Fission yeast homolog of murine Int-6 protein, encoded by mouse mammary tumor virus integration site, is associated with the conserved core subunits of eukaryotic translation initiation factor 3.

The murine int-6 locus, identified as a frequent integration site of mouse mammary tumor viruses, encodes the 48-kDa eIF3e subunit of translation initiation factor eIF3. Previous studies indicated that the catalytically active core of budding yeast eIF3 consists of five subunits, all conserved in eukaryotes, but does not contain a protein closely related to eIF3e/Int-6. Whereas the budding yeast genome does not encode a protein closely related to murine Int-6, fission yeast does encode an Int-6 ortholog, designated here Int6. We found that fission yeast Int6/eIF3e is a cytoplasmic protein associated with 40 S ribosomes. FLAG epitope-tagged Tif35, a putative core eIF3g subunit, copurified with Int6 and all five orthologs of core eIF3 subunits. An int6 deletion (int6Delta) mutant was viable but grew slowly in minimal medium. This slow growth phenotype was accompanied by a reduction in the amount of polyribosomes engaged in translation and was complemented by expression of human Int-6 protein. These findings support the idea that human and Schizosaccharomyces pombe Int-6 homologs are involved in translation. Interestingly, haploid int6Delta cells showed unequal nuclear partitioning, possibly because of a defect in tubulin function, and diploid int6Delta cells formed abnormal spores. We propose that Int6 is not an essential subunit of eIF3 but might be involved in regulating the activity of eIF3 for translation of specific mRNAs in S. pombe.


Mouse mammary tumor virus (MMTV)
integrates into the mouse genome and frequently causes mammary tumors. The sites of MMTV integration were identified to delineate the molecular basis of the tumorigenesis (see Refs. 1 and 2 for review). Except for int-6, the int loci encode growth factors or transmembrane receptors. Thus, altered expression of these Int proteins leads to mammary tumorigenesis presumably by affecting signal transduction pathways controlling cell growth.
By contrast, the molecular basis for tumorigenesis caused by integration into int-6 is unclear. Int-6 encodes a ubiquitously expressed, 52-kDa protein, which corresponds to the 48-kDa eIF3e subunit 2 of eukaryotic translation initiation factor 3 (eIF3) (3). int-6 is the site of MMTV integration in at least three independently isolated mouse tumors; in all cases, MMTV integration into an intron of the int-6 gene resulted in expression of truncated int-6 mRNA species (4). Therefore, MMTV integration into int-6 may produce a dominant negative allele, because of expression of a truncated, unregulatable form of the protein. Alternatively, the truncation may simply disrupt the biological function of Int-6, and tumorigenesis would result from reduced int-6 gene dosage.
Although Int-6 is a stable component of mammalian eIF3, it is not known whether it is required for the activity of eIF3 in translation initiation (3). Furthermore, two laboratories (5-7) have localized a significant proportion of Int-6 in the nucleus, particularly in a nuclear compartment called the promyelocytic leukemia (PML) nuclear body (8). Nuclear localization of an Int-6 ortholog was also reported in Arabidopsis thaliana (9). These results raised the possibility that integration into int-6 affected the role of Int-6 in the nucleus but not in the cytoplasm as a component of eIF3. To identify the biological role of Int-6 and decipher the consequences of its truncation, it was important to establish the role of Int-6 in translation initiation as a part of eIF3. Thus far, eIF3 has been purified from mammalian cells (10,11), wheat germ (12,13), A. thaliana (14), and budding yeast Saccharomyces cerevisiae (15)(16)(17). Mammalian eIF3 associates with the 40 S ribosome and stimulates the binding of both mRNA and Met-tRNA i Met (10,11). Purified S. cerevisiae eIF3 was also shown to stimulate Met-tRNA i Met binding to 40 S ribosomes in vitro (16,17).
