A protein interaction map of Drosophila melanogaster.

Drosophila melanogaster is a proven model system for many aspects of human biology. Here we present a two-hybrid-based protein-interaction map of the fly proteome. A total of 10,623 predicted transcripts were isolated and screened against standard and normalized complementary DNA libraries to produce a draft map of 7048 proteins and 20,405 interactions. A computational method of rating two-hybrid interaction confidence was developed to refine this draft map to a higher confidence map of 4679 proteins and 4780 interactions. Statistical modeling of the network showed two levels of organization: a short-range organization, presumably corresponding to multiprotein complexes, and a more global organization, presumably corresponding to intercomplex connections. The network recapitulated known pathways, extended pathways, and uncovered previously unknown pathway components. This map serves as a starting point for a systems biology modeling of multicellular organisms, including humans.

A localized GTPase exchange factor, Bud5, determines the orientation of division axes in yeast.

GTPases are widespread in directing cytoskeletal rearrangements and affecting cellular organization. How they do so is not well understood. Yeast cells divide by budding, which occurs in two spatially programmed patterns, axial or bipolar [1-3]. Cytoskeletal polarization to form a bud is governed by the Ras-like GTPase, Bud1/Rsr1, in response to cortical landmarks. Bud1 is uniformly distributed on the plasma membrane, so presumably its regulators, Bud5 GTPase exchange factor and Bud2 GTPase activating protein, impart spatial specificity to Bud1 action [4]. We examined the localizations of Bud5 and Bud2. Both Bud1 regulators associate with cortical landmarks designating former division sites. In haploids, Bud5 forms double rings that encircle the mother-bud neck and split upon cytokinesis so that each progeny cell inherits Bud5 at the axial division remnant. Recruitment of Bud5 into these structures depends on known axial landmark components. In cells undergoing bipolar budding, Bud5 associates with multiple sites, in response to the bipolar landmarks. Like Bud5, Bud2 associates with the axial division remnant, but rather than being inherited, Bud2 transiently associates with the remnant in late G1, before condensing into a patch at the incipient bud site. The relative timing of Bud5 and Bud2 localizations suggests that both regulators contribute to the spatially specific control of Bud1 GTPase.

Multigenerational cortical inheritance of the Rax2 protein in orienting polarity and division in yeast.

Diploid yeast cells repeatedly polarize and bud from their poles, probably because of highly stable marks of unknown composition. Here, Rax2, a membrane protein, was shown to behave as such a mark. The Rax2 protein itself was inherited immutably at the cell cortex for multiple generations, and Rax2 was shown to have a half-life exceeding several generations. The persistent inheritance of cortical protein markers would provide a means to couple a cell's history to the future development of a precise morphogenetic form.

Cyk3, a novel SH3-domain protein, affects cytokinesis in yeast.

Cytokinesis requires the wholesale reorganization of the cytoskeleton and secretion to complete the division of one cell into two. In the budding yeast Saccharomyces cerevisiae, the IQGAP-related protein Iqg1 (Cyk1) promotes cytokinetic actin ring formation and is required for cytokinesis and viability [1-3]. As the actin ring is not essential for cytokinesis or viability, Iqg1 must act by another mechanism [4]. To uncover this mechanism, a screen for high-copy suppressors of the iqg1 lethal phenotype was performed. CYK3 suppressed the requirement for IQG1 in viability and cytokinesis without restoration of the actin ring, demonstrating that CYK3 promotes cytokinesis through an actomyosin-ring-independent pathway. CYK3 encodes a novel SH3-domain protein that was found in association with the actin ring and the mother-bud neck. cyk3 null cells had misshapen mother-bud necks and were deficient in cytokinesis. In the cyk3 null strain, actin rearrangements associated with cytokinesis appeared normal, suggesting that the phenotype reflects a defect in secretory targeting or septal synthesis. Deletion of either cyk3 or hof1 alone results in a mild cytokinetic phenotype [5-7], but deletion of both genes resulted in lethality and a complete cytokinetic block, suggesting overlapping function. Thus, Cyk3 appears to be important for cytokinesis and acts potentially downstream of Iqg1.

