Human Bin3 complements the F-actin localization defects caused by loss of Hob3p, the fission yeast homolog of Rvs161p

To further investigate the function of BAR adaptor proteins, we identified a mammalian homolog of Rvs161p, termed Bin3 (Bridging INtegrator-3), and a Schizosaccharomyces pombe homolog of Rvs161p, termed Hob3p (H omolog O f B in3 ). Analysis of hob3 D mutants revealed an important role for Hob3p in regulation of F-actin localization, as was found for Rvs161p. The F-actin localization defect of hob3 D mutants was completely rescued by human BIN3 and partially rescued by RVS161, raising the possibility that Bin3 regulates F-actin localization in mammalian cells.


Introduction
BAR (Bin/Amphiphysin/Rvs domain) adaptor proteins, which include proteins encoded by the mammalian genes Amphiphysin, BIN1, and BIN2 and the Saccharomyces cerevisiae genes RVS167 and RVS161, are characterized by a unique N-terminal region termed the BAR domain. While their exact functions are largely unknown, BAR adaptor proteins appear to integrate signal transduction pathways that regulate membrane dynamics, F-actin cytoskeleton, and nuclear processes, roles that are highlighted in the nomenclature of two recently identified members of the family (Bridging INtegrators or BIN proteins).
Both genes encoding BAR adaptor proteins in S. cerevisiae were initially identified in a genetic screen for mutants that lost viability upon nutrient starvation [1; 2]. Subsequent work revealed that Rvs167p and Rvs161p form a physiological complex that regulates F-actin localization, cell polarity, bud formation, and endocytosis [2][3][4][5][6][7]. Rvs161p is also important for karyogamy, the nuclear fusion process which follows mating [8]. A variety of Rvs-interacting proteins were identified that are consistent with Rvs161p and Rvs167p functions in F-actin regulation, lipid metabolism, cell cycle integration, and nuclear processes [9][10][11][12][13]. Despite the importance of Rvs161p and Rvs167p in these diverse functions, RVS161 and RVS167 genes are not required for viability.
Three genes encoding BAR adaptor proteins have been described in mammalian cells.

ROUTHIER et al.
Two of these genes, amphiphysin and BIN1, encode structural orthologs of Rvs167p, whereas the third gene, BIN2, encodes a structurally unique protein. The expression patterns of each gene suggest different physiological roles: BIN1 is widely expressed whereas amphiphysin and BIN2 are tissue-restricted in their expression. The product of the amphiphysin gene, which was identified by virtue of its biochemical properties [14], is a neuronal adaptor protein that regulates synaptic vesicle endocytosis [15]. The restricted pattern of amphiphysin expression argues that its physiological function is limited to the specialized processes of synaptic vesicle recovery. In a similar way, BIN2 expression is restricted to hematopoietic cells. BIN2 function is undefined but appears to be nonredundant with other mammalian BAR proteins [16]. The BIN1 gene has a complex function(s) suggested by its diverse patterns of alternate splicing. BIN1 splice isoforms have been identified by virtue of interaction with the c-Myc oncoprotein, structural similarity to amphiphysin, interaction with the nuclear tyrosine kinase c-Abl, and characterization of the BIN1 gene itself [17][18][19][20][21][22][23]. Brain-specific isoforms, alternately termed amphiphysin II or amphiphysin-like isoforms, are exclusively cytosolic and can influence endocytosis [15].
However, only brain isoforms include regions required for interaction with key components of the endocytosis machinery [24]. Thus, it is unclear whether Bin1 participates in endocytosis outside the brain. Nuclear functions are suggested by the ability of muscle-specific and ubiquitous isoforms to localize to the nucleus and to functionally associate with the c-Myc and c-Abl proteins [17; 19; 25] containing appropriate nutritional supplements when necessary [39]. Expression from pREP2 plasmids was achieved by growing cells to early log phase in medium containing 0.06 mM thiamine, washing the cells 3x in thiamine-free medium, and resuspending the cells in the same medium. Budding yeast were grown in YPAD or SC medium lacking the appropriate nutritional supplements, in some cases with the addition of 6% (w/v) NaCl [40].
Immunofluorescence. Exponential phase S. pombe cultures were stained for F-actin as described [41] using AlexaFluor 488-conjugated phalloidin (Molecular Probes). Nuclei were stained with DAPI. Images were captured on a Nikon Eclipse TE300 microscope fitted with a Nikon Plan Fluor 100X objective using a Toshiba 3CCD camera. Images were manipulated using Image Pro Plus version 4.0 software (Media Cybernetics). Endocytosis. Exponential phase cultures of S. pombe cells were assayed for uptake of the lipophilic styryl dye FM4-64 (Molecular Probes) as described [42].
Northern Analysis. MTNI and MTNII human multiple tissue Northern blots obtained from Clontech (Palo Alto CA) were hybridized to 32 P-labelled probes for Bin3, Bin1, and amphiphysin I generated by the random priming method as per the vendor's instructions. The Bin1 and amphiphysin I probes have been described [14; 27]. The Bin3 probe was a 32 Plabelled 600 bp BamHI-BglII fragment of the human BIN3 cDNA. Hybridization of a Northern blot of RNA isolated from a panel of tumor cell lines, cultured and processed as described previously [17; 31], was performed using the BIN3 probe and a β-tubulin probe to normalize the blot.

