Estrogen Controls Vitamin D₃-Mediated Resistance to Experimental Autoimmune Encephalomyelitis by Controlling Vitamin D₃ Metabolism and Receptor Expression¹

Multiple sclerosis (MS)³ is a genetically and immunologically complex neurodegenerative disease whose etiology and pathogenesis are poorly understood (1). The characteristic CNS lesions of MS show T lymphocyte and mononuclear cell infiltration, focal demyelination, and axonal injury and loss, which have been interpreted as evidence that autoimmune-mediated damage is causal in MS pathogenesis. Genetic susceptibility to MS has been unambiguously linked to the human MHC (2). However, high rates of MS discordance in monozygotic twins (3), geographical gradients in MS prevalence (4), and alteration of MS risk by migration (5) all suggest that environmental factors influence MS risk at a population level. In fact, when genetic factors are held constant, modifiable environmental factors appear to set the disease threshold and may hold the key to preventing ∼80% of MS cases (6). The strong inverse correlation between sunlight exposure and MS prevalence throughout the world indicates that sunlight exposure is probably one of these modifiable environmental risk factors (4, 5, 7, 8).

The vitamin D endocrine system probably mediates the protective effects of sunlight exposure in MS (9). UV light is required for an early step in the biosynthesis of 1,25-dihydroxyvitamin D₃ (1,25-(OH)₂D₃), a secosteroid hormone (10). The unique dependence of 1,25-(OH)₂D₃ synthesis on UV light, and the discovery of vitamin D receptors (VDR) in activated T lymphocytes (11, 12), led us to propose that sunlight’s protective effects in MS might reflect the need for 1,25-(OH)₂D₃ to control the T lymphocyte-mediated autoimmune responses that are pathogenic in MS (9, 13).

Since we first proposed this hypothesis, a diverse and compelling body of supporting evidence has emerged from human studies (14). For example, high levels of serum 25-hydroxyvitamin D₃ (25-(OH)D₃), which is biologically inactive, correlated with low MS risk, fewer relapses, and significantly less disability (15, 16, 17, 18, 19, 20, 21). Conversely, low 25-(OH)D₃ levels correlated with high MS risk, frequent relapses, and significantly higher disability. In an ongoing vitamin D₃ intervention study, investigators noted a trend toward fewer relapses and stable or improved disability in the supplemented patients compared with the unsupplemented MS patients (22). There is a 3–4 mo lag between the seasonal decreases in UV light exposure and circulating 25-(OH)D₃ levels and the increases in MS relapses (17, 19, 20). This lag suggests that seasonal fluctuations in UV light exposure could contribute to the relapsing-remitting MS disease phenotype. Importantly, low circulating levels of biologically active 1,25-(OH)₂D₃ correlated with a progressive disease course (21). Consistent with the possibility that reduced 1,25-(OH)₂D₃ synthesis might be causal in MS progression, a recent report described three patients with vitamin D-dependent rickets type I who all developed MS (23). Vitamin D-dependent rickets type I is a very rare hereditary disease caused by mutation of the Cyp27b1 gene encoding the enzyme that converts inactive 25-(OH)D₃ into 1,25-(OH)₂D₃. The probability of three coincident vitamin D-dependent rickets type I and MS cases by chance alone is ∼3.4 × 10⁻⁹, arguing strongly for a cause-effect relationship between Cyp27b1 mutation, reduced 1,25-(OH)₂D₃ synthesis, and MS disease. Finally, a clinical trial found that MS patients had 80% fewer relapses and no disability progression while they were receiving 1,25-(OH)₂D₃ treatment, compared with their relapse frequency and disability progression rates before and after treatment (24). Taken together, these human studies strongly support our hypothesis that low 25-(OH)D₃ and 1,25-(OH)₂D₃ levels increase MS disease risk, severity, and progression (9).

There is a rapidly increasing female gender bias in MS that is not understood (25, 26, 27, 28, 29, 30). This female gender bias involves sex hormones, because it becomes apparent after sexual maturity (31). It is not clear why the female-to-male sex ratio rises with increasing estrogen synthesis in sexually mature females, because estrogen is protective in both human MS and rodent experimental autoimmune encephalomyelitis (EAE) (32, 33, 34, 35, 36). In our EAE studies, we found a link between female sex hormones and vitamin D₃. High-level vitamin D₃ supplementation inhibited EAE in intact adult female mice, but not in ovariectomized (OVX) females or males (37). Thus, vitamin D₃ provided a female-specific and sex hormone-dependent protective effect in EAE. Interestingly, high serum 25-(OH)D₃ levels correlated with a reduced MS risk in women but not in men (38). If vitamin D₃ were to provide a female-specific and sex hormone-dependent protective effect in MS as in EAE, then this protective effect would become evident at sexual maturity, and after puberty, vitamin D₃ insufficiency would increase MS risk in a female-specific manner.