Molecular linkage underlying microtubule orientation toward cortical sites in yeast.

Selective microtubule orientation toward spatially defined cortical sites is critical to polarized cellular processes as diverse as axon outgrowth and T cell cytotoxicity. In yeast, oriented cytoplasmic microtubules align the mitotic spindle between mother and bud. The cortical marker protein Kar9 localizes to the bud tip and is required for the orientation of microtubules toward this region. Here, we show that Kar9 directs microtubule orientation by acting through Bim1, a conserved microtubule-binding protein. Bim1 homolog EB1 was originally identified through its interaction with adenomatous polyposis coli (APC) tumor suppressor, raising the possibility that an APC-EB1 linkage orients microtubules in higher cells.

Cell cycle programs of gene expression control morphogenetic protein localization.

Genomic studies in yeast have revealed that one eighth of genes are cell cycle regulated in their expression. Almost without exception, the significance of cell cycle periodic gene expression has not been tested. Given that many such genes are critical to cellular morphogenesis, we wanted to examine the importance of periodic gene expression to this process. The expression profiles of two genes required for the axial pattern of cell division, BUD3 and BUD10/AXL2/SRO4, are strongly cell cycle regulated. BUD3 is expressed close to the onset of mitosis. BUD10 is expressed in late G1. Through promotor-swap experiments, the expression profile of each gene was altered and the consequences examined. We found that an S/G2 pulse of BUD3 expression controls the timing of Bud3p localization, but that this timing is not critical to Bud3p function. In contrast, a G1 pulse of BUD10 expression plays a direct role in Bud10p localization and function. Bud10p, a membrane protein, relies on the polarized secretory machinery specific to G1 to be delivered to its proper location. Such a secretion-based targeting mechanism for membrane proteins provides cells with flexibility in remodeling their architecture or evolving new forms.

Cell polarity in yeast.

Subcellular asymmetry, cell polarity, is fundamental to the diverse specialized functions of eukaryotic cells. In yeast, cell polarization is essential to division and mating. As a result, this highly accessible experimental system serves as a paradigm for deciphering the molecular mechanisms underlying the generation of polarity. Beyond yeast, cell polarity is essential to the partitioning of cell fate in embryonic development, the generation of axons and their guidance during neuronal development, and the intimate communication between lymphocytes within the immune system. The polarization of yeast cells shares many features with that of these more complex examples, including regulation by both intrinsic and extrinsic cues, conserved regulatory molecules such as Cdc42 GTPase, and asymmetry of the cytoskeleton as its centerpiece. This review summarizes the molecular pathways governing the generation of cell polarity in yeast.

Novel Cdc42-binding proteins Gic1 and Gic2 control cell polarity in yeast.

Cdc42p, a Rho-related GTP-binding protein, regulates cytoskeletal polarization and rearrangements in eukaryotic cells, but the effectors mediating this control remain unknown. Through the use of the complete yeast genomic sequence, we have identified two novel Cdc42p targets, Gic1p and Gic2p, which contain consensus Cdc42/Rac interactive-binding (CRIB) domains and bind specifically to Cdc42p-GTP. Gic1p and Gic2p colocalize with Cdc42p as cell polarity is established during the cell cycle and during mating in response to pheromones. Cells deleted for both GIC genes exhibit defects in actin and microtubule polarization similar to those observed in cdc42 mutants. Finally, the interaction of the Gic proteins and Cdc42p is essential, as mutations in the CRIB domain of Gic2p that eliminate Cdc42p binding disrupt Gic2p localization and function. Thus, Gic1p and Gic2p define a novel class of Cdc42p targets that are specifically required for cytoskeletal polarization in vivo.

