Cloning, characterisation and comparative analysis of a starch synthase IV gene in wheat: functional and evolutionary implications

A. Cloning and characterisation of SSIVb

Identification and characterization of a wheat SSIV cDNA

An expressed sequence tag (EST) [Genbank:BT009276] homologous to plant SSs and glycogen GSs at the amino acid level was identified from a developing wheat seedling cDNA. Using a 5'RACE-like procedure, contiguous cDNA fragments were amplified from wheat libraries that collectively gave a 3386 bps-long sequence containing a full-length ORF of 2806 bp [Genbank:AY044844] (Figure 1). The amino acid sequence deduced from this ORF contains 914 residues and a predicted molecular mass of 103.1 kDa. A putative cleavage site between amino acids 45 and 46 was identified using the ChloroP neural network [32] cleavage at which would result in a mature protein of molecular mass 98.3 kDa. Analysis of the sequence showed that it was most homologous (87%) to OsSSIVb [Genbank:AAQ82623].

The SSIV cDNA sequence allows it to be unambiguously identified as a SS. The starch catalytic and glycosyltransferase domains characteristic of SSs are present in the C-terminus (Figure 1), and conservation of sequence identities at the two putative ADPglucose binding motifs, i.e. KVGGL and KTGGL, was found between this wheat cDNA and the rice SSIV and SSIII Classes. Expression of the 968-bp EST and 2100-bp RACE products in E. coli produced polypeptides of the expected size. In the latter case, glycosyltransferase activity could be detected, albeit at a low level [see Additional file 1]. Like all other SSs the N-terminus of TaSSIVb is unique; an SSIV-specific region, from amino acids 1–405, was detected and within this region two coiled-coil domains and a 14-3-3-protein recognition site were identified (Figure 1).

Phylogenetic analysis of starch and glycogen synthases

The deduced amino acid sequence of the wheat SSIV cDNA was compared to those of 36 other SSs and 3 GSs, two from Synechocystis and the other from Agrobacterium tumefaciens, using Clustal W [33, 34]. The five SS classes clustered into two groups (Figure 2), an arrangement also reported by others [6, 35]. For ease of description, GBSSI, SSI and SSII will be called the Group A SSs, and SSIII and SSIV will be called the Group B SSs, respectively. Within the Group A SSs, GBSSI appears to diverge from SSI and SSII. Ostreococcus does not have a SSIV gene but has three isoforms of SSIII, with the OtSSIIIa most closely resembling other SSIV proteins. Synechocystis PCC 6803 is one of the few cyanobacteria identified from BLAST analysis to have two GS isoforms, one of which repeatedly clustered with the Group B SSs, and the other repeatedly with the Group A SSs. This was found in all phylogenetic trees created whether using neighbour-joining, minimum evolution or maximum parsimony on both nucleotide and protein sequences.

Genomic organisation of TaSSIVb

A partial genomic clone of SSIV [Genbank:DQ400416)] was sequenced from Triticum aestivum using PCR. Three overlapping contiguous genomic clones (See Figure 3) together gave a sequence of 7,141 bp. The promoter region could not be amplified in our hands presumably because of the high GC content of the 5' end of the gene. The TaSSIVb gene contains 16 exons separated by 15 introns, a structure which is similar to that reported in all SS isoforms cloned to date [36]. Although intronic sequences are generally only weakly similar among plant SSs, there is high identity (ranging from 51.1 to 78.3%) between introns four to nine of TaSSIVb and OsSSIVb while the identity between the introns of the two OsSSIV isoforms averaged only 20% (data not shown). Of note is the similarity of sequence found in intron-4 among all the SSIV isoforms from wheat, rice and Arabidopsis (40 to 47%).

Chromosomal location of TaSSIVb

The chromosomal location of the TaSSIVb gene was determined by Southern blot analysis of wheat nullisomic-tetrasomic and ditelosomic lines [37]. Three bands of approximately 12 kb, 10 kb and 6 kb hybridized to the SSIV probe (data not shown) in nullisomic-tetrasomic lines, indicating that TaSSIVb is a single-copy gene, one gene present on the homeologous group I chromosome of each of the A, B, and D genomes. Additional mapping experiments with ditelosomic lines showed that the 10 kb band was missing in the 1AS lines, the 6 kb was missing in the 1BS line and the 12 kb band was missing in the 1DS lines (Figure 4) which is consistent with a gene on the long arms of the group I chromosomes.