Results
Bin3 encodes a widely expressed BAR adaptor protein related to Rvs161p. Sequences encoding Bin3, a novel human BAR adaptor protein, were identified using Rvs161p to search the EST database with the TBLASTN algorithm. Sequence analysis of full-length cDNA clones identified in this manner revealed that Bin3 was a protein of 253 residues in length and was comprised solely of a BAR domain, like Rvs161p (Fig. 1a, 1b). The Bin3 BAR domain was 27% identical to Rvs161p but less than 24% identical to other BAR domains (Table 1). Northern analysis of mouse tissue RNAs was performed to compare the Bin3 expression pattern to that of other mammalian BAR adaptor genes. A single mRNA species of ~2.2 kb was detected at similar levels in all embryonic and adult tissues examined, except for brain where Bin3 mRNA was undetectable (Fig. 2a). This wide expression pattern of Bin3 was similar to Bin1, which was widely expressed, but contrasted with amphiphysin, which was expressed primarily in brain, and Bin2, which was expressed primarily in hematopoetic cells. Since Bin1 expression was frequently decreased in malignant cells, Bin3 expression was determined in a panel of human tumor cell lines. All cell lines tested expressed Bin3 (Fig. 2b). We concluded that Bin3 was a widely expressed, BAR adaptor protein that was structurally most similar to S. cerevisiae Rvs161p.

S. pombe hob3+ encodes a BAR adaptor protein related to Rvs161p and Bin3.
Sequences encoding Hob3p (Homolog of Bin3) were identified using Rvs161p to search the S.

