Distinct roles for multiple Src family kinases at fertilization

At fertilization, the eggs of all species studied to date exhibit a transient rise in intracellular free Ca²⁺ that is necessary to establish the polyspermy block and for the activation of development (reviewed by Stricker, 1999; Runft et al., 2002). The mechanism by which this rise in Ca²⁺ is induced remains unknown. Src family kinases (SFKs) have been implicated as a required component of the Ca²⁺ release pathway at fertilization in echinoderms, ascidians and certain vertebrates (reviewed by Jaffe et al., 2001; Runft et al., 2002; Sato et al., 2000b; Sato et al., 2004). The current working hypothesis in the echinoderm, ascidian, fish and frog models is that sperm somehow trigger the activation of an egg SFK, which directly or indirectly activates phospholipase Cγ (PLCγ), leading to inositol trisphosphate (IP₃) production and release of Ca²⁺ through IP₃ receptors in the endoplasmic reticulum (ER) of the egg.

Much of the experimental evidence in support of the model comes from studies in the echinoderms, which offer the ability to conduct large-scale biochemistry on exquisitely synchronous populations of eggs and the ease of single-cell microinjection experiments. The necessity of SFKs and PLCγ in echinoderm egg activation has been demonstrated using pharmacological inhibitors (Lee and Shen, 1998; Shearer et al., 1999; Shen et al., 1999; Abassi et al., 2000) and dominant-interfering strategies (Carroll et al., 1997; Carroll et al., 1999; Shearer et al., 1999; Giusti et al., 1999b; Giusti et al., 2003; Kinsey and Shen, 2000; Runft et al., 2004). Injection of active Src protein triggers Ca²⁺ release in starfish eggs, but cannot rescue eggs that have been injected with dominant-negative PLCγ, placing the SFK upstream of PLCγ (Giusti et al., 2000).

There is evidence for a similar SFK-PLCγ-IP₃ egg activation pathway in ascidians (Runft and Jaffe, 2000) and zebrafish (Wu and Kinsey, 2001; Wu and Kinsey, 2002; Kinsey et al., 2003). In Xenopus, inhibition of PLC or SFK activity have been reported to block egg activation (Sato et al., 1998; Sato et al., 2000a; Glahn et al., 1999; Tokmakov et al., 2002) and both SFK and PLCγ activity are increased at fertilization (Sato et al., 1996; Sato et al., 1999; Sato et al., 2000a; Sato et al., 2002). Likewise, in mammals, pharmacological studies have implicated PLC and tyrosine kinase activity in the Ca²⁺ release pathway at fertilization (DuPont et al., 1996). However, in amphibians and mammals, SH2-mediated dominant interference of PLCγ has no effect on sperm-induced Ca²⁺ release at fertilization (Runft et al., 1999; Mehlmann et al., 1998), indicating that if PLCγ is involved, it is regulated differently than in echinoderms and ascidians. A recent report (Kurokawa et al., 2004) indicates that specific pharmacological (PP2, lavendustin A) and peptide inhibitors (peptide A) of SFKs have no effect on Ca²⁺ release in mouse eggs and that injection of recombinant Src protein or Src mRNA encoding a constitutively active form of Src does not stimulate Ca²⁺ release. Furthermore, inhibition of SFK activity in rat eggs using PP2 or SU6656 suggests that SFK may play a role downstream of Ca²⁺ release (Talmor-Cohen et al., 2004). The presence of multiple SFKs in Xenopus (Steele et al., 1989a; Steele et al., 1989b; Steele et al., 1990; Sato et al., 2004) and mammalian eggs (Kurokawa et al., 2004; Talmor et al., 1998; Talmor-Cohen et al., 2004) might result in compensatory or combinatorial effects and may help to explain why it has been difficult to demonstrate conclusively their role in egg activation. Clearly, the precise pathway and regulatory aspects of Ca²⁺ release in eggs of these vertebrate species requires further investigation.

The possibility of multiple SFKs and potential species differences also underscores the importance of evaluating the precise molecular mechanism used in the echinoderm model, particularly in regard to identifying individual signaling components and testing their roles in the pathway. Identifying the SFK(s) involved in the switch-like signaling event at fertilization is also necessary to understand the initial, upstream activation trigger (whether it is sperm- or egg-derived). Most metazoans have more than one Src-type kinase (Ottilie et al., 1992; Hughes, 1996; Hanks, 2003) and it has been demonstrated in a number of cellular signaling contexts that multiple SFKs can have overlapping roles within the same signaling module (Abram and Courtneidge, 2000; Courtneidge, 1994; Brown and Cooper, 1996; Thomas and Brugge, 1997; Nel, 2002). This raises the question of whether multiple SFKs participate in the egg activation pathway.

