Cloning and Expression of a 5*-Iodothyronine Deiodinasefrom the Liver of Fundulus heteroclitus*

CARLOS VALVERDE-R, WALBURGA CROTEAU, GARY J. LAFLEUR, JR.,AUREA OROZCO, AND DONALD L. ST. GERMAIN

Universidad National Autonoma de Mexico (C.V.-R., A.O.), Mexico 04510 D. F.; Departments ofMedicine and Physiology (W.C., D.L.S.), Dartmouth Medical School, Lebanon, New Hampshire 03756;and The Whitney Laboratory (G.J.L.), University of Florida, St. Augustine, Florida 32086

ABSTRACTRecent molecular cloning studies in mammals and amphibians

have demonstrated that the types I, II, and III deiodinases constitutea family of selenoproteins of critical importance in metabolizing T4 toactive (i.e. T3) and inactive (i.e. rT3) metabolites. In several tissues ofteleost fish, various deiodinase processes have been described, but thestructural and functional characteristics of these enzymes and theirrelationship to the deiodinases present in higher vertebrates remainsuncertain.Using a complementaryDNA library derived from the liverof the teleost Fundulus heteroclitus, we have identified a complemen-

tary DNA that codes for a deiodinase with functional characteristicsvirtually identical to those of the mammalian and amphibian type IIdeiodinase. Sequence analysis demonstrates a high degree of homol-ogy at both the nucleotide and predicted amino acid levels between theFundulus clone and these previously characterized type II enzymes,including the presence of an in-frame TGA codon that codes for sel-enocysteine. These findings demonstrate that the deiodinase familyof selenoproteins has been highly conserved during vertebrate evo-lution and underscores their importance in the regulation of thyroidhormone action. (Endocrinology 138: 642–648, 1997)

THE THYROID axis in fish is similar in many respects tothat in higher vertebrates. Such similarities extend to

certain actions of thyroid hormones involved in the regula-tion of development, growth, and reproduction in this spe-cies (1–4). Furthermore, thyroid hormone receptors in fishare highly homologous to those present in mammals andshare the same preferential affinity for T3 as compared withother iodothyronines (5, 6). However, there is evidence thatthe thyroid follicles in trout and other teleosts secrete pre-dominantly, if not exclusively, T4 (5, 7), implying that cir-culating and tissue levels of the metabolically more activecompound T3 are derived primarily in these species from the59-deiodination of T4 in extrathyroidal tissues.The deiodination of iodothyronines is catalyzed by a fam-

ily of selenoenzymes that have differing catalytic propertiesand are expressed in both tissue-specific and developmen-tally-specific fashions (8). Information concerning the bio-chemical properties of these enzyme are derived primarilyfrom studies inmammals and amphibians. Two deiodinases,the types I (DI) and II (DII), serve an activating role byconverting T4 to T3 by 59-deiodination, whereas the type IIIdeiodinase (DIII) facilitates 5-deiodination, which convertsT4 and T3 to inactive metabolites (rT3 and 3, 39-diiodothy-ronine (T2), respectively). Similar processes of deiodinationhave also been reported in fish (5, 9). In particular, both DI-and DII-like activity have recently been reported in certainteleost species [Orozco, A., J. Silva, and C. Valverde-R, sub-mitted for publication and (10)]. For example, our recent

studies in Fundulus heteroclitus, a small estuarine teleost na-tive to North America (11), have shown that the liver of thisspecies contains high levels of DII-like activity (10).Complementary DNAs (cDNAs) for the DI, DII, and DIII

from several mammalian and amphibian species have re-cently been identified (12–20). To date, however, little isknown of the structural and molecular features of the en-zymes catalyzing deiodination in fish. Given the apparentprimacy in fish of extrathyroidal 59-deiodination in the gen-eration of circulating and tissue T3, we sought to identify acDNA that codes for a fish 59-deiodinase. We report hereinthe successful identification and expression of such a cDNAfrom a F. heteroclitus liver cDNA library.

Materials and MethodscDNA library screening

A l gt10 cDNA library was prepared using poly(A)1 RNA derivedfrom a pool of five livers frommale F. heteroclitus treated with estradiol-17b. The preparation of this library has previously been described (21)and used tissue from estrogen-treated animals so as to enhance therepresentation of vitellogenin cDNAs. The library was screened byplaque hybridization under low stringency conditions according to themethods of Lees et al. (22). The first 305 nucleotides of the coding regionof an amphibian DII cDNA (RC59DII) and a 714-nucleotide rat DI cDNA(6b-short) that encompasses 92% of the coding region of that enzymewere used together as probes in the initial screening protocol. Detailsconcerning the isolation and structures of these cDNAs have previouslybeen published (19, 23). Positive plaques were detected by autoradiog-raphy and purified by additional rounds of screening using the hybrid-ization conditions described above and either the amphibian DII or ratDI cDNAs separately as probes. cDNA inserts from the positive cloneswere amplified by PCR using vector-based primers. (PCR amplificationwas necessitated by the loss of the EcoRI restriction enzyme sites in thel gt10 vector during the library construction process.) The PCR reactionmixture of a selected clone (designated FhDII) was then subjected toagarose gel electrophoresis and the reaction product purified using theQIAquick Gel Extraction Kit (Qiagen, Chatsworth, CA) and subcloned

Received August 15, 1996.Address all correspondence and requests for reprints to: Donald L. St.

