Cellular dynamics and genomic identity of centromeres in Rice Blast

A series of well-synchronized events mediated by kinetochore-microtubule interactions ensure faithful chromosome segregation in eukaryotes. Centromeres scaffold kinetochore assembly and are among the fastest evolving chromosomal loci in terms of the DNA sequence, length, and organization of intrinsic elements. Neither the centromere structure nor the kinetochore dynamics is well studied in plant pathogenic fungi. Here, we sought to understand the process of chromosome segregation in the rice blast fungus, Magnaporthe oryzae. High-resolution confocal imaging of GFP-tagged inner kinetochore proteins, CenpA and CenpC, revealed an unusual albeit transient declustering of centromeres just before anaphase separation in M. oryzae. Strikingly, the declustered centromeres positioned randomly at the spindle midzone without an apparent metaphase plate per se. Using chromatin immunoprecipitation followed by deep sequencing, all seven centromeres were identified as CenpA-rich regions in the wild-type Guy11 strain of M. oryzae. The centromeres in M. oryzae are regional and span 57 to 109 kb transcriptionally poor regions. No centromere-specific DNA sequence motif or repetitive elements could be identified in these regions suggesting an epigenetic specification of centromere function in M. oryzae. Highly AT-rich and heavily methylated DNA sequences were the only common defining features of all the centromeres in the Rice Blast. Detailed gene synteny analyses helped identify and compare the centromere regions in distinct isolates of M. oryzae and its related species Magnaporthe poae. Overall, this study identifies unusual centromere dynamics and maps the centromere DNA sequences in the top model fungal pathogen M. oryzae, which causes severe losses in global rice production. Significance Magnaporthe oryzae is an important fungal pathogen that causes an annual loss of 10-30% rice crop due to the devastating blast disease. In most organisms, kinetochores are arranged either in the metaphase plate or are clustered together to facilitate synchronized anaphase separation of chromosomes. In this study, we show that the initially clustered kinetochores separate and position randomly prior to anaphase in M. oryzae. Centromeres, identified as the site of kinetochore assembly, are regional type without any shared sequence motifs in M. oryzae. Together, this study reveals atypical kinetochore dynamics and identifies functional centromeres in M. oryzae, thus paving the way to define heterochromatin boundaries and understand the process of kinetochore assembly on epigenetically specified centromere loci.


Introduction 1
Faithful chromosome segregation is one of the essential processes that is required for 2 maintaining genome integrity in dividing cells. This process is successfully carried out by the 3 attachment of microtubules, emanating from opposite spindle poles, to the proteinaceous multi-4 subunit structure, the kinetochore, that is pre-assembled onto centromeres (1, 2). The centromere 5 forms a crucial part of this machinery and yet, it is one of the most rapidly evolving loci in 6 eukaryotic genomes (3,4). On the contrary, the proteins that bind to centromere DNA are 7 evolutionary conserved (2). Centromere DNA shows a wide diversity in the length and 8 composition of the underlying DNA sequence. A few fungal species, like Saccharomyces 9 cerevisiae, harbor centromeres that are less than 400 bp comprising of conserved DNA sequence 10 elements to form point centromeres (5). Most others possess regional centromeres that span from 11 few kilobases to several megabases. Unlike point centromeres, regional centromeres in an 12 organism often do not share any conserved DNA sequence features. For example, the regional 13 centromeres in Schizosaccharomyces pombe and Candida tropicalis have a homogenized central 14 core flanked by inverted repeats (6, 7). Likewise, the regional centromeres in Cryptococcus 15 neoformans possess specific retrotransposons that are present randomly therein (8). In contrast, 16 centromeres in Candida albicans, Candida lusitaniae, and Candida dubliniensis differ between 17 all chromosomes and lack a conserved DNA sequence element (9-11). Centromeres in 18 filamentous fungi like Neurospora crassa, on the other hand, span long stretches of repetitive 19 DNA but lack a consensus sequence or pattern (12,13). Metazoans and plants also have regional 20 centromeres that are up to few megabases long, and mostly consist of repetitive DNA or 21 transposons (14-16).

22
Despite this sequence divergence, centromeres in most studied organisms are bound by a 1 centromere-specific histone H3 variant CENP-A /CenH3/Cse4, also known as the hallmark of 2 centromere identity (4, 17). CENP-A forms the foundation of the kinetochore assembly and is 3 essential for cell viability in all organisms studied until date. Evolutionary conservation of 4 CENP-A along with other kinetochore proteins also provides an efficient tool to identify 5 centromeres. Additionally, studies with fluorescently-labeled inner kinetochore proteins such as 6 CENP-A or CENP-C/Cen-C/Mif2 has led to an understanding of spatial dynamics of the 7 kinetochore within the nucleus (18)(19)(20)(21)(22). These studies established that kinetochores in most yeast 8 species are clustered throughout the nuclear division, and unlike metazoan CEN, do not align on 9 a metaphase plate. However, more recently, some variations to the metaphase plate or 10 kinetochore clustering have been reported revealing the diversity in this phenomenon.