The eIF3 purified from budding yeast consists of five proteins, eIF3a (TIF32), eIF3b (PRT1), eIF3c (NIP1), eIF3g (TIF35), and eIF3i (TIF34), homologous to mammalian eIF3 subunits and does not contain an ortholog of human eIF3e/ Int-6 (17,18). Moreover, the complete S. cerevisiae genome sequence does not encode a protein closely related to Int-6. Thus, an Int-6-related protein does not seem to be involved in the functions of eIF3 in the yeast Saccharomyces. Besides Int-6, mammalian eIF3 contains three other noncore subunits, eIF3d (p66), eIF3f (p47), and eIF3h (p40), which are conserved in plants, Caenorhabditis elegans, and Drosophila melanogaster but absent in S. cerevisiae (19). Thus, it is possible that the activities of eIF3 from higher eukaryotes are modulated by association of the five core subunits with Int-6 and the other eIF3 subunits not found in S. cerevisiae. Interestingly, the recent progress in sequencing the entire genome of fission yeast Schizosaccharomyces pombe has revealed that it encodes the four orthologs of noncore eIF3 subunits, in addition to orthologs of the five core subunits (14). Thus, S. pombe eIF3 may resemble mammalian eIF3 in containing several polypeptides in addition to the five core subunits present in budding yeast.
In this study, we examined the role of the S. pombe protein, Int6, closely related to murine Int-6. We found that Int6 is predominantly localized in the cytoplasm, unlike mammalian and plant Int-6. We also provide evidence that Int6 is a stable component of fission yeast eIF3 and is functionally homologous to human Int-6 protein. Unlike tif35 ϩ , which encodes an essential core eIF3g ortholog, int6 ϩ is dispensable. The int6⌬ mutant grew slowly in minimal medium, however, and this slow growth phenotype was accompanied by reduction in the amount of polyribosomes engaged in translation. Our results implicate Int-6 homologs in translation initiation and lay the groundwork for investigating the cytoplasmic function of Int-6 using fission yeast as a model organism.

MATERIALS AND METHODS
Plasmid and Yeast Strains-The DNA segment containing the int6 ϩ open reading frame (ORF) was amplified by polymerase chain reaction with oligonucleotides 1 and 2 (Table I) from genomic DNA, digested with NdeI and XhoI, and subcloned into the NdeI and SalI sites of pREP41 (20) or pREP41HA N (21), producing pREP-int6 or pRHA-int6, respectively. The tif35 ϩ ORF was similarly amplified with oligonucleotides 3 and 4, digested with NdeI and BglII, and subcloned into the NdeI and BamHI sites of pRF41, a derivative of pREP41 containing the coding sequence for the FLAG epitope, inserting it between the BamHI and SmaI sites of pREP41. The resultant pREP41-tif35-FLAG plasmid encodes eIF3g (Tif35) FLAG-tagged at its C terminus under the control of the nmt1 thiamine-repressible promoter.
The DNA segment encompassing the human int-6 ORF was amplified by polymerase chain reaction with oligonucleotide 5 and a T7promoter primer from pETp48 3 (the T7 promoter sequence lies upstream of the human int-6 ORF in pETp48, with a BamHI site in between) and digested with BamHI. The resulting 1.3-kilobase BamHI fragment was used to replace the S. pombe int-6 ϩ ORF in pRHA-int6 to generate pRHA-hInt-6.
Purification of the Putative eIF3 Complex Containing FLAG-Tif35/ eIF3g-Transformants of JW350 (int6 -3HA) carrying pREP-tif35-FLAG, or pREP41 as a control, were grown in 2 liters of EMM-leu-B1 to A 600 ϭ ϳ10 and harvested by centrifugation. After washing with sterilized water, cells (ϳ10 g) were suspended in 15 ml of buffer A (25) and homogenized with a French Press (Spectronic Unicam). After clearing the extract by two rounds of centrifugation (13,000 ϫ g for 20 min and 23,000 ϫ g for 20 min), all ϳ15 ml of WCE (ϳ750 mg of total protein) were centrifuged in quick seal tubes in a Beckman 70.1 Ti rotor at 55,000 rpm for 2 h at 4°C. The ribosomal pellets were suspended in 5 ml of buffer B (same as buffer A described by (25), except containing 350 mM KCl), followed by centrifugation in polycarbonate tubes in a Beckman TLA 100.4 rotor at 80,000 rpm for 1 h at 4°C. The resulting ribosomal salt wash (RSW) fraction was immediately mixed with 100 l of M2 anti-FLAG affinity resin (Sigma) for 2 h at 4°C. After washing four times with 1 ml of buffer B, the bound proteins were eluted in 200 l of buffer B containing 0.4 mg/ml FLAG peptide (Sigma). The resulting FL-Tif35 preparation contained ϳ30 ng/l of putative eIF3 complex.
