The transcriptome of the early life history stages of the California Sea Hare Aplysia californica

Abstract

Aplysia californica is a marine opisthobranch mollusc used as a model organism in neurobiology for cellular analyses of learning and behavior because it possesses a comparatively small number of neurons of large size. The mollusca comprise the second largest animal phylum, yet detailed genetic and genomic information is only recently beginning to accrue. Thus developmental and comparative evolutionary biology as well as biomedical research would benefit from additional information on DNA sequences of Aplysia. Therefore, we have constructed a series of unidirectional cDNA libraries from different life stages of Aplysia. These include whole organisms from the egg, veliger, metamorphic, and juvenile stages as well as adult neural tissue for reference. Individual clones were randomly picked, and high-throughput, single pass sequence analysis was performed to generate 7971 sequences. Of these, there were 5507 quality-filtered ESTs that clustered into 1988 unigenes, which are annotated and deposited into GenBank. A significant number (497) of ESTs did not match existing Aplysia ESTs and are thus potentially novel sequences for Aplysia. GO and KEGG analyses of these novel sequences indicated that a large number were involved in protein binding and translation, consistent with the predominant biosynthetic role in development and the presence of stage-specific protein isoforms.

Introduction

Opisthobranch molluscs of the genus Aplysia, and most especially the California Sea Hare, Aplysia californica, are important model systems in the study of the molecular and cellular basis of neural memory and learning (Kandel, 2001). The utility of this model system derives from a number of factors, most notably the relatively small number and large size of cells in its nervous system, allowing repeatable identification of neurons and direct linkage of these with specific behaviors. Despite the importance of Aplysia californica as a neurological model, genomic and other DNA sequence information has been relatively slow to accumulate for this species. This pace is also surprising given the interesting phylogenetic position of the Aplysia. Sequencing efforts on over 30 bilaterian species in the deuterostomes and ecydosozoa have been undertaken but Aplysia californica will be the first member of the Lophotrochozoa to have its entire genome sequenced (Aplysia Genome Sequencing Project, 2009, Genbank Accession AASC00000000.2 GI:225542573). In advance of the completed annotation of this full genome project, there is other important and interesting DNA sequence information available. The complete mitochondrial genome of A. californica is known (Knudsen et al., 2006), and molecular phylogenies at the order and genus level have been constructed based on mtDNA and other sequences (Medina and Walsh, 2000; Medina et al., 2001; 2005). More recently, two rather comprehensive expressed sequence tag (EST) projects have focused exclusively on the adult neural transcriptome in A. californica (Moroz et al., 2006) and A. kurodai (Lee et al., 2008).

A. californica is a simultaneous hermaphrodite and reproduces by copulation, and the life cycle consists of a fertilized egg, which develops in a capsule in a benthic egg mass, and then hatches into a free-swimming veliger larvae. Veliger larvae metamorphose to a juvenile stage, resembling a miniature adult, which grow and mature to the adult form. This life cycle has been reproduced in captivity (Kriegstein et al., 1974) and thus the model can be bred and manipulated in a number of ways (see Capo et al., 2009, for a recent example). One benefit of this closed life cycle is the ability to obtain cultured individuals from all life stages, and thus there is potential to further employ A. californica as a developmental model. Since there is no published information specifically on the transcriptome of the early developmental stages of this species, we undertook an EST project for these developmental stages in order to possibly identify genes not detected in earlier EST projects and which might be expressed only in the early development stages of these animals. Specifically, we analyzed ESTs from eggs, veliger larvae, metamorphic individuals, juveniles and adults and obtained 5507 quality-filtered ESTs that clustered into 1988 unigenes, which are annotated and deposited into GenBank. In general, these new sequences indicate that the expression of a large number of known genes has likely not yet been archived or reported for A. californica, and this new information likely relates from the fact that different developmental phases and tissues (i.e., non-neural) have been used.

