Thyroid-hormone-dependent development of the neuroretina has principally been described in amphibia. Here, we show by in situ hybridisation that mRNAs coding for three distinct thyroid hormone receptors (TRs), TRα and two TRβ variants, are differentially expressed during chick retinal development. We isolated a cDNA for a novel N-terminal variant of chick TRβ (cTRβ) that is predominantly expressed in retinal development. Interestingly, in its N-terminal A/B domain cTRβ2 is 70% homologous to the rat pituitaryspecific TRβ. Expression of cTRβ mRNA was high at embryonic day 6 (Eli) in the retinal outer nuclear layer (ONL) and decreased to low levels at hatching. Mrna for the previously described chick β receptor, cTRβ0, was expressed at low levels in both the ONL and the inner nuclear layer (INL) after E10. In contrast, cTRa expression occurred in the ONL, INL and ganglion cell layer at intermediate and later stages. Finally, cTRβ2 confers a stronger irans-activation of reporter gene transcription than cTRβ0. The distinctive kinetics and localisation of TRa and β gene expression suggest cell- and stage-specific functions for TRs, both individually and in combinations, in chick neuroretinal development.

It is well established that thyroid hormones (triiodothyronine, T3; thyroxine, T4) are essential for development of the vertebrate central nervous system (CNS) (McCarrison, 1908; Gesell et al., 1936; Smith et al., 1957). Study of cretinism in man and hypothyrodism in animal models has pointed to a critical period in later embryogenesis and early neonatal life during which hormone is required (Legrand, 1984; Dussault and Ruel, 1987). More recently, it has been shown that vertebrates produce various forms of T3 receptors (TRs) from two related genes, TR α - and β (Sap et al., 1986; Weinberger et al., 1986; Thompson et al., 1987; Benbrook and Pfahl, 1987; Izumo and Mahdavi, 1988; Murray et al., 1988; Forrest et al., 1990; Yaoita et al., 1990), and that TRs belong to the nuclear hormone receptor family which act as ligand-dependent transcription factors (for review, see Green and Chambón, 1988) Differences in the structure of the N terminus and DNA-binding domain (Murray et al., 1988; Forrest et al., 1990) suggest potential differences in the transactivating properties for TRβ- and β. Furthermore, it has been shown that the TRβ- and βgenes are differentially expressed during chick and rat brain development (Bradley et al., 1989; Forrest et al., 1990; Strait et al., 1990), with TRβ mRNA being abruptly up-regulated during the known period of T3 requirement, whereas TRa-mRNA increases gradually, suggesting distinct developmental functions for TRα- and TRβ (Forrest et al., 1991).

Most reports of thyroid hormone requirement in development of the nervous system pertain to the brain. However, we also recently demonstrated expression of TRβ and βgenes in the embryonic chick eye (Forrest et al., 1990). Thyroid hormone is known to trigger major reorganisation and development of the amphibian eye during metamorphosis (for review, see Hoskins, 1990) but few investigations have addressed the role of this hormone in avian or mammalian eye development. In the present report, we demonstrate by in situ hybridisation analysis that, in the chick embryo, specific retinal layers express unique and changing patterns of mRNA for α and β thyroid hormone receptors during development. Furthermore, we have cloned a cDNA encoding a TRβ N-terminal variant that is selectively and developmentally expressed in the embryonic retina. Our results indicate specific developmental functions for TRs in retinal development and we discuss a model for developmental control involving cell- and stagespecific functions for combinations of different TRs.

RNA preparation and analysis

Poly(A)-selected RNA was prepared from tissues of brown Leghorn chick embryos and analysed by northern blot hybridisation using probes specific for cTRα and β and glyceraldehyde-3-phosphate dehydrogenase as previously described (Forrest et al., 1990). The probes used in RNAse protection assays were transcribed in vitro from an XhoI-linearized subclone of pSK-β2 and hybridised to mRNA as described (Forrest et al., 1990).

Isolation of cTRβ2 cDNA clones

First strand cDNA enriched for cTRβ sequences was synthesised from 2μg of poly (A)-selected RNA from eyes of 9 day chick embryos using a cTRβ-specific oligonucleotide primer (nucleotides 1179-1225 in Forrest et al., 1990). Second strand cDNA was made and blunt-ended with T4 DNA polymerase by standard methods (Forrest et al., 1990). 3’ ends of 0.5 μg of this product were G-tai]ed using dGTP and terminal deoxynucleotidyl transferase as described (Innis et al., 1990). PCR amplified products representing 5’ ends of cTRβ mRNAs were obtained using a specific antisense cTRjS primer (nucleotides 520–546 in Forrest et al., 1990) and an upstream primer complementary to the G-tailed end that contained an EcoRI site (5’-TCAATGAATTC(C)l5-3’). A second PCR amplification was performed on 10 βl of the first PCR reaction products using an internal downstream primer (nucleotides 486-512 in Forrest et al., 1990) and the same upstream primer as in the first reaction. The PCR products were digested with EcoRI and PstI (a Pstl site is located internally in the common cTRβ sequence; nucleotide 419 in Forrest et al., 1990), and cloned into the Bluescript plasmid Positive clones were identified by colony hybridisation and were sequenced by the dideoxy method. The novel cTRβ2 sequence obtained facilitated a second, confirmatory PCR amplification from the original E9 eye cDNA, this time using a cTRβ2-specific internal primer (nucleotides 235–281 in Fig. 1A). These products were digested with EcoRI and Dral (a Dral site is located in the cTRβ2 sequence at nucleotide 229 in Fig. 1A), subcloned and sequenced to verify the cTRβ2 5’ sequence.

Fig. 1.

