Background

The early-diverging eudicot family Circaeasteraceae (Ranunculales) sensu APG IV [1] contains only the two monotypic genera Circaeaster Maxim. and Kingdonia Balf.f. & W.W. Smith, which were historically treated as separate families (Circaeasteraceae and Kingdoniaceae) (e.g. [2,3,4,5,6]). Kingdonia has also been placed in Ranunculaceae in the past (e.g. [7,8,9,10]). Circaeaster agrestis Maxim. can be found in China and the Himalayas, whereas Kingdonia uniflora Balf.f. & W.W. Smith is endemic to China. Fossil fruits somewhat similar to those of Circaeaster have been reported from the mid-Albian of Virginia, USA [11, 12], while no fossil record is known for Kingdonia. Both species are herbs growing at high elevations, and possess the same distinctive dichotomous venation, which is very rare among angiosperms.

The Ranunculales sensu APG IV [1] form a well-supported clade comprised of seven families: Berberidaceae, Circaeasteraceae, Eupteleaceae, Lardizabalaceae, Menispermaceae, Papaveraceae and Ranunculaceae. To date, complete plastomes have been sequenced for representatives from all of these families except Circaeasteraceae [13,14,15,16,17,18,19,20,21,22,23,24]. These and plastomes from other eudicot families have helped to successfully resolve phylogenetic relationships among early-diverging eudicots, including among most families of Ranunculales (e.g.[17,18,19, 23, 24]). This is highly promising given that the relationships among many of these families had been poorly to moderately resolved in previous studies utilizing only a few genes (e.g. [25,26,27,28,29,30,31,32,33,34,35,36,37,38]). In previous phylogenetic studies of Ranunculales based on only a few genes, Circaeasteraceae has been resolved as sister to Lardizabalaceae, but only with weak or moderate support [25, 26, 29, 32, 36, 38, 39].

Over the past decade, knowledge of the organization and evolution of angiosperm plastomes has rapidly expanded [40, 41]. Plastomes of most flowering plants possess a typical quadripartite structure with two Inverted Repeat regions (IRA and IRB) separating the Small and Large Single-Copy regions (SSC and LSC) [42]. Nevertheless, deviations from this canonical structure have been found with increasing frequency as the pace of plastome sequencing has exploded in recent years. For example, the length of the IR region has been found to vary significantly in some plant groups (e.g. [43,44,45]), and Sun et al. [23] documented six major “IR types” among 18 early-diverging eudicot plastomes, representing 12 of the 13 early-diverging eudicot families. Reconstruction of the ancestral IR gene content suggests that 18 genes were likely present in the IR region of the ancestor of eudicots [23], although representatives from Circaeasteraceae were absent from this study. Likewise, large inversions have been detected throughout the plastome in an increasing number of taxa (e.g. [46,47,48,49,50]). However, no obvious large inversions or rearrangements have been detected from early-diverging eudicot plastomes. Finally, gene loss (including pseudogenization) has also been found to be widespread among angiosperm plastomes, especially in species whose plastomes are highly rearranged [51].

To characterize plastome structural evolution in early-diverging eudicots and to better understand relationships within Ranunculales, we sequenced the complete plastomes of both extant species of Circaeasteraceae and included these two plastomes in a larger phylogenetic analysis including representatives of all major lineages of angiosperms. Consistent with previous work, we find that these complete plastome sequences improve support for phylogenetic relationships among Ranunculales, including the position of Circaeasteraceae. Moreover, we report several significant plastome structural changes, including a large inversion and several gene loss events.

Results

Plastome assemblies

Illumina paired-end sequencing produced 474,002 and 1,092,236 raw reads for Circaeaster and Kingdonia, respectively. The mean coverage of the plastome was 392.3× for Circaeaster, and 926.4× for Kingdonia. For both Circaeaster and Kingdonia, assembly yielded a single scaffold comprising the entire plastome sequence. Junction regions between the IR and Single-Copy regions were confirmed by PCR and Sanger sequencing (C8-C11 and K5-K8 in Additional file 1). Assembly statistics are presented in Table 1.