There is evidence for multiple SFKs in sea urchin eggs based on data using antibodies directed against vertebrate SFKs (cf. Kinsey, 1996; Abassi et al., 2000) and on partial SFK-like cDNA sequences identified in homology-based RT-PCR screens (Wessel et al., 1995; Kinsey, 1997; Sakuma et al., 1997). A Src-type kinase (AcSrc) has been described in eggs of the sea urchin Anthocidaris crassispina, but its function is not yet known (Onodera et al., 1999). We recently described the cloning and characterization of an SFK from the sea urchin Strongylocentrotus purpuratus, designated SpSFK1, which functions in initiating Ca²⁺ release at fertilization (Giusti et al., 2003). However, extensive screening of a macroarrayed sea urchin 7-hour cDNA library using AcSrc1 or SpSFK1 did not result in identification of additional SFK cDNAs (Giusti et al., 2003) and additional SFK-like sequences have not yet been identified through the sea urchin genome project (http://seaurchin.caltech.edu/genome/). Thus, in no model system to date has there been an inclusive study that both identifies all egg SFKs and that tests each specifically for their role in egg activation.

Here, we describe the cloning of three SFK-like sequences expressed in starfish (bat star; Asterina miniata) eggs. Specific antibodies were generated against each of the SFKs, and were used to evaluate the presence and activity profiles of the SFK proteins in eggs. To test the necessity of these proteins, dominant-interfering SH2 domains were designed for each SFK and assessed for the ability to inhibit sperm-induced Ca²⁺ release when microinjected into eggs.

Adult Asterina miniata were collected from the Santa Barbara Channel or Bodega Bay, CA and maintained in open system aquaria at 14-16°C. Ovary and testis were obtained, oocytes matured and sperm prepared for fertilization as described (Runft et al., 2004; Foltz et al., 2004).

A ³²P-labeled random primed RT-PCR generated probe corresponding to 156 bp within the kinase domain of an A. miniata SFK was used to screen a macroarrayed A. miniata oocyte cDNA library (Kalinowski et al., 2003). The probe was obtained by RT-PCR using degenerate oligonucleotides based upon conserved amino acids in the kinase domain of human Src (GenBank accession number P12931) corresponding to amino acids 385-389 and 446-452. First-round amplification was carried out on A. miniata oocyte cDNA using a 3′RACE adaptor oligonucleotide and the forward primer followed by nested amplification using the degenerate primers. Resulting fragments were ligated into pBluescript KS for sequencing (Iowa State Sequencing Facility, Ames, IA). A 156 bp fragment exhibited high homology to Src family kinases in a BLAST search (www.ncbi.nlm.nih.gov/BLAST). Gene-specific primers were generated based on that sequence, which amplified a specific 156 bp fragment from A. miniata oocyte cDNA that showed highest homology to vertebrate SFKs, spanning the most conserved region of the kinase domain. This fragment was used as the probe to screen the arrayed library. Later analysis (see below) demonstrated that this sequence corresponds to amino acids 425-477 of AmSFK3.

Sequencing of cDNA clones identified in the library screen revealed that eight cDNA clones encoded tyrosine kinase domains. Of these, two identical clones contained a 1849 bp insert, which encoded an entire open reading frame of an SFK, designated AmSFK1. A third cDNA encoded a partial open reading frame overlapping the 3′ end of AmSFK1. A fourth cDNA clone contained a 1070 bp insert, which encoded part of the SH2 domain and all of the kinase domain of a distinct SFK, designated AmSFK2. Two other identical clones contained a 1226 bp insert, which contained approximately 400 bp of the extreme 3′ end of the open reading frame (part of the kinase domain) of yet a third SFK, designated AmSFK3. A single additional clone contained a 2062 bp insert that encoded the entire open reading frame of AmCsk (C-terminal Src kinase), and one clone contained the entire open reading frame of a receptor-type tyrosine kinase.

Rapid amplification of cDNA ends (RACE) with a Smart RACE cDNA Amplification kit (Clontech, Palo Alto, CA) was used to identify the entire open reading frames of the AmSFK2 and AmSFK3 cDNAs. RACE products were separated on a 1.2% agarose gel and bands greater in size than that of the remaining predicted open reading frame were gel-purified, cloned into the TOPO pCR2.1 vector (Invitrogen, Carlsbad CA) and sequenced. Clones containing overlapping sequence with the AmSFK2 and AmSFK3 partial cDNAs isolated from the library screen were confirmed by sequencing on both strands (Iowa State Sequencing Facility, Ames, IA). Open reading frames were assigned using the GCG map function (Genetics Computer Group, Madison WI). New oligonucleotides were then designed within the 5′ UTR of the 5′ RACE products and the 3′ UTR of the cDNA sequences in order to generate a full-length physical clone. Three different clones from three independent PCR reactions for both AmSFK2 and AmSFK3 were cloned into the TOPO pCR2.1 vector and sequenced in both directions. The PCR reactions using these oligonucleotides generated a 1914 bp physical clone of AmSFK2 and a 2348 bp physical clone of AmSFK3. The complete sequences have been deposited in the GenBank database (accession numbers: AmCsk, AY518773; AmSFK1, AY518774; AmSFK2,AY518775; and AmSFK3, AY518776).

