) and then incubated in Ca²⁺ transport buffer, as described previously (10). In the intestinal regional compensation experiment, tissues were also taken from the jejunum. The outer luminal and inner serosal compartments had equal initial concentrations (0.25 mM) of nonradioactive Ca²⁺, then ⁴⁵Ca²⁺ was added to the outer mucosal compartment. After 1 h, the active accumulation of ⁴⁵Ca²⁺ in the inner serosal compartment was calculated as a ratio of the final concentration of (⁴⁵Ca²⁺ serosal/⁴⁵Ca²⁺ mucosal compartments) then normalized to that of virgins fed starch. Ca²⁺ accumulation in the tissue was expressed as nanomoles per milligram and normalized to that of virgins fed starch (10). Paracellular Ca²⁺ transport was determined in uneverted sacs, so ⁴⁵Ca²⁺ was added to the serosal side.

Fructose and glucose uptake

Four 1-cm jejunal segments were made into everted sleeves, mounted on rods, and preincubated for 5 min in Krebs-Ringer bicarbonate (KRB), as described previously (12). Two segments each were then incubated in 50 mM glucose or fructose KRB solutions containing tracer concentrations of ¹⁴C-glucose or ¹⁴C-fructose, respectively.

Phosphate uptake

Intestinal P_i transport was determined in two consecutive 4-cm segments of medial jejunum using the previously described everted gut sac assay (10). Briefly, the everted intestinal segment for determining total P_i transport was incubated for 1 h in Na⁺-containing P_i transport buffer (1.2 mM Pi), while the adjoining segment for determining Na⁺-independent P_i transport was incubated in Na⁺-free transport buffer. The active (total P_i less Na⁺-independent Pi) transport of ³³P_i in the serosal fluid was expressed as a ratio of the final concentration of (³³P_i serosal/mucosal compartments) and normalized to the ratio obtained from sacs of virgin rats fed starch.

Measurements of serum clinical parameters

Following earlier work (10), blood urea nitrogen (BUN) concentrations were determined using QuantiChrom urea assay kit (BioAssay Systems, Hayward, CA, USA), while P_i concentrations were determined using QuantiChrom P_i assay kit. The total serum Ca²⁺ concentrations were determined by previously described techniques using flame atomic absorption spectrophotometry (Perkin Elmer Model 603; Perkin Elmer, Norwalk, CT, USA; ref. 10). Glucose and fructose concentrations in the serum were determined using glucose and fructose assay kits, respectively (Biovision Research, Mountain View, CA, USA).

Vitamin D and PTH assays

Following earlier work (10), serum 1,25-(OH)₂D₃ levels were measured by enzyme immunoassay [ImmunoDiagnostic Systems (IDS), Fountain Hills, AZ, USA]. Briefly, serum samples were delipidated, and 1,25-(OH)₂D₃ was immunoextracted before the assay. Serum 25-(OH)D₃ levels were measured directly (IDS). PTH was determined using a 2-site sandwich, intact enzyme-linked immunosorbent assay (ELISA; Immunotopics, San Clemente, CA, USA).

Western blot analysis

Western blot analysis was performed using 50 μg of intestinal or renal protein extracts following earlier work (10). For CaBP9k (Swant Swiss Antibodies, Marly, Switzerland), 4–20% Tris-HCl gel (Bio-Rad, Hercules, CA, USA) was used. For CYP27B1, CYP24A1, and VDR (all from Santa Cruz Biotechnology, Santa Cruz, CA, USA) and CaBP28K (kindly provided by Dr. Sylvia Christakos, UMDNJ), a 12% Tris-HCl gel was used. All membranes were stripped and reprobed with β-actin antibody (Chemicon International, Billerica, MA, USA).

Real-time PCR

Total RNA was isolated, reverse transcription was done, and real-time PCR was performed using Mx3000P (Stratagene, La Jolla, CA, USA), as described previously (13). The control group was the virgin rats fed starch. The reference gene was α-elongation factor 1, EF1α whose expression is independent of maternal status and dietary carbohydrate (12).

Previously published primer sequences and annealing temperatures for CaBP9k, GLUT2, NaPi2b, NCX1, PMCA1, and TRPV6 are listed in ref. 10, and those for EF1α, GLUT5, and SGLT1 are listed in ref. 12. Primers for the following genes are (forward; reverse; annealing temperature): CaBP28k: 5′-GAAGGAAAGGAGCTGCAGAA-3′; 5′-TTCATCTCAGGTGATAGCTCCA-3′; 56°C; CYP24A1: 5′-TGGATGAGCTGTGCGATGA-3′; 5′-TGCTTTCAAAGGACCACTTGTTC-3′; 55°C; CYP27B1: 5′-CAACTCGGGGGTTAACTAACAG-3′; 5′-AAGCATGGAAGGATCAGTGG-3′; 57°C; TRPV5: 5′-GCTTTCCTCCAAGAAGATAGAGG-3′; 5′-GGGTTGTCCATATTTCTTCCAC-3′; 56°C; and VDR: 5′-TATTCTCCAAGGCCCACACT-3′; 5′-CGGATGGTTCCATCATGTCT-3′; 56°C.

Histology and immunohistochemistry

Immunohistochemistry was performed on paraffin slides using rabbit anti-mouse CYP27B1 (Chemicon International; 1:200 in 1% BSA in PBS) or rabbit anti-rat GLUT5 (Chemicon International; 1:500 in 1% BSA in PBS) overnight at 4°C, then washed in PBS. The secondary antibody was goat anti-rabbit IgG labeled with Cy3 (Chemicon International; 1:100), diluted in 1% BSA in PBS, and applied for 1 h at 24°C. The stained sections were examined at ×20 or ×40 magnification with a laser scanning confocal microscope (Bio-Rad; Radiance 2100). All images of sections being compared were obtained with the same settings of the microscope. Nonspecific staining with secondary antibodies only was consistently negligible.