Polysome Profile Analysis-S. pombe cells were treated for polysome analysis exactly as described for S. cerevisiae (23). For immunoblot analysis of the gradient samples, the WCEs were layered on 15-40% sucrose gradients, centrifuged at 39,000 rpm for 4.5 h, and separated into twenty 0.6-ml fractions. A portion of the first fifteen fractions from the top was precipitated with 10% trichloroacetic acid, followed by SDS-PAGE and immunoblotting.
Mass Spectrometry-We resolved the FLAG-Tif35 complex by SDS-PAGE, and each of the five peptides (except p40 and p39 analyzed together) was subjected to in gel trypsin digestion and matrix-assisted laser desorption-ionization-time-of-flight mass spectrometry at Integrative Proteomics, Inc. (Toronto, Canada). The observed mass spectrum of its tryptic peptides was compared with the predicted spectra of all the proteins in the data base using the ProFound program.

RESULTS
The Phenotypes of S. pombe Cells Lacking the int6 ϩ Gene-Fission yeast S. pombe encodes a hypothetical protein (Gen-Bank TM accession number CAA22813) homologous to murine eIF3e/Int6. This protein, designated Int-6 here, is 45% identical and 57% similar to murine Int6. Because this extent of similarity is only slightly less than that between murine Int-6 and related proteins in C. elegans (52% identical) and D. 3 J. Clayton and K. Asano, unpublished material. Fission Yeast Int6 is a Part of eIF3 melanogaster (62% identical), we suspected that Int6 is the S. pombe homolog of mammalian Int-6.
First we examined the effect on cell growth of deleting the int6 ϩ gene. We transformed a diploid S. pombe strain with an int6 deletion construct containing a selectable kanamycin resistance gene (kan) in place of the int6 ϩ ORF and confirmed the deletion of one allele of chromosomal int6 ϩ by Southern blot analysis. When the resulting heterozygous diploid strain was allowed to sporulate, we observed normal formation of tetrads with four viable ascospores. As expected, two spores in every tetrad were Kan R , indicating the presence of the int6⌬::kan allele in 50% of the ascospores. These results indicate that int6 ϩ is not essential (data not shown).
We found that the haploid int6⌬ strain designated JW346 grew normally in rich medium (YES) with a doubling time of ϳ3-4 h at 30°C in liquid medium but grew slowly in minimal medium (EMM) at 30°C with a doubling time of ϳ7 h compared to ϳ3 h for the isogenic wild type strain JY450. This reduction in the growth rate was exacerbated by prolonged incubation in EMM for Ͼ24 h. When haploid int6⌬ cells were shifted to minimal medium, we observed that a large fraction (Ͼ70%) of these cells became smaller and a significant fraction (ϳ5-10%) of them exhibited displacement of cell wall material (arrowhead in Fig. 1A) after only 6 h of incubation. In a substantial number of cells, the nucleus was displaced from the medial region, and during mitosis, divided nuclei did not separate well (asterisk in Fig. 1B). As a consequence, after 24 h in EMM, ϳ16% of septated int6⌬ cells had undergone unequal nuclear partitioning, such that one daughter received two nuclei and the other received none (I in Fig. 1B) or that one of the daughter nuclei was lodged in the middle of the new septum (II in Fig. 1B). These events were not observed with wild type cells (Ͻ0.1% of septated cells) and were observed less frequently (ϳ4% of septated cells) with int6⌬ cells grown in YES medium.
Thus, int6⌬ appears to impair steps in mitosis, especially the positioning of daughter nuclei. It is noteworthy that mutants deficient in tubulin function exhibit unequal nuclear positioning and are hypersensitive to microtubule-depolymerizing drugs such as thiabendazole (26). We found that int6⌬ cells also are hypersensitive to thiabendazole (data not shown), suggesting that their nuclear positioning defect might result from defects in tubulin function. When the diploid int6⌬ cells were subjected to meiosis and sporulation, Ͼ70% of the zygotes formed spores with two nuclei (asterisk in Fig. 1C) and/or abnormally shaped spores (arrowhead in Fig. 1C). Thus, the int6 deletion has pleiotropic phenotypes.
