Vitamin D₃ Metabolism in Human Glioblastoma Multiforme: Functionality of CYP27B1 Splice Variants, Metabolism of Calcidiol, and Effect of Calcitriol

The classic metabolism of the secosteroid hormone vitamin D₃ to its active form followed by several degradation pathways is well described. Vitamin D₃ is hydroxylated in the liver to 25-hydroxyvitamin D₃ [25-(OH)D₃ or calcidiol]. The further hydroxylation to the most active metabolite of vitamin D₃, 1α,25-dihydroxyvitamin D₃ [1α,25-(OH)₂D₃ or calcitriol], is mediated by CYP27B1 in the kidney. 24-Hydroxylation by 1α,25-dihydroxyvitamin D₃-24-hydroxylase (CYP24) is the first step of the main inactivating pathway, resulting in metabolites with decreased activity. The effects of calcitriol on target cells are mediated via binding to the nuclear vitamin D receptor (VDR) that functions as a ligand-dependent transcription factor (1). However, there is increasing evidence for the involvement of nongenomic pathways, mediated by a proposed membrane receptor (2, 3).

Besides renal CYP27B1 activity, extrarenal CYP27B1 expression has been described in a variety of tissues (4). In vitro, several nonrenal cells, including microglial cells (5), produce calcitriol from its precursor. In addition, several cancer cells show CYP27B1 activity (e.g., human non–small cell lung carcinoma cells; ref. 6). Local production of calcitriol has been postulated to play an autocrine or paracrine role in vitamin D–mediated growth control (7).

As shown for numerous normal and cancer cell lines, calcitriol is at high concentrations (10⁻⁹ to 10⁻⁶ mol/L) an antiproliferative and prodifferentiating agent that induces apoptosis and inhibits cell migration (8). The antiproliferative properties of calcitriol, as described above, were also shown for central nervous system tumors, including glioblastoma cell lines (9). Beneficial effects of the vitamin D₃ metabolite 1α-hydroxyvitamin D₃ have been reported for the treatment of glioblastoma multiforme in a small phase II clinical study (10). Its properties in regulating cancer cell growth make the hormone of potential interest in the management of tumors and lead to the development of a broad variety of analogues with weaker calcemic side effects (11).

Previously, we reported gene amplification and mRNA splice variants of the gene encoding the P450 cytochrome 25-hydroxyvitamin D₃-1α-hydroxylase in human glioblastoma multiforme (12).

Gene amplifications are frequent in glioblastoma multiforme, including an amplicon at chromosome 12q13-15 that contains the gene for CYP27B1. The relatively high amplification frequency (16%) of the CYP27B1 gene (12)⁴

As already described for other cytochrome P450 genes (13), alternative splicing can play a role in regulating the enzyme level and may cause tissue-specific variations in healthy cells. A number of studies show that alternative splicing occurs frequently in human cancer cells (e.g., breast and ovarian cancer and glioblastoma multiforme; refs. 14–16).

We used both a descriptional and a functional approach to clarify the role of vitamin D₃ metabolism in glioblastoma multiforme.

Tumor tissues. Tumor samples were kindly provided by the Department of Neurosurgery and were used with consent of the patients. All tumor samples originated from grade 4 glioblastoma multiforme, as determined by the Department of Neuropathology (Homburg, Germany).

Cell culture. COS-1 cells and rat glioblastoma multiforme cell line C6 were purchased from the American Type Culture Collection (Manassas, VA). Cell lines TX3868 and TX3095 were established from glioblastoma and xenografted on mice. The remaining glioblastoma multiforme cell lines were established from cytologic diagnosed glioblastoma multiforme without xenografting. All glioblastoma multiforme cell cultures were maintained in DMEM/Ham's F12 (DMEM/F12) medium (Invitrogen, Karlsruhe, Germany) supplemented with 1 μg/L penicillin/streptomycin (PAA, Coelbe, Germany) and 10% FCS (PAA). Cells were transferred to serum-free DMEM/F12 medium containing the following additives: 1 μg/L penicillin/streptomycin, 1× insulin/transferrin/selenium (ITS-X, Invitrogen), and 1% bovine serum albumin (Sigma, Taufkirchen, Germany) 24 hours before enzyme assays. The following media were used for proliferation assays: serum-free medium as described above or serum-reduced medium DMEM/F12 supplemented with penicillin/streptomycin, ITS-X, bovine serum albumin, and 2% charcoal/dextran-treated FCS (HyClone, Erembodegem-Aalst, Belgium).

