Life-cycle-generation-specific developmental processes are modified in the immediate upright mutant of the brown alga Ectocarpus siliculosus

The brown algae evolved multicellularity independently of animals and higher plants, and they include several lineages that rival higher plants in their complexity (de Reviers,2003). Brown algae develop from cells that are released into the surrounding seawater (de Reviers,2002), and the early stages of development are therefore easily observable. The growth of the initial cell can be markedly different in different species, involving either unipolar or bipolar germination (emergence of either one or two germ tubes, respectively). Moreover, bipolar germination can result in different growth patterns. When germination produces a symmetric filament, this leads to the development of a prostrate, basal structure before the erect thallus is formed (termed mediate differentiation or heterotrichy,see Table 1). By contrast, if an asymmetric structure is formed following germination, this leads to the immediate development of an erect thallus without the formation of a prostrate, basal structure (immediate differentiation)(Pedersen, 1984; Fletcher, 1987).

Asymmetric cell divisions are defined in terms of developmental fate, so that a cell division may be asymmetric even if there is no morphological difference between the daughter cells at the time of division, provided that each cell adopts a different cell fate(Horvitz and Herskowitz, 1992; Scheres and Benfey, 1999). Similarly, a symmetric division is defined as one that generates daughter cells that are destined to acquire the same developmental fate(Morrison and Kimble, 2006). Zygotic/initial cell divisions frequently elaborate polarity that underlies a major body axis [Berleth and Chatfield,2002(http://www.aspb.org/publications/arabidopsis/); Schneider and Bowerman, 2003; Huynh and St Johnston, 2004]. Within the brown algae, efforts to dissect the processes of polarisation and the first cell division have concentrated on the Fucales, where it has been shown that polarity can be established in response to a range of external vectors with unidirectional light being the strongest stimulus(Quatrano, 1997).

The filamentous brown alga Ectocarpus siliculosus (Dillwyn)Lyngbye has been proposed as a general model for the brown algae(Peters et al., 2004a) and its genome has been sequenced at Genoscope(http://www.genoscope.cns.fr/spip/Ectocarpus-siliculosus,740.html). The life cycle of E. siliculosus involves an alternation between two macroscopic generations that differ morphologically (sporophytes produce few laterals and develop from a branched prostrate base, whereas gametophytes are more richly branched and devoid of a prostrate base)(Kornmann, 1956; Müller, 1964). In this study, we show that, unlike the sporophyte, which is formed by mediate differentiation following bipolar germination and symmetric division of the initial cell (Peters et al.,2004b), the E. siliculosus gametophyte exhibits an asymmetric initial cell division and immediate differentiation of an erect thallus. Therefore, the alternation of generations in E. siliculosusinvolves an alternation between two fundamentally different patterns of initial cell division: symmetric and asymmetric.

A spontaneous mutant, immediate upright (imm), exhibited several phenotypic traits characteristic of the gametophyte generation during the sporophyte generation of its life cycle, including asymmetric initial cell division. This mutant produced functional meiospores, demonstrating that symmetric initial cell division is not essential for an individual to become a functional sporophyte. However, specific alterations in the expression of generation-specific genes were detected in this mutant using a microarray approach, indicating that imm is a bona fide life cycle mutant and hence that the phenotypic traits that are modified in imm are normally under life cycle control in wild-type algae.

The Ectocarpus siliculosus strains used were meiotic offspring of a field sporophyte collected in 1988 in San Juan de Marcona, Peru. The gametophyte clone used for genome sequencing, Ec 32, is from the same lineage(Table 2). The strains were cultivated in 7-8 ml polystyrene Petri dishes, occasionally on glass cover-slips, normally at 15°C in white fluorescent light of 10-30 μmol m^-2s^-1 photon fluence rate. Additional light regimes used were: natural diffuse light at 5°C, dim (1 μmol m^-2s^-1) white light at 15°C and bright (10-30μmol m^-2s^-1) white light at 10 and 20°C. Day length was 10 hours light: 14 hours dark, except for the 5°C condition in which day length varied according to the season. The culture medium was half-strength, Provasoli-enriched (Starr and Zeikus, 1993), autoclaved seawater. For the study of germination, spores or gametes were allowed to settle on cover-slips. Zygotes were produced and isolated as described by Peters et al.(Peters et al., 2004b). The sex of gametophytes was determined by microscopic observation of zygote formation in hanging-drop preparations(Kawai et al., 2005) in which the strains to be tested had been combined with fertile thalli of male and female reference strains. Genetic analysis of meiotic offspring followed Müller (Müller,1991).

