INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Diatoms are one of the most abundant eukaryotic unicellular algae known, with as many as 20,000 extant species distributed among marine and freshwater ecosystems (27, 36). A defining characteristic of diatoms is their siliceous cell wall, or frustule. The physical and developmental constraints associated with the generation of this relatively inflexible cell wall lead to one of the more intriguing aspects of the diatom life cycle: each mitotic division results in the formation of two differently sized daughter cells, one that is the same size as the parent and one that is slightly smaller. Over successive generations, therefore, the mean cell size of a population decreases and the standard deviation about this mean increases (26, 35).

The most common manner of escaping this trend of diminishing cell size is through sexual reproduction (reviewed in reference 12). It is generally believed that once a cell decreases in size to a diameter of less than about 30 to 40% of the maximum diameter for a given species (11), it can be induced by a wide variety of environmental triggers (see, e.g., references 2, 10, 18, and 41) to exit the mitotic cycle and initiate sexual reproduction. Centric diatoms are monoecious, and thus both sperm and eggs can be formed within a single clone (11). It remains unclear how the sperm and egg find one another, although some have hypothesized that the eggs produce pheromone-like compounds to attract sperm (see, e.g., reference 10). Once the sperm and egg do recognize one another, a series of signalling events must allow the sperm to gain entry past the egg frustule so that the gametes can fuse to create the diploid zygote, or auxospore. The auxospore then breaks free of its frustule and forms a postauxospore cell that is able to generate an entirely new frustule and thus a cell many times larger than either parent (30, 33). Importantly, these newly formed large cells rapidly resume asexual reproduction and are essentially "immune" to further induction triggers until an appropriately small cell size is obtained and external triggers can once again elicit the sexual cycle. Thus, those newly generated large cells whose genotypes convey selective advantages can be quickly dispersed throughout a population (see, for example, reference 1).

Sexual reproduction and the accompanying generation of genetic diversity in diatoms are therefore intimately coupled to the control of cell size. The mechanisms underlying this coupling remain mysterious, however. As a first step toward teasing apart the steps required for the transition from vegetative growth to sexual reproduction, I have begun to identify sexual reproduction-specific genes in the centric diatom Thalassiosira weissflogii. An extremely sensitive PCR-based cDNA subtraction technique was used to identify, for the first time, genes that are transcribed within the first few hours of entry into the sexual cycle, a period that precedes the major morphological changes associated with gamete formation (2). The identification of sexual reproduction-specific genes in diatoms will ultimately permit the generation of molecular markers specific to sexually reproducing cells and thus will allow a determination of when and where sexual reproduction occurs in the field, a question that has been difficult to address by more traditional techniques.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Culture conditions. Clonal isolates of T. weissflogii Grun. clone Actin (from the Culture Collection of Marine Phytoplankton, Bigelow Laboratories for Ocean Sciences) were obtained by plating cells on f/2 enriched seawater (20) solidified with 1.5% agar (Meer Corporation) and then transferring individual colonies to liquid f/2 medium. The size distributions of the isolates were determined with a Coulter Multisizer II (Coulter Corporation). A number of isolates were chosen based on their size distributions and maintained in exponential growth in a semicontinuous batch culture at 20°C and 120 microeinsteins of continuous illumination · m² · s¹ (1). Each isolate was tested for responsiveness to a sexual induction trigger. The induction signal used for all experiments involves an interruption of exponential growth in continuous light with 12 hours of darkness (2). Once dark-exposed cultures are returned to continuous light, induced cells enter relatively synchronously into the sexual cycle (2). Approximately 24 h after a return to continuous light, aliquots of the induced cultures were examined microscopically to determine whether sexual stages were present. A culture that formed sexual stages in response to the dark induction was defined as responsive. A culture that did not form sexual stages in response to the dark induction trigger was defined as unresponsive.

