Introduction

Ginkgo is regarded as the oldest living tree species on the Earth dating back to at least the early Jurassic period, more than 170 million years ago (Zhou and Zheng 2003 ). Ginkgo species were once common in Asia, North America, and Europe, but vanished from most regions during the Ice Age, leaving a single surviving species, G. biloba, in China. From there, Ginkgo spread to Japan and Korea. At the end of the seventeenth century Ginkgo seeds from Japanese trees were introduced to Europe and from there it spread to different countries in the Western world (Strømgaard and Nakanishi 2004). Most, if not all, modern Ginkgo trees in Europe and North America are descendants of just a few Japanese trees.

Owing to the presence of unique terpenoids, ginkgolides and bilobalide, and flavonoids, Ginkgo extracts are used for treatment of, e.g. vascular disorders and age-related dementias (Braquet 1997; Strømgaard and Nakanishi 2004). In the course of research for accumulating such compounds in in vitro cultures (Laurain et al. 1993, 1997; Balz et al. 1999) we have identified the green eukaryotic algae present inside Ginkgo cells. These algae had been taken into culture and were identified by small subunit ribosomal RNA (ss rRNA) analysis to belong to the green algal genus Coccomyxa (Trémouillaux-Guiller et al. 2002). Coccomyxa is a terrestrial coccal alga known as photobiont in some lichens (Friedl and Büdel 1996; Lohtander et al. 2003), but has also been described as intracellular parasite of bivalves (Gray et al. 1999). Within Ginkgo host cells, the alga resides in an immature “precursor” state: nucleus and mitochondria are not observed, and diffuse electron-dense areas mark thylakoid-like membranous structures of a nonfunctional chloroplast (Fig. 1a). Mature algae with eukaryotic traits and a normal functional chloroplast (Fig. 1b) were observed almost exclusively in host cells undergoing necrosis. Such algae could be easily cultured in inorganic, ammonium-containing medium allowing identification and characterization of the endosymbionts (Trémouillaux-Guiller et al. 2002).

Fig. 1
figure 1

Electron micrographs of two endosymbiotic algae freshly escaped from in vitro-cultured Ginkgo biloba cells. a Immature precursor alga showing within its cytoplasm large lipid droplets (black arrows) and electron-dense material (white arrows), which subsequently evolved into thylakoids. b Mature alga with a nucleus (black arrowhead), mitochondria (white arrows), a cup-shaped chloroplast (white arrowhead), and lipid droplets (black arrows). Scale bar = 1.0 μm

Full size image

As existence of a green algal endosymbiont in G. biloba was never reported or observed in planta, we used endosymbiont-specific sequences of ribosomal RNA genes to address the following questions: does the endosymbiont naturally occur in Ginkgo trees, in which tissues can it be identified, and is the symbiont ubiquitously distributed in Ginkgo trees worldwide?

Materials and methods

Sampling

Juvenile or fertilized G. biloba L. ovules were harvested from female trees in Botanical or private gardens, as well as from field or alley trees from different geographic locations (i.e., China, France, Italy, Japan, and the United States) and surface-sterilized before excising zygotic embryos of 6–22 mm in length from the gametophyte tissues. Microsporophylls were taken from male trees and sterilized, as were all plant samples, by successive treatment with 70% (v/v) ethanol for 1 min, 10 or 20% (v/v) Domestos commercial bleach solution [0.5–2% (v/v) NaOCl; Lever Bros., Warrington, UK] for 5 min under vigorous shaking, and with sterilized distilled water, before being quickly frozen in liquid nitrogen. Details on the geographic location, time of harvest and type of tissue of all G. biloba specimens studied are shown in Table 1.

Table 1 Details on the geographic location, time of harvest, and type of tissue isolated from G. biloba specimens
Full size table

DNA isolation

Approximately 300–500 mg of either sterilized frozen G. biloba tissues or of sedimented cell pellet of the endosymbiont (isolate CMS-93) cultured in MS medium (Murashige and Skoog 1962) were ground to fine powder. DNA was extracted with the DNeasy Plant minikit (Qiagen). In total, 18, 17, and 11 independent DNA extractions were performed from embryos, and from reproductive and vegetative tissues, respectively.