To determine the response of the gametophyte and the sporophyte to a light stimulus, algae were grown at low density in 5 cm (7-8 ml) Petri dishes under unidirectional white light. The orientation of germination was scored according to four quadrants (i.e. towards the light, away from the light or in one of the two quadrants perpendicular to the light). Algae that germinated into the quadrant away from the light were scored as exhibiting negative phototropy.

A microarray was constructed by spotting PCR amplified inserts from two suppression subtraction hybridisation (SSH) libraries, enriched for genes expressed preferentially during the sporophyte and gametophyte generations,respectively, onto glass slides. The SSH libraries were produced in a differential screen between the partheno-sporophyte and gametophyte generations of the life cycle using the PCR Select kit (Clontech, Mountain View, CA, USA). Oligonucleotides corresponding to flanking vector sequences were used to amplify inserts from 600 clones randomly chosen from each library and these 1200 PCR products were spotted in triplicate on each microarray. RNA was extracted as described by Apt et al.(Apt et al., 1995) and its quality was monitored on a Bioanalyser using an RNA 6000 Nano Assay kit(Agilent, Santa Clara, CA, USA). Labelled cDNA targets were synthesised from 15 μg of total RNA with an oligo(dT) primer and by indirect labelling using the CyScribe cDNA Post Labelling kit (Amersham, Piscataway, NJ, USA). Hybridised microarrays were scanned with a GenePix 4000 (Molecular Devices Corporation, Downingtown, PA, USA) and analysed using GenePix Pro 5.1. All hybridisations were carried out in triplicate. Data corresponding to a set of 132 probes that did not show differential expression during the sporophyte and gametophyte generations were used to normalise the data obtained from the different hybridisation experiments. The average value for the coefficient of variation was 0.29 for the wild-type partheno-sporophyte samples and 0.35 for the imm mutant partheno-sporophyte samples. Full descriptions of the array design, probe sequences, RNA extraction, cDNA labelling, hybridisation and normalisation protocols are available via the ArrayExpress database(accession number E-MEXP-848). Statistical analysis was carried out using the Statistical Analysis of Microarrays (SAM) method(Tusher et al., 2001) in the TIGR MeV package, version 3.1. The cDNA fragments corresponding to probes that had been shown to be significantly, differentially expressed by this analysis were sequenced and compared with the EST and genomic sequence data available for Ectocarpus siliculosus.

Total RNA was extracted from diploid sporophytes, partheno-sporophytes and gametophytes of both wild-type and imm strains using the Plant RNeasy extraction kit (Qiagen, Courtaboeuf, France), and treated with RNAse-free DNAse-I according to the manufacturer's instructions (Qiagen). The concentration and quality of the RNA was determined by spectrophotometry and agarose gel electrophoresis. Two milligrams of total RNA were reverse-transcribed using the Superscript II RT kit (Life Technologies,Gaithersburg, MD, USA) according to the manufacturer's instructions.