Chromatin immunoprecipitation (ChIP)

After being thawed, samples were prepared for immunoprecipitation, as described previously (14). Sonicated solutions were incubated with anti-VDR (Santa Cruz), with anti-acetylated histone H3 or H4 (Millipore, Bedford, MA, USA), and rabbit IgG (control for nonspecific DNA:protein binding). The PROMO database (http://alggen.lsi.upc.es) was utilized for in silico analysis at 10% dissimilarity to the canonical VDR response elements (VDREs). Primers for ChIP were designed using Primer3 (http://primer3.sourceforge.net) and targeted as close as possible to previously identified VDRE. Regions within 300–400 bp from the cis element of nuclear receptors are able to detect binding as previously shown (15). Primers for ChIP assay of CABP9k are as follows (promoter region: forward; reverse): −3872 to −3797: 5′-ACAGGCCTTGTGTAGTCCAAGT-3′; 5′-GGGAATCCGAAGTTTAAGGTCT-3′; −1309 to −1241: 5′-CAACACAGGAAGCAGAAGGAG-3′; 5′-CCTCCCACTCACTATAAAGACCA-3′; −1049 to −963: 5′- CCATTCCCCTACAGTCATTCA-3′; 5′-GTCCCTGGAACTTATACTGATGCT-3′; and −76 to +60: 5′-TTCATAATCAGGGTGGTGTGTC-3′; 5′-TCAGACAGGAACAGGTGAGGT-3′. Primers for TRPV6 are −4923 to −4821: 5′-GAGCCAAACAGTCTCCCACTA-3′; 5′-TCCTCTTTCCTCCACTTACTTCC-3′; −1295 to −1229: 5′-CCCACATACACATATGCAAGACC-3′; 5′-TGCAAGCTTTGAGTTGACTCTG-3′; −586 to −516: 5′-CTACGTAACCCACCTACCACTAGAA −3′; 5′-AGGCTGGTGAGTCAGGGTATAA-3′; −217 to −131: 5′-GTGCCAAGAGTCTTTCAAGTGAG-3′; 5′-GAGCTGGTTCCTGGACTTTG-3′ and −62 to +2: 5′-GCCTTGGTAGGCAGGACTTT-3′; 5′-CCTCCCTCCTCCCAAATTAC-3′.

Determinations of bone mineral content (BMC), bone mineral density (BMD), and bone mechanical strength

Whole-body BMD and BMC were evaluated using dual energy X-ray absorptiometry (Lunar PIXImus; GE Medical Systems, Madison, WI, USA), as described previously (10). After whole-body measurements, the left and right femora were dissected, and muscle fibers were removed. The left femur and the femoral neck were scanned for BMC and BMD. The right femur was used for bone mechanical testing, as described previously (16). Proximal and distal ends were embedded in Field's metal (Alfa Aesar, Ward Hill, MA, USA) and torsionally tested to failure at a rate of 2.0 deg/s. Polar moment of inertia (about the centroidal axis, represented by the longitudinal axis of the bone), torque to failure, structural stiffness, and ultimate shear stress were obtained through standard equations, modeling each femur as a hollow ellipse.

Statistical analyses

Data are presented as means ± se. If an initial 2-way ANOVA indicated a significant effect of lactation and/or diet, a 1-way ANOVA followed by LSD test (StatView, Abacus Concepts, Piscataway, NJ, USA) were used to determine differences among means. The P values from the 2-way ANOVA are in Supplemental Table S1, those from the 1-way are mentioned in the text. Simple linear regressions with the mean of each group were used to examine the relationship between 1,25-(OH)₂D₃ levels and intestinal Ca²⁺ transport, between 1,25-(OH)₂D₃ levels and CYP27B1 expression, and between CYP24A1 or CYP27B1 expression and PTH levels.

RESULTS

Lactation increased body weight and food intake

Throughout gestation, food intake was the same in all 6 groups (Supplemental Fig. S2). Shortly after parturition, lactating rats gradually increased their food intake, such that by the end of the lactation, they were eating ∼3 times more than virgins. Since diet had no effect on feeding rate, differences in caloric intake did not confound results. During gestation, body weight of the lactating rats increased by 60% (from ∼200 to 330 g), that of virgins increased by only 35% (from ∼200 to 270 g). By the end of lactation, the body weights of dams and virgins were similar (∼310 g) and were ∼150% of original. Dietary carbohydrate had no significant effect on body weight.

Ca²⁺, sugar, and P_i transporter activity and expression change with lactation and diet

Both lactation and diet significantly affected active transepithelial Ca²⁺ transport in the duodenum (Fig. 1A and Supplemental Table S1). Rates of duodenal transepithelial Ca²⁺ transport were ∼60% lower in dams fed fructose compared to those fed glucose or starch (P=0.008) and were, in fact, similar to those of virgins. Likewise, Ca²⁺ accumulation in the duodenal mucosa increased in dams relative to virgins but failed to increase in dams fed a fructose diet (P=0.003; Fig. 1B). This suggests that dietary fructose prevented a lactation-induced increase in Ca²⁺ transport.

Figure 1.

Effects of dietary sugars on intestinal Ca²⁺ transport in lactating and virgin rats. A) Active transepithelial Ca²⁺ transport from the luminal to the basolateral compartment was initially expressed as a ratio of [⁴⁵Ca²⁺ inside (basolateral)/⁴⁵Ca²⁺ outside (luminal)] of everted gut sacs obtained from the duodenum of lactating dams and virgin rats fed a glucose (G), fructose (F) or starch (S) diet for 6 wk. B) Ca²⁺ uptake by the intestinal cells was calculated as the accumulation of ⁴⁵Ca²⁺ dpm · mg⁻¹ · min⁻¹ of duodenum. Data in panels A and B were normalized to levels in virgins fed S. Bars represent relative transport rate (n=5–8/group). C) Active transepithelial Ca²⁺ transport was measured using everted intestinal sacs from the duodenum or proximal jejunum of dams fed a G, F, or S diet for 6 wk. Data were normalized relative to transepithelial Ca²⁺ transport in the duodenum of pregnant dams fed S (n=4/group). D) Paracellular Ca²⁺ transport was measured in the same rats in panel C, using noneverted intestinal sac from the proximal jejunum. Data were normalized to paracellular Ca²⁺ transport in the jejunum of pregnant dams fed S. Values represent means ± se. Differences (P<0.05) among means, indicated by differences in superscript letters, were analyzed by 1-way ANOVA LSD after a 2-way ANOVA showed significant diet or lactation effects. ^♦P < 0.05 vs. virgins.

We then verified that the fructose-induced decrease in duodenal Ca²⁺ transport in dams was not compensated by increases in Ca²⁺ transport in the more distal regions because jejunal Ca²⁺ transport was, for all diets, less than that in the duodenum (P<0.01; Fig. 1C). However, dietary fructose still tended to decrease jejunal Ca²⁺ transport. The fructose-induced decrease was also not compensated for by increases in paracellular Ca²⁺ transport (P>0.50; Fig. 1D).

To determine whether changes in transport parallel changes in mRNA expression, we next show that mRNA levels of the apical Ca²⁺ transporter, TRPV6 (Fig. 2A), the intracellular Ca²⁺-binding protein, CaBP9k (Fig. 2B), and the basolateral Ca²⁺ transporters PMCA1 (Fig. 2C) increased with lactation while, surprisingly, the expression of the second basolateral Ca²⁺ transporter NCX1 decreased (Fig. 2D). Dams fed fructose clearly failed to increase expression of TRPV6 (P<0.05) and CaBP9k (P<0.001). Levels of PMCA1 and NCX1 mRNA did not vary with diet. Changes in levels of CaBP9k protein paralleled those of its mRNA (Fig. 2B). Results suggest that the fructose-induced reduction of transepithelial Ca²⁺ transport stems from the failure of fructose-fed dams to stimulate intestinal expression of TRPV6, CaBP9k, and maybe even PMCA1.