Table 1 Comparison of the plastid genomes of Circaeaster agrestis and Kingdonia uniflora
Full size table

Structure and gene content of the Circaeaster and Kingdonia plastomes

The plastome size of Circaeaster agrestis is 151,033 bp and that of Kingdonia uniflora is 147,378 bp (Fig. 1). Both plastomes possess the typical quadripartite structure of angiosperms, although both also contain several remarkable structural rearrangements. Most notably, a large ~49 kb inversion in the LSC region, including all genes from trnQ-UUG to rbcL/accD (accD is absent from Circaeaster) is present in both plastomes (Figs. 1, 2). In addition, both plastomes also share a much smaller inversion (~3.5 kb) involving all four genes from atpB to trnV-UAC (Figs. 1, 2). Circaeaster also possesses a number of other unique structural changes, including a ~ 3.5 kb inversion involving all four genes from psaI to petA (Figs. 1, 2) and a highly unusual IR structure. Specifically, the following changes have occurred within the IR of Circaeaster: (1) ndhB, rps7, and the 3′ end of rps12 have shifted to a position between trnN-GUU and ycf1 (compared to their typical positions between trnL-CAA and trnV-GAC in nearly all other angiosperms); (2) rpl32 and trnL-UAG are within the IR (rather than in the SSC region as in nearly all other angiosperms), and (3) ycf1 is almost entirely outside the IR (rather than having ~1000 bp of ycf1 within the IR, as is more typical of angiosperms). Within Kingdonia, the IR/SSC boundary has shifted to include all of ycf1, rps15, ΨndhH, and ΨndhA. In both plastomes, there are unusual arrangements of rpl32 and trnL-UAG, which in almost all other angiosperms are found adjacent to each other on the same strand within the SSC. The endpoints of these inversions were confirmed in both plastomes via PCR and Sanger sequencing using custom-designed primers (Additional files 1 and 2).

Fig. 1
figure 1

Plastome maps of Circaeaster agrestis and Kingdonia uniflora. IR, inverted repeat; LSC, large single-copy region; SSC, small single-copy region; Inv1, large inversion 1; Inv2, large inversion 2; Inv3, large inversion 3. The IR rearrangement is also indicated

Full size image
Fig. 2
figure 2

ProgressiveMauve alignment among Akebia, Circaeaster and Kingdonia showing the structural rearrangements in Circaeaster and Kingdonia. Colored blocks represent locally collinear blocks (LCBs) and are connected by lines to similarly colored LCBs, indicating homology

Full size image

Overall, Circaeaster and Kingdonia were found to possess the typical gene and intron complement of angiosperms, with a few notable exceptions (Table 1). Both plastomes contain 30 tRNA genes and four rRNA genes, as is typical in angiosperms. The plastome of Circaeaster agrestis has 77 protein-coding genes and one pseudogene (ΨinfA, which is truncated to a length of 36 bp); accD is absent. The plastome of Kingdonia uniflora only has 70 protein-coding genes due to the loss or pseudogenization of nearly all ndh genes: four genes (ndhC, ndhF, ndhI and ndhG) were absent and five (ΨndhA, ΨndhB, ΨndhD, ΨndhH and ΨndhK) were identified as pseudogenes. More specifically, the second exon is absent from ΨndhA and ΨndhB, ΨndhD is severely truncated to 18 bp (vs. 1503 bp in Circaeaster), ΨndhH is truncated to 618 bp (vs. 1182 bp in Circaeaster) in length, and ΨndhK is only 237 bp (vs. 669 bp in Circaeaster) in length. The Ks values of these pseudogenes were calculated between Circaeaster and Kingdonia (Additional file 3). A total of 32 and 14 repeats ≥30 bp in length were found in the plastome of Circaeaster and Kingdonia, respectively (Additional files 4 and 5). For comparison, the number of repeats ≥30 bp in seven other Ranunculales species are as follows: (1) 17 in Akebia trifoliata (Thunb.) Koidz.; (2) 24 in Epimedium sagittatum (Sieb. & Zucc.) Maxim.; (3) 17 in Euptelea pleiosperma Hook.f. & Thomson; (4) 29 in Mahonia bealei (Fortune) Pynaert; (5) nine in Nandina domestica Thunb.; (6) nine in Papaver somniferum L.; and (7) eight in Stephania japonica (Thunb.) Miers (Additional file 6).