Total RNA was isolated from ∼2.5 g unfertilized oocytes or blastula, gastrula and bipinnaria stage A. miniata embryos (Kalinowski et al., 2003). PolyA⁺ mRNA was selected using the Micro Poly(A) Pure Kit (Ambion, Austin, TX); 1 μg was separated on a formaldehyde-containing gel and transferred to Hybond N⁺ nylon membrane (Amersham Pharmacia Biotech, San Francisco, CA). [α-³²P]dCTP randomly primed cDNA probes corresponding to 162 bp of the 5′ UTR of the AmSFK1 cDNA, 109 bp of the 3′ UTR of the AmSFK2 cDNA, or 265 bp of the 3′ UTR of the AmSFK3 cDNA were hybridized overnight at 68°C and washed under high stringency conditions (Sambrook and Russell, 2001).

A phylogenetic tree was generated using the sequences published (Hughes, 1996) except for CeSrc (GenBank accession number NP493502). Additional echinoderm sequences are AcSrc1 (AB016815), SpSFK1 (AY063749), SpAbl (AY063748). The SH3, SH2 and kinase domains were aligned using GCG PileUp then imported into the PAUP^* 4.0 beta software system (Swofford, 1998) and subjected to maximum-parsimony analysis. The phylogenetic tree was constructed by the neighbor-joining method on the basis of the proportion of amino acid differences (Saitou and Nei, 1987). All characters are estimated with equal weight in PAUP^* and a gap was considered missing. The robustness of the nodes in the parsimony analysis was tested with bootstrap analysis. Confidence limits were tested by bootstrapping the trees 100 times, and the `best' tree from that analysis is shown.

Recombinant protein was expressed in the TNT Coupled Reticulocyte Lysate System (Promega, Madison, WI) using the full-length AmSFK cDNA plasmids as templates in the presence of [³⁵S]methionine. Samples were prepared for electrophoresis by suspension in Laemmli sample buffer or were diluted in RIPA buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 2% Triton X-100; 0.1% SDS, 1 mM Na₃VO₄, 1 mM PMSF, 10 μM each of leupeptin, aprotinin and benzamidine), immunoprecipitated and resolved through a 12% polyacrylamide gel. Gels were stained, treated with EN³HANCE (NEN-DuPont), dried and exposed to Fuji (Tokyo, Japan) Super RX film. Films were scanned and figures compiled using Adobe Photoshop (Adobe Systems, Seattle, WA).

The unique domains of each AmSFK protein (see Fig. 1) (AmSFK1 amino acids 1-69; AmSFK2 1-77; AmSFK3 1-121) were amplified by PCR and ligated into pET28b (Novagen, Madison, WI). The constructs were sequenced to confirm in-frame fusion and to rule out PCR incorporation error. Proteins were expressed in E. coli strain BL21(DE3) (Novagen, Madison, WI), and purified via the His₆ tag at the C-terminus as described (Runft et al., 2004). Concentrated, purified fusion protein was used as an immunogen in two hens for each and IgY was purified from immune egg white (Aves Labs, Tigard, OR). The rabbit polyclonal anti-cSrc antibody (SRC2) was from Santa Cruz Biotechnology (Santa Cruz, CA).

For immunoblot analyses, eggs and embryos were either lysed in Triton X-100 solubilization buffer (1% Triton X-100, 50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM Na₃VO₄, 1 mM PMSF, 10 μM each of leupeptin, aprotinin and benzamidine) or were fractionated as previously described to generate cytosolic and total membrane fractions (Giusti et al., 1999a; Belton et al., 2001). After determining the concentration (BCA method; PierceEndogen), proteins were separated on 10 or 12% polyacrylamide gels (see Figure legends). Blots were probed with 6.1 μg/ml AmSFK1 IgY, 10.4 μg/ml AmSFK2 IgY or 10.4 μg/ml AmSFK3 IgY in blocking buffer containing 5% dry milk. HRP-conjugated secondary antibody (goat anti-chicken IgY, Aves Labs, Tigard, OR) was used at 0.05 ng/ml in the blocking buffer containing 5% dry milk. Antibody detection was with the PicoWest SuperSignal enhanced chemiluminescence kit (PierceEndogen).

