Transgenic Analysis Reveals that Thyroid Hormone Receptor Is Sufficient To Mediate the Thyroid Hormone Signal in Frog Metamorphosis

Thyroid hormone (T3) has long been known to be important forvertebrate development and adult organ function. Whereas thyroidhormone receptor (TR) knockout and transgenic studies of micehave implicated TR involvement in mammalian development, theunderlying molecular bases for the resulting phenotypes remainto be determined in vivo, especially considering that T3 isknown to have both genomic, i.e., through TRs, and nongenomiceffects on cells. Amphibian metamorphosis is an excellent modelfor studying the role of TR in vertebrate development becauseof its total dependence on T3. Here we investigated the roleof TR in metamorphosis by developing a dominant positive mutantthyroid hormone receptor (dpTR). In the frog oocyte transcriptionsystem, dpTR bound a T3-responsive promoter and activated thepromoter independently of T3. Transgenic expression of dpTRunder the control of a heat shock-inducible promoter in premetamorphictadpoles led to precocious metamorphic transformations. Molecularanalyses showed that dpTR induced metamorphosis by specificallybinding to known T3 target genes, leading to increased localhistone acetylation and gene activation, similar to T3-boundTR during natural metamorphosis. Our experiments indicated thatthe metamorphic role of T3 is through genomic action of thehormone, at least on the developmental parameters tested. Theyfurther provide the first example where TR is shown to mediatedirectly and sufficiently these developmental effects of T3in individual organs by regulating target gene expression inthese organs.

INTRODUCTION

Top

Introduction

Thyroid hormone receptors (TRs) are dual-function transcriptionalregulators (60). On positive hormone response elements in theabsence of thyroid hormone (T3), TRs repress T3 response genes,and in the presence of ligand, TRs activate these genes. Thedual effects of TR are accomplished by recruiting mutually exclusivesets of coregulators to target promoters (39, 63). Corepressorcomplexes comprising nuclear corepressor (N-CoR) or silencingmediator of retinoid and thyroid hormone receptors (SMRT) withhistone deacetylase 3, TBL1/TBLR1, and GPS2 associate with unligandedTR and deacetylate histones (18, 27, 28, 32, 53, 61, 62). Thepresence of T3 induces a conformational change in TR, promotingthe release of corepressor complexes and recruitment of coactivatorcomplexes that increase histone acetylation (5, 26, 35, 39,41, 63). Despite the wealth of knowledge of the molecular mechanismsof TR action in vitro, relatively little is known of the roleof TR and associated mechanisms in vivo during vertebrate development.

TR genes comprise two genetic loci, ${alpha}$ and ß, each withmultiple splice variants, and knockout studies of various subsetsof TR gene products in mice reveal a large diversity of developmentaleffects (10-12, 15, 16, 57). Some phenotypes may be explainedbased on differential tissue expression of TR ${alpha}$ and TRß.For example, TR expression in the ear is predominantly TRß,and TRß but not TR ${alpha}$ knockouts affect hearing. However,other phenotypes cannot be explained based on differential expressionalone. Equal amounts of TR ${alpha}$ and TRß are expressed inthe intestine, yet only TR ${alpha}$ knockouts have a dramatic effecton intestine development. Also, double knockouts are less severethan mutations or treatments that eliminate T3, which leavesopen the possibility that some of the developmental roles ofT3 may be due to nongenomic effects of T3, which have been welldocumented (8). In addition, the molecular bases of these variousdevelopmental effects have not been clarified, and the resolutionof these questions requires an in vivo model system.

Frog metamorphosis provides an attractive model to address thedevelopmental roles of TR at the molecular level. The ease andspeed of making transgenic tadpoles allow rapid evaluation ofeffects of transgenes on tadpole development, and the molecularmechanisms of these effects can be readily assessed becauseof the large size of the free-living tadpoles. More importantly,thyroid hormone physiology in frogs exists in a simplified statecompared to that in mice, where maternal T3 is always present.In frogs, T3 is absent from premetamorphic tadpoles (56), andpostembryonic developmental events that take place during metamorphosisare completely dependent upon T3. In addition, metamorphosiscan be induced precociously by the addition of exogenous T3(45). Thus, all TRs are in the unliganded state before metamorphosis,and at the climax of metamorphosis or upon addition of T3 tothe rearing water, the TRs become ligand bound. Previous workwith frogs has validated in vitro cell culture models of TRfunction in development, in that TR is constitutively boundto target promoters in different tissues in vivo (43). In addition,corepressors N-CoR, SMRT, and TBLR1 are present on the promotersof T3 response genes TRß and basic leucine zippertranscription factor (TH/bZIP) in premetamorphic tadpoles (42,53). Furthermore, transgenesis using a dominant negative TRhas been used to show that T3-induced activation or derepressionis necessary for gene regulation and morphological changes duringmetamorphosis (3, 7, 44).

In the present study, we addressed whether TR activation issufficient for the changes in gene regulation and morphologyin metamorphosis. We developed a dominant positive TR (F-dpTR),which has a FLAG tag and VP16 activation domain fused to theN terminus of TR and the receptor-interacting domain (RID) ofXenopus N-CoR fused to the C terminus. We showed that this mutantreceptor strongly activates positively regulated promoters inthe absence of ligand. In tadpoles transgenic for the F-dpTRunder heat shock control, metamorphic changes were induced afterheat shock, even in the absence of T3, suggesting that TR issufficient to mediate these metamorphic effects of thyroid hormone.