Structure of the chicken TRβ2 N-terminal variant. (A) Nucleotide and predicted amino acid sequence of cTRβ2 variant N terminus. Two in-frame stop codons upstream of the indicated initiation codon are underlined. The cTRβ0 and β 2sequences are identical downstream of nucleotide 347. Amino acid differences in the rat TRβ2 are shown beneath the chicken sequence (note that the published rTRβ2 N terminus extends 38 amino acids beyond the start of the cTRβ2 reading frame). The dashed line indicates an amino acid insertion in cVRμl relative to rTRβ2. The arrows at the start of the cTRβ2 sequence indicate 5’ ends of different cDNA clones obtained. (B) Comparison of chick TR β0 and β N terminal variants and rat TRβ2. DNA- and T3-binding domains are indicated and amino acid numbering is shawm below each receptor. Percentage similarities in protein sequences are shown. (C) In vitro translated cTRβ2 protein. [55S]methionine-labelled products were analysed by SDS-PAGE and autoradiography. Lanes: M, marker proteins (relative molecular masses in thousands); β2, product of re-constructed cTR β2 cDNA clone (see Materials and methods); p0, cTRβproduct; 0, unprogrammed lysate.

Fig. 1.

Structure of the chicken TRβ2 N-terminal variant. (A) Nucleotide and predicted amino acid sequence of cTRβ2 variant N terminus. Two in-frame stop codons upstream of the indicated initiation codon are underlined. The cTRβ0 and β 2sequences are identical downstream of nucleotide 347. Amino acid differences in the rat TRβ2 are shown beneath the chicken sequence (note that the published rTRβ2 N terminus extends 38 amino acids beyond the start of the cTRβ2 reading frame). The dashed line indicates an amino acid insertion in cVRμl relative to rTRβ2. The arrows at the start of the cTRβ2 sequence indicate 5’ ends of different cDNA clones obtained. (B) Comparison of chick TR β0 and β N terminal variants and rat TRβ2. DNA- and T3-binding domains are indicated and amino acid numbering is shawm below each receptor. Percentage similarities in protein sequences are shown. (C) In vitro translated cTRβ2 protein. [55S]methionine-labelled products were analysed by SDS-PAGE and autoradiography. Lanes: M, marker proteins (relative molecular masses in thousands); β2, product of re-constructed cTR β2 cDNA clone (see Materials and methods); p0, cTRβproduct; 0, unprogrammed lysate.

In vitro expression of cTRβ2 protein

An improved translational initiation site conforming to a Kozak consensus sequence was introduced into cTRβ2 and β0 by PCR mediated mutagenesis (to be reported elsewhere). The full open reading frame of cTRβ2 was then re-constructed by substituting an EcoRI-EcoRV fragment from cTRβ2 into a plasmid containing cTRβ0 (Forrest et al., 1990). Recombinants were verified by DNA sequencing and protein products analysed by in vitro transcription and translation as described (Forrest et al., 1990).

In situ hybridisation analysis

Sections (8 or 10 μm thick) of embryonic brown Leghorn chicks were prepared and hybridised in situ with oligonucleotide probes which were 3’-end-labelled with terminal deoxynucleotidyl transferase and [α35S] dATP as described (Forrest et al., 1991). Antisense oligonucleotides specific for the unique 5’ sequences of cTRβ0 and β2 mRNAs were derived from nucleotides 22–67 of cTRjβ0(Forrest et al., 1990) and nucleotides 109-156 of cTRβ2 (Fig. 1A). Antisense oligonucleotides specific forcTRo and the common 3’ region ofcTRβ0 and β2 mRNAs were previously described; the control probe for background hybridisation levels was the complementary sense sequence of the common cTRβ probe (Forrest et al., 1991). The specificity of all oligonucleotide probes was first established by northern blot hybridisation, and then tested by in situ hybridisation where optimal wash conditions were determined to be as reported (Forrest el al., 1991) with the modification that the last wash for 20 min at 55°C was in 0.18 × SSC (instead of 0.3 × SSC) and included β -mercaptoethanol and N-lauryl sarcosine. Autoradiography employed Amersham (Tmax film or Kodak NTB2 emulsion, and sections were stained with cresyl violet for histological examination as described (Forrest et al., 1991).

Tissue culture and CAT-assays

CV-1 cells were seeded at 1×106 cells per 10 cm dish in DMEM containing 8% fetal calf serum and antibiotics. Transfections were performed as described (Sap et al., 1990). The cells were incubated for 40 hours in T3-depleted medium to which 0 or 25 nM T3 had been added. CAT assays (90 minute reaction time) were performed using standard procedures. The pSG5 plasmid used for the expression of cTRβ0 and β2 is driven by the SV 40 early promoter (Green et al., 1988). Eyes for organ culture were dissected from E6 chick embryos, rinsed in PBS and incubated for 3 days in DMEM containing antibiotics and 10% feta] calf serum that was depleted of T3. Cultures were maintained in the absence or presence of added T3 at a final concentration of 25 nM, and medium was changed every 24 hours.

Cloning an eye-specific variant TRβ cDNA

We have previously shown that TRα- and β genes are expressed during chick eye development (Forrest et al., 1990). Furthermore, we identified an mRNA that is predominantly expressed in the early developing chick eye and that coded for a putative N-terminal variant of TR$. To compare the roles of different TRs in eye development, we first obtained cDNA clones of the unknown 5’ sequences of the variant TR/l mRNA using PCR. Complementary DNA enriched for TRβ sequences was made from embryonic day 9 (E9) eye RNA, and 5’ sequences were amplified from this using the anchored PCR approach (see Materials and methods). Twenty of the longest clones obtained were sequenced and shown to encode a variant receptor (designated cTRβ2), which possessed an N-terminal A/B domain of 107 amino acids (Fig. 1A), distinct from the 2 amino acids in this region of the previously characterised chick TRβ cDNA (now called cTRβ0). The sequenced part of the common region was identical to that of cTRβ0 from nucleotide 346 (Fig. 1A). Since previous analyses had indicated that the C-terminal coding region of TRβ mRNA was invariant in development of 14 different chick tissues (Forrest et al., 1990), we were confident that the new variant receptor differed only in the N terminus from cTR)30. The N-terminal region of cTRβ2 was found to have 70% amino acid homology to the N-terminal domain of the rat TRβ2, isolated from a pituitary cell line (Hodin et al.,1989) However, cTRβ is 38 residues shorter and also carries an amino acid (Glu) insertion at codon 138 as compared to the rat sequence (Fig. 1A,B). This raised the issue whether or not the cDNA clones that we obtained contain truncated versions of the cTRβ2 coding sequence. Several observations argue against this: (i) 3 independent cDNA clones had identical stop codons in the 5’-untranslated sequence, in frame with the proposed initiating ATG; (ii) the mouse TRβ2 is also about 38 amino acids shorter than the rTRβ2 in the N-terminus (Wood et al., 1991; and (iii) our screening of ∽70 cTRβ2 clones revealed none that was longer than the sequence shown in Fig. 1A, instead, 7 of the 10 sequenced clones extended to within the first 20 nucleotides.