Gene amplification and sequencing

Two sets of primers were designed to specifically amplify endosymbiont DNA sequences from Ginkgo tissues by PCR. The first set (5′-TCACGCCTGGGCTCCCCAG-3′ and 5′-GAACCGCCGGAAGCCGCCA-3′) with an expected amplification product size of 352 bp was targeted against regions of the ss rRNA gene with sufficient sequence heterogeneity among symbiont (GenBank accession number AJ302939) and host (D16448). The second primer set with an expected amplification product size of 387 bp was targeted against a region close to the end of the ss rRNA gene (5′-GTTAAACCCTCCCACCTAG-3′) and to the terminal region of ITS1 (5′-TGGTTTGAAGGCAGCGCTT-3′) of the symbiont (AJ880282). Amounts of DNA template from Ginkgo tissues were up to 1.5 μg compared to 1–10 ng from the cultured alga used as control. Among the tested DNA polymerases, either the Invitrogen Taq polymerase (Invitrogen Life Technology) or the TaKaRa Ex Taq polymerase (TaKaRa Biomedicals, Saint-Germain-en-Laye, France) worked best for the PCR reactions. Up to 20 PCR reactions were performed for each DNA isolated from embryos and reproductive and vegetative tissues to check the presence of the respective 352 and 387 bp PCR bands. Amplification products were sequenced either by Genome Express (Grenoble, France) with a 3730 XLTM Sequencer (Applied Biosystems) or by ourselves with an ABI Prism 310 Genetic Analyzer (Applied Biosystems).

Results

PCR amplification and sequence comparisons

We designed two sets of primers specifically to amplify the endosymbiont DNA from Ginkgo tissues by PCR. The first set of primers was targeted to regions of the endosymbiont ss rRNA genes with sufficient sequence heterogeneity compared to the corresponding Ginkgo sequence (Chaw et al. 1993). With these primers, different tissues collected from several Ginkgo trees from Asia, Europe, and North America were assayed for presence of the endosymbiont. In order to obtain reasonable PCR products, it was necessary to enhance DNA quantities from G. biloba tissue extracts up to 1,000-fold compared to the endosymbiont control DNA. Moreover, the choice of particular Taq polymerase was critical for obtaining a positive result. Using optimized PCR conditions, DNA samples taken from reproductive tissues such as microsporophylls, pollen grains, ovules, and prothallus, as well as from complete zygotic embryos yielded the expected 352 bp band for all Ginkgo specimens studied (Fig. 2a, b). Sequences obtained from the amplification products were completely identical among each other and to those from the cultured symbionts (Table 1). For vegetative tissues, the presence of the endosymbiont could be confirmed so far only for cotyledons and young leaves (data not shown). However, the negative PCR reactions for older leaves might be caused by a dilution effect of the fast growing leaves and thus by very low endosymbiont amounts in these tissues. Failure to obtain PCR products from organs that are utmost exposed to their environment such as leaves, on the other hand, serves as a reliable negative control to exclude exogenous contamination as the source for the presence of the alga.

Fig. 2
figure 2

Molecular detection of endosymbiotic algae in various Ginkgo biloba tissues collected from different geographic locations. a, b A 352 bp fragment of the ss rRNA gene sequence amplified with symbiont-specific PCR primers from total DNA extracts from Italian embryo (IE), French microsporophyll (FM), French prothallus (FP), and Chinese embryo (CE), and from French prothallus (Pr), ovule (Ov), microsporophyll (Mi), and pollen grains (Po). c, d A 387 bp fragment of the symbiont ITS1 sequence was amplified from embryos collected in China (CE), Italy (IE), France (FE), and the United States (AE), and from prothallus (Pr), ovule (Ov), pollen grains (Po), and microsporophyll (Mi) from French trees. DNA of cultured symbionts (strain CMS-93) as control c S size marker (100 bp ladder)