Figure 2.

Effects of dietary sugars on expression of intestinal Ca²⁺ transporters in lactating and virgin rats. mRNA and protein expression levels were measured in the duodenum of lactating and virgin rats that had been fed a glucose (G), fructose (F), or starch (S) diet for 6 wk. A) mRNA level of TRPV6. B) mRNA and protein levels of CaBP9k (with β-actin as loading and transfer control for the Western blot). C) mRNA levels of PMCA1. D) mRNA levels of NCX1. mRNA levels were analyzed by real-time PCR using EF1α as a reference. Data were normalized to levels in virgin rats fed S. Attempts to determine protein levels of TRPV6, PMCA1, and NCX1 failed. Values represent means ± se (n=5–8/group). Differences (P<0.05) among means analyzed and presented as in Fig. 1.

Sugars

To establish that the effect of fructose on Ca²⁺ transport was specific, we show that in the same rats used in Fig. 1, the intestinal uptake of glucose (Fig. 3A) and the expression level of the intestinal Na⁺/glucose cotransporter (SGLT1; Fig. 3C) were independent of lactation and of diet. Lactation also had no effect on fructose uptake (Fig. 3B) and the mRNA abundance of the intestinal fructose transporter GLUT5 (Fig. 3D). In marked contrast to its inhibitory effect on Ca²⁺ transport, dietary fructose markedly enhanced rates of intestinal fructose uptake (P=0.04). Expression of GLUT5, but not that of the basolateral glucose/fructose transporter GLUT2 (Fig. 3E), varied with diet (P=0.002) but not with maternal status.

Figure 3.

Effects of dietary sugars on intestinal glucose and fructose uptake in lactating and virgin rats. A, B) Glucose (A) and fructose (B) uptakes were measured from the jejunum of lactating dams and virgin rats fed a glucose (G), fructose (F), or starch (S) diet for ∼6 wk. C–E) Effects of dietary sugar and lactation were also determined on the mRNA levels of the intestinal apical Na⁺/glucose cotransporter SGLT1 (C), the intestinal apical fructose transporter GLUT5 (D), and the intestinal basolateral glucose and fructose transporter GLUT2 (E). Bars, analyses, and normalization as in Fig. 1.

Phosphate

We also examined P_i as it is critical for bone formation. Active transepithelial P_i transport in the jejunum was almost nonexistent in virgins (Fig. 4A) but increased significantly with lactation but did not vary with diet. Likewise, mRNA expression of the intestinal Na⁺-dependent phosphate cotransporter, Napi2b, increased significantly with lactation but not with diet (Fig. 4B).

Figure 4.

Effects of dietary sugars on intestinal P_i absorption in lactating and virgin rats. Transepithelial P_i transport was measured using everted intestinal sacs from the proximal jejunum of lactating and virgin rats fed a glucose (G), fructose (F), or starch (S) diet for ∼6 wk. A) Active Na⁺-dependent transepithelial P_i transport from the luminal to the basolateral compartments was calculated as the total transepithelial P_i transport subtracted by Na⁺-independent P_i transport. Because of the low rates of active P_i transport in adult rats (44), which fluctuated around 0, we could not express rates relative to virgin rats. B) NaPi2b mRNA expression was analyzed by real-time PCR using EF1α as a reference. Bars, analyses, and normalization as in Fig. 1.

Blood chemistry and hormone levels

Lactation reduced serum levels of Ca²⁺ by ∼20% in all three groups of dams (Table 1 and Supplemental Table S1). There was a modest effect of diet, as serum level of Ca²⁺ was significantly higher in glucose-fed dams than in those fed fructose or starch. In contrast, lactation and diet had no effect on serum levels of Pi. Serum fructose levels varied with diet, but not lactation, as fructose consumption increased blood fructose levels in both dams and virgins. For serum glucose levels, there was a significant interaction between diet and maternal status, suggesting that the response to diet may depend on whether the rats were lactating. Thus, virgin rats fed fructose for 6 wk were significantly (P<0.05) more hyperglycemic when compared to virgin rats fed starch or glucose.

Table 1.

Fructose effects on blood chemistry: phosphate, calcium, glucose, fructose, 25-(OH) D₃, 1,25-(OH)₂ D₃ and parathyroid hormone concentrations in the blood as a function of lactation and of diet

Diet	Dam			Virgin			Significance
	Glucose	Fructose	Starch	Glucose	Fructose	Starch	Status	Diet	Diet × status
Phosphate (mg/dl)	5.4 ± 0.6	4.9 ± 0.5	5.1 ± 0.3	6.2 ± 0.7	6.2 ± 0.6	5.2 ± 0.4	0.43	0.487	0.784
Calcium (mg/dl)	11.7 ± 0.1b	10.1 ± 0.2c	10.0 ± 0.1c	12.8 ± 0.1a	12.1 ± 0.4a	12.0 ± 0.2a	<0.001	<0.05	0.862
Glucose (mM)	8.6 ± 1.3b	8.5 ± 0.8b	8.9 ± 1.5b	9.4 ± 0.8b	13.2 ± 1.2a	10.4 ± 0.8a,b	0.018	0.257	0.048
Fructose (mM)	ND	1.1 ± 0.4	ND	ND	0.8 ± 0.2	ND	0.34	<0.05	0.025
25-(OH) D3 (nM)	59 ± 5	59 ± 2	56 ± 2	49 ± 4	47 ± 7	49 ± 5	0.011	0.911	0.827
1,25-(OH)2 D3 (pM)	643 ± 134a	301 ± 59b	559 ± 88a	305 ± 55b	240 ± 34b	294 ± 53b	0.006	0.051	0.298
PTH (pg/ml)	346 ± 69b	781 ± 153a	406 ± 79b	229 ± 72b	242 ± 74b	202 ± 53b	0.034	0.412	0.519

Data are means ± se; n = 5–8/group. Significance determined by 2-way ANOVA. Superscript letters refer to results of LSD post hoc tests after 1-way ANOVA (P<0.05). Means with different superscript letters are significantly different. ND, not detectable.

Among the hormones involved in the regulation of intestinal Ca²⁺ uptake and bone resorption, 25-(OH)D₃, 1,25-(OH)₂D₃, and PTH were each up-regulated in dams compared to virgins. Diet significantly affected serum levels of 1,25-(OH)₂D₃ but not those of 25-(OH)D₃. There is a dramatic decrease in levels of 1,25-(OH)₂D₃ in dams fed fructose compared to those fed glucose or starch. Interestingly, fructose tended to induce a 2-fold increase in PTH (P=0.07, 1-way ANOVA).

