A large-scale analysis of mRNAs expressed by primary mesenchyme cells of the sea urchin embryo

The primary mesenchyme cells (PMCs) of the sea urchin embryo have been a powerful experimental system for the analysis of morphogenesis at the cellular level. The optical transparency of the sea urchin embryo and the ease with which PMCs can be isolated and manipulated both in vivo and in vitro have led to a detailed understanding of PMC behavior during gastrulation and later embryogenesis (reviewed by Gustafson and Wolpert, 1967; Okazaki, 1975a; Solursh, 1986; Ettensohn et al., 1997; Ettensohn, 1999).

The PMCs are the sole descendants of the large micromeres, four blastomeres that form near the vegetal pole of the 32-cell stage embryo. The progeny of the large micromeres become incorporated into the epithelial wall of the blastula near the center of the vegetal plate. At the beginning of gastrulation, these cells undergo a conversion from an epithelial to a mesenchymal phenotype. They become motile and ingress into the blastocoel, migrating on the inner surface of the gastrula wall by means of numerous filopodia (Gustafson and Wolpert, 1967; Malinda et al., 1995; Miller et al., 1995). PMC filopodia interact with a complex mixture of extracellular matrix (ECM) molecules that form a thin basal lamina lining the blastocoel cavity. The PMCs gradually accumulate in a characteristic ring-like pattern near the equator of the embryo, guided by substrate-associated cues that arise progressively during the blastula and gastrula stages. As the PMCs migrate, their filopodia fuse, forming long cables that link the cells in a syncytial network (Okazaki, 1965; Hodor and Ettensohn, 1998). Within these filopodial cables, the PMCs secrete the crystalline rods (spicules) that constitute the elaborate larval skeleton (Decker and Lennarz, 1988; Wilt, 1999). These cellular events have been described in considerable detail. Indeed, there is (arguably) a more complete understanding of the morphogenetic behavior of PMCs than that of any other population of embryonic cells.

An elucidation of molecular mechanisms that underlie PMC morphogenesis has lagged behind our understanding of PMC behavior at the cellular level. Recent studies have pointed to molecular changes that accompany ingression (Miller and McClay, 1997; Hertzler and McClay, 1999) and a PMC substrate molecule has been cloned – the proteoglycan core protein-like molecule, ECM3 (Hodor et al., 2000). Two other ECM molecules, pamlin and ECM18, have been identified that may also play a role in PMC migration (Katow, 1995; Berg et al., 1996). Approximately 15 gene products expressed specifically by PMCs, or enriched in these cells, have been cloned. These include four spicule matrix proteins, SM50 (Benson et al., 1987), SM30 (George et al., 1991), PM27 (Harkey et al., 1995) and SM37 (Lee et al., 1999a); the cytoskeletal proteins α-spectrin (Wessel and Chen, 1993), profilin (Smith et al., 1994) and actin CyIIa (Cox et al., 1986); the cell surface protein MSP130 (Leaf et al., 1987); an ETS-family transcription factor (Kurokawa et al., 1999), several collagens (Angerer et al., 1988; Wessel et al., 1991; Suzuki et al., 1997); and lamin B (Holy et al., 1995). PMCs also express at least two β-integrins (Marsden and Burke, 1997; Marsden and Burke, 1998). In some cases, the functions of these molecules have been partly defined. For example, spicule matrix proteins are integral components of the skeletal rods and play an important role in regulating the process of biomineralization (Wilt, 1999). Secretion of collagen by the PMCs appears to provide a necessary microenvironment for skeletogenesis (Blankenship and Benson, 1984; Wessel et al., 1991), perhaps by regulating the presentation of growth factors that control the expression of specific spicule matrix protein genes (see Ettensohn et al., 1997).

To gain a more detailed understanding of the molecular basis of PMC morphogenesis, we have carried out a large-scale analysis of mRNAs expressed by these cells during gastrulation. We took advantage of the fact that PMC precursors, the micromeres of the 16-cell stage embryo, can be isolated in large quantities and cultured in vitro under conditions that allow the cells to undergo a normal program of differentiation (Okazaki, 1975b; Harkey and Whiteley, 1983). This analysis has led to the identification of candidate regulators of cell migration, cell fusion and skeletogenesis, and to the cloning of key components of signaling pathways that have been shown to function in a variety of contexts in the early embryo.

Adult Strongylocentrotus purpuratus were purchased from Marinus (Long Beach, CA). Gametes were obtained by intracoelomic injection of 0.5 M KCl. Micromeres were isolated and cultured according to a protocol provided by Steve Benson (personal communication). Briefly, eggs were fertilized in 10 mM para-aminobenzoic acid and rinsed twice with fresh artificial seawater (SW). The fertilized eggs were cultured at 15°C. At the four-cell stage, the seawater was replaced with Ca²⁺-free seawater (CF-SW). At the 16-cell stage, fertilization membranes were removed by passing the embryos through 53 μm Nitex mesh. The embryos were then rinsed several times with Ca²⁺/Mg²⁺-free seawater at 4°C (50× the packed embryo volume/rinse) and suspended in CF-SW (15× the packed embryo volume). The embryos were dissociated by pipetting using a 9 inch (23 cm) Pasteur pipette. The dissociated cells were loaded on a 3-30% sucrose gradient at 4°C and separated at 1 g for 40 minutes. The micromeres, which formed a clear band one-quarter to one-half inch (5-12 mm) below the top of the gradient, were drawn off and plated at a density of 2×10⁴ cells/cm² on 100 mm tissue culture dishes. After the cells attached to the plates they were rinsed three times with sterile SW and cultured in sterile SW supplemented with 2% horse serum and 1× penicillin-streptomycin-glutamine (Gibco Life Technologies). The cells were cultured at 15°C, until sibling embryos reached the mid-gastrula stage, when they were collected for RNA isolation (below). About 70% of the cultured cells were positive when immunostained with monoclonal antibody 6e10, which recognizes the PMC-specific cell surface glycoprotein MSP130. The remaining cells were presumably large micromeres that did not differentiate into PMCs, small micromere derivatives and cells derived from contaminating mesomeres and macromeres.

Total RNA was extracted from cultured cells using Trizol reagent (Gibco Life Technologies). Poly(A)+ RNA was isolated using a MicroPoly(A)Pure kit (Ambion). cDNA was synthesized using an oligo(dT) primer and cloned directionally into the pSPORT plasmid vector following the manufacturer’s instructions (Gibco Life Technologies). The average insert size was 1.5-2.0 kb. The library was arrayed in 384-well plates using a Genetix Q-Bot robot.

Plasmid template DNA was prepared from individual clones isolated from wells of the 384-well plates. Cells were grown overnight in 400 μl of LB medium and lysed under alkaline conditions (Birnboim and Doly, 1979). Lysates were cleared using Millipore lysate clearing plates (catalog code, MANANLY50) and DNA was purified using Millipore multiscreen glass fiber filter plates (catalog code, 52EM108M8). Most steps in the plasmid isolation procedure were carried out robotically using Beckman Multimek 96 pipetting robots and automated filtration stations.

