1,25-Dihydroxyvitamin D₃ Is a Positive Regulator for the Two Anti-Encephalitogenic Cytokines TGF-β1 and IL-4¹

Multiple sclerosis (MS)⁴ is a chronic autoimmune disease of unknown etiology affecting the central nervous system (CNS). Epidemiologic studies show that MS is more prevalent in temperate latitudes than at equatorial latitudes (1, 2, 3). The environmental factor that explains the link between geography and MS, however, has been elusive. In 1974, Goldberg first suggested that the amount of vitamin D available in the environment either from sunshine exposure or from the diet might affect the prevalence of MS (4). Ultraviolet irradiation of the skin is the major and ultimate source of vitamin D. Areas with low supplies of vitamin D correlate with geographic regions associated with a high risk for MS. Conversely, the prevalence of MS is lower where vitamin D is abundant, as in sunny climates, high altitudes, and areas with diets rich in fish oil (4, 5, 6).

We have recently provided strong experimental evidence that 1,25-dihydroxyvitamin D₃ (1,25-(OH)₂D₃), the functional metabolite of vitamin D₃, can greatly reduce or eliminate the incidence and severity of experimental autoimmune encephalomyelitis (EAE), an animal model of MS (7). EAE is mediated by CD4⁺ T cells, and more specifically Th1 cells that recognize proteins in the CNS (8). Th1 cells secreting IFN-γ and TNF-α are associated with EAE in mice (9, 10) and with MS in humans (11). Neutralization of TNF in vivo decreases the severity of EAE in mice (12). Interestingly, neutralization of IFN-γ in vivo and the use of IFN-γ knockout mice suggest a beneficial role for IFN-γ in EAE (13, 14). A number of T cell products and the development of Th2 cells specific for CNS proteins are associated with the suppression of EAE in mice (15). In particular, TGF-β1 treatment of mice exhibiting signs of EAE has been shown to be beneficial (16). Conversely, neutralization of TGF-β1 in vivo increases the severity of EAE (17). Finally, based on work with EAE, TNF-α inhibitors and exogenous TGF-β1 have been developed as new immunotherapies for MS (18).

There are at least two possible mechanisms whereby vitamin D might prevent or decrease the severity of EAE. The functional metabolite or hormonal form of vitamin D, 1,25-(OH)₂D₃, has been shown to inhibit the in vitro proliferation of T cells and to decrease the in vitro production of two Th1 cytokines, IFN-γ and TNF-α (19, 20, 21). The first possibility is that 1,25-(OH)₂D₃ is negatively regulating encephalitogenic T cells and the cytokines they produce. A second possibility is that 1,25-(OH)₂D₃ is positively regulating anti-encephalitogenic cells and the cytokines they produce. These potential mechanisms are not mutually exclusive.

Experiments are described below that were designed to probe the immunobiologic mechanism(s) underlying the treatment and prevention of EAE by 1,25-(OH)₂D₃. We measured four cytokines pivotal for the outcome of EAE: IFN-γ, TNF-α, IL-4, and TGF-β1. Our results support the hypothesis that 1,25-(OH)₂D₃ acts primarily by increasing the anti-inflammatory cytokines IL-4 and TGF-β1.

The B10.PL mice were produced in our colony using breeding pairs obtained from The Jackson Laboratory (Bar Harbor, ME). During this time, the mice were maintained on Purina Chow that provides abundant amounts of vitamin D. At 6 to 7 wk of age, male and female mice were fed experimental purified diets described previously (7) that contained no vitamin D. Although the diets do not contain vitamin D, the mice were housed in rooms with fluorescent light that catalyzes vitamin D production in skin. Therefore, the mice were not vitamin D deficient. The amount of calcium in the diet was uniformly 0.25%. Mice were maintained for 1 wk on this experimental diet before use in experiments. Experiments used no less than 5 and up to 10 mice in each treatment group. The complete experiment with all time points were repeated at least two times but as many as seven times.

