•The filamentous brown alga Ectocarpus has a complex life cycle, involving alternation between independent and morphologically distinct sporophyte and gametophyte generations. In addition to this basic haploid–diploid life cycle, gametes can germinate parthenogenetically to produce parthenosporophytes. This article addresses the question of how parthenosporophytes, which are derived from a haploid progenitor cell, are able to produce meiospores in unilocular sporangia, a process that normally involves a reductive meiotic division.
•We used flow cytometry, multiphoton imaging, culture studies and a bioinformatics survey of the recently sequenced Ectocarpus genome to describe its life cycle under laboratory conditions and the nuclear DNA changes which accompany key developmental transitions.
•Endoreduplication occurs during the first cell cycle in about one-third of parthenosporophytes. The production of meiospores by these diploid parthenosporophytes involves a meiotic division similar to that observed in zygote-derived sporophytes. By contrast, meiospore production in parthenosporophytes that fail to endoreduplicate occurs via a nonreductive apomeiotic event.
•Our results highlight Ectocarpus’s reproductive and developmental plasticity and are consistent with previous work showing that its life cycle transitions are controlled by genetic mechanisms and are independent of ploidy.
Recent phylogenetic studies have indicated that the ancestral brown macroalgal life cycle involved sexual alternation between two isomorphic multicellular generations, the diploid sporophyte and the haploid gametophyte, linked by meiosis and karyogamy (Cho et al., 2004; Kawai et al., 2007; Phillips et al., 2008). However, variations on this basic life cycle have arisen independently several times. In the kelps, for instance, the gametophyte is microscopic, but still develops independently of the sporophyte. In the Fucoids and Ascoseira, there is no second multicelluar generation, the haploid gametes are produced directly from a syncytial structure that forms within the sporophyte (Jensen, 1974; de Reviers, 2003). Life cycle variations can even be seen within a single order – the Ectocarpales, for example, contains not only families with isomorphic generations (Acinetosporaceae, Ectocarpaceae), but also families with strongly heteromorphic generations in which either the sporophyte (Scytosiphonaceae) or the gametophyte (Chordariaceae, Adenocystaceae) is microscopic (Peters & Ramirez, 2001).
The Ectocarpales’ eponymou alga, the filamentous Ectocarpus siliculosus, also has slightly heteromorphic sporophyte and gametophyte generations (Papenfuss, 1935), although both consist of similarly sized thalli with branched filaments (Supporting Information Fig. S1a,b). In the field, these generations are difficult to tell apart, but they can be distinguished in culture, as sporophytes form relatively compact thalli which attach to the substratum, whereas gametophytes form diffuse thalli that attach less strongly and hence tend to float in the medium or, in some cases of mixed-phase thalli, develop vegetatively on parthenogenetic sporophytes (Kornmann, 1956; Müller, 1963, 1967; Peters et al., 2008).
The sporophyte generation in Ectocarpus can reproduce in two ways – first, through plurilocular (i.e. divided into many compartments) sporangia, in which spores are formed by asexual mitosis and grow to give sporophytes which are clones of their parent (Fig. S1e), and, second, through unilocular (i.e. composed of a single compartment) sporangia, in which a single meiosis, followed by subsequent mitotic divisions, gives rise to at least 50 meiospores (Fig. S1a) (Knight, 1929; Müller, 1975). An additional complication is the presence of heteroblasty, a process in which a small proportion of meiospores develop into sporophytes instead of gametophytes (Müller, 1967).
