ARTICLE

TOP ARTICLE LITERATURE CITED

Germinating Ceratopteris richardii spores have proven to be a useful single-cell system for analyzing light and gravity regulation of development (Cooke et al., 1995; Banks, 1999; Chatterjee and Roux, 2000; Murata and Sugai, 2000), but the lack of a gene sequence database and of a reliable way to suppress the expression of specific genes in these cells have made molecular genetic studies difficult. This report describes the production of an expressed sequence tag (EST) library for C. richardii and the use of that information in an RNA interference (RNAi) approach to suppress expression of target genes during spore germination and development.

The data and discussion presented here expand on the recent demonstration of RNAi in spores of the fern Marsilea vestita (Klink and Wolniak, 2000) by detailing the sequence dependence, dose dependence, and specificity of this phenomenon in C. richardii. It also provides original findings on the effectiveness of single-stranded RNA in suppressing the expression of specific genes. The EST and RNAi information provided greatly increase the utility and value of C. richardii for studying an array of basic phenomena in single-cell spores including photomorphogenesis, gravity-directed polarity development, and sex determination and development.

The germination of C. richardii spores is photoreversibly induced by red light (Cooke et al., 1987), and subsequent polarity development is oriented by the vector of gravity during a limited window between 6 to 18 h after the initiation of germination (Edwards and Roux, 1994). Several events indicating polarization can be distinguished during early spore development. A polar calcium current, directed by gravity, peaks during the period of axis fixation. Disruption of this current with calcium channel antagonists reduces the ability of the spore to orient its polar axis parallel to the vector of gravity (Chatterjee et al., 2000). The calcium current is followed by downward nuclear migration. This repositioning of the nucleus sets up an asymmetric cell division, creating a bottom cell that emerges as the rhizoid, and an upper cell that develops into the gametophyte thallus (Edwards and Roux, 1998).

To facilitate an analysis of the genetic machinery needed for polarity development in C. richardii spores and to create a large source of sequence information, we conducted a cDNA sequencing project and deposited the resulting ESTs in GenBank. RNA isolated from spores 20 h after light initiation of germination (approximately 24 h before the first cell division) was used for a commercially prepared cDNA library (Invitrogen, Carlsbad, CA). Randomly chosen clones from this library were sequenced, and 3,587 of the resulting single-pass sequences were submitted to dbEST (GenBank accession nos. BE640669-BE643506 and BQ086920-BQ087668). The average read length of the submitted, vector-trimmed sequences is 748 bases.

The identities of the cDNAs in this collection were determined using BLASTX (http://www.ncbi.nlm.nih.gov/BLAST/; Altschul et al., 1997) against the Arabidopsis proteome (http://www.Arabidopsis.org). Nearly 70% of the ESTs have significant homology to Arabidopsis entries, and more than 33% of the EST annotations in the C. richardii collection are for genes that are represented more than once (Table I). Given this level of redundancy, we estimate that the set of unique ESTs represents approximately 20% to 25% of total mRNA population at this stage of development. Current annotations for the C. richardii EST collection can be found on-line (http://www.esb.utexas.edu/roux/).

Because of the importance of calcium signaling to normal polarity development in C. richardii, we selected ESTs corresponding to five calcium signaling genes, calmodulin (CaM), CaM domain protein kinase (CDPK), two distinct annexins, and profilin, a monomeric actin binding protein whose activity is calcium dependent (Kovar et al., 2000), to fully sequence and characterize. Complete sequences for these genes were determined and deposited in GenBank as follows: annexin 1, AnnCr1 (AF308588); annexin 2, AnnCr2 (AF308589); CaM, CrCaM1 (AF510075); CDPK, CrCPK1 (AY138479); and profilin, CrPRO1 (AY102169). Upon request, all novel materials described in this publication will be made available in a timely manner for noncommercial research purposes.

With the EST information available, a reverse genetics approach to knock out or suppress the expression of known, sequenced target genes became feasible. RNA silencing, or RNAi, is a widely reported and well-known phenomenon by which the introduction or formation of double-stranded RNA (dsRNA) induces degradation of homologous mRNA (for review, see Sharp, 2001; Hannon, 2002). This technique has been applied successfully in plants (Chuang and Meyerowitz, 2000) and was first extended to fern spores by Wolniak and coworkers to study spermiogenesis in M. vestita spp. gametophytes (Klink and Wolniak, 2000; Klink and Wolniak, 2001; Tsai and Wolniak, 2001). Because of the relative ease of synthesizing dsRNA constructs and introducing the dsRNA into the spore, this approach seemed attractive as a potentially rapid and efficient method to suppress gene expression in C. richardii spores (Fig. 1).