EAE was induced exactly as described (7). Briefly, mice were immunized s.c. with 400 μg of guinea pig myelin basic protein (MBP) in CFA. In addition, on the day of immunization and 2 days later the mice received an i.p. injection with 200 ng of pertussis toxin (List Biological Laboratories, Campbell, CA) in saline.

EAE severity was scored daily by two blinded individuals using the following scoring system: 0 = no symptoms; 1 = limp tail; 2 = paraparesis; 3 = hind limb paralysis; 4 = fore and hind limb paralysis; and 5 = moribund. Intermediate scores of ±0.5 were used for mice with symptoms in between any two scores. Twenty-five to thirty percent of control B10.PL mice developed severe EAE and were sacrificed. These animals were scored as 5 on the day they were sacrificed. The future EAE scores do not include scores from sacrificed mice.

For the first experimental design, the 1,25-(OH)₂D₃ treatment was started the day before EAE induction. Females were fed 1,25-(OH)₂D₃ at 50 ng/day and the males were fed 1,25-(OH)₂D₃ at 200 ng/day. These doses of 1,25-(OH)₂D₃ were found previously to be the minimum needed to prevent EAE completely (Ref. 7, and M. T. Cantorna, J. Humpal-Winter, and H. F. DeLuca, unpublished observations). A second experimental design was to start 1,25-(OH)₂D₃ treatment when the tails of immunized mice first became limp (first signs of EAE, about day 7). One group of mice was treated with an i.p. injection of 1,25-(OH)₂D₃ (300 ng) in 0.1 ml of propylene glycol, and the other group of mice was injected with propylene glycol alone (controls). Both male and female mice injected with 1,25-(OH)₂D₃ were maintained on the experimental diet to which 1,25-(OH)₂D₃ was added to provide 50 ng per mouse daily. Control animals were provided the experimental diet throughout. A third experimental design was to allow mice to develop severe EAE (severity score of 2.5). Mice with EAE severity scores of 2.5 were randomly placed into two groups. One group of mice was treated with an i.p. injection of 1,25-(OH)₂D₃ (300 ng) in 0.1 ml of propylene glycol, and the other group of mice was injected with propylene glycol alone (controls). Some of the 1,25-(OH)₂D₃-injected and control-injected mice were used 24 h later for cytokine transcript analyses. The others were maintained on the experiment diet (controls) or on the same diet supplemented with 1,25-(OH)₂D₃ to provide 50 ng per mouse daily.

Axillary, brachial, and inguinal lymph nodes (LN) were collected from control- and 1,25-(OH)₂D₃-treated mice at days 3, 7, 10, 14, and 21 postimmunization. These LNs were chosen because they were enlarged and drained the site of immunization. Collected LNs were disrupted manually using a 23-gauge needle and a pair of forceps. Total cell numbers in the LNs were determined by counting the number of lymphocytes from control- and 1,25-(OH)₂D₃-treated mice and dividing by the number of mice in the group. Flow cytometry of fluorescent-labeled cell populations (Thy-1, class II, CD4, and CD8) was done on LN cells from control- and 1,25-(OH)₂D₃-treated mice. For cytokine PCR analysis, LN cells were saved for total cellular RNA isolation. For cytokine protein determination, equal numbers of cells from control- and 1,25-(OH)₂D₃-treated mice were cultured in HL-1 serum-free medium (BioWhittaker, Walkerville, MD) supplemented with 2-ME (50 μM), glutamine (2 mM), penicillin (10 U/ml), and streptomycin (10 μg/ml). To stimulate cytokine production, lymphocytes were restimulated with 25 μg/ml of MBP for 24 to 72 h. Nonrestimulated LN cells did not produce cytokines. Supernatants were harvested and assayed for cytokine production using cytokine-specific ELISAs.

