Plant materials and DNA isolation.

Plants were field grown in Palmer, AK. Young, actively growing leaves were collected from 58 clonal rhubarb accessions on 22 June 2011 and shipped for overnight delivery to the USDA-ARS National Clonal Germplasm Repository in Corvallis, OR. The accessions included 53 Rheum ×rhababarum and one additional accession each of: R. officinale, R. palmatum, R. palmatum ‘Rubra’, R. palmatum var. tanguticum, and an unknown Rheum species (Table 1). Within 24 h of receiving the leaves, 30 to 50 mg of leaf tissue from each accession was placed in a cluster tube, frozen in liquid nitrogen, and stored at –80 °C as described by Gilmore et al. (2011).

Table 1.

Fifty-three Rheum ×rhababarum accessions and five Rheum species representatives evaluated from the U.S. Department of Agriculture–Agricultural Research Service National Plant Germplasm System (NPGS).^z

View Table

DNA sequencing and SSR discovery.

Rheum ×rhababarum accession RB14 ‘Timperley Early’ was one of eight Illumina^® sequencing libraries that were pooled and sequenced in one lane with the Illumina^® Genome Analyzer II at the Oregon State University Center for Genome Research and Biocomputing (CGRB). Each library had a different bar-coded adapter to allow assignment of sequences to the correct plant. The remaining seven libraries were representatives of Peonia species and hybrids (Gilmore et al., 2013). A single Illumina^® library was prepared from DNA of RB14 ‘Timperley Early’ and consisted of: DNA sonication [Bioruptor XL (BR_XL); Diagenode, Denville, NJ]; DNA strand end repair; addition of an adenine base to the 3′ ends; ligation of a custom Illumina^® sequencing adapter containing a 4-bp barcode (AGGT) to the DNA fragments; purification of the ligation products to remove any unligated adapters or adapter–adapter products using a low melting point 1.5% agarose gel (Lonsa NuSieve, Rockland, ME); and excision of a 350-bp DNA band from the gel smear with a scalpel (5-Prime SafeXtractor-25; Fisher, Waltham, MA). The adapter-modified DNA fragments that had adapter molecules on both ends were selectively enriched by polymerase chain reaction (PCR) for 18 cycles. The PCR products were purified using a QIAquick PCR Purification Kit (Qiagen, Valencia, CA) and then separated on a 1.5% agarose gel to verify library size and to visually estimate the fragment concentration. The extracted DNA was quantified, and the 260/280 absorbance ratio was determined using a ultraviolet-Vis spectrophotometer (NanoDrop ND-1000; Thermo Scientific, Wilmington, DE). The library size was determined with a Bioanalyzer (Agilent 2100; Agilent Technologies, Wilmington, DE) at the CGRB. After diluting and pooling in equimolar amounts, the bar-coded Rheum library and seven other libraries (7.5 pM) (Gilmore et al., 2013) were submitted for paired-end 80-bp sequencing in a single lane by the Illumina^® Genome Analyzer II at the CGRB. Illumina^® Version 3.0 reagents were used for cluster generation and sequencing, image acquisition, and base calling using the Illumina^® pipeline Version 1.5. The resulting microreads were sorted by barcodes using a custom PERL script BCSORT (Knaus, 2011). Sorted reads were searched only for di-nucleotide motifs. The di-nucleotide motif guidelines were microreads containing at least four perfect repeats. Both nucleotides were represented at least four times and had fewer than eight ambiguous (N) bases. Paired-end microsatellite-containing reads were joined into a single sequence (contig) by concatenating read 1, the reverse complement of read 2, and separated by 50 Ns. This was used to identify the break between microreads. A stringent filter was applied to identify microreads with microsatellites located near the center of the sequence. The output was filtered for redundant sequences (identity 95% or greater) to a single unique microread by clustering the reads using the program cd-hit-454 (Niu et al., 2010). This stringent filter yielded the largest possible flanking sequences for subsequent primer design (Jennings et al., 2011).

Microsatellite marker development.

