The complete chloroplast genome of Ginkgo biloba reveals the mechanism of inverted repeat contraction.

We determined the complete chloroplast genome (cpDNA) of Ginkgo biloba (common name: ginkgo), the only relict of ginkgophytes from the Triassic Period. The cpDNA molecule of ginkgo is quadripartite and circular, with a length of 156,945 bp, which is 6,458 bp shorter than that of Cycas taitungensis. In ginkgo cpDNA, rpl23 becomes pseudo, only one copy of ycf2 is retained, and there are at least five editing sites. We propose that the retained ycf2 is a duplicate of the ancestral ycf2, and the ancestral one has been lost from the inverted repeat A (IR(A)). This loss event should have occurred and led to the contraction of IRs after ginkgos diverged from other gymnosperms. A novel cluster of three transfer RNA (tRNA) genes, trnY-AUA, trnC-ACA, and trnSeC-UCA, was predicted to be located between trnC-GCA and rpoB of the large single-copy region. Our phylogenetic analysis strongly suggests that the three predicted tRNA genes are duplicates of trnC-GCA. Interestingly, in ginkgo cpDNA, the loss of one ycf2 copy does not significantly elevate the synonymous rate (Ks) of the retained copy, which disagrees with the view of Perry and Wolfe (2002) that one of the two-copy genes is subjected to elevated Ks when its counterpart has been lost. We hypothesize that the loss of one ycf2 is likely recent, and therefore, the acquired Ks of the retained copy is low. Our data reveal that ginkgo possesses several unique features that contribute to our understanding of the cpDNA evolution in seed plants.

Introduction

Ginkgo (Ginkgo biloba L.), also known as maidenhair tree, is a well-known living gymnosperm fossil with edible seeds, medicinal efficacy, and ornamental value (Pang et al. 1996). Fossil records suggest that during the late Mesozoic and early Tertiary era (ca. 120–60 Ma), the genus Ginkgo reached its highest species diversity and was widespread in the Northern Hemisphere (Gong et al. 2008). Today, ginkgo is the only living species left within the family Ginkgoaceae, and its natural habitat is restricted to small areas in China (Shen et al. 2005).

The presence of two large inverted repeats (IRs) is one of the most remarkable features in the chloroplast genomes (cpDNAs). In land plants, dynamic expansion/contraction of IRs has been previously reported in some lineages, such as Apioideae (Plunkett and Downie 2000), monocots (Wang et al. 2008), ferns (Wolf et al. 2010), and Pinaceae which have extremely reduced IRs (Lin et al. 2010). The fluctuating lengths of IRs contribute to increase/decrease of cpDNA sizes and can be utilized to address phylogeny but with the need of caution (Wolf et al. 2010).

Using gene mapping and cross-hybridization methods, Palmer and Stein (1986) constructed the first cpDNA map of ginkgo and reported its IR length of approximately 17 kb. Apparently, the IR of ginkgo is significantly shorter than those of most angiosperms (ca. 20–28 kb, Chumley et al. 2006) and Cycas taitungensis (ca. 25 kb, Wu et al. 2007), which indicates that ginkgo has experienced an IR contraction. However, the mechanism of IR contraction in ginkgo remains unclear.

Therefore, this study aimed to 1) elucidate the cpDNA organization of ginkgo with reference to other gymnosperms and 2) expand our understanding of cpDNA diversity and evolution as part of our long-term gymnosperm cpDNA evolutionary study. Here, we report several unique characteristics of ginkgo cpDNA, propose the underlying mechanism of its IR contraction, and discuss the evolution of an unusual transfer RNA (tRNA) gene cluster.

Materials and Methods

Genomic DNA Extraction

Genomic DNA (gDNA) was extracted from fresh young leaves of a ginkgo plant in the greenhouse of Academia Sinica by use of a CTAB-based protocol (Stewart and Via 1993). The purity and integrity of the extracted gDNA were measured and judged by the OD 260/280 ratio and gel electrophoresis, respectively. The gDNA with a 260/280 ratio greater than 1.8 was collected for polymerase chain reaction (PCR) experiments.

Amplification and Sequencing

The cpDNA fragments were amplified using a long-range PCR method with LA Taq (Takara Bio Inc., Shiga, Japan) and specific primers (supplementary table 1, Supplementary Material online). Amplicons were purified (260/280 ratio = 1.8–2.0; 260/230 ratio > 2), then sequenced by use of an Illumina GA IIx sequencer (YOURGENE BIO SCIENCE Co., New Taipei City, Taiwan). We trimmed short reads (73 bp) of paired-end sequencing using CLC Genomic Workbench 4.9 (CLC Bio, Aarhus, Denmark) with an error probability <0.05 and then assembled these trimmed reads in the same software without any reference information. Regions with <200× coverage depth were trimmed off manually, and these trimmed regions were considered as gaps. Finally, the average coverage depth of contigs is approximately 2080×, which is greatly larger than the proposed minimum coverage depth for cpDNAs (30×) (Straub et al. 2012). All gaps between contigs were filled with sequences of specific PCR products.