A single sequencing reaction was carried out with each template using dideoxy chain terminators and a T7 primer, which provided sequence from the 5′ ends of the directionally cloned cDNAs. Sequencing reactions were resolved using ABI 3700 DNA analyzers. The average length of readable sequence was 600-800 nucleotides. Several clones of interest were subsequently sequenced fully on both DNA strands using an ABI 377 sequencer at the University of Pittsburgh School of Medicine DNA Sequencing Facility.

DNA sequences were loaded into an Oracle database and subjected to quality control using Phred (Ewing and Green, 1998; Ewing et al., 1998). Sequences were trimmed to remove contaminating vector (pSport), Escherichia coli and S. purpuratus repetitive sequences. Sequences that included >200 bp of Q20 bases were considered of high quality and subjected to BLASTX searching against the public non-redundant protein databases in GenBank.

Whole-mount in situ hybridization was performed as described previously (Guss and Ettensohn, 1997), with minor modifications. Tween-20 (0.1%) was included in all wash solutions to prevent embryos from adhering to the walls of microfuge tubes. After hybridization with digoxigenin-labeled probes, embryos were washed with 0.1 × SSC (rather than 1 × SSC) to reduce nonspecific staining.

A total of 8293 high-quality sequences were generated and subjected to BLASTX analysis. The initial data for each clone consisted of a single sequencing reaction primed at the 5′-most end of the cDNA, although a number of clones were later sequenced fully (see below). All 8293 DNA sequences can be accessed through GenBank (Accession Numbers BG780044-BG789442) or through the sea urchin genome project (http://sea-urchin.caltech.edu:8000/genome/databases).

BLASTX analysis showed that of the 8293 ESTs, 1629 were strong matches (P<10⁻⁷) to previously identified proteins (Table 1). The frequency of matches was therefore 1629/8293 (about 0.20). Further analysis of the 1629 matches showed that they represented 771 distinct proteins. Of the 771 proteins, 116 were matches to ESTs (mostly from Caenorhabditis elegans and Homo sapiens), while the remaining 655 were matches to proteins that have been characterized to varying extents. In some cases, matches were to proteins that have been characterized only with respect to their pattern of expression, while other matches were to proteins with well-defined biochemical and cellular functions. A complete list of the 655 proteins, grouped according to major cellular function, is shown in the Appendix. The great majority of proteins were identified only once in our analysis, although a few highly abundant proteins were identified many times (Table 2). The average number of hits/protein was 2.1 (1629/771).

About 80% of the ESTs fell into the ‘no match’ category (6664/8293 sequences). By examining a random sample of 315 sequences in this category, we estimated that a very small fraction (3%, or ∼200 total sequences) could be accounted for by poor sequence data (i.e. sequences with >10% unreadable bases) that were not eliminated in the initial sequence screening. Approximately 6% of the cases in the ‘no match’ category (∼400 total sequences) contained no insert or one shorter than ∼50 nucleotides when analyzed by BLASTN and therefore represented an artifact of library construction. A similar proportion of the ‘no match’ cases (6.8%, or ∼450 total sequences) were rRNA sequences. The most commonly identified sequence in this class was mitochondrial 16S rRNA, a super-abundant polyadenylated transcript (see also Davidson, 1986; Poustka et al., 1999). In addition, approximately 2% of the ESTs (∼200 total) represented sequences similar to untranslatable, interspersed repetitive sequences that have been identified in the egg (Costantini et al., 1978). Taken together, these findings indicate that a relatively small fraction, about 18% of the ‘no match’ category, can be accounted for by these classes of transcripts. The remaining 82% are therefore likely to represent other untranslated sequences and bona fide proteins that did not match entries in GenBank.

We arrived at a conservative estimate of the number of protein-coding sequences in the ‘no match’ category by comparing the distribution of maximum open reading frame (ORF) lengths of sequences in this category to the distribution of maximum ORF lengths in untranslated sequences (see also Lee et al., 1999b). Because the PMC library was oligo(dT)-primed, we assumed that most clones in the ‘no match’ category represented 3′ UTR sequences. We analyzed all S. purpuratus genes in GenBank, as of 7/11/2000, for which 3′-UTR sequences >600 nucleotides were available (39 genes). We divided these 3′-UTRs into 120 non-overlapping segments with an average length of 800 nucleotides, a value chosen to match the average read length of a random sampling of ‘no match’ sequences (described below). The longest ORF in each fragment was determined and the distribution of these values is plotted as a histogram in Fig. 1A. Most of the ORFs were quite short and in no case did we find an ORF greater than 350 nucleotides in length. For comparison, the maximum ORF lengths of a sample of 120 ESTs that yielded strong BLAST matches (P<10⁻⁷) were also plotted (Fig. 1A). The average length was much greater and the distribution only slightly overlapped that of the 3′-UTR ORFs.

We then chose 120 ESTs at random from the ‘no-match’ category, excluding those with no inserts, low-quality sequences and rRNA sequences. The average read length of these sequences was 796 nucleotides. As above, we determined the length of the longest ORF in each sequence and plotted these as a histogram (Fig. 1B). Because of the directional cloning strategy used to construct the library, the great majority (>95%) of strong BLAST matches to known proteins were in one orientation (reading frames +1, +2 and +3) and the same must be true of cryptic protein-coding sequences in the ‘no match’ population. We therefore restricted our analysis of ORFs in the ‘no match’ sequences to the three positive reading frames. As expected, most sequences in this category contained only short ORFs (100-200 nucleotides) but longer ORFs were also apparent in the population. 20/120 clones (16.7%) contained an ORF equal to or longer than 350 nucleotides.

We used the 16.7% value as an estimate of the fraction of ESTs in the ‘no match’ category that represented bona fide coding sequences. This is likely to be a conservative estimate. As shown in Fig. 1A, a small but significant fraction (22.5%) of protein-coding ESTs (i.e. those with strong BLAST matches) had maximum ORF lengths of <350 nucleotides. Undoubtedly, some cryptic protein-coding sequences also had maximum ORF lengths shorter than 350 nucleotides. Nevertheless, by using the 16.7% value, and after eliminating from consideration poor sequences, clones with very short inserts, rRNA sequences, and sequences similar to untranslatable, repetitive mRNA (18% of the sequences in the ‘no match’ category, as described above), we estimate that 913 sequences in the ‘no match’ population represent bona fide protein-coding sequences. If these cryptic proteins exhibit, on average, the same prevalence distribution as proteins identified by BLAST matches, then 913 sequences would represent 435 distinct proteins (2.1 hits/protein). By adding this value to the 771 distinct proteins already identified by BLAST analysis, we conservatively estimate that the EST database contains sequences corresponding to ∼1200 different proteins.

The EST analysis identified a number of cDNA clones that encoded especially strong candidates for regulators of PMC morphogenesis. As a first step in the further characterization of such gene products the complete sequences of several clones were determined.

Clone 03-0233 contained a large (>4 kb) insert that encoded 1079 amino acids of a molecule with significant similarity to vertebrate fibronectin. The closest BLAST match was to bovine fibronectin (P=3×10⁻⁴¹). The ORF encoded a large C-terminal portion of the sea urchin protein which consisted of nine tandem fibronectin Type III repeats followed by a 200-300 amino acid region at the extreme C terminus that was not similar to known proteins at a significant level. An RGDT sequence was identified within the sixth Type III repeat. This tetrapeptide has a cell-binding function in vertebrates (Pierschbacher and Ruoslahti, 1984).