SSIVmRNA expression

The size of the TaSSIVb transcript was determined by northern blotting. A band of approximately 3.4 kb was detected, which is close to the size of the predicted full-length SSIV transcript based on the PCR-amplified fragments. A faint band of higher molecular mass (~3.7 kb) that hybridized to the probe could also be seen (Figure 5A).

The expression pattern of transcripts encoded by TaSSIVb was then determined by semi-quantitative comparative reverse transcription (RT)-PCR (Figure 5B). Expression was highest in leaf and embryo and lower in endosperm where it decreased steadily as the organ matured. This result was reproducible over four different experiments. Expression of the TaSSIIIa transcript was also compared to that of TaSSIVb because of the close phylogenetic relationship between these isoforms (Figure 5B). The TaSSIIIa transcript is strongly expressed in the endosperm, but could be detected in leaf (upon long exposure, data not shown) and embryo albeit at significantly lower levels. Similar expression profiles were obtained in rice with OsSSIVb and OsSSIIIa [38–40].

Expression of SSIV leaf transcript by semi-quantitative comparative RT-PCR under different light conditions

Because TaSSIVb is preferentially expressed in leaf it is possible that transcript levels could be responsive to changes in the light-dark regimen and that it could be regulated by the circadian clock. TaGBSSIb was included in this study for comparison because it was reported to be unresponsive to light when analyzed by northern blotting [41]. The TaSSIVb, TaSSIIIa and TaGBSSIb transcripts (Figures 6A &6B) were all expressed at high levels in response to light, but after 24 h of continuous darkness (at time point 4 in Figure 6B), transcripts were barely detectable. This indicates that in wheat the circadian clock does not regulate expression of these genes since during the extended dark period, mRNA levels do not follow the pattern expected during the 16 h light period.

B. Global sequence analysis of SS Isoforms

Comparative protein modeling

To further our understanding of the wheat SSIVb gene product and SSs in general, the sequences of this, and 31 SSs and 3 GSs from a broad range of species were analysed to identify common as well as unique features that may underscore the basis for conserved functions as well as differences in SS metabolic action in planta. To gain insight on the structural features of SSs, we also compared SS sequences specifically to that of AgtGS for which a crystal structure is available.

3D Structure of SSIV

A homology model for the C-terminus (amino acids 406–914) of TaSSIVb was built by threading its sequence onto the 3D structure of AgtGS (Figure 7). The resulting model is in good agreement with those developed for AgtGS and AtSSIII [26, 42]. The two different α-β-α Rossmann-domains of the GT-B family are apparent as is the large cleft which contains the ADP-binding pocket towards the C-terminal side [43]. Residues that take part in catalysis and/or substrate-binding depicted on the model are elaborated in Table 1, which shows amino acids that are conserved across GS and SSs, and Table 2, which highlights those amino acids that are not highly conserved. These residues were identified based on published studies on AgtGS and, to a lesser extent, based on site mutagenesis studies of EcGS [18, 19, 44]. These amino acids are all brought into close proximity of the ADP molecule in response to the predicted change in conformation of the enzyme in going from an opened, relaxed form to a closed state [43].

	SSs: Group A			SSs: Group B
AgtGS	GBSSI	SSI	SSII	SSIII	SSIV
Lys15	Lys91	Lys153	Lys322	Lys1194	Lys435
Gly18	Gly94	Gly156	Gly325	Gly1197	Gly438
Asp138	Asp229	Asp285	Asp447	Asp1309	Asp558
His163	His258	His314	His476	His1337	His586
Asn246	Asn347	Asn408	Asn562	Asn1385	Asn671
Ile297	Val/Ile400	Ile460	Ile619	Val/Ile1437	Val/Ile724
Arg299	Arg402	Arg462	Arg621	Arg1439	Arg726
Gly327	Gly431	Gly490	Gly649	Gly1467	Gly754
Tyr354	Phe458	Phe517	Phe676	Phe/Tyr1499	Tyr784
Asn355	Ser/Asn459	Ser/Asn518	Ser677	Asp1500	Ap785

Amino acids that are part of the catalytic site in AgtGS were identified in previous studies [18, 19, 43, 44] and the equivalent residues in plant SSs were identified after alignment of AgtGS using ClustalW [34]. Only AgtGS amino acids that were conserved across all of the SS isoforms are shown. Data was extracted from the alignment shown in Figure 8.

