Abstract

Materials and Methods

RT-PCR.

cDNA was synthesized from S. pombe RNA preparations using Reverse Transcriptase (BRL) according to the manufacturer's instructions. cDNA samples were analyzed using ade6⁺, ura4⁺, hsp16⁺, or SPBC19C7.04c gene primers in multiplex or competitive PCR strategies (see legend of Fig. ​Fig.44).

Open in a separate window

Fig 4.

ORF microarray expression profiling analysis of dcr1Δ cells. (A) The Venn diagram shows the number of genes whose expression was increased >1.5-fold in two independent experiments (red and green circles; see Materials and Methods). The yellow intersection shows the set of 10 reproducibly identified genes of which two (hsp16⁺ and SPBC19C7.04c) were increased >3 fold. The complete data sets from the ORF microarray expression analyses are available at www.ki.se/biovetenskap/karl. The green circle corresponds to microarray ratios 1, and the red circle corresponds to microarray ratios 2. (B) The scatter plot shows nonprocessed hybridization intensity data from dcr1Δ and WT samples plotted against each other for the gene sets identified in A. The filled yellow points show the reproducibly identified genes, and points representing hsp16⁺ and SPBC19C7.04c are shown by the white arrows. The diagonal green lines represent 2-fold changes. (C) RT-PCR analysis of the up-regulated genes, hsp16⁺ and SPBC19C7.04c. The top band in each PCR reaction represents the ura4⁺ control gene and the bottom band represents hsp16⁺ or SPBC19C7.04c genes, as indicated. The fold increase of gene expression is indicated below each lane. For A–C, the strains used were Hu679 (Δ) and Hu303 (wt).

Results and Discussion

Database searches with the human Dicer sequence, which we initially found in a yeast two-hybrid screening using 5-lipoxygenase as a bait (25), identified homologous sequences in a variety of organisms, except the budding yeast Saccharomyces cerevisiae. Notably, the recently sequenced genome of the fission yeast S. pombe (26) contains a homologous gene. The gene SPCC584.10c (SWISS-PROT accession no. Q09884), here designated dcr1⁺, is 4,125 bp long and encodes a protein of 1,374 aa with an N-terminal helicase domain containing an ATP-binding site motif ³²MRTGAGKT³⁹ and a ¹⁴⁵DECH¹⁴⁸ box, and tandem C-terminal ribonuclease (RNase) III signatures [¹¹²⁰QQLEFLGDA¹¹²⁸ (RIIIb) and ⁹³⁰DRLEFYGDC⁹³⁸ (RIIIa)] (divergent residues are underlined). No other ORF sequences combining these putative functional domains could be identified in the S. pombe genome, unlike the situation in Drosophila, which has at least two Dicer genes (Dcr-1 and Dcr-2) (4). Alignment of the multiple Dicer sequences revealed a relatively high sequence similarity (see Fig. 5, which is published as supporting information on the PNAS web site, www.pnas.org). In this alignment, conservation of the ATP-binding site, the DECH box, and the second RNase III motif (RIIIb) is noteworthy, whereas the first RNase III motif (RIIIa) in dcr1⁺ is slightly shifted toward the C terminus, as compared with the same motif in the mammalian proteins. The domain structure of dcr1⁺ also differs from the other Dicer proteins in that it appears not to contain a Piwi/Argonaute/Zwille (PAZ) domain (27). This domain, whose exact function remains unknown, is also present in Drosophila Argonaute (28).