Clone 0016_B2_H02 contained the entire ORF of a protein related to fibrinogen. This protein was 308 amino acids in length and consisted of an N-terminal signal sequence, a short (∼70 amino acid) segment without significant similarity to known proteins, and a C-terminal fibrinogen-related domain (FRD). This globular domain is found at the C terminus of a variety of extracellular proteins in both vertebrates and invertebrates, including fibrinogens β and γ, angiopoietins, tenascin, ficolins, the product of the Drosophila scabrous gene, and some lectins (see Xu and Doolittle, 1990; Conklin et al., 1999; Gokudan et al., 1999).

Clone 0026_B2_F03 contained a single long ORF encoding more than 1200 amino acids with significant similarity to vertebrate and invertebrate nidogen/entactin. The closest match by BLAST analysis was to human nidogen (P=3×10⁻⁶³). Based on comparison with the human sequence, the sea urchin clone appears to lack only the N-terminal-most 10-20 amino acids of this protein.

Clone 0016_B2_D04 encoded a full-length protein with significant similarity to vertebrate and invertebrate osteonectin (SPARC, BM-40). The closest BLAST match was to osteonectin/SPARC from Caenorhabditis elegans (P=4×10⁻³⁶). Sea urchin osteonectin is 270 amino acids in length, a size similar to that of osteonectins from other organisms (250-310 amino acids). Like other osteonectins, the sea urchin protein has a putative signal sequence, is acidic (calculated pI=4.35) and relatively rich in cysteines.

Clone PM990802-08-0472 encoded a full-length homolog of the small GTPase Rac. The closest BLAST match was to human Rac1 (4×10⁻⁹⁴). Sea urchin Rac is 194 amino acids in length, two amino acids shorter than human Rac1. The N-terminal two-thirds of sea urchin Rac and human Rac1 are identical at 118/121 positions, and the two proteins show ∼90% amino acid identity overall.

Clone PM990802-06-0460 encoded a full-length protein highly similar to human tetraspanin NET-5 (P=3×10⁻³⁹). The sea urchin protein exhibited the characteristic organization of tetraspanins; three putative transmembrane domains near the N terminus and a fourth near the C terminus, with a large extracellular loop between the third and fourth transmembrane domains. Both the human tetraspanin NET-5 and its sea urchin counterpart are 239 amino acids in length.

Clone 0016_A2_H04 encoded a full-length protein most similar to human tetraspanin NET-7 (P=5×10⁻³⁵). The sea urchin protein is 243 amino acids in length, five amino acids longer than human tetraspanin NET-7. It contained the distinctive spacing and number of transmembrane domains described above.

Clone PM990802-03-0379 encoded a C-terminal fragment, 520 amino acids in length, of a protein highly similar to DOCK180/Myoblast city. The closest BLAST match was to human DOCK180 (P=4×10⁻⁵⁴).

Clone PM990802-08-0413 encoded the N-terminal two-thirds (∼650 amino acids) of a sea urchin homolog of a discoidin-domain receptor tyrosine kinase. The closest BLAST match was to human DDR1 (TrkE) (P=6×10⁻⁷⁵). The sea urchin protein has the characteristic domain organization of this class of receptor tyrosine kinase: an N-terminal signal sequence and discoidin domain, a central transmembrane domain, and a cytoplasmic protein tyrosine kinase domain.

Clone 0022_B2_H02 encoded the C-terminal 226 amino acids of a previously unidentified spicule matrix protein. The C-terminal 90 amino acids of the protein were organized in 28 tandem copies of a distinctive repeat element of the form P-X-Y, where X is N, F or T (usually N), and Y is Q, N, T, A or R (usually Q). The presence of tandem copies of a proline- and/or glycine-rich repeat is a common feature of many spicule matrix proteins, although the primary sequence of the repeat and the copy number varies between these proteins (Katoh-Fukui et al., 1991; Livingston et al., 1991; Harkey et al., 1995; Lee et al., 1999a). The remainder of the SM50-related sequence showed a high degree of similarity to the N-terminal region of SM50 (P=1×10⁻⁴¹), which contains a C-lectin-like domain (see Harkey et al., 1995). The two proteins were 60% identical over this 140-amino acid region. The SM50-related protein therefore exhibited the distinctive two-domain structure characteristic of other spicule matrix proteins. The alignment of the SM50-related protein with the N-terminal region of SM50 suggested that the first ∼60 amino acids are missing from the SM50-related clone.

Clone 0014_A1_A09 contained the complete ORF of a small (186 amino acid) protein similar to C-type lectins from several organisms. The closest BLAST match (3×10⁻²⁰) was to echinoidin, a C-type lectin identified in Anthocidaris crassispina (Giga et al., 1987), but the degree of amino acid identity was sufficiently low (34%) to make it doubtful these proteins are homologues. Moreover, another C-lectin similar to echinoidin has been identified in S. purpuratus (Smith et al., 1996) and is clearly distinct from the protein identified here. While the C-lectin we identified lacks the obvious repeat elements of other spicule matrix proteins, it exhibits several features that suggest it may belong to this class of proteins: (1) it includes an N-terminal signal sequence and is presumably secreted; (2) it includes a C-lectin domain, as do the previously identified spicule matrix proteins (Harkey et al., 1995; Killian and Wilt, 1996); and (3) it is expressed at high levels specifically by PMCs, as shown by whole-mount in situ hybridization (Fig. 2).

Clone 0025_B2_A08 encoded a large C-terminal region (609 amino acids) of a protein closely related to, but distinct from, the PMC-specific cell surface glycoprotein MSP130 (P=4×10⁻⁴⁸). The MSP130-related 1 sequence aligned over its entire length with S. purpuratus MSP130 at an overall amino acid identity level of 35-40%. The alignment included all regions of the MSP130 protein, except the N-terminal-most ∼90 amino acids (the corresponding amino acids are missing from MSP130-related 1, which is a partial clone) and amino acids 226-378, which correspond to the second glycine-rich domain of MSP130 (Parr et al., 1990). This domain is absent from the MSP130-related 1 protein. Like MSP130, MSP130-related 1 includes 14-16 hydrophobic amino acids at the extreme C terminus that may function as a GPI-anchor domain (Parr et al., 1990). We also identified several cDNAs encoded by a second gene, MSP130-related 2, that is clearly distinct from both MSP130 and MSP130-related 1. MSP130 and the MSP130-related proteins therefore represent a small gene family consisting of at least three members.

The expression patterns of 20 mRNAs were examined by in situ hybridization. Of these, we found that eight were expressed exclusively or predominantly by PMCs (Fig. 2). Included in this group were two transcription factors (ERG and aristaless), two extracellular matrix proteins (fibronectin and fibrinogen-related protein), two cell surface proteins (MSP130-related 1 and NET-7), and two new putative spicule matrix proteins (SM50-related and C-lectin). Probes against the other 12 mRNAs showed more general labeling patterns consistent with expression in PMCs, as well as other cell types in the embryo.

