The Identification of Two Drosophila K Homology Domain Proteins

Sam68 is a member of a growing family of RNA-binding proteins that contains an extended K homology (KH) domain embedded in a larger domain called the GSG (GRP33,Sam68, GLD1) domain. To identify GSG domain family members, we searched data bases for expressed sequence tags encoding related portions of the Sam68 KH domain. Here we report the identification of two novel Drosophila KH domain proteins, which we termed KEP1 (KH encompassingprotein) and SAM. SAM bears sequence identity with mammalian Sam68 and may be the Drosophila Sam68 homolog. We demonstrate that SAM, KEP1, and the recently identifiedDrosophila Who/How are RNA-binding proteins that are able to self-associate into homomultimers. The GSG domain of KEP1 and SAM was necessary to mediate the RNA binding and self-association. To elucidate the cellular roles of these proteins, SAM, KEP1, and Who/How were expressed in mammalian and Drosophila S2 cells. KEP1 and Who/How were nuclear and SAM was cytoplasmic. The expression of KEP1 and SAM, but not Who/How, activated apoptotic pathways inDrosophila S2 cells. The identification of KEP1 and SAM implies that a large GSG domain protein family exists and helps redefine the boundaries of the GSG domain. Taken together, our data suggest that KEP1 and SAM may play a role in the activation or regulation of apoptosis and further implicate the GSG domain in RNA binding and oligomerization.

The K homology (KH) 1 domain is a protein module consisting of 70 -100 amino acids that was originally identified as a repeated sequence in heterogeneous nuclear ribonucleoprotein K (1). The KH domain is an RNA binding motif that is thought to make direct protein-RNA contacts with a three-dimensional ␤␣␣␤␤␣-fold (2). Alignment of KH domains from various proteins (2, 3) reveals a subfamily of KH domains including mammalian Sam68 (4,5), Artemia salina GRP33 (6), Caenorhabditis elegans GLD-1 (7), mouse Qk1 (8), Xenopus Qk1 (9), mammalian SF1 (10), and the Drosophila Who/How (11)(12)(13). All these proteins contain a single extended KH domain included within a larger protein domain of ϳ200 residues called the GSG (GRP33, Sam68, GLD-1) domain (7). This domain is also called STAR for signal transduction and activator of RNA (14). The KH domain embedded in the GSG domain is approximately 26 amino acids longer than other KH domains with extra amino acids located in loops 1 and 4 (2). Although the function of the GSG domains remains unknown, this protein module has been shown to be required for self-association and RNA binding (15)(16)(17)(18).
There is considerable genetic evidence from various species supporting the physiological role of the KH domain. In humans, gene lesions that prevent expression of the KH protein FMR1 result in the fragile X mental retardation syndrome (19,20), the most common form of heritable mental retardation (21). The particular significance of the KH domain was implicated by a point mutation altering a conserved isoleucine 304 to asparagine in the second KH domain of FMR1 (22). The point mutation alters the structure of the KH domain (2) and severely impairs RNA binding activity (23). In C. elegans, GLD-1 is a cytoplasmic protein required for germ cell differentiation (24 -26) and alteration of glycine 227, within the KH domain, results in a recessive tumorous germ line phenotype (7). The structure of the KH domain predicts that this conserved glycine forms part of the RNA binding surface (2) and mutation of the corresponding residue in Sam68 abolishes RNA binding (16). In mice, the quaking viable mutation severely impairs myelination and as a result, mice develop a rapid tremor at postnatal day 10 (27). A missense mutation in the GSG domain of Qk1 is embryonic lethal (8). This point mutation has been shown to prevent homodimerization and may be the reason for the lethality observed in mice (18). The Drosophila Bicaudal C (Bic-C) contains five KH domains and gene lesions that truncate the Bic-C protein or a point mutation that replaces glycine 295 with an arginine in the third KH domain are strong alleles, leading to defects in RNA binding, oogenesis, and anteriorposterior embryonic patterning (28,29). Drosophila Who/How is expressed in muscle and animals that possess weak mutations either die or survive as adults with defects in wing position (11)(12)(13). A point mutation in the KH domain of Who/How has been identified and results in flies with "wings held out" (11).