Phylogenetic analyses

The final 79-gene, 99-taxon alignment used for ML analyses was 62,238 bp in length after character exclusion. The best partitioning scheme identified under the Bayesian information criteria (BIC) using relaxed clustering analysis in PartitionFinder (ln L = −1,173,388.00241; BIC 2353123.26941) contained 35 partitions. The tree with the highest ML score (ln L = −1,178,285.119460) produced by the 35-partition ML analysis (Fig. 3) shared an identical topology with the best tree from unpartitioned analysis (ln L = −1,200,753.541175) (Additional file 7), except for the relationships among Trochodendrales, Buxales and Gunneridae. The 35-partition analysis supported the sister relationship between Buxales and Gunneridae, but the support value was low (52%); while the unpartitioned analysis strongly supported the sister relationship between Trochodendrales and Gunneridae. Within Ranunculales, Eupteleaceae was found to be the earliest-diverging lineage, and Papaveraceae was sister to the clade comprised of Berberidaceae, Ranunculaceae, Menispermaceae, Lardizabalaceae and Circaeasteraceae. Lardizabalaceae and Circaeasteraceae formed a strongly supported clade that was sister to the clade of Berberidaceae, Ranunculaceae and Menispermaceae.

Fig. 3
figure 3

Phylogram of the best tree determined by RAxML for the 79-gene, 99-taxon data set using the 35-partition scheme recovered as optimal by PartitionFinder. Numbers associated with branches are ML bootstrap support values. Branches with no bootstrap values listed have 100% bootstrap support

Full size image

Discussion

Plastome structure and gene content

The unusual structural rearrangements and gene losses (especially the loss of the ndh genes) detected in the two Circaeasteraceae plastomes represent the first reported among early-diverging eudicot plastomes, and hence shed important insight into the evolution of early eudicot plastomes. The fact that two of the observed inversions (the ~49 kb and ~3.5 kb inversions) are shared by Circaeaster and Kingdonia suggests that they predate the evolutionary split between these two genera. Although uncommon, relatively large inversions have been detected in a number of other angiosperm lineages and often serve as useful phylogenetic markers [50, 52,53,54]. Some of the best examples of relatively large inversions that are synapomorphies for clades of flowering plants include the 22.8 kb inversion shared by all Asteraceae except Barnadesioideae [53, 55], the 78 kb inversion shared by all Fabaceae subtribe Phaseolinae [56], and the 36 kb inversion present in all core genistoid legumes [50, 57]. Highly rearranged plastome structures also characterize a number of other angiosperm lineages, such as Campanulaceae, Geraniaceae, and the IR-lacking clade of Fabaceae, and these are associated with greatly elevated rates of molecular evolution and large numbers of short repeats [58]. In some cases the endpoints of large plastome inversions have been found to be associated with short inverted repeats (sIRs), although we did not detect sIRs in association with the inversion endpoints in Circaeaster or Kingdonia.

The IR regions of Circaeaster and Kingdonia are also structurally unique among angiosperms, with several rearrangements. The most unusual of these involves the transposition of the ndhB, rps7 and 3′ end of the rps12 gene to a point near the junction of the IRB and the SSC region (Fig. 1). These three genes form a transcriptional operon [59] and this operon is not disrupted in Circaeaster, nor does its transposition interrupt adjacent operons. The IR/SSC endpoints themselves are also rearranged in Circaeaster, with rpl32 and trnL-UAG within the IR and non-adjacent to ndhF, unlike almost all other angiosperms [54]. The IR boundaries of Kingdonia are also unusual for their expansion to include several genes that are normally in the SSC (e.g. ycf1, rps15), resulting in a much smaller and rearranged SSC (which is also influenced by the loss of ndh genes; see below). The exact sequence of rearrangements that could account for the unusual IR arrangements of Circaeasteraceae is clearly complicated and hence difficult to reconstruct. IR expansion and contraction is well-known in a number of other angiosperm lineages (e.g. [45]), including within early-diverging eudicots, which have been found to possess a number of expansions and contractions [23]. Given the expansions and rearrangements observed in Circaeaster and Kingdonia, neither fit into any of the six IR types for early-diverging eudicots delineated in Sun et al. [23], and thus we designate two new early-diverging eudicot IR types for Circaeaster (type G) and Kingdonia (type H) (Fig. 1).