To characterize the function of dcr1⁺, a dcr1Δ strain was created (see Table ​Table1).1). To document the effect of Dicer deletion on growth, colony growth assays were performed at 20, 25, 30, or 36°C (Fig. ​(Fig.11B). Growth assays in liquid cultures were also performed at 25°C (Fig. ​(Fig.11C). The dcr1Δ strain exhibited a temperature-sensitive phenotype, as compared with the WT strain; we observed a slight impairment of cell growth at 30 and 36°C, but a more severe growth defect at lower temperatures (20 and 25°C). The temperature-sensitive slow growth phenotype of dcr1Δ cells is suggestive of a possible defect in mitosis or other cell division-related phenomena. To investigate whether chromosome segregation was affected in dcr1Δ cells, their sensitivity to TBZ was investigated. dcr1Δ cells displayed a marked supersensitivity to TBZ, not growing at low concentrations of the drug that were well tolerated by WT cells (Fig. ​(Fig.11D, rows 1 and 4). Next, functional complementation of the slow growth and TBZ sensitivity phenotypes of dcr1Δ cells was investigated. We used plasmid constructs, in which the S. pombe dcr1⁺ or human Dicer genes were inserted downstream of the nmt1 promoter (see Materials and Methods). The dcr1Δ strain was transformed, and the transformants were grown on selective media at 25, 30, or 36°C. As shown in Fig. ​Fig.11 B and D, plasmid expression of dcr1⁺ rescued the growth defect and TBZ sensitivity of the dcr1Δ cells. Importantly, a partial functional rescue of dcr1Δ cells was achieved by episomal expression of human Dicer (see Fig. ​Fig.11D, row 3), indicating that the function of this new class of proteins has been conserved during evolution.

Open in a separate window

Fig 1.

Domain structure of Dcr1p, documentation of the temperature-sensitive, slow-growth sensitivity to TBZ phenotypes of the dcr1Δ strain, and functional complementation by human Dicer. (A) Predicted domain structure of Dcr1p deduced from the amino acid sequence of Dicer (SWISS-PROT accession no. Q09884). (B) Growth of dcr1Δ (Δ) and WT (wt) cells transformed with a vector expressing S. pombe Dcr1, or empty vector only, at 20, 25, 30, or 36°C. Cells were serially diluted at 5-fold dilutions and plated on minimal medium plates. Plates were photographed after 4 days of growth. Strains were Hu655 (Δ) and Hu111 (wt). (C) Growth of dcr1Δ or WT cells in liquid YES [supplemented yeast extract (rich)] medium at 25°C. Cultures were incubated with shaking at 25°C. Cell density (OD₆₀₀) was monitored. (D) Growth of WT (wt) and dcr1Δ (Δ) cells transformed with a vector expressing S. pombe Dcr1 (S.p. dcr1⁺), human Dicer (H.s. Dicer), or empty vector only, plated on minimal medium plates containing 20 μM thiamine and 0 or 15 μg/ml TBZ as indicated. Strains were Hu655 (Δ) and Hu111 (wt). Cells were serially diluted at 5-fold dilutions. Plates were photographed after 4 days of growth at 30°C.

To visualize the mitotic defect of dcr1Δ cells, anti-tubulin immunofluorescence (IF) microscopy was performed. Clearly, dcr1Δ cells displayed a high incidence of lagging chromosomes on anaphase spindles (Fig. ​(Fig.22A), and this defect was more pronounced at 25°C (33%, n = 100) than at 30°C (16% lagging chromosomes, n = 100). For WT cells, no lagging chromosomes were observed. This phenotype is generally indicative of a defective centromeric heterochromatin function, such as in swi6Δ cells (24), and has been shown to be due to loss of centromeric cohesion (29). To test whether formation of centromeric heterochromatin depends on dcr1⁺, the transcription of an inserted ade6⁺ marker gene [cen1-otrR(SphI):ade6⁺] was measured by RT-PCR (Fig. ​(Fig.22B). We observed that ade6⁺ was transcribed from the centromere in dcr1Δ cells, but not in WT control cells, indicating that heterochromatin was disrupted in dcr1Δ cells. In Drosophila, multiple integrated copies of transgenes produce transcripts that are silenced posttranscriptionally by RNAi (30). To test whether the S. pombe centromeric multiple copy repeats themselves also produce transcripts that may be degraded by Dcr1p, we carried out RT-PCR analysis of the noncoding centromere otr-dg repeats in WT and dcr1Δ cells. Our RT-PCR analysis showed that otr-dg transcripts were not detectable in RNA preparations from WT cells, but accumulated to high levels in dcr1Δ cells (Fig. ​(Fig.22C). Our RT-PCR analysis of the integrated ade6⁺ marker gene centromere repeats are in agreement with the recently reported Northern and RT-PCR analysis of integrated ura4⁺ marker gene and centromeric repeats (15). Also recently, small RNAs corresponding to S. pombe centromeric heterochromatic repeats were detected (16). The accumulation of long RNA molecules corresponding to heterochromatic repeats that we observed in dcr1Δ cells indicates that Dcr1p is normally required for posttranscriptional degradation of these long RNAs via dsRNA substrates and, thus, Dcr1p should be capable of degrading such dsRNA substrates in vitro.

