Transcribing RNA Polymerase II Is Phosphorylated at CTD Residue Serine-7

See allHide authors and affiliations

Science  14 Dec 2007:
Vol. 318, Issue 5857, pp. 1780-1782
DOI: 10.1126/science.1145977


RNA polymerase II is distinguished by its large carboxyl-terminal repeat domain (CTD), composed of repeats of the consensus heptapeptide Tyr1-Ser2-Pro3-Thr4-Ser5-Pro6-Ser7. Differential phosphorylation of serine-2 and serine-5 at the 5′ and 3′ regions of genes appears to coordinate the localization of transcription and RNA processing factors to the elongating polymerase complex. Using monoclonal antibodies, we reveal serine-7 phosphorylation on transcribed genes. This position does not appear to be phosphorylated in CTDs of less than 20 consensus repeats. The position of repeats where serine-7 is substituted influenced the appearance of distinct phosphorylated forms, suggesting functional differences between CTD regions. Our results indicate that restriction of serine-7 epitopes to the Linker-proximal region limits CTD phosphorylation patterns and is a requirement for optimal gene expression.

Differential phosphorylation of CTD residues of the large subunit of eukaryotic RNA polymerase II (Pol II) occurs during the transcription cycle and appears to orchestrate the recruitment, activation, and displacement of various factors involved in transcription and mRNA processing (1, 2). A variety of kinases have been identified, with phosphorylation activity directed toward the amino acids tyrosine-1 (Abl1/2), serine-2 (CTDK1, CDK9, and DNA-PK), serine-5 (ERK1/2 and CDK7-9), and serine-7 (DNA-PK) (3). The mammalian CTD is >99% conserved across species and possesses almost double the length of its yeast counterparts (4). A minimum length of CTD is required to support the growth of yeast or mammalian cells. However, this is dependent on the number and position of consensus and nonconsensus repeats, which suggests that CTD function is composed of both sequence and length (58). Of the 52 mammalian CTD repeats, 21 obey the consensus sequence and lie largely proximal to the Linker region (fig. S1). The distal C-terminal region deviates from this consensus, predominantly at position 7. These nonconsensus repeats may affect the binding of specific factors or may serve to prevent phosphorylation at the position of deviation. Indeed, studies in vivo suggest that they are equivalent to consensus repeats for functions such as splicing of the fibronectin extra domain I exon (9) but not for maintenance of long-term cell viability (5, 6).

To investigate the role of the CTD repeat structure on its phosphorylation, we have established a system that allows the comparison of CTDs of different lengths and repeat compositions in vivo. Recombinant polymerases are engineered with a point mutation conferring resistance to α-amanitin, allowing the endogenous polymerase to be inhibited (and degraded) after addition of α-amanitin but without affecting recombinant polymerase activity (5, 10). Monoclonal antibodies (mAbs) were produced against the CTD phosphoserine epitopes Ser2-P, Ser5-P, and Ser7-P (Fig. 1A and fig. S2). In preparing these antibodies, we considered earlier findings that showed that the functional unit of the CTD is not the heptad repeat itself but is in a sequence lying within heptapeptide pairs (11, 12). Thus, in the production and testing of our antibodies, a panel of di-heptapeptides with various modifications was used. Analysis of our antibodies and commercially available antibodies revealed that some recognition profiles were limited by modifications on neighboring repeats. For example, the α-Ser7-P antibody (4E12) is affected by upstream, but not downstream, Ser5-P (Fig. 1A).

Fig. 1.

CTD phosphorylation is affected by its length and composition. (A) Survey of CTD phosphopeptides used for characterization of mAbs H5, H14, 4F8, 3E10, 3E8, and 4E12. High (+++), medium (++), low (+), and no (–) reactivity. (B) Western blots of protein extracted from cell lines expressing WT, or CTDs of varying length. Membranes were screened with antibody to Rpb1 to reveal both exogenously expressed α-amanitin resistant polymerase, and any remaining endogenous polymerase (upper panel). Dual labeling was performed with antibody to Rpb1 (green) in addition to antibody to phospho-CTD (red) against Ser2-P (3E10), Ser5-P (3E8), and Ser7-P (4E12), as indicated. White bands indicate saturation of red signal. (C) Analysis of phosphorylated forms from synthetic CTD mutants. The number of repeats and the residue at position 7 in proximal (p) and distal (d) positions of the CTD is shown. Synthetic mutants containing 24 or 48 consensus repeats with Ser7 (S) or its substitution for alanine (A) confirm the specificity of the 4E12 mAb. Chimeras of consensus repeats (lanes labeled S) and repeats with Ser7 mutated to alanine (A), glutamic acid (E), and threonine/lysine (T/K) affect the appearance of a Ser5-P/Ser7-P reactive band when Ser7 is absent in the Linker proximal region. Mutants containing T/K-rich sequences at position 7, proximal to the Linker, result in degradation of polymerase to the CTD-less, Pol IIb form.