	SSs: Group A			SSs: Group B
AgtGS	GBSSI	SSI	SSII	SSIII	SSIV
Thr16	Thr92	Ser/Thr154	Thr323	Val1195	Val436
Gln140	His231	His287	His449*	Ser1311①	Gln560
Ile162	Ile257	Ile313	Ile475	Ile1336	Cys585②
Phe167	Tyr/Phe262	His-318	His480	Tyr/Phe1341	Tyr590②
Ile297	Val/Ile400	Ile460	Ile619	Val/Ile1437	Val/Ile724
Ser298	Gly401	Gly461	Gly620	Ser/Thr1438	Ser/Thr725
Gly329	Gly433	Gly492	Gly651	Ala1469	Ser756②
Gly353	Arg/Lys457	Gly516	Gly675	Arg/Lys/Tyr/Thr1498	Lys783②
Asn355	Ser/Asn459	Ser/Asn518	Ser677	Asp1500	Asp785②
Ser359	Ala463	Ala/Ser522	Ala681	Ser1504	Ser789②
Thr381	Ile485	Asn554	Asn703	Ser/Thr1526	Thr811

Amino acids identified in AgtGS as part of the ADPglucose pocket and are involved in binding and/or catalysis that were not conserved across all of the SS isoforms are shown. Non-conservative amino acids changes that could affect enzyme properties are emboldened. Data was extracted from the alignment shown in Figure 8. Species with an amino acid at the residue other than that in the table are indicated as follows: ①OtSSIIIa/c where the residue is a glutamine. ②CrSSIV. *OsSSIIa – Tyr

Analysis of the conserved domains

The length and homology of the conserved starch catalytic (GT-5) and the glycosyltransferase-1 (GT-1) domains of some SSs and GSs were compared to those found in prokaryotic GSs for which a crystal structure exists. The GT-5 domain (Pfam PF08323) was similar among the different SS and GS isoforms with the range of Expect values (E) from 6.5e^-10 to 2.5e^-140 compared to that of AgtGS (data not shown). However there was less similarity in the GT-1 region (Pfam PF00534), with the Group B SSs differing most. E-values were 0.5 to 4.2e^-3 in the Group B SSs and from 2e^-7 to 4.8e^-16 in the Group A SSs when compared to that in EcGS.

SS secondary structure

The secondary structures of the different wheat SSs were compared to AgtGS using the Jnet prediction program [45, 46]. Some structural features specific to each SS sub-family among the different species could be identified. For example, an additional β-strand is found between β1 and α1 in the Group A SSs that is absent in the SpGSs, AgtGS and the Group B SSs, (Figures 8, 9, 10, 11). The hydrophilic "tail" region at the end of the C-terminus that is unique to GBSSI proteins [15] was found to adopt a helical structure. This helical region is also found in SpGS1 but not in SpGS2 (Figure 11).

There are two Loops containing important residues (H163 and Gly329, Table 1; discussed below) that have different lengths in the SSs sub-families. The first Loop, referred to as the 380s Loop [42, 43] (and see Figures 7 and 9), is a stretch of 35 amino acids containing a small antiparallel sheet (β8-β9-β10) and one helix (α5) in AgtGS but only one predicted helix in the SSs: GBSSI, SSI, SSII and SSIV. In the wheat SSIII, this polypeptide stretch is reduced to 7 amino acids and does not include a helix. This region is highly variable at all levels (nucleotides, amino acids, secondary structure) among the SSs sub-families. The second Loop is located in motif VI, which contains Gly329 in AgtGS, and is shown in green on Figures 7 and 10. It is longer in SSIII and SSIV than in the Group A SSs and presents few homologies when compared to AgtGS.

Active site residues

Amino acid residues that make up the highly conserved regions of SSs and GSs initially found by Gao et al., (1998) [27] may now be extended as more SS and GS sequences are available. Residues involved in ADPglucose and/or glycogen/starch binding have also been identified in AgtGS which could also be instructive in identifying those defining amino acids needed for catalysis [43] (Figures 8, 9, 10, 11).