Fructose reduces VDR binding and histone H4 acetylation

Because TRPV6 and CaBP9k expression is regulated by 1,25-(OH)₂D₃ primarily through the VDR, we investigated in vivo whether fructose inhibited VDR binding to their promoter regions. We only used dams that, in contrast to virgins, exhibited large-magnitude, fructose-induced changes in expression of Ca²⁺-transporting and -binding proteins. Although fructose feeding specifically decreased intestinal expression of TRPV6 and CaBP9k (Fig. 2), as well as the circulating level of 1,25-(OH)₂D₃ (Table 1), the mRNA expression and protein level of the VDR in the intestine were independent of diet (P=0.89) in dams (Fig. 5A). This suggests that fructose effects on TRPV6 and CaBP9k expression were not confounded by changes in levels of the VDR. By in silico screening, we located the putative VDREs in the promoter regions of both genes in rats (Fig. 5B). Using ChIP, we discovered that VDR binding to the promoter of TRPV6 was significantly dependent on diet and promoter region examined (2-way ANOVA; Supplemental Table S1). We then analyzed the effect of diet on VDR binding within each promoter region and found that it decreased significantly between −1429 and −1295 bp of the TRPV6 promoter in fructose-fed dams (Fig. 5C and Supplemental Table S1). This VDRE affected by diet almost exactly matches the most proximal VDRE in humans, in terms of position (17).

Figure 5.

Binding of the VDR to promoter regions of TRPV6 and CaBP9k in lactating dams. A) mRNA and protein levels of the VDR in the duodenum of dams fed glucose, fructose, or starch diet for 6 wk. VDR mRNA levels were normalized to EF1α mRNA levels, while β-actin was used as loading and transfer control in Western blots. Relative mRNA levels were then normalized to those of dams fed starch. B) Schematic representation of the approximate location (solid rectangles) of the VDREs within the TRPV6 and CaBP9k promoters and of the primers (arrows) used for ChIP assays performed using anti-VDR antibodies. C, D) ChIP signals were detected by real-time RT-PCR using the primers to detect binding in the promoter regions of TRPV6 (C) and CaBP9k (D). Sonication of our samples generated DNA fragments of 250–500 bp. Because primers within 300–400 bp from the cis element of nuclear receptors are able to detect binding (15), primers designed as indicated in panel B should allow us to detect VDR interactions with the promoter regions of interest. ChIP signals were normalized to input signals. PCR product size was confirmed by electrophoresis on a 2% agarose gel. Only one band appears with the primer sets used (Supplemental Fig. S7). Bars are means ± se (n=3), with each n being analyzed twice and results averaged. Differences (P<0.05) among means within a region, indicated by differences in superscript letters, were analyzed by 1-way ANOVA LSD after an initial 2-way ANOVA showed significant diet or promoter region effects.

VDR binding to the promoter of CaBP9k was also highly dependent on diet but did not vary among the different promoter regions examined. Here, the negative effect of fructose on VDR binding was greatest in the proximal region of the CaBP9k promoter, between −70 and +60 bp (Fig. 5D). Binding of IgG to the promoter regions of TRPV6 and CaBP9k was ≤ 0.02% of input (Supplemental Fig. S3A, B, respectively), suggesting that the binding of the VDR antibody to the TRPV6 and CaBP9k promoter regions was specific.

Histone acetylation accompanies transcription of some genes, including TRPV6 activation by 1,25-(OH)₂D₃ (18). Acetylated histone H4 binding in vivo to the promoter of TRPV6 and CaBP9k was highly dependent on diet (Supplemental Table S1). Decreases in TRPV6 expression (Fig. 2A) and VDR binding (Fig. 5C) in dams fed fructose were associated with a marked fructose-induced reduction (P<0.05 by 1-way ANOVA) of acetylation of histone H4 in each TRPV6 promoter region examined (Fig. 6A). Because fructose-induced decreases in H4 acetylation were less dramatic in the CaBP9k promoter, no significance was detected in individual promoter regions by 1-way ANOVA (Fig. 6B). H3 acetylation in the TRPV6 and CaBP9k promoters was also analyzed but it was not affected by diet (data not shown).

Figure 6.

Effect of diet on histone H4 acetylation in the TRPV6 and CaBP9k promoter regions of lactating dams. ChIP assays were performed using anti-acetylated histone H4 antibody in the small intestine of lactating dams fed glucose, fructose, or starch diet for 6 wk. ChIP signals were detected by real-time RT-PCR using the primers for designated promoter regions of TRPV6 (A) and CaBP9k (B). ChIP signals were normalized to input signals. Bars are means ± se (n=3 or 4), with each n analyzed 2 times. Differences (P<0.05) among means within a region are indicated by differences in superscript letters, as described in Fig. 5.

Fructose induces renal hypertrophy but decreases renal synthesis of 1,25-(OH)₂D₃

We examined the effects of fructose on the kidney since it is the major organ system involved in 1,25-(OH)₂D₃ synthesis that decreases with fructose. Dietary fructose significantly increased the kidney mass in virgins and dams (P<0.05, data not shown). The renal somatic index varied with lactation and with diet (Fig. 7A). Dietary fructose increased renal somatic indices of dams (P=0.0073) and tended to increase those of virgins (P=0.07). Despite the hypertrophy of the kidney (Fig. 7B), renal excretion did not appear to be impaired by dietary fructose or be altered by lactation since there were no significant changes in levels of BUN (Supplemental Fig. S4) that were normal (<30 mg/dl) and similar to previous work (10).

Figure 7.

Effect of diet on the kidney morphology of lactating dam and virgin rats. Kidneys were obtained from dams (D) or virgin (V) rats fed a glucose (G), fructose (F), or starch (S) diet for 6 wk. A) Kidney somatic index. Bars are means ± se; n=5–8. Differences (P<0.05) among means are indicated by differences in superscript letters, as analyzed by 1-way ANOVA LSD after an initial 2-way ANOVA showed diet effects. B) Hematoxylin-and-eosin staining of a transverse section at the longest point of the right kidney of a rat representing each diet group. Length (horizontal arrow with grid) and width (vertical arrow) of the VS kidney were determined initially, and then the same-sized arrows superimposed on the sections were obtained from DG, DF, DS, VG, and VF rats, to depict differences in size. Results indicate that the representative, randomly chosen VS kidney is smaller than those of dams, particularly that of dams fed fructose.