View larger version (70K): [in this window] [in a new window]   Figure 1.   Model of RNAi in C. richardii. dsRNA is added to the germination medium and taken up by the spore where it triggers silencing of homologous endogenous mRNA. The paradigm for RNAi action includes processing of dsRNA into 21 to 25 nucleotide small interfering RNAs (siRNAs) that guide subsequent degradation of mRNA by the RNA-induced silencing complex (RISC; Hannon, 2002).

To test the efficacy of this approach, we synthesized a number of dsRNA constructs for the five genes selected from the EST library. DNA templates for the transcription reactions were made by PCR using forward and reverse primers that included 5' T7 RNA polymerase promoter sequence (5'-TAATACGACTCACTATAGGGAGACCAC-3'). The resulting DNA fragments included about 200 bp corresponding to the target gene, typically from the deduced open reading frame (ORF), and were used for in vitro transcription reactions (Ampliscribe T7, Epicentre Technologies, Madison, WI) in which both the sense and antisense strands were transcribed simultaneously and self-annealed during the reaction incubation (Kennerdell and Carthew, 1998). To introduce these dsRNAs into the C. richardii spore, 4 mg of surface-sterilized spores were resuspended in 250 µL of liquid spore germination media (25 mM MES, pH 6.0; one-half-strength Murashige and Skoog) containing the dsRNA, and the spores were immediately placed in light to initiate the germination process (Klink and Wolniak, 2001).

Treatments with constructs of dsRNA derived from CrCaM1, CrCPK1, and CrPRO1 all at a concentration of 0.1 mg mL¹ specifically suppressed steady-state mRNA levels of the corresponding gene without altering the expression pattern of any of the other genes assayed (Fig. 2A). The specificity of this effect was more rigorously tested by attempting to suppress the expression of one of two highly similar annexin genes, AnnCr1 and AnnCr2. Constructs of dsRNA were synthesized targeting the 3' end of the ORF regions of these two genes. Within the construct sequence, each of the dsRNAs shared approximately 80% identity with the nucleotide sequence of the non-targeted genes. Both of these dsRNA constructs are able to effectively reduce the steady-state annexin mRNA levels when compared with untreated spores, and this reduction is specific for the targeted annexin gene (Fig. 2B). These results indicate that RNAi is an effective approach to specifically suppress expression of even closely related genes in C. richardii. A significant advantage of this system over other plant systems is the relative ease of delivery of the dsRNAs into the target cells, because the spores readily take up the dsRNA from the germination media.

View larger version (31K): [in this window] [in a new window]   Figure 2.   Sequence-specific and concentration-dependent suppression of gene expression in C. richardii by treatment with dsRNA. A and B, dsRNA specifically suppresses gene expression. Spores either were treated with 0.1 mg mL1 of various dsRNA constructs or were untreated (lane C). C, Suppression of gene expression is concentration dependent. Spores were treated with various concentrations of dsRNA. The constructs targeted the gene for which expression was assessed. D, Effectiveness of suppression varies between antisense (asRNA) and dsRNA and the targeted region of the message. Spores were treated with either asRNA or dsRNA at a concentration of 0.1 mg mL1 homologous to either predominantly the ORF or the 3'-untranslated region (3'-UTR) of AnnCr1 or AnnCr2. The dsRNAs used in A through C corresponded to the following nucleotide regions of the cDNA sequences in GenBank: dsCaM, 297 to 497 (AF510075); dsCDPK, 1,492 to 1,695 (AY138479); dsPRO, 345 to 546 (AY102169); dsAnn1, 804 to 1,013 (AF308588); dsAnn2, 968 to 1,185 (AF308589). D, The RNAs used for asAnn1 and dsAnn1 constructs based on AnnCr1 cDNA sequence (accession no. AF308588) from bp 1,039 to 1,250, including 98 bp in the ORF and 102-bp 3'-UTR. asAnn2 and dsAnn2 constructs based on AnnCr2 cDNA sequence (accession no. AF308589) from bp 918 to 1,118, including 163 bp in the ORF and 48-bp 3'-UTR. For all experiments, dry spores were placed in germination medium with or without various dsRNA constructs and allowed to develop for 24 h in continuous light. Total RNA was then isolated from the spores, and gene expression levels were assessed using reverse transcriptase (RT)-PCR. To isolate RNA, harvested spores were ground in equal parts RNA grinding buffer (1.0 M Tris-HCl, pH 7.3; 5.0 mM EDTA, pH 8.0; and 1% [w/v] SDS) and acidic phenol:chloroform:IAA (pH 5.2; 25:24:1, v/v), yielding an aqueous layer that was removed and extracted once with chloroform:IAA (24:1, v/v). Using glycogen as a coprecipitant, the RNA was precipitated in ethanol, resuspended in diethyl pyrocarbonate-treated water, and treated with DNase I. First-strand cDNA synthesis was carried out using SuperScript II Reverse Transcriptase (Invitrogen) according to the manufacturer's directions, and the resulting first-strand cDNA was used as template in PCR reactions to assess the presence or absence of various genes. For all panels, the RT-PCR product of a C. richardii actin gene (accession no. BE643189) was used as an internal control, and the primers used for RT-PCR and the corresponding size products are as follows: CrCaM1 (5'-ATGGCTGAGCAACTCACCACT-3'; 5'-TCACTTTGAAAGCATCATCCTCA-3'; 449 bp); CrCPK1 (5'-AATACATTTGGACGTGAAGAG-3'; 5'-GAAACAGCCACGAAAACTGCT-3'; 2,252 bp); CrPRO1 (5'-ATGTCTTGGAATGGGTA TGTTGA-3'; 5'-GTACCAAAATCAAGAGCCGTTC-3'; 630 bp); AnnCr1 (5'-AATCCGGTACCTGACACGAA-3'; 5'-TTACAGCATCTTCCTTCGCA-3'; 1,000 bp); AnnCr2 (5'-TTACAGTCCCCAATCCGGTA-3'; 5'-AATTCATAATGGAAGTTTTGTCA-3'; 866 bp); and actin (5'-TGCATTGGACTATGAACAGGA-3'; 5'-GTATGACGAGTCAGGGCCAT-3'; 446 bp).