The results of two other sea urchin cDNA sequencing projects, each considerably smaller in scale than the present study, have recently been reported (Lee et al., 1999b; Poustka et al., 1999). Following a strategy essentially similar to that described here, Lee et al. (Lee et al., 1999b) examined 956 ESTs from an arrayed cDNA library generated using S. purpuratus cleavage-stage poly(A)+ RNA. Using criteria similar to ours, they identified 232 ESTs with significant matches to known protein-coding sequences in GenBank. These 232 ESTs were found to represent 153 different proteins. The average number of hits/protein was therefore about 1.5, less than the value we obtained (2.1 hits/protein). This difference is undoubtedly partly due to the larger sample size of our study, as the probability that a given EST will match a previously identified protein increases with the sample size. Another likely contributing factor is the relatively lower diversity of the pool of mRNAs expressed by a specific cell type (in this case, PMCs) from a later developmental stage (Davidson, 1986).

One significant difference between the study by Lee and co-workers and the present work was the method used to prime cDNA synthesis (oligo(dT) versus random priming). Oligo(dT) priming undoubtedly led to a relatively greater representation of 3′UTR sequences in our analysis. Nevertheless, the overall frequency of matches (fraction of total sequences that were significantly similar to previously identified protein coding sequences) was only slightly lower in our study (0.20 versus 0.24). One reason the difference may not have been greater is that the level of rRNA contamination was lower in the PMC library (6-7% versus 14%), probably due at least in part to the method of priming. We may also have had relatively less contribution from untranslatable, interspersed-repeat-containing poly(A)+ RNA. These sequences are on average at least five times as long as mRNAs and would tend to be more highly represented in cDNA libraries generated by random priming than those produced by oligo(dT) priming (Davidson, 1986; Lee et al., 1999b). A minor, but useful, feature of the oligo(dT) priming strategy and 5′ orientation of sequencing was that when N termini of proteins were identified by BLAST analysis; such clones nearly always contained the complete coding sequences of the corresponding proteins.

Poustka et al. (Poustka et al., 1999) used oligonucleotide fingerprinting to generate a normalized cDNA collection representing about one-third of all genes expressed in the fertilized egg of S. purpuratus. Starting with an oligo(dT)-primed cDNA library generated from poly(A)+ RNA, 21,925 clones were fingerprinted by hybridization with 217 different 8-mer oligonucleotide probes and grouped into 6291 clusters (corresponding to different transcripts) ranging in size from 1 to 265 clones. In a pilot analysis, the 5′ ends of representative clones from 711 clusters were sequenced and the sequences of 90 clones (12.7%) were found to show significant similarity to 80 distinct proteins in the databases (P<10⁻⁵). The potential advantage of a fingerprinting approach is that by grouping cDNAs into clusters before selecting clones for sequencing, the probability of resequencing prevalent mRNA species repeatedly is greatly reduced.

In our study, we were able to identify a large number of different proteins without any normalization of the cDNA library, simply by sequencing large numbers of clones. Nucleic acid hybridization studies indicate that there are some 8500 diverse mRNA species at the gastrula stage, assuming a mean length of 2 kb (Galau et al., 1974; Davidson, 1986). Not all these mRNAs are expressed by PMCs, as many transcripts expressed at the gastrula stage have tissue-specific distributions (Kingsley et al., 1993). If we accept, for the sake of argument, that 5000 different mRNA species are expressed by PMCs at the midgastrula stage, then our EST analysis identified approximately one quarter of those gene products. The average number of hits/protein was still quite low (2.1). If the sample size were increased further, it would become progressively more difficult to identify new mRNAs, and some method of selectively enriching for rare sequences would probably be required to obtain a complete catalog of expressed genes. Nevertheless, a more comprehensive catalog of genes expressed by PMCs could certainly be obtained simply by additional high-throughput sequencing. Such an approach would be facilitated to a modest extent by first performing filter hybridization using gene-specific probes to identify those clones in the arrayed library that correspond to highly abundant sequences (rRNAs, cytochrome C oxidase subunit I, MSP130, etc.), which together represent 10-15% of the clones in the library, and then eliminating those from further analysis.

The gene products that emerged from the EST analysis appear to mirror closely the cellular composition of the cDNA library. Of the 21 gene products identified more than eight times in our analysis (Table 2), four are known to be expressed specifically by PMCs (MSP130, PM27, SM37 and SM50) and the others are proteins with housekeeping functions that are likely to be expressed by many cell types, including PMCs. Moreover, every gene product currently known to be expressed exclusively or primarily by PMCs at the gastrula stage (including SM30, profilin, spectrin, collagens, lamin B, etc.) was identified at least once in our analysis. We have also shown by in situ hybridization that many of the proteins identified for the first time through our sequencing analysis are expressed primarily or exclusively by PMCs (Fig. 2). Based on these observations and our determination of the purity of the cell population used to generate the library, we expect that the great majority of proteins identified in our analysis are expressed by PMCs. Nevertheless, a small number of mRNAs were also identified that are unlikely to be expressed by these cells. Clones were identified that encoded various members of the Spec gene family (10 cases), arylsulfatase (four cases), and hatching enzyme (three cases). All are abundant mRNAs expressed specifically by presumptive or definitive ectoderm cells. Therefore, independent methods will be required to confirm that any specific protein identified in our analysis is expressed by PMCs.

We chose to study PMC gene expression at the equivalent of the mid-gastrula stage. Analysis of proteins synthesized by cultured micromeres by two dimensional gel electrophoresis indicates that the major transition in the molecular program of differentiation of the cells occurs prior to that stage, approximately concomitant with ingression (Harkey and Whiteley, 1983). Most proteins that are upregulated at ingression continue to be synthesized throughout later development. This pattern of protein expression is consistent with studies demonstrating that most major morphogenetic activities of the PMCs are activated by the early to mid-gastrula stage but persist much later in development. For example, the ability of the cells to migrate directionally in response to guidance information in the blastocoel is clearly established by the mid-gastrula stage, when the subequatorial ring forms, and persists at least until the late gastrula stage (Ettensohn, 1990). PMCs first become fusogenic at the early gastrula stage but remain capable of fusing with other PMCs throughout embryogenesis (Hodor and Ettensohn, 1998). Thus, by focussing on the population of mRNAs expressed by PMCs at the mid-gastrula stage, we are very likely to include most of the gene products that regulate the major morphogenetic activities of these cells.

Because our library was not normalized, the frequencies with which we identified specific mRNAs should reflect their relative abundance within the sequence population (Lee et al., 1999b). A potential limitation is that transcripts with unusually long or short 3′-UTRs could be under- or over-represented, respectively, in the pool of ESTs that yielded matches to known protein-coding sequences. Nevertheless, of the 21 proteins identified more than eight times, four are terminal differentiation gene products of PMCs (MSP130, PM27, SM37 and SM50) and the remainder have general housekeeping functions. All these proteins might therefore be expected to be expressed at high levels by PMCs. It has been estimated that there are >200 SM50 mRNA molecules per PMC and about 16 PM27 mRNA molecules per PMC at peak expression levels during gastrulation (Killian and Wilt, 1989; Harkey et al., 1995). The single most frequently identified sequence in the EST analysis, mitochondrial 16S rRNA (169 hits) has been shown by independent methods to be the most prevalent poly(A)+ RNA species in the embryo (Davidson, 1986; Poustka et al., 1999). Finally, four of the other proteins in the collection of 21 (cyclins A and B, α-tubulin, and the small subunit of ribonucleotide reductase) were also among a subset of sequences identified multiple times in a random-primed, cleavage stage cDNA library (Lee et al., 1999b). Based on all these considerations, it seems likely that the frequency with which a specific sequence was identified in our EST analysis provides a good indication of the prevalence of the corresponding mRNA in PMCs, at least in most cases. The fact that the great majority of proteins were identified only once or twice in the EST pool indicates that most mRNAs expressed by PMCs are in the moderate-to-low prevalence class.