In this study, we report the cloning of two new Drosophila GSG domain family members that we have called SAM and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  KEP1. SAM bears homology to the mammalian Sam68 and may be its homolog, whereas KEP1 is a unique family member. SAM is a cytoplasmic protein expressed in early embryos and female flies. KEP1 is a nuclear protein that is expressed at all stages of Drosophila development. KEP1 and SAM are RNAbinding proteins that bind homopolymeric RNA in vitro and self-associate into multimers. We further show that extracts from KEP1-and SAM-transfected Drosophila S2 cells trigger chromatin condensation and apoptotic body formation, in vitro events known to be caspase-dependent (30,31).

EXPERIMENTAL PROCEDURES
DNA Constructs and Cloning of KEP1, SAM, and Who/How-To obtain a full-length KEP1 cDNA clone, we screened a 0 -2-h embryonic Drosophila library constructed in the ZAP vector. A 600-base pair BamHI fragment isolated from expressed sequence tag LD14468 (Gen-Bank accession AA439802, Genome Systems) was used as a probe. The longest clone, 9 -12, was obtained and contained a 1.4-kb insert. The coding region of this clone was PCR-amplified using oligonucleotide 5Ј-CGC GAA TTC GAT AAA AAT GGA AAC CCC AAG C-3Ј and T7 promoter primer. The DNA fragment was digested with EcoRI and XhoI (located at the 3Ј of the insert) and subcloned in the corresponding sites in HA-and Myc-Bluescript (16,32). The expressed sequence tag LD08190 (GenBank accession AA264447, Genome Systems) contained the entire coding region of SAM. This region was PCR-amplified using the T7 promoter primer and 5Ј-CGC GAA TTC TAT GGC GGA AAG AAA TCG AAT G-3Ј. The amplified DNA fragment was digested with EcoRI and XhoI and subcloned in HA-and Myc-Bluescript (16,32). The entire open reading frame of Who/How was amplified by PCR using an expressed sequence tag (GenBank accession AA439173, Genome Systems). The oligonucleotides utilized for the PCR were 5Ј-CGG GAA TTC TAT GAG TGT CTG TGA GAG CAA AG-3Ј and 5Ј-CTC GAA TTC TTA CTG TAT CTC AAT GAA ACC-3Ј. The resulting DNA fragment was digested with EcoRI and subcloned in HA-or Myc-Bluescript (16,32). The DNA inserts were sequenced manually with Sequenase (U. S. Biochemical Corp.) or by the Sheldon Biotechnology Institute automated sequencing facility (McGill University). All sequence data were compilations of multiple reads on both strands. GSG constructs were generated by PCR using the following oligonucleotides. The forward oligonucleotide for all the SAM constructs was 5Ј-AAA GAA TTC ATG ACC GAG AAG TAC GAC CGC-3Ј, whereas the reverse oligonucleotides used to generate deletions were 5Ј-GCG GAA TTC GAC ATT GTT GCC TCC CTG CAT GTG-3Ј, 5Ј-GCG GAA TTC TTA TGG ACA TCA CTT TGT-3Ј, and 5Ј-GCG GAA TTC ATC CAT GTC CAT GAG TGC C-3Ј for SAM:1-264, SAM:1-306, and SAM:1-345, respectively. The DNA encoding KEP1-224 was generated by PCR using 5Ј-CGC GAA TTC GAT AAA AAT GGA AAC CCC AAG C-3Ј and 5Ј-GCG GAA TTC CTA CTT CTT GTC GAA GAC AGA TCT GTA GGC C-3Ј and Myc-KEP1 as DNA template. The DNA fragment was digested with EcoRI and subcloned in the corresponding site in Myc-Bluescript. The DNA for SAM:233-428 was PCR-amplified with T7 promoter primer and 5Ј-CGC GAA TTC CAG ACA GTA ACG ACA TTA TCC GAC AG-3Ј using LD08190 as a DNA template. The amplified DNA fragment was digested with EcoRI and XhoI and subcloned in Myc-Bluescript.
