The SH3p4/Sh3p8/SH3p13 protein family: Binding partners for synaptojanin and dynamin via a Grb2-like Src homology 3 domain

The yeast two-hybrid screen was performed using the protocol of Vojtek et al. (14). Briefly, a portion of the rat synaptojanin cDNA encoding amino acids 1,015–1,308 (the proline-rich tail of the 145-kDa isoform of synaptojanin) was subcloned into the pLexA vector, creating a fusion protein with the lexA DNA binding domain. This construct includes a 16-aa alternative splice found between amino acids 1,140 and 1,155. The yeast strain L40 (partial genotype MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacZ) was transformed with this construct to create the strain L40:pLexAPro. Expression of the fusion protein was confirmed by Western blot of cell lysates. As a positive control, L40:pLexAPro was transformed with the SH3 domain of human amphiphysin I subcloned into the pACT vector, generating a fusion of the amphiphysin SH3 domain with the GAL4 activation domain. Double transformants were selected and screened for His⁺ lacZ⁺ phenotype. Activation of the reporter genes in double transformants demonstrated that the L40:pLexA strain was expressing a functional ligand for the amphiphysin SH3 domain. L40:pLexAPro was then transformed with a rat brain cDNA library maintained in the pGAD10 vector (CLONTECH). Of approximately 70 × 10⁶ clones plated, 108 were His⁺ and 30 demonstrated expression of the lacZ⁺ reporter gene by a filter assay. These 30 clones represented 11 distinct cDNAs as determined by restriction analysis of the cDNA inserts. The clones isolated from the primary screen were screened a second time by a mating assay, allowing us to discard clones that activated reporter genes when paired with a pLexA:Lamin construct. Six distinct clones fulfilled the requirements for specificity of this assay. These cDNAs were sequenced entirely on both strands.

Five micrograms of the rat brain cDNA library described above was used as a template to amplify the missing 5′ regions of the SH3p4 and the SH3p13 cDNAs. 5′ and 3′ primers were directed against the pGAD10 vector and the cDNA to be amplified, respectively. Standard PCR conditions were used with the exception that a 5:1 mixture of Taq (Promega) and PFU (Stratagene) polymerases was used. PCR products were subcloned into the pGex 4T2 vector and sequenced.

Commercial multiple tissue Northern blots were purchased from CLONTECH and probed with dsDNA derived from clones isolated in the two-hybrid screen. Probes for SH3p4 and SH3p13 transcripts were PCR products of 530 and 1,300 nucleotides, respectively. The SH3p8 probe was an EcoRI fragment from the library plasmid that consists of the entire cDNA (see Fig. 1A). Fifty nanograms of gel-purified probe was labeled using a random primed labeling kit (Boehringer Mannheim), and blots were probed under standard high-stringency conditions (15) and exposed for 24 hr. Blots were dehybridized by boiling in 0.1× standard saline citrate for 5 min and reprobed after verification of dehybridization.

Portions of the coding regions of the SH3p4/8/13 proteins corresponding to the SH3 domains and variable regions of SH3p4, SH3p8, and SH3p13 were amplified by PCR using Vent polymerase (NEB, Beverly, MA). PCR fragments were digested by appropriate restriction enzymes and subcloned into the pGex4T series of vectors (Pharmacia) or the pET21a vector (Invitrogen) that expresses the recombinant protein with a polyhistidine tag. Constructs were verified by sequencing. The host strain BL21(DE3) was transformed with the resulting expression vectors, and the recombinant protein was purified on the appropriate affinity matrix using standard methods.

Triton X-100 extracted brain homogenate was prepared as follows. Male Sprague–Dawley rats were euthanized and decapitated. One brain was homogenized in 10 ml of homogenization buffer (20 mM Hepes, pH 7.4/300 mM sucrose/150 mM NaCl) to which a protease inhibitor cocktail was added (4 μg/ml each of pepstatin, aprotinin, leupeptin, and phenylmethylsulfonyl fluoride). Total homogenate was spun at 2,600 rpm for 10 min at 4°C in an SS-34 rotor (Beckman). Twenty percent Triton X-100 (wt/vol) was added to the supernatant to a final concentration of 1%. After a 30-min incubation at 4°C the sample was spun for 1 hr at 35,000 rpm at 4°C in a Ti-70 rotor (Beckman). The supernatant from the high-speed centrifugation was saved as brain extract.