ROUTHIER et al.
pombe genome (Fig. 1a). Hob3p was 264 residues in length, 56% identical to S. cerevisiae Rvs161p, and 29% identical to Bin3 throughout its entire sequence (Fig. 1b). In contrast, the Hob3p BAR domain sequences were less than 26% identical to the BAR domain sequences in Bin1, Bin2, and amphiphysin (Table 1). Like Rvs161p and Bin3, Hob3p was comprised solely of a BAR domain, without the additional C-terminal sequences found in Rvs167p or known mammalian BAR adaptor proteins (Fig. 1b). The similarity of BAR sequences and lack of non-BAR sequences suggest that Rvs161p, Bin3, and Hob3 comprise a subfamily within the family of BAR adaptor proteins. The other BAR adaptor protein encoded by the S. pombe genome was identifed. The structure and characterization of this predicted protein, which was most similar to Rvs167p, will be described elsewhere 2 .
hob3∆ mutants have a cell division defect. We began by studying S. pombe hob3+ since fission yeast genetics allowed us to rapidly characterize Hob3p function. Using standard methods, haploid S. pombe strains were made where the entire coding region of hob3+ was replaced with the kanMX6 cassette, conferring resistance to G418 [34]. Southern analysis confirmed construction of a strain with the hob3∆ allele (Fig. 3a). Examination of hob3∆ mutants revealed a fraction of cells that were longer than hob3+ cells and contained more than two nuclei (Fig. 3b). In particular, about 9% of hob3∆ cells from an actively-growing culture contained more than two nuclei, with most of these elongated cells containing four nuclei (n = 108). In contrast, no cells with more than two nuclei were observed in a parallel culture of hob3+ cells (n = 125). Calcofluor staining showed that septal material separated most of the nuclei in hob3∆ cells (Fig. 3b). Furthermore, the hob3∆ cells contained increased amounts of Calcofluor-ROUTHIER et al. shown) and stopped dividing at lower cell density than hob3+ cells (Fig. 3c). Overexpression of hob3+ had no effect on hob3+ cells (data not shown). Hence, hob3+ had a role in cell division in fission yeast at the level of septation, with some proportion of cells failing to separate following septum formation and accumulating increased levels of septal material.
In particular, F-actin patches were found equally distributed along the entire length of mononuclear cells, with 86% of such mononuclear cells (n = 88) exhibiting this staining pattern versus 4% of cells from a hob3+ strain (n = 115). Second, medial F-actin rings and patches were rarely observed in hob3∆ mutants with two nuclei. In hob3∆ mutant cells with 2 nuclei, ROUTHIER et al. hob3∆ mutants respond normally to nutrient and osmotic stress. RVS161 was discovered in a screen for mutants that had reduced viability upon starvation for glucose, nitrogen or sulfur [1]. In these experiments, rvs161∆ mutants had a 35% reduction in cell viability after 48 hours in N005 low nitrogen medium. Further analysis revealed that rvs161∆ mutants showed dramatic morphologic changes in response to high salt and low nitrogen media, and more significant reductions in cell growth when shifted to media with high salt [48]. Based on these results, the response of hob3∆ mutants to nutrient and osmotic stress was tested. This analysis revealed that hob3∆ mutant cells were relatively insensitive to lack of nitrogen or elevated/decreased ROUTHIER et al. temperature, as assayed by growth on plates (Fig. 5). Furthermore, microscopic inspection of hob3∆ mutant cells following temperature, osmotic, or nutrient shift did not reveal detectable differences in cell morphology (data not shown). Based on these results, we concluded that hob3∆ mutants, unlike rvs161∆ mutants, respond like hob3+ cells to changes in temperature, osmolarity, and nutrients.  Hob3p, Bin3, and Rvs161p share common functions. To test this hypothesis, we determined whether ectopic expression of Bin3 and Rvs161p could rescue the defects of hob3∆ mutants. For this purpose, plasmids were constructed where Hob3p, Bin3, Rvs161p, and Rvs167p were expressed using the thiamine-repressible nmt1 promoter of S. pombe [33]. These plasmids or a control plasmid were then introduced into hob3∆ mutants and the fraction of elongated cells and F-actin staining patterns were quantified (Table 2, Fig. 7). As expected, ectopic expression of hob3+ complemented the cell elongation and F-actin defects in hob3∆ cells whereas the control vector had no effect. Interestingly, Bin3 expression also corrected the cell elongation and Factin defects of hob3∆ mutants while Rvs161 expression partially corrected the defects of hob3∆ mutants. In particular, hob3∆ mutant cells expressing Rvs161p were not elongated and contained easily detectable medial F-actin in mitotic cells. Rvs161p expression failed, however, to correct the mislocalization of F-actin patches in hob3∆ mutants. Rescue of hob3∆ mutants by Bin3 and Rvs161p was specific for these BAR adaptor proteins since Rvs167p expression did not correct the defects of hob3∆ mutants. We conclude that Bin3 and Rvs161p, but not Rvs167p, at least partially rescue the F-actin localization defects of hob3∆ mutants arguing that these proteins can perform similar functions.

hob3+, but not Bin3, complements the osmotic sensitivity of S. cerevisiae rvs161∆
mutants. Since both RVS161 and BIN3 complemented the F-actin localization defects of hob3∆ mutants, we tested whether expression of BAR-containing proteins could rescue a S. cerevisiae rvs161∆ mutant. Complementation was tested by the ability of a rvs161∆ strain to grow on synthetic dropout medium containing 6% NaCl. As expected, rvs161∆ cells lacking a plasmid or ROUTHIER et al.
containing a control plasmid failed to grow on media with 6% NaCl (Fig. 8). In contrast, rvs161∆ mutants expressing Rvs161p or Hob3p, but not Bin3 or Bin1, grew similarly to RVS161 cells (Fig 8). All strains which received a plasmid grew on synthetic dropout medium lacking 6% NaCl (data not shown). We conclude that, due to the lesser divergence between the Rvs161p and Hob3p proteins, Hob3p was able to complement the osmolarity defect of rvs161∆ null cells, but that the greater extent to which Rvs161p and Bin3 have diverged precluded complementation by Bin3.