Mammalian green fluorescent protein (GFP) fusion proteins were generated as follows. Myc-SAM and Myc-KEP1 plasmids were digested with EcoRI and XhoI. The DNA fragments encoding SAM and KEP1 were subcloned in the EcoRI and SalI site of pEGFP-C1 (CLONTECH). GFP-Who/How was generated by digesting Myc-Who/How with EcoRI, and the DNA fragment was subcloned into the EcoRI site of pEGFP-C1. The Drosophila GFP plasmids were generated by digesting the mammalian GFP constructs with NheI and SmaI. The ends of the DNA fragments were made blunt with the Klenow fragment of DNA polymerase I. The DNA fragments were subcloned into the SmaI site of pRM Ha3 (33).
Northern Blot Analysis-RNA was prepared from early (0 -4 h), mid to late (4 -20 h) embryos, first instar larvae, adult males and females, and ovaries, by phenol-chloroform extraction and ethanol precipitation as described previously (34). Poly(A) ϩ RNA was isolated by oligo(dT)cellulose chromatography (type III, Sigma) and quantitated by measuring absorbance at 260 nm. For preparation of the Northern blots, 10 g of poly(A) ϩ RNA denatured with glyoxal, electrophoretically separated on a 1% agarose gel, and transferred to Gene Screen filters. The KEP1 and SAM 32 P probes were generated by PCR using the above described oligonucleotides with KEP1-Myc-Bluescript and SAM-Myc-Bluescript as DNA templates. Sizes of the hybridized bands were estimated by comparison to the migration distances of commercial marker RNAs (0.24-to 9.5-kb RNA ladder, Life Technologies, Inc.).
Protein Expression and Analysis-HeLa cells were transfected with the vaccinia virus T7 expression system and lysed as described previously (32). Samples were analyzed on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Immunoblotting was performed using the following monoclonal antibodies: anti-Myc 9E10 (35) and anti-hemagglutinin (HA; BAbCO). The secondary antibody was goat anti-mouse conjugated to horseradish peroxidase (Organon Teknika-Cappel, Durham, NC), and chemiluminescence was used for protein detection (NEN Life Science Products).
RNA Binding Analysis-Homopolymeric RNA binding was performed using poly(U) (Amersham Pharmacia Biotech), poly(A), poly(C), and poly(G) Sepharose beads (Sigma) in lysis buffer supplemented with 2 mg/ml heparin (Sigma) for 1 h at 4°C. The beads were washed and analyzed as described (36). Chemical cross-linking studies were preformed as described previously (16).
Transfection in NIH 3T3 Cells and Apoptosis-NIH 3T3 cells were plated 12 h before transfection typically at a density of 10 5 cells/22-mm 2 coverslip (Fisher Scientific Co.). Cells were transfected with DNA constructs encoding GFP alone, GFP-SAM, GFP-KEP1, or GFP-Who/How using LipofectAMINE PLUS reagent (Life Technologies, Inc.). Twelve, 24, or 36 h after transfection, the cells were fixed with 4% paraformaldehyde in phosphate-buffered saline for 15 min and permeabilized with 1% Triton X-100 in phosphate-buffered saline for 5 min. The nuclei were stained with 3 g/ml 4,6-diamidino-2-phenylindole (DAPI). The morphology of transfected cells was examined with fluorescence microscopy, and cells with morphological features such as nuclear condensation and fragmentation were considered apoptotic.