RNA isolation. Samples were collected for total RNA isolation 5 h after dark-induced cultures were returned to continuous light. Ten liters of induced cultures at approximately 7 × 10⁴ cells · ml¹ were filtered through 1.2-µm-pore-size Millipore filters, and the filtered cells were either frozen at 70°C before processing or else processed immediately. Total RNA was isolated essentially as described by Kirk and Kirk (24). Briefly, approximately 10 ml of lysis buffer (50 mM Tris [pH 8], 0.3 M NaCl, 2% sodium dodecyl sulfate [SDS], 15 mM EGTA, and 1.5% freshly added diethyldithiocarbamic acid) was used per 3.5 × 10⁸ filtered cells, and the cells were incubated at 37°C for 30 min with intermittent vortexing. Cell debris was removed by centrifugation at 10,000 × g for 10 min, and 2 M KCl was added to the resulting supernatant to achieve a final concentration of 0.23 M KCl. The mixture was incubated on ice for 15 min and centrifuged at 10,000 × g for 10 min. A 1 M Tris (pH 9) solution was added to the resulting supernatant to achieve a final concentration of 34 mM, and then the supernatant was extracted twice with Tris-buffered phenol (Amresco). Nucleic acids were precipitated with ethanol at 20°C, and the pellet was resuspended in 4 ml of water. RNA was precipitated overnight on ice by the addition of an equal volume of 4 M LiCl. The RNA was pelleted at 10,000 × g for 10 min and resuspended in water or TE (10 mM Tris [pH 7.6]-1 mM EDTA). Total RNA was quantified with a GeneQuant RNA/DNA calculator (Pharmacia). Poly(A)-selected RNA was isolated according to the manufacturer's instructions by using the Oligotex mRNA Isolation Kit (Qiagen).

cDNA subtraction. cDNAs were generated and subtracted according to the manufacturer's instructions by using the PCR-Select cDNA Subtraction Kit (Clontech). Briefly, the mRNA isolated from the responsive culture and the unresponsive culture was reverse transcribed by using avian myeloblastosis virus reverse transcriptase (Clontech) in two separate reactions to generate two populations of double-stranded cDNAs, one representative of the genes transcribed in the responsive culture and one representative of the genes transcribed in the unresponsive culture. Both sets of cDNAs were restriction digested with RsaI to generate cDNA fragments. The digested cDNAs from the responsive culture were split into two aliquots. Adapter 1 (supplied with the kit) was ligated to the cDNAs from one aliquot; adapter 2R (supplied with the kit) was ligated to the cDNAs from the other aliquot. A ligation efficiency test was performed to ensure that at least 25% of the cDNAs from each aliquot possessed the appropriate adapter. For the efficiency test, two gene-specific control primers were designed to amplify the carbonic anhydrase gene recently cloned from T. weissflogii (34). The carbonic anhydrase-specific PCR primers are ACCTCGATATGGAGACTCTTC (forward) and CCCATTCCCATTTCTTCATCG (reverse).

cDNA screening. To determine if the subtracted library contained clones up-regulated in the responsive culture, 117 colonies were randomly chosen and the cDNA insert from each was PCR amplified with the adapter-specific primers. Approximately 0.5 µg of each insert DNA was denatured by addition of an equal volume of 0.6 N NaOH and was then spotted onto duplicate maximum-strength Nytran Plus (Schleicher and Schuell) membranes. The membranes were incubated in 0.5 M Tris (pH 7.5) for 4 min and allowed to air dry, and the DNA was UV cross-linked to the membrane with a UV Stratalinker 1800 (Stratagene). The replicate membranes were then hybridized to either a forward-subtracted or a reverse-subtracted cDNA probe. These probes were generated by first sequentially digesting the populations of reverse- or forward-subtracted cDNAs with RsaI, SmaI, and EagI to ensure complete removal of the adapters. The digested cDNAs were separated from the adapters with a Qiaquick column (Qiagen), and the cDNAs were fluoroscein labeled by using the Random Prime Labeling and Signal Amplification System for the Fluorimager (Vistra). The hybridization, wash, and signal detection conditions used were those suggested in the Random Prime Labeling kit. Only those inserts that hybridized to the forward-subtracted probe but not to the reverse-subtracted probe were screened further. A subset of the inserts from these positive clones was PCR amplified and labeled with the Random Prime Labeling and Signal Amplification System. Approximately 0.5 µg of the unsubtracted cDNAs from the responsive and unresponsive cultures were denatured as before, spotted onto duplicate Nytran membranes, and separately probed with the individually labeled cDNAs.

DNA sequencing and generation of full-length cDNA clones. Plasmid DNA was prepared from positive clones with the Qiagen Mini Prep Kit and sequenced with the ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction Kit with AmpliTaq DNA polymerase by using a combination of adapter-specific and gene-specific primers. DNA sequencing was performed on an Applied Biosystems 373A DNA Sequencer. Sequence data were compiled and analyzed with the Wisconsin Package, version 10.0, of the Genetics Computer Group (GCG), Madison, Wis. (9).