The gametophyte generation, on the other hand, has only one reproductive structure, the plurilocular gametangium, in which male or female gametes are produced through mitosis (Fig. S1b). Despite their morphological similarity, male and female gametes are physiologically different, with females producing a pheromone that attracts males (Müller, 1967, 1978). Gametes can fuse to form a diploid zygote (Fig. S1c) or, alternatively, can develop parthenogenetically in the absence of fertilization (Fig. S1e) to give rise to parthenosporophytes (Fig. S1f). The ploidy of Ectocarpus parthenosporophytes has attracted particular interest in the past (Müller, 1967), because these organisms are derived from a haploid gamete cell and yet can perform all the functions typical of a diploid, zygote-derived sporophyte, including the production of meiospores in unilocular sporangia (a process that usually involves a reductive meiotic division). Müller (1967) proposed that parthenosporophytes endoreduplicate at some point in their development to produce diploid parthenosporophytes. However, measurements of ploidy were technically difficult at that time and this issue has remained unresolved. Endoreduplication, a modified cell cycle in which nuclear DNA is replicated without concomitant cell division, leads to an increase in cell ploidy and is a key component of growth and development in many sessile organisms, including brown algae (Garbary & Clarke, 2002), red algae (Goff & Coleman, 1990) and flowering plants (Nagl, 1976; Sugimoto-Shirasu & Roberts, 2003).
Accordingly, the current investigation describes the nuclear cell cycle (ploidy) transitions associated with the parthenogenetic cycle in Ectocarpus. We show that the first cell cycle in parthenosporophytes of Ectocarpus may be either mitotic or endoreduplicative. In contrast with the sexual cycle, which involves alternation of meiosis and karyogamy, we present evidence that the asexual cycle involves parthenogenesis followed either by endoreduplication and meiosis or by an apomeiotic event if endoreduplication does not occur. These observations corroborate the idea that life cycle transitions in Ectocarpus are independent of ploidy, but are controlled instead by genetic mechanisms. We therefore finish by providing an overview of the genes which regulate the mitotic and endoreduplicative cell cycle, by classifying the core cell cycle components in the Ectocarpus genome.
Materials and Methods
Algal strains and culture conditions
Ectocarpus siliculosus (Dillwyn) Lyngbye strains (Table S1) were taken from parthenogenetic sporophytes maintained as stock cultures in 8 ml polystyrene Petri dishes at 15°C in white fluorescent light with a photon fluence rate of 10–30 μmol m−2 s−1 and with a 10 h : 14 h light : dark cycle. The culture medium was half-strength, Provasoli-enriched (PES), autoclaved seawater (Starr & Zeikus, 1993). Our experiments concentrated on the Peruvian Ectocarpus male strain which was used for complete genome sequencing (Cock et al., 2010), and which is genetically distant from the European strains used previously to describe the life history of Ectocarpus (Müller, 1963, 1967). The Peruvian strain has the accession number CCAP 1310/4 in the Culture Collection of Algae and Protozoa (CCAP), and Ec 32 in the Roscoff Ectocarpus Culture Collection; the European strains have the accession numbers CCAP 1310/148 and 1310/149.
Comparing Ectocarpus life cycle progression in different strains
Culture and manipulation methods were the same for all strains. To follow sporophyte and parthenosporophyte development from gametes, male and/or female gametophytes were obtained by culturing meiospores released from unilocular sporangia isolated from stock parthenosporophyte cultures. Sexually mature gametophytes were placed in 35 mm Petri dishes and left in the dark overnight at 13°C. The following morning, these were transferred to a strongly lit bench (100 μmol photons m−2 s−1) at room temperature (20°C) to stimulate the synchronous release of gametes. Once released, male gametes were either mixed with female gametes in a suspended drop (Peters et al., 2004) or were allowed to settle alone and to develop as parthenosporophytes. The development and morphology of meiospores, mitospores, gametes and zygotes were then monitored using either an Olympus SZ61 stereomicroscope (http://www.olympus.fr) or a Zeiss Axio Observer inverted microscope equipped with a × 63/1.40 n.a. oil immersion lens (http://www.zeiss.fr). All growth data are presented as mean ± SE.