The variable amino acid in the Lys-X-Gly-Gly-Leu motif in the SS and GS sequences was first examined. This motif, which occurs at the highly conserved Lys15 in AgtGS, has long been considered as part of the ADPglucose active site in SSs and GSs [20, 21]. Although this hypothesis has since been brought into question [23], our observations show a high level of intra-group conservation at this residue. It is a valine (non-polar) in Group B SSs, and a threonine or a serine (both polar) in the Group A SSs (Table 2; Figure 8). This appeared true in all 5 SS classes and for all 33 species compared (data not shown). Interestingly, like the Group B SSs, SpGS1 and SpGS2 both have non-polar residues (valine and alanine, respectively) at this site.

Variation in other amino acids believed to play a role in ADPglucose binding in AgtGS, may indicate specificities that help define the individual SS classes (Table 2). For example, in AgtGS, van der Waals forces among Ser298, Ser359 and Tyr354 are thought to stabilize the adenine heterocycle when ADPglucose binds the enzyme [43]. In the Group A SSs, glycine (aliphatic) is substituted for the polar Ser298 in Motif V (Figure 10). It has also been predicted that the hydroxyl (-OH) side chain of threonine at residue 381 (Thr381) would form a hydrogen bond with the O₂ atom of the ribose moiety of ADPglucose [43]. However, in all of the GBSSIs analyzed, Thr381 is replaced by isoleucine, which has an aliphatic hydrophobic side chain and cannot form hydrogen or ionic bonds with other groups. Members of the GT5 family have two highly conserved glycines (Gly327 and Gly329 in AgtGS) in Motif VI (Figure 10). Interestingly, the G-X-G motif is not conserved in the GT3 family that uses UDPglucose [43]. The Gly329 residue is replaced by serine (polar) in the SSIV sequences and such a substitution would likely affect the functional properties of the protein. In AgtGS, His163 and Phe167 are involved in glycogen binding [43] and would be predicted to be found in the GT-5 domain of SSs. However, a previous study stated that His163 is not present in SSs [42]. Our analysis shows that His163 is indeed present and highly conserved among SSs (Figure 9 and Table 2). In contrast, the equivalent Phe167 residue in the SSI and SSII sequences examined are basic (His318 and His380, respectively) and not the aromatic amino acids found in AgtGS, the Group B SSs and GBSSIs (Figure 9, Table 2). The only exception we found within the SSII family was OsSSIIa, where the conserved tyrosine (aromatic) is evident.

Relationship between genomic organisation and location of key amino acids

It has been suggested that if the splicing site of a particular intron coincides at the same amino acid position among members of a group of orthologous proteins, then there is ancient relatedness and conservation of functionality for that amino acid residue [47–50]. The basis for this idea is that proteins are built up through modular exonic segments spliced together using introns and if an amino acid serves an important or critical functional role, these residues will remain relatively unaffected by evolutionary changes in splicing and/or exon re-arrangements.

The positions of introns relative to the positions of specific amino acids were compared within the SS subfamilies to look for evidence of evolutionary conservation. In most cases, these positions are conserved among orthologous SS isoforms (Figure 12). However, two consecutive intron positions are conserved among all Group B SSs, and one position is identical for the SSI and SSII sequences examined (Figure 9). Both of these occurrences are within regions of interest within the predicted amino acid sequence. The conserved glutamine identical in SSI and SSII is within the 380s Loop between Motifs I and II (Figure 12), and the consecutive positions for SSIII and SSIV (Group B SSs) occurs within Motif I at the variable residue in the K-X-G-G-L motif between the lysine and valine residues (Figure 8). The amino acid sequences surrounding the conserved splicing positions are shown in blue boxes in Figures 8 and 9.

Gene organisation and chromosomal location

The intron-exon arrangement of the TaSSIVb genomic clone isolated (Figure 3) was similar to orthologous genes from rice and Arabidopsis. Of interest was the sequence of intron-4 which was highly similar between orthologues. This intron is in a region that when spliced and translated, produces the SSIV-specific N-terminal region. Introns are generally neutral to selection and undergo rapid change during evolution [49], therefore high sequence similarity between orthologous introns indicate a functional constraint during evolution [52].

The chromosomal location of TaSSIVb was also determined. It is positioned on the long arms of homeologous group I chromosomes and is syntenic with the OsSSIVb gene which maps to Chromosome 5. This differs from the location of many of the genes involved in starch biosynthesis in wheat which co-localise with homeologous group VII chromosomes [36] and to a lesser extent the group II chromosomes [41]. In contrast, TaSSIVb and TaSSIIIa are on chromosome I, perhaps highlighting the distinct evolutionary origin of these genes as compared to the other SSs [53]. However, TaSSIVb and TaSSIIIa map to opposite arms of chromosome I therefore chromosomal location offers few clues as to the evolutionary history of these two genes relative to each other.

