ABSTRACT
The erbA[alpha] gene encodes two [alpha]-thyroid hormone receptor isoforms, TR[alpha]1 and TR[alpha]2, which arise from alternatively processed mRNAs, erbA[alpha]1 ([alpha]1) and erb [alpha]2 ([alpha]2). The splicing and alternative polyadenylation patterns of these mRNAs resemble that of mRNAs encoding different forms of immunoglobulin heavy chains, which are regulated at the level of alternative processing during B cell differentiation. This study examines the levels of erbA[alpha] mRNA in eight B cell lines representing four stages of differentiation in order to determine whether regulation of the alternatively processed [alpha]1 and [alpha]2 mRNAs parallels the processing of immunoglobulin heavy chain mRNAs. Results show that the pattern of [alpha]1 and [alpha]2 mRNA expression is clearly different from that observed for immunoglobulin heavy chain mRNAs. B cell lines display characteristic ratios of [alpha]1/[alpha]2 mRNA at distinct stages of differentiation. Furthermore, expression of an overlapping gene, Rev-ErbA[alpha] (RevErb), was found to correlate strongly with an increase in the ratio of [alpha]1/[alpha]2 mRNA. These results suggest that alternative processing of erbA[alpha] mRNAs is regulated by a mechanism which is distinct from that regulating immunoglobulin mRNA. The correlation between RevErb and erbA[alpha] mRNA is consistent with negative regulation of [alpha]2 via antisense interactions with the complementary RevErb mRNA.
Thyroid hormone receptors (TRs) mediate the cellular response to thyroid hormone (T3) by regulating target gene transcription (1 -3 ). In all vertebrates TRs are the products of two different genes, erbA[alpha] and erbA[beta] (4 ). The mammalian erbA[alpha] gene produces two mRNAs, erbA[alpha]1 ([alpha]1) and erbA[alpha]2 ([alpha]2), through alternative processing of the 3'-end of its pre-mRNA transcript (5 -7 ). These mRNAs give rise to receptor isoforms with antagonistic functions. [alpha]1 codes for the [alpha]-thyroid hormone receptor (TR[alpha]1), whereas [alpha]2 codes for an orphan nuclear receptor (TR[alpha]2) which does not bind T3 (6 ,7 ). TR[alpha]2 competes with TR[alpha]1 and TR[beta] for specific DNA binding sites, thereby antagonizing T3 action (8 ). Because the erbA[alpha] gene produces both a transcriptional activator (TR[alpha]1) and its specific inhibitor (TR[alpha]2), regulation of the alternative processing of [alpha]1 and [alpha]2 mRNA may provide an important mechanism for determining the cellular response to thyroid hormone.
Alternative processing of the 3'-end of erbA[alpha] RNA transcripts involves competition between splicing and polyadenylation. Polyadenylation at an upstream site yields [alpha]1 mRNA, whereas processing from a 5' splice site (ss) within the final exon of [alpha]1 to a downstream 3' ss produces [alpha]2 mRNA. The levels of [alpha]1 and [alpha]2 vary in different tissues and at different developmental stages (9 ,10 ). However, the mechanisms which regulate expression of the alternatively processed mRNAs are not well understood.
Two general models have been described for the regulation of alternative splicing and polyadenylation (11 ). In some cases alternative processing is regulated by transcript-specific factors which alter the efficiency of either splicing or polyadenylation. In other instances, the activity of one or more constitutive components of the mRNA processing apparatus is altered. In the latter case, the processing of many unrelated transcripts may be affected.
The organization of the erbA[alpha] gene and the alternative mRNA processing of its transcripts appear similar to that of the immunoglobulin (Ig) heavy chain genes. The Ig heavy chain gene produces two functionally distinct mRNAs which encode heavy chains for the membrane-bound (mb) and secreted (sec) forms of Ig. Like [alpha]1, Ig sec mRNA processing utilizes an upstream polyadenylation site. Similar to [alpha]2, Ig mb mRNA processing utilizes a 5' ss in the last common exon to splice to a downstream exon. Regulation of Ig heavy chain mRNA processing in B lymphocytes has been well characterized. In early stages of B cell development downstream splicing of Ig mb mRNA is equal to or greater than that of Ig sec mRNA processing. At later stages upstream polyadenylation of Ig sec mRNA predominates (12 ). Several studies suggest that the balance between splicing and polyadenylation required for processing of Ig mb and sec mRNAs is regulated by a change in the level of a general polyadenylation factor (13 -16 ). In view of these findings, it is possible that erbA[alpha] mRNA processing may parallel that of Ig heavy chain mRNA during B cell development.
