Cloning and Expression of a cDNA for a Mammalian Type III Iodothyronine Deiodinase
1995; Elsevier BV; Volume: 270; Issue: 28 Linguagem: Inglês
10.1074/jbc.270.28.16569
ISSN1083-351X
AutoresWalburga Croteau, Susan L. Whittemore, Mark J. Schneider, Donald L. St. Germain,
Tópico(s)Vitamin D Research Studies
ResumoThe type III iodothyronine deiodinase metabolizes the active thyroid hormones thyroxine and 3,5,3’-triiodothyronine to inactive compounds. Recently, we have characterized a Xenopus laevis cDNA (XL-15) that encodes a selenoprotein with type III deiodinase activity (St. Germain, D. L., Schwartzman, R., Croteau, W., Kanamori, A., Wang, Z., Brown, D. D., and Galton, V. A.(1994) Proc. Natl. Acad. Sci. U. S. A. 91, 7767-7771). Using the XL-15 as a probe, we screened a rat neonatal skin cDNA library. Among the clones isolated was one (rNS43-1) which contained a 2.1-kilobase pair cDNA insert that manifested significant homology to both the XL-15 and the G21 rat type I deiodinase cDNAs, including the presence of an in-frame TGA codon. Expression studies demonstrated that the rNS43-1 cDNA encodes a protein with 5-, but not 5’-, deiodinase activity that is resistant to inhibition by propylthiouracil and aurothioglucose. Northern analysis demonstrated a pattern of tissue expression in the rat consistent with that of the type III deiodinase and site directed mutagenesis confirmed that the TGA triplet codes for selenocysteine. We conclude that the rNS43-1 cDNA encodes the rat type III deiodinase and that the types I and III deiodinases present in amphibians and mammals constitute a family of conserved selenoproteins important in the metabolism of thyroid hormones. The type III iodothyronine deiodinase metabolizes the active thyroid hormones thyroxine and 3,5,3’-triiodothyronine to inactive compounds. Recently, we have characterized a Xenopus laevis cDNA (XL-15) that encodes a selenoprotein with type III deiodinase activity (St. Germain, D. L., Schwartzman, R., Croteau, W., Kanamori, A., Wang, Z., Brown, D. D., and Galton, V. A.(1994) Proc. Natl. Acad. Sci. U. S. A. 91, 7767-7771). Using the XL-15 as a probe, we screened a rat neonatal skin cDNA library. Among the clones isolated was one (rNS43-1) which contained a 2.1-kilobase pair cDNA insert that manifested significant homology to both the XL-15 and the G21 rat type I deiodinase cDNAs, including the presence of an in-frame TGA codon. Expression studies demonstrated that the rNS43-1 cDNA encodes a protein with 5-, but not 5’-, deiodinase activity that is resistant to inhibition by propylthiouracil and aurothioglucose. Northern analysis demonstrated a pattern of tissue expression in the rat consistent with that of the type III deiodinase and site directed mutagenesis confirmed that the TGA triplet codes for selenocysteine. We conclude that the rNS43-1 cDNA encodes the rat type III deiodinase and that the types I and III deiodinases present in amphibians and mammals constitute a family of conserved selenoproteins important in the metabolism of thyroid hormones. The varying patterns of thyroid hormone metabolism occurring in different tissues and during different stages of development are dictated in large part by the selective expression of one or more of several iodothyronine deiodinases. During fetal development in many species, including mammals, the predominant metabolic process involves the catalytic removal of iodine at the 5-, or the chemically equivalent 3-, position on the tyrosyl, or inner, ring of thyroxine (T4)1 1The abbreviations used are: T4thyroxineT33,5,3’-triiodothyronineT23,3’-diiodothyroninerT33,3’,5’-triiodothyronine5-D5-deiodinase5’-D5’-deiodinasePCRpolymerase chain reactionkbkilobase pair(s). and 3,5,3’-triiodothyronine (T3). This 5-deiodinase (5-D) activity results in the formation of the essentially inactive metabolites 3,3’,5’-triiodothyronine (reverse T3, rT3) and 3,3’-diiodothyronine (T2), respectively(1Germain St., D.L. Wu S.-Y. Visser T.J. Thyroid Hormone Metabolism: Molecular Biology and Alternative Pathways. CRC Press, Ann Arbor, MI1994: 45-66Google Scholar). The type III deiodinase, which functions exclusively as a 5-D, is highly expressed in both mammalian placenta and several fetal tissues and thus likely contributes significantly to this metabolic process(2Visser T.J. Schoenmakers C.H.H. Acta Med. Austriaca. 1992; 19: 18-21PubMed Google Scholar). After birth, the expression of type III 5-D is more limited and in rats occurs primarily in the central nervous system and skin, with particularly high levels noted in the latter tissue during the neonatal period(3Huang T. Chopra I.J. Beredo A. Solomon D.H. Chua Teco G.N. Endocrinology. 