Gametes were collected and nuclei were isolated by suspension in nuclei buffer [30 mM MgCl, 120 mM trisodium citrate, 120 mM sorbitol, 55 mM 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES) pH 8, 5 mM EDTA supplemented with 0.1% (v/v) Triton X-100 and 5 mM sodium bisulfite; pH 8.0], and their DNA content was measured immediately by flow cytometry. Between 600 and 13, 200 nuclei were analyzed in each sample. Gametes were considered to be haploid and were used as an internal reference for the determination of ploidy (Peters et al., 2004). Nuclei from adult Ectocarpus were isolated by cutting the filaments with a razor blade and adding nuclei buffer. The nucleic acid-specific stain SYBR Green I (http://www.invitrogen.com) was used at a final dilution of 1 : 10 000. Samples were analyzed using a FACSort flow cytometer (http://www.bsbiosciences.com).
Fluorescence imaging of nuclei in single cells
The DNA contents of individual nuclei were quantified by either epifluorescence or two-photon microscopy (Helmchen & Denk, 2005) at various times during early development. All images were analyzed using Image J (http://rsbweb.nih.gov/ij/).
For epifluorescence measurements of nuclear DNA levels, gametophytes and/or sporophytes were loaded for 15 min in vivo with 100 μM Hoechst 33342 (http://www.invitrogen.com). Nuclear DNA was then visualized with the epifluorescence facility on either a Zeiss Axio Observer or an Olympus BX60 microscope, using 4′,6-diamidino-2-phenylindole (DAPI) filter sets. The nuclear area was used as a proxy for DNA content (Lee et al., 1995).
For two-photon microscopy, cells were loaded for 15 min in vivo with 100 μM Hoechst 33342 (http://www.invitrogen.com) and then imaged with a Zeiss LSM 510 confocal/two-photon microscope (http://www.zeiss.co.uk) whilst being superfused with a gravity perfusion system. Loaded cells were imaged using a C-Apochromat × 63/1.2 n.a. water immersion objective. In all cases, 512 × 512 pixel images were acquired with a pixel dwell time of 2.56 μs and, unless stated, eight-line summation. Hoechst 33342 was excited at 765 nm and emitted light bandpass filtered between 435 and 485 nm. Nuclear volume – as calculated from the projections of up to 10 different confocal nuclear sections – was used as a proxy for DNA content (Bothwell et al., 2008). Mean filament DNA content was calculated from the nuclei in the one- to three-filament cells nearest the base of each sporangium, looking at 8–12 separate filaments for each life cycle stage; the corresponding mean sporangial daughter cell DNA content was calculated from 15–30 sporangial nuclei. Significance testing was carried out using one-way ANOVA in PASW Statistics (http://www.spss.com).
Core cell cycle genes in the Ectocarpus genome were identified by PFAM, IPR and keyword searches of the gene models predicted by Eugène (Foissac et al., 2008), and checked using BLAST searches against reference sequences from humans, the flowering plant Arabidopsis (Vandepoele et al., 2002), the green algae Chlamydomonas reinhardtii (Bisova et al., 2005) and Ostreococcus tauri (Robbens et al., 2005), and the diatom Phaeodactylum tricornutum (Huysman et al., 2010). The resulting candidate sequences were manually aligned against reference datasets using Jalview 2.4 (Clamp et al., 2004), and obvious errors were rectified if corrections were supported by the Ectocarpus expressed sequence tag (EST) or Tiling libraries (Cock et al., 2010). Their evolutionary histories were then inferred from phylogenetic analyses in MEGA4 (Tamura et al., 2007) using either neighbor-joining (Saitou & Nei, 1987) or maximum parsimony (Eck & Dayhoff, 1966), both with bootstrap support (Felsenstein, 1985), and with all positions containing gaps having been eliminated from the dataset.
Geographically distinct Ectocarpus strains follow similar life cycles under laboratory conditions
All Ectocarpus strains used (Table S1) shared essentially the same life cycle and developmental forms, taking c. 3 months to complete and including heteroblasty and parthenogenesis. Parthenosporophytes take c. 48 h to germinate – about twice as long as sexually generated, heterozygous, sporophytes (Le Bail et al., 2008) – but show no difference in the timing of maturation, from a horizontal to an upright form, or plurilocular sporangia production, both of which happen c. 15 d after prostrate thalli are transferred to high light conditions. Heterozygous sporophytes, however, produce unilocular sporangia significantly more quickly than do parthenosporophytes (Fig. S2).