Germain, M.D., Dartmouth Medical School, One Medical Center Drive,Lebanon, New Hampshire 03756. E-mail: stgermain@dartmouth.edu.

* These studies were supported by the NIH in the form of GrantsDK-42271 and the Norris Cotton Cancer Center Core Grant CA-23108.

0013-7227/97/$03.00/0 Vol. 138, No. 2Endocrinology Printed in U.S.A.Copyright © 1997 by The Endocrine Society

642

into pBluescript using the pCR-Script Cloning Kit (Stratagene, La Jolla,CA). The cDNA insert was sequenced on both strands using vector-based and gene-specific primers and an automated sequencing systemwith fluorescent dye terminators (Applied Biosystems, Foster City, CA).

Radiolabeled cDNA probes for library screening and Northern anal-ysis were prepared as previously described (20).

The hydrophobicity profile of the FhDII proteinwas performed usingthe Kyte-Doolittle methodwith theMacDNASIS Pro computer program(Hitachi Software Engineering, San Bruno, CA), whereas prediction ofthe transmembrane domain was made using the TopPred II program(24).

Preparation of a chimeric FhDII cDNA

A chimeric cDNAwas constructed by splicing part of the 39-untrans-lated region of the full length rat DIII cDNA (rNS43-1), which containsan active selenocysteine insertion sequence (SECIS) element, to the 39end of the coding region of the FhDII cDNAusing overlap PCRmethodsanalogous to those previously described (20). The entire coding regionof the FhDII cDNA remained intact in this construct. The splice regionof the chimeric cDNA was sequenced to ensure the accuracy of theconstruction method.

Expression studies in COS-7 cells

For expression in COS-7 cells, the FhDII chimera cDNA was sub-cloned into the pcDNA3 mammalian expression vector (Invitrogen, SanDiego, CA). cDNAs for the rat DI (G21, kindly provided byDrs.M. Berryand P. R. Larsen, BostonMA), rat DII (rBAT1-1 chimera), and amphibianDIII (XL-15) were subcloned into the same vector as previously de-scribed (20). COS-7 cells were cultured and transfected as previouslydescribed (20), then maintained in culture medium for 48 h beforeharvesting. After aspiration of the medium, cell monolayers werewashed twice with PBS, and the cells then scraped from the dish, pel-leted, and sonicated in 0.25 m sucrose, 0.02 m Tris/HCl, pH 7.4.

59D and 5D activities were determined in COS-7 cell sonicates ac-cording to published methods (25, 26). For the 59D assay, the reactionbuffer contained 1 mm EDTA. In kinetic studies using rT3 and T4, 59Dactivity was determined during a 1-h incubation in a 50 ml reactionmixture volume containing 2 mg of sonicate protein and either 0.5–6 nm125I-rT3 or

125I-T4 as substrate. In these reactions, dithiothreitol at aconcentration of 20 mm was used as cofactor. The extent of substrateutilization was less than 52%. To correct for this degree of substrateutilization in kinetic studies, average substrate concentrations duringthe incubation mixture were used in the analysis as detailed by Lee andWilson (27). Kinetic constants were determined from double reciprocalor Eadie-Hofstee plots (28).

In other experiments, the 59-deiodinase activity in sonicates fromCOS-7 cells transfected with the G21 rat DI cDNA (13), the BAT1-1 ratDII cDNA (20) or the FhDII cDNA were determined at different assayincubation temperatures, or in the absence or presence of PTU (10–100mM) or aurothioglucose (0.01–10 mm). In these assays, deiodinase ac-tivity was measured during a 1 h (for temperature studies) or 2 h (forPTU and aurothioglucose studies) incubation using 1.5 nm 125I-rT3 assubstrate and 20 mm dithiothreitol as cofactor. Substrate utilization wasless than 35% in these studies.

Initial studies employing T3 as substrate used 1 nm 125I-T3, 50 mmdithiothreitol as cofactor, and a 2-h incubation period. For kinetic studieswith T3, a 1-h incubation period, 25 ml reaction volume containing 20 mgof sonicate protein, 20 mm DTT, and 0.5–15 nm or 15–1000 nm 125I-T3were used. Substrate utilization was less than 25%.