Oligonucleotide sequences were designed for each gene using both Primer Express TM1.0 (PE Applied Biosystems, Foster City, CA, USA) and Oligo 4.0(Rychlik and Rhoads, 1989). cDNAs were amplified using the IQ Sybrgreen supermix (Biorad Laboratories,Hercules, CA, USA) on a Chromo4 System thermocycler (BioRad Laboratories). The amplification efficiency was tested using a genomic dilution series and was always at least 80%. The specificity of amplification was checked with a dissociation curve. Details of the oligonucleotides used can be provided on request. E. siliculosus genomic DNA was used as a quantification reference. A dilution series ranging from 37 to 48671 copies of the E. siliculosus genome was prepared and tested for each gene amplification. The EF1α gene was chosen as a constitutively expressed control based on tests carried out on three Ectocarpus genes (encoding actin,EF1α and α-tubulin). Normalisation was carried out using the EF1α data and the geNorm method (Vandesompele et al., 2003). The normalised data were expressed as the mean±s.d. calculated from three independent biological experiments.

The sexual life cycle of E. siliculosus involves an alternation between two independent generations: the sporophyte and the gametophyte(Fig. 1). To compare the early development of these two generations, gametophytes were raised from meiospores of a heterozygous, field-isolated, sporophyte (strain Ec 17) and their development in culture compared with that of sporophytes that were also derived from the Ec 17 strain (see Table 1 for a glossary of the terms used in this section). A total of 120 gametophytes were isolated and all exhibited the same pattern of early development. Germination was bilateral and an asymmetric division of the initial meiospore cell resulted in the occurrence of different cell types in the two emerging germ tubes. The first germ tube developed into a rhizoidal filament 3-5 μm in diameter (Fig. 2A), whereas the second germ tube formed an upright filament(Fig. 2B). The rhizoid had a wavy appearance and, as it originated 2-7 days before the emergence of the second germ tube, it was often quite long by the time that the upright filament elongated (Fig. 2C). The cells of the upright filament were characterised by broader (initially 10μm in diameter but enlarging to 20-40 μm in higher thallus parts) and regularly cylindrical cells (Fig. 2B,D). The upright filaments produced laterals from the distal end of filament cells (Fig. 2E) to form a richly branched thallus (Fig. 2F). After 3-5 weeks in white light, gametophytes became fertile with plurilocular gametangia forming at terminal or lateral positions on the upright filaments (Fig. 2G). On older gametophytes, additional rhizoids were formed as laterals from the basal ends of upright filaments (Fig. 2H,I). Gametophytes adhered only weakly to glass or polystyrene,easily becoming detached and floating off into the medium. The overall development of the gametophyte was of the immediate differentiation type because the erect thallus developed directly without prior establishment of a prostrate basal structure.

E. siliculosus sporophytes can be produced in a number of different ways (Fig. 1): (1)via gamete fusion and zygote production; (2) by mito-spores produced by the plurilocular sporangia of sporophytes; (3) from a minority of the meio-spores produced by unilocular sporangia (the majority producing gametophytes); or (4)by parthenogenesis from settled unfertilised gametes (referred to as partheno-sporophytes). The early development of sporophytes produced by each of these alternative pathways was compared with the early development of the gametophyte generation of the same strain, described above. All sporophytes showed the same pattern of development, except that the growth of partheno-sporophytes was slower during the first few days. Germination was bipolar, as in the gametophyte. However, in contrast to the situation observed in the gametophyte, the two daughter cells of the sporophyte initial cell exhibited symmetric cell fates, producing the two ends of a symmetric prostrate filament (Fig. 3A-E). Because of these identical cell fates, the initial cell division in the sporophyte is defined as symmetric despite the morphological asymmetry of the initial cell at the time of cell division(Fig. 3B). This follows the definition given by Morrison and Kimble(Morrison and Kimble,2006).

