In a previous study in which the T locus of soybean was identified as the flavonoid 3'-hydroxylase gene (F3'H), we reported the partial genomic sequences of three F3'H alleles of the soybean lines, Williams 43 (T, tawny), XB22A (T*, tawny) and its gray trichome isoline, 37609 (t*). We postulated that a large intron existed in the 5' region of the F3'H genes and that it prevented PCR amplification of the contiguous full length F3'H genomic sequence using standard methods. In order to obtain the full sequence of soybean F3'H, we isolated a BAC clone from cultivar PI437654 with the T/T genotype. The 64 kb of BAC sequence (to be detailed later) revealed a 4.3 kb intron in the F3'H gene. This information was utilized to synthesize primers intended to amplify the promoter and Intron-1 genomic regions of F3'H alleles in XB22A, 37609, 37643, 33745 and Williams 43 soybean lines (Table 1). The ~2 kb promoter region of F3'H in all those lines was identical although there were several differences from the PI437654 cultivar. The 4.1 kb Intron-1 sequence was also determined for the F3'H alleles of Williams 43 (T, tawny trichomes), XB22A (T*, tawny trichomes) and the stable gray-trichome isoline, 37609 (t*). The three F3'H full length genomic sequences: Williams 43 (T: 8,680 bp), XB22A (T*: 8,678 bp) and its isoline 37609 (t*: 29,609 bp) were entered in GenBank with accession numbers, EU190438, EU190439 and EU190440 respectively.

The dramatically differing sizes between the tawny (T and T*) and gray (t*) trichome F3'H alleles above was found to be due to a 20.5 kb insertion in Intron-1 in the 37609 mutant isoline (Figure 1). PCR amplification of the 20.5 kb insertion required the use of higher melting temperature oligonucleotide primers (Z37F and IN663R) and optimized conditions for LA Taq polymerase (TaKaRa Bio Inc.) as described in the Methods section. As shown in Figure 1, a large fragment of ~23 kb was amplified from genomic DNA from both of the stable gray trichome lines 37609 and 37643 as well as from the mutable tawny/gray trichome line, 33745 (Table 1). The same primers also amplified a smaller 2.6 kb fragment from the mutable 33745 line and those lines with the dominant alleles, Williams 43 and XB22A. The 2.6 kb fragment is the Intron-1 portion comprised between the two primers used in the PCR reactions. The insertion maps ~2.4 kb to the left of the IN663R primer (Figure 2).

The ~23 kb PCR fragment was purified from 1% seaplaque agarose gel with the aid of GELase agarose gel-digesting preparation (Epicentre Biotechnologies), cut with HindIII and cloned in a pGem vector. Fortunately, two of the resulting clones that hybridized to a labeled probe prepared with an aliquot of the GELase purified 23 kb fragment, with sizes 199 bp, and 3,620 bp turned out to be the insertion's borders, each containing portions of intron sequence, a CACTA inverted repeat and the adjacent three base target duplication (ATA).

A 37-mer reverse primer (TGM23R) designed to the insertion's right border found in the 3.6 kb clone was paired with the Z37F primer upstream of the insertion's left border to amplify a 17 kb DNA fragment (Figure 2). An attempt at cloning the17 kb fragment into the pJAZZ-OC vector using the BigEasy v2.0 linear cloning kit from Lucigen Corp. failed to clone it in one piece but many partial clones ranging in size from 1 to 7 kb were recovered. Five different clones were sequenced and assembled to reveal a 20,544 bp CACTA transposable element that was named Tgmt* (Figure 2). This sequence was entered in GenBank as part of the t* allele genomic sequence from the 37609 line (Acc. No. EU190440). Previous studies of soybean clones containing CACTA ends analyzed seven distinct sequences (Tgm1-Tgm7) ranging in size from 1.6 to 12 kb that were believed to be portions of deletion derivatives of a larger active element (Rhodes and Vodkin, 1985 and 1988). Only Tgm4 and Tgm5 sequences had a 1 kb segment with 39% similarity to the ORF1 of maize Spm/En transposable element tnpD gene. The larger Tgmt* element we have cloned in this report has the capacity to encode all known functions required of an active and autonomous CACTA element as will be described later.