11
Kinetochores remain clustered throughout the cell cycle in two well-studied ascomycetes, S. 12 cerevisiae, and C. albicans (23,24). In S. pombe, kinetochores undergo a brief declustering 13 during mitosis but remain clustered otherwise (18,25). Another ascomycete, Zymoseptoria 14 tritici, shows multiple kinetochore foci instead of a single cluster during interphase although 15 their localization dynamics during mitosis remains unexplored (26). On the other hand, the cells 16 of a basidiomycete C. neoformans display multiple foci of kinetochores in interphase, but 17 kinetochores gradually cluster during mitosis (19,22). Even the phenomenon of 18 centromere/kinetochore clustering is observed in Drosophila that depends on centric chromatin 19 rather than specific DNA sequences (27). 20 Besides CENP-A, several other chromatin features are known to be associated with 21 centromeres. For example, centromeres are devoid of genes/ORFs and exhibit a significantly low 22 level of polyA transcription as compared to the rest of the genome (8, 28). Furthermore, 23 6 centromeres in many organisms are heterochromatic in nature and harbor the heterochromatic 1 marks like H3K9di/trimethylation and DNA methylation (8,13,29). A preference for AT-rich 2 DNA sequence is evident for centromere formation in some organisms (13,(30)(31)(32). It is 3 important to note that none of these features exclusively define centromeres and, in most cases, 4 the importance of an individual factor in defining centromere loci is not well understood.

5
However, the presence of such features on discrete chromosomal loci may pave the way for 6 predicting centromeres in organisms in which genome tractability is difficult.

7
Magnaporthaceae is a family of ascomycete fungi comprising of many important plant 8 pathogenic species including Magnaporthe oryzae and Magnaporthe poae. M. oryzae causes the 9 devastating blast disease in cereal crops including rice, wheat, barley and millet (33, 34). M. 10 poae is responsible for summer-patch disease in turf grasses (35). M. oryzae, also known as rice 11 blast, is a constant threat to agriculture-based economies due to significant damage to rice 12 harvests. In recent years, rice blast has also emerged as a model pathosystem for studying host-13 pathogen interactions due to the availability of the genome sequence, fully characterized 14 infection cycle, genetic tractability and economic significance of the fungus (36, 37). However, 15 even with the availability of the genome sequence and annotated assembly, the 16 centromere/kinetochore identity of the rice blast fungus remains unexplored or poorly defined.

17
Here, we first studied and characterized orthologs of CENP-A and CENP-C, two well-conserved 18 kinetochore proteins, to understand the kinetochore dynamics in this organism and used these 19 kinetochore proteins as tools to identify bona fide centromeres.

1
Kinetochores are clustered during interphase in M. oryzae 2 A subset of putative kinetochore proteins was previously annotated in M. oryzae (12). We 3 expanded the list further by identifying putative orthologs of the remaining conserved 4 kinetochore proteins using in silico predictions (SI Appendix, Table S1). Multiple sequence 5 alignment established the identity of at least two most conserved inner kinetochore proteins: 6 CenpA (MGG_06445, a homolog of CENP-A) and CenpC (MGG_06960, a homolog of CENP-7 C) (SI Appendix, Figure S1) in M. oryzae. CenpA and CenpC of M. oryzae share 73% and 42% 8 sequence identity with their N. crassa counterparts CenH3 and CEN-C, respectively. Next, we 9 functionally expressed the GFP-tagged CenpA and CenpC from their native genomic loci in the 10 wild-type Guy11 strain of M. oryzae. GFP-CenpA and CenpC-GFP signals appeared as a single 11 dot-like and co-localized on chromatin, marked by mCherry-tagged histone H1 ( Figure 1A, 1B).