Full size image

A common ancestry of endosymbionts in modern Ginkgo trees

As outlined above, most Ginkgo trees have a separate evolutionary history of only about 300 years. If the ancestor of these trees already contained the endosymbiont, we would not expect to find differences within the conserved ss rRNA gene even in geographically separate specimens. The ITS region of the ribosomal cistron is much more variable and widely used for phylogenetic inference at low taxonomic levels, including biogeographic studies (Álvarez and Wendel 2003). A second primer set was therefore directed against the ITS1 region of the alga. This region is similar to only about 70% between the endosymbiont and the lichen-forming Coccomyxa glaronensis (Lohtander et al. 2003), while their ss rRNA genes share a sequence similarity of 99.5%. Using this primer set, we obtained amplification products of the expected size 387 bp for several tissues of Ginkgo specimens from different geographic locations (Fig. 2c, d). All ITS1 sequences amplified from Ginkgo tissues again completely matched the sequence from the cultured symbiont (Table 1). This could imply a recent origin of the event that led to the establishment of this intracellular symbiosis. Alternatively, and more plausibly, the observed genetic homogeneity of endosymbiont populations might be caused by the short evolutionary history most Ginkgo trees share. According to this idea, the Japanese ancestor(s) of modern Ginkgo trees, from which seeds were first brought to Europe and their offspring then distributed all over the world, did already contain the endosymbiont. Even the variable ITS1 DNA did not accumulate mutations within the following 300 years of separate evolution till today. Investigation of wild-growing Ginkgo that might still exist in China should resolve this question.

Discussion

The kind of symbiosis described here between a cryptic green eukaryotic alga and a higher plant is unprecedented. Intracellular symbioses usually have co-evolved to the benefit of at least one partner, resulting in host/symbiont relationships defined as commensalism when neutral to the host, as parasitism when injurious to the host, and as mutualism when beneficial for both partners (Parniske 2000). The type of symbiosis in the Ginkgo/Coccomyxa association is unknown but likely to be beneficial for at least the host. Unicellular algae are widely found as intracellular symbionts in diverse groups of organisms such as invertebrates and protists (Huss 1999). Commonly in this kind of symbiosis, the algae complement their host’s diet by providing photosynthetically fixed energy-rich carbon compounds, eventually becoming indispensable for some hosts, such as corals (Lewis and Coffroth 2004). Such a role for the Ginkgo endosymbiont can be ruled out for two reasons: Ginkgo itself is photoautotrophic, and the in situ precursor forms of the endosymbiont are likely to be photosynthetically inactive (Fig. 1a). During the development from precursors into mature algae, transition forms were observed by transmission electron microscopy (TEM) representing different maturation degrees (Trémouillaux-Guiller et al. 2002). Although mature algae showed the typical eukaryotic organelles (Fig. 1b), neither nucleus nor mitochondria or plastids could be identified in the immature precursors (Fig. 1a). Instead, diffuse electron-dense areas were visible, which can be interpreted as thylakoid-like membranous structures of an obviously nonfunctional chloroplast. Upon maturation they evolved into thylakoid membranes, and chloroplasts became visible. An interesting phenomenon of membrane reorganization could be followed by TEM, although biogenesis of plastids is classically admitted to follow the division of pre-existing organelles. Moreover, TEM pictures showed numerous storage bodies, probably of lipid nature, in the cytoplasm of endosymbionts as well as in Ginkgo host cells (Trémouillaux-Guiller et al. 2002). Similar storage bodies were observed in Coccomyxa phycobionts of ascolichens, while they were absent in cultured isolates (Peveling and Galun 1976). This suggests a possible involvement of the endosymbiont in metabolic pathways of its host. Abundant lipid biosynthesis is indicated by mitochondria with tubular cristae present inside mature Ginkgo symbionts, implying a lipid nature of the storage bodies (Prince 1999).

How could an alga become an inherent part of a tree like Ginkgo? Our hypothesis is linked to the unique fertilization process of Ginkgo. As a lichen alga, Coccomyxa without doubt sometimes lands upon a pollination droplet, which covers the micropyle of young Ginkgo ovules. The alga would be retracted together with the pollen inside the pollen chamber, where it may survive the 4–5 months until fertilization (Hori and Miyamura 1997; Soma 1997). Considering the size of Coccomyxa, which is about 20 times smaller than a Ginkgo spermatozoid, it seems possible that an algal cell fortuitously could have entered the egg cell along with sperm. Once in the egg cell, it had to escape digestion like any other intracellular symbiont, adjust reproduction rate, and become an inherent part of the host. Undoubtedly, such a process leading to a stable endosymbiosis is a rare event, but recurrently exemplified by the diverse endosymbiotic associations found in nature. First detected in in vitro cell cultures of Ginkgo tissues, we now could demonstrate its existence in planta in geographically dispersed Ginkgo trees.