Groups of five to eight control and five to eight 1,25-(OH)₂D₃-treated mice were sacrificed and perfused with 10 to 15 ml of sterile saline. The treatments were started the day before immunization and both groups of mice were sacrificed when the controls reached EAE severity scores of ≥2.5. The spinal cords were either saved for histopathology or the spinal cords and brains were pooled and manually homogenized in saline using a Dounce homogenizer. The debris was allowed to settle and the cells were resuspended in a 70% Percoll (Pharmacia, Piscataway, NJ) solution. The cell/Percoll suspension was overlaid with a 30% Percoll solution and spun for 15 min at 500 × g. Lymphocytes were collected from the 30/70% interface (22, 23). The lymphocytes were washed twice and relayered onto Histopaque 1083 (Sigma, St. Louis, MO) to remove contaminating red cells and additional debris. The lymphocytes were again collected from the interface, washed twice, and saved for the isolation of total cellular RNA.

Spinal cords from unimmunized B10.PL mice or from B10.PL mice treated with 1,25-(OH)₂D₃ or controls were removed and fixed in 10% formalin. Paraffin-embedded tissue sections were prepared and stained in luxol fast blue and periodic acid Shiffs reagent by the University of Wisconsin Veterinary Science Department. Stained sections were viewed by two individuals who were blinded as to the source of the spinal cords. Inflammation was scored as follows: 0, no sign of inflammation; 1, mild inflammation; 2, discrete lesions with substantial inflammation; and 3, multiple lesions with extensive inflammation.

ELISAs were used to measure IFN-γ, TNF-α, and TGF-β1 (24). The IFN-γ ELISA was performed exactly as described (25). The capture and detection Abs for the TNF-α ELISA were purchased from Genzyme (Cambridge, MA). Color development was with avidin-β-d-galactosidase and p-nitrophenyl-β-d-galactosidase. Murine TGF-β1 was detected using the human TGF-β1 kit exactly as described by Promega (Madison, WI). The ELISA detection limits were 1 ng/ml IFN-γ, 50 pg/ml TNF-α, and 500 pg/ml TGF-β1.

Cells for mRNA analysis were dissolved in acid guanidinium thiocyanate, and total RNA was isolated by the phenol chloroform extraction method (26). Total cellular RNA was reverse transcribed using oligo(dT) primers, according to the manufacturer’s protocols (Promega) and quantitated by competitive PCR. Primers and mimic DNA specific for glyceraldehyde-3-phosphate dehydrogenase (G3PDH), IL-4, IFN-γ, TNF-α, and TGF-β1 were obtained from Clontech Laboratories (Palo Alto, CA) (27, 28). Competitive cDNA mimics, which included the G3PDH, IL-4, TNF-α, IFN-γ, and TGF-β1 primer sequences adjoining a neutral DNA segment, were serially diluted and added to test cDNA aliquots (26, 27). The authentic product to mimic bp sizes were 983/600 for G3PDH, 306/544 for IL-4, 333/500 for IFN-γ, 354/500 for TNF-α, and 525/390 for TGF-β1. The mixture was amplified under predetermined optimal conditions and the products were resolved by 1.5% agarose gel electrophoresis and ethidium bromide stained. The cytokine bands were identified by size with respect to m.w. standards. The mimic DNA dilution that yielded a band with a fluorescence intensity that matched the cytokine band was used to calculate cytokine cDNA copy number. The G3PDH transcript quantitation served as a control for reverse transcription efficiency. Values are reported as cytokine cDNA copies per 1000 copies of G3PDH cDNA.

Where possible, values reported were averages from multiple mice or experiments. Because of the variability in EAE induction, peak severity, and cytokine gene expression from one experiment to another, some values were reported as the values from one representative experiment of two to seven experiments. Again where possible statistical analyses were done using a statistics program for the Macintosh computer (STATVIEW STUDENT). The unpaired two group Student’s t test (and confirmed using the Mann-Whitney U test) was done and values of ρ < 0.05 were considered statistically significant.

