Alternative splicing of pre-mRNAs of Arabidopsis serine/arginine-rich proteins: regulation by hormones and stresses

Authors


For correspondence (fax 970 491 0649; e-mail reddy@colostate.edu).

Summary

Precursor mRNAs with introns can undergo alternative splicing (AS) to produce structurally and functionally different proteins from the same gene. Here, we show that the pre-mRNAs of Arabidopsis genes that encode serine/arginine-rich (SR) proteins, a conserved family of splicing regulators in eukaryotes, are extensively alternatively spliced. Remarkably about 95 transcripts are produced from only 15 genes, thereby increasing the complexity of the SR gene family transcriptome by six-fold. The AS of some SR genes is controlled in a developmental and tissue-specific manner. Interestingly, among the various hormones and abiotic stresses tested, temperature stress (cold and heat) dramatically altered the AS of pre-mRNAs of several SR genes, whereas hormones altered the splicing of only three SR genes. These results indicate that abiotic stresses regulate the AS of the pre-mRNAs of SR genes to produce different isoforms of SR proteins that are likely to have altered function(s) in pre-mRNA splicing. Sequence analysis of splice variants revealed that predicted proteins from a majority of these variants either lack one or more modular domains or contain truncated domains. Because of the modular nature of the various domains in SR proteins, the proteins produced from splice variants are likely to have distinct functions. Together our results indicate that Arabidopsis SR genes generate surprisingly large transcriptome complexity, which is altered by stresses and hormones.

Introduction

The hallmark of most eukaryotic protein-coding nuclear genes is that the coding region is interrupted by non-coding intervening sequences called introns (Sharp and Burge, 1997). In the completed genome sequence of the model plant Arabidopsis about 80% of protein-coding nuclear genes are predicted to contain one or more introns. The number of introns in Arabidopsis genes varies from 1 to 77 (Reddy, 2001). The introns in the primary transcript (precursor mRNA) are removed and the coding sequences (exons) are joined in the nucleus to generate functional mRNAs, which are transported to the cytoplasm. Studies have shown that introns are often required for the full expression of genes (Rose, 2004; Rose and Beliakoff, 2000). Precursor mRNAs (pre-mRNAs) with multiple introns can undergo alternative splicing (AS) to produce structurally and functionally different proteins from the same gene (Black, 2003). In non-plant systems, AS plays important roles in various aspects of development (Jeanteur, 2003). Several recent studies indicate that AS is a main source of transcriptome and proteome diversity in metazoans (Graveley, 2001; Johnson et al., 2003; Maniatis and Tasic, 2002). The alignment of expressed sequence tags (ESTs) and full-length cDNAs with the completed genome sequence of Arabidopsis has permitted the global analysis of AS. Such analyses have revealed that about 12–20% of Arabidopsis nuclear genes produce multiple RNAs (Haas et al., 2003; Iida et al., 2004; Wang and Brendel, 2006).

The precise excision of introns from nuclear pre-mRNA and the joining of exons takes place in the spliceosome, a large RNA–protein complex that contains five small nuclear ribonucleoprotein particles (snRNPs) and a large number of non-snRNP proteins (Burge et al., 1999). The assembly of the spliceosome involves a series of RNA–RNA, RNA–protein and protein–protein interactions. Comprehensive analysis of the human spliceosome indicates that it consists of over 300 proteins (Rappsilber et al., 2002; Zhou et al., 2002). Serine/arginine-rich (SR) proteins are part of the spliceosome and play critical roles in both constitutive splicing and AS of pre-mRNAs (Manley and Tacke, 1996). Members of this family of proteins are highly conserved with one or two RNA recognition motifs (RRMs) and an arginine/serine-rich (RS) domain. These proteins participate in RNA–protein and protein–protein interactions and function as splicing regulators at multiple stages of spliceosome assembly (Graveley, 2000; Graveley et al., 1999; Sanford et al., 2003). The RRM sequence is necessary for binding to RNA, whereas the RS region is required for interaction with either other SR proteins or cellular proteins. Furthermore, it was recently shown in animal systems that the RS domain in RS proteins interacts directly with the branchpoint in pre-mRNA to promote pre-spliceosome assembly (Shen and Green, 2004). The functions of SR proteins have been well characterized in mammals and Drosophila melanogaster (Jeanteur, 2003). They modulate AS pathways either in a tissue-specific manner or at a particular developmental stage (Baker, 1989; Sanford et al., 2003). SR proteins bind exonic splicing enhancers or silencers (ESE/ESS), recruit other spliceosomal proteins such as U2 auxiliary factor (U2AF) and U1 snRNP to weak splice sites and promote splicing (Graveley, 2000; Maniatis and Tasic, 2002). Some SR proteins (e.g. human SRp38) inhibit splicing, whereas others antagonize the splicing inhibitory activity of the heterogeneous nuclear RNP (hnRNP) proteins that bind ESSs (Sanford et al., 2003; Shin et al., 2004). In addition to their roles in pre-mRNA splicing, recent studies indicate a role for SR proteins in mRNA export, mRNA stability and translation (Huang and Steitz, 2001; Lemaire et al., 2002; Sanford et al., 2004).

In humans, there are 10 SR proteins with a molecular mass ranging from 20 to 75 kDa (Graveley, 2000; Sanford et al., 2003). Published reports on SR proteins and analysis of the completed Arabidopsis genome sequence have revealed that there are 19 SR genes (Kalyna and Barta, 2004). The molecular mass of Arabidopsis SR proteins ranges from 21 to 45 kDa. Some plant SR proteins have been shown to complement the S100 extract that is deficient in splicing (Lopato et al., 1996a,b, 1999b), and one SR protein promoted a switch in splice site (Lazar et al., 1995). A few plant SR proteins have been shown to alter the splicing of pre-mRNAs in vivo (Gao et al., 2004; Isshiki et al., 2006; Kalyna and Barta, 2004; Kalyna et al., 2006; Lopato et al., 1999b). Several plant SR proteins have been shown to be localized to nuclear speckles (Ali and Reddy, 2006; Ali et al., 2003; Docquier et al., 2004; Fang et al., 2004; Lorkovic et al., 2004a; Tillemans et al., 2005). Five Arabidopsis SR proteins (SR45, SR33/SCL33, SRZ21/RSZ21, SRZ22/RSZ22 and SR1) interact with U1 snRNP70K, one of the U1 snRNP-specific proteins implicated in regulating basic splicing and AS of pre-mRNAs (Golovkin and Reddy, 1996, 1998, 1999; Lopato et al., 2002; Lorkovic et al., 2004b). Interestingly, several SR genes in plants and animals are known to undergo AS (Ge and Manley, 1990; Iida and Go, 2006; Jumaa et al., 1997; Kalyna et al., 2003, 2006; Komatsu et al., 1999; Lazar and Goodman, 2000; Lejeune et al., 2001; Lopato et al., 1999b; Screaton et al., 1995; Wang and Brendel, 2006).