Isolation of total RNA and mRNA, reverse transcription-PCR, and nested touchdown reverse transcription-PCR. Isolation of total RNA and mRNA, classic reverse transcription-PCR (RT-PCR), and nested touchdown RT-PCR from tissues and cell culture were done as described previously (12, 17). Normal brain RNA was obtained from Clontech (Palo Alto, CA).

Real-time reverse transcription-PCR. Real-time RT-PCR was carried out using a Light Cycler (Roche Diagnostics, Mannheim, Germany) by measuring the binding of the fluorescence dye SYBR green I (DNA Master SYBR green I, Roche Diagnostics) to double-stranded DNA. The crossing points (Cp, beginning of the PCR exponential phase) for each reaction were determined by the Second Derivative Maximum algorithm and arithmetic baseline adjustment. We used the RelQuant 1.01 quantification software using the principle of efficiency-corrected calibrator normalized relative quantification. Standard curves were assessed in triplicates and confirmed in two independent experiments (see Supplementary Fig. S1). For quantification, the relative ratio of target gene (CYP27B1, VDR, CYP24) to reference gene (β₂-microglobulin) for each sample and for the calibrator was calculated (Cp median, Cpm). The target reference ratio of each sample was then divided by the target reference ratio of the calibrator and corrected for PCR efficiency (ΔCp median, ΔCpm). We determined the variance of ΔCpms for the calibrator of each investigated gene (ΔCp median variance) to show reproducibility.

All PCR reactions were done in duplicate for each cDNA sample. Primer sequences and optimized reaction specificities for each primer pair are given in the Supplementary Table S1. The program was as follows: initial melting for 10 minutes at 95°C, 50 cycles consisting of melting for 10 seconds at 95°C, annealing for 5 seconds at 60/62°C, and elongation for 5 seconds at 72°C with a single fluorescent reading at the end of each cycle. The program was completed with a melting curve analysis. To exclude contamination with DNA, we did Alu-PCR and minus reverse transcriptase control.

PCR and cloning of normal CYP27B1 and splice variants into bacterial and mammalian expression plasmids. The cDNA sequences of the normal enzyme and the splice variants Hyd-V2, Hyd-V3, and Hyd-V4 were generated by RT-PCR as previously described (12). The inserts were amplificated using primers 5′Hyd (5′-CGGGATCCACCCAGACCCTCAAGTACG-3′) containing a BamHI restriction site and 3′Hyd (5′-CCCAAGCTTTAGGGGAAGATGTATACC-3′) containing a HindIII restriction site (5 minutes 94°C, 1 minute 94°C, 45 seconds 57°C, 1.5 minutes 72°C for 26 cycles, 10 minutes 72°C). PCR products were gel purified and extracted with the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). Following BamHI/HindIII digestion, the inserts were ligated into the pQE-30 expression vector in frame to the 6× His-Taq. The recombinant plasmids were transformed in Escherichia coli strain M15(pREP4).

The inserts of CYP27B1, Hyd-V2, and Hyd-V4 cloned in pGEM-T Easy were amplified using primers 5′HydStart (5′-ATGACCCAGACC CTCAAGTACGC-3′) and 3′HydStop (5′-CTATCTGTCCAAAAACTGTAGGTTGATGCTC-3′). PCR products were gel-purified and cloned into the mammalian expression vector pcDNA3.1-TOPO (Invitrogen).

Recombinant expression of normal CYP27B1 and three splice variants. Transformed E. coli cells were grown in Luria-Bertani medium with kanamycin (25 μg/mL) and ampicillin (100 μg/mL).

Protein expression was induced at an absorbance of 0.6 at 600 nm with isopropyl-l-thio-B-d-galactopyranoside in a final concentration of 1 mmol/L. Samples of the bacterial cultures were obtained before induction and 1, 2, 3, and 4 hours after induction of protein expression by isopropyl-l-thio-B-d-galactopyranoside. Bacterial lysates were subjected to SDS-PAGE and electroblotted onto polyvinylidene difluoride-Membrane (Millipore, Bedford, MA). Expression of the recombinant proteins was shown via detection of the His-Taq with Ni-NTA-Conjugate (Qiagen).

Generation and purification of polyclonal antibodies. The synthesis of three CYP27B1-specific antibodies was carried out using antigenic regions of the reported human amino acid sequence. Two antibodies were generated using peptides located in exon 1 (Ab401, amino acids 1-14, MTQTLKYASRVFHR and Ab461, amino acids 35-48, RSLADIPGPSTPS). The peptide for Ab462 is located in exon 5 (amino acids 267-281, HVERREAEAAMRNG). Peptide synthesis, generation, and purification of polyclonal antibodies were done as described previously (18). Protein lysates from tissue samples and cell cultures were prepared as described previously (18, 19).