Expression analysis

The expression of TaSSIVb mRNA was highest in non-endosperm tissue (Figure 5B) and showed fewer marked changes in response to varying light-dark conditions as compared to TaSSIIIa (Figure 6B). The TaSSIVb gene may be classified as a "steady expresser" [39]. If protein amount and activity reflects message levels then the TaSSIVb has a housekeeping role in starch biosynthesis in wheat endosperm as its transcript accumulation levels did not coincide with the period of high carbon flux to starch in the endosperm.

A circadian clock does not regulate TaSSIVb or TaSSIIIa transcript levels (Figure 6B). The lack of appropriate elements in the promoter regions of the rice homologues OsSSIVb and OsSSIIIa support this result. Induction of expression of TaSSIIIa and TaSSIVb during the light period may be a response to sugar accumulation as observed in IbGBSSI and OsGBSSIb [54, 55]. In our hands, TaGBSSIb was not circadian clock-regulated which is at odds with the expected expression pattern based on the presence of the relevant cis-element in the OsGBSSIa homologue. However, Vrintren et al. [39] reported that TaGBSSIb expression was not only independent of a circadian clock, but also non-responsive to light.

Sequence analysis

The second aim in this work was to advance existing knowledge of structure-function relation of starch synthases including the SSIV gene cloned. Known motifs in AgtGS predicted to play a role in catalysis and/or binding were compared with SSs analysed. If these sequences are invariant among SSs from diverse species, this may indicate that the corresponding amino acids may function similarly, as conserved sequence motifs suggests an evolutionary relationship between proteins. Structural analysis and alignment of the predicted amino acid sequences of SS and GSs revealed several amino acids potentially crucial in determining specific SS activity. These residues are presented in Table 1. By extension, amino acids within conserved motifs that are distinct between SS isoforms may at least be partially explained, by the important differences in functionality. These will be discussed in turn.

Proposed role for 380s Loop

There is evidence to suggest that some SSs can catalyse the initiation and elongation of glucan chains. AgtGS is able to prime glycogen synthesis independent of a glycogenin protein and Roldan et al. [11] drew a comparison between the priming functions of AgtGS protein and that of SSIV in Arabidopsis based on the phenotype of the AtSSIV mutant. Indeed, some SSs are able to prime starch synthesis in vitro [28]. To explain this possible self-priming function a 2-point insertion mechanism was elaborated by Guan & Keeling [57] whereby an SS would bind two ADPglucose molecules at different sites and α-1,4-glucan chain synthesis would occur de novo by the transfer of the glucose moiety of one ADPglucose to the other. Considering the position of Lys15 and the conserved Gly18 in Motif, I far from the catalytic core of the enzyme, it is initially difficult to see how they could determine primer specificity. However, these residues are invariant in the SS/GS family, suggestive of a functional role. One possibility is that after ADPglucose binding at the active site within the cleft or ADPglucose-binding pocket, the closed structure which results brings the α-1 and β-1 sheets containing the K-X-G-G-L motif and any primer it binds, e.g. another ADPglucose molecule or glucan chain, into proximity of the active site. If the 380s Loop functions in SSs in the way proposed for phosphorylases [42, 58–60], then the movement of this Loop during the open-close transition after ADPglucose binding may be an important part of the mechanism that drives the specificity of action of the SSs within the catalytic region. This mechanism would not require that the 380s Loop directly bind a carbohydrate. The only region that is consistently different among all SSs is the 380s Loop and these differences may possibly be translated, although indirectly, into different interactions with substrate and/or glucan primer.

Although it is suggested here that the 380s Loop could have a role in determining SS specificity, it does not rule out the possibility that other regions of SSs may also contribute or may be more important for SS functionality. In this study we focused our comparisons exclusively on the two-lobes of the C-terminus containing the GT1 and GT5 domains. It has been established that this region is responsible for catalytic activity and also, differences in substrate preference, affinity and reaction velocity between SSs [15, 23, 24]. However the function of the N-terminus in potentially defining SS activity is not yet known. This region may help to confer SS unique action by influencing or facilitating their interaction with other starch biosynthetic enzymes and/or substrate [29, 30]. The formation of starch biosynthetic enzyme complexes especially could modify the activity of each SS isoform and their relative contribution to starch glucan assembly.