In this study, we performed comprehensive analyses of pre-mRNA splicing of all 19 Arabidopsis SR genes. We analyzed the expression and splicing patterns of SR genes in different tissues and in seedlings of different ages. In addition, we analyzed AS of SR genes in response to various hormones (auxin, abscisic acid and cytokinin), environmental stresses (heat, cold and salt), methyl jasmonate, hydrogen peroxide and glucose. Our results show that most SR genes (15 out of 19) undergo AS and produce two or more transcripts. At least 95 transcripts are produced from 15 SR genes. Furthermore, the AS of several SR genes is significantly altered by various environmental stresses, whereas hormones influenced the splicing of only three SR genes. Sequence analysis of splice variants revealed the included/excluded regions in each transcript and their effect on encoded proteins.

Results

Expression and splicing of SR genes in different organs and pollen

The number of introns in Arabidopsis SR genes ranges from five to 12 (Table S1). Intron size varies from 73 nucleotides in SR30 to 1100 nucleotides in SCL30. Although several plant SR genes are known to undergo AS (Gao et al., 2004; Gupta et al., 2005; Iida and Go, 2006; Isshiki et al., 2006; Kalyna et al., 2003, 2006; Lazar and Goodman, 2000; Lopato et al., 1999b; Wang and Brendel, 2006), the full extent and the regulation of the AS of SR genes are poorly understood. To determine the AS of pre-mRNAs encoding all SR genes in Arabidopsis, we performed comprehensive analyses of expression and splicing patterns of all 19 SR genes using RT-PCR in different organs and pollen. Gene-specific primers (Table S1) corresponding to the first and last exon (or second exon and penultimate exon in a few cases) of each gene were used in RT-PCR. To test the specificity of the primers we performed PCR with Arabidopsis genomic DNA. As shown in Figure S1(a), each primer set produced a single band of expected size, indicating that the primers are specific to each gene. Prior to preparing the first-strand cDNA, DNAase-treated RNA was used as a template in PCR to determine the genomic DNA contamination in the RNA. No amplification was observed with all the primer sets (Figure S1b), indicating that the RNA used in RT-PCR analysis is free of DNA contamination.

To determine either the organ- or the tissue-specific expression and AS of SR genes, we performed a detailed analysis of the exression of SR genes as well as splicing variation in roots, stems, leaves, inflorescence and pollen. The first-strand cDNA prepared from these organs and pollen was used in PCR. As shown in Figure 1, all 19 genes are expressed in all tissues tested, suggesting that none of the SR genes is a pseudogene. Surprisingly, 15 out of 19 (∼80%) SR genes exhibited AS, whereas only four genes (SRZ21/RSZ21, SRZ22/RSZ22, RSZ22a and SCL28) produced a single transcript in all the tissues tested (Figure 1a and Table 1). Over 90 transcripts are produced from 15 genes, thereby increasing the transcriptome complexity by more than six-fold. The sizes of all the alternatively spliced transcripts of SR genes are presented in Table 1. In genes that produced multiple transcripts, in most cases, one splice variant was more abundant than the others (e.g. SR34, SR34a, RS40, RS41, RSZ32, RSZ33, SCL30 and SCL30a). Interestingly, the relative levels of the different isoforms of some SR genes varied considerably among different tissues (Figure 1a,b). Quantification of all transcripts from alternatively spliced genes in tissues and pollen revealed that several SR genes (e.g. RS31, RS40, RS41, RSZ32 and SCL30a) are expressed at low levels in pollen (Figure S2).

Figure 1.

Figure 1.

 Analysis of the expression and splicing of serine/arginine-rich (SR) genes in root, stem, leaf, inflorescence and pollen.
(a) Gel pictures showing the expression and splicing pattern of all 19 Arabidopsis SR genes. DNAse-treated RNA from different tissues was used to prepare first-strand cDNA. An equal quantity of first-strand cDNA was used in PCR with primers specific to each SR gene. Sequences of forward and reverse primers used in PCR are presented in Table S1. An equal quantity of template in each reaction was verified by amplifying a constitutively expressed cyclophilin. The name of the SR gene is shown on the left of each panel. Asterisks indicate the transcripts that encode full-length proteins. DNA sizes in bp are indicated by arrows.
(b) Quantification of splice variants in tissues. Each splice variant was quantified and normalized to cyclophilin level and presented as the percentage of all isoforms in a sample. If an isoform is not detectable in any tissue, but appeared under other conditions, then quantification data corresponding to that isoform is not shown.

Figure 1.

Figure 1.

 Analysis of the expression and splicing of serine/arginine-rich (SR) genes in root, stem, leaf, inflorescence and pollen.
(a) Gel pictures showing the expression and splicing pattern of all 19 Arabidopsis SR genes. DNAse-treated RNA from different tissues was used to prepare first-strand cDNA. An equal quantity of first-strand cDNA was used in PCR with primers specific to each SR gene. Sequences of forward and reverse primers used in PCR are presented in Table S1. An equal quantity of template in each reaction was verified by amplifying a constitutively expressed cyclophilin. The name of the SR gene is shown on the left of each panel. Asterisks indicate the transcripts that encode full-length proteins. DNA sizes in bp are indicated by arrows.
(b) Quantification of splice variants in tissues. Each splice variant was quantified and normalized to cyclophilin level and presented as the percentage of all isoforms in a sample. If an isoform is not detectable in any tissue, but appeared under other conditions, then quantification data corresponding to that isoform is not shown.

Table 1.   A summary of the sizes of all splice variants from Arabidopsis serine/arginine-rich (SR) genes
 SR geneNo. productSize of the productsFull-length cDNA sizeExpected size of product with full-length transcripta
  1. aThis number represents the total number of spliced products observed from each gene in tissues, seedlings and under stress conditions. The number in parenthesis indicates the number of splice variants that could not be cloned for sequencing.