The cells of the prostrate filament became rounder and their cell walls thickened as they became older (i.e. in the central region of the filament; Fig. 3E,F). Laterals with the same morphology as the initial filament were produced from the rounded cells,and grew along the surface of the substratum(Fig. 3F) or up into the medium. Upright filaments with cylindrical cells (20-30 μm diameter)developed on the prostrate base after 4-5 weeks to produce an erect thallus(Fig. 3G,H). The upright filaments usually emerged from older regions of the prostrate base. The mature sporophyte consisted of a well-developed, prostrate basal system from which emerged the upright filaments (Fig. 3I). Compared with gametophyte germlings, the prostrate sporophyte thalli were not easily detached from either polystyrene or glass surfaces. Plurilocular sporangia developed on upright filaments after 7 weeks, in terminal or lateral positions (Fig. 3J). Occasionally, plurilocular sporangia developed on the prostrate structure, in some cases even before the emergence of upright filaments (not shown). Unilocular sporangia formed only on upright filaments,in terminal or lateral positions, a week after the plurilocular sporangia(Fig. 3K). The upright filaments were branched (not shown), although to a lesser extent than those of a gametophyte, and, in older thalli, also produced rhizoids (not shown). Rhizoids originated only from the upright filaments and were not produced by the filaments of the prostrate structure(Fig. 3H). Overall, the sporophyte exhibited mediate differentiation because a prostrate, basal structure was formed before the development of the erect thallus (i.e. the upright filaments).

In summary, observation of the early development of the two generations of the E. siliculosus life cycle showed that they exhibited markedly different patterns of development. These two patterns of development involved either asymmetric or symmetric divisions of the initial cells of gametophytes or sporophytes, respectively. The gametophyte formed two types of vegetative structures (rhizoids and upright filaments) and one kind of reproductive structure (plurilocular gametangia), the sporophyte three types of vegetative structures (prostrate filaments, rhizoids and upright filaments) and two kinds of reproductive structure (plurilocular and unilocular sporangia).

Initial cell (zygote) photopolarisation has been extensively studied in members of the Fucales group (Corellou et al., 2005; Robinson and Miller, 1997; Fowler et al.,2004; Brownlee and Bouget,1998). In this group, photopolarisation causes the zygote to germinate in a negatively phototropic manner, emitting a rhizoid on the side opposite the light source. The markedly different germination patterns of the E. siliculosus sporophyte and gametophyte led us to test their responses to unidirectional light. When grown in unidirectional light, the majority of individuals of both the sporophyte and gametophyte generations germinated in a negatively phototropic manner(Fig. 4). However, this response was significantly less marked in the sporophyte generation, where only 73% of individuals germinated away from the light, compared with 93% for the gametophytes (χ²=25.68, P<0.001).

The ability of E. siliculosus to produce partheno-sporophytes from unfertilised, haploid gametes allows the phenotypic effects of mutations to be detected directly in both the gametophyte and the (partheno-)sporophyte generations. Phenotypic analysis of partheno-sporophytes derived from the 120 gametophytes produced by the Ec 17 sporophyte identified one gametophyte strain (Ec 137) that produced partheno-sporophytes with an aberrant germination pattern. The partheno-sporophytes of this strain exhibited asymmetric division of the initial cell rather than the usual pattern of symmetric division of the initial cell (exhibited by the partheno-sporophytes derived from its sibling gametophytes). The same aberrant developmental phenotype was seen in diploid sporophytes homozygous for the mutant locus (see below for the production of such thalli). The aberrant germination pattern involved the production of a first germ tube that developed into a thin (3-5μm diameter) filament that resembled the rhizoid produced by the germinating gametophyte (compare Fig. 5A with Fig. 2A). The second germ tube developed into an upright filament (initially 10 μm in diameter, but widening to 20-30 μm in the upper part of the filament),again similar to that produced by germinating gametophytes (compare Fig. 5B with Fig. 2D). Based on this phenotype, the mutant was named immediate upright (imm). The aberrant germination pattern was observed for all developing gametes of Ec 137 and was not modified by cultivation at 5, 10 or 20°C.

The aberrant early development of the imm mutant had a significant effect on the morphology of the alga later in development. After 6 weeks growth, the imm mutant had produced a well-developed erect thallus attached to the substratum with rhizoids(Fig. 5C), whereas wild-type sporophytes had, after the same period of growth, developed a dense,prostrate, basal system from which only the first upright filaments emerged(Fig. 3I). This morphological difference between wild-type and imm sporophytes was already visible under the light microscope after 1-2 weeks growth and the two types of growth habit could be distinguished with the naked eye after 4-5 weeks. This is illustrated by the macroscopic views shown in Fig. 5D,E.

