Identification and Characterization of a Synaptojanin 2 Splice Isoform Predominantly Expressed in Nerve Terminals*

We have previously identified synaptojanin 1, a phosphoinositide phosphatase predominantly expressed in the nervous system, and synaptojanin 2, a broadly expressed isoform. Synaptojanin 1 is concentrated in nerve terminals, where it has been implicated in synaptic vesicle recycling and actin function. Synaptojanin 2A is targeted to mitochondria via a PDZ domain-mediated interaction. We have now characterized an alternatively spliced form of synaptojanin 2 that shares several properties with synaptojanin 1. This isoform, synaptojanin 2B, undergoes further alternative splicing to generate synaptojanin 2B1 and 2B2. Both amphiphysin and endophilin, two partners synaptojanin 1, bind synaptojanin 2B2, whereas only amphiphysin binds synaptojanin 2B1. Sequence similar to the endophilin-binding site in synaptojanin 1 We synaptojanin 2B with as yet uniden-tified partners in developing spermatids and acts on a pool of phosphoinositides the of their cytoskele-ton. both amphiphysin 1 (49) and endophilin 3 (SH3p13) (14, 73), which have here to to synaptojanin 2B2, are also in testis, although their subcellular localization Our present observations the notion that alternative splicing generates various forms of synaptoja-F IG . 8. Active Rac1 GTPase affects the intracellular localization of synaptojanin 2 isoforms, but not of syn- aptojanin 1 (145-kDa form). COS-7 cells were transfected singly with ex- pression plasmids encoding GFP-synap-tojanin 1 ( A ), GFP-synaptojanin 2A ( D ), GFP-synaptojanin 2B1 ( G ), or GFP-synaptojanin 2B2 ( J ). Alternatively, cells were cotransfected with expression plasmids encoding the HA-tagged Rac1 G12V dominant active mutant and GFP-synaptojanin 1 ( B and C ), GFP-synapto- janin 2A ( E and F ), GFP-synaptojanin 2B1 ( H and I ), or GFP-synaptojanin 2B2 ( K and L ). The cells were observed by confocal laser scanning microscopy to detect localizations of synaptojanin isoforms by GFP fluorescence ( A , B , D , E , G , H , J , and K ) and Rac1 by indirect immunofluorescence using an anti-HA antibody ( C , F , I , and L ).

Synaptojanins are a family of phosphoinositide phosphatases with a unique three-domain structure: an N-terminal region homologous to yeast Sac1p (1, 2), a central inositol 5-phosphatase domain, and a C-terminal region mediating protein-protein interactions such as Src homology 3 (SH3) 1 domain binding (Ref. 3; for a review, see Ref. 4). The identification and analyses of synaptojanin homologs in lower eukaryotes confirmed that this family of enzymes plays a role of fundamental importance in phosphoinositide metabolism (5)(6)(7). Synaptojanin 1, 2 the founding member of this family of enzymes, is a protein predominantly expressed in brain, where it is highly enriched in nerve terminals (8,9). The C-terminal region of synaptojanin 1 was reported to interact with the SH3 domains of several proteins implicated in the regulation of clathrinmediated synaptic vesicle endocytosis and actin organization at nerve terminals, including amphiphysins (10 -12), endophilins (13,14), intersectins (15), and syndapins (pacsins) (16 -19). These protein-protein interactions are thought to regulate the intracellular localization and activity of synaptojanin 1 (for reviews, see Refs. 20 and 21). The generation and phenotypic analyses of synaptojanin 1-deficient mice (22) and the characterization of Caenorhabditis elegans unc-26 mutants harboring mutations in the synaptojanin gene (23), together with studies of lamprey synapses in which synaptojanin function is acutely perturbed (24), have demonstrated a critical role of synaptojanin 1 in nerve terminal function. They have shown that synaptojanin 1 regulates a pool of phosphoinositides in nerve terminals that affect actin function and a dynamic equilibrium between assembly and disassembly of clathrin coats during synaptic vesicle recycling (for reviews, see Refs. [25][26][27]. We (28) and others (29) have previously described another mammalian synaptojanin (synaptojanin 2) that is expressed as multiple alternative splice variants in various tissues. Our characterization of the first identified form of synaptojanin 2 (henceforth referred as synaptojanin 2A) indicated that this isoform is recruited to mitochondria through the interaction with the PDZ domain of the mitochondrial outer membrane protein OMP25 and thereby affects the distribution and morphology of these organelles (30). Complementary DNA sequences for other splice variants of synaptojanin 2 lack the motif responsible for the interaction with OMP25 (29,31), which strongly suggests that other splice isoforms of synaptojanin 2 have different interacting proteins and functions.