Usually, gene content is highly conserved among photosynthetic angiosperm plastomes [50, 60], but in Circaeasteraceae, a number of genes have been lost or pseudogenized, each of which has been found to be lost repeatedly across angiosperms. For example, the loss of accD in Circaeaster is mirrored in a number of other lineages where accD is pseudogenized or absent, e.g., grasses [61], Lobeliaceae [62], Campanulaceae [52, 63, 64], Acorus [65], Oleaceae [66], Pelargonium [67], and Lolium perenne [68]. Likewise, infA is also known to be a pseudogene in numerous other angiosperms, including two Ranunculales (Ranunculus macranthus and Epimedium sagittatum) [23], tobacco [69], and numerous rosids [70]. Whether accD or infA have been transferred to the nucleus in Circaeaster is unknown.

The loss or pseudogenization of nearly all ndh genes from Kingdonia has also been observed in a number of other land plants. The ndh genes encode subunits of the plastid NDH (NADH dehydrogenase-like) complex, which permits cyclic electron flow associated with Photosystem I and hence facilitates chlororespiration [71, 72]. Although the NDH complex is widely retained across land plants, it has been found to be dispensable under optimal growth conditions and the plastid ndh genes have been lost in a number of autotrophic and heterotrophic lineages [72, 73]. For example, the plastid ndh loci have been lost or pseudogenized en masse from parasitic plants such as Orobanchaceae and Cuscuta (Convolvulaceae) [74,75,76], from mycoheterotrophs including several orchids [77] and Petrosavia (Petrosaviaceae) [78], and from autotrophs including Gnetales, conifers, Najas (Hydrocharitaceae), Carnegiea (Cactaceae), and Erodium (Geraniaceae) [79,80,81,82,83,84]. It is not clear whether the ndh genes in Kingdonia have been transferred to the nucleus or whether their loss represents the complete loss of the NDH complex, but in any case Kingdonia is the only known early-diverging eudicot that has experienced ndh pseudogenization and loss.

Moreover, the loss of the ndh genes in Kingdonia accounts for the smaller overall size of the Kingdonia plastome and may have also played an indirect role in the expansion of the IR of Kingdonia. The complete loss of ndhF, which normally occupies a position immediately adjacent to the IRB/SSC boundary, may have led to instability of the IR/SSC boundaries, leading to rearrangements of the SSC and IR. This hypothesis is supported by other recent studies in orchids [77] and Najas flexilis [82] where ndhF loss is associated with shifts in the IR/SSC boundary.

Phylogeny of Ranunculales

The circumscription of Ranunculales was long controversial (e.g. [3, 4, 9, 10, 85, 86]), but molecular phylogenetics has clarified the delimitation of Ranunculales to Berberidaceae, Circaeasteraceae, Eupteleaceae, Lardizabalaceae, Menispermaceae, Papaveraceae, and Ranunculaceae [1, 5, 6, 26, 29, 31, 32, 36, 38]. While the expansion of Circaeasteraceae to include Kingdonia is accepted by a majority of taxonomists [1], the rank and position of Kingdonia have long been in dispute [38]. The complete plastome sequence data strongly support the sister relationship between Kingdonia and Circaeaster, in concordance with previous molecular results [25, 26, 31, 32, 36, 38, 87]. The two inversions and the rare, open dichotomous leaf venation shared by these taxa are good synapomorphies that additionally support the placement of Circaeaster and Kingdonia in one family.

Conclusions

The plastomes of Circaeaster agrestis and Kingdonia uniflora provide the first reference genome sequences for Circaeasteraceae, which will enrich the sequence resources of plastomes in early-diverging eudicots. The unusual rearrangements including large inversions and unusual IR structure detected in the Circaeasteraceae plastomes will help us better characterize plastome structural evolution in early-diverging eudicots. Phylogenetic analyses of the 79-gene, 99-taxon data set confirmed the position of Circaeasteraceae in Ranunculales, with maximum support as sister to Lardizabalaceae. The two Circaeasteraceae plastomes will also be of benefit for further phylogenomic analyses within early-diverging eudicots.