Sequence analyses revealed features that are conserved among transposable elements of the Spm/En family. Tgmt* possesses nearly identical 13 bp CACTA inverted repeats with three reciprocal mismatches and features a three-base-target site duplication, ATA. It also contains asymmetric, reiterated direct and inverted sequence motif in the subterminal regions capable of forming 12 stem-loop structures in the right border and two in the left border (Figure 3). The largest stem-loop structures were formed by direct and inverted repeated sequence motif 17 bp long. This sequence motif could be divided in two sub-motifs, a 7 bp motif (TTGGCAG) present in all 14 stem-loop structures and a 10 bp motif (AATCTTACAG) that was missing or incomplete in some of the stem-loops. See an alignment of all direct stem repeated sequences (read from 3'-end to 5'-end) in Figure 4.

A similarly complex pattern of stem-loop structures was described for Tgm1 15. In this instance the length of the repeated sequence motif (17–20 bp) was more uniform (see Additional file 1: Alignment of Tgm1 subterminal direct repeats). The consensus sequence of these direct repeats from Tgm1 was compared to the one deduced from Tgmt*, Tgm-Express1 (25; see Additional file 2: Alignment of Tgm-Express1 subterminal direct repeats) and Tgmw4m (GenBank Acc. No.: EU068464; see Additional file 3: Alignment of Tgmw4m subterminal direct repeats). Surprisingly, the Tgm1 sequence motif diverged from all three others in 6 nucleotide positions highlighted in red in Figure 5. The subterminal direct repeat motif of En-1 is smaller (12 bp) and except for a TCTTA domain its sequence is different from the motif of the soybean Tgmt* transposon family (Tgm-Express1and Tgmw4m) analyzed. The extended sequence divergence of Tgm1 is intriguing and suggests the existence of a second CACTA transposon family in soybean. This variability of subterminal repeats extends to other motifs reported. Figure 5 shows an alignment of subterminal repeats from 12 transposon sequences emphasizing both, their divergence and slight commonality. They all seem to have a GC rich and an AT rich portion. Whether or not these are recognition sites for the DNA-binding proteins remains to be determined.

Softberry- FGeneSH gene prediction web-based algorithms using Medicago truncatula as training set http://www.softberry.com found one gene with 21 exons in the negative DNA strand relative to the F3'H gene (t*) where Tgmt* inserted. Thus, all sequence analysis description and depiction of Tgmt* Gene-1 is that of the reverse complement. The predicted Gene-1 first exon starts at base pair 6,123 and its polyA at 19,332 bp (Table 2). This gene could transcribe a 6,585 bp mRNA that would translate a 2,194 amino acid (aa) product. The derived aa sequence subjected to NCBI Non-redundant protein sequence database BLAST found similarity to many putative CACTA transposon proteins. The highest similarities (59% and 53%) were to Vitis vinifera hypothetical proteins CAN82870 and CAN66891 over a length of 1060 and 1400 aa respectively. The Tgmt* predicted protein had two putative conserved domains, a transposase family tnp2 domain [+](pfam02992) (E value 1e-80) located between 265 – 493 aa and a TNP1/EN/SPM transposase domain [+](pfam03017) (E value: 0.006) located at 1,493 – 1,534 aa. A search for similar domain architectures with NCBI CDART (Conserved Domain Architecture Retrival Tool) found two sequences from Oryza sativa (Os03g0714800) 27 to have the most closely related domain architectures. In contrast, the two V. vinifera hypothetical proteins with the highest similarity to Tgmt* Gene-1 product had only one transposase domain, the tnp2 domain.

Based on the molecular analysis of Tgmt* DNA sequence with web-based gene prediction algorithms and the extent of the similarity of the predicted gene to the well characterized gene of the En-1 autonomous CACTA transposable element, it was presumed that Tgmt* Gene-1 should be expressed in soybean lines where the putative autonomous Tgmt* element could be active. An initial attempt at determining Gene-1 expression was a search for RNAs hybridizing to Gene-1 DNA probes on RNA blots with samples extracted from multiple tissues of four soybean lines varying at the T locus, Williams 43 (T), XB22A (T*), 37609 (t*) and 33745 (t^m) (Table 1). No clear, significant hybridization was detected in any of the RNA blots with any of the tested DNA probes which together covered the region of Gene-1 encoding Exons 1–7, with the tnp2 and TNP1 transposase domains (data not shown).