Materials and methods

Animals

All developmental stages of A. californica were obtained from the National Resource for Aplysia at the University of Miami, supported by National Center for Research Resources of the U.S. National Institutes of Health. The developmental phases and the stages within each phase used in this study adhere to the nomenclature of Kriegstein (1977) and were: (1) EGGS consisting of fertilized egg strands collected each day from the date laid through pre-hatching, encompassing all embryonic stages. After RNA extraction (see below), the samples were pooled; (2) VELI consists of free-swimming veliger larvae, stages 1-5 veliger larvae; (3) JMAC consists of metamorphosing larvae, stages 6-8 in transition from veliger to settled juveniles; (4) JUVE is made from pre-sexually mature settled juveniles, stages 9-11; and (5) ACAN is made from neural tissue only from adult animals of a range of ages from young adult through senescence, stages 12 and 13. The neural tissue includes all five major pairs of ganglia: buccal, cerebral, pleural, pedal and abdominal. Movies of each developmental phase can be accessed at http://aplysia.miami.edu/life-cycle.html.

RNA extraction/purification, cDNA synthesis, library construction

Total RNA was extracted with TriZol Reagent (Sigma) from whole animals, except for the adult neural samples that consisted of complete CNS. For EGGS, JUVE and ACAN, mRNA was purified with the PolyATtract Kit (Promega) and used for cDNA synthesis (Just cDNA, Stratagene). For the JMAC and VELI samples, the total RNA was limited and was therefore amplified with the MessageAmp Kit (Ambion) prior to cDNA synthesis. The cDNAs were unidirectionally cloned into the EcoRI and XhoI sites of the pSMART vector (Lucigen, Middleton, WI) and used to electrotransform the highly electrocompetent E. cloni® cells (Lucigen, Middleton, WI) to generate plasmid cDNA libraries with the maximum number of clones and less than 1% non-recombinants. The pSmart-cDNA vector expresses kanamycin-resistance and has a terminator on both sides of the cDNA insertion site preventing expression of cDNA resulting in a stable library. Normalization was not employed so a series of subtractions were performed after each round of sequencing to remove the high copy number clones.

Sequencing and subtraction

A portion of each library was plated with kanamycin and clones were randomly selected for high-throughput (96 well format) sequence analysis. Clones were grown at 37°C overnight in 1 mL Superbroth medium with kanamycin and a portion of each culture was archived as a glycerol stock frozen at −80 °C. Aliquots of the cultures were amplified by PCR using vector primers (Sf forward: GTGAGTTGATTGCAGTCCAGTTACGCT and Sr reverse: GGTCAGGTATGATTTAAATGGTCAGTG) under the following conditions: 0.2 mM dNTPs, 10 pmoles of each primer, 1 unit of Taq (0.2 μL), and reaction buffer (50 mM Tris HCl, pH 9.2 (25 °C), 16 mM (NH₄)₂SO₄, 2.25 mM MgCl₂, 2 % (v/v) DMSO, 0.1 % (v/v) Tween 20). Two-step thermal cycle conditions were used (94 °C for 10 min; then 35 cycles of 94 °C for 30 s followed by 70 °C for 4 min; then 72 °C for 15 min). PCR products were purified with AmPure binding (Agencourt, Beverly MA, USA) and ethanol washes and eluted in distilled water. PCR products were subjected to dideoxy-dye-terminator sequencing using Big Dye chemistry (Applied Biosystems) from the 5′ end of the unidirectional inserts using a vector primer (T7Smart: TAATACGACTCACTATAGGGGAAG). Sequencing reactions were purified by CleanSEQ binding (Agencourt) and ethanol washes, and eluted in 0.5 mM Tris, pH 8.0 and analyzed on an Applied Biosystems 3730 DNA Sequence Analyzer.