The deduced cTRβ2 open reading frame of 476 amino acid residues was reconstructed by replacing the TRβ0 5’ coding sequence for that of TRβ2, and the protein product characterised by in vitro transcription and translation. Fig. 1C shows the size of the product upon SDS-PAGE (61×103) to be in the range of the predicted relative molecular mass (55×103).

The new chick TRβwas designated as cTRβ2 due to its general homology’ to rat TRβ2. No chick TRβ related to the mammalian TRβ1 (Weinberger et al., 1986; Murray et al., 1988) has been identified. We have therefore renamed the original chick TRβ (Forrest et al., 1990) cTRβ since its short N-terminal region (2aa) is unrelated to that of the other TRs.

Developmental switching of expression of TRp N-terminal variant mRNAs in retina

We next compared expression of mRNAs for both cTRβ2 and β0 in chick eye development by RNAse protection assay, since we had shown previously that both mRNAs were present in embryonic eye at E16 (Forrest et al., 1990). Surprisingly, expression of mRNA was relatively high as early as E6 but displayed a progressive decrease at developmental stages later than E16 to minimal levels after hatching (Fig. 2). The level of β2 mRNA appeared to be generally low during development. However, due to the relative decline of β2 mRNA levels, the ratio of β0: β2 mRNA increased at later stages. Thus after E16, β0 mRNA became relatively more abundant and this pattern was maintained in the hatched chick, at least until day 19. Additional analyses with independent probes specific for β0 and β2mRNAs confirmed the above results (data not shown).

Fig. 2.

RNAse protection analysis of developmental switching of expression of mRNAs for cTRβ variants in eye. The probe corresponding to the 5’ sequences of cTRβ mRNA has been described previously (Forrest et a]., 1990). DNA, DNA-binding domain; the putative splice point after the second codon of TRβO is shown by a dotted vertical line. 5’UT, 5’ untranslated sequence. The sizes of expected protected bands representing mRNAs for TR/SO (360) and variant TRβ2 (333) are indicated in nucleotides. Lanes: M, HaeIII -digested pBR322 markers, tRNA, probe hybridised with carrier tRNA and no sample RNA; E6 -E19, P19 indicate embryonic (E) and post-hatched (P) chick ages in days for different RNA samples.

Fig. 2.

RNAse protection analysis of developmental switching of expression of mRNAs for cTRβ variants in eye. The probe corresponding to the 5’ sequences of cTRβ mRNA has been described previously (Forrest et a]., 1990). DNA, DNA-binding domain; the putative splice point after the second codon of TRβO is shown by a dotted vertical line. 5’UT, 5’ untranslated sequence. The sizes of expected protected bands representing mRNAs for TR/SO (360) and variant TRβ2 (333) are indicated in nucleotides. Lanes: M, HaeIII -digested pBR322 markers, tRNA, probe hybridised with carrier tRNA and no sample RNA; E6 -E19, P19 indicate embryonic (E) and post-hatched (P) chick ages in days for different RNA samples.

Developmental expression of cTRβ2 mRNA in outer nuclear layer of the retina

In considering functions for different receptors, it was important to investigate whether both TRβ variants were expressed in the same or different cells. We used oligonucleotide probes specific for the 5’ variant regions of cTRβ 0 and β2 mRNAs to localise expression of the respective mRNAs in developing brain and eye by in situ hybridisation (Fig. 3). First, the developmental and tissue specificity of these probes was established by northern blot analysis essentially as described previously, with β2 mRNA showing the expected restriction in expression (Forrest et al., 1991). Both β 0 and β2 probes detected a 7 kb mRNA in embryonic eye, indicating that the 5’ sequence variation did not greatly alter the TRβ mRNA sizes (data not shown).

Fig. 3.

In situ hybridisation analysis of cTR β2 and β 0 mRNA expression in the developing chick nervous system. (A) The origin of antisense oligonucleotide probes unique to TRβ2 and β 0 mRNAs (β 2 and β 0) and common to both TRβ mRNAs (β), and of the sense strand probe used as a measure of background hybridisation (see Materials and methods). Receptor structures are indicated schematically relative to cDNA clones; DNA and T3, DNA- and T3-binding domains respectively. The vertical dashed line indicates the point upstream of which TRβ2 and β sequences diverge. (B) Comparison of cTRβand β O mRNA expression in the E10 brain and retina. Film autoradiographs (2 week exposures) are of adjacent parasagittal head sections, with a line drawing of the sections for anatomical identification shown in the right-hand panel; CBL, cerebellum; RET, retina; TEL, telencephalon. Signals appear white. White arrowheads in autoradiographs indicate signals in the ventricular epithelium of the cerebellum. (C) Higher magnification localisation of cTR β2 mRNA expression in the retinal outer nuclear layer, shown for stages E10 and E7. Left-hand panels show cresyl-violet-stained sections for histological identification and right-hand panels dark-field views of corresponding adjacent sections hybridised with β 2 and βO probes and exposed to photographic emulsion for 13 weeks. Signals appear as white grains. P, pigment layer (note that this layer produces artefactua) signa) under dark-field microscopy, but not in the film autoradiographs in Fig. 3B above); ON, outer nuclear layer; IN, inner nuclear layer; GC, ganglion cell layer. White arrowheads indicate the ON in the darkfield views.

