Chromosomal Organization and Sequence Diversity of Genes Encoding Lachrymatory Factor Synthase in Allium cepa L.

Lachrymatory factor synthase (LFS) catalyzes the formation of lachrymatory factor, one of the most distinctive traits of bulb onion (Allium cepa L.). Therefore, we used LFS as a model for a functional gene in a huge genome, and we examined the chromosomal organization of LFS in A. cepa by multiple approaches. The first-level analysis completed the chromosomal assignment of LFS gene to chromosome 5 of A. cepa via the use of a complete set of A. fistulosum-shallot (A. cepa L. Aggregatum group) monosomic addition lines. Subsequent use of an F(2) mapping population from the interspecific cross A. cepa × A. roylei confirmed the assignment of an LFS locus to this chromosome. Sequence comparison of two BAC clones bearing LFS genes, LFS amplicons from diverse germplasm, and expressed sequences from a doubled haploid line revealed variation consistent with duplicated LFS genes. Furthermore, the BAC-FISH study using the two BAC clones as a probe showed that LFS genes are localized in the proximal region of the long arm of the chromosome. These results suggested that LFS in A. cepa is transcribed from at least two loci and that they are localized on chromosome 5.

The onion nuclear genome is notable for its great size, 17 pg or 15 Gbp per 1C (Arumuganathan and Earle 1991; Bennett and Leitch 1995; Ricroch ), one of the largest among cultivated plants. There are limited genomic resources available in onion due to its huge genome size, and genetic studies are further complicated because it is a biennial, out-crossing, and highly heterozygous species. However, an EST resource and PCR-based map (Kuhl ; Martin ) has been recently developed. Combination with these resources and chromosome addition lines (Shigyo ) has revealed chromosomal location and genetic map position of genes responsible for important properties, such as carbohydrate accumulation (McCallum ; Masuzaki ; Yaguchi ) and flavonoid biosynthesis (Masuzaki ). The physical distribution of AFLP markers along Allium chromosomes has been studied via the integration of recombination and physical maps in a trihybrid population, A. cepa × (A. roylei × A. fistulosum) (Khrustaleva ). Direct physical mapping of genes on onion chromosomes is limited due to the genome abundance with repetitive elements (Stack and Comings 1979; Pearce ). Fluorescence in situ hybridization (FISH) was successfully applied for the detection of specific loci using large genomic clones as probes mostly in plant species with small gene-rich genomes, such as Arabidopsis thaliana (Kornneef ) or rice (Jiang ). However, in some cases, using repetitive DNA that blocks probe hybridization of repetitive sequences allows for the detection of genes inserted in a bacterial artificial chromosome (BAC) (Lamb ; Szinay ).

The most distinctive attribute of onion is the tearing property conferred by lachrymatory factor (LF; propanethial S-oxide). LF is formed from 1-propenylsulphenic acid by lachrymatory factor synthase (LFS) (Imai ). 1-propenylsulphenic acid is a putative reaction product derived from 1-propenyl cysteine sulfoxides (trans-PRENCSO) by alliinase. Suppressing the LFS gene turns off the conversion of 1-propenylsulfenic acid to LF, which in turn increases the yield of thiosulfinates thought to be responsible for the flavor and beneficial health properties of onion. Eady previously demonstrated large shifts in organosulfur secondary compound profiles in onions in which LFS activity was suppressed by RNAi. These studies suggest that LFS is an important target for molecular breeding in onion. LFS cDNAs have been cloned from other five lachrymatory Allium species (A. ampeloprasum, A. cepa Aggregatum group, A. chinese, A. fistulosum, and A. porrum), and all recombinant proteins from them showed LFS enzymatic activity. In addition, strong homologies were observed in these LFS cDNA, and no homologous sequences were yielded in GenBank searches (Imai ; Masamura ). These results suggested that the LFS gene is strongly conserved among lachrymatory Allium species and is only distantly related to proteins in other higher plant taxa. Because LF is such a bioactive and distinctive compound, it is likely that strong selective forces have acted on LFS genes during evolution and domestication of Allium.