After each round of ten 96-well plates of sequence, subtraction was performed to reduce the complexity of the library by removing the previously characterized clones using methods of Oleksiak et al. (2001). This was done on the adult neural and juvenile metamorphic libraries and results in the increase of the relative abundance and identification of rarer clones. 5′ biotin-labeled antisense cDNAs were produced by PCR of previously characterized clones using vector specific primers (T7Smart: TAATACGACTCACTATAGGGGAAG and SR-Biotin: B-GGTCAGGTATGATTTAAATGGTCAGT). The PCR products were hybridized at a 100x molar excess to the DNA from the cDNA libraries in the presence of blocking oligomers that prevented hybridization to vector sequences (T7SmartC: CAGGATCCGAATTCGTCGACGAATATCTTCCCCTATAGTGAGTCGTATTA; SRC: GGATATCACTGACCATTTAAATCATACCTGACC) and oligo dA to block common oligo dT sequences in the antisense cDNAs. After a 24-h hybridization, genes in the library that bound to these biotin-labeled PCR products were removed with the use of magnetized, streptavidin coated beads. The remaining DNA was used to transform competent cells to form the new subtracted libraries.

Sequence analysis and annotation

We define quality ESTs as sequence reads remaining after low quality (phred score < 20) ends were removed using a 5bp sliding window, and vector sequences were removed using cross_match (Ewing and Green, 1998; Ewing et al., 1998). Only quality ESTs of length > 100 bp were retained for further analysis. ESTs from all stages were pooled for clustering into contigs after the definition of Staden (1980), which comprised of both clustered EST and un-clustered singletons, using CAP3 (Huang and Madan, 1999). Both individual ESTs and clustered contigs were subjected to BLAST analysis (NCBI BLAST; Altschul et al., 1997) against protein and sequence databases using ad hoc Perl scripts based on the BioPerl programming interface (Stajich et al., 2002) to run parallel BLAST analysis on a compute cluster using the Sun Grid Engine load management system. To identify matches with previously published EST sequences, individual ESTs were BLASTed against the subset of dbEST (Boguski et al., 1993) that excluded human and mouse ESTs, which was further partitioned by separating Aplysia from non-Aplysia ESTs. The tBLASTx (translated nucleotide against translated nucleotide) program was used to identify mores distantly related ESTs.

An expect-value cutoff of 1e-8 was used as the retention criterion for BLAST hits (see below for discussion of this cutoff). GO and KEGG annotations were done with the annot8r software package (Schmid and Blaxter, 2008). As part of this analysis, individual ESTs were BLASTed against subsets of UniProt (The UniProt Consortium, 2008) databases optimized for identification of Gene Ontology (GO, The Gene Ontology Consortium, 2000) or KEGG pathways (Kanehisa et al., 2008) using the BLASTx (translated nucleotide against protein) program. As suggested in the program documentation, the expect-value cutoff of 1e-8 was used for retention of BLAST matches. The top 10 best hits for each EST were used to assign GO_slim terms and KEGG pathways.

Contigs corresponding to EST clusters and singletons were also used for BLAST analysis against the complete Uniprot (The UniProt Consortium) and PFAM (Finn et al., 2006) protein databases with the BLASTx program. To get a general sense of which EST clusters matched annotated proteins or protein domains (see table 1, figure 1), we used a more permissive expect-value cutoff of 1e-3. The taxonomic ranks Kingdom, Phylum and Class (or nearest equivalent) of the best BLAST hits (figure 1) were identified by recursion with an ad hoc Perl script through the taxonomy database downloaded from NCBI (Wheeler et al., 2000). In order to identify EST clusters that do not correspond to previously published Aplysia EST sets, the contigs were BLASTed against the non-Aplysia subset of dbEST described above, using the more restrictive expect-value cutoff of 1e-8 as recommended in Schmid and Blaxter (2008).

Figure 1.

Taxonomic divisions of Swissprot and Trembl proteins with BLAST alignments to Aplysia californica EST contigs (e <= 10e⁻³). The three layers correspond to kingdom (or nearest equivalent), phylum and class, respectively. For phylum and class, only sections of at least 5% of the total are labeled.

