Altered nucleosome positions in maize haplotypes and mutants of a subset of SWI/SNF‐like proteins

Abstract

Chromatin remodelers alter DNA–histone interactions in eukaryotic organisms and have been well characterized in yeast and Arabidopsis. While there are maize proteins with similar domains as known remodelers, the ability of the maize proteins to alter nucleosome position has not been reported. Mutant alleles of several maize proteins (RMR1, CHR101, CHR106, CHR127, and CHR156) with similar functional domains to known chromatin remodelers were identified. Altered gene expression of Chr101, Chr106, Chr127, and Chr156 was demonstrated in plants homozygous for the mutant alleles. These mutant genotypes were subjected to nucleosome position analysis to determine whether misregulation of putative maize chromatin proteins would lead to altered DNA–histone interactions. Nucleosome position changes were observed in plants homozygous for chr101, chr106, chr127, and chr156 mutant alleles, suggesting that CHR101, CHR106, CHR127, and CHR156 may affect chromatin structure. The role of RNA polymerases in altering DNA–histone interactions was also tested. Changes in nucleosome position were demonstrated in homozygous mop2‐1 individuals. These changes were demonstrated at the b1 tandem repeats and at newly identified loci. Additionally, differential DNA–histone interactions and altered gene expression of putative chromatin remodelers were demonstrated between different maize haplotypes.

1. INTRODUCTION

Eukaryotic DNA is organized into a compact arrangement of DNA and proteins referred to as chromatin. The basic unit of chromatin is the nucleosome, which is made up of 150 base pairs of DNA wrapped 1.65 times around a histone octamer that is made up of two each H2A, H2B, H3, and H4 histones. DNA is further packaged with the addition of histone H1 and scaffold proteins to linker DNA between nucleosomes. While these DNA–histone interactions are in part determined by the DNA sequence itself, they can also be determined by chromatin remodeling complexes (reviewed by Jiang & Pugh, 2009) that utilize ATP hydrolysis to generate energy for changing DNA–histone interactions (reviewed by Kwon & Wagner, 2007).

Chromatin remodeling is the ATP‐dependent noncovalent alteration of DNA–histone interactions (reviewed by Kwon & Wagner, 2007). Chromatin remodeling is considered biologically significant because it is associated with changes in transcriptional activity and other phenotypes (reviewed by Kwon & Wagner, 2007). One of the four major families of chromatin remodeling complexes is switching defective/sucrose nonfermenting (SWI/SNF) (reviewed by Clapier & Cairns, 2009). The motor proteins of SWI/SNF belong to the SNF2 family of proteins and are characterized by their conserved SNF2_N and Helicase_C domains (Eisen, Sweder, & Hanawalt, 1995; Flaus, Martin, Barton, & Owen‐Hughes, 2006). Several of the SNF2 motor proteins have demonstrated chromatin remodeling ability, including S. cerevisiae (Sc) ScSNF2 (Laurent, Treich, & Carlson, 1993), ScRAD54 (Alexeev, Mazin, & Kowalczykowski, 2003; Petukhova, Van Komen, Vergano, Klein, & Sung, 1999), and A. thaliana (At) AtDDM1 (Brzeski & Jerzmanowski, 2003). AtDDM1 has also been shown to function in the maintenance of DNA methylation, in genomewide transposon and repetitive sequence methylation, and in the spread of silencing‐related histone modifications (Ito et al., 2015; Jeddeloh, Bender, & Richards, 1998; Jeddeloh, Stokes, & Richards, 1999).

Maize (Zea mays) proteins highly similar to the Arabidopsis AtDDM1 (Jeddeloh et al., 1998), AtDRD1 (Kanno et al., 2004), and AtCLSY1 (Smith et al., 2007) have been identified. These include the homologs CHR101 and CHR106 (Li et al., 2014), CHR127 and CHR156, and RMR1 (Hale, Stonaker, Gross, & Hollick, 2007) and CHR167 (Haag et al., 2014). While the biochemical activity of these proteins has not been reported, various phenotypes associated with mutant alleles of some proteins have been described. Among these, the extensively studied RMR1 is highly similar to AtCLSY1 and has been implicated in the spontaneous epigenetic silencing of a transgene (McGinnis, Springer, Lin, Carey, & Chandler, 2006) and the regulation of b1 and pl1 transcript abundance (Hale et al., 2007). It has also been found to interact with RNA polymerase IV (PolIV), along with CHR167 (Haag et al., 2014) and AtCLSY1 (Zhang et al., 2013). The maize proteins CHR101 and CHR106 are highly similar to AtDDM1 (Li et al., 2014). Similar to phenotypes described for Arabidopsis ddm1 homozygous plants (Jeddeloh et al., 1998, 1999), maize mutants defective for CHR106 exhibit reduced CHH and CHG (H = A, C, or T) methylation (Li et al., 2014). The maize proteins CHR127 and CHR156 are highly similar to AtDRD1, but mutant alleles have not been uncovered in published forward genetic screens based on disrupted epigenetic silencing in maize (Dorweiler et al., 2000; Hollick & Chandler, 2001; Hollick, Kermicle, & Parkinson, 2005; Madzima, Mills, Gardiner, & McGinnis, 2011; Sidorenko et al., 2009; Stonaker, Lim, Erhard, & Hollick, 2009). Similar to its Arabidopsis homolog (Law et al., 2010), CHR127 interacts with RNA polymerase V (PolV) (Haag et al., 2014). These phenotypic similarities suggest that these maize proteins may have similar functions to their Arabidopsis homologs.

While chromatin remodelers have been extensively studied in several organisms, the activities of such proteins have not been investigated in maize yet. Therefore, this study investigated nucleosome positioning and its relationship to transcriptional regulation by focusing on the transcription start site (TSS) regions of a random selection of 400 classical genes on a microarray. Although sequencing has become a favored approach with the increasing availability of high‐throughput sequencing, this microarray provided a cost‐effective way to identify changes at several loci in uncharacterized genotypes. Herein, several mutant maize alleles of putative chromatin remodelers were identified and tested with nucleosome protection assay to identify nucleosome position changes.

2. MATERIALS AND METHODS

2.1. Phylogeny

The Saccharomyces Genome Database (SGD, http://www.yeastgenome.org/) was used to acquire the amino acid sequences and identifiers (ID) for ScSNF2 (SGD ID S000005816) and ScRAD54 (SGD ID S000003131). The Arabidopsis amino acid sequences were acquired from The Arabidopsis Information Resource (TAIR, https://www.arabidopsis.org/), along with the Arabidopsis Genome Initiative (AGI) identifiers for AtDDM1 (AGI ID At5g66750), AtCLSY1 (AGI ID At3g42670), and AtDRD1 (AGI ID At2g16390). These proteins were queried against Zea mays using NCBI's BLASTP (Altschul, Gish, Miller, Myers, & Lipman, 1990) to identify the maize proteins within the top five percent identity for each query. The Pfam SNF2_N and Helicase_C domains of the yeast, Arabidopsis, and selected maize proteins were identified by BLASTP and MUSCLE Aligned using the software Geneious (http://www.geneious.com/). The percent identities at each amino acid were downloaded and averaged to determine domain conservation.

Phylogenetic tree was generated, and the number of amino acid substitutions was determined in the Helicase_C domains using MEGA7 (Kumar, Stecher, & Tamura, 2016). The evolutionary history was inferred by using the maximum‐likelihood method based on the JTT matrix‐based model (Jones, Taylor, & Thornton, 1992). The Archaea Haloarcula californiae ATP‐dependent helicase (NCBI accession WP_007188160) was used as the outgroup.

