Research Article: Five pillars of centromeric chromatin in fungal pathogens

Date Published: August 23, 2018

Publisher: Public Library of Science

Author(s): Vikas Yadav, Lakshmi Sreekumar, Krishnendu Guin, Kaustuv Sanyal, Donald C Sheppard.


Partial Text

DNA sequence features provide the necessary template to act as a binding platform for kinetochore proteins. The genus Candida, which harbors several pathogenic species, presents a diverse array of centromere types. Candida glabrata carries point genetic centromeres, much like the 125-bp DNA sequence that serves as a fully functional point centromere in the budding yeast Saccharomyces cerevisiae [3–5]. Typically, genetic centromeres have specific and conserved DNA sequence motifs and confer mitotic stability to otherwise unstable plasmids carrying an autonomous replicating sequence (ARS) during cell division. Despite high-structural homology in DNA sequence elements, the point centromeres of C. glabrata are not fully functional in S. cerevisiae, suggesting that centromere function is species-specific [5, 6]. Short regional genetic centromeres of Candida tropicalis comprise a central core flanked by inverted repeats, similar to those of the fission yeast Schizosaccharomyces pombe [3, 7]. The sequence and orientation of these repeats are important for centromere function. Due to the presence of inverted repeats, the centromeres in C. tropicalis can acquire a hairpin loop-like secondary structure that might be crucial for kinetochore assembly. Candida albicans and Candida dubliniensis, on the other hand, possess unique and different centromere DNA sequences on each of their chromosomes [8, 9]. While C. tropicalis centromeres can stabilize an ARS plasmid, indicating a role of DNA sequence in centromere identity, the same does not hold true for C. albicans. Cryptococcus neoformans, a basidiomycetous pathogen that diverged from C. albicans more than 900 million years ago, harbors large regional centromeres that are rich in centromere-specific retroelements [10–12]. While the presence of such retroelements hints toward the functional dependence on DNA sequence in centromere function, more studies are needed to explore such links. Some fungal centromeres possess specific DNA sequence features. For example, Candida lusitaniae (teleomorph Clavispora lusitaniae) and Malassezia sympodialis centromeres are present in highly AT-rich regions of the genome but lack any easily detectable conserved sequence motifs or repeats [13, 14]. Whether any AT-rich DNA sequence can act as a centromere in these organisms, similar to what is observed in diatoms [15], remains an open question.

The conserved centromere-specific histone H3 variant CENP-A/Cse4 is specifically present at all fungal centromeres identified to date but is largely excluded from other regions of the genome. Mucor circinelloides and Phycomyces blakesleeanus are notable exceptions in this regard, as they have no obvious CENP-A homologs, even though their centromeres are not yet physically mapped [16]. CENP-A is considered as the epigenetic determinant of centromere identity, as these molecules can seed the formation of a functional centromere in most organisms. This is supported by the fact that ectopic CENP-A incorporation can result in neocentromere formation, which is activated when an endogenous centromere becomes nonfunctional [17]. Although it is not well understood how CENP-A acts as the epigenetic determinant of the centromere, structural properties like a longer alpha N-terminal (αN) helix and the L1 loop region and biophysical features of the CENP-A nucleosome array, such as higher condensation properties, might be crucial for this role [18]. The process of CENP-A incorporation has been studied in C. albicans. Like other species, the CENP-A chaperone Holliday junction recognition protein/suppressor of chromosome mis-segregation 3 (HJURP/Scm3) is found to be crucial for CENP-A loading in C. albicans (our unpublished results). Regional centromeres harbor canonical histone H3 along with CENP-A. Post-translational modifications of histone H3 are crucial in forming a functional kinetochore [19]. Biochemical studies revealed the presence of heterochromatin histone marks such as dimethylation of histone H3 lysine 9 (H3K9diMe) across the centromeres of C. neoformans [10]. Apart from histone marks, DNA methylation has also been observed at the centromeres in C. neoformans, but its functional significance is unclear yet [10].