Fig. 3.

In situ hybridisation analysis of cTR β2 and β 0 mRNA expression in the developing chick nervous system. (A) The origin of antisense oligonucleotide probes unique to TRβ2 and β 0 mRNAs (β 2 and β 0) and common to both TRβ mRNAs (β), and of the sense strand probe used as a measure of background hybridisation (see Materials and methods). Receptor structures are indicated schematically relative to cDNA clones; DNA and T3, DNA- and T3-binding domains respectively. The vertical dashed line indicates the point upstream of which TRβ2 and β sequences diverge. (B) Comparison of cTRβand β O mRNA expression in the E10 brain and retina. Film autoradiographs (2 week exposures) are of adjacent parasagittal head sections, with a line drawing of the sections for anatomical identification shown in the right-hand panel; CBL, cerebellum; RET, retina; TEL, telencephalon. Signals appear white. White arrowheads in autoradiographs indicate signals in the ventricular epithelium of the cerebellum. (C) Higher magnification localisation of cTR β2 mRNA expression in the retinal outer nuclear layer, shown for stages E10 and E7. Left-hand panels show cresyl-violet-stained sections for histological identification and right-hand panels dark-field views of corresponding adjacent sections hybridised with β 2 and βO probes and exposed to photographic emulsion for 13 weeks. Signals appear as white grains. P, pigment layer (note that this layer produces artefactua) signa) under dark-field microscopy, but not in the film autoradiographs in Fig. 3B above); ON, outer nuclear layer; IN, inner nuclear layer; GC, ganglion cell layer. White arrowheads indicate the ON in the darkfield views.

Fig. 3A shows that, in the brain at E10, cTRβ 0 mRNA was mainly detected in the ventricular epithelium of the early cerebellar outgrowth, but also at very low levels in the retina. The cTRβ signal displayed a reciprocal pattern: expression was readily detected only in the retina, but with low’ signals also in the ventricular epithelium of the cerebellum. Expression of neither TRβ mRNA was obvious elsewhere in the CNS at E10, as expected from previous studies (Forrest et al., 1991). An antisense probe complementary to the common 3’ coding region of both mRNAs detected signals of about equal intensity in both the retina and cerebellar epithelium, confirming the results with the unique probes. The sense strand control probe produced only very low background in all areas.

Dark-field analysis at higher magnification demonstrated that, in the retina at E10, cTR; β 2 mRNA was localised in the outer nuclear layer (ONL), the site of the developing photoreceptor cells (Fig. 3C). Expression was not detected in the inner nuclear layer (INL) or ganglion cell layer (GC). The cTRβ 0 signal was coincident with that of rβ 2 in the ONL, but was at considerably lower levels, being only slightly over background. Expression of β 2 mRNA showed the same restriction to the ONL from early stages (E4, E7) until E15 (not shown). Similarly, in the cerebellar outgrowth both mRNAs were apparently co-expressed in the differentiating epithelium, although in this case β) mRNA always predominated. Thus, in both the retina and cerebellum, both mRNAs were co-expressed, but with distinct differences in the ratio of β0: β 2 mRNAs depending on the tissue and developmental stage.

Differential expression of TRα and β genes in the developing retina

Trans-activation by both TRβ - and β through a common T3RE has been demonstrated experimentally (Graupner et al., 1989; M.S., D F., B.V., unpublished data), suggesting possible physiological interactions; we therefore analysed expression of the TRα gene in the retina. Fig. 4A comparesTRaand flmRNA expression in the E9 brain and eye by in situ hybridisation. TRa mRNA was widely expressed in brain, as shown previously (Forrest et al., 1991), and was also apparent in the retina, but in a clearly different pattern than that of TRα. Like TRβ, TRa mRNA was present in the ONL but was also in the INL and GC (shown for stage E15 in Fig. 4B). However, at earlier stages (E5, Fig. 4B), TRα signal was scarcely detectable in any layer of the retina. In contrast, at E5, TRβ signals were clear, and were clustered over cells of the ONL. Later in development, at E15, TRβ mRNA was also detected in the INL as well as the ONL. After E15 the β signal represents an increasing predominance of β0 relative to β2 mRNA as shown by RNAse protection analysis (Fig. 2). It should be noted that TRamRNA detected here is likely to code for normal T3-binding receptors, since the non-T3-binding forms of TRa found in mammals (Izumo and Mahdavi, 1988; Miyajtma et al., 1989) apparently do not occur in the chicken (Forrest et al.,1990)

Fig. 4.

Developmental expression of TR α and β mRNAs in retina and brain. (A) Film autoradiographs of parasagittal sections through E9 brain and eye hybridised with probes specific for cTR α (Forrest et al., 1991), β (common probe, β, in Fig. 3A), and control probe, C (Fig. 3A). Sections lie slightly closer to the head mid line than those in Fig. 3: CBL, cerebellum; TEL, telencephalon. (B) High magnification light field analysis in the developing retinal layers at E5 and E15. Sections were hybridised and exposed to emulsion as in Fig. 3C. Signals in light-field view appear as black silver grains over stained cells (again note that the pigment produces artefactual signals). Abbreviations are as in Fig. 3C. The black arrowhead indicates typical clustering of TRβsignals over ON cells in the E5 retina.

Fig. 4.

Developmental expression of TR α and β mRNAs in retina and brain. (A) Film autoradiographs of parasagittal sections through E9 brain and eye hybridised with probes specific for cTR α (Forrest et al., 1991), β (common probe, β, in Fig. 3A), and control probe, C (Fig. 3A). Sections lie slightly closer to the head mid line than those in Fig. 3: CBL, cerebellum; TEL, telencephalon. (B) High magnification light field analysis in the developing retinal layers at E5 and E15. Sections were hybridised and exposed to emulsion as in Fig. 3C. Signals in light-field view appear as black silver grains over stained cells (again note that the pigment produces artefactual signals). Abbreviations are as in Fig. 3C. The black arrowhead indicates typical clustering of TRβsignals over ON cells in the E5 retina.