Evolutionary relationship between cyanobacterial GSs and SSs

Analysis of algal SSs and cyanobacterial GS proved informative in helping to understand the function and relationship between the classes of these enzymes. Of particular interest, Ostreococcus tauri does not have SSIV but three isoforms of SSIII [6]. Granule initiation does not seem to be essential in Ostreococcus as this alga partitions its starch granules from mother to daughter cells by binary fission [6]. If SSIV is part of the biological mechanism involved in granule initiation [11], it is also possible that OtSSIIIa performs a similar function. Sequence analysis of Synechocystis PCC 6803 GSs also clarified our view of SS relationships. Synechocystis has two forms of GS, one of which, SpGS2, is closely related to the Group B SSs while the other, SpGS1, is more similar to the Group A SSs (Figure 2). Interestingly, the C-terminal extension found only in GBSSI polypeptides is also present in SpGS1. This raises intriguing possibilities about the evolutionary relationship between SpGS1 and the GBSSIs, and, when it is considered that all members of the Group A SSs have the ability to bind to the starch granule, the functional association of SpGS1 with the Group A SSs as a whole.

Phylogenetic analysis of the origins of starch biosynthetic genes in higher plants and green and red algae suggests there to be a complex set of evolutionary events with vertical and lateral gene transfers between host and endosymbiont cells [35]. In spite of this it is widely accepted that genes from a cyanobacterial endosymbiont are now present in plant host nuclei [61, 62] and some of these genes have retained much of their sequence identity. The clear separation of GSs into two distinct clades, and sequence similarities between them and SSs from one group or the other, suggests that the Group A and Group B SSs evolved directly from the two independent GS types. The SSIV group could have evolved from SpGS2, and subsequent gene duplication events might have led to the evolution of SSIIIs. One exon each in the wheat and Arabidopsis Class III SSs (exons III and I, respectively), contains a significant proportion of the coding region that is present in SSIII but not SSIV or other SSs [53]. It is possible that the Class III SSs arose as a result of duplication of the SSIV followed by a single insertion event of a gene sequence represented by these exons. The resulting difference in gene structure, as well as subsequent changes selected for during evolution, might have led to two distinct gene products that are sufficiently diverged to constitute two different gene classes. High sequence similarity of intronic sequences and intron-amino acid positions is supportive of the notion that SSIII and SSIV followed a common evolutionary path.

Plant material

Wheat (Triticum aestivum L. cv. Hi-Line) plants were grown in a greenhouse at a minimum temperature of 15°C, with supplemented lighting to give a photoperiod of 16 h during the winter. Genomic DNA of Chinese Spring nullisomic-tetrasomic lines, were a kind gift from Dr. Petra Wolters (DuPont-Pioneer, Crop Genetics) and Dr. David Benscher (Department of Plant Breeding, Cornell University). We thank Prof. Jon Raupp (Kansas State University) for seeds of Chinese Spring ditelosomic lines.

Materials

ADP-[U-¹⁴C] glucose was purchased from Amersham Pharmacia (Piscataway, NJ). Dowex 1 × 8 resin, ADPglucose and amylopectin were purchased from Sigma (St. Louis, Missouri).

RNA extraction

RNA was extracted from various tissues of wheat using TRIzol reagent (Invitrogen, Carlsbad, CA). Tissues included embryos from caryopses harvested at 25 DPA, endosperm (including the aleurone layer) from caryopses harvested at approximately 4–7 DPA, 5–10 DPA, 10–16 DPA, 16–20 DPA, 20–25 DPA, 26–32 DPA, whole caryopses harvested at 3, 7 and 14 DPA, and leaves, roots and the aerial parts of 14-day seedlings. Poly (A)+ RNA was purified from total RNA using Pharmacia QuickPrep RNA purification kit (Amersham Pharmacia Biotech, Inc., Piscataway, NJ).

cDNA library construction and EST analysis

First-strand cDNA synthesis was catalysed by Superscript II Reverse Transcriptase (Invitrogen, Carlsbad, CA) with a nucleotide mixture containing 5-methyl dCTP. The cDNA library was ligated 5'→3' into an EcoRI/Xho I-digested pBluescript SK+ vector according to the manufacturer's protocol (Stratagene, Cedar Creek, TX) and maintained in Escherichia coli DH10B cells (Life Technologies, Carlsbad, CA). The SSIV EST [Genbank:BT009276] was identified by sequencing randomly picked bacterial colonies from a developing wheat seedling cDNA library made using a previously described method [63].

