Correlation analyses of ISW2 expression with expressions of nucleosomal histones
Histone 1 is involved in epigenetic silencing, and its expression and density are linked to the silencing activity (Willcockson et al., 2021). Isw2 binds DNA and histone 4 (Donovan et al., 2021). To investigate correlations between histones and the putative MoIsw2 for our strain, we used published expression data, from the course of rice leaf infection, from many different published experiments, as we have done previously for other genes (Zhang et al., 2019) (70-15). The MoHIS1 expression increases steeply with MoISW2 gene expression. The putative MoISW2 is also found to be co-regulated with a putative MoHIS4 in the downloaded data (Fig. 1). Thus, it could mean that the putative MoISW2 (Li et al., 2021) is a true MoISW2. The expression of MoHIS2B is related to the overall growth rate (Zhang et al., 2019) and is correlated with MoISW2, but not with a steep slope, indicating that a smaller amount of MoHis2B is needed for the DNA synthesis per cell. Similarly, for the expression of MoHIS3 but not for MoHIS2A, that are not significantly correlated with the putative MoISW2. (Fig. S1). The other histones expressions were less correlated with the putative MoISW2 (Fig. S1). Isw2 works together with ITC1 and His4; thus the expression of these three putative genes should be correlated in M. oryzae, and they are nicely correlated in the data from plant infection (Fig. S2).
Thus, the putative MoISW2 gene likely encodes a putative MoIsw2 protein involved in targeted local chromatin compaction working with MoHis1 and MoHis4 (Fazzio et al., 2005; Donovan et al., 2021; Willcockson et al., 2021). From now on, we skip "putative" as an attribute and investigate if MoIsw2 has the expected functions as an Isw2 and what other effects it likely has on the biology of the fungus. Also, the local chromatin compaction caused by MoIsw2 should be dynamic and more decisive the more the gene and the protein are expressed.
ChIP-seq analysis to find likely intergenetic palindromic DNA binding motifs
We used a MoISW2-GFP construct from our previous study (Li et al., 2021) to perform a ChIP-seq analysis to find conserved DNA binding motifs for the binding of MoIsw2 to M. oryzae DNA sequences. We looked especially for palindromic motifs in the sequences interacting with MoIsw2-GFP since our hits in the ChIP-seq data showed hits mainly for intergenic sequences (Supplemental Data 1), where transposable elements (TEs) are commonly located with TE target sites that are often palindromic (Linheiro and Bergman, 2008). In addition, TEs are involved in stress adaptation, and host specialization of M. oryzae (Chadha and Sharma, 2014; Yoshida et al., 2016), and our previous work showed MoIsw2 to be especially needed when plant host innate immune responses are known to be high (Li et al., 2021). In addition, interacting with a palindromic sequence that is likely to form hairpin structures in DNA (Ganapathiraju et al., 2020) can make "non-sliding" anchors for MoIsw2 to pull nucleosomes closer. We found the most common palindromic motif in 196 of our ChIP-seq DNA sequences using the MEME website, and there, the MEME search engine to search for common motifs (Fig. 2A. Supplemental Data 2). This motif has similarities with a staggered-cut palindromic target site involved in transposon cut and paste (Linheiro and Bergman, 2008).
Using a TOMTOM search at the MEME website, we could see that the found palindromic motif has similarities to a known Myb-binding protein containing DNA binding motif from humans (Fig. 2B Supplemental Data 3). The large number of palindromic hits and their alleged similarities with a human Myb since ISW2 contains a Myb-SANT domain (Li et al., 2021) made us focus on these 196 hits to investigate how and if genes are regulated around these sites in the genome to test if MoISW2 is involved in local chromatin compaction around the sites of these palindromic DNA motifs for MoISW2 DNA binding.