Isolation of genomic clones. Total genomic DNA was isolated from T. weissflogii by filtering approximately 5 · 10⁶ cells onto a 0.45-µm-pore-size cellulose filter. The cells were scraped from the filter and incubated at 60°C for 1 h in 300 µl of lysis buffer (10 mM TE [pH 7.5]-0.5% SDS-100 µg of proteinase K/ml); 50 µl of 5 M NaCl and 40 µl of 10% cetyltrimethylammonium bromide (CTAB) in 0.7% NaCl was then added, and the mixture was incubated at 65°C for an additional 10 min. The DNA was purified by using the column and wash reagents provided with the Qiagen DNeasy Plant Mini Kit according to the manufacturer's instructions. Gene-specific PCR primers were used to amplify genomic fragments. These fragments were cloned into pGEM-T and transformed into TOP10 Escherichia coli cells. Recombinant plasmid DNA was isolated and sequenced as described above.

Reverse transcriptase PCR (RT-PCR). Two hundred nanograms of poly(A)-selected RNA isolated from the responsive and the unresponsive cultures were reverse transcribed in two separate reactions by using Moloney murine leukemia virus reverse transcriptase (Gibco BRL), along with the primers and conditions for first-strand synthesis provided with the SMART cDNA Synthesis kit (Clontech). Gene-specific forward and reverse PCR primers that had been designed to cross an intron were used to PCR amplify first-strand cDNAs. PCR products were analyzed by agarose gel electrophoresis.

Detection of homology and sequence alignment. Homology between the predicted amino acid sequences and those present in the GenBank database was detected by using BLAST 2.0, provided by the National Center for Biotechnology Information (32a). Multiple amino acid alignments were performed by using the ClustalW 1.7 programs (2a). Corresponding DNA sequences were aligned by hand.

Nucleotide sequence accession numbers. The GenBank accession numbers for the Sig sequences are AF154499, AF154500, and AF154501.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of responsive and unresponsive isolates. Clonal cultures of T. weissflogii representing a range of size distributions were tested for responsiveness to a dark induction trigger (2). Two clones that displayed different size distributions and widely different responses to the induction cue were chosen (Fig. 1). Since the focus of this study was to identify genes expressed during the early stages of sexual reproduction, mRNA was isolated from the two induced cultures 5 h after their return to continuous light, a time that precedes the obvious morphological changes associated with sperm formation (2). Twenty-four hours after the return to continuous light, approximately 40% of the cells observed in a remaining aliquot of the responsive culture were readily identifiable male gametes, whereas about 1% of the cells in the unresponsive culture were male gametes. Rare auxospores (data not shown) were observed only in the responsive culture, suggesting that even more cells within this culture initiated the sexual cycle, since some of the "vegetative" cells must have been eggs.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Coulter size distributions of the responsive (thin line) and unresponsive (thick line) isolates used for sexual induction and a simplified schematic of the resulting life cycle features of the responsive (left side) and unresponsive (right side) cells. The mean cell diameter of the responsive culture was 10.8 µm, and the mean cell diameter of the unresponsive culture was 12.9 µm.

Multiple genes are up-regulated during the onset of sexual reproduction. One hundred seventeen individual clones from the sexual gene-enriched library (forward-subtracted) were screened with probes made from either the sexual (forward-subtracted) cDNAs or the vegetative (reverse-subtracted) cDNAs to determine the percentage of the library that contained cDNAs up-regulated during the onset of sexual reproduction. Of the 117 cDNAs tested, 25 hybridized specifically to the forward-subtracted probe and 15 hybridized to both the forward- and reverse-subtracted probes, indicating that these cDNAs had somehow "slipped through" the subtraction process. The vast majority of the cDNAs, however, hybridized to neither probe at a detectable level, suggesting that cDNAs of this class are expressed at relatively low levels (data not shown). Nineteen of the 25 strongly hybridizing cDNAs were used as individual probes against the total (unsubtracted) unresponsive and responsive cDNA populations to confirm that these particular cDNAs were truly up-regulated in the sexual culture. Eleven of these cDNAs appeared to be expressed at higher levels in the responsive cultures, although two of the clones were later shown to correspond to the same gene. Thus, a total of 10 of the 19 cDNAs tested appeared to be up-regulated in the responsive culture (Fig. 2), although some cDNA clones, such as clones 56 and 52, appeared to be only slightly up-regulated, while others, such as clones 29 and 71, appeared to be strongly up-regulated.

View larger version (79K): [in this window] [in a new window]   FIG. 2.   Comparison of the steady-state levels of transcription of 10 differentially expressed cDNA clones. Spots of approximately 0.5 µg of the total cDNAs isolated from the unresponsive (U) and responsive (R) cultures 5 h after the dark-induced cultures were returned to continuous light are shown.

