RNA-mediated trans-communication can establish paramutation at the b1 locus in maize

Abstract

Paramutation is the epigenetic transfer of information between alleles that leads to the heritable change of expression of one allele. Paramutation at the b1 locus in maize requires seven noncoding tandem repeat (b1TR) sequences located approximately 100 kb upstream of the transcription start site of b1, and mutations in several genes required for paramutation implicate an RNA-mediated mechanism. The mediator of paramutation (mop1) gene, which encodes a protein closely related to RNA-dependent RNA polymerases, is absolutely required for paramutation. Herein, we investigate the potential function of mop1 and the siRNAs that are produced from the b1TR sequences. Production of siRNAs from the b1TR sequences depends on a functional mop1 gene, but transcription of the repeats is not dependent on mop1. Further nuclear transcription assays suggest that the b1TR sequences are likely transcribed predominantly by RNA polymerase II. To address whether production of b1TR-siRNAs correlated with paramutation, we examined siRNA production in alleles that cannot undergo paramutation. Alleles that cannot participate in paramutation also produce b1TR-siRNAs, suggesting that b1TR-siRNAs are not sufficient for paramutation in the tissues analyzed. However, when b1TR-siRNAs are produced from a transgene expressing a hairpin RNA, b1 paramutation can be recapitulated. We hypothesize that either the b1TR-siRNAs or the dsRNA template mediates the trans-communication between the alleles that establishes paramutation. In addition, we uncovered a role for mop1 in the biogenesis of a subset of microRNAs (miRNAs) and show that it functions at the level of production of the primary miRNA transcripts.

Fig. 1.

Transcription from the b1TRs is not reduced in mop1-1 but is reduced by inhibitors of DNA-dependent RNA polymerases. (A) Map of the RNA probes used for nuclear run-on analyses. Open arrows depict parts of the sixth and seventh b1TRs required for paramutation. The black box indicates the sequence immediately downstream of the repeats. No transcription was detected upstream of the repeats (4); thus, that region was not tested in these experiments. Black paired arrows below the repeats indicate forward and reverse RNA probes used in relation to this drawing. The location of the four LNA probes used for Northern blot analysis (Fig. 2B) is indicated with four lines below repeat 7. (B) The b1 repeat transcription in B′ immature ears that are WT or homozygous for mop1-1. The results for the genotypes are indicated with open (WT) or solid (mop1-1) histograms for each forward (F) or reverse (R) probe, For each of the three biological replicates, raw counts were normalized to the Ubiquitin2 probe; SD is shown as bars within each histogram. (C) Transcription results after treatment with actinomycin D, a drug that inhibits all DNA-templated RNA synthesis. B′ and B′ mop1-1 samples not treated (no inhibitor) and B′ mop1-1 samples treated with 50 and 100 μg/mL actinomycin D are shown. The percent transcription from inhibitor-treated relative to no-inhibitor–treated control is indicated above each group of histograms. (D) Transcription results after treatment with α-amanitin, a drug that most strongly inhibits Pol-II transcription. The percent transcription from inhibitor-treated relative to no-inhibitor samples is indicated above each pair of histograms. For both C and D, young sheaths were used and transcription is shown for probes with the strongest signals in untreated samples. Transcription of control genes, Ubiquitin2 (transcribed by Pol-II) and 18S (transcribed by Pol-I), is shown separately to accommodate their high incorporation rates. The significance of the differences between control and treated samples was tested using an exact binomial probability calculation with the null hypothesis that drug treatments do not affect transcription and the alternative hypothesis that treatments reduce transcription (49). With either actinomycin treatment (P = 0.002) or α-amanitin treatment (P = 0.002), nine of nine probes demonstrated reduced transcription with drug treatment.