A distinguishing feature of the mammalian erbA[alpha] gene locus is the presence of a third gene, Rev-ErbA[alpha] (RevErb), encoded on the DNA strand opposite erbA[alpha] (17 ,18 ). The 3' exon of RevErb overlaps with the [alpha]2-specific 3' exon but not with [alpha]1 sequence. The unusual organization of these genes results in [alpha]2 and RevErb mRNAs which are complementary at their 3'-ends allowing the possible formation of antisense/sense RNA duplexes. Such basepairing interactions between the RevErb and [alpha]2 mRNAs could negatively regulate [alpha]2 mRNA levels, and therefore offer a transcript-specific mechanism by which [alpha]1 and [alpha]2 mRNA levels are regulated.
In this study, we examine the expression of [alpha]1 and [alpha]2 mRNA from B cells representing different stages of differentiation. We find that the thyroid hormone receptor, TR[alpha]1, and the orphan receptors TR[alpha]2 and RevErb are expressed at all stages of differentiation but at varying levels in the different cell lines. Our results indicate that the regulation of alternative RNA processing of [alpha]1 and [alpha]2 is distinct from that of Ig sec and mb RNA processing regulation. However, changes in the relative levels of [alpha]1 and [alpha]2 correlate strongly with variations in levels of RevErb mRNA.
All cell lines are from the mouse B cell lineage and were grown as previously described (13 ,14 ,19 ,20 ). 70Z/3.12 represents a pre-B cell stage line with equal amounts of sec and mb IgM mRNA (20 ). WEHI-231 is an early B cell with equal amounts of sec and mb IgM (20 ). The M12 cell line represents an early B cell which has lost its endogenous heavy chain but expresses approximately equal amounts of sec and mb IgM mRNA when transfected with an IgM gene (13 ). The A20 and 2PK3 cell lines are memory B cells and produce about equal quantities of sec and mb IgG heavy chain mRNA (14 ). 4T001 is a plasmacytoma cell line which secretes large amounts of [gamma]2b, [kappa] molecules of IgG (14 ). S194 is a plasmacytoma cell line which has lost its endogenous heavy chain but produces a large excess of sec over mb mRNA when transfected with an IgM gene (13 ). J558L is a plasmacytoma cell line which has lost its endogenous [alpha] heavy chain but when transfected with an IgG gene sec mRNA is expressed in excess over mb mRNA (19 ).
The erbA[alpha] probe used for northern analysis was prepared from a 600 nt XbaI-SacI DNA fragment excised from plasmid p[alpha]2HN. p[alpha]2HN contains an EcoRI-HincII fragment which is common to both [alpha]1 and [alpha]2. This fragment was isolated from plasmid p[alpha]2[Delta]C-Sac/stop (21 ) and subcloned in pBluescript KS+ (Stratagene). The probe was uniformly labeled with [[alpha]-32P]dCTP by random oligonucleotide-primed synthesis (22 ). For RNase protection assays, a single-stranded antisense riboprobe of [alpha]1 and [alpha]2 common sequence was prepared from pB3EOP which contains the 162 nt PstI-Eco0109I erbA[alpha] DNA fragment cloned between the ApaI-PstI sites of pBluescript KS+. This fragment of the erbA[alpha] gene spans the [alpha]2-specific 5' ss within the 3'-most exon of [alpha]1. pB3EOP was linearized with XbaI and transcribed with T3 RNA polymerase to produce a 209 nt probe. Of this probe, 162 nt are complementary to [alpha]1 mRNA and 135 nt are complementary to [alpha]2 mRNA. The RevErb riboprobe was made from pB4E6 which contains 303 nt of a BglII-Bsu36I fragment from pB4-1 (17 ) cloned into pBluescript KS+. pB4E6 was linearized with XbaI and transcribed with T3 RNA polymerase to produce a 360 nt probe, 130 nt of which are complementary to RevErb exon 6. RNA probes were uniformly labeled with [[alpha]-32P]UTP and purified by electrophoresis (22 ). Plasmids p[alpha]2[Delta]C-Sac/stop and pB4-1 were generously provided by M.A. Lazar, University of Pennsylvania.
