Analysis of small RNA in fission yeast; centromeric siRNAs are potentially generated through a structured RNA

Abstract

The formation of heterochromatin at the centromeres in fission yeast depends on transcription of the outer repeats. These transcripts are processed into siRNAs that target homologous loci for heterochromatin formation. Here, high throughput sequencing of small RNA provides a comprehensive analysis of centromere-derived small RNAs. We found that the centromeric small RNAs are Dcr1 dependent, carry 5'-monophosphates and are associated with Ago1. The majority of centromeric small RNAs originate from two remarkably well-conserved sequences that are present in all centromeres. The high degree of similarity suggests that this non-coding sequence in itself may be of importance. Consistent with this, secondary structure-probing experiments indicate that this centromeric RNA is partially double-stranded and is processed by Dicer in vitro. We further demonstrate the existence of small centromeric RNA in rdp1Delta cells. Our data suggest a pathway for siRNA generation that is distinct from the well-documented model involving RITS/RDRC. We propose that primary transcripts fold into hairpin-like structures that may be processed by Dcr1 into siRNAs, and that these siRNAs may initiate heterochromatin formation independent of RDRC activity.

Figure 1

Characteristics of the small RNA populations of wild-type and rpb7G150D mutant cells in comparison to Ago1-associated siRNAs (Buhler et al, 2008). Small RNAs without a perfect match in the S. pombe genome or those matching tRNA and rRNA were removed. (A) Start nucleotide distribution of small RNA. The 5′ nucleotide on the x-axis and the relative abundance of small RNAs on the y-axis in wild-type and rpb7G150D mutant cells compared with Ago1-associated siRNAs. (B) Size distribution of small RNA, length in nucleotides on the x-axis and relative abundance on the y-axis in wild-type, rpb7G150D mutant cells and Ago1-associated siRNAs. (C) Percentages of genomic distributions of small RNA reads in wild-type cells, rpb7G150D mutant cells and Ago1-associated siRNAs.

Figure 2

Distribution of small RNA of wild-type cells at the centromeres. (A) Schematic on-scale representation of the centromeres of chromosome I, II and III based on the S. pombe GeneDB. Vertical black arrows represent tRNAs with standard one letter abbreviations representing amino-acid specificity. The green bars represent the KpnI restriction fragment that was shown to be necessary for centromere function (Baum et al, 1994). Red horizontal arrows represent RevCen; orange horizontal arrows represent the region within the dh element with a translocation similar to RevCen. A 700-bp region is deleted from the dh cluster at the left arm of centromere I and the sequence in between dg and dh siRNA clusters is deleted in centromere III. Each arrow represents a sequenced, small RNA match; thicker arrows indicate multiple sequenced small RNAs. Owing to the high degree of sequence similarity, most sequenced centromeric small RNA have perfect match to several clusters within the centromeres. These sequences have been plotted at each perfect match, that is, more than once per sequence. (B) Histogram of small RNA strand distribution at the otr2R-dh and otrR2-dg siRNA clusters. The x-axis depicts ratio of forward to reverse strand siRNAs (log₁₀ scale). All bars are significantly different from 1; ^***P<0.001; ^**P<0.01. (C) Representation of typical centromeric siRNA clusters, here exemplified by the clusters at otr2R-dh and otr2L-dg. Matching sequenced small RNAs are represented by arrows according to their position, orientation and length; pink=14–19 bases, red=20–21 bases, green=22–23 bases, blue=24–25 bases, and grey >25 bases. Hotspots of siRNAs have been named in roman numerals, the most abundant siRNA, ‘IV', was sequenced 400 times and has been cropped. The red horizontal arrow depicts the location of the RevCen sequence and roman numerals in red indicate the siRNA hotspots that match to both dg and dh siRNA clusters.