MATERIALS AND METHODS

Top

Materials and Methods

Molecular cloning. A series of 11 PCRs was used to make the dominant positive constructs.In reaction 1, a FLAG-tagged activation domain of VP16 (herpessimplex virus type 1; amino acids [aa] 413 to 490) was createdby PCR amplification using the primers DRB 42 and DRB 43 (seeTable 1 for primer sequences) from the template CGI-DBEcR (giftfrom David Johns, Johns Hopkins University). These primers removedthe natural XbaI site from VP16. Full-length (aa 2 to 419) (reaction2) and two dominant negative (aa 2 to 400) (reactions 3 and4) Xenopus TR ${alpha}$ A (xTR ${alpha}$ A) minus the initial methionine and stopcodon were amplified by using primers DRB 44 and 45, DRB 44and 46, and DRB 44 and 49b from the template pSP64[poly(A)]-TR ${alpha}$ A(59). Two RIDs (aa 1680 to 2357 and 1988 to 2357) (reactions5 and 6) from Xenopus N-CoR were amplified from pSP64-N-CoR-4with primers DRB 47 and 48 and DRB 50b and 48 (42). The FLAG-taggedVP16 activation domain was attached to the TRs by mixing PCRproducts from reactions 1 and 2 with primers DRB 42 and 45 tomake F-VP-TR (reaction 7), PCR products from reactions 1 and3 with primers DRB 42 and 46 (reaction 8), and PCR productsfrom reactions 1 and 4 with primers DRB 42 and 49b (reaction9). To attach the RIDs to the FLAG-VP16-dnTRs, PCRs were performedusing products from reactions 5 and 8 with primers DRB 42 and48 to make F-dpTR-l (reaction 10) and products from reactions6 and 9 with primers DRB 42 and 48 to make F-dpTR (reaction11). PCR products from reactions 7, 10, and 11 were cloned intopSP64RI (S. Sato, National Institute of Diabetes and Digestiveand Kidney Diseases) by using EcoRI for making mRNA, and theF-dpTR from reaction 11 was subcloned by using AgeI and EcoRIinto pCGHG, a double promoter transgenesis vector (13). Theconstructs F-TR-l and F-TR-s were made by digesting F-dpTR-land F-dpTR with EagI, which removed the fragment containingVP16, and religating.

View this table:
[in this window]
[in a new window]
TABLE 1. Primers for cloning

Animals. Adult frogs were purchased (Nasco, Fort Atkinson, Wis.) foroocyte assay and transgenesis. Transgenesis by restriction enzyme-mediatedintegrations was carried out as described previously (3, 31).Wild-type and transgenic tadpoles were reared at room temperaturein 1 mM methimazole to block endogenous T3 synthesis (4, 6).The tadpoles were fed daily, and water was changed every otherday. Premetamorphic tadpoles (stage 45 to 54) (36) were heatshocked at 33°C for 5 min on day 0 and then heat shockedfor 1 h at 33°C on days 1 through 10. For comparison, somewild-type tadpoles at stage 52 were treated with 5 nM T3 for3 days (47). Tadpoles were not fed during the treatment period.Animals were photographed under bright field and fluorescencewith a green fluorescent protein (GFP) filter, and the imageswere merged to show GFP expression in transgenic tadpoles harboringthe double promoter transgenesis vector (13).

Transcription assay of the oocyte. Oocytes were prepared according to established procedures (52,58). Wild-type and mutant TR and retinoid X receptor (RXR) mRNAsfor oocyte injections were synthesized with mMESSAGE mMACHINE(Ambion) from the mutant TR templates (the present study) andpSP64[poly(A)]-TR ${alpha}$ A and pSP64[poly(A)]-RXR (59). For the luciferaseassay, 23 nl of water or a 50/50 mixture of 50 ng of RXR mRNA/µland 50 ng of the indicated TR mRNAs/µl was injected intothe oocyte cytoplasm. Then, 13.4 nl of a 66/33 mixture of 25ng of pGL-TRE luciferase (TRE-luc)/µl (1) and 5 ng ofphRG-thymidine kinase (TK)/µl (Promega) was injected intothe nucleus. After overnight incubation at 18°C, six oocyteswere lysed together by pipetting in 90 µl of 1x lysisbuffer from the dual-luciferase assay kit (Promega), and 7 µlof oocyte lysate was used for each assay with 37 µl ofsubstrate and 37 µl of Stop and Glo buffer according tothe manufacturer's instructions. Assays were performed in triplicate.Western blotting, immunoprecipitation, and chromatin immunoprecipitation(ChIP) from oocytes were performed as previously described (52,53).

Histology and TUNEL assay. Tadpole intestines were fixed in 4% paraformaldehyde for 2 hat room temperature, cryoprotected in 0.5 M sucrose-60% phosphate-bufferedsaline (PBS), and embedded in Tissue-Tek OCT medium. Cryosections10 µm in size were made and stained with methyl green-pyronineY (Muto) or processed for the terminal deoxynucleotidyltransferase-mediateddUTP-biotin nick end labeling (TUNEL) assay. For TUNEL, sectionson slides were refixed in 4% paraformaldehyde for 15 min atroom temperature and then rinsed in PBS. The sections were permeabilizedwith 20 µg of protease K/ml in 10 nM Tris, pH 8, for 10min at room temperature. After sections were rinsed in PBS,endogenous peroxidase activity was quenched for 20 min with3% H₂O₂ in 90% methanol. The TUNEL reaction with and withoutterminal deoxynucleotidyl transferase (TdT) was carried outat 37°C for 90 min in TdT buffer (Oncogene), 0.016 mM biotin-14-dATP(Invitrogen), and 0.3 U of TdT/ml (Invitrogen). Then, the sectionswere rinsed in PBS, blocked for 10 min with 1% bovine serumalbumin, and incubated with ABC Elite Ready-To-Use solution(Vector Labs). After 30 min, antibody binding was detected witha DAB staining kit (Vector Labs).