In this study, as a model of a functional gene in a huge genome, we determined the genome organization of LFS genes by sequence analysis, genetic mapping, and physical methods to understand evolution of LFS in Allium and contribute to targeting molecular breeding and mutagenesis approaches for manipulating onion quality.

Materials and Methods

Genetic analyses of LFS gene by using monosomic addition line and mapping population

The plant materials were a complete set of A. fistulosum–shallot monosomic addition lines [2n = 2x + 1 = 17, FF+1A (plant number 130), FF+2A (141), FF+3A (5), FF+4A (10), FF+5A (26), FF+6A (308), FF+7A (324), FF+8A (240) ] and parental control plants, Japanese bunching onion (A. fistulosum cv. Kujyo-hoso, 2n = 2x = 16, FF) and shallot (A. cepa Aggregatum group ‘Chiang Mai’, 2n = 2x = 16, AA) (Shigyo ). They were grown in an experimental field at Yamaguchi University (34°N, 131°E).

Genomic DNA was extracted from the frozen base of leaf sheath tissues by DNeasy Plant Mini Kit (Qiagen, Hilden, Germany). PCR were performed a 25 µl reaction mixture containing 2.5 µl of template, 1:10 or 1:100 diluted cDNA or genomic DNA (20 ng/µl); 0.125 µl of Taq polymerase [AmpliTaq GOLD (5 U/µl), Applied Biosystems, Foster City, CA]; 2.5 µl 10× PCR buffer; 1.5 µl of MgCl2 (25 mM); 0.5 µl of forward primer (25 µM) and 0.5 µl of reverse primer (25 µM) of primer set (cepaLFS); and 2.5 µl of dNTP mixture (2 mM). Nucleotide sequences of the primers are shown in Table 1. PCR was carried out in GeneAmp 2400 or GeneAmp 9600 (Applied Biosystems) with the following amplification program: an initial heating to activate the Taq polymerase at 94° for 10 min, followed by 35 cycles at 94° for 1 min, 65° or 68° for 1 min, 72° for 1 min, and then a final elongation at 72° for 10 min. The PCR products were detected by electrophoresis in 2% agarose gels.

Table 1

Previously unpublished primer sets used in this study

Primer Set	Marker Type	GenBank Accession Number	Forward Primer Reverse Primer
cepaLFS	—	AB089203	ACAAAGCCAGAGCAAGCATGGACA
			CTGCAAACCTCTTCGATTTTCTGACCTATC
LFS5	Heteroduplex	AB089203	GCACTAGAACTTGCAAAAAGCA
			TGAGATAGGTCAGAAAATCGAAGA
ACP052	SSCP	CF445004 CF445805 CF445805	TTCCCTCCTCACTCCCTACA
			CGACCACAAACACAAGCAAC
ACM295	SSR	CF445600	AGATCCGTCCCATGAAACT
			GATCCGCTTCTGAAATCTCG
ACM021	SSR	CF448154	AAAACCCTCAACATCTCACTCC
			TCTCTTCTTCCTCGTCCTGC
ACM076	SSR	CF449018	ATTAGAAACATCCATCGCCG
			CGCGATCATCATTTTCCATA
ACP003	SSCP	BE205590	AAGCTCTTAAAGCTGCTGATGG
			ATGCACGATAGCACAAGACATC
LFSorf	—	AB089203	ACAAAGCCAGAGCAAGCATGGACA
			CTGCAAACCTCTTCGATTTTCTGACCTATC
LFShaplo	—	AB089203	ATAGTGGAGGGTCCTGAGCA
			ACACAACACTCAGTCTTACTTATT

DNA templates and genetic map data from the interspecific Allium cross A. cepa × A. roylei were used as described by Van Heusden , b). Primer sets used in this study are shown in Table 1. Design, PCR, and analysis methods for SSCP and SSR markers were described previously (McCallum , 2007, 2008). Linkage analysis was performed using JoinMap ver. 4 (Van Ooijen 2006).