A novel gene family is expressed during sexual reproduction. The 10 positive cDNA clones were sequenced, and three of these partial cDNAsclones 42, 71, and 78displayed similar features. The genes corresponding to these three clones are now referred to as Sig, for sexually induced gene. Sig1 corresponds to cDNA clone 42, Sig2 corresponds to cDNA clone 71, and Sig3 corresponds to cDNA clone 78. Gene-specific RACE PCR primers were designed to amplify the 5' and 3' ends of these cDNAs and thus allow the full-length amino acid sequences to be predicted. In each instance, the initiator methionine was assumed to be the first methionine of the longest open reading frame. The full-length Sig1 cDNA is 2,432 bp, with a 5' untranslated region (UTR) of 23 bp and a 3' UTR of 56 bp; the full-length Sig2 cDNA is 1,395 bp, with a 5' UTR of 20 bp and a much longer 3' UTR of 320 bp; the full-length Sig3 cDNA is 906 bp, with a 5' UTR of 47 bp and a 3' UTR of 198 bp. The Sig1 genomic sequence contains four introns at cDNA positions (from the 5'-most end of the cDNA) 186, 709, 1198, and 1985. The Sig2 and Sig3 genomic sequences each contain a single intron at cDNA positions 123 and 144, respectively (Fig. 3A). The sizes of these introns are remarkably homogeneous, ranging only from 77 to 86 bp. Moreover, the introns are more A+T rich (G+C content of 26 to 39%) than the coding regions (G+C content of 47 to 49%).

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Structure and expression of Sig1, Sig2, and Sig2. (A) Schematic of the genomic structures of the Sig1, Sig2, and Sig3 transcription units. Rectangles, exons; lines, introns; solid rectangles, untranslated RNA. Arrows indicate approximate locations of the PCR primers used for RT-PCR. (B) Two hundred nanograms of poly (A)-selected RNA isolated from the responsive (R) and unresponsive (U) cultures 5 h after the dark-induced cultures were returned to continuous light were reverse transcribed. Total genomic DNA (G) or equal amounts of the first strand-cDNAs (R or U) were PCR amplified with the primers specific to Sig1, Sig2, and Sig3 shown in panel A. Molecular weight markers (in kilobases) are shown on the left. PCR primers specific to carbonic anhydrase (Cal) were used to PCR amplify equal amounts of the first-strand cDNAs, and PCR products were analyzed on a second gel; molecular weight markers for this gel are shown on the right.

View larger version (61K): [in this window] [in a new window]   FIG. 4.   Alignment and structure of the predicted amino acid sequences of SIG1, SIG2, and SIG3. (A) Alignment of the predicted amino acid sequences; amino acid numbering, beginning with the initiator methionine, is shown to the right of each line. Dashes, alignment gaps; solid diamond, potential cleavage site of the signal sequences; boldfaced N's, potential N-linked glycosylation sites; areas highlighted in black, amino acid identity domains I through V. The location of RGD in SIG1 is indicated by asterisks. The EF hand in SIG1 and the ATP/GTP binding site in SIG2 are boxed. (B) Schematic of the structure and orientation of domains I through V within the SIG polypeptides. The five different patterns represent the five different domains, arranged from left to right with domain I leftmost. Only domain I of SIG2 and domain III in each polypeptide do not contain an EGF-like motif. Open rectangles, nonconserved regions. In SIG1, the asterisk above domain I indicates the location of the RGD motif. In SIG2, the asterisk indicates the location of the ATP/GTP binding motif. The short vertical lines above SIG1 and SIG2 indicate the locations of potential N-linked glycosylation sites.

The SIG polypeptides display homology to the EGF-like domain of the vertebrate extracellular matrix protein tenascin. The SIG polypeptides display strong homology to the EGF-like domain of a family of extracellular matrix glycoproteins known as tenascin. Each member of the tenascin family (tenascin X, R, and C) is composed of four functional domains that together promote cell-cell interactions during different stages of development (7, 15). The strongest homology is observed between SIG3 and EGF-like domains of tenascin X from mice (GenBank accession no. 2564958), humans (4), and cows (13). Since SIG3 is composed primarily of domains I through V only (Fig. 4), and because these five domains occur in the same order in each SIG polypeptide (Fig. 4), composites of the amino acid sequences corresponding to the five domains for each polypeptide were constructed and compared to tenascin X. Sequence homologous to the entire length of the five domains is found in six overlapping repeats throughout the EGF-like domains of mouse and human tenascin X and in four overlapping repeats throughout the EGF-like domain of bovine tenascin X (Fig. 5). This strongly conserved homology between a diatom and a vertebrate protein again suggests that the five domains within the SIGs may reflect functional domains.