Cytoplasmic RNAs were isolated by detergent lysis and phenol/chloroform extraction as previously described (19 ). The RNA was passed over oligo(dT) columns to isolate poly(A)+ RNA. Electphoresis of 1-3 µg of poly(A)+ RNA on a 1.0% agarose-0.22 M formaldehyde gel was run in buffer containing 0.22 M formaldehyde. The RNA was transferred by capillary action to Nytran (Schleicher and Schuell, Inc.) after which the RNA was UV-irradiated and hybridized to labeled probes (23 ). RNA size markers (GIBCO-BRL) were included on the gel and stained with methylene blue after transfer to Nytran (22 ).
Labeled probes were hybridized to 10-30 µg target RNAs as previously described (22 ). The resulting hybrids were digested with 1.5 µl RNase (1 mg/ml RNase A, 20 000 U/ml RNase T1, Ambion, Inc.) at 30oC for 1 hr. Protected RNAs were denatured in formamide and resolved on 5.5% polyacrylamide-urea gels. Band intensities were quantified by radioanalytic scanning with an AMBIS 100 Image Analyzer. Background was subtracted by using regions of identical size located immediately above each experimental band. RevErb/[alpha]2 ratios were computed from experiments in which probes were prepared in parallel using identical mixes of labeled and unlabeled nucleotides to ensure identical specific activities. The specific activity of the probes ranged from 9 * 108 to 2 * 109 c.p.m./nmol between experiments. Molar ratios were calculated after correcting for length and composition of the protected RNA fragments.
In mammals, a single locus codes for the two alternatively processed erbA[alpha] mRNAs and the overlapping RevErb mRNA (Fig. 1 A). To determine if the erbA[alpha] mRNAs are expressed in B lymphocytes, we assayed [alpha]1 and [alpha]2 mRNA levels in two mouse B cell lines representing different stages of differentiation. The cell line 70Z/3 is a tumor line arrested in the pre-B cell stage (20 ) and J558L is a myeloma cell line representing a late-stage B cell or plasma cell line (19 ). The C6 rat astrocytoma cell line, which is well-characterized for erbA[alpha] mRNA levels (9 ), was also analyzed. Northern blot experiments using an erbA[alpha] specific probe show the expected 5.5 kb [alpha]1 mRNA and 2.6 kb [alpha]2 mRNA which correspond to the documented sizes of C6 cell [alpha]1 and [alpha]2 mRNA (Fig. 1 B) (17 ). These results demonstrate that both erbA[alpha] mRNAs are expressed at early and late stages of B cell differentiation.
We next asked whether the balance between alternative processing of [alpha]1 and [alpha]2 mRNA parallels that of Ig sec and mb mRNA at different stages of B cell differentiation. Cytoplasmic mRNA was analyzed from eight mouse cell lines representing four different stages of B cell development. In addition to the pre-B and plasmacytoma cell lines described above, mRNA from two mature B cell lines [WEHI-231 (20 ) and M12 (13 )], two cell lines representing memory B cells [2PK3 and A20 (14 )], and two additional plasma cell lines [4T001 (14 ) and S194 (13 )] were examined to determine the levels of expression of [alpha]1 and [alpha]2 mRNA. RNase protection assays of mRNA from these eight cell lines show that the pre-B cell line, 70Z/3, has the highest ratio of [alpha]1/[alpha]2 mRNA, 3.6, indicating a predominance of upstream polyadenylation over downstream splicing (Fig. 2 A and B). The lowest [alpha]1/[alpha]2 ratios are seen in the plasma cell lines, S194 and J558L and the mature B cell line, M12, which demonstrates an excess of [alpha]2 splicing over polyadenylation. The ratios of [alpha]1/[alpha]2 mRNA in the different B cell lines are clearly distinct from those seen for sec and mb mRNA from the Ig heavy chain gene (Fig. 2 B). For example, the levels of Ig sec (polyadenylated) mRNA levels are nearly equal to those of Ig mb (spliced) mRNA at early stages of B cell differentiation but increase sharply as cells switch almost exclusively to production of Ig sec mRNA in plasma cells. Although changes in the levels of erbA[alpha] mRNAs during B cell differentiation are opposite from the Ig mRNA alternative processing pattern, B cell lines representing distinct developmental stages display characteristic ratios of [alpha]1/[alpha]2 mRNA (Fig. 2 B, Table 1 ).