Here we describe the characterization of two synaptojanin 2 alternative splice variants, synaptojanins 2B1 and 2B2. We show that synaptojanin 2B shares many properties with synaptojanin 1, including the biochemical interaction with several endocytic SH3 domain-containing proteins and the predominant localization at nerve terminals in brain. These results suggest that synaptojanin 2B may have overlapping functions with synaptojanin 1 at synapses and may at least in part compensate for the lack of synaptojanin 1 function in synaptojanin 1-deficient neurons (22). In addition, we show that synaptojanin 2B is also abundantly expressed in adult testis, where it is associated with the spermatid manchette, suggesting that synaptojanin 2B may take part in the germ cell morphological differentiation during spermiogenesis. We also provide evidence that active Rac1 GTPase specifically affects the intracellular localization of various forms of synaptojanin 2, but not of synaptojanin 1.

EXPERIMENTAL PROCEDURES
Identification of Human and Rat Synaptojanin 2B cDNAs-Our previous Northern blot analysis of synaptojanin 2 detected multiple transcripts in various tissues and indicated the presence of multiple alternatively spliced forms of synaptojanin 2 (28). BLAST (32) searches against the sequence data bases revealed the presence of a cDNA sequence derived from human brain highly homologous to rat synaptojanin 2A (KIAA0348, kindly provided by Dr. T. Nagase, Kazusa DNA Research Institute, Kisarazu, Japan) (33). The major difference between rat synaptojanin 2A and the predicted protein from human brain KIAA0348 cDNA is the assigned initiator methionine. Although the 5Ј-portion of the KIAA0348 sequence is highly homologous to the corresponding region of rat synaptojanin 2A and could potentially encode a large part of the putative Sac1 domain, the translation was terminated by a stop codon at nucleotides 971-973. The methionine codon at nucleotides 1049 -1051 was predicted to be the initiator codon for the longest possible open reading frame in KIAA0348, which thus apparently encodes a protein of 1113 amino acids with a truncated Sac1 domain followed by an inositol 5Ј-phosphatase domain and a prolinerich region. To verify the presence of the messenger RNA for the human synaptojanin 2-related protein lacking the Sac1 domain, we synthesized two oligonucleotide primers based on KIAA0348 sequence flanking the intervening termination codon, 5Ј-CACCAGGTGAGCTGCTG-3Ј, corresponding to nucleotides 359 -375, and 5Ј-CTCGAGCGCGATGAAGCT-3Ј, corresponding to the complement of nucleotides 1091-1108, and performed PCR on the cDNAs reverse-transcribed from human lung, peripheral leukocyte, placenta, or human brain. A single major product of 0.7 kilobases was amplified from all the tissue cDNAs used as template. The PCR product from each tissue source was subcloned into the pCR2.1 vector (Invitrogen) and sequenced. All these products were found to be identical to the corresponding region of KIAA0348, except for the absence of 57 base pairs corresponding to nucleotides 970 -1026 of KIAA0348. By inspection of the sequence, we noticed that KIAA0348 has direct repeats of 12 base pairs in nucleotides 958 -969 and nucleotides 1015-1026. Based on these results, we concluded that KIAA0348 has an artifactual insertion in nucleotides 970 -1026 presumably arising from an aberrant rearrangement or duplication during the cloning procedure. After removing the artifactual insertion, the corrected KIAA0348 sequence (GenBank TM /EBI accession number AF039945) was then found to encode a protein of at least 1443 amino acids with an N-terminal Sac1 domain, although the open reading frame is incomplete at the 5Ј-end and lacks the region corresponding to the most amino-terminal 50 amino acids.
A 2.1-kilobase XbaI fragment from KIAA0348 cDNA spanning the region encoding the proline-rich sequence in human synaptojanin 2B was radiolabeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) and used as a probe for screening a rat brain ZAPII cDNA library (Stratagene) following standard techniques to obtain rat synaptojanin 2B cDNA (34). The screening yielded cDNAs for two forms of rat synaptojanin 2B, designated synaptojanins 2B1 and 2B2. Accordingly, we refer to a protein encoded by the corrected KIAA0348 cDNA as human synaptojanin 2B2 (see Fig. 1).
Bacterial Fusion Protein Production-The pMalC2 constructs for production of the maltose-binding protein (MBP)-and His 6 -tagged proline-rich carboxyl-terminal regions of rat synaptojanins 1 and 2A (amino acids 1015-1317 of synaptojanin 1 and amino acids 1030 -1248 of synaptojanin 2A) were as described (28). The DNA fragments encoding the proline-rich regions of rat synaptojanins 2B1 and 2B2 (amino acids 1030 -1451 of synaptojanin 2B1 and amino acids 1030 -1496 of synaptojanin 2B2) were generated by PCR on the cDNAs. They were then subcloned into the pMalC2 vector (New England Biolabs Inc.) to produce the dually tagged proline-rich regions of synaptojanins with MBP at the N termini and His 6 at the C termini as described (28). The MBPand His 6 -tagged proteins were expressed in Escherichia coli BL21 and purified by two consecutive affinity chromatography steps on amylose resin (New England Biolabs Inc.) and nickel-nitrilotriacetic acid-agarose (QIAGEN Inc.) following the manufacturers' instructions.
