| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Biochemistry and Molecular Biophysics (G.K.W., H.T.L.D., T.B., G.H., C.E.D., M.R.H.) and Microbiology and Immunology (S.F.S., R.M.B., J.J.M.), College of Medicine, University of Arizona, Tucson, Arizona 85724; Department of Zoology and the Division of Life Sciences (L.A.M., J.H.Y.), University of Toronto at Scarborough, Ontario, Canada M1C 1A4
Address all correspondence and requests for reprints to: G. Kerr Whitfield, Department of Biochemistry & Molecular Biophysics, University of Arizona College of Medicine, 1501 North Campbell Avenue, P.O. Box 245042, Tucson, Arizona 85724. E-mail: kerr{at}medbioc.arizona.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
VDRs have been characterized from mammals (6, 7, 8), birds (9), Xenopus laevis (10), zebrafish (GenBank accession no. AAF21427), and two VDRs from Japanese flounder, Paralichthys olivaceus (11). All of these species represent vertebrates with calcified endoskeletons. True VDRs are unknown in nonchordate species; the currently known nonchordate protein with the highest similarity to VDR is the ecdysone receptor (12), found in insects and crustaceans, with 27–30% amino acid sequence identity to human VDR (hVDR). Before the recent cloning of zebrafish VDR and flounder VDRa and VDRb, it was speculated that the 1,25(OH)2D3-VDR system originated with terrestrial tetrapods (10). Such an origin of VDR, per se, has become untenable with the cloning of fish VDRs; also, it is known that a variety of fish species, including a shark (Prionace glauca) and lamprey (Entosphenus japonicus) (13), contain appreciable levels of plasma 1,25(OH)2D3, ranging from 28–274 pg/ml (0.07–0.66 nM; Refs. 13, 14, 15, 16, 17). Nevertheless, in the absence of studies to determine the exact physiologic role of 1,25(OH)2D3 in fish, it is still possible that participation of VDR and 1,25(OH)2D3 in calcium homeostasis may indeed date from a time when animals could no longer access an unlimited supply of calcium from their aquatic environment (10).
Lampreys represent, with hagfishes, jawless (agnathan) fishes with the most ancient lineage among extant vertebrates (18, 19, 20). The placement of lampreys in the evolution of vertebrates is not completely clarified, but recent phylogenies, based on new finds of primitive lower Cambrian vertebrates (e.g. Ref. 20), suggest that lampreys belong to a group of soft-bodied vertebrates that diverged before the development of calcified structures (19, 20). Given this tentative phylogenetic position, plus the above-cited evidence that lampreys contain significant circulating levels of 1,25(OH)2D3, the active hormonal form of vitamin D, the present study was undertaken to determine whether lampreys contain a functional VDR. In this report, we describe three distinct VDR cDNA clones obtained from both larval and juvenile specimens of sea lamprey, Petromyzon marinus, that appear to reflect differences in mRNA splicing. In a limited survey of tissues, skin and mouth represent major sites of VDR mRNA expression in juvenile lamprey. Relatively little VDR mRNA was found in juvenile intestine, unlike the situation in mammals and birds. Lamprey VDR (lampVDR) is functional in that it binds 1,25(OH)2D3 with high affinity and transactivates a transfected reporter construct containing a VDRE derived from the human CYP3A4 gene (21), which encodes a cytochrome P450-containing enzyme (CYP) implicated in detoxification of xenobiotics (22). Therefore, we present the first evidence that the appearance of VDR in the vertebrate lineage may actually predate calcified tissues, and we postulate that its original, noncalcemic role might have been, at least in part, the induction of P450 detoxification enzymes.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
PCR using degenerate primers
First-strand cDNA was made from protospleen poly(A)+ RNA (1 µg) using an RT-PCR kit with supplied oligo(deoxythymidine) primers (Stratagene Corp., La Jolla, CA). For the design of degenerate VDR primers for PCR, two areas of amino acid identity among human, mouse, rat, and chicken VDRs were found near the N and C termini. A sense primer encoding part of the open reading frame near the N terminus, corresponding to codons 33–40 of hVDR (-GFHFNAMT-), was 5'-GGNTTYCAYTTYAAYGCNATGAC-3' (where N is a mixture of all four bases and Y is a mixture of both pyrimidines). An antisense primer, corresponding to codons 394–401 near the C terminus of hVDR (-NEEHSKQY-), was 5'-TAYTGYTTNGARTGYTCYTCRTT-3' (abbreviations as above, with R being a mixture of both purines). The PCR contained 2 µl of the first-strand cDNA reaction (described above) along with 500 ng of each degenerate primer and 2.5 U Taq polymerase (Life Technologies, Inc., Rockville, MD) in a 50-µl reaction volume. PCR conditions were as follows: presteps, 94 C for 1 min and 78 C for 3 min, during which the Taq polymerase was added; 40 cycles of 94 C for 30 sec, 55* C for 30 sec, and 72 C for 1 min 20 sec; and a final incubation at 72 C for 10 min, followed by storage at 4 C. The annealing temperature (indicated by an asterisk) was decreased from 55 C to 43 C in 0.3 C increments over 40 cycles.