The binding of MoIsw2 regulates the genes closest to the motif binding site
We found that the196 motifs for binding sites were primarily intergenic, and we investigated the regulation of the genes closest to these 196 MoIsw2 binding sites to see if MoIsw2 restricts the regulation of the genes closest to the DNA binding site, as would be expected (Donovan et al., 2021). We searched all ChIP-seq sequences against the NCBI database and found good hits for 176 of the 196 motifs. Most of these hits were intergenic (113), while the rest (63) were in the promoter region. We next performed an RNAseq for the DMoisw2 strain and the background Ku80 strain, where we investigated the regulation of the gene with the hit in the promoter region or the closest gene to the intergenic binding site (Fig. 3). Only genes expressed at all in our datasets were included in the analysis. There was a general upregulation of the closest genes to the MoIsw2 binding site in theDMoisw2 mutant (Fig. 3A). For these genes, the DMoisw2 /Ku80 expression ratio is lower when the MoIsw2 is predicted to bind directly in the promoter region than for the closest gene of a predicted intergenic binding. When the absolute (positive or negative) regulation was considered, on the other hand, both types of genes with MoIsw2 binding in the promoter had similar absolute regulation indicating that many genes with MoIsw2 binding their promoter region are repressed in DMoisw2 (Fig. 3B). The latter supports MoIsw2 as an Isw2 protein that creates a local nucleosome condensation at specific nucleosomes (Donovan et al., 2021).
Several genes close to the MoIsw2 binding site were neither expressed in DMoisw2 nor Ku80, indicating that these genes could be permanently non-expressible potential pseudogenes without biological roles. We only use the term pseudogene in its limited sense (Cheetham et al., 2020) for genes that look like genes without any roles and are not expressed under many natural conditions. Since the overuse of the pseudogene term does not make sense, we follow the nomenclature suggested (Cheetham et al., 2020). Since the non-existence of a role can never be firmly tested experimentally, we prefer to call them potential pseudogenes. Our observation that these genes are or have become potential pseudogenes during evolution is exciting since some genes are annotated as avirulence genes (see Table 1). The data and analyses for this and the next section is available in a supplemental file (Supplemental Data 4)
Local gene regulations around MoIsw2 palindromic DNA binding sites fit the Isw2 specific DNA binding model and can likely be used to estimate the NRL size around the binding site.
To further investigate local regulation around the binding sites with the found palindromic motif, we investigated our RNAseq data and ordered the MGG_codes for the genes in the order of the genes on the supercontigs (Supercontig order downloaded from BROAD). We found that genes upstream and downstream, the genes closest to the binding site (+8/-8 genes), were generally differently regulated, and DMoisw2 differed from the background Ku80 in this respect.
Because of the targeted local nucleosome condensation, the gene regulation around the Isw2 binding site (Donovan et al., 2021) DMoisw2/W.T. regulation should be in waves around the DNA attachment point due to the likely interference pattern (phasing) between nucleosome repeat lengths (NRLs) and the average distance between gene promoter sequences, if the average gene distance is similar for most genes (Chereji et al., 2018). We first calculated the average gene distance in M.oryzae using the downloaded data from BROAD and found that the average distance is 4659bp for all genes ordered on supercontigs with a relatively small standard error of 203 bp (SEM=4%). We then searched for the best fit of a sinus function of the DMoisw2/Ku80 transcription using our gene expression data (RNAseq) of the +8/-8 genes surrounding the MoIsw2 binding site. We used MS Excel Solver to fit a reflecting sinus wave, reflecting with a minimum in the MoIsw2 attachment point. We used a sinus function since DNA are wrapped around and slides around nucleosomes of similar sizes, with a constant overall NRL (van Holde, 1989; Cutter and Hayes, 2015; Donovan et al., 2021; Willcockson et al., 2021) and NRLs sizes depending on the organism, cell type and cell status (van Holde, 1989). We found a general up-regulation at the point of DNA-interaction (Fig. 3) and waves of regulation on each side of the DNA-binding site for MoIsw2 probably due to changed nucleosome phasing affecting the in vivo transcription (Chereji et al., 2018) differently in DMoisw2 and the background Ku80. By searching for the best fit with the minimum wavelength (Fig. 4) that can create the observed data, we could approximate that the predicted average NRLs close to the MoIsw2 DNA attachment should be 211bp with an SEM of 9bp. That roughly agrees with Donovan et al. (2021) and what could be expected for not highly condensed NRLs (van Holde, 1989).
