Cloning of a cDNA for the type II iodothyronine deiodinase.

Three types of iodothyronine deiodinase have been identified in vertebrate tissues. cDNAs for the types I and III have been cloned and shown to contain an in-frame TGA that codes for selenocysteine at the active site of the enzyme. We now report the cloning of a cDNA for a type II deiodinase using a reverse transcription/polymerase chain reaction strategy and RNA obtained from Rana catesbeiana tissues. This cDNA (RC5′DII) manifests limited but significant homology with other deiodinase cDNAs and contains a conserved in-frame TGA codon. Injection of capped in vitro synthesized transcripts of the cDNA into Xenopuslaevis oocytes results in the induction of deiodinase activity with characteristics typical of a type II deiodinase. The levels of RC5′DII transcripts in R. catesbeiana tadpole tail and liver mRNA at stages XII and XXIII correspond well with that of type II deiodinase activity but not that of the type III activity in these tissues. These findings indicate that the amphibian type II 5′-deiodinase is a structurally unique member of the family of selenocysteine-containing deiodinases.

Three types of iodothyronine deiodinase have been identified in vertebrate tissues. cDNAs for the types I and III have been cloned and shown to contain an inframe TGA that codes for selenocysteine at the active site of the enzyme. We now report the cloning of a cDNA for a type II deiodinase using a reverse transcription/ polymerase chain reaction strategy and RNA obtained from Rana catesbeiana tissues. This cDNA (RC5DII) manifests limited but significant homology with other deiodinase cDNAs and contains a conserved in-frame TGA codon. Injection of capped in vitro synthesized transcripts of the cDNA into Xenopus laevis oocytes results in the induction of deiodinase activity with characteristics typical of a type II deiodinase. The levels of RC5DII transcripts in R. catesbeiana tadpole tail and liver mRNA at stages XII and XXIII correspond well with that of type II deiodinase activity but not that of the type III activity in these tissues. These findings indicate that the amphibian type II 5-deiodinase is a structurally unique member of the family of selenocysteine-containing deiodinases.
Intracellular concentrations of the thyroid hormones, T 4 1 and T 3 , are profoundly influenced by the activity of three iodothyronine deiodinases, classified as types I, II, and III (1). In mammals, the type I enzyme (5DI) catalyzes 5Ј-deiodination (5ЈD), the removal of iodine from the 5Ј (or 3Ј) positions of T 4 and its derivatives. The enzyme can also catalyze 5-deiodination (5D), the removal of an iodine located at either the 5 (or 3) positions of iodothyronines, but does so efficiently only with sulfated iodothyronine substrates (2). The type II enzyme (5DII) also catalyzes 5ЈD, but it is readily distinguished from the 5DI by its kinetics, substrate specificity, sensitivity to propylthiouracil (PTU) and aurothioglucose (AThG) (1,3), and response to thyroid status (1). The type III enzyme (5DIII) catalyzes primarily 5D activity (1), a process that results in derivatives with little or no thyromimetic activity (1).
The primary function of the types I and II deiodinases is to convert T 4 to its metabolically more active derivative, T 3 . However, the tissue distribution and physiological roles of the two enzymes are very different. The principal role of the 5DI in mammals is to provide a source of plasma T 3 by deiodination of T 4 in peripheral tissues such as liver and kidney. In contrast, the 5DII is responsible for the majority of the intracellular T 3 in tissues such as the pituitary, brain, and brown fat by mediating local deiodination of T 4 and is considered to be of major importance in regulating thyroid hormone action in these tissues (1,3). The 5DII also plays a major role during development. 5DII is the principal 5Ј-deiodinase expressed in the mammalian fetus, and it is notable that 5DII activity in brain peaks in the neonatal period, the time that is critical for thyroid hormone-dependent development in this tissue (4). Moreover, 5DII is the only 5Ј-deiodinase present in the developing frog in which the orderly progression of developmental processes is dependent on the ability to attain appropriate intracellular levels of T 3 (5). Thus, the 5DII appears to play an essential role in intracellular T3 production in those circumstances where thyroid hormone-dependent processes take on critical significance.