Production of Synaptojanin Isoform-specific Antibodies-A mouse monoclonal antibody against synaptojanin 1 (AC1) and a rabbit polyclonal antibody against synaptojanin 2A were as described (9,28). A portion of the synaptojanin 2B-specific C-terminal region (amino acids 1210 -1346 and 1255-1391 in rat synaptojanins 2B1 and 2B2, respectively) was amplified by PCR on synaptojanin 2B2 cDNA and subcloned into pGEX4T-1 and pMalC2. The GST fusion protein from the pGEX construct was expressed in E. coli DH5␣ and purified by glutathione-Sepharose 4B column chromatography following standard procedures. The GST fusion protein was used to immunize two rabbits to generate anti-synaptojanin 2B antisera. The MBP fusion protein from the pMal construct was expressed in E. coli DH5␣ and purified on amylose resin. The MBP fusion protein was run on a 12% SDS-polyacrylamide gel and transferred onto a polyvinylidene difluoride membrane (Immobilon, Millipore Corp., Bedford, MA). The membrane was stained with Ponceau S (Sigma), and the band containing the MBP fusion protein was excised and used for affinity purification of the antibody from the antisera as described (37,38).
Protein Binding Experiments-Blot overlay assays with GST fusion proteins were performed as previously described (35). The PCR-generated DNA fragment encoding the proline-rich regions of rat synaptojanin 2B2 (amino acids 1030 -1496) was subcloned into pBTM116 to express the LexA fusion protein in the yeast strain L40 (39). The pBTM116 constructs for expression of the LexA fusion proteins of the C-terminal regions of rat synaptojanins 1 and 2A were as described (14,30). The yeast two-hybrid interaction assay was carried out using these plasmids as described (30).
Affinity Chromatography-The peptides B2N (CDDADLMTLKLEL-EVAGNFRHRSP) and B2C (CSRSLSVPNRPRPPHPPQRPPPPT), corresponding to the N-and C-terminal 23 amino acids of a 2B2-specific insert sequence in rat synaptojanin 2B2 with an additional N-terminal cysteine residue (see Fig. 3A), respectively, were synthesized and coupled to SulfoLink coupling gel (Pierce) via the N-terminal cysteine residues according to the manufacturer's instructions. Preparation of rat brain Triton X-100 extract and affinity chromatography on the peptides coupled to gel beads were performed as described (30).
Preparation and Immunoblot Analysis of Germ Cell and Sertoli Cell Primary Cultures from Rat Testis-Testes from 21-day-old Wistar rats were dissected, and Sertoli cells and germ cells were isolated and grown in culture according to established procedures (40). Cells at 3 days in culture were harvested and solubilized in SDS sample buffer (41). Aliquots of each sample (30 g of protein) were subjected to SDSpolyacrylamide gel electrophoresis and immunoblotting with rabbit polyclonal anti-synaptojanin 2B antibody (1:200). The immunoreactive bands were visualized using horseradish peroxidase-conjugated goat secondary antibody (Pierce) and an enhanced chemiluminescent detection kit (Amersham Pharmacia Biotech).
Expression of Synaptojanin Isoforms and Rho/Rac GTPases in Mammalian Cells-Full-length cDNA for each rat synaptojanin isoform was subcloned into the pEGFP-C1 vector (CLONTECH) by a PCR-mediated procedure to express the N-terminally GFP-tagged protein in mammalian cells. RhoA and Rac1 cDNAs were amplified from human placenta by reverse transcriptase-PCR. The cDNAs encoding constitutively GTPbound forms of RhoA (G14V) and Rac1 (G12V) were generated by PCR on RhoA and Rac1 cDNAs, respectively. The cDNAs encoding native and active forms of RhoA and Rac1 were subcloned into the pHA vector (30) to express HA epitope-tagged proteins in mammalian cells. Trans-fection and immunofluorescence analysis of COS-7 cells were performed as described (42).
Production of the Recombinant Sac1 Domain of Rat Synaptojanin 2-The region coding the Sac1 homology domain of rat synaptojanin 2 (amino acids 1-522) was amplified by PCR and subcloned in a modified pFastBac1 baculovirus transfer vector (Life Technologies, Inc.) downstream of the coding sequence of GST (43). The point mutations corresponding to sac1-8, sac1-10, and sac1-22 alleles (44) were introduced FIG. 1. Primary structures of rat synaptojanin 2B isoforms. A, amino acid sequence comparison of synaptojanin isoforms. Amino acid sequences of rat synaptojanin 1 (SJ1; 145-kDa form) and rat and human 2B2 (SJ2B2) isoforms were aligned by the Clustal method (78). The 46-amino acid sequence present in synaptojanin 2B2 but missing in synaptojanin 2B1 is underlined. Amino acid residues are numbered on the left. The sequence data for rat synaptojanins 2B1 and 2B2 and human synaptojanin 2B2 are available from the DDJB/GenBank TM /EMBL Data Bank under accession numbers AY034050, AY034051, and AF039945, respectively. B, schematic representation of the domain structures of synaptojanin isoforms. Note that the most C-terminal portion of synaptojanin 2A is replaced in synaptojanin 2B by a longer proline-rich sequence and that the 2B2 form of synaptojanin has a region missing in the 2B1 form.
