Novel exon combinations generated by alternative splicing of gene fragments mobilized by a CACTA transposon in Glycine max

Background The recent discoveries of transposable elements carrying host gene fragments such as the Pack-MULEs (Mutator-like transposable elements) of maize (Zea mays), rice (Oryza sativa) and Arabidopsis thaliana, the Helitrons of maize and the Tgm-Express of soybeans, revealed a widespread genetic mechanism with the potential to rearrange genomes and create novel chimeric genes affecting genomic and proteomic diversity. Not much is known with regard to the mechanisms of gene fragment capture by those transposon elements or the expression of the captured host gene fragments. There is some evidence that chimeric transcripts can be assembled and exist in EST collections. Results We report results obtained from analysis of RT-PCR derived cDNAs of the Glycine max mutant flower color gene, wp, that contains a 5.7-kb transposon (Tgm-Express1) in Intron 2 of the flavanone 3-hydroxylase gene (F3H) and is composed of five unrelated host gene fragments. The collection of cDNAs derived from the wp allele represents a multiplicity of processed RNAs varying in length and sequence that includes some identical to the correctly processed mature F3H transcript with three properly spliced exons. Surprisingly, the five gene fragments carried by the Tgm-Express1 were processed through complex alternative splicing as additional exons of the wp transcript. Conclusion The gene fragments carried by the Tgm inverted repeat ends appear to be retained as functional exons/introns within the element. The spliceosomes then select indiscriminately the canonical intron splice sites from a pre-mRNA to assemble diverse chimeric transcripts from the exons contained in the wp allele. The multiplicity and randomness of these events provide some insights into the origin and mechanism of alternatively spliced genes.

host gene fragments. It carries intronic and exonic regions of five genes: unknown protein (UP), cell division cycle 2 (CDC2), fructose-6-phosphate 2-kinase/fructose-2-6biphosphatase (FPK), malate dehydrogenase (M) and cysteine synthase (CS) [1]. Little is known about how any of those transposons or retroelements acquired the gene pieces but there is some evidence that they are transcribed creating chimeric cDNAs that exist in EST (expressed sequence tag) collections [4][5][6][7]. Of 475 Pack-MULEs identified on chromosomes 1 and 10 of rice via computer searches, 5% were transcribed based on exact matches to full-length cDNAs [4]. Most of the transcripts (91%) were initiated from promoters at the TIRs (terminal inverted repeats) or within the element while 9% of the transcripts initiated outside the element. Three chimeric transcripts were seen in an RNA blot probed with both a sh2 and a Helitron insertion fragment probes [8]. A single 2,620 bp chimeric transcript spanning the entire Helitron including several gene fragments contained within the element has also been described [6]. The promoter was predicted to reside upstream of the Helitron insertion site.
We present here an array of 12 distinct alternatively spliced chimeric transcripts obtained via RT-PCR that were derived from the soybean wp mutant allele in which the second intron of the F3H1 gene is interrupted by the 5.7 kb transposon containing five captured host gene-fragments. The chimeric transcripts analyzed were more abundant in seed coats than in cotyledons and ranged in size from 3,108 bp to a correctly processed one of 1,422 bp that was identical to the transcript derived from the wild type Wp allele. All transcripts isolated initiated at the F3H1 gene (Wp, wp) promoter that is strongly expressed in seed coats.
Alternative splicing is a common regulatory mechanism in higher eukaryotes and the mechanisms governing it have been studied extensively in mammalian systems but sparingly in plants [9,10]. In general, the splicing pattern of a multiexon pre-mRNA can be altered in many ways. Exons that are always spliced and included in the mature mRNA are known as constitutive exons. However, mechanistic decisions of the splicing components can result in exons that are included at times but excluded at others times from the mature mRNA. These are referred to as cassette exons. There are also occurrences of 5' and 3' alternative splice sites altering the length of some exons. In addition, the failure to remove an intron, referred to as intron retention, is also found. Genes whose pre-mRNAs have multiple locations of alternative splicing produce a family of related proteins with different allosteric regulation, protein localization, or enzymatic activity [9].
We show that the exon/intron regions of gene fragments carried by the Tgm-Express1 of the wp allele are alterna-Illustration of the effect of wp on flower and seed coat phenotypes or stable pink flower of plants with wp genotype (right panel) in lines LN89-5320-6 (i i RtW1Wp) and LN89-5322-2 (i i RtW1wp) both of which have yellow seed coats. In soybean I (CHS), R and T (F3'H) are three independent loci that control pigmentation in seed coats and W1 (F5'3'H) and Wp (F3H) were described as flower color markers, but all five loci seem to be encoding genes of the anthocyanin and proanthocyanidin pathways. Mutant alleles of those loci (i, i i , r, t, w1 and wp) affect flower, seed coat, hypocotyle or pubescence coloration [1,28,30,31]. (B) Imperfect black color of seed coats of plants with iRtW1Wp genotype (left panel) as contrasted with the lighter shaded seed coats of plants with iRtW1wp genotype (right panel). Effect on the seed coat phenotype was revealed by crossing the wp allele into lines having the recessive i allele that allows spatial pigmentation of the entire seed coat [24]. The cracks on both seed coat types result from an epistatic effect of t [31]. (C) Black seed coats of plants with iRTW1Wp genotype where the T allele drives the synthesis of cyanidins (left panel) contrasted with the lighter seed coats of plants with iRTW1wp genotype (right panel). (D) Abbreviated schematic representation of the three branches leading to the synthesis of the three anthocyanin classes and the genes encoding the enzymes relevant to the present study.
tively spliced and assembled with the constitutive exons of the F3H1 gene to generate an array of chimeric transcripts encoding a variety of open reading frames (orfs). Analysis of the derived amino acid sequence from the 12 distinct wp chimeric cDNAs predicted putative chimeric orfs varying in length and frame locations.
The splicing machinery at times eliminates all extraneous (cassette) exons and introns of the Tgm-Express1 element to generate a full length transcript identical to that of the wild type gene and thus likely functional. The number of F3H molecules may be extensive enough to allow the synthesis of sufficient anthocyanin pigment that could account for the pink flower and the lighter seed coat phenotypes in the wp lines ( Figure 1). On the other hand, the more complex chimeric transcripts containing cassette exons (UP, CDC2, FPK, M and CS) from the captured gene fragments in the transposon may upon translation generate products that could interfere with function of the wild type host-gene counterparts leading to secondary phenotypes. Whether any of the novel exon combinations derived from alternative splicing of the mobile exons of the Tgm-Express1 element create new phenotypes is unknown. However, there is growing evidence from both plant [11,7] and animal [12,13] systems that repeat sequences derived from mobile elements play a significant role in generation and evolution of novel genes and exons. The wp locus in soybean is a unique example of an insertional mutation in the act of de novo generation of fused, multiple chimeric exons through inclusion or exclusion of cassette exons carried by the element into the affected gene.