Since Isw2 interacts with histone 4 (Donovan et al., 2021), we made a few attempts to detect MoIsw2 interaction with a putative MoHis4 protein using a yeast two-hybrid assay. We could not detect any interaction solid and long enough to give a positive result. Thus, there might be no interaction, or the interaction is very transient; the latter is likely (see discussion).
Palindromic sequences between genes often target transposon sequences, and MoISW2 DNA binding the palindromic DNA motif can influence close-by avirulence genes expression.
Palindromic DNA are typical mobile elements of retrotransposons. Avirulence proteins are pathogenicity-related fungal proteins that the plant senses and mounts innate immunity reactions towards. These genes are generally situated close to known retrotransposons in the M. oryzae genome (Yoshida et al., 2016). Most of the 196 ChIP-seq sequences with palindromic binding motifs were found in known mobile elements (transposons).
Sub-hypothesis from the above observation: If avirulence gene expressions in M. oryzae are affected by MoISW2 targeted nucleosome condensation, then transposable elements interacting with MoIsw2 could be directly involved in the pathogen-plant arms race between pathogen pathogenicity and plant resistance. Moreover, these genes should be adjacent to the MoIsw2 binding site, and their expression should be affected differently depending on their closeness to the binding site.
To test this hypothesis, we first made a list of all known avirulence genes noted for M. oryzae at NCBI (Table 1). We found 16 avirulence genes of different types and the avirulence cluster for cytochalasan type compound biosynthesis (Collemare et al., 2008; Song et al., 2015) containing 12 more genes. The cytochalasan cluster is specifically activated at early hours post infection HPI during penetration and produces a secondary metabolite recognized by the R gene Pi33 in resistant rice cultivars (Collemare et al., 2008). Avr-PWL1 and Avr-PWL2 (Dioh et al., 2000), together with the previous, makes 18 classic avirulence type genes excluding the cytochalasan gene cluster genes (Table 1).
Most of the genes in the cytochalasan cluster are very close to the MoIsw2 palindromic DNA binding motif site, and several are differently expressed in 70-15 and 98-06 (Table 1), which might lead to the production of different final metabolites from the gene cluster even if the metabolites have similar functions as virulence factors.
Six of the other avirulence genes are not expressed in either strain (Table 1). Of these 6, half of the genes are close to the MoIsw2 binding site, and 3 are further away in 70-15. Two Avir genes (MGG_17614 and MGG_15611) are on supercontigs with unknown gene-order in our data, so it is unsure how far they are from a MoIsw2 binding site. The positioning of the MoIsw2 binding site close to retrotransposons and the differential regulation and expression variation make it likely that MoIsw2 specific targeting nucleosome binding is instrumental for stabilizing avirulence gene expression appropriately, concerning fungal-experienced plant host immune reactions. The data and analyses for this section is available in a supplemental file (Supplemental Data 5)
Table 1. Comparison of the position of avirulence genes in relation to the MoIsw2 palindromic DNA binding motif site in strain 70-15, as well as expression of the avirulence genes in MoISW2 knockout compared to the background; and expression of the same genes during infection of rice for strain 70-15 and strain 98-0 from published data. Yellow-marked cells; are genes that MoIsw2 downregulates in 70-15. Orange marked cells; genes with no expression could be potential pseudogenes in 70-15 and/or 98-0. Green marked cells; are genes that are only expressed during infection by one of the strains
ID.
|
Annotation
|
Within or outside the +8/-8 genes around the MoIsw2 binding site
|
More In KO (X)
|
Absent expression in both strains
(X)
|
Less in KO.