Measurement of Phosphoinositide Phosphatase Activity-Preparation of 3 H-labeled phosphoinositides and measurement of phosphoinositide phosphatase activity were performed as described (45,46).
Immunofluorescent Microscopy of Rat Brain and Testis-Adult male Wistar rats were fixed by intracardiac perfusion of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Cryostat sections (10-m thickness) were cut from the brain and testis. The sections were immunostained as described (47,48) using rabbit polyclonal anti-synaptojanin 2B antibody, mouse monoclonal antibody against amphiphysin 1 (clone 3) (49), or mouse monoclonal anti-␣-tubulin antibody (Sigma) as the primary antibody and rhodamine-conjugated goat anti-rabbit IgG or Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes, Inc., Eugene, OR) as the secondary antibody. Where indicated, F-actin was detected using Alexa 488-conjugated phalloidin (Molecular Probes, Inc.). Tissue sections were examined through an Axiophot epifluorescent microscope or an LSM 510 confocal laser scanning microscope (both form Carl Zeiss, Inc.).
Miscellaneous Procedures-Procedures for SDS-polyacrylamide gel electrophoresis and immunoblotting were as described (41,50). Protein concentration was determined by the method of Bradford (51) with bovine serum albumin as a standard. Standard techniques were used for recombinant DNA manipulations (34). PCR was performed using Vent polymerase (New England Biolabs Inc.) following the manufacturer's instructions. Mouse monoclonal anti-HA antibody was obtained from BabCo (Richmond, CA) and used for detection of the HA epitope. Rabbit polyclonal antibodies against endophilins 1 and 2 were as described (14). Oligonucleotide and peptide synthesis and nucleotide sequencing were performed at the Yale Keck Biotechnology Facility and the RIKEN BSI Advanced Technology Development Center.

Molecular Cloning of Two Forms of Synaptojanin 2B-We
conducted data base searches and PCR-mediated screening as described under "Experimental Procedures" and have identified two alternatively spliced forms of rat synaptojanin 2 distinct from synaptojanin 2A. We shall refer to these two forms of synaptojanin 2 as synaptojanins 2B1 and 2B2. As deduced from the cDNA sequences, rat synaptojanins 2B1 and 2B2 have 1451 and 1496 amino acids with predicted molecular masses of 160 and 165 kDa, respectively. We have previously shown that synaptojanin 2A has a C-terminal motif that binds to the PDZ domain of OMP25 (27). In the synaptojanin 2B isoforms, the C-terminal end of rat synaptojanin 2A corresponding to amino acids 1209 -1248 is replaced by a 248-amino acid sequence with multiple potential SH3 domain-binding regions, and the PDZ domain-binding motif is absent. The difference between the 2B1 and 2B2 forms is the absence in 2B1 and the presence in 2B2 of a 46-amino acid sequence. Based on the sequence com-  (30) or the SH3 domain of rat endophilin 1 (SH3p4) (lower half) (14). Activation of HIS3 by the interactions was assayed for growth on the medium lacking histidine.

TABLE I
Binding properties of synaptojanin isoforms Protein interactions for the C-terminal regions of synaptojanin isoforms were determined in this report and our previous studies (10,14,30,35). Positive and negative interactions are denoted by ϩ and Ϫ, respectively.
Synaptojanin isoform parison, an alternatively spliced form of murine synaptojanin 2 previously reported (29) is likely to correspond to synaptojanin 2B1. Likewise, the human KIAA0348 cDNA (33) may be derived from a human homolog of synaptojanin 2B2 (Fig. 1). Protein Binding Properties of the Proline-rich Regions of Synaptojanin Isoforms-Several SH3 domain-containing proteins such as GRB2, amphiphysins 1 and 2, and SH3p4/8/13 (endophilins 1-3, respectively) have been identified as binding proteins for the proline-rich region of synaptojanin 1 (145-kDa form) (10 -14, 35). The EH domains of EPS15 were shown to bind to the multiple NPF motifs in the extended proline-rich region of the 170-kDa form of synaptojanin 1 (9,52,53). Recently, the most carboxyl-terminal sequence of synaptojanin 2A was found to associate with the PDZ domain of OMP25 (30).
Interaction of the C-terminal portions of various forms of synaptojanin with these proteins is thought to regulate their intracellular localization and hence the phosphoinositide pools on which they act and the cellular processes in which they participate (for recent reviews, see Refs. 20 and 21). As an initial step to address the possible function of synaptojanin 2B, we therefore tested the protein binding properties of the proline-rich regions of the synaptojanin 2B isoforms in comparison with those of synaptojanins 1 and 2A.
