Transcriptome-mining for single-copy nuclear markers in ferns

Abstract

Background: Molecular phylogenetic investigations have revolutionized our understanding of the evolutionary history of ferns-the second-most species-rich major group of vascular plants, and the sister clade to seed plants. The general absence of genomic resources available for this important group of plants, however, has resulted in the strong dependence of these studies on plastid data; nuclear or mitochondrial data have been rarely used. In this study, we utilize transcriptome data to design primers for nuclear markers for use in studies of fern evolutionary biology, and demonstrate the utility of these markers across the largest order of ferns, the Polypodiales.

Principal findings: We present 20 novel single-copy nuclear regions, across 10 distinct protein-coding genes: ApPEFP_C, cryptochrome 2, cryptochrome 4, DET1, gapCpSh, IBR3, pgiC, SQD1, TPLATE, and transducin. These loci, individually and in combination, show strong resolving power across the Polypodiales phylogeny, and are readily amplified and sequenced from our genomic DNA test set (from 15 diploid Polypodiales species). For each region, we also present transcriptome alignments of the focal locus and related paralogs-curated broadly across ferns-that will allow researchers to develop their own primer sets for fern taxa outside of the Polypodiales. Analyses of sequence data generated from our genomic DNA test set reveal strong effects of partitioning schemes on support levels and, to a much lesser extent, on topology. A model partitioned by codon position is strongly favored, and analyses of the combined data yield a Polypodiales phylogeny that is well-supported and consistent with earlier studies of this group.

Conclusions: The 20 single-copy regions presented here more than triple the single-copy nuclear regions available for use in ferns. They provide a much-needed opportunity to assess plastid-derived hypotheses of relationships within the ferns, and increase our capacity to explore aspects of fern evolution previously unavailable to scientific investigation.

Figure 1. Schematic diagrams of the ten nuclear genes for which we developed fern-specific primers.

(A) ApPEFP_C; (B) CRY2; (C) CRY4; (D) DET1; (E) gapCpSh; (F) IBR3; (G) pgiC; (H) SQD1; (I) TPLATE; (J) transducin. Each subset of the figure represents one protein-coding locus, using the most closely related Arabidopsis thaliana homolog as the template. The coding sequence is measured (in base pairs) along the bottom of the thickened horizontal line, with each locus wrapping onto a new line every 2000 base pairs, when necessary. Intron location, number, and length (in base pairs in Arabidopsis) are given above the line. Also shown below the line are the priming locations for each of the markers we developed. For gapCpSh, intron locations are based on Arabidopsis gapCp1: the first two exons of Arabidopsis gapCp2 are each one codon shorter than in gapCp1.

Figure 2. Maximum likelihood phylograms for each region, including only those taxa that were successfully sequenced from our 15-taxon genomic DNA test set.

Bold branches indicate strong support (≥70% bootstrap support). Scale bars are in units of substitutions per site. In the taxon names, “C.” and “P.” refer to Cystopteris and Polypodium, respectively. These phylograms are unrooted, but oriented as if rooted by the Cyatheales (or our best guess, when the Cyatheales accession did not sequence successfully), when space permits.

Figure 3. Combined data maximum likelihood phylogram of our 15-taxon genomic DNA test set.

Analyses were performed under our best-fitting model (model 3, see Table 3). Bold branches indicate strong support (≥70% bootstrap support); internal branches are labeled A – L for ease of discussion.

Figure 4. Flowchart of our transcriptome-mining pipeline.

Figure 5. Example of our sequence-merging protocol.

(A) In this schematic of a transcriptome alignment, aligned sequence fragments are indicated by the horizontal bars. Included are four fragments (colored) from our focal accession, which group together in the maximum parsimony tree. However, the two fragments from the 5’ end of the protein (in red) have some base pair conflicts with each other, as do the fragments from the 3’ end (in blue). Since the two sets of fragments do not overlap, and they group in the same area of the MP tree, it is not possible to determine which 5’ fragment belongs with which 3’ one. In this case we merged the sequences arbitrarily (B). The resulting alignment retains the full nucleotide data for primer-design purposes, but the relationships at the tips of the tree may be erroneous due to the two potentially chimaeric sequences.