For time-course samples, a small sample of eggs was fertilized in a beaker with gentle mixing and samples were removed at various times, quickly centrifuged and the seawater removed. The eggs or zygotes were lysed immediately in ice-cold Triton X-100 buffer by passage through a 27.5-gauge needle. The time point indicates the time between sperm addition and lysis and was typically ±5 seconds. Clarified total egg lysates (1 ml of 1 mg/ml) were incubated with 2 μg of AmSFK1, AmSFK2 or AmSFK3 IgY antibody for 2 hours at 4°C. Anti-chicken IgY conjugated to agarose beads was added (25 μl of a 50% slurry; Promega or Aves Lab) and samples were incubated for an additional 2 hours. Immune complexes were collected by centrifugation and washed four times in lysis buffer and then in ice-cold kinase buffer (10 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl₂, 5 mM MnCl₂). The kinase reaction was started by addition of 20 μl kinase buffer containing 2.5 μM ATP and 10 μCi [γ-³²P]ATP (3000 Ci/mmol, 10 mCi/ml; DuPont NEN). For the tight time-course experiments, the washed immune complexes were resuspended in 25 μl of 40 mM HEPES, pH 7.5, 10 mM MgCl₂, 3 mM MnCl₂, 10% glycerol, 1 mM DTT and 7 μCi [γ-³²P]ATP (Annerén et al., 2004). In some experiments, 0.5 μg of an SFK substrate was added: the optimal peptide substrate for Src fused to the C-terminus of the green fluorescent protein, GFP₂₃₅IYGEFG (Yang et al., 1999; Giusti et al., 2003). Kinase reactions were performed at 30°C for 15 minutes and stopped by addition of 5 μl 5× sample buffer and heating at 95°C for 10 minutes. Samples were separated on 12% polyacrylamide SDS gels. Gels were stained, dried and exposed to X-ray film. Scanned films were imported into Adobe Photoshop to produce the figures. Tyrosine phosphorylation of the GFP₂₃₅IYGEFG substrate was quantified from scanned images of films using the NIH Image program (http://rsb.info.nih.gov/nih-image/). Each time point was normalized to that of the unfertilized (time 0) time point, which was arbitrarily given a value of 1. Means, standard deviations and statistical significance were calculated using InStat (GraphPad Software, San Diego, CA).

The SH2 domains of AmSFK1 (amino acids 126-235), AmSFK2 (amino acids 128-230) and AmSFK3 (amino acids 179-282), were amplified by PCR using primers that introduced the restriction sites EcoRI (5′) and XhoI (3′). AmSFK1 SH2 and AmSFK2 SH2 domains were directionally cloned in-frame into pGEX-6P-3 and AmSFK3 SH2 into pGEX-6P-2 (Stratagene, San Diego, CA) and sequenced on both strands. The SH2 domain of S. purpuratus Abl (SpAbl), was used as a negative control (Giusti et al., 2003).

Glutathione S-transferase (GST) SH2 domain fusion proteins were expressed in E. coli JM109 and purified using glutathione-sepharose (Pharmacia) as previously described (Carroll et al., 1997; Carroll et al., 1999; Giusti et al., 1999a). The beads were washed extensively in PBS, and the eluted proteins were spin-dialyzed and concentrated in PBS. Aliquots were frozen in liquid N₂ and stored at -80°C.

Quantitative microinjections were made using mercury-filled micropipettes (see Jaffe and Terasaki, 2004). Injection volumes were 3-5% of the egg volume (90 pl or 150 pl for the 3000 pl A. miniata oocyte). For Ca²⁺ measurements, 10 μM Calcium Green 10 kDa dextran (Molecular Probes, Eugene, OR) was injected into the oocytes, alone or together with the indicated protein. Injected oocytes were matured with 1-methyladenine (1-MA; Sigma Chemical Co., St. Louis, MO) and then inseminated with sperm diluted 1:10,000. Calcium Green fluorescence was measured during fertilization using a photodiode and recorded using Scope software (AD Instruments, Grand Junction, CO) as previously described (Giusti et al., 2003; Runft et al., 2004). Experiments were performed at 16°C. Means, standard deviations and statistical significance were calculated using InStat (GraphPad Software, San Diego, CA).

A macro-arrayed A. miniata oocyte cDNA library (Kalinowski et al., 2003) was screened with an RT-PCR-generated probe representing a portion of the kinase domain of an A. miniata SFK. One C-terminal Src Kinase (Csk) cDNA (designated AmCsk) and three Src family kinase cDNAs (designated AmSFK1, AmSFK2, and AmSFK3) were isolated. AmCsk has a 1299 bp open reading frame and a predicted M_r of 50.3K. Characterization of this Src regulatory kinase will be described elsewhere. AmSFK1 has a 1548 bp open reading frame and a predicted M_r of 57.8K; AmSFK2 has a 1560 bp open reading frame and a predicted M_r of 58.9K; and AmSFK3 has a 1689 bp open reading frame and a predicted M_r of 63.1K. Each of the SFK cDNAs exhibits the canonical Src family domain structure (Brown and Cooper, 1996): a unique N-terminal domain, SH3, SH2 and kinase domains, as well as the conserved regulatory tyrosines (Y416 and Y527 of human Src) characteristic of SFKs (Fig. 1) (Xu et al., 1999).

These AmSFKs and other echinoderm SFKs (AcSrc1, SpSFK1) are most closely related to the Src A gene family, although none of them is a member of this group based on phylogenetic comparisons (Fig. 2). Table 1 provides sequence comparisons of the AmSFKs with the human Src family members. AmSFK2 appears to be a more ancestral SFK as it groups with other invertebrate and early metazoan SFKs (Fig. 2). For example, it is 56-68% identical and 73-81% similar to Drosophila melanogaster Dsrc41, Dsrc42A and the sponge SRK4 whereas AmSFK1 and AmSFK3 share greater homology with the vertebrate SrcA and B family members (Table 1, Fig. 2). In addition, AmSFK3 contains a unique domain that is nearly twice as large as that of AmSFK1 and AmSFK2 (121 amino acids versus 69 and 77 amino acids, respectively).

