We discovered that a pink flower locus (Wp) of soybean encoded a flavanone 3- hydroxylase gene (F3H1) by differential screening of a cDNA soybean microarray using RNAs from mutant pink (wp) and standard purple (Wp) flower isolines 1. We also showed that the Tgm-Express1 transposon insertion impaired expression of the mutated gene and that the F3H1 gene was strongly expressed in the seed coats but not in cotyledons 1.

Analysis of the wp allele expression by RT-PCR with a pair of F3H1 outermost 5' and 3'-primers revealed a bizarre pattern of amplification resulting in a variety of cDNA sizes from both seed coat and cotyledon RNAs (Figure 2). The broad bright band of PCR product obtained with the seed coat samples (Figure 2A) represents multiple size bands. Shorter exposure photograph of that same gel revealed at least 4 distinct bands (left most lane, Figure 2A). The wp transcriptional activity between the two tissues, cotyledons and seed coats, could be deduced from the difference in the intensity of the PCR products obtained from the two wp RNA sets (Figure 2A and 2B). Even though no hybridization to a F3H probe was apparent by RNA blots with cotyledon RNAs of either genotype (Wp and wp) 1, RT-PCR showed the existence of 1.4 kb transcripts representing the mature F3H1 gene (data not shown) and the aberrant larger transcripts from mutant line RNAs (Figure 2B).

Cloning the larger sized RT-PCR cDNAs from plants homozygous for the wp allele resulted in a surprising array of alternatively spliced transcripts. Sequence analysis of the multiple size cDNAs cloned from both seed coat and cotyledons revealed multiple transcripts derived from the wp allele containing the wild type gene (F3H1) exons (1, 2, 3) plus varying portions of exonic and intronic regions of the gene fragments captured by the Tgm-Express1 element that interrupts Intron 2 in the wp allele (Figure 3).

Figure 3A shows the schematic representation of the genomic sequence of the wp allele with the Tgm-Express1 insertion in Intron 2. The number of gene fragments contained within the element and their exons (solid colored boxes) and introns (striped boxes) were revealed upon sequence analysis of the multiple transcripts derived from wp expression in seed coats (See additional file 1: Seed coat wp RT-PCR cDNA sequence alignment) and cotyledons (See additional file 2: Cotyledons wp RT-PCR cDNA sequence alignment). There were exonic portions of five distinct genes: unknown protein (UP), cell division cycle 2 (CDC2), fructose-6-phosphate 2-kinase/fructose-2-6-biphosphatase (FPK), malate dehydrogenase (M) and cysteine synthase (CS). Some of the intronic regions could be assigned to specific genes (one color stripes) while others (two colors) could not. The solid black line between the solid arrow heads (inverted repeats) could be introns or intergenic DNA regions. Including the latter, all marked intronic regions conform to the canonical 5'GT donor and 3'AG acceptor splice sites.

Figure 3B is a graphic summary of seven distinct RT-PCR cDNA clones derived from seed coat RNAs. The larger clones (wp-25s, wp-22s, wp-28s, wp-9s and wp-12s) contain beside the three exons of the F3H1 gene, all cassette exons of the Tgm-express1 gene fragments and varying intron pieces. The smaller clones (wp-4s and wp-15s) had only the F3H exons (wp-15s) or the F3H exons and three cassette exons (wp-4s). Sequence data from these clones have been deposited with the EMBL/GenBank Data Libraries under accession numbers: EF100865 (wp-25s), EF100866 (wp-22s), EF100867 (wp-28s), EF100868 (w-p9s), EF100869 (wp-12s), EF100870 (wp-4s), EF100871 (wp-15s).

Likewise, Figure 3C shows six cDNA clones obtained via RT-PCR from cotyledon RNAs. As in the case of the seed coat derived cDNA clones, the larger cotyledon cDNA clones (wp-9c, wp-8c, wp-2c and wp-13c) contained some intron fragments besides the three F3H exons and cassette exons from the Tgm-Express1 element. The smaller clones (wp-12c and wp-6c) contained only exons, the three F3H exons and the five cassette exons correctly spliced. The latter two clones diverged only by 61 bp mostly due to two splicing errors in wp-6c deleting 15 bp at the beginning of Exon 2 and 47 bp at the CDC2/FPK exons junction (See additional file 2: Cotyledons wp RT-PCR cDNA sequence alignment). Of the six cotyledon cDNAs cloned, only one (wp-8c) was identical to one (wp-22s) of the seed coat cDNA clones. Sequence data from these clones have been deposited with the EMBL/GenBank Data Libraries under accession numbers: EF100872 (wp-9c), EF100873 (wp-8c), EF100874 (wp-2c), EF100875 (wp-13c), EF100876 (wp-12c), and EF100877 (wp-6c).