Western blot of tissue and cell culture lysates. Proteins were subjected to 12.5% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore). Filters were blocked and incubated with primary preabsorbed antibody at a final dilution of 1:375. Monoclonal mouse anti-α-tubulin antibody (Sigma) was used in a 1:10,000 dilution. Blocking experiments were with primary antibody preabsorbed with 100-fold excess of immunizing peptide. BLAST-Algorithm was to exclude homology of the used peptides to other cytochrome P450. Densitometric analysis was done with BioDocAnalyze (Biometra, Göttingen, Germany).

Measurement of CYP27B1 activity in transfected COS-1 cells and glioblastoma multiforme cell cultures. CYP27B1 activity was determined as described previously (20–23). Glioblastoma multiforme cell cultures were seeded in 60-mm² plates at 1.2 × 10⁶ or 1.5 × 10⁶ cells per dish and grown in serum-free medium to 90% confluency. COS-1 cells were seeded in 60-mm² plates at 5 × 10⁵ cells per dish in DMEM/F12 containing FCS. Cells were transfected with pcDNA3.1 plasmids as described above with Effectene Transfection Reagent (Qiagen). The general transfection efficiency was 30 ± 2.4% as quantified by fluorescence-activated cell sorting analysis (Becton Dickinson, Heidelberg, Germany). Twenty-four hours after transfection, cells were washed with serum-free medium and incubated with 1 μmol/L 25-(OH)D₃ (Biomol, Hamburg, Germany) and 0.1 μCi/mL [³H]25-(OH)D₃ (187 Ci/mmol, Amersham, Freiburg, Germany) for 4 and/or 24 hours at 37°C. Media and cells were collected for extraction of vitamin D metabolites with methanol/chloroform (1:4, v/v). The extracts were dried, redissolved in 10 μL hexane, and separated on glass-backed silica-coated high-performance TLC (HPTLC) plates (Kieselgel 60 F₂₅₄, Merck, Darmstadt, Germany) in dichloromethane/isopropanol (9:1, v/v). Standard contained 25-(OH)D₃, [³H]25-(OH)D₃, and 1α,25-(OH)₂D₃ (Biomol). Conversion of tritiated 25-(OH)D₃ was quantified after exposure to Fuji imaging plates using a phosphoimager (BAS-250, Fuji, Stamford, CT). Nonradioactive standards were made visible using UV light. All data were calculated from three independent experiments, each done in duplicates and were normalized against equal cell numbers.

Calcitriol/calcidiol treatment of cell cultures. If not mentioned otherwise, cells were seeded for proliferation assays at 1,000 cells per well into 96-well plates. After 12 hours, cells were maintained 24 hours in serum-free or serum-reduced DMEM/F12. Calcitriol was added in concentrations ranging from 10⁻¹² to 10⁻⁶ mol/L and calcidiol in a dose of 2.5 × 10⁻⁸ mol/L. Cells treated with solvent (ethanol) were included as controls. The medium was replaced every other day containing fresh vitamin D₃ metabolites. After 6 days (144 hours), cell proliferation was determined using WST-1 Reagent (Roche Diagnostics) after manufacturers' instructions.

For real-time PCR, cells were seeded at 6.4 × 10⁶ cells per well into 100-mm² plates. Subconfluent cells were washed and cultured for 24 hours in serum-free medium. Cells were treated with 10⁻⁸ mol/L calcitriol dissolved in ethanol or with solvent alone for 16 hours followed by a calcitriol-free incubation period of 32 hours. Forty-eight hours after treatment, cells were collected and RNA was isolated as described previously.

New splice variants of CYP27B1 in glioblastoma multiforme. We recently reported several new splice variants of CYP27B1 mRNA (17). After cloning further variants from the glioblastoma multiforme cell line TX3868 (Fig. 1), we sequenced 16 splice variants including variant Hyd-V3 with deletion of a part of exon 8; variant Hyd-V2 with deletion of the entire exons 4 and 5; two variants Hyd-V1 and Hyd-V7 with deletion of exons and insertion of introns, respectively; variant Hyd-V4 with insertion of a part of intron 2; and variants Hyd-V5, Hyd-V6, Hyd-V8 to Hyd-V16 with insertions of one or several entire introns. The majority of the sequence variations results in frameshifts and premature termination signals. The only variant retaining the heme-binding and the ferredoxin-binding sites is Hyd-V4, containing a 27 AS in-frame insertion of intron 2. Northern blot analysis of glioblastoma multiforme cell line TX3868 identified two signals at 2.0 and 4.0 kb corresponding in size to variant Hyd-V2 (2,085 bp) and to variant Hyd-V15 (3,904 bp; data not shown).