cDNAs for the type I enzyme of rat (6), dog (7), and human (8) have been cloned. These cDNAs contain an in-frame TGA coding for selenocysteine, which is necessary for maximal enzyme activity (6). Three cDNAs for the type III enzyme have also been cloned; we have shown that XL-15, a cDNA isolated by Wang and Brown (9) from a Xenopus laevis tadpole tail cDNA library, encodes a 5DIII (10) and, using XL-15 as a probe, we have isolated 5DIII cDNAs for Rana catesbeiana (11) and rat (12). These cDNAs exhibit significant sequence homology to the mammalian 5DI cDNAs including the in-frame TGA codon, which codes for selenocysteine.
Isolation of a cDNA for a 5DII has yet to be reported. To this end we predicted that this enzyme would share significant sequence homology with other deiodinases. Close examination of the sequences of the known 5DI and 5DIII cDNAs revealed that, although the overall similarity between the two types is relatively low, there are three limited regions that are highly conserved. One is near the TGA codon that codes for selenocysteine, and the other two are approximately 60 and 230 bp 3Ј of this codon. We hypothesized that these regions would also be conserved in the 5DII gene.
This hypothesis proved to be correct. In the present report we describe the cloning of a cDNA for the 5DII of R. catesbeiana using a reverse transcription/polymerase chain reaction (RT/ PCR) strategy, oligonucleotide primers based on the sequences of these conserved regions, and RNA from R. catesbeiana tissues that contain relatively high levels of type II 5ЈD activity. Once a portion of the putative coding region of the 5DII cDNA was obtained, gene-specific primers were used to synthesize the 3Ј-and 5Ј-ends of the cDNA using rapid amplification of cDNA ends (RACE) procedures (13). The resulting cDNA (RC5ЈDII) contains the conserved TGA codon and codes for a protein with characteristics typical of a 5DII. * This work was supported by the following grants from the National Institutes of Health: HD 09020 and HD 27706 (to V. A. G.), DK 07508 (to K. B. B.), and DK 42271 (to D. L. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) L42815.

Animals-R. catesbeiana tadpoles, stages XX-XXIV (Taylor and
Kollros (14)), were obtained from Charles D. Sullivan, Inc., Nashville, TN. Maintenance of tadpoles and preparation of total and poly(A) ϩ RNA from their tissues were carried out as described previously (15).
Synthesis of a Partial cDNA for the 5DII from R. catesbeiana Tissues-The sequences of the primers used in the synthesis of the cDNA for the 5DII of R. catesbeiana are shown in Table I. Lettered primers (A, B, C, and D) are based on the conserved sequences of the 5DI and 5DIII cDNAs; numbered primers (1-6) are specific to the RC5ЈDII cDNA. Their locations relative to each other in RC5ЈDII are shown in Fig. 1A. A schematic representation of the RT/PCR-based cloning strategy is shown in Fig. 1B. Total RNA obtained from the hindlimb of tadpoles was reverse transcribed using primer C, and the resulting cDNA was amplified by PCR using primers A and C. The following cycling conditions were used in this and all subsequent PCR reactions: 94°C for 45 s; 52°C for 45 s; 72°C for 60 s, with a 2-s extension at 72°C for each of 30 cycles. This was followed by a 10-min extension at 72°C. The 110-bp product obtained was reamplified with primers A and B to yield a 98-bp product whose sequence, excluding the primers, exhibited 46% identity with the corresponding sequence in the R. catesbeiana 5DIII cDNA (RC5D). Then skin RNA from stage XXIII tadpoles was reverse transcribed using the oligo(dT) adapter primer from the 3Ј-RACE kit (Life Technologies, Inc.), and the resulting cDNA was amplified using primers 1 and D to yield a 238-bp product, whose sequence was 48% identical to that of RC5D. This sequence information revealed that the 32 nucleotides in the "conserved" region used to design primers B and C are 84% identical to those in RC5D.
