https://orcid.org/0000-0001-9749-7647
Figures
Abstract
Exceptions to Mendelian inheritance often highlight novel chromosomal behaviors. The maize Pl1-Rhoades allele conferring plant pigmentation can display inheritance patterns deviating from Mendelian expectations in a behavior known as paramutation. However, the chromosome features mediating such exceptions remain unknown. Here we show that small RNA production reflecting RNA polymerase IV function within a distal downstream set of five tandem repeats is coincident with meiotically-heritable repression of the Pl1-Rhoades transcription unit. A related pl1 haplotype with three, but not one with two, repeat units also displays the trans-homolog silencing typifying paramutations. 4C interactions, CHD3a-dependent small RNA profiles, nuclease sensitivity, and polyadenylated RNA levels highlight a repeat subregion having regulatory potential. Our comparative and mutant analyses show that transcriptional repression of Pl1-Rhoades correlates with 24-nucleotide RNA production and cytosine methylation at this subregion indicating the action of a specific DNA-dependent RNA polymerase complex. These findings support a working model in which pl1 paramutation depends on trans-chromosomal RNA-directed DNA methylation operating at a discrete cis-linked and copy-number-dependent transcriptional regulatory element.
Author summary
Although switching genes “on” and “off” in a coordinated fashion is essential to organismal development and homeostasis, some genes ‐ or specific forms thereof ‐ display unusual switching behaviors where the “off” state is sexually transmitted to offspring. Further, only “off” states persist when both “on” and “off” states are combined in heterozygous condition. Examples of this unusual behavior, known as paramutation, exist for certain transgenes and/or endogenous genes in both plants and animals but the chromosome features triggering this switching and the responsible cellular machinery are poorly understood. Here, we used long-read DNA sequencing, comparative genomics, DNA -DNA interaction assays, mutant analyses, and cytosine methylation profiling to identify a switchable regulatory sequence associated with paramutation at the maize purple plant1 (pl1) locus. These key sequences are embedded within a 5-copy tandem repeat array of mostly transposable elements ~14 kilobases distal to the pl1 gene. The “off” state is associated with these sequences producing so-called “small interfering RNAs” (siRNAs) that may direct the recruitment of chromatin-modifying enzymes. Our findings support a molecular model for paramutation and highlight the regulatory importance of repeated sequences found in large eukaryotic genomes.
Citation: Deans NC, Talbot J-ERB, Li M, Sáez-González C, Hövel I, Heavens D, et al. (2024) Paramutation at the maize pl1 locus is associated with RdDM activity at distal tandem repeats. PLoS Genet 20(5): e1011296. https://doi.org/10.1371/journal.pgen.1011296
Editor: Ortrun Mittelsten Scheid, Gregor Mendel Institute of Molecular Plant Biology, AUSTRIA
Received: January 9, 2024; Accepted: May 8, 2024; Published: May 30, 2024
Copyright: © 2024 Deans et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Sequence data from this article can be found in the GenBank, SRA, and GEO data libraries under the following accessions, Pl1-Rhoades haplotype sequence (Genbank OQ405024), 4C libraries (SRA SRR23352217-SRR23352222), Pl-Rh, Pl´, and Pl-Rh / Pl´ cob and seedling sRNA libraries (SRA SRR15400896-SRR15400912), B73 sRNAs (GEO GSE52103), rpd1-1 and rmr1-1 sRNAs (SRA SRR15410418-SRR15410422), and dcl3-2 sRNAs (SRA SRR16938685-SRR16938686). All other relevant data are within the manuscript and its Supporting Information files.
Funding: Research activities were funded by awards to J.B.H. from the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service (www.usda.gov; 99-35301-7753, 2001-35301-10641, and 2006-35304-17399), the National Science Foundation (www.nsf.gov; MCB-0419909, -0920623, -1715375) and The Ohio State Foundation (www.osu.edu). I.H. was supported by funding of the Research Priority Area Systems Biology of the University of Amsterdam (www.uva.nl). This study and N.C.D. research activity was supported in part by resources and technical expertise from the Georgia Advanced Computing Resource Center (www.gacrc.uga.edu), a partnership between the University of Georgia’s Office of the Vice President for research and Office of the Vice President for Information Technology. Additional support of research activities were provided by the Comprehensive Cancer Center and the National Institutes of Health (www.nih.gov) under grant number P30 CA016058 The views expressed are solely those of the authors and are not endorsed by the sponsors of this work. The sponsors and funders did not play any role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: All rmr materials and their uses are covered by U.S. patent 8134047 assigned to The Regents of the University of California. The authors claim no other known competing interests.
Introduction
Patterns of transcriptional regulation governing normal development are consistently replicated from one generation to the next. However, certain trans-homolog interactions known as paramutations facilitate heritable changes in regulation–canonically observed as an active allele being heritably suppressed upon exposure to a repressed allele in trans (reviewed in [1,2]). Susceptible alleles are “paramutable” while repressed derivatives have “paramutagenic” potential. Despite the potential evolutionary significance of such behaviors, the chromosome organization and nuclear feature(s) that define these dynamic properties remain largely unknown.
Paramutation-like behaviors have been described for endogenous alleles in several species including maize (Zea mays) [3–7], tomatoes (Solanum lycopersicum) [8], Arabidopsis (Arabidopsis thaliana) [9], mice (Mus musculus) [10], and potentially humans [11]. Specific transgenes displaying paramutation-like behaviors are also documented in petunia [12], tobacco (Nicotiana tobaccum) [13], Arabidopsis [14,15], mice [16,17], flies (Drosophila melanogaster) [18,19], and worms (Caenorhabditis elegans) [20–24]. Many of these pan-eukarya examples implicate roles for small RNA (sRNA)-based chromatin modifications (reviewed in [25]) highlighting an ancient mechanism of genome control having potential for the transgenerational inheritance of novel regulatory programs [26].
In maize, the first paramutation examples were reported for specific red1/colored1 (r1) alleles [3] which encode basic helix-loop-helix (bHLH) transcription factors conferring seed and plant color [27,28] and an allele of the paralogous booster1/colored plant1 (b1) locus [4]. Other alleles exhibiting paramutation are found at the purple plant1 (pl1) [6] and pericarp color1 (p1) [5] loci that encode R2R3 MYB-type transcription factors required for plant and cob pigmentation, respectively [28]. An additional example occurring at the low phytic acid1 (lpa1) locus [7] shows that paramutations in maize are not confined to genes encoding transcriptional regulators of pigment production.
Forward genetic screens for mediators of paramutation (mop) of the B1-Intense (B1-I) allele [29,30] and factors required to maintain repression (rmr) of the Pl1-Rhoades allele [31,32] identify both overlapping and locus-specific functions. Several mop and rmr loci encode orthologs of proteins involved in an Arabidopsis RNA-directed DNA methylation (RdDM) pathway. RdDM utilizes two plant-specific DNA-dependent RNA polymerases (RNAPs IV and V) and accessory proteins to guide de novo cytosine methylation. RNAP IV indirectly produces mostly 24 nucleotide (24nt) RNAs that, in complex with Argonaute proteins, help recruit a de novo cytosine methyltransferase to nascent long-non-coding RNAP V transcripts (reviewed in [33]). rmr6/mop3 / rpd1 encodes the maize RNAP IV largest subunit, RPD1 [34,35], and rmr7/mop2 encodes one of at least two second largest subunits shared between RNAPs IV and V, RP(D/E)2a [30,36,37]. RPD1 and RP(D/E)2a are orthologous to the Arabidopsis NUCLEAR RNA POLYMERASE D1 (NRPD1) and NRP(D/E)2, respectively. Maize has three genes producing second largest subunits that, according to proteomic data, constitute distinct RNAP IV and RNAP V subtypes [37]. mop1 encodes an ortholog of RNA-DEPENDENT RNA POLYMERASE2 (RDR2) [38] that converts nascent RNAP IV transcripts to a double-stranded substrate for a dicer-like3 (DCL3) endonuclease that generates primarily 24nt RNAs. rmr5 encodes the sole maize DCL3 [39].
The other three RMR proteins identified to date, (RMR1, RMR2, and RMR12/CHD3a), also influence 24nt RNA biogenesis, but neither RMR1 nor RMR12/CHD3a are required to mediate b1 paramutation or to maintain repressed B1-I states ‐ denoted B´ [32,40]. RMR1 is a likely SNF2-type ATPase, potentially orthologous to Arabidopsis CLASSY3 and 4 [32], that co-immunoprecipitates with RPD1 in undifferentiated callus cells [37]. RMR2 is a pioneer protein whose potential Arabidopsis orthologs have no known function [41], and RMR12/CHD3a is a chromodomain helicase DNA-binding3 (CHD3) protein orthologous to the Arabidopsis nucleosome-remodeling protein, PICKLE [40]. Both PICKLE and RMR12/CHD3a influence genome-wide distributions of 24nt RNA production [40,42].
Additional proteins implicated in the repression of B´ include a putative maize ortholog of Arabidopsis ARGONAUTE9 (AGO9) (a member of the AGO4 clade), ZmAGO104, [43] and a novel protein isoform encoded by the unstable factor for orange1 (ufo1) locus [44,45]. A dominant Ufo1-1 allele enhances the expression of both repressed B´ and p1 alleles through an unknown mechanism [44].
One working model for paramutation in maize envisions 24nt RNAs produced from a paramutagenic allele serving as a diffusible substance directing cytosine methylation and/or repressive histone modifications to an active paramutable allele leading to heritable repression. Chromatin modifications, including 5-methylcytosine (5mC), or the 24nt RNAs themselves, may serve as a meiotically heritable mark of paramutagenic alleles [2,25,46]. This model is supported by 1) the associations of MOP and RMR proteins with 24nt RNA biogenesis, 2) transgenic shRNA-mediated induction of b1 paramutation [47], 3) increased 5mC levels at paramutagenic r1 [48], p1 [44,49,50], and b1 alleles [51], and 4) transcriptional repression of paramutagenic B1-I [52], P1-rr [53], and Pl1-Rhoades [54] alleles. However, paramutations at pl1 can still occur with reduced frequencies in rmr1-1/rmr1-2 and rmr2-1 mutants [32,41], paramutation at b1 still occurs in a parent-of-origin specific manner in the rmr7-1 mutant [36], sRNA profiles of active (B-I) and repressed (B´) B1-I states appear similar [47], no 5mC differences have so far been found between active (Pl-Rh) and repressed (Pl´) Pl1-Rhoades states [32,54], and–with the possible exception of RP(D/E)2a –no known MOP or RMR proteins represent RdDM-type components acting downstream of 24nt biogenesis. These findings contrast the expectations of a canonical RdDM-like model.
In maize, paramutagenic haplotypes are associated with repetitive sequences. At r1, the repeats include entire r1-coding regions plus flanking intergenic sequences [55–57], while B1-I and P1-rr have intergenic repeats acting as transcriptional enhancers [58,59]. A hepta-repeat of an otherwise unique 853bp sequence found ~100kb 5´ of the B1-I coding sequence (CDS) acts as both an enhancer and an efficient inducer of B1-I paramutation [59,60]. These findings indicate that the b1 hepta-repeat can both promote expression of B-I and confer paramutagenic activity to B´. At both r1 and b1, repeat numbers directly correlate with paramutagenic strength [57,59–61] but the nature of this relationship remains speculative [62].
Although the sequences mediating paramutation at Pl1-Rhoades have remained unknown, a Pl1-Rhoades || pl1-B73 recombinant derivative, pl1-R30, defines a region > 6kb 3´ of the pl1 CDS required for conferring strong pigmentation and for paramutagenicity [63]. pl1-R30 contains the Pl1-Rhoades CDS and a promoter-proximal CACTA-type DNA transposable element (TE) fragment of the doppia subclass [56,63,64] but confers uniformly weak pigmentation similar to that of pl1-B73. 5mC patterns at this doppia fragment are dependent on both RPD1 and RMR1 [32] but appear similar in both Pl-Rh and Pl´ states [54]. This TE fragment likely directs seed-specific expression because 1) a similar doppia fragment appears to serve as a seed-specific promoter in the paramutable R-r:standard haplotype [56]; 2) the non-paramutable Pl1-Blotched (Pl1-Bh) allele that has near-identical sequence to Pl1-Rhoades over at least 3.5kb, including the doppia sequences [65,66], confers blotchy seed pigmentation [65]; and 3) Pl1-Rhoades becomes seed expressed after multi-generational absence of RPD1 [63]. Thus, RNAP IV is responsible for the tissue-restricted expression of Pl1-Rhoades, and other genes [67,68], potentially by repressing regulatory TEs [68,69]. Importantly, the non-paramutable nature of both pl1-R30 and Pl1-Bh indicate the doppia fragment by itself is insufficient for paramutagenic function.