The EST analysis identified a large number of new, potential regulators of PMC morphogenesis that will be attractive candidates for further study. For example, we identified several proteins that have been shown to play a role in mediating cell-cell fusion in other developing systems. These include the small GTPase Rac, three members of the tetraspanin family, and DOCK180/myoblast city. Tetraspanins are a recently identified family of four-pass transmembrane proteins that function in multiple cellular processes. These proteins regulate integrin function and play a role in the fusion of myoblasts and gametes (Hemler, 1998; Tachibana and Hemler, 1999; Kaji et al., 2000). Genetic and biochemical studies have shown that Rac and DOCK180/myoblast city are important regulators of myoblast fusion and interact directly with one another (Luo et al., 1994; Erickson et al., 1997; Kiyokawa et al., 1998; Nolan et al., 1998; Frasch and Leptin, 2000). We found that NET-7 is expressed at high levels specifically by PMCs, supporting the view that this protein has a special function in these cells.

Several proteins were identified that have been implicated in the regulation of filopodial motility. These include three proteins that regulate actin polymerization and the formation of filopodia and other cell protrusions: Arp3 (P=2×10⁻⁸²) (cloned previously from Hemicentrotus pulcherrimus; GenBank Accession Number, AB016822 ), N-WASP (P=4×10⁻¹⁰) and cdc-42 (P=7×10⁻¹⁶) (Miki et al., 1998; Carlier et al., 1999; Rohatgi et al., 1999; Borisy and Svitkina, 2000). In addition, we identified putative sea urchin homologs of the cell surface proteoglycan syndecan (P=1×10⁻⁷) and syntenin (P=3×10⁻⁵⁴), a cytoplasmic protein that interacts with the C-terminal region of syndecan (Grootjans et al., 1997). Syndecans have been implicated in a variety of processes related to cell adhesion, signaling and motility, including the formation of filopodia (Woods and Couchman, 1998; Granes et al., 1999).

The major biosynthetic activity of the PMCs is the secretion of the calcareous skeleton. We identified two new candidate spicule matrix proteins, SM50-like and a C-lectin, and showed that both were expressed specifically by PMCs. In addition, we identified a discoidin domain receptor (DDR) tyrosine kinase that might function in skeletogenesis. DDRs are an ancient class of receptor tyrosine kinase that have recently been found to act as collagen receptors, undergoing a slow autophosphorylation and activation in response to that ligand (Shrivastava et al., 1997; Vogel et al., 1997; Vogel, 1999; Vogel et al., 2000). A variety of evidence (reviewed by Ettensohn et al., 1997) indicates that PMCs must interact with a self-produced collagenous substrate in order to synthesize spicules, probably in part through the activation of the SM30 gene, and sea urchin DDR is a candidate for mediating such an interaction. Finally, we identified two proteins closely related to MSP130. MSP130 is a novel, GPI-linked protein that appears to function in facilitating Ca²⁺ import (Farach-Carson et al., 1989). We have found that MSP130 and MSP-related proteins form a small gene family consisting of at least three members. cDNAs encoding MSP130-related 2 were identified 11 times in our analysis, indicating that this mRNA is expressed at high levels by PMCs. MSP130-related 1 was identified only once, but in situ hybridization analysis suggests that this mRNA is also abundant (Fig. 2).

We cloned at least seven new ECM molecules from the sea urchin: perlecan (P=9×10⁻⁶⁷), fibronectin (P=3×10⁻⁴¹), fibrinogen-related protein (P=2×10⁻⁴⁴), fibrillin (P=9×10⁻⁴³), F-spondin (P=3×10⁻¹⁹), nidogen (P=1×10⁻¹³⁵) and osteonectin (P=4×10⁻³⁶). One of these, fibronectin, has been implicated in PMC migration in several previous studies (Fink and McClay, 1985; Katow and Hayashi, 1985; Katow et al., 1990). These studies relied on probes against vertebrate fibronectin, however, and sea urchin fibronectin had proven refractory to cloning for many years. Our findings resolve the long-standing issue of whether sea urchins have fibronectin and will allow further analysis of the function of the endogenous protein. We also identified several ECM-degrading enzymes that might function in PMC ingression, migration, or skeletogenesis, including membrane-type matrix metalloprotease 15 (P=8×10⁻⁴⁶), matrix metalloprotease 1 (collagenase) (P=6×10⁻³⁰), heparanase (P=5×10⁻¹⁸) and a metalloelastase (P=6×10⁻³⁰).

A large number of transcription factors emerged from the EST analysis, most of which were isolated as full-length clones. These included aristaless (P=8×10⁻³⁰), MTA1 (P=0), interleukin enhancer binding factor 2/NF45 (P=1×10⁻¹²⁵), Sox11 (P=5×10⁻²⁷), Sox21 (P=6×10⁻³⁸), MCG4 (P=1×10⁻⁵⁸), MED7 (P=2×10⁻⁶⁶), AP-1/c-jun (P=3×10⁻³²), HEX (P=3×10⁻¹⁹) and ERG (P=1×10⁻¹¹⁰; a partial ERG sequence from sea urchin was previously reported by) (Qi et al., 1992). Several of these factors are expressed selectively by mesodermal cells in other systems (e.g. aristaless, NF45 and ERG) and two, MTA1 and ERG, have been implicated in the regulation of cell movements in metastatic and embryonic cells (Nicolson and Moustafa, 1998; Herman et al., 1999; Vlaeminck-Guillem et al., 2000). In situ hybridization analysis showed that at least two of these transcription factors, ERG and aristaless, are expressed predominantly or exclusively by PMCs.

The EST analysis also identified many components of conserved signaling pathways that play important roles in early sea urchin development. With respect to the Wnt signaling pathway (reviewed by Kikuchi, 2000; Peifer and Polakis, 2000), we identified (1) a frizzled-related protein most similar to mouse frizzled-1 (P=4×10⁻²⁷); (2) axin, a key scaffolding protein that regulates the phosphorylation and degradation of β-catenin (P=5×10⁻⁴⁷); (3) a Wnt protein most similar to vertebrate Wnt-8 (P=6×10⁻⁸⁴); (4) protein kinase B/Akt, a kinase that phosphorylates and inactivates GSK3 (P=1×10⁻¹¹⁵); and (5) regulatory subunit B of protein phosphatase 2A (PP2A), another component of the multi-protein complex that regulates β-catenin phosphorylation (P=3×10⁻⁶¹). Differential nuclearization of β-catenin along the animal-vegetal axis plays an important role in patterning the early sea urchin embryo (reviewed by Davidson et al., 1998; Angerer and Angerer, 2000; Ettensohn and Sweet, 2000). The identification of these components of the β-catenin pathway will facilitate further analysis of the regulation of β-catenin nuclearization.