Screening of BAC clones containing LFS gene and its molecular characterization

The partial BAC library of onion (Suzuki ) was used for the PCR screening with onion LFS-specific primers amplifying a 459 bp fragment comprising most of the LFS ORF. Nucleotide sequence of the primer set (LFSorf) used are shown in Table 1. PCR was performed with Ampli-Taq GOLD (Applied Biosystems) for 40 cycles of denaturation for 1 min at 94°, annealing for 1 min at 58°, and extension for 1 min at 72°, followed by a final extension for 7 min at 72°.

BAC DNA purified by QIAGEN plasmid Midi kit (Qiagen) was digested with EcoRI or HindIII and separated on 0.7% agarose gel. After electrophoresis, DNA was transferred to HybondN+ membrane (GE Healthcare UK Ltd., Buckinghamshire, UK). The digoxigenin (DIG)-labeled LFS-specific DNA probe (452 bp) was prepared by PCR reaction using PCR DIG Labeling Mix (Roche Diagnostics, Manheim, Germany). Hybridization was carried out in 5× SSC, 50% formamide at 38.5° overnight. The membrane was washed in 0.5× SSC, 0.1% SDS at 65° for 30 min. After blotting and washing, we detected the positive signal with fluorescence according to the instruction of the DIG fluorescent detection Kit (Roche Diagnostics) with ECF (GE Healthcare UK Ltd.) as the substrate.

Purified BAC clone 2E8/10 was nebulized and shotgun-cloned into vector pSmartHC, and 752 clones were sequenced from both ends (mean sequence length ca. 700 bp) by GATC Biotech Ltd Germany (Konstanz, Germany http://www.gatc-biotech.com). Trace files were assembled using the Staden package ver. 1.6 (Staden 1996) and Sequencher (Gene Codes Corporation, Ann Arbor, MI). The shotgun sequencing of BAC clone 4F10/155 was performed at Hokkaido System Science Co. Ltd. (Sapporo, Japan) with the GS FLX Titanium system (454 Life Sciences, Roche, Branford, CT). The GS FLX Titanium generated ca.8.5 million bases (about 30 thousand reads), giving 85-fold coverage of the insert DNA of the clone. The reads were assembled into contigs with GS De novo Assembler software (454 Life Sciences). The contigs (∼20 kb) containing LFS from both BAC clones were annotated using BLAST (Altschul ) searches against GenBank plant protein and nucleotide databases. BLASTN comparison between contigs was visualized using genoPlotR (Guy ). Tandem repeats were identified using EMBOSS quicktandem (Rice ). Sequences were submitted to GenBank (accession nos. JN798503 and JN798504).

BAC FISH analysis

Mitotic chromosomes were prepared from young root meristems of bulb onion (A. cepa cv. Khalcedon). Onion Cot-100 DNA was prepared as described by Zwick with some modification. Total genomic DNA was isolated using CTAB method (Rogers and Bendich 1988) and was sonicated to a fragment size of about 1 kb. Sheared DNA was denatured in 0.3 M NaCl at 95° for 10 min, and then it was re-annealed at 65° for 31 h 40 min. The exact renaturation time was calculated taking into account the initial DNA concentration and percentage of G-C fraction in onion genome. The remaining ssDNA was digested with S1 endonuclease (Fermentas, Burlington, ON, Canada, http://www.fermentas.com) with final concentration 1 U/µg for 90 min at 37°. The reaction was stopped and DNA was extracted by adding 300 µl chloroform/isoamyl alcohol (24:1). Purified BAC DNA was labeled with digoxigenin (DIG)-11-dUTP by Nick-translation standard protocol (Roche Diagnostics,). Slide pretreatment was performed according to common FISH procedure. Hybridization was carried out with the mixture consisted of 50% (v/v) deionized formamide, 10% (w/v) dextran sulfate, 2× SSC, 0.25% sodium dodecyl sulfate, 12.5 ng/µl probe DNA, and 0.3 µg/µl Cot-100. Stable sites of the probe hybridization were detected with anti-DIG FITC antibody. Chromosomes were counterstained with DAPI in Vectashield antifade (Vector Laboratories, Burlingame, CA). Slides were examined under a Zeiss Axio Imager microscope (Carl Zeiss MicroImaging, Jena, Germany). Selected images were captured using an Axio Cam MRm digital camera. Image processing and thresholding were performed using AxioVision. Final image optimization was performed using Photoshop (Adobe Inc., San Jose, CA). Detailed experimental conditions are presented in supporting information, File S1.