1985; 117: 2106-2113Crossref PubMed Scopus (82) Google Scholar). In contrast, the type I deiodinase is highly expressed in the liver, kidney, and thyroid gland of adult rats and can function as both a 5’-deiodinase (5’-D) and a 5-D. However, it efficiently catalyzes the 5-deiodination of sulfated iodothyronine substrates only (4Moreno M. Berry M.J. Horst C. Thomas R. Goglia F. Harney J.W. Larsen P.R. Visser T.J. FEBS Lett. 1994; 344: 143-146Crossref PubMed Scopus (60) Google Scholar) and its expression is low in most tissues during development(5Wu S. Fisher D.A. Polk D. Chopra I.J. Wu S. Thyroid Hormone Metabolism, Regulation and Clinical Implications. Blackwell Scientific Publications, Boston1991: 293-320Google Scholar). thyroxine 3,5,3’-triiodothyronine 3,3’-diiodothyronine 3,3’,5’-triiodothyronine 5-deiodinase 5’-deiodinase polymerase chain reaction kilobase pair(s). We have recently reported the characterization of the XL-15 cDNA(6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar), which was isolated by Wang and Brown from a Xenopus laevis tadpole tail cDNA library (7Wang Z. Brown D.D. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 11505-11509Crossref PubMed Scopus (297) Google Scholar) and encodes a type III 5-D. Of note, this amphibian cDNA, like the previously isolated mammalian type I deiodinase cDNAs(8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar, 9Mandel S.J. Berry M.J. Kieffer J.D. Harney J.W. Warne R.L. Larsen P.R. J. Clin. Endocrinol. Metab. 1992; 75: 1133-1139Crossref PubMed Google Scholar, 10Toyoda N. Harney J.W. Berry M.J. Larsen P.R. J. Biol. Chem. 1994; 269: 20329-20334Abstract Full Text PDF PubMed Google Scholar), contains an in-frame TGA codon which encodes selenocysteine at the enzyme's active site. The structural features of the mammalian type III deiodinase are uncertain, and several lines of evidence have suggested that this subclass of enzymes does not contain selenocysteine. Thus, the rat type III deiodinase is resistant to inhibition by propylthiouracil and gold compounds such as aurothioglucose(11Santini F. Chopra I.J. Hurd R.E. Solomon D.H. Chua Teco G.N. Endocrinology. 1992; 130: 2325-2332PubMed Google Scholar), properties which Berry et al.(12Berry M.J. Kieffer J.D. Harney J.W. Larsen P.R. J. Biol. Chem. 1991; 266: 14155-14158Abstract Full Text PDF PubMed Google Scholar) have suggested are consistent with the presence of cysteine at the enzyme's catalytic site. In addition, type III activity is maintained at essentially normal levels in the brain(13Meinhold H. Campos-Barros A. Walzog B. Köhler R. Müller F. Behne D. Exp. Clin. Endocrinol. 1993; 101: 87-93Crossref PubMed Scopus (77) Google Scholar), placenta(14Chanoine J. Alex S. Stone S. Fang S.L. Veronikis I. Leonard J.L. Braverman L.E. Pediatr. Res. 1993; 34: 288-292Crossref PubMed Scopus (30) Google Scholar), and skin (15Veronikis I.E. Alex S. Emerson C.H. Fang S.L. Wright G. Braverman L.E. in 68th Meeting of the American Thyroid Association. 1994; (Chicago, IL, Abstr. 69)Google Scholar) of selenium-deficient rats, whereas type I deiodinase activity in the liver and kidney is markedly decreased. And finally, protein labeling studies using 75Se have failed to identify candidate selenoproteins in rat tissues manifesting type III activity (11Santini F. Chopra I.J. Hurd R.E. Solomon D.H. Chua Teco G.N. Endocrinology. 1992; 130: 2325-2332PubMed Google Scholar). In the present report we describe the isolation from a rat neonatal skin cDNA library of a cDNA that manifests significant sequence homology to the mammalian type I and amphibian type III deiodinase cDNAs, including the presence of an in-frame TGA codon. X. laevis oocytes injected with RNA transcripts derived from this cDNA express abundant 5-D activity with characteristics of the type III enzyme. For the preparation of RNA, skin samples from 7-14-day-old neonatal rats and their corresponding dams were harvested and homogenized as described previously(16Germain St., D.L. Morganelli C.M. J. Biol. Chem. 1989; 264: 3054-3056Abstract Full Text PDF PubMed Google Scholar). Poly(A)+ RNA was isolated by two cycles of chromatography on oligo(dT)-cellulose and used for injection into X. laevis oocytes and cDNA library construction. RNA from other tissues used for Northern analysis was prepared by the same methods. In one experiment, rats were rendered hyperthyroid by the injection of 50 μg T3 subcutaneously/100 g body weight for 4 days prior to sacrifice. Poly(A)+ RNA from neonatal rat skin was used with oligo(dT) primers to prepare double-stranded cDNA using reagents from the Choice kit (Life Technologies, Inc.). Following the addition of EcoRI linkers, this material was used to construct a cDNA library in the Lambda-Zap II vector according to the manufacturer's instructions (Stratagene, La Jolla, CA). Screening of the library was performed using plaque hybridization under low stringency conditions according to the methods of Lees et al.