Here, we investigated the female-specific and sex hormone-dependent protective effect of vitamin D₃ in EAE in more detail. To test the hypothesis that 17β-estradiol (E₂) might regulate vitamin D₃ metabolism and/or VDR expression in the CNS, OVX female mice were implanted with E₂ or placebo pellets and evaluated for vitamin D₃-mediated EAE resistance, vitamin D₃ metabolism, and VDR expression in the CNS. Our data show for the first time that E₂ enables vitamin D₃ to decrease EAE risk and severity in female mice through enhancement of VDR transcription and function in the CNS. We propose a model for the functional synergy between the vitamin D and estrogen endocrine systems, and we discuss our model in the context of the relapsing-remitting phenotype, the female gender bias, and vitamin D₃-based strategies to decrease MS risk and disease severity.

The B10.PL-H2^uH2-T18^a/(73NS)/SnJ (hereafter B10.PL) and C57BL/6 (hereafter B6) mice were obtained from The Jackson Laboratory. Some B10.PL mice were bred in the pathogen-free mouse colony of the Department of Biochemistry from The Jackson Laboratory-derived breeding pairs. Mice were housed at 25°C with a 12 h light-dark cycle and 40–60% humidity. The drinking water was provided ad libidum. Before experiments, the mice were fed commercial mouse chow containing 0.33 μg/day of vitamin D₃ and 1% calcium (Lab Diet no. 5008; PMI Nutrition International). The Institutional Animal Care and Use Committee approved the experimental protocols. All animal experimentation was conducted in accordance with accepted standards of humane animal care.

The synthetic diet was formulated to contain all essential nutrients except vitamin D₃ exactly as we described (37). The vitamin D₃ (cholecalciferol; Acros Organics) was dissolved in absolute ethanol (1 mg/ml) and stored in the cold. It was added to the synthetic diet in an amount calculated to provide 0 (−D diet) or 1 μg/day (+D diet) of vitamin D₃ based on a measured daily diet consumption of 4.0 g dry weight of diet per mouse (37). Groups of mice (sex-matched; age 6–8 wk) were fed the +D or −D diet continuously beginning on the first day of each study. Fresh synthetic diet was prepared weekly, stored at 4°C, and provided to the mice three times per week.

OVX or SHAM surgery was performed 1 wk after synthetic diet feeding began. During the surgery, a controlled release pellet containing E₂ (3 mg; 0.1 mg of E₂) or a placebo was implanted dorsally s.c. using a 12-gauge trochar, as described by the manufacturer (Innovative Research of America).

EAE was induced 3 wk after synthetic diet feeding began. The myelin basic protein (MBP), isolated from guinea pig spinal cords and dissolved in 0.1 M acetic acid (4 mg/ml), was emulsified with an equal volume of CFA containing heat-killed Mycobacterium tuberculosis H37 Ra (4 mg/ml). Each anesthetized B10.PL mouse was injected s.c. with 0.2 ml of emulsion (400 μg of MBP) in three sites on the hind flanks. Each mouse was injected i.p. with 200 ng of pertussis toxin (List Biological Laboratories) in 0.1 ml of sterile PBS solution on the day of MBP immunization and again 2 days later. EAE was induced in B6 mice using the same protocol, except that a peptide from mouse myelin oligodendrocyte protein (MOG_35–55; sequence MEVGWYRSPFSRVVHLYRNGK) replaced MBP, and the emulsion injection volume was 0.1 ml (200 μg of MOG_35–55 peptide).

EAE disease severity was evaluated daily as follows: 0, normal; 0.5, partial loss of tail tone and cool tail temperature; 1, limp tail; 1.5, partial paralysis of one hind limb; 2, partial paralysis of both hind limbs; 2.5 partial paralysis of one hind limb and complete paralysis of one hind limb; 3, paralysis of both hind limbs; 4, hind and fore limb paralysis; 5, moribund or dead. This disease severity scale, originally published in 1974 (39), uses lower disease severity numbers for equivalent degrees of disability than do present disease severity scales (40). Accordingly, caution must be used in making direct comparisons between EAE disability curves presented herein and elsewhere, without referring to the disease severity scale used by the investigators.

Blood samples were obtained from representative mice before synthetic diet feeding, 1 day before immunization, and at the conclusion of the study. The blood was clotted, centrifuged, and the decanted serum was frozen at −70°C before analysis. When the mice were euthanized, perfusion was performed and the spinal cords were removed as described (41). The spinal cords were flash frozen with liquid nitrogen and stored at −70°C before RNA or 1,25-(OH)₂D₃ extraction.