Overall, we isolated 12 different transcripts synthesized from the wp allele. These are a good representation of the chimeric transcripts generated by the spliceosome machinery in the tissues examined. We conclude that the most abundant transcripts shown by the discrete bands in figure 2A (left lane) have been cloned based on their size. The four most abundant bands are between 2 and 3 kb in size as are 11 of the 12 different cloned cDNAs. Our results also demonstrated that alternative splicing at the wp allele occurs in two tissues, one (the seed coats) in which the F3H promoter is highly expressed and another (cotyledons) in which it is not.

The amino acid sequences derived from the cDNA sequences of seed coat and cotyledon wp-cDNA clones shown in Figure 3B and 3C, varied significantly from clone to clone and consequently the putative open reading frames (orfs) of these chimeric transcripts. A search for orfs consisting of more than 100 amino acids (aa) found that many were chimeric (Figure 4 and additional file 3: Seed coat wp RT-PCR cDNA derived amino acid sequences and open reading frames, and additional file 4: Cotyledon wp RT-PCR cDNA derived amino acid sequences and open reading frames). Of interest were two putative chimeric orfs present in several of the wp-cDNAs. One was composed of approximately 210 bp (70 aa) of the UP exon fragment and 192 bp (64 aa) of the CDC2 exon fragment. It was present in seed coat transcripts wp-25s, -22s, -28s, -9s, -12s, -4s, and cotyledon transcripts wp-9c, -8c, -2c, -13c, -12c, -6c, always in a (+) frame. In half the clones this orf appears as just described (seed coat wp-25s, -22s, -12s and cotyledon wp-9c, -8c, -2c) (Figure 4-D). In the other half the orf is part of a larger chimeric orf containing also the F3H1 Exon 1 and Exon 2 sequences (Figure 4-A and 4C) (seed coat wp-28s, -9s, -4s and cotyledon wp-13c, -6c, -12c). A second chimeric orf predicted for five of the wp-cDNAs (seed coat wp-25s, -22s, -9s and cotyledon wp-9c, -8c) consisted of approximately 103 bp (35 aa) of FPK/M intronic fragment and 212 bp (70 aa) of FPK exonic region, always in one of the three (-) frames (Figure 4-E). A related chimeric orf containing FPK exonic sequence of approximately 231/199 bp (77/66 aa) appears in seed coat wp-12s clone and cotyledon wp-13c and wp-6c, all three in a (+) frame (Figure 4-F).

The products of these chimeric orfs may not serve enzymatic functions per se but if translated, they could potentially affect the function of wild type proteins synthesized from the intact host genes (UP, CDC2 or FPK). In addition to the chimeric orfs, we also found a cDNA that reconstituted the F3H1. The seed coat wp-15s cDNA clone in the (+3) frame contained an orf of 394 amino acids identical to the orf derived from the functional allele F3H1 of the purple flower isoline (Wp) (Figure 4-B). The product of this F3H1orf has the full potential to be translated into a functional F3H enzyme.

To analyze the expression of the host genes related to the exons captured by the Tgm-Express1 element, we amplified the cassette exons from the seed coat derived wp-12 cDNA clone (Figure 3B) to generate a chimeric radiolabeled probe that would hybridized to all RNAs with homology to the probe's exon fragments. These include those transcripts derived from the related host genes as well as the chimeric transcripts expressed from the wp mutant allele. This probe contained sequence fragments with similarity to an unknown protein (UP), cell division cycle 2 (CDC2), fructose-6-phosphate 2-kinase/fructose-2-6-biphosphatase (FPK), malate dehydrogenase (M) and cysteine synthase (CS) genes.

Figure 5 shows the hybridization of the chimeric radiolabelled probe to an RNA blot containing flower bud, seed coat and cotyledon RNAs from the wp, Wp and wp^m isolines. It appears that the hybridization signal in the seed coats from the wp line that contains the Tgm-Express1 insertion is greater than the signal in the Wp line that lacks the insertion, suggesting that the chimeric fragments within the wp allele are detected in the RNA blots along with hybridization to the host genes. In cotyledons on the other hand, hybridization levels to RNAs from both isolines Wp and wp were similar. Since we have previously shown that the F3H1 promoter is strongly expressed in seed coats but with practically undetectable expression in the cotyledons by RNA blot analysis 1, we deduced that the hybridization seen with cotyledon RNAs is likely to represent expression from one or more of the host genes. We have RNA blot data showing transcripts hybridizing to a soybean FPK cDNA clone (Gm-c1023-5325) of a size similar to those of the chimeric transcripts and that were more highly expressed in cotyledons of all three isolines than in seed coats (data not shown). This would explain the hybridization results observed for the cotyledons of the Wp where no chimeric transcripts should be synthesized.