Transfection in Drosophila Schneider 2 Cells and Condensation of HeLa Nuclei-Drosophila Schneider 2 (S2) (Invitrogen) cells were maintained at room temperature in Schneider media (Life Technologies, Inc.) supplemented with 10% fetal calf serum. Prior to transfection, 2 ϫ 10 6 S2 cells were plated and grown for 6 h. The cells were transfected by using the calcium phosphate precipitation method as suggested by the supplier. In brief, cells were incubated with DNA for 4 h and washed twice with complete S2 media prior to induction by addition of CuSO 4 to a final concentration of 0.5 mM. For chromatin condensation assay, HeLa nuclei were freshly isolated essentially as described (30) with the following modification; the buffer composition was 10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 10 mM ATP, 0.5 mM phenylmethanesulfonyl fluoride, and 5 g/ml each of leupeptin, pepstatin, and aprotinin. S2 cells were harvested 18 h after the addition of CuSO 4 . The cells were collected, washed twice with ice-cold phosphate-buffered saline, and resuspended in 125 l of 0.25 M Tris, pH 7.4. Cells were lysed by three freeze/thaw cycles (dry ice-ethanol/37°C). Cell debris was removed by centrifugation for 10 min at 10,000 rpm at 4°C. One hundred l of each lysate was added to 30 l of isolated HeLa nuclei (ϳ2.0 ϫ 10 6 nuclei), and incubated at 37°C for up to 3 h. Aliquots of nuclei were removed, streaked onto coverslips, and stained with DAPI. The presence of condensed nuclei and apoptotic bodies were scored as nuclei undergoing apoptosis.

Cloning of Two Novel Drosophila KH Domain Proteins-To
identify GSG related family members, the expressed sequence tag data base was searched with the BLAST program (37) using the amino acid sequence of the KH domain of Sam68 (32). Two Drosophila sequence tags were identified that contained conserved regions similar to the Sam68 KH domain. One protein we called KEP1 for KH domain encompassing protein 1. The mRNA for this protein has an open reading frame of 960 nucleotides and encodes a protein of 320 amino acids (Fig. 1A). The amino acid sequence of this protein does not resemble any of the known GSG domain proteins. In contrast, the other protein was called SAM because it has a high sequence similarity with mammalian Sam68 and therefore potentially represents the Drosophila Sam68 homolog. The mRNA for SAM encodes a protein of 428 amino acids (Fig. 1B). The primary amino acid sequence of KEP1 and SAM reveal that these proteins are members of the GSG domain family. Alignment of SAM with mouse (32) and human Sam68 (4) revealed a 50% sequence identity in the GSG domain (Fig. 1C), whereas the closest other Sam68 family member, GRP33, has a 38% sequence identity with Sam68. In addition, SAM has a cluster of 12 tyrosines at its carboxyl terminus like Sam68. In addition to the KH domain, Sam68 has two RGG sequences at its amino terminus that might function as an RGG box and may be the site of arginine methylation (38). The function of this potential Sam68 RNA binding motif is unknown (16). SAM also has an RGG sequence like Sam68; however, it is localized at the carboxyl terminus of the molecule. In contrast to Sam68 (32, 39 -41), SAM does not contain the multiple proline-rich motifs that serve as binding sites for SH3 domains.
Alignment of KEP1 and SAM with Other GSG Domain Family Members-The KH domain boundaries used for this alignment were taken from Musco and co-workers (2). The GSG domain amino-terminal boundary begins with the lysine at position 4 of GRP33 as defined by Jones and Schedl (7). Defining the COOH terminus of the GSG was more difficult and this region was previously called the CGA (carboxyl-terminal GSG domain associated) region (7). Now with the presence of other family members, it is clear that the GSG domain extends to the COOH terminus of the previously called CGA domain. Based on the data presented here, we define the whole region including the CGA as the GSG domain. Alignment of KEP1 and SAM with Who/How, Qk1, GLD-1, Sam68, GRP33, and SF1 demonstrates that 14 amino acids in the GSG domain are identical between all members (Fig. 2, *). Twenty-one amino acids are conservative substitutions (:), 39 residues (∧) are found in at least five of the seven members, and 30 residues are found in four of the seven members (ϩ). This indicates that the GSG domain and the embedded KH domain are highly conserved between species. Of particular note in the KH domain, the GXXXGXXG sequence predicted to form the RNA binding site (2), is fully conserved between all members. In addition, the lengths of the KH domain loop 1 and 4 are identical between members except for GLD-1, Qk1, and Who/How, which have a shorter loop 4 lacking two residues (Fig. 2).