  2. Splice variants that are shown in bold and italics are not sequenced.

1SR3051) 773 2) 852 3)1018 4) 1111 5) 11901144 773
2SR1/SR3471) 880 2) 960 3) 1197 4) 1233 5)1267 6) 1313 7)16191237 880
3SR34a41) 1005 2) 1116 3) 1299 4) 139910981005
4SR34b91) 906 2) 1079 3) 1146 4) 1254 5)1284 6) 1286 7) 1435 8) 1480 9) 175515131286
5RS3181) 522 2) 915 3) 1035 4) 1086 5) 1200 6) 1320 7) 1404 8)14911118 522
6RS31a91) 701 2) 795 3) 1033 4) 1257 5) 1406 6) 1500 7) 1507 8) 1578 9) 16721007 701
7RS407 (1)1) 1029 2) 1096 3) 1286 4) 1612 5) 1679 6) 1867 7) 85011541029
8RS418 (2)1) 1340 2) 1420 3) 1435 4) 1666 5) 1865 6) 2240 7) 1500 8) 160014101340
9SRZ21/RSZ2111) 6371358 637
10SRZ22/RSZ2211) 456 982 456
11SRZ22a11) 900 976 900
12RSZ3271) 671 2) 788 3) 889 4) 927 5) 1006 6) 1276 7) 14971441 889
13RSZ336 (2)1) 947 2) 1087 3) 1165 4) 1305 5) 1350 6) 170013691165
14SC3531) 1006 2) 1196 3) 198912661006
15SR33/SCL338 (1)1) 993 2)1066 3) 1119 4) 1180 5) 1227 6) 1668 7) 1834 8)135011571066
16SCL305 (3)1) 554 2) 1003 3) 820 4) 1100 5)12001101 554
17SCL30a7 (2)1) 782 2) 836 3) 903 4) 983 5) 1505 6) 1200 7) 13001198 782
18SCL2811) 9071085 907
19SR4521) 301 2) 3201474 320

There are four alternative splicing factor/splicing factor 2 (ASF2/SF2)-like genes (SR30, SR34, SR34a and SR34b) in Arabidopsis (with either 11 or 12 introns) and all produced multiple transcripts (Figure 1a). The largest splice variant of SR30 (isoform 4) was more abundant in root, stem and leaf. In contrast, in pollen the smallest transcript (isoform 1) was much more abundant than the other splice variants (Figure 1a,b). Six transcripts were produced from SR1/SR34 and isoform 1 is more abundant in all tissues. The largest transcript of SR34a (isoform 4) was more abundant in pollen compared with the organs (Figure 1a,b). SR34b generated seven alternatively spliced products, and all seven were visible in inflorescence. Three of these (isoforms 3, 6 and 7) were abundant in root, stem, leaf and inflorescence, whereas the smallest transcript of SR34b (isoform 1) was more abundant in pollen (Figure 1b).

Four SR genes (RS31, RS31a, RS40 and RS41) that belong to a group with no animal counterparts have five introns in each. Isoform 3 of RS31 was present only in pollen (Figure 1b). RS31a produced seven transcripts in all organs and eight transcripts in pollen (Figure 1a,b). RS40 produced six products, and the smallest transcript was the most abundant of all (Figure 1b). Eight splice variants were generated from RS41, and the smallest transcript was the most prominent in all tissues.

Five Arabidopsis SR proteins (SRZ21/RSZ21, SRZ22/RSZ22, RSZ22a, RSZ32 and RSZ33) have one or more zinc knuckles, and three of them (SRZ21/RSZ21, SRZ22/RSZ22 and RSZ22a) with one zinc knuckle are similar to animal 9G8 SR protein (Kalyna and Barta, 2004; Reddy, 2004). The 9G8-like SR genes with five introns produced a single transcript in all organs (Figure 1a). RSZ32 produced seven splice variants, and the smallest splice form was abundant in all tissues. RSZ33 generated five transcripts, and the level of the most abundant splice variant (smallest isoform) varied considerably among tissues (Figure 1a,b).

Five SR genes (SC35, SR33/SCL33, SCL30, SCL30a and SCL28) in Arabidopsis with between five and eight introns encode proteins similar to SC35 in animals (Table S1). SC35 produced three splice variants, and two of these three were present in all organs and pollen. Interestingly, the levels of different isoforms of SR33/SCL33 varied among tissues (Figure 1a,b). SCL30a produced seven transcripts in root, leaf, stem and inflorescence. SCL28 with six introns produced a single transcript in all organs, but the expression of this transcript is very low in pollen (Figure 1b). SR45, a plant-specific gene with 11 introns, produced two transcripts in all tissues (Figure 1a). Together, these results indicate that the relative levels of splice variants of several SR genes changed significantly in different tissues.

Expression and splicing of SR genes in seedlings of different ages

In most SR genes, the expression and splicing pattern in seedlings of different ages has not been studied previously. To analyze the expression and the number of alternatively spliced products in seedlings, and to determine if the splicing patterns change as seedling age, we analyzed the expression of all 19 genes in 3-, 5-, 10- and 15-day-old (3d, 5d, 10d and 15d) seedlings. The expression of each gene, represented as the total of all isoforms at each stage, is presented in the Figure S3. All SR genes are expressed in seedlings. Interestingly, the expression of some SR genes (SR34b, RS31, RS41, RSZ33, SCL 30 and SCL30a) is lower in seedlings compared with that found in organs, whereas SR34a expression in seedlings is higher than that found in organs. The number of alternatively spliced products is the same as in other tissues with a few exceptions (Figures 1a and 2a). For example, in seedlings fewer products are seen with RS41, SC35, SCL30 and SCL30a (compare Figure 1a with Figure 2a). Interestingly, for several genes either the number of spliced products or the ratios of different transcripts changed with the age of the seedlings. In the case of SR30 the abundance of the shortest transcript decreased, whereas the longest transcript increased from 3d to 15d seedlings (Figure 2b). All six splice variants of SR34 were present in all seedling stages. However, one transcript (isoform 5) increased in 15d seedlings (Figure 2b). In the case of SR34b one transcript (isoform 7) increased, whereas isoform 3 decreased, as the age of seedling increased (Figure 2b). Isoform 4 of RS31 increased with the age of the seedling (Figure 2b). Interestingly one transcript (isoform 2) of SR33/SCL33 decreased, whereas the others (isoforms 3 and 6) increased, with age (Figure 2b). In summary the AS pattern of several SR genes is changed markedly with the age of the seedlings, whereas it is either unaltered or marginally altered in other genes.

Figure 2.

Figure 2.

 Alternative splicing of serine/arginine-rich (SR) genes in 3-, 5-, 10- and 15-day-old seedlings.
(a) RT-PCR results of the splicing of SR genes in seedlings. For details on the RT-PCR and primer sets used see the legend to Figure 1. Cyclophilin amplification was used to demonstrate an equal quantity of template in each PCR reaction. The number on top of each lane indicates the age of the seedling. The names of the SR genes are given on the left of each panel.
(b) Quantification of splice isoforms in seedlings. Each splice variant was quantified and normalized to cyclophilin and shown as the percentage of all isoforms in a sample. If an isoform is not detectable in any of these developmental stages then the quantification data corresponding to that isoform is not shown.

Figure 2.

Figure 2.