Despite the resemblance of the imm partheno-sporophyte to the wild-type gametophyte, further analysis showed that it retained its sporophyte identity. Mature imm partheno-sporophytes produced not only plurilocular but also unilocular sporangia, reproductive structures that are only produced by the sporophyte generation(Fig. 5F,G). Both of these structures can either emerge directly from a filament or be borne on a short branch, in both the wild type and in the imm mutant. Mating combinations with gametophyte reference strains showed that the spores released from the plurilocular sporangia were indeed spores and not gametes(they were incapable of fusing with fertile gametes of either sex). Even during early development, and despite their resemblance to the gametophyte, imm partheno-sporophytes exhibited several features that were typical of the wild-type sporophyte. For example, imm partheno-sporophytes adhered firmly to the substratum and were difficult to detach. Moreover, when grown in unidirectional light, the negative phototropic response was less marked than that detected normally for the gametophyte and more similar to that of the wild-type sporophyte, with only 68% of the individuals tested germinating away from the light (Fig. 4). It was also noted that, although the first germ tube developed as a filament in the same manner as in the gametophyte, it was less wavy in appearance (compare Fig. 5Bwith Fig. 2C,D).

The meio-spores produced in the unilocular sporangia of the immpartheno-sporophytes developed into phenotypically normal gametophytes that produced functional gametes (data not shown). All asexually produced sporophytes, derived either from mito-spores produced in the plurilocular sporangia of the imm partheno-sporophytes or via the gametophyte generation by parthenogenetic germination of gametes, exhibited the imm phenotype.

Taken together, these experiments showed that the imm mutant carried a heritable character that induces a modification of the early development of the sporophyte, causing it to resemble the gametophyte morphologically. The imm mutation had no observable effect on the development of the gametophyte generation. The appearance of the immmutation in one of 120 gametophytes derived from the Ec 17 sporophyte indicates that this mutation arose spontaneously in this generation of gametophytes.

The imm mutant (strain Ec 137) was crossed with a sister gametophyte (Ec 25) whose gametes showed a wild-type pattern of parthenogenetic development (see Fig. 6 for a summary of the crosses carried out for the genetic analysis). Three zygotes were raised from this cross and the resulting sporophytes all showed a wild-type pattern of early development. The development of one of these sporophytes (strain Ec 372) was followed to maturity and no differences were observed compared with the developmental pattern of a wild-type sporophyte (data not shown). This indicated that the imm mutation was recessive and was complemented by the wild-type allele in the diploid sporophyte.

To analyse the segregation of the imm locus in subsequent generations, 15 unilocular sporangia were isolated individually from the diploid sporophyte Ec 372. These sporangia, each of which contained more than 100 meio-spores (derived from a single meiosis followed by at least 5 mitotic divisions), produced 15 `families' of gametophytes, which again exhibited a germination pattern typical of wild-type gametophytes. For each of these families, between 15 and 31 gametophytes were sub-isolated and their sex and the germination pattern of their gametes (i.e. the partheno-sporophyte generation) were recorded (see Table S1 in the supplementary material). Overall, 155 gametophytes produced partheno-sporophytes with a wild-type germination pattern and 191 produced partheno-sporophytes with the imm germination pattern. These figures are consistent with a 1:1 segregation ratio and Mendelian inheritance of a single-locus recessive mutation (χ²=3.746, the deviation from 1:1 has a probability greater than 0.05).

Several additional crosses were performed to analyse further the inheritance of the imm mutation. Using the gametophytes derived from the diploid sporophyte Ec 372, crosses were performed with all possible combinations of wild-type and imm mutant gametes(Fig. 6). Eleven diploid sporophytes, homozygous for imm, were raised from the cross Ec 419(imm) × Ec 420 (imm). Without exception, they showed the imm germination pattern, indicating full penetrance of the imm mutation in this diploid context (provided the dominant,wild-type IMM allele is absent). From one of these sporophytes (Ec 428), we isolated five unilocular sporangia. From each of the families of gametophytes that developed from them, we selected six gametophytes and recorded their sex and the germination pattern of their gametes. Regardless of their sex, all of these gametophytes produced partheno-sporophytes with the imm germination pattern.