Expression of the KEP1 and SAM Transcripts-The pattern of expression of KEP1 and SAM was investigated by Northern blot analysis (Fig. 3). The SAM gene produces four identifiable transcripts: a major transcript of 2.5 kb, and less abundant mRNAs of 2.8, 4.0, and 4.6 kb. The longest transcript is the least abundant, but like Sam68, uniformly expressed throughout development. The three shorter transcripts are detectable only in adult females, ovaries, and early (0 -4 h) embryos (Fig.  3, lanes 1-6), suggesting that these SAM mRNAs are synthesized during oogenesis and are contributed maternally to the embryo. The presence of multiple SAM transcripts suggests that the SAM gene is alternatively spliced, as it has recently been demonstrated for Sam68 (42). The KEP1 RNA is expressed at high levels throughout all stages of development except for larvae, in which its relative level is reduced. The major transcript from KEP1 is 1.6 kb, although a 2.0-kb form is also observed. Two smaller transcripts (1.3 and 0.6 kb) are specific to males (Fig. 3, lanes 7-12). These findings show that KEP1 and SAM have a different pattern of expression and most likely perform completely separate functions in the flies.
KEP1, SAM, and Who/How Are RNA-binding Proteins-To determine whether KEP1 and SAM are RNA-binding proteins, we epitope tagged each protein at the amino terminus with a short sequence from c-Myc that is recognized by the hybridoma antibody 9E10 (35). Since specific RNA target sequences for these members are unknown, we utilized homopolymeric RNA. The plasmids encoding these proteins were transfected in HeLa cells, and the cells were lysed in lysis buffer supplemented with 2 mg/ml heparin. The cell lysates were incubated with control, poly(G), poly(U), poly(C), and poly(A) Sepharose beads. The bound proteins were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with anti-Myc antibodies. SAM migrated with an approximate molecular mass of 50 kDa and bound both poly(U) and poly(A) Sepharose (Fig. 4, lanes  1-6). This homopolymeric RNA binding specificity was similar to that observed for Sam68 and GRP33 (16,40). KEP1 migrated with an approximate molecular mass of 40 kDa and bound poly(U) Sepharose (Fig. 4, lanes 7-12). This binding specificity was similar to that observed for GLD-1 (16). We also cloned the cDNA for Who/How (11)(12)(13), the other known Drosophila GSG domain protein, and examined its ability to bind homopolymeric RNA in vitro. Who/How migrated with an approximate molecular mass of 55 kDa and selectively bound poly(A) Sepharose (Fig. 4, lanes 13-18). These findings show that KEP1, SAM, and Who/How are RNA-binding proteins and bind homopolymeric RNA with different specificities.