Here we document the Pl1-Rhoades haplotype structure and molecular changes occurring within a distal tandem repeat that are associated with paramutation. A 192kb assembly of long-read sequences of bacterial artificial chromosomes (BACs) containing the Pl1-Rhoades CDS revealed a 2092bp sequence present in five tandem direct copies beginning ~14kb 3´ of the pl1 polyadenylation site (PAS). Structure comparisons indicate that three repeat copies are present in another paramutable pl1 haplotype from the CML52 inbred. Circular chromosome conformation capture (4C) results show these repeats are in close proximity to a bait sequence located between the 5´ doppia fragment and Pl1-Rhoades transcription start site (TSS). One or more of these repeats differ in both nuclease sensitivity and polyadenylated RNA levels between active Pl-Rh and repressed Pl´ states. We also show that an otherwise unique portion of the repeat unit produces RNAP IV-dependent 24nt RNAs only in the Pl´ state, identifying a ~200bp segment as having differential RNAP IV recruitment between Pl-Rh and Pl´. We found that this precise positioning of sRNA biogenesis is defined by CHD3a. In contrast, no significant sRNA differences were observed at the doppia fragment or elsewhere in the 192kb haplotype. The differential sRNA abundances also coincide with differing 5mC patterns. Our findings are consistent with a model in which Pl1-Rhoades paramutation involves a maize-specific RNA-directed DNA methylation-like mechanism acting at a copy-number-dependent distal cis-regulatory element.
Results
The Pl1-Rhoades haplotype contains a penta-repeat
To extend the existing Pl1-Rhoades reference sequence [40], six Pl1-Rhoades-containing BAC clones were sequenced, and resulting long-reads were assembled into a 192kb contig (Fig 1 and see S1 Table; Genbank OQ405024). Comparison to the B73 AGPv4 reference genome [70] highlighted two major structural variations downstream of the Pl1-Rhoades CDS, including the absence of a B73-specific gene model (Zm00001d037119) and the presence of five tandem repeats, each having an identical 2092bp sequence, located between ~14kb 3´ of the PAS and ~100bp 3´ of the next, convergent gene model (Zm00001d037120)(Fig 1D). A CENSOR-based annotation of the repeated sequence [71] identified mostly TE fragments with two small DNA TEs (one highly repetitive DNA-7_ZM and one relatively unique PIF-Harbinger-type) flanking a 390bp unique non-coding region hereafter referred to as the unique subregion (USR)(see S1A Fig; see S1 File; 135,133–135,522 in Genbank sequence OQ405024). Southern blot hybridizations (see S1B–S1D Fig) with a radiolabeled USR probe showed that both pl1-B73 and pl1-R30 have single similar-sized BglII fragments but no ~2kb fragments diagnostic of the tandem repeats. Additionally, the ~4kb pl1-B73 and pl1-R30 NsiI fragments contrast the >12kb Pl1-Rhoades fragment encompassing the five repeats, hereafter referred to as the penta-repeat. Together with previous results [63], these data indicate the pl1-R30 recombination event occurred between ~6 and 12kb 3´ of the Pl1-Rhoades PAS (see S1B–S1C Fig). Importantly, recombinant loss of the penta-repeat in pl1-R30 is coincident with the absence of both high pl1 gene expression and paramutagenicity [63].
(A) BAC clone contigs containing Pl1-Rhoades assembled from long-read sequences. (B) Pl1-Rhoades haplotype representation with arrows identifying annotated B73 AGPv4 gene models. Bracketed line represents the region used for GEvo alignments with pl1-B73 (D), Pl1-CML52 (E), and pl1-NC350 (F). (C) Images of anther phenotypes conferred by the respective active and repressed states of the Pl1-Rhoades and Pl1-CML52 alleles. pl1-CO159 is a recessive non-functional allele used here to illustrate the pigmentation potential of Pl1-CML52. (D-F) Shaded areas represent regions of synteny, black boxes represent gene models, and magenta boxes immediately upstream of the pl1 gene models represent a doppia transposon fragment. Green shading and black arrows represent sequences found as a penta-repeat in Pl1-Rhoades.
Repeat copy numbers correlate with paramutagenicity
So far, the only pl1 alleles shown to have or acquire paramutagenic activities are Pl1-Rhoades and Pl1-CML52 [63,66], the latter being present in one of the 25 founder lines (CML52) of a nested association mapping (NAM) population [72]. Both alleles can condition strong pigmentation of anthers and plant body ([6]; Fig 1C). We queried all founder line long-read assemblies [73] with the USR sequence and found all contain >80% of this sequence with >97% identity, but only the CML52 and NC350 lines contain more than a single intact repeat unit. The CML52 haplotype consists of two pl1 CDSs (Fig 1E) distinguished by a 6bp deletion and three 3bp insertions unique to Pl1b-CML52, and four synonymous and two conservative non-synonymous bp substitutions. The 5´-most gene model (Pl1b-CML52) also has a large annotated 3´ UTR intron, while the other (Pl1a-CML52) is identical to Pl1-Rhoades, aside from a 32bp insertion into the adjacent doppia fragment [63]. Two tandem copies of the repeat are found precisely where the five are located in Pl1-Rhoades and a third solo copy lies just upstream of Pl1a-CML52 (Fig 1E). Both the solo repeat unit and the first of the downstream tandem repeats are identical to those in Pl1-Rhoades while the second downstream repeat unit has a SNP.
The NC350 pl1 haplotype is also similar to Pl1-Rhoades, but only has two repeats (Fig 1F). The CML52, NC350 and Pl1-Rhoades haplotypes are nearly identical over ~20kb encompassing the doppia fragment and first two tandem repeats. Differences include a CML52- and NC350-shared 32bp doppia insertion, 4 CML52-specific SNPs, a single NC350-specific SNP within the doppia fragment, one Pl1-Rhoades-specific SNP, and an intergenic poly-T sequence with lengths of 8, 9, and 10nt in Pl1-Rhoades, NC350, and CML52, respectively. Given the structural similarity of the NC350 haplotype with others having paramutagenic potential, we used an established pedigree analysis with a linked genetic marker [74] to test whether it could acquire paramutagenic activity following exposure to Pl´ (see S1 Methods). With one possible recombinant exception, no paramutagenic activity co-segregated with the NC350 pl1 haplotype (see S1 Methods and S2 Table). Similar tests showed that none of seven other NAM founder line pl1 haplotypes containing single repeat units acquired paramutagenic activity (see S1 Methods and S2 Table). Because the CML52 pl1 haplotype becomes paramutagenic following exposure to Pl´ [63], we infer that haplotypes having more than two of these repeats have paramutation-like behaviors. Additionally, paramutagenicity of the CML52 haplotype is less than that of Pl1-Rhoades (see S2 Table; [63]) consistent with repeat copy numbers correlating with paramutagenic strength.
Repeat sequences associate with the Pl1-Rhoades promoter
The extended sequence of the Pl1-Rhoades haplotype enabled a 4C analysis to identify potential distal enhancers. Using a bait sequence located 20nt 5´ of the Pl1-Rhoades TSS, we generated and sequenced 4C libraries of DpnII ‐ NlaIII fragments derived from nuclei of both inner seedling leaves (S) and inner husk leaves (H) harvested from Pl-Rh / Pl-Rh plants (see S3 Table). In both tissues pl1 mRNA is highly expressed relative to Pl´ / Pl´ genotypes (see S2 Fig; [54]). Sequence tags were first mapped to the Pl1-Rhoades haplotype with the penta-repeat collapsed to a single unit and ranked by abundance for each library, excluding those within 2kb of the bait sequence (see S4 Table). The highest tag density overlapped the repeat sequence, with three of the five most abundant H tags and two of the top five S tags residing in the repeat region. Mapping all 4C tags to the B73 AGPv4 genome indicated that many of the most abundant tags represented highly repeated sequences. We therefore repeated the 4C tag analysis by 1) excluding reads multiply-mapping to the genome, 2) mapping this filtered set to the Pl1-Rhoades haplotype, 3) ranking their abundances (see S5 Table), and 4) displaying these in 2kb bins (Fig 2). The highest tag density again overlapped the repeat sequence with the most abundant tags from both tissues (see S1 and H1) located within the USR (Fig 2E). These data indicate that the penta-repeat is physically close to the Pl1-Rhoades promoter in multiple tissues of Pl-Rh plants in which pl1 mRNA is abundant, including inner husk leaves where the gene is highly transcribed [54].
Uniquely-mapping reads per million (rpm) 4C tags in 2kb bins across the Pl1-Rhoades haplotype: schematic combined datasets (A), inner seedling leaves (S) (B), and inner husk leaves (H) (C). (D) Schematic identifying positions and ranks of the five most abundant 4C tags (blue bars) ligated to the bait sequence (magenta bar) in S and H samples excluding 2kb on either side of the bait. Black arrows represent annotated gene models. A single copy of the penta-repeat (green triangles) is displayed in (E) and shows 4C tag positions and ranks (as above), repetitive index (heat map) per 24nt sliding windows, and annotated TE features. Cyan box represents a unique subregion (USR). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
The penta-repeat is targeted by CHD3a
CHD3a specifies genome-wide locations of sRNA biogenesis and is required to maintain both somatic and meiotic repression of Pl´ states [40]. To further investigate the potential involvement of the penta-repeat in Pl1-Rhoades regulation, we aligned existing day-8 seedling-derived 18-30nt RNA reads (excluding those multiply-mapping to the genome; see Methods) from Pl´ Chd3a non-mutant and chd3a-3 mutant siblings to the Pl1-Rhoades haplotype and looked for differential sRNA patterns. In Pl´ Chd3a profiles, the primary sRNA source across the repeat is within the USR (Fig 3). In both Chd3a non-mutants and chd3a-3 mutants, the USR sRNA reads are primarily 24nt (24mers)(79% and 73% respectively) and these 24mer fractions are not significantly different between genotypes (p = 0.27, 2-sample t-test). In the chd3a-3 mutant profiles, however, a 48% reduction (p = 0.02, 2-sample t-test) of USR sRNAs is observed and this reduction is associated with a novel peak (15.8-fold increase) of sRNAs mapping to the flanking PIF-Harbinger-type TE (p = 0.03, 2-sample t-test)(Fig 3). Again, 24mers represent the majority of sRNAs mapping to this TE in both mutants (77.7%) and non-mutants (67.8%), and these fractions do not differ significantly between genotypes (p = 0.56, 2-sample t-test). Because 24nt RNAs are diagnostic of RNAP IV action [34], these findings imply that CHD3a action normally positions RNAP IV at the USR in a paramutagenic Pl´ state. In the absence of CHD3a, RNAP IV is still recruited to penta-repeat sequences but now acts on a DNA TE adjacent to the USR. These data indicate that one or more penta-repeat units are a target of both RNAP IV and CHD3a actions.
Alignments of uniquely-mapping 18-30nt reads from libraries representing single Chd3a / Chd3a (A-B), and chd3a-3 / chd3a-3 (C-E) Pl´ 8-day seedlings across a single repeat unit (F) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
USR nuclease sensitivity and RNAP II-derived RNA levels differ between Pl-Rh and Pl´ states
We reasoned that because the USR is targeted by CHD3a, differences in chromatin structure might distinguish Pl-Rh and Pl´. To test this, we measured relative USR nuclease sensitivity by micrococcal nuclease (MNase)-qPCR. After using a mild MNase digestion to produce a gel-separated ladder of nucleosome fragments, mono- and dinucleosome-bound DNA was isolated from pooled isogenic day-8 Pl-Rh (Pl-Rh / Pl-Rh) and Pl´ (Pl´ / Pl´) seedlings (3 biological replicates each) and the presence of USR sequences was quantified using four tiled amplicons (Fig 4). The central USR region (amplicons P2 and P3) was significantly more associated with mono- and dinucleosomes in the Pl-Rh state compared to Pl´ (p<0.05; two-sample t-test)(Fig 4B). From this result we infer that the Pl´ USR is either associated with a less accessible and therefore less nuclease sensitive chromatin structure, or it is more nucleosome free and hence more nuclease sensitive than that of Pl-Rh. Regardless, changes in USR MNase accessibility in day-8 seedlings are associated with differences in seedling and plant pigmentation [6].
(A) Relative locations of amplicons (black lines) used in (B). (B) Fold DNA relative to Pl´ samples of mono- and dinucleosome bound DNA after light MNase digestion measured by qPCR in triplicate Pl-Rh and Pl´ seedlings. * p < 0.05. Error bars are s.e.m.