 Alternative splicing of serine/arginine-rich (SR) genes in 3-, 5-, 10- and 15-day-old seedlings.
(a) RT-PCR results of the splicing of SR genes in seedlings. For details on the RT-PCR and primer sets used see the legend to Figure 1. Cyclophilin amplification was used to demonstrate an equal quantity of template in each PCR reaction. The number on top of each lane indicates the age of the seedling. The names of the SR genes are given on the left of each panel.
(b) Quantification of splice isoforms in seedlings. Each splice variant was quantified and normalized to cyclophilin and shown as the percentage of all isoforms in a sample. If an isoform is not detectable in any of these developmental stages then the quantification data corresponding to that isoform is not shown.

Effect of hormones and environmental stresses on the expression and splicing of SR genes

Little is known about the effect of hormones and various stresses on the AS of SR genes (Lazar and Goodman, 2000). To analyze the effect of hormones and stresses on the expression and splicing pattern of SR genes we either treated 2-week-old seedlings with various hormones or exposed them to various stresses and performed RT-PCR. We used three hormones [abscisic acid (ABA), auxin (indole acetic acid; IAA) and cytokinin (6-benzyl aminopurine, BA)], several stresses [salt (NaCl), heat and cold) and chemicals [hydrogen peroxide (H2O2) and methyl jasmonate (MJ)] that are known to activate defense responses. In addition we have also used glucose, which has been shown to act as a signal and regulate gene expression in plants (Price et al., 2004). All hormones and stresses used here are known to induce the expression of specific genes in Arabidopsis. To ensure that all our treatments were effective, we first performed RT-PCR analyses with primers specific to genes (see Table S2) that are induced by these signals. In our analysis we used RD29A (known to be induced by ABA, cold and NaCl) (Ishitani et al., 1997), IAA1 (induced by auxin) (Park et al., 2002), ARR4 (induced by cytokinin) (D'Agostino et al., 2000), HSP18 (induced by heat shock) (Lohmann et al., 2004), HXK1 (induced by glucose) (Price et al., 2004), PR1 (induced by H202) (Green and Fluhr, 1995) and lipoxygenase (induced by methyl jasmonate) (Bell and Mullet, 1993) as positive controls. As shown in Figure S4, the expression of positive control genes is induced by either hormones or stresses, suggesting that the seedling treatments have worked.

The expression and splicing pattern of several SR genes are affected by one or more treatments (Figure 3a,b). The expression of each SR gene (i.e. the total of all isoforms) in each treatment is presented in Figure S5. Almost all treatments reduced the level of SR33/SCL33 expression (Figure S5). SR30 gene expression was affected by heat and glucose. In heat-treated seedlings, one additional transcript (isoform 2) was observed and the level of the top product (isoform 4) decreased (Figure 3b). In glucose-treated seedlings the level of isoform 1 increased, whereas the isoform 4 level decreased. However, no additional spliced variants were observed (Figure 3a). The expression and splicing pattern of SR1/SR34 is altered by hormones, heat and cold (Figure 3b and Figure S5). ABA and BA reduced the level of isoform 3, whereas cold and heat suppressed one splice variant (isoform 4) of SR1/SR34. Heat and cold showed a dramatic effect on the splice pattern of SR34b (Figure 3a,b). One additional splice form (isoform 9) of SR34b appeared in heat-treated seedlings, whereas some isoforms (3, 6, 7 and 8) were reduced (Figure 3a,b). The level of two transcripts of SR34b (isoform 1 and 2) is increased in seedlings treated with heat and cold. Furthermore, hormones and NaCl reduced the level of isoform 3 (Figure 3b). In seedlings exposed to cold, three splice variants (isoforms 3, 5 and 6) of RS31 were increased (Figure 3b). In the case of RS31a, heat increased isoform 1 and reduced the level of isoforms 4, 5 and 8 (Figure 3b). Isoform 1 of RS40 was significantly increased, whereas isoforms 2 and 4 disappeared in cold-treated seedlings (Figure 3b). Heat also reduced the level of some isoforms (isoforms 4–6) of RS40. Several splice variants (isoforms 4–6) of RS41 were reduced in heat-treated seedlings (Figure 3b). Three splice variants (isoforms 5–7) of RSZ32 disappeared in heat. Heat treatment suppressed isoforms 3 and 4 of RSZ33, whereas glucose reduced the levels of isoforms 2 and 3 (Figure 3b). Hormones and several stresses affected the splicing of SR33/SCL33. Isoform 2 was greatly decreased by ABA, IAA, BA, NaCl and cold, whereas its level was increased by heat and glucose (Figure 3b). Cold increased the level of isoform 3 and glucose increased the level of isoforms 1 and 2 and reduced the level of isoforms 3 and 6. SCL30a produced seven transcripts in the control and all other treatments, except in heat-treated samples. In heat-treated seedlings only one isoform (the smallest transcript) was present, and the other six isoforms were reduced (Figure 3b). Hydrogen peroxide and methyl jasmonate affected neither the expression nor splicing pattern of the SR genes (Figure S6). To test if the observed results with glucose were caused by an osmotic effect, we treated the seedling with mannitol and analyzed the splicing pattern of three genes (SR30, SR1/SR34 and SR33/SCL33) that showed a glucose effect and another gene (RSZ32) that is not affected by glucose. As shown in Figure S7(a,b) the glucose effect on the splicing pattern was not observed with mannitol, suggesting that the glucose results are not caused by an osmotic effect.

Figure 3.

Figure 3.

 Effect of hormones and environmental stresses on the expression and alternative splicing of serine/arginine-rich (SR) genes.
(a) Gel pictures of splicing products. Two-week-old seedlings were treated with 20 μm abscisic acid (ABA) for 2 h, 10 μm indoleacetic acid (IAA) for 2 h, 5 μm 6-benzyl aminopurine (BA) for 1 h, 200 mm NaCl for 6 h, heat at 38°C for 6 h, cold at 4°C for 24 h and 7% glucose for 3 h as described in Experimental procedures. RNA from control and treated samples was used for RT-PCR.
(b) Quantification of splice variants. Each splice variant was quantified and normalized to cyclophilin and presented as the percentage of all isoforms in a sample. If an isoform is not detectable in any treatment then the quantification data corresponding to that isoform is not shown.

Figure 3.

Figure 3.

 Effect of hormones and environmental stresses on the expression and alternative splicing of serine/arginine-rich (SR) genes.
(a) Gel pictures of splicing products. Two-week-old seedlings were treated with 20 μm abscisic acid (ABA) for 2 h, 10 μm indoleacetic acid (IAA) for 2 h, 5 μm 6-benzyl aminopurine (BA) for 1 h, 200 mm NaCl for 6 h, heat at 38°C for 6 h, cold at 4°C for 24 h and 7% glucose for 3 h as described in Experimental procedures. RNA from control and treated samples was used for RT-PCR.
(b) Quantification of splice variants. Each splice variant was quantified and normalized to cyclophilin and presented as the percentage of all isoforms in a sample. If an isoform is not detectable in any treatment then the quantification data corresponding to that isoform is not shown.