Five diploid sporophytes were raised from the cross Ec 421 (IMM)× Ec 423 (IMM) and all showed the wild-type germination pattern. From one of these sporophytes (Ec 432) we isolated five unilocular sporangia and again analysed the sex and gamete germination patterns of six gametophytes from each family. Regardless of sex, all showed the wild-type germination pattern.

Finally, from the reciprocal crosses Ec 419 (imm) × Ec 423(IMM) and Ec 421 (IMM) × Ec 420 (imm) we raised nine and three zygotes, respectively, and these all showed a wild-type germination pattern. A sporophyte from each cross (Ec 429 and Ec 430,respectively) was raised and eight (for Ec 429) and 11 (for Ec 430) families of gametophytes corresponding to single unilocular sporangia were isolated. The mutant germination pattern was observed in the germlings from ∼50% of the 18-20 gametophytes analysed per family.

In none of the above analyses did a single gametophyte produce populations of partheno-sporophytes with both types of germination pattern.

The analyses of progeny from the three imm/IMM heterozygous sporophytes (Ec 372, Ec 429 and Ec 430) were used to test for linkage between IMM and the sex locus. For this, the phenotypes of the progeny from each unilocular sporangium were used to determine the segregation pattern of the two loci during the single meiotic event that had occurred in each sporangium (see Table S2 in the supplementary material). No statistical deviation from a 1:1:1:1 ratio was detected in any of the three crosses (e.g. for strain Ec 372, χ²=0.266, P>0.9), indicating that IMM segregated independently of the sex locus.

Taken together, these data demonstrate that the imm mutation was stably inherited as a recessive Mendelian factor through several generations of the life cycle and that the IMM locus was not linked to the sex locus.

The sporophyte generation of the imm mutant exhibited several morphological features typical of the gametophyte generation during its early development, such as asymmetric initial cell division, and yet remained functionally a sporophyte (producing spores but not gametes). To determine whether these similarities with the gametophyte generation were due to a partial conversion of the sporophyte into a gametophyte, we assayed the expression of two libraries of sporophyte- and gametophyte-upregulated genes(isolated by suppression subtraction hybridisation) in the partheno-sporophyte of the imm mutant using a microarray approach. cDNA fragment inserts from 600 clones from each of the two SSH libraries were arrayed on glass slides and hybridised with fluorescently labelled cDNA. Fig. 7A shows that the relative abundances of the transcripts corresponding to the sequences spotted on the microarray were similar in two independent total RNA samples from wild-type partheno-sporophytes of the strain Ec 432. By contrast, when cDNA from the partheno-sporophyte stage of the imm mutant was compared with cDNA from the same wild-type partheno-sporophyte, there was a marked tendency for a subset of the transcripts corresponding to the gametophyte-expressed SSH library to be more abundant in the mutant cDNA sample and for a subset of the transcripts corresponding to the sporophyte-expressed SSH library to be less abundant (Fig. 7B). A statistical test, carried out using the Statistical Analysis of Microarrays(SAM) method (Tusher et al.,2001), identified 80 clones whose corresponding transcripts were significantly more abundant in the imm partheno-sporophyte RNA sample and 62 clones whose transcripts were significantly less abundant. Seventy-nine of the 80 clones corresponding to genes upregulated in the imm mutant(99%) were from the gametophyte-expressed SSH library, whereas 60 of the 62 clones corresponding to downregulated genes (97%) were from the sporophyte-expressed SSH library (Fig. 7C). Taken together with the similarity between the early morphogenesis of the imm mutant sporophyte and that of the wild-type gametophyte (Fig. 3), these data provide convincing evidence that there is a partial switch from the sporophyte to the gametophyte developmental programme in the immmutant sporophyte.