We next evaluated whether Pl1-Rhoades expression also corresponded with RNAP II-dependent USR-derived RNA levels using oligo(dT)-primed qRT-PCR on RNA isolated from triplicate isogenic Pl-Rh and Pl´ individuals. We chose male flowers for this analysis since Pl1-Rhoades-dependent pigment production is much stronger in male flowers than in day-8 seedlings. Poly A+ USR-containing RNA levels were 15-fold higher (p = 0.027, two-sample t-test) in Pl-Rh versus Pl´ individuals (Fig 5). Thus, differences in chromatin accessibility at USR sequences producing CHD3a-dependent sRNAs in seedlings coincides with differential USR RNA abundances that positively correlate with Pl1-Rhoades expression.
Mean fold RNA (2-ΔΔCt)(±s.e.m) across the USR measured by qRT-PCR normalized to gapdh levels in biological triplicate Pl-Rh and Pl´ male flowers. Error bars are s.e.m.
Paramutagenic states are associated with an RNAP IV complex generating USR 24nt RNAs
All known MOP and RMR factors influence 24nt RNA production, or locations thereof, therefore we looked for paramutation-specific sRNA profiles between biologically triplicate isogenic Pl-Rh and Pl´ individuals. Six combined Pl-Rh and Pl´ 4cm cob sRNA datasets (see S6 Table for library statistics) were first genome filtered to exclude multiply-mapping reads (see Methods) and then used to call uniquely-mapping 18-30nt sRNA clusters across the Pl1-Rhoades haplotype using ShortStack [75]. Normalized read abundances from each dataset were then compared for each cluster. Thirty clusters were identified–all dominated by 24mers, mean 80%–(see S2 File; see S3 Fig) with clusters 24–25 overlapping the repeat region (24 immediately before the USR and 25 overlapping the USR and adjacent region) and 12–13 covering parts of the doppia fragment (see S4 Fig). Only cluster 25 differed significantly, with average sRNA abundances 9.9-fold higher in Pl´ compared to Pl-Rh libraries (see S2 File; p = 0.003, 2-sample t-test). Day-8 seedling sRNA abundances were also compared across these same 30 clusters (see S2 File) and only cluster 25 was significantly different; 4.4-fold higher in the Pl´ libraries (p = 0.002, 2-sample t-test). Cluster 25 also has the most abundant sRNA reads in Pl´ and Pl-Rh / Pl´ datasets from both cob and seedling. Thus, in both vegetative and reproductive tissues, the Pl´ state is associated with increased sRNA production specifically from a subregion of one or more penta-repeat units.
Visualizing both cob and seedling sRNA read alignments, we found abundant USR reads in both Pl´ and Pl-Rh / Pl´ heterozygotes and very few of these reads in Pl-Rh libraries (see Figs 6, S5 and S6). Thus, the significant differences in cluster 25 read abundances noted above predominantly reflect the USR profiles. To further characterize these USR-derived sRNAs, the reads were collapsed to their 5´-most coordinate and the mean reads per million (rpm) were graphed in 10nt bins for each tissue and genotype (Fig 6H–6M), colored by length and separated by strand. In Pl´ and Pl-Rh / Pl´ cobs and seedlings, the reads primarily represent 24mers from both strands at similar frequencies. These 24mers are absent in Pl-Rh cobs and seedlings indicating the change in sRNA abundance between allele states largely represents changes in 24mer production. Interestingly, in cob profiles, we noted an antisense-specific bias for 23mers that may represent a strand bias for RNAP IV transcription [76]. The Pl-Rh / Pl´ heterozygotes also produced USR 24mers at levels similar to those of Pl´ individuals indicating that the formerly Pl-Rh allele may already be a source of 24nt RNA biogenesis in day-8 seedlings. In contrast, 4cm cob sRNA libraries from the B73 inbred had few reads representing the corresponding repeat region in the B73 genome (see S7 Fig).
Alignments of uniquely-mapping 18-30nt reads from libraries representing single Pl-Rh (A), Pl´ (B), and Pl-Rh / Pl´ (C) immature cob and seedling (D-F) across a single repeat unit (G) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray). Mean rpm over the USR in 10nt bins is shown in stacked bar graphs for cob Pl-Rh (H), Pl´ (I), and Pl-Rh / Pl´ (J), and seedling Pl-Rh (K), Pl´ (L), and Pl-Rh / Pl´ (M) genotypes. Reads colored by length mapping to the plus or minus stands are shown above and below the line respectively.
We next previewed the entire penta-repeat unit sRNA landscapes by including both unique and multi-mapping reads from Pl-Rh, Pl´ and Pl-Rh / Pl´ in cob and seedling (Methods). In all datasets, profiles were dominated by a similar multimapping read cluster precisely over the repetitive DNA-7_ZM TE, dwarfing the adjacent USR read peak (see S8 Fig). While the genomic source of these TE sRNAs is unknown, they could contribute to recruiting RNAP IV or V in the Pl´ state.
We expected the USR 24mers should be sourced by RNAP IV because RPD1 is responsible for nearly all maize 24nt RNA production [34]. To test this, we compared sRNA reads from ~4cm cob libraries representing two combined Pl´; Rpd1 / rpd1-1, and one rpd1-1 / rpd1-1 sibling sample (Fig 7A and 7B). Because 24nt RNAs are largely depleted in rpd1-1 mutants, other size classes are relatively more abundant in sRNA datasets; to compensate, we normalized the fraction of each size class in both mutant and non-mutants over the relative fraction of 21nt RNAs which are unaffected by loss of RNAP IV [34]. Less than 10% of genome-wide 24mers remained in the rpd1-1 mutant library (see S7 Table) and few USR sRNA reads of any size were represented (Fig 7B). We therefore conclude that the majority of USR sRNAs depend on RNAP IV, though the specific RNAP IV subtype(s) acting there remains unknown.
Alignments of uniquely-mapping 18-30nt reads from libraries representing Rpd1 / rpd1-1 (two combined) (A), rpd1-1 / rpd1-1 (B), Rmr1 / rmr1-1 (C), rmr1-1 / rmr1-1 (D), Rmr2 / rmr2-1 (E), rmr2-1 / rmr2-1 (F), Dcl3 / dcl3-2 (G), and dcl3-2 / dcl3-2 (H) immature cob across a single repeat unit (I) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
Peptides of RMR1 and its paralog CHR167 co-purify with epitope-tagged RPD1 [37], therefore it is possible that these two putative helicases define additional functional RNAP IV subtypes to those distinguished by alternate second largest subunits. Because ethidium-bromide staining of gel-fractionated sRNAs indicated that RMR1 is required for ~68% of all immature cob 24nt RNAs [32], we compared ~4cm cob sRNA profiles from single Pl´; Rmr1 / rmr1-1 and rmr1-1 / rmr1-1 siblings. In contrast to the rpd1 mutant profiles, in the rmr1-1 mutant, genome-wide normalized 24mers (see S7 Table; see Methods) were ~48% of those of their heterozygous sib, and these reductions were almost exclusive to repetitively-mapping 24mers. Nearly all USR sRNA reads were depleted in the rmr1-1 mutant library (Fig 7C and 7D), indicating that one or more penta-repeat USRs are targeted by an RMR1-specific RNAP IV subtype in this reproductive tissue.
We also reanalyzed existing 4cm cob sRNA datasets from rmr2-1 and dcl3-3 mutants and heterozygous non-mutant siblings that indicated RMR2 and DCL3 account for ~65% and ~95% of all 24mers, respectively [39,41]. Abundant USR-derived sRNA reads were found in both non-mutants but were depleted in the corresponding rmr2-1 and dcl3-3 mutant libraries (Fig 7E–7H). Thus, in all the rmr mutants we profiled, there is a loss of USR-derived 24mers, even when abundant genome-wide 24mers remained in rmr1-1 (48%) and rmr2-1 (35%) mutants. These findings highlight a subset of RNAP IV-dependent 24mers whose USR-specific biogenesis is associated with paramutagenic Pl´ states.
USR cytosine methylation patterns distinguish Pl-Rh and Pl´ states
Since Pl-Rh and Pl´ are distinguished by differential USR sRNA production, we examined whether these also correspond with differing 5mC patterns. Seedling-derived genomic DNA samples isolated from single isogenic Pl-Rh, Pl´, and Pl-Rh / Pl´ individuals were treated with APOBEC (NEB) to deaminate unmethylated cytosines and then the USR segment where differences in sRNA production and MNase-sensitivity were observed was PCR amplified. Resulting amplicons were then cloned and Sanger-sequenced. We observed that this USR segment in Pl-Rh has almost no 5mC in any context while ~88% of cytosines in CG and 55–70% in CHG–yet few (6–7%) in CHH–contexts were methylated in both Pl´ and Pl-Rh / Pl´ genotypes (Fig 8). These data show that reduced expression and paramutagenic action of the Pl´ state correlate with elevated 5mC levels at most, if not all, of the five USR copies.
Cytosine methylation profiles of individual PCR amplicons of genomic DNA from Pl-Rh (A), Pl´ (B) and Pl-Rh / Pl´ (C) day-8 seedlings representing the bracketed region of the USR shown in (D). Circles (open: unmethylated, closed: methylated) represent cytosines in CG (red), CHG (blue), and CHH (green) contexts. Black arrows in (D) represent PCR primers used to generate the sequenced amplicons.
Discussion
Our findings are consistent with the Pl1-Rhoades penta-repeat acting as both a distal transcriptional enhancer and as a feature required for paramutagenic function. The only abundance difference of uniquely mapping sRNAs distinguishing paramutable Pl-Rh and paramutagenic Pl´ across 192kb occurs at one or more USR copies within the penta-repeat. When Pl1-Rhoades is in the highly expressed Pl-Rh state, portions of the USR display differential nuclease sensitivity, increased polyadenylated RNA levels, few 5mCs, and physical association with sequences immediately upstream of the Pl1-Rhoades TSS. From these findings, we conclude that the USR has regulatory significance for the Pl1-Rhoades haplotype. Furthermore, these data indicate that the USR is RNAP II-transcribed in the Pl-Rh state but RNAP IV-transcribed in the Pl´ state. Thus, this small non-coding region within the penta-repeat context appears to serve as a switchable distal element specifying meiotically heritable regulatory states.
Although challenges remain in defining features of active plant enhancers, chromatin accessibility, histone acetylation, absence of 5mC, and promoter proximity appear diagnostic [77]. Both the b1 hepta-repeat and proximal P1-rr repeated sequences can promote transcriptional activation in transgenesis-based reporter assays [58,60]. Compared to the repressed B´ state, the b1 hepta-repeat sequences in the active B-I state 1) are more DNase I sensitive, 2) are more represented in a FAIRE-based protein-depleted DNA fraction, 3) are associated with elevated H3 acetylation [51,78], 4) have reduced 5mC levels [51], and 5) physically associate with the b1 CDS at higher frequencies [79]. The hepta-repeat junction sequences are also, like metazoan enhancers (reviewed in [80–82]), bi-directionally RNAP II-transcribed, but this occurs at similar levels in both the B-I and B´ states [47]. Likewise, sRNA levels so far appear similar across the hepta-repeat [38,47] whereas the abundance of both penta-repeat polyadenylated transcripts and RNAP IV-dependent sRNAs distinguish Pl-Rh from Pl´. The P1-rr enhancer is also transcribed at similar rates between allele states, but like Pl1-Rhoades, sRNA levels appear higher in the paramutant state [50]. These distinct characteristics point to variations in the mechanism driving paramutation at different loci; this intriguing finding demands further investigation.
The polyadenylated USR-containing RNAs associated with the active Pl-Rh state indicate this region is transcribed, but the origin(s), direction(s), and length(s) of these RNAs remains unknown. Because neither Arabidopsis RNAP IV nor RNAP V produce polyadenylated RNAs [83–87], we infer that these USR-containing transcripts are RNAP II-derived. Whether this transcription reflects a more open chromatin nucleosome-free configuration region, however, remains equivocal. Because the B73 AGPv4 annotated PAS of the adjacent gene immediately downstream of the penta-repeat is less than 100bp from the end of the last repeat, one idea is that pre-termination transcription continues into the repeat array and a downstream PAS is used. Alternatively, the penta-repeat RNAs could initiate within USR sequences or from its resident TE fragments. In either case, a nascent RNA containing USR sequences and/or the adjacent highly repetitive DNA-7_ZM TE could serve as an essential scaffold transcript for the initial recruitment of RNAP V via sRNA-directed argonaute complexes [88]. Hence, a paramutable Pl-Rh state may require such polyadenylated penta-repeat RNAs for paramutation initiation while a paramutagenic Pl´ state likely uses RNAP V-sourced transcripts to maintain sRNA biogenesis.