(X)
|
Only in 70-15
(X)
|
Only in 98-0
(X)
|
Expressed in both strains and in 70-15 infection data
|
Expression in WT 70-15 infection data
|
MGG_12447
|
cytochalasan
|
Within
|
X
|
|
|
|
|
X
|
X
|
MGG_08386
|
cytochalasan
|
Within
|
|
X
|
|
|
|
|
|
MGG_08377
|
cytochalasan
|
Within
|
X
|
|
|
|
|
X
|
X
|
MGG_08378
|
cytochalasan
|
Within
|
X
|
|
|
|
|
X
|
X
|
MGG_08380
|
cytochalasan
|
Within
|
|
|
|
|
X
|
|
|
MGG_08381
|
cytochalasan
|
Within
|
|
|
|
|
X
|
|
ND
|
MGG_08384
|
cytochalasan
|
Within
|
X
|
|
|
|
|
X
|
X
|
MGG_08389
|
cytochalasan
|
Within
|
|
|
X
|
|
|
x
|
X
|
MGG_08390
|
cytochalasan
|
Within
|
|
|
|
|
|
X
|
X
|
MGG_08391
|
cytochalasan
|
Within
|
|
|
|
|
X
|
|
|
MGG_15927
|
cytochalasan
|
Within
|
|
|
|
|
X
|
|
|
MGG_15928
|
cytochalasan
|
Within
|
X
|
|
|
X
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Avirulence genes
|
|
|
|
|
|
|
|
|
MGG_07199
|
ATR13, RxLR effector
|
Outside
|
|
|
|
|
|
X
|
X
|
MGG_10556
|
Avr_Pii
|
4 genes away
|
|
X
|
|
|
|
|
|
MGG_17614
|
Avr_Pii
|
nonordered
|
|
|
X
|
X
|
|
|
ND
|
MGG_13283
|
Avr_Pik
|
Outside
|
|
X
|
|
|
|
|
|
MGG_15972
|
Avr_Pik
|
nonordered
|
X
|
|
|
|
|
X
|
X
|
MGG_03029
|
Avr_Pita1
|
Outside
|
|
|
X
|
|
|
X
|
X
|
MGG_07038
|
Avr_Pita1
|
Outside
|
|
|
X
|
|
|
X
|
X
|
MGG_09617
|
Avr_Pita1
|
1 genes away
|
|
X
|
|
|
|
|
|
MGG_10927
|
Avr_Pita1
|
13 genes away
|
X
|
|
|
|
|
X
|
X
|
MGG_03808
|
Avr_Pita1 like
|
Outside
|
|
X
|
|
|
|
|
|
MGG_15370
|
Avr_Pita1 like
|
1genes away
|
X
|
|
|
|
|
X
|
X
|
MGG_17611
|
Avr_Pita1 middle
|
nonordered
|
X
|
|
|
|
|
|
ND
|
MGG_14981
|
AVR_PiTA2
|
Outside
|
|
X
|
|
|
|
|
|
MGG_15212
|
AVR_Pita2
|
4 genes away
|
|
|
|
|
X
|
|
|
MGG_18041
|
Avr_Piz-t
|
nonordered
|
|
|
|
|
|
X
|
ND
|
MGG_03685
|
Avr-Pi54
|
Outside
|
X
|
|
|
|
|
X
|
X
|
MGG_13863
|
Avr-PWL1
|
2genes away
|
|
|
|
|
X
|
|
|
MGG_07398
|
Avr-PWL2
|
Outside
|
|
|
|
X
|
|
|
|
RNA-seq data of DMoisw2 and Ku80 indicates that MoIsw2 regulates the balance between glycolytic growth with biosynthesis, and aerobic catabolism
For DMoisw2 grown on MM-medium, 339 genes were significantly upregulated compared to the background Ku80. The two largest significantly overrepresented gene categories were for secondary metabolism and DNA-binding, while most of the other categories are related to DNA synthesis and DNA-related activities (Fig. 5A). These are the genes MoISW2 likely negatively regulates in Ku80. Most of the downregulated genes in the DMoisw2 are genes involved in mitochondrial electron transport and other mitochondrial processes. In other words, our results indicate that MoIsw2 is involved in regulating the balance between DNA synthesis, which is safest without ATP generation by mitochondrial respiration that generates Radical Oxygen Species (ROS) that is a necessary biproduct of respiratory metabolism (Klevecz et al., 2004).