Cloning and sequencing
PCR products were subjected to agarose gel electrophoresis, excised from the gel, and purified using a Qiaex II agarose gel extraction kit (QIAGEN, Inc., Valencia, CA). Purified PCR products (7.5 µl) were cloned into T-vector (Promega Corp., Madison, WI) and sequenced using a T7 Sequenase 2.0 kit (Amersham Pharmacia Biotech Inc., Piscataway, NJ). Alternatively, preparations of T-vector harboring PCR products were sent to an in-house core facility for automated sequencing.
5'- and 3'-rapid amplification of cDNA ends (RACE)
A cDNA library was constructed from larval protospleen poly(A)+ RNA (1 µg) using a Marathon cDNA amplification kit (CLONTECH Laboratories, Inc., Palo Alto, CA). 5' and 3' RACE reactions were then performed using the AP1 primer from the kit in combination with VDR-specific primers designed from partial lampVDR sequence data. The successful 5' and 3' primers are underlined in Fig. 1A
(the 5' primer was the antisense complement of the underlined sequence). To identify the lamprey-specific product, 5'-RACE reactions were subjected to agarose gel electrophoresis and then blotted onto an Immobilon Ny+ membrane (Millipore Corp., Bedford, MA) by capillary action. As the hybridization probe, a 210-bp NcoI-AflIII fragment including 149 bp from the 5'-end of the original lamprey clone was biotinylated using a biotin high prime kit (Roche Molecular Biochemicals, Indianapolis, IN), and then 4 µl of this probe was incubated with the Immobilon blot at 42 C overnight in 10 ml of a solution containing 5x SSPE (1x SSPE = 0.18 M NaCl; 10 mM NaxPO4, pH 7.4; 1 mM EDTA), 5x Denhardt’s solution, and 50% formamide. Washes were 2 x 5 min at room temperature in 2x SSPE/0.1% sodium dodecyl sulfate and then 2 x 15 min at 65 C in 0.1x SSPE, 0.1% sodium dodecyl sulfate. Detection of hybridizing bands was carried out using 5-bromo-4-chloro-3-indolylphosphate and 4-nitro blue tetrazolium (both reagents from Roche Molecular Biochemicals) as previously published (23).
|
PCR products were resolved on agarose gels, excised, purified, and cloned into T-vector as described above. After confirmatory sequencing, the T-vector clone was digested with EcoRI and BamHI, and the lamprey insert of 1347 bp, which lacked insert A (see Fig. 1B
, top), was purified on an agarose gel and cloned into predigested pSG5 vector (24). Site-directed mutagenesis of the pSG5-lampVDR to remove insert B was performed using the oligonucleotide primer 5'-GGCTATCTGCCTCTTCTCTCCCG ACCGGCCTGGCGTACAAGACCG, and its complement, corresponding to 22–23 bases on either side of the insert (see Fig. 1B, bottom), in a PCR-based protocol using a QuikChange kit (Stratagene Corp.). The resulting clone is referred to as lampVDR.
Cloning of lampVDR from juvenile tissues
First-strand cDNA was prepared from juvenile intestine, skin, and mouth poly(A)+ RNA as described above for larval protospleen and then subjected to PCR amplification using primers flanking insert B in the larval VDR clones: forward primer 5'-CATTGAAATCATCATCCTCCGC-3', corresponding to bp 738–759 of the lampVDR cDNA (Fig. 1), and reverse primer 5'-ATCTCTCCACTTCGTCCATCCC-3', corresponding to the antisense of bp 1244–1264. Likewise, primers flanking insert A were designed, consisting of forward primer 5'-GCAACATCACCAAGGACAACCG-3', corresponding to bp 237–258, and reverse primer 5'-GCATTTCTACATTCAGGACTGCCATC-3', corresponding to the antisense of bp 518–543. The absence of inserts A and B was confirmed by sequencing of independent clones from mouth and skin using each primer pair.