As shown in Fig. 2A, GST fusion proteins of the SH3 domains of amphiphysins 1 and 2 bound to the recombinant C-terminal regions of rat synaptojanins 1 (145-kDa form), 2B1 and 2B2, whereas GST fusion proteins of the SH3 domains of endophilins 1-3 bound to the C-terminal regions of synaptojanins 1 and 2B2, but not of synaptojanin 2B1, by blot overlay method. Consistent with the absence of NPF motifs in the proline-rich regions of synaptojanin 1 (145-kDa form), 2A, 2B1, and 2B2, none of these proteins showed a detectable binding to a GST fusion protein of the EH domains of EPS15 under the conditions employed. We also tested the interaction between the proline-rich C-terminal region of synaptojanin 2B2 and OMP25 by the yeast two-hybrid assay. No activation of the HIS3 reporter gene was detected in the yeast strain carrying the synaptojanin 2B2 and OMP25 constructs (Fig. 2B, upper half), whereas activation was observed in the strain carrying the synaptojanin 2B2 construct together with the SH3 domain of endophilin 1 (lower half). Based on these and the previous results (see Fig. 2 in Ref. 30), we concluded that neither synaptojanin 1 nor synaptojanin 2B interacts with OMP25, as expected from the absence in synaptojanins 1 and 2B of the C-terminal motif for OMP25 PDZ domain binding (Fig. 2B). As summarized in Table I, the protein binding properties of the synaptojanin 2B forms are very different from that of synaptojanin 2A and are similar to that of synaptojanin 1.
Synaptojanin 2B2 Has an Endophilin-binding Motif-We (24, 54) 3 and others (79) have recently shown that endophilin specifically binds synaptojanin 1 through the interaction of the SH3 domain of endophilin with a short amino acid sequence in the C-terminal region of synaptojanin 1. Inhibition of the interaction in living synapses results in remarkable abnormalities in synaptic morphology and function, demonstrating its physiological importance (24,54). Protein binding experiments have shown that the SH3 domains of endophilins specifically bind to synaptojanin 2B2, although the two forms of synaptojanin 2B differ only with regard to the presence or absence of a 3

FIG. 3. Synaptojanin 2B2 has an endophilin-binding motif. A, synaptojanin (SJ) 2B2-specific insert sequences from rat and human synaptojanin 2B2
and an endophilin-binding motif in rat synaptojanin 1 (PP19) (24, 54) are aligned. The shared amino acid residues are shown by black boxes. The synthetic peptides B2N and B2C cover the N-and C-terminal halves of the rat synaptojanin 2B2-specific insert sequence, respectively, as shown at the top. B, endophilins were affinity-purified by the peptide within the synaptojanin 2B2-specific sequence. A Triton X-100 extract of rat brain was affinity-purified onto the column conjugated with either the N-terminal half (B2N) or the C-terminal half (B2C) of the synaptojanin 2B2-specific insert sequence. Bound proteins were released by Laemmli SDS sample buffer (41). Rat brain extract and the eluate from the peptide columns were either stained with Coomassie Brilliant Blue (left panel) or analyzed by immunoblotting with antiendophilin 1 antibody (middle panel) or anti-endophilin 2 antibody (right panel). Molecular mass markers are indicated in kilodaltons on the left. The 55-kDa protein in both the B2N and B2C eluates was purified by the unconjugated control column and thus was considered nonspecific. 46-amino acid sequence (Fig. 1). The C-terminal part of the 2B2-specific region is rich in proline residues and bears a striking similarity to the endophilin-binding motif of synaptojanin 1 (Fig. 3A) (24,54,79). 2 To test the protein binding property of the putative endophilin-binding site in synaptojanin 2B2, two peptides, B2N and B2C, corresponding to the N-terminal 23 amino acids and C-terminal 23 amino acids of the 2B2-specific insert sequence, respectively, were synthesized and coupled to gel beads via the N-terminal ends. The beads were incubated with rat brain extract, and the bound protein was eluted by SDS sample buffer. As shown in Fig. 3B, peptide B2C specifically purified 40-and 45-kDa proteins from rat brain, which were recognized by anti-endophilin 1 and anti-endophilin 2 antibodies, respectively. No binding was detected between peptide B2N and the endophilins under the conditions used. These results indicate that rat synaptojanin 2B2 interacts with the SH3 domains of the endophilins via the endophilin-binding site present in the C-terminal part of the 2B2-specific sequence. We also noted that comparison of the deduced amino acid sequences of human and rat synaptojanin 2B2 indicates that within the 2B2-specific region, the C-terminal half is more highly conserved than the N-terminal half (Fig. 3A). This sequence conservation may further imply the functional importance of the endophilin-binding site in the synaptojanin 2B2 form.
Phosphoinositide Phosphatase Activity of Synaptojanin 2B-The key characteristic of synaptojanins is the presence of a central domain homologous to inositol 5-phosphatase and an N-terminal domain homologous to the product of the yeast SAC1 gene. These two domains are identical in synaptojanins 2A and 2B because they are upstream of the region of alternative splicing (Fig. 1B). The 5-phosphatase domain, like the 5-phosphatase domain of synaptojanin 1 (3,46,55), was previously shown to be enzymatically active specifically on the D5 phosphoryl group of inositol polyphosphates, PI(4,5)P 2 and PI(3,4,5)P 3 (28,29).