Next, we examined the intracellular localization of Int6. For this purpose, we constructed a haploid yeast strain (JW346) encoding the int6 protein fused to GFP, in place of the wild type int6 protein (see "Materials and Methods"). JW346 grew indistinguishably from its isogenic wild type parent (JY450) both in rich and minimal media (data not shown), indicating that GFP-Int6 is functional. As shown in Fig. 2, fluorescence was observed throughout the cytoplasm and was absent from the nucleus. Therefore, Int6 is predominantly localized in the cytoplasm.

Int6 Is Part of a High Molecular Mass Complex
That Associates with 40 S Ribosomes-Because eIF3 associates with 40 S ribosomes, it was important to examine whether Int6 is associated with 40 S ribosomes in S. pombe cells. We constructed strain JW350 containing an HA epitope-tagged int6 allele (int6 -3HA) in place of the chromosomal wild type allele. HA-Int6 is functional, because JW350 grew indistinguishably from the wild type both in rich and minimal media (data not shown). As a control, we analyzed the S. cerevisiae strain KAY50 encoding an HA-tagged form of eIF3i subunit TIF34 (HA-TIF34), because it was shown that eIF3 is associated with 40 S subunits in budding yeast (27). Cell extracts prepared from these strains in the presence of cycloheximide were resolved by sucrose gradient velocity sedimentation, and the resulting A 254 absorbance profiles indicated that the ribosomal species in S. pombe cosedimented exactly with the corresponding ribosomal species in S. cerevisiae (Fig. 3, A and B, upper panels). Immunoblot analysis of the gradient fractions showed that the majority of HA-Int6 sedimented with the bulk of 40 S subunits in S. pombe extracts, consistent with its physical association with 43-48 S preinitiation complexes (Fig. 3A, lanes 7 and 8). As expected, a large fraction (ϳ50%) of the HA-TIF34 in S. cerevisiae extracts sedimented at exactly the same location (Fig. 3B,  lanes 7 and 8).
We also found that a subpopulation of both HA-Int6 (Fig. 3A,  lanes 4 and 5) and HA-TIF34 (Fig. 3B, lanes 4 and 5) sedimented in fractions just preceding the 40 S ribosome, suggesting that these proteins are present in high molecular mass complexes. Immunoblotting of the S. cerevisiae fractions with antibodies against the budding yeast eIF3b (PRT1) supports the notion that the high molecular mass complex containing HA-TIF34 at fractions 4 -5 is the eIF3 complex of ϳ600 kDa (17,18,28). As expected, the S. cerevisiae eIF2 complex of 124 kDa sedimented in fractions closer to the top of the gradient, with a peak in fraction 3, whereas the bulk of eIF1 (13 kDa) peaked in fraction 2 (Fig. 3B, bottom two panels). Thus, the sedimentation pattern of HA-Int6 strongly suggests that Int6 is associated with a high molecular mass complex, most likely FIG. 1. Phenotypes of int6⌬ cells. A, JY450 (int6 ϩ ) or JW346 (int6⌬) cells, cultured in YES, were shifted to EMM and incubated for 6 h, followed by staining with calcofluor to visualize the septa. The arrowhead indicates the displaced cell wall material. b, cells cultured in EMM for 24 h were stained with DAPI to visualize the nuclei and with a small amount of calcofluor to visualize the septa. The asterisk indicates an example of displaced nuclei, and the arrowheads indicate examples of unequal nuclear partitioning in septated cells. C, the int6⌬ cells were subjected to sporulation on synthetic sporulation agar and stained with DAPI. The asterisk indicates a binucleate fused spore, and the arrowhead indicates an abnormally shaped spore.
Fission Yeast Int6 is a Part of eIF3 eIF3, which interacts tightly with the 40 S ribosome.