Metabolism of calcidiol by CYP27B1 and splice variants. We determined the percentile production of calcitriol in transfected COS-1 cells. We selected Hyd-V4, the only variant containing the functional domains necessary for enzymatic activity and Hyd-V2, as an example for the remaining splice variants with deletions of functional domains. The latter splice variants are not likely to have any enzymatic activity. We subcloned the coding sequences of the normal CYP27B1 (1α-Hyd) and the splice variants Hyd-V2 and Hyd-V4 into pcDNA3.1 and expressed them in mammalian COS-1 cells. The cDNA of Hyd-V4 was also cloned in reverse orientation to serve as control. RT-PCR confirmed the expression of the corresponding mRNAs in the transfected cells. HPTLC separation of tritiated calcidiol was used to determine the relative percentile production of calcitriol over a 4-hour incubation period. HPTLC analysis of COS-1 cells transfected with control vector showed a peak at 49 mm coincident with authentic calcidiol standard. COS-1 cells transfected with pcDNA3.1/1α-Hyd showed peaks at 49 and 21 mm coincident with authentic calcidiol and calcitriol standards (Fig. 2A). In COS-1 cells transfected with the normal CYP27B1 cDNA sequence, there was a mean 14.2% conversion from calcidiol to calcitriol (Fig. 2B). COS-1 cells transfected with splice variants Hyd-V2 and Hyd-V4 showed no higher CYP27B1 activity than control cells transfected with Hyd-V4 in reverse orientation. An elevated incubation period of 24 hours revealed similar results (data not shown). Additionally, peaks at 11 and 36 mm were detected, that were referred to in Discussion.

Quantification of CYP27B1 expression in glioblastoma multiforme biopsies and cell cultures. We assessed the expression of the CYP27B1 gene relative to the housekeeping gene β₂-microglobulin by real-time RT-PCR using primers from exon 1 of the CYP27B1 gene. We analyzed glioblastoma multiforme biopsies with and without gene amplification, glioblastoma multiforme cell cultures with and without amplification, and a normal human brain sample (Table 1). A 2-fold increase of gene expression in comparison with normal brain was considered as elevated expression. CYP27B1 expression was elevated in 8 of 10 tumor biopsies with gene amplification (80%), in 4 of 10 tumor biopsies without gene amplification (40%), in 5 of 5 cell lines with gene amplification (100%), and in 4 of 5 cell lines without gene amplification (80%). In total, 12 of 20 glioblastoma multiforme biopsies and 9 of 10 glioblastoma multiforme cell cultures showed an elevated gene expression in comparison to normal human brain. Furthermore, a clear increase of CYP27B1 expression was detected in five of six cell cultures in comparison with the corresponding primary tumor sample (H246, H366, H545, H546, and H556).

Antibodies against CYP27B1 and variants of CYP27B1. We generated antibodies Ab401 and Ab461 for the detection of all CYP27B1 variants using peptides located in exon 1. Additionally, we generated antibody Ab462 using a peptide located in exon 5 (4). We tested the specificity of the CYP27B1 antibodies using CYP27B1 wild type protein and the splice variants Hyd-V2, Hyd-V3, and Hyd-V4, all of them expressed as 6× His fusion proteins in E. coli. The expression of the recombinants was confirmed with a Ni-NTA-conjugate (see Supplementary Fig. S2). The purified antibodies Ab401 and Ab461 identified all four recombinant proteins including the normal protein at 56 kDa, the variant Hyd-V2 at 23 kDa, the variant Hyd-V3 at 49 kDa, and the variant Hyd-V4 at 59 kDa. Antibody Ab462 detected the normal CYP27B1 recombinant protein and the variants Hyd-V3 and Hyd-V4 but not the variant Hyd-V2 that shows a deletion of exons 4 and 5. These results are consistent with the localization of the peptides used to generate the antibodies (see Supplementary Fig. S3).

Additional control experiments on glioblastoma multiforme protein lysates were done with antibodies preabsorbed with 100× immunizing peptide and with preimmune sera (see Supplementary Fig. S4).