Re-sequencing of LFS amplicons

DNA was isolated as described previously (McCallum ) from Allium roylei, and the following onion genotypes selected based on previous genetic diversity (McCallum ) and pungency phenotypes: ‘W202A’, ‘Texas Grano 438’, ‘BYG15-23’, ‘Alisa Craig 43’, ‘Colossal Grano’, ‘Early Longkeeper P12’, ‘W429A’, ‘Houston Grano’, ‘Tearless F1’, and ‘Faridpuri’. PCR reactions were performed in a 15 µl volume and contained 0.5 µM of primers LFS5L and R (Table 1), 200 µM of dNTPs, 1.5 mM of Mg2+, 10–30 ng of template DNA, 0.375 U of ThermoPrimeTaq polymerase (Thermo Scientific, MA) and 1× the manufacturer’s ReddyMix PCR buffer. Reactions were performed in a Gene Amp PCR System 9700 thermocycler (Applied Biosystems). Cycling conditions were 2 min at 95° followed by a touchdown that included four cycles of 30 sec at 95°, 30 sec at 62° to 59°, in which this annealing temperature decreased by 1° each cycle, and 30 sec at 72°. This touchdown was followed by 36 cycles of 30 sec at 95°, 30 sec at 58°, and 30 sec at 72°, and a final extension of 10 min at 72°. PCR products were visualized by electrophoresis on a 1% LE agarose plus 1% NuSieve (FMC Bioproducts, Rockland, ME) agarose gel stained with ethidium bromide.

PCR products were then ligated into the pCR 4 TOPO Vector using a TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA) and transformed into DH5α by heat shock. Twelve colonies for each of the 12 onion lines were selected after growth on LB containing amplicilin 100 µg/ml. These were streaked on amplicilin 100 µg/ml and tested by PCR (see above). Eight colonies containing LFs amplicons from each of the 12 lines were used to inoculate a 1 ml LB (ampicillin 100 µg/ml) overnight culture. Plasmids were then isolated using a Perfectprep Plasmid Isolation Kit (Eppendorf, Hamburg, Germany). LFS plasmid inserts were then Sanger-sequenced with M13 forward primer using Big Dye Terminators v3.1 (Applied Biosystems). Sequencing reactions were run on an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems). Traces were trimmed and aligned using Geneious software (Drummond ) and submitted to GenBank (accession nos. HQ738844–HQ738918). Alignments were cropped to the region corresponding to bases 368–632 of GenBank accession no. AB089203, and maximum likelihood trees were constructed using PHYML (Guindon and Gascual 2003) with 100 bootstraps. Nucleotide diversity calculations were performed using DnaSP (Librado and Rozas 2009).

Identification of LFS transcripts by 454 sequencing

Raw 454 flowgram files were used from GenBank BioProject accession no. PRJNA60277 for shoot transcriptome sequencing of doubled haploid onion line ‘CUDH2150’ (Cornell University; GenBank BioSample no. 138247) and ‘Nasik Red’ (USDA-ARS PI 271311; GenBank BioSample no. 138248). Reads were aligned to the LFS reference sequence (AB089203) using Roche gsMapper software, using minimum overlap identity of 98% in the assembly step and minimum overlap length of 300 for ‘DH2150’ (Titanium reads) and 150 for ‘Nasik Red’ (GS-FLX reads).