Thyroid hormone does not alter levels of TRα and β mRNA in cultured retinal cells

Expression of TRβ mRNAs has been shown to be regulated by T3: in tadpoles, TRβ mRNA is up-regulated by T3 (Yaoita and Brown, 1990), whereas in rat pituitary GH3 cells, T3 slightly-up-regulates TRβ1 mRNA and down-regulates TRβ2 mRNA (Hodin et al., 1989). The chick embryonic thyroid gland begins secreting thyroxine at around E10/E11 (Wentworth and Ringer, 1986); thus, the levels of cTRβ2 mRNA, which gradually decrease until hatching, varied inversely with hormone levels, suggesting that hormone may down-regulate β2 mRNA levels. We investigated this possibility by culturing retinas from E6 embryos in the absence or presence of added T3 for 3 days. Northern blot analysis of mRNA from these cultures revealed that neither TRα nor β mRNA levels were subject to obvious T3-dependent variations in levels (Fig. 5). The type of TRβ mRNA detected was identified as that for TRβ2 by RNAse protection analyses (Fig. 2) and also by PCR analysis of the RNA samples used in Fig. 5 (not shown).

Fig. 5.

Northern blot analysis of expression of TRα and β mRNA in T3-treated retina of E6 chick embryos. Poly(A)-selected RNA (6 μg lanc) samples were hybridised with probes for TR α, TRβand as a control probe, glyceraldehyde-3-phosphate dehydrogenase, GAD.

Fig. 5.

Northern blot analysis of expression of TRα and β mRNA in T3-treated retina of E6 chick embryos. Poly(A)-selected RNA (6 μg lanc) samples were hybridised with probes for TR α, TRβand as a control probe, glyceraldehyde-3-phosphate dehydrogenase, GAD.

Transcriptional activation by TRβ N-terminal variants

Transcriptional activation by cTR α has been shown previously (Sap el al., 1989; Forman et al., 1989). To study trans-activât ion by cTRβ2, we performed cotransfection assays with CV-1 cells using as reporter, the plasmid pBL2-CAT containing a dimer T3REp (a palindromic T3 response element, Glass et al., 1988) upstream of the TK promoter of the CAT gene, together with plasmids expressing either cTRβ2 or β 0. The plasmids expressing the receptors were constructed so that both cDNA clones contained the same consensus sequence upstream of the initiating ATG (CAC-CATG) for obtaining optimal and as equal translation of the resulting mRNAs as possible without changing amino acid sequences. Transfected cells were maintained in the presence or absence of T3 and CAT assays were then performed on cell extracts. Fig. 6 shows that the plasmid expressing cTRβ2 gave an about 3-fold greater induction compared to cTR β0. The same results were obtained with NIH 3T3 cells.

Fig. 6.

Transcriptional activation by TRβ2 and β 0 in CV-1 cells. Cells were transfected with plasmids expressing the respective receptors together with the pBL2-CAT reporter plasmid containing a dimer T3REp. The relative values shown represent the ratio of acetylated chloramphenicol to acetylated chloramphenicol in the CAT-assay. The results shown represent an average of 3 individual experiments.

Fig. 6.

Transcriptional activation by TRβ2 and β 0 in CV-1 cells. Cells were transfected with plasmids expressing the respective receptors together with the pBL2-CAT reporter plasmid containing a dimer T3REp. The relative values shown represent the ratio of acetylated chloramphenicol to acetylated chloramphenicol in the CAT-assay. The results shown represent an average of 3 individual experiments.

Developmental and tissue-specific N-terminal variants of TRβ

The TRs encoded by the c-erb.-A/TR α or and β genes differ considerably in N-terminal structure (Murray et al., 1988; Forrest et al., 1990). Further N-terminal variants of cTR α apparently arise by alternative translational initiation within the same mRNA (Goldberg et al., 1988; Bigler and Eisenman, 1988). The TR α and β genes possess a related exon structure in the rat, chick, Xenopus and man (Zahraoui and Cuny, 1987; Lazar et al., 1989; Yaoita et al., 1990; Sakurai et al., 1990; Laudet et al., 1991). The point of divergence in the N termini of cTRβ2 and β 0 that we have detected coincides with a known exon junction in the TRβgenes in Xenopus and man, suggesting the involvement of alternative splicing, although elucidation of the precise mechanism awaits cloning and characterisation of the 5’ region of the cTRβgene. However, alternative promoter usage is not excluded, perhaps as an additional mechanism, and could explain the extremely restricted expression of cTRβ 2 mRNA, with respect to its tissue- and developmental stage-specificity. Indeed, both mechanisms have been invoked in the generation of N terminal variants of retinoic acid receptor fi (Zelent et al., 1991). The striking developmental switching of cTRβ2 and β 0 mRNAs is probably regulated by several factors. Embryonic T3 production is unlikely to be the sole factor since we find no evidence for T β -regulation of cTRβ 2 mRNA levels in the retina in culture, although the rTRβ 2 mRNA has been reported to be down-regulated by T3 in pituitary GH3 cells (Hodin et al., 1989).

The finding that cTRβ 2 activates transcription of a reporter gene with greater efficiency than cTRβ0 indicates the importance of the TR N terminus in transactivation. The unique N terminal structures of TRβ variants could, for example, be necessary for interaction with cell- or developmental stage-specific factors in determining target gene activation, as suggested for N-terminal variants of the chick progesterone receptor (Tora et al., 1988). Further versatility in the transcriptional response to T3 may be mediated by cTRa N-terminal variants. The shorter form initiating only 14 codons upstream of the DNA-binding region lacks the phosphorylation sites at Ser 12 and Ser 28 of the longer version (Goldberg et al., 1988; Glineur et al., 1989). Although the function of these sites has not yet been elucidated for cTRa, Ser 28 is essential for the transforming activity’ of the mutant TR α product encoded by the v-erbA oncogene (Glineur et al., 1990). The role of phosphorylation in control of TRaactivity merits further study, particularly since the N terminus of cTR/G also possesses putative phosphorylation sites; for casein kinase II at amino acid positions 28, 37 and 75 and for protein kinase C at position 106.