HA-Int6/eIF3e Resides in a Multisubunit Complex Containing FLAG-Tif35/eIF3g and Four Other Proteins Homologous to Core eIF3 Subunits-Next, we attempted to establish a physical interaction between Int6 and eIF3 subunits in S. pombe. We focused first on a predicted S. pombe protein of 31.4 kDa, designated Tif35/eIF3g (GenBank TM accession number CAA18400), that is an ortholog of the 32-kDa subunit of S. cerevisiae eIF3 encoded by TIF35 (18,29,30) and of the 44-kDa subunit of mammalian eIF3 (31) (Table II). We found that deletion of tif35 ϩ in S. pombe is lethal, indicating that tif35 ϩ is an essential gene, as expected for a core eIF3 subunit (data not shown; see "Materials and Methods"). We constructed a plasmid (pREP-tif35-FLAG) encoding C-terminally FLAG-tagged Tif35 under a thiamine-repressible promoter and showed that it complemented the lethality associated with tif35⌬ in the absence of thiamine but not in its presence (data not shown). A transformant of int6 -3HA strain JW350 carrying pREP-tif35-FLAG grew as rapidly as did the parental strain JW350 car-rying an empty vector in minimal medium lacking thiamine, and we confirmed that FLAG-Tif35 was expressed in WCE prepared from this transformant grown under these conditions (Fig. 4A, lanes 2 and 3). As shown in Fig. 4A, FLAG-Tif35 in the WCE was immunoprecipitated with anti-FLAG antibodies (panel I, lanes 5 and 6), and HA-Int6 was coimmunoprecipitated only when FLAG-Tif35 was being expressed in this strain (lanes 4 and 5). Likewise, HA-Int6 and FLAG-Tif35 were coimmunoprecipitated specifically with anti-HA antibodies (panel II, lane 5). These results indicate a physical interaction between HA-Int6 and FLAG-Tif35, presumably in the context of eIF3.
Encouraged by this finding, we attempted to affinity purify the S. pombe eIF3 complex. We prepared a RSW fraction (that should be enriched for initiation factors) from the transformant of JW350 (int6 -3HA) carrying pREP-tif35-FLAG after growth in minimal medium lacking thiamine. The FLAG-Tif35-containing complex was purified from the RSW with anti-FLAG affinity resin. As shown in Fig. 4B, lane 2, Coomassie staining of the purified complex eluted from the resin revealed seven major polypeptides (p105, p85, p70, p40, p39, p32, and p20) and several other minor constituents (p64, p60, p46, p43, and p36). None of these proteins were contained in the fraction prepared from a control untagged tif35 ϩ strain (lane 1). Immunoblotting of the purified complex revealed that the 60-kDa minor constituent (p60) reacted with anti-HA antibodies (lanes 3 and 4), whereas p39, p32, and p20 reacted with anti-FLAG antibodies (lanes 5 and 6). These results suggest that p60 and p39 are Int6-HA and FL-Tif35, respectively, and that p32 and p20 are degradation products of FL-Tif35. Identification of p60 as HA-Int6 was supported by preparing the FL-Tif35 complex from a control strain that carries pREP-tif35-FLAG but contains untagged int6 ϩ and finding a 58-kDa protein instead of p60 in SDS-PAGE (data not shown).
Because functional eIF3 complexes purified from S. cerevisiae contained only five core subunits eIF3a, eIF3b, eIF3c, eIF3g, and eIF3i (16,17), we suspected that the five major polypeptides in the FL-Tif35-containing complex, p105, p85, p70, p40, and p39 would correspond to the predicted S. pombe proteins, eIF3a, eIF3b, eIF3c, and eIF3i, in addition to (FL-) Tif35/eIF3g itself (see the Introduction). To test this idea directly, we processed each of the five polypeptides for mass spectrometry (see "Materials and Methods"), and we identified all five S. pombe orthologs of S. cerevisiae core eIF3 subunits and the heat shock protein Hsp70 (Table II). Copurification of Hsp70 with FL-Tif35 was unexpected but occurs frequently in affinity purification of protein complexes from different organisms. 4 These results suggest strongly that the FLAG-Tif35containing complex corresponds to S. pombe eIF3.
The presence of HA-Int6 in a purified complex containing FL-Tif35 indicates that the HA-Int6/FL-Tif35 interaction we observed in cell extracts (Fig. 4A) was tight enough to with-4 B. Cox, personal communication.