With respect to the Notch pathway (reviewed by Artavanis-Tsakonas et al., 1999), we identified for the first time in the sea urchin a putative Notch ligand, a homolog of the protein Delta (P=1×10⁻¹³⁵). Notch signaling is required for specification of non-skeletogenic mesoderm (NSM), and activation of Notch depends on inductive signals from micromere progeny (Sherwood and McClay, 1999; Sweet et al., 1999). It has therefore been speculated that micromere descendants might express a ligand for the Notch receptor (Sweet et al., 1999). The identification of sea urchin Delta in our PMC EST analysis supports this hypothesis. The EST analysis also led to the identification of sea urchin homologs of two proteins that regulate the post-translational processing of Notch, TNFα-converting enzyme (P=3×10⁻¹⁸) and presenilin I (P=3×10⁻³¹) (Chan and Jan, 1999; Brou et al., 2000).

Considerable progress is currently being made in elucidating molecular pathways that pattern the early sea urchin embryo (see Davidson et al., 1998; Angerer and Angerer, 2000; Ettensohn and Sweet, 2000). The specification of PMC fate normally requires the presence of β-catenin in micromeres (Emily-Fenouil et al., 1998; Wikramanayake et al., 1998; Logan et al., 1999) and may also require the zygotic activation of an ETS transcription factor in the micromere lineage at the late blastula stage (Kurokawa et al., 1999). PMC specification is also linked to a change in the properties of the vegetal cortex at the eight-cell stage and/or the unequal cell division that produces the micromeres (reviewed by Ettensohn and Sweet, 2000). An apparently complete program of PMC specification can also be elicited in other cells of the early embryo, in some lineages even as late as the mid-late gastrula stage, by experimentally perturbing cellular interactions (Ettensohn, 1992; McClay and Logan, 1996). Ultimately, both normal and regulative pathways of PMC fate specification must be linked to the specific activation of the downstream effector molecules that execute the remarkable morphogenetic program of PMCs.

Alix (ALG-2-interacting protein)

Anaphase-promoting complex, subunit 10

Bcl-X (apoptosis regulator)

BTG1 (B-cell translocation gene 1 protein)

BUB-3

Cdc5

Cdc6

Cdc45

Cdc47

Chromodomain helicase DNA binding protein 3

Cyclin A

Cyclin B

Cyclin C

Cyclin D-interacting protein

Cyclin K

Cyclin 1

Cyclin-dependent kinase 2

Cyclin-dependent kinase 3

Cyclin-dependent kinase 8

Cyclin-dependent kinase inhibitor 3

DAD (‘defender against cell death’) 1

DNA helicase

DNA ligase III

DNA polymerase, α subunit

DNA polymerase, β subunit

DNA polymerase, epsilon subunit

ERCC-6 (excision-repair protein)

Fizzyl

GADD45

Histone acetyltransferase

Histone H1 (cleavage stage)

Histone H1 (embryonic)

Histone H2A (cleavage stage)

Histone H2A variant

Histone H2B (cleavage stage)

Histone H2B (embryonic)

Histone H3 (embryonic)

Inner centromere protein

Mad2 (spindle assembly checkpoint protein)

MOB1 (mitosis and ploidy protein)

Nim (‘never-in-mitosis’)-related kinase

PCD6 (programmed cell death protein)

Prohibitin

PRB1 (pRB-associated protein)

RAD1

RAD21

RAD23

RAD51

RCC1 (regulator of chromatin condensation 1)

RecQ (DNA helicase)

Replication origin recognition complex, subunit 4

RBP2 (pRB binding protein 2)

SUDD protein

Topoisomerase

UV-damaged DNA binding factor

XRCC (‘x-ray repair cross-complementing’

protein) 3

ACK protein kinase

Adenylyl cyclase

Axin

BMP2/4 (univin)

Calmodulin

CAM kinase I

Casein kinase I, α

Casein kinase I, γ

Cysteine-rich FGF receptor

Δ

DVR-1 (Vg1-like)

Frizzled-1

G protein, β1 subunit

GTPase-activating protein

HP28 (PDGF-associated protein)

Inositol 1,4,5-triphosphate-binding protein

IRE1 (ER kinase)

JNK protein kinase

MAPKAPK-4

MAPK phosphatase

Myotubularin (dual specificity phosphatase)

Notch- like protein

Nucleoside diphosphate kinase B

PERK (ER ser/thr kinase)

Phosphatidyl inositol-4-phosphate-5-kinase

Phosphorylase B kinase, α subunit

Phosphorylase B kinase, γ subunit

Phosphotyrosyl phosphatase activator

Pim-3/KID-1 kinase

Pleiotrophin (heparin-binding growth factor)

Pre-B-cell colony-enhancing factor

Presenilin I

Protein kinase, 5'-AMP-activated

Protein kinase B/Akt

Protein kinase C inhibitor, 14-3-3 protein

Protein phosphatase 1, γ subunit

Protein phosphatase 2A, 74 kDa regulatory subunit

Protein phophatase 2A inhibitor (SET protein)

Protein phosphatase 2C

Protein phosphatase 4, regulatory subunit 1

Protein phosphatase with EF-hands

Protein ser/thr kinase 11 (PAR-4)

Protein ser/thr kinase, RING3

Protein tyrosine kinase 9

Protein tyrosine phosphatase

Protein tyrosine phosphatase receptor interacting

protein (liprin)

PTEN tumor suppressor

Rac

Raf

Rap1b

Regulator of protein phosphatase 4

Rho1

Scavenger receptor, cysteine-rich

SMAD5

Src-type protein kinase

SpAN protease

TNFα converting enzyme

TRAF (tumor necrosis factor receptor associated factor)

TRAP170 (thyroid hormone receptor associated factor)

Wnt-8

Vav-2 oncogene

ABC transporter

ADP/ATP translocase

Ammonium transporter

Anion exchange protein, AE-2

Annexin VI

Cationic amino acid transporter

Cystinosin

Glutamate receptor (AMPA-type)

Glycine transporter

L-amino acid transporter, LAT-1

Lysosomal proton pump ATPase

Lysosomal proton pump ATPase, δ subunit

MTRP (Golgi 4-transmembrane spanning transporter)

N-type Ca²⁺ channel, α1 subunit

Na⁺/Ca²⁺ exchange protein

Na⁺/H⁺ exchange regulatory factor 2

Na⁺/K⁺ ATPase, α chain

Na⁺/K⁺ ATPase, β chain

Na⁺/phosphate cotransporter

Organic anion transporter

Organic cation transporter

Porin

Proline transporter

SAP97 (discs-large homolog, PDZ protein)

SLOB protein (K⁺ channel-interacting protein)

Sodium bicarbonate transporter

Sodium-dependent phosphate transporter, type II

Stomatin

Sulfonylurea receptor 2B

Tetracycline transporter-like protein

Abp1 (SH3P7)

Actin, muscle-specific

Actin, cytoskeletal

Actin-binding LIM protein

Actin-like protein, 13E

Actin-like protein, BAF53

Α-actinin

Anillin

Ankyrin

p21-Arc

p41-Arc

Arp1/centractin

Arp3

Arp6p

Attractin (CUB family of adhesion/guidance molecules)

Bicaudal-D

Bystin

Canoe

Cdc10 (septin 1)

Cdc42

Cdc42-interacting protein 4

Centrin

Cofilin

Crumbs

Cut1

Del-1 (integrin-binding protein)