There were two differences between the life cycles of Peruvian and European strains. First, contrary to previous observations on strains from Naples and Split (Müller, 1963), temperature reduction (from 20°C to 14°C) did not stimulate the production of unilocular sporangia in the Peruvian Ec 32 strain, which was able to produce unilocular sporangia at temperatures of up to 20°C. Second, sporophyte maturation is stimulated in the Peruvian strains by high light intensity. Modification of the light regime from 10–30 μmol photons m−2 s−1 to 60–80 μmol photons m−2 s−1 resulted in 19 ± 9% of the individuals producing upright filaments, compared with 4 ± 4% in the control sample that remained at 10–30 μmol photons m−2 s−1 continuous light (P <0.001, n =197).
Parthenogenetically derived sporophyte populations contain C, 2C and 4C nuclei
Flow cytometry was used to analyze ploidy in adult Ectocarpus thalli from several different strains: Ec 494 (male), Ec 560 (female), Ec 32 (male) and Ec 568 (female). For a typical DNA histogram, one peak will represent cells in the G1 phase of the cell cycle and a second, with twice the channel value, will represent post-DNA replication cells in the G2/M phase. Sexually mature gametophytes followed this pattern, consisting primarily of 1C nuclei, with a small peak of 2C nuclei (Fig. 1a). Gametes produced by mitosis from these gametophytes contained 1C nuclei only (Fig. 1b), indicating, as expected, that these cells are arrested in G1 and are not dividing.
Twenty-day-old parthenosporophyte samples (that is, collections of many clonal, but separate, parthenosporophyte filaments) that had developed from these 1C gametes contained not only 1C nuclei, but also many more 2C nuclei than were seen in gametophytes (Fig. 1c). Moreover, the relative proportions of 1C and 2C nuclei varied between different parthenosporophyte samples, and we therefore observed two flow cytometry distribution patterns in the 13 collections of parthenosporophyte filaments analyzed. Nine of these 13 samples contained a large C peak with a smaller, but substantial 2C peak (Fig. 1c); four of the 13 samples, however, contained more 2C than C nuclei, and all four samples also exhibited a 4C peak (Fig. 1d). This is strong evidence that these samples, each a collection of several individual filaments, contained at least some parthenosporophytes with cells that had 2C G1 DNA contents and in which, therefore, endoreduplication had occurred. For comparison, we also analyzed populations of 20-d-old diploid sporophytes derived by gamete fusion and zygote development (strains Ec 339 and Ec 569). Most of the nuclei from this population were 2C, but a small peak of 4C nuclei was also detected (Fig. 1e). These peaks in diploid sporophytes presumably correspond to cells in G1 and G2/M, respectively.
There are two possible explanations for the co-existence of 1C and 2C G1 nuclei in parthenosporophyte populations (Fig. 1d): either individual Ectocarpus parthenosporophytes are mosaics of 1C and 2C nuclei, or some parthenosporophytes are wholly haploid and others wholly diploid. To distinguish between these two possibilities, we stained nuclei in > 15-d-old sporophytes and parthenosporophytes, with gametophytes as an additional reference (Fig. 1f–i). Microscopic analysis of the nuclei in multiple individuals provided no evidence for mosaic ploidy: all of the cells of an individual thallus had the same DNA content (Fig. 1g,h). We therefore interpret our flow cytometric data to indicate that parthenosporophyte populations include not only wholly haploid individual parthenosporophyte filaments, with cells in G1 producing the peak at 1C, but also a proportion of wholly diploid individuals, whose cells in G1 produce large 2C peaks and whose cells in G2 produce smaller 4C peaks (Fig. 1d).