Gene-specific UTR probes were used to evaluate the mRNA expression pattern for each of the AmSFKs. The AmSFK1 transcript is ∼4.5 kb and is present at roughly equal levels in oocytes throughout early development to the bipinnaria larval stage (Fig. 3, left panel). The AmSFK2 transcript is ∼9 kb in eggs. However, a second transcript of ∼8.5 kb appears at the blastula stage, the time of increased zygotic transcription. The 9 kb message (presumably maternal) is present throughout early development through to gastrula stage at roughly equal levels, though decreasing slightly with time and is substantially reduced by the bipinnaria larval stage. The AmSFK2 8.5 kb zygotic message is highly expressed at blastula stage, then decreases as early development progresses through the bipinnaria stage (Fig. 3, middle panel). The AmSFK3 transcript is ∼3.5 kb in oocytes, but levels decreased substantially by the bipinnaria stage (Fig. 3, right panel).

In order to characterize the proteins encoded by the AmSFK transcripts, antibodies directed against the unique N-terminal domains of all three proteins were developed. Each antibody recognized the appropriate recombinant protein immunogen on blots (data not shown). In order to demonstrate rigorously that the predicted open reading frame of each cDNA produced a protein of the predicted molecular weight and that the antibodies were specific for each AmSFK protein, immunoprecipitations of in vitro transcribed and translated proteins were conducted. Recombinant proteins of M_r ∼58K for AmSFK1, ∼59K for AmSFK2 and ∼63K for AmSFK3 were produced (Fig. 4), indicating that each of the predicted open reading frames represents a contiguous transcript that generates a protein of the predicted molecular weight. Immunoprecipitations of the in vitro transcription and translation reactions using the antibodies raised against the unique domain of each SFK revealed that the antibodies specifically recognized the cognate full-length protein (Fig. 4). In addition, a polyclonal antibody raised against the C-terminal 18 amino acids of human Src (sc-18; see Fig. 1) was tested in the same immunoprecipitation assay. This antibody recognizes the vertebrate Src A proteins Src, Yes and Fyn and has been used previously to evaluate SFK presence and activity in echinoderm eggs (Giusti et al.., 1999a; Abassi et al.., 2000). This vertebrate Src A antibody precipitated AmSFK1, but not AmSFK2 or AmSFK3 protein.

Immunoblotting revealed that all three AmSFK proteins are present in eggs (Fig. 5) and throughout embryonic development (data not shown). The antibody made against AmSFK1 recognizes a protein of ∼58 kDa that is enriched in the membrane fractions of eggs (Fig. 5A). The AmSFK2 protein (∼59 kDa) is present in the membrane fraction and to a somewhat lesser degree, in the cytosolic fraction (Fig. 5B). The AmSFK3 protein (∼63 kDa), like AmSFK1, appears to be enriched in the membrane fractions of eggs (Fig. 5C). The results of nine different experiments indicated that there was no significant fertilization-dependent change in the fractionation pattern of the three AmSFK proteins (data not shown).

Overexpression of an SFK SH2 domain can have a specific, dominant-interfering effect on the endogenous kinase (Roche et al., 1995; Giusti et al., 1999b; Giusti et al., 2003; Abassi et al., 2000; Kinsey and Shen, 2000; Runft and Jaffe, 2000), inhibiting the function of the endogenous SFK by competing specifically for binding to interacting proteins (Kinsey et al., 2003). To test the necessity of each AmSFK in the Ca²⁺ release pathway, purified GST SH2 domain fusion proteins of each AmSFK were expressed in bacteria and purified (Fig. 6). The SH2 domain fusion proteins were pre-tested in affinity interactions with egg lysates to ensure that they were capable of binding to tyrosine phosphorylated proteins (data not shown). The SH2 domains were then co-injected with the calcium indicator dye, Calcium Green dextran (CaG), into A. miniata oocytes. Oocytes were matured, sperm were added to the chambers and Ca²⁺ release was monitored (Table 2; Fig. 7). Control eggs injected with either 10 μM Calcium Green alone or SpAbl SH2 domains (25 μM final concentration in the cytoplasm) showed normal Ca²⁺ release (Fig. 7A,B; Table 2). Of the nine oocytes injected with the AmSFK1 SH2 domain (1 mg/ml or 25 μM final concentration in the cytoplasm), three showed statistically significant delayed and reduced Ca²⁺ release (Fig. 7C), and six showed no Ca²⁺ release over the duration of the recording (Fig. 7D; Table 2). Of the nine oocytes injected with the AmSFK2 SH2 domain (25 μM), all showed a very slight delay in the initiation of Ca²⁺ release versus that of the SpAbl SH2 control-injected oocytes, but this was not statistically significant (Fig. 7F; Table 2). Of the eight oocytes injected with the AmSFK3 SH2 domain (25 μM), four showed statistically significant delayed and reduced Ca²⁺ release (Fig. 7G), and four showed no Ca²⁺ release over the duration of the recording (Fig. 7H; Table 2). In every case, even when a complete block of Ca²⁺ release was observed, the eggs showed an action potential, indicating that sperm-egg fusion had occurred (McCulloh and Chambers, 1992; Swann et al., 1992). Eggs that exhibited a delayed or blocked Ca²⁺ release typically were polyspermic (data not shown) and, as can be seen by the sample traces (Fig. 7), there were often amplitude and wave duration effects as well.