When the 1,25-(OH)₂D₃ treatment was started at the same time as EAE induction, the spinal cord sections looked identical to unimmunized B10.PL spinal cords (data not shown). The absence of inflammatory infiltrates from spinal cord sections of 1,25-(OH)₂D₃-treated mice reflected the absence of EAE symptoms (7). When 1,25-(OH)₂D₃ was started at the first symptoms of EAE, the histology scores from control mice were significantly (ρ = 0.0001) higher than those from 1,25-(OH)₂D₃-treated mice (Table I). The differences between the control- and 1,25-(OH)₂D₃-treated animal histology scores at day 14 posttreatment were less striking and statistically insignificant (Table I). In all cases, the histopathology scores reflected the visual EAE severity scores in our mice (Table I).

At various times postimmunization, the LNs draining the site of injection from four to eight mice were collected, pooled, and counted. The number of cells recoverable in the LNs of control mice started between 1 and 2 × 10⁷ cells/mouse and doubled by 10 to 14 days after immunization (Fig. 1,A). Sampling on days 14 and 21 was repeated three and four times, respectively. LNs of control mice yielded 2.6 ± 1.3 (10⁷) cells/mouse on day 14 and 3.9 ± 0.2 (10⁷) cells/mouse on day 21 postimmunization. The number of cells recoverable from 1,25-(OH)₂D₃-fed mice did not increase with time postimmunization (Fig. 1,A). LNs of 1,25-(OH)₂D₃-fed mice yielded 1.1 ± 0.5 (10⁷) cells/mouse on day 14 and 1.8 ± 0.7 (10⁷) cells/mouse on day 21 postimmunization. The control cell yields were significantly (ρ = 0.008) higher than 1,25-(OH)₂D₃ cell yields by day 21 postimmunization. 1,25-(OH)₂D₃ treatment inhibited the in vivo expansion of LN cells both when 1,25-(OH)₂D₃ treatment was started at the time of immunization (Fig. 1,A) and 1 wk after immunization (Fig. 1 B). Although the total number of cells recoverable from the LN of 1,25-(OH)₂D₃-treated mice was fewer than from controls, the proportion of CD4⁺, CD8⁺, Thy-1+7, and class II⁺ cells were not different in the control- and 1,25-(OH)₂D₃-treated animals (data not shown).

IFN-γ and TNF-α secretion was quantitated by assaying the amount of protein secreted in response to MBP in vitro. IFN-γ secretion was first detected 7 days postimmunization and peaked 10 days postimmunization (Fig. 2,A). IFN-γ production from the LN cells of 1,25-(OH)₂D₃-treated mice was the same or in one case higher than IFN-γ secretion from the LN cells of control mice (Fig. 2,B). The magnitude of IFN-γ secreted varied greatly from experiment to experiment but the amount of IFN-γ secreted in cells from 1,25-(OH)₂D₃-treated mice was in all but one case (Fig. 2,B, 15 of 16 experiments) identical to that of controls. The number of IFN-γ transcripts was the same or lower (statistically not different) in CNS samples from 1,25-(OH)₂D₃-treated mice when compared with CNS samples from controls (Table II). Overall, the IFN-γ response of the LN cells reflected the IFN-γ response of cells from the CNS.

T cell-derived (MBP-specific) TNF-α secretion was first detected after 14 days of immunization, and TNF-α decreased thereafter (data not shown). On day 14 postimmunization, cells from control mice secreted 242 ± 15 pg/ml of TNF-α while cells from 1,25-(OH)₂D₃-treated mice secreted 194 ± 59 pg/ml of TNF-α. Although the differences were small, the amount of TNF-α secretion by LN cells from control mice was always higher (10 of 10 experiments) than that from 1,25-(OH)₂D₃-treated mice (data not shown). In the CNS, the number of TNF-α transcripts detectable from 1,25-(OH)₂D₃- and control-treated mice was low and not significantly different (Table II).