Synthesis of the 3Ј-and 5Ј-Ends of RC5ЈDII Using the RACE Procedure-The 3Ј-and 5Ј-RACE procedures were carried out as recommended by the manufacturer (Life Technologies, Inc.) with one exception. Since in our experience the universal amplification primer (UAP), which is used in both procedures, routinely makes multiple products, a modified UAP (UAP*) was designed. The 12 nucleotides at the 5Ј-end, which are only necessary if the kit's cloning procedure is employed, were eliminated and, with the aid of a computer program (OLIGO TM 4.0, National Biosciences, Inc., Plymouth, MN), the remaining UAP sequence, 5Ј-GGCCACGCGTCGACTAGTAC-3Ј, was modified to 5Ј-GTCCACGCATCGACTAGTA-3Ј. The 3Ј-end of RC5ЈDII was synthesized using the cDNA obtained from skin RNA (see above) as template. Two rounds of PCR, the first with primers 2 and UAP* and the second with nested primer 3 and UAP*, yielded a product of about 1000 bp. The overlapping 5Ј-end was made from skin RNA, which was reverse transcribed with primer 6, and the resulting cDNA was subjected to two rounds of PCR, the first with primers UAP* and 5 and the second with UAP and 4, to yield a 600-bp product. The two halves of RC5ЈDII, which overlapped by 92 bp, were joined to yield the full-length RC5ЈDII cDNA by overlap extension PCR (16). This reaction mixture contained equimolar amounts of the two products and the UAP* primer. The approximately 1500-bp product of this PCR was cloned into pBluescript using the PCR-Script kit (Stratagene, La Jolla, CA) and sequenced. To check for potential PCR errors, the entire cDNA was synthesized three times in separate RT/PCR reactions. Three errors were found, none of which was in the coding region.
Expression of RC5ЈDII in X. laevis Oocytes-In vitro synthesized, capped RNA transcripts of the putative 5DII cDNA were prepared using the MEGAscript kit (Ambion, Austin, TX) and injected into stage 5-6 X. laevis oocytes (50 ng of RNA/oocyte). Groups of oocytes were also injected with capped transcripts synthesized from a rat 5DI cDNA (G21, provided by Drs. Marla Berry and Reed Larsen, Boston, MA). The oocytes were incubated for 4 days in L-15 medium and then harvested, and membrane fractions were prepared as described by Sharifi and St. Germain (17), and 5ЈD and 5D activities were measured according to published methods (17,18). For the 5ЈD assay, the substrate was 1 nM [ 125 I]rT 3 with 0 -120 nM nonradioactive rT 3 , and the cofactor was 20 mM dithiothreitol; for the 5D assay, 1 nM [ 125 I]T 3 was used as substrate and 50 mM dithiothreitol as cofactor. [ 125 I]Iodothyronines (Dupont de Nemours & Co., Boston, MA) were purified by chromatography using Sephadex LH-20 (Sigma) before use. In some experiments the effect of PTU (0.001-0.1 mM) and AThG (0.1-100 M) on 5ЈD activity was examined.
Functional Analysis of RC5ЈDII by Deletion and Mutation-RC5ЈDII was truncated in the 3Ј-untranslated region at position 906 using DraI. Mutants of RC5ЈDII were made in which the in-frame TGA codon (bp 380 -382) was changed either to a TAA (stop) or to a TGT (cysteine) codon. The sequences of these mutants were verified. The ability of capped in vitro synthesized RNA transcripts of these cDNAs to induce 5ЈD activity following injection into X. laevis oocytes was then determined.
Analysis of RNA-Samples of poly(A) ϩ RNA from tissues of premetamorphic tadpoles (stages X-XII) and tadpoles undergoing metamorphic climax (stages XXIII-XXIV) were examined for the presence of RC5ЈDII-specific transcripts by Northern and slot blot analyses using methods previously described (19). Hybridizations and washings were carried out, respectively, at 42 and 50°C. Signals were visualized by autoradiography and quantitated with the 628E PhosphorImager (Molecular Dynamics). Data are reported as arbitrary densitometric units.

