Potential role of Arabidopsis PHP as an accessory subunit of the PAF1 transcriptional cofactor

Abstract

Paf1C is a transcriptional cofactor that has been implicated in various transcription-associated mechanisms spanning initiation, elongation and RNA processing, and is important for multiple aspects of development in Arabidopsis. Our recent studies suggest Arabidopsis Paf1C is crucial for proper regulation of genes within H3K27me3-enriched chromatin, and that a protein named PHP may act as an accessory subunit of Paf1C that promotes this function.

The sheer complexity of mechanisms participating in transcriptional regulation is staggering. For example, the protein responsible for transcription of most eukaryotic genes, RNA polymerase II (Pol II), consists of about a dozen subunits, and directly depends on interactions with several other protein factors such as Transcription Factor II D (TFIID) and Mediator, themselves made up of scores of subunits, to engage transcription upon appropriate genes. Subsequently, Pol II is faced with the task of negotiating chromatin structure during elongation, a process that involves several additional proteins and mechanisms. All of these factors must cooperate with gene-specific cis elements and transcription factors to drive production of functional mRNA at the appropriate time and in the appropriate cells. Within this mechanistic milieu, the role of the conserved Paf1C (Paf1 complex) cofactor has been enigmatic. In various studies spanning the past 10 years, Paf1C has been implicated in facilitating ubiquitination of histone H2B, promoting interaction of histone methyltransferases with chromatin at active genes, and assisting in RNA processing.^1–7 With such apparent diversity in function, it has been difficult to propose one specific role for Paf1C in transcription. Adding to the puzzle, our recent studies demonstrated that in Arabidopsis, Paf1C is required for proper regulation of genes characterized by a specific “chromatin signature”—specifically, nucleosomes carrying histone H3 trimethylated at lysine 27 (H3K27me3), and that this activity is promoted by the Paf1C subunit PHP.^8,9

Of the several subunits of Arabidopsis Paf1C, PHP seems atypical for a number of reasons. First, loss of any of the “core” Paf1C subunits VIP4, VIP5, VIP6 or ELF7 led to mutual loss of the other subunits as well as PHP, possibly a result of protein destabilization. In contrast, loss of PHP did not affect accumulation of these other Paf1C subunits (Fig. 1).⁹ One explanation is that PHP is not a “core” component of Paf1C that helps maintain structural integrity of the complex, but rather may be peripherally associated. Second, loss of PHP did not recapitulate the developmental phenotypes caused by loss of other Paf1C components. Mutations in VIP3, VIP4, VIP5, VIP6 or ELF7 lead to developmental pleiotropy, including abnormal leaf and flower development and early flowering. The latter defect is associated with loss of expression of the flowering inhibitor, MADS-domain transcription factor FLC. However, we found that the effect of loss of PHP is limited to early flowering. Indeed, transcriptional profiling revealed that only a very limited number of genes, including FLC, are misregulated in a php mutant. Mostly, these genes constitute a subset of those misregulated in other mutants for Paf1C.⁹ These observations support the idea that PHP assist Paf1C at a specific subset of genes. Such a role would be analogous to that proposed for the human counterpart of PHP, designated Parafibromin, in recruiting the human Paf1C (hPAF) to genes targeted by the Wnt signaling pathway.¹⁰

Figure 1.

Immunoblot analysis of Arabidopsis Paf1C proteins in various genetic backgrounds. Antibodies generated against specific Paf1C proteins (PHP, ELF7, VIP3, VIP4, VIP5 and VIP6) were used to analyze total cell lysate from wild-type plants (Col-0 and Col:FRI) and Paf1C mutants (php-1/-2, elf7, vip3, vip4, vip5 and vip6). The asterisk in the VIP5 panel indicates an immunocrossreactive protein. Immunoreactivity for histone H3 is shown as a loading control (from Park et al.⁹).

We previously showed that genes targeted for regulation by Paf1C in Arabidopsis exhibit strong enrichment for H3K27me3.⁸ Our subsequent analysis of PHP extended those results, and revealed that the subset of genes that are dependent on PHP show the most intense enrichment for this modification (Fig. 2).⁹ Thus, we proposed that PHP is an accessory subunit of Paf1C, required for full activity of Paf1C within the specialized chromatin environment set up by extensive H3K27me3. H3K27me3 is an elaboration in higher eukaryotes best characterized as a mechanism of repression of developmentally important genes by the so-called Polycomb-group (PcG) machinery. However, recent genome-wide chromatin mapping studies have found that H3K27me3 is pervasive in the genome, and is neither limited to developmentally important genes nor to genic regions.^8,11,12

Figure 2.

Enrichment for H3K27me3 across genes that are upregulated (solid red, left part) or downregulated (solid green, right part) in a strong php mutant. Positional signals across the transcriptional unit were averaged for most protein coding genes. Genes that are misregulated in the Paf1C mutant vip3 are shown as a green or red dashed line. The averaged positional signal for most protein coding genes is shown as a solid black line; the 95^th percentile confidence intervals are depicted as dashed black lines (from Park et al.⁹).

Proposing a model by which PHP might facilitate Paf1C activity in this manner is complicated by the diversity in transcription-related molecular mechanisms in which Paf1C apparently participates. An obvious possibility is that PHP helps Pol II negotiate H3K27me3-enriched chromatin during the elongation phase of transcription. Recently, it was shown that H3K27me3-enriched genes tend to be associated with promoter-proximal pausing of Pol II,¹³ a newly recognized mechanism of transcriptional control that appears conserved and widespread.^14,15 Simply, PHP may be involved in moderating this pausing.

An independent analysis of the Arabidopsis PHP gene by Yu and Michaels¹⁶ includes two complementary findings worth highlighting. First, an epitope-tagged PHP protein accumulated in the nucleus and immunoprecipitates targeting this modified PHP were enriched for FLC DNA segments spanning the transcribed region. This suggests that PHP directly associates with the chromatin of at least one targeted gene, and is consistent with a role for PHP in transcriptional elongation. Second, they observed that loss of PHP largely suppressed the late-flowering phenotype and ectopic FLC expression conferred by mutation in the FCA, FLK and FY, but not FLD, FVE or LD, genes. These two classes of gene (FCA, FLK and FY; FLD, FVE and LD) are believed to act in mechanisms of FLC suppression that are at some level separable.^17,18 The observed effect of loss of PHP could be explained if FCA, FLK and/or FY somehow limited activity of the PHP gene or protein. At least both FCA and FY participate in pre-mRNA 3′-end processing, and it is interesting to note that Parafibromin physically associates with the Cleavage and Polyadenylation Specificity Factor (CPSF) and Cleavage stimulation Factor (CstF) complexes that are crucial for pre-mRNA 3′-end processing.¹⁹

The proposal that PHP helps Paf1C to negotiate H3K27me3-enriched chromatin is difficult in light of the fact that PHP is well conserved with its counterpart CDC73 in budding yeast, an organism that lacks H3K27me3. However, H3K27me3 may simply be an elaboration of chromatin facilitating a conserved higher-order repressive structure, and it is at this level that PHP may function. Further insight into the mechanisms of PHP awaits a better understanding both of Paf1C function, and of how H3K27me3 acts in transcriptional suppression.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.