Open in a separate window

Fig 2.

Lagging chromosomes, centromere silencing defects, and transcription of noncoding otr-dg centromeric repeats in dcr1Δ cells. (A) Cells grown at 25°C were stained with TAT1 (α-tubulin, green) and 4′,6-diamidino-2-phenylindole (DAPI; DNA, blue). (Left) A normal anaphase wt cell. (Right) A lagging chromosome in a dcr1Δ cell. Strains were Hu655 (Δ) and Hu111 (wt). (B) Silencing defects of dcr1Δ. (Upper) Strategy for competitive RT-PCR. The top band (707 bp) represents the ade6⁺ gene inserted in centromere [cen1-otrR(SphI): ade6⁺] and the bottom band (555 bp) represents ade6⁺-DN/N allele at the endogenous ade6⁺ locus amplified using the same set of primers. Strains were Hu913 (Δ) and Fy2002 (wt). (C) Accumulation of transcribed centromeric repeats in dcr1Δ cells. (Upper) Strategy for RT-PCR. The position of the primers are indicated on the schematic representation of cen1. Strains were Hu679 (Δ) and Hu303 (wt).

To test the enzymatic activity of the Dcr1 protein in vitro, it was tagged with the HA epitope (see Materials and Methods). The Dcr1-HA-tagged protein was fully functional, as judged by growth and TBZ assays (unpublished data). Dcr1-HA could be immunoprecipitated from S. pombe cell homogenates by using anti-HA antibody, as shown by immunoblot analysis (Fig. ​(Fig.33A), and these immunoprecipitates exhibited dsRNase activity (Fig. ​(Fig.33 B and C). In this assay, an ≈23-nt band was produced by Dcr1-HA immunoprecipitates when incubated with the human let-7 microRNA precursor (Fig. ​(Fig.33B, lane 3), but not by immunoprecipitates prepared from dcr1Δ cells (Fig. ​(Fig.33B, lane 4). Similar results were obtained when using a 500-bp (Fig. ​(Fig.33C) dsRNA substrate. Thus, the S. pombe Dicer enzyme is capable of processing dsRNA in vitro.

Open in a separate window

Fig 3.