The mechanism of dominant interference predicts a dose-dependent effect, which was tested by injecting varying amounts of the AmSFK1 and AmSFK3 SH2 domain proteins. Of the eggs injected with 2.5 μM AmSFK1 SH2 domain, all six exhibited slightly delayed, but statistically significant Ca²⁺ release (Fig. 7E; Table 2). Of the seven eggs injected with 2.5 μM AmSFK3 SH2 domain, four showed significant delayed and reduced Ca²⁺ release (Fig. 7I) whereas three showed no Ca²⁺ release (Fig. 7J; Table 2). Injection of the AmSFK3 SH2 domain protein at 0.25 μM revealed that all five oocytes showed normal Ca²⁺ release (Fig. 7K; Table 2).

In order to assess their enzymatic activity, each AmSFK was immunoprecipitated from lysates of unfertilized and fertilized eggs and immune complex kinase assays (ICKAs) were conducted (Fig. 8). As controls for input and equal recovery, the lysate input and a small portion of the immunoprecipitate prior to the kinase assay were blotted and probed with the appropriate AmSFK antibody (Fig. 8C and data not shown). These assays were carried out in a standard Src kinase assay buffer with cold ATP present (Fig. 8A). AmSFK1 immune complexes showed a consistent and statistically significant fertilization-inducible tyrosine kinase activity (Fig. 8A). Maximal labeling of the substrate occurred at ∼2 minutes post sperm addition and then decreased over time (Fig. 8B,D). By 5 minutes (and then on to 60 minutes) post sperm addition, the levels of AmSFK1 activity returned to that of prefertilization levels, whereas the amount of AmSFK1 protein in the egg remained steady (Fig. 8C; data not shown). However, under these conditions, we were unable to detect reproducible kinase activity of AmSFK2 or AmSFK3 in unfertilized or fertilized eggs (up to 60 minutes post sperm addition), although in some experiments AmSFK3 activity appeared to be stimulated extremely rapidly and transiently.

We were concerned that the antibodies might be interfering with the kinase activity of AmSFK2 and AmSFK3. To assess this possibility, recombinant proteins were expressed in transcription-translation coupled reactions, immunoprecipitated and subjected to kinase assays. All three AmSFKs exhibited robust kinase activity including autophosphorylation; both AmSFK1 and AmSFK3 phosphorylated the Src peptide and polyGluTyr substrates whereas AmSFK2 targeted only polyGluTyr (data not shown). Thus, at least in vitro, all three AmSFKs exhibit kinase activity.

We then pursued the possibility that the lack of reproducible detection of the activity of AmSFK2 and AmSFK3 was due to a sensitivity issue. In order to increase the sensitivity of the assay, the ICKAs were conducted in the absence of unlabeled (carrier/driver) ATP. AmSFK2 activity, including autophosphorylation, was never detected under these (or any other) conditions. However, AmSFK3 activity was detected at the earliest time point that could be obtained reproducibly (10 seconds post-sperm addition) and peaked at 30 seconds, just as AmSFK1 activity was beginning to rise (Fig. 9). By 2 minutes post-sperm addition, at the peak of AmSFK1 activity, AmSFK3 activity had returned to prefertilization levels (Fig. 9C). Immune complex kinase assays in which the exogenously added substrate was omitted indicated that both AmSFK1 and AmSFK3 were labeled, presumably by autophosphorylation (Fig. 9B).

Because of the accumulating evidence that SFKs play a central role in regulating the initiation of Ca²⁺ release at fertilization in echinoderms and potentially other species, and the precedence that multiple SFKs often function together within the same signaling cascade, we aimed to determine if there were multiple SFKs expressed in echinoderm oocytes. There have been several reports suggesting that echinoderms eggs have multiple SFKs, but this has been limited to analysis of RT-PCR fragments and the use of vertebrate antibodies generated against vertebrate proteins to detect multiple activities (Satoh and Garbers, 1985; Kamel et al., 1986; Wessel et al., 1995; Kinsey, 1996; Sakuma et al., 1997; Giusti et al., 1999a; Giusti et al., 1999b; Onodera et al., 1999; Abassi et al., 2000). Here, we describe the identification of three A. miniata oocyte SFK-like cDNAs (AmSFK1, AmSFK2 and AmSFK3) in addition to one C-terminal Src Kinase (AmCSK). Antibodies to each AmSFK revealed that the proteins are present in eggs. This fits with the general model that the Src/Csk regulatory circuit arose during early metazoan evolution (Miller et al., 2000). The AmSFK1 and AmSFK3 proteins group more closely with the vertebrate SrcA gene family, whereas AmSFK2 is a more primitive SFK, grouping with sponge SFKs, as well as with the unique mouse Srm and human Frk, which are thought to be ancestral SFKs (Hughes, 1996). None of the AmSFKs groups within the SrcA or SrcB subfamilies, indicating that the expansion of the Src-type kinases probably occurred after the echinoderms diverged on the deuterostome branch.