Complex aberrant expression of the flower color mutant gene wp
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).

Seed Coats Cotyledons
A B + + -sequence alignment). There were exonic portions of five distinct genes: unknown protein (UP), cell division cycle 2 (CDC2), fructose-6-phosphate 2-kinase/fructose-2-6biphosphatase (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), 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 Schematic representation of the wp recessive allele and the novel exon combinations generated in its transcribed RNAs  Figure 2A) was also spliced out to form the FPK exon in the chimeric transcripts ( Figure 2B and C). A smaller FPK intron flanked by 15 bp exon fragment (narrow orange block not named) at the 5'end ( Figure 2A) is not always spliced out ( Figure 2B and C).

bp = 19 mm
Seed coats wp -RT -PCR cDNA clones 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.

Open reading frames of chimeric, multi-exon wp transcripts
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  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.

Expression of host genes with homology to the Tgm-Express1 captured gene fragments
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. The chimeric orfs were generated from several of the cDNAs sequenced and they are listed underneath each orf class and also the frame in each one of the clones. * The orf from the wp-6c clone was 5 aa shorter. ** The orf from the wp-6c was 2 aa longer. This probe contained sequence fragments with similarity to an unknown protein (UP), cell division cycle 2 (CDC2), fructose-6-phosphate 2-kinase/fructose-2-6biphosphatase (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.

Schematic of relevant chimeric and non-chimeric orfs gener-ated by the wp allele
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.