Results and Discussion

A total of five directional cDNA libraries were constructed, containing expressed sequences from all major phases/stages of the A. californica life cycle. Each library consists of between 300,000 and 500,000 clones with an average insert size between approximately 800 and 1000 base pairs. From these libraries a total of 7971 sequence reads were generated from randomly selected clones, and after culling reads of vector only, or of <100bp, or below a phred score of 20, the number of quality ESTs retained was 5507, ranging from 100 to 900 bp in length. (Supplementary Figure 1 indicates the EST length distribution per developmental phase.) The sequence summary and average read length for the five libraries is given in Table 1a. Sequences are available in GenBank under the following accession numbers: EGGS GR213105-GR213928; VELI GR213929-GR214651; JMAC GR215461-GR217669; JUVE GR214652-GR215460; ACAN GR212169-GR213104. Following clustering analysis (Table 1b), there were 1988 total contigs (1438 singletons and 550 clusters). 1444 contigs have at least one significant (e <= 1e-3) BLAST hit to UniProt or PFAM proteins (608 contigs) or publicly available ESTs from NCBI (1276 contigs). 1173 of the EST contigs identified in this study match known Aplysia ESTs and, of phylogenetic relevance, 742 are possibly clade-specific, as they have no other BLAST matches with ESTs or annotated proteins from other species. 497 of the contigs match previously reported ESTs from other species but are novel for Aplysia. Of the 608 contigs matching annotated UniProt or PFAM proteins, 230 are novel in the sense that their best BLAST matches are proteins from other species. The remaining 378 correspond to known Aplysia proteins.

Table 1a.

Length distribution of quality-trimmed reads by developmental phase for ESTs of Aplysia californica

Library	Total sequences	Quality ESTs retained1	Percent of Total	Average length (bp)
Egg	960	824	85.8	405
Veliger	877	723	82.4	520
Metamorphosing	3264	2213	67.8	507
Juvenile	960	810	84.4	457
Adult	1910	937	49.1	472

¹EST reads of length > 100bp and phred > 20

Table 1b.

Contig statistics including length distribution and annotation information for singleton and non-singleton clusters

Contigs1	Total	Length (bp)			Average ESTs/contig
		Min	Max	Average
>1 EST	550	104	2110	659	7.59
Singletons	1438	101	849	455	1
Annotated2	608	101	2110	601	4.97
Aplysia ESTs3	1193	103	1554	488	2.18
Other ESTs3	497	143	787	409	1.86
Unmatched	617	101	1429	456	1.51

¹Using Staden (1980) definition of contig, which includes singletons. Only contigs of length > 100 bp were used in the analysis.

²“Annotated” refers to contigs matching either Uniprot (Swissprot and TrEMBL) proteins or PFAM proteins with an e-value <= 1e-3

³BLAST matches (e < 1e-3) to NCBI’s “est_other” database, with Aplysia californica ESTs filtered into a separate category.

Annotation of the top 15 largest clusters (containing the most ESTs) from the normalized libraries by developmental stage (Table 2) reveals expected abundances in genes associated with energy transfer, ribosomal function and other housekeeping genes. (A complete list of annotations available for all ESTs is available in Supplementary Table 1.) Of greater interest however is the annotation of the top 15 largest clusters (containing the most ESTs) that have no publicly available Aplysia EST match but have significant similarity to known proteins in UniProt or PFAM databases (Table 3). (A complete list is available in Supplementary Table 2.) The majority of these abundant ESTs have not been observed in previous Aplysia EST libraries (two matched prior Aplysia sequences where whole coding regions of genes were obtained). In general, these new sequences indicate that the expression of a large number of known genes has likely not yet been archived or reported for A. californica, and this new information likely relates from the fact that different developmental phases and tissues (i.e., non-neural) have been used. Notably, at least in this top 15 list (Table 3), there are also many genes commonly seen as abundant in EST projects (e.g., housekeeping genes such as elongation factor, actin, tubulin, ribosomal proteins, and components of mitochondrial energy pathways), suggesting that Aplysia may have isoforms of these common genes specific to particular developmental phases.

Table 2.