An IQGAP-related protein controls actin-ring formation and cytokinesis in yeast.

BACKGROUND: Proteins of the IQGAP family have been identified as candidate effectors for the Rho family of GTPases; however, little is known about their cellular functions. The domain structures of IQGAP family members make them excellent candidates as regulators of the cytoskeleton: their sequences include an actin-binding domain homologous to that found in calponin, IQ motifs for interaction with calmodulin, and a GTPase-binding domain. RESULTS: The genomic sequence of Saccharomyces cerevisiae revealed a single gene encoding an IQGAP family member (denoted IQGAP-related protein: Iqg1). Iqg1 and IQGAPs share similarity along their entire length, with an amino-terminal calponin-homology (CH) domain, IQ repeats, and a conserved carboxyl terminus. In contrast to IQGAPs, Iqg1 lacks an identifiable GAP motif, a WW domain, and IR repeats, although the functions of these domains in IQGAPs are not well defined. Deletion of the IQG1 gene resulted in lethality. Cellular defects included a deficiency in cytokinesis, altered actin organization, aberrant nuclear segregation, and cell lysis. The primary defect appeared to be a cytokinesis defect, and the other problems possibly arose as a consequence of this initial defect. Consistent with a role in cytokinesis, Iqg1 co-localizes with an actin ring encircling the mother-bud neck late in the cell cycle -a putative cytokinetic ring. IQG1 overexpression resulted in premature actin-ring formation, suggesting that Iqg1 activity temporally controls formation of this structure during the cell cycle. CONCLUSIONS: Yeast IQGAP-related protein, Iqg1, is an important regulator of cellular morphogenesis, inducing actin-ring formation in association with cytokinesis.

Rac and Cdc42 induce actin polymerization and G1 cell cycle progression independently of p65PAK and the JNK/SAPK MAP kinase cascade.

Rac and Cdc42 regulate a variety of responses in mammalian cells including formation of lamellipodia and filopodia, activation of the JNK MAP kinase cascade, and induction of G1 cell cycle progression. Rac is also one of the downstream targets required for Ras-induced malignant transformation. Rac and Cdc42 containing a Y40C effector site substitution no longer intact with the Ser/Thr kinase p65PAK and are unable to activate the JNK MAP kinase pathway. However, they still induce cytoskeletal changes and G1 cell cycle progression. Rac containing an F37A effector site substitution, on the other hand, no longer interacts with the Ser/Thr kinase p160ROCK and is unable to induce lamellipodia or G1 progression. We conclude that Rac and Cdc42 control MAP kinase pathways and actin cytoskeleton organization independently through distinct downstream targets.

Generation of cell polarity in yeast.

Yeast cells constitute an excellent system for studying cell polarity. They polarize by means of internally programmed patterns of cell division; they polarize chemotropically towards a partner during mating; and they utilize polarity to segregate cell-fate determinants during division. In the past year, considerable progress has been made towards increasing our understanding of the molecular mechanisms underlying each of these processes.

Human Ste20 homologue hPAK1 links GTPases to the JNK MAP kinase pathway.