(17Lees J.A. Saito M. Vidal M. Valentine M. Look T. Harlow E. Dyson N. Helin K. Mol. Cell. Biol. 1993; 13: 7813-7825Crossref PubMed Scopus (380) Google Scholar). The first 732 nucleotides of the XL-15 cDNA, which includes most of the coding region, was used as a probe. Positive plaques were detected by autoradiography and purified by additional rounds of screening using the same hybridization conditions. cDNA inserts were sequenced on both strands using gene-specific primers and an automated sequencing system with fluorescent dye terminators (Applied Biosystems, Foster City, CA). Stage 5-6 X. laevis oocytes were isolated and each microinjected as described previously (16Germain St., D.L. Morganelli C.M. J. Biol. Chem. 1989; 264: 3054-3056Abstract Full Text PDF PubMed Google Scholar) with 5-50 ng of either poly(A)+ RNA or 50 ng of in vitro synthesized capped RNA transcripts prepared using the MEGAscript kit (Ambion, Austin, TX). After injection, oocytes were incubated for 4 days in Barth's medium (for determination of 5-D activity) or L-15 medium (for determination of 5’-D activity), then harvested, membrane fractions prepared as described previously (18Sharifi J. Germain St., D.L. J. Biol. Chem. 1992; 267: 12539-12544Abstract Full Text PDF PubMed Google Scholar) and deiodinase activity determined according to published methods(18Sharifi J. Germain St., D.L. J. Biol. Chem. 1992; 267: 12539-12544Abstract Full Text PDF PubMed Google Scholar, 19Germain St., D.L. Croteau W. Mol. Endocrinol. 1989; 3: 2049-2053Crossref PubMed Scopus (12) Google Scholar). In kinetic studies, 5-D activity was determined using 1-20 nM [125I]T3 and 50 mM dithiothreitol as cofactor. Kinetic constants were determined from double reciprocal plots. 125I-Labeled iodothyronines used as substrates were obtained from E. I. Du Pont de Nemours & Co. and purified by chromatography using Sephadex LH-20 (Sigma) prior to use. In other experiments, the deiodinase activities in oocyte membrane preparations were determined in the absence or presence of propylthiouracil (10-1000 μM) or aurothioglucose (0.01-10 μM). 5’-D activity was measured using 67 nM [125I]rT3 as substrate and 20 mM dithiothreitol as cofactor. The 5-D assay utilized 1 nM [125I]T3 and 20 or 50 mM dithiothreitol. Protein concentrations were determined by the method of Bradford (20Bradford M.M. Anal. Biochem. 1976; 72: 248-254Crossref PubMed Scopus (217544) Google Scholar) with reagents obtained from Bio-Rad. The G21 cDNA was kindly provided by Drs. M. Berry and P. R. Larsen (Boston, MA). The reverse transcriptase-polymerase chain reaction (PCR) was performed as described previously(21Davey J.C. Schneider M.J. Galton V.A. Dev. Genet. 1994; 15: 339-346Crossref PubMed Scopus (21) Google Scholar). In brief, 0.5 μg of poly(A)+ RNA from rat neonatal skin or placenta was digested with 0.1 unit of RNase-free DNase I (Life Technologies, Inc.), then reverse-transcribed using Superscript™ II reverse transcriptase (Life Technologies, Inc.) and the rNS43-1-derived antisense oligonucleotide designated 1148AS (5’-GTCACTTGTCCCTTGGTTTT-3’) as primer. Following treatment with RNase H (Life Technologies, Inc.), 5 μl of the reaction mixture (total volume, 30 μl) was used directly in PCR. PCR was performed using the hot start method with five pairs of rNS43-1-derived oligonucleotides as primers used in different reactions. In all cases, 668AS (5’-GATGCGCTGGCTCTGGAA-3’) was used as the antisense primer. Sense primers included Bluescript/M13 reverse, 133S (5’-GGGCGGCTGTGTGAGT-3’), 226S (5’-CAGGGAGACCAGAAAGCAGAG-3’), 262S (5’-GGTCGGAGAAGGTGAAGGG-3’), and 319S (5’-CTCCCTGCTGCTTCACTCT-3’). Reaction conditions included 32 cycles of 94°C × 1 min, 55°C × 45 s, and 72°C × 1 min and a final 10-min extension period. Reaction products were separated on a 3% agarose gel and stained with ethidium bromide. The PCR reaction products separated by agarose gel electrophoresis were transferred by capillary blotting to nylon membranes (MagnaCharge, MSI, Inc., Westboro, MA) and probed using the 32P-radiolabeled rNS43-1-derived 430AS oligonucleotide (5’-GGAAATGCTTGCGGATG-3’). Following washing, membranes were autoradiographed for 10-60 min. Northern analysis was performed using 5-10 μg of poly(A)+ RNA/sample as described previously (22O'Mara B.A. Dittrich W. Lauterio T.J. Germain St., D.L. Endocrinology. 1993; 133: 1715-1723Crossref PubMed Scopus (88) Google Scholar), except that hybridization was performed at 42°C and the blots were subsequently washed twice at room temperature in 2 × SSPE (1 × SSPE: 0.15 M NaCl, 10 mM NaH2PO4, 1 nM EDTA, pH 7.4), 0.1% SDS × 10 min; once at room temperature in 0.1 x SSPE, 0.1% SDS × 10 min; then once at 42°C in 0.1 x SSPE, 0.1% SDS × 60 min prior to autoradiography. In some experiments, blots were also washed an additional time at 60°C in 0.1 x SSPE, 0.1% SDS. Site-directed mutagenesis was performed using the Altered Sites mutagenesis system according to the manufacturer's instructions (Promega). All mutations were confirmed by DNA sequence analysis using the methods described above. For each construct, oocytes were injected with 50 ng of capped RNA transcripts synthesized in vitro. Four days later 5-D activity was determined as described above. Activity was compared with that obtained in uninjected oocytes and those injected with an equivalent amount of RNA synthesized from the wild type cDNA in the pAlter vector (Promega). The choice of tissue for constructing a cDNA library was based on expression studies in X. laevis oocytes (Fig. 1). Injection into oocytes of 5 ng of poly(A)+ RNA isolated from neonatal rat skin or rat placenta induced significant amounts of 5-D activity which manifested a Km (using T3 as substrate) of ~3 nM, consistent with the properties of a type III deiodinase(1Germain St., D.L. Wu S.