Parthenosporophytes can generate haploid meiospores from 1C basal cells
Given the existence of wholly haploid parthenosporophytes, we were interested in whether meiospore production could occur in their unilocular sporangia, because the production of meiospores in the unilocular sporangia of diploid, sexual, sporophytes involves meiotic cell division, which is impossible in a haploid individual. How, then, do haploid parthenosporophytes produce meiospores? One possibility is that endoreduplication occurs just before meiosis in the unilocular sporangia borne by haploid thalli, although this would be in contrast with previous observations in the closely related pheophyte, Alaria esculenta, in which endoreduplication occurs after, rather than before, sporangial production (Garbary & Clarke, 2002). Alternatively, haploid unilocular sporangia mother cells may produce daughter cells by an apomeiotic mechanism.
To resolve this question, multinucleate plurilocular and unilocular sporangia from individual parthenosporophytes were examined using two-photon microscopy. Diploid sporophytes, whether derived sexually from zygotes (Fig. 2a,b) or asexually through parthenogenesis (Fig. 2c,d), produce plurilocular sporangia (Fig. 2a,c) whose daughter cells have the same ploidy as the filament cells at the base of the sporangia from which the sporangia are produced (Fig. 2g; n =8–12 filaments). However, the daughter cells of unilocular sporangia (Fig. 2b,d) in diploid sporophytes and parthenosporophytes have only half the DNA content of their basal filament cells (Fig. 2g; n =8–12 filaments). By contrast, in haploid parthenosporophytes, both plurilocular sporangia (Fig. 2e) and unilocular sporangia (Fig. 2f) have daughter cells with the same ploidy as their basal filament cells (Fig. 2g; n =8–12 filaments). These results indicate that endoreduplication is not essential for the production of unilocular sporangia, but can restore normal meiosis in parthenosporophytes. They also, in passing, confirm that both haploid and diploid parthenosporophytes are capable of producing mitospores in plurilocular sporangia (Müller, 1967), and suggest that this occurs through a mitotic mechanism in both cases.
The ploidy of adult parthenosporophytes is determined during the first cell division
To investigate when nuclear ploidy increases following parthenogenesis, we used the fluorescent dye Hoechst 33342 (Bothwell et al., 2008) to follow ploidy changes during the first 10 d of Ectocarpus sexual sporophyte and parthenosporophyte development (Fig. 3). Unfertilized gametes had fluorescent pronuclear cross-sectional areas of 4.0 ± 1.0 μm2 (Fig. 3a; n =17 gametes). When male and female gametes were mixed, fertilization occurred immediately to give a zygote which contained two pronuclei (Fig. 3b) and which secreted a birefringent cell wall within 6 h of gamete fusion. Pronuclear fusion then followed between 12 and 48 h after fertilization, giving rise to zygotic nuclei with a significantly larger (P <0.001) cross-sectional area of 9.4 ± 1.4 μm2 (Fig. 3c; n =17 zygotes).
Next, we verified that individual Ectocarpus filaments may develop asexually and still be wholly diploid by staining nuclei of either male or female gametes with Hoechst 3342 and allowing them to develop parthenogenetically (Fig. 3d–g). Parthenogenetic activation was confirmed by the appearance of a birefringent cell wall, usually within 12 h of gamete release, followed by germination c. 48 h after gamete release. By 7 d after germination, two subpopulations of parthenosporophytes could be distinguished (Fig. 3e): one subpopulation of parthenogenetic germlings had nuclear cross-sectional areas of 5.8 ± 0.9 μm2 (n =30 germlings), which were not significantly different from those of unfertilized gametes, but the second subpopulation had significantly higher (P <0.001) nuclear cross-sectional areas of c. 10.5 ± 1.4 μm2 (n =12 germlings), or twice the DNA content of unfertilized gametes (Fig. 3e).