Enzyme activity of the Dcr1 protein. The dcr1Δ and Dcr1-HA-tagged strains were grown and harvested, and the epitope-tagged Dcr1-HA protein was immunoprecipitated with anti-HA antibody (20 μg; clone 12CA5, Roche) and Protein A-Agarose (Roche). (A) Dcr1-HA protein expression. After several washes, the protein was eluted from the beads by boiling in loading buffer, and analyzed by immunoblotting using anti-HA antibody. (B–C) Detection of dsRNase activity. Twenty-three-nucleotide degradation products are indicated by arrows. ³²P-labeled dsRNA substrates were prepared, as described in refs. 20 and 36. After several washes, the beads were incubated with ³²P-labeled (B) 103-nt human let-7 microRNA precursor substrate (37) or (C) 500-bp Coactosin-like Protein (CLP) (nucleotides 150–649; accession no. L54057) dsRNA, in 10 μl of assay buffer (20 mM TrisHCl/5 mM MgCl₂/1 mM DTT/1 mM ATP/5% SuperaseIn, pH 7.5) for 1 h (A and B) or 6 h (A–C) at 30°C. E, empty Protein A-Agarose beads; M indicates a 10-nt RNA size marker (Ambion, Austin, TX) prepared with [γ-³²P]ATP. Reactions were stopped on ice, and an equal volume of gel loading buffer was added. After heating at 85°C for 5 min, the samples were analyzed by electrophoresis on a 10% polyacrylamide/7M urea gel run in Tris-borate-EDTA (TBE) buffer. The gel was fixed and dried, and the ³²P-labeled RNA products were detected by autoradiography. The 20- to 30-nt region is shown, with the full-length gel on the right.

To investigate whether Dcr1p is also required for silencing of genes in euchromatin, we have performed an initial exploration of the genes whose expression may be regulated, directly or indirectly, by Dcr1p, using a microarray expression profiling approach (Y. Xue, S. Haas, T. Driss, D. Marechal, M. Vingron, M., K.E., and A.W., unpublished work). Gene expression in the dcr1Δ cells was assessed and compared with that of the WT strain (Fig. ​(Fig.44A). Two genes, hsp16⁺ and SPBC19C7.04c, were found to be consistently up-regulated by >3-fold in dcr1Δ cells as compared with WT cells at 25°C (P < 0.05, t test). To confirm the up-regulation of these two genes, an RT-PCR experiment was carried out (Fig. ​(Fig.44C). This experiment showed that hsp16⁺ and SPBC19C7.04c genes were up-regulated 3.6- and 6.2-fold, respectively, in dcr1Δ cells. Interestingly, DNA matrix comparisons identified a common denominator to these two genes. Thus, an 11-nt DNA sequence (5′-GAGGACGTTCA-3′) is present in four locations in the entire S. pombe genome: in opposite directions in the ORFs of hsp16⁺ and SPBC19C7.04c, in one noncoding region (SPAC26H5), and in the gene encoding a putative U3 snoRNA component, which, in other organisms, is known to be required for rRNA processing (31). The presence of this conserved motif in ORFs and the fact that both genes containing this motif were strongly up-regulated in dcr1Δ cells are suggestive of a mechanism in which Dcr1p acts posttranscriptionally to degrade mRNA recognized by this motif.

We have shown that dcr1Δ cells display lagging chromosomes during anaphase, supporting a new role for Dicer in chromosome segregation and centromere silencing to maintain centromeric heterochromatin. Very recently, Dcr1, Ago1 (Argonaute), and Rdp1 (RNA-dependent RNA polymerase) were shown to be important for heterochromatin formation and centromere silencing by promoting H3 lysine 9 methylation in fission yeast (15). In addition, small RNA corresponding to S. pombe centromeric heterochromatin have been detected (16). These findings indicate that S. pombe is a suitable model for understanding biological roles for RNAi. Interestingly, an Argonaute homologue was isolated in a genetic screen as csp mutants (centromere suppressor of position effect) that alleviate centromeric silencing and cause defects in chromosome segregation very similar to those we report for dcr1Δ in this study (ref. 32; R. Allshire, personal communication). Thus, both Dicer and components of the RNA-induced silencing complex (RISC) are implicated in S. pombe centromere silencing, chromosome segregation, and centromere integrity. Examples of phenotypic abnormalities after inactivation of Dicer genes in other species are the unregulated cell division in Arabidopsis floral meristems (33) and sterility in C. elegans (6, 8, 34). We speculate that, as in other species, aberrant small RNA formation underlies the effects of dcr1⁺ inactivation in S. pombe, and that a functional Dicer is required for normal chromosome segregation and centromere integrity in other eukaryotes.

Supplementary Material

Acknowledgments

Abbreviations

References