Mapping the RNA Binding and Self-association Regions of KEP1 and SAM-We investigated whether the GSG domains of KEP1 and SAM were necessary and sufficient for RNA binding. A series of SAM deletions were created to find the minimal region required for RNA binding. The minimal region required to bind RNA was the SAM NH 2 -terminal 345 amino acids, encompassing the entire GSG plus 100 amino acids COOH-terminal to the GSG domain (Fig. 5A). A SAM protein containing amino acids 223-428 was unable to associate with RNA (Fig. 5A, SAM:233-428) further demonstrating that the SAM GSG domain was required for RNA binding. SAM:233-428 contains one RGG repeat and an RGG box, a type of RNA binding motif (43), usually contains greater than five RGG repeats. Our data demonstrate that the single RGG repeat is not sufficient to bind RNA (SAM:233-428) and does not seem to cooperate with the KH domain because SAM:1-345, devoid of the RGG repeat, bound with similar relative affinities as wild-type SAM (Fig. 5A). These findings show that the SAM GSG domain is necessary but not sufficient for RNA binding. In contrast to SAM, the GSG domain of KEP1 was sufficient to mediate RNA binding. A KEP1 truncation protein encompass-ing the entire GSG domain bound poly(U) homopolymeric RNA (Fig. 5A, KEP:1-224).
We investigated whether KEP1, SAM, and Who/How could self-associate by co-immunoprecipitation studies. HeLa cells transfected with HA-SAM and Myc-SAM, HA-KEP1 and Myc-KEP1, HA-Who/How and Myc-Who/How were immunoprecipitated with control or anti-Myc antibodies. The bound proteins were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with anti-HA antibodies. SAM, KEP1, and Who/How all self-associated into homomultimers (data not shown). The deletion constructs for KEP1 and SAM were tested for their ability to self-associate with wild-type KEP1 or SAM in co-immunoprecipitation assays in transfected HeLa cells. SAM:1-306 and SAM:1-345 were able to associate with wildtype SAM (Fig. 5B, lanes 9 and 12), whereas SAM:1-264 and SAM:233-428 were unable to associate with SAM (lanes 6 and  15). These data suggest that the GSG domain is required for self-association and show that the minimal region for SAM self-association resides in amino acids 1-306. Amino acids 245-306 located at the COOH terminus of the GSG domain may stabilize the GSG domain or interactions mediated by the GSG domain. The KEP1 GSG domain was sufficient to associate with wild-type KEP1, demonstrating that the KEP1 GSG domain can mediate the self-association (Fig. 5B, lane 21).
The Induction of Apoptosis by SAM and KEP1-To determine the cellular function of SAM, KEP1, and Who/How, we first expressed these proteins as GFP fusions in mammalian fibroblast cells. GFP alone was localized both in the cytoplasm and nucleus, GFP-Who/How and GFP-KEP1 were found exclusively in the nucleus, and GFP-SAM was cytoplasmic (Fig. 6A, GFP panels). Twelve hours after transfection, all "green" cells looked normal and healthy (data not shown). However, 36 h after transfection the KEP1-expressing cells exhibited morphological changes characteristic of apoptosis, including cell shrinkage, cytoplasm condensation, and membrane blebbing, not observed with GFP alone, GFP-SAM, or GFP-Who/How (Fig. 6A). The apoptotic cells were scored based on the presence of chromatin condensation and fragmentation as observed by DAPI staining (Fig. 6A). Nuclear morphological changes revealed by DAPI staining correspond to terminal deoxynucleotidyltransferase-mediated fluorescein-dUTP nick end labeling (TUNEL, Ref. 18).
The levels of apoptosis induced by KEP1 protein was quantitated by randomly counting cells and expressing the number of apoptotic cells as a percentage of transfected (green) cells. NIH 3T3 cells were transfected with plasmids expressing GFP, GFP-KEP1, GFP-SAM, or GFP-Who/How. A small fraction of GFP, GFP-SAM, or GFP-Who/How expressing cells were apoptotic, and this fraction remained steady (Fig. 6B, ϳ10 -15%). In contrast, ϳ70% of the KEP1-transfected cells were apoptotic at 36 h (Fig. 6B). These data demonstrate that the expression of KEP1 in fibroblast cells induces apoptosis.