Interestingly, in those parthenosporophytes which had undergone cell division to the two-cell stage (Fig. 3f; n =13 germlings) or which were allowed to germinate for 14 d to the 5–10-cell stage (Fig. 3g; n =16 germlings), all the nuclei in a single filament had similar DNA contents (Fig. 3f,g). This agrees with our earlier finding that all the cells in any given individual filament are of the same ploidy (Fig. 1e–h). Although one reason for these findings might be that all the cells in any given filament pass through G2/M phase in synchrony, we dismiss this explanation because we, and other researchers (Müller, 1967; Le Bail et al., 2008), have only ever seen single cells (typically tip cells; Fig. 3h) undergoing cell division in growing filaments at any one time. Instead, therefore, we infer that any increase in G1 DNA content occurs before the first cell division.
Although there is increasing evidence that endoreduplication of this sort can be driven by an increase in cell size, and although the level of ploidy is often closely correlated with cell size in land plants (John & Qi, 2008), red algae (Goff & Coleman, 1990; Garbary & McDonald, 1998) and in at least one species of brown macroalga, A. esculenta (Garbary & Clarke, 2002), we found no correlation between cell area and ploidy level (Fig. S2b) in Ectocarpus parthenosporophytes, nor did the inhibition of DNA replication by the addition of aphidicolin, a DNA polymerase inhibitor, significantly affect the cell size over the first 48 h of parthenosporophyte development (Fig. S2c).
The Ectocarpus core cell cycle machinery displays evolutionary novelty
To identify genes that represent potential molecular regulators of life cycle ploidy, the Ectocarpus genome sequence was searched for core cell cycle genes (Vandepoele et al., 2002; Bisova et al., 2005; Robbens et al., 2005), namely cyclins and cyclin-dependent kinases (CDKs), together with those regulators of mitotic cyclin–CDK complexes which allow plant cells to switch between mitotic and endoreduplicative cell cycles (Inzé & De Veylder, 2006).
The Ectocarpus genome contains six classes of CDK, five of which are orthologous to the green lineage CDK classes A, C, D, H and I/F (Fig. 4, Table 1), and one of which, here called CDKB-like, is currently known only in the brown algae (Huysman et al., 2010). Ectocarpus has one CDKA1, which was classified by both the presence of the canonical PSTAIRE cyclin-binding domain common to all eukaryotic CDKAs (Robbens et al., 2005) and its overall phylogenetic affinity with CDKAs from other species (Fig. 4). Similar characteristic cyclin-binding motifs (Bisova et al., 2005; Robbens et al., 2005), in conjunction with overall sequence homologies and phylogenetic affinities, allowed the classification of the remaining Ectocarpus CDKs, and showed that they possess hitherto unrecorded and diverse cyclin-binding motifs (Table S2, cf. Huysman et al., 2010). Of particular interest are the presence of the brown algal-specific CDKA2 (Huysman et al., 2010) and the hitherto unreported brown algal CDKH sequence, which is more closely related to a Chlamydomonas sequence than to any Arabidopsis sequence (Bisova et al., 2005), and which hints at the differential patterns of CDK loss which have occurred during the evolution of the green lineage (Table S3).
Table 1. The 41 Ectocarpus core cell cycle genes characterized in this study
Gene locus ID
Evidence for sequence
The locus ID is the identification number of the gene in the Ectocarpus genome sequence (genome accession numbers CABU01000001–CABU01013533, FN647682–FN629242, FN649726–FN649760). EST, expressed sequence tag.
Gaps; genome incomplete
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Gaps; genome incomplete
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Full EST, Tiling
Partial EST, Tiling
Partial EST, Tiling
Similar sequence homology comparisons show that Ectocarpus cyclins fall into seven classes, orthologous to the well-characterized eukaryotic cyclins A, B and D, which are known to regulate cell cycle activity (Fig. 5, Table 1), and the more poorly characterized cyclins F, H, L and T (Fig. 5, Table 1). Again, the distribution of cyclin families across eukaryotic taxa can help to shed light on their evolution and, in this context (Bisova et al., 2005), we note the presence of H- and F-type cyclins, but the lack of C-type cyclins (Fig. 5, Tables 1, S3), and the lack of any expansion of the diatom-specific cyclins which have been discovered recently in Phaeodactylum (Huysman et al., 2010).