125I-labeled iodothyronines were obtained from du Pont de Nemours(Boston, MA) and purified by chromatography using Sephadex LH-20(Sigma, St. Louis, MO) before use. Protein concentrations were deter-mined by the method of Bradford (29) with reagents obtained fromBio-Rad (Richmond, CA).

RNA Preparation and northern analysis

RNA was prepared as previously described (21) from livers of maleF. heteroclitus. Poly(A)1 RNA was isolated by two cycles of chromatog-raphy over oligo(dT)-cellulose (Collaborative Biomedical Products, Bed-ford, MA). RNA gel electrophoresis, transfer to nylon membranes, hy-

bridization and washing of Northern blots were performed aspreviously described for rat tissues (30) with the final wash performedat 60 C. The FhDII cDNA was used as the probe, and the blots wereexposed to x-ray film for 1 week.

Results

The F. heteroclitus cDNA library was initially screened atlow stringency with radiolabeled probes derived from thecoding regions of both a rat DI and a R. catesbeiana DIIcDNAs. Four positive clones were isolated, all of which hy-bridized only with the RC59DII cDNA probe on subsequentscreenings. No clones reacting with the rat DI probe wereidentified. (The effects of estradiol treatment on deiodinaseexpression in F. heteroclitus liver are at present unknown butcould have influenced the relative abundance in the libraryof either theDI- orDII-type cDNAs.) Each of the four positiveisolates were demonstrated by PCR amplification to containa cDNA insert of the same size [;1 kilobase pair (kb)]. Thenucleotide sequence of one of these cDNAs (designated Fh-DII) was determined and is shown in Fig. 1A. An openreading frame, 798 bp in length, extends from nucleotides#132-929 and codes for a protein with a predicted molecularmass of 29.6 kDa. Included in the open reading frame is anin-frame TGA triplet at codon #134 (nucleotide nos. 531–533)that is predicted, like the other deiodinase cDNAs isolated todate (13–20), to code for selenocysteine. A comparison of theFhDII with DII cDNAs isolated from R. catesbeiana (19), rat,and human (20) reveals 65–68% nucleotide homologywithinthe coding region and the same percentage of amino acididentity (Fig. 1B). In contrast, homology to DI and DIIIcDNAs from several mammalian and amphibian species isconsiderably lower (less than 50% nucleotide conservationand only 30% amino acid identity). Nonetheless, severalstructural features of the FhDII protein are common to alltypes of deiodinases including, the highly conserved regionsurrounding the selenocysteine residue, two histidine resi-dues at amino acid nos. 166 and 186 (31), and a highlyhydrophobic region in the amino terminus that is predictedto form a membrane spanning domain encompassing resi-due nos. 8–28 (Fig. 1C).Although the 39 end of the FhDII cDNA contains a stretch

of 15 adenosines, no upstream polyadenylation signal ispresent, and hence thismay not represent a true poly(A)1 tailbut rather may have resulted during library constructionfrom the oligo(dT)-priming of first strand synthesis from aregion of the messenger RNA (mRNA) rich in adenosines.This is relevant because the 39-untranslated region of theFhDII clone is only 53 nucleotides in length and does notcontain a consensus SECIS element as required for the in-corporation of selenocysteine during translation (32). In theabsence of such an element, attempts at expression of thisclone would be predicted to result in a truncated, inactiveprotein in that translation would terminate at the TGAcodon. This indeed proved to be the case in that transfectionof COS-7 cells with the FhDII cDNA failed to induce expres-sion of deiodinase activity. Thus, to express and characterizethe full length FhDII protein, a chimeric cDNA was con-structed by replacingmost of the short 39 untranslated regionof the FhDII clone with the corresponding region of theNS43-1 rat DIII cDNA, which contains an active SECIS ele-

FUNDULUS 59-IODOTHYRONINE DEIODINASE 643

ment. The chimeric cDNA was then transfected into COS-7cells for functional studies.Initial studies in COS-7 cell sonicates demonstrated that

transfection with the FhDII chimeric cDNA induced signif-icant levels of 59-deiodination (52% deiodination of 1 nmradiolabeled rT3 during a 2-h incubation at 37 C; velocity 529 fmol/minzmg protein). In comparison, expression andassay of the G21 rat DI cDNA in the same experiment in-duced a considerably higher level of activity (velocity 5 283fmol/minzmg protein), a finding consistent with previousactivity comparisons between other DII cDNAs and the G21clone (20). [Wehave shownpreviously that COS-7 cells trans-fected with the empty pcDNA3 vector contain undetectablelevels of 59-deiodinase activity (20).]Kinetic analysis of the FhDII-induced 59-deiodinase activ-