Sequence alignments and phylogenetic comparisons
Sequences for VDRs and other related receptors were obtained from GenBank, often by using BLAST searches (http://www.ncbi.nlm.nih.gov/BLAST/). Sequences were aligned using Clustal W, version 1.8 (25). Aligned sequences were then used to construct cladograms in Clustal W but also in Puzzle (version 4.0.2; Ref. 26) and in the Protpars routine from Phylip (version 3.573c; Ref. 27). The insect ultraspiracle protein from Locusta migratoria (accession no. AAF00981.1) and the nhr-8 protein from Caenorhabditis elegans (accession no. AAB88373.1) were used as outgroups.
Transfection and transactivation assays
The pSG5 plasmids (25–50 ng/well) harboring cDNAs for lampVDR, hVDR, or pSG5 vector without cDNA insert were transfected into COS-7 monkey kidney cells (60,000/1.88 cm2 well in a 24-well plate), using a calcium phosphate coprecipitation procedure as previously described (28). All transfections included a human GH reporter plasmid (1.0 µg/well) linked to two or more copies of VDREs from vitamin D-regulated genes. VDREs used for this analysis included the rat osteocalcin VDRE (29) as well as two distinct VDREs from the human CYP3A4 gene, namely a distal VDRE (21) of the DR3 type, with the sequence 5'-GGGTCAgcaAGTTCA-3' (two hexanucleotide half-sites in uppercase), and a proximal VDRE of the ER6 type (21, 30), with the sequence 5'-TGAACTcaaaggAGGTCA-3' (upstream half-site is everted). Some transfections also contained 50 ng/well of a TIF plasmid expressing Danio rerio (zebrafish) RXR
(31). Cells were incubated for 24 h with 1,25(OH)2D3 or ethanol vehicle, and then culture media were assayed for secreted GH using a kit from Nichols Institute (San Juan Capistrano, CA).
Hormone-binding assay
The pSG5 plasmids (10 µg) harboring cDNAs for lampVDR, hVDR, or pSG5 without cDNA insert were transfected into COS-7 cells (3 x 106 cells/150-mm plate) as described above but without the VDRE reporter construct. After a 48-h incubation without hormone, cells were washed twice with PBS, and then cell lysates were prepared as described in Nakajima et al. (32). The ability of expressed lampVDR to bind [3H]-1,25(OH)2D3 was assessed as described in Whitfield et al. (28), with the following modifications: 10–30 µl of lysate were combined with 30 µl rat liver nuclear lysate, prepared according to Nakajima et al. (33), in 200 µl total volume. [Rat liver contains both RXR and RXRß (34) to serve as potential heterodimeric partners for VDR.] The 1,25(OH)2-26,27-[3H]dimethyl-vitamin D3 (Amersham Pharmacia Biotech, original specific activity 163 Ci/mmol) was diluted to 18 Ci/mmol with unlabeled 1,25(OH)2D3 and then further diluted with 90% ethanol/5% isopropanol. After a 15-min incubation on ice, 10 µl of the appropriate dilutions were added to achieve the following approximate final concentrations in nM: 0.1, 0.2, 0.4, 0.8, and 1.6. Binding reactions were incubated overnight on ice. Unbound hormone was removed by incubation for 15 min at 4 C with 80 µl of a 3% suspension of dextran-coated charcoal (Sigma-Aldrich Corp., St. Louis, MO) prepared in 0.15 M NaCl, 0.015 M NaN3, sodium phosphate buffer, pH 7 (0.085 M Na2HPO4 and 0.055 M NaH2PO4), and 0.1% gelatin. Samples were centrifuged at 10,000 x g for 2 min, and supernatants (200 µl) were taken for scintillation counting. Data were transformed for Scatchard analysis using LIGAND (35), a program originally written in BASIC and adapted for the Macintosh by Robert E. Williams (Xoma Corp., Santa Monica, CA).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A 3' RACE reaction yielded a 488-bp product containing the 3' end of the lampVDR coding region, an in-frame stop codon, and 353 bases of 3'-UTR. Numerous 5' RACE reactions using a variety of primers finally produced a 260-bp product that hybridized to a lampVDR probe (see Materials and Methods). This 5'-RACE product contained the 5' end of the lampVDR coding region and 31 bases of 5'-UTR. In addition, two in-frame stop codons (Fig. 1A, enclosed in boxes at top) were found within the 5'-UTR, making it highly likely that the complete coding region of the lamprey cDNA clone had been obtained. High-fidelity PCR was then performed using primers in the 5'- and 3'-UTRs (primer sequences given in Materials and Methods). Two independent PCRs each produced a 1349-bp product, which was cloned into T-vector for confirmatory sequencing.