-Y. Visser T.J. Thyroid Hormone Metabolism: Molecular Biology and Alternative Pathways. CRC Press, Ann Arbor, MI1994: 45-66Google Scholar, 3Huang T. Chopra I.J. Beredo A. Solomon D.H. Chua Teco G.N. Endocrinology. 1985; 117: 2106-2113Crossref PubMed Scopus (82) Google Scholar). In contrast, injection of 50 ng/oocyte of poly(A)+ RNA isolated from the skin of lactating dams induced little or no 5-D activity, a finding in agreement with the low levels of activity previously noted in homogenates prepared from this tissue(3Huang T. Chopra I.J. Beredo A. Solomon D.H. Chua Teco G.N. Endocrinology. 1985; 117: 2106-2113Crossref PubMed Scopus (82) Google Scholar). Screening of a neonatal skin library by plaque hybridization under low stringency conditions with the coding region of the XL-15 amphibian type III deiodinase cDNA as a probe resulted in the identification and isolation of seven different clones. Partial DNA sequencing of several of the cDNA inserts from these clones revealed areas that had significant homology to the XL-15 cDNA. Two clones designated rNS27-1 and rNS43-1, containing 1.5- and 2.1-kb inserts, respectively, were selected for detailed expression studies and complete sequencing. Functional characterization of the protein products of the isolated cDNA was carried out using the X. laevis oocyte translational assay system. Oocytes injected with RNA synthesized in vitro using clone rNS43-1 as template expressed high levels of 5-D activity as determined by the conversion of [125I]T3 to [125I]T2. Such activity was dependent on the presence of relatively high concentrations of dithiothreitol (Fig. 2), requiring 50 mM dithiothreitol for maximal activity. In contrast, type I 5’-D activity induced by the injection of RNA derived from the G21 cDNA was near-maximal with only 0.1 mM dithiothreitol in the reaction mixture. Unlike the 5’-D activity of the type I deiodinase(18Sharifi J. Germain St., D.L. J. Biol. Chem. 1992; 267: 12539-12544Abstract Full Text PDF PubMed Google Scholar), glutathione at concentrations of 5 and 50 mM did not support 5-D activity in oocyte membranes injected with rNS43-1-derived RNA (data not shown). Using 50 mM dithiothreitol as cofactor, kinetic analysis of the rNS43-1-induced 5-D activity revealed a Km value for T3 of 1 nM. Using rT3 as substrate and 20 mM dithiothreitol as cofactor, no 5’-D activity was detected in rNS43-1-injected oocyte membrane fractions. Injection into oocytes of RNA derived from clone rNS27-1, which lacks sequences in the 5’ region of the open reading frame of rNS43-1 (see below), did not induce 5-D activity (data not shown). A comparison was made of the sensitivity of the rNS43-1 5-D and the G21 type I 5’-D to inhibition by propylthiouracil and aurothioglucose. As expected(8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar), the G21-induced 5’-D activity was highly sensitive to inhibition by propylthiouracil with a concentration of 100 μM inhibiting activity by essentially 100% (Fig. 3A). In contrast, the induced rNS43-1 5-D activity was resistant to high concentrations of propylthiouracil. Similarly, the G21 deiodinase was at least 100-fold more sensitive than the rNS43-1 deiodinase to the inhibitory effects of aurothioglucose; 50% inhibition of G21-induced type I activity occurred at approximately 0.01 μM aurothioglucose, whereas 1 μM was required to obtain the same level of inhibition of the rNS43-1-induced activity (Fig. 3B). The studies depicted in Fig. 3 were performed using different concentrations of dithiothreitol in the 5’-D and 5-D assays. Because the sensitivity of these processes to propylthiouracil can be influenced by the concentration of cofactor(23Leonard J.L. Rosenberg I.N. Endocrinology. 1978; 103: 2137-2144Crossref PubMed Scopus (133) Google Scholar, 24Goswami A. Rosenberg I.N. Endocrinology. 1986; 119: 916-923Crossref PubMed Scopus (15) Google Scholar), we determined in other experiments that at equivalent concentrations of dithiothreitol (20 mM), the marked disparity in the inhibitory effects of propylthiouracil on the G21 and rNS43-1 deiodinases persist (data not shown). A diagram of the rNS27-1 and rNS43-1 cDNAs is shown in Fig. 4. The sequence of the rNS27-1 cDNA is identical to nucleotides 549-2081 of the rNS43-1 cDNA. The latter cDNA contains an open reading frame of 834 nucleotides encoding a protein of 278 amino acids with a predicted molecular mass of 31.6 kDa. Of note is an in-frame TGA codon, presumably coding for selenocysteine, beginning at nucleotide 740 and located in a region of high homology to the XL-15 protein and the mammalian type I deiodinases (see below). A search of GenBank™ for homologous nucleotide sequences identified the XL-15 cDNA, previously isolated mammalian type I deiodinase cDNAs (from rat, human, dog), and a 283-nucleotide partial cDNA isolated from a human skeletal muscle library (accession number Z19511) which shows 80% sequence identity to nucleotides 1-196 of the rNS43-1 cDNA. The identity of this human cDNA has not been established. Because of the unexpected homology of a portion of the rNS43-1 cDNA to a human skeletal muscle cDNA, and the finding that previously isolated cDNAs for other deiodinases have relatively short 5’-untranslated regions (6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 7Wang Z. Brown D.D. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 11505-11509Crossref PubMed Scopus (297) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar, 9Mandel S.J. Berry M.J. Kieffer J.D. Harney J.W. Warne R.L. Larsen P.R. J. Clin. Endocrinol. Metab. 1992; 75: 1133-1139Crossref PubMed Google Scholar, 10Toyoda N. Harney J.W. Berry M.J. Larsen P.R. J. Biol. Chem. 1994; 269: 20329-20334Abstract Full Text PDF PubMed Google Scholar, 11Santini F. Chopra I.J. Hurd R.E. Solomon D.H. Chua Teco G.N. Endocrinology. 1992; 130: 2325-2332PubMed Google Scholar, 12Berry M.J. Kieffer J.D. Harney J.W. Larsen P.R. J. Biol. Chem. 1991; 266: 14155-14158Abstract Full Text PDF PubMed Google Scholar, 13Meinhold H. Campos-Barros A. Walzog B. Köhler R. Müller F. Behne D. Exp. Clin. Endocrinol. 1993; 101: 87-93Crossref PubMed Scopus (77) Google Scholar, 14Chanoine J. Alex S. Stone S. Fang S.L. Veronikis I. Leonard J.L. Braverman L.E. Pediatr. Res. 1993; 34: 288-292Crossref PubMed Scopus (30) Google Scholar, 15Veronikis I.E. Alex S. Emerson C.H. Fang S.L. Wright G. Braverman L.E. in 68th Meeting of the American Thyroid Association. 1994; (Chicago, IL, Abstr. 69)Google Scholar, 16Germain St., D.L. Morganelli C.M. J. Biol. Chem. 1989; 264: 3054-3056Abstract Full Text PDF PubMed Google Scholar, 17Lees J.A. Saito M. Vidal M. Valentine M. Look T. Harlow E. Dyson N. Helin K. Mol. Cell. Biol. 1993; 13: 7813-7825Crossref PubMed Scopus (380) Google Scholar, 18Sharifi J. Germain St., D.L. J. Biol. Chem. 1992; 267: 12539-12544Abstract Full Text PDF PubMed Google Scholar, 19Germain St., D.L. Croteau W. Mol. Endocrinol. 1989; 3: 2049-2053Crossref PubMed Scopus (12) Google Scholar, 20Bradford M.M. Anal. Biochem. 1976; 72: 248-254Crossref PubMed Scopus (217544) Google Scholar, 21Davey J.C. Schneider M.J. Galton V.A. Dev. Genet. 1994; 15: 339-346Crossref PubMed Scopus (21) Google Scholar, 22O'Mara B.A. Dittrich W. Lauterio T.J. Germain St., D.L. Endocrinology. 1993; 133: 1715-1723Crossref PubMed Scopus (88) Google Scholar, 23Leonard J.L. Rosenberg I.N. Endocrinology. 1978; 103: 2137-2144Crossref PubMed Scopus (133) Google Scholar, 24Goswami A. Rosenberg I.N. Endocrinology. 1986; 119: 916-923Crossref PubMed Scopus (15) Google Scholar nucleotides), reverse transcriptase-PCR was conducted to determine the extent to which sequences in the 5’ region of the rNS43-1 cDNA were derived from the corresponding rat deiodinase mRNA. The experimental approach is depicted in Fig. 5A. Reverse transcription was carried out using poly(A)+ RNA from rat neonatal skin or rat placenta and the rNS43-1-derived antisense oligonucleotide 1148AS as a primer. The resulting cDNA was then used in the PCR with a number of different primer pairs derived from the rNS43-1 cDNA sequence (see “Materials and Methods” and Fig. 5A). In addition, to detect any contamination with the rNS43-1 plasmid during the procedures, a primer pair which included the Bluescript/M13 reverse sequencing primer was also used. PCR products were analyzed by agarose gel electrophoresis and ethidium bromide staining followed by transfer of the products to nylon membranes and Southern blotting using the nested oligonucleotide 430AS as a probe. Expected amplicon sizes for each primer pair, based on the rNS43-1 sequence, were 882 (rev/668AS), 553 (133S/668AS), 460 (226S/668AS), 424 (262S/668AS), and 367 (319S/668AS) nucleotides. As shown in Fig. 5B, control PCR reactions using the rNS43-1 plasmid as template (lanes 7-11) demonstrated the expected size products for all primer combinations, thus confirming the usefulness of these primer pairs for amplifying the desired rNS43-1-related sequences. In contrast, only primer pairs 262S/668AS and 319S/668AS gave demonstrable products of the correct size by ethidium bromide staining (data not shown) and Southern blotting when cDNA derived from rat neonatal skin (lanes 1-5) or rat placenta (lanes 12-16) were used as templates in the PCR. No amplification was noted with the primer pair rev/668AS using either cDNA preparation (lanes 1 and 12, respectively). Nor was amplification noted when (a) reverse transcriptase (lanes 6, 17, and 18) or (b) RNA (lanes 19-21) were omitted from the initial cDNA synthesis step or when (c) PCR reaction tubes contained water in place of an aliquot of the cDNA synthesis