served between a-1 and P) is totally different from 
the corresponding region in a-2. This variation ap- 
parently results in the inability of a-2 to bind thy- 
roid hormone or its analogues. However, this form 
retains the ability to bind TREs in a number of 
genes, including those encoding the rat growth 
hormone and TSH subunits. From these and other 
data it appears that the initial a-subunit gene tran- 
script is differentially spliced to yield either the a-1 
or a-2 mRNAs. Surprisingly, one of the spliced 
products is not a bona fide THR, because it lacks 
the ability to bind thyroid hormone. Similar rat a-2 
forms were described by Dr. T. Mitsuhashi (brain) 
and Drs. S. Izumo and V Mahdavi (heart). Dr. Chin 
showed that a-2 is the most abundant form in 
brain. 
Dr. Chin has investigated the potential biological 
role of a-2 in the regulation of thyroid hormone- 
responsive genes. In particular, the ability of a-2 to 
bind putative TREs in such genes suggested a po- 
tential mechanism for interactions with thyroid 
hormone-binding forms. First, the presence of a-2 
with a-1 in the clonal GH^ cell line indicates the 
potential for the two forms to interact in the same 
cell. Second, in collaboration with Drs. Ronald 
Koenig, E Reed Larsen (HHMI), and David Moore at 
Brigham and Women's Hospital and Massachusetts 
General Hospital, cotransfection experiments were 
performed in which a-2 was introduced with bona 
fide THRs into cells lacking functional receptors. 
The ability of each THR or related forms, separately 
or together, to transactivate genes was assessed. 
When a 5 -flanking region reporter gene construct 
was used, a-2 alone did not transactivate rat 
growth hormone gene expression. However, in 
combination with rat a-1 or human (3, a-2 could in- 
hibit the transactivation capability of these bona 
fide THRs. Thus it appears that two THR-related 
forms derived from a single transcript via alternate 
splicing possess competitive biologic actions. This 
suggests that a-2 might be a natural inhibitor of 
thyroid hormone activity. 
In the study of the c-erbAa gene locus in the rat, 
the carboxyl-terminal protein-coding regions for a- 
1 and a-2 were shown to be encoded by adjacent 
exons. In fact, the penultimate 3 '-exon of the rat c- 
erbAa gene encodes both a common region (i.e., a 
region that is identical in both a-1 and a-2) and a 
region specific for a-1. Hence, by alternate donor 
site choice, variable splicing can lead to either a-1 
or a-2 in a tissue-specific manner. Further examina- 
tion of the rat c-erbAa locus revealed yet another 
interesting finding. The transcripts in various tis- 
sues could be detected using a sense probe of a 
portion of the a-2-specific exon, suggesting tran- 
scription of the opposite strand. Such analyses re- 
vealed the presence of a gene (Rev-erbAa) that is 
transcribed from the opposite strand of the rat c- 
erbAa locus with the 3 '-exons of the Rev-erbA and 
a-2 transcriptional units sharing 269 common 
bases that represent respective coding regions. Sur- 
prisingly, this opposite-strand transcript encodes 
another member (504 aa) of the c-erbA thyroid/ste- 
roid hormone receptor family, which also includes 
the retinoic acid and vitamin D receptors. Rev- 
erbAa, which does not bind thyroid hormone or 
TREs, is as related to THRs as it is to retinoic acid 
and vitamin D receptors. It is speculated that Rev- 
erbAa might be functional with an unknown ligand 
and undefined target genes, as well as the potential 
regulation of a-l/a-2 transcript production and 
mRNA translation via antisense nucleic acid mecha- 
nisms. The conservation of this genomic arrange- 
ment in humans and in rodents, as confirmed by 
this laboratory and by Dr. N. Miyajima, and the 
>95% homology between rat and human Rev- 
erbAa suggest the possible biological importance of 
this curious transcript and/or product. 
Yet another putative THR form, (3-2 (514 aa), was 
cloned from the rat pituitary GH^ library. The (3-2 is 
nearly identical to rat P-1 (461 aa) (homologue of 
human placental c-erbA^) from a region just amino- 
terminal to the DNA-binding domain through to 
the carboxyl terminus. The only difference is ob- 
served in the amino-terminal ends of the molecules 
involving the A/B region of members of the thy- 
roid/steroid hormone receptor family. The (3-2 is ca- 
pable of trans-activating thyroid hormone-respon- 
sive genes but without apparent quantitative 
differences among the putative THRs. Remarkably, 
the P-2 mRNA is expressed only in the pituitary 
gland. This contrasts with the observation that a-1, 
a-2, and (3-1 are expressed in nearly all other tis- 
sues, although in varying ratios. Thus marked tis- 
sue-specific expression of (3-2 is notable. 
This work has revealed the marked heterogeneity 
of THR and related forms. Indeed, there are at least 
three bona fide THRs, as judged by their ability to 
bind thyroid hormone and their analogues in the 
correct fashion and to transactivate putative thyroid 
hormone-responsive genes. A fourth variant form, 
a-2, related to a-1, may be a natural inhibitor of 
thyroid hormone action in various cells. The de- 
tails of the role and regulation of the opposite- 
strand transcript encoding Rev-erbAa require more 
elucidation. 
Continued 
42 