Fission Yeast Int6 is a Part of eIF3
stand the high salt buffer used to strip initiation factors from the ribosomes in preparing the RSW. The fact that the amount of Int6 was substoichiometric compared with the amounts of the five core subunits, including FL-Tif35, may indicate that only a portion of eIF3 contains Int6. However, it is possible that Int6 is a stoichiometric component of eIF3 but was partially removed from the core complex during the RSW preparation. Based on the results shown in Figs. 3 and 4 and Table II, we propose that Int6 is physically associated with at least a portion of the eIF3 complexes in S. pombe.
Deletion of int6 ϩ Diminishes Translation Initiation in Minimal Medium-As shown in Fig. 1, deletion of int6 ϩ was not lethal and, instead, showed pleiotropic phenotypes including slow growth in EMM and defects in mitosis and meiosis. To determine whether the int6⌬ slow growth phenotype is associated with reduced rate of translation initiation, we analyzed the polysome profiles in cell extracts prepared from strains JW346 (int6⌬) and JY450 (int6 ϩ ) grown in minimal medium. As shown in Fig. 5, wild type cells had a sizable polysome content, characterized by a polysome to monosome ratio (P/M) of 4.7 (or 67% polysomes compared with the total amount of ribosomes), whereas the int-6⌬ mutant contained a reduced amount of polyribosomes, with a P/M ratio of 1.5 (or 47% polysomes compared with the total amount of ribosomes). The reduced P/M ratio in the int6⌬ mutant is indicative of a moderate reduction in the rate of translation initiation. In principle, the diminished rate of initiation could be a secondary consequence of reduced growth rate. However, we showed pre-    1 and 4) or pREP41-tif35-FLAG (lanes 2 and 5) and the transformant of JY450 (int6 ϩ ) carrying pREP41-tif35-FLAG (lanes 3 and 6) were grown in thiamine-depleted EMM for 16 h to induce FLAG-Tif35 expression from pREP41-tif35-FLAG. Cell extracts were prepared and subjected to immunoprecipitation (CoIP) with an anti-FLAG antibody (panel I) or anti-HA antibody (panel II). Immnoprecipitates were separated by SDS-PAGE and examined for the presence of the indicated proteins by immunoblotting using anti-FLAG or anti-HA antibodies (lanes 4 -6). One-fourteenth of the total crude extracts were separated and analyzed similarly (lanes 1-3). B, FL-Tif35 complex purified from the RSW contains HA-Int6. The FLAG-Tif35 complex (ϩ, lanes 2, 4, and 6) and the control eluate prepared in parallel from a strain carrying the vector pREP41 (Ϫ, lanes 1, 3, and 5) were prepared as described under "Experimental Procedures." A portion of these samples was subjected to SDS-PAGE, followed by Coomassie staining (lanes 1 and 2) or immunoblotting with anti-HA or anti-FLAG antibodies (lanes 3-6). Proteins specifically associated with FL-Tif35 in addition to HA-Int6 were labeled by their sizes in kDa.

Fission Yeast Int6 is a Part of eIF3
viously that a slow growth rate in budding yeast caused by respiratory deficiency did not alter the P/M ratio (32).
Genetic Evidence That S. pombe Int6 Is Homologous to Mammalian Int-6 -Finally, we examined whether S. pombe Int6 was functionally homologous to the mammalian Int-6/eIF3e protein. We constructed plasmid pRHA-hInt-6 for expression of HA-tagged human Int-6 under the thiamine-repressible promoter and investigated whether this plasmid could complement the slow growth phenotype associated with the int6⌬ mutation. As controls, we used the empty vector pREP41HA N, pREP-int6 encoding untagged S. pombe Int6, and pRHA-int6 encoding HA-tagged S. pombe Int6 (see "Materials and Methods"). As shown in Fig. 6, the int6⌬ cells carrying the vector alone grew less rapidly on EMM than the isogenic wild type cells carrying the same plasmid (rows 1 and 2). This slow growth phenotype was complemented fully by untagged S. pombe int6 (row 3) or partially by HA-tagged int6 (row 4). Interestingly, the HA-tagged human int-6 and HA-tagged S. pombe int6 complemented the slow growth phenotype of int6⌬ to similar degrees (row 4 and 5). Thus, human Int-6 can substitute at least partially for the function of S. pombe Int6. Because the slow growth phenotype of int6⌬ cells is associated with a reduced rate of translation initiation (Fig. 5), these results suggest that human Int-6 protein can function in place of S. pombe Int6 in translation initiation.