Figure 3

Validation and analyses of siRNAs by northern blots. (A) Single oligonucleotide probes, antisense, sense, or nearby to siRNAs IV, XII, XXII and VI were used for detection of small RNAs from wild-type or dcr1Δ cells. SnoRNA 58 was used as loading control. (B) and (C) Analyses of 5′-termini of small RNAs by enzymatic reactions followed by northern blots. (B) Terminator exonuclease digests monophosphorylated 5′-termini and (C) guanylyltransferase (GTase) caps di- or triphosphorylated 5′-termini. The control oligonucleotides are 5′-triphosphorylated RNA oligonucleotide (PPP—) and 5′-monophosphorylated RNA oligonucleotide with a blocked 3′-end (P—X) (Ule et al, 2005). The blots were probed sequentially with specific oligonucleotide probes and with a random-primed probe spanning the whole siRNA cluster (dh +dg siRNAs). (D) Analysis of Ago1-purified siRNA 5′-ends by enzymatic reactions followed by northern blot as described above.

Figure 4

The RevCen fragment. (A) Representation of siRNA from wild type (WT) and rpb7G150D cells at otr2L-dg according to their position and orientation. (B) Histogram of siRNA size distributions from WT and rpb7G150D cells, siRNA length in nucleotides on x-axis, and percentage of number of siRNA on y-axis. (C) Sequence alignment of the RevCen promoter region characterized in Djupedal et al. (2005) in the dg repeats from all centromeres. Coordinates are: otr1L-dg from 3752406, otr1R-dg from 3778228, otr2L-dg from 1604242, otr3L-dg from 1073094, and otr3R-dg from 1137620. The region is not present in otr2R-dg. The arrow indicates the direction of transcription. (D) Sequence alignment showing the 340-nt translocation present in all dh and dg elements from the three centromeres, one dh and dg repeat per chromosome arm is shown as they are identical. The siRNAs VII and VIII are shown in boxes with orientation. Coordinates are: otr1R-dh from 3781697, otr2L-dh from 1611493, otr2R-dh from 1636391, otr3L-dh from 1088967, otr3R-dh from 1109980, otr1L-dg from 3763484, otr1R-dg from 3784652, otr2L-dg from 1604792, otr3L-dg from 1082192, and otr3R-dg from 1116691. The region is not found in otr1L-dh and otr2R-dh.

Figure 5

Secondary structure determination of RevCen. (A) Secondary structure probing of in vitro transcribed, [γ-³²P]ATP 5′-end-labelled RevCen RNA analysed on polyacrylamide gels. Partial RNA cleavages were performed as described in Materials and methods section. An OH ladder and a T1 ladder were used to assess cleavage positions. The positions of G residues are marked on the right. (B) Refined prediction of the RevCen secondary structure using the Mfold software using constraints from structural probing data in (A). The siRNAs VII and VIII are indicated in light and dark orange. The grey shade indicates the part of RevCen, which could not be resolved. Nucleotides are highlighted with different colours to show probe-dependent cleavage, as indicated in the colour key. (C) Cleavage of RevCen by human Dicer in vitro. RNA fragments ranging from 21–33 nt were formed when treating RevCen with recombinant human Dicer. RNA was labeled after incubation. The increasing dose of Dicer is indicated. Incubation times were as follows: 5 min for lanes 1 and 4; 30 min for lanes 2 and 5; 2 h for lanes 3, 6 and 7. Concentrations of Dicer: lane 7: 0 units; lanes 1–3: 0.1 units and lanes 4–6: 1.0 units.

Figure 6

Reduction of centromeric siRNAs in rdp1Δ cells compared with wild type (WT), here displayed at otr2L-dg. Representation of the RevCen stem loop, in sense orientation, containing the siRNAs VII and VIII. Matching sequenced siRNAs in wild type and in rdp1Δ are represented by arrows according to their position, orientation, and length; pink=14–19 bases, red=20–21 bases, green=22–23 bases, blue=24–25 bases, and grey >25 bases. Each arrow represents a region with a matching siRNA; thicker arrows indicate regions with multiple siRNAs on a log scale.

Figure 7

Detection of centromeric siRNAs in H3K9R cells as compared with isogenic wild type at cen I otr1L-dg. Each arrow represents a sequenced RNA, colour code as in Figure 2C. Position within chromosome on x-axis is given in kb. Location of RevCen is marked with red arrow.