CYP27B1 protein in glioblastoma multiforme biopsies and glioblastoma multiforme cell cultures. Using Western blotting, we analyzed the expression of the CYP27B1 gene in 15 glioblastoma multiforme biopsies, two glioblastoma multiforme cell cultures, and three normal human tissues including brain. All antibodies detected a signal at 56 kDa that corresponds to the size of the normal CYP27B1 protein. We also identified additional signals that likely represent splice variants. The predicted protein sizes of the known variants correspond to the sizes of the proteins detected by Western blot analysis. Antibody Ab401 identified proteins at 14.1 kDa likely representing Hyd-V1, Hyd-V9, and Hyd-V16. Antibody Ab461 identified proteins at 22.7 kDa likely representing variant Hyd-V2. Both antibodies identified proteins at 45.5 and 46 kDa likely representing Hyd-V3 and Hyd-V5, respectively, and a protein at 58.9 kDa likely representing Hyd-V4. The variants containing an insertion of intron 1 were not represented by a protein of the corresponding size (9.2 kDa). Antibody Ab462 identified proteins at sizes between 45 and 75 kDa likely representing splice variants Hyd-V3, Hyd-V4, and Hyd-V5 and the normal enzyme. Additionally, all three antibodies detected proteins that are larger than the expected protein sizes (Fig. 3A-C).

The antibodies detected minor differences in the intensity of specific bands. Single differences as for example the increased intensity of the 56-kDa band in sample H205 likely depend on different sensitivity of the antibodies and will not be addressed in detail (Fig. 3A). In general, antibodies Ab401 and Ab461 identified a similar overall expression pattern in glioblastoma multiforme biopsies and glioblastoma multiforme cell cultures but a slightly different expression pattern in normal tissues. Antibody 462 identified an expression pattern different from the pattern found by the antibodies Ab401 and Ab461. However, antibody Ab462 confirmed the similarity of the expression pattern among glioblastoma multiforme biopsies found by the antibodies Ab401 and Ab461. Likewise, antibody Ab462 confirmed the difference between the expression pattern of glioblastoma multiforme samples and normal tissues. Using anti-tubulin antibody for normalization, we did densitometric analysis of CYP27B1 protein in glioblastoma multiforme biopsies. Although the intensity of single bands was different in some cases, antibodies 401, 461, and 462 did not reveal major differences in the overall expression pattern. In summary, our analysis of the normal CYP27B1 protein and its variants showed an expression profile that is (i) similar among glioblastoma multiformes, (ii) different between glioblastoma multiformes and normal human tissues, and (iii) unrelated to the CYP27B1 gene amplification status.

CYP27B1 activity in glioblastoma multiforme cell cultures. We tested glioblastoma multiforme cell cultures for CYP27B1 activity by analyzing the conversion of calcidiol. The conversion was normalized against plates without cells to exclude possible background oxidation of calcidiol. In total, we analyzed nine glioblastoma multiforme cell cultures including five glioblastoma multiforme cell cultures with CYP27B1 gene amplification and four cell cultures without gene amplification. We also included the human glioblastoma multiforme cell line U373 and rat C6 cell line with unknown amplification status of the CYP27B1 gene. The relative percentile production of radiolabeled metabolites of calcidiol was determined over a 4-hour incubation period. A typical HPTLC separation of tritiated calcidiol metabolism in medium alone and in glioblastoma multiforme cells is shown in Fig. 4A. A product peak at 49 mm was coincident with authentic calcidiol standard, whereas the peak of a more polar enzymatic product at 23 mm slightly differed to authentic calcitriol, showing a peak at 21 mm. All glioblastoma multiforme cell cultures showed conversion from calcidiol to this unidentified metabolite with a mean value of 3.5% (Fig. 4B). In detail, the glioblastoma multiforme cell lines with CYP27B1 gene amplification showed a mean conversion value of 2.9% with a low SD and the glioblastoma multiforme cell lines without gene amplification showed a mean conversion value of 4.1% with a higher SD. An extended incubation time of 24 hours resulted in similar results (data not shown).

Effect of calcitriol and calcidiol on glioblastoma multiforme cell lines. We examined the effect of calcitriol on cell proliferation in nine human glioblastoma multiforme cell lines and rat glioblastoma multiforme cell line C6. Melanoma cell line MeWo, known as sensitive for calcitriol (24), served as control. We confirmed a growth inhibition of 56.3 ± 8.8% in the latter cell line by 10⁻⁷ mol/L calcitriol. Glioblastoma multiforme cell lines were cultivated in serum-free medium, unless insufficient proliferation required the use of serum-reduced medium. For C6 cells cultured in serum-free medium, a cell density–specific effect was detectable. Seeding 400 cells, there was a dose-dependent growth inhibition by calcitriol. In contrast, seeding 800 cells, there was a proliferative effect of calcitriol at all concentrations. The proliferative effect was maximal at low doses (10⁻¹⁰ mol/L; Fig. 5A).