The serum E₂ was assayed using an enzyme immunoassay kit (DSL 10-4300; Diagnostic Systems Laboratories). Enzyme immunoassay kits were also used to quantify the serum 25-(OH)D₃ (Alpco Diagnostics) and 1,25-(OH)₂D₃ (Immunodiagnostic Systems). The spinal cords were extracted and the extracts assayed in duplicate for 1,25-(OH)₂D₃ as we described previously (37). A spike-recovery control was performed with each spinal cord extraction, the percentage recovery was calculated, and a recovery correction factor was applied to the experimental data.

The GAPDH, IFN-γ, VDR, Cyp27b1, and Cyp24a1 transcript abundance was measured in the spinal cord by real-time PCR as described (37). The amplification and detection were accomplished with a 7500 Fast real-time PCR system (Applied Biosystems) and SYBR Advantage quantitative PCR premix (Clontech). Published primers for the GAPDH, IFN-γ, and Cyp27b1 were used (42, 43). The VDR primers were forward 5′-GCA ACA GCA CAT TAT CGC CAT-3′ and reverse 5′-TAC GTC TGC ACG AAT TGG AGG-3′. The CYP24A1 primers were forward 5′-ACC CCC AAG GTC CGT GAC ATC-3′ and reverse 5′-CCA GTT GGT GGG TCC AGG TAA GG-3′. Primers were purchased from Integrated DNA Technologies or Invitrogen. To generate a standard curve, cDNA representing each specific amplicon was gel purified and quantified by absorbance at 260 nm. Each real-time PCR included reactions with serially diluted standard cDNA. The standard cDNA copy number, calculated from the absorbance and the dilution, was plotted vs the threshold cycle (C_t). The transcript copy number in each unknown sample was determined from C_t by reference to the appropriate standard curve. The data were calculated as transcript abundance relative to GAPDH as the internal standard.

Individual mice were analyzed and the means and SD or SEM were calculated for each group of mice. The group sizes are given in the table and figure legends. The significance of differences between the group means was determined using the Mann-Whitney U test (n ≤ 16), Student’s t test (n > 16), or χ² test (binomial data); p < 0.05 was considered significant.

The hypothesis that E₂ controls vitamin D₃-mediated EAE resistance through regulation of vitamin D₃ metabolism and/or VDR expression in the CNS predicts that E₂ repletion would restore vitamin D₃-mediated EAE resistance in OVX females. Therefore, we first determined an E₂ repletion level that would not inhibit EAE in OVX mice independently of supplementary vitamin D₃. Implanting 0.36 mg of E₂ timed-release pellets into OVX B10.PL females inhibited EAE independently of vitamin D₃ (44), so we repeated this study with 0, 0.1, or 0.36 mg of E₂ implants. Two weeks after OVX and pellet implantation, the mice were immunized with MBP and monitored daily for clinical EAE signs. Uterine tissue was collected and weighed at the end of the study as an indicator of total E₂ exposure (Fig. 1,A). The SHAM control mice had uterine weights of 0.11 ± 0.05 mg. The OVX mice with placebo pellets had lower uterine weights than did SHAM controls (Fig. 1,A), confirming E₂ depletion. The OVX mice with 0.1 mg of E₂ pellets had uterine weights equal to the SHAM controls, indicating normal E₂ levels. Finally, the OVX mice with 0.36 mg of E₂ pellets had higher uterine weights than did SHAM controls, indicating high-level E₂ repletion (44). Because high-level E₂ repletion decreased the EAE incidence, peak severity, and cumulative disability independently of supplementary vitamin D₃ (Fig. 1, B and C), but low-level E₂ repletion alone had no effect, we used low-level E₂ repletion in all subsequent experiments.