Transcripts for IL-4 were first detected in the LN 21 days after immunization. Two weeks of 1,25-(OH)₂D₃ treatment increased IL-4 transcripts 25-fold over the controls (Fig. 3). Although the overall magnitude of the IL-4 response varied from experiment to experiment, this was a highly reproducible result. In six of six experiments, long-term 1,25-(OH)₂D₃-treatment increased IL-4 transcripts 3–25-fold higher then controls. IL-4 transcripts were detectable in samples from 1,25-(OH)₂D₃-treated but not control CNS samples. This was a significant difference (Table II, ρ = 0.0002).

TGF-β1 transcripts were first detected in 1,25-(OH)₂D₃-treated mice at 14 days postimmunization (Fig. 4). Eightfold more TGF-β1 transcripts were detected in the LNs of 1,25-(OH)₂D₃-treated mice than controls 21 days after immunization (Fig. 4). Although the overall magnitude of the response varied from experiment to experiment, this was a highly reproducible result. In seven of seven experiments, TGF-β1 transcripts were 4- to 24-fold higher in LN of mice treated long-term with 1,25-(OH)₂D₃ compared with controls. Similarly, TGF-β1 transcripts were 5- to 6-fold higher (significant ρ = 0.05) in CNS samples from 1,25-(OH)₂D₃-treated mice than in controls (Table II). Because TGF-β1 is extensively posttranscriptionally regulated, we attempted to confirm our transcript measurements by quantitating TGF-β1 protein. MBP-specific TGF-β1 protein was undetectable from LN samples of 1,25-(OH)₂D₃-fed and control-fed mice at any time postimmunization (data not shown). LPS stimulation of resident peritoneal cells from 1,25-(OH)₂D₃-treated mice did make detectable TGF-β1 protein (630 ± 70 pg/ml), while peritoneal cells from control-treated mice did not (≤500 pg/ml).

1,25-(OH)₂D₃ treatment of mice with severe EAE caused a rapid decline in their symptoms (Fig. 5) compared with controls. Three of 10 controls but only 1 of 10 1,25-(OH)₂D₃-treated mice developed severe EAE (scored 5) and were sacrificed within 3 to 5 days of treatment. By day 6 post-1,25-(OH)₂D₃-treatment, the remaining 7 controls had more severe EAE (ρ ≤ 0.05) than the 9 remaining 1,25-(OH)₂D₃-treated mice. The 7 control mice underwent a short-lived recovery at day 11 posttreatment (significant ρ = 0.01, then day 10 EAE scores), followed by a relapse. The 1,25-(OH)₂D₃-treated mice maintained a low level of symptoms as long as the diet was continuously supplemented with 1,25-(OH)₂D₃. Twenty-four hours after 1,25-(OH)₂D₃ treatment, the LN and serum were collected from 5 individual mice per treatment group. Serum calcium was raised slightly in the 1,25-(OH)₂D₃-treated mice, confirming the effectiveness of the treatment (Table III). IFN-γ and TNF-α transcripts were not significantly different following 1,25-(OH)₂D₃ treatment (Table III). IL-4 transcripts were fourfold higher (although statistically not different) in the 1,25-(OH)₂D₃-treated mice compared with controls. TGF-β1 transcripts were fivefold higher and significantly (ρ ≤ 0.04) higher in mice receiving a single injection of 1,25-(OH)₂D₃ 24 h before sacrifice.

Our results provide strong evidence that in vivo 1,25-(OH)₂D₃ treatment arrests the expansion of cells in the LNs of mice with EAE. In addition, histopathologic evaluation of spinal cords revealed less inflammation in 1,25-(OH)₂D₃-treated mice compared with controls. The various 1,25-(OH)₂D₃ treatments did not affect the amount of IFN-γ or TNF-α made on a per cell basis at any stage of EAE disease. However, control mice had as much as twice the number of mononuclear cells in the LNs than 1,25-(OH)₂D₃-treated mice, and the net effect for control mice would be a Th1 response that is twice that of the 1,25-(OH)₂D₃-treated mice. We conclude that 1,25-(OH)₂D₃ is not negatively regulating the Th1 cells that drive the encephalitogenic process, although this hormone may down-regulate the Th1 response through a modest effect on cellular numbers.