Glioblastoma multiforme cell line TX3868 in serum-free medium showed a comparable proliferative calcitriol effect in several doses. There was no significant effect detectable in three other glioblastoma multiforme cell lines (H346, H366, and TX3095; Fig. 5B). In serum-reduced medium, TX3868 also showed a proliferative response to calcitriol at a low dose (10⁻¹⁰ mol/L). A high dose of calcitriol caused a proliferation of TX3095. The other cell lines did not show significant dose-dependent effects of calcitriol. (Fig. 5C).

An analogue treatment with 2.5 × 10⁻⁸ mol/L calcidiol resulted in a proliferative effect on C6 cells cultivated in serum-free medium and seeded at a density of 400 or 800 cells per well. Likewise, we found a proliferative effect for cell culture TX3095 also cultivated in serum-free medium. Cell culture H366 did not show an effect, and cell cultures H346 and TX3868 showed a low antiproliferative effect (Fig. 5D). In serum-reduced medium, we found a proliferative effect of calcidiol in three of nine glioblastoma multiforme cultures and an antiproliferative effect in the remaining six glioblastoma multiforme cell cultures (Fig. 5E).

Quantification of vitamin D receptor and CYP24 expression in glioblastoma multiforme biopsies and cell cultures. We quantified the mRNA expression of the VDR gene and the CYP24 gene in glioblastoma multiforme biopsies and glioblastoma multiforme cell cultures that were used for quantification of CYP27B1. The expression of the genes was normalized against the housekeeping gene β₂-microglobulin (Table 1) and compared with normal brain.

Real-time RT-PCR showed higher mRNA expression of the VDR gene in normal brain than in glioblastoma multiforme biopsies and cell cultures, including glioblastoma multiformes without detectable VDR expression. Out of 20 tumors, only tumor H366 showed a higher expression of the VDR gene in comparison with normal brain (8.5-fold). For the other 19 glioblastoma multiforme biopsies, the expression of the VDR gene was 13.8-fold lower than in normal brain. There was no significant difference in the expression level of the VDR gene between tumors with and tumors without 12q13-14 gene amplification of CYP27B1. Cultures including H366 showed a low expression level of the VDR gene in comparison to normal brain (18.8-fold), with two cell lines lacking any detectable VDR expression.

Real-time RT-PCR showed a strong CYP24 expression in 13 of 20 glioblastoma multiforme biopsies (43-fold higher than in normal brain). In seven biopsies, no expression was detectable. Out of 10 glioblastoma multiforme cell cultures, only culture H246 showed a significant CYP24 expression (76-fold higher than in normal brain). Both, biopsies and cell cultures, showed no correlation between expression of CYP24 and the amplification status of the CYP27B1 gene.

Influence of calcitriol on CYP27B1, VDR, and CYP24 expression. Glioblastoma multiforme cell line T3868, carrying a CYP27B1 gene amplification, was treated with 10⁻⁸ mol/L calcitriol. The expression of CYP27B1, CYP24, and VDR was determined after 24 hours employing real-time RT-PCR.

The expression of CYP27B1 was elevated 4.9 times after calcitriol treatment in comparison to the solvent control (Fig. 6). The expression of the VDR was not significantly increased in comparison with normal brain. Treatment of T3868 with 10⁻⁸ mol/L calcitriol revealed a 24.7-fold elevated expression of CYP24 in comparison with normal brain. By contrast, in cells treated with vehicle alone no CYP24 expression was detectable.

This study contributes to a more complete picture of CYP27B1 variants in glioblastoma multiforme. A majority of splice variants, including Hyd-V6 to Hyd-V8 and Hyd-V10 to Hyd-V15, contain intron 1. These splice variants and splice variants Hyd-V2, Hyd-V3, Hyd-V5, Hyd-V9, and Hyd-V16 result in truncated proteins with deletions of protein domains essential for enzymatic activity, including oxygen-, hormone-, heme-, and ferredoxin-binding sites (25, 26). In agreement with these data, we did not find enzymatic activity in COS-1 cells expressing Hyd-V2. In addition, we did not find residual enzymatic activity in COS-1 cells expressing variant Hyd-V4, which contains domains essential for enzymatic activity but shows an in-frame insertion of 27 AA of intron 2. This insertion may confound the enzymes' conformation and therefore its activity.