ity using T4 and rT3 as substrates demonstrated low Kmvalues (0.5 and 1.0 nm, respectively) typical of a DII (Fig. 2).The Vmax value using rT3 as substrate was approximatelytwice the value observedwith T4, resulting inVmax/Km ratiosthat were equivalent for both substrates. Thus, the FhDIIenzyme appears to catalyze the 59-deiodination of T4 and rT3with equal efficiency.The ability of the FhDII deiodinase to catalyze 5-deiodi-

nation was investigated using T3, radiolabeled on the outerring at the 39- or chemically equivalent 59-position, as sub-strate and an ascending paper chromatography method forthe quantitation of reaction products (33). No radiolabeled T2was detected as a product in these reactions indicating that

catalysis at the 5-position is not a feature of this enzyme (Fig.3). Such results are again consistent with the properties of aDII. As a positive control, abundant T2 formation (;90%conversion of T3 to T2) was obtained in COS-7 cell sonicates

FIG. 2. Kinetic analysis of the expressed FhDII chimera deiodinaseactivity. Analysis by double reciprocal plot using T4 or rT3 as sub-strates in COS-7 cell sonicates. Vmax values are expressed in units ofactivity where 1 U 5 1 pmol/min z mg protein.

FIG. 1. A,Nucleotide and predicted amino acid sequence for the FhDII cDNA. The in-frameTGA codon is designated as coding for selenocysteine(SeC). B, Comparison of the deduced amino acid sequence of the FhDII protein with the sequence of the R. catesbeiana (RC59DII), rat (rBAT1–1) and human (hZ44085) DII proteins. 3 5 selenocysteine. Dots indicate identical residues, whereas dashes indicate gaps in the sequence.C, Hydrophobicity plot of the FhDII protein.

644 FUNDULUS 59-IODOTHYRONINE DEIODINASE Endo • 1997Vol 138 • No 2

expressing the XL-15 amphibian DIII cDNA. An unexpectedfinding in these studies was that considerable iodide (rep-resenting 18% of the radioactive T3 substrate added) wasformed in the FhDII assay mixture indicating that outer ring,or 59-, deiodination of T3 was occurring (Fig. 3). In contrast,no iodide formation occurred in the XL-15 sonicate mixture.Confirmation of the ability of the FhDII enzyme to 59-

deiodinate T3 was sought by using ion exchange chroma-tography to separate the reaction products (25). As shown inFig. 4, 40% of the T3 added as substrate was deiodinated atthe 59-position by the Fundulus enzyme, whereas minimal orno iodide formation was noted in sonicates from cells trans-fected with cDNAs for the rat DI (G21), the rat DII (rBAT1-1chimera), or an amphibianDIII (XL-15). (As noted above, andpreviously (20), these sonicates demonstrated very high lev-els of 59-deiodinase activity (rat DI and rat DII) when rT3 wasused as substrate, or 5-deiodinase activity (XL-15) when as-sayed using T3, indicating that the relevant deiodinases wereexpressed in these sonicates.)The results of a kinetic analysis of 59-deiodinase activity

with T3 as substrate is shown using an Eadie-Hofstee plot inFig. 5 and compared with the kinetic data for T4 cited above.The Km for T3 was considerably higher (340 nm) than for T4or rT3, and although a higher Vmax value was observed, theVmax/Km ratio was markedly lower than for the other twosubstrates (0.005 for T3 vs. 0.4 for T4 and rT3).The sensitivity of the FhDII deiodinase to inhibition by

6n-propyl-2-thiouracil (PTU) and aurothioglucose (AThG) isshown in Fig. 6. Analogous to amphibian and mammalianDIIs (19, 20), the Fundulus enzyme, when compared with theexpressed rat DI (G21), was markedly resistant to inhibitionby PTU and 10-fold less sensitive to the effects of AThG.Because F. heteroclitus is a poikilotherm, the temperature

sensitivity of the FhDII deiodinase was determined in vitro.As shown in Fig. 7, maximal 59-deiodinase activity was ob-served at 37 C, similar to the pattern observed with theexpressed rat DI and DII enzymes.A Northern blot of total and poly(A)1 RNA derived from

F. heteroclitus liver was probed with the FhDII cDNA (Fig. 8).In the lane containing total RNA, hybridization occurred

predominantly to a 1.3-kb RNA species, with a fainter bandat 6 kb noted. In the poly(A)1 lane, hybridization occurredonly to the larger RNA species.

Discussion

Using a cDNA library prepared from F. heteroclitus liverRNA, we have isolated the first cDNA coding for a fishdeiodinase and have demonstrated using a chimeric cDNAconstruct that the catalytic properties of the expressed en-zyme are remarkably similar to the previously characterizedamphibian and mammalian DII enzymes (19, 20). Further-

FIG. 4. 59-Deiodinase activity determined during a 2-h incubationusing 125I-T3 (1 nM) as substrate and dithiothreitol (50 mM) as cofac-tor. Iodide formation was quantified using ion exchange chromatog-raphy (25) of reactionmixtures utilizing sonicates of COS-7 cell trans-fected with empty pcDNA3 vector (Cont.), G21 rat DI, rBAT1-1chimera rat DII, XL-15 amphibian DIII, or FhDII chimera cDNAs.Values represent the means of closely agreeing duplicate assay de-terminations and are expressed as the percentage of the substrate T3deiodinated. Values are normalized for the differing amounts of pro-tein in the sonicate preparations.