BACKGROUND: The Rho-related GTP-binding proteins Cdc42 and Rac1 have been shown to regulate signaling pathways involved in cytoskeletal reorganization and stress-responsive JNK (Jun N-terminal kinase) activation. However, to date, the GTPase targets that mediate these effects have not been identified. PAK defines a growing family of mammalian kinases that are related to yeast Ste20 and are activated in vitro through binding to Cdc42 and Rac1 (PAK: p21 Cdc42-/Rac-activated kinase). Clues to PAK function have come from studies of Ste20, which controls the activity of the yeast mating mitogen-activated protein (MAP) kinase cascade, in response to a heterotrimeric G protein and Cdc42. RESULTS: To initiate studies of mammalian Ste20-related kinases, we identified a novel human PAK isoform, hPAK1. When expressed in yeast, hPAK1 was able to replace Ste20 in the pheromone response pathway. Chemical mutagenesis of a plasmid encoding hPAK1, followed by transformation into yeast, led to the identification of a potent constitutively active hPAK1 with a substitution of a highly conserved amino-acid residue (L107F) in the Cdc42-binding domain. Expression of the hPAK1(L107F) allele in mammalian cells led to specific activation of the Jun N-terminal kinase MAP kinase pathway, but not the mechanistically related extracellular signal-regulated MAP kinase pathway. CONCLUSIONS: These results demonstrate that hPAK1 is a GTPase effector controlling a downstream MAP kinase pathway in mammalian cells, as Ste20 does in yeast. Thus, PAK and Ste20 kinases play key parts in linking extracellular signals from membrane components, such as receptor-associated G proteins and Rho-related GTPases, to nuclear responses, such as transcriptional activation.

Bud10p directs axial cell polarization in budding yeast and resembles a transmembrane receptor.

BACKGROUND: The budding yeast Saccharomyces cerevisiae can bud in two spatially programmed patterns: axial or bipolar. In the axial budding pattern, cells polarize and divide adjacent to the previous site of cell separation, in response to a cell-division remnant, which includes Bud3p, Bud4p and septin proteins. This paper investigates the role of an additional component of the cell-division remnant, Bud10p, in axial budding. RESULTS: The sequence of Bud10p predicts a protein that contains a single trans-membrane domain but lacks similarity to known proteins. Subcellular fractionations confirm that Bud10p is associated with membranes. Bud10p accumulates as a patch at the bud site prior to bud formation, and then persists at the mother-bud neck as the bud grows. Towards the end of the cell cycle, the localization of Bud10p refines to a tight double ring which splits at cytokinesis into two single rings, one in each progeny cell. Each single ring remains until a new concentration of Bud10p forms at the developing axial bud site, immediately adjacent to the old ring. Certain aspects of Bud10p localization are dependent upon BUD3, suggesting a close functional interaction between Bud10p and Bud3p. CONCLUSIONS: Bud10p is the first example of a transmembrane protein that controls cell polarization during budding. Because Bud10p contains a large extracellular domain, it is possible that Bud10p functions in a manner analogous to an extracellular matrix receptor. Clusters of Bud10p at the mother-bud neck formed in response to Bud3p (and possibly to an extracellular cue, such as a component of the cell wall), might facilitate the docking of downstream components that direct polarization of the cytoskeleton.

A mechanism of Bud1p GTPase action suggested by mutational analysis and immunolocalization.

BACKGROUND: Yeast cells polarize, bud, and divide in either of two genetically programmed patterns: axial or bipolar. The Saccharomyces cerevisiae gene BUD1 (also known as RSR1) encodes a Ras-related GTPase critical for selection of a bud sites in these patterns. To distinguish between possible mechanisms of Bud1p action, we have examined the function and subcellular localization of Bud1p in a variety of mutant situations. RESULTS: Bud1p has 57% identity to H-ras, except for an 81 amino-acid insertion near the carboxyl terminus. Mutation of the proposed BUD1 effector domain produces a protein which can neither support normal patterns of budding nor interact with CDC24, which encodes a likely Bud1p effector. A version of Bud1p deleted for the 81 amino-acid unique region is essentially wild-type. Immunofluorescence and cell fractionation indicate that Bud1p remains associated with the membrane throughout its GTPase cycle. Both potential effectors of Bud1p, Bem1p and Cdc24p, are also membrane associated even in the absence of Bud1p, suggesting that Bud1p is not required to dock these proteins from the cytosol but may couple these proteins and others within the plane of the plasma membrane. CONCLUSIONS: Based upon observations reported here and elsewhere, we propose a novel mechanism of Bud1p GTPase action. Like Ras, Bud1p GTPase is constitutively associated with the plasma membrane; however, concentrated activities of Bud5p GDP-GTP exchange factor and Bud2p GTPase-activating protein at the future bud site promote rapid cycling of Bud1p between GTP- and GDP-bound conformations in a spatially restricted manner. Local GTPase cycling serves to efficiently nucleate complexes between polarity establishment functions that direct cytoskeletal polarization towards the bud site.