Protein and RNA extraction, Western blotting, and RT-PCR. Tadpole organs were crushed in lysis buffer (10 mM Tris-HCl[pH 7.5], 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, completemini EDTA-free protease inhibitor tablet [Roche], 1 mM dithiothreitol)and then forced through a syringe after an equal volume of 2xsodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)loading buffer was added. The protein lysate was boiled, runon an SDS-PAGE gel, and subjected to anti-N-CoR and anti-FLAG(Stratagene) Western blotting (52, 53). RNA from tails and intestineswas isolated with TRIzol reagent (Invitrogen). Reverse transcription(RT)-PCR was carried out as described previously using the sameprimers (3).

In vivo chromatin immunoprecipitation. Isolation of chromatin from tadpole organs was performed asdescribed previously (43, 53). Immunoprecipitation (IP) wascarried out by using anti-FLAG (Stratagene) and anti-acetylhistone H4 (Upstate) antibodies. Preclearing and IP for anti-FLAGantibodies were done with protein G Sepharose beads (AmershamBiosciences) rather than protein A Sepharose beads. After reversecross-linking, DNA was purified with a PCR purification kit(QIAGEN). A quantitative PCR was carried out to analyze theChIP DNA samples; each sample was run in duplicate on an ABI7000 (Applied Biosystems) using gene-specific primers and FAM(6-carboxyfluorescein)-labeled Taq-man probes (Applied Biosystems)(Table 2). To ensure the validity of the quantitative PCR, thefollowing control experiments were done. For each assay, sixtwofold serial dilutions from a large batch of ChIP input DNAmixed from tails and intestines prepared especially for thepurpose of serving as standards were used for quantificationof the experimental samples. The calculated standard curvesranged in slope from –3.30 to –3.70, where theoreticalamplification has a slope of –3.32. Also included in eachrun was a no-template control where pure water was added tocontrol for PCR product contamination. In addition, a sampleof 0.1 ng of tadpole genomic DNA/µl was used to assessthe consistency of the standard curve calculated from the serialdilutions between runs. The coefficient of variation for thecontrol samples was less than 5% for each primer probe set used.Results from the control sample and experimental ChIP samplesfell within the range of the standard curve. Over 10 side-by-sidecomparisons of results from radioactive gel electrophoresisand quantitative PCR using the same ChIP DNA samples gave thesame qualitative conclusions (data not shown).

View this table:
[in this window]
[in a new window]
TABLE 2. Primers and probes for quantitative PCR

RESULTS

Top

Results

Construction of dominant positive TR. Because no known TR mutants are dominant positive, we designedmutant TRs to enable high levels of gene induction without ligandand maintain native promoter binding. As a first step, we wantedto mutate TR to block endogenous corepressor binding. Unfortunately,mutations within the TR ligand-binding domain and hinge regionthat inhibit interaction with corepressors also reduce naturalheterodimerization with the endogenous TR binding partner, 9-cis-retinoidX receptor (RXR), which would ensure native promoter bindingaffinity and specificity (33, 64). An alternative way to blockTR-mediated repression is to use dominant negative forms ofN-CoR (9). However, this strategy requires highly overexpressedamounts of dominant negative N-CoR in vivo, is not specificfor TR, and is undesirable because two proteins would be neededfor subsequent transgenesis. We decided to fuse the RID of XenopusN-CoR to the C terminus of TR. The N-CoR and TR ${alpha}$ moieties areexpected to bind intramolecularly, thus preventing gene repressiondue to the binding of native corepressor complexes to the TR ${alpha}$ moiety in the absence of T3. Because helix 12 of TR reducescorepressor binding (34), we fused the RID of Xenopus N-CoRto a dominant negative TR ${alpha}$ (dnTR) lacking the C-terminal helix12 to promote the intramolecular association. Two differentfragments containing the RID (21, 55) of Xenopus N-CoR weremade and used to generate two dnTR fusion proteins (Fig. 1A).The long version, F-TR-l, had an N-terminal extension from theRID in hopes that it would be able to adopt the proper TR interactionconformation. For the shorter version, F-TR-s, we estimatedthe minimal number of amino acids required for the N-CoR bindingsurface of TR to fit the corresponding corepressor nuclear receptorbinding motif of N-CoR RID I. The N-terminal FLAG tag was addedto wild-type and mutant TRs for ease of detection.

View larger version (21K):
[in this window]
[in a new window]
FIG. 1. The fusion of the RID of N-CoR to a dominant negative TR leads to partial derepression. (A) Diagram of TR fusion proteins. The fusion constructs F-TR-l and F-TR-s include a long and short fragment of Xenopus N-CoR containing the RID. The black boxes labeled 3, 2, and 1 represent the corepressor nuclear receptor binding motifs (I/LXXI/VI) of N-CoR that directly bind TR (21). The dominant negative Xenopus TR ${alpha}$ (dnTR ${alpha}$ ) part of the fusion protein corresponds to Xenopus TR ${alpha}$ without the terminal helix 12 (3, 44). Both fusion proteins also contain an N-terminal FLAG tag. (B) The fusion proteins have reduced abilities to repress T3 target genes in the frog oocyte transcription system. The plasmids for the dual luciferase reporter system, where the firefly luciferase was driven by the T3-inducible Xenopus TRßA promoter (TRE-luc) and the Renilla luciferase was driven by the T3-independent TK promoter, were injected into the oocyte nucleus. The mRNAs for RXR and FLAG-tagged wild-type TR ${alpha}$ (F-TR), F-TR-l, or F-TR-s were injected into the cytoplasm. After overnight incubation in the absence of T3, the oocytes were harvested for luciferase assay, and the relative expression of the reporter firefly luciferase over the control Renilla luciferase was determined. The experiment was repeated with similar results. (C) Western blot showing the expression of the receptor proteins in the frog oocyte. Protein extracts from luciferase assay from lanes 1 to 4 as shown in panel B were separated by SDS-PAGE, and FLAG-tagged wild-type and mutant TRs were detected by anti-FLAG Western blotting. Molecular weight markers are shown, and the asterisk indicates a cross-reacting band.