Amplification of LFS haplotypes in the FF+5A

PCR was performed an 30 µl reaction mixture containing 3.0 µl of genomic DNA (20 ng/µl) purified from two different plants [FF+5A(26) and FF+5A(71)] of FF+5A or shallot parent; 0.15 µl of Taq polymerase (ExTaq(5 U/µl); TakaraBio, Ohtsu, Japan); 3.0 µl 10× Ex buffer; 0.15 µl of forward primer (25 µM) and 0.15 µl of reverse primer (25 µM) of primer set (LFShaplo); and 2.4 µl of dNTP mixture (2.5 mM). Nucleotide sequences of the primers are shown in Table 1. PCR was carried out in GeneAmp 2400 or GeneAmp 9600 (Applied Biosystems) with the following amplification program: an initial heating to activate the Taq polymerase at 94° for 3 min, followed by 35 cycles of 94° for 1 min, 64.9° for 1 min, and 72° for 1 min. The 5 µl of PCR reactions were subjected to restriction endonuclease digestion with NsiI and SphI enzymes, and the fragments were analyzed by electrophoresis in 2.0% agarose gels.

RESULTS

Chromosomal mapping of LFS gene in A. cepa

From the genomic DNA of alien monosomic addition lines, the expected size amplicon was observed only in the alien monosomic addition line FF+5A (Figure 1). Therefore, we assigned LFS of A. cepa to chromosome 5A. Genetic mapping of polymorphisms detected by heteroduplex analysis of LFS amplicons in the A. cepa × A. roylei interspecific cross revealed co-segregation with chromosome 5 markers (Figure 2). The LFS marker was linked to several other markers developed to onion ESTs, most notably the SNP marker ACP052. This was designed to partial ESTs showing homology to N-terminal regions of group I sucrose transporters (Braun and Slewinski 2009). The RFLP marker API66C-E5 mapped in the ‘BYG15-23 × AC43’ intraspecific onion population (King ) was revealed by the cDNA probe API66 (GenBank accession no. BE205593.1), which is homologous to sucrose transporter group I and III proteins. It is not known yet whether ACP052 and API66 markers target the same gene or linked duplications. We previously showed close linkage of API66 markers to a well-supported QTL affecting bulb dry matter (Galmarini ; Martin ; Masuzaki ; McCallum ), suggesting that the LFS loci are in close linkage with this QTL.

Figure 1

Amplification of LFS from AMALs. M, molecular size marker (100 bp ladder); FF, A. fistulosum; 1A–8A, eight different A. fistulosum–shallot monosomic additions; AA, shallot.

Figure 2

Genetic mapping of LFS heteroduplex marker LFS5 to chromosome 5 in the A. cepa × A. roylei population and alignment with the onion linkage map (BYG15 × AC43) of Martin . The scale denotes the recombination distance in Kosambi units. AFLP markers in the interspecific map have been deleted for clarity.

Selection and characterization of BAC clones containing LFS gene

PCR screening of the BAC library revealed 8 positive clones, from a total of 48,000 clones corresponding to 0.32 genome equivalent (Suzuki ). Southern blot analysis of the positive clones was carried out using a DIG-labeled LFS probe (Figure 3). In HindIII digestion, all 8 clones showed a similarly sized signals, but EcoRI digestion showed a larger size in 4F10/155 compared with the other 7 clones. This observation suggested that there are at least two LFS loci in the onion genome.

Figure 3

Southern blot analyses of eight LFS-containing BAC plasmid with LFS probe. HindIII and EcoRI indicate digestion with HindIII or EcoRI, respectively. M lane shows 1 kb ladder marker lane of the gel used on this Southern blot.

Sequence comparison between two LFS containing BAC clones

Sanger and 454 shotgun sequencing of clones 2E8/10 and 4F10/155 provided contigs of 19 and 7, respectively, that were longer than 500 bp. From Sanger shotgun analysis of 2E8/10, a LFS ORF sequence was detected in a contig containing 8623 bp, and a sequence of 20,577 bp (from upper 13 kbp to lower 7 kbp of LFS ORF) was built by assembling contigs with 5–6x coverage. In the 454 contigs (total of 138,394 bp) of 4F10/155, an LFS ORF sequence was observed in the longest contig containing 71,391 bp.