Two hundred micrograms of recombinant glutathione S-transferase (GST) fusion proteins was incubated with glutathione-Sepharose beads for 2–3 hr and washed three times with PBS containing 1% Triton X-100 (wt/vol). The beads were equilibrated with homogenization buffer and then added to 10 ml of brain extract. After 3 hr of incubation at 4°C the beads were washed four times for 20 min in PBS/1% Triton X-100 (wt/vol). Proteins on the beads were eluted into SDS/PAGE sample buffer, separated by SDS/PAGE, and either visualized by Coomassie brilliant blue stain or processed for Western blotting.

Three pairs of rabbits were immunized with GST fusion proteins comprising COOH-terminal fragments of SH3p4 (97 amino acids), SH3p8 (113 amino acids), or SH3p13 (103 amino acids), respectively. Sera were affinity-purified on GST fusion proteins immobilized on a 1-ml HiTrap column (Pharmacia). Polyclonal antibodies raised against dynamin I (DG1) and glutamic acid decarboxylase were previously described (16, 17). A mAb raised against the proline-rich tail of synaptojanin (AC1) was a kind gift of Amy Hudson and Christof Haffner in our laboratory. An anti-synaptophysin mAb (C7.2) was a kind gift of R. Jahn (Yale University), and anti-dynamin mAb (HUDY-1) was purchased from Signal Transduction Laboratories (Lake Placid, New York).

One milliliter of the Triton X-100 brain extract used for affinity chromatography (see above) was precleared by incubation at 4°C for 2 hr with 10 μg of nonimmune rabbit IgGs followed by two sequential precipitations with 100 μl of protein A-Sepharose. Five micrograms of affinity-purified antibodies then were added to the precleared extract and incubated for 3 hr at 4°C. Immune complexes were precipitated with 50 μl of protein A Sepharose and eluted into 200 μl of SDS/PAGE sample buffer.

Cultured hippocampal neurons were prepared as previously described (18). Immunofluorescence of brain tissue and cultured hippocampal neurons was performed as described (18, 19). SDS/PAGE and Western blotting were performed according to the methods of Laemmli et al. (20) and Towbin et al. (21), respectively. Preparation of total brain homogenate and gel overlays with recombinant fusion proteins (far Western blots) were performed as previously reported (10).

A portion of the rat synaptojanin cDNA encoding the proline-rich tail of the 145-kDa isoform (amino acids 1,015–1,308) was used as the bait in a two-hybrid screen to identify interacting partners from a rat brain library. Both the bait construct and the brain library were transformed into the yeast strain L40, which contains two selectable markers for the two-hybrid interaction: the yeast HIS3 and the bacterial lacZ genes. The screen led to the isolation of 11 clones encoding five distinct cDNAs, which activated both reporter genes in the presence of the bait construct and not in the presence of a control plasmid.

Sequencing revealed that four cDNAs encoded SH3 domain containing proteins. Three of the cDNAs encoded rat homologues of recently identified mouse proteins SH3p4, SH3p8, and SH3p13 (13). Comparison of these cDNAs with the sequence of the mouse homologues revealed that two identical clones contained the entire ORF of Sh3p8, including 3′ and 5′ untranslated regions (Fig. 1A). The clones corresponding to SH3p4 (two identical clones) and Sh3p13 (three identical clones) lacked the starting methionine but extended beyond the COOH terminus, which encodes the SH3 domain. The other SH3 domain containing protein isolated by the screen (two independent clones containing the COOH-terminal fragment of the molecule) was amphiphysin II. This protein interacts in vitro with synaptojanin and dynamin (22) and has been extensively characterized by our lab and others (13, 22, 23). Because SH3p4, SH3p8, and Sh3p13 were uncharacterized, we focused our further analysis on this protein family.

The missing 5′ regions of the ORFs of SH3p4 and SH3p13 were cloned by PCR (see Materials and Methods) and found to correspond to the sequences of the mouse homologues present in the database. The complete ORFs of rat SH3p4, SH3p8, and SH3p13 encode proteins of 352, 368, and 347 amino acids, respectively. An alignment of the three sequences (Fig. 1B) revealed that they are highly similar to each other (68% average identity) with two blocks of strong similarity corresponding to the NH₂-terminal two-thirds of the molecule, which is predicted to form a coiled-coil structure, and to the COOH-terminal SH3 domains. These two regions are linked by a sequence that is substantially divergent among family members. A high degree of identity is observed between rat and mouse proteins along their entire sequences (96% average identity). The three proteins do not contain hydrophobic stretches predicted to form transmembrane regions.