For a very long time, the centromere locus was considered heterochromatic and transcriptionally inert. While centromere regions are generally transcription poor, landmark studies in several yeast species revealed that small interfering RNAs (siRNAs) derived from pericentromeric regions are necessary for centromere function [20, 21]. These studies indicated that centromere transcription is permissible and has a functional significance. A pan–fungal analysis of RNA interference (RNAi) proteins revealed that a few species, including C. glabrata and Ustilago maydis, have lost all of the proteins required for functional RNAi during the course of evolution, whereas species including C. albicans and C. tropicalis harbor a cryptic RNAi machinery [22]. A recent study in the pathogenic Cryptococcus species complex correlated the loss of RNAi with the length of centromeres, thereby proposing that RNAi helps in maintaining long repetitive, transposon-rich centromeres [10]. Whether the RNAi machinery has a functional significance in the centromere biology of this species complex and other fungal pathogens remains unexplored. Apart from RNAi, centromere transcription can also play a functional role through long noncoding RNA (ncRNA). Indeed, pervasive levels of transcription have been documented in various fungal species [23, 24]. However, the transcripts generated from the centromeres are significantly low in number compared to the rest of the genome, as shown in the Cryptococcus and Ustilago species complex [10].

Unlike metazoans, most fungal species have early replicating centromeres [25–27]. This temporally distinct replication timing not only allows better tolerance toward replication stress but also ensures proper kinetochore assembly at the centromeres. In C. albicans, centromeres replicate early in every synthesis phase (S-phase) and are associated with an early firing replicating origin [26]. Additionally, the formation of a neocentromere advances the replication time of the flanking region by activating an early replicating origin. This proximity effect was explained by a replication-coupled repair mechanism in a kinetochore-dependent manner [28]. Centromere-proximal origins stall randomly at the centromere, leading to accumulation of single-stranded DNA, which then recruits the homologous recombination proteins such as Rad51 and Rad52. These proteins physically interact with CENP-A in C. albicans and load it at the site of stalled replication forks; that is, centromeres. How this process is regulated to occur only during S-phase remains unknown, with possible implications for the CENP-A chaperone Scm3. Based on studies in many other nonpathogenic fungal species, the physical proximity of a partitioning locus (centromere) and an initiator site (replication origin) is relevant when one dissects the functional aspects of genome maintenance. However, evidence toward this connection is just beginning to emerge.

Most fungal centromeres are clustered near spindle pole bodies (SPBs). This association may result in folding back of chromosomes and positioning them such that telomeres are juxtaposed in the interphase nucleus, giving rise to the Rabl conformation [29]. This phenomenon has been shown to occur in both animal and plant pathogens including C. albicans, C. tropicalis, and Fusarium graminearum. Using chromosome conformation capture (3C) experiments, clustered centromere DNA regions were shown to be present in close spatial proximity, leading to physical interactions between different centromeres [30–32]. It has been proposed that the clustering of centromeres aids in determining the site of centromere formation in these organisms. According to this hypothesis, a part of the nucleus is enriched with a pool of CENP-A proteins to form a CENP-A–rich zone or CENP-A cloud [33, 34]. It was proposed that the region of a chromosome that is near this CENP-A cloud would attract a higher level of CENP-A and thus serves as a preferred site for centromere formation. In S. cerevisiae, for example, a locally enriched population of accessory CENP-A molecules at pericentric chromatin has been shown to serve as a reservoir for rapid incorporation of CENP-A in the event of premature eviction from centromeres [35]. Further evidence supporting the CENP-A cloud hypothesis stems from studies in C. albicans in which neocentromeres were formed close to the native centromere in most cases [34]. In addition, neocentromeres in C. albicans change the spatial location to be a part of the centromere cluster by 3C experiments [31]. Interestingly, centromeres were found to be unclustered in premitotic C. neoformans cells that eventually cluster at the onset of mitosis [36]. Whether this centromere clustering also arises as a result of physical interactions among centromeres is not yet known.