2.2. Plant materials

Seeds for the wild types B73 and W22 were acquired from the Maize Genetics Cooperation Stock Center (http://maizecoop.cropsci.uiuc.edu/) and have been maintained through self and sibling pollination. Seeds for chr101‐m1 (B73 background), chr101‐m3 (B73 background), chr106‐m1 (B73 background), and chr106‐T11 (W22 background) were generously donated by the Springer Lab (University of Minnesota) and were originally described by Li et al. (2014). The 5′UTR transposon insertion alleles chr127‐m1 (mu1048254) and chr156‐m1 (mu1055648) (McCarty & Meeley, 2009; Settles et al., 2007) were identified through MaizeGDB (http://www.maizegdb.org/) and were acquired from the Maize Genetics Cooperation Stock Center. The chr127‐m1 and chr156‐m1 genotypes are in the W22 background and have been maintained through self or sibling pollinations. For rmr1‐1 (mixed background), seed stocks from segregating families of homozygous mutants and heterozygotes were used (Hollick, Patterson, Asmundsson, & Chandler, 2000). For mop2‐1 (W23/K55 background), homozygous mutants and wild types from segregating families were used (Sidorenko et al., 2009). For each plant, seedling leaf tips were collected for genotyping purposes. The eighth leaf of each plant was also collected when the plants were at their V9 stage of development to maintain consistency due to the effects of plant development and differentiation. These leaves were immediately flash frozen in liquid nitrogen and stored at −80°C until nuclei or RNA extraction.

2.3. DNA extractions and genotyping

Genomic DNA was isolated from approximately 2‐cm² leaf tips of seedlings using 96‐well plate CTAB protocol (Labonne, Dorweiler, & McGinnis, 2013). The genotyping primers and detection methods are described for each genotype in Table S2. Sanger sequencing was performed at the Molecular Core Facility (Department of Biological Sciences, Florida State University).

2.4. RNA extraction and qRT‐PCR

Plant tissue was ground in liquid nitrogen, and then, RNA was extracted using TRI Reagent (Molecular Research Center, Inc.) according to the manufacturer's instructions. The resulting RNA was DNaseI (Promega) treated at 37°C for 30 min and then cleaned using RNA clean and concentrator‐25 (Zymo Research). Superscript III (Invitrogen) was used to reverse transcribe 1 ug of RNA into cDNA according to the manufacturer's instructions. cDNA was used as a template for qRT‐PCR using gene‐specific primers and Ubiquitin‐conjugating enzyme (Sekhon et al., 2011) or 45S rRNA (Hale, Erhard, Lisch, & Hollick, 2009) primers for internal control (Table S3) with SYBR Green Master Mix (Applied Biosystems) per the manufacturer's instructions on a 7500 Fast Real‐Time PCR System (Applied Biosystems). Relative expression changes were calculated using 2^−∆∆Ct (Livak & Schmittgen, 2001). All data represent two technical replicates.

2.5. Nuclei isolation

Nuclei were isolated from 10 g of leaf tissue as described previously (Vera et al., 2014) with several modifications. Frozen, ground tissue was fixed in ice‐cold modified Apel buffer (20 mm Tris‐HCl pH 7.8, 250 mm sucrose, 5 mm MgCl₂, 5 mm KCl) and 1% formaldehyde by gently stirring for 10 min on ice. Fixation was stopped by stirring with 125 mm glycine for 5 min. The cells were lysed by continuing to stir on ice for 5 min with 0.25% Triton X‐100 (EMD Millipore) and 0.1% BME. This mixture was filtered through two layers of Miracloth (Calbiochem) and suspended over 50% Percoll density gradient media (GE Healthcare) in Apel buffer. The nuclei were floated by centrifugation at 2000 g for 15 min at 4°C in a swinging‐bucket centrifuge. The band of nuclei floating on top of the Percoll were transferred to new ice‐cold tubes and resuspended in a minimum of twice the nuclei‐band volume of MNase digestion buffer (50 mm Tris‐HCl pH 7.5, 320 mm sucrose, 4 mm MgCl₂, and 1 mm CaCl₂). Nuclei were pelleted by centrifugation at 500 g for 15 min, resuspended in 1 ml of MNase digestion buffer, aliquoted and flash frozen in liquid nitrogen, and stored at −80°C until use.

2.6. Nuclei quantification, MNase digestion, and nucleosomal DNA isolation

Quantification of nuclei, MNase digestion, and mono‐/dinucleosomal DNA purification were performed as described previously (Labonne et al., 2013).

2.7. Analysis of nucleosomal DNA by microarray

Digested nucleosomal DNA and reference undigested DNA were labeled with Cy3 and Cy5 (NimbleGen), respectively, according to the manufacturer's instructions. The labeled DNA were then hybridized to a ~400 classical gene transcription start site microarray (Labonne et al., 2013) and probed according to the manufacturer's instructions (NimbleGen) in the Molecular Core Facility (Department of Biological Sciences, Florida State University). DEVA software (NimbleGen) generated normalized ratio GFF files that were analyzed using R (R Development Core Team) and the DrawGff package (http://fincher.github.io/DrawGff/).

Initially, to eliminate highly polymorphic regions, the raw GFF files for the genomic DNA of B73 and W22 were compared. 364 of the 3000‐bp probe regions were identified with no significant difference (Student's t test, p > .05) for at least 50% of their probes. Only these 364 regions were included in subsequent analysis. To determine the statistical significance of the identified changes in nucleosome position, Student's t test statistical analysis was performed at each probe using the raw GFF files for the nucleosomal and reference data. Regions having a significant difference (Student's t test, p ≤ .05) between wild‐type and mutant nucleosomal DNA while having no significant difference (Student's t test, p > .05) between the wild‐type and mutant reference sequences through a minimum of 140‐bp continuous probe coverage were identified.

2.8. Analysis of nucleosomal DNA by qPCR

Five of the loci (Stc1_NP1, Tdy1_NP1, Bx1_NP1, Tub2_NP1, and Hm1_NP1) with robust nucleosome position differences between the genotypes identified by microarray analysis were also tested by alternate MNase‐qPCR method (Fig. S4a,b). Two ng of MNase‐digested nucleosomal and reference undigested genomic DNA, prepared as described above, was analyzed using qPCR. Primers spanning the region of significant nucleosome position change were designed for Fdx3_NP1, Stc1_NP1, Tdy1_NP1, Hm1_NP1, Bx1_NP1, and Tub2_NP1 (Table S4). A locus (Mwp1‐GRMZM2G082264) with a consistent nucleosome position region between genotypes was identified from the microarray analysis, and it was also tested by MNase‐qPCR (Fig. S5) as an internal control. The relative nucleosome position changes for the other loci were calculated using 2^−∆Ct (=2^{−[Ct(MNase)−Ct(gDNA)]}) and normalizing each 2^−∆Ct to that of Mwp1 (Han et al., 2012). All data represent two technical replicates.

2.9. Motif search

The fasta‐get‐markov program was acquired through the MEME Suite (http://meme-suite.org/) and was used to generate the appropriate Markov model from the transcription start site regions of the ~400 classical genes of the microarray used in this study. This Markov model was used as the background when searching for shared sequence motifs at the altered nucleosome position loci. Regions of the DNA sequences with significant nucleosome position change, as determined by the microarray analysis, were searched for the presence of common motifs using MEME (Bailey & Elkan, 1994) of the MEME Suite. Motif‐1 was further analyzed by submitting it to the TOMTOM (Gupta, Stamatoyannopolous, Bailey, & Noble, 2007) application for comparison to known maize and Arabidopsis transcription factor binding motifs (O'Malley et al., 2016; Yu et al., 2015) using a significance of E‐value ≤0.05. Motif‐1 was also scanned at the b1 tandem repeats, accession number AY078063 (Stam, Belele, Ramakrishna et al., 2002) using FIMO (Grant, Bailey, & Noble, 2011) of the MEME Suite.