To test the effectiveness of the N-CoR moieties to inhibit repressionby the TR moiety, we used the constructs F-TR-l and F-TR-s inthe oocyte transcription assay, where we can study gene regulationby TR in the context of chromatin (58). We expressed RXR togetherwith FLAG-tagged wild-type TR (F-TR), F-TR-l, or F-TR-s by microinjectingtheir mRNAs into the oocyte cytoplasm. As a reporter, we microinjectedinto the oocyte nucleus the plasmid containing the firefly luciferaseunder the control of the T3-inducible promoter of the XenopusTRßA gene (40) together with a control plasmid containingthe Renilla luciferase under the control of the T3-independentTK promoter. The luciferase assay of the oocyte extract afterovernight incubation showed that as expected, F-TR repressedbasal activity in the absence of ligand (Fig. 1B, lanes 1 and2). On the other hand, both F-TR-l and F-TR-s were fourfoldless effective in repressing the promoter (Fig. 1B, lanes 3and 4). Under the experimental conditions, similar or slightlyhigher levels of mutant TR were expressed (Fig. 1C), indicatingthat the N-CoR moiety in F-TR-l and F-TR-s fusion proteins wasable to reverse much of the repression caused by the TR moietyof the fusion protein.

To achieve gene activation in the absence of ligand, we fusedthe VP16 activation domain (hereafter VP16) to the N terminusof wild-type TR, F-TR-l, and F-TR-s to make F-VP-TR, F-dpTR-l,and F-dpTR (Fig. 2A). The ability of these fusion proteins toactivate gene expression in the absence of T3 compared to ligandedwild-type receptor was tested in the oocyte transcription system.Again, F-TR in the absence of ligand repressed transcriptionand in the presence of T3 activated transcription above basallevels (Fig. 2B, lanes 1 to 3). All three VP16 fusion proteinsinduced gene activation in the absence of ligand (Fig. 2B, lanes4 to 6). Because the expression level of the mutants was higherthan that of F-TR (Fig. 2C), quantitative comparisons betweenwild-type and mutant TRs are not possible. Nevertheless, geneupregulation by mutant TRs was well above basal transcriptionlevels, and the N-CoR moiety in F-dpTR-l and F-dpTR increasedthe activation ability compared to the F-VP-TR fusion protein.As F-dpTR appeared to be the most active TR, we chose it forthe rest of the studies.

View larger version (24K):
[in this window]
[in a new window]
FIG. 2. The transactivation domain of VP16 fused to wild-type TR, F-TR-l, and F-TR-s induces gene activation in the absence of T3. (A) Diagram of TR fusion proteins. The activation domain of VP16 was inserted between the FLAG tag and the N terminus of wild-type Xenopus TR ${alpha}$ , F-TR-l, and F-TR-s (Fig. 1) to produce F-VP-TR, F-dpTR-l, and F-dpTR, respectively. (B) The VP16 fusion proteins activate transcription in Xenopus oocytes. The oocytes were injected as described in the legend to Fig. 1, except with the indicated receptor mRNAs. Luciferase activity was measured after overnight incubation in the presence or absence of 10 nM T3. Note that all TR fusion proteins not only failed to repress the promoter but also induced transcription above basal levels in the absence of T3. The experiment was repeated with similar results. (C) Western blot showing the expression of the receptor proteins in the frog oocyte. Protein extracts from the luciferase assay from lanes 1 to 6 in panel B were separated by SDS-PAGE and detected by anti-FLAG Western blotting. For unknown reasons, the VP16 fusion proteins were expressed at higher levels than the FLAG-tagged wild-type TR when the same amount of mRNA was injected, but this result does not affect the conclusion that they activate the promoter in the absence of T3. Molecular weight markers are shown.

The results from the transcription assays suggested that endogenouscorepressors were at least partially excluded from binding tomutant TR containing the N-CoR moieties. To directly addressthis, we performed a coimmunoprecipitation experiment usinganti-FLAG antibody on lysates from oocytes injected with F-TRand RXR or F-dpTR and RXR incubated with or without T3. As expected,N-CoR was coimmunoprecipitated with F-TR only in the absenceof T3 (Fig. 3A, lanes 2 and 3). In contrast, there was littleN-CoR coimmunoprecipitated with F-dpTR in the presence or absenceof T3 (Fig. 3A, lanes 4 and 5), corroborating the transcriptiondata that N-CoR is largely excluded from functioning with themutant receptors.