Regulation of DNA-binding genes closest to the MoIsw2 binding site
Since 65 genes above were classified as DNA-binding and were among the enriched genes upregulated in DMoisw2 (Fig. 5A), we counted the number of genes positioned between these genes and the MoIsw2 binding sites for all the DNA-binding genes. The products of some of these genes could be transcription factors directly under the control of MoISW2. We found that the most upregulated DNA binding genes in the DMoisw2 are physically closest to the MoIsw2 palindromic DNA binding motif site, indicating that deleting MoISW2 gives better DNA access for the promoters of DNA-binding genes situated close to the MoIsw2 palindromic DNA binding motif site (Fig. 6).
Table 2. Genes close to the MoIsw2 binding motif and most upregulated in DMoisw2 compared to the background Ku80-strain with annotation from NCBI
|
MGG_02762
|
MGG_02762. ATP-dependent RNA helicase DED1; Belongs to the DEAD box helicase family
|
MGG_06470
|
MGG_06470, DNA repair helicase RAD25 (835 aa).
|
MGG_05948
|
MGG_05948 zinc knuckle domain-containing protein. This domain is a zinc-binding domain of the form CxxCxxxGHxxxxC from various species. It is found in the MPE1 protein from Saccharomyces cerevisiae which is a component of the cleavage and polyadenylation factor (CPF) complex important for polyadenylation-dependent pre-mRNA 3'-end formation
|
MGG_01990
|
b-ZIP transcription factor IDI-4 (induces autophagic cell death in Podospora)
|
MGG_02429
|
KOG4062 6-O-methylguanine-DNA methyltransferase MGMT/MGT1, involved in DNA repair Replication, recombination and repair https://jgi-myco-web-4.jgi.doe.gov/annotator/servlet/jgi.annotation.Annotation?pDb=Lasov1&pStateVar=View&pProteinId=680852&pViewType=protein
|
MGG_04428
|
Zinc finger transcription factor ace1
|
MGG_05995
|
MGG_05995, Magnaporthe oryzae 70-15 hypothetical protein (244 aa), translin family protein, Translin family (PF01997). If Translin here is a review about that https://link.springer.com/article/10.1007/s12038-019-9947-6
|
MGG_04429
|
ATP-dependent DNA helicase MPH1; ATP-dependent DNA helicase is involved in DNA damage repair by homologous recombination and genome maintenance.
|
Of the DNA-binding genes most suppressed by MoIsw2 activity and closest to the MoISW2 binding site, only two encode conventional TFs; the others are mainly involved in DNA repair (Table 2) needed for DNA synthesis and growth.
To get further information about the genes MoIsw2 nucleosome interactions directly can regulate, we investigated which genes were situated +8/-8 genes from the MoISW2 binding sites (Supplemental Data S4) in the M. oryzae genome using FunCat classification. In the background possessing MoIsw2, these genes are more likely to be regulated (positive or negative). These genes can be expected to be most influenced by MoIsw2 chromatin-modifying activities. Interestingly, significantly enriched genes that are physically close to the MoIsw2 binding sites are gene categories indicative of interaction with the abiotic and biotic environment, while significantly depleted genes are related to DNA synthesis and growth. The depleted genes are, in principle, indicative of housekeeping. Thus, combined with the above, it appears that MoISW2 is a switch between growth (DNA synthesis) and interaction with the environment.