3. RESULTS

3.1. Identification of putative maize chromatin remodelers

The SNF2 family proteins share conserved SNF2_N and Helicase_C domains (Eisen et al., 1995) and have demonstrated abilities to alter nucleosome position (NP) in yeast (Alexeev et al., 2003; Laurent et al., 1993; Petukhova et al., 1999) and in Arabidopsis (Brzeski & Jerzmanowski, 2003). An alignment analysis (Altschul et al., 1990) was used to identify putative maize chromatin remodelers by querying the database with SNF2 family proteins from S. cerevisiae and A. thaliana. Maize proteins with E‐values below 0.01 were identified and those with the highest percent identities were selected for further analysis (Table S1). These proteins included the unnamed GRMZM2G083138, GRMZM2G102625, and GRMZM2G108166 proteins and six previously characterized proteins: CHR101, CHR106, CHR127, CHR156, RMR1, and CHR167. Genotypes mutant for CHR101 and its homolog CHR106 have published DNA methylation phenotypes (Li et al., 2014). RMR1, CHR167 and CHR127 interact with RNA polymerases IV or V (Haag et al., 2014) and have another homolog CHR156 (Hale et al., 2007). On average, the identified maize proteins were 32.0% identical to the known yeast and Arabidopsis proteins. As expected for two plant species, Arabidopsis and maize were typically more identical, at an average of 33.4%, than maize and yeast with an average identity of 29.9%.

Based upon similarity to the Pfam domains (Finn et al., 2016), SNF2_N and Helicase_C domains were identified for the maize, Arabidopsis, and yeast chromatin proteins (Figure 1). The conservation of these domains among the maize, Arabidopsis, and yeast proteins was investigated through their MUSCLE alignments (Edgar, 2004). The pairwise identities indicated that the Helicase_C domains are more conserved, with an average 37% identity, than the SNF2_N domains, with an average 24% identity (Fig. S1a,b).

Figure 1.

Putative chromatin remodelers and their domains. Scaled diagram of the Zea mays, S. cerevisiae (Sc), and A. thaliana (At) proteins. The white arrows indicate the protein‐coding regions. Shaded blocks indicate Pfam‐identified SNF2_N (dark gray), Helicase_C (black), and additional non‐conserved domains (light gray). Transposon insertions (▽) and point mutations (*) are indicated with the allele names. A scale of 200 amino acids (aa) is indicated

Because of strong conservation, the Helicase_C domain was a good candidate for phylogenetic analysis using MEGA7's (Tamura, Stecher, Peterson, Filipski, & Kumar, 2013) maximum‐likelihood (JTT model) method. The evolutionary divergence of the maize and Arabidopsis protein was determined based on the number of amino acid substitutions per site (aas). The maize, Arabidopsis, and yeast proteins were divided into two subgroups. Subgroup 1 included AtDDM1 and the maize homologs CHR101 and CHR106, which diverged 0.10 aas from each other and an average of 0.21 aas from AtDDM1 (Table 1). This group also included GRMZM2G102625, GRMZM2G083138, ScSNF2, and ScRAD54 (Fig. S2). Subgroup 2 included RMR1, CHR167, CHR127, CHR156, GRMZM2G108166, AtDRD1, and AtCLSY1 (Fig. S2). Phylogenetic analysis determined a 0.10 aas divergence between CHR127 and CHR156 and an average divergence of 0.39 aas of these proteins from AtDRD1 (Table 1). CHR101, CHR106, CHR127, CHR156, RMR1, CHR167, GRMZM2G102625, GRMZM2G083138, and GRMZM2G108166 were identified as potential chromatin remodeling proteins and were selected for further analysis.

Table 1.

Amino acid substitutions per site in the Helicase_C domain

Helicase_C	AtDDM1	RMR1	CHR101	CHR106	CHR127	CHR156	ScRAD54	ScSNF2	GRMZM 2G083138	GRMZM 2G102625	AtDRD1	AtCLSY1	GRMZM 2G108166	CHR167	WP_007188160
AtDDM1	0.00
RMR1	1.34	0.00
CHR101	0.19	1.26	0.00
CHR106	0.22	1.25	0.10	0.00
CHR127	1.87	0.87	1.73	1.71	0.00
CHR156	1.82	0.84	1.65	1.64	0.10	0.00
ScRAD54	1.13	1.42	1.07	1.09	1.65	1.57	0.00
ScSNF2	0.73	1.54	0.73	0.73	2.06	1.89	1.25	0.00
GRMZM2G083138	1.24	1.57	1.29	1.27	1.67	1.64	0.33	1.27	0.00
GRMZM2G102625	0.55	1.55	0.53	0.50	1.79	1.73	1.23	0.40	1.14	0.00
AtDRD1	1.83	0.91	1.70	1.75	0.37	0.40	1.66	1.91	1.75	1.78	0.00
AtCLSY1	1.41	0.97	1.38	1.44	0.99	0.95	1.34	1.67	1.57	1.61	1.01	0.00
GRMZM2G108166	1.60	0.99	1.55	1.61	0.91	0.89	1.41	1.64	1.61	1.75	0.89	0.47	0.00
CHR167	1.30	0.47	1.36	1.39	0.79	0.85	1.34	1.53	1.46	1.56	0.91	1.12	0.96	0.00
WP_007188160	2.00	1.90	1.92	1.97	2.02	1.99	2.09	1.98	1.87	1.92	2.00	1.93	1.96	1.82	0.00

The number of amino acid substitutions per site between sequences of the Helicase_C domains of each maize, A. thaliana (At), and S. cerevisiae (Sc) proteins are shown. Duplicate values in table were left blank.

3.2. Identification of alleles with altered gene expression of maize putative chromatin proteins

To further investigate the function of the putative maize chromatin proteins, alleles of Chr101, Chr106, Chr127, Chr156, and Rmr1 were identified in reverse and forward genetics resources (Hollick & Chandler, 2001; McCarty & Meeley, 2009; Settles et al., 2007; Till et al., 2004). At the time of the search, alleles could not be identified for Chr167, GRMZM2G102625, GRMZM2G083138, and GRMZM2G108166. However, mutant alleles for both subgroups of the phylogenetic tree were identified. Mutant alleles of chr101 and chr106 include transposon insertions in the 5′untranslated region (UTR, chr101‐m1) and exons (chr101‐m3, chr106‐m1), as well as a nonsense point mutation (chr106‐T11, Figure 1). Plants homozygous for these alleles were already demonstrated to have altered DNA methylation, and chr101 chr106 double mutants exhibited lethality (Li et al., 2014). The gene expression level of Chr101 and Chr106 had not been reported for these genotypes. Based upon real‐time quantitative reverse transcription‐PCR (qRT‐PCR) analysis, gene expression was significantly reduced in all four homozygous mutants compared to wild type (Figure 2a). Gene expression of Chr101 was reduced to approximately half of the wild type in both the chr101‐m1 and chr101‐m3 homozygous mutants. The homozygous mutants chr106‐m1 and chr106‐T11 appeared to be null alleles with the gene expression of Chr106 being reduced to nearly zero (Figure 2a).

Figure 2.