mixture (data not shown). These results strongly suggest that sequences in the rNS43-1 cDNA insert located 5’ to oligonucleotide 262S are not present in the type III deiodinase mRNA from neonatal skin and placenta. At present, the identity of these sequences is uncertain; their presence in the rNS43-1 cDNA may represent a cloning artifact secondary to the inclusion of two unrelated cDNA inserts in this clone. The nucleotide and deduced amino acid sequence of the rNS43-1 cDNA are shown in Fig. 6. Uppercase letters represent the sequences proven for the type III deiodinase and start at the point corresponding to oligonucleotide 262S. The 261 nucleotides of unknown origin in the region 5’ to this are shown in lowercase letters. Northern analysis of rat tissues was performed using the rNS27-1 cDNA, which contains only deiodinase cDNA sequences, as a probe (Fig. 7). Using poly(A)+ RNA from neonatal skin and placenta, a single RNA species of 2.2 kb in size was noted. No hybridization was detected to RNA derived from liver, kidney, or the skin of postpartum (lactating) rats, all tissues which contain little or no type III deiodinase activity. In the cerebral cortex of euthyroid adult rats, a more complex hybridization pattern was noted with several faint bands visualized. Of note, the two larger RNA species of approximately 3.3 and 3.6 kb appeared more abundant after T3 administration to induce hyperthyroidism. Type III activity in this tissue was also increased with T3 administration (data not shown). In both skeletal muscle and cerebral cortex, a smaller band at 1.6 kb was faintly visible. These hybridization patterns did not significantly change when blots were washed at 60°C. Previous studies of the rat type I deiodinase and the XL-15 type III deiodinase have demonstrated that the presence of selenocysteine is critical for the catalytic activity of the protein(6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar). Thus, substitution of cysteine for selenocysteine in these proteins reduces expressed catalytic activity by approximately 76-90%, in spite of increased translational efficiency of the cysteine mutant(25Berry M.J. Maia A.L. Kieffer J.D. Harney J.W. Larsen P.R. Endocrinology. 1992; 131: 1848-1852Crossref PubMed Scopus (109) Google Scholar). To investigate the importance of selenocysteine to the functional activity of the rNS43-1-encoded deiodinase, site-directed mutagenesis of the TGA codon was performed. Conversion of this triplet to a TAA stop codon, or a TTA leucine codon, abolished expression in oocytes. The TGT cysteine mutant was also essentially inactive, manifesting less than 5% of wild type 5-D activity in four experiments. To ensure that this lack of activity of the cysteine mutant did not result from some other unplanned alteration in structure, the mutant cDNA was sequenced in its entirety and found to be otherwise identical to the rNS43-1. In addition, RNA derived from a second isolated cysteine mutant clone also exhibited a comparably low level of activity when expressed in X. laevis oocytes. A comparison of the deduced amino acid sequence of the rNS43-1 protein with that of other deiodinases (6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar) revealed 58% sequence identity and 73% similarity with the XL-15 type III deiodinase, but only 39 identity and 57% similarity to the rat G21 type I deiodinase. On alignment of the deiodinase protein sequences published to date(6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar, 9Mandel S.J. Berry M.J. Kieffer J.D. Harney J.W. Warne R.L. Larsen P.R. J. Clin. Endocrinol. Metab. 1992; 75: 1133-1139Crossref PubMed Google Scholar, 10Toyoda N. Harney J.W. Berry M.J. Larsen P.R. J. Biol. Chem. 1994; 269: 20329-20334Abstract Full Text PDF PubMed Google Scholar), several regions of conservation in the mid- and carboxyl-terminal portions of the molecules are evident (Fig. 8). High homology is noted in the regions surrounding the selenocysteine (designated “X” at amino acid 144 in r5DIII [rNS43-1]) and surrounding a histidine residue (amino acid 176) demonstrated by Berry et al.(26Berry M. J. Biol. Chem. 1992; 267: 18055-18059Abstract Full Text PDF PubMed Google Scholar) to be essential for type I deiodinase activity. Amino acid identity in these two regions is 83 and 75%, respectively. The structural and functional data presented herein, as well as the patterns of tissue expression as determined by Northern analysis, indicate that the rNS43-1 cDNA encodes the rat type III 5-D. Of paramount importance is the finding that this cDNA contains an in-frame TGA codon. Although we have not directly demonstrated that the rNS43-1-encoded enzyme is a selenoprotein, the mutagenesis studies, the strong amino acid conservation present in all the deiodinases in the region of this codon, and preliminary studies identifying a selenocysteine insertion element in the 3’-untranslated region of the cDNA2 2B. Moyer and D. L. St. Germain, unpublished observations. provide compelling evidence that the TGA encodes selenocysteine rather than a termination signal. Thus, the type I and type III deiodinases constitute a family of structurally related selenoproteins. To date, the type II 5’-D has not yet been purified, nor have cDNAs been isolated. The structural characteristics of the type II enzyme therefore remain uncertain. The finding that the rNS43-1 cDNA encodes a selenoprotein has important implications for the study of selenoproteins in general and deiodinases in particular. First, previous studies by Berry et al.