PCR primers were designed for each SSR-containing singleton cluster and contig sequence with BatchPrimer3 (You et al., 2008). Default settings were used except for product size, which was increased to a maximum of 300 bp to the 5′ end of each forward primer; an M13 sequence (TGTAAAACGACGGCCAGT) was added to allow for an economic method of fluorescent labeling of PCR products following the procedure outlined by Schuelke (2000). The 97 M13-tagged forward primers and corresponding reverse primers, along with universal fluorescently labeled M13 (-21) forward primers (WellRED D2, D3, or D4) were ordered from Integrated DNA Technologies (IDT, San Diego, CA). The primers were tested for amplification in four rhubarb accessions, RB7, RB10, RB12, and RB45, and products were separated by 2% agarose gel electrophoresis and visualized with ethidium bromide. Thermocycler amplification of the M13-tagged SSRs was performed with a touchdown PCR consisting of an initial denaturing step of 94 °C for 3 m, then 10 cycles of 94 °C for 40 s, 62 °C for 45 s (decreasing the annealing temperature by 1.0 °C per cycle), and 72 °C for 45 s followed by 20 cycles of 94 °C for 40 s, 52 °C for 45 s, and 72 °C for 45 s; eight cycles of 94 °C for 40 s, 53 °C for 45 s, and 72 °C for 45 s; and a final extension of 72 °C for 30 min. The 15-μL PCR reaction mix contained: 3 μL of 5× GoTaq DNA Polymerase Buffer (Promega Corp., Madison, WI); 1.2 μL of 2.5 mm dNTPs; 1.2 μL of 25 mm MgCl₂; 0.075 μL of 5U/μL GoTaq DNA Polymerase (Promega Corp.); 0.18 μL of 10 μM forward primer; 0.75 μL of 10 μM reverse primer; 0.75 μL of 10 μM M13-fluorescent tag, WellRED D2, D3, or D4; and 1.5 μL of 3 ng·μL⁻¹ template DNA.

Ninety SSR markers generated a PCR product, which were then evaluated for polymorphism in eight rhubarb accessions, ‘Crimson Delicious’ (RB7), ‘Coulter MacDonald’ (RB10), Penn. State 3 (RB12), ‘New Zealand’ (RB19), ‘Parson’s Crimson’ (RB32), ‘Crimson Cherry’ (RB39), ‘Lober Blut’ (RB41), and R. officinale (RB45). The PCR products were separated by 2% agarose gel electrophoresis and visualized with ethidium bromide. Of those 90, 75 that appeared polymorphic on agarose were screened by capillary electrophoresis using the CEQ 8000 (Beckman Coulter, Brea, CA) to confirm polymorphism and evaluate ease of scoring.

Fingerprinting and genetic diversity evaluation.

Twenty-five primer pairs were selected for fingerprinting and genetic diversity assessment of the 58 accessions in the rhubarb collection (Tables 2 and 3). The PCR products were pooled into 12 multiplexes before capillary electrophoresis to maximize the number of PCR fragments separated in one capillary channel. Each multiplex contained fragments that did not overlap in size and had different fluorescent colors. Allele sizing and visualization were performed using the fragment analysis module of the CEQ 8000 software (Beckman Coulter). Alleles were scored by fitting peaks into bins that were less than one nucleotide wide. The exact copy number of each allele was not inferred in the rhubarb individuals as a result of unknown ploidy. Therefore, individuals were scored for the presence or absence of each allele and each allele was treated as a separate locus.

Table 2.

Primer names and sequences (5′-3′), number of fragments (A), and observed size range at 25 di-nucleotide repeats in 58 Rheum accessions.^z

View Table

Table 3.

List of 29 additional simple sequence repeat primers screened for polymorphism in eight Rheum ×rhabarum accessions ‘Crimson Delicious’ (RB7), ‘Coulter MacDonald’ (RB10), Penn. State 3 (RB12), ‘New Zealand’ (RB19), ‘Parson’s Crimson’ (RB32), ‘Crimson Cherry’ (RB39), ‘Lober Blut’ (RB41), and R. officinale (RB45) either by 2% agarose gel electrophoresis (AGE) or by capillary electrophoresis (CE).^z

View Table

Frequency-based genetic distance was computed with PowerMarker Version 3.25 (Liu and Muse, 2005) using the proportion of shared alleles. The distance matrix was used for cluster analysis using the neighbor joining (NJ) algorithm. Statistical significance was determined using 1000 bootstrap iterations.

Microsatellite locus discovery.