The sequence presented in Fig. 1A represents a composite of nine independent, overlapping PCR products, seven partial sequences, and two high-fidelity cDNAs containing the entire coding region. The open reading frame, discounting the two out-of-frame insertions, predicts a protein of 406 amino acids, compared with a range of 420–453 residues for other known VDRs (6, 7, 8, 9, 10, 11). Figure 1B displays the two insertions found in larval clones. Insert A (31 bp), found only in the original clone produced by PCR with degenerate primers, occurred precisely at the junction between exons 3 and 4 in the human VDR gene (36). Insert B (41 bp) occurs at a location exactly matching the junction between exons 8 and 9 in the rat (37) and human (36) VDR genes and was found in all five clones obtained from the original larval protospleen. A minor fraction of rat VDR cDNA has been reported (37) to contain a much larger insertion (retained intron) in exactly the same position as insert B, resulting in expression of a truncated VDR with dominant negative activity.
Expression of lampVDR mRNA in juvenile lamprey tissues
To show that lampVDR transcripts exist in which both insertions are lacking, we prepared poly(A)+ RNA from three tissues dissected from juvenile lampreys. Skin, mouth, and intestinal RNAs were subjected to RT-PCR using two sets of primers (Fig. 2). One primer set matched sequences in lampVDR-flanking insert A (Fig. 2A), and the second primer set flanked insert B (Fig. 2B). PCR products from both reactions were cloned and sequenced. As can be seen in Fig. 2A, the sequences of PCR products from both mouth and skin were nearly identical with the original larval clone except that insert A was missing from both clones; one clone, that from skin, contained a Thr codon in place of a Met codon at position 98 (three codons past where insert A would have occurred). Likewise, the PCR products seen in Fig. 2B also matched the original lamprey clone except that insert B was missing from both clones, and one clone, this time from mouth, contained a different codon at position 371, encoding Gly instead of Ser. Also, both clones differed with the original larval clone at codon 360, which was ACC (Thr) in the larval form but GCC (Ala) in both juvenile clones.
|
, lower right panel).
|
and data not shown).
General comparison of lampVDR with related nuclear receptors
The deduced amino acid sequence of lampVDR showed an overall amino acid identity of 59–62% with other vertebrate VDRs (Fig. 3). The highest overall identity to lampVDR among other vertebrate VDRs is seen with flounder VDRa (62.2%), and the lowest identity is with X. laevis VDR (58.4%). By comparison, human VDR exhibits 67% or greater identity with other vertebrate VDRs (excluding lampVDR). Thus, lampVDR represents the most divergent member of the known VDRs. The amino acids that show identity to other VDRs tend to occur in regions of known functional significance. Accordingly, the zinc finger DNA-binding domain (DBD, black shading) in lampVDR contains 58 of 66 residues (87.9%) that are identical with those in human VDR. Also, of 14 residues known to contact the 1,25(OH)2D3 ligand in the hVDR ligand-binding domain crystal (39) (these residues are localized in the gray shaded areas in Fig. 3 and are circled in Fig. 1A), 13 residues (92.9%) are identical in lampVDR. The heterodimerization interface with RXR (hatched in Fig. 3) is less well defined in VDRs, with the only relevant crystal data being from a heterodimer of mouse RXR with the all-trans-retinoic acid receptor, hRAR. By analogy to this crystal, the general regions involved in heterodimerization (hatched in Fig. 3) display from 57–63% identity when lampVDR is compared with known VDRs from other vertebrate species (see also Fig. 7, below).
|
Assessment of 1,25(OH)2D3 hormone binding by lampVDR
Larval lampVDR cDNAs in a pSG5 expression construct, both containing and lacking insert B, were transfected into VDR-deficient COS-7 cells to test the ability of expressed lampVDR to bind radiolabeled 1,25(OH)2D3, using pSG5-hVDR as a positive control. LampVDR expressed from a cDNA that retained insert B was unable to bind hormone at the tested concentrations of 1,25(OH)2D3 (0.1–1.6 nM), showing only background levels of [3H]1,25(OH)2D3 in the protein-bound fraction, which, in COS-7 cells, typically constitute less than 500 receptors/cell (data not shown). When COS-7 cultures were transfected with either insertless lampVDR or an hVDR positive control; however, the number of high-affinity, saturable receptors was much higher, averaging 5.4 ± 2.7 x 105 receptors/cell for lampVDR and 3.0 ± 0.4 x 105 receptors/cell for hVDR (±SD, n = 3). Moreover, as illustrated in Fig. 4B, the insertless lampVDR cDNA expressed a protein that bound 1,25(OH)2D3 with high affinity. In three independent experiments, dissociation constants for 1,25(OH)2D3 of lampVDR ranged from 0.1 to 1.4 nM and were typically three times greater than the dissociation constants determined for matched hVDR controls (Fig. 4A). Thus, lampVDR can be considered a bona fide VDR based on its high affinity for the 1,25(OH)2D3 hormonal ligand.