Generation of the full-length SSIVsequence by 5'RACE of wheat cDNA libraries

DNA was prepared from each wheat library using a DNA miniprep kit (Qiagen, Valencia, CA). An aliquot of DNA (1 ng.ul^-1) was used as a template in separate PCR reactions with 10 μM each of T3 vector primer (5'-GCC AAG CTC GGA ATT AAC CCT CA-3') and a gene-specific primer complementary to the 5' end of EST [Genbank:BT009276]. PCR amplification was done with GC-Advantage polymerase mix (Clontech, Palo Alto, CA). Cycling parameters were 94°C for 1 min; 10 cycles of 94°C for 30 s, 68°C for 30 s, 72°C for 4 min, then 25 cycles of 94°C for 30 s, 63°C for 30 s, 72°C for 4 min, and an extension time of 72°C for 7 min in a PE 9700 Thermocycler (Perkin Elmer, Norwalk, CT). The products of the PCR reaction were diluted 1:100 and the DNA re-amplified using a nested primer and the vector primer SK-long (5'-GCC GCT CTA GAA CTA GTG GAT CCC CCG GGC TGC AGG A-3'). PCR products were gel-purified (Qiaquick, Qiagen, Valencia, CA) and subcloned into pCR 2.1 (TOPO-TA, Invitrogen, Carlsbad, CA) or pGEM-T (Stratagene, Cedar Creek, TX) vectors. At least four transformants containing an insert were selected and sequenced in both directions using an ABI 3700 capillary Sequencer with the ABI BigDye terminator chemistry (The PE Corporation, Foster City, CA). The longest PCR product was used to design primers for additional 5'-RACE. The antisense oligonucleotide primers used in conjunction with a vector primer to amplify contiguous sequences of the entire cDNA were (please refer to Figure 1): for Sequence A (nucleotides 1 to 1023), 5'-GCA GAT CAT GAT TGT GGT CCA ATA-3' and 5'-AGC ATG CTC TAC TTG GTT TGC TG-3'; for Sequence B (nucleotides 991 to 1778), 5'-AAC AAA TGC AGT TTG CCA GTC A-3' and 5'-CTG GAT GCT GTG GCT CAA TAA A-3'; and for Sequence C (nucleotides 1658 to 2194), 5'-TGC AAG GGT TCC ATG TAT CTG TG-3' and 5'-CCC TCT GAG CGA ACC TCT AGT GC-3'. To amplify fragment 1600 bp-RACE, primers 5'-CGC CCA AGC AGA TAC GAT GAA A-3' and 5'-TCA TAC TTC AAA ATC AGC CGG A-3' were used. To amplify fragment 2100 bp-RACE, the primers were identical to those used to amplify Sequence C and the PCR conditions similar except that Advantage polymerase mix was used and the nested PCR was done using a 1:200 dilution of the primary PCR reaction as template.

Northern and Southern hybridisations

Northern analysis was performed with 0.5 μg poly (A+) RNA extracted from developing wheat caryopses, separated on a 1.4% formamide-formaldehyde agarose gel and blotted onto HyBond N+ (Amersham Pharmacia Biotech, Piscataway, NJ) according to the manufacturer's protocol. The cDNA probe was a 550 bp fragment from a PstI and EcoRI digest of clone 1600 bp-RACE in TOPO-TA (Invitrogen, Carlsbad, CA). The probe was labelled using RadPrime DNA Labelling System (Invitrogen, Carlsbad, CA). Hybridisation was done using PerfectHyb Plus solution (Sigma, St. Louis, MO) at 65°C according to the manufacturer's protocol. The membrane was washed twice for 20 min in 2 × SSC and 1% (w/v) SDS at 65°C, and five times for 20 min in 0.1 × SSC and 0.1% (w/v) SDS at 65°C, until the background radioactivity was almost zero. Southern analysis was performed with genomic DNA extracted from leaves of Triticum aestivum cv. Chinese Spring nullisomic-tetrasomic and ditelosomic lines using standard techniques. A 20 μg aliquot of each sample was digested with EcoRV and resolved on a 0.8% agarose gel. Blotting, hybridisation and washing of the filter was as described for RNA analysis.