Development of the retina and thyroid hormone receptors

Selective developmental expression of cTRβ 2 mRNA in the early retinal outer nuclear layer suggests a very specific role for the |82 receptor. The neuroblasts of the sensory layer of the chick retina proliferate extensively to form by E5 the outer and inner nuclear layers (ONL and INL) (Romanoff, 1960). The early ONL contains the developing rod and cone photoreceptor cells. Differentiating rods and cones are apparent in the ONL by E10, and by E15 produce the photopigmentcontaining outer segment discs (Grim, 1982). It is informative to reflect that during amphibian metamorphosis T3 is required for changes in eye development such as in photopigment composition and retinal circuitry (for review, see Hoskins, 1990). These studies support a direct role for TRs in mediating retinal development and also indicate candidate genes that may be subject to regulation by cTRβ2 in chick eye development, for example, photopigment or other early differentiation markers of rods and cones. In contrast, the later developmental increase in the ratio of cTR/D:β2 mRNAs in the retinal nuclear layers points to some later maturation function for cTRβ 0, perhaps in synaptogenesis or myelination which occur in the pre-hatching embryonic period (Griin, 1982). The more widespread expression of TRa’mRNA over all retinal layers suggests that TRa mediates more general events, possibly in retinal metabolism (Romanoff, 1960), or may have a modulatory function on the activity of other TRs.

It is curious to note that rTR β2 which is highly related to cTRβ2, is reported to be pituitary-specific, where a possible role is in feedback regulation of thyroxine production (Hodin et al., 1989). Thus, in mammalian and avian species, a related variant of TRβ appears to be employed in two distinct, yet very specialised functions. It would be interesting to analyse expression of TRβ 2 in the chick pituitary and, conversely, in mammalian retinal development to ascertain the degree of overlap of receptor function in different species.

Implications for developmental regulation of target genes

The expression patterns for cTR α, β0 and β 2 change during retinal development and could give rise to changing combinations of TRs in the same cell suggesting various mechanisms of developmental control (Fig. 7). Thus, specific combinations of TRs maybe as important as individual TRs in determining target gene specificity, perhaps involving competitive or cooperative interactions as suggested previously (Forrest et al., 1991). We have shown by in vitro DNA-binding assays that cTR α and β0 can form both homodimers and heterodimers (unpublished data), suggesting that particular dimers could have cell- and stage-specific functions in development. A model for developmental control by different combinations of TRs could be reminiscent of that proposed for other vertebrate DNA-binding proteins, such as the homeo-box genes in development of the murine neuroectoderm (Kessel and Gruss, 1990).

Fig. 7.

Summary of changing patterns of TRa and )3 variant mRNAs during retinal development. Retinal layers are shown at stages E5, E10 and E19, with the pigmented epithelium as a solid black band; ONL, outer nuclear layer; INL, inner nuclear layer, GC, ganglion cell layer. Location and approximate levels of specific TR mRNA signals are based on the results in Figs 3 and 4, and additional data not shown A key to the symbols representing different TR mRNAs is shown beneath the figure. The figure is not intended to be strictly quantitative per cell but rather to show the changing expression trends and to indicate the possible homodimers and heterodimers which could form in the cehs of different retinal layers in development.

Fig. 7.

Summary of changing patterns of TRa and )3 variant mRNAs during retinal development. Retinal layers are shown at stages E5, E10 and E19, with the pigmented epithelium as a solid black band; ONL, outer nuclear layer; INL, inner nuclear layer, GC, ganglion cell layer. Location and approximate levels of specific TR mRNA signals are based on the results in Figs 3 and 4, and additional data not shown A key to the symbols representing different TR mRNAs is shown beneath the figure. The figure is not intended to be strictly quantitative per cell but rather to show the changing expression trends and to indicate the possible homodimers and heterodimers which could form in the cehs of different retinal layers in development.

Finally, since our results indicate specific functions for TRs in retinal development, it is intriguing to ask why eye deformities have generally not been described in congenital thyroid disorders. In contrast, impaired development of the brain is well-documented in human cretinism and in hypothyroid animals (Legrand, 1984; Dussalt and Ruel, 1987) and specific expression of TR α and β genes correlates well with T3-dependent developmental events in the brain (Forrest et al., 1991). One explanation consistent with our findings would be that early retinal development is independent of embryonic thyroid gland function. Thus, TRs expressed in the early retina could operate in the absence of embryonic hormone perhaps as T3-independent repressors as suggested from trany-activation studies (Damm et al., 1989) or, alternatively, may be responsive to low levels of maternal hormone present in the yolk (Sechman and Bobek, 1988). In conclusion, we have identified in the chick embryonic retina a novel and well-defined system in which to analyse in detail the developmental functions of different TRs and of thyroid hormone.

We thank Dr Joanna Wroblewski and Dr Johan Thyberg for cryostat and microscope facilities. We are grateful to Dr P. Chambón for gift of the pSG5 plasmid. This project was funded by grants from the Swedish Cancer Society (RMC), the Natural Science Foundation (NFR), The Swedish Technical board for Development (ST1JF), The Beijer Foundation and funds at the Karolinska Institute.