Levels of CYP27B1 mRNA varied with lactation and with diet (Fig. 8A). CYP27B1 mRNA and protein expression increased in dams fed glucose and starch (P<0.006) but not in dams fed fructose, whose CYP27B1 expression levels were low and remained similar to those of the virgin rats. Thus, a high intake of fructose prevented the lactation-induced increase in expression of CYP27B1 mRNA and protein. In contrast, levels of 24-hydroxylase CYP24A1 mRNA and protein remained independent of maternal status and diet (Fig. 8B). In the renal cortical area of kidneys from dams fed glucose or starch, CYP27B1 expression is higher in the proximal tubules (Fig. 8C, yellow solid arrows in panels DG and DS) than in the distal tubules (Fig. 8C, white solid arrows) or glomeruli (Fig. 8C, blue solid arrows). In dams fed fructose (Fig. 8C, panel DF), there is reduced expression of CYP27B1 in the proximal tubule (Fig. 8C, dotted yellow arrow), such that expression in the proximal became similar to that in the glomerulus (Fig. 8C, dashed blue arrow). Activity of 1α-hydroxylase, obtained as a ratio of 1,25-(OH)₂D₃/25-(OH)D₃, varied with lactation but not with diet (Fig. 8D). A high fructose intake is associated with a nonincrease in 1α-hydroxylase activity, because dams fed glucose or starch had >2-fold increases (P<0.02) in activity when compared to that of virgins and of dams fed fructose.

Figure 8.

Effect of diet on renal enzymes involved in the synthesis and degradation of 1,25-(OH)₂D₃ in lactating and virgin rats. Kidney homogenates were obtained from dams (D) and virgin rats fed a glucose (G), fructose (F), or starch (S) diet for 6 wk. A) mRNA and protein levels of 1α-hydroxylase CYP27B1. mRNA levels were normalized to EF1α mRNA levels, while β-actin was used as loading and transfer control in Western blots. B) mRNA and protein levels of 24-hydroxylase CYP24A1. mRNA expression data were normalized relative to levels in virgins fed S. C) Immunolocalization of CYP27B1 in the kidney of dams. Yellow solid arrows indicate 1α-hydroxylase occupying the basolateral cytosolic pole of cells lining the proximal tubules (p) in panels DG and DS. White arrows indicate more modest quantities of 1α-hydroxylase from distal tubules (d), while blue arrows indicate its virtual absence from glomeruli (g) of panels DG and DS. In dams fed fructose (panel DF), distal tubular and glomerular cells also have low abundance of CYP27B1, and there is reduced expression (dotted yellow arrows) of CYP27B1 in the proximal tubule. CYP27B1 immunolocalization for virgin rats is not shown since no effect of dietary fructose was observed (see panel A). D) Relative 1α-hydroxylase activity expressed as ratio of 1,25-(OH)₂D₃ /25-(OH)D₃. Bars, analyses, and normalization as in Fig. 1.

Fructose specifically reduces expression of Ca²⁺-transporting and -binding proteins in the kidney

To determine whether fructose affected organ systems other than the small intestine, we examined the effect of fructose on the expression of renal Ca²⁺ transporters, since 1,25-(OH)₂D₃ also regulates Ca²⁺ transport in the kidney. Renal cells reabsorb Ca²⁺ through an enterocyte-like system regulated by 1,25-(OH)₂D₃. TRPV5, CaBP28k, PMCA1 and NCX1 expression each varied with lactation and tended to vary with diet, with a significant interaction between maternal status and diet for TRPV5 (Supplemental Table S1). Subsequent 1-way ANOVA indicated that for TRPV5 (P<0.017) and CaBP28k (P<0.015), dietary fructose prevented lactation-induced increases in expression (Fig. 9A, B). For PMCA1 (P<0.09) and NCX1 (P<0.07), there was a tendency for fructose to inhibit the lactation response. Thus, decreased expression levels of renal Ca²⁺-transporting and -binding proteins suggest a possible reduction in renal Ca²⁺ reabsorption in fructose-fed dams.

Figure 9.

Effects of dietary sugar on expression of renal Ca²⁺ transporters in lactating and virgin rats. mRNA and protein expression levels were measured in the kidney homogenates of lactating dams and virgin rats fed a glucose (G), fructose (F), or starch (S) diet for 6 wk. A, B) mRNA levels of TRPV5 (A) and protein levels of CaBP28k (B). mRNA levels were normalized to EF1α mRNA levels, while β-actin was used as loading and transfer control in Western blots. C) mRNA levels of PMCA1. D) mRNA levels of NCX1. All data were normalized to levels in virgins fed S. Bars are means ± se (n=5–8/group). Differences (P<0.05) among means were analyzed and shown as described Fig. 1.

The effect of lactation on the renal Ca²⁺ transport system is specific, because lactation has no effect on GLUT5 expression (Supplemental Fig. S5). The inhibitory effect of dietary fructose on renal Ca²⁺ transporters is also specific, because fructose induces renal GLUT5 expression. GLUT5 was expressed mainly in the proximal straight tubule, as was also shown by Ebert et al. (33), and its up-regulation by fructose was limited to those specific cells (Supplemental Fig. S5; compare panels DF and VF to DG, DS, VG, and VS).

Bone mineral density and mechanical properties

The physiological demands of lactation may have been met by compromising bone quality and mechanical properties. The BMC of the whole body decreased by 20–40% in dams compared to virgins (Table 2). Dietary fructose exacerbated the lactation effect, because it led to a 10–30% decrease in BMC. Since body weights did not change, the source of decreases in BMC was likely a similar decrease in BMD. Indeed, BMD of the whole body decreased by 20–30% as a function of lactation, and by an additional ∼10% with dietary fructose. It is not clear why dams fed glucose also decreased in BMD as with dams fed fructose. Similar results were observed for the isolated whole femur, a mainly cortical bone, and for the femoral neck, a more trabecular area. Total femur BMC and BMD were ∼30% lower in fructose- and glucose-fed dams compared to those fed starch. Total femur BMC and BMD were also ∼30% less in fructose- and glucose-fed dams when compared to those of virgins fed the same diets. Interestingly, in the femoral neck, a more pronounced lactation effect was observed so that even starch-fed dams displayed a significant decrease in BMD (P<0.05).

Table 2.