CDK–cyclin complexes may be regulated in three ways. First, by phosphorylation or dephosphorylation of either the CDK or cyclin partners. During mitosis, these processes are mediated by the Wee1 kinase, which is a negative regulator of CDK/cyclin in both animals and plants, and by the Ccd25 phosphatase, which has been found so far only in animals. Analysis of the Ectocarpus genome shows that, like green plants, it contains Wee1 but no obvious orthologues of Cdc25 (Tables 1, S3).
The second way to inhibit CDK–cyclin complexes is through directed proteolysis by the anaphase-promoting complex, an E3 ubiquitin ligase which targets CDKA for degradation and which is also present in Ectocarpus, together with its activator CDH1 (Tables 1, S3). The third and final way to regulate CDK–cyclin activity is through the binding of regulatory proteins, such as the RBR/E2F/DP/DEL pathway, the cyclin-dependent kinase subunits (CKS) and the cyclin-dependent kinase inhibitors (CKI). In flowering plants, the major class of CKIs are the Kip-related proteins, or KRPs, and, interestingly, although the Ectocarpus genome contains copies of CKS, RBR, DEL, E2F and DP proteins, no KRPs were found (Tables 1, S3), although it should be noted that these proteins are poorly conserved between organisms and may be present in a form that we are currently unable to predict.
Cell biological plasticity underlies life cycle plasticity in Ectocarpus
Ectocarpus, like many brown macroalgae, has a complicated life cycle in which nuclear ploidy and developmental phases may be uncoupled. However, the development of any organism is intimately interconnected with its cell cycle (Meijer & Murray, 2001). Although parthenogenesis, endoreduplication and heteroblasty have all been described previously in Ectocarpus (Müller, 1967), the ways in which these cell biological mechanisms are coordinated to grant Ectocarpus its extreme developmental plasticity are still a matter for debate.
In this study, we have shown that endoreduplicated, diploid, parthenosporophytes can produce haploid meiospores in unilocular sporangia by a reductive and, presumably, meiotic division (Fig. 2d). However, nonendoreduplicated, haploid, parthenosporophytes may also produce haploid spores in unilocular sporangia through a nonreductive, presumably apomeiotic, cell division (Fig. 2f). Hence, endoreduplication can restore meiosis, but is not necessary for meiospore production, suggesting that haploid parthenosporophytes are able to produce spores through an apomeiotic process. Although apomeiosis has been reported previously in Ectocarpus (Müller, 1967), as well as in other brown algae (Kuhlenkamp & Müller, 1985; Müller, 1986), it has only been described during diploid stages of the life cycle and in association with a loss of sexuality, in which diploid individuals failed to produce haploid daughter cells in unilocular sporangia, even though they entered meiotic prophase. Our results, by contrast, suggest that apomeiosis can occur in haploid individuals and can give rise to daughter cells (apomeiospores) that are capable of growing into viable haploid gametophytes.
Individual Ectocarpus parthenosporophytes may be haploid or diploid, but not mosaics
Endoreduplication is a common event during development in a wide range of eukaryotes, and endopolyploidy, presumably following endoreduplication, has been reported in a number of brown macroalgal species (Deshmukhe & Tatewaki, 1993; Le Gall et al., 1996; Garbary & Clarke, 2002). Earlier work in Ectocarpus, using chromosome counts to estimate ploidy, suggested that parthenosporophyte populations contained both C and 2C G1 nuclei, with approximately two-thirds of any given population being C and one-third being 2C (Müller, 1967). Our results (Figs 1, 3) suggest that this population structure is established during early development and that the ploidy of adult parthenosporophytes depends on whether or not endoreduplication occurs during their first cell division (Fig. 3). Thus, all the nuclei in any given parthenosporophyte have the same ploidy (Fig. 1f,g), and we found no evidence for the mosaic development reported in Laminaria saccharina and Alaria esculenta (Garbary & Clarke, 2002). This may be related to the different extent of cellular differentiation in different macroalgal species; in Ectocarpus, there is only a limited range of cell types and all cells are roughly the same size. In L. saccharina and A. esculenta, there are a wider range of cell types and sizes in any given plant and, although the association between cell size and cell ploidy is not absolute, it is likely that mosaic development may reflect this wider range of cell types.