Discoidin-domain receptor tyrosine kinase

DOCK 1/myoblast city

Dynactin, subunit p25

Dynein heavy chain

Dynein heavy chain, isotype 3A

Dynein light chain

Dyskerin/CBF5 (centromere/MT-binding protein)

EWAM (actin binding protein)

Filamin

Flamingo (protocadherin)

Flightless I

Interaptin

Kelch-motif protein

Kinesin-like protein 1

Kinesin heavy chain

LASP-1

MAP (77 kDa)

Moesin

Myosin heavy chain, nonmuscle

Myosin heavy chain kinase, β

Myosin light chain kinase

N-WASP

Outer dense fiber protein 2

PAK-interacting exchange factor

Profilin

Prominin

p55-related MAGUK protein (multiple PDZ-domain protein)

Roadblock

Semaphorin VIa

Slit-2

Spectrin, α chain

Spectrin, β chain

Syndecan-2

Syntenin (syndecan-binding PDZ protein)

Talin

Tektin B1

Tensin

Tetraspan CD-53

Tetraspan NET-4

Tetraspan NET-5

T-plastin

Tropomyosin

Tubulin, α

Choriogenin

Coagulation factor V

Collagen, α2 (IV)

Collagen, α3 (IV)

ECM18

Fibrillin

Fibrinogen-related

Fibronectin

Fibropellin Ia

Fibropellin Ib

Fibropellin II

F-spondin

Glypican (HSPG)

Heparanase

HLC-32

Hyalin

Laminin, α chain

Laminin-like protein

Matrix metalloprotease 1 (MMP-1, collagenase)

Matrix metalloprotease 15 (MT2MMP-15, membrane-type)

Metalloelastase

Nidogen/entactin

Osteonectin

Perlecan

Acetyl-CoA-acyltransferase A

Acetyl-serotonin N-methyltransferase

Aconitate hydratase

Acyl-CoA dehydrogenase, long chain

Acyl-CoA oxidase, subunit II

Acyl-CoA-synthetase

ADE-2 (multifunctional protein)

Adenosylhomocysteinase

Adenylosuccinate lyase

ADP ribosyltransferase

Adenylosuccinate synthetase

Alanine aminotransferase

Aldehyde dehydrogenase 4

Aldose-1-epimerase

Aminocyclopropane carboxylate deaminase

Aminocyclopropane carboxylate synthase

AMP deaminase

Arginine methyltransferase

Arylsulfatase

Aspartate transaminase

ATPase N2B

ATP synthase, αsubunit

ATP synthase, β subunit

ATP synthase, γ subunit

ATP synthase F0 subunit 6

ATP synthase coupling factor 6

cGMP-specific phosphodiesterase

Chondroitin-6-sulfotransferase

CoA-thioester hydrolase

CTP synthase

Cytochrome B

Cytochrome B5

Cytochrome C

Cytochrome C-1 (heme protein precursor)

Cytochrome C oxidase, subunit I

Cytochrome C oxidase, subunit II

Cytochrome C oxidase, subunit III

Cytochrome C oxidase, subunit VIa

Cytochrome C reductase, Complex III, subunit 2

Cytochrome C reductase, Complex III, subunit 6

Cytochrome C reductase, iron-sulfur subunit

Cytochrome P450 monooxygenase

Deoxyribonuclease I

Diacylglycerol acyltransferase

2,4-Dienoyl-CoA-reductase

Dihydroflavonol-4-reductase

Dihydrofolate reductase

Electron transfer flavoprotein, β subunit

Endo-β1,4-glucanase

Enhancer of Rudimentary protein

Esterase

Fructose biphosphate aldolase

Galactosylceramide sulfotransferase

GalNAc-α2,6-sialyltransferase

GalNAc transferase I

GalNAc transferase II

GlcNAc sulfotransferase

GlcNAc transferase I

GlcNAc transferase II

Glutamate-cysteine ligase, regulatory subunit

Glutamine synthetase

Glutathione-S-transferase

Glycerol kinase

Glycine hydroxymethyltransferase

Glycogen debranching enzyme

GMP synthase

GPI anchor biosynthesis protein

HemeA:farnesyltransferase

Hexosaminidase

Hexosamindase B, β subunit

Holocytochrome C synthetase

Hydrolase, α/β

8-Hydroxyguanine glyosylase

Hydroxyisobutyrate dehydrogenase

Hydroxymethylglutaryl-CoA-synthase

Isocitrate dehydrogenase

Lipoic acid synthetase

Lipoyl transferase

Malate dehydrogenase

Maltase

Methyl sterol oxidase

NADH-dependent glutamate synthase

NADH deydrogenase, α/β subcomplex, 8 kDa subunit

NADH dehydrogenase, Complex I, B14 subunit

NADH dehydrogenase, Complex I, 15 kDa subunit

NADH dehydrogenase, Complex I, 19 kDa subunit

NADH dehydrogenase, Complex I, 20 kDa subunit

NADH dehydrogenase, Complex I, 39 kDa subunit

NADH dehydrogenase, Complex I, 75 kDa subunit

NADH dehydrogenase, subunit 1

NADH dehydrogenase, subunit 2

NADH dehydrogenase, subunit 4

NADH dehydrogenase, subunit 5

NAD(P) transhydrogenase

NAD(P)H steroid dehydrogenase

N-arginine dibasic convertase

Nitrogen fixation protein

Ornithine decarboxylase

Ornithine decarboxylase antizyme

Peptidyl-prolyl cis-trans isomerase

Peroxide reductase

Phosphate transfer protein B

Phosphatidylinositol transfer protein

Phosphoadenosine phosphosulfate synthase

Phospholipase B

Phospholipid scramblase, TRA1

Pyruvate dehydrogenase phosphatase

Retinoic acid hydrolase

Ribonucleotide reductase, small subunit

Selenophosphate synthase

Serine hydroxymethyltransferase

Sterol 14-α demethylase

Succinate dehydrogenase

Succinyl-CoA-synthetase, α subunit

Sucrose isomaltase

Tricarboxylate carrier

tRNA pseudouridine synthase A

Uronyl 2-sulfotransferase

Uroporphyrin decarboxylase

ADP ribosylation factor 1

ADP ribosylation factor 4

ADP ribosylation factor-directed GTPase activating protein

Amphiphysin

AP17 (clathrin coat assembly protein)

BET3

Clathrin, heavy chain

Coatomer, β subunit

Coatomer, γ subunit

Copine I

Copine III

Cytohesin

ER lumen protein retaining receptor

EXO70

Exportin(tRNA)

Importin α4

Importin β3 (RanBP5)

Importin 7 (RanBP7)

NSF-1

Nucleoporin p58

Prenylated Rab acceptor 1

Rab1

Rab2

Rab6

Rab7

Rab10

Rab11b

Rab21

Rab23

Rab26

RAMP4

RanBP1

Sar-1a

Sec13

Sec23

Sec24

Sec61

SNAP-25-interacting protein

Sorting nexin 4

Sorting nexin 12

Synaptophysin

Syntaxin 7

Transitional endoplasmic reticulum ATPase

Aminopeptidase A

Aminopeptidase N

Aqualysin

Capn7

Carboxypeptidase A

Cathepsin C

Cathepsin D

Cathepsin L

Chymotrypsin inhibitor 2

ClpX

Cullin 1

DnaJ

HSP70

HSP90

HSP90-binding protein, p23

HSP, 97 kD

HSPCO14

HSPCO17

Legumaine

Lysosomal carboxypeptidase

Prefoldin 1

Proteasome, δ chain

Proteasome, ε chain

Proteasome, 26S, subunit S3

Proteasome, 26S, subunit 6

Proteasome, 26S, subunit 7

Proteasome, 26S, subunit 9

Proteasome, 26S, subunit 10b

PRT1

Skp1

Smt3A (ubiquitin-like protein)