FIG. 5. Kinetic data of the expressed FhDII chimera deiodinase ac-tivity using T3 as substrate and analyzed by Eadie-Hofstee plot. Thesame COS-7 cell sonicate preparation used in the experiments shownin Fig. 2 was used in this study. For comparison, the kinetic dataderived using T4 as substrate, and previously shown in Fig. 2, isplotted on this figure as a dashed line. Vmax values are expressed inunits of activity where 1 U 5 1 pmol/minzmg protein.

FIG. 3. Deiodinase reaction products formed during a 2-h incubationwith 125I-T3 (1 nM) as substrate and dithiothreitol (50mM) as co-factorin sonicates of transfected COS-7 cells. Shown is an autoradiographof the ascending paper chromatography strips used for the productanalysis (33). Duplicate strips are shown for each sonicate. Control,tissue-free reaction mixtures; FhDII, sonicates of COS-7 cells trans-fected with the FhDII chimera; XL-15, sonicates of COS-7 cells trans-fected with the XL-15 amphibian DIII cDNA. The location of T3, T2and iodide peaks are shown in the margin.

FUNDULUS 59-IODOTHYRONINE DEIODINASE 645

more, the FhDII cDNA and protein are highly homologousto these DII enzymes from other species.The F. heteroclitus is an excellent model for conducting

these studies in that it possesmany features typical of teleosts(21). In addition, thyroid function has been studied in thisspecies and a marked seasonal variation in T4 secretion hasbeen noted (34), as is true of many fish species (35, 36). Inaddition, we have recently characterized the deiodinase ac-tivity in the F. heteroclitus liver and have demonstrated thepresence of both DI and DII-like activity (10). Studies of59-deiodination in liver homogenates of other fish species bydifferent investigators have yielded somewhat conflictingdata on the characteristics of the deiodinase activities presentin this tissue. However, Eales and colleagues (37, 38) haveconsistently demonstrated a low Km, DII-like activity in the

liver of rainbow trout, and we have recently confirmed thisfindings and also demonstrated the presence of DI-like ac-tivity in this tissue (Orozco, A., J. Silva, and C. Valverde-R,submitted for publication). It is notable that the functionalcharacteristics of the FhDII deiodinasematch closelywith theDII-like activities defined in Fundulus and trout liver andother tissues, suggesting that this cDNA, and its homo-logues, which are presumably present in other teleost spe-cies, indeed code for these enzymes.The functional activity of the FhDII as expressed in COS-7

cells is typical of a DII (39). Thus, this enzyme demonstratesa Km value for T4 and rT3 in the nanomolar rangewhen using

FIG. 6. Sensitivity of the G21 rat DI and FhDII chimera deiodinasesto the inhibitory effects of (A) PTU and (B) aurothioglucose. COS-7cells were transfectedwith either theG21 or the FhDII chimera cDNAand assays performed in cell sonicates in the absence or presence ofthe concentrations of inhibitors shown. Values of 100% represent theresults of control incubations performed in aliquots of the same son-icate preparations in the absence of inhibitors.

FIG. 7. A comparison of the thermal activation profiles of the rat DI(G21), rat DII (rBAT1-1 chimera) and FhDII chimera deiodinases.59-Deiodinase activity was determined in vitro in sonicates of trans-fected COS-7 cells at the temperatures indicated. Results representthe mean of closely agreeing duplicate assay determinations. Valuesof 100% represent the maximally observed rates of deiodination thatoccurred at 37 C for each of the sonicate preparations and amountedto 20–30% of the substrate present in the assay mixture. Activities atother temperatures are expressed as a percentage of this maximalvalue for each sonicate.

FIG. 8. Northern analysis using the FhDII cDNA as a probe of total(20 mg) and poly(A)1 RNA (2 mg) derived from male F. heteroclitusliver.