The next experiments investigated possible interactions between vitamin D₃ and low-level E₂ repletion in OVX mice. Female B10.PL mice were fed diets with (1 μg/day; +D) or without (0 μg/day; −D) vitamin D₃. This +D diet provided 3-fold the amount of vitamin D₃ in standard laboratory mouse chow. After 1 wk of −D or +D diet feeding, OVX or SHAM surgery was performed and 0 or 0.1 mg of E₂ pellets was implanted. MBP immunizations were performed 2 wk after surgery. The serum E₂ levels were quantified 7, 24, and 48 days after surgery, and uterine weights were measured 48 days after surgery (Fig. 2,A). In the OVX/placebo mice, the E₂ level declined significantly relative to the SHAM controls 24 days after surgery (p < 0.005), but rose by 48 days after surgery due to E₂ synthesis by the adrenal glands (45). In the OVX/E₂ mice, the E₂ level was elevated 7 days after surgery (p < 0.005), but matched the SHAM controls 24 and 48 days after surgery. Unexpectedly, when the E₂ data were analyzed by dietary vitamin D₃ group, there was significantly more E₂ in the +D OVX/placebo serum samples than in the −D OVX/placebo samples (p < 0.03) (Fig. 2,B). This trend was also evident in the SHAM and OVX/E₂ mice, but it did not reach significance due to the variability in serum E₂ measurements. The uterine weights confirmed that the OVX mice had lower total E₂ exposure than did the SHAM controls, whereas the 0.1 mg of pellet provided a total E₂ exposure equal to the SHAM controls (Fig. 2 C).

The serum 25-(OH)D₃ levels were quantified at the end of the study. The groups ingesting the −D diet had significantly less serum 25-(OH)D₃ than did the groups ingesting the +D diet (p < 0.01; Table I). The groups did not differ significantly with respect to terminal weights or serum calcium levels. These data are consistent with our published data (37), confirming that this level of vitamin D₃ supplementation does not cause hypercalcemia.

Having confirmed that the serum E₂ and 25-(OH)D₃ levels differed between the respective pellet and dietary groups as intended, we next analyzed the EAE data. Consistent with our published report, the vitamin D₃ significantly reduced the incidence 28% (p < 0.05), delayed disease onset by 10 days (p < 0.05), decreased the peak severity ∼50% (p < 0.05), and diminished the cumulative disease severity ∼70% (p < 0.05) in the SHAM females, but provided no benefits whatsoever in the OVX/placebo mice (Fig. 3 and Table II) (37). Extending our published results, we found that diestrus level E₂ repletion restored all of the protective effects of vitamin D₃ with respect to EAE incidence, onset, peak severity, and cumulative disease, although it had no independent effect on EAE disease.

In B10.PL mice immunized with MBP, EAE follows a relapsing-remitting disease course (46), whereas in B6 mice immunized with MOG_35–55 peptide, EAE follows a chronic progressive disease course (47). To learn whether E₂ and vitamin D₃ would cooperate to inhibit chronic progressive EAE as they did to inhibit relapsing-remitting EAE, we repeated the OVX experiments in B6 mice. Female B6 mice were fed +D or −D diets, OVX or SHAM surgery was performed, and 0 or 0.1 mg of E2 pellets was implanted in the OVX mice. The MOG_35–55 immunizations were performed 2 wk after surgery. In the SHAM B6 mice, the vitamin D₃ reduced the cumulative disease severity by 43% (p < 0.05) compared with the −D SHAM group, and there were trends toward lower incidence and decreased peak severity that did not reach significance (Table II). However, the vitamin D₃ provided no benefits whatsoever in the OVX/placebo B6 mice. Comparing the OVX B6 mice with 0.1 mg of E₂ implants to the OVX B6 mice with placebo implants, we saw a delay in disease onset and a 36% reduction in the cumulative disease (p < 0.02) that were attributable to the E₂ implants independently of dietary vitamin D₃. However, comparing the +D OVX/E₂ mice to the −D OVX/E₂ mice, we found that the +D diet further reduced the cumulative disease by 37% (p < 0.001), and there was a trend toward lower peak severity. The animals ingesting the +D diet had significantly higher spinal cord 1,25-(OH)₂D₃ levels than did the animals ingesting the −D diet (p < 0.005; Table III), although the serum 1,25-(OH)₂D₃ levels were not different between the groups (data not shown). These data confirm our report that 1,25-(OH)₂D₃ synthesis occurred in the inflamed spinal cord, independently of 1,25-(OH)₂D₃ synthesis in the kidney (37). In summary, E₂ repletion provided a benefit that was independent of dietary vitamin D₃ in B6 mice, and additionally, E₂ repletion restored the ability of dietary vitamin D₃ to inhibit EAE disease.

In our previous study, vitamin D₃ failed to inhibit MBP-induced EAE in the intact and castrated male B10.PL mice, indicating that male sex hormones did not impede vitamin D₃-mediated EAE resistance (37). Here we examined whether E₂ implants would enable vitamin D₃-mediated EAE resistance in intact male mice. Male B6 mice were fed diets with 0, 1, or 2 μg/day vitamin D₃, immunized with MOG_35–55 peptide to model the progressive form of MS often seen in men, and evaluated daily for EAE signs. The vitamin D₃-fed groups were not significantly different from the 0 μg/day vitamin D₃ controls for any EAE parameter (Table IV), consistent with our published data for male B10.PL mice (37). Implanting E₂ pellets in the males significantly increased their serum E₂ levels and inhibited EAE induction, as others reported previously (48). However, E₂ supplementation did not enable vitamin D₃ to further inhibit any EAE parameter in the males.