(12Berry M.J. Kieffer J.D. Harney J.W. Larsen P.R. J. Biol. Chem. 1991; 266: 14155-14158Abstract Full Text PDF PubMed Google Scholar) have demonstrated that mutagenesis of the selenocysteine in the type I deiodinase to cysteine renders that enzyme insensitive to propylthiouracil and gold compounds. This finding suggested that the unique biochemical properties of selenocysteine confer sensitivity to these inhibitors. However, the present studies, as well as our previous demonstration that the XL-15 deiodinase is resistant to these agents (6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar), indicates that factors other than the presence of selenocysteine, such as those related to protein structure or kinetic mechanisms, must dictate insensitivity to propylthiouracil and aurothioglucose in the type III deiodinases. Second, it has previously been demonstrated that type III deiodinase activity in rat brain, placenta, and skin are largely preserved in the face of nutritional selenium deprivation severe enough to markedly lower type I protein and activity levels in the liver and kidney (13Meinhold H. Campos-Barros A. Walzog B. Köhler R. Müller F. Behne D. Exp. Clin. Endocrinol. 1993; 101: 87-93Crossref PubMed Scopus (77) Google Scholar, 14Chanoine J. Alex S. Stone S. Fang S.L. Veronikis I. Leonard J.L. Braverman L.E. Pediatr. Res. 1993; 34: 288-292Crossref PubMed Scopus (30) Google Scholar, 15Veronikis I.E. Alex S. Emerson C.H. Fang S.L. Wright G. Braverman L.E. in 68th Meeting of the American Thyroid Association. 1994; (Chicago, IL, Abstr. 69)Google Scholar). Our demonstration that the type III deiodinase in this species is a selenoprotein indicates that selenium stores in the former tissues are preserved in the face of moderate selenium deficiency. Although prior studies of selenium turnover have demonstrated that the brain, like the thyroid gland, testes, and other endocrine organs (27Behne D. Hilmert H. Scheid S. Gessner H. Elger W. Biochim. Biophys. Acta. 1988; 966: 12-21Crossref PubMed Scopus (399) Google Scholar, 28Buckman T.D. Sutphin M.S. Eckhert C.D. Biochim. Biophys. Acta. 1993; 1163: 176-184Crossref PubMed Scopus (29) Google Scholar, 29Chanoine J. Braverman L.E. Farwell A.P. Safran M. Alex S. Dubord S. Leonard J.L. J. Clin. Invest. 1993; 91: 2709-2713Crossref PubMed Scopus (141) Google Scholar), has a marked propensity to conserve selenium, the data presented herein are the first to suggest that placenta and skin share this property. This is of particular interest given that placental selenium stores must be accumulated de novo during pregnancy, a circumstance which presumably should predisposes this organ to selenium deficiency in the setting of nutritional selenium deprivation. Third, the inability of others using 75Se to label a candidate type III deiodinase protein of 32 kDa in appropriate tissues (e.g. brain, placenta) points out the insensitivity of these methods to identify selenoproteins(11Santini F. Chopra I.J. Hurd R.E. Solomon D.H. Chua Teco G.N. Endocrinology. 1992; 130: 2325-2332PubMed Google Scholar, 27Behne D. Hilmert H. Scheid S. Gessner H. Elger W. Biochim. Biophys. Acta. 1988; 966: 12-21Crossref PubMed Scopus (399) Google Scholar). This is not surprising given that little is known about the rates of entry of selenium into various tissues, selenium pool sizes, or the abundance and turnover rates of most selenoproteins. The importance of selenocysteine to the catalytic activity of the deiodinases is reaffirmed by the present studies. Indeed, unlike the type I enzymes and the XL-15 deiodinase which retain a significant, albeit substantially reduced (10-24%), level of activity when cysteine is substituted for selenocysteine(6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar), the cysteine mutant of the rNS43-1 cDNA is essentially inactive. Given that activity was determined using an in vivo expression system, it is uncertain whether this lack of activity reflects an intrinsically inactive protein or whether expression of the protein was impaired due to a decreased half-life of the injected RNA or the cysteine mutant protein. However, the latter two possibilities appear unlikely given that the cysteine mutants of other deiodinases induce significant levels of activity in X. laevis oocytes(6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar). Northern analysis revealed hybridization of the type III deiodinase probe to multiple species in different tissues. In neonatal skin