Insights into centromeric transcription in mitosis

ABSTRACT

The major role of RNA polymerase II (RNAP II) is to generate mRNAs. I recently uncovered a novel function of RNAP II in chromosome segregation in mitosis, installing the cohesin protector, Shugoshin, at centromeres. Here I will discuss the current understanding of RNAP II-dependent centromeric transcription in mitosis.

The centromere is a specialized chromatin structure with specific DNA sequences that constitutes the base for assembly of the kinetochore, which in turn binds microtubules to mediate chromosome segregation. Compromised centromere and kinetochore functions will lead to chromosome missegregation, which contributes to aneuploidy. Although their DNA sequences vary across species, centromeres serve to play a similar and highly conserved function. Budding yeast cells have simple centromeres with ∼120 base pairs; in contrast, higher organisms, such as human, have more complicated centromeres with hundreds of thousands of base pairs, many of which are repetitive DNA sequences.¹ The core unit of the repeats in human cells is a DNA sequence of 171 base pairs, which does not encode any proteins and is designated as α-satellite DNA.¹ Centromeres comprise two portions, centromere cores (hereafter, centromeres) and pericentromeres flanking centromeres. Although centromeres and pericentromeres are heterochromatic and, thus, transcriptionally silent, increasing evidence indicates that active transcription occurs in both centromeres and pericentromeres.^2,3 In fission yeast pericentromeres, RNA Polymerase II (RNAP II)-dependent transcripts are processed by the RNAi pathway and recruit heterochromatin factors, such as HP1, to establish the heterochromatin-specific histone modifications, thus facilitating the heterochromatin formation.² The centromeric chromatin is characterized by the histone H3 variant CENP-A, upon which kinetochores assembly is initiated.⁴ The centromeric transcription itself and its transcripts (likely α-satellite RNA) have been shown to facilitate the deposition of CENP-A.^3,5-8 Thus, the RNAP II-dependent transcription is important for maintenance of the centromere identity.

When cells enter mitosis, most of the transcriptional factors and RNA polymerases, including RNAP II, are released from chromosomes.⁹ Thus, transcriptional activities in mitosis are largely silent. Interestingly, I, together with others, found that centromeres are actively transcribed by RNAP II.^10,11 Several lines of evidence support this notion. Firstly, the presence of RNAP II at kinetochores in mitosis can be verified by antibody-based fluorescence microscopy: two antibodies raised against phosphor-CTD of Rpb1,^10,11 the largest subunit of RNAP II, and the GFP antibodies against ectopically expressed GFP-tagged Rpb2,¹⁰ the second largest subunit of RNAP II. Secondly, application of the transcriptional inhibitor α-amanitin resulted in significantly weakened centromeric cohesion and decreased CENP-C localization.^10,11 Considering the specificity of α-amanitin toward RNAP II,^12,13 it is very likely that the phenotypes were derived from the inhibition of transcriptional activities of RNAP II. Thirdly, several groups, including us, took advantage of the transcription run-on assay and proved that the mitotic chromosomes, especially at centromeres, are able to be transcribed, at least by RNAP II.^10,11 Taken together, I conclude that transcription occurs at centromeres in mitosis.

Why do cells maintain transcription at centromeres during mitosis? A recent study demonstrated that inhibition of RNAP II activities by α-amanitin increased the anaphase cells with lagging chromosomes that exhibited the reduced levels of CENP-C present at kinetochores. The result suggests that the centromeric transcription helps maintain centromeric proteins at centromeres.¹¹ Given the fact that CENP-A anchors CENP-C at centromeres,^14,15 the decreased levels of CENP-C could be a result of the defects in maintaining CENP-A at centromeres in response to transcription inhibition. Surprisingly, I did not detect any obvious decrease in CENP-A levels in response to α-amanitin in mitosis. Alternatively, centromeric transcription in mitosis might somehow specifically affect CENP-C localization at centromeres. In contrast, I found that inhibition of transcription by α-amanitin significantly impaired centromeric cohesion. I further demonstrated that the cohesion defects were a result of mislocalization of Sgo1, the cohesin protector, in response to transcription inhibition.¹⁰ Sgo1, together with PP2A, localizes at inner centromeres, referred to the regions between two sister centromeres (Fig. 1), to protect the centromeric cohesion in mitosis, thus preventing chromosome missegregation and chromosome instability.^16-20 Our recent immunofluorescence microscopy demonstrated that Sgo1 is sequentially recruited to kinetochores and inner centromeres.¹⁰ The kinetochore and centromeric receptors for Sgo1 are histone H2A phosphor Thr120 (H2A pT120) and cohesin, respectively.^19,21 How is Sgo1 installed at inner centromeres after initial kinetochore recruitment? Our data strongly suggests that RNAP II transcription plays an important role in this process. Mitosis-specific inhibition of RNAP II transcription impaired the installment of Sgo1 at inner centromeres and retained it at kinetochores, which finally resulted in centromeric cohesion defects.¹⁰ Thus, I discovered a novel function in RNAP II transcription in centromeric cohesion protection and chromosome segregation.

Figure 1.

Putative model of Sgo1 installment at centromeres by RNAP II. When cells enter mitosis, Sgo1 and RNAP II are recruited to H2A-pT120 at kinetochores. Elongation of RNAP II or RNA transcripts displaces histones from chromatin, which might allow dislodgement of Sgo1 from H2A-pT120. The dislodged Sgo1 might bind to and travel together with RNAP II to the inner centromere, where Sgo1 binds to cohesin in a phosphorylation-dependent manner. The centromere is comprised of hundreds of thousands of repetitive DNA sequences. The kinetochore is a multiple-protein complex that assembles on some centromeric DNA. The inner centromere is referred to the specific cohesin-binding domain between two sister centromeres.