It is not clear why diestrus-level E₂ repletion restored the protective effects of vitamin D₃ in OVX females immunized to induce EAE. Two possibilities are E₂-mediated enhancement of in situ 1,25-(OH₂)D₃ accumulation through control of the Cyp27b1 or Cyp24a1 genes encoding the synthetic and inactivating enzymes, respectively, or E₂-mediated enhancement of the VDR in the CNS. The 1,25-(OH₂)D₃ synthesis and inactivation rates are proportional to the Cyp27b1 and Cyp24a1 transcripts, respectively (49). Therefore, to evaluate the propensity for hormone accumulation, we analyzed the Cyp27b1 and Cyp24a1 transcripts. As before, female B10.PL mice were fed +D or −D diets, OVX or SHAM surgery and placebo or E₂ pellet implantation was performed, and the mice were immunized with MBP. Ten days postimmunization (disease onset), spinal cord RNA was isolated and reverse transcribed, and transcripts were quantified by quantitative real-time PCR. Optimal PCR primers and reaction conditions were selected to amplify the transcripts with >90% efficiency (50). We recorded the threshold PCR cycles (C_t) at which the Cyp27b1 and Cyp24a1 amplicons were detected. The C_t is inversely proportional to transcript abundance, so the relative abundance of the two transcripts was expressed as ΔC_t (C_t for Cyp24a1 minus the C_t for Cyp27b1) (Fig. 4 A). In nonimmunized mice, the C_t for Cyp24a1 was ∼30 and the C_t for Cyp27b1 was ∼28–29, so ΔC_t was ∼1–2, regardless of dietary vitamin D₃. Assuming equal amplification efficiencies and fluorescence intensities, this ΔC_t indicates that Cyp27b1 transcripts were slightly more abundant than Cyp24a1 transcripts. From standard curves we estimated that the Cyp27b1-to-Cyp24a1 transcript ratio was ∼4:1 in naive mice.

In MBP-immunized SHAM and OVX mice, the C_t for Cyp24a1 shifted to ∼32–33, while the C_t for Cyp27b1 remained ∼28–29, so ΔC_t was ∼3–4, regardless of dietary vitamin D₃ (Fig. 4,A). This increase in the C_t for Cyp24a1 from ∼30 in naive mice to ∼32–33 in MBP-immunized mice indicates that Cyp24a1 gene expression decreased with MBP immunization, increasing the Cyp27b1-to-Cyp24a1 transcript ratio to ∼8:1. These data are consistent with our previous report showing decreased Cyp24a1 gene expression during CNS inflammation in female mice (37). There was a trend toward further reduction in Cyp24a1 gene expression in OVX mice with E₂ implants that did not reach significance. Thus, the Cyp27b1-to-Cyp24a1 transcript ratio might be >8:1 in these mice due to E₂-mediated Cyp24a1 suppression. Decreased Cyp24a1 gene expression during CNS inflammation would favor in situ 1,25-(OH₂)D₃ accumulation, consistent with the results presented in Table III.

We next investigated possible E₂-mediated enhancement of VDR gene expression in the CNS. Spinal cord samples collected 10 days postimmunization (disease onset) were analyzed for VDR and GAPDH transcripts. In samples from unimmunized controls, the C_t for GAPDH and VDR were ∼20 and ∼29, respectively, regardless of dietary vitamin D₃. So, ΔC_t (C_t for VDR minus the C_t for GAPDH) was ∼9, indicating very low VDR gene expression (10–15 VDR copies/1000 GAPDH copies) (Fig. 4, B and C). In samples from the MBP-immunized SHAM mice ingesting the +D diet, the VDR C_t decreased to ∼24, indicating a ∼32-fold increase in transcript abundance. This increase in VDR transcript abundance was also present in the MBP-immunized OVX mice with E₂ implants ingesting the +D diet, but it was absent in the −D SHAM group, the −D and +D OVX groups, and in the −D OVX group with E₂ implants. Thus, high-level VDR transcription in the spinal cord required an inflammatory stimulus, dietary vitamin D₃, and a source of E₂. Graphing the spinal cord VDR ΔC_t at the time of EAE disease onset vs the cumulative EAE disease at the end of the 28-day study revealed a linear correlation between these parameters with an R² equal to 0.78 (Fig. 4 D). This correlation was observed in two independent studies performed one year apart. We conclude that diestrus-level E₂ repletion restored the protective effects of vitamin D₃ in OVX females immunized to induce EAE, through decreased Cyp24a1 gene expression and increased VDR gene expression. These two actions enhanced in situ 1,25-(OH₂)D₃ accumulation and VDR function in the CNS, contributing to the resolution of inflammation and reduction of EAE disease.