Nucleotide sequence comparison of the approximately 20 kbp contigs from both BAC exhibited high similarity over 10 kbp flanking the intronless LFS gene. Notable areas of lower similarity within this region were the region immediately upstream of the LFS gene, including its promoter, and a 1.5 kbp insertion in 2E8/10 showing BLASTX similarity to polyproteins (Figure 4). Both contigs shared a 445 bp direct repeat in 1 kbp upstream and a region of homology to other onion BAC sequences in 8–14 kbp upstream of LFS. Regions 3–6 kbp downstream showed BLASTX homology to plant Ty1-copia–like elements. This same general pattern of degenerated retroviral elements and transposons was reported in earlier onion BAC sequencing by Jakše . Sequence comparison in the region corresponding to LFS transcripts revealed five differences at sites corresponding to the following locations in GenBank accession no. AB094593: a nonsynonymous substitution 203A > T conditioning the mutation V50D in the translation, two synonymous variants (441T > C and 474C > T) in the coding region, and two variants in the 3′UTR (596A > G and 654T > C).

Figure 4

Sequence similarity and annotation of LFS flanking regions of 2E8/10 and 410/155. Gray-scale shading denotes sequence similarity over 80% detected by BLASTN alignment of sequences. Black arrow indicates intronless LFS ORF. Gray sidebar denotes a 454 bp direct tandem repeat. Gray bar denotes insertion in 2E810 sequence with BLASTX homology to polyproteins, and gray arrows denote regions with BLASTX similarity to Ty1-copia–like sequences. Black lines denote regions with BLASTN similarity to other onion BAC sequences.

Chromosomal localizations of BAC clones

The DIG-labeled BAC clones, 2E8/10 and 4F10/155, were hybridized to the mitotic metaphase chromosome preparation of A. cepa. By using the Cot-100 fraction to block the repetitive sequence hybridization, both BAC clones generated distinct signals from a single pair of somatic metaphase chromosomes (Figure 5A). The karyotype analysis revealed that signals were located on the proximal region of the long arm of chromosome 5 (Figure 5B). The identity of this chromosome was established based on its chromosome size and the position of its centromere [relative chromosome length 12.7 ± 1.0, centromere index 48.7 ± 0.7; De Vries (1990)]. Each BAC’s location from the centromere was measured (20 chromosomes from 10 metaphases per each BAC), and their relative positions were estimated. The position of 2E8/10 was 0.31 ± 0.03 and the position of 4F10/155 was 0.32 ± 0.03 (Figure 5C). No statistically significant difference was found between the position of hybridization signals in the two BAC clones (Student t-test, td = 0.195, tst = 2.04, n = 40, P = 0.05).

Figure 5

BAC-FISH using two LFS bearing clones (2E8/10 and 4F10/155) as a probe. (A) Mapping of BAC clones on onion mitotic metaphase chromosomes: Bar: 10 μm. (B) FISH karyotypes and (C) chromosome 5 extracted from a number of mitotic metaphases, which possess hybridization signal (green) in the proximal region of the long arm.

Sequence variation in LFS amplicons and transcripts

Mapping leaf transcript 454 reads from the doubled haploid line ‘CUDH2150’ to A. cepa LFS cDNA sequences (GenBank accession nos. AB094593 and AB089203) with stringent criteria revealed 119 matching reads containing three variants at intermediate frequencies (Table 2). The same variants were also observed in ‘Nasik Red’ and formed two haplotypes. In a previous study (McCallum ), we surveyed allelic variation at multiple SSR loci and failed to detect any heterozygosity in ‘CUDH2150’, suggesting that the observed variants are transcripts from duplicated LFS genes. BLAST searches of GenBank EST division revealed that, in addition to ESTs with the same haplotype as AB094593 (‘haplotype 1’), three onion ESTs exhibited the alternate haplotype (‘haplotype 2’) at these three sites (accession nos. CF451348, FK935151, and FK936343).

Table 2

Counts of 454 cDNA reads from ‘CUDH2150’ and ‘Nasik Red’ cDNA 454 cDNA sequencing classified by variants

			Total Depth		Variant Frequency %
Sitea	AB089203	Variant	CUDH2150	Nasik Red	CUDH2150	Nasik Red
439	T	C	106	37	41	49
594	A	G	75	39	48	23
652	T	C	62	31	50	26

Site positions are relative to GenBank accession no. AB089203.