In all, 48.5 million bar-coded sequence reads were generated in the Illumina^® flow cell lane that contained one rhubarb library and seven peony libraries. The rhubarb library comprised 1.3 million reads compared with a high of 14 million bar-coded reads for Paeonia rockii and a low of 2.8 million reads for P. delavayi var. lutea × P. ×suffruticosa that were included in the same lane (Gilmore et al., 2013). The DNA used to prepare the rhubarb Illumina^® library was viscous, indicating poor quality, which may have led to fewer sequence reads for rhubarb than was previously found for the seven peony libraries (Gilmore et al., 2013). The number of primer pairs designed to amplify di-nucleotide SSR repeats was proportional to the number of bar-coded sequence reads obtained from each library. Whereas only 97 primer pairs were designed from the rhubarb library that generated the smallest number of sequences, 2219 primer pairs were designed from the library that generated the largest number of sequence reads, P. rockii. Still, at an estimated reagent and sequencing cost of ≈$215, we obtained many new SSRs in rhubarb, a crop in which only 10 SSRs were previously described (Chen et al., 2009).

Of the 97 rhubarb primer pairs designed to amplify di-nucleotide motifs present in the sequenced library, 25 primer pairs were polymorphic and were evaluated further in this study. Of the remaining 72 primer pairs, 22 failed to amplify or produced unclear results and were discarded; 18 were monomorphic; three were monomorphic in the R. ×rhababarum accessions but had a unique allele within R. officinale; seven were polymorphic but were not evaluated further; nine were polymorphic but failed to amplify in one or more of the eight tested accessions; and 13 generated polymorphic DNA fragments based on agarose gel analysis but need further evaluation by capillary electrophoresis to confirm polymorphism (Table 3). Therefore, observed and potential polymorphism in the 29 primer pairs that exhibited polymorphism in the seven R. ×rhababarum test sample panel used needs further evaluation and is provided (Table 3). It is possible that a larger number of SSR markers could be generated from this library by optimizing PCR amplification conditions for the 22 primer pairs that failed to amplify and by mining the sequence data for tri- and tetra-nucleotide repeats. The raw sequence data from this study are available at the sequence read archive at the National Center for Biotechnology Information (NCBI), accession number SRA054037 (NCBI, 2012).

Microsatellite marker diversity.

The 25 SSR primer pairs generated 314 fragments for an average of 12.56 fragments per SSR primer pair in the 58 accessions (Table 2). The number of fragments, A, ranged from three for Rhe57 to 30 for Rhe72 and on average was larger than the 4.9 previously reported by Zhang et al. (2008). This difference in A is not surprising given the small number of samples (12 R. tanguticum) used by Zhang et al. (2008) compared with the 58 accessions in this study. Ploidy may also have been a factor. However, A is comparable to the 14.6 obtained in the 114 R. tanguticum from 10 geographically separate populations evaluated with seven SSRs (Chen et al., 2009).

Fingerprinting and genetic diversity in rhubarb.

Based on NJ cluster analysis using the proportion of shared allele distance for 25 SSRs, each of the 58 accessions was uniquely identified (Fig. 1). Even the nine sets of accessions that had similar or identical names had unique SSR profiles (Table 1). Accessions from all but four (‘Strawberry’, ‘Crimson Red’, ‘Prince Albert’, and ‘German Wine’) of these similarly named individuals were grouped together in the dendrogram indicating a close relationship (Fig. 1). One unnumbered ‘Strawberry’ from Palmer did not group with the other two accessions designated as ‘Strawberry’ (RB34 and RB61) and had a pairwise genetic distance of 0.176 and 0.170 to RB34 and RB61, respectively. This finding suggests that the unnumbered ‘Strawberry’ accession is mislabeled as speculated by personnel at the ASPGB (Fig. 1). The genetic distance was 0.296 between the two ‘Crimson Red’ (RB13, RB22) accessions, 0.239 between the two ‘Prince Albert’ (RB24, RB61), and 0.298 between the two ‘German Wine’ morphotypes (RB30, RB31). The distances indicated that these three similarly named accessions were genetically different. Genetic distance among these 58 accessions ranged from 0.012 between ‘Sutton Seedless’ (RB47) and ‘Cooper’ (RB35) to 0.44 between ‘Paragon’ (RB52) and R. palmatum (RB50) and averaged 0.113 (Fig. 1). Rhubarb is clonally propagated so the genetic distance should be zero among plants of the same genotype. However, rhubarb readily outcrosses, is wind-pollinated, and the seeds easily germinate. Contamination from seedlings could explain the genetic variation we observed among these similarly named accessions. These results agree with Kuhl and DeBoer’s (2008) suggestion that the two ‘German Wine’ or ‘Prince Albert’ or ‘Crimson Red’ accessions could have arisen from seedlings. Mislabeling could also have occurred during exchange of these rhubarb clones.