Sequence analyses of splice variants

To identify the included/excluded regions in the splice variants and to determine the effects of these on predicted proteins we cloned and sequenced the splice variants. Schematic diagrams of sequenced transcripts and predicted protein from each splice variant are presented in Figure 4. The details on AS events in splice variants are provided in Table 2. In determining AS events, the isoform that encodes the full-length protein (the top-most isoform in Figure 4) for each gene was used as a reference transcript. If a portion of the intron was retained, and did not involve alternative 5′ or 3′ splice selection, we classified it as an alternative exon. However, if a portion of an intron was retained because of an alternative 5′ or 3′ splice site we include these events as alt5′ or alt3′ splicing events, respectively. For example, in isoform 3 of SR30 the splicing event is the result of an alternative exon, whereas in isoform 4 the AS is the result of an alternative 3′ splice site (Figure 4a). An intron is considered as retained only if the full intron is retained. Furthermore, if the same splicing event occurred in more than one splice variant from a gene it was considered as a single event. For example, in SR30 the retention of the 3rd intron in isoforms 2 and 5 was calculated as a single intron retention event. Similarly, the AS of intron 10 of SR30 by an alternative 3′ splice event in isoforms 4 and 5 was counted as a single event. If an exon was present in the reference transcript but not in the splice variant (e.g. the absence of the 11th exon in isoform 2 of SR34b or the 4th exon in isoform 1 of SCL30a), then it was considered as exon skipping (Figure 4). If an AS event of an intron is the result of an alternative 5′ as well as 3′ splice site (see the splicing of the 2nd intron in isoform 7 of RS31a or of the 10th intron in isoform 7 of SR34) it was grouped under the alt5′ and alt3′ category. Analysis of all splice variants revealed that most were generated by intron retention, followed by an alternative exon and an alternative 3′ site. The generation of splice variants because of exon skipping and alternative 5′ and 3′ splice sites was rare (Figure 4 and Table 2). In general, intron retention was found to be prevalent in plants (Ner-Gaon et al., 2004; Wang and Brendel, 2006). None of the splice variants showed mutual exclusion of introns. In most cases, AS events occurred in the coding region (Table 2). AS affected the untranslated regions in only eight splice variants (Table 2). Interestingly, AS resulted in 53 distinct predicted proteins that ether lacked one or more modular domains or contained truncated domains with additional new stretches of amino acids (Figure 4 and Table 2). In four cases (SR34b, SR33/SCL33, SCL30a and SR45), a stretch of amino acids were either deleted or added in the middle without affecting the reading frame (Figure 4). A number of splice variants derived from the AS (about 60 splice variants) of SR genes harbor a premature termination codon (PTC) located more than 50 nucleotides upstream of an exon–exon junction, and meet the criterion of non-sense-mediated decay (NMD) targets (Green et al., 2003) (Table 2).

Figure 4.

 Schematic diagrams of alternatively spliced transcripts from serine/arginine-rich (SR) genes. Genes encoding alternative splicing factor/splicing factor 2 (ASF/SF2)-like proteins (a), SC35-like proteins (b), plant-specific arginine/serine-rich (RS) subfamily proteins (c), zinc knuckle family SRs (d) and plant-specific SR45 (e). The name of the SR gene is shown on the left of each panel. The schematic diagram for each gene shows the gene structure and its alternatively spliced mRNA isoforms (the numbers below each isoform indicate the size of the transcript in nucleotides). Isoforms are numbered in ascending order according to the size (isoform 1 represents the smallest transcript). In all cases except SR34b, SR33, SCL30a and SR45 the smallest transcript (isoform 1) encodes the full-length protein. Predicted proteins from splice variants are shown to the right of each isoform. Exons are filled rectangles and introns are thin lines. Black rectangles represent constitutively spliced exons, whereas the red rectangles indicate the included regions in splice variants. The vertical black arrowheads and asterisks show the start and stop codons, respectively; horizontal green and red arrowheads above and below the gene structures indicate the position of forward and reverse primers, respectively. Sequences of forward and reverse primers used to amplify the products are shown in Table S1. In the schematics of predicted proteins, numbers to the right are the number of amino acids in the protein. RRM, RNA recognition motif; RS, arginine/serine-rich domain; PSK, a domain rich in proline, serine and lysine. The blue rectangles indicate stretches of amino acids that are not present in functional SR proteins.

Table 2.   Summary of alternative splicing events in serine/arginine-rich (SR) genes
No.Gene nameAlt. 5′-SAlt. 3′-SAlt. 5′- and Alt. 3′Alt.exonIRESNMDLocation of alternative splicingNo. predicted proteinsProtein sizes
5′-UTR3′-UTRCoding
  1. Alt. 5′-S, alternative 5′- splice site; Alt. 3′-S, alternative 3′- splice site; Alt. 5′ and Alt. 3′, Alternative 5′ and 3′- splice sites; Alt. exon, alternative exon; IR, intron retention; ES, exon skipping; 5′-UTR, 5′- untranslated region; 3′-UTR, 3′- untranslated region; NMD, potential candidates for non-sense-mediated decay.

  2. Splicing events in two isoforms (one event in isoform 3 of SR34 and one event in isoform 2 of SR33/SCL33) are not included in the table as they do not fall into the listed categories.

  3. aNumbers in bold indicate the size of the full-length protein.

1SR3001011040043268a, 256, 92
2SR3411110060065303, 285, 272, 266, 257
3SR34a01011002012300, 297
4SR34b05012280285278, 178, 168, 97, 37
5RS3111013070073255, 71, 72
6RS31a10114060084250, 241, 53, 42
7RS4010021042043350, 60, 38
8RS4101013040055356, 354, 76, 55, 50
9RSZ3201015050065284, 85, 54, 42, 36
10RSZ3301001030033290, 60, 55
11SC3500006000112303, 119
12SR33/SCL3310031050065287, 263, 109, 84, 74
13SCL3000010000012262, 104
14SCL30a01011131044280, 262, 85, 75
15SR4501000000012414, 407