|
, compare right two bars with left two bars). In contrast, hVDR mediated a robust (40.8x) transcriptional response from the osteocalcin VDRE (Fig. 5A), as shown in many previous studies (reviewed in Ref. 1). When tested on an everted repeat (ER6) element from the human CYP3A4 gene (Fig. 5B), lampVDR appeared to mediate a small, but not significant, transcriptional response to 10-8 M 1,25(OH)2D3, again in contrast to hVDR, which showed a nearly 18-fold induction by 1,25(OH)2D3. However, lampVDR was able to mediate a significant functional response (3.91x) from the CYP3A4 distal VDRE of the DR3 type (Fig. 5C). This response was enhanced to 4.5x (Fig. 5D) by the cotransfection of zebrafish RXR. In three independent experiments using zebrafish RXR, the reporter gene up-regulation mediated by lampVDR in the presence of 10-8 M 1,25(OH)2D3 was 4.54-, 5.21-, and 4.50-fold, compared with values with the insertless pSG5 expression vector of 1.17-, 1.34-, and 1.44-fold, respectively. LampVDR also appeared to be capable of inhibiting reporter gene transcription in the absence of hormone (e.g. Fig. 5D, white bar marked with ***).
|
). The resulting cladogram confirms that lampVDR is in subgroup NR1I containing known VDRs (1I1) as well as PXRs (1I2) and CARs (1I3). An expanded analysis of lampVDR with a more extensive sampling of group 1I nuclear receptors is shown in Fig. 6B. This analysis includes receptors from mammals, one avian species, one reptile, two amphibians, and three bony fishes. Two VDR sequences, from G. gecko (reptile) and B. marinus (amphibian), were obtained in our laboratory and are here reported for the first time. In addition, the analysis includes the closest BLAST hits to human and lampVDRs from the genomes of tunicate (Ciona intestinalis), fruitfly (Drosophila melanogaster), and nematode (C. elegans). Among the subgroup 1I receptors, lampVDR unambiguously groups with other VDRs. Numerous other phylogenetic analyses have been performed using different selections of receptors and different regions of the amino acid alignment, but, as shown here, lampVDR always forms the basal member of the VDR grouping and does not group with either the PXRs or CARs (data not shown).
|
. The first of these regions (left side of figure) is the N-terminal region that contains a docking site for general transcription factor IIB (TFIIB) in hVDR (42) and the chicken thyroid receptor- (cTR; Ref. 43). In both cases, the interacting region contains two or more basic residues (indicated by circled plus symbols and enclosed by brackets for hVDR and cTR), flanked on the N-terminal side by a single acidic residue (indicated by a boxed minus symbol). The sequence of lampVDR in this region contains only one of the two basic amino acids that are otherwise conserved in all vertebrate VDRs (missing basic residue and other selected sequence divergences highlighted in yellow). Also, lampVDR is similar to the PXRs in that the putative TFIIB docking site is flanked on the N-terminal side by an extensive stretch of acidic residues. These differences between lampVDR and other vertebrate VDRs could conceivably impact the transactivation potential of lampVDR when tested in a mammalian system. Indeed, a common polymorphic variant of hVDR has been reported in which the removal of a negatively charged residue from the N terminus enhances transcriptional activity (42, 44).
The second region analyzed is a C-terminal extension (CTE) of the DNA-binding region (Fig. 7, center). This region contains clusters of basic amino acids (shown in red) that are conserved between lampVDR and hVDR in a pattern that is distinctive for each subgroup of group 1 receptors (45).
The third region examined in Fig. 7 is a portion of the C-terminal ligand-binding domain that has been shown to form part of the dimerization interface in a cocrystal of hRAR with mouse RXR (46). Unlike the previously discussed two regions, the sequence of lampVDR in this putative dimer interface is quite divergent from that of other VDRs (differing residues shown in yellow) and also related receptors, PXRs and CARs (Fig. 7 and data not shown).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
cladogram); 2) the next closest receptor by BLAST analysis, PXR, shows less than 44% identity to lampVDR; 3) lampVDR binds 1,25(OH)2D3 with high affinity; and 4) lampVDR is capable of mediating 1,25(OH)2D3-dependent transactivation from a VDRE-containing reporter gene. However, our data cannot exclude the possibility that the lamprey genome may harbor a second VDR gene, such as is found in pufferfish and flounder. The present report of a bona fide VDR from an agnathan reveals that this nuclear receptor has a more ancient origin than previously suspected. Complete cDNAs have been reported of apparent VDR homologs from teleost fishes, and, taken together with previous reports of high-affinity 1,25(OH)2D3 binding in species such as trout (47), eel (16), and cod (17), provide evidence that teleost fishes possess functional VDRs. However, lampreys represent a far more ancient group of fishes than teleosts, having diverged from other vertebrates over 450 million years ago (18, 19, 20).