Gene expression of putative chromatin protein‐encoding genes in different genotypes. (a) Relative expression levels of the different chromatin proteins (above bars) were determined by qRT‐PCR and calculated as 2^−∆∆Ct. Gene expression was determined in wild type (black bar) and homozygous mutant (white bar) individuals, and expression level of the non‐mutant gene in the appropriate wild genotype was set to 1 for each comparison. The data represent averages ±SD (error bars) of three replicates that were normalized to Ubiquitin‐conjugating enzyme. Student's t test (p ≤ .05) was used to identify statistically significant changes between mutant and WT (*). (b) Relative expression levels of Chr106 in B73 (black bar) and W22 (white bar) plants were determined by qRT‐PCR and calculated as 2^−∆∆Ct. The data represent averages ±SD (error bars) of three pools of three biological replicates that were normalized to Ubiquitin‐conjugating enzyme. Student's t test (p ≤ .05) was used to identify statistically significant change between the haplotypes (*)

For Chr127 and Chr156, 5′UTR transposon insertion alleles (chr127‐m1, chr156‐m1) were identified (Figure 1). These alleles did not have any observable vegetative or reproductive phenotypes (data not shown), possibly due to chr127‐m1 and chr156‐m1 being mutations with weak expressivity or because of genetic redundancy. Based on qRT‐PCR analysis, the chr156‐m1 homozygous mutant exhibited an approximately one third reduction from wild‐type gene expression (Figure 2a), and the chr127‐m1 homozygous mutants exhibited approximately one third increased gene expression compared to wild type. Elevated gene expression due to transposon insertion has been demonstrated to lead to phenotypic abnormalities (Barkan & Martienssen, 1991; Robbins, Sekhon, Meeley, & Chopra, 2008), indicating that this allele might be useful for studying CHR127 function.

For RMR1, the well‐characterized rmr1‐1 mutant allele was chosen. It includes a point mutation in its Helicase_C domain (Figure 1) and was previously characterized as a loss‐of‐function allele with detectable phenotypes (Hale et al., 2007). While this allele did not show any significant changes in gene expression (data not shown), the molecular lesion presumably alters protein structure and function.

3.3. The inbred lines B73 and W22 exhibit differential nucleosome positions

The mutant alleles of chr106, chr106‐m1 and chr106‐T11, were isolated from mutant stocks of different genetic backgrounds, B73 and W22 (Li et al., 2014). Genetic diversity has been demonstrated between the maize inbred Stiff Stalk haplotypes, including B73, and Non‐Stiff Stalk haplotypes, including W22 (Liu et al., 2003). Nucleosome positions of B73 and W22 were compared to each other to identify haplotype‐specific differences. The nucleosome position comparison was performed using microarray analysis of nucleosome‐sized DNA relative to undigested DNA in a 3000‐bp window of the transcription start site (TSS) region of 400 classical genes (Labonne et al., 2013).

As this analysis is highly sensitive to sequence polymorphisms, and the different haplotypes are known to be polymorphic (van Heerwaarden, Hufford, & Ross‐Ibarra, 2012), regions with substantial genetic variation between the relevant genotypes were excluded from this and all subsequent analysis, leaving 364 loci that are amenable to study with this technique. To detect nucleosome position changes, the log2 ratio of the signal intensities of the different genotypes was plotted using an R script (Druliner et al., 2013) and was observed for changes between haplotypes. The hybridization signal pattern for each genotype was compared against the hybridization signal of the same genotype genomic DNA to identify “peaks” corresponding to strongly positioned nucleosomes and “valleys” indicating a lack of positioned nucleosomes in that genomic position. In comparing nucleosome positions between the two haplotypes, nucleosome positions were consistent at most genomic locations, with a microarray‐wide Pearson correlation of 0.85. However, there were some discrete changes at specific loci, suggesting localized changes in chromatin structure.

Between B73 and W22, eight loci with statistically significant nucleosome position changes longer than 140 bp were detected (Figure 3; Table 2). A positioned nucleosome was gained in W22 at two loci (Gpc1_NP1 and Cyc7_NP1) and in B73 at the other six loci. As differential expression of some genes has been demonstrated between the two haplotypes (Dong et al., 2016), gene expression of Chr106 was analyzed in B73 and W22. The qRT‐PCR comparison of the two haplotypes demonstrated the significantly reduced gene expression of Chr106 in W22 compared to B73 (Figure 2b), suggesting that there is differential chromatin protein activity between haplotypes of maize.

Figure 3.

Nucleosome position differences between the B73 and W22 haplotypes. Comparison of the nucleosome position in the haplotypes B73 (gray line) and W22 (black line). Genes are all oriented according to a left‐to‐right transcriptional direction, with the nucleotide position in kb relative to the transcription start site (0) indicated on the x‐axes. Relative fluorescent signals are indicated on the y‐axes (Signal (log2 ratio)). Three pools of three biological replicates were averaged for each plot

Table 2.

Maize loci with detectable nucleosome position differences between B73 and W22 haplotypes

Gene name	Gene ID	NP change name	Haplotype with NP gain	Motif‐1
Glb2	GRMZM2G026703	Glb2_NP1	B73
Chi1	GRMZM2G155329	Chi1_NP1	B73
Zfl2	GRMZM2G180190	Zfl2_NP1	B73
Ba1	GRMZM2G397518	Ba1_NP1	B73
Bap2	GRMZM2G152655	Bap2_NP1	B73
Gpc1	GRMZM2G046804	Gpc1_NP1	W22	1
Tpi3	GRMZM2G018177	Tpi3_NP1	B73
Cyc7	GRMZM2G310115	Cyc7_NP1	W22

The names and IDs of maize genes with demonstrated nucleosome position (NP) change between the B73 and W22 haplotypes are listed. The haplotype with the gained nucleosome position (haplotype with NP gain) is indicated as well as the number of Motif‐1 occurrences.

3.4. Nucleosome position is altered in plants homozygous for mutant alleles of putative chromatin proteins

One possible phenotype associated with disrupted chromatin remodeler activity is a change in nucleosome position in transcriptional regulatory regions of the genome (reviewed by Jiang & Pugh, 2009). To identify potential chromatin structure phenotypes, wild‐type plants and homozygous mutants for rmr1, chr101, chr106, chr127, and chr156 were compared using microarray analysis as described for the B73 and W22 haplotypes. Nucleosome positions were consistent between genotypes at most genomic locations, with a microarray‐wide Pearson correlation average of 0.93 between mutant and wild‐type genotypes. However, there were some discrete changes at specific loci, suggesting that loss of the respective protein activity may induce localized, rather than genomewide changes in chromatin structure. Nucleosome position patterns were compared for each mutant genotype against its corresponding wild‐type nucleosome position map and statistically significant nucleosome position changes longer than 140 bp were detected. Among all the genotypes tested, a total of 23 loci with altered nucleosome positions were identified (Table 3). Because a genotype‐ or lineage‐specific transposon or other sequence polymorphism could lead to some of the detected differences, a control of nucleosomal DNA by genomic DNA was used to minimize false positives. Additionally, the use of multiple alleles with distinct genetic pedigrees is also expected to reduce the likelihood of this, suggesting that many or most of these differences in nucleosome position are true, genotype‐specific changes. These loci did not seem to belong to any one consistent gene ontology (GO) group (data not shown), suggesting that the effects were not limited to any specific categories of biological function.

Table 3.