646 FUNDULUS 59-IODOTHYRONINE DEIODINASE Endo • 1997Vol 138 • No 2

DTT as the thiol cofactor, and it catalyzes the 59-deiodinationof both these substrates with approximately equal efficiency.Furthermore, the FhDII deiodinase is resistant to inhibitionby high concentrations of PTU and demonstrates diminishedsensitivity to AThG as compared with the mammalian DI.The structural features of the FhDII correlate well with thesefunctional characteristics in that the cDNA and predictedamino acid sequences are most homologous to the DII sub-family of deiodinases. In particular, the FhDII cDNA con-tains an alanine at amino acid no. 132, which is two residuestoward the amino terminus from the selenocysteine. Thepresence of an alanine at this position has been a feature ofall of the DII proteins characterized to date (19, 20). In con-trast, the DI and DIII enzymes contain a cysteine in thislocation. The functional significance of this difference instructure is as yet uncertain.An unexpected structural feature of the rat and humanDII

cDNAs is the presence of a second in-frame TGA codonlocated just 59 to an unambiguous TAG (rat) or TAA (human)stop codon (20). Whether this TGA codes for a second sel-enocysteine residue in these proteins is uncertain, and coulddepend on the selenium status of the cell. This downstreamTGA codon is not present in the FhDII cDNA, however. Inthis respect, the FhDII more closely resembles the R. cates-beianaDII, which also contains only a single TGA codon (19).This suggests that incorporation of a second selenocysteineis not essential for DII activity, a finding that we have re-cently confirmed by demonstrating that mutagenesis of thesecond TGA codon to TAA in the rat DII cDNAdoes not alterthe functional activity of this enzyme (Croteau and St. Ger-main, unpublished data). However, the essential nature ofthe first selenocysteine to the catalytic activity of this familyof enzymes has clearly been demonstrated by us (15, 16, 19)and other investigators (13, 40); mutagenesis of this residueto cysteine either renders the enzyme inactive or reduces thecatalytic efficiency by 1000-fold. The presence of selenocys-teine in the FhDII enzyme demonstrates its conservationthrough vertebrate evolution, and thus further emphasizesthe requirement of this rare amino acid for efficient catalysisof the reductive deiodination reaction.Incorporation of selenocysteine into proteins occurs dur-

ing translation and requires the presence of a specific stemloop structure (the SECIS element) in the 39-untranslatedregion of the mRNA (41). The 1-kb FhDII cDNA is shorterthan the corresponding 1.3- and 6-kb RNA species identifiedin Fundulus liver by Northern analysis and appears to havea truncated 39-untranslated region that lacks such a SECISelement. Thus the relative efficiency of translation of theFhDII when utilizing its native SECIS element is unknownand may differ from that of the chimeric cDNA constructused in these studies.The hybridizing 1.3-kb RNA seen in the total RNA sample

was not present when poly(A)1 RNA was used indicatingthat this short RNA lacks a poly(A)1 tail. Its presence sug-gests that alternative mRNA processing may be involved inregulating the expression of DII in this tissue. The 6-kb spe-cies, however, is similar in size to the predominant DII RNAspecies identified in samples of poly(A)1 RNA preparedfrom several mammalian tissues (20) but is larger than thepredominant 1.5-kb mRNA present in R. catesbeiana (19). In

contrast, the mRNAs coding for the DI and DIII enzymes areapproximately 2 kb in size (12, 16).The FhDII deiodinase demonstrates a unique functional

activity: namely, its ability to catalyze, albeit with relativelylow efficiency, the 59-deiodination of T3. Such activity has notpreviously been noted in any of the deiodinases character-ized to date, including the amphibian RCDII cDNA (V. A.Galton, personal communication). Thus, comparative stud-ies with deiodinases from other species may provide inter-esting insights into the structure-function correlates of thisunique catalytic activity. Notably, 59-deiodination of T3 hasbeen observed in teleost tissue homogenates; Pimlott andEales (37) have described “weak” 59-deiodination of T3 inrainbow trout liver homogenates incubated at 20 C, the high-est temperature at which they performed their assays. Thephysiological significance of this catalytic activity, however,may be relatively minor given the marked preference of theFhDII to utilize T4 and rT3 as substrates.Early studies conducted by Leatherland (42) in trout liver

homogenates suggested that the optimal temperature for59-deiodination in this tissue was 20 C. In these studies, nothiol co-factors were included in the trout liver reaction mix-tures and the substrate concentration was so high (1.4 mm T4)as to preclude distinguishing between DI and DII activity.Thus, a comparison of these data with ours is problematic.However, more recent data from our laboratory have dem-onstrated that DII activity as determined in vitro in trout liverhomogenates is maximal at 25 C (Orozco, A., J. Silva, and C.Valverde-R, submitted for publication). It was, therefore,somewhat surprising to find that the thermal activity profileof the FhDII deiodinase was the same as that of the mam-malian DI and DII enzymes with maximal activity noted at37 C. Such differences between the temperature sensitivitiesof the trout and Fundulus enzymes could reflect their differ-ent habitats in that the latter species tends to reside inwarmerwaters. Similar observations showing maximal activity of anamphibianDII at 37Chave beenmade in tissue homogenatesfrom R. catesbeiana (V. A. Galton, personal communication).In summary, although previous studies have documented

that there are important species differences in the tissue anddevelopmental patterns of expression of the DII, the presentstudy demonstrates that the structural and functional char-acteristics of the DII protein have been highly conservedduring evolution. This provides additional evidence that thisenzyme serves an important physiological role. The avail-ability of the FhDII cDNA should facilitate further investi-gations into the role of thyroid hormones and their metabolicfate in fish.