Although splice variants of some Arabidopsis SR genes have been reported previously (Kalyna et al., 2003, 2006; Lazar and Goodman, 2000; Lopato et al., 1999b), many new splice variants were uncovered in our study (Figure 4). Of the five SR30 variants, three that correspond to isoforms 1, 3 and 4 were reported previously (Lopato et al., 1999b). By northern analysis, two of these were found to be differentially regulated in tissues and during development. The seven transcripts of SR1/SR34 are similar to published results (Lazar and Goodman, 2000; Lopato et al., 1999b). It has been reported that heat increases one isoform (isoform B, which corresponds to isoform 4 in this study). In our case, heat decreased this transcript and other transcripts (Figure 3). This difference could be likned to the age of the plant and/or the time of exposure to heat. We treated 2-week-old seedlings for 6 h, whereas in the published report rosette plants were treated for 3 days. Recently, three isoforms of RS31 were reported (Kalyna et al., 2006). Our study revealed five additional isoforms (Figure 4c). Previously northern analysis with RS41 showed a single transcript (Lopato et al., 1996b), and our RT-PCR analyses revealed eight transcripts (six were sequenced). All four previously reported splice variants of RSZ33 are identified. In this study we showed heat and cold regulation of some of these isoforms (Figure 3b). Using northern analysis RSZ33 was reported to be expressed only in roots and flowers, whereas its promoter–reporter fusion showed expression in shoot meristem, leaf primordia and embryos (Kalyna et al., 2003; Lopato et al., 2002). In our RT-PCR analysis, we also found RSZ33 expression in stems and leaves. We have repeated this analysis several times and sequence verified the products derived from these tissues. The reason for this difference could be that RT-PCR is more sensitive compared with northern analysis and the GUS reporter system. Similarly, published northern data showed two products of RS31 in leaf and stem (Lopato et al., 1996b), whereas our RT-PCR analysis showed three products. Again, this is likely to be aresult of the increased sensitivity of the RT-PCR. Analysis of the splice variants of SR genes that are altered by heat and cold revealed that a given stress can affect more than one AS event. For example, heat caused intron retention (e.g. isoform 2 of SR30 and isoform 9 of SR34b), reduced the level of isoforms that used the alternative 3′ splice site (e.g. isoform 5 of SR34b), the alternative 5′ splice site (e.g. isoform 4 of RS40) or that contained an alternative exon (e.g, isoform 5 of RS40). Similarly, more than one type of AS event was observed in splice variants that are affected by cold stress (e.g. intron retention in isoform 6 of RS31 and RS40, alternative 5′ splice site in isoform 4 of RS40, alternative 3′ splice site in isoform 3 of RS31 and alternative exon in isoforms 2 and 3 of RS40). These results indicate that multiple AS events are affected by stresses.

Discussion

Arabidopsis SR genes are expressed in all organs tested, but the splicing pattern is regulated in a tissue-specific manner and by development

All SR genes are expressed in the tissues tested, but with some expressed at very low levels in pollen (Figure 1a and Figure S2). Our results do not exclude the possibility that some plant SR proteins are expressed in a cell-specific manner. In fact using promoters of SR34/SR1 and RSZ33 fused to the GUS reporter, expression of SR proteins in specific cell types has been reported (Kalyna et al., 2003; Lopato et al., 1999a). For some genes (e.g. SR30, SR34, RS31 and SR33/SCL33) different splice variants were obvious in pollen (Figure 1a,b). Furthermore, the levels of individual splice variants in pollen differed from other tissues (see SR30, RS31a and SR33/SCL33). Some SR genes produced the same number of bands in all tissues. However, the relative levels of transcripts varied between tissues and over different developmental stages (Figures 1 and 2). Some splice variants are expressed either specifically in a tissue (Figure 1) or at a particular developmental stage (see SR33/SCL33 in Figure 2). The results presented in Figures 1 and 2 clearly show that the splicing pattern of SR genes is tightly regulated, and this leads to differences in the abundance of splice variants in tissues and at different developmental stages. The preferential expression of a splicing variant of an SR gene either at a developmental stage or in a specific tissue suggests a role for that variant at that developmental stage or in that specific tissue.

Alternative splicing increases the transcriptome complexity of SR proteins by six-fold

Our comprehensive analysis of the AS of pre-mRNAs of all SR genes in different organs, developmental stages and in response to various signals revealed that about 80% of SR genes (15 out of 19 genes) undergo AS and produce as many as nine splice variants from a single gene. A total of 95 transcripts are produced from 15 Arabidopsis SR genes, thereby increasing the transcriptome complexity from these genes by six-fold (Tables 1 and 2). It is likely that there are even more splice variants as our analysis does not amplify the transcripts that lack the exons to which we have made primers. Recent limited analysis of the AS of rice SR genes has shown that eight SR genes produce 17 splice variants (Isshiki et al., 2006). However, it is not known if either hormones or stresses regulate the splicing of rice SR genes. The splice variants that we observed in Arabidopsis are about four times more numerous than those found in rice. The average size of introns in Arabidopsis is 173 nucleotides. However, all SR genes, with the exception of SC35, that are alternatively spliced have long introns (between 400 and 1100 nucleotides), suggesting that genes with long introns are likely to undergo AS. However, some short introns are also alternatively spliced (Figure 4). Genome-wide computational analysis of AS revealed that about 12–20% of Arabidopsis genes are alternatively spliced (Haas et al., 2003; Iida et al., 2004; Wang and Brendel, 2006). Our results suggest that either the SR genes are unique in producing a large number of splice variants or current estimates of alternative spliced genes in Arabidopsis are underestimated because of the small set of available ESTs/cDNAs in Arabidopsis compared with humans. Global analysis of the AS of all Arabidopsis genes with splicing-sensitive arrays, as has been carried out with humans, should uncover the possibilities (Johnson et al., 2003).

A few reports indicate that AS plays a role in flowering and disease resistance against viral and bacterial plant pathogens (Reviewed in Reddy, 2004). The functional significance of the alternatively spliced products of SR genes is currently not known. However, from the sequence analysis of splice variants (Figure 4 and Table 2) a number of variants are predicted to produce either truncated proteins as a result of in-frame translation termination codons or extended proteins that lack one or more functional domains. As all SR proteins are modular and contain distinct domains [one or more RRMs, one or more SR regions and the region rich in proline, serine, lysine (PSK domain)], the truncated or extended proteins that lack one or more domains may have altered function. For example, proteins that lack one of the domains may either be localized differently or be non-functional but still interact with some spliceosomal proteins and function as dominant negative regulators. Also, it is possible that alternatively spliced variants, especially those that differ only in the 5′ and 3′ untranslated region (see SR34a, RS40 and SCL30a in Figure 4) may have altered RNA stability. In metazoans, many transcripts with a termination codon located more than 50 nucleotides upstream of an exon–exon junction are targets of NMD (Green et al., 2003; Lewis et al., 2003). A large number of splice variants (about 60, see Table 2) derived from AS harbor a PTC and meet the criterion of NMD targets. These splice variants may be degraded selectively by NMD, which is likely to regulate the abundance of different transcripts. In humans a third of alternatively spliced variants are candidates for NMD (Green et al., 2003; Lewis et al., 2003). It has been shown that such regulation operates in the case of SC35, a splicing factor (Sureau et al., 2001). Plants, like non-plant systems, appear to have the capability to detect and degrade transcripts with PTCs (Hori and Watanabe, 2005; Isshiki et al., 2001).