Lactation and fructose effects on mineral content and mechanical properties of femora from dams and virgin rats fed high levels of glucose, fructose, or starch

Diet	Dam			Virgin			Significance
	Glucose	Fructose	Starch	Glucose	Fructose	Starch	Status	Diet	Status × diet
Body BMC (mg)	5.54 ± 0.13c,b	5.26 ± 0.26c	6.86 ± 0.28a	7.74 ± 0.63a	6.84 ± 0.65a,b	7.97 ± 0.63a	<0.0005	0.046	0.564
Body BMD (g/cm2)	0.115 ± 0.005d,c	0.114 ± 0.002d	0.130 ± 0.003c,b	0.153 ± 0.007a	0.141 ± 0.007b,a	0.155 ± 0.008a	<0.0001	0.056	0.457
Total femur BMC (mg)	0.268 ± 0.011b	0.283 ± 0.08b	0.351 ± 0.023a	0.365 ± 0.042a	0.373 ± 0.026a	0.369 ± 0.033a	0.001	0.057	0.108
Total femur BMD (g/cm2)	0.113 ± 0.002b	0.122 ± 0.003b	0.143 ± 0.006a	0.158 ± 0.01a	0.158 ± 0.006a	0.160 ± 0.08a	<0.0001	0.231	0.247
Neck femur BMC (mg)	0.007 ± 2E−4b	0.007 ± 1.8E−4b	0.008 ± 2.5E−4a	0.009 ± 0.001a	0.009 ± 4.5E−4a	0.009 ± 0.001a	<0.0001	0.288	0.728
Neck femur BMD (g/cm2)	0.119 ± 0.002c	0.130 ± 0.002c	0.145 ± 0.006b	0.170 ± 0.014a	0.166 ± 0.006a	0.174 ± 0.01a	<0.0001	0.123	0.381
Stiffness (N · mm)	284 ± 21a,b	241 ± 13b	313 ± 11a	338 ± 36a	324 ± 22a	300 ± 10a	0.020	0.321	0.070
Torsional rigidity (N · mm/deg)	17727 ± 1935	16693 ± 1963	16449 ± 2121	19991 ± 3367	19468 ± 2195	16948 ± 1847	0.326	0.868	0.278
Shear stress (MPa)	105 ± 7c,b	96 ± 6c	131 ± 8a	127 ± 7a,b	119 ± 8a,b	139 ± 13a	0.021	0.011	0.653
Shear modulus (mm4)	3385 ± 278	3479 ± 332	3662 ± 443	4052 ± 108	4142 ± 473	4359 ± 527	0.046	0.771	0.999

Data are means ± se; n = 5–8 per group Significance determined by 2-way ANOVA. Superscript letters refer to results of LSD post hoc tests after 1-way ANOVA (P<0.05). Means with different superscript letters are significantly different.

Biomechanical testing was then used to assess the structural properties of the femur (Table 2). Femurs from dams had significantly reduced stiffness indices compared to those of the virgins. There were also lactation-related decreases in shear stress and shear modulus but not in torsional rigidity. There were no significant dietary effects except in indices of stiffness which tended to decrease by ∼15% in dams, and the decrease was worse with fructose feeding. Likewise, shear stress tended to decrease by ∼30% in fructose- compared to starch-fed dams, and by ∼15% in fructose-fed compared to starch-fed virgins.

Correlations among factors regulating intestinal Ca²⁺ transport rates and blood 1,25-(OH)₂D₃ levels

Since CYP24A1 expression did not change, CYP27B1 expression may regulate circulating concentrations of 1,25-(OH)₂D₃, which, in turn, modulate intestinal Ca²⁺ transport; hence, we examined a simple regression between the average amount of Ca²⁺ accumulated in enterocytes with mean 1,25-(OH)₂D₃ levels and between 1,25-(OH)₂D₃ levels and mean relative CYP27B1 expression. Because of large blood or tissue requirements for hormonal and transport assays, not all analyses could be done in the same animal; hence, we plotted average instead of individual values to one another. Our data demonstrated a highly significant (P=0.0009) slope between accumulation of Ca²⁺ in intestinal cells (Fig. 10A) or transduodenal transport of Ca²⁺ (P=0.003, not shown) to serum levels of 1,25-(OH)₂D₃. In turn, we found serum levels of 1,25-(OH)₂D₃ to be tightly linked (P=0.0014) to CYP27B1 (Fig. 10B) but not CYP24A1 (P>0.88, not shown) expression, indicating that the decrease in serum level of the active form of vitamin D₃ is likely due to impaired synthesis and not increased degradation. Increased 1,25-(OH)₂D₃ synthesis may be mediated by PTH-induced increases in levels of CYP27B1. However, there was no relationship (P=0.88) between expression of CYP27B1 (Fig. 10C) or levels of 1,25-(OH)₂D₃ (P>0.5; not shown) with PTH, suggesting a fructose-induced disruption in PTH control of calcitriol synthesis.

Figure 10.

Changes in rates of intestinal Ca²⁺ transport varied with levels of 1,25-(OH)₂D₃ and abundance of 1α-hydroxylase. A) Variations in mean enterocyte Ca²⁺ accumulation are strongly correlated with average concentrations of 1,25-(OH)₂D₃ in rats. B) Average circulating levels of 1,25-(OH)₂D₃ are tightly correlated with mean relative mRNA expression levels of CYP27B1. C) Relative expression of CYP27B1 mRNA is independent of PTH levels. Regression coefficients and significance (P) are in boxes.

DISCUSSION

Our study showed that consumption of high amounts of fructose impairs Ca²⁺ homeostasis during rat lactation by preventing increases in active intestinal Ca²⁺ absorption, as well as in expression of intestinal and renal Ca²⁺ transporters and binding proteins, important adaptations designed to meet lactation-related increases in Ca²⁺ requirements. This inhibition of adaptive increases in active Ca²⁺ transport appears to be driven by a fructose-induced reduction in levels of 1α-hydroxylase, thereby reducing renal synthesis and blood levels of 1,25-(OH)₂D₃.

Regulation of intestinal TRPV6 and CaBP9k by fructose and lactation

Because of the rapid growth and large number of offspring, rat lactation is associated with modest reductions in serum Ca²⁺ that normally lead to a strong induction of intestinal active Ca²⁺ transport (ref. 19 and Table 1). We discovered that excessive fructose consumption has a strong inhibitory effect on the lactation-induced increases in active intestinal Ca²⁺ transport. This effect is most likely mediated by fructose-induced reductions in levels of TRPV6, CaBP9k, and perhaps even of PMCA1.

While it had been accepted that TRPV6 and CaBP9k play a major role in intestinal active Ca²⁺ transport, recent studies using TRPV6-, CaBP9k-, or the double TRPV6/CaBP9k-knockout mice did not display a complete loss of intestinal Ca²⁺ absorption (11), suggesting that other mechanisms may play a role, such as the L-type channel Ca(v)1.3 that seems vitamin D independent (20). Indeed, total intestinal Ca²⁺ absorption in CaBP9k-null mice is normal and similar to that of wild type, due likely to compensatory increases in expression of Ca²⁺ transporter TRPV6 and PMCA1b (21). TRPV6-null mice displayed a partial reduction (∼60%) of total Ca²⁺ absorption (22), indicating the singular importance of this channel in intestinal Ca²⁺ transport. In addition, the adaptive increases in active Ca²⁺ transport induced by a low Ca²⁺ diet in TRPV6-null mice was 25% lower than that in wild type. The double TRPV6/CaBP9k-knockout mice exhibited modest but still significant active Ca²⁺ transport. Moreover, enhancement of Ca²⁺ transport by low-Ca²⁺ diets or 1,25-(OH)₂D₃ injections was 50% lower in these mutants than in the wild-type mice (11). In our study, both TRPV6 and CaBP9k were reduced specifically by fructose feeding in lactating females without compensatory increases in PMCA1 and NCX1 expression. Thus, increased TRPV6 and/or CaBP9k expression is likely necessary to achieve a full induction of active Ca²⁺ transport in response to dietary Ca²⁺ insufficiency, vitamin D supplementation (11), and lactation (present work).