MoIsw2 is likely involved in the regulation of core secondary metabolism genes
Interactions with the environment (biotic and abiotic) and secondary metabolism are two categories that, together with the regulation of avirulence genes and closeness to transposons, could indicate that MoISW2 is instrumental in creating Non-Conserved Regions (NCRs) in M. oryzae. These gene categories positioning in NCRs have been characterized in Fusarium. graminearum compared to conserved regions containing housekeeping genes (Zhao et al., 2014). To investigate the positioning of secondary metabolite genes, we used the Anti-SMASH website and searched the complete MoDNA sequence for core secondary metabolite genes, so we are not dependent on previous annotations from the literature. We found 54 core secondary metabolite genes (Supplemental Data 6). Of these genes, 19 are positioned +8/- 8 genes from the MoIsw2 binding sites, and 25 are within the double-distance from the binding site. Using the Fisher exact test (Table S1) and testing against random distribution, the P-value for the null hypothesis that the distribution is random is 2.18E-08 for the 19 genes closest to the MoIsw2 binding site and 1.93E-08 for the double gene distance from the binding site. In other words, there is a highly significant overrepresentation of secondary metabolite genes close to the MoIsw2 palindromic DNA binding motif site. Most of the 19 closest genes are also differentially regulated by MoIsw2, indicating that MoIsw2 seems to regulate their expression (Fig. 8) actively.
MoIsw2 binding to DNA sequences closest to or inside genes
We now turn back to the rest of the ChIP-seq data. We have above shown that genes under MoISW2 control that are more expressed in the background Ku80 strain compared to the DMoisw2 strain are enriched for gene classes characteristic for secondary metabolism and biomass growth (anabolism) (Fig. 5A), while the downregulated genes in the mutant are genes connected to aerobic metabolism and stress (Fig. 5A).
In the MoIsw2 ChIP-seq data, we investigated the FunCat classification of all genes with sequence hits. First, we removed double hits not to count the same gene twice. We then investigated the list of genes closest to the MoIsw2 palindromic DNA binding motif site. These genes should be targeted by MoIsw2 and might be under positive or negative control from MoIsw2. According to the previous analysis, they should not belong to growth (anabolism) but be genes active when oxygen is consumed and the substrates oxidized (catabolism) (Klevecz et al., 2004; Machné and Murray, 2012). In addition, the binding sequences close to avirulence genes and retrotransposons, so their regulation can shift depending on retrotransposon transpositions. Since ChIP-seq only shows potential binding to DNA in vivo, we cannot say if these genes are up or downregulated, only that regulation influenced by MoIsw2 can occur. Genes hits closest to the MoIsw2 palindromic DNA binding motif site are mainly enriched for secondary metabolite genes and other gene classes essential for biotic interactions (Fig. 9A), while specific gene categories for biomass growth and housekeeping (anabolism) are depleted (Fig. 9B).
Genes with MoISW2 binding their promoter, or in their exons, or introns are other genes likely regulated by MoIsw2 nucleosome positioning activity during anabolism/catabolism shifts. These genes are many more than those close to MoIsw2 palindromic DNA motif sites, and we detect more gene classes (Fig. 10A), including detoxification, signaling (cyclic nucleotide-binding), disease, and defense as classes important in biotic and abiotic interactions with the environments. Depleted are again gene classes characteristic for anabolism (biomass growth) (Fig. 10B)
Finally, as shown in Fig. 11, the genes closest to ChIP-seq hits in intergenic regions, not containing the MoIsw2 palindromic DNA binding motif site, are again genes enriched for secondary metabolite genes and genes involved in abiotic and biotic interactions (Fig. 11A). In contrast, genes needed for biomass growth are again depleted (Fig. 11B).