The 15 most abundant transcripts by developmental stage. Only transcripts matching previously reported Aplysia californica ESTs are shown. Abundance is measured as the number of EST reads/contig. The best BLAST hit (E-value cutoff is <= 1e-8), where applicable, is shown for each transcript. Precedence for the best BLAST hits is Swissprot->Trembl->PFAM->SMART. A complete list of transcripts is shown in supplementary table 1

Transcript	Egg	Veli.	Meta.	Juve.	Adult	Total	Database	ID	Description
229	1	0	533	0	0	534
27	34	41	212	72	2	361	swissprot	P48893	ATP synthase subunit a - Albinaria coerulea (Land snail)
355	3	1	59	6	2	71	swissprot	P48884	Cytochrome b - Albinaria coerulea (Land snail)
249	0	1	43	0	0	44	trembl	Q6Q0B0	NADH-ubiquinone oxidoreductase chain 3 - Aplysia californica (California sea hare)
117	1	0	2	24	13	40	swissprot	O57592	60S ribosomal protein L7a - (Japanese pufferfish) (Takifugu rubripes)
364	2	1	31	1	2	37	swissprot	P48897	NADH-ubiquinone oxidoreductase chain 1 - Albinaria coerulea (Land snail)
386	2	1	3	23	8	37	swissprot	P55935	40S ribosomal protein S9 - Drosophila melanogaster (Fruit fly)
326	2	8	24	0	0	34	swissprot	Q9YGM8	Caveolin-1 - Fugu rubripes (Japanese pufferfish) (Takifugu rubripes)
294	0	4	26	0	0	30	swissprot	Q2IBG8	Caveolin-1 - Eulemur macaco macaco (Black lemur)
212	0	1	28	0	0	29	swissprot	P55769	NHP2-like protein 1 - Homo sapiens (Human)
245	0	0	27	0	0	27	trembl	Q26424	C-myc proto-oncogene homolog protein - Crassostrea virginica (Eastern oyster)
227	0	1	24	1	0	26	swissprot	O21001	NADH-ubiquinone oxidoreductase chain 2 - Branchiostoma lanceolatum (Common lancelet) (Amphioxus)
418	0	3	13	4	1	21	swissprot	P48889	Cytochrome c oxidase subunit 2 - Albinaria coerulea (Land snail)
154	0	2	16	3	0	21	swissprot	O96647	60S ribosomal protein L10 - Bombyx mandarina (Wild silk moth) (Wild silkworm)
382	1	0	16	0	3	20	swissprot	O46543	Ubiquitin - Ovis aries (Sheep)
393	0	1	17	2	0	20	swissprot	Q9YGQ1	Elongation factor 1-beta - Gallus gallus (Chicken)

Table 3.

The 15 most abundant novel transcripts that match no previously reported Aplysia californica ESTs. Abundance is measured as the number of EST reads/contig. The best BLAST hit (E-value cutoff is <= 1e-8), where applicable, is shown for each transcript. Precedence for the best BLAST hits is Swissprot->Trembl->PFAM->SMART. A complete list of novel transcripts is in supplementary table 2

Transcript	Egg	Veli.	Meta.	Juve.	Adult	Total	Database	ID	Description
88	33	10	83	74	56	256
75	27	26	79	25	43	200	trembl	A7S783	Predicted protein - Nematostella vectensis (Starlet sea anemone)
74	20	8	18	9	4	59	swissprot	P48887	Cytochrome c oxidase subunit 1 - Albinaria coerulea (Land snail)
65	0	0	5	0	51	56	swissprot	P01362	ELH precursor [Contains: Beta-bag cell peptide - Aplysia californica (California sea hare)
85	16	3	4	4	16	43	swissprot	Q964E0	Actin, cytoplasmic - Biomphalaria tenagophila (Bloodfluke planorb)
91	1	4	2	3	23	33	swissprot	P42577	Soma ferritin - Lymnaea stagnalis (Great pond snail)
459	0	3	12	12	5	32	swissprot	P48891	Cytochrome c oxidase subunit 3 - Albinaria coerulea (Land snail)
324	5	0	21	2	1	29	swissprot	P05214	Tubulin alpha-3 chain - Mus musculus (Mouse)
313	13	3	2	3	7	28	swissprot	Q92005	Elongation factor 1-alpha - Danio rerio (Zebrafish) (Brachydanio rerio)
56	1	4	11	5	4	25	swissprot	P46777	60S ribosomal protein L5 - Homo sapiens (Human)
61	6	3	5	3	5	22	swissprot	O42248	Guanine nucleotide-binding protein subunit beta-2-like 1 - Danio rerio (Zebrafish)
52	0	0	4	0	17	21	swissprot	Q10998	Cerebral peptide 1 precursor - Aplysia californica (California sea hare)
388	13	0	7	0	0	20	swissprot	P68371	Tubulin beta-2C chain - Homo sapiens (Human)
449	4	0	0	14	2	20	pfam	pfam08208	pfam08208, RNA_polI_A34, Yeast RNA polymerase I subunit RPA34.5. This is a family of yeast proteins. Subunit A34.5 of RNA polymerase I is a non-essential subunit which is thought to help Pol I overcome topological constraints imposed on ribosomal DNA during the process of transcription..