We next examined the ability of Who/How, SAM, and KEP1 to induce apoptosis in Drosophila S2 cells. Although the S2 cells have never been demonstrated to undergo the typical morphological changes associated with apoptosis such as nuclear condensation, nuclear bodies, and membrane blebbing, activation of apoptosis in these cells can be measured indirectly by mixing cell extracts with isolated HeLa nuclei (30). This assay has been utilized to show that S2 cells treated with cycloheximide or transfected with reaper have increased caspase activity (31). To determine whether Who/How, SAM, and KEP1 activated the apoptotic pathway, S2 cells were mocktransfected or transfected with GFP-Who/How, GFP-SAM, and GFP-KEP1. As visualized by using the GFP moiety, approximately 10 -15% of the S2 cells were transfected; by using a  1-6) and KEP1 (lanes 7-12). Size markers are indicated in kilobases (kb). specific KEP1 antibody, we have estimated the amount of KEP1 protein to be 5-6 times higher than endogenous KEP1 in transfected cells (data not shown).
The transfection of SAM in Drosophila S2 cells demonstrated that it was cytoplasmic, whereas KEP1 and Who/How were both nuclear, consistent with our previous data in NIH 3T3 cells (Fig. 6B and data not shown). Cellular extracts were isolated from the different transfected S2 cells and mixed with isolated HeLa nuclei for 3 h. In the presence of extracts from mock-and GFP-Who/How-transfected cells, HeLa nuclei did not exhibit significant morphological changes over the duration of the experiments (Fig. 7A). In contrast, extracts from KEP1and SAM-transfected S2 cells induced chromatin condensation and formation of apoptotic bodies (Fig. 7A). The levels of apoptosis induced by SAM and KEP1 was quantitated by randomly counting HeLa nuclei and expressing the number of apoptotic nuclei as a percentage of total nuclei. Approximately 5% of the HeLa nuclei were undergoing apoptosis when mixed with cellular extracts from S2 cells mock-or GFP-Who/How-transfected and ϳ25-40% of the HeLa nuclei were undergoing ap-  1-6), , and Who/How (lanes [13][14][15][16][17][18] were expressed in HeLa cells. Cell lysates were lysed with lysis buffer supplemented with 2 mg/ml heparin, and incubated with control Sepharose or poly(G), poly(U), poly(C), or poly(A) Sepharose. The bound proteins were separated on SDS-polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with anti-Myc antibodies. The molecular mass markers are shown on the left in kDa.
optosis when mixed with cellular extracts from S2 cells transfected with SAM or KEP1. These data suggest that the apoptotic pathways are activated in S2 cells by the overexpression of SAM or KEP1. DISCUSSION We have identified two GSG domain protein family members in Drosophila, SAM and KEP1. SAM is a cytoplasmic protein that has 50% sequence identity with Sam68 in the GSG domain and may represent the Drosophila homolog of Sam68. In addition to amino acid similarities in the GSG domain, SAM also contains a tyrosine-rich COOH terminus that may serve as a substrate for tyrosine kinases as has been shown for Sam68 (4, 32, 39 -41). SAM is an RNA-binding protein, and it bound both poly(U) and poly(A) homopolymeric RNA in vitro. This binding specificity is similar to that observed with Sam68 and GRP33 (16). SAM self-associated and deletion studies demonstrated that the SAM GSG domain was required for both RNA binding and self-association like Sam68 (16). Although SAM has many similarities with Sam68 it is clearly different. SAM is cytoplasmic, whereas Sam68 is predominantly nuclear (data not shown and Ref. 4). The localization of SAM suggests that it may be involved in some aspect of translation or mRNA stability. The presence of SAM in ovaries is consistent with this hypothesis because translation control is an important mechanism of gene regulation in that tissue (44).