We also examined whether the human Int-6 encoded by pRHA-hInt-6 can complement the nuclear partitioning defect we observed in Fig. 1B. We grew the int6⌬ cells carrying the vector alone, pRHA-int6 or pRHA-hInt-6, employed in Fig. 6, in EMM for 24 h and counted the number of septated cells that underwent unequal nuclear partitioning. The frequency of abnormal nuclear partitioning with both pRHA-int6 and pRHA-hInt-6 was reduced to Ͻ25% of the level of the vector control in each of three independent experiments. Thus, HA-tagged forms of fission yeast and human Int-6 can partially complement the nuclear partitioning defect associated with int6⌬. Based on these results, we conclude that human Int-6 protein can function in place of fission yeast Int6 in mitosis as well as in translation initiation. DISCUSSION In this study, we characterized a fission yeast protein Int6 closely related to murine Int-6 protein, encoded by a frequent MMTV integration site. We showed that Int6 is cytoplasmic (Fig. 2) and associates in vivo with 40 S ribosomes (Fig. 3). Affinity purification of the putative eIF3 complex of S. pombe directed against FLAG epitope-tagged Tif35/eIF3g revealed the presence of Int6 and the five proteins homologous to the core subunits of budding yeast eIF3 (Fig. 4 and Table II). This is the first evidence that fission yeast eIF3 contains the five core proteins that comprise budding yeast eIF3 (17,18) and also occur in mammalian eIF3 (19). In addition to the five core subunit orthologs, we also identified the fission yeast ortholog of eIF3f (PIR accession number T40490, 33 kDa) as the 36-kDa polypeptide in the FLAG-Tif35 complex shown in Fig. 4B (data not shown). Thus, physical interactions between the core and noncore subunits appear to be conserved between the fission yeast and mammalian eIF3 complexes. Furthermore, the complementation of the int6⌬ slow growth phenotype by human int-6 suggests that S. pombe Int6 is functionally homologous to mammalian Int-6 ( Fig. 6).
Despite the evidence for physical interaction between Int6 and the putative core eIF3 subunits, int6 ϩ is not an essential gene, and we observed only a moderate reduction in the rate of protein synthesis initiation in int6⌬ cells (Fig. 5). Thus, Int6 is dispensable for the essential activities of eIF3. moe1 ϩ , encoding the S. pombe homolog of human eIF3d subunit p66, is also nonessential (33), although its association with S. pombe eIF3 remains to be determined. Presumably, the essential functions of S. pombe eIF3 depend only on the five core subunit orthologs, of which (at least) eIF3i (34) and Tif35/eIF3g are essential.
We found that int6⌬ cells showed several phenotypes, including frequent unequal nuclear partitioning during mitosis in haploids, abnormal spore formation in diploids (Fig. 1), and hypersensitivity to the microtubule-depolymerizing drug, thiabendazole. If Int6 is involved in general translation initiation, how can deletion of int6 ϩ lead to such specific phenotypes? One possible explanation is that a moderate reduction in eIF3 function caused by int6⌬ severely impaired translation of a subset of mRNAs with relatively poor initiation regions that compete poorly for ribosomes and other factors with mRNAs containing optimal initiation sites. Those mRNAs whose translation was strongly reduced in int6⌬ cells would include mRNAs encoding proteins critical for nuclear partitioning or sporulation. An alternative explanation for the pleiotropic int6⌬ phenotype is that Int6 is involved in gene-specific translational control, e.g. by facilitating recruitment of a specific subset of mRNAs to the eIF3 bound to 40 S ribosomes.