View larger version (34K):
[in this window]
[in a new window]
FIG. 3. F-dpTR has drastically reduced ability to associate with N-CoR in vivo, correlating with its failure to induce histone deacetylation in the absence of T3, and competes with wild-type TR for transcription regulation. (A) Coimmunoprecipitation revealed dramatically weakened interaction between N-CoR and F-dpTR. Xenopus oocytes were injected with RXR and F-TR or F-dpTR and treated with or without T3. After overnight incubation, anti-FLAG antibodies were added to oocyte lysates to immunoprecipitate the receptor complexes. Immunoprecipitates were run on SDS-PAGE and blotted with anti-N-CoR or anti-FLAG antibodies. Note that N-CoR associated with wild-type TR in the absence but not the presence of T3 (lanes 2 and 3), whereas little N-CoR bound to F-dpTR independently of T3 (lanes 4 and 5). The weak signals in both lane 4 and lane 5 were due to residual binding of endogenous N-CoR to the TR moiety or immunoprecipitation background. Lane 1 is a no-injection control, and the asterisk indicates a cross-reacting band. (B) Coexpression of F-dpTR prevents histone deacetylation caused by wild-type TR in oocytes. Oocytes were injected with the reporter DNA and the indicated receptor mRNAs as described in the legend to Fig. 1 and incubated overnight in the absence of T3. ChIP was carried out on oocyte homogenates with antibodies against acetylated H3 or FLAG, and the ChIP DNA was detected by using a primer set flanking the TRE region in the reporter DNA. Note that the F-dpTR was bound to the TRE and prevented histone deacetylation (lane 3) as seen with wild-type TR alone (lane 2). (C) F-dpTR induces transcription even in the presence of unliganded wild-type TR. Oocytes were injected with reporter plasmids and the indicated receptor mRNAs as described in the legend to Fig. 1. Wild-type TR and RXR were injected at a concentration of 50 ng/µl, and dpTR was injected at 10, 50, and 200 ng/µl. After an overnight incubation in the absence of T3, the relative luciferase activity was measured. Note that the mutant receptor can completely counter the repression by wild-type TR and at higher mutant TR concentrations induce transcription above basal levels.

Because tadpoles, unlike oocytes, express endogenous TR, a finaltest for the feasibility of using F-dpTR in transgenesis wasto determine whether the mutant TRs could function in a dominantpositive fashion. We coinjected mRNAs encoding wild-type TRwithout FLAG tag and F-dpTR together with RXR and the reporterDNA into oocytes. After overnight incubation, we assayed luciferaseactivity and found that F-dpTR was able to activate the promoterin oocytes even in the presence of unliganded wild-type TR (Fig.3C). In addition, we subjected some of the oocytes to a ChIPassay with anti-acetylated histone H3 or anti-FLAG antibodies.PCR analysis of the precipitated DNA showed that the regionencompassing the thyroid hormone response element (TRE) of thereporter plasmid had a basal level of acetylated histones thatwas reduced in the presence of wild-type TR (Fig. 3B, lanes1 and 2). However, when F-dpTR was coinjected with wild-typeTR, the level of histone acetylation was restored (Fig. 3B,lane 3). The anti-FLAG ChIP assay showed that F-dpTR was boundto the TRE region when coexpressed with TR (Fig. 3B, lane 3).These results indicate that F-dpTR can bind T3-regulated promotersin vivo, even in the presence of wild-type TR, and functionas a dominant positive transcriptional activator.

Transgenic expression of F-dpTR induces metamorphic events. To investigate whether TR is sufficient to mediate the metamorphiceffects of T3, we introduced F-dpTR into developing animalsthrough transgenesis. We used a double promoter transgenesisconstruct (13). First, to prevent potential deleterious effectson embryonic development, we placed F-dpTR under the controlof a heat shock-inducible promoter, which has little basal expressionin the absence of heat shock (13). The other promoter in thetransgenesis construct was the ${gamma}$ -crystalline promoter drivingthe expression of GFP in the eye, allowing easy sorting of thetransgenic and nontransgenic sibling tadpoles when they areobserved under UV light (13).

Frog transgenesis was performed by restriction enzyme-mediatedintegration, and sibling transgenic and wild-type tadpoles werereared in methimazole to block synthesis of endogenous T3. Inthe presence of methimazole, tadpoles stop developing at thelimb bud stage but continue growing. Wild-type and transgenictadpoles were heat shocked at 33°C for 1 h per day for 4to 10 days. During heat shock treatments, which induced expressionof the transgene, the transgenic tadpoles stopped growing andbegan transforming, similar to metamorphic changes that arenormally induced by T3 treatment of premetamorphic tadpoles.For example, gill resorption and limb bud outgrowth and differentiationwere observed in transgenic tadpoles, similar to T3-treatedwild-type animals, whereas wild-type tadpoles without T3 treatmentretained the larval phenotype regardless of heat shock treatments(Fig. 4). Transgenic tadpoles survived about 10 days of heatshock, and thus further metamorphic progress, such as tail shorteningor forelimb emergence, was not observed. We observed no deleteriouseffects on wild-type tadpoles after weeks of heat shock, indicatingthat lethality in heat-shocked transgenic animals was due toincomplete and/or precocious tissue morphogenesis. This resultis also similar to that observed during T3-induced metamorphosis,where partially transformed tadpoles die before tail resorptiontakes place. Roughly 50% of the transgenic tadpoles (11 of 19)began to metamorphose after heat shock, a result most likelydue to differences in transgene expression levels, as seen inother studies (3, 44).

View larger version (84K):
[in this window]
[in a new window]
FIG. 4. Expression of F-dpTR in transgenic tadpoles initiates metamorphosis. Wild-type tadpoles (wt) and sibling tadpoles transgenic for F-dpTR (Tg) under the control of a heat shock-inducible promoter were reared together in methimazole to block endogenous T3 synthesis and were heat shocked daily for 8 days. For comparison, wild-type tadpoles were treated with 5 nM T3 for 3 days. At day 0, tadpoles were the same size and at the same stage (stage 52 to 54). Note that after 8 days of heat shock, transgenic and wild-type animals without T3 treatment were morphologically distinct, as highlighted by the gills (white brackets) and the hind limbs (white arrowheads), and the morphology of the transgenic animal looked similar to that of the T3-treated wild-type tadpoles. The color differences among animals are due to photography and do not affect the conclusion. This experiment was repeated more than 10 times with similar results.