mRNA detection using comparative Reverse Transcription-PCR

Reverse transcription of wheat RNA, (treated with DNase I – RNase-free, Ambion, Austin, TX) was performed with 0.1–1.0 μM of antisense gene-specific oligonucleotide primer (MWGBiotech, High Point, NC) at 42°C for 30 min with an RNA-PCR kit (Roche Molecular Systems, Branchburg, NJ). PCR was done using the reverse transcription reaction as a template and Advantage-2 GC Polymerase mix (Clontech, Palo Alto, CA) with the addition of the sense primer, using the following PCR conditions: TaSSIIIa and TaSSIVb, 94°C for 30 s, 55°C for 60 s, 72°C for 90 s for 25 cycles and for TaGAPDH and TaGBSSIb, 94°C for 15 s, 59°C for 30 s, 72°C for 60 s for 22 cycles. Each cycle had a final extension of 72°C for 7 min. All reactions were done in a PE 9700 Thermocycler. The primers used to amplify the SSIV transcript were 5'-GATACAAACGGCACTGGACGAA-3' and 5'-CTCCATTTTTAGCTCGCGTTCT-3'. To amplify TaSSIIIa [Genbank:AF258608] the primers used were 5'-AATGGGATGTTTGGCGTTGG-3' and 5'-GTATAAGGGACCGGGATAAAA T-3'. To amplify TaGBSSIb [Genbank:AF109395] the primers used were 5'-GTTCTGCCTTTTGTGCCTTGCT-3' and 5'-GCAGCATCAGTTCCTCCTCCAT-3. To amplify TaGAPDH [Genbank:AF251217] the primers used were 5'-TCTCCAACGCTAGCTGCACCAC-3' and 5'-GGAAGTCAGTGGAAACAAGGTC-3'. Primer pairs were used at a final concentration of 0.2 μM with the exception of the TaGAPDH pair, which was 0.1 μM in the PCR reaction. The amounts of all products responded linearly to RNA input and were as follows: TaSSIIIa, 50 ng and for TaSSIVb, TaGAPDH and TaGBSSIb, 400 ng. Each reaction was repeated at least four times. Ethidium bromide-stained PCR products were visualized with an Eagle-Eye II Imager (Stratagene, La Jolla, CA). Each PCR product was sequenced completely to confirm its identity. A product was not observed in RNase-treated controls, or when RNA was used as a template in a PCR reaction.

Isolation of genomic DNA

The SSIV genomic clone was isolated by using PCR on genomic DNA from Triticum aestivum leaves (line N1BT1D) using specific primers designed from the cDNA sequence [Genbank:AY044844]. Two Taq polymerase enzymes (and their corresponding buffers) were used: Expand Long template PCR system (Roche Molecular Systems, Branchbury, NJ) and Advantage-2 GC Polymerase mix GC-2 (Clontech, Palo Alto, CA). Each PCR reaction contained 250 ng genomic DNA, 0.3 mM of each primer and 0.25 U of Taq DNA polymerase. Amplification was done with 1 denaturation cycle at 94°C for 2 min, 30 cycles of 20 sec at 94°C, 30 sec at 62°C and 4 min at 68°C, followed by a final cycle of extension at 68°C for 10 min in a Techne Gradient PCR Cycler (Techne Inc., Burlington, NJ). The PCR reaction was loaded and separated on a 0.8% (w/v) agarose gel. The visualized bands were extracted from the gel and the purified fragments cloned into the TOPO-XL vector (Invitrogen, Carlsbad, CA) and sequenced as described earlier.

DNA and protein sequence analysis

Computer analyses of all cDNA sequences were done using the program VectorNTI (Invitrogen, Carlsbad, CA). Similarity searches of nucleotide and protein sequences were performed with BLASTP using the non-redundant database at NCBI [64] Sequence alignments were accomplished using ClustalW [34] on the European Bioinformatics Institute's server and subsequent phylogenetic analyses were computed with MEGA3 [65]. The conserved domains within the predicted amino acid sequences were identified using the Pfam database [66] ProDom [67] and Simple Modular Architecture Research Tool (SMART) [68–70]. Prediction of the structural relationship between protein sequences of unknown 3D structure against proteins with a known 3D structure were made using SAM-T02 [71, 72]). The Jnet server was used to predict secondary structure [45, 46]. Protein models were constructed using T.I.TO: Tool for Incremental Threading Optimization at [73, 74].