12
Further, co-localization of CenpA and CenpC signals confirmed their overlapping spatial 13 positions in both mycelia and conidia ( Figure 1C). Clustering of kinetochores is a hallmark 14 feature of many yeast and fungal genera. Such clustered kinetochores are often found in close 15 proximity to the spindle pole bodies (SPBs) (38). We localized SPBs by tagging Alp6 (ortholog 16 of S. cerevisiae Spc98) with mCherry and observed that SPBs localize close to the clustered 17 GFP-CenpA signals in M. oryzae ( Figure 1D). These results indicate that kinetochore 18 localization during interphase in M. oryzae is similar to that observed in other ascomycetes. Our  During interphase, the microtubules are mostly localized throughout the cytoplasm (SI 8 Appendix, Figure S2B). Live-cell imaging during mitosis revealed dispersed GFP-CenpA signals 9 localized along the mitotic spindle ( Figure 2A and Movie S1). Strikingly, the declustered dot-10 like signals of GFP-CenpA then segregated into two halves in a non-synchronous manner. Once 11 segregated, the signals began to cluster again and localized as two bright foci close to poles of 12 the mitotic spindle. To further probe the dynamics of kinetochore segregation, we performed 13 high-resolution imaging in mitotic cells expressing GFP-CenpA ( Figure 2B, C, Movie S2 and 14 S3). We observed that while the GFP-CenpA signals were spread out, they were localized in 15 pairs, most likely representing the segregated kinetochore signals ( Figure 2B, time 00:32). We 16 were able to count fourteen discrete spots of GFP-CenpA corresponding to 14 kinetochores of 17 the seven duplicated chromosomes. These results suggest that kinetochores in M. oryzae remain 18 largely unclustered during mitosis. It was further supported by co-localization of GFP-CenpA 19 with a SPB markerAlp6-mCherry during the mitotic stages ( Figure 2D). In pre-mitotic cells, we 20 observed two duplicated spots of Alp6-mCherry that co-localized with replicated clustered GFP-21 CenpA signals. During mitosis, GFP-CenpA signal localized as multiple puncta scattered in 22 between the two SPBs represented by Alp6-mCherry. After the division, the GFP-23 CenpA/kinetochores clustered again and localized adjacent to the SPBs (SI Appendix, Figure   1 S2C, and Movie S4). Taken together, we conclude that kinetochores decluster during mitosis in 2 M. oryzae, and align along the mitotic spindle. Furthermore, we infer that an equatorial plate 3 alignment of the kinetochores is not evident in M. oryzae, indicating a lack of a well-defined 4 metaphase plate. Similar dynamics of the kinetochore and microtubules were observed in M. 5 oryzae cells during pathogenic development and in planta conditions (SI Appendix, Figure S3, 6 Movie S5, and S6). Based on these observations, we propose a schematic model for the 7 kinetochore and SPB dynamics during the mitotic cycle in rice blast where kinetochore 8 clustering-declustering dynamics is most likely dependent on their direct link to the SPBs 9 ( Figure 2E). During mitosis, this link is likely broken, and the clustering is thus perturbed. We 10 infer that such timely and dynamic kinetochore clustering/declustering is crucial for proper 11 chromosome segregation in M. oryzae.

13
Kinetochore protein binding identifies regional centromeres in M. oryzae 14 CenpA binding is a hallmark of functional centromeres in eukaryotes (4, 15). We used 15 GFP-CenpA as a tool for molecular identification of centromeres in the M. oryzae genome. We  Table 1 and SI Appendix, Figure S4). The CenpA binding spans a 57 to 109 kb region 21 suggesting that M. oryzae possesses large regional centromeres. The centromere identity of these 22 regions was further validated by binding of another evolutionarily conserved independent 23 kinetochore protein CenpC. ChIP-qPCR using the fungal strain expressing CenpC-GFP (SI 1 Appendix, Figure S5A) confirmed specific overlapping binding of CenpA and CenpC on each of 2 these seven CEN regions. We also observed an additional region of 1200 bp on Contig 4 apart 3 from the seven distinct peaks in CenpA ChIP-seq analysis. The enriched peak was found to be 4 present on the gene encoding the vacuolar morphogenesis protein AvaB (MGG_01045). Using 5 specific ChIP-qPCR primers for this region, we confirmed that the aforementioned CenpA 6 enrichment on Contig 4 was likely an artifact (SI Appendix, Figure S5B). Overall, the binding of 7 two independent kinetochore proteins at seven long regions confirmed that these are indeed 8 authentic centromeres of the corresponding chromosomes in M. oryzae.