Maize loci with detectable nucleosome position differences in different genotypes

Gene name	Gene ID	NP change name	Genotypes with altered NP	Genetic background	NP in mutants	% mCHHa	CHHi locationa	More correlated genotype to NOL	Motif‐1
Fdx3	GRMZM2G053458	Fdx3_NP1	chr101‐m1 chr101‐m3 chr106‐T11	B73 B73 W22	Gain Gain Gain	NAb	NAb	M M M	2 2 1
Adf1	GRMZM2G117603	Adf1_NP1	chr101‐m1 chr101‐m3	B73 B73	Loss Loss	66.7	5′ end	M M
Pdc2	GRMZM2G038821	Pdc2_NP1	chr101‐m1 chr101‐m3	B73 B73	Loss Loss	32.3	3′ end	M M
Pdc2	GRMZM2G038821	Pdc2_NP2	chr101‐m1 chr101‐m3	B73 B73	Gain Gain	32.3	3′ end	M M
Stc1	GRMZM2G177098	Stc1_NP1	chr106‐T11 chr127‐m1	W22 W22	Loss Loss	25.0	3′ end	M M
Tpi3	GRMZM2G018177	Tpi3_NP1	chr106‐T11	W22	Gain	NAb	NAb	W
Tdy1	GRMZM2G321778	Tdy1_NP1	chr127‐m1	W22	Gain	33.3 77.7	3′ end 5′ end	M	2
Bx1	GRMZM2G085381	Bx1_NP1	chr156‐m1	W22	Loss	NAb	NAb	M
Bx1	GRMZM2G085381	Bx1_NP2	chr156‐m1	W22	Loss	NAb	NAb	M	1
Tub2	GRMZM2G334899	Tub2_NP1	chr156‐m1	W22	Loss	NAb	NAb	M
Hm1	GRMZM5G881887	Hm1_NP1	chr156‐m1	W22	Loss	77.8	3′ end	W	1
Pac1	GRMZM2G058292	Pac1_NP1	chr156‐m1	W22	Loss	NAb	NAb	M	3
Adh2	GRMZM2G098346	Adh2_NP1	chr156‐m1	W22	Gain	52.9	3′ end	W	1
Ms26	GRMZM2G091822	Ms26_NP1	chr156‐m1	W22	Loss	NAb	NAb	M
Ms26	GRMZM2G091822	Ms26_NP2	chr156‐m1	W22	Loss	NAb	NAb	M
Ts1	GRMZM2G104843	Ts1_NP1	chr156‐m1	W22	Loss	100 72.2	3′ end 5′ end	M
PhyC2	GRMZM2G129889	PhyC2_NP1	chr156‐m1	W22	Loss	89.5	5′ end	M
Bt1	GRMZM2G144081	Bt1_NP1	chr156‐m1	W22	Loss	43.2 45.7	3′ end 5′ end	M
Sam2	GRMZM2G154397	Sam2_NP1	chr156‐m1	W22	Loss	79.5	3′ end	M
Mads1	GRMZM2G171365	Mads1_NP1	chr156‐m1	W22	Loss	43.2	3′ end	W	1
Lhcb1	GRMZM2G351977	Lhcb1_NP1	chr156‐m1	W22	Loss	NAb	NAb	M	1
Lhcb1	GRMZM2G351977	Lhcb1_NP2	chr156‐m1	W22	Loss	NAb	NAb	M
Pl1	GRMZM2G701063	Pl1_NP1	chr156‐m1	W22	Loss	NAb	NAb	M

The names and IDs of maize genes with demonstrated nucleosome position (NP) change in one or more genotypes are listed. The genotypes within which NP differences were detected, along with the type of NP change (gain–gain of a positioned nucleosome or loss–loss of a positioned nucleosome) in each genotype, are indicated. For the NP change names, at loci with multiple NP change regions, the more 5′ region is designated as NP1 and the more 3′ region is designated as NP2. The percentage of methylation of CHH islands (% mCHHi) and the location of the CHHi relative to the genes is indicated. Similarity of NP in mutant (M) and wild type (W) to Nucleosome Occupancy Likelihood (NOL) was determined using comparison of the Pearson correlation coefficients. The genotypes more correlated with NOL are indicated as well as the number of Motif‐1 occurrences.

^aData from (Li et al., 2015).

^bNo data in (Li et al., 2015).

The phylogenetic subgroup 1 proteins analyzed in this study were CHR101 and CHR106. Microarray analysis of nucleosome position in chr101‐m1 compared to the wild‐type B73 genotype identified nucleosome position changes at the 3000‐bp region around the TSS of three different genes, including two nucleosome position changes near Pdc2 (Figure 4; Table 3). Changes in nucleosome position were designated according to the locus name and nucleosome position change number; for example, Adf1_NP1 refers to the first detected change at the Adf1 locus. At the Adf1_NP1 and Pdc2_NP2 regions, a positioned nucleosome was lost in homozygous mutant plants, suggesting that a well‐positioned nucleosome had been removed. On the other hand, the gain of a positioned nucleosome was demonstrated in homozygous mutant plants at Fdx3_NP1 and Pdc2_NP1 region, suggesting that a well‐positioned nucleosome has been added. It is possible that the two changes at Pdc2 actually reflect a shifted nucleosome from one region to another rather than an unrelated loss at one region and gain at another, consistent with the Arabidopsis DDM1 moving nucleosomes rather than removing them (Brzeski & Jerzmanowski, 2003). For each of these loci, qualitatively similar changes were observed in plants homozygous for either chr101‐m1 or chr101‐m3 (Figure 4), which supports the notion that this nucleosome position phenotype is associated with loss of CHR101 activity. Where there is a quantitative difference, chr101‐m3 is always more similar to wild type than chr101‐m1 although the chr101‐m3 mutation is more likely to disrupt the protein coding capacity of Chr101.

Figure 4.

Nucleosome position changes in chr101‐m1 and chr101‐m3 mutants. Comparison of the nucleosome position in wild type B73 (solid gray line), homozygous chr101‐m1 (solid blue line), and homozygous chr101‐m3 (dashed blue line) plants. Genes are all oriented according to a left‐to‐right transcriptional direction, with the nucleotide position in kb relative to the transcription start site (0) indicated on the x‐axes. Relative fluorescent signals are indicated on the y‐axes (Signal (log2 ratio)). Three pools of three biological replicates were averaged for each plot, except for chr101 m3, which includes one pool of three biological replicates

The microarray analysis of chr106‐T11 compared to the wild‐type W22 genotype identified three loci with altered nucleosome positioning. At Stc1_NP1, the loss of a positioned nucleosome was observed in homozygous mutant plants, suggesting nucleosome removal or relocation at these loci. At Fdx3_NP1 and Tpi3_NP1, the gain of a positioned nucleosome was demonstrated in chr106‐T11 homozygous plants compared to wild type (Figure 5a; Table 3) indicating the acquisition of a well‐positioned nucleosome. Due to the small number of loci with differences in chr106‐T11 homozygous plants, chr106‐m1 homozygous plants were tested by focused qPCR of micrococcal nuclease (MNase)‐digested samples at Fdx3_NP1 and Stc1_NP1 rather than microarray. The Tpi3_NP1 region was not amenable to qPCR analysis. Although statistical significance was not demonstrated, the nucleosome position changes observed in chr106‐T11 at Fdx3_NP1 and Stc1_NP1 were also observed in chr106‐m1 (Figure 5b), suggesting that the nucleosome position changes at Fdx3_NP1 and Stc1_NP1 are associated with the loss of CHR106 activity.

Figure 5.

Nucleosome position (NP) changes in chr106‐T11 and chr106‐m1 mutants. (a) Comparison of the nucleosome position in wild‐type W22 plants (black line) and plants homozygous for chr106‐T11 (green line). Genes are all oriented according to a left‐to‐right transcriptional direction, with the nucleotide position in kb relative to the transcription start site (0) indicated on the x‐axes. Relative fluorescent signals are indicated on the y‐axes (Signal (log2 ratio)). Three pools of three biological replicates were averaged for each plot. (b) Relative 2^‐∆∆Ct values (Relative NP) for wild type B73 (black bar) and homozygous chr106‐m1 (white bar) plants at Fdx3_NP1 and Stc1_NP1. The data for B73 represent averages ±SD (error bars) of three pools of three biological replicates and for chr106‐m1 one pool of three biological replicates

The phylogenetic subgroup 2 proteins analyzed in this study were RMR1, CHR127, and CHR156. In chr127‐m1, changes in nucleosome positioning were identified for two loci (Table 3). A positioned nucleosome was lost at Stc1_NP1 in chr127‐m1 and gained at Tdy1_NP1 (Figure 6). Although the microarray was initially designed to study the 5′ end of genes, the small size of the Tdy1 gene (819 bp) made it possible to include the entire gene in the 3000‐bp region of study, and the nucleosome position change observed at this locus in chr127‐m1 was detected in the 3′UTR region, suggesting that the nucleosome position changes in this genotype were not limited to transcription start sites. Nucleosome position changes were detected for 13 genes in plants homozygous for chr156‐m1. Of these 13 genes, three had multiple nucleosome position changes, which contributed to a total of 16 discrete nucleosome position changes detected in this genotype (Figure 7; Table 3). In the homozygous chr156‐m1 plants, positioned nucleosome was gained at Adh2_NP1 compared to wild type, while it was lost at the other 15 regions.