How does RNAP II transcription facilitate Sgo1 relocating from kinetochores to inner centromeres? Firstly, I have shown that Sgo1 physically interacts with RNAP II.¹⁰ Thus, Sgo1 could travel together with RNAP II and gradually approach inner centromeres where it binds with cohesin (Fig. 1). The distance from kinetochores to inner centromeres is about 0.5 μm in human cells. Thus, Sgo1 must undergo dynamic movement to reach inner centromeres from kinetochores. This notion is partially supported by our FRAP experiments showing that chromatin-bond Sgo1 was dynamically exchanging with the soluble pool.¹⁰ Therefore, RNAP II might function as a carrier to bring Sgo1 toward inner centromeres. Secondly, in mitosis, chromatins are highly compacted into chromosome structures. Due to the compacted structures, it might be difficult for Sgo1 to be directly loaded onto centromeric cohesin, which is embedded in chromosomes. RNAP II transcription might facilitate this process by loosening the compacted chromosome structures via transiently displacing histones from chromatins (Fig. 1). As a matter of fact, RNAP II transcription has been shown to alter chromatin structures in interphase.^22,23 Of course, these two possibilities are not mutually exclusive. When RNAP II transcription loosens the compacted chromatin structures, Sgo1 is either diffused into centromeres by itself or together with RNAP II or both (Fig. 1). Another interesting point is that Sgo1 might be recruited to kinetochores together with cohesin, and then transported together to inner centromeres where the Sgo1-cohesin complex exchanges with centromeric cohesin itself, thus finishing the loading of Sgo1. Thirdly, does RNA function in Sgo1 relocation from kinetochores to inner centromeres? A long non-coding α-satellite RNA with specific length has been identified and shown to maintain the centromere identity in interphase.⁷ But it is unclear whether this type of RNA or other ones existing at mitotic centromeres facilitate the Sgo1 relocation from kinetochores to inner centromeres. I have shown that Sgo1 directly interacts with RNA using the same region that binds to H2A pT120. I attempted to separate these two binding activities, but I failed to obtain the separation-of-function Sgo1 mutants. Therefore, in order to reveal the functional importance of RNA in Sgo1 relocation, more rigorous mutagenesis is needed to separate the two binding activities. Regardless, I uncovered a novel function of RNAP II transcription in sister-chromatid cohesion regulation. Recently, it has been demonstrated that centromeric cohesion is weakened in cancer cells.²⁴ So it would be interesting to investigate whether the RNAP II transcription-dependent Sgo1 installment at inner centromeres has been compromised in cancer cells.

Bub1, an important mitotic kinase, phosphorylates histone H2A pT120 to recruit Sgo1 to kinetochores.^19,21 I was wondering whether this is also a mechanism to recruit or maintain RNAP II at kinetochores in mitosis. Depletion of Bub1 or pharmacological inhibition of Bub1 kinase activity was sufficient to dislodge RNAP II from kinetochores, which inactivates the transcription at kinetochores.¹⁰ Thus, Bub1 is a major regulator of centromeric transcription in mitosis. As the currently known histone substrate of Bub1 is histone H2A, I reason that H2A pT120, in addition to recruiting Sgo1, may also serve an epigenetic mark to recruit RNAP II to kinetochores. If this is true, a further question is how H2A pT120 recruits or maintains RNAP II at centromeres during mitosis.

The first possibility is that H2A pT120 serves as a transcriptional initiating mark to recruit transcriptional factors or directly contributes to the recruitment of RNAP II (Fig. 2). The CTD serine 5 of RNAP II will be phosphorylated when it is in the state of initiation.^25,26 I did not detect any signal on immune-staining of the CTD serine 5; instead, I detected a signal on immune-staining of the CTD serine 2, the elongation signal for RNAP II. Therefore, I favor the possibility that H2A pT120 serves as a histone mark to promote elongation of RNAP II. How does H2A pT120 contribute to the elongation of RNAP II? Firstly, H2A pT120 could recruit chromatin remodeling factors to help open the compacted mitotic chromatin structures, thus promoting RNAP II elongation. Thus, in the future, it will be interesting to identify any chromatin remodeling complexes that are associated with H2A pT120. Secondly, it is possible that H2A pT120 recruits histone tail modifiers to modify the neighboring histone tails, thus promoting transcription (Fig. 2). Some epigenetic histone markers that promote active transcription, such as H3K4 dimethylation, have been identified in centromeric nucleosomes.²⁷ It will be tempting to test if these histone tail modifications are also regulated by Bub1 and H2A pT120. Finally, Bub1 might also phosphorylate other histone tails to promote transcription (Fig. 2). Therefore, it is necessary to firstly verify if H2A pT120 is important for promoting transcription, like Bub1 kinase.

Figure 2.

Possible mechanisms of Bub1-depenent RNAP II transcription. A detailed description can be found in the main text.

Accumulating evidence suggests that transcription is present at centromeres and plays important roles in centromere functions. In interphase, RNAP II transcription and its transcripts, likely α-satellite RNAs, help deposit or maintain CENP-A at centromeres, thus maintaining the centromere identity. But the molecular mechanisms are not understood. In mitosis, RNAP II transcription functions to install Sgo1 at centromeres. Bub1 and its kinase activity are essential for the kinetochore localization of RNAP II likely through H2A pT120. In future it will be tempting to understand the molecular mechanisms of Bub1-dependent transcription. Because H2A pT120 is not mitosis-specific and Bub1 kinase activity is constantly active during cell cycle,²⁸ it will be also tempting to address whether Bub1-dependent H2A T120 phosphorylation also regulates transcription globally in interphase. Knowing so will further help understand the function of this regulation.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgment

I thank Dr. Yanchang Wang for critical reading of this manuscript