We have presented evidence that female mice with disrupted ovarian hormone production lost vitamin D₃-mediated resistance to EAE and, conversely, that low-level E₂ repletion restored vitamin D₃-mediated resistance to EAE in OVX female mice. The low-level E₂ repletion did not inhibit EAE disease independently of dietary vitamin D₃. The vitamin D₃ in the +D diet used here was 3-fold higher than laboratory chow, and it yielded serum 25-(OH)D₃ levels that were ∼1.6-fold higher than those in chow-fed mice. The vitamin D₃-mediated EAE resistance was manifested as a lower incidence, later onset, decreased peak severity, and diminished cumulative disease severity in the MBP/B10.PL model, and as a diminished cumulative disease severity in the MOG/B6 model. None of these vitamin D₃-mediated benefits were evident in male mice, with or without E₂ supplementation. Our data showed that high-level VDR gene expression and function in the spinal cord required an inflammatory stimulus, E₂, and sufficient dietary vitamin D₃ to support in situ 1,25-(OH)₂D₃ synthesis. To our knowledge, this is the first evidence of synergy between a sex hormone and vitamin D₃ in the control of an autoimmune disease, and the first evidence that E₂ is essential for VDR gene expression and function in the inflamed CNS.

One mechanism of synergy between vitamin D₃ and E₂ appears to be vitamin D₃ enhancement of E₂ biosynthesis. We found that vitamin D₃-supplemented OVX mice had 2-fold more serum E₂ than did unsupplemented mice. Our data are consistent with published data showing that VDR-targeted female mice had uterine hypoplasia and impaired folliculogenesis, because a lack of estrogen synthase in the ovary decreased E₂ biosynthesis (51). E₂ supplementation reversed these defects. Moreover, 1,25-(OH)₂D₃ enhanced the transcription of the Cyp19 gene encoding estrogen synthase in glial cells (52) and placental trophoblasts (53). Estrogen synthase (also termed aromatase), the rate-limiting enzyme in the formation of estrone and estradiol from the C19 androgens androstenedione and testosterone, is expressed in the gonads, adrenals, brain, and adipose tissue (45). Thus, 1,25-(OH)₂D₃ and VDR-dependent enhancement of E₂ biosynthesis could be one mechanism allowing the estrogen and vitamin D₃ endocrine systems to function synergistically in women.

A second mechanism of synergy between the estrogen and vitamin D₃ endocrine systems appears to be E₂ suppression of Cyp24a1 gene expression, leading to 1,25-(OH)₂D₃ accumulation, and enhancement of VDR gene expression in females. Our new data show E₂-dependent transcriptional activation and function of the VDR gene in the spinal cord during an inflammation. This is true in other tissues as well. In osteoblasts, VDR gene expression decreased with E₂ deprivation and increased with E₂ supplementation (54, 55). Moreover, E₂-mediated up-regulation of the VDR gene was also reported in the duodenal mucosa (56), where reduced VDR gene methylation correlated with transcriptional activation, elevated VDR protein, increased responsiveness to endogenous 1,25-(OH)₂D₃, and greater resistance to colonic carcinogenesis (57, 58). Liver cells (59) and breast cells (60, 61) also showed E₂-mediated transcriptional activation of the VDR gene. In breast cells, an estrogen receptor (ER)-mediated mechanism controlled the VDR gene expression (60). Estrogen-responsive promoter elements were identified immediately upstream of exon 1c in the human VDR gene (61). Collectively, these data suggest that a second general mechanism of synergy between the estrogen and vitamin D₃ endocrine systems is E₂-mediated transcriptional activation of the VDR gene. Linking the first and second mechanisms together yields an amplification loop: 1,25-(OH)₂D₃ enhances E₂ biosynthesis by VDR-mediated up-regulation of estrogen synthase, and E₂ enhances 1,25-(OH)₂D₃ synthesis and function by ER-mediated down-regulation of Cyp24a1 and up-regulation of VDR. These two mechanisms and possibly others allow the vitamin D₃ and estrogen endocrine systems to function synergistically in women. Large increases in maternal serum estriol and 1,25-(OH)₂D₃ levels occur during pregnancy due to placental production of these hormones (62). Thus, the synergy between the vitamin D₃ and E₂ endocrine systems may have evolved to support successful reproduction through maternal immune tolerance for fetal Ags.