In addition, the hybridizing RNAs shown in Figure 5 are large transcripts of size similar to those recovered by RT-PCR suggesting that most likely there are no internal transcription initiation sites within the Tgm-Express1 element to generate truncated smaller transcripts.

The Glycine max cultivars and isolines used for this study were: LN89-5320-6 (Wp, purple-flowered plants), LN89-5322-2 (wp, a stable line with pink flowered plants), and LN89-5320-8-53 (wp^m, a mutable line with chimeric flower colors or sectors of pink and purple flowers). Each is homozygous for the indicated alleles of the Wp locus. The origin, genetics, and isolation of the wp allele have been described previously 23251. Plants were grown in the greenhouse. Seed coats dissected from seeds at varying stages of development, cotyledons of various stages of seed development, flower buds, stems, mature leaves and roots from two week old plants, and shoot tips (meristems surrounded by primordial leaves) were frozen in liquid nitrogen, freeze dried (Multi-dry lyophilazer; FTS systems), and stored at -20°C. For seed coat's (cotyledon's) developmental stages, seeds were divided into the following groups according to the fresh weight of the entire seed: 10–25 mg, 25–50 mg, 50–75 mg.

Total RNA was isolated from seed coats and other soybean tissues using a phenol-chloroform and lithium chloride precipitation method 2526. RNA was stored at -70°C until used.

cDNA copies of the F3H1 gene from the three isolines (LN89-5320-6, LN89-5322-2 and LN89-5320-8-53) were amplified from a first-strand cDNA pool synthesized using 1 μg of seed coat or cotyledon total RNA and the Superscript first strand synthesis system for reverse transcriptase (RT)-PCR with oligo (dT)_12–18 primers (Invitrogen, San Diego). The total RNAs used for these RT-PCR reactions were treated with DNAaseI using Ambion's DNA-free kit and concentrated in Microcon YM-30 columns (Millipore, Bedford, MA). For each RNA sample, parallel reactions were allowed in the absence of Superscript (- controls) to assess the extent of DNA contamination. The sequences of the two primers used were: 5'-GCATTGCATTCTGCTATTTAATTCC-3' (7F) and 5'-AAAGACAGTGCCACTTATTTTCATT-3' (1428R). These primers map at the 5' and 3' ends of the F3H1 gene respectively. The numbering correspond to the base pair of the F3H1 gene cDNA sequence (Figure 3A)

Oligonucleotide primers were synthesized on an Applied Biosystems (Foster City, CA) model 394A DNA synthesizer at the Keck Center, a unit of the University of Illinois Biotechnology Center. PCR reactions were performed by an initial denaturation step at 94°C for 2 min followed by 30 cycles of denaturing at 94°C for 30 sec, annealing at 56°C for 1 min, extension at 68°C for 9 min, to end with a 10 min extension at 72°C. High-fidelity and -efficiency ExTaq (Takara Bio Inc. Otsu, Japan) polymerase was used at 0.75 units per 50 μl reaction. Amplified cDNAs were separated from oligonucleotides with a QIAquick PCR Purification kit (QIAGEN), cloned into pGem-T easy and sequenced in an ABI 3730 × l (Applied Biosystems, Inc. Foster City, CA) at the Keck Center.

RNA (10 μg/sample) was electrophoresed in a 1.2% agarose-3% formaldehyde gel 27. Size-fractionated RNAs were transferred to Optitran-supported nitrocellulose membrane (Midwest Scientific, Valley Park, MO) by capillary action as described in Sambrook et al. (1989) 27 and cross-linked with UV light (Stratagene, La Jolla, CA). Nitrocellulose RNA blots were prehybridized, hybridized, washed, and exposed to Hyperfilm (Amersham, Arlington Heights, IL) as described by Todd and Vodkin (1996) 28. All the RNA blot results presented are from autoradiographs exposed for 7 days.

Cloned DNAs used as probes were PCR amplified, electrophoresed, and purified from the agarose using the QIAquick gel extraction kit (QIAGEN, Valencia, CA). DNA concentration of the final eluate was determined with a NanoDrop (NanoDrop Technologies, Inc. Rockland, DE). Purified DNA fragments (25–250 ng) were labeled with [a-³²P]dATP by random primer reaction 29.