The budding yeast SAC1 gene was initially identified as a gene whose mutation displays an allele-specific suppression of yeast actin mutations and a bypass suppression of mutations in the SEC14 gene. The SEC14 gene encodes a major phosphatidylinositol/phosphatidylcholine transfer protein in budding yeast (1,2,56). Guo et al. (46) demonstrated that the Sac1 domains from yeast Sac1p, Inp52p, and Inp53p and human synaptojanin 1 exhibit a potent phosphoinositide phosphatase activity that hydrolyzes phosphate from PI(3)P, PI(4)P, and PI(3,5)P 2 , but not from PI(4,5)P 2 . Our recent work shows that Radiolabeled phosphoinositides were incubated for 1 h at 37°C with the indicated form of the purified proteins (10 ng each; wild-type (WT), D388N, G435V, and R460H) or without any Sac1 protein (Control). The reaction products were extracted, deacylated, and resolved by high performance liquid chromatography (45). The indicated values represent the levels of the remaining substrates of three independent experiments (mean Ϯ S.D.). The inset shows the levels of PI(3,5)P 2 on an expanded scale. Sac1 proteins are widely conserved in eukaryotic genera (45). The recombinant Sac1 domain from rat Sac1 has phosphoinositide phosphatase activity with the same unique substrate specificity as yeast Sac1p, and the enzyme activity is similarly affected by the point mutations corresponding to those identified in yeast sac1 alleles (44). Expression of rat Sac1 in the yeast strain deleted of SAC1 relieved various phenotypes associated with loss of Sac1p function, and PI(4)P phosphatase activity was found to be necessary for rescue. This observation demonstrates that PI(4)P phosphatase activity represents a major aspect of the physiological function of yeast Sac1p and mammalian Sac1.
As the catalytic activity of the Sac1 domain of synaptojanin 2 has not yet been investigated, we examined its phosphoinositide phosphatase activity and the effects of mutations corresponding to yeast sac1 alleles. We introduced into the Sac1 domain of rat synaptojanin 2 (amino acids 1-522) D388N, G435V, and R460H mutations, corresponding to yeast sac1-8, sac1-10, and sac1-22 alleles, respectively (Fig. 4A) (44), and expressed native and mutant forms of the Sac1 domain of rat synaptojanin 2 in insect cells. The phosphatase activity of the purified recombinant proteins was measured using labeled phosphoinositides as substrate. As shown in Fig. 4B, the Sac1 domain of rat synaptojanin 2 exhibited phosphoinositide phosphatase activity capable of hydrolyzing phosphate from PI(3)P, PI(4)P, and PI(3,5)P 2 in vitro. PI(4,5)P 2 was not hydrolyzed under the conditions used. The phosphatase activity of the D388N mutant protein was barely detectable. The phosphatase activity was reduced to a lesser extent by the G435V mutation and not affected appreciably by the R460H mutation. These results indicate that the Sac1 domain in rat synaptojanin 2 has phosphoinositide phosphatase activity with the same substrate spectrum as the Sac1 domains of other proteins so far reported. The yeast sac1-like mutations produce very similar effects on the phosphatase activity of the Sac1 domains from rat Sac1 and synaptojanin 2 (compare with Fig. 9B in Ref. 45). It is of note that in both Sac1 domains from rat Sac1 and synaptojanin 2, the mutation corresponding to the sac1-8 allele, replacement by an asparagine of an aspartic acid residue conserved in various Sac1 domains (57), has the most pronounced effect and yields catalytically inactive proteins. This finding may be useful for the characterization of other Sac1 domain-containing proteins of unknown function (57).
Rat Synaptojanin 2B Is Concentrated at Nerve Terminals in Brain and at the Spermatid Manchette in Testis-To characterize the tissue and subcellular distribution of the synaptojanin 2B isoforms, we raised an antibody against the B-specific proline-rich C-terminal region (Fig. 1B). The reactivity of this antibody with the synaptojanin 2B forms was confirmed by its recognition of rat synaptojanin 2B fusion proteins produced in E. coli. The anti-synaptojanin 2B antibody was immunoreactive with the synaptojanin 2B forms and did not display crossreactivity with either synaptojanin 1 or 2A (data not shown). This antibody was therefore used to determine the tissue distribution of rat synaptojanin 2B by immunoblot analysis of various adult rat tissues. A 160-kDa band was detected primarily in protein extracts from rat brain and testis, in good agreement with the predicted molecular mass of synaptojanin 2B (Fig. 5).
We next used this antibody to investigate the cellular and subcellular localization of synaptojanin 2B in these tissues by indirect immunofluorescence. Following immunostaining of rat brain sections, synaptojanin 2B immunoreactivity was very similar to that produced by amphiphysin 1, an endocytic protein in presynaptic nerve terminals (Fig. 6), as well as by a variety of antibodies directed against other proteins localized primarily in the presynaptic compartment (data not shown). These results suggest an overlapping localization of synaptojanins 2B and 1 (3,8). Moreover, they speak to a physiological significance of the interaction of the synaptojanin 2B isoforms with amphiphysin and endophilin because the three proteins are concentrated in the same compartment (10,13,14,49,58).