Figure 6.

Nucleosome position changes in chr127‐m1 mutants. Comparison of the nucleosome position in wild type W22 (black line) and homozygous chr127‐m1 (red line) plants. Genes are all oriented according to a left‐to‐right transcriptional direction, with the nucleotide position in kb relative to the transcription start site (0) indicated on the x‐axes. Relative fluorescent signals are indicated on the y‐axes (Signal (log2 ratio)). Three pools of three biological replicates were averaged for each plot

Figure 7.

Nucleosome position changes in chr156‐m1 mutants. Comparison of the nucleosome position in wild type W22 (black line) and homozygous chr156‐m1 (orange line) plants. Genes are all oriented according to a left‐to‐right transcriptional direction, with the nucleotide position in kb relative to the transcription start site (0) indicated on the x‐axes. Relative fluorescent signals are indicated on the y‐axes (Signal (log2 ratio)). Three pools of three biological replicates were averaged for each plot

Analysis of plants homozygous for rmr1‐1 did not reveal any loci with significantly altered nucleosome positions compared to wild type, although two genes, b1 and pl1 (Hale et al., 2007) that have been shown to have higher levels of transcript abundance in rmr1‐1, were included on the microarray.

3.5. Similar nucleosome position changes were detected in multiple genotypes at the same loci

Analysis of nucleosome positioning identified some of the same loci with nucleosome position changes in different genotypes. Tpi3_NP1 changes occurred between the B73 and W22 haplotypes (Fig. S3a), and between chr106‐T11 and its wild type W22. While the gene expression of Chr106 was reduced in W22 compared to B73 (Figure 2b), the lowest level of Chr106 gene expression was observed in chr106‐T11 (Figure 2a). At Tpi3_NP1, the most positioned nucleosome was in B73 and the least positioned nucleosome was in W22 (Fig. S3a), and homozygous chr106‐T11 plants had an intermediate level of nucleosome positioning between that of B73 and W22. Nucleosome position changes were also shared between chr101‐m1, chr101‐m3, and chr106‐T11 at Fdx3_NP1 (Fig. S3b). This type of change could be indicative of overlapping or cooperative function between homologous proteins.

One locus also had phenotypes in plants deficient for both phylogenetic subgroups 1 and 2 proteins. A positioned nucleosome was lost at Stc1_NP1 in chr127‐m1 in a similar manner as observed in chr106‐T11 (Fig. S3c). Common nucleosome position changes in multiple genotypes could indicate that nucleosome position at these loci is particularly susceptible to perturbation or that nucleosome position at these loci requires the coordinated action of multiple proteins or regulatory pathways. CHR127 has been shown to interact with PolV (Haag et al., 2014), while chr106‐T11 has published DNA methylation phenotypes (Li et al., 2014). Transcriptional gene silencing pathways have been indicated to utilize both chromatin proteins and plant‐specific polymerases to alter DNA methylation (reviewed by Wendte & Pikaard, 2016). One possible mechanism that could explain the shared nucleosome position changes at Stc1_NP1 might require CHR127 for chromatin remodeling to allow PolV access to regulated loci and CHR106 for the coordinated alteration of DNA methylation at loci with PolV transcription.

3.6. CHH islands are highly methylated near loci with altered nucleosome positions in maize mutants with altered gene expression of chromatin proteins

In maize, RNA‐directed DNA methylation (RdDM) pathway‐dependent short sequences rich in asymmetric cytosine methylation (CHH) islands have been identified between some genes and transposable elements (TEs) and have been suggested to function as barriers between euchromatic and heterochromatic regions of the genome (Gent et al., 2013; Li et al., 2015). Loci with altered nucleosome positions in homozygous chr101‐m1, chr106‐T11, chr127‐m1, and chr156‐m1 mutants were analyzed for proximity to CHH islands using published data (Li et al., 2015). Although a genomewide analysis showed that only 57% of CHH islands have a 20% or higher methylation level (Li et al., 2014), all CHH islands identified at loci with differential nucleosome position in the mutant genotypes of this study had methylation levels above 20% in B73 wild type (Table 3).

3.7. Nucleosome position correlates more with intrinsic sequence signals in wild‐type plants than in chr101‐m1 and chr127‐m1

One important determinant of nucleosome position lies in the intrinsic properties of the DNA itself (reviewed in Fincher & Dennis, 2011). Nucleosome mapping microarray experiments have been used to computationally generate nucleosomal predictions, termed nucleosome occupancy likelihood (NOL) for the maize B73 genome (Fincher et al., 2013). Because DNA sequence information is the only input variable for this analysis, the predicted positions of nucleosomes generated by the NOL are related to the primary sequence of the analyzed loci. Chromatin remodelers are thought to override the sequence‐dependent positioning of nucleosomes, making comparisons with NOL useful in estimating whether nucleosome position is sequence‐ or protein‐directed.

The Pearson correlation coefficient was used to compare nucleosome position of the wild haplotypes and homozygous mutants to NOL. For all four of the loci with changes in nucleosome position in the chr101‐m1 mutants, the NOL had a higher correlation with the position detected in the mutant genotype (Table 3). This is consistent with sequence‐dependent nucleosome positioning in genotypes with reduced levels of CHR101. Similarly, for two of the three nucleosome position change regions in the chr106‐T11 genotypes, the plants homozygous for the mutant allele had a higher correlation to NOL, suggesting that in the wild‐type plants these nucleosomes are positioned away from sequence signals (Table 3).

For the two loci with changes in nucleosome position in the chr127‐m1 homozygous plants, the NOL also had a higher Pearson correlation with the position detected in the mutants (Table 3), indicating a sequence‐dependent positioning of nucleosomes in the chr127‐m1 mutant. Notably, Chr127 gene expression is higher in homozygous chr127‐m1 plants, which could result in Chr127 overexpression rather than loss of function. The changes in nucleosome position in plants homozygous for chr127‐m1 suggest that it may be a loss‐of‐function allele in spite of the elevated Chr127 transcript levels detected by qRT‐PCR.

The most nucleosome position changes were identified in chr156‐m1, and at these loci, 13 of the 16 regions (81%) had a higher correlation between mutant and NOL (Table 3), suggesting protein‐directed nucleosome position in wild‐type plants. The other three regions indicated a higher correlation between wild type and NOL suggesting distinct types of loci‐dependent effects in this genotype.

3.8. Nucleosome position at some loci is similar in chr127‐m1 and mop2‐1

Because CHR127 is known to interact in a PolV complex with other proteins including mediator of paramutation2 (MOP2) (Haag et al., 2014), it seemed plausible that CHR127 might function in a PolV‐mediated regulatory pathway. Thus, CHR127‐mediated changes in nucleosome position might also be dependent upon MOP2 function. Nucleosome position was assayed in mop2‐1 homozygous plants at Stc1_NP1 and Tdy1_NP1, where nucleosome position phenotypes had been noted in chr127‐m1. The nucleosome position at Stc1_NP1 and at Tdy1_NP1 showed similar changes in mop2‐1 (Figure 8a,b) as observed in chr127‐m1 homozygous plants (Figure 6).

Figure 8.