Discussion
While the exact role of the N-terminal fold of the BAR family proteins is unknown, it is apparent that BAR family proteins are nonredundant in function. However, the BAR family may be subdivided based on structural considerations. Thus, a subset of BAR family members contain a C-terminal SH3 domain. In some cases, as for the Bin1-binding c-Abl oncoprotein, the protein partner responsible for interaction is known [19]. Another subset of BAR family members contain domains known to be involved in binding components of the endocytotic machinery and vesiculation [14; 22; 50]. Other domains, such as the c-Myc binding domain of Bin1 [17], are unique within the BAR family. The proteins described in the present study are characterized by a lack of identifiable functional domains outside of the BAR N-terminal fold.
In combination with the greater homology exhibited between members of this subset and their ability to cross-complement, this suggests to us that they form a bona fide subfamily within the BAR family of proteins. The inability of other BAR-containing proteins to complement defects ROUTHIER et al.
in the expression of these proteins, even in the case of proteins native to the same organism, supports this notion 2 [48]. While it is known that some members of the BAR family are able to interact, such as the yeast Rvs161p and Rvs167p proteins [7; 51], and amphiphysin and the brain isoform of Bin1 [22], the possibility of homo-or heterotypic interactions between other BAR family members remains to be determined.  [52][53][54][55]. sep2+ was identified in a screen for mutants with increased resistance to lysing enzymes, while sep12+ was isolated by application of the diploid enrichement screen of Chang et al [56]. It is worth noting that a fraction of sep2 cells contained double septa, which yielded two daughter cells and an anucleate minicell upon cleavage. Neither double septa nor anucleate cells were observed in cultures of hob3∆. It was shown that cultures of sep12 mutant cells contained 64% hyphae, a much larger percentage than the typical 10-15% observed in hob3∆ cultures. In addition, sep12 cells were sterile, while hob3∆ mated with normal kinetics. spn1+ is a member of the S. pombe septin family of proteins. As such, it has a role in promoting septation of fission yeast. However, a role in actin patch movement has not been predicted. rlc1+ encodes a myosin regulatory light chain which associates with the yeast ROUTHIER et al.
Myo2p and Myo3p gene products. Although the morphology of hob3∆ cells closely resembled that of rlc1 mutant cells, the latter were found to be cold-sensitive for growth, a condition not seen in hob3∆ cultures. Interestingly, Rvs161p, Hob3p and Bin3 are homologous to unconventional myosins; Rvs161p is 25% identical to Myo1p, the sole type II unconventional myosin of budding yeast; Hob3p is 20% identical to Myo2p, one of two myosins in S. pombe, and Bin3 is 24% identical to human type VI unconventional myosin. BLAST analysis of the Bin3 protein assigns unconventional type VI myosins as the most highly homologous non-BAR polypeptides.
The mislocalization of F-actin patches by hob3∆ null cells has been previously observed in mutants of the Arp2/3 complex of S. pombe, as well as in mutants of other, known actininteracting proteins such as the products of the cdc3+ and cdc8+ genes, which encode profilin and tropomyosin, respectively [44; 45; 57; 58]. However, these mutants do not display the linear, multiseptated morphology characteristic of hob3∆ cells. In addition, loss of Cdc3p, Cdc8p or Arp3p function is lethal, whereas loss of Hob3p is not. Given that the major defect in these mutants is probably actin-related, and that perturbations of actin organization generally result in gross morphological defects throughout the cell cycle, it is not surprising that the hob3+ gene is not essential, nor do hob3∆ cells exhibit the profound morphological abnormalities observed in more severe cases of loss of actin organization, such as in the cdc3 mutant [58].
Nevertheless, a defect in F-actin patch movement is apparent in hob3∆ cells. It remains to be determined whether Hob3p is directly associated with actin or with actin-binding proteins such as profilin, tropomyosin or the Arp2/3 complex.

ROUTHIER et al.
We observed cross-species complementation of the hob3∆ F-actin defect by the budding yeast homolog Rvs161p , and by the mammalian homolog Bin3. A partial rescue of the F-actin localization defect by Rvs161p was observed insofar as the majority of dividing cells regained medial F-actin staining, but failed to correctly localize F-actin patches during interphase. On the other hand, Bin3 was able to rescue both loss of medial F-actin and localization of F-actin to cortical patches during interphase. It was noted that budding yeast Rvs167p, which is not a member of the subfamily of BAR proteins defined by Rvs161p, Hob3p and Bin3, was unable to rescue the F-actin defect of hob3∆ cells. An alternative explanation for the partial complementation observed with Rvs161p and the lack of complementation seen in the case of Rvs167p could be due to decreased steady-state levels of these proteins in S. pombe. However, we have confirmed the presence of either Bin1 or Bin3 polypeptides in hob3+ and hob3∆ S. pombe strains transformed with pREP2-based expression vectors. Furthermore, we were able to ascertain that the resulting Bin1 polypeptide failed to correct the F-actin defect observed in hob3∆ cells (data not shown). We thus favor the interpretation that Bin3 is the mammalian homolog of Rvs161p and Hob3p, but that there exists a degree of divergence in this gene during evolution.     Cell density was determined by counting appropriate dilutions of cells with the aid of a hemocytometer. Viability was determined by a colony formation assay on solid YE medium.
All measurements were obtained in triplicate.       Table 2. Quantification of F-actin distribution in hob3∆ mutants expressing BARcontaining polypeptides. Cells were grown to exponential phase in the medium indicated, fixed, and stained with AlexaFluor 488-phalloidin and DAPI. Cells were counted and distributed according to their nuclear content and F-actin staining pattern.
The number (n) of cells analyzed is indicated in parentheses. N/A, not applicable.