RESULTS AND DISCUSSION
The nucleotide and deduced amino acid sequences of RC5ЈDII are shown in Fig. 2A. RC5ЈDII is a 1459-bp cDNA with an open reading frame extending from bp 11 to 802 and an in-frame TGA codon at bp 380 -382 that, by analogy with the cDNAs for the 5DI and 5DIII, is likely to code for selenocysteine. A schematic comparison of the RC5ЈDII protein with those of R. catesbeiana type III and the rat type I is shown in Fig. 2B. The areas of homology are indicated, and it is also  noted that the proteins exhibit a hydrophobic region at the amino-terminal end and two histidine residues 3Ј to the TGA. In the rat 5DI, these histidines have been shown to be critical for 5ЈD activity (20).
5ЈD activity was induced in X. laevis oocytes after injection of capped RNA transcripts obtained by in vitro transcription of RC5ЈDII (Fig. 3A). No 5D activity was detected (data not shown). The 5ЈD activity was not inhibited by PTU (0.1 mM), but the percent deiodination of the [ 125 I]rT 3 was greatly reduced in the presence of 3 nM non-radioactive rT 3 , indicating that the enzyme manifests a low K m for this substrate. We have shown previously that the 5ЈD activity in tadpole tissues exhibits a low K m and is resistant to inhibition by PTU (5,21), characteristics that are typical of the type II 5ЈD activity described in mammals (1). In contrast, capped transcripts of the 5DI cDNA, G21, induced activity that was highly sensitive to PTU and appeared to have a relatively high K m . In addition, the 5ЈD activity induced by the RC5ЈDII transcripts was relatively insensitive to inhibition by AThG (Fig. 3B). The 5ЈD activity induced in oocytes by RC5ЈDII transcripts was consid-erably lower than that induced by transcripts of the G21 cDNA. The reason for this difference is not known. However, 5ЈD activity is only minimally induced in X. laevis oocytes by rat brown adipose tissue mRNA, a tissue containing considerable type II 5ЈD activity, 2 and it is possible that other factors important for type II 5ЈD activity are not optimal in this oocyte system.
In view of the indirect evidence suggesting that the mammalian 5DII is not a selenoprotein (3), including the fact that it catalyzes 5ЈD activity that is relatively insensitive to PTU and AThG, the possibility that the protein coded by RC5ЈDII is not a selenoprotein or that selenocysteine is not involved in the activity of the enzyme was investigated. It was found that 5ЈD activity in oocytes injected with capped transcripts derived from mutant RC5ЈDII cDNAs, where the TGA codon had been changed to TAA (stop) or TGT (cysteine), was essentially the same as levels obtained in uninjected oocytes. Furthermore, no induction of activity was observed after injection of transcripts derived from RC5ЈDII truncated at bp 906, suggesting that the deleted 3Ј-untranslated region contains a selenocysteine insertion sequence. These findings provide strong evidence that RC5ЈDII codes for a protein with selenocysteine at its catalytic site. Previous studies by Berry and Larsen (3) had led to the suggestion that sensitivity of a deiodinase to PTU and AThG could serve as a marker for the presence of selenocysteine at the enzyme's active site. That this is not a valid criterion is demonstrated by the presence of the TGA coding for selenocysteine in the RC5ЈDII cDNA described herein and in three recently isolated type III deiodinase cDNAs, all of which encode enzymes resistant to PTU and AThG (10 -12). Thus other structural properties of these enzymes or differences in kinetic  3. A, induction of 5ЈD activity in X. laevis oocytes following injection of capped RNA transcripts (50 ng/oocyte) synthesized in vitro from the RC5ЈDII and the 5ЈDI (G21) cDNAs. Control oocytes received no injection. Reaction mixtures contained 45, 49.5, and 54 g of membrane protein from control, RC5ЈDII, and G21 mRNA-injected oocytes, respectively. B, sensitivity to aurothioglucose of the 5ЈD activity induced in X. laevis oocytes by RC5ЈDII and G21 transcripts. Values are corrected for the 5ЈD activity (ϳ2%) observed in uninjected oocytes. Reaction mixtures contained 48 and 21 g of membrane protein from the RC5ЈDII and G21 mRNA-injected oocytes, respectively. mechanisms may dictate PTU and AThG sensitivity.