These findings leave the question about the biological meaning of the alternative splicing of CYP27B1. A shift to the expression of splice variants followed by nonsense-mediated mRNA decay might cause a lower expression level of mRNA coding the active enzyme (27). However, we found that the majority of CYP27B1 splice variants is not degraded on RNA level but likely exists as proteins without enzymatic activity. The expression of CYP27B1 may be a steady state of fine-tuned equilibrium of stable and unstable mRNAs and active and inactive protein products. In this context, the expression of a wide variety of inactive CYP27B1 protein variants may imply a reduction in the relative amount of active enzyme.

CYP27B1 is ubiquitously expressed in extrarenal cells (4) and the pattern of regulation of CYP27B1 expression differs between kidney and extrarenal cells (6). In this and in previous studies, we report specific CYP27B1 mRNA variants in normal and in tumor tissues (12, 17, 28). The newly generated antibodies detected similar patterns among glioblastoma multiformes and different patterns between glioblastoma multiformes and normal human tissues, indicating a tissue-specific mode of CYP27B1 expression. Such a tissue-specific expression pattern has been suggested as prominent sign for the role of alternative splicing as regulation mechanism (29).

To evaluate the biological meaning of CYP27B1 gene amplification in glioblastoma multiforme, we determined the mRNA and protein expression levels. We found an elevated mRNA expression in glioblastoma multiforme biopsies with gene amplification in contrast to tumors without gene amplification. Western blot experiments did not confirm an elevated expression in glioblastoma multiforme biopsies with gene amplification. This inconsistency may reflect a cellular mechanism that levels out mRNA expression differences that were due to amplification. Alternatively, this inconsistency may be caused by different sensitivity of the quantification methods.

The CYP27B1 mRNA expression was elevated in the majority of glioblastoma multiforme cell cultures in comparison with the corresponding tumor biopsy. This difference may be in part explained by the high heterogeneity of glioblastoma multiforme tissues in comparison to cell cultures that are less heterogeneous due to clonal selection (30). Selective pressure in favor of an elevated CYP27B1 expression in glioblastoma multiforme cell cultures may indicate a potential significance of CYP27B1 activity for neoplastic growth in glioblastoma multiforme.

Recently, the 3-epimerization of calcidiol was shown in various cultured cells, including osteosarcoma, colon adenocarcinoma, and hepatoblastoma cells. In addition, it was shown that 3-epi-25-(OH)D₃ is metabolized by CYP27B1 to 3-epi-1α,25-(OH)₂D₃ (31). The 3-epimere of calcitriol is known to be more stable than calcitriol but to have lower antiproliferative and differentiation inducing activity for various cell lines (31–34). This decrease in biological activity can be explained by a reduced binding affinity to the VDR (31, 33, 35). The conversion to this metabolite has been reported to be tissue specific and to depend on the degree of proliferation (33, 36). The mobility of the unidentified metabolite of calcidiol with a peak at 23 mm may possibly indicate a 3-epimere of calcitriol in glioblastoma multiforme cell lines. The possible formation of 3-epi-calcitriol would in turn indicate CYP27B1 activity in glioblastoma multiforme cell lines. However, final proof of a 3-epimere of calcitriol requires experiments that show comigration of standard 3-epi-1α,25-(OH)₂D₃.

Besides the peak at 21 mm corresponding to calcitriol, CYP27B1-transfected COS-1 cells show further peaks likely resulting from the conversion of calcidiol in COS-1 cells. These peaks may correspond to 24,25-(OH)₂D₃ (36 mm), 3-epi-1α,25-(OH)₂D₃ (23 mm), and 1α,24,25-(OH)₃D₃ (11 mm) and thus possibly indicate a background CYP24 activity in COS-1 cells and an active vitamin D₃ metabolism, which would be expected for a renal cell line. The small peaks at 21 mm in untransfected cells and in COS-1 cells transfected with control vector may be due to endogenous CYP27B1 activity or to possible oxidation of calcidiol to 19-nor-10-oxo-25(OH)D₃. It is conceivable that these processes occur in all samples.

It is well known that calcitriol can both inhibit and stimulate growth of normal and tumor cells. These effects have been reported to be dependent on the doses of calcitriol and on the presence of growth factors in the culture medium (8, 37). Calcitriol can induce glioma cell death in a dose-dependent manner in rat glioma cell line C6 (38, 39) and in human glioblastoma cell lines (40) but not in primary astrocytes (41). Calcitriol sensitivity varies between different glioblastoma multiforme cell cultures (42). We confirmed the antiproliferative mode of action of calcitriol at high doses (10⁻⁶ mol/L) in C6. We showed that calcitriol can have a proliferative effect at low doses (10⁻¹⁰ mol/L) in these cells. Whereas the majority of human glioblastoma multiforme cell lines were calcitriol insensitive, we also found a few cell lines exhibiting a proliferative response to calcitriol treatment, providing first evidence for a mitogenic effect of calcitriol on glioblastoma multiforme cell lines.