In addition to external morphological changes in response toheat shock in the absence of T3, we examined internal metamorphicchanges. We focused on the intestine because it is one of thebest-studied organs and because intestinal remodeling involvesapoptosis of larval epithelial cells and proliferation of adultprecursor epithelial cells (22, 48). First, we examined celldeath using the TUNEL assay in heat-shocked wild-type and transgenictadpoles as well as wild-type tadpoles treated with T3. After4 days of heat shock, the intestinal epithelium of wild-typeanimals showed no TUNEL labeling, whereas the intestinal epitheliumof F-dpTR transgenic animals but not the muscle or connectivetissue exhibited numerous apoptotic nuclei, as in T3-treatedwild-type tadpoles (Fig. 5). Second, histological changes reflectingcell proliferation were examined by using methyl green-pyronineY staining (24). After 10 days of heat shock, the intestinesof wild-type animals were larval in character, i.e., the epitheliumhad a thin muscle layer and little connective tissue (Fig. 6A).In heat-shocked transgenic animals, remodeling was evident,as reflected by the thicker muscle and connective tissue layersand by nests of proliferating adult precursor cells (Fig. 6B).

View larger version (72K):
[in this window]
[in a new window]
FIG. 5. Expression of F-dpTR causes larval intestinal epithelial cells to undergo apoptosis. Wild-type and F-dpTR transgenic tadpoles were reared in methimazole to block metamorphosis and then heat shocked. A TUNEL assay to detect apoptotic cells was carried out on cross sections of intestine after 4 days of heat shock. For comparison, wild-type tadpoles were treated with 5 nM T3 for 3 days before the TUNEL assay. Wild-type tadpoles without T3 treatment showed no TUNEL labeling (A), whereas the intestine of F-dpTR transgenic and T3-treated tadpoles had many stained cells, indicating cell death (B and C, black arrows). Lack of staining in the sections performed without terminal TdT revealed the specificity of the reaction (D to F). Brackets indicate boundaries of muscles (m) and larval intestinal epithelium (epi) facing the intestinal lumen (not marked is a thin layer of connective tissue present between the muscles and epithelium). The color differences among sections were due to photography but do not affect the conclusion. This experiment was repeated three times with similar results. Bars, 25 µm.

View larger version (100K):
[in this window]
[in a new window]
FIG. 6. Expression of F-dpTR initiated the development of the adult intestine. Wild-type and sibling transgenic tadpoles were reared in methimazole to block metamorphosis and then heat shocked for 10 days. For comparison, wild-type tadpoles were treated with 5 nM T3 for 3 days. The intestines were isolated, and sections were stained with methyl green-pyronine Y. Note that wild-type intestine remained typical of tadpole intestine, as seen by the presence of a thin muscle layer, little connective tissue, and no adult intestinal precursor cells (A). Histology of the F-dpTR and T3-treated wild-type intestines revealed increases in muscle layer thickness, proliferation of connective tissue, and appearance of adult epithelial precursor cells (B and C). Staining intensity differed among the sections due to staining conditions but does not affect the conclusions. This experiment was repeated four times with similar results. Bars, 25 µm.

F-dnTR induces metamorphosis by activating the T3-dependent gene regulation pathway through direct binding to endogenous T3 target genes. To characterize the molecular nature of the metamorphosis inducedby F-dpTR, we examined the expression of genes regulated duringnatural and T3-induced metamorphosis. First, we isolated RNAfrom whole bodies of early premetamorphic tadpoles (stage 45)after 3 days of heat shock and analyzed the expression of TH/bZIP,a direct, ubiquitous T3-induced gene (25). As expected, it wasnot expressed in heat-shocked, wild-type premetamorphic tadpolesbut was expressed during metamorphosis (stage 62) (Fig. 7A).TH/bZIP was also expressed in all four premetamorphic transgenictadpoles after heat shock (Fig. 7A). Interestingly, each ofthe four transgenic animals from this experiment exhibited adifferent level of TH/bZIP induction, consistent with variationin the ability to elicit a phenotypic response by heat shockin transgenic tadpoles. Because all animals in this study wereF₀ generated from the transgenic procedure directly, individualtransgenic animals likely differ in transgene copy number andinsertion site.

View larger version (53K):
[in this window]
[in a new window]
FIG. 7. Transgenic expression of F-dpTR leads to the activation of known T3-regulated genes in the absence of T3. (A) The T3 response gene, TH/bZIP, is upregulated in whole bodies of early premetamorphic tadpoles (stage 45). Total RNA was extracted from whole bodies of individual stage (St.) 45 wild-type (wt) and transgenic (Tg) tadpoles after 3 days of heat shock and subjected to RT-PCR. No expression was detected in wild-type tadpoles, whereas variable levels of upregulation were observed in four different F-dpTR transgenic tadpoles, and this range of expression encompassed that seen in the tails of wild-type tadpoles at the climax of metamorphosis (stage 62). The gene rpl8 was used as an RNA-loading control. (B) F-dpTR regulates gene expression as in T3-treated wild-type tadpoles. Wild-type and F-dpTR transgenic tadpoles blocked at stage 52 by methimazole were heat shocked for 4 days, and RT-PCR analysis was carried out on RNA isolated from tails and intestines for the expression of T3 response genes, with the T3-independent gene rpl8 as an internal control. For comparison, the same analysis was carried out on intestines from wild-type tadpoles treated with 5 nM T3 for 3 days. The following genes were analyzed: TRß and TH/bZIP, ubiquitously expressed, direct T3 response genes; stromelysin-3 (ST3), a fibroblast-specific, direct T3 response gene; sonic hedgehog (xhh), an intestinal, epithelium-specific, direct T3 response gene; and bone morphogenic protein 4 (BMP-4), a fibroblast-specific, late upregulated gene. The experiments were repeated with similar results for six transgenic tadpoles showing a phenotype similar to that seen in Fig. 4.