Benbrook
,
D.
and
Pfahl
,
M.
(
1987
).
A novel thyroid hormone receptor encoded by a cDNA clone from a human testis library
.
Science
238
,
788
791
.
Bigler
,
J.
and
Eisenman
,
R. N.
(
1988
).
c-erbA encodes multiple proteins in chicken erythroid cells
.
Molec. cell. Biol
.
8
,
4155
4161
.
Bradley
,
D. J.
,
Votingm
,
W. S.
and
Weinberger
,
C.
(
1989
).
Differential expression of trand thyroid hormone receptor genes in rat brain and pituitary
.
Proc. nam. Acad. Sci. U.S.A
.
86
,
72507254
.
Damm
,
K.
,
Thompson
,
C. C.
and
Evans
,
R. M.
(
1989
).
Protein encoded by v-erbA functions as a thyroid-hormone receptor antagonist
.
Nature
339
,
593
597
.
Dussault
,
J. H.
and
Ruel
,
J.
(
1987
).
Thyroid hormones and brain development
.
Ann. Rev. Physiol
.
49
,
321
334
.
Forman
,
B. M.
,
Yang
,
C.
,
Au
,
M.
,
Casanova
,
J.
,
Ghysdael
,
J.
and
Samuels
,
H. H.
(
1989
).
A domain containing a leucin-zipper-like motif mediates novel in vivo interactions between the thyroid hormone and retinoic acid receptors
.
Mol. Endocrinol
.
3
,
16101626
.
Forrest
,
D.
,
Hall Book
,
F.
,
Persson
,
H.
and
Vennstrbm
,
B.
(
1991
).
Distinct functions for thyroid hormone receptots a- and (Jin brain development indicated by differentia! expression of receptor genes
.
EMBOJ
.
10
,
269
275
. .
Forrest
,
D.
,
Sjsberg
,
M.
and
Vennstrfim
,
B.
(
1990
).
Contrasting developmental and tissue-specific expression of a and fl thyroid hormone receptor genes
.
EMBO J
9
,
1519
1528
Gesell
,
A.
,
Amatruda
,
C. S.
and
Culotta
,
C. S.
(
1936
).
Effect of thyroid therapy on the mental and physical growth of cretinous infants
.
American J. of Diseases of Children
52
,
1117
1138
.
Glass
,
C. K.
,
Holloway
,
J. M.
,
Devary
,
O. V.
and
Rosenfeld
,
M. G.
(
1988
).
The thyroid hormone receptor binds with opposite transcriptional effects to a common sequence motif in thyroid hormone and estrogen response elements
.
Cell
54
,
313
323
.
Gllneur
,
C.
,
Baiolly
,
M.
and
Ghysdael
,
J.
(
1989
).
The c-erbA a- encoded thyroid hormone receptor is phosphorylated in its amino terminal domain by casein kinase II
.
Oncogene
4
,
1247
1254
.
Gllneur
,
C.
,
Zenke
,
M.
,
Beug
,
H.
and
Ghysdael
,
J.
(
1990
).
Phosphorylation of the v-erbA protein is required for its function as an oncogene
.
Genes Dev
.
4
,
1663
1676
.
Goldberg
,
V-
,
Gllneur
,
C.
,
Gesqulere
,
J.-C.
,
Ricouart
,
A.
,
Sap
,
J.
,
Vennstrdm
,
B.
and
Ghysdael
,
J.
(
1988
).
Activation of protein kinase C or cAMP-dependent protein kinase increases phosphorylation of the c-erbA-encoded thyroid hormone receptor and of the v-erbA-encoded protein
.
EMBO J
7
,
2425
2433
.
Graupner
,
G.
,
Wills
,
K. N.
,
Tzukertnan
,
M.
,
Zhang
,
X.-k.
and
Pfahl
,
M.
(
1989
).
Dual regulatory role for thyroid-hormone receptors allows control of retinoic-acid receptor activity
.
Nature
340
,
653656
Green
,
S.
and
Chambón
,
P.
(
1988
).
Nuclear receptors enhance our understanding of transcription regulation
.
Trends in Genetics
4
,
309
314
.
Green
,
S.
,
Isseman
,
L
, and
Sheer
,
E.
(
1988
).
A versatile In vivo and in vitro eukaryotic expression vector for protein engineering
.
Nucl. Acids Res
16
,
369
.
Gron
,
G.
(
1982
).
The development of the vertibrate retina: a comparative survey
. In
Adv. Anatom., Embryol and Cell Biol
, vol.
78
, pp.
7
85
.
Berlin
:
Springer-Verlag
.
Hodin
,
R. A.
Lazar
,
M. A.
,
Wintman
,
B.
Darting
,
D. S.
,
Koenig
,
R. J.
,
Larsen
,
P. R.
,
Moore
,
D. D.
and
Chin
,
W. W.
(
1989
).
Identification of a thyroid hormone receptor that is pituitaryspecific
.
Science
244
,
76
79
.
Hoskins
,
S. G.
(
1990
).
Metamorphosis of the amphibian eye
.
J. Neurobiol
.
21
,
970
989
.
Lnnls
,
M. I.
,
Gelfand
,
D.
Ji-Sninsky
,
J. J.
and
White
,
T. J
,
(
1990
).
RACE: Rapid amplification of cDNA ends
. In
PCR Protocols. A Guide to Methods and Applications
, pp.
28
-
38
.
San Diego
:
Academic Press
, INC.
Izumo
,
S.
and
Mahdavi
,
V.
(
1988
).
Thyroid hormone receptor a isoforms generated by alternative splicing differentially activate myosin HC gene transcription
.
Nature
334
,
539
542
.
Kessel
,
M.
and
Gruss
,
P.
(
1990
).
Murine developmental control genes
.
Science
249
,
374
379
.
Laudel
,
V.
,
Begue
,
A.
,
Henry-Dutholt
,
C.
,
Joubel
,
A.
,
Martin
,
P.
,
Stehlln
,
D
, and
Saule
,
S.
(
1991
).
Genomic organization of tbe human thyroid hormone receptor a (c-erbA-1) gene
.
Nucl. Acids Res
.