The relationship of sRNAs with endogenous alleles undergoing paramutation has to this point remained unclear. Here we found that the appearance of USR-specific 24nt RNAs in both early seedlings and immature cobs precisely correlates with the reduced plant pigment phenotypes and paramutagenic action diagnostic of Pl´ states [6]. The similar USR sRNA profiles observed between Pl´ and Pl-Rh / Pl´ seedlings indicate that RNAP IV is likely already recruited to the naive Pl-Rh allele state early in development. Our mutant sRNA analyses show that loss of sRNA biogenesis from a discrete USR segment is coincident with increased Pl1-Rhoades transcription (rmr2-1, rpd1-1, and dcl3-3), increased Pl1-Rhoades mRNA-stability (rmr1-1) and, meiotically-heritable reversion of Pl´ to Pl-Rh states (rmr1-1, rmr2-1, rpd1-1 and dcl3-3) [31,32,39,41,74].
Because 24nt USR RNA production depends on RNAP IV and other RMR proteins in an RdDM-like mechanism, it is reasonable to posit that RNAP IV-dependent chromatin changes at the USR impede Pl1-Rhoades transcription. Given that rmr1-1 mutant cobs lose most USR-specific 24mers, this indicates that the RNAP IV subtype occupying the USR in immature cobs is associated with RMR1 and not its CHR167 paralog [37]. The redistribution of sRNAs seen in chd3a mutants argues that CHD3a specifies where the seedling-based RNAP IV complexes function. Since CHD3 proteins are associated with nucleosome remodeling [89], and the Arabidopsis CHD3 PICKLE can move nucleosomes in vitro [90], it is likely that the chd3a-3 mutant penta-repeat sRNA profile reflects changes in nucleosome positioning that impact RNAP IV localization. The appearance of 5mC, coincident with USR sRNA biogenesis, agrees with a model in which CHD3a-positioned nucleosomes direct RNAP IV-dependent 24nt RNA production that, analogous to Arabidopsis RdDM, specifies de novo 5mC patterns.
We found a correlation between 5mC appearance at the USR sequences and the Pl´ state. 5mC presence appears generally associated with alleles displaying paramutagenic behaviors, including 1) at the 5´ ends of divergently oriented r1 genes in R-r:standard repressed states [48], 2) within repeats flanking the p1 CDS in paramutagenic p1 haplotypes [5,44,49,58], 3) at the b1 hepta-repeat in the B´ state–with the exception of specific cytosines being more methylated in B-I [51,59,60,78], 4) the repressed phenotype conferred by a paramutagenic lpa1-241 allele is partially compensated by 5-azacytidine treatment [7] implicating 5mC in repression, and 5) in Arabidopsis autotetraploids, a spontaneously silenced epiallele of a partially-repeated hygromycin resistance transgene displays paramutagenic properties coincident with increased 5mC and 24mers primarily across the repeat region [14,91]. Moreover, 6) in tomato, the paramutagenic sulfurea allele also has increased 5mC in the promoter region and part of the 5´ CDS [92]. Changes in 5mC and sRNA production are also common when divergent genomes are combined [9,93,94], but whether these are causative of, or responsive to, changes in gene regulation remains unknown. Although 5mC presence is a unifying hallmark of paramutagenic allele states, 5mC alone is insufficient to predict paramutagenicity.
Arabidopsis RdDM facilitates de novo cytosine methylation in CG, CHG, and CHH contexts (reviewed in [33,95]), and once established, these modifications can be maintained in the absence of RdDM by a maintenance methyltransferase (MET1; CG) and a pair of chromomethyltransferases (CMT3; CHG, and CMT2; CHH) [96] that are recruited to methylated H3K9 [97]. Because maize lacks a CMT2 ortholog [96], mCHH presence is assumed to reflect ongoing RdDM [98]. Interestingly, mCHH is specifically found in an enhancer region of the paramutagenic P1-rr´ allele, but not in the stably silenced, non-paramutagenic, P1-prTP allele [44]. Indeed, maize has little genome-wide mCHH (~5%); its presence is more correlated with sRNA production than with mCG or mCHG [94], and loss of MOP1/ZmRDR2, RP(D/E)2a, or RPD1 results in genome-wide reductions [98]. Although most maize mCHH depends on active RdDM, the CMT3 orthologs encoded by zmet2 and zmet5 are required to maintain some mCHH [98]. Since most of our sequenced USR amplicons from Pl´ and Pl-Rh / Pl´ seedlings are devoid of mCHH, we conclude that active RdDM is not occurring in most of the nuclei we sampled. Therefore, the high mCG (88%) and mCHG (55–70%) levels we observed mostly reflect maintenance methylation. While initial recruitment and trans-homolog communication of paramutation may depend on an RdDM-like mechanism primarily occurring in cells not sampled by our analysis, 5mC maintenance pathways are likely required for somatic and intergenerational stability of repressed allele states.
While 5mC could be a meiotically heritable feature specifying paramutagenicity, it remains curious that, so far, no downstream (effector) proteins in an RdDM-like pathway have been identified in the mop- and rmr-based genetic screens. In Arabidopsis, sRNAs may direct the action of the SUVH2 and SUVH9 lysine methyltransferases to specify H3K9me2 [99], which can be maintained in a feedforward loop with mCHG through the actions of CMT3 and SUVH4/5/6 [100]. Arabidopsis SAWADEE HOMEODOMAIN HOMOLOGUE1 (SHH1) recruits RNAP IV to H3K9me2 [101]. A related maize ortholog, ZmSHH2a, also binds methylated H3K9 with the highest affinity for H3K9me1 [102]. ZmSHH2a is complexed with RP(D/E)2a and MOP1/ZmRDR2 [37], therefore Wang et al. [102] propose that ZmSHH2a forms a connection between RNAP IV recruitment and H3K9me1. Thus, similar to eukaryotes such as fission yeast that lack any cytosine methylation machinery, H3K9me could represent a heritable mark as well [103]. Indeed, loss of either tomato CMT3 or SUVH4 orthologs lead to reactivation of a paramutagenic sulfurea allele [104]. Additionally, because CHD3 proteins affect the regulation of loci marked by H3K27me3 [105–107], and CHD3a influences penta-repeat 24nt RNA production, it is possible that H3K27 modifications are also involved in maintaining a meiotically heritable epigenetic mark distinguishing repressed Pl1-Rhoades states. Loss of MOP1/ZmRDR2 or RPD1 results in a reduction of both H3K9 and H3K27 dimethylation at the b1 hepta-repeat, suggesting a connection between these marks and b1 paramutation [78]. Indeed, recent work in Drosophila implicates H3K27me3 and inter-chromosomal contacts in facilitating paramutation at the Fab2L transgene [19].
Gametic sRNAs may also contribute to the inheritance of repressed Pl´. In other metazoan examples of paramutation, it is the cytoplasmic transmission of Argonaute-bound sRNAs that confers heritable paramutagenic action (reviewed in [25]). Indeed, Argonaute-bound sRNAs are sufficient to initiate RdDM in Arabidopsis [88] and shRNA constructs can induce the change of B-I to B´ in transgenic maize [47]. In maize, however, paramutagenicity exclusively co-segregates with specific alleles [4,6,58,108]. Additionally, when aneuploid sperm cells are generated from haploid Pl´ gametophytes–as can be accomplished using B-A chromosomes [109]–only sperm cells contributing Pl´ chromosomes transmit paramutagenic function [110]. This result indicates that any sperm-transmitted sRNAs are by themselves insufficient to induce paramutation.
We found that, similar to specific r1 and b1 haplotypes, repeat copy numbers correlate with paramutagenic activities [57,59,61]. Importantly, the presence of one or two USRs in non-paramutable pl1 haplotypes indicates that more repeats are required to obtain paramutagenic properties. In r1 paramutation, repeated r1 coding regions and flanking sequences need not be adjacent, or in cis, to promote paramutagenic function [61]. Because the NC350 and CML52 haplotypes share identical 3´ repeats, we expect that either the solo repeat found upstream of Pl1a-CML52 and/or the additional Pl1b-CML52 gene confers its paramutagenic properties. In addition, the pl1-bol3 haplotype containing three pl1 gene copies–none having the doppia fragment [111]–is reportedly paramutagenic [112]. The pl1-bol3 downstream region remains uncharacterized, however, so whether any paramutagenic activity is dependent on pl1 gene copy number or other cis-regulatory features is unclear. Our findings are consistent with the emerging consensus that repeated sequences are essential for paramutagenic function.
The distinct sequences and structural arrangements of the B1-I and Pl1-Rh distal repeats may be related to the observed differences in allele stabilities and locus-specific RMR requirements. Neither RMR1 nor RMR12 repress B´ [32,40]. Each individual unit of the b1 hepta-repeat (853bp) consists entirely of unique sequence [59], while the 2092bp penta-repeat unit is mainly consisting of TE fragments with an embedded 390bp unique sequence. Unlike B´, where heritable reversions to a fully expressed B-I state have not been observed [52,113,114], the Pl´ state is less stable and can revert to Pl-Rh when either hemizygous [115], heterozygous with certain recessive alleles [66,115] or in the persistent absence of RMR1 [31,41], RPD1 [74], CHD3a [40], or DCL3 [39]. Whether the size, copy number, location, and/or composition of these b1 and pl1 repeats contribute to the observed differences in allele behaviors and mechanistic requirements remains to be explored.
Depending on the phenotypic trait affected, metastable alleles having paramutation-like properties may be desirable or detrimental to breeding efforts. With long-read-based whole genome assemblies now resolving repeated sequences, it should be possible to predict which alleles of genes having agronomic importance might show similar dynamic behaviors. Such efforts will require ATAC-seq profiles and Hi-C interaction maps to identify repeat-associated enhancers, and sRNA distributions to highlight regions of ongoing RdDM. Additionally, RNA-seq profiles of RNAP IV-deficient plants would help validate the regulatory significance of such repeats. Crop improvement programs can now identify and manage these examples of heritable epigenetic variation.
Methods
Genetic materials
Inbred lines A619, A632, and B73 were sourced from the North Central Regional Plant Introduction Station (USDA-ARS, Ames, IA). The Pl1-Rhoades haplotype introgressed into A619 ([74]; herein designated A619 Pl1-Rhoades; 99.25% A619 for Pl-Rh and 98.5% A619 for Pl´) came from a previously described color-converted W23 stock (herein designated W23 Pl1-Rhoades) [6] developed by Ed Coe, Jr. (USDA-ARS, University of Missouri, Columbia, MO) and maintained in several lines obtained from the Maize Genetics Cooperation Stock Center (MGCSC)(USDA-ARS, University of Illinois, Urbana, IL) and Vicki Chandler (University of Oregon, Eugene, OR). A T6-9 (043–1) [116] interchange chromosome (T Pl1-Rhoades) having linkage between Pl1-Rhoades (6L) and a recessive 9S waxy1 allele (wx1) was introgressed into B73 (herein designated B73 T Pl1-Rhoades; 98.44%) as previously described [66,74] and into A619 (herein designated A619 T Pl1-Rhoades; ~97% A619) and A632 (herein designated A632 T Pl1-Rhoades; 99.25% A632). The BAC library and 4C analysis were derived from Pl-Rh type A619 Pl1-Rhoades plants. qRT-PCR was performed on W23 Pl1-Rhoades, A619 Pl1-Rhoades, and B73 T Pl1-Rhoades individuals. MNase assays were done with A619 Pl1-Rhoades seedlings. 5mC analyses and sRNA libraries were made from isogenic homozygous Pl-Rh, Pl´ and heterozygous (Pl-Rh / Pl´) cobs and seedlings of B73 T Pl1-Rhoades parentage (progenies 161453, 161483, 161482, 170671, 170673, 170674). Additional sRNA libraries were derived from immature cobs of Rpd1/rpd1-1 and rpd1-1/rpd1-1 siblings from progeny 80031 (~25% W23, 25% A632) and Rmr1/rmr1-1 and rmr1-1/rmr1-1 siblings from progeny 80886 in A632 (BC4F3). For mutant analyses, all non-mutant siblings were homozygous for Pl1-Rhoades in a Pl´ state while mutants had darky colored anthers consistent with derepression of Pl1-Rhoades.