To examine gene regulation in more detail, RNA from tails andintestines of tadpoles whose metamorphosis was blocked withmethimazole was analyzed for several genes with varying tissueexpression profiles and T3 dependence (Fig. 7B). The directT3 response genes (i.e., genes containing TREs), TRßand TH/bZIP (14, 40), were upregulated in heat-shocked transgenictadpoles but not in wild-type animals. Another direct T3 responsegene, stromelysin-3 (37), and the late response gene, bone morphogenicprotein 4 (BMP-4) (23), which are expressed in fibroblasts,were upregulated in tails and intestines only in transgenictadpoles subjected to heat shock treatments. Furthermore, theintestine-specific, direct T3 response gene, sonic hedgehog(xhh) (49), was induced by the F-dpTR transgene after heat shock.Interestingly, F-dpTR did not induce xhh in the tail, as observedduring T3 treatment (49). These results are in perfect agreementwith those observed during natural and T3-induced metamorphosis(Fig. 7B) (14, 23, 37, 40, 49), indicating that F-dpTR specificallyinduced the endogenous T3-dependent gene regulation pathwayto initiate metamorphosis in transgenic animals.

One advantage of using frog metamorphosis is the ability tostudy directly the mechanisms of nuclear receptor function atthe molecular level in vivo during development by using a ChIPassay. First, we confirmed the expression of the transgene afterheat shock in transgenic tadpoles by using anti-FLAG Westernblotting, which recognizes the FLAG tag of F-dpTR (Fig. 8A).Next, we used a ChIP assay with the anti-FLAG antibodies todetermine whether F-dpTR regulates T3 response genes throughdirect binding to known TREs. For this purpose, we focused ontwo T3-regulated promoters, TRß and TH/bZIP (14, 40),using nuclei isolated from the tails and intestines of heat-shockedwild-type and transgenic tadpoles. Quantitative PCR analysisshowed that the transgenic F-dpTR bound to the TREs in bothtissues at both promoters in the transgenic animals and thatonly low levels of nonspecific binding were observed in wild-typeanimals, as expected (Fig. 8B and C). Furthermore, this bindingby F-dpTR was specific to the TREs of the two T3-dependent genesin the transgenic but not the wild-type animals, because inthe transgenic animals, F-dpTR did not bind to TRßexon 5, which lacks a TRE (Fig. 8D).

View larger version (27K):
[in this window]
[in a new window]
FIG. 8. ChIP assay using quantitative PCR shows that F-dpTR binds specifically to TRE of T3 target promoters in transgenic animals. After 4 days of heat shock, proteins from the head region were isolated from wild-type and transgenic tadpoles for Western blotting, and nuclei from tails and intestines were isolated for the ChIP assay. (A) Western blotting revealed the expression of F-dpTR in transgenic but not wild-type tadpoles after heat shock. The proteins were separated by SDS-PAGE and detected with anti-FLAG antibody. (B to D) ChIP from isolated nuclei with anti-FLAG antibodies followed by quantitative PCR shows that F-dpTR binds specifically to TRE of T3 target promoters in transgenic animals. Quantitative PCR was carried out on purified ChIP DNA. The anti-FLAG ChIP showed background levels of 0.5% or less of input DNA in wild-type (wt) tadpoles, compared to 1 to 10% in transgenic (Tg) tadpoles for the T3 response genes TRß and TH/bZIP (B and C). Less than 0.5% of input DNA was immunoprecipitated in both wild-type and transgenic tadpoles for TRß exon 5 in the coding region (D), demonstrating the specificity of F-dpTR binding to chromatin targets in vivo. These experiments were repeated two to four times with similar results.

Because T3-induced transcription of target genes is associatedwith increases in local histone acetylation, we examined histoneacetylation levels at the TRE regions of TRß and TH/bZIP.A ChIP assay with anti-acetylated histone H4 showed that histoneacetylation increased by two- to fourfold, although for unknownreasons, there was considerable variation in the intestine (Fig.9). Thus, F-dpTR may also utilize increased histone acetylationto activate target gene expression, similar to T3-bound TR.

View larger version (13K):
[in this window]
[in a new window]
FIG. 9. Histone acetylation levels at T3 target promoters in tails and intestines increase after heat shock in dpTR transgenic tadpoles. Wild-type and transgenic tadpoles were heat shocked for between 4 and 10 days before chromatin was isolated from intestines and tails for anti-acetyl-H4 ChIP. Acetylation levels were higher at the TRß (A) and TH/bZIP (B) promoters in both the intestines and the tails. The error bars are from duplicate PCR samples of a representative ChIP experiment. The values below the bars represent the average ± standard error for the increase (n-fold) in acetylation levels between wild-type and dpTR transgenic tadpoles from three to five independent ChIP experiments.