Ectocarpus is also unusual in undergoing endoreduplication so early in development. In most eukaryotes studied to date, endoreduplication occurs during postembryonic, rather than embryonic, development and is associated with specialization and differentiation (Gendreau et al., 1997; Sugimoto-Shirasu & Roberts, 2003; John & Qi, 2008). However, in a broad variety of arthropods, mammals and plants, endoreduplication is associated with species which, like Ectocarpus, have a small genome and a short life cycle (Nagl, 1976; Edgar & Orr-Weaver, 2001; Barow & Meister, 2003). This is presumably because diploid development provides gene redundancy, buffers epigenetic defects, allows double strand breaks to be repaired by homologous and nonhomologous recombination (Aylon & Kupiec, 2004), and may also allow an optimal balance between organellar and nuclear DNA levels, enhancing metabolic capacity (Sugimoto-Shirasu & Roberts, 2003). Although the adaptive benefits of endoreduplication in Ectocarpus, if any, remain unclear, such protective effects may be important as Ectocarpus lives in the intertidal zone and is exposed to regular and multiple biotic and abiotic stresses.
The core cell cycle machinery of Ectocarpus describes candidate networks which may regulate the onset of endoreduplication
There is growing evidence that an endoreduplicative endocycle may be triggered in a broad range of eukaryotic organisms, including animals, green plants and fungi, by inhibition of mitotic CDK–cyclin complexes following DNA replication (Inzé & De Veylder, 2006; John & Qi, 2008). Understanding such a switch, in which the whole cell cycle is deregulated, rather than just a single gene, requires a knowledge of the relevant network components and structure and, in this context, our annotation of the Ectocarpus cell cycle machinery provides a first step towards this goal, revealing several proteins that are thought to cooperate in the control of endoreduplication in other organisms (Inzé & De Veylder, 2006).
More broadly, the availability of this dataset of Ectocarpus cell cycle regulators lays the foundation for future work aimed at determining the molecular basis of the remarkably plastic ploidy states observed during different facets of the Ectocarpus life cycle. The relevant cell cycle machinery in Ectocarpus is similar to that seen in the closely related diatoms (Table S3), although there are novelties in both the sequences (Fig. 4, Table S2) and distribution of various cell cycle families, when compared with the cell cycle complement of Arabidopsis, Chlamydomonas, Ostreococcus and Phaeodactylum (Figs 4, 5; Table S3). The most notable disjunctions between Ectocarpus and Phaeodactylum, which are both heterokonts, is the expansion of the diatom-specific cyclin family and the loss of M- and T-type cyclins in Phaeodactylum (Huysman et al., 2010), but not in Ectocarpus, together with the presence in Ectocarpus of I-type CDKs, which have previously only been described in green plants.
Further study of these patterns is beyond the scope of our current investigation, but will be expected to shed more light on the evolution of eukaryotic development and multicellularity, especially when multicellular organisms, such as Ectocarpus and Arabidopsis, may be compared with their unicellular relatives (Cock et al., 2010).
We thank Génoscope for access to the assembled Ectocarpus genome, the Bioinformatics group at the Universiteit Gent for automatic gene predictions, Hervé Moreau of the Oceanological Observatory in Banyuls for help in annotating cell cycle genes, and Laurence Dartevelle for assistance with the cultures of Ectocarpus. J.H.B. was funded by a Leverhulme Trust Early Career Fellowship. This work was supported by the Centre National de la Recherche Scientifique and the University Pierre and Marie Curie. Finally, the manuscript was greatly improved by the critical reading of three anonymous referees, to whom we are extremely grateful.