Acknowledgments

The authors wish to thank M. Marion Byrne for providing the F.heteroclitus liver cDNA library and Jennifer Davey, Mark Schneider,Kathryn Becker, and Valerie Anne Galton for their interest and helpfuldiscussions during the course of these studies.

References

1. Barrington EJW 1975 An Introduction to General and Comparative Endocri-nology, ed 2. Clarendon Press, Oxford, pp 176–178

2. McNabb FMA 1992 Thyroid Hormones. Prentice Hall, Inc.3. Leatherland JF 1987 Thyroid hormones and reproduction. In: Norris DO, Jones

FUNDULUS 59-IODOTHYRONINE DEIODINASE 647

RE (eds) Hormones and Reproduction in Fishes, Amphibians, and Reptiles.Plenum Press, New York, pp 411–431

4. Higgs DA, Fagerlund UHM, Eales JG, McBride JR 1982 Application of thy-roid and steroid hormones as anabolic agents in fish culture. Comp BiochemPhysiol 73B:143–176

5. Eales JG 1985 The peripheral metabolism of thyroid hormones and regulationof thyroidal status in poikilotherms. Can J Zool 63:1217–1231

6. Weirich RT, Schwartz HL, Oppenheimer JH 1987 An analysis of the inter-relationships of nuclear and plasma triiodothyronine in the sea lamprey, laketrout, and rat: evolutionary considerations. Endocrinology 120:664–677

7. Grau EG, Helms LHM, Shimoda SK, Food CA, LeGrand J, Yamanski K 1986The thyroid gland of the Hawaiian parrotfish an its uses as an in vitro modelsystem. Gen Comp Endocrinol 61:100–108

8. St. Germain DL 1994 Iodothyronine deiodinases. Trends Endocrinol Metab5:36–42

9. Mol K, Kaptein E, Darras VM, de Greef WJ, Kuhn ER, Visser TJ 1993Different thyroid hormone-deiodinating enzymes in tilipia (Oreochromis nil-oticus) liver and kidney. FEBS Let 321:140–144

10. Orozco A, Linser PJ, Valverde-RC Salinity modifies hepatic outer-ring deio-dination (ORD) activity in Fundulus heteroclitus. Program of the 18th Confer-ence of Comparative Endocrinologists Rouen France 1996 (Abstract)

11. Nelson JS 1994 Fishes of the World, ed 3. John Wiley & Sons, Inc., New York,pp 269–272

12. St. GermainDL,DittrichW,Morganelli CM, CrynsV 1990Molecular cloningby hybrid arrest of translation in Xenopus laevis oocytes: identification of acDNA encoding the type I iodothyronine 59-deiodinase from rat liver. J BiolChem 265:20087–20090

13. Berry MJ, Banu L, Larsen PR 1991 Type I iodothyronine deiodinase is aselenocysteine-containing enzyme. Nature 349:438–440

14. Mandel SJ, Berry MJ, Kieffer JD, Harney JW, Warne RL, Larsen PR 1992Cloning and in vitro expression of the human selenoprotein, type I iodothy-ronine deiodinase. J Clin Endocrinol Metab 75:1133–1139

15. St. Germain DL, Schwartzman R, Croteau W, Kanamori A, Wang Z, BrownDD, Galton VA 1994 A thyroid hormone regulated gene in Xenopus laevisencodes a type III iodothyronine 5-deiodinase. Proc Natl Acad Sci USA 91:7767–7771. Correction. Proc Natl Acad Sci USA 91:11282, 1994

16. CroteauW,Whittemore SL, SchneiderMJ, St. Germain DL 1995 Cloning andexpression of a cDNA for a mammalian type III iodothyronine deiodinase.J Biol Chem 270:16569–16575

17. Becker KB, Schneider MJ, Davey JC, Galton VA 1995 The type III 5-deio-dinase in Rana catesbeiana tadpoles is encoded by a thyroid hormone-respon-sive gene. Endocrinology 136:4424–4431

18. SalvatoreD, LowSC, BerryM,MaiaAL,Harney JW, CroteauW, St. GermainDL, Larsen PR 1995 Type 3 iodothyronine deiodinase: cloning, in vitro ex-pression, and functional analysis of the placental selenoprotein. J Clin Invest96:2421–2430

19. Davey JC, Becker KB, Schneider MJ, St. Germain DL, Galton VA 1995Cloning of a cDNA for the type II iodothyronine deiodinase. J Biol Chem270:26786–26789