Combining these tools, we compared the phosphorylation of wild-type (WT) CTD with that of different lengths of consensus repeats (Fig. 1B). If all repeats are equally accessible to CTD kinases, we should expect intensities of phosphorylation signals for WT and mutants 1 to 8 proportional to CTD length. Dual labeling of membranes with α-Rpb1 antibody (green signal) (mAb Pol3/3 recognizes an epitope outside the CTD) and with α-phospho-CTD antibody (red signal) reveals forms of different mobility—the rapidly migrating, unmodified IIa form and the slower, modified IIo form. For mutants containing 16 to 24 consensus repeats (mutants 3 to 5), the majority of Pol II is not efficiently phosphorylated and accumulates in the IIa form (green). Mutants 1 and 2 are no longer visible at this time point because they are unable to support their own expression. Within the IIo form, Ser2-P appears in a sharp, slow migrating band, whereas in longer CTDs (mutants 6 to 8), Ser5-P appears largely in a band (white saturation) migrating between the Ser2-P band and IIa, which suggests that at least two populations of phosphorylated CTD exist in vivo at any time: Ser2-P alone and Ser5-P alone. These data are supported by both the recognition profiles of the antibodies and previous work showing a shift in IIo to a faster migrating form upon treatment with a Ser2-kinase inhibitor (13). Antibody raised against Ser7-P revealed the existence of this epitope in vivo, which is distributed among the major Ser2-P and Ser5-P reactive bands. The epitope is lacking from the Ser5-P band that appears just above the IIa form. Strong reactivity of α-Ser7-P is detectable for a band between IIa and IIo (Fig. 1B, *). Furthermore, although Ser2-P and Ser5-P appear in all mutants, Ser7-P appearsonlyinmutants with more than 24 repeats (mutants 4 to 8).

To investigate the effect of nonconsensus repeats on the distribution of phosphorylation, a panel of CTD mutants (fig. S3) was analyzed for their reactivity against phospho-CTD antibodies (Fig. 1C). α-Ser7-P does not recognize a mutant lacking Ser7 (48xS7A) but strongly recognizes mutants containing Ser7 substituted with glutamic acid (S7E), indicating either that this antibody recognizes a CTD conformation or that S7E can structurally mimic Ser7-P for antibody recognition. Furthermore, replacement of Ser7 with alanine prevents recognition of the intermediate band between IIa and IIo (Fig. 1C, *) by α-Ser5-P, suggesting that this form may be Ser7-P–dependent.

Because deviations from serine at position 7 in the WT CTD are concentrated in its distal region, chimeras were produced to assess the effect of proximal and distal positioning of nonconsensus repeats. The two chimeras of consensus repeats, and repeats containing S7E substitutions, produce a form that migrates between IIa and IIo (Fig. 1C, ‡). The proximal positioning of nonconsensus repeats (S7A and S7T/K) affects the appearance of a form similar in mobility to the intermediate IIo Ser5/7-P–reactive band (Fig. 1, B and C, *) seen in mutants of >35 pure consensus repeats. (Fig. 1, B and C; for more detail, see fig. S4).

To determine whether Ser7 phosphorylation is a physiological event during the transcription cycle, chromatin immunoprecipitation (ChIP) experiments were conducted. A detailed example is shown for the T cell receptor beta (TCRβ) gene locus (Fig. 2A). Ser7 was phosphorylated on transcribing Pol II, appearing strongly at the promoter and increasing toward the 3′ region of TCRβ (Fig. 2B). The differences in Ser2 phosphorylation that we observe, compared with earlier data, may result from the antibodies used, because the H5 antibody preferentially recognizes repeats with phosphorylated Ser2 and Ser5 (Fig. 1A).

Fig. 2.

ChIP analysis of CTD modifications. (A) Scale drawing of the rearranged TCRβ gene, with black boxes indicating exons of the variable segment V12-3, of the joining region cluster J1, and of the constant region C1. Gray boxes on top show relative location of ChIP quantitative real-time polymerase chain reaction (QPCR) amplicons 1 to 10. (B) QPCR analysis of ChIP analyses in Jurkat cells with antibodies against Rpb1 and indicated phospho-CTD antibodies. The relative location of the transcriptional start site (arrow) and pA (lollipop) are indicated. (C) Transcriptional activity of CTD mutants. QPCR of c-myc mRNA using primers specific for the mature, spliced form. To control for the effects of α-amanitin treatment, values were normalized against a 7SK RNA control for the same sample. Myc expression is shown relative to WT levels. Labeling as in Fig. 1B; ø, without recombinant Pol II. Error bars, SD.

Given that Ser7 is phosphorylated across TCRβ and all other genes tested (GAPDH, RPLPO, and RPS27), the ability of synthetic polymerases to transcribe and produce mature mRNA from the c-myc (Fig. 2C) and pes1 genes (fig. S5) was analyzed. The effect on c-myc and pes1 mRNA levels of Ser7 substitution to E or K/T appears dependent on its position, either proximal or distal to the Linker, suggesting again that functional differences exist between these regions. Substitution of Ser7 to the non-phosphoacceptor, alanine, did not obviously affect mRNA levels, nor did it affect the long-term growth of cell lines, although viability was compromised (fig. S6). This may be due to the effect of this mutation on small nuclear RNA genes (14).

ChIP experiments revealed that S7E-containing mutants do not stably associate with any of the genes tested, providing an explanation for the deficit in mRNA observed for mutants containing S7E in the Linker-proximal region. Mutants containing either 48 consensus or S7A repeats appear to be recruited to protein coding genes at similar levels (14).

We conclude that the nature of the amino acid at position 7 of the Linker-proximal CTD region is important in steps involved in the stable association of Pol II with class II genes. Accumulation of Ser7-P in the 3′ region of the TCRβ gene may suggest a role in transcription and/or 3′ RNA processing of some protein-coding genes. We are now able to expand previous models for the cycle of CTD modification across genes that are transcribed by RNA polymerase II (15), not only to show how potential phosphorylation patterns change from 5′ to 3′ regions across a gene but also to speculate as to the region of the CTD in which they occur. Phosphorylation of Ser7 in the proximal part of CTD and replacement of Ser7 by other amino acids in the distal part of CTD may constitute an added layer of gene regulation by mammalian RNA polymerases.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article