In testis, synaptojanin 2B was found to be partially associated with the bundles of microtubules surrounding the nucleus in the elongating spermatids in the seminiferous epithelium (Fig. 7). This microtubular array corresponds to the manchette, a transient structure implicated in nuclear shaping in the elongating spermatids during spermiogenesis (Ref. 59 and references therein). Interestingly, synaptojanin 2B was not located at the axoneme, another prominent microtubular structure in spermatids (data not shown). To further confirm that synaptojanin 2B is a component of the manchette, we performed immunoblot analysis of Sertoli cell and germ cell primary cultures from rat testis (40). Anti-synaptojanin 2B antibody detected the 160-kDa band in protein extracts from germ cells, but produced no detectable signal on blots of extracts from Sertoli cells (data not shown), thus supporting the results of the immunofluorescence study.
Rac1 GTPase Regulates the Intracellular Localization of Synaptojanin 2 Isoforms-While our characterization of the synaptojanin 2B isoforms was in progress, Malecz et al. (60) reported that the GTP-bound form of the Rac1 GTPase, known to be implicated in endocytosis (61), interacts with human synaptojanin 2 (2B2 according to our nomenclature) and thereby causes the translocation of synaptojanin 2B2 from the cytosol to the plasma membrane. Prompted by their finding, we investigated the effects of Rac1 on the subcellular localization of various synaptojanin isoforms by transient expression experiments in COS-7 cells. In transfected cells, synaptojanin 2B2 was localized diffusely in the cytoplasm when expressed alone (Fig. 8J) or with the native form of Rac1 (Ref. 60 and data not shown). As reported by Malecz et al. (60), synaptojanin 2B2 translocated to membrane ruffles upon coexpression with the constitutively GTP-bound form of Rac1 (Fig. 8, K and L). The activated form of Rac1 exerted a similar effect on the intracellular distribution of other synaptojanin 2 variants (Fig. 8, D-I). FIG. 5. Rat synaptojanin 2B is expressed in brain and testis. The tissue distribution of rat synaptojanin 2B was determined by Western blotting of total protein extracts (50 g each) from various adult rat tissues using anti-rat synaptojanin 2B antibody. Molecular mass markers are indicated in kilodaltons on the left. Note that synaptojanin 2B immunoreactivity (160-kDa band) was predominantly expressed in brain and testis. The amounts of lower molecular mass immunoreactive proteins varied among sample preparations, and they are likely to represent degradation products of the 160-kDa protein. In marked contrast, the intracellular distribution of synaptojanin 1 (145-kDa form) was unaffected by coexpression of either form of Rac1 GTPase (Fig. 8, A-C). Lamaze et al. (61) reported that activated forms of Rac1 and RhoA GTPases similarly inhibit transferrin receptor-mediated endocytosis in HeLa and A431 cells. We therefore performed similar coexpression experiments in COS-7 cells using native and constitutively active forms of RhoA GTPase and detected no obvious effects on the subcellular localization of any form of synaptojanin tested (data not shown). These results confirm and extend the finding by Malecz et al. (60) and indicate that Rac1 may be a specific regulator of the intracellular localization of various forms of synaptojanin 2, but not of synaptojanin 1 (145-kDa form), the major synaptojanin isoform at nerve terminals of the adult brain (55). DISCUSSION Our previous studies of the properties of mammalian synaptojanin 1 and the generation and analyses of synaptojanin 1-deficient mice (22) have demonstrated that phosphoinositide metabolism by synaptojanin 1 at nerve terminals is essential for normal synaptic vesicle recycling by clathrin-mediated endocytosis and proper synaptic transmission. Similar results were obtained in C. elegans, where mutations in unc-26, which codes for the single synaptojanin in this organism, affect synaptic vesicle recycling at multiple stages (23). In all these mutant animals, synaptic vesicle recycling is impaired, but not eliminated. We also observed that in the giant synapses of lamprey, an acute loss of function in synaptojanin causes not only a partial block of synaptic vesicle recycling, but also anom-FIG. 6. In brain, synaptojanin 2B is localized at nerve terminals. Double immunofluorescence micrographs of rat brain sections stained with a rabbit polyclonal antibody for synaptojanin 2B (A, C, and E) and a mouse monoclonal antibody for amphiphysin 1 (B, D, and F) are shown. Immunoreactivity appears as white. A/B and C/D show low power views of the brain stem and cerebral cortex, respectively. E and F show a region of the brain stem at higher magnification. Note in A/B and E/F the bright spots outlining neuronal cell bodies and proximal dendrites. These spots represent immunoreactive axon terminals. Note in C and D the negative profile of a pyramidal cell neuron (center of the two fields), completely surrounded by small immunoreactive puncta, which reflect the close apposition of synapses in this brain region. Other negative profiles represent additional neuronal cell bodies and blood vessels. The scale bar corresponds to 19 m in A and B, 25 m in C and D, and 10 m in E and F.

FIG. 7. In testis, synaptojanin 2B is localized at the spermatid manchette.