Annotation

We used DOGMA (Wyman et al. 2004) and BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to annotate protein coding, ribosomal RNA (rRNA), and tRNA genes. All tRNA genes were further verified by their structures predicted by tRNAscan-SE 1.21 (Schattner et al. 2005).

Examination of RNA-Editing Sites

RNA extraction and reverse transcription polymerase chain reaction experiments involved use of the Plant Total RNA Miniprep Purification Kit (Gene Mark Co., Taiwan) and the RevertAid First Strand cDNA Synthesis Kit (Fermentas Inc., Glen Burnie, MD), respectively. The obtained cDNAs were used as PCR templates for examining specific RNA-editing sites.

Estimation of Synonymous Rates (Ks)

Ks values of genes were estimated by use of PAL2NAL 1.3 (Suyama et al. 2006). Amborella genes were used as the reference.

Phylogenetic Analyses

Thirty-five tRNA sequences of ginkgo cpDNA were aligned by use of ClustalW (Thompson et al. 1994). The aligned sequences were used to construct a maximum-likelihood (ML) tree with a General time reversible + Gamma + Proportion Invariant (GTR+G+I) model and 1,000 bootstrapping analyses in MEGA 5 (Tamura et al. 2011).

Results and Discussion

Characteristics of Ginkgo cpDNA

The cpDNA of ginkgo (Accession number: AB684440) is a circular molecule of 156,945 bp with a pair of IRs separated by large single-copy (LSC) and small single-copy regions (fig. 1), which agrees well with the restriction mapping of Palmer and Stein (1986), although the total lengths slightly differ. We found that the shortened IR previously noted by Palmer and Stein (1986) is due to the complete loss of the ycf2 from the IRA. We identified 120 unique genes in ginkgo cpDNA: 81 protein-coding genes, 35 tRNA genes, and 4 rRNA genes. A total of 14 genes are duplicated, including three protein-coding genes, six tRNA genes, and four rRNA genes in the IR, as well as one tRNA gene in the LSC region. Thirteen protein-coding genes and eight tRNA genes have introns. The overall AT content is 60.4% (protein-coding genes, 61.1%; tRNA genes, 46.1%; rRNA genes, 44.7%; introns, 60.2%; intergenic spacers, 63.2%). We detected five C-to-U RNA-editing sites and experimentally verified them at the initial codons of petL and rps8 and the terminal codons of petL, rps4, and ndhC.

F

Complete cpDNA map of Ginkgo biloba. Genes inside and outside the circle are transcribed clockwise and counterclockwise, individually. *: genes with introns. Ψ: pseudogenes.

Comparisons of Ginkgo and a Cycad cpDNAs

Because cycads and ginkgo are the two most ancient lineages of gymnosperms, we compared their cpDNA features. The cpDNA organizations of both ginkgo and C. taitungensis are similar (table 1), except that ginkgo has only a single copy (SC) of ycf2 and its rpl23 has become pseudo, and Cycas lost the trnT-GGU originally located between psbD and trnE-UUC in the LSC region (fig. 1 in Wu et al. 2007). These events led to a downsizing of ginkgo cpDNA. In addition, ginkgo cpDNA contains a specific cluster of three novel tRNA genes (trnSeC-UCA, trnC-ACA, and trnY-AUA) that are located between the rpoB and the trnC-GCA of the LSC region (fig. 1).

Table 1

Comparison of cpDNA Features between Ginkgo and Cycas

Features	Ginkgo biloba	Cycas taitungensis
Size (bp)	156,945	163,403
LSC	99,221	90,216
SSC	22,258	23,039
IR	17,733	25,074
% AT content	60.4	60.5
% Coding genes	54.6	57.2
RNA-editing sitesa	5	37b (Chen et al. 2011)
Total number of genes	134	133
Protein-coding genes	84	87
Duplicated genes	14	15
tRNA genes	42	38
rRNA genes	8	8
Genes with introns	21	21

NOTE.—SSC, small single copy.

Editing sites with experimental verification.

Partial editing sites are included.