To further investigate the observed modifications in gene expression in the imm mutant, the SSH clones corresponding to genes that had been identified as significantly up- or downregulated in the imm mutant partheno-sporophyte compared with the wild-type partheno-sporophyte were sequenced. The sequences of the SSH clones were compared with each other, with a collection of 26,000 E. siliculosus EST sequences generated from an immature partheno-sporophyte cDNA library and with a 7× assembly of the E. siliculosus genome sequence (A.F.P., D.S., S.M.C., B.C., N. Aiach,B. Segurens, B. Setterblad, J. Weissenbach, P. Wincker and J.M.C.,unpublished). This analysis showed that the 142 differentially regulated SSH clones corresponded to 40 different genes (27 downregulated genes and 13 upregulated genes). These genes are listed in Table 3 (and the corresponding SSH clones are listed in Table S3 in the supplementary material). All but four of the genes that were downregulated in the imm mutant compared with the wild type were represented in the collection of 26,000 partheno-sporophyte ESTs, whereas only one of the 13 upregulated genes was represented in this collection of sequences (Table 3). This supports the data from the microarray analysis indicating that a group of genes that are normally expressed during the gametophyte stage are upregulated in the imm partheno-sporophyte, whereas a group of sporophyte-generation genes are correspondingly downregulated. When the selected upregulated and downregulated genes were compared with public databases, over half (23 genes) either did not match sequences in the database or matched only hypothetical proteins(Table 3).

Table 3 lists the genes that were identified, by the microarray analysis, as being significantly downregulated (IDW genes) or upregulated (IUP genes) in the immpartheno-sporophyte compared with a wild-type partheno-sporophyte. The abundances of mRNAs corresponding to ten of the IDW genes and to nine of the IUP genes were measured in the sporophyte and gametophyte generations of both the imm mutant and wild-type strains using quantitative PCR.

The results for the wild-type partheno-sporophyte and gametophyte samples showed that most of the IDW genes exhibited a sporophyte-specific or sporophyte-preferential pattern of expression, whereas most of the IUP genes exhibited a gametophyte-specific or gametophyte-preferential pattern of expression (Fig. 8). The difference between the expression levels in the two generations was very marked in most cases, with a maximum fold difference of 6975 for IUP9. Altogether, the quantitative PCR analysis allowed the validation of 16 genes (nine for the sporophyte and seven for the gametophyte)that showed a differential pattern of expression between the two generations(Fig. 8).

We compared the expression levels of these genes in the partheno-sporophyte and gametophyte of the imm mutant with the values obtained for the wild type (Fig. 8). With only a few exceptions (IDW7, IDW9, IDW11, IUP8 and IUP9), the genes were expressed at a similar level in imm and wild-type gametophytes. This is consistent with the fact that we did not detect any visible phenotype of the imm mutation in the gametophyte generation. In addition, in the imm partheno-sporophyte, most of the genes assayed showed an expression level that was either comparable with that of the wild-type gametophyte (IDW1, IDW2, IDW3, IUP1) or was intermediate between the abundances assayed in the two generations of the wild type (IDW4, IDW5,IDW6, IDW7, IDW8, IUP2, IUP3, IUP4, IUP5, IUP6, IUP7, IUP8, IUP9). For the last five genes, the transcripts were markedly less abundant in the imm partheno-sporophyte compared with the wild-type gametophyte but they were, nonetheless at least 28 times more abundant than in the wild-type partheno-sporophyte (see Table S4 in the supplementary material). The quantitative PCR, therefore, confirmed that there was an increase in the abundance of the transcripts of gametophyte-expressed genes and a decrease in the abundance of sporophyte-expressed genes in the immsporophyte.

Quantitative PCR analyses were carried out on partheno-sporophytes because,as mentioned above, we observed no morphological or developmental differences between partheno-sporophytes and diploid sporophytes derived from gamete fusions. This was the case for both the wild-type Ec 32 strain and for the imm mutant. To further verify that the two stages of the life cycle behaved in the same manner, we also measured transcript abundances for five of the genes tested in Fig. 8 at both stages for both the wild type and the imm mutant. The results of this experiment (see Fig. S1 in the supplementary material) indicate, as expected, that partheno-sporophytes and zygote-derived sporophytes exhibit very similar patterns of gene expression and that the deregulation of generation-specific genes observed in the imm partheno-sporophyte also occurred in a imm/imm diploid sporophyte.