All three AmSFK proteins are expressed in starfish eggs. What role(s) do each of these SFKs play in fertilization or perhaps early development, especially given that echinoderm eggs (like eggs of most species) contain stored mRNAs and proteins for use in later development? The SH2 domain-based dominant-interference strategy was applied to test the necessity for each AmSFK, indicating that both AmSFK1 and AmSFK3 (but not AmSFK2) inhibit Ca²⁺ release at fertilization in starfish eggs in a specific and dose-dependent manner. This implicates both of these SFKs in signaling during egg activation. AmSFK2, on the other hand, is apparently not necessary for early egg activation events. It could play a role as an adaptor or scaffold protein, functioning in a way that does not require enzymatic activity and that is not inhibited by SH2-mediated interference. Although unusual, this is not unprecedented for SFKs and other non-receptor tyrosine kinases (Xu and Littman, 1993; Kaplan et al., 1995; Lin et al., 1997; Fincham and Frame, 1998; Foti et al., 2002; Carragher et al., 2003). However, AmSFK2 is probably playing a different, later role in early starfish development. The lack of detectable AmSFK2 kinase activity in eggs and zygotes and the observation that there appears to be a switch from maternal to zygotic forms of AmSFK2 mRNA at blastula-gastrula transition is consistent with this.

Because the dominant-interference method is currently the only method available to test necessity (to date, RNAi and morpholino antisense-mediated knockdown attempts of maternal proteins in echinoderm eggs have been unsuccessful), it is important to establish the specificity of the method. Injection of the SH2 domains of various non-Src vertebrate proteins have no effect on Ca²⁺ release in starfish eggs, but the SH2 domains of vertebrate Src and Fyn are inhibitory (Giusti et al., 1999b). In the current work, neither the SpAbl nor AmSFK2 SH2 domain have an effect on Ca²⁺ release, and the inhibitory effect of both the AmSFK1 and AmSFK3 SH2 domains is dose-dependent. AmSFK1 SH2 domains inhibit strongly at 25 μM, but not at 2.5 μM. AmSFK3 SH2 domains inhibit strongly at 25 μM and at 2.5 μM, but not at 0.25 μM. Recently, it was elegantly demonstrated that the mechanism of this dominant-interfering effect is probably due to disruption of protein-protein interactions and not through direct inhibition of kinase activity of the targeted SFK (Kinsey et al., 2003). These observations are consistent with the fact that SH2 domains from different proteins are selective in their substrate specificity (cf. Nollau and Mayer, 2002). In terms of relative affinity for target peptides, they vary at most, within a single order of magnitude (Ladbury, 2000; Kuriyan and Cowburn, 1997), but in the cell, even this small difference is probably sufficient to exert biological specificity (Pawson, 2004). Comparison of the SH2 domains from each of the AmSFKs reveals that they share the general structural features of SrcA family members, but that there are some differences in specific residues that could account for differential binding (see Fig. 1). In particular, the AmSFK2 SH2 domain exhibits distinct differences compared to AmSFK1 and AmSFK3 in the βD region that could influence binding (Fig. 1) (see Kuriyan and Cowburn, 1997). Our preliminary studies of the gross analyses of tyrosine phosphorylated proteins that bind to each SH2 domain are consistent with a model with distinct subsets of interacting proteins that account for the differential dominant-interfering effects seen in the eggs injected with SH2 domains.

There is also the possibility that the SH3 domain (that recognizes Pro-X-X-Pro motifs in target binding partners) of SFK also plays a role in substrate selectivity (cf. Broome and Hunter, 1996; Summy et al., 2000) (reviewed by Pawson and Nash, 2003). In sea urchin eggs, injection of the unique and SH3 domains of Xenopus Fyn caused a small delay in Ca²⁺ release at fertilization (Kinsey and Shen, 2000). However, our preliminary data suggest that microinjection of the SH3 domains of AmSFK into starfish oocytes have no effect on sperm-induced Ca²⁺ release and that injection of tandem SH3-SH2 domains is no more potent than injection of SH2 domains alone.

Although the dominant-interference tests suggest that both AmSFK1 and AmSFK3 are required for triggering Ca²⁺ release at fertilization, it is possible that both inhibitory SH2 domains affect the same SFK. This can occur if the two AmSFKs share substantial substrate specificity within their respective SH2 domains (see above). A model in which only AmSFK1 is required would predict that although AmSFK3 SH2 domains might inhibit AmSFK1 activity, they would not be as effective as AmSFK1 SH2 domains. Because the AmSFK3 SH2 domains have a stronger inhibitory effect (roughly an order of magnitude) than AmSFK1 SH2 domains, we do not favor the `AmSFK1 only' model. It is also possible that only AmSFK3 is required in the Ca<sup>2+</sup> release pathway and that the AmSFK1 SH2 domains exert their inhibitory effect on AmSFK3. Although we cannot rule this out, the `one kinase only model' does not account for the observation that both AmSFK3 and AmSFK1 kinase activities are rapidly and transiently stimulated (by at least four- to fivefold) at fertilization (see below). Therefore, our current working model is that both AmSFK1 and AmSFK3 play a role in and are required for normal initiation of Ca²⁺ release at fertilization in starfish eggs. In this `both AmSFk1 and AmSFK3' model, each respective SH2 domain specifically inhibits its own cognate full-length endogenous kinase by competing for binding partners and the activity of the other AmSFK is incapable (both spatially and temporally) of overcoming the block of its partner kinase to efficiently stimulate sperm-induced Ca²⁺ release.