View Full Size

Neighbor joining (NJ) dendrogram of 58 Rheum accessions based on 25 simple sequence repeats (SSRs) using the proportion of shared allele distance. Bootstrap support based on 1000 iterations is indicated above each branch if it exceeded 50%. Five groups (‘Cherry Red’, ‘Canada Red’, ‘Parson’s Crimson’, ‘Valentine’, and ‘Sutton’) with bootstrap support 72% or greater were named following Kuhl and DeBoer (2008) in the amplified fragment length polymorphism (AFLP)-based unweighted pair group method with arithmetic mean (UPGMA) dendrogram. Three additional groups with strong bootstrap support were also observed: ‘Ben Vanderweel’ (72%), ‘MacDonald’ (74%), and ‘Coulter MacDonald’ (82%).

Citation: Journal of the American Society for Horticultural Science 139, 1; 10.21273/JASHS.139.1.22

• Download Figure
• Download figure as PowerPoint slide

The majority of the well-supported groups (72% or greater bootstrap support) in this study was also well-supported based on AFLP analysis of 37 culinary rhubarb accessions and four species representatives (greater than 90% bootstrap support; Kuhl and DeBoer, 2008), suggesting close genetic relationships. These five well-supported groups were named following Kuhl and DeBoer (2008) and include: 1) ‘Cherry Red’; 2) ‘Canada Red’; 3) ‘Parson’s Crimson’; 4) ‘Valentine’; and 5) ‘Sutton’. Three additional groups with strong bootstrap support were also observed: ‘Ben Vanderweel’ (72%), ‘MacDonald’ (74%), and ‘Coulter MacDonald’ (82%) (Fig. 1). The Palmer Genebank obtained ‘Coulter MacDonald’, ‘Coulter McDonald’, and ‘Plum Hutt’ from the University of Minnesota. These cultivars originated from the Morden Experiment Station, Morden, Manitoba, Canada. ‘Kremlin Red’ also grouped with these accessions. This ‘Kremlin Red’ accession was found near the original USDA building (commonly referred to as the “Kremlin” building) in downtown Palmer, AK, and may have been brought from Minnesota by one of the 55 founding families that colonized Palmer, AK, in 1935 (Estelle, 2009).

There is limited historical and pedigree information about these rhubarb accessions and this severely limits discussion of the observed clustering. Furthermore, the deep branching observed in the dendrogram that was generated from binary analyses of SSR data across varying levels of ploidy in Rheum precludes inference of relationships among these accessions. Still, comparable well-supported grouping based on Kuhl and DeBoer’s (2008) AFLP-based analyses and the current SSR analyses indicates close relationships between cultivars that are present in these groups.

Similarities in clustering were observed between the SSR-based dendrogram (Fig. 1) and the unweighted pair group method with arithmetic mean dendrogram generated for 36 culinary rhubarb cultivars and two rhubarb species based on 19 morphological and phytochemical characters (Pantoja and Kuhl, 2009). Many accessions from well-supported groups in this study also grouped together based on morphological and phytochemical traits with some exceptions: ‘Moore’s Red-Right-Thru’ was in the ‘Cherry Red’ group; ‘Canada Red’ and ‘Coulter MacDonald’ grouped together; ‘Minn No. 8’ grouped with ‘Parson’s Crimson’, whereas ‘Strawberry’ grouped with ‘The Sutton’ and ‘German Wine’. Based on morphological and phytochemical differences, ‘Sutton’ and ‘The Sutton’ were different and did not group together (Kuhl and DeBoer, 2008); however, they grouped closely in the ‘Sutton’ group with a genetic distance of 0.038 based on SSR analysis (Fig. 1). On the other hand, ‘Mammoth’ and ‘Cooper’ could not be separated based on morphological and phytochemical characters but were easily separated in the ‘Sutton’ group based on SSR analysis (Fig. 1). Therefore, morphologically similar cultivars can be differentiated with SSR markers.