In S. cerevisiae, reduced protein synthesis rates caused by mutations in eIF3 subunit genes TIF34, PRT1, or TIF32/RPG1 result in cell cycle arrest primarily in the G 1 phase, although the tif34 mutation additionally impedes the G 2 /M transition (29,35,36). In contrast, int6⌬ has more profound effects on the completion of mitosis (Fig. 1B) than on the start of the cell cycle, in favor of the model that Int6 is involved specifically in translational control in G2. In keeping with this idea, deletion of moe1 ϩ encoding another noncore eIF3 subunit homolog impairs microtubule function and confers resistance to microtubule-destabilizing reagents (33), suggesting a link between Moe1 (eIF3d) and Int6 (eIF3e) in mitosis. Recently, one of the mis (minichromosome instability) mutations in fission yeast was shown to impair 18 S ribosomal RNA biogenesis, suggesting that a decrease in general protein synthesis rates would impair chromosome segregation (37). Understanding the molecular basis for mitotic defects caused by disruption of the translation machinery will require further examination.
While this manuscript was under review, two articles were published reporting the characterization of the Int-6 ortholog in S. pombe (38,39). Crane et al. (38) proposed that S. pombe Int-6 is a part of eIF3, based on coimmunoprecipitation of epitope-tagged eIF3i and eIF3b orthologs with Int-6-HA. We demonstrated this point by affinity purification (Fig. 4B) and subsequent immunoblot (Fig. 4B) and mass spectrometry (Table II) analyses of a FLAG-Tif35/eIF3g-containing complex. This purified complex contained seven of the nine eIF3 subunit FIG. 6. Human Int-6/eIF3e can complement the slow growth phenotype of int6⌬ cells. Transformants of JY450 (int6 ϩ ) or JW346 (int6⌬) carrying the empty vector pREP41HA N (rows 1 and 2) or those of JW346 carrying pREP-int6 (row 3), pRHA-int6 (row 4), or pRHA-hInt-6 (row 5) were grown in SC medium lacking leucine at 30°C for 1.5 days. Equal A 600 units and 1 ϫ 10 Ϫ1 or 1 ϫ 10 Ϫ2 dilutions of these amounts, were spotted on EMM lacking thiamine and leucine and incubated at 30°C for 3 days. W. T., wild type.
Fission Yeast Int6 is a Part of eIF3 orthologs encoded in fission yeast. We also showed that the amount of polysomes engaged in protein synthesis was reduced significantly in the int6⌬ mutant (Fig. 5). These results are consistent with the polysome analysis of Bandyopadhyay et al. (39), who also observed a moderately reduced rate of total protein synthesis in the int6⌬ mutant, as measured by amino acid incorporation. Together, these data provide convincing biochemical evidence that Int6 enhances the rate of translation initiation in vivo. These articles also reported interesting phenotypes associated with overexpression and deletion of int6 ϩ . Overexpression of int6 ϩ conferred multi-drug resistance, dependent on the transcription activator Pap1 (38), whereas int6⌬ cells were hypersensitive to caffeine (39). However, human Int-6 protein did not function in place of fission yeast Int6 to confer drug resistance when overexpressed, nor did it complement the caffeine sensitivity of int6⌬ cells. In contrast, the nuclear partitioning defect we observed in int6⌬ cells (Fig. 1B) was complemented by expression of human Int-6 protein. Thus, the role of Int-6 in mitosis may be conserved between fission yeast and mammals.
Finally, what are the implications of our results for MMTV tumor biology? Our hypothesis that the deletion of int6 ϩ has an impact on translation of a specific subset of mRNAs is consistent with the generation of tumors, which would contain normal rates of protein synthesis but display protein expression profiles different from that of normal cell lines. If unequal nuclear division observed for S. pombe int6⌬ haploids also happens frequently for mice carrying the MMTV genome inserted at one allele of int-6, this might increase the degree of aneuploidy and, hence, the frequency of tumorigenesis (40). However, it remains possible that disruption of the nuclear function of Int-6 is responsible for tumorigenesis by altering the transcription rates of certain genes. PML, protein the major constituent of the PML nuclear bodies where Int-6 resides, has been implicated as a tumor suppressor and transcriptional regulator (8). Although fission yeast Int6 was not localized in the nucleus (Fig. 2), the role of Int6 in activating multidrug resistance appears to be independent of the role of Int6 in translation (38). Understanding the role of Int-6 by using S. pombe as a model organism will contribute greatly to solving this complicated problem.