It has been previously reported that incubation of prostate cancer cells with calcidiol has the same antiproliferative effect as calcitriol (43). It is possible that the formation of calcitriol from calcidiol may be involved in the calcidiol-mediated effects. Magrassi et al. suggested an antiproliferative effect of calcidiol on glioblastoma multiforme cells (40). We treated a subset of glioblastoma multiforme cell lines with a dose of 2.5 × 10⁻⁸ mol/L calcidiol. Like for calcitriol, we also found antiproliferative and proliferative effects of calcidiol. The effects of calcidiol may be mediated by calcidiol itself or by vitamin D₃ metabolites that are derived from calcidiol, as discussed by Lou et al. for prostatic stromal cells (44).

Several classic vitamin D₃ modes of action are described to be mediated through VDR-dependent transcriptional regulation. The mRNA expression of the VDR gene in glioblastoma multiforme biopsies and glioblastoma multiforme cell cultures was very low or undetectable in comparison to normal brain. Treatment of glioblastoma multiforme cell culture T3868 with 10⁻⁸ mol/L calcitriol did not reveal a significant increase in VDR expression. We found, however, a dose-dependent proliferative calcitriol effect on glioblastoma multiforme cell lines. The lack of significant VDR expression in glioblastoma multiforme cell lines indicates other pathways to play a role in vitamin D₃ metabolite–mediated effects on cell growth. Baran et al. hypothesized a membrane receptor responsible for fast nongenomic vitamin D₃ metabolites signaling (45). Magrassi et al. and Elias et al. proposed a receptor specific for vitamin D₃ compounds with low affinity to the nuclear VDR (46, 38).

Of the glioblastoma multiforme biopsies, 65% of samples show expression of CYP24, known to be regulated by calcitriol (47), indicating an active 24-hydroxylation pathway in glioblastoma multiforme, probably depending on the calcitriol level in these tumors. By contrast, in the majority of glioblastoma multiforme cell cultures, we did not find significant CYP24 expression. This may be due to a low calcitriol concentration in culture medium, a notion that is supported by the observation that the addition of 10⁻⁸ mol/L calcitriol to the culture medium resulted in an induction of CYP24 gene expression in T3868.

Several studies reported a resistance of CYP27B1 gene expression to the negative regulatory effects of calcitriol for cells derived from extrarenal sites (48, 49). Calcitriol treatment of glioblastoma multiforme cell culture T3868 carrying a CYP27B1 amplification resulted in an increase of CYP27B1 transcription in contrast to vehicle treated cells. Because a VDRE has not been identified thus far in the CYP27B1 promoter, there is no evidence for a transcriptional regulation of CYP27B1 via the VDR pathway. Other mechanisms must explain the enhancement of CYP27B1 gene expression after calcitriol treatment. Regarding the existence of splice variants in glioblastoma multiforme, it is not possible to make conclusions about an increased CYP27B1 activity.

This study provides first evidence for different modes of action of calcitriol in glioblastoma multiforme. Depending on cell environment and possibly on different molecular profiles of the vitamin D₃ metabolism, there seem both proliferative and antiproliferative effects. The mechanisms behind these effects are likely to be rather complex: (i) Gene amplification or simply increased CYP27B1 expression, resulting not necessarily in a higher level of active enzyme but in a deregulation of transcription/translation of the enzyme. (ii) Deregulation of the tissue-specific expression of splice variants in glioblastoma multiforme, resulting in a higher level of inactive enzyme. (iii) A special catabolism of calcitriol in glioblastoma multiforme, resulting in a change of calcitriol levels or in the formation of additional vitamin D₃ metabolites with unknown effects. (iv) An elevated expression of CYP24 in contrast to normal brain that causes an increased degradation of calcitriol. (v) Low expression of VDR and the expression of another, not yet identified receptor, mediating proliferative effects.

Specifically, in the light of the large number of splice variants, any therapeutically concept should take into account the complexity of the vitamin D₃ metabolism in glioblastoma multiforme. The investigation of escape mechanisms in glioblastoma multiforme may result in new ways to manipulate glioblastoma multiforme growth employing effects of vitamin D₃ metabolites and therefore may lead to new modes of therapy.