What are the mechanistic roles of the starfish SFKs? Both AmSFK1 and AmSFK3 appear to be required for, and play a role in early egg activation events. AmSFK3 kinase activity peaks at 30 seconds, whereas AmSFK1 kinase activity peaks at 2 minutes post sperm addition. AmSFK3 activity is very rapid and transient, and detectable only under conditions where sensitivity is maximized. AmSFK1 activity, in contrast, is very robust. Interestingly, labeling of the SFK proteins themselves, presumably via autophosphorylation, was detected only using the more sensitive method and probably reflects a situation in which the bulk of the SFK isolated in the immune complex from the egg is in a stable, already phosphorylated condition.

A simple model that explains both the dominant-interference results and the fertilization-induced kinase activity is that a small amount of AmSFK3 is stimulated very rapidly at fertilization (perhaps localized to the site of sperm-egg interaction). This then leads (directly or indirectly) to activation of AmSFK1, which amplifies the signal. This is consistent with sperm-induced activation of a kinase in the pathway leading to Ca²⁺ release through PLCγ activation. Although Ca²⁺ release occurs within tens of seconds of insemination, the response is switch-like and thus there are technical limitations to determining the precise threshold of SFK activity needed: the `peak' of kinase activity does not necessarily represent the threshold value (Shearer et al., 1999; Runft et al., 2004).

Multiple SFKs and related tyrosine kinases functioning within the same pathway is a common occurrence. An example of overlapping cellular function of SFK activity in a cell activation scenario is T-cell receptor (TCR) engagement and the concomitant activation of the T-cell (Clements and Koretsky, 1999; Nel, 2002; Nel and Slaughter, 2002; Cannons and Schwartzberg, 2004). In this case, different SFKs are activated sequentially (Burkhardt et al., 1994; Huang and Wange, 2004) and there is an important role for adaptor proteins (Chan et al., 2003; Latour et al., 2003; Jordan et al., 2003), which create docking sites for numerous other SH2 domain-containing proteins including PLC-γ1. This gives rise to the calcium transient necessary to stimulate T-cell activation fully (Schaeffer and Schwartzberg, 2000; Lewis et al., 2001; Miller and Berg, 2002; Nel, 2002; Nel and Slaughter, 2002; Singer and Koretzky, 2002).

The rapid activation of protein tyrosine kinase activity in T cells is similar to the kinetics of protein tyrosine kinase activity observed in echinoderm (reviewed by Kinsey, 1997; and see Kamel et al., 1986; Ciapa and Epel, 1991; Moore and Kinsey, 1995; Kinsey, 1996; Giusti et al., 1999a; Abassi et al., 2000) and frog (reviewed in Sato et al., 2000b) (see Sato et al., 1999; Sato et al., 2000a) eggs at fertilization. Similarly, as in T cells, PLC-γ is also tyrosine phosphorylated and activated in echinoderm and frog eggs at fertilization (De Nadai et al., 1998; Rongish et al., 1999; Shearer et al., 1999; Runft et al., 2004; Sato et al., 2000a; Tokmakov et al., 2002). Although SFK protein and SFK activity can be detected in affinity interactions using starfish egg PLCγ SH2 domains (Giusti et al., 1999a; Runft et al., 2004), we have been unable to establish definitively if indeed AmSFK1 or AmSFK3 directly interacts with and phosphorylates PLCγ. Now that the egg SFKs have been identified and reagents are available, this and other detailed mechanistic questions regarding the Ca²⁺ release pathway will be more amenable to investigation.

The work presented here demonstrates that multiple SFKs can operate in egg activation pathways, emphasizing that the entire complement of SFKs in the egg must be identified and evaluated to test their roles in the sperm-induced egg activation pathway. Doing so, in any species, will lead to new strategies to identify the upstream activators of the Ca²⁺ release pathway: the so-called `activation triggers', whether they are egg- or sperm-derived.

The authors gratefully acknowledge the advice, expertise and stimulating discussions with Laurinda Jaffe, John Lew and Andy Giusti throughout the project. We especially thank the anonymous reviewer who, in addition to providing many insightful comments, encouraged us to pursue the more sensitive kinase assays. Thanks also to Stu Feinstein, Michelle Roux and Linda Runft for critical comment and review of the manuscript; to Ian Townley for help with the time-course sample processing; and to Shane Anderson, Dave Farrar and Terry Marchiando for animal collection and sea water system maintenance at UCSB. F.J.O. was a UC Regents Graduate Fellow and J.G. was supported in part by a UCSB Graduate Dissertation Fellowship. This work was supported by a grant from the NSF to K.R.F.