BAC identification and sequencing
Snap-frozen developing cobs (progenies 91353 and 91363) were submitted to Amplicon Express (Pullman, WA) to construct a ~3.7X coverage pCC1 BAC-based library and representative Hybond-N+ (Amersham Biosciences) filter arrays. 73,728 clones were screened by formamide-based hybridization to a random-primed 32P radiolabeled 0.9kb HindIII fragment from the pJH1 plasmid [54]. BAC DNA from ten clones was isolated using a Qiagen miniprep kit and evaluated for pl1 sequences using pl1-specific PCR amplification (see S8 Table for primers). Six PCR-positive clones (60_P21, 25_M18, 51_M22, 53_O24, 161_K19, 192_C12) were cultured with Epicentre Biotechnologies (Madison, WI) CopyControl BAC Induction Solution and BAC DNA was isolated with modification to an established protocol [117] including 1μl of 100mg/ml RNaseA in Solution I and adding a final phenol:chloroform extraction followed by chloroform extraction and isopropanol precipitation. BAC ends were Sanger sequenced by the Vincent J. Coates Genome Sequencing Laboratory (UC Berkeley) as per Poulsen and Johnson [118] (see S8 Table for primers) using 2μg of BAC template and 3 pmol of primer, in 20μl Sanger sequencing reactions with the following thermocycle steps: 3 min at 96°C followed by 99 cycles of 10 sec at 96°C, 10 sec at 50°C, 4 min at 60°C. BLASTn aligned the respective end sequences of all six BACs to 6L positions flanking the pl1 gene model. All six BAC clones were later prepared and sequenced on the Pacific Biosciences (PacBio) platform at The Genome Analysis Centre, TGAC, Norwich, UK (Now Earlham Institute).
BAC assembly and annotation
Raw PacBio reads (see S1 Table for statistics) were filtered and assembled into partial or complete linear contigs at The Genomic Analysis Centre (see S3 File). BAC backbone sequences were identified by BLASTn and removed. The resulting genomic sequences were aligned to each other and the existing Pl1-Rhoades reference using the Multiple Align feature of Geneious (v6.1.8; [119]). Together, these BAC sequences extended the Pl1-Rhoades reference sequence to 192kb, with six independent BACs supporting the sequence over the pl1 coding region and four supporting the sequence over the penta-repeat. One of the BAC sequences (60_P21) was unable to be assembled into a single contig, so only those fragments which overlap the consensus sequence of the other BAC sequences were included. A fragment that only overlapped the consensus by 1kb consisting entirely of Gypsy transposon sequence was also excluded. Gene models were predicted by BLASTing the assembled BAC sequence to cDNA molecules from the B73 AGPv4 genome [70]. Transposons and TE fragments were predicted using CENSOR on April 26, 2022 ([71]; see S4 File).
Haplotype comparisons
~70kb extracted from the Pl1-Rhoades BAC contig containing the Pl1-Rhoades allele and surrounding region was aligned to the corresponding region in the B73 AGPv4 genome [70] and the draft CML52 and NC350 genomes [73] with GEvo [120] using the default settings with a minimum high-scoring segment pair size of 400nt. The resulting alignments were manually redrawn using Adobe Illustrator at scale.
4C analysis
Tissue was collected from biological triplicate inner husk leaves at silking and from inner stem leaves from A619 Pl1-Rhoades (Pl-Rh) V5 seedlings. Interacting chromatin was isolated from these tissues as described for 3C experiments [79,121,122]. In short, 2-3g tissue was crosslinked in 1X PBS with 2% formaldehyde for 1 hour. Nuclei were resuspended in 1.2X Buffer B (Roche) and treated with 0.2% (w/v) SDS followed by 2% (w/v) Triton X-100 and digested with 400 U DpnII (NEB, 50U/μl) overnight. DpnII was heat-inactivated, and for intra-molecular ligation of crosslinked digested DNA, the sample was incubated in a volume of 7ml with 100U T4 ligase (Promega, 20U/μl) at 16°C for 5 hours, followed by 45 min at RT. Chromatin was de-crosslinked and ligated DNA (3C sample) was recovered by precipitation and the pellet dissolved in 150μl of 10mM Tris-HCl pH 7.5. To generate 4C samples, the 3C samples were digested and ligated essentially as described [123]. The DNA was digested with 50U NlaIII (NEB, 5U/μl), NlaIII was heat-inactivated and the digested DNA was re-ligated overnight in a total volume of 14ml as described above. The ligated DNA was precipitated, dissolved and column purified (QIAquick PCR purification kit- Qiagen). To produce 4C libraries, for each sample, 8 PCR reactions were performed. 100ng of 4C DNA served as template in each inverse PCR reaction. PCR was conducted with the Expand Long Template PCR System (Roche, #11681834001) using primers fused to P5 (forward) and P7 (reverse) illumina sequencing adapters; a forward primer immediately upstream of the Pl1-Rhoades CDS, and a reverse primer at the 5´ end of the doppia transposon fragment upstream of Pl1-Rhoades facing away from the gene (see S8 Table). The program used was 2min 94°C; [30sec 94°C, 30sec 52°C, 2min 72°C] 35x; 10min 72°C, 12°C ∞. Proper 4C amplicons contained: P5 ‐ forward primer (bait sequence) ‐ DpnII ‐ interacting sequence ‐ NlaIII ‐ reverse primer ‐ P7. The 8 PCR reactions per sample were pooled and purified with the Qiaquick PCR purification kit (Qiagen). Library concentrations were measured using the Nanodrop spectrophotometer, mixed in equal concentrations and sequenced on an Illumina 2000 platform by BGI Tech Solutions (Hongkong) Co., producing single-end 49 bp reads. Replicates were processed in different lanes.
Reads were first filtered to those containing the DpnII cutsite and the 3 preceding nt of the forward primer to ensure the DpnII site was in the correct context using the BBTools (https://jgi.doe.gov/dataand-tools/bbtools/) function bbduk (options: k = 7 mink = 7 hdist = 1 adapter: CTCGATC). The resulting reads were then trimmed up to the DpnII cutsite using bbduk (options: ktrim = l k = 16 mink = 12 hdist = 2 maxlen = 28, adapter: TACGCCGGCGGAGCTC). Since each read is 49bp, setting the maximum length to 28 ensured that any read where the forward primer could not be recognized due to sequencing errors would not be retained. The NlaIII site and reverse primer, if present, were then trimmed using bbduk (options: ktrim = r k = 12 mink = 7 hdist = 1 minlen = 8, adapter: CATGAGCTATGA). Setting the minimum length to 8 ensured the reads contained an interacting sequence, and eliminated potential primer dimers; reads with longer interacting sequences were not expected to contain the NlaIII site, so presence of this site was not required. Lastly the reads were quality filtered and trimmed using bbduk (options: qtrim = r maq = 20 minlen = 8). After these filtering steps, the third replicate from each tissue had significantly fewer reads than the other two libraries (see S3 Table) and these were excluded from further analyses. The remaining four libraries were aligned to the B73 AGPv4 genome [70] using Bowtie (v0.12.8; options: -v 0 ‐‐best -m 1 -S) [124]. Those reads which either mapped uniquely or did not map were aligned to a collapsed version of the Pl1-Rhoades BAC-based sequence containing only one copy of the penta-repeat, allowing no mismatches or multiply -mapping reads, using Bowtie (v0.12.8; options: -v 0 ‐‐best -m 1 -S) [124]. All clean reads (not filtered to the genome) were also aligned to the assembled BAC sequences as above. All alignments were collapsed to their 5´ most coordinate and converted to BED format using awk. The abundance of tags at each position of the BAC sequence was determined using BEDTools genomecov [125]. Only positions with counts were retained, and 2kb on either side of the bait region were excluded to avoid artifacts. The resulting counts were normalized to rpm and ranked by total abundance (see S4 and S5 Tables).
Chromatin profiling
Mono and dinucleosome-bound DNA was obtained by isolating nuclei from six pooled day-8 A619 Pl1-Rhoades (Pl-Rh) and (Pl´) [74] seedlings in triplicate and fractionating MNase-treated chromatin as described [126]. A range of MNase concentrations was tested to identify a mild treatment which produced a full ladder representing different numbers of nucleosomes bound to DNA following gel electrophoresis. An amount of nuclei yielding ~20μg DNA were incubated for 4 minutes at 37°C with 20U MNase (Thermo Scientific EN0181) in MNase reaction buffer (50mM Tris pH 7.5, 320mM sucrose, 4mM MgCl2, 1mM CaCl2) before proteinase K treatment and DNA extraction as described [126]. The bands corresponding to mono and dinucleosomes were extracted from the gel, pooled, and three technical replicate qPCR reactions were performed on three biological replicates. Abundance of each USR amplicon was normalized to abundance of alanine aminotransferase (alt4; Zm00001d014258) (see S8 Table for primers). DNA was amplified with SensiMix SYBR No ROX (Bioline), and data was collected using an Eppendorf Mastercycler EP Gradient S thermocycler. Cycle threshold (Ct) values were calculated using the noiseband option in Eppendorf Mastercycler EP Realplex V2.2 software.
qRT-PCR
To quantify pl1 mRNA, total RNA was isolated from biological duplicates of Pl-Rh and Pl´ inner seedling and husk leaves from both A619 Pl1-Rhoades and B73 T Pl1-Rhoades lines. cDNA was generated, and qRT-PCR was performed using an Applied Biosystems 7500 Real-Time PCR System on technical duplicates. pl1 mRNAs were normalized to those from actin1 (Zm00001d010159) (see S8 Table). To quantify USR transcripts, total RNA was isolated from male flowers (florets) of triplicate Pl-Rh and Pl´ type W23 Pl1-Rhoades individuals [6] just prior to anthesis. Three florets from a single individual were pooled per sample. RNA was isolated, cDNA was generated, qRT-PCR was performed, and fold expression was calculated as described [40] except the cycle threshold (Ct) values were established using the CalQplex option in Eppendorf Mastercycler EP Realplex V2.2 software. Candidate enhancer transcripts were amplified with P3_F and P3_R primers (see S8 Table) and normalized to gapdh which was amplified using previously published primers ([127]; see S8 Table). Raw Ct values are provided in S5 File.
sRNA sequencing
Biological triplicate day-8 seedling and 4cm immature cob B73 T Pl1-Rhoades (Pl-Rh), (Pl´) and (Pl-Rh / Pl´) sRNA libraries (SRA accessions SRR15400896-SRR15400912) were made from TRizol-extracted total RNA using the Bioo Scientific (Austin, TX) Nextflex Small RNA-Seq Kit v3 as described [40], and 150bp paired-end sequencing was performed by Novogene Co. Ltd. on an Illumina HiSeq4000 platform. Only two seedling Pl-Rh / Pl´ libraries were successfully sequenced. Between 37M and 198M reads were produced per library (see S6 Table for library statistics). Two B73, two Rpd1 / rpd1-1, one rpd1-1 / rpd1-1, one Rmr1 / rmr1-1, and one rmr1-1 / rmr1-1 4cm immature cob libraries (GEO GSE52103 and SRA SRR15410418-SRR15410422) were prepared as described [41]. Briefly, TRizol-extracted total RNA were separated by PAGE, sRNAs were gel-extracted, and libraries were prepared according to the Illumina sRNA Sequencing Guide. Adapters were trimmed from rmr mutant and heterozygous cob libraries, deduplicated, and quantified using a Perl script. B73 cob libraries were filtered and trimmed using the BBTools (https://jgi.doe.gov/dataand-tools/bbtools/) function bbduk (options: ktrim = r k = 18 mink = 11 hdist = 1 maq = 10 minlen = 18 maxlength = 30); B73 immature cob libraries were not deduplicated. Pl-Rh, Pl´, Pl-Rh / Pl´ and previously published Chd3a / Chd3a and chd3a-3 / chd3a-3 ([40]; GEO GSE158990) raw reads were filtered using bbduk with options (ktrim = r k = 18 mink = 11 hdist = 1 minlen = 26 maxlength = 38). Only mate 1 was used from paired-end libraries. The resulting trimmed reads were converted to .fasta format using seqtk (https://github.com/lh3/seqtk) and deduplicated using the FASTX-Toolkit command fastx_collapser (http://hannonlab.cshl.edu/fastx_toolkit/index.html). The Nextflex Small RNA-Seq Kit v3 adapters have 4N tags flanking the insert. These tags were removed using seqtk (options: trimfq -b 4 -e 4). bbduk was used to filter rpd1, rmr1, and rmr2 immature cob datasets by length (options: minlen = 18 maxlen = 30). rpd1 and rmr1 whole genome mapping statistics (see S7 Table) were obtained as previously described for rmr2 and dcl3 [39,41]. Percent sRNAs of each size class were then normalized by the ratio of the percent 21mers in the mutants to the percent 21mers the corresponding heterozygotes since 21mer levels appear unchanged in these mutants [34,69].