9
A detailed analysis revealed that the seven centromeres in M. oryzae comprise of highly 10 AT-rich sequences (≥67%) ( Figure 3B and Table 1). The centromeres in M. oryzae harbor a few 11 repetitive elements, mostly retrotransposons belonging to long terminal repeat (LTR) elements 12 ( Figure 3B, Dataset S1). However, these elements are neither exclusive to the centromeres nor  oryzae. Additionally, we also infer that centromeres in M. oryzae are large, regional and lie 22 within transcriptionally-poor 5mC-rich DNA regions of the genome. CenpA ChIP-seq reads on to the MG8 assembly. This analysis revealed seven distinct peaks, one 9 on each chromosome ( Figure 4A, SI Appendix, Figure S7B, and Table S2). We also observed 10 two additional CenpA-enriched regions in the unassembled Supercontig8.8 of MG8 assembly for 11 70-15 (SI Appendix, Figure S7C). Additionally, the identified centromere on chromosome 7 in 12 this assembly mapped to the same region that was previously predicted to harbor the centromere 13 based on genetic analysis (44). 14 Next, we analyzed the recently published PacBio genome sequence/assembly of the M. 15 oryzae field isolate FJ81278 (39) to identify the centromere sequences and compare them with 16 the 70-15 assembly. Mapping of CenpA ChIP-seq reads revealed nine distinct peaks in the 17 FJ81278 genome assembly (SI Appendix, Figure S8, and Table S2). Three of these enriched 18 regions were present at the end of three separate contigs (Contig 3, 14 and 16). By comparing 19 genome assemblies of 70-15 and FJ81278, we concluded that contigs 3 and 14 are most likely 20 parts of the same chromosome and the CenpA-enriched regions observed in these two contigs 21 represent a single centromere (CEN4). Synteny analysis also revealed that the CenpA peaks in 22 Contig 11 and 16 belong to the same chromosome. However, Contig11 of FJ81278 assembly 23 seems to be mis-assembled, since a part of this contig does not show synteny with any region of 1 the 70-15 genome. Thus, we excluded this centromere (CEN7) region from further analysis. 4B and SI Appendix, Figure S9). However, a major part of the centromere sequences was found 7 to be missing from the 70-15 genome assembly as compared to Guy11 and FJ81278. It is 8 important to note that the MG8 version of the 70-15 genome assembly is not complete and 9 harbors a number of gaps. We believe that some of the centromere sequences are part of the  (35). We were able to identify eight putative centromere regions 2 across the M. poae genome based on this in-silico analysis (SI Appendix, Figure S11, and Table   3 S3). Three of these eight putative CEN regions were present at the end of different contigs. Since  Magnaporthe, as reported in several fungal species complex before (6,8,10,12,45,46).

2
Kinetochores cluster together in a single locus at the nuclear periphery in many fungi.

3
This locus is often referred to as the CENP-A-rich zone or CENP-A cloud (47, 48). It has been 4 proposed that such a nuclear subdomain with a high concentration of CENP-A favor centromere 5 seeding on the chromosomal regions in close proximity to it, in the absence of a centromere-6 specific DNA sequence. In most budding yeasts, kinetochores are clustered throughout the cell 7 cycle except in C. neoformans, which shows clustered kinetochores only during mitosis (19).

8
The kinetochore dynamics in M. oryzae is found to be similar to the "fission" yeast rather than 9 that of the budding yeast species. It is possible that mitotic declustering of kinetochores is a 10 feature of all yeasts/fungi that divide by septum formation. However, a more detailed analysis of 11 kinetochore behavior in filamentous fungi like N. crassa and Z. tritici will be useful to establish  budding yeast, S. cerevisiae, is also highly AT-rich (54). A recent study reports the presence of 1 AT-rich centromeres of varying lengths in diatoms (30). Furthermore, the 171-bp alpha satellite 2 repeat DNA present in human centromeres is also AT-rich in nature (55). Overall, these results 3 suggest that AT-richness favors centromere function in many organisms. Intriguingly, in vitro 4 experiments suggest that CENP-A binds with a lower affinity to an AT-rich DNA sequence (56).

5
In contrast, the same study also revealed that the CENP-A chaperone Scm3 has a higher affinity 6 towards AT-rich sequences. With more AT-rich centromeres being characterized, identifying the 7 exact role of AT-rich sequences in centromere function is critical. with the host. In this study, we find that the centromere location is close to these repeat clusters 20 in some but not all chromosomes. Our results raise the possibility that centromere sequences in 21 M. oryzae are prone to repeat-mediated evolution.

22
A comparison between two M. oryzae isolates, Guy11 and FJ81278, revealed that while 1 the overall CEN DNA sequence between the two isolates is very similar, the repeat content at the 2 centromeres of orthologous chromosomes varies widely. It is known that centromere DNA 3 sequence among isolates of N. crassa can be different (12). The CEN sequences identified here 4 would pave the way for a more detailed comparative analysis of centromeres in diverse isolates 5 of M. oryzae. Such analyses will provide valuable insights into centromere evolution in this 6 species and the potential impact of host factors on this process. A comparative genome analysis 7 between M. oryzae and M. poae revealed the presence of a higher density of repeats in the latter.

8
Overall, these results suggest that while the centromere DNA sequence properties, not the DNA 9 sequence per se, remain conserved in this species complex, the centromere architecture is 10 divergent and might have been shaped by the repeat elements. Further studies will provide more 11 insights into the evolution of centromere DNA sequences and its possible link to host adaptation 12 and variability in virulence within the Magnaporthe species complex.

15
Wild-type M. oryzae strain Guy11 (MAT1-2; a kind gift from Lebrun group, France) was 16 used as the parent strain for all the experiments conducted in this study (except for the results 17 shown in SI Appendix, Figure S3, Movie S5 and S6 that were performed using B157 strain). The 18 strains thus validated and used in this study are listed in SI Appendix,