Pseudogenization of rpl23

In addition to the dysfunctional tufA reported by Wu et al. (2007), rpl23, which is retained in many land plants and is near the junction of IRB and LSC regions, becomes pseudo in ginkgo (viz. Ψrpl23). In gymnosperms, loss of rpl23 was previously reported in gnetophyte cpDNAs (Wu et al. 2009). The Ψrpl23 of ginkgo has a truncated 5′ region as compared with the functional rpl23 of Cycas. These data suggest that rpl23 was independently lost from these two gymnosperm lineages.

trnH-GUG as an Evolutionary Footprint Caused by IR Contraction

As mentioned previously, loss of an ycf2 copy downsized the IRs and the cpDNA of ginkgo. Therefore, two questions are raised: whether the IRA of ginkgo originally had a syntenic ycf2 copy, as is found in most seed plants, and if the ginkgo did lose the ycf2 copy from the ancestral IRA, to what extent the retained ycf2 copy evolved. To answer these questions, we first compared the boundaries of IRs among two representatives of ferns and five representatives of IR-containing seed plants, including Cycas (representative of cycads), ginkgo, Gnetum (representative of gnetophytes), Amborella (representative of basal angiosperms), and Nicotiana (representative of eudicots) (fig. 2).

F

(A) Comparison of IR boundaries among two representative ferns (Psilotum nudum and Angiopteris evecta), gymnosperms (Cycas taitungensis, Ginkgo biloba, and Gnetum parvifolium), and angiosperms (Amborella trichopoda and Nicotiana tabacum). (B) Hypothetical scenario illustrating IR contraction in ginkgo cpDNA. The ancestral IRs of gymnosperms should have expanded to include a trnH-GUG, and then ycf2 was lost from the ancestral IRA during ginkgo evolution. The evolutionary footprints, two trnH-GUG, and IR contraction are indicated. The tree topology was modified from Wu et al. (2011). *: genes with introns. JLA: junction between LSC and IRA; JLB: junction between LSC and IRB regions.

Because the IRs of leptosporangiate ferns (e.g., Adiantum and Alsophila) independently expanded to encompass rps7, 3′rps12, and ycf2 (Wicke et al. 2011), two eusporangiate ferns (Psilotum and Angiopteris) that retain ancestral cpDNA organizations (Gao et al. 2011) were included to simplify evolutionary inferences. As compared with the IR boundaries of ferns, those of all seed plants, except ginkgo, expanded to include ycf2 sequences (fig. 2). This indicates that duplication of ycf2 is a common trait among the cpDNAs of seed plants. Wu et al. (2007) proposed that the ycf2 of IRB was duplicated from that of IRA. Of note, IRs of both cycads and gnetophytes retain a trnH-GUG. In contrast, this tRNA gene is absent from the IRs of both ferns and angiosperms, which suggests that duplication of trnH-GUG is gymnosperm specific. Because the cpDNA of ginkgo has two respective trnH-GUG sequences near its IR boundaries, each of the ancestral IRs of ginkgo should have expanded to include a trnH-GUG sequence, and subsequently the IRs were contracted by loss of at least the ycf2 sequence from the IRA (fig. 2). As a result, the trnH-GUG that adjoins the current IRA could be considered an evolutionary footprint due to the contraction of the ancestral IRA.

IR Contraction Has No Effect on the Substitution Rate of the Retained ycf2

Perry and Wolfe (2002) discovered that in IR-containing legumes, the synonymous rates (Ks) of IR genes are 2.3-fold lower than those of SC genes, whereas in IR-lacking legumes, the mean Ks of formerly IR-residing genes are 1.3-fold higher than those of the remaining genes. The authors concluded that in IR-lacking cpDNAs, decreased copy number rather than intrinsic properties directly elevates the Ks of genes formerly residing in IRs.

With the conclusion of Perry and Wolfe (2002), one should expect an accelerated Ks in the retained ycf2 of ginkgo cpDNA. Figure 3 shows comparisons of the ycf2 and the rest of the IR genes among seven available IR-containing gymnosperm cpDNAs. The Ks values are largely variable among lineages, with the highest in Ephedra of the gnetophytes. To exclude the lineage effect, the Ks of ycf2 was divided by the mean Ks of the rest of the IR genes in respective lineages (the obtained ratios for Cycas 2.74; Bowenia 3.24; Zamia 2.79; Ginkgo 3.38; Ephedra 3.03; Welwitschia 3.79; Gnetum 3.57). Two-tailed Z-test results revealed no difference between ratios for ginkgo and other gymnosperms (P = 0.29). Therefore, in ginkgo cpDNA, the event of losing an ycf2 copy is likely recent, and the retained copy accumulates few mutations.