The lack of interaction between E₂ and vitamin D₃ in males was puzzling, since males express ER, and E₂ is an effective regulator of EAE in males (63). We suspect that that there may be differential regulation of the Cyp24a1 gene between males and females. We previously reported that males expressed higher levels of Cyp24a1 mRNA in the CNS than do females, and they did not show complete repression of Cyp24a1 gene expression during inflammation, as did females (37). This resulted in a failure to accumulate 1,25-(OH)₂D₃, which could undermine benefits derived from E₂-mediated enhancement of VDR expression.

The synergy between 1,25-(OH)₂D₃ and E₂ as inhibitors of EAE is reflected in the similar mechanisms by which these two hormones reportedly inhibit the disease (34, 64). Both hormones inhibit the accumulation of inflammatory cells in the CNS by decreasing chemokines, chemokine receptors, and cell migration (64, 65). The protective functions of the two hormones also depend on regulatory lymphocytes and antiinflammatory cytokines (66, 67, 68, 69). Finally, the two hormones have documented neuroprotective functions that involve regulation of neurotrophins and neurotrophin receptors (34, 70, 71). To explain how E₂ enabled vitamin D₃-mediated inhibition of EAE, we suggest that E₂ and ER-α may enhance VDR expression in CD4⁺ T cells through estrogen-responsive promoter elements in the VDR gene (61).

An important question remaining to be answered is whether 1,25-(OH)₂D₃ exerts direct, E₂-independent protective effects in EAE, or only indirect effects attributable to increased E₂ synthesis. Similarly unanswered is the question whether E₂ exerts direct, VDR-independent protective effects in EAE, or only indirect effects attributable to increased VDR synthesis. There are reported differences in the mechanisms of action of these two hormones that suggest independent action. For example, 1,25-(OH)₂D₃ reversed the signs of severe acute EAE (41), whereas E₂ did not (48). Also, E₂ inhibited EAE in mice with a targeted disruption of the IL-10 gene (72), whereas 1,25-(OH)₂D₃ did not (69). ER-α expression was necessary for E₂ to inhibit EAE (73, 74). ER-α is expressed in neurons, glia, oligodendrocytes, lymphocytes, macrophages, and dendritic cells (34), but its expression in T lymphocytes was not necessary for E₂-mediated inhibition of EAE (75). In contrast, our unpublished chimera data show that VDR expression is necessary in T lymphocytes for 1,25-(OH)₂D₃-mediated inhibition of EAE (C. G. Mayne, J. A. Spanier, and C. E. Hayes, unpublished data). These hormone receptor expression studies suggest that 1,25-(OH)₂D₃ may primarily target T lymphocytes, whereas E₂ may primarily target non-T cells. Thus, although the two hormones E₂ and 1,25-(OH)₂D₃ show many similar mechanisms of action suggesting synergy between them, they also show significant differences in mechanism suggesting each hormone can act independently. Additional experiments are underway to define the synergistic and independent functions of E₂ and 1,25-(OH)₂D₃ more precisely.

The functional synergy between the vitamin D₃ and E₂ endocrine systems may be a driver for the relapsing-remitting MS disease phenotype. Seasonal fluctuations in UV light exposure and thus circulating 25-(OH)D₃ levels show the same periodicity as seasonal fluctuations in MS attacks (15, 16, 17, 18, 19, 20, 21). The fact that seasonal changes in UV light/vitamin D₃ preceded changes in MS attack rates by ∼3–4 mo supports a causal link between UV light/vitamin D₃ fluctuations and changes in the MS attack rates. Data showing beneficial effects of 25-(OH)D₃ only in females is also consistent with this hypothesis (37, 38). Seasonal increases in UV light exposure and serum 25-(OH)D₃ and spinal cord 1,25-(OH)₂D₃ levels could trigger the protective amplification loop described above. The increased E₂ biosynthesis and E₂-mediated enhancement of VDR expression and function would enable 1,25-(OH)₂D₃ and E₂ to activate antiinflammatory mechanisms that drive MS disease into remission. Conversely, seasonal declines in UV light exposure and serum 25-(OH)D₃ and spinal cord 1,25-(OH)₂D₃ levels could interfere with the amplification loop and allow inflammatory mechanisms to ignite and precipitate MS attacks.