Fig. 2.

siRNAs associated with the b1TRs that mediate paramutation. (A) siRNAs matching the b1TRs (b1TR-siRNAs) from small RNA libraries. The seven tandem repeats found in B-I and B′ are depicted as open arrows relative to the b1 coding region (located ∼100 kb downstream of the repeats). Repeat 1 is expanded to show the location of the b1TR-siRNAs identified from deep sequencing [ref. ; Gene Expression Omnibus (GEO) database accession nos. GSM306487 and GSM306488]. For simplicity, the mapping is shown only for the first repeat, with details of where each siRNA mapped summarized in Table S1. A dotted line inside the repeat unit indicates the AT-rich (72% AT-rich) region within the 3′-end of the tandem repeats. This region is not drawn to scale because no siRNAs were found in the libraries that matched the AT-rich region. The siRNAs are shown as bars above or below the repeat unit, representing the strand of DNA to which they match. Different colors represent different siRNA sizes, as indicated in the key. Two siRNAs observed only in the B′ mop1-1 libraries are shown as dotted bars. The siRNA observed in both libraries is labeled with an asterisk. All other siRNAs were observed only in the WT library. Locations of LNA probes VC1657, VC1658, VC1659, and VCFB (Materials and Methods) used for the Northern blot analysis (B) are shown below the repeat. (B) Northern blot analysis using 100 μg of small RNA-enriched fractions from immature ears. The genotypes are B-I and B′, each of which has seven tandem repeats and participates in paramutation (B-I is highly transcribed, and B′ has very low transcription) (50); B′ mop1-1, which does not participate in paramutation because of the mop1-1 mutation (5, 20); and the recessive b allele, which has a single copy of the repeat unit and does not participate in paramutation (21). Each b1TR probe used for hybridization is indicated above each panel. rRNA is shown as a loading control. Ethidium bromide staining of the small RNA-enriched fraction, which monitors global siRNA levels, is shown at the bottom of each panel.

Fig. 3.

Transgene-generated b1TR-siRNAs recapitulate key features of paramutation. (A) Diagram of the 35S:b1IR construct harboring a single tandem repeat unit (arrows) cloned as an inverted repeat that produces b1TR-siRNAs. (B and C) Crossing scheme used to test paramutation effects of 35S:b1IR on B-I. Progeny classes are diagrammed with the informative classes in bold, and the results are summarized below for each progeny class. (B) Test for establishment of paramutation. All B-I plants that inherited the transgene were silenced, as evidenced by reduced pigment levels. The phenotypes are illustrated in Fig. S1. We indicate this silent state as B′*. No spontaneous paramutation was observed in the nontransgenic B-I siblings. (C) Test for heritability of the B′* state and its ability to induce paramutation in the absence of the transgene. Only nontransgenic siblings are shown. (D) Northern blot to detect b1TR-siRNAs in three each 35S:b1IR transgenic and control nontransgenic siblings. Fifty micrograms of the small RNA-enriched fraction was probed with VC1658 (forward). Half of the amount of RNA and a 10-fold reduction of exposure time (12 h vs. 5 d) were used relative to results in Fig. 2B. rRNA is shown as a loading control.

Fig. 4.

The mop1-1 mutation influences biogenesis of certain miRNAs. (A and B) Levels of each indicated miRNA family were monitored by Northern blot analysis using 5′-end–labeled DNA oligonucleotides complementary to the mature miRNA and 20 μg of the small RNA fraction from B′ immature ears that were WT (wt) or homozygous for the mop1-1 (mop1) mutation. Levels of rRNA and U6 are shown as loading controls. The numbers indicate the mean abundance and SD of miRNAs in mop1-1 relative to the WT control after scanning and normalization for loading for three biological replicates. (C) Levels of pri-miRNAs were reduced in the mop1-1 mutant. RT-PCR analysis of pri-miR156 levels in WT and mop1-1. Total RNA was extracted from immature ears. Analysis of actin served as a loading control showing that equivalent amounts of RNA were tested in all reactions. The numbers indicate the mean abundance and SD of pri-miRNAs in mop1-1 relative to the WT control from three biological replicates.