Does ‘splicing noise’ contribute to enhanced transcriptome complexity?

As described above, one reason for the generation of such a large number of splice variants from a small set of genes might be that these splice variants have important functions in generating SR proteome diversity and/or regulating the abundance of functional transcripts by regulated unproductive splicing and translation (RUST) (Soergel et al., 2006). However, the other alternative explanation for this high transcriptome complexity is that pre-mRNA splicing in plants is inherently ‘sloppy’, and the multiple spliced variants are generated as a result of ‘missplicing’ or ‘splicing noise’. However, several observations indicate that this is not the case. First, we analyzed the splicing of another gene family (pectate lyases), which consists of 26 genes with multiple introns that are of the size found in SR genes. Interestingly, pre-mRNAs of none of the pectate lyase genes are alternatively spliced in different tissues and in response to stresses and hormones (data not shown), suggesting that the observed splice variants within SR genes are not a result of ‘missplicing’ or ‘splicing noise’. Second, in most SR genes only one or two exons or introns contributed to all splice variants where different regions of the same exon or intron were retained (Figure 4). Third, predicted proteins from spliced variants lack one or more known functional modules (e.g. RRM, RS or PSK domain), indicating the importance of these variants (Figure 4). Fourth, the AS of SR genes is regulated in a tissue- and stage-specific manner and by various signals (stresses and hormones) (Figures 1–3). Fifth, the position of alternatively spliced introns is conserved across evolutionarily distant plants (Arabidopsis, rice and ferns) (Iida and Go, 2006; Isshiki et al., 2006; Kalyna et al., 2006; Wang and Brendel, 2006). Finally, in one case where the function of the AS of an SR gene was analyzed in transgenic plants, the observed phenotypic effects were seen with the plants transformed with the genomic clone, but not with the cDNA (Kalyna et al., 2003), suggesting that the splice variants produced from the intron-containing gene are important for its function.

Abiotic stresses have a dramatic influence on the alternative splicing of pre-mRNA of SR proteins

Plants, unlike animals, are sessile and thus have to cope with adverse environmental stresses such as heat, cold and salinity. Hence, plants have evolved sophisticated mechanisms to adapt to these environmental stresses by altering their gene expression patterns. Abiotic stresses have been shown to rapidly alter the transcriptome in plants by either inducing or repressing the expression of specific genes (Thomashow, 1999; Ulm et al., 2004). Stress-induced genes have been extensively analyzed for steady state mRNA accumulation. Some of the effects of stresses on gene expression could result from the regulation of splicing machinery. However, very little is known about the effects of stresses on the splicing of genes involved in pre-mRNA splicing (Lazar and Goodman, 2000; Reddy, 2001). In addition, virtually nothing is known about the effect of plant hormones on pre-mRNA splicing. Our results clearly show that heat and cold strongly alter the AS of most SR genes, whereas hormones affected the AS of three SR genes (SR1/SR34, SR34b and SR33/SCL33) (Figure 3b). Hence, some of the heat- and cold-induced changes in gene expression may be caused by the AS of SR genes. New splice variants have appeared in the presence of some stresses (Figure 3). In addition, some splicing products are either increased or decreased by abiotic stresses, particularly by heat and cold (Figure 3). These results will have important implications in understanding the mechanisms of gene regulation under abiotic stresses. Altered ratios of splice variants in response to stresses may have a role in the adaptation of plants to these stresses. Regulation of the AS of some SR genes (SR1/SR34, SR34b and SR33/SCL33) by hormones indicates a role for these SR proteins in some of the processes controlled by hormones.

Our results suggest stresses such as heat and cold alter the expression of SR proteins, which in turn alters the splicing of other pre-mRNAs including SR pre-mRNAs. In support of this, it has been recently reported that cold and other stresses affect the AS profiles of Arabidopsis genes (Iida et al., 2004). Also, it has been shown that the manipulation of expression of one SR protein alters the splicing of its own pre-mRNA and other SR genes (Kalyna et al., 2003, 2006; Lopato et al., 1999b). Low temperatures have been shown to influence the splicing of non-spliceosomal genes (Mastrangelo et al., 2005). It is likely that these effects are mediated by stress-induced changes in the splicing patterns of SR genes. The mechanisms involved in the stress regulation of the AS of plant SR genes remain to be elucidated. In metazoans, it has been shown that changes in the phosphorylation status of SR proteins affect protein–protein interaction and their function. Phosphorylation of SR proteins is required for spliceosome assembly and splice site selection (Cao et al., 1997; Mermoud et al., 1994). Furthermore dephosphorylation of SR proteins is also necessary for the later stages of splicing, suggesting that the phosphorylation and dephosphorylation cycles of SR proteins play a critical role in splicing (Murray and Jarrell, 1999). In humans, SRp38 represses splicing in response to heat shock. Heat shock dephosphorylates SRp38 and the dephosphorylated form causes a general inhibition of splicing (Shin et al., 2004). It is possible that stresses alter the phosphorylation status of SR proteins, which in turn affects the splicing of either their own or other pre-mRNAs of SR proteins. Plant SR proteins have been shown to be phosphorylated in vitro by LAMMER-type protein kinase (Golovkin and Reddy, 1999; Savaldi-Goldstein et al., 2000). AFC2, a LAMMER-type kinase, phosphorylates four Arabidopsis SR proteins (SRZ21/RSZ21, SRZ22/RSZ22, SR33/SCL33 and SR45) and the interaction between the kinase and SR33/SCL33 is modulated by the phosphorylation status of these proteins (Golovkin and Reddy, 1999). Overexpression of a LAMMER-type protein kinase (PK12) from tobacco in Arabidopsis altered the splicing pattern of specific endogenous genes (SR30, SR1/SR34 and U1-70K) and the expression of several other genes, and caused defects in development (Savaldi-Goldstein et al., 2003). In Arabidopsis, the intranuclear distribution of several SR proteins is regulated by phosphorylation and stresses such as heat and cold (Ali et al., 2003; Tillemans et al., 2005). Hence, some of the observed effects of heat and cold stresses on splicing could be the result of an altered distribution of SR protein in the nucleus.