These results suggested that if the transposase gene was expressed it did so at very low levels, and thus we opted for the more sensitive reverse transcriptase polymerase chain reaction technique (RT-PCR) to assay Gene-1 expression. Figure 6 shows the amplification products obtained using three primer pairs. Each pair amplified a different portion of the Gene-1 region with the tnp2 and TNP1 domains. The RT-PCR reactions shown were carried out with RNAs from three isolines that were derived from a single rogue soybean plant that had shown variegation in trichome color in a field breeding program: XB22A (T*), 37609 (t*) and 33745 (t^m) (Table 1). The negative controls (-) were reactions where the cDNA synthesis step was allowed in the absence of superscript (See Methods). Interestingly, the 37609 line with the recessive allele (t*) from which Tgmt* was isolated did not appear to express Gene-1. In contrast, XB22A (T*) and 33745 (t^m) isolines seem to have retained a putatively active Tgmt* expressing the transposase region of Gene-1. This differential pattern of RT-PCR amplification amongst the isolines was repeated with several other primer sets tested covering other regions of Gene-1 (data not shown). Figure 7 shows results of additional examples of PCR reactions performed with cDNAs synthesized from RNA of the mutable isoline 33745 (t^m) and that generated larger or more complex DNA fragment patterns. Lane 2 of Figure 7A shows the range of amplification products (~2- 5.5 kb) obtained with two primers (1 and 7) each at either end of the predicted Gene-1. Many of the products amplified from cDNAs of this mutable line, 33745 (t^m), were cloned and sequenced and the analysis and assembly of the cDNA sequences provided a clear picture of Tgmt* expression which is shown in Figure 8. The location of all primer pairs used in the PCR reactions which results are shown in Figures 6 and 7 were marked in Figure 8 with small arrows numbered from 1–7. In certain instances the arrows were placed near the clones (Cl. 23, 14 and 47) to simplify the diagram.

The sequences of 17 cDNA fragments were used to map all the exons and introns on the genomic Tgmt* sequence. The exon-intron boundaries obey the canonical GT-AG rule 29 in the majority of the cDNA clones. See Additional files 5 and 6 where exon sequences of cDNA clones have been aligned with the Tgmt* genomic sequence. The sizes and locations of the cDNA exons are listed in Table 2 and highlighted in bold are 17 of them that are almost identical in size and location to exons predicted by the Softberry- FGeneSH program. Two of the predicted Exons (No. 11 and 15) were portions of larger exons in the cloned cDNAs (No. 11 and 19). However, predicted Exons No. 17 and 19 towards the 3' end of the gene were not part of the cloned cDNAs and since we recovered 11 different cDNA clones from this region of the gene it can be assume with certainty that predicted Exons-17 and -19 were errors of the FGeneSH program. A full length cDNA of 7,554 bp spanning the predicted Gene-1 could be assembled with all the exon sequences of the cDNA clones (see Additional file 7: Tgmt* Gene-1 precursor transcripts sequence). Most likely, transcripts of this size were synthesized in vivo given the RT-PCR results obtained and the locations of the primers used on the gene sequence (Figures 7 and 8).

Of most relevance were the sequences of the amplified cDNAs with primers at the ends of the gene (1 and 7) because they revealed splicing of the first 346 bp (1/9 portion) of Exon-1 with Exon-11 to generate transcripts 2 – 2.5 kb in size (Figure 8, Mosaic cDNAs, Cl. 37 – 39; see Additional file 8: Tgmt* Mosaic-transcript sequence). The splicing of the 346 bp of Exon-1, ending in ATAAT, did not occur at the intron-exon junction of Exon-11 but rather 23 bp into Intron-10 right border, adding the 23 bp of intron sequence in the synthesis of mosaic transcripts. The Intron-10 sequence up-stream of the splice site also ends on ATAAT, the same motif found up-stream of Exon-1 splice site (see Additional file 9: Tgmt* genomic sequence (20,544 bp)). Whether or not this sequence motif is involved in the recognition and splicing mechanism to create the mosaic gene is not known.