20. CroteauW, JCD, Galton VA, St GermainDL 1996 Cloning of themammaliantype II iodothyronine deiodinase: a selenoprotein differentially expressed andregulated in the human brain and other tissues. J Clin Invest 98:405–417

21. LaFleur Jr. GJ, Byrne BM, Kanungo J, Nelson LD, Greenberg RM, WallaceRA 1995 Fundulus heteroclitus vitellogenin: the deduced primary structure ofa piscine precursor to noncrystalline, liquid-phase yolk protein. J Mol Evol41:505–521

22. Lees JA, Saito M, Vidal M, Valentine M, Look T, Harlow E, Dyson N, HelinK 1993 The retinoblastoma protein binds to a family of E2F transcriptionfactors. Mol Cell Biol 13:7813–7825

23. O’Mara BA, Dittrich W, Lauterio TJ, St. Germain DL 1993 Pretranslationalregulation of type I 59-deiodinase by thyroid hormones and in fasted anddiabetic rats. Endocrinology 133:1715–1723

24. Claros MG, von Heijne G 1994 TopPred II: an improved software for mem-brane protein structure predictions. Computer applications in the bioscience:CABIOS 10:685–686

25. Sharifi J, St. Germain DL 1992 The cDNA for the type I iodothyronine59-deiodinase encodes an enzyme manifesting both high Km and low Kmactivity. J Biol Chem 267:12539–12544

26. St. Germain DL, Croteau W 1989 Expression of phenolic and tyrosyl ringiodothyronine deiodinases inXenopus laevis oocytes is dependent on the tissuesource of injected poly(A)1 RNA. Mol Endocrinol 3:2049–2053

27. Lee H-Y, Wilson IB 1971 Enzymatic parameters: measurement of V and Km.Biochem Biophys Acta 242:519–522

28. Hofstee BHJ 1952 On the evaluation of the constants Vm and Km in enzymereactions. Science 116:329–331

29. Bradford MM 1976 A rapid and sensitive method for quantitation of micro-gram quantities of protein utilizing the principle of protein-dye binding. AnalBiochem 72:248–254

30. DePalo D, Kinlaw WB, Zhao C, Engelberg-Kulka H, St. Germain DL 1994Effect of selenium deficiency on type I 59-deiodinase. J Biol Chem269:16223–16228

31. Berry M 1992 Identification of essential histidine residues in rat type I iodo-thyronine deiodinase. J Biol Chem 267:18055–18059

32. Low SC, Berry MJ 1996 Knowing when not to stop: selenocysteine incorpo-ration in eukaryotes. Trends Biochem Sci 21:203–208

33. GaltonVA,Hiebert A 1987 The ontogeny of the enzyme systems for the 59-and5-deiodination of thyroid hormones in chick embryo liver. Endocrinology120:2604–2610

34. BergO,GorbmanA,KobayashiH 1959 The thyroid hormones in invertebratesand lower vertebrates. In: Gorbman A (ed) Comparative Endocrinology. JohnWiley & Sons, Inc., New York, pp 302–319

35. Osborn RH, Simpson TH, Youngson AF 1978 Seasonal and diurnal rhythmsin thyroidal status in the rainbow trout, Salmo gairdneri Richardson. J Fish Biol12:531–540

36. Eales JG, Fletcher GL 1982 Circannual cycles of thyroid hormones in plasmaof winter flounder (Pseudopleuronectes americanus Walbaum). Can J Zool60:304–309

37. Pimlott NJ, Eales JG 1983 In vitro study of hepatic iodothyonine deiodinationin rainbow trout, Salmo gairdneri. Can J Zool 61:547–552

38. Frith SD, Eales JG 1996 Thyroid hormone deiodination pathways in brain andliver of rainbow trout, Oncorhynchus mykiss. Gen Comp Endocrinol101:323–332

39. Leonard JL 1991 Biochemical basis of thyroid hormone deiodination. In: WuS (ed) Thyroid Hormone Metabolism: Regulation and Clinical Implications.Blackwell Scientific Publications, Boston, pp 1–28

40. Berry MJ, Maia AL, Kieffer JD, Harney JW, Larsen PR 1992 Substitution ofcysteine for selenocysteine in type I iodothyronine deiodinase reduces thecatalytic efficiency of the protein but enhances its translation. Endocrinology131:1848–1852

41. BerryMJ, Banu L, ChenY,Mandel SJ, Kieffer JD,Harney JW, Larsen PR 1991Recognition of UGA as a selenocysteine codon in type I deiodinase requiressequences in the 39 untranslated region. Nature 353:273–276

42. Leatherland JF 1981 Conversion of L-thyroxine to triiodo-L-thyronine in rain-bow trout (Salmo gairdneri) liver and kidney homogenates. Comp BiochemPhysiol 69B:311–314

648 FUNDULUS 59-IODOTHYRONINE DEIODINASE Endo • 1997Vol 138 • No 2