F

Comparison of synonymous rates (Ks) between ycf2 and the remaining IR genes. Three genes (ndhB, rps7, and 3′rps12) in cycads and ginkgo were selected to represent the remaining IR genes, and only rps7 and 3′rps12 were sampled from gnetophytes because ndhB was lost from the IRs of gnetophytes.

Although ycf2 is essential for plants (Drescher et al. 2000), retaining two ycf2 copies seems unnecessary because several lineages, such as lower land plants, lycophytes, eusporangiate ferns, conifers, and legumes, have only one ycf2 copy. Of note, natural ginkgo populations show a low level of genetic variance (Shen et al. 2005), which suggests that ginkgo had experienced population bottlenecks in the past. Thus, loss of an ycf2 copy might initially occur in an individual of ginkgo, but the bottleneck effect that ginkgo experienced later homogenized the genomic content of all individuals, with only one ycf2 retained. Further studies are needed to investigate whether retaining two ycf2 copies in a cpDNA is advantageous or not.

Duplications of trnC-GCA Occurred at Least Twice

We detected three adjacent tRNA genes (trnY-AUA, trnC-ACA, and trnSeC-UCA) in the same orientation, as well as trnC-GCA (a syntenic tRNA gene of all land plant cpDNAs), in the region between petN and rpoB of the LSC in ginkgo. We exclude the possibility that the three clustered tRNA genes derived from horizontal transfers because of no DNA-importing system in chloroplasts (Smith 2011). Intriguingly, these three clustered tRNA genes have high sequence similarity with trnC-GCA (fig. 4): the sequence similarities between trnC-GCA and trnY-AUA, trnC-ACA, and trnSeC-UCA are 82.1%, 85.1%, and 80.8%, respectively. The ML tree depicted in figure 4 shows that almost all synonymous tRNA species are clustered with each other or to one another and that trnY-AUA, trnC-ACA, trnSeC-UCA, and trnC-GCA are grouped as a monophyletic clade (bootstrapping value = 70%), in which trnC-ACA and trnSeC-UCA form a subclade (bootstrapping value = 64%). This result suggests that the three clustered tRNA genes are duplicates of trnC-GCA, and they might derive from at least two duplication events. The tandem duplicated trnF-GAA copies found in the cpDNAs of Brassicaceae were characterized by several parallel gains and losses (Koch et al. 2005). However, the duplicated tRNA genes that we reported here may not be specific to ginkgo or inherited from the common ancestor of ginkgophytes because cpDNAs of extinct ginkgo lineages, for example, G. adiantoides and G. yimaensis (Zhou and Zheng 2003), are unavailable. Interestingly, trnSeC-UCA was also annotated in the cpDNA of Adiantum (Wolf et al. 2003), a leptosporangiate fern, but it is not syntenic with that of ginkgo cpDNA. Gao et al. (2009) proposed that in the Adiantum cpDNA, trnR-CCG was substituted by trnSeC-UCA because the former is not essential. In contrast, in ginkgo cpDNA, trnC-GCA coexists rather than is replaced by its duplicates, possibly because trnC-GCA is vital for plant cell development (Legen et al. 2007). In addition, whether trnSeC-UCA of ginkgo cpDNA is functional and what is its evolutionary significance require further scrutiny.

F

Phylogenetic relationships of the three clustered tRNA genes uniquely found in ginkgo cpDNA. (A) Alignment of trnC-GCA, trnY-AUA, trnC-ACA, and trnSeC-UCA sequences. (B) Unrooted ML tree based on all 35 tRNA genes encoding in ginkgo cpDNA. The trnC-GCA and the three clustered tRNA genes are in yellow shadow. Tree branches leading to synonymous tRNA species have the same colors. Values along branches denote bootstrapping values estimated from 1,000 replicates (only values ≥50% are shown).

Conclusions

We elucidated that the shortened IR of ginkgo cpDNA is a consequence of IR contraction, and the contraction mainly resulted from loss of one ycf2 copy from the IRA. The presence of two trnH-GUG, one near the junction of LSC-IRA and the other upstream of ycf2, are considered as footprints of IR contraction. Unexpectedly, the Ks of the retained ycf2 copy is nonaccelerated, which suggests that the loss might be recent in ginkgo evolution. Moreover, we found a unique cluster of three tNRA genes upstream of trnC-GCA in ginkgo cpDNA. The duplicated relationships between the three clustered tRNA genes and trnC-GCA are evident on the basis of their high sequence similarity and phylogenetic evaluation. However, the evolutionary impact of this tRNA gene cluster needs further investigation.

Supplementary Material

Supplementary table 1 is available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).