An elevated 1,25-(OH)₂D₃ level followed by activation of the VDR is one of the main mechanisms by which intestinal Ca²⁺ transport is induced during lactation. In VDR-null mice, the expression levels of CaBP9k and TRPV6, but not of PMCA1 and NCX1, are considerably reduced compared to their wild-type littermates (2, 23, 24), indicating involvement of 1,25-(OH)₂D₃ and VDR-dependent mechanisms in the regulation of CaBP9k and TRPV6 expression.

Our finding that inhibition of H4, but not H3, acetylation accompanies 1,25-(OH)₂D₃-mediated reductions in TRPV6 and CaBP9k expression, as well as in VDR binding to CaBP9k and TRPV6 promoters, is similar to an earlier finding that histone H4 acetylation of the mouse TRPV6 promoter is positively induced by 1,25-(OH)₂D₃ in vivo (18). Nevertheless, lactating VDR-null mice still induced TRPV6 expression even if the level of expression of this gene was lower than that of lactating wild-type mice (2), suggesting a modest regulatory role by other hormones. In male mice, induction of intestinal TRPV6 and CaBP9k expression and of active Ca²⁺ transport by 1,25-(OH)₂D₃ is amplified by injections of prolactin. However, even if prolactin itself modestly increases TRPV6 mRNA levels, it has no effect on CaBP9k expression (25).

Dietary fructose can also perturb active Ca²⁺ transport by reducing activity of transporters. Since fructose reduces ATP levels (26), low ATP levels may, in turn, reduce the rate of basolateral Ca²⁺ transport since ATP depletion has been shown to inhibit NCX1 activity in various cell types (27), perhaps secondary to perturbations in activity of the Na⁺-K⁺-ATPase that affects membrane potential. Moreover, ATP is required by cAMP-dependent protein kinase, which phosphorylates and increases the activity of PMCA1 (28). Finally, TRPV6 activity is positively modulated by PiP2, whose synthesis is ATP dependent (29). Therefore, the activity of 3 of the 4 major Ca²⁺ transporters would likely be perturbed by fructose-induced reductions in cytosolic ATP concentrations. However, since chronic fructose intake reduces binding to VDREs of TRPV6 and CaBP9k and because Ca²⁺ transport is a linear function of serum levels of 1,25-(OH)₂D₃, the fructose effect is most likely driven by perturbations in the vitamin D system.

Regulation of renal TRPV5 and CaBP28k by fructose and lactation

By reabsorbing Ca²⁺ from glomerular filtrates, the kidney plays a major role in Ca²⁺ homeostasis. Although TRPV5 represents the rate-limiting step in Ca²⁺ reabsorption, Ca²⁺-binding proteins are also important. Indeed, CaBP28k-null mice overexpress CaBP9k, suggesting that when one isoform is deficient, the other must increase to maintain the rate of passage of Ca²⁺ across the cytosol (30). In our study, however, CaBP9k is modestly expressed in the kidney and is not affected by maternal status or diet (Supplemental Fig. S6); hence, CaBP9k expression does not compensate for fructose-induced decreases in abundance of CaBP28k.

The fructose-mediated decrease in 1,25-(OH)₂D₃ is also likely to reduce TRPV5 and CaBP28k expression. A deficiency in levels of TRPV5 and CaBP28k induced compensatory increases in levels of intestinal TRPV6 and CaBP9k (30), thus ensuring high rates of intestinal Ca²⁺ absorption when renal Ca²⁺ reabsorption is compromised. In our study, however, no such compensation occurred, because the fructose-induced reduction in expression of Ca²⁺-transporting and -binding proteins decreased simultaneously in both the intestine and kidney. Decreases in TRPV5 and CaBP28k expression typically result in calciuria, so that fructose consumption should increase urinary Ca²⁺ loss in dams (30). This may explain the loss of BMD observed in dams fed fructose since their ability to meet the elevated Ca²⁺ requirement of lactation was compromised by the failure to increase rates of intestinal absorption and renal reabsorption of Ca²⁺.

Effect of fructose on synthesis of 1,25-(OH)₂D₃

After considering differences in methods, as well as in dietary and blood Ca²⁺ levels, the 1,25-(OH)₂D₃ concentrations observed in virgin rats (∼275 pM, Table 1) are rather similar to those previously observed (19, 31). Even though lactation increased by ∼ 20% 25-(OH)D₃ levels, fructose-fed dams might not have converted this precursor to 1,25-(OH)₂D₃, perhaps because of fructose-induced reductions in CYP27B1 expression and in 1α-hydroxylase activity. Moreover, CYP24A1 remained unchanged among the three diets, suggesting that degradation rate via 24-hydroxylase did not change with diet. Other studies have shown that reductions in calcitriol concentrations in rats were driven also by reductions in synthesis and not degradation rates (31).

The effects of fructose and lactation on 1,25-(OH)₂D₃ levels are puzzling, as these defy classical means of CYP27B1 regulation. By feedback inhibition, CYP27B1 expression and activity are thought to be negatively regulated by 1,25-(OH)₂D₃ levels (Fig. 11A), so that high 1,25-(OH)₂D₃ levels decrease CYP27B1 expression. Our data show instead a strong positive correlation between CYP27B1 expression and 1,25-(OH)₂D₃ levels, indicating that this negative feedback loop is not working. Second, PTH, which is induced by low blood Ca²⁺ concentrations, is classically a positive regulator of 1,25-(OH)₂D₃. We found, however, that changes in PTH levels are negatively correlated with those of 1,25-(OH)₂D₃ abundance. Thus, in our model of lactating rats, the fructose-induced increase in PTH concentrations is incongruent with the fructose-induced decrease in 1,25-(OH)₂D_3. Moreover, the high level of PTH and the low level of 1,25-(OH)₂D₃, each condition a potent inducer of CYP27B1, instead failed to enhance CYP27B1 expression and reestablish a level of 1,25-(OH)₂D₃ adequate to support increased Ca²⁺ requirements during lactation (Fig. 11B). Overall, the data suggest that fructose might lead to dysregulation of the PTH-1,25-(OH)₂D₃ regulatory loop.

Figure 11.