Here, we show that the sporophyte and gametophyte generations of E. siliculosus exhibit markedly different patterns of early development and correlate this difference with the mode of division of the initial cells during each generation. The early stages of gametophyte development involve asymmetric division of the initial cell and immediate differentiation, whereas the sporophyte arises via a symmetric initial cell division and exhibits mediate differentiation.

Explaining the stability of haploid-diploid life cycles has been particularly problematic because, depending on the niche of the organism,either the haploid or the diploid generation is expected to present certain advantages and theoretical models predict that this should lead to a dominance of one generation over the other under a wide range of conditions(Mable and Otto, 1998). It has been suggested, however, that if the two generations are adapted to different ecological niches, this could stabilise a haploid-diploid life cycle (Stebbins and Hill, 1980; Willson, 1981)and this suggestion is supported by more recent theoretical work(Hughes and Otto, 1999). It is possible that the difference in morphology between the wild-type sporophyte and gametophyte of Ectocarpus reflects an adaptation to different ecological niches, with the dense, more robust thallus of the sporophyte,particularly its prostrate base, being better adapted for persisting in less favourable conditions during most of the year and the more fragile gametophyte being short-lived and adapted for producing gametes over a period of a few weeks. Future work on the ecology of the two generations of the Ectocarpus life cycle might therefore provide a means to test theoretical hypotheses concerning the stability of haploid-diploid life cycles.

The gametophyte generation of the imm mutant was phenotypically wild type, but the sporophyte generation exhibited a gametophyte-like pattern of germination and early development. Microarray analysis of gene expression in the mutant strain showed that the morphological resemblance of the imm sporophyte to a gametophyte was correlated with a specific upregulation of a subset of the genes that are normally expressed during the gametophyte generation and a corresponding downregulation of a subset of sporophyte-generation genes. This analysis, therefore, provided evidence that the mutant was affected in processes that are regulated during the life cycle. In addition, phenotypic analysis of the imm mutant showed that the development of mature sporophytes with the appropriate reproductive structures(unilocular and plurilocular sporangia) was not contingent on the wild-type sporophyte pattern of early development (involving symmetric division of the initial cell and mediate differentiation). These two, temporally separate parts of the sporophyte developmental programme were genetically uncoupled in the imm mutant. Hence, not only does E. siliculosus exhibit alternation between developmental programmes involving symmetric and asymmetric initial cell divisions during the sporophyte and gametophyte phases of the life cycle, respectively, but both symmetric (in the wild type) and asymmetric (in imm individuals) initial cell divisions can give rise to the sporophyte generation. Moreover, because the imm mutation represents a single locus, germination and subsequent cell division patterns might have been modified quite easily during evolution in the brown algal lineage. This may help explain the variety of germination patterns observed in different brown algal species.

Microarray analysis identified 40 genes whose level of expression was modified in the imm mutant. This number is an underestimate because only a small subset of the Ectocarpus gene set was represented on the array. These data suggest that IMM may be regulating a cascade of downstream genes to mediate its influence on early sporophyte development. Analysis of the sequences of the genes whose expression patterns were altered in the imm mutant provided only limited information about the molecular events underlying the observed phenotypes.

Current efforts aimed at developing E. siliculosus as a model species (Peters et al., 2004a; Coelho et al., 2007; Charrier et al., 2008) are expected to provide a means to investigate, at the molecular level, the mechanisms behind the phenomena described in this study.

This work was supported by the Centre National de Recherche Scientifique,the Région Bretagne, the European Union network of excellence Marine Genomics Europe, the French GIS Marine Genomics, the University Pierre and Marie Curie, and by a long-term European Molecular Biology Organisation fellowship to S.M.C.