The affinity of lampVDR for 1,25(OH)2D3 is slightly lower than that of human VDR. This could be due to variation in a residue known to be a ligand contact in the human receptor (39), namely Val-418, which is a leucine in lampVDR. It is perhaps worth noting that the levels of circulating 1,25(OH)2D3 reported for a Japanese lamprey species, E. japonicus (13), are, at 274 pg/ml, approximately 7- to 8-fold higher than the average 1,25(OH)2D3 levels of 33 pg/ml (48) and 36 pg/ml (49) reported for normal human subjects. Thus, the lower ligand-binding affinity of lampVDR determined in the present study would seem to be compatible with the higher circulating levels of 1,25(OH)2D3, at least in one lamprey species.
Additional functional testing revealed that lampVDR is capable of modest but reproducible and significant 1,25(OH)2D3-dependent transactivation from a DR3-type responsive element found in the human CYP3A4 gene (Fig. 5, C and D). LampVDR did not show measurable activity from either the rat osteocalcin VDRE, also of the DR3 type, or ER6-type element in the human CYP3A4 gene. The fact that lampVDR transcriptional activity is of a lower magnitude (5x), compared with hVDR (20x) when tested in a mammalian cell transfection system, could be due to several potential factors. One possibility is that lampVDR may not interact well with mammalian proteins required for transactivation, such as with TFIIB or the endogenous (monkey) RXR isoforms in the host cells used for transfection. As noted above, the amino acid sequence of lampVDR is quite divergent from that of other VDRs in a region corresponding to a dimerization domain in hRAR (Fig. 7, bottom right). This possibility is further supported by gel mobility shift analyses, which demonstrated that lampVDR was incapable of forming a detectable complex with human RXR on either the rat osteocalcin VDRE or mouse osteopontin VDRE (with the sequence GGTTCAcgaGGTTCA) (50), under conditions that allowed for strong complex formation by hVDR (data not shown). That mammalian RXRß or RXR
isoforms are also not ideal partners for lampVDR is strongly suggested by transactivation experiments performed in HeLa cells (containing mainly human RXRß) (51) or HT-29 cells (containing mainly human RXR) (52), which also showed no activity with the rat osteocalcin VDRE reporter construct (data not shown). The additional observation that zebrafish RXR was able to boost transactivation by lampVDR (Fig. 5D) is consistent with the fact that three recently isolated lamprey RXR isoforms show slightly higher amino acid sequence identities to zebrafish RXR (72%, 73%, and 72%), compared with identities to human RXR of 70%, 70%, and 69%, respectively (Manzon, L. A., and J. H. Youson, unpublished data).
The evident difficulty in heterodimerization between lampVDR and mammalian RXRs may help explain the selectivity of lampVDR with respect to transactivation. The CYP3A4 DR3 (GGGTCAgcaAGTTCA) (taken from Ref. 21), which supported measurable transcriptional activation by lampVDR (Fig. 5, C and D), represents a nearly perfect high-affinity site for VDR-RXR binding, as determined by binding of hVDR-RXR heterodimers to random oligonucleotides (reported as PuGGTCAxxgPuGTTCA, where Pu is either A or G) (53, 54). Thus, a perfect VDRE may be able to compensate in some way for an imperfect heterodimeric partner, perhaps because of the very high affinity of VDR and RXR for their respective DR3 half-elements.
Another, more speculative explanation relates to the type of gene from which the VDRE was taken. Rat osteocalcin is a Gla-containing, calcium-binding protein expressed in bone that likely has no counterpart in the boneless lamprey. On the other hand, CYP3A4 is an enzyme whose function is believed to be detoxification of xenobiotics. Recent studies have shown that VDR can not only regulate human CYP3A4 in intestine (21, 41) but can also bind to certain toxic ligands such as the secondary bile acid, lithocholic acid, presumably serving as a sensor to activate detoxification of these compounds in mammals (55). This novel activity of VDR is similar to reported actions of PXR and CAR (56), perhaps reflecting the close evolutionary relationship of these receptors (Fig. 6). It has been reported further that a protein in C. elegans, nhr-8, whose amino acid sequence shows a similarity to the VDR/PXR/CAR grouping (57), plays a role in detoxification of colchicine and chloroquine in the nematode gut (58). This observation suggests that activation of detoxification pathways may be a very ancient function of group 1I receptors (58).