SUMO-1 activating enzyme, subunit 1

TCP-1, δ subunit

TCP-1, η subunit

Tubulin-specific chaperone

Ubiquitin

Ubiquitin carboxyterminal hydrolase

Ubiquitin-conjugating enzyme E2

Ubiquitin-conjugating enzyme

Ubiquitin-specific protease 3

Ubiquitin-specific protease 14

eIF-1α

eIF-1β

eEF-2

eIF-2C

eIF-2G

eIF-3, subunit 8

eIF-4A

eIF-5

Nanos/Xcat2

NAT1 translational repressor

Ribosomal protein, 60S subunit, L3

Ribosomal protein, 60S subunit, L5

Ribosomal protein, 60S subunit, L7a

Ribosomal protein, 60S subunit, L10

Ribosomal protein, 60S subunit, L11

Ribosomal protein, 60S subunit, L15

Ribosomal protein, 60S subunit, L30

Ribosomal protein, 60S subunit, L44

Ribosomal protein, 40S subunit, S3

Ribosomal protein, 40S subunit, S4

Ribosomal protein, 40S subunit, S15a

Ribosomal protein P0

Signal peptidase

Signal recognition particle, 14 kDa protein

Signal recognition particle, 54 kDa protein

Signal sequence receptor, β subunit

Signal sequence receptor, δ subunit

Signal sequence receptor, γ subunit

SUI1

tRNA synthetase, arginyl

tRNA synthetase, asparaginyl

tRNA synthetase, glycyl

tRNA synthetase, phenylalanyl

tRNA synthetase, valyl

WHO translational regulator

Abstrakt (DEAD box protein)

AU-rich RNA-binding protein

BAT1 (nuclear ATP-dependent RNA helicase)

CIRP (cold-inducible RNA-binding protein)

Cleavage stimulation factor, 50 kDa subunit

Crooked-neck protein

GRY-RBP (RNA-binding protein)

hnRNP protein F

hnRNP protein K

hnRNP protein L

hnRNP protein R

IGF-II mRNA-binding protein

Lark (RNA-binding protein)

MVP-100 (major vault protein)

NOP4

NSAP1 (RNA-binding protein)

Poly(A)-binding protein

Rae1

Ribonuclease

Ribonuclease P

RNA helicase, p68

RNase L inhibitor

RRP5

SAP (‘spliceosome-associated protein’)-130

SAP-145

SAP-155

SF3a (splicing factor)

Sm D3 (small nuclear ribonucleoprotein)

SnRNP assembly defective 1

SnRNP protein B

Splicing factor, arginine/serine-rich 7

Splicing factor, CC1.3

Splicing factor, KH-type

Splicing factor, polyU-binding

Splicing factor, SC 35

Splicing factor, proline/glutamine-rich

SR protein

SS-A, 60 kD ribonucleoprotein

U3 snRNP protein, 55 kDa

U5 snRNP protein, 116 kDa

U5 snRNP protein, dim1

U5 snRNP protein, Prp8

Zinc-finger RNA-binding protein

C-lectin

MSP130

MSP130-related 1

MSP130-related 2

PM27

SM30

SM37

SM37-related

SM50

SM50-related

AMY-1

AND-1

AP1/c-Jun

Aristaless

CHD 1 (chromodomain-helicase-DNA-binding protein)

DR1

ERG

Eyes absent

Glucose-regulated repressor

HEX

Hexamer-binding protein

HMED7

HMG1

HMG2

Interleukin enhancer binding factor 2

MAX-like bHLHZIP protein

MCG4

MTA-1

NFkB

NF-X1/shuttle craft

Nuclear receptor co-repressor 1

p100 transcriptional coactivator

p300 transcriptional cofactor JMY

RNA polymerase II, subunit RPB4

Scaffold attachment factor B

SNF2α

SNF2-related CBP activator protein

SoxB1

Sox11

Sox21

SpShr2

Stage-specific activator protein (SSAP)

SWI/SNF complex, 170 kDa subunit

TATA-binding protein-related factor 2

TFIIH

TFIIS

Zinc-finger protein 2

Zinc-finger protein 84

Zinc-finger protein 184 (Kruppel-related)

Zinc finger protein, KRAB

Zinc-finger protein, MEX-1

Zinc-finger protein, OZF

Acinus L protein

Adrenal gland protein (lozenge-like)

Amyloid-β(A4) precursor

Androgen-induced protein

Aut1p

BCRP1 (breast cancer resistance protein)

BING4 (WD40 protein)

Brain protein I3

Brain protein 44

Butyrate response factor 1

Calcyclin-binding protein

Calcyphosine

Cell surface antigen 4F2

Coiled-coil protein

Degenerative spermatocyte protein (transmembrane)

EDRK-rich factor 2

EF-hand protein

EGF-repeat-containing transmembrane protein

Egg receptor for sperm

F protein

Fatvg

GARP (‘glutamic acid-rich protein’)

Glucose-regulated protein, 170 kDa

GOB4 (cement gland protein )

Growth arrest-specific gene 11

HAN11 (WD-repeat protein)

Hatching enzyme

HDL-binding protein/vigilin

HOOK1

Human surface glycoprotein

Huntingtin-interacting protein

Hydroxyproline-rich glycoprotein

IgGFc-binding protein

JTV1 protein

Lamin B

Lamin B receptor

LDL receptor-related protein 2

Leukemia virus receptor

LYAR (growth-regulating nucleolar Zn-finger protein)

MAMA

Meiosis-specific nuclear structural protein 1

Melastatin (down-regulated in metastatic melanoma)

Metallothionein

Nasopharyngeal epithelium-specific protein 1

Neuralized

Neuronal protein 15.6

Neuropathy target esterase (NPE)

Nuclear phosphoprotein p150TSP

Nucleolar protein P120

NMP200 (nuclear matrix protein)

Nuclear protein np95

Out-at-first

Oxysterol binding protein

Parkin

Peroxisomal biogenesis factor

p22, calcium-binding (EF hand) protein

Phosphoprotein α 4

PIWI

Polyposis locus protein 1

Polyprotein

Pregnancy-induced growth inhibitor

Prostate cancer overexpressed gene 1

Protein B

Rcd (‘required for cell differentiation’) 1

Reduced expression in cancer

Reticulocalbin

Reverse transcriptase-like protein

SART-1 tumor antigen

Selenoprotein W

SMCY

Spastin

Spec1a

Spec2c

Spec2d

Stromal interaction molecule 1

Testis-specific Zn-finger protein

TRABID protein

Transposase

Traube

Trichohyalin

Tubby

Tumor-suppressing subtransferable candidate 1

Wolf-Hirschhorn syndrome candidate 1

This work was supported by NIH Grant HD24690 (C. A. E.), NSF Grant IBN-9817988 (C. A. E.), and by a grant from the Stowers Institute for Medical Research.