The splicing of all the many exons to form the mosaic transcripts as well as the full length precursor transcripts must be cumbersome judging by the multiplicity of related cDNAs amplified with a given set of primers (Figure 7 A, lanes 2 and 3; Figure 8, mosaic gene transcripts (Cl. 37–39) and transcripts amplified with primers 5 and 7, Cl. 47-43). Some of the cDNAs retained entire introns such as Cl. 47 or intron portions in Cl. 44 and 53 (Figure 8). Exon-19 was spliced out in six of the cDNA clones sequenced (Cl.44-43 and 38 and 39). This erratic splicing of exons was observed earlier in transcripts from another complex soybean CACTA transposon,Tgm-Express1 containing multiple host-gene fragments 2530.

Nevertheless, a putative full length mosaic cDNA of 2,572 bp was cloned and sequenced (Figure 8 and see Additional file 8: Tgmt* Mosaic-transcript sequence). The splice site between Exon-1 (first 1/9 portion of the exon) and Exon-11 is indicated with an asterisk in Figure 8. This mosaic transcript may encode the DNA-binding protein that is required for excision of these CACTA transposons in soybean in the same fashion TNPA does it for the En-1 element of maize. NCBI/BLAST/blastx (version 2.2.18) found similarities to some predicted proteins from V. vinifera, Arabidopsis and O. sativa in the non-redundant protein database but not to maize. This result supports previous observations that proteins that share little similarity, such as TNPA and TNP1, may have a similar function, that of binding to the subterminal repeats, regions that vary considerably amongst the different CACTA transposons 31.

In summary, Tgmt* expresses the predicted Gene-1 in XB22A (T*) and the mutable 33745 (t^m) lines, but not in the stable, gray trichome 37609 (t*) isoline from which it was cloned. A multiplicity of mRNAs, were transcribed from Gene-1 and the putative largest transcript assembled was 7,554 bp in length. In addition, mosaic transcripts of different sizes were also synthesized and the largest was 2,572 bp. The precursor mRNA codes for a transposase with a tnp2 and TNP1 domains in Exons-1 and 5 respectively. The 2,572 mosaic transcript may encode a DNA-binding protein with a function similar to that of TNPA of the maize En-1 and TNP1 of A. majus Tam3, but with very limited amino acid sequence similarity.

In order to determine the abundance of Tgmt* sequences in the soybean genome, an initial NCBI/BLAST/blastn search of the nucleotide collection (nr/nt) database optimized for highly similar sequences was run using either the entire 20,544 bp Tgmt* sequence, the 5,078 bp comprising Exon-1 through Exon-5 with the transposases ORF1 and ORF2 or the 7,778 bp starting at Exon-11 through Exon-24. The transposase portion of the element had 84% identity with 98% coverage to a Glycine max clone gmw1-45m6, (AC166742.25), 75% identity over a 58% coverage to two Glycine tomentella clones gtt1-62b6 (AC188785.18) and gtt1-188p11 (AC183809.15), 75%identity and 26% coverage to a Glycine max clone BAC GM_WBb080D (EF533700.1), and finally, 76% identity over 17% coverage, to the soybean Tgm5 transposable element (X13528.1). However, no highly similar sequences were found when the BLAST search was performed with the second half of Gene-1 sequence (7,778 bp). Two additional sequences, one from Glycine max cultivar T322 dihydroflavonol-4-reductase 2 (DFR2) gene, Intron II and transposon Tgmw4m (EU068464.1) and a second one from Glycine max cultivar T321 dihydroflavonol-4-reductase 2 (DFR2) gene, DFR2-w4-dp allele, disrupted promoter region and transposon Tgmw4m (EU068463.1) had 99% identity to the right and left border of Tgmt* over 1,325 and 975 bases, respectively, indicating that these may be deletion derivatives from an element similar to Tgmt*.