MNase‐qPCR analysis of nucleosome position (NP) in homozygous mop2‐1 plants. Comparison of relative 2^−∆∆Ct values (Relative NP) for wild type Mop2 (black bar) and homozygous mop2‐1 (white bar) plants at Stc1_NP1 (a), Tdy1_NP1 (b), and b1tr (c). The data represent averages ±SD (error bars) of three pools of three biological replicates. Student's t test (p ≤ .05) was used to identify statistically significant changes between mutant and WT (*)

While MOP2 is the second largest subunit of PolIV and PolV (Sidorenko et al., 2009), mediator of paramutation1 (MOP1) appears to interact only with PolIV (Haag et al., 2014). Previous studies have analyzed the nucleosome position of the homozygous mop1‐1 mutant plants and wild‐type siblings, and the nucleosome position changes observed for mop1‐1 (Labonne et al., 2013) were not observed in chr127‐m1. An additional statistical analysis (Student's t test) confirmed that nucleosome position changes observed in both mop2‐1/mop2‐1 and chr127‐m1/chr127‐m1 plants were not detectable in mop1‐1/mop1‐1 plants (data not shown). These results indicate that nucleosome position at Stc1 and Tdy1 is likely associated with a PolV complex that includes MOP2 and CHR127 rather than a PolIV complex that would also include MOP1.

3.9. A positioned nucleosome is lost in mop2‐1 at the b1 tandem repeats

MOP2 and other genetically interacting proteins are required for appropriate transcriptional regulation of the b1 gene (Alleman et al., 2006; Belele et al., 2013; Hale et al., 2007; Hollick et al., 2005; Sidorenko et al., 2009; Sloan, Sidorenko, & McGinnis, 2014). Two epialleles of the b1 gene, B‐I (high expressing) and B' (low expressing), are associated with distinct chromatin structure at a regulatory region 100‐kb upstream of the b1 gene (Haring et al., 2010; Louwers et al., 2009; Stam, Belele, Dorweiler, & Chandler, 2002), and changes in gene expression of b1 in different tissues have been shown to be associated with changes in nucleosome position in these regulatory regions (Haring et al., 2010). To determine whether MOP2‐dependent changes in the gene expression of b1 are associated with changes in nucleosome position, MNase‐qPCR was performed in mop2‐1 homozygous mutants at the distal b1 tandem repeats (b1tr) with characterized chromatin structural changes (Haring et al., 2010). Nucleosome position was compared in Mop2 B' wild type, and mop2‐1 homozygous mutant plants with upregulated B', in a W23/K55 background. A positioned nucleosome was lost in the mop2‐1 homozygous plants compared to wild‐type plants (Figure 8c), suggesting that reduced MOP2 function is associated with changes in nucleosome position at the b1 regulatory regions. Because PolV regulation of b1 is allele‐specific (Stam, Belele, Dorweiler, et al., 2002) and most genotypes do not have the appropriate b1 allele for the analysis of b1 gene expression and chromatin structure, the effects of the other predicted chromatin proteins on b1tr were not tested, and no nucleosome position difference was detected at the 3000‐bp region around the TSS of the b1 gene on the microarray (data not shown).

3.10. Gene expression of Stc1 is reduced in both the chr127‐m1 mutants and in the loss‐of‐function mop2‐1 mutants

The nucleosome position changes at b1tr in mop2‐1 homozygous plants (Figure 8c) were similar to those observed in the wild‐type plant tissues where B‐I gene expression is highest (Haring et al., 2010), suggesting that the PolV‐MOP2 regulatory pathway coordinately influences gene expression and nucleosome position. To determine whether MOP2‐ and CHR127‐associated nucleosome positions at Stc1_NP1 and Tdy1_NP1 are also associated with specific gene expression levels, qRT‐PCR analysis of transcript abundance was performed. The loss of a positioned nucleosome downstream of the TSS at Stc1_NP1 in chr127‐m1 (Figure 6) and mop2‐1 (Figure 8a) was associated with significantly reduced gene expression of Stc1 in both chr127‐m1 and mop2‐1 compared to their respective wild types (Figure 9a). The gain of a positioned nucleosome in the 3′UTR at Tdy1_NP1 in chr127‐m1 (Figure 6) and mop2‐1 (Figure 8b) was associated with significantly increased gene expression of Tdy1 in chr127‐m1, but there was no significant change in mop2‐1 (Figure 9b), even though the nucleosome position phenotype was observed in both the chr127‐m1 and mop2‐1 mutants (Figures 6 and 8b). Thus, nucleosome position and gene expression changes were not always observed together and not always in a consistent manner with the loss of silencing effect observed at b1 in mop2‐1 homozygous plants.

Figure 9.

Gene expression of Stc1 and Tdy1 in homozygous chr127‐m1 and mop2‐1 plants. Relative expression levels of Stc1 (a) and Tdy1 (b) were determined by qRT‐PCR and calculated as 2−^∆∆Ct. Gene expression was determined in wild‐type (black bar) and homozygous mutant (white bar) individuals. The data represent averages ±SD (error bars) of three pools of three biological replicates that were normalized to 45S rRNA. Student's t test (p ≤ .05) was used to identify statistically significant changes between mutant and WT (*)

3.11. Nucleosome position changes are associated with changes in gene expression in chr156‐m1 mutants, but the changes do not occur in mop2‐1 mutants

CHR127 and CHR156 have 86% sequence identity, but only CHR127 seems to interact with PolV (Haag et al., 2014). Two loci with differential nucleosome positioning in chr156‐m1 (Hm1_NP1 and Bx1_NP1) were also tested for nucleosome position change in plants homozygous for mop2‐1. Neither of the loci had a significant nucleosome position change in mop2‐1 (data not shown), suggesting that nucleosome position phenotypes are shared between mop2‐1 and chr127‐m1, but not mop2‐1 and chr156‐m1. The loss of positioned nucleosomes in homozygous chr156‐m1 plants (Figure 7) was associated with increased gene expression of Bx1 and Hm1 as determined by qRT‐PCR (Figure 10a,b).

Figure 10.

Gene expression of Bx1 and Hm1 in chr156‐m1. Relative expression levels of Bx1 (a) and Hm1 (b) were determined by qRT‐PCR and calculated as 2^−∆∆Ct. Gene expression was determined in homozygous wild type (W22, black bar) and homozygous chr156‐m1 (white bar) plants. The data represent averages ±SD (error bars) of three pools of three biological replicates that were normalized to 45S rRNA. Student's t test (p ≤ .05) was used to identify statistically significant changes between mutant and WT (*)

3.12. The plant‐specific C2C2‐D of transcription factor binding motif is present at several nucleosome position change regions and at the b1 tandem repeats

The nucleosome position changes in multiple genotypes and comparisons with NOL suggested that DNA sequence was a relevant factor in nucleosome position at the tested loci. To determine whether different chromatin proteins target specific DNA sequences, loci with differential nucleosome positions were analyzed for DNA sequence motifs using the MEME application (Bailey & Elkan, 1994) of the MEME Suite (Bailey et al., 2009). This motif search application requires a background Markov model. As the molecular analysis by microarray was TSS focused, the Markov model used in the analysis was generated from the combined TSS sequences represented on the microarray.

Through microarray analysis of nine genotypes, a total of 30 distinct loci with nucleosome position changes in any genotype were detected. All 30 loci were subjected to a group analysis by MEME using an E‐value of 0.05 as a threshold to identify motifs. One motif (Motif‐1) was identified at nine loci (Figure 11a; Tables 2 and 3). As seven of these loci are those with altered nucleosome position in the mutants of phylogenetic subgroup 2 proteins, Motif‐1 seems to have some specificity toward the putative chromatin proteins that belong to phylogenetic subgroup 2.

Figure 11.