The functional synergy between the vitamin D₃ and E₂ endocrine systems may also be a driver of the increasing female bias in MS. An equal number of women and men were afflicted with MS until about 1950, when prevalence surveys first revealed a 1.4:1 female bias (76). Recent estimates put the sex ratio between ∼3:1 (28) and ∼5:1 (30), depending on the decade of birth. An increase in the female bias of MS has been observed in the United States (26, 77), Canada (28, 30), Australia (78), the United Kingdom (29, 79), Norway (27, 80), and Sardinia (81), due to an increased incidence among women, rather than to a decreased incidence among men. One explanation for the female bias could be X-linked risk factors, but extensive X chromosome linkage studies have not revealed MS susceptibility loci (82), and the rapidity of the female-to-male sex ratio increase argues against a genetic origin (28). Another explanation for the female bias in MS invokes sex hormone-determined differences in immune responsiveness (25). This explanation is consistent with data showing that the female bias becomes apparent only after sexual maturity (31), but it is not consistent with data showing a rapid female-specific increase in MS risk (26, 27, 28, 29, 30, 77, 78, 79, 80, 81).

The proposed functional synergy between the vitamin D₃ and E₂ endocrine systems could explain the rapidly increasing female-to-male sex ratio in MS. The female bias in MS would become evident at puberty, because with increasing E₂ biosynthesis would come increasing responsiveness to sunlight and vitamin D₃ via the amplification loop, with E₂ enhancing 1,25-(OH)₂D₃ synthesis and function by ER-mediated down-regulation of Cyp24a1 and up-regulation of VDR, and 1,25-(OH)₂D₃ enhancing E₂ biosynthesis by VDR-mediated up-regulation of estrogen synthase. Because high UV exposure and vitamin D₃ supplies apparently reduce the MS risk in women but not in men (37, 38), it follows that diminishing sunlight exposure and vitamin D₃ supplies due to lifestyle changes could drive a disproportional increase in MS among women, establishing a female gender bias. Moreover, the lifestyle changes that have decreased overall UV light exposure and serum 25-(OH)D₃ levels in women over the last half century (83) could be driving the rapidly increasing female gender bias. Examples of significant lifestyle changes are increased numbers of women in the workforce, decreased outdoor activity, increased sun avoidance, and use of sunscreens (which inhibit vitamin D₃ biosynthesis). We suggest that the vitamin D₃ and E₂ functional synergy hypothesis, rather than the X-chromosome or sex hormone hypothesis, is most consistent with recent data on the rapid rise in MS among women.

We have proposed that there is functional synergy between the vitamin D₃ and E₂ endocrine systems, and that this synergy is causally related to the relapsing-remitting MS disease phenotype and the increasing female bias in MS prevalence. Our data and the mechanisms we have proposed to explain the data have very significant implications for MS. Specifically, inadequate sunlight exposure and low vitamin D₃ supplies may undermine the beneficial effects of estrogens, to the extent that these activities depend on enhancement of VDR expression and function. Moreover, inadequate E₂ biosynthesis due to ovarian failure or menopause may undermine the beneficial effects of vitamin D₃, to the extent that these activities depend on enhancement of E₂ synthesis. Combined vitamin D₃ and E₂ deprivation, such as one might expect for older female MS patients with limited mobility, could have a devastating synergistic effect, triggering the evolution of relapsing-remitting MS into a chronic-progressive disease course with rapid accumulation of disability.

MS is a devastating neurodegenerative disease that imposes heavy burdens on patients, on families, and on health care systems throughout the world. At an estimated lifetime cost in excess of $2.2 million per MS case, the implications of the sustained increases in female cases to the world’s strained health care systems are staggering (28). In this context, it is encouraging that modifiable environmental factors appear to set the disease threshold and may hold the key to preventing the vast majority of MS cases (6). Sunlight exposure and vitamin D₃ supplies appear to be those modifiable environmental risk factors (9, 13). If health care providers were to monitor serum 25-(OH)D₃ levels, especially in girls and women who are genetically related to an individual with MS, and prescribe enough sunlight exposure and/or vitamin D₃ supplementation to maintain >100 nmol/L of serum 25-(OH)D₃ throughout the year, an estimated 90% of MS cases might be prevented (15). For men and women already afflicted with MS, intermittent 1,25-(OH)₂D₃ pulse dose therapy (F. E. Nashold, R. A. Derks, and C. E. Hayes, manuscript in preparation) in the context of sufficient natural E₂ in young women or E₂ replacement therapy in postmenopausal women might activate antiiinflammatory mechanisms that drive MS disease into remission and significantly decrease the cumulative disability. The overwhelming body of evidence suggests that these intervention strategies could dramatically reduce the impact of MS on patients, on families, and on our health care systems.

We thank Dr. Halina Offner for discussion on the research and critically reading the manuscript. We acknowledge Anna R. Raines for technical assistance, Dr. Peter Crump for statistical advice, and the Biochemistry Animal Care staff for careful animal husbandry.

The authors have no financial conflicts of interest.