In a-c, the rat testis section was doublestained with a rabbit polyclonal antibody for synaptojanin 2B (red) (SJ2B; a) and a mouse monoclonal antibody for ␣-tubulin (green) (b), and both images were merged in c. In d-f, the section was stained with a rabbit polyclonal antibody for synaptojanin 2B (red) (d), and F-actin was visualized with Alexa 488-conjugated phalloidin (green) (e). Both images were merged in f. Fluorescence was detected by confocal laser scanning microscopy. Note the colocalization of synaptojanin 2B immunoreactivity with the tubulin immunoreactivity surrounding the nuclei in elongating spermatids corresponding to the manchettes (arrowheads), but not with the tubulin immunoreactivity extending into the flagella corresponding to the axonemes or with F-actin. The scale bar corresponds to 10 m. aly in actin organization as well (24). In the present report, we have described the properties of synaptojanin 2B, which might be involved in the regulation of neuronal functions in conjunction with synaptojanin 1.
The colocalization at nerve terminals of synaptojanins 1 and 2B (Fig. 5) (8,9), their similar biochemical and protein interaction properties such as binding to amphiphysins and endophilins (Figs. 2 and 3) (10 -14), and the virtually indistinguishable profiles of phosphoinositide phosphatase activities of their Sac1 homology domains (Fig. 4) (46) and 5-phosphatase domains (3,28,29,55) suggest that these isoforms may have overlapping functions at nerve terminals. It is hence conceivable that in the synaptojanin 1-deficient neurons, synaptojanin 2B could compensate for certain functions of synaptojanin 1 and could at least in part account for the residual synaptic vesicle recycling capacity (22). We have previously shown that synaptojanin 2B is indeed expressed in synaptojanin 1-deficient neurons (see Fig. 1C in Ref. 22). On the other hand, there are also some important differences between the properties of synaptojanins 1 and 2B, suggesting that synaptojanin 2B may have additional roles in non-neuronal cells and presumably in neurons under physiological conditions. This view is supported by recent studies of the synaptojanin-like gene products in Saccharomyces cerevisiae. In this budding yeast, three synaptojanin-like gene products have both overlapping and distinct functions (5)(6)(7)(62)(63)(64)(65).
Most significantly, transient expression experiments (Fig. 8) indicate that Rac1 specifically affects the intracellular distribution of synaptojanin 2 forms, but not of synaptojanin 1, in a GTP-dependent manner (our results and Ref. 60). It was hypothesized that the inhibitory effect of activated Rac1 on endocytosis (61) may be attributed to the translocation of synaptojanin 2B2 to the plasma membrane and to the resulting degradation of phosphoinositides (60), which are required for clathrin coat assembly (66, 67) and actin nucleation (68, 69).
However, it was shown that activated forms of both Rac1 and RhoA GTPases similarly inhibit transferrin receptor-mediated endocytosis in HeLa and A431 cells (61), suggesting that other mechanisms must come into play. The selective interaction of synaptojanin 2B with Rac1 implies that the intracellular localizations of synaptojanins 1 and 2B might be regulated by different mechanisms at nerve terminals. Thus, either synaptojanins 1 and 2B may act on distinct phosphoinositide pools within nerve terminals, or they may act on the same phosphoinositide pools, but in response to different signals. Multiple synaptojanins may be required to attain the fine-tuning of phosphoinositide levels that regulate neuronal functions.
In addition to the predominant localization at nerve terminals in brain, synaptojanin 2B was found to exhibit an unexpected localization in adult testis at the spermatid perinuclear manchette. The manchette is thought to be involved in the nuclear shaping of spermatids during spermiogenesis (Refs. 70 and 71; for a review, see Ref. 59). It is therefore of interest that Seet et al. (31) and Cremona et al. (72) mapped murine and human synaptojanin 2 genes to the chromosomal regions where several genes implicated in spermiogenesis are localized. Since we failed to detect binding of synaptojanin 2B to microtubules by in vitro protein binding experiments (data not shown) and by morphological analysis of transfected cells (Fig. 8), the association of synaptojanin 2B with the manchette is likely to be mediated by its binding to microtubule-associated constituents. We speculate that synaptojanin 2B interacts with as yet unidentified partners in developing spermatids and acts on a pool of phosphoinositides that affect the organization of their cytoskeleton. Interestingly, both amphiphysin 1 (49) and endophilin 3 (SH3p13) (14,73), which have been shown here to bind to synaptojanin 2B2, are also abundantly expressed in testis, although their subcellular localization has not yet been determined.
Our present observations further substantiate the notion that alternative splicing generates various forms of synaptoja- nin with different C-terminal regions and therefore unique protein binding properties, thus targeting the synaptojanin isoforms to specific subcellular compartments. These different isoforms act on phosphoinositide pools implicated in the regulation of diverse cellular processes, including cytoskeletal organization and membrane trafficking (for reviews, see Refs. 25, 27, and 74). Additionally, our results indicate that regulation of the intracellular distribution by Rac1 GTPase is a shared but specific property of synaptojanin 2 isoforms (Fig. 8). It will be important to explore the possible involvement of these synaptojanin isoforms in many biological processes in which Rac GTPases have been reported to participate (for recent reviews, see Refs. [75][76][77].