and placenta only a single transcript was identified that was somewhat larger in size than the 1.8-kb size of the rNS43-1 type III deiodinase sequences. In the cerebral cortex, however, larger hybridizing species were observed, and a low intensity 1.6-kb band was noted in both cerebral cortex and skeletal muscle. Whether such species represent alternatively processed transcripts or products of a related gene remains uncertain. Of note, relatively high levels of 5-D activity have been measured in fetal rat skeletal muscle(30Huang T. Chopra I.J. Boado R. Solomon D.H. Chua Teco G.N. Pediatr. Res. 1988; 23: 196-199Crossref PubMed Scopus (52) Google Scholar). In the adult rat, however, levels in this tissue are only 2% of that found in cerebral cortex or skin(30Huang T. Chopra I.J. Boado R. Solomon D.H. Chua Teco G.N. Pediatr. Res. 1988; 23: 196-199Crossref PubMed Scopus (52) Google Scholar). A comparison of the deduced amino acid sequences of the five mammalian and amphibian deiodinase proteins reported to date (6Germain St., D.L. Schwartzman R. Croteau W. Kanamori A. Wang Z. Brown D.D. Galton V.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91 (Correction): 7767-7771Proc. Natl. Acad. Sci. 1994; 91 (Correction): 11282Crossref PubMed Scopus (122) Google Scholar, 8Berry M.J. Banu L. Larsen P.R. Nature. 1991; 349: 438-440Crossref PubMed Scopus (758) Google Scholar, 9Mandel S.J. Berry M.J. Kieffer J.D. Harney J.W. Warne R.L. Larsen P.R. J. Clin. Endocrinol. Metab. 1992; 75: 1133-1139Crossref PubMed Google Scholar, 10Toyoda N. Harney J.W. Berry M.J. Larsen P.R. J. Biol. Chem. 1994; 269: 20329-20334Abstract Full Text PDF PubMed Google Scholar) indicates that the rNS43-1 protein is more closely related to the amphibian XL-15 enzyme than to the mammalian type I deiodinases. This finding is consistent with the similar functional characteristics of the 5-D activity displayed by these two proteins. In contrast, the activity of the rNS43-1 deiodinase differs markedly from that of the mammalian type I enzymes in that it manifests no 5’-D activity and is able to catalyze efficiently the 5-deiodination of non-sulfated iodothyronine substrates. Although little primary sequence homology is present in the amino-terminal portions of the deiodinases (Fig. 8), all contain a region of marked hydrophobicity, which in the rNS43-1 protein encompasses the initial 42 amino acid residues. Toyoda et al.(31Toyoda N. Berry M. Larsen P.R. in 76th Annual Meeting of the Endocrine Society. 1994; (Anaheim, CA, Abstr. 4)Google Scholar) have provided evidence that the analogous hydrophobic region in the rat type I deiodinase serves to anchor the protein in the endoplasmic reticulum. Given that the type III deiodinases are also membrane proteins, such regions in the rNS43-1 and XL-15 proteins likely serve a similar function. In other studies, these investigators have recently demonstrated that mutation of a phenylalanine at position 65 in the human and rat type I deiodinases results in marked inefficiency of these enzymes in utilizing rT3 as a substrate for 5’-deiodination(10Toyoda N. Harney J.W. Berry M.J. Larsen P.R. J. Biol. Chem. 1994; 269: 20329-20334Abstract Full Text PDF PubMed Google Scholar), but does not affect the inner ring deiodination of appropriate sulfated substrates(4Moreno M. Berry M.J. Horst C. Thomas R. Goglia F. Harney J.W. Larsen P.R. Visser T.J. FEBS Lett. 1994; 344: 143-146Crossref PubMed Scopus (60) Google Scholar). The observation that the rNS43-1 and XL-15 deiodinases lack a phenylalanine at the corresponding locations provides further evidence that such a residue is not critical for 5-deiodination of T4 or T3. The ability of the mammalian type III deiodinase to convert T4 and T3 to metabolically inactive metabolites, as well as its patterns of expression and regulation in the placenta during fetal development, and in the adult brain, suggest that it plays a protective role in preventing exposure of tissues to excessive levels of active thyroid hormones(1Germain St., D.L. Wu S.-Y. Visser T.J. Thyroid Hormone Metabolism: Molecular Biology and Alternative Pathways. CRC Press, Ann Arbor, MI1994: 45-66Google Scholar). This concept is strengthened by the observations of Silva and Matthews (32Silva J.E. Matthews P.S. J. Clin. Invest. 1984; 74: 1035-1049Crossref PubMed Scopus (100) Google Scholar) that in the cerebral cortex of neonatal rats the residency time of T3, a parameter likely dependent in large part on the activity of the type III deiodinase, is critical to thyroid hormone homeostasis in this organ. The isolation of the rNS43-1 cDNA thus provides an important tool for the further study of iodothyronine metabolism. In particular, the development of animal models with targeted deletion or overexpression of the type III deiodinase gene will allow a direct assessment of its physiologic role in thyroid hormone metabolism and a better understanding of thyroid hormone action during development and in selected tissues such as the brain.
Referência(s)