We next examined the concentration of dsRNA required to knock down the expression levels of the respective genes. Although concentrations of dsRNA down to 0.01 mg mL¹ can be effective in reducing expression levels, the most consistent suppression was seen at 10-fold higher concentrations of 0.1 mg mL¹ (Fig. 2C). Because the major events in polarity development occur within the first 96 h of gametophyte development, the duration of the suppression effect was examined. Spores continuously treated with dsRNA for CaM exhibit suppression through at least 120 h after light initiation of germination (data not shown). These results suggest that RNAi may be used to study events of gravity-directed polarity development in C. richardii including the polar Ca²⁺ current, nuclear migration, asymmetric cell division, and tip growth, which occur over the first 120 h of development.

Given the ease of uptake of dsRNA into C. richardii spores, it became of interest to learn whether single-stranded antisense RNA (asRNA) could also be taken up, and if so, to compare the relative effectiveness of dsRNA and asRNA constructs in suppressing gene expression. For these tests, both the dsRNA and asRNA constructs used were based on sequences bridging the ORF-3'-UTR border of the two closely related annexin cDNA sequences.

For AnnCr1, but not for AnnCr2, the antisense construct was more effective than a dsRNA construct based on the same sequence. The asRNA for AnnCr1 specifically and completely knocked out AnnCr1, whereas the corresponding dsRNA construct did not reduce AnnCr1 expression levels (Fig. 2D). In contrast, the dsRNA Ann2 construct was much more effective than the single-stranded construct in reducing the level of steady-state message for its targeted annexin (Fig. 2D). A significant difference between the AnnCr1 and AnnCr2 constructs is that the AnnCr1 RNAs target proportionately more of the 3'-UTR than the AnnCr2 constructs. The AnnCr2 constructs conversely contain more ORF sequence than those targeting AnnCr1. These results suggest that dsRNA constructs are more effective at suppressing expression when they are based on the ORF of a gene than when based on the 3'-UTR. However, it appears that antisense constructs do not share the same limitation.

The antisense results are in contrast to those reported previously in the M. vestita system. In M. vestita, asRNAs targeting centrin required 10-fold greater concentrations than the corresponding dsRNA constructs to be effective (Klink and Wolniak, 2001). One possible explanation for this conflicting result is that the centrin constructs used in the M. vestita studies were based solely on the ORF region of the targeted gene. Our results also indicate that the antisense Ann1 construct is more effective than the antisense Ann2 construct, which targets only one-half as much of the 3'-UTR (Fig. 2D).

A new family of expressed RNAs, termed micro-RNAs, has recently been discovered in a wide variety of eukaryotes including plants and may represent a novel way of regulating gene expression (Grosshans and Slack, 2002; Reinhart et al., 2002; Llave et al., 2002). These small RNAs appear to be complimentary to the 3'-UTR of target transcripts and are very effective at decreasing message levels for a particular endogenous gene (Lai, 2002). One intriguing possibility is that the antisense constructs we used in this study are effective because they include a micro-RNA sequence within their 3'-UTR.

Together, these results describe a complete system allowing a rapid and efficient assessment of gene function during an important developmental stage. Among the more than 2,300 different ESTs in the C. richardii library are many genes that are likely candidates for involvement in signaling. The methodology described here provides a facile yet effective approach for testing the role of these genes in light and gravity signaling and in basic cellular processes such as polarity development and cell division.