DISCUSSION

Top

Discussion

Thyroid hormone regulation of postembryonic development occursin all vertebrates, and thyroid hormone deficiency has profoundeffects during human gestation, leading to cretinism, a diseasecharacterized by short stature and mental retardation (20).Natural TR mutations are the basis for most incidences of resistanceto thyroid hormones syndrome, where patients exhibit a widevariety of abnormalities (17). Furthermore, these diseases andthe high occurrence of hypothyroidism due to various etiologieshave led to broad-based research on the molecular mechanismsof TR in transducing the T3 signal. In vitro cell culture studieshave formed the basis for our molecular understanding of TRfunction (5, 26, 35, 39, 41, 63). Genetic studies of mice, comprisingknock-ins of mutant TRs and knockouts of TR isoforms, revealmultiple developmental roles for TRs (10). However, due to thelack of a model system amenable to molecular studies of generegulation in development, knowledge of the role of TR in developmentis limited. Similar developmental events dependent on thyroidhormone occur both in mammals and in frogs (50), such as intestineremodeling during weaning and metamorphosis (38, 48, 54). Therefore,molecular studies of free-living tadpoles undergoing T3-dependentmetamorphosis are valuable additions to our basic knowledgeof the role of TR in vertebrate development.

Frog metamorphosis is entirely dependent upon T3 and involvescell proliferation and apoptosis in nearly every organ in thetadpole (45). In the absence of T3, tadpoles continue to growto a large size and fail to undergo any morphological changes.In addition, if premetamorphic tadpoles are treated with exogenousT3 before endogenous T3 production, precocious metamorphosisensues. These effects of T3 are believed to be mediated by TR.Recent molecular and genetic studies in vivo have provided somesupport for this idea. A dominant negative form of TR with adeletion of the C-terminal activation domain, when overexpressedin transgenic tadpoles, blocks T3-induced metamorphosis andfails to release corepressors in the presence of ligand, therebydemonstrating that TR function is necessary for transformation(3). These studies, however, leave one important question unanswered,i.e., whether TR is sufficient to mediate all the developmentaleffects of T3, especially considering the well-documented, nongenomiceffects of T3 (8).

Our present data from a dominant positive TR suggest that thegenomic action of TR is sufficient for metamorphosis, at leastfor all the changes analyzed in this study. Transgenic tadpolesbearing F-dpTR under heat shock-inducible control undergo externaland internal morphological changes indistinguishable from thoseof wild-type tadpoles treated with T3 (45). In addition, T3response genes with or without TRE-containing promoters expressedubiquitously or in specific tissues all responded to heat shockinduction of F-dpTR, a result similar to that seen when T3 isadded to wild-type animals (46). Furthermore, these actionsof F-dpTR took place via direct, in vivo binding to chromatin,i.e., F-dpTR bound to TRE-containing promoters and failed tobind coding regions. Similar to tadpoles undergoing precociousinduction by T3, transgenic tadpoles expressing F-dpTR diedbefore completing metamorphosis. These identical morphologicaland gene expression changes observed in F-dpTR transgenic animalsand T3-treated premetamorphic tadpoles suggest that TR activationis sufficient for these events and that TR is the sole mediatorof the T3 signal in these processes.

As mentioned above, T3 is known to affect postembryonic developmentin mammals. Genetic analyses of transgenic and knockout micehave resulted in various phenotypes that are not readily explainableat the molecular level based on knowledge from cell culturestudies (10-12, 15, 16, 57). For example, mice lacking TR ${alpha}$ , TRß,or both have fewer developmental defects than those lackingT3. In addition, mice with dominant negative TRß (dnTRß),which mimics resistance to thyroid hormone syndrome in humans,and dnTR ${alpha}$ show different phenotypes (19, 29, 30, 51, 65), andin some cases, a number of known T3 response genes in differenttissues were found to have different expression levels in micewith dnTRß and dnTR ${alpha}$ and in wild-type mice (29, 30).However, the results from mice with dnTR ${alpha}$ and dnTRßwere quite different and could not be explained simply basedon the regulation mechanisms obtained from in vitro or cellculture studies. A number of potential mechanisms could explainthe different phenotypes and differences in gene expressionfound in the various TR knockout and mutant animals. These includepotential nongenomic effects of T3, different expression profilesand levels of different receptors, distinct isoform-dependentin vivo signaling pathways, and/or indirect effects throughaltered circulating T3 titer.

Given the similarity between postembryonic development in mammalsand frog metamorphosis (2, 45, 50), our results here suggestthat the effects of T3 during mammalian postembryonic developmentare also mediated by TR and that the nongenomic action of T3is not likely to have a major impact during this period of mammaliandevelopment. On the other hand, it is unclear whether the variousdevelopmental phenotypes observed in TR knockout and transgenicmice are caused directly or indirectly by TR-mediated gene regulationin the affected tissues and/or organs. The methodologies andreagents reported here should be useful for future studies toaddress such issues, especially in probing the role of tissue-tissueinteractions in TR-dependent developmental events. For example,epithelium-mesenchyme interactions are of critical importancein organogenesis and postembryonic development. The use of dpTRin concert with tissue-specific promoters allows the opportunityto "activate" one tissue layer of interest a time, thereby allowingone to determine whether cell-cell interaction plays a rolein the corresponding developmental process.

ACKNOWLEDGMENTS

We thank David Johns for the gift of plasmid.

FOOTNOTES

* Corresponding author. Mailing address: Building 18T, Rm 106, LGRD, NICHD, NIH, Bethesda, MD, 20892. Phone: (301) 402-1004. Fax: (301) 402-1323. E-mail: shi{at}helix.nih.gov.

D.R.B. and A.T. contributed equally to this work.

REFERENCES

Top