The ontogenic profiles of 5ЈD and 5D activities in R. catesbeiana tadpole tail and liver are very different; 5ЈD activity is minimal in tadpole tail until the onset of metamorphic climax when it increases markedly reaching a maximum by stage XXIII, while liver is devoid of 5ЈD activity at all stages of development (21). In contrast, 5D activity is present in both tail and liver during premetamorphosis, and in liver it is greatly reduced when the tadpoles reach metamorphic climax (22). To obtain additional evidence concerning the identity of RC5ЈDII, the size and relative abundance of RC5ЈDII-related transcripts in tail and liver poly(A) ϩ RNA obtained from premetamorphic (stage X-XII) and metamorphosing (stage XXIII) tadpoles were determined by Northern analysis (Fig. 4). No RC5ЈDII transcripts were detected in liver mRNA at either stage of development. This finding is consistent with the absence of 5ЈD activity in this tissue at all stages of the life cycle (5,21). However, a major RNA species of approximately 1.5 kb was detected in tail RNA; two minor species were just discernible at approximately 1.8 and 2.2 kb. The 2.2-kb species may represent cross-hybridization with RC5D transcripts since comparable blots probed with RC5D exhibit a signal at 2.2 kb (data not shown). In other blots a minor species was also detected at approximately 7.4 kb. The level of the 1.5-kb species was much higher in RNA from metamorphosing tail than in that from premetamorphic tail. Thus the profile of RC5ЈDII transcripts in liver and tail corresponds closely to that of 5ЈD but not 5D activity in these tissues. The increase in RC5ЈDII mRNA species in tail during metamorphic climax was quantified using slot blot analysis. Densitometric analysis indicated that the hybridization signals in tadpole tail at stages X-XII and stages XXIII-XXIV were, respectively, 280 Ϯ 50 (S.E.) and 5151 Ϯ 345 units (p Ͻ 0.001). As with the Northern blot, no signal was observed in liver RNA. Reprobing of the blot with RC5D revealed that levels of RC5D transcripts, which were clearly evident in both liver and tail at both stages of development, were not increased during metamorphic climax (data not shown). Thus, the observed increase in the level of RC5ЈDII-related transcripts on the blot cannot be attributed even in part to cross-hybridization of RC5ЈDII with RC5D transcripts.
Additional evidence that RC5ЈDII is an amphibian type II deiodinase is provided by our recent identification of rat and human homologues of this cDNA. Both are highly homologous to RC5ЈDII within the coding region (rat, 71%; human, 73%). Furthermore, the tissue distribution of their related mRNA transcripts, as determined by Northern analysis, is characteristic of the mammalian type II enzyme. 3 We thus conclude that RC5ЈDII is the cDNA for the 5DII in R. catesbeiana. The characteristics of the deiodinase for which it codes are comparable with those of the 5ЈD activity in R. catesbeiana tadpole tissues (5) and in mammalian brain, pituitary, and brown fat (1). The fact that type I 5ЈD activity has not been detected in tissues of this amphibian species (5, 21) makes it highly unlikely that RC5ЈDII codes for a form of the type I deiodinase. In addition, the data strongly suggest that the type II deiodinase coded by RC5ЈDII is a selenoprotein. We have compared the protein sequences deduced from the seven cloned deiodinase cDNAs, and it is evident that the RC5ЈDII protein has limited but significant homology with both the type I and the type III enzymes. Thus the amphibian type II deiodinase represents a structurally unique member of this family of selenocysteine-containing enzymes.