Motif‐1 enriched in loci with differential nucleosome position. (a) Diagram of the sequence of Motif‐1. (b) The location of Motif‐1 (red box) along repeats 2 through 7 of the b1 tandem repeats (b1tr) as determined by FIMO. Diagram is to scale with Motif‐1 representing 12 nucleotides

Because MOP2 and CHR127 appear to have some overlapping function and Motif‐1 was present at Tdy1_NP1, Motif‐1 was compared with the regulatory sequence of b1, which is a series of tandem repeats (Stam, Belele, Dorweiler, et al., 2002). Analysis of Motif‐1 against b1tr using FIMO (Grant et al., 2011) of the MEME Suite identified a significant (q‐value ≤0.05) match. This match was present in six of the seven tandem repeats of b1tr (Figure 11b), which are associated with transcriptional regulation of b1 (Stam, Belele, Dorweiler, et al., 2002).

Although SWI/SNF chromatin remodelers interact with and modify chromatin, direct DNA interaction has not been demonstrated, and many chromatin remodelers have been shown to be recruited by transcription factors (reviewed by Kwon & Wagner, 2007). To determine whether Motif‐1 is a transcription factor recognition motif, it was compared to previously identified maize and Arabidopsis transcription factor motifs (O'Malley et al., 2016; Yu et al., 2015) using the MEME Suite's motif comparison tool (TOMTOM, Gupta et al., 2007). The most similar previously identified maize putative transcription factor binding site was ID0180 (E‐value ≤0.05), although very little information is available about the biological function of ID0180. The comparison to Arabidopsis revealed that the most similar previously identified motif is the plant‐specific C2C2‐D of transcription factor recognition motif (E‐value ≤0.05). These results could suggest that the chromatin proteins of this study and a C2C2‐Dof‐like transcription factor may act at Motif‐1 sequences to modify nucleosome position and facilitate transcriptional regulation. It can also be suggested that the motif corresponding to ID0180 could be a C2C2‐Dof transcription factor recognition sequence.

4. DISCUSSION

Using a microarray‐based nucleosome mapping approach, changes in nucleosome positioning were detected in genotypes with mutant alleles for four predicted chromatin‐related proteins (CHR101, CHR106, CHR127, and CHR156). All of the chromatin proteins tested in this study are homologous with known members of SWI/SNF complexes in other species. The majority of the loci with nucleosome position changes in homozygous mutants showed evidence of nucleosomes being moved or held away from sequence‐dependent positions by functional chromatin proteins. These results are consistent with the human hSWI/SNF proteins, which move nucleosomes away from their sequence‐favored positions (Sims, Baughman, & Schnitzler, 2008; Sims, Lane, Ulyanova, & Schnitzler, 2007), and suggest a role for nucleosome positioning for all the maize chromatin proteins analyzed in this study, except potentially RMR1.

Interestingly, although plants homozygous for rmr1‐1 have phenotypes related to the transcriptional regulation of gene expression, suggesting that some functionality is disrupted by this mutation, no changes in nucleosome position were observed in plants homozygous for rmr1‐1. This could be because RMR1 does not function to modify or maintain nucleosome positioning, because RMR1 regulates loci other than the ones investigated, or because this particular allele of RMR1 does not disrupt the specific function of RMR1 in nucleosome positioning. RMR1 possesses the domains common to chromatin remodelers, including a Helicase_C domain, but it is possible that RMR1‐dependent regulatory mechanisms do not involve changes in nucleosome position and these domains instead support functions other than chromatin remodeling, like resolving the double helix during transcription. In support of this, two genes known to be transcriptionally regulated by RMR1 (Hale et al., 2007; Hollick & Chandler, 2001) were on the microarray and did not exhibit differential nucleosome position in rmr1‐1 homozygous plants. It is also possible that the rmr1‐1 mutation may not be sufficient to impede RMR1‐mediated nucleosome positioning, although the phenotypes associated with rmr1‐1 suggest that this is a deleterious allele. It is also possible that analysis of additional tissues, developmental stages, or regions not on the microarray may reveal RMR1‐dependent nucleosome position phenotypes.

Consistent with prior comparisons of gene expression between maize haplotypes (Dong et al., 2016), this analysis revealed differential gene expression of Chr106 between B73 and W22. This study also identified nucleosome position changes between B73 and W22 in regions with minimal sequence polymorphisms. Shared nucleosome position changes between the two haplotypes and in the homozygous chr106‐T11 mutants, and reduced Chr106 expression in W22 suggests that differential gene expression of chromatin proteins may influence haplotype‐specific phenotypes.

CHR101 and CHR106 are highly similar to one another, and this analysis detected both shared and distinct changes in nucleosome position in chr101 and chr106 genotypes. Previous research has demonstrated the lethality of double mutants and some phenotypic differences in genomewide DNA methylation between the chr101 and chr106 genotypes (Li et al., 2014), suggesting that CHR101 and CHR106 may have some distinct roles. Collectively, this suggests that while CHR101 and CHR106 might coordinate activities at some loci, they may also have distinct and important locus‐specific functions.

Sequence similarity with a known Arabidopsis transcription factor binding motif suggests that a C2C2‐Dof type transcription factor may interact with some loci with changes in nucleosome positioning. Motif‐1 was detected in Fdx3_NP1, Tdy1_NP1, Bx1_NP2, Hm1_NP1, Pac1_NP1, Adh2_NP1, Lhcb1_NP1, Mads1_NP1, and Gpc1_NP1, which seem to have nucleosome position regulated by CHR101, CHR106, CHR127, and CHR156. This may indicate that a C2C2‐Dof type transcription factor recruits these chromatin proteins to regulated loci.

It has been proposed that RdDM activity at CHH islands maintains silencing of transposable elements by forming an epigenetic barrier that blocks transcription from adjacent genes from spreading into the transposable elements (Li et al., 2015). This would mean that loss of RdDM might not change gene expression, although maize mutants thought to be defective in RdDM have demonstrated changes in gene expression of b1 (Alleman et al., 2006; Belele et al., 2013; Hale et al., 2007; Hollick et al., 2005; Sidorenko et al., 2009; Sloan et al., 2014) and other genes (Jia et al., 2009). The fact that some nucleosome position changes were associated with changes in gene expression and some were not, combined with the presence of CHH islands near some nucleosome position change regions, suggests that RdDM, nucleosome position, and gene expression may intersect in distinct ways at different loci. Although a change in one characteristic is not consistently correlated with a change in the others, these characteristics frequently overlap at apparently regulated loci.

The chr127‐m1 allele used in this study is associated with increased gene expression of Chr127 compared to wild type, while mop2‐1 homozygous plants are loss‐of‐function mutants (Sidorenko et al., 2009). Transposon insertions in the 5′UTR with elevated gene expression are sometimes associated with loss‐of‐function phenotypes such as hypomethylation (Prelich, 2012; Robbins et al., 2008), but could also lead to phenotypes caused by overexpression of a functional protein. The fact that homozygous chr127‐m1 plants share the same phenotype as plants homozygous for the loss‐of‐function mop2‐1 alleles suggests that the transposon insertion in chr127‐m1 also results in loss of function.

This work has demonstrated altered nucleosome positions in several maize genotypes, including inbred haplotypes and mutant genotypes. Nucleosome position changes were observed at genes encoding proteins with a range of predicted biological functions. Collectively, these results suggest that nucleosome position may contribute to the phenotypic diversity observed in maize haplotypes and may be useful in exploiting and understanding this important crop plant. MNase‐sequencing approaches using these and other mutant genotypes would likely identify additional loci that are impacted by these mechanisms.

AUTHOR CONTRIBUTIONS

K.M.M. and L.K.S. designed the experiments. L.K.S. performed the experiments. K.M.M. and L.K.S. interpreted the data and wrote the article.

Supporting information

 

 

 

Stroud LK, McGinnis KM. Altered nucleosome positions in maize haplotypes and mutants of a subset of SWI/SNF‐like proteins. Plant Direct. 2017;00:1–17. 10.1002/pld3.19