19
,
1105
1112
.
Lazar
,
M. A.
,
Hodin
,
R. A.
,
Darling
,
D. S.
and
Chin
,
W. W.
(
1989
).
A novel member of the thyroid/steroid receptor family is encoded by the opposite strand of the rat c-erbAa transcriptional unit
.
Molec. ceil. Bio!
.
9
,
1128
1136
.
Legrand
,
J.
(
1984
).
Effects of thyroid hormones on central nervous system development
. In
Neurobehavioral Teratology
, (ed.
J.
Yanai
), pp.
331
-
363
.
Amsterdam
:
Elsevier Science Publishers.
Mccarrison
,
R.
(
1908
).
Observations on endemic cretinism in the Chitral and Gilgil valleys
.
Lancet 1275-1280
.
Miyajlma
,
N-
,
Horiuchi
,
R.
,
Shibuya
,
Y.
,
Fukushige
,
S.-L
,
Matsuburn
,
K.-I.
,
Toyoshima
,
K.
and
Yamamoto
,
T.
(
1989
).
Two erbA homologs encoding proteins with different T3 binding capacities are transcribed from opposite DNA strands of the same genetic locus
.
Cell
57
,
31
39
.
Murray
,
M. B”
Zllz
,
N. D”
Mccreary
,
N. L.
,
Macdonald
,
M. J.
and
Towie
,
H. C.
(
1988
).
Isolation and characterization of rat cDNA clones for two distinct thyroid hormone receptors
.
J. Biol. Chem
.
263
.
12770
-
12777
. ’
Romanoff
,
A. L.
(
1960
).
The organs of special sense
. In
The Avian Embryo
, vol.
2
,
pp 363
-
427
. New York: The MacmiHian Company.
Sakurai
,
A.
,
Nakai
,
A.
and
Degroot
,
L. J.
(
1990
).
Structural analysis of human thyroid hormone receptor ft gene
.
Mol. Cell. Endocrinol
.
71
,
83
91
.
Sap
,
J.
,
De Magistrls
,
L.
,
Stunnenberg
,
H.
and
Vennstrbm
,
B.
(
1990
).
A major thyroid hormone response element in the third intron of the rat growth hormone gene
.
EMBO J
.
9
,
887
896
.
Sap
,
J.
,
Muñoz
,
A.
,
Damm
,
K.
,
Goldberg
,
Y.
,
Ghysdael
,
J.
,
Leutz
,
A.
,
Beug
,
H.
and
Vennstrdm
,
B.
(
1986
).
The c-erb-A protein is a high-affinity receptor for thyroid hormone
.
Nature
324
,
635
640
.
Sap
,
J.
,
Muñoz
,
A.
,
Schmitt
,
J.
,
Stunnenberg
,
H.
and
Vennstrôm
,
B.
(
1989
).
Repression of transcription mediated at a thyroid hormone response element by the v-erb-A oncogene product
.
Nature
340
,
242
244
.
Sechman
,
A.
and
Bobek
,
S.
(
1988
).
Presence of iodothyronines in the yolk of the hen’s egg
.
General and comparative Endocrinol
.
69
,
99105
.
Smith
,
D. W.
,
Blizzard
,
R. M.
and
Wilkins
,
L.
(
1957
).
The mental prognosis in hypothyroidism in infancy and childhood
.
Pediatrics
19
,
1011
1022
Strait
,
K. A.
,
Schwartz
,
H. L.
,
Perez-Castillo
,
A.
and Oppenheimer. J. H
. (
1990
).
Relationship of c-erbA mRNA content to tissue triiodothyronine nuclear binding capacity and function in developing and adult rats
.
J. biol. Chem
.,
265
,
10514
10521
.
Tora
,
L-
,
Gronemeyer
,
H.
,
Turcotte
,
B
,,
Gaub
,
M.-P.
and
Chambón
,
P.
(
1988
).
The N-terminal region of the chicken progesterone receptor specifies target gene activation
.
Nature
333
,
185
188
Weinberger
,
C.
,
Thompson
,
C. C.
,
Ong
,
E. B.
,
Lebo
,
R.
,
Gruol
,
D. J.
and
Evans
,
R. M.
(
1986
).
The c-erb-A gene encodes a thyroid hormone receptor
.
Nature
324
,
641
646
.
Wentworth
,
B. C.
and
Ringer
,
R. K.
(
1986
).
Thyroids
. In
Avian Physiology
. (ed. P.D. Sturkie), pp. 452-464.
Berlin
:
SpringerVerlag
.
Wood
,
W. M.
,
Ooran
,
K. W.
,
Gordon
,
D. F.
and
Ridgway
,
E. C.
(
1991
).
[solation and charactenzation of mouse complementary DNAs encoding a and thyroid hormone receptors from thyrotrope cells: the mouse pituitary-specific pl isoforra differs at the amino terminus from the corresponding species from rat pituitary tumor cells
.
Molecular Endocrinology
5
,
1049
1061
.
Yaolta
,
Y.
and
Brown
,
D. D.
(
1990
).
A correlation of thyroid hormone receptor gene expression with amphibian metamorphosis
.
Genes Dev
.
4
,
1917
1924
.
Yaolta
,
Y
,,
Shi
,
Y.-B.
and
Brown
,
D. D.
(
1990
).
Xenopus laevis o- and P thyroid hormone receptors
.
Proc. natn. Acad Set. U.S.A
.
87
,
7090
7094
.
Zahraoni
,
A.
and
Cony
,
G.
(
1987
).
Nucleotide sequence of the chicken proto-oncogene e-erb.4 corresponding to domain 1 of verbA Eur
.
J. Biochem
.
166
,
63
69
.
Zelent
,
A.
,
Mendelsohn
,
C.
,
Kastner
,
P.
,
Krust
,
A.
,
Gander
,
J.-M.
,
Ruffenach
,
F.
,
Leroy
,
P.
and
Chambón
,
P.
(
1991
).
Differentially expressed isoforms of the mouse retinoic acid receptor p are generated by usage of two promoters and alternative splicing
.
EMBO
.
10
,
71
81
.

The nucleotide sequence described in this paper has been submitted to the EMBL database under the accession number X62642.