The taxonomic distribution of positive matches to protein databases appears in Figure 1. While there will be a certain degree of bias in this distribution directly related to the number of available annotated sequences, it is interesting that 60% of the matches are to genes from non-mammalian organisms. While the matches to molluscs are currently 14% of the total, this number is likely to increase as more sequences are added and annotated. (Note in regard to molluscan-specific sequences that the data in Supplementary Tables 1 and 2 are further filtered to only molluscan species in Supplementary Tables 3 and 4.)

A gene ontology (GO) analysis of the annotated ESTs was performed. A breakdown of these distributions across various developmental phases appears in Table 4. Among the cellular components, the nuclear component is highest in eggs and veliger. In terms of molecular function, the protein and nucleotide binding classifications are highest for the eggs. In general, the categories shift in abundances, albeit in complicated patterns, again indicative of shifting requirements and the likely existence of isoforms that are specific to developmental stages.

Table 4.

Five most abundant Gene Ontology (GO) terms by developmental stage. Numbers are the percentage of individual, unclustered ESTs that matched the indicated GO term for each stage in each of the three broad categories. GO terms were assigned using the ten best matching Uniprot proteins identified by BLAST analysis of EST sequences (e <= 1e-8) using annot8r (Schmid and Blaxter, 2008)

	GO term	Egg	Veliger	Metamorph.	Juvenile	Adult
Biological Process	Translation (GO:0006412)	6.53	6.62	8.62	15.72	11.28
	Transport (GO:0006810)	3.03	2.27	3.46	5.12	1.78
	Translational elongation (GO:0006414)	4.78	3.59	5.23	5.83	6.06
	Metabolic process (GO:0008152)	1.11	0.38	0.15	2.65	0.83
	Transcription (GO:0006350)	1.59	0.19	0.29	0.53	1.07
Molecular  Function	Protein binding (GO:0005515)	13.55	14.54	12.54	10.45	9.6
	Structural constituent of ribosome (GO:0003735)	3.24	11.35	10.9	11.69	9.83
	RNA binding (GO:0003723)	4.12	8.51	9.38	6.19	5.76
	Nucleotide binding (GO:0000166)	8.69	3.19	2.46	2.61	5.42
	Metal ion binding (GO:0046872)	2.06	5.67	6.33	6.19	4.75
Cellular Component	Cytoplasm (GO:0005737)	13.6	8.03	4.35	7.92	12.12
	Nucleus (GO:0005634)	6.44	5.35	4.43	2.64	3.52
	Ribonucleoprotein complex (GO:0030529)	5.9	8.03	9.16	12.22	11.27
	Intracellular (GO:0005622)	5.01	7.54	6.98	11.94	10.55
	Ribosome (GO:0005840)	4.29	7.79	7.13	12.36	11.15