In summary, our results show previously unrecognized transcriptome complexity from pre-mRNAs of SR genes and splicing regulation by abiotic stresses. Remarkably, AS increases the transcriptome complexity of SR genes by about six-fold. Furthermore, AS is differentially regulated during development, in different tissues and in response to environmental stresses, suggesting that the AS of pre-mRNAs of SR genes and the regulation of AS are an important part of the regulatory circuitry of expression of SR and other genes. The dramatic effect of heat and cold and some hormones on the AS of specific SR genes suggests that part of the dynamic changes in the transcriptome of plants in response to abiotic stresses could be caused by changes in SR protein levels and their isoforms. Furthermore, the SR genes that produce altered splice variants in response to stresses provide a unique opportunity to use them as models to dissect the mechanisms by which stress signals modulate AS. It would also be interesting to determine if the alternatively spliced transcripts with PTC differ in their stability compared with transcripts that encode full-length proteins. The results presented here provide a framework for the future functional analysis of alternatively spliced transcripts in stress responses and of various aspects of plant growth and development.

Experimental procedures

Plant material

Seedlings (Arabidopsis thaliana ecotype Columbia) were grown on MS medium (Gibco BRL, http://www.invitrogen.com/) with 1% sucrose at 22°C with 16-h/8-h or 14-h/10-h light/dark cycles. For analysis of splicing at different stages, 3d, 5d, 10d and 15d seedlings were collected and total RNA was extracted. To analyze expression and splicing of SR genes in different tissues of Arabidopsis, root, stem, leaf, inflorescence and pollen were collected from 5-week-old plants. Pollen was collected as described previously (Golovkin and Reddy, 2003).

Seedling treatments

Two-week-old seedlings grown on MS agar plates were used for all treatments. Seedlings were transferred to small Petri dishes containing 10 ml of 1/2 strength MS medium either with or without 20 μm ABA for 2 h, 10 μm IAA for 2 h, 5 μm BA for 1 h, 200 mm NaCl for 6 h, and either 7% glucose or mannitol for 3 h. For heat treatment, seedlings were placed in MS as above and placed at 38°C for 6 h. For cold treatment, seedlings were placed in the medium and incubated at 4°C for 24 h. For methyl jasmonate treatment, seedlings (grown at 22°C, 14-h light/10-h dark) were incubated in MS medium containing 100 μm methyl jasmonate for 6 h. As methyl jasmonate was dissolved in ethanol and the final incubation medium contained 0.1% ethanol, seedlings incubated in MS medium containing 0.1% ethanol were used as a control. Hydrogen peroxide treatment of seedlings (grown at 22°C, 14-h light/10-h dark) was performed by incubating the seedlings in 100 μm H2O2 for 6 h in the dark. Control seedlings were incubated in MS medium in the dark. After the treatments seedlings were collected and frozen in liquid nitrogen for RNA extraction.

Genomic PCR

Arabidopsis genomic DNA was isolated using the cetyl trimethyl ammonium bromide (CTAB) method (Golovkin and Reddy, 1996). Genomic DNA (150 ng) was used for PCR amplification in a reaction volume of 20 μl. All PCR reactions were performed using Takara EX TaqTM polymerase (http://www.takara-bio.co.jp/). The following PCR conditions were used for amplification: initial denaturation was performed at 94°C for 2 min, followed by 29 cycles at 94°C for 30 sec, 56°C for 30 sec and 72°C for 1 min. The final extension was performed at 72°C for 10 min. The amplified products were resolved in 1% agarose gels.

RT-PCR analysis

Total RNA was extracted using an RNeasy Plant Mini Kit (Qiagen, http://www.qiagen.com/), quantified spectrophotometrically at 260 nm and treated with DNase1 according to the manufacturer's instructions (Invitrogen). RNA from pollen was isolated using the TRIzol method with modifications (Golovkin and Reddy, 2003). Prior to RT-PCR, DNAse-treated samples were used in PCR with primers corresponding to SR genes. No amplification of SR genes was observed in DNAse-treated RNA samples, indicating that the RNA used in our analysis has no genomic DNA contamination. DNAse-treated RNA (1.5 μg) was used to synthesize first-strand cDNA with an oligo (dT) primer in a 20-μl reaction volume using SuperScriptII (Invitrogen). One-twentieth of the first-strand cDNA was used for PCR amplification in a reaction volume of 20 μl. Gene-specific primers of the 19 SR genes and control genes that were used for PCR amplification are presented in Tables S1 and S2, respectively. For SR genes, in most cases, the primers correspond to the first and last exon of each gene (Figure 4). Primers for all genes were designed using the Primer3 Input (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi) software. For each gene, preliminary PCRs were performed at different cycles (15, 20, 25, 30 and 35) to determine the linear range of amplification. Based on these analyses, depending on the gene, PCRs were run for 15, 25, 30 or 35 cycles. Twenty-five cycles were used for SRZ21/RSZ21 and SR45 genes, 35 cycles were used for SCL28 and 30 cycles were used for all other genes. Fifteen and 25 cycles were used for RD29A and hexokinase1, respectively. The following PCR conditions were used: initial denaturation was performed at 94°C for 2 min, followed by the respective number of cycles at 94°C for 30 sec, 56°C for 30 sec and 72°C for 1 min. The final extension was performed at 72°C for 10 min. The amplified PCR products were resolved by electrophoresis in 1% agarose gels. All PCR reactions were performed using the Takara EX TaqTM polymerase.

Quantification of transcripts

All mRNA isoforms were quantified using the NIH ImageJ program (http://rsb.info.nih.gov/nih-image/). The level of each isoform in each sample was normalized using cyclophilin, the expression of which was shown to be unchanged in organs and in response to stress (Ditt et al., 2001; Lippuner et al., 1994; Nicot et al., 2005; Yasuno and Wada, 1998). The levels of all isoforms for each gene in each sample were added to calculate the total transcript level. The relative level of each isform in a sample was calculated by dividing the level of each isoform by the sum of all isoforms. The relative level of each isoform is presented as the percentage of all isoforms in a sample.

Cloning and sequencing of splice variants

After PCR amplification with gene-specific primers, each PCR mixture consisting of alternatively spliced products was size fractionated on a 1% agarose gel. Individual bands corresponding to different sizes were gel-purified with the Qiagen gel purification kit (Qiagen). Each PCR product was cloned into pCR2.1-TOPO® vector according to the manufacturer's instructions (Invitrogen). At least 4–6 independent clones for each alternatively spliced product were sequenced. Sequence analysis was performed using Spidey (http://www.ncbi.nlm.nih.gov/IEB/Research/Ostell/Spidey/) and manual inspection.

Acknowledgements

We thank Dr K.V.S.K. Prasad for performing some of the seedling treatments and Dr Irene Day for her comments on the manuscript. This work was supported by a grant from the Department of Energy (DE-FG02-04ER15556) to ASNR.

Ancillary