The present phylogenetic analyses confirm that PXRs and CARs are the closest relatives of VDR. Indeed, a standardized nuclear receptor nomenclature system places VDRs and PXRs/xONRs as the sister groups NR1I1 and NR1I2 (2, 59). Another striking similarity between VDRs and the PXR/ONR grouping exists as well because these receptors are the only known members of the superfamily to heterodimerize with RXR on a DR3-type responsive element (60, 61). The ability of VDRs and PXRs to preferentially bind DR3-type elements is perhaps reflected in the conservation of the amino acid residues in the DBD. A cocrystal of the human thyroid hormone receptor-ß and human RXR- DBDs on a thyroid hormone DNA response element (62) has revealed that residues just C-terminal to the zinc finger DNA-binding motifs in the thyroid hormone receptor make very significant DNA contacts. Further, of 10 such contacts, six involve positively charged residues. When the corresponding regions of VDRs and PXRs are examined (Fig. 7, center panel; see also Ref. 45), it can be seen that clusters of positively charged residues occur in the CTE of the DBD. The conservation of these positively charged clusters between VDRs (including lampVDR) and PXRs is striking and entirely consistent with the above-cited observation that these receptors, alone among known nuclear receptors, bind to DR3-type DNA elements (45, 60, 61, 62).
This high degree of relatedness suggests that VDRs (NR1I1) and the PXR/xONR group (NR1I2) diverged relatively recently in evolutionary history, presumably by means of gene duplication (reviewed in Ref. 63). The analysis in Fig. 6B suggests that such a duplication may have taken place before the divergence of vertebrates from nonvertebrate chordates, perhaps during one of two genome-wide duplications proposed to have transpired in early vertebrate evolution (63). This conclusion is based on the presence of separate VDR-like and PXR-like sequences in the pufferfish genome and also on the observation that related sequences from tunicate, fruitfly, and nematode seem to predate the divergence of VDR from PXRs. However, this conclusion must remain tentative because of the uncertainty of whether lampreys or other groups, such as cephalochordates, contain PXR-like sequences. Regarding CARs (NR1I3), the absence of any CAR-like sequences in fish suggests that CAR may be a tetrapod innovation. The exact timing of these divergences must await the analysis of sequences from other taxa. Nevertheless, the present analysis points to an origin of VDR at the dawn of vertebrates. What, then, might have been its ancestral function?
Studies of mice in which the VDR has been ablated support the traditional view that VDR, with its 1,25(OH)2D3 ligand, plays a crucial role in the formation of calcified structures (bones and teeth) by acting as a hypercalcemic hormone to promote both calcium and phosphate absorption from the intestine as well as calcium and phosphate retention at the kidney (4, 64). Other studies of VDR null mice have also revealed a role for VDR in skin, mainly in the mammalian hair cycle (5). Translation of these observations in mice to the physiology of lampreys is made difficult by the absence in the latter species of a calcified skeleton and teeth, as well as hair. Nevertheless, related functions may exist in lampreys, or in their ancestors, that may be regulated by VDR.
There are fossil groups of jawless fishes that possessed extensive dermal plates, thought to have contained a calcium phosphate derivative (18, 19). One interpretation (reviewed in Ref. 18) holds that lampreys are descendants of these armored agnathans and that extant lampreys have secondarily lost the ability to form these calcified structures. Two published observations would tend to support this view: an extant lamprey species, Ichthyomyzon unicuspis, has been reported to possess calcified cartilage in the head region (65), and living (but not dead) tissues from P. marinus adult specimens (head region) have been shown to be capable of calcifying the extracellular matrix when supplied with sufficient calcium and phosphate in organ culture (66).
If it is true that modern lampreys are descended from organisms that had calcified head plates, then the ability to form calcified structures (including, presumably, VDR action to mediate calcium homeostasis) may still be latent in lampreys but not actually expressed. This lack of VDR activity could be due to a lower level of lampVDR expression in key tissues such as intestine and kidney or an inactivation of lampVDR at key times in development. It is conceivable that both of these mechanisms might be at work in P. marinus because we have observed both a low level of intestinal expression of lampVDR in the juvenile as well as inactivating insertions in the larval protospleen. A more complete examination of larval juvenile and adult tissues will be required to fully examine this hypothesis.