TGF-β1 is a potent modulator of cell growth (29). This cytokine inhibits the proliferation and differentiation of T and B cells (29), and antagonizes the effects of specific inflammatory cytokines, including TNF-α, IFN-γ, IL-1, IL-2, and IL-6 (30). More specifically, TGF-β1 is a critical regulator of autoimmune disease (30). Exogenous TGF-β1 can delay the onset and decrease the severity of EAE and arthritis in mice (16, 30, 31). Conversely, neutralization of TGF-β1 in vivo increases the severity of these two autoimmune diseases in the mouse (17, 30). In the present investigation, supplementation with 1,25-(OH)₂D₃ increased the number of transcripts of TGF-β1 mRNA in mononuclear cells both from sensitized LN and from affected central nervous tissue. This phenomenon was apparent in LN cells regardless of the state of disease at the time the hormone was first administered. Moreover, 1,25-(OH)₂D₃ was effective within only 24 h, an outcome that suggests a direct influence on cytokine production. Our current hypothesis, therefore, is that the actions of 1,25-(OH)₂D₃ in preventing and treating EAE may be in part a result of positive regulation of TGF-β1.

We did not detect an MBP-specific increase in TGF-β1 protein in vitro, but LPS stimulated TGF-β1 protein secretion in peritoneal exudate cells of mice given 1,25-(OH)₂D₃ supplements. It therefore seems likely that in our experiments TGF-β1 is coming from a macrophage and not from a T cell. LN cells are greater then 90% lymphocytes and have few macrophage cells relative to peritoneal cells. This is important in view of the finding that macrophage cells constitutively produce vitamin D receptor, while among T cells, activation is required for expression of the vitamin D receptor (32, 33). It seems likely that in EAE, 1,25-(OH)₂D₃ may be up-regulating the production of TGF-β1 by macrophage cells.

Long-term in vivo 1,25-(OH)₂D₃ treatment increased IL-4 transcripts in the LN and CNS compared with controls. Our data suggest that the effect of 1,25-(OH)₂D₃ on IL-4 may be indirect. The increase in IL-4 24 h after 1,25-(OH)₂D₃ injection was not significantly different from controls (Table III). We hypothesize that 1,25-(OH)₂D₃-driven TGF-β1 production by macrophage cells might make the T cell microenvironment more conducive to Th2-type cell differentiation. Experiments are underway to address this point.

We have previously shown that 1,25-(OH)₂D₃ can prevent and diminish the severity of EAE (7). The present results suggest that 1,25-(OH)₂D₃ acts by positively regulating the anti-encephalitogenic cytokines IL-4 and TGF-β1. We were unable to obtain results consistent with a direct effect of 1,25-(OH)₂D₃ on the encephalitogenic Th1 effector cell in terms of either IFN-γ or TNF-α expression. As a result of our 1,25-(OH)₂D₃ treatments, the numbers of cells recoverable from the LN and the number of inflammatory lesions discernible in histopathology sections were dramatically reduced. This could be due, in part, to increased production of TGF-β1, a cytokine known to inhibit T and B cell proliferation (16, 17). Based on the ability of TGF-β1 to ameliorate EAE (16), we hypothesize that 1,25-(OH)₂D₃-mediated up-regulation of TGF-β1 production might be the immunobiologic mechanism underlying the efficacy of 1,25-(OH)₂D₃ for halting the progression of EAE.

We thank Dr. Robert Fritz at the Medical College of Wisconsin (Milwaukee) for scoring our histopathology sections. We also thank Faye Nashold and Jean Humpal-Winter for their technical assistance and Pat Salzwedel for her expertise in animal husbandry.