Methods

Taxon sampling, chloroplast DNA isolation, high-throughput sequencing

Fresh leaves of Circaeaster agrestis were collected from Shennongjia, Hubei Province, China, in 2015, and from Kingdonia uniflora in Meixian, Shanxi Province, China, in 2016. Voucher specimens (Circaeaster agrestis: Y.X. Sun 1510; Kingdonia uniflora: Y.X. Sun 1606) were deposited at the Herbarium of Wuhan Botanical Garden, Chinese Academy of Sciences (HIB). For both species, high-quality chloroplast DNA was obtained following the plastid DNA extraction method of Shi et al. [88]. The sequencing libraries were constructed and quantified following the methods introduced by Sun et al. [23]. For both plastomes, a 500-bp DNA TruSeq Illumina (Illumina Inc., San Diego, CA, USA) sequencing library was constructed using 2.5–5.0 ng sonicated chloroplast DNA as input. Libraries were quantified using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and by real-time quantitative PCR. Libraries were then multiplexed, and 2 × 125 bp sequencing was performed on an Illumina HiSeq 2000 platform at the Beijing Genomics Institute.

Plastome assembly, annotation, and structural analyses

Following Sun et al. [23], duplicate reads, adapter-contaminated reads, and reads with more than five Ns were filtered out. Remaining, high-quality reads were assembled into contigs with a minimum length of 1000 bp using CLC Genomics Workbench with default parameters, except for a word size value of 60.

Plastomes were annotated using DOGMA [89] and through comparison with the sequences of published early-diverging eudicot plastomes. Physical maps were drawn using GenomeVx [90], followed by subsequent manual editing with Adobe Illustrator CS5. Boundaries for tRNAs were identified with tRNAscan-SE 1.21 [91] and confirmed by comparison with available early-diverging eudicot plastome sequences. The finished genomes were deposited in GenBank (Table 1).

To investigate plastome structural evolution, whole plastome alignment between Circaeasteraceae and representatives of other early-diverging eudicot families was performed with ProgressiveMauve v 2.4.0 [92], including only one copy of the IR (IRB), and locally collinear blocks (LCBs) were identified. Because the 18 reported early-diverging eudicot plastomes in Sun et al. [23] share the same gene order, and because Circaeasteraceae was resolved as sister to Akebia in present research, the Akebia plastome was used as the reference sequence for ProgressiveMauve comparisons. mVISTA [93] was employed to generate sequence identity plots. The number and location of repeat elements in the plastomes of Circaeaster and Kingdonia as well as seven other Ranunculales species (Akebia trifoliata, Epimedium sagittatum, Euptelea pleiosperma, Berberis bealei, Nandina domestica, Papaver somniferum and Stephania japonica) were determined by REPuter [94], with a minimum size of 30 bp and a Hamming distance of 1. Before performing the analysis, one copy of the IR was removed.

Phylogenetic analyses

All protein-coding regions were extracted from the plastomes of Circaeaster and Kingdonia. These sequences were then added manually to the 97-taxon alignment of Sun et al. [23], resulting in a data set with complete coverage of early-diverging eudicot families sensu APG IV [1]. GenBank information for all plastomes used for present phylogenetic analyses can be found in Additional file 8. Regions of ambiguous alignment and sites with more than 80% missing data were excluded from the alignment.

Maximum likelihood (ML) analyses were conducted using RAxML version 7.4.2 [95], under the general time-reversible (GTR) substitution model and the Γ model of rate heterogeneity. We conducted both unpartitioned and partitioned analyses. PartitionFinder version 1.1.1 [96] was used to select the best-fit partitioning scheme, considering all 237 possible gene-by-codon position partitions (79 genes × 3 codon positions). For both ML analyses, a single set of branch lengths for all partitions was used. Ten independent ML searches were conducted and bootstrap support was estimated with 1000 bootstrap replicates.