The search was extended to the 7× draft sequence assembly Glyma0 from the Joint Genome Inititive (JGI version 12/6/07, see methods). The blastn search with the Tgmt* entire sequence (20.5 kb), found high level of similarity to 52 scaffolds (E value = 0.0). To determine how many of the 52 scaffolds had the transposase and the exons of the mosaic transcript, two blastn searches with the 5,078 bp (Exon-1 through Exon-5) and the 7,778 bp (Exon-11 through Exon-24) were performed. The transposase exons found high similarity (E value = 0.0) to 45 scaffolds while the exons of the mosaic transcript found high similarity (E value = 0.0) to 15 scaffolds. Scaffolds with high similarities to both halves of Tgmt* Gene-1 were, Scaffold_4, _81, _24, and _5. Thus, it appears that the soybean genome of cultivar Williams has at least 4 regions with extensive sequence similarity to Tgmt*.

We examined also the regions of similarity in the Glyma0 soybean genome (JGI 7× draft sequence assembly) to the Tgm5 transposase sequence (Ac. No. X13528.1) and the equivalent sequence region of Tgmt* (932 bp) using the Phytozome Glycine max (v3.0) BLAST search http://www.phytozome.net/soybean.php. We found that the two transposons had similarities to two different sets of scaffolds. Tgm5 transposase had the highest similarities to scaffolds_38, _20, _10, _229, _7, _12, _97 and _31 while Tgmt* transposase had the highest similarities to scaffolds_24, _4, _81, _6, _5 and _15. This distinction is additional supporting evidence that Tgm5 and Tgmt* are elements representing two different CACTA families in soybean.

A closer examination of the regions of similarities in each one of the scaffolds revealed a large number of partial transposase sequence copies in each one of those scaffolds when the 932 bp segment carrying the tnp2 domain was used as query. For Tgm5 scaffolds_38, _20, _10, _229, _7, _12, _97 and _31 there were 50, 37, 47, 6, 31, 23, 6 and 31 copies respectively. For Tgmt* scaffolds_24, _4, _81, _6, _5 and _15 there were 27, 55, 13, 53, 55 and 3 copies respectively. However most of these were relatively short regions not containing the element ends. Thus it appears that some regions of the 932 bp orf that carries the conserved tnp2 domain are widely dispersed in the soybean genome. These observations of moderately high copy number is supported by the intense hybridization signals of varying sizes obtained when this region of the element is used as a probe on genomic DNA blots. Furthermore, this region also hybridized to many clones when used to screen genomic libraries (unpublished observations).

As previously discussed, to fully determine the molecular defect of the F3'H allele (t*) with the gray trichome phenotype, it required the cloning of a full length F3'H gene including its promoter. With such an aim, two Glycine max BAC libraries, one from Williams 82 (GMWBa libarary) and the other from the more distantly related plant introduction line PI437654 (Clemson University Genomic Institute) were screened. No clones were obtained from the Williams 82 GMWBa library but a BAC clone (71B1) containing a full length F3'H gene copy was isolated from the PI437654 library. Partial sequence of the F3'H gene in 71B1 clone was determined initially from PCR amplified fragments and later confirmed and extended with sequences resulting from a shotgun library of the entire BAC clone. These sequences were assembled into three contigs, most likely arranged as shown in Figure 9 (Acc. No.: EU721743). Softberry FGeneSH gene prediction and annotation web-based program identified and placed the full F3'H copy (1,590 bp) of the gene in contig-2. In addition, two smaller fragments with F3'H similarity (504 and 354 bp) mapped to contig-1. All three F3'H sequences are marked with red arrows in Figure 9. All other genes found in 71B1 clone were also annotated and their size and relative location in the three contigs are depicted in Figure 9 and listed in the annotation Table 3. As noted, other full copy genes in this BAC clone were three located in tandem in contig-3 encoding functions of a retrotransposon (gag-pol polyprotein and envelop-like protein). The assembly of the sequences resulting from the shotgun library did not overlap the three contigs shown in Figure 9 and their order could not be determine. With the recent release of the 7× draft sequence assembly (JGI version 12/6/07) (see Methods) from the soybean genome of cultivar Williams 82, a BLASTn search was run with the sequences of the three contigs from 71B1 BAC clone. It was determined that contigs-1 and -2 are adjacent in scaffold_83 and contig-3 sequence is an insertion in the middle of contig-1. For that reason contig-3 was drawn above contig-1 in Figure 9. Because the sequences with high similarity to the three contigs mapped closely together in scaffold_83, we can presume conservation of this region in the two cultivars, Williams 82 and PI437654 (see Additional file 10: Alignment of sequences of 3 contigs from 71B1 BAC clone in the 7× draft sequence assembly (JGI)).