Sanger sequencing of LFS amplicons from a range of diverse onion germplasm and A. roylei revealed these sequence variants as well as rarer ones (Figure 6). The most commonly observed haplotype was that matching AB094593 (‘haplotype 1’), and this was amplified from all onion populations surveyed. Trees based on alignment of cDNA and genomic sequences (Figure 7) revealed clear clustering of onion sequences into two groups corresponding to haplotypes 1 and 2 observed in 454 sequencing and an outgroup containing other Allium species. The variant distinguishing AB089203 from other LFS sequences (A570T) was not observed in any other reads, suggesting this is a PCR or sequencing error. Four singleton reads with single base differences to the haplotype 1 sequence were observed, one of which (HQ738863) conditioned a nonsynonymous M > I mutation. By contrast, several well-supported haplotypes were observed within the ‘haplotype 2‘ group corresponding to BAC 2E8/10. And a nonsynonymous variant (HQ738883, V > G at 166) was also found in the group. Average nucleotide diversity was higher in the haplotype 2 group (π = 0.0042) than in haplotype 1 (π = 0.0028).

Figure 6

Alignment of cDNA sequences from onion and related Allium species with haplotypes observed in region corresponding to bases 370–658 of GenBank accession no. AB089203 in genomic PCR amplicons of LFS genes from diverse onion germplasm. Accession numbers marked by an asterisk denotes haplotypes based on sequence variants supported only by a single read, which may represent PCR errors.

Figure 7

Cladogram of LFS sequences based on consensus maximum likelihood tree of LFS sequence alignments with 100 bootstraps.

The out-group containing related Allium species also contained two sequences amplified from A. cepa ‘Faridpuri’ and ‘AC43’ (HQ738857 and HQ738882,). Notably, these sequences exhibit a deletion corresponding to bases 578–581 of AB089203 and other variants in the 3′ UTR that observed in LFS cDNA sequences from A. roylei, A. chinense, and A. fistulosum. HQ738882 contains a premature stop (G450A) within the putative coding region. These may represent rare or ancestral alleles but could also be products amplified from other LFS loci.

The hypothesis of selective neutrality in this sequence region in the two paralog groups was tested using Tajima’s D. This test revealed a significant deviation from neutral expectations for the haplotype 1 group (D = −1.91; N = 55 sequences; P < 0.05) but not for the haplotype 2 group (D = −0.23; N = 21 sequences; P > 0.10).

Existence of distinctive LFS haplotypes in the FF+5A

The A594G and T652C variants, which distinguish the two haplotype groups, condition NsiI and SphI restriction sites, respectively, and we used this to test for existence of distinctive haplotypes in the FF+5A. This revealed both NsiI and SphI digestion of the LFS product amplified from this line but no evidence for double digestion (Figure 8). This confirms that the two putative paralogous loci are both located on onion chromosome 5.

Figure 8

Restriction analysis of LFS PCR products from chromosome 5 AMALs. Lanes 1–4: FF+5A (plant number 26); lanes 5–8: FF+5A (71); lanes 9–12: AA (‘Chiang Mai’). (−) indicates uncut PCR product, N indicates NsiI digested, S indicates SphI digested, and N+S indicates double digested by NsiI + SphI. M shows 100 bp ladder marker. The two AMALs were the sibs obtained from a single cross.

DISCUSSION

By the use of multiple approaches, this study has determined that LFS in onion is transcribed from at least two loci and that they are localized on chromosome 5. Southern blotting of eight LFS-bearing clones revealed two distinct EcoRI RFLP patterns, but assignments using alien addition lines and mapping analysis in an interspecific cross placed the gene only on chromosome 5. Furthermore, the BAC-FISH study showed co-localization of these BACs. Under the optical limit of a 1.4 numerical aperture for a conventional microscope objective, a maximum distance of only 0.2 µm can be resolved. de Jong reported that mitotic metaphase FISH can resolve 4–5 Mbp in a tomato. The level of metaphase chromosome condensation in Allium is about five times higher than in tomato metaphase chromosomes (Khrustaleva and Kik 2001). Thus, 2E8/10 and 4F10/155 clones can be distant from each other up to 25 Mbp and still be located in the same position on the mitotic metaphase chromosome. Because onion is highly heterozygous, a possible interpretation of these results is heterozygosity at the LFS locus. This possibility was eliminated through transcriptome sequencing in a doubled haploid genotype, which also revealed two distinct sequence variants. This confirms that LFS is transcribed from two distinct loci.