Discussion
Plant transposable elements have long been known to cause changes in gene expression as a result of insertion or deletion in coding regions and gene promoters and their effect on RNA processing [14]. However, more than 20 years of research on tracking the movements of individual transposable elements at the molecular level has revealed only a few examples that these elements were capable of transposing non-element associated coding regions from the host genome. A part of a host gene was carried by a Mutator (Mu) related element [15] and another by the Bs1 retroviral element [16]. More recently, with the availability of entire genome sequences for rice and Arabidopsis, many Mu-like elements associated with fragments of host cellular genes have been found in those two plant species [17,18,4]. Bioinformatics analysis of the rice genome uncovered 898 intact retrogenes of which 380 were predicted to have chimeral protein coding sequences and sev-eral of these were confirmed by expression data [7]. A novel family of maize transposons, the Helitrons, has been found recently to be embedded with portions of cellular gene fragments [8,19,5,6]. The Tgm-Express1, a member of the CACTA family, was shown to carry four genic fragments [1]. An additional exonic region, UP, was identified upon analysis of the RT-PCR cDNA sequences obtained in the present study. More significantly, we show also that the wp allele carrying the Tgm-Express1 spawns an array of chimeric transcripts resulting from alternative splicing events at this locus, some of which lead to novel open reading frames.
If one were to envision an evolutionary advantageous mechanism for the novo generation of mutations, it likely would be one that would alter the splicing of a gene's constitutive exons to produce novel proteins without the complete loss of the wild-type protein. The wp pink flower mutation of soybean that we describe offers an example of how a transposon insertion in the intron of the wild-type gene (F3H1) could generate an array of alternatively spliced transcripts with potential to be translated into novel proteins without totally losing the ability to synthesize the wild-type transcript and encoded protein.  tively spliced with F3H1 exons. Sequence analysis of multiple RT-PCR derived cDNAs from two different tissues, the seed coat and the cotyledon, revealed the processing of at least 12 distinct putative transcripts, 11 of which contained besides the F3H exons (1, 2 and 3) varying numbers of cassette exons with or without small host genefragment intronic portions. In addition, transcripts with only the constitutive F3H exons were also synthesized from the wp mutant allele. Although observations of element removal from maize exons [20,21] and introns [22] has previously been observed, the splicing of the very large wp intron of 6.44 kb (consisting of the 718 bp F3H intron 2 and the 5722 bp Tgm-Express1 transpon within it) is an unusual event. Plant introns are generally short and splicing of extremely large introns is rare [10].
In alternative splicing processes in other eukaryotic systems, splicing of the pre-messenger RNA required five ribonucleoproteins and numerous accessory proteins to form four complexes (E, A, B and C) to join two adjacent exons with the release of an intervening sequence [9]. It is remarkable to see the variety of processed transcripts that were created from the wp mutant allele leading to the precise splicing of the F3H and the cassette exons from within the Tgm-Express1 element ( Figure 3B and 3C, seed coat wp-4s, wp-15s and cotyledon wp-12c and wp-6c, for example). Of greater interest also is the proper processing of the three CDC2 and two FPK exons in some of the transcripts that results in fusion of unrelated exons.
Analysis of the derived amino acid sequence from the 12 distinct wp chimeric cDNA sequences predicted putative chimeric open reading frames (orfs) varying in location and length within the cDNAs (Figure 4). Alternative splicing of the cassette exons within Tgm-Express1 creates premature termination codons in many of the transcripts which could lead to their being targets for nonsense-mediated mRNA decay, the surveillance mechanism that degrades selectively nonsense mRNAs [23]. However the larger 5'-end chimeric orfs (seed coats wp-28s, -9s, and cotyledons wp-13c and wp-6c) may have a higher chance to be translated (Figure 4-A). This orf is 418 aa, 24 aa longer than the one like the F3H wild type orf of 394 aa (Figure 4-B).
The correctly processed transcript, represented by the isolated wp-15s clone (Figure 4-B), would allow the synthesis of functional F3H protein which could explain the existence of residual anthocyanin pigment in the pink flowers and the lighter colored seed coats in plants with wpwp genotype ( Figure 1). F3H function is a required step in the synthesis of all three branches of anthocyanin and proanthocyanidins ( Figure 1D). The results presented here demonstrate that wp is not a null mutation and that correctly spliced F3H1 transcripts are synthesized in sufficient amounts to allow the synthesis of anthocyanin pigment coloring the pink flowers ( Figure 1A) and seed coats of plants with the i, t, wp ( Figure 1B) and i, T, wp, ( Figure 1C) genotypes. The fact that wp is not a null mutation suggests that the residual anthocyanins being synthesized in the wp flowers are most likely the same as in the Wp purple flowered isolines. The only difference between the colors of the two phenotypes may be the amount of pigment being synthesized.
We cloned two similar F3H genomic sequences (F3H1 and F3H2) and found that the F3H2 gene is not expressed in the tissues discussed here [1]. Therefore, the residual flower and seed coat color-phenotypes displayed by wp allele (Figure 1  ). Since both the host genes and the wp chimeric transcripts appear to be weakly expressed in the tissues examined ( Figure 5 and data not shown), any polypeptide fragments translated from the putative orfs, might interfere with the assembly and function of their active host protein counterparts. Thus, the CDC2 and FPK polypeptide pieces translated from the wp mutant allele may have potential to interfere with the functional enzymes translated from intact CDC2 or FPK host genes. Intriguingly, the second phenotype manifested in the wp mutant plants is seeds with lower oil and higher protein content than the Wp plants [2,3]. Inhibition of key metabolic enzymes such as FPK could potentially influence the direction of metabolic flux resulting in decreased fatty acid metabolism and linked increases in protein synthesis.

Conclusion
The multiplicity of transcript isoforms described here add an additional layer of complexity reinforcing the tremendous potential these gene-fragment-loaded-transposon elements such as Tgm-Express, Pack-MULEs, Helitrons, and retroelements can have not only in disrupting or modifying gene function but in the creation of new or modified genes leading to an increase in plant genome and possibly proteome diversity. Analysis of human and other verte-brate genomes [12,13] revealed that recently evolved exons are more likely to be alternatively spliced cassette exons originating from highly repeated DNA elements including transposons, SINEs, LINEs, and Alu repeats, emphasizing the importance of mobile elements in creating diversity during evolution of both animal and plant species.

Plant material and genotypes
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 [2,3,25,1]. 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.

RNA extraction, purification and cDNA synthesis
Total RNA was isolated from seed coats and other soybean tissues using a phenol-chloroform and lithium chloride precipitation method [25,26]. RNA was stored at -70°C until used.

Primer synthesis, PCR reaction conditions, cDNA cloning and DNA sequencing
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.
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-32 P]dATP by random primer reaction [29].