A total of 1394 EST sequences have significant similarity to KEGG orthologs (e <= 1e-8) that are assigned to a total 168 KEGG pathways. The top 20 most abundant KEGG pathways represented by our EST project are presented in Table 5 categorized by developmental phase. Some of the pathways that appear to be over represented in the earlier developmental phases are the signaling pathways (GnRH and MAPK), with the most notable being insulin signaling, melanogenesis and calcium signaling in the metamorphic stage. Extensive differences in gene expression profiles might be expected in this library due to the widespread changes in morphology and life style occurring during metamorphosis when pelagic, planktivorous, veliger larvae become benthic, herbivorous slugs. This developmental transformation is accompanied by rapid growth, increased pigmentation, alterations in the morphology of the shell, and changes in the central nervous system (Heyland and Moroz, 2006). Also notable is the emphasis on oxidative phosphorylation in the earlier life history stages (Table 5). This observation might be due to inclusion of all tissues from these stages (as opposed to only neural tissue from the adults) and would represent, for example, locomotory tissue genes.

Table 5.

20 most abundant KEGG pathways represented in EST clusters. Numbers indicate the percentage of individual, unclustered ESTs that matched the indicated KEGG pathway components for each stage. KEGG pathways were assigned based on the ten best matching Uniprot protein identified by BLAST analysis of EST sequences (e <= 1e-8) using annot8r (Schmid and Blaxter, 2008)

Pathway	Egg	Veliger	Metamorph.	Juvenile	Adult
Ribosome	5.37	12.70	10.58	21.95	16.80
Oxidative phosphorylation	8.35	19.68	24.18	19.32	5.99
Starch and sucrose metabolism	2.68	1.59	0.82	3.19	1.04
Regulation of actin cytoskeleton	4.47	1.27	1.32	2.06	4.04
Purine metabolism	3.28	1.27	0.82	1.31	2.47
Focal adhesion	3.13	2.54	2.27	1.69	2.60
Gap junction	4.17	2.22	4.72	1.13	2.21
GnRH signaling pathway	0.75	2.22	3.21	0.19	0.26
MAPK signaling pathway	1.04	2.22	2.77	0.94	1.04
Glycolysis / Gluconeogenesis	0.60	1.90	0.25	0.38	1.04
Antigen processing and presentation	2.24	0.95	0.31	1.13	2.34
Vibrio cholerae infection	2.38	2.22	2.27	2.25	2.60
Insulin signaling pathway	0.89	2.54	2.77	0.38	0.26
Melanogenesis	0.60	1.90	2.77	0.00	0.91
Wnt signaling pathway	0.60	1.90	2.33	0.00	1.04
Ubiquitin mediated proteolysis	0.89	0.00	0.06	1.13	0.26
PPAR signaling pathway	0.45	0.32	0.19	0.75	0.78
Calcium signaling pathway	0.60	2.22	2.58	0.00	0.65
Alanine and aspartate metabolism	1.34	1.27	0.44	3.00	5.34
Tight junction	2.68	0.32	0.19	2.25	2.60

The gap junction pathway members are more highly represented in the embryonic and metamorphic phases, although no definitive gap junction ESTs were identified. Gap junctions play a role in early development and in tissue differentiation in mammals and may play a similar role in Aplysia development. The Wnt signalling pathway is over represented during Aplysia metamorphosis. Wnt signalling plays a role in establishment of the anterior-posterior axis and posterior growth in all bilaterian organisms (Martin and Kimelman, 2009). Furthermore, Wnt signalling has been shown to be important in myogenesis in C. elegans (Amin et al., 2009) and may play a similar role in Aplysia metamorphosis from a free swimming veliger to a crawling juvenile animal.

In summary, the current EST sequencing and characterization study has increased the amount of transcriptome data for the early life phases of Aplysia californica. This information will facilitate genome annotation (Haas et al., 2002) and a variety of post-genomic studies, e.g., RNA interference and microarray analyses. Furthermore, these publicly available EST sequences will be essential to link the genome with the proteome. Furthermore, understanding ontogenic gene expression changes may be much more critical for animals undergoing metamorphosis such as Aplysia than for directly developing animals. Perhaps after examination of a number of these metamorphic groups at the transcriptomic level are we likely to be able to understand how these complex life cycle changes can be regulated and managed. These developmental patterns will be especially critical to genome annotation if entire groups of ESTs are only seen in pre-juvenile stages.