KEP1 is a 40-kDa RNA-binding protein that bound poly(U) homopolymeric RNA. It is a ubiquitously expressed nuclear protein that most likely self-associates into homotrimers, as determined from the molecular mass of the chemical crosslinked complex in HeLa cells (data not shown). KEP1 crosslinked complexes were able to bind homopolymeric RNA in vitro, supporting the idea that GSG domain-containing proteins bind RNA as multimers (16). The self-association property is now shared by eight different GSG domain proteins including Sam68, GRP33, mouse Qk1-7, GLD-1 (16,18), Xqua (17), and SAM, KEP1, and Who/How (this report). All GSG domain proteins identified thus far have been shown to associate with RNA in vitro (4,9,10,16). The multimer formation may be required for high affinity binding to their target RNA sequences. The GSG domain-containing family of proteins is growing rapidly (for a review, see Ref. 14). Drosophila partial sequences of 27 amino acids in the KH domain were cloned for four quaking-related proteins (12). The sequence of these four proteins has been recently reported, and the proteins were named QKR58E1, E2, E3, and B (45). QKR54B and QKR58E3 are SAM and KEP1, respectively. QKR58E1 and QKR58E2 are novel GSG domain family members. Taken together, these findings demonstrate that Drosophila has at least five GSG domain-containing family members.
The Drosophila Who/How is expressed in invaginating mesoderm during early embryogenesis (11)(12)(13). The expression of Who/How subsequently becomes restricted to the myogenic lineage including the cardioblasts and myoblasts (11,12). Weak alleles are embryonic lethal or cause wing defects because of improper myotube migration and attachment (11). These data implicate Who/How in muscle cell determination and a number of functions including control of heart rate in late embryogenesis (13). The absence of apoptosis in Who/Howtransfected cells is consistent with the idea that Who/How plays a role in differentiation. Although we have recently shown that Qk1-7 induces apoptosis in NIH 3T3 cells (18), it will be necessary to examine whether the Who/How mouse equivalent, Qk1-5, also induces apoptosis. Our data confirm the nuclear localization of Who/How (13) and demonstrate that Who/How is an RNA-binding protein that can bind RNA as a multimer (data not shown). A missense mutation has been identified in the KH domain of Who/How (11), and this point mutation did not affect the ability of Who/How to self-associate and bind homopolymeric RNA (18).
The expression of KEP1 in NIH 3T3 induced apoptosis, with greater than 70% of the transfected cells undergoing apoptosis 36 h after transfection. SAM-and Who/How-transfected NIH 3T3 cells were healthy and had no morphological changes associated with apoptosis. When expressed in Drosophila S2 cells, both KEP1 and SAM induced apoptotic activity. It remains unclear at this time why SAM did not induce apoptosis in NIH 3T3 cells as it did in Drosophila S2 cells, the more physiologically relevant cell line for SAM. By using an in vitro system, we observed that extracts from Drosophila S2 cells transfected with KEP1 or SAM could trigger dramatic morphological changes characteristic of apoptosis in isolated HeLa nuclei. These changes are identical to those induced by cycloheximide-treated S2 extracts (data not shown). Since cycloheximide has been demonstrated to activate caspases (31) and morphological changes in HeLa nuclei are caspase-dependent (30,31), we suspect that KEP1 and SAM induce apoptosis by activating the Drosophila caspases such as drICE (31) and DCP-1 (46). The apoptosis induced by the overexpression of KEP1 and SAM did not require additional manipulations such as the reduction in serum as observed with c-Myc (47). In this regard, SAM and KEP1 are similar to Reaper, HID, and GRIM, where the overexpression alone is sufficient to induce apoptosis in insect cells (48,49).
The identification of two new Drosophila GSG domain family members that we called SAM and KEP1 demonstrates the existence of a large family of GSG domain proteins. KEP1 and SAM are RNA-binding proteins that bind homopolymeric RNA in vitro and self-associate into multimers. Although our results strongly indicate a putative role for KEP1 and SAM in the regulation apoptosis, it remains to be determined whether these proteins have a toxic effect such as perturbing the normal cellular processes or are directly involved in caspase-mediated apoptosis. Since the name SAM is used for a protein domain (sterile alpha motif), we propose the name Sam50 for the Drosophila SAM protein.