The observation of LFS-like sequences in onion closely resembling those from related Allium species may provide some further scope to study the evolution of LFS in Allium. These sequences may be rare or ancestral alleles at either of the major loci, but they could also be cross-amplified from other LFS loci. It is possible that such sequences could represent ancestral copies of the gene family and indicate that gene duplication is still ongoing in the onion genome. The LFS gene is small and intronless. In plants, small and intronless genes are frequently plant- or lineage-specific (Jain ). It is therefore plausible that LFS is such an example of a novel, genus-specific gene family that has arisen recently and provided a strong selective advantage. A survey by PCR across diverse germplasm suggested lower nucleotide diversity in haplotype group 1 in the region surveyed compared with the haplotype group 2 corresponding to BAC 2E8/10 locus. This is suggestive of possible differences in selective constraints; however, wider sampling of nucleotide diversity is required to test the role of purifying selection and gene conversion on these loci.

Gene duplication is known to be an important source of evolutionary innovation and adaptation (Des Marais and Rausher 2008). Mapping studies have suggested extensive duplication in onion. King reported that 21% of 91 cDNA probes of RFLP detected more than one segregating RFLP, of which 53% were unlinked and 47% were linked. These findings indicated that gene duplication has occurred with high frequency in the onion genome; the ratio of the linked duplicated RFLP was higher than in other plants. Our results demonstrate that LFS genes exist as functional linked duplicated genes, implying a low possibility for obtaining null mutants through mutant screening. However, BAC sequence analysis and a PCR experiment to distinguish two haplotypes indicated that 2E8/10 was ‘haplotype 2’ and 4F10/155 was ‘haplotype 1’, and they showed a large difference in upper flanking 1.2 kbp region of LFS between the BACs. These observations of variation in the promoter regions suggest that there may be differential expression between the paralogs. Kim speculated similarly in pink onion. More detailed surveys of paralog-specific expression might reveal mutants with lower LFS activity. The identification of the complete genomic sequence surrounding the two paralogs will now enable wider surveys for natural mutants by deep sequencing, like the variant (HQ738882) possessing premature stop that we identified in ‘Faridpuri’. Using such genotypes as a material for further mutant induction or breeding, we may achieve production of LFS null onion. A nonsynonymous variant found in this article, HQ738883 (V > G), may have enzymatic activity equivalent to wild-type LFS, as we have shown a deletion of nine C-terminal amino acids (160–169) had no effect on the enzymatic activity (Masamura ). It remains unclear whether the nonsynonymous variant HQ738863 (M > I) has enzymatic activity. However, because LFS catalyzes an intramolecular H+ substitution reaction, this variant, which changes between nonpolar amino acids, might retain the enzymatic activity of wild-type onion LFS. For the same reason, we speculate that the variant (V50D) found in 4F10/155 has a low potential for loss of enzymatic activity.

It has been reported that a well-supported QTL affecting bulb dry matter and a putative sucrose transporter gene are located on chromosome 5 (Galmarini ; Martin ; Masuzaki ; McCallum ). Dry matter is an important trait affecting firmness and storability of onion bulb and, therefore, survival under human or natural selection. It is notable that LFS is located also in this region, sufficiently close that we would expect some effects on LFS through linkage drag. Consequently this region of chromosome 5 could be an important domestication region in the onion genome, and it would be fruitful for further functional and crop evolutionary studies. Our findings provide a highly relevant case study of a duplication leading to multiple functional loci encoding a gene with important adaptive traits in onion.