18-30nt reads from all datasets were aligned to known maize rRNA and tRNA sequences (provided by Dr. Blake Meyers, Donald Danforth Plant Sciences Center) using Bowtie (v0.12.8; options: -v 0 ‐‐best -m 1 -S) [124]. Those sequences not mapping to tRNA/rRNA were considered “clean” and aligned to the B73 AGPv4 genome [70] using Bowtie (v0.12.8; options: -v 0 ‐‐best -m 1 -S) [124]. Those reads (excluding the B73 libraries) that either mapped uniquely or did not map were aligned using ShortStack [75] to a modified version of the Pl1-Rhoades haplotype sequence with the five repeats collapsed to a single copy to allow reads which could map to any repeat to map uniquely. ShortStack clusters were called using the six libraries from Pl-Rh and Pl´ cobs since these were the deepest datasets (ShortStack options: ‐‐mismatches 0 ‐‐mmap n ‐‐nohp ‐‐mincov 152 ‐‐pad 50), and sRNAs from all other libraries were quantified at these defined loci using the ShortStack option ‐‐locifile. Minimum coverage of 152 represents 0.5 rpm for those datasets, and the non-default pad of 50 (minimum distance between independent clusters) was chosen to prevent ShortStack from artificially collapsing independent clusters into a single locus because of the depth of these libraries. Reads mapping uniquely to each nucleotide of the composite unique subregion were collapsed to their 5´ most nucleotide and separated by strand and length using awk, quantified using the samtools function samtools-depth [128], and normalized to rpm. Normalized counts are displayed in 10nt bins (Fig 6H–6M). In parallel, all clean reads from Pl-Rh, Pl´, and Pl-Rh / Pl´ were mapped to the collapsed BAC sequence using Bowtie (v1.3.1; options: -v 0 ‐‐best). Representative alignments are shown for each tissue and genotype in S8 Fig.
5mC profiling
Genomic DNA was isolated from B73 T Pl1-Rhoades (Pl-Rh), (Pl´) and (Pl-Rh / Pl´) aerial day-8 seedling tissue. ~160ng DNA was mixed with 30ng unmethylated lambda DNA and sheared using a Branson sonicator (two 15 sec pulses at 7watts with 45 sec rest), and unmethylated cytosines were converted to uracils using the NEBNext Enzymatic Methyl-seq Conversion Module (NEB #E7125) followed by PCR using standard Taq DNA polymerase (NEB)(see S8 Table) to amplify fragments within the USR and lambda control. Lambda_F and R were used on the Pl-Rh and Pl´ samples while Lambda_F2 and R2 were used on Pl-Rh / Pl´ sample. The PCR products were then purified using AxyPrep Mag PCR Clean-Up Kit (Axygen MAG-PCR-CL-5). All PCR amplicons ‐ with exception of the lambda amplicon from the Pl-Rh / Pl´ sample ‐ were cloned into pGEM-T easy vectors (Promega A1360) and transformed into competent NEB5-alpha E. coli (NEB #C2987H). Colonies were selected by blue white screening and purified plasmid DNA was Sanger sequenced at The Genomics Shared Resource in the OSU Comprehensive Cancer Center. A 123bp region with full coverage high quality sequence in all amplicons was analyzed by Kismeth [129] for cytosine methylation. For the Pl-Rh / Pl´ sample, Sanger sequencing was performed on the lambda amplicons directly. Complete cytosine conversions were seen for all lambda controls (see S9 Table).
Supporting information
S1 Fig. Structures of pl1 haplotypes.
(A) Structure of a single copy of the Pl1-Rhoades penta-repeat. Cyan box: unique subregion (USR) used as a hybridization probe. Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray). BglII (B) and NsiI (N) restriction sites producing hybridizing fragments in the Pl1-Rhoades (B) and pl1-B73 haplotypes (C) are shown. Black boxes represent gene models, large arrows represent the penta-repeat sequences, cyan boxes represent the USR, and green shading represents the region of recombination for pl1-R30. (D) Southern blot of Pl1-Rhoades (Pl1-Rh), pl1-B73, and pl1-R30 genomic DNA digested with BglII and NsiI and probed with a radiolabeled USR fragment.
https://doi.org/10.1371/journal.pgen.1011296.s001
(PDF)
S2 Fig. Pl1-Rhoades mRNA expression profiles.
Mean fold pl1 mRNA (2-ΔΔCt) in Pl-Rh and Pl' inner seedling and husk leaves relative to inner husk leaves from Pl' individuals measured by qRT-PCR normalized to actin1 levels in biological duplicate plants from A619 and B73 lines.
https://doi.org/10.1371/journal.pgen.1011296.s002
(PDF)
S3 Fig. sRNA clusters across the Pl1-Rhoades haplotype.
(A) Locations of 30 clusters called by ShortStack across the Pl1-Rhoades haplotype (blue bars) with clusters 24 and 25 which overlap the penta-repeat (B) highlighted. Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
https://doi.org/10.1371/journal.pgen.1011296.s003
(PDF)
S4 Fig. Immature cob doppia sRNA profiles.
Alignments of uniquely-mapping 18-30nt reads from libraries representing single Pl-Rh (A-C), and Pl' (D-F) immature cobs across the doppia fragment upstream of the Pl1-Rhoades coding sequence and the 5' region of Pl1-Rhoades exon 1 (G) in reads per million (rpm). (H) Clusters called by ShortStack.
https://doi.org/10.1371/journal.pgen.1011296.s004
(PDF)
S5 Fig. Immature cob penta-repeat sRNA profiles.
Alignments of uniquely-mapping 18-30nt reads from libraries representing single Pl-Rh (A-C), Pl' (D-F), and Pl-Rh / Pl' (G-I) immature cobs across a single repeat unit (J) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
https://doi.org/10.1371/journal.pgen.1011296.s005
(PDF)
S6 Fig. Seedling penta-repeat sRNA profiles.
Alignments of uniquely-mapping 18-30nt reads from libraries representing single Pl-Rh (A-C), Pl' (D-F), and Pl-Rh / Pl' (G-H) seedlings across a single repeat unit (I) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
https://doi.org/10.1371/journal.pgen.1011296.s006
(PDF)
S7 Fig. B73 sRNA profiles across penta-repeat region.
Alignments of uniquely-mapping 18-30nt reads from libraries representing single B73 inbred (A-B) immature cobs across B73 sequence (C) equivalent to the Pl1-Rhoades penta-repeat repeat unit (D) in reads per million (rpm) with structural differences highlighted (dotted lines). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray). Only transposons unique to B73 are labeled in (C).
https://doi.org/10.1371/journal.pgen.1011296.s007
(PDF)
S8 Fig. Multimapping penta-repeat sRNA profiles.
Alignments of all unique and multimapping 18-30nt reads from libraries representing single Pl-Rh, Pl', and Pl-Rh / Pl' cob (A-C) and seedlings (D-F) across a single repeat unit (G) in reads per million (rpm). Arrows represent DNA transposons (light gray), Helitrons (black), and LTR retrotransposons (dark gray).
https://doi.org/10.1371/journal.pgen.1011296.s008
(PDF)
S2 Table. Paramutagenic properties of pl1 haplotypes from NAM founder lines.
https://doi.org/10.1371/journal.pgen.1011296.s010
(DOCX)
S3 Table. Library statistics for 4C datasets.
https://doi.org/10.1371/journal.pgen.1011296.s011
(XLSX)
S6 Table. Library statistics for sRNA datasets.
https://doi.org/10.1371/journal.pgen.1011296.s014
(DOCX)
S7 Table. rmr mutant sRNA genome mapping statistics.
https://doi.org/10.1371/journal.pgen.1011296.s015
(XLSX)
S9 Table. Enzymatic cytosine conversion efficiency of unmethylated lambda DNA.
https://doi.org/10.1371/journal.pgen.1011296.s017
(DOCX)
S1 File. Fasta-formatted Pl1-Rhoades haplotype sequence.
https://doi.org/10.1371/journal.pgen.1011296.s019
(TXT)
S3 File. Fasta-formatted BAC sequences containing Pl1-Rhoades.
https://doi.org/10.1371/journal.pgen.1011296.s021
(TXT)
S4 File. CENSOR-based, gff3-formated, transposon annotations.
https://doi.org/10.1371/journal.pgen.1011296.s022
(TXT)
Acknowledgments
We thank the Maize Genetics Cooperation Stock Center (MGCSC)(USDA-ARS, University of Illinois, Urbana, IL) and North Central Regional Plant Introduction Station (USDA-ARS, Ames, IA) for contributing valuable genetic materials. We are grateful to Drs. Blake Meyers (Donald Danforth Plant Sciences Center) and Stacey Simon (University of Delaware) for constructing and sequencing the immature cob sRNA libraries from the various rmr mutants and non-mutant siblings. We thank Dr Mario Caccamo (NAIB East Malling Research) for his assistance in sequencing the Pl1-Rhoades BACs (sequencing was delivered via the BBSRC National Capability in Genomics (BB/J010375/1) at the Earlham Institute by members of the Genomics Pipelines Group), Drs. Pawel Krajewski and Dimitrios Zisis (Institute of Plant Genetics, Polish Academy of Sciences) for advice with the 4C data analysis, and Ludek Tikovsky and Harold Lemereis for taking care of the maize plants (Swammerdam Institute for Life sciences, University of Amsterdam). NCBI, Gramene, and MaizeGDB provided critical sequence and data resources. Maize nurseries were supported in part by the University of California College of Natural Resources Oxford Facilities Unit, the Ohio Agricultural Research and Development Centers Waterman Agricultural and Natural Resources Laboratory, and The Ohio State University College of Arts & Sciences Biological Sciences Greenhouse.
References
- 1. Brink RA. Paramutation. Annu Rev Genet. 1973;7: 129–152. pmid:4361265
- 2. Chandler VL. Paramutation’s properties and puzzles. Science. 2010;330: 628–629. pmid:21030647
- 3. Brink RA. A genetic change associated with the R locus in maize which is directed and potentially reversible. Genetics. 1956;41: 872–889. pmid:17247669
- 4. Coe EH. A regular and continuing conversion-type phenomenon at the B locus in maize. Genetics. 1959;45: 828–832. pmid:16590451
- 5. Das OP, Messing J. Variegated phenotype and developmental methylation changes of a maize allele originating from epimutation. Genetics. 1994;136: 1121–1141. pmid:8005419
- 6. Hollick JB, Patterson GI, Coe EH, Cone KC, Chandler VL. Allelic interactions heritably alter the activity of a metastable maize pl allele. Genetics. 1995;141: 709–719. pmid:8647404
- 7. Pilu R, Panzeri D, Cassani E, Badone FC, Landoni M, Nielsen E. A paramutation phenomenon is involved in the genetics of maize low phytic acid1-241 (lpa1-241) trait. Heredity. 2009;102: 236–245. pmid:18781168
- 8. Hagemann R. [Somatic conversion in Lycopersicon esculentum Mill]. Z Vererbungsl. 1958;89: 587–613.
- 9. Greaves IK, Groszmann M, Wang A, Peacock WJ, Dennis ES. Inheritance of trans chromosomal methylation patterns from Arabidopsis F1 hybrids. Proc Natl Acad Sci USA. 2014;111: 2017–2022. pmid:24449910
- 10. Grandjean V, Gounon P, Wagner N, Martin L, Wagner KD, Bernex F, et al. The miR-124-Sox9 paramutation: RNA-mediated epigenetic control of embryonic and adult growth. Development. 2009;136: 3647–3655. pmid:19820183
- 11. Bennett ST, Wilson AJ, Esposito L, Bouzekri N, Undlien DE, Cucca F, et al. Insulin VNTR allele-specific effect in type 1 diabetes depends on identity of untransmitted paternal allele. Nat Genet. 1997;17: 350–352. pmid:9354805
- 12. Meyer P, Heidmann I, Niedenhof I. Differences in DNA-methylation are associated with a paramutation phenomenon in transgenic petunia. The Plant Journal. 1993;4: 89–100. pmid:8220478
- 13. Matzke MA, Moscone EA, Park YD, Papp I, Oberkofler H, Neuhuber F, et al. Inheritance and expression of a transgene insert in an aneuploid tobacco line. Molecular Genetics and Genomics. 1994;245: 471–185.
- 14. Mittelsten Scheid O, Afsar K, Paszkowski J. Formation of stable epialleles and their paramutation-like interaction in tetraploid Arabidopsis thaliana. Nat Genet. 2003;34: 450–454. pmid:12847525
- 15. Zheng Z, Yu H, Miki D, Jin D, Zhang Q, Ren Z, et al. Involvement of multiple gene-silencing pathways in a paramutation-like phenomenon in Arabidopsis. Cell Rep. 2015;11: 1–8. pmid:25981044
- 16. Herman H, Lu M, Anggraini M, Sikora A, Chang Y, Yoon BJ, et al. Trans allele methylation and paramutation-like effects in mice. Nat Genet. 2003;34: 199–202. pmid:12740578
- 17. Rassoulzadegan M, Grandjean V, Gounon P, Vincent S, Gillot I, Cuzin F. RNA-mediated non-mendelian inheritance of an epigenetic change in the mouse. Nature. 2006;441: 469–474. pmid:16724059
- 18. de Vanssay A, Bougé A-L, Boivin A, Hermant C, Teysset L, Delmarre V, et al. Paramutation in Drosophila linked to emergence of a piRNA-producing locus. Nature. 2012;490: 112–115. pmid:22922650
- 19. Fitz-James MH, Sabarís G, Sarkies P, Bantignies F, Cavalli G. Interchromosomal contacts between regulatory regions trigger stable transgenerational epigenetic inheritance in Drosophila. bioRxiv:2023.07.13.548806v3 [Preprint]. 2023[cited 2024 February 29]. Available from http://www.biorxiv.org/content/doi:10.1101/2023.07.13.548806.