B lymphocyte cell lines provide a unique system for investigating the molecular basis of developmentally regulated alternative processing. Such processing has been intensively studied for Ig heavy chain genes during B cell differentiation. In this study we show that erbA[alpha] mRNAs are expressed at different levels in B cell lines representing specific stages of development. The ratio of [alpha]1/[alpha]2 mRNA varies in a manner which suggests stage-specific regulation of mRNA levels, but the pattern of expression is clearly different from the variations in sec/mb mRNA levels. Characterization of mRNA levels from the overlapping RevErb gene shows that expression of RevErb mRNA correlates with changes in the [alpha]1/[alpha]2 ratio. These observations have important implications both for regulation of alternative processing and for the role of thyroid hormone during B cell development.
Molecular analysis of Ig heavy chain mRNA processing has shown that regulation of polyadenylation and splicing depends on changes in the levels of essential processing factors (14 ,16 ). Of particular interest to this study is the observation that unrelated pre-mRNAs which undergo similar 3'-end processing are differentially processed throughout B cell differentiation in the same manner as Ig sec and mb heavy chain mRNAs (13 ,15 ). On the basis of these results, erbA[alpha] mRNA processing would also be predicted to parallel that of Ig heavy chain mRNAs, with the [alpha]1/[alpha]2 ratio lowest in pre-B cells, highest in plasmacytoma cells and at intermediate levels in mature and memory B cells. Our results show a different and more complex pattern of variation in the [alpha]1/[alpha]2 expression than predicted. Mature B cells and plasma cells both have relatively low [alpha]1/[alpha]2 ratios while higher ratios are observed for memory cells and the pre-B cell line. These results suggest that the balance between alternative splicing and polyadenylation of erbA[alpha] mRNAs is regulated in a manner independent of Ig heavy chain alternative mRNA processing.
One possible mechanism for regulating expression of erbA[alpha] mRNAs involves interactions with the RevErb mRNA. Since this overlapping mRNA is complementary to [alpha]2, but not [alpha]1, variations in RevErb expression may differentially affect [alpha]1 and [alpha]2 expression. The results presented here are consistent with such a relationship. Despite wide variations in the levels at which these mRNAs are expressed in the eight lymphocyte cell lines analyzed (Table 1 ), a strong correlation is evident between the ratio of RevErb/[alpha]2 and the ratio of [alpha]1/[alpha]2 (Fig. 3 ). Specifically, when the ratio of RevErb/[alpha]2 is between 0.4 and 1.2, the [alpha]1/[alpha]2 ratio is <1.5, and when the RevErb/[alpha]2 is >2.0 the [alpha]1/[alpha]2 ratio is >= 2.5. Thus, an increase from 1.5 to 2.0 in the RevErb/[alpha]2 ratio may represent a threshold associated with a 2- to 3-fold increase in [alpha]1/[alpha]2. Since the ratio of RevErb/[alpha]2 varies over a wider range than the ratio of [alpha]1/[alpha]2 and there is no evident correlation in the variation of RevErb and [alpha]2 levels, interactions between these genes or their mRNA products may be modulated by additional factors.
We thank Mitch A. Lazar for providing recombinant materials, for helpful discussions and for a critical reading of the manuscript. This research was supported by National Institutes of Health grants #DK48034 to S.H.M. and #GM50145 to C.M. and a National Science Foundation grant #MCB-9507513 to M.L.P. M.L.H. was supported by a pre-doctoral fellowship from the Arthur J. Schmitt Foundation.
REFERENCES