Schematic diagram of disruptions in Ca²⁺ homeostasis caused by fructose. A) In dams fed starch (or glucose) diet, lactation stimulates increases in 25-(OH)D₃ and CYP27B1 expression, which, in turn, should lead to increases in circulating levels of 1,25-(OH)₂D₃ and boost rates of intestinal Ca²⁺ transport (thick gray arrow), thereby addressing the increased Ca²⁺ demand for lactation. This enhancement of intestinal Ca²⁺ transport minimized lactation-induced decreases in serum levels of Ca²⁺ and in BMD (thin gray arrow). B) Under fructose feeding, despite a normal increase in levels of 25-(OH)D₃, the circulating levels of 1,25-(OH)₂D₃ remained low due to decreases in CYP27B1 expression. Consequently, rates of intestinal Ca²⁺ transport stayed low, which were not sufficient (thin gray arrow) to meet the demand for Ca²⁺ during lactation. This leads to a greater reduction of serum Ca²⁺ levels and to greater bone resorption (thick gray arrow) stimulated by an increase in PTH levels. Thus, dietary fructose appears to perturb the regulation of renal vitamin D synthesis (dashed line).

In this context, since PTH and feedback inhibition by its product 1,25-(OH)₂D₃ do not seem to regulate CYP27B1, what does? Among the obvious candidates are the phosphaturic compound fibroblast growth factor 23 (FGF-23; ref. 32) and the nuclear factor-κ B (NF-κB; ref. 33), candidates whose levels should be evaluated in future work. The regulation of CYP27B1 by FGF-23 would be more likely associated with an increase of CYP24A1 (32). NF-κB is known to bind directly to the promoter region of CYP27B1 and to subsequently repress its expression (33). Interestingly, chronic fructose feeding increases hepatic and aortic NF-κB activity (34). In addition, fructose consumption induced an inflammatory response in the kidney (35) presumably via NF-κB. Thus, in our hyperphagic lactating females, a high-fructose diet may activate renal NF-κB, thereby inhibiting the expression of CYP27B1.

We (Supplemental Fig. S5) and others (35) have shown that dietary fructose induces a robust increase in renal GLUT5 expression. Because postprandial fructose tends to accumulate in the kidney (36), this increase in serum fructose levels associated with a marked increase in GLUT5 in the kidney raises a question about the potential direct role of GLUT5 in mediating the effect of fructose on renal 1,25-(OH)₂D₃ synthesis.

Sweeteners and bone resorption during lactation

Lactation is already associated with a decrease in BMC, BMD, and bone mechanical strength, as the skeleton acts as a ready reservoir from which minerals can be mobilized rapidly. Overconsumption of rapidly metabolized sugars may exacerbate the effect of lactation on bone health. Dietary sucrose has been shown to reduce the mechanical strength and BMC of rat bones (37, 38), and excessive consumption of soft drinks laden with sugars may decrease BMD in humans (39, 40). However, the mechanisms underlying the detrimental effect of sugars on bone remain completely unexplored. Our work suggests that this detrimental effect may, in part, be attributed to bone resorption to limit perturbations in blood Ca²⁺ levels when, under conditions of physiological and nutritional stress, a high fructose intake prevents adaptive increases in intestinal absorption and renal reabsorption of Ca²⁺. Indeed, in fructose-fed dams, the decrease in serum Ca²⁺ levels did dramatically increase PTH concentrations that could activate bone resorption (Fig. 11). The failure of reduced levels of 1,25-(OH)₂D₃ and of increased levels of PTH to stimulate 1,25-(OH)₂D₃ synthesis meant that intestinal Ca²⁺ transport and renal Ca²⁺ reabsorption in lactating rats could not be stimulated to rates sufficient to maintain normal Ca²⁺ homeostasis. This fructose-induced inhibition of lactation-induced increases in calcitriol synthesis would then lead to “transient” low serum levels of Ca²⁺ causing increases in blood PTH that could eventually lead to bone resorption (Fig. 11).

In the starch-fed rats, we observed a moderate, lactation-induced decrease in BMD as well as in serum Ca²⁺ levels. In those fed fructose, the lactation-induced decrease in serum Ca²⁺ levels was similar to those fed starch, but the decrease in BMD was much greater. This difference between starch- and fructose-fed rats may be due to the fact that, whereas Ca²⁺ transport increased in starch-fed rats, it did not in fructose-fed, and may have forced the latter to obtain Ca²⁺ from bone, thereby reducing BMD. Interestingly, dietary glucose is also associated, like dietary fructose, with a decrease in bone quality but not in serum Ca²⁺. Although in this study, glucose feeding was not associated with any perturbation in mineral transport nor in any change in PTH level, there may be other factors involved that affect BMD. Hyperglycemia, type 2 diabetes, and even a prediabetic state are associated with weaker bones (41). In our study, the rats fed glucose during lactation had normal fasting serum levels of glucose but because of the high levels of glucose in the diet, have likely undergone transient daily postprandial increases in serum levels of insulin that have been recently shown to promote bone resorption (42).

Perspectives

There has been a dramatic surge in the United States in per capita consumption of high-fructose corn syrup (HFCS), from 0.23 kg/person-yr in 1970 to almost 30 kg in 1998. When consumption of HFCS is pooled with that of other fructose sources (e.g., sucrose), the per capita amount of fructose consumed is 50–60 g/d (26). This dramatic increase in dietary fructose consumption is tightly correlated with increased incidence in type II diabetes, hypertension, and obesity (43). Here, we suggest that excessive fructose consumption may affect Ca²⁺ transport, 1,25-(OH)₂D₃ levels, and bone quality. Since fructose caused a deficiency in 1,25-(OH)₂D₃ without perturbing 25-(OH)D₃ levels (the major determinant of vitamin D status in humans; ref. 9), it may be essential to distinguish in humans the effect of this sugar on levels of both metabolites. Future work should evaluate whether treatment of fructose-fed postweaning rats with 1,25-(OH)₂D₃ restores normal TRPV5/6 and CaBP9/28k expression in kidney and intestine.

The AIN recommends 73% (AIN 93M diet) or 63% (AIN 93G) carbohydrate levels in rodent diets. Since this is among the initial studies evaluating the effects of dietary fructose on Ca²⁺ homeostasis, we used animal models with an unusual demand for Ca²⁺, and a diet containing the fructose equivalent of the recommended carbohydrate levels in lactating rats—one certainly high enough to ensure a fructose effect, if any. It is possible that in humans with lower carbohydrate requirements, a lower fructose level may be sufficient to trigger similar symptoms. Obviously, this study must be evaluated employing diets with lower fructose levels mixed with other carbohydrates and utilizing other physiological situations that perturb Ca²⁺ balance, such as neonatal growth, hypovitaminosis D, or low dietary Ca²⁺.