- 20. Shirayama M, Seth M, Lee HC, Gu W, Ishidate T, Conte D, et al. piRNAs initiate an epigenetic memory of nonself RNA in the C. elegans germline. Cell. 2012;150: 65–77. pmid:22738726
- 21. Luteijn MJ, Van Bergeijk P, Kaaij LJT, Almeida MV, Roovers EF, Berezikov E, et al. Extremely stable Piwi-induced gene silencing in Caenorhabditis elegans. EMBO J. 2012;31: 3422–3430. pmid:22850670
- 22. Seth M, Shirayama M, Gu W, Ishidate T, Conte D, Mello CC. The C. elegans CSR-1 argonaute pathway counteracts epigenetic silencing to promote germline gene expression. Dev Cell. 2013;27: 656–663. pmid:24360782
- 23. Sapetschnig A, Sarkies P, Lehrbach NJ, Miska EA. Tertiary siRNAs mediate paramutation in C. elegans. PLoS Genet. 2015;11: e1005078. pmid:25811365
- 24. Shukla A, Perales R, Kennedy S. piRNAs coordinate poly(UG) tailing to prevent aberrant and perpetual gene silencing. Current Biology. 2021;31: 4473–4485. pmid:34428467
- 25. Hollick JB. Paramutation and related phenomena in diverse species. Nat Rev Genet. 2017;18: 5–23. pmid:27748375
- 26. Heard E, Martienssen RA. Transgenerational epigenetic inheritance: Myths and mechanisms. Cell. 2014; 157: 95–109. pmid:24679529
- 27. Ludwig SR, Habera LF, Dellaporta SL, Wessler SR. Lc, a member of the maize R gene family responsible for tissue-specific anthocyanin production, encodes a protein similar to transcriptional activators and contains the myc-homology region. Proc Natl Acad Sci USA. 1989;86: 7092–7096. pmid:2674946
- 28. Dooner HK, Robbins TP, Jorgensen RA. Genetic and developmental control of anthocyanin biosynthesis. Annu Rev Genet. 1991;25: 173–99. pmid:1839877
- 29. Dorweiler JE, Carey CC, Kubo KM, Hollick JB, Kermicle JL, Chandler VL. mediator of paramutation1 is required for establishment and maintenance of paramutation at multiple maize loci. Plant Cell. 2000;12: 2101–2118. pmid:11090212
- 30. Sidorenko L, Dorweiler JE, Cigan AM, Arteaga-Vazquez M, Vyas M, Kermicle J, et al. A dominant mutation in mediator of paramutation2, one of three second-largest subunits of a plant-specific RNA polymerase, disrupts multiple siRNA silencing processes. PLoS Genet. 2009;5: e1000725. pmid:19936058
- 31. Hollick JB, Chandler VL. Genetic factors required to maintain repression of a paramutagenic maize pl1 allele. Genetics. 2001;157: 369–378. pmid:11139517
- 32. Hale CJ, Stonaker JL, Gross SM, Hollick JB. A novel Snf2 protein maintains trans-generational regulatory states established by paramutation in maize. PLoS Biol. 2007;5: 2156–2165. pmid:17941719
- 33. Matzke MA, Kanno T, Matzke AJM. RNA-directed DNA methylation: the evolution of a complex epigenetic pathway in flowering plants. Annu Rev Plant Biol. 2015;66: 243–267. pmid:25494460
- 34. Erhard KF, Stonaker JL, Parkinson SE, Lim JP, Hale CJ, Hollick JB. RNA polymerase IV functions in paramutation in Zea mays. Science. 2009;323: 1201–1205. pmid:19251626
- 35. Sloan AE, Sidorenko L, McGinnis KM. Diverse gene-silencing mechanisms with distinct requirements for RNA polymerase subunits in Zea mays. Genetics. 2014;198: 1031–1042. pmid:25164883
- 36. Stonaker JL, Lim JP, Erhard KF, Hollick JB. Diversity of Pol IV function is defined by mutations at the maize rmr7 locus. PLoS Genet. 2009;5: e1000706. pmid:19936246
- 37. Haag JR, Brower-Toland B, Krieger EK, Sidorenko L, Nicora CD, Norbeck AD, et al. Functional diversification of maize RNA Polymerase IV and V subtypes via alternative catalytic subunits. Cell Rep. 2014;9: 378–390. pmid:25284785
- 38. Alleman M, Sidorenko L, McGinnis K, Seshadri V, Dorweiler JE, White J, et al. An RNA-dependent RNA polymerase is required for paramutation in maize. Nature. 2006;442: 295–298. pmid:16855589
- 39. Narain AS, Liao IT, Talbot J-ERB, Deans NC, Hollick JB. A dicer-like3 protein affects paramutation at multiple loci in Zea mays. bioRxiv:2023.09.29.560153v1 [Preprint]. 2023[cited 2024 February 29]. Available from http://www.biorxiv.org/content/doi:10.1101/2023.09.29.560153.
- 40. Deans NC, Giacopelli BJ, Hollick JB. Locus-specific paramutation in Zea mays is maintained by a PICKLE-like chromodomain helicase DNA-binding 3 protein controlling development and male gametophyte function. PLoS Genet. 2020;16: e1009243. pmid:33320854
- 41. Barbour J-ER, Liao IT, Stonaker JL, Lim JP, Lee CC, Parkinson SE, et al. required to maintain repression2 is a novel protein that facilitates locus-specific paramutation in maize. Plant Cell. 2012;24: 1761–1775. pmid:22562610
- 42. Yang R, Zheng Z, Chen Q, Yang L, Huang H, Miki D, et al. The developmental regulator PKL is required to maintain correct DNA methylation patterns at RNA-directed DNA methylation loci. Genome Biol. 2017;18: 103. pmid:28569170
- 43. Aubert J, Bellegarde F, Oltehua-Lopez O, Leblanc O, Arteaga-Vazquez MA, Martienssen RA, et al. AGO104 is a RdDM effector of paramutation at the maize b1 locus. PLoS One. 2022;17. pmid:36040902
- 44. Sekhon RS, Wang PH, Sidorenko L, Chandler VL, Chopra S. Maize Unstable factor for orange1 is required for maintaining silencing associated with paramutation at the pericarp color1 and booster1 loci. PLoS Genet. 2012;8: e1002980. pmid:23055943
- 45. Wittmeyer K, Cui J, Chatterjee D, Lee T, Tan Q, Xue W, et al. The dominant and poor penetrant phenotypes of the maize mutation Unstable factor for orange1 are caused by DNA methylation changes at a linked transposon. Plant Cell. 2018;30: 3006–3023. pmid:30563848
- 46. Teixeira FK, Colot V. Repeat elements and the Arabidopsis DNA methylation landscape. Heredity. 2010;105: 14–23. pmid:20461104
- 47. Arteaga-Vazquez M, Sidorenko L, Rabanal FA, Shrivistava R, Nobuta K, Green PJ, et al. RNA-mediated trans-communication can establish paramutation at the b1 locus in maize. Proc Natl Acad Sci U S A. 2010;107: 12986–12991. pmid:20616013
- 48. Walker EL. Paramutation of the r1 locus of maize is associated with increased cytosine methylation. Genetics. 1998;148: 1973–1981. pmid:9560410
- 49. Goettel W, Messing J. Paramutagenicity of a p1 epiallele in maize. Theor and Appl Genet. 2013;126: 159–177. pmid:22986680
- 50. Sidorenko L V, Chandler VL, Wang X, Peterson T. Transcribed enhancer sequences are required for maize p1 paramutation. Genetics. 2024;226: iyad178. pmid:38169343
- 51. Haring M, Bader R, Louwers M, Schwabe A, van Driel R, Stam M. The role of DNA methylation, nucleosome occupancy and histone modifications in paramutation. Plant Journal. 2010;63: 366–378. pmid:20444233
- 52. Patterson GI, Thorpe CJ, Chandler VL. Paramutation, an allelic interaction, is associated with a stable and heritable reduction of transcription of the maize b regulatory gene. Genetics. 1993;135: 881–894. pmid:7507455
- 53. Sidorenko L, Chandler V. RNA-dependent RNA polymerase is required for enhancer-mediated transcriptional silencing associated with paramutation at the maize p1 gene. Genetics. 2008;180: 1983–1993. pmid:18845841
- 54. Hollick JB, Patterson GI, Asmundsson IM, Chandler VL. Paramutation alters regulatory control of the maize pl locus. Genetics. 2000;154: 1827–1838. pmid:10747073
- 55. Eggleston WB, Alleman M, Kermicle JL. Molecular organization and germinal instability of R-stippled maize. Genetics. 1995;141: 347–360. pmid:8536982
- 56. Walker EL, Robbins TP, Bureau TE, Kermicle J, Dellaporta SL. Transposon-mediated chromosomal rearrangements and gene duplications in the formation of the maize R-r complex. EMBO J. 1995;14: 2350–2363. pmid:7774593
- 57. Panavas T, Weir J, Walker EL. The structure and paramutagenicity of the R-marbled haplotype of Zea mays. Genetics. 1999;153: 979–991. pmid:10511572
- 58. Sidorenko L, Peterson T. Transgene-induced silencing identifies sequences involved in the establishment of paramutation of the maize p1 gene. Plant Cell. 2001;13: 319–335. pmid:11226188
- 59. Stam M, Belele C, Dorweiler JE, Chandler VL. Differential chromatin structure within a tandem array 100 kb upstream of the maize b1 locus is associated with paramutation. Genes Dev. 2002;16: 1906–1918. pmid:12154122
- 60. Belele CL, Sidorenko L, Stam M, Bader R, Arteaga-Vazquez MA, Chandler VL. Specific tandem repeats are sufficient for paramutation-induced trans-generational silencing. PLoS Genet. 2013;9: e1003773. pmid:24146624
- 61. Kermicle JL, Eggleston WB, Alleman M. Organization of paramutagenicity in R-stippled maize. Genetics. 1995;141: 361–372. pmid:8536983
- 62. Martienssen RA. Maintenance of heterochromatin by RNA interference of tandem repeats. Nat Genet. 2003;35: 213–214. pmid:14593407
- 63. Erhard KF, Parkinson SE, Gross SM, Barbour J-ER, Lim JP, Hollick JB. Maize RNA polymerase IV defines trans-generational epigenetic variation. Plant Cell. 2013;25: 808–819. pmid:23512852
- 64. Cone KC, Cocciolone SM, Moehlenkamp CA, Weber T, Drummond BJ, Tagliani LA, et al. Role of the regulatory gene pl in the photocontrol of maize anthocyanin pigmentation. Plant Cell. 1993;5: 1807–1816. pmid:8305873
- 65. Cocciolone SM, Cone KC. Pl-Bh, an anthocyanin regulatory gene of maize that leads to variegated pigmentation. Genetics. 1993;135: 575–588. pmid:7694886
- 66. Gross SM, Hollick JB. Multiple trans-sensing interactions affect meiotically heritable epigenetic states at the maize pl1locus. Genetics. 2007;176: 829–839. pmid:17435245
- 67. Parkinson SE, Gross SM, Hollick JB. Maize sex determination and abaxial leaf fates are canalized by a factor that maintains repressed epigenetic states. Dev Biol. 2007;308: 462–473. pmid:17612519
- 68. Erhard KF, Talbot J-ERB, Deans NC, McClish AE, Hollick JB. Nascent transcription affected by RNA polymerase IV in Zea mays. Genetics. 2015;199: 1107–1125. pmid:25653306
- 69. Hale CJ, Erhard KF, Lisch D, Hollick JB. Production and processing of siRNA precursor transcripts from the highly repetitive maize genome. PLoS Genet. 2009;5: e1000598. pmid:19680464
- 70. Jiao Y, Peluso P, Shi J, Liang T, Stitzer MC, Wang B, et al. Improved maize reference genome with single-molecule technologies. Nature. 2017;546: 524–527. pmid:28605751
- 71. Kohany O, Gentles AJ, Hankus L, Jurka J. Annotation, submission and screening of repetitive elements in Repbase: RepbaseSubmitter and Censor. BMC Bioinformatics. 2006;7: 474. pmid:17064419