An alternative interpretation of vertebrate evolution (reviewed in Ref. 19) is that lampreys diverged from other vertebrates before the advent of calcified tissues. If this view is correct, then the occurrence of VDR in a lamprey is evidence that VDR phylogenetically predates calcified structures. This does not totally preclude a role for lampVDR in calcium homeostasis because calcium levels might require regulation for other physiologic reasons. However, a role for VDR in skin differentiation is well established (5), and lampreys, although lacking hair, do have specialized skin structures such as keratinized teeth and mucous glands (67). Whether VDR regulates the development or maintenance of either teeth or mucous glands in lampreys is unknown, although the (qualitatively) high expression of normally spliced lampVDR mRNA in juvenile skin and mouth tissue is consistent with such a possibility.
As introduced above, another candidate function for lampVDR is as a sensor for endogenous or exogenous toxins and an inducer of CYP enzymes to detoxify them. An implication of this idea for VDR is that the toxins themselves, rather than (or in addition to) 1,25(OH)2D3, are ligands for VDR. Indeed, human and rodent VDRs have recently been shown to bind to and be activated by toxic bile salt derivatives, including lithocholic acid and 3-keto lithocholic acid (55). Whether lampVDR binds to the lamprey versions of bile acids or their derivatives has not yet been tested.
Yet a third possibility for a VDR function in lampreys is suggested by many studies documenting a role of the 1,25(OH)2D3-VDR system in the regulation of immune function. In mammals, VDR is involved in differentiation and regulation of certain immune cell types, including T-cells (see reviews in Refs. 68, 69). Although lampreys appear to lack the defining characteristics of the adaptive or combinatorial immune system in higher vertebrates, namely immunoglobulins, T cell receptors, or recombination activator genes (70), larval lampreys do possess cells in either protospleen (71) or gut epithelium (72, 73) that not only morphogenetically resemble lymphocytes but, in the case of the gut-associated cells, also express genes that are known to be crucial for lymphocyte differentiation in higher vertebrates (73, 74, 75). Our attempts to isolate lampVDR from juvenile intestine and larval peripheral blood lymphocytes have been either equivocal (intestine) or consistently negative (peripheral blood lymphocytes, data not shown). Also, as noted above, larval intestine yielded lampVDR transcripts that appear to contain inactivating insertions. Nevertheless, it might be worthwhile to examine more thoroughly the possibility of lampVDR expression in larval and/or juvenile lymphocyte-related cells, or in the adult fat column, in which lymphoid-like cells have been shown to reside (76).
In summary, the modern actions of VDR in terrestrial animals relating to a mineralized skeleton for locomotion and hair growth for protection from the sun are clearly not relevant in the sea lamprey. Instead, it must be hypothesized that VDR plays a different and, perhaps, more fundamental role in this ancient vertebrate. A full appreciation of the evolution of the VDR/PXR/CAR nuclear receptor grouping will require further studies in a variety of vertebrates with an ancient lineage as well as in key organisms such as protochordates and echinoderms. The presence of the related nhr-8 receptor in C. elegans, as noted above (57, 58), indicates that the evolutionary history of the VDR/PXR/CAR grouping could be very ancient indeed.
|
Acknowledgments |
|---|
expression plasmid, and J. B. Whitfield for helpful discussions regarding the phylogenetic analysis. The novel nucleotide sequences presented in this manuscript have been submitted to GenBank under the following accession numbers: lampVDR, AY249863; geVDR, AY254096; tdVDR, AY268062. |
Footnotes |
|---|
Abbreviations: CAR, Constitutive androstane receptor; CTE, C-terminal extension; cTR, chicken thyroid receptor-; CYP, cytochrome P450-containing enzyme; DBD, DNA-binding domain; 1,25(OH)2D3, 1,25-dihydroxyvitamin D3; hRAR, human retinoic acid receptor; hVDR, human vitamin D receptor; lampVDR, lamprey vitamin D receptor; ONR, orphan nuclear receptor; PXR, pregnane X receptor; RACE, rapid amplification of cDNA ends; RXR, retinoid X receptor; SSPE, sodium chloride/sodium phosphate/EDTA; TFIIB, transcription factor IIB; UTR, untranslated region; VDR, vitamin D receptor; VDRE, vitamin D responsive element; xONR, Xenopus ONR.
Received October 23, 2002.
Accepted for publication February 3, 2003.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, 25-dihydroxyvitamin D in biological fluids. Methods Enzymol 67:522–528[Medline]