To verify the interaction between the SH3p4/8/13 proteins and synaptojanin, the COOH-terminal moieties (variable region plus SH3 domain) of the SH3p4/8/13 proteins were expressed as GST fusion proteins and used to probe brain extract in an overlay (far Western blot) assay. As shown in Fig. 2A each of the SH3 domains binds primarily to bands with electrophoretic mobilities of ≈100 and 150 kDa. These bands also are recognized by the SH3 domain of amphiphysin I and previously have been identified as dynamin I and synaptojanin (10). Accordingly, bands of the same motility were recognized by anti-dynamin I and anti-synaptojanin antibodies (Fig. 2B). Note that in comparison to the SH3 domain of amphiphysin I these SH3 domains label synaptojanin relatively more intensely than dynamin I.

The fusion proteins comprising the COOH-terminal regions of SH3p4 and SH3p13 also were used to affinity-purify binding proteins from brain extract. As shown in Fig. 3 A and B, these fusion proteins specifically purified two bands that by Western blotting were shown to be dynamin I and synaptojanin, respectively (Fig. 3C).

Having established that the SH3p4/8/13 proteins bind to synaptojanin and dynamin I, we characterized the cellular and subcellular distribution of these proteins.

The pattern of expression of SH3p4, SH3p8, and SH3p13 mRNAs in different tissues was determined by Northern blot analysis (Fig. 4). The message for SH3p4 (single band around 3 kb) is expressed almost exclusively in brain, with extremely low levels detectable in kidney (Fig. 4A). The Sh3p8 message (single band at 2.4 kb) is found in all tissues (Fig. 4B). The message for SH3p13 (single band around 1.8 kb) is most abundant in testis, but also detected at lower levels in brain and liver (Fig. 4C).

Antibodies were raised against GST fusion proteins of C-terminal fragments of the SH3p proteins comprising the variable domains plus the SH3 domains. Two sera were used for further experiments due to their high titers and specificity for the SH3p4/8/13 proteins. Serum 1 that was raised against SH3p13 proved to be specific for SH3p13 whereas serum 2 that was raised against SH3p4 recognized all three SH3p4/8/13 proteins by Western blot (Fig. 5A). Antibodies affinity-purified from serum 1 recognized a single band of 40 kDa in brain and affinity-purified antibodies from serum 2 recognized a doublet of 40 and 45 kDa in brain (Fig. 5B). These electrophoretic mobilities agree with the predicted molecular masses of the SH3p4/8/13 proteins. Based on a comparison of data obtained with serum 1 and serum 2 as well as other sera (data not shown), the 40-kDa immunoreactivity corresponds to the SH3p4 and SH3p13 proteins whereas the 45-kDa immunoreactivity corresponds to SH3p8. In conclusion, all three proteins are expressed in brain although only SH3p4 seems to be brain specific.

Western blot analysis of primary subfractions of rat brain demonstrated that all three proteins were present in both soluble and particulate fractions. The overall distribution of these proteins during subcellular fractionation was similar to that of dynamin I and synaptojanin (data not shown). We next investigated by immunofluorescence whether the cellular and subcellular localization of the SH3p4/8/13 proteins in brain is consistent with an interaction in situ with synaptojanin and dynamin I.

Immunostaining of rat brain sections (Fig. 6) and cultured hippocampal neurons (Fig. 7) was performed using the antibodies affinity-purified from serum 1 and serum 2 (see Fig. 5 A and B). Both in brain sections and cultured neurons immunostaining was mainly concentrated in nerve terminals as demonstrated by a colocalization with immunoreactivity for synaptophysin, a marker of presynaptic nerve terminals (24, 25). A diffuse cytosolic staining also was observed in both brain sections and cultured neurons, consistent with the existence of a soluble pool of the SH3p4/8/13 proteins. Generally the immunoreactivity for the SH3p4/8/13 proteins was highly overlapping with the immunoreactivity for dynamin I, synaptojanin, and amphiphysin I immunostaining.

To further confirm the association of the SH3p4/8/13 proteins with synaptojanin and dynamin I in the neuronal cytoplasm we investigated whether immunoprecipitation of the SH3p4/8/13 proteins from brain extract resulted in coprecipitation of these two proteins. As shown in Fig. 8, a Western blot analysis of the immunoprecipitates formed by antibodies affinity-purified from serum 2 demonstrated that both dynamin I and synaptojanin, but not a cytosolic control protein, glutamic acid decarboxylase, were present in a complex with the SH3p4/8/13 proteins. Note that the immunoprecipitate is relatively more enriched with synaptojanin in comparison to dynamin I consistent with the preferential binding of these proteins to synaptojanin (see above).