View larger version (113K): [in this window] [in a new window]   FIG. 5.   Alignments of composites of the predicted amino acid sequence from domains I through V for SIG1, SIG2, and SIG3 with the EGF-like domains of tenascin X from mice (GenBank accession no. 2564958), humans (4), or cows (13). Alignment gaps are indicated by dashes. Identical and similar amino acids (similarly charged acidic [D and E] or basic [R, K, and H] residues or uncharged [M, L, V and A] residues) are highlighted in black. Amino acids 19 to 206 of SIG3 are included. Amino acids 391 to 575 (Ms1), 298 to 451 (Ms2), 236 to 389 (Ms3), 484 to 637 (Ms4), 546 to 732 (Ms5), and 132 to 296 (Ms6) of the mouse sequence, amino acids 283 to 467 (Hu1), 407 to 591 (Hu2), 221 to 405 (Hu3), 500 to 681 (Hu4), 146 to 312 (Hu5), and 593 to 748 (Hu6) of the human sequence, and amino acids 403 to 619 (Bov1), 279 to 463 (Bov2), 558 to 744 (Bov3), and 142 to 308 (Bov4) of the bovine sequence are included.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The ability of centric diatoms to form anisogamous sperm and eggs was first suspected in the mid-1930s (see, e.g., references 3 and 19) and eventually confirmed some 15 years later (45). Since then, the life cycles of numerous centric diatoms have been described based on both laboratory and field observations. A common feature of many of these studies, however, is a certain amount of luck, because easily observable sexual events are rarely found. Thus, most of our knowledge of the impact and extent of diatom sexual reproduction that occurs in the field is based on inference (see, e.g., reference 6). Consequently, our understanding of how the sexual cycle is induced, how gametes find one another, and how sexual events influence the genetic diversity of field populations remains quite limited. The approach taken here has been to identify the genes expressed during the early stages of sexual reproduction in the sexually manipulable diatom T. weissflogii (1, 2, 43) with the goals of both understanding the onset of sexual events in more detail and developing molecular markers for monitoring these events in the field.

Sexual reproduction-specific gene expression. Sexual reproduction in diatoms is often triggered when vegetative cells are exposed to unfavorable growth conditions (12), an event that likely induces expression of a wide variety of genes, some of which will be integral to the onset of sexual reproduction and some of which will be part of a more general response to stress. The key step, then, in the identification of genes specifically required for sexual reproduction was to eliminate the stress response genes from analysis. The fact that the ability to undergo sexual reproduction in diatoms is coupled to the obtainment of an appropriate cell size provided an ideal means for distinguishing between the two response types. Within a given species of diatom, large cells tend to "ignore" sexual induction triggers and continue to divide asexually, while small cells tend to respond to the same induction triggers by forming gametes. This implies that a similar suite of stress response genes is expressed whether small or large cells are exposed to an induction trigger. In contrast, sexual reproduction genes should be expressed only when appropriately sized small cells are exposed to the induction trigger. Furthermore, because the small cells of T. weissflogii produce gametes in a relatively synchronous manner once a dark-induced culture is returned to continuous light (2), a snapshot of the pattern of gene expression occurring during the transition to sexual reproduction could be obtained.

Potential involvement of the extracellular matrix in gamete recognition. Despite the relatively small number of genes analyzed thus far, a new gene family, known as Sig, consisting of at least three members strongly up-regulated during the onset of sexual reproduction, has been identified. The three polypeptides predicted to be encoded by these genes display a number of common features: they each contain three or more EGF-like motifs; they each have five highly conserved domains that display strong homology to the EGF-containing domain of the vertebrate extracellular matrix glycoprotein tenascin X; they each possess a signal sequence; and they each appear to lack transmembrane domains. Hundreds of (predominantly animal) proteins that contain the EGF-like, cysteine-rich motif (5, 8) have been identified. These EGF-containing proteins fall into four general categories: growth factors, transmembrane receptors or adhesion molecules, soluble secreted proteins, and extracellular matrix proteins. Because of the similarity of the SIG polypeptides to the extracellular matrix glycoprotein tenascin X and the absence of transmembrane domains, the SIG polypeptides are hypothesized to be components of the extracellular matrix.