We reported earlier that the F3'H gene appeared to be single copy in soybean 26. The results of a BLAST search of the 7× draft sequence assembly (JGI version 12/16/07, see Methods) from the soybean genome of cultivar Williams with the Williams F3'H gene sequence (Ac. No.: EU190438) and the sequence of BAC clone 71B1, all pointing to a single Scaffold_83 with only one full F3'H gene copy, are further evidence that F3'H is a single copy gene in soybean. These results also explain the gray trichome phenotype of the soybean line in which the large Tgmt* element inserted in Intron-1 of the single copy of the F3'H gene in soybean.

The Glycine max cultivars and isolines used for this study were: Williams (T, tawny trichomes) XB22A (T*, tawny trichomes), 37609 (t* gray trichomes), 37643 (t*, gray trichomes) and 37345 (t^m, with variegated hilum tawny and gray trichomes). Each is homozygous for the indicated alleles of the T locus. The origin, genetics, and discovery of the T allele as F3'H has been described previously 26. Plants were grown in the greenhouse. Shoot tips (meristems surrounded by primordial leaves), seed coats and cotyledons dissected from seeds at varying stages of development were frozen in liquid nitrogen, freeze dried (Multi-dry lyophilazer; FTS systems), and stored at -20°C. The seed coats and cotyledons used in this study were those of seeds which fresh weight of the entire seed was 25–50 mg.

Total RNA was isolated from shoot tips, seed coats and cotyledons using a phenol-chloroform and lithium chloride precipitation method 3637. RNA was stored at -70°C until used.

cDNA copies of the Tgmt* predicted gene(s) from three of the isolines (XB22A, 37609, 33745) and Williams 43 were amplified from a first-strand cDNA pool synthesized using 1 μg of cotyledon total RNA and the Superscript first strand synthesis system for reverse transcriptase (RT)-PCR (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 primer pairs used in RT-PCR reactions shown in Figure 6 were:

(1) Primer No. 3 5'-GGGACATCTGAGAATGAC-3' (TGME3-43F)

Primer No. 4 5'-AACAACATACAATCCCAT-3' (TGME3-36R)

(2) Primer No. 1: 5'-AACAGCACGCATAACTGAAGA-3' (TGM8-14677R)

Primer No. 2: 5'-AACTGGTGTCGTACTCCC-3' (43-43-FR)

(3) Primer No. 5: 5'-GGCTGAAATAATATCAGG-3' (TGME-36NZ-F)

Primer No. 6: 5'-CATCATATTTAGCATCTG-3' (43-43-R2)

Primer pairs of RT-PCR reactions shown in Figure 7A and 7B were:

(A1) Primer No. 1: 5'-AACAGCACGCATAACTGAAGA-3' (TGM8-14677R)

Primer No. 6: 5'-TTAGCATCTGTTATTTCTAATAGC-3' (TRP-1-F) ≅ (43-43-R2)

(A2) Primer No. 1: 5'-AACAGCACGCATAACTGAAGA-3' (TGM8-14677R)

Primer No.7: 5'-TGTGAAGTCATATAGAGTGGC-3' (TGM8-1741F)

(A3) Primer No. 5: 5'-GGCTGAAATAATATCAGG-3' (TGME-36NZ-F)

Primer No.7: 5'-TGTGAAGTCATATAGAGTGGC-3' (TGM8-1741F)

(B) Primer No. 3 5'-GGGACATCTGAGAATGAC-3' (TGME3-43F)

Primer No. 6: 5'-TTAGCATCTGTTATTTCTAATAGC-3' (TRP-1-F) ≅ (43-43-R2)

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. For small DNA fragment amplification, 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 Ex Taq (Takara Bio Inc. Otsu, Japan) polymerase was used at 0.75 units per 50 μl reaction. To amplify the larger DNA fragments, PCR reaction conditions were as follows: initial denaturation step at 94°C for 2 min followed by 30 cycles of 94°C for 30 sec and 68°C for 10 min, and ending with a 10 min extension at 72°C. LA Taq polymerase (Takara Bio Inc. Otsu, Japan) was used at 0.75 units per 50 μl reaction.