- 72. Liu K, Goodman M, Muse S, Smith JS, Buckler E, Doebley J. Genetic structure and diversity among maize inbred lines as inferred from DNA microsatellites. Genetics. 2003;165: 2117–2128. pmid:14704191
- 73. Hufford MB, Seetharam AS, Woodhouse MR, Chougule KM, Ou S, Liu J, et al. De novo assembly, annotation, and comparative analysis of 26 diverse maize genomes. Science. 2021;373: 655–662. pmid:34353948
- 74. Hollick JB, Kermicle JL, Parkinson SE. Rmr6 maintains meiotic inheritance of paramutant states in Zea mays. Genetics. 2005;171: 725–740. pmid:16020780
- 75. Johnson NR, Yeoh JM, Coruh C, Axtell MJ. Improved placement of multi-mapping small RNAs. G3. 2016;6: 2103–2111. pmid:27175019
- 76. Loffer A, Singh J, Fukudome A, Mishra V, Wang F, Pikaard CS. A DCL3 dicing code within Pol IV-RDR2 transcripts diversifies the siRNA pool guiding RNA-directed DNA methylation. Elife. 2022;11: e73260. pmid:35098919
- 77. Schmitz RJ, Grotewold E, Stam M. Cis-regulatory sequences in plants: Their importance, discovery, and future challenges. Plant Cell. 2022;34: 718–741. pmid:34918159
- 78. Hövel I, Bader R, Louwers M, Haring M, Peek K, Gent JI, et al. RNA-directed DNA methylation mutants reduce histone methylation at the paramutated maize booster1 enhancer. Plant Physiol. 2024;kiae072. pmid:38366582
- 79. Louwers M, Bader R, Haring M, van Driel R, de Laat W, Stam M. Tissue- and expression level-specific chromatin looping at maize b1 epialleles. Plant Cell. 2009;21: 832–842. pmid:19336692
- 80. Shlyueva D, Stampfel G, Stark A. Transcriptional enhancers: From properties to genome-wide predictions. Nat Rev Genet. 2014;15(4): 272–286. pmid:24614317
- 81. Kim TK, Hemberg M, Gray JM. Enhancer RNAs: A class of long noncoding RNAs synthesized at enhancers. Cold Spring Harb Perspect Biol. 2015;7: a018622. pmid:25561718
- 82. Andersson R, Sandelin A. Determinants of enhancer and promoter activities of regulatory elements. Nat Rev Genet. 2020;21: 71–87. pmid:31605096
- 83. Blevins T, Podicheti R, Mishra V, Marasco M, Wang J, Rusch D, et al. Identification of Pol IV and RDR2-dependent precursors of 24 nt siRNAs guiding de novo DNA methylation in Arabidopsis. Elife. 2015;4: e09591. pmid:26430765
- 84. Li S, Vandivier LE, Tu B, Gao L, Won SY, Li S, et al. Detection of Pol IV/RDR2-dependent transcripts at the genomic scale in Arabidopsis reveals features and regulation of siRNA biogenesis. Genome Res. 2015;25: 235–245. pmid:25414514
- 85. Zhai J, Bischof S, Wang H, Feng S, Lee T, Teng C, et al. A one precursor one siRNA model for Pol IV-dependent siRNA biogenesis. Cell. 2015;163: 445–455. pmid:26451488
- 86. Wierzbicki AT, Haag JR, Pikaard CS. Noncoding transcription by RNA polymerase Pol IVb/Pol V mediates transcriptional silencing of overlapping and adjacent genes. Cell. 2008;135: 635–648. pmid:19013275
- 87. Wendte JM, Pikaard CS. The RNAs of RNA-directed DNA methylation. Biochim Biophys Acta Gene Regul Mech. 2017;1860: 140–148. pmid:27521981
- 88. Sigman MJ, Panda K, Kirchner R, McLain LL, Payne H, Peasari JR, et al. An siRNA-guided ARGONAUTE protein directs RNA polymerase V to initiate DNA methylation. Nat Plants. 2021;7: 1461–1474. pmid:34750500
- 89. Flaus A, Martin DMA, Barton GJ, Owen-Hughes T. Identification of multiple distinct Snf2 subfamilies with conserved structural motifs. Nucleic Acids Res. 2006;34: 2887–2905. pmid:16738128
- 90. Ho KK, Zhang H, Golden BL, Ogas J. PICKLE is a CHD subfamily II ATP-dependent chromatin remodeling factor. Biochim Biophys Acta. 2013;1829: 199–210. pmid:23128324
- 91. Bente H, Foerster AM, Lettner N, Mittelsten Scheid O. Polyploidy-associated paramutation in Arabidopsis is determined by small RNAs, temperature, and allele structure. PLoS Genet. 2021;17. pmid:33690630
- 92. Gouil Q, Novák O, Baulcombe DC. SLTAB2 is the paramutated SULFUREA locus in tomato. J Exp Bot. 2016;67: 2655–2664. pmid:26957563
- 93. Groszmann M, Greaves IK, Albertyn ZI, Scofield GN, Peacock WJ, Dennis ES. Changes in 24-nt siRNA levels in Arabidopsis hybrids suggest an epigenetic contribution to hybrid vigor. Proc Natl Acad Sci USA. 2011;108: 2617–2622. pmid:21266545
- 94. Regulski M, Lu Z, Kendall J, Donoghue MTA, Reinders J, Llaca V, et al. The maize methylome influences mRNA splice sites and reveals widespread paramutation-like switches guided by small RNA. Genome Res. 2013;23: 1651–1662. pmid:23739895
- 95. Law JA, Jacobsen SE. Establishing, maintaining and modifying DNA methylation patterns in plants and animals. Nat Rev Genet. 2010;11: 204–220. pmid:20142834
- 96. Zemach A, Kim MY, Hsieh PH, Coleman-Derr D, Eshed-Williams L, Thao K, et al. The Arabidopsis nucleosome remodeler DDM1 allows DNA methyltransferases to access H1-containing heterochromatin. Cell. 2013;153: 193–205. pmid:23540698
- 97. Stroud H, Do T, Du J, Zhong X, Feng S, Johnson L, et al. Non-CG methylation patterns shape the epigenetic landscape in Arabidopsis. Nat Struct Mol Biol. 2014;21: 64–72. pmid:24336224
- 98. Li Q, Eichten SR, Hermanson PJ, Zaunbrecher VM, Song J, Wendt J, et al. Genetic perturbation of the maize methylome. Plant Cell. 2014;26: 4602–4616. pmid:25527708
- 99. Parent JS, Cahn J, Herridge RP, Grimanelli D, Martienssen RA. Small RNAs guide histone methylation in Arabidopsis embryos. Genes Dev. 2021;35: 841–846. pmid:34016690
- 100. Du J, Johnson LM, Groth M, Feng S, Hale CJ, Li S, et al. Mechanism of DNA methylation-directed histone methylation by KRYPTONITE. Mol Cell. 2014;55: 495–504. pmid:25018018
- 101. Zhang H, Ma ZY, Zeng L, Tanaka K, Zhang CJ, Ma J, et al. DTF1 is a core component of RNA-directed DNA methylation and may assist in the recruitment of Pol IV. Proc Natl Acad Sci USA. 2013;110: 8290–8295. pmid:23637343
- 102. Wang Y, Zhou X, Luo J, Lv S, Liu R, Du X, et al. Recognition of H3K9me1 by maize RNA-directed DNA methylation factor SHH2. J Integr Plant Biol. 2021;63: 1091–1096. pmid:33913587
- 103. Ragunathan K, Jih G, Moazed D. Epigenetic inheritance uncoupled from sequence-specific recruitment. Science. 2015;348(6230): 1258699. pmid:25831549
- 104. Martinho C, Wang Z, Ghigi A, Buddle S, Barbour F, Yarur A, et al. CHROMOMETHYLTRANSFERASE3/KRYPTONITE maintains the sulfurea paramutation in Solanum lycopersicum. Proc Natl Acad Sci USA. 2022;119: e2112240119. pmid:35324329
- 105. Aichinger E, Villar CBR, Farrona S, Reyes JC, Hennig L, Köhler C. CHD3 proteins and Polycomb group proteins antagonistically determine cell identity in Arabidopsis. PLoS Genet. 2009;5: e1000605. pmid:19680533
- 106. Zhang H, Rider SD, Henderson JT, Fountain M, Chuang K, Kandachar V, et al. The CHD3 remodeler PICKLE promotes trimethylation of histone H3 lysine 27. Journal of Biological Chemistry. 2008;283: 22637–22648. pmid:18539592
- 107. Carter B, Bishop B, Ho KK, Huang R, Jia W, Zhang H, et al. The chromatin remodelers PKL and PIE1 act in an epigenetic pathway that determines H3K27me3 homeostasis in Arabidopsis. Plant Cell. 2018;30: 1337–1352. pmid:29802212
- 108. Brink RA, Brown DF, Kermicle J, Weyers WH. Locus dependence of the paramutant R phenotype in maize. Genetics. 1960;45: 1297–1312. pmid:17248000
- 109. Roman H. Mitotic nondisjunction in the case of interchanges involving the B-type chromosome in maize. Genetics. 1947;32: 391–409. pmid:17247252
- 110. Hollick JB. Paramutation: a trans-homolog interaction affecting heritable gene regulation. Curr Opin Plant Biol. 2012;15: 536–543. pmid:23017240
- 111. Pilu R, Piazza P, Petroni K, Ronchi A, Martin C, Tonelli C. pl-bol3, a complex allele of the anthocyanin regulatory pl1locus that arose in a naturally occurring maize population. Plant Journal. 2003;36: 510–521. pmid:14617081
- 112. Pilu R. Paramutation phenomena in plants. Semin Cell Dev Biol. 2015;44: 2–10. pmid:26335267
- 113. Coe EH. The properties, origin, and mechanism of conversion-type inheritance at the B locus in maize. Genetics. 1966;53: 1035–1063. pmid:17248307
- 114. Chandler VL, Eggleston WB, Dorweiler JE. Paramutation in maize. Plant Mol Biol. 2000;43: 121–145. pmid:10999400
- 115. Hollick JB, Chandler VL. Epigenetic allelic states of a maize transcriptional regulatory locus exhibit overdominant gene action. Genetics. 1998;150: 891–897. pmid:9755217
- 116. Longley AE. Breakage points for four corn translocation series and other corn chromosome aberrations. USDA-ARS Crop Research Bulletin. 1961;34–16: 1–40.
- 117. Sambrook J, Russell DW. Isolation of BAC DNA from small-scale cultures. Cold Spring Harb Protoc. 2006. pmid:22485431
- 118. Poulsen TS, Johnsen HE. BAC end sequencing. Methods Mol Biol. 2004;255: 157–161. pmid:15020822
- 119. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, et al. Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics. 2012;28: 1647–1649. pmid:22543367
- 120. Lyons E, Freeling M. How to usefully compare homologous plant genes and chromosomes as DNA sequences. Plant Journal. 2008;53: 661–673. pmid:18269575
- 121. Hövel I, Louwers M, Stam M. 3C Technologies in plants. Methods. 2012;58: 204–211. pmid:22728034
- 122. Weber B, Jamge S, Stam M. 3C in Maize and Arabidopsis. Methods in Molecular Biology. 2018;1675: 247–270. pmid:29052196
- 123. Splinter E, de Wit E, van de Werken HJG, Klous P, de Laat W. Determining long-range chromatin interactions for selected genomic sites using 4C-seq technology: From fixation to computation. Methods. 2012;58: 221–230. pmid:22609568
- 124. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009;10: R25. pmid:19261174
- 125. Quinlan AR, Hall IM. BEDTools: A flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26: 841–842. pmid:20110278
- 126. Labonne JDJ, Dorweiler JE, McGinnis KM. Changes in nucleosome position at transcriptional start sites of specific genes in Zea mays mediator of paramutation1 mutants. Epigenetics. 2013;8: 398–408. pmid:23538550
- 127. Thompson BE, Basham C, Hammond R, Ding Q, Kakrana A, Lee T-F, et al. The dicer-like1 homolog fuzzy tassel is required for the regulation of meristem determinacy in the inflorescence and vegetative growth in maize. Plant Cell. 2014;26: 4702–4717. pmid:25465405
- 128. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25: 2078–2079. pmid:19505943
- 129. Gruntman E, Qi Y, Slotkin RK, Roeder T, Martienssen RA, Sachidanandam R. Kismeth: Analyzer of plant methylation states through bisulfite sequencing. BMC Bioinformatics. 2008;9. pmid:18786255