In most instances, amplified DNAs 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. However, the larger, 23 kb PCR product (primers: Z37F, 5'-ATTTGAAACGCGTGGTGCCTGCATTTAAAGACAA

TTT-3' and IN663R, 5'-ATCACCTCCATCACCATAGCCTTAAACTCATCAGCCC-3') was extracted from 1% Seaplaque agarose gel in TA buffer with the aid of GELase Agarose Gel-Digesting Preparation (Epicenter Biot. Madison, WI) following the manufacturer's protocol. Two equal fractions (200 ng) of the 23 kb DNA fragment extracted in this fashion was utilized to prepare a radiolabel probe and in HindIII restriction digests. The resulting HindIII fragments were cleaned with QIAquick PCR Purification kit (QIAGEN) and cloned into CIP dephosphorylated pGem vector cut with HindIII. A DNA ligation kit Ver.2.1 (TaKaRa) aided the cloning step. The resulting 23 kb HindIII clones were verified in DNA blots hybridized to the radiolabel 23 kb probe.

The conditions used to optimize the cloning of a 17 kb DNA fragment included the phosphorylation of primers Z37F (sequence above) and TGM23R (5'-GTGATTTGATAGAACAAGGTACGTAAAAGCTGAAAC-3') with T4-polynucleotide kinase (Lucigen Co. Middleton, WI) prior to the PCR amplification reaction. The products of this reaction were extracted from a 1% low-melt Seaplaque agarose gel with QIAEX II gel extraction Kit (QIAGEN) and cloned into pJAZZ-OC vector using the BigEasy v2.0 Linear Cloning Kit (Lucigen Co. Middleton, WI). Multiple size fragments were cloned and verified by sequencing.

RNA (10 μg/sample) was electrophoresed in a 1.2% agarose-3% formaldehyde gel 38. 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) 38 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) 39.

Cloned DNAs used as probes were PCR amplified, electrophoresed, and purified from the agarose using the QIAquick gel extraction kit (QUIAGEN, 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 40.

A shotgun library of BAC clone 71B1 from G. max PI437654 cultivar was constructed at the Keck Center for Comparative and Functional Genomics (University of Illinois). BAC DNA was randomly sheared and cloned into the pCR4Blunt-TOPO vector using the Topo Shotgun Subcloning Kit (Invitrogen). Sequencing of the clones was done using the Big Dye Terminator chemistry (Applied Biosystems, Foster City, CA). Base-calling and quality assessment were performed automatically using PHRED 41. High-quality sequences were assembled with PHRAP to order the contigs. PCR amplification and direct sequencing of subclones was used to close gaps with an end result of three contigs that could not be overlapped.

Sequencing of all other clones was done at the Keck Center (University of Illinois Biotechnology Center)

Blast searches with sequences of the Tgmt* transposon and the three contigs of 71B1 BAC clones were extended to the 7× draft sequence assembly Glycine max 0 (Glyma0) from the Joint Genome Initiative (JGI) released 12/06/07. This 7× assembly consisted of 3,317 sequences with a total of 996,173,606 letters which was made accessible to us by the Matt Hudson Laboratory (U. of Illinois at Urbana/Champaign) http://stan.cropsci.uiuc.edu/blast/blast.html. We also searched the JGI Glyma0 soybean genome assembly using the Phytozome: Glycine max (v3.0) BLAST web-site http://www.phytozome.net/search.php?show=blast. Note that the scaffolds numbers are predicted to change once the final 8× soybean genome assembly is completed.

Sequence data from this article can be found in the EMBL/GenBank data libraries under accession numbers: [EU190438: F3'H (T) allele in Williams43]; [EU190439: F3'H (T*) allele in XB22A]; [EU190440: F3'H (t*) mutant allele in 37609 isoline]; [EU721743: 71B1 BAC clone].