Patterns of bud-site selection in the yeast Saccharomyces cerevisiae.

Cells of the yeast Saccharomyces cerevisiae select bud sites in either of two distinct spatial patterns, known as axial (expressed by a and alpha cells) and bipolar (expressed by a/alpha cells). Fluorescence, time-lapse, and scanning electron microscopy have been used to obtain more precise descriptions of these patterns. From these descriptions, we conclude that in the axial pattern, the new bud forms directly adjacent to the division site in daughter cells and directly adjacent to the immediately preceding division site (bud site) in mother cells, with little influence from earlier sites. Thus, the division site appears to be marked by a spatial signal(s) that specifies the location of the new bud site and is transient in that it only lasts from one budding event to the next. Consistent with this conclusion, starvation and refeeding of axially budding cells results in the formation of new buds at nonaxial sites. In contrast, in bipolar budding cells, both poles are specified persistently as potential bud sites, as shown by the observations that a pole remains competent for budding even after several generations of nonuse and that the poles continue to be used for budding after starvation and refeeding. It appears that the specification of the two poles as potential bud sites occurs before a daughter cell forms its first bud, as a daughter can form this bud near either pole. However, there is a bias towards use of the pole distal to the division site. The strength of this bias varies from strain to strain, is affected by growth conditions, and diminishes in successive cell cycles. The first bud that forms near the distal pole appears to form at the very tip of the cell, whereas the first bud that forms near the pole proximal to the original division site (as marked by the birth scar) is generally somewhat offset from the tip and adjacent to (or overlapping) the birth scar. Subsequent buds can form near either pole and appear almost always to be adjacent either to the birth scar or to a previous bud site. These observations suggest that the distal tip of the cell and each division site carry persistent signals that can direct the selection of a bud site in any subsequent cell cycle.

Role of Bud3p in producing the axial budding pattern of yeast.

Yeast cells can select bud sites in either of two distinct spatial patterns. a cells and alpha cells typically bud in an axial pattern, in which both mother and daughter cells form new buds adjacent to the preceding division site. In contrast, a/alpha cells typically bud in a bipolar pattern, in which new buds can form at either pole of the cell. The BUD3 gene is specifically required for the axial pattern of budding: mutations of BUD3 (including a deletion) affect the axial pattern but not the bipolar pattern. The sequence of BUD3 predicts a product (Bud3p) of 1635 amino acids with no strong or instructive similarities to previously known proteins. However, immunofluorescence localization of Bud3p has revealed that it assembles in an apparent double ring encircling the mother-bud neck shortly after the mitotic spindle forms. The Bud3p structure at the neck persists until cytokinesis, when it splits to yield a single ring of Bud3p marking the division site on each of the two progeny cells. These single rings remain for much of the ensuing unbudded phase and then disassemble. The Bud3p rings are indistinguishable from those of the neck filament-associated proteins (Cdc3p, Cdc10p, Cdc11p, and Cdc12p), except that the latter proteins assemble before bud emergence and remain in place for the duration of the cell cycle. Upon shift of a temperature-sensitive cdc12 mutant to restrictive temperature, localization of both Bud3p and the neck filament-associated proteins is rapidly lost. In addition, a haploid cdc11 mutant loses its axial-budding pattern upon shift to restrictive temperature. Taken together, the data suggest that Bud3p and the neck filaments are linked in a cycle in which each controls the position of the other's assembly: Bud3p assembles onto the neck filaments in one cell cycle to mark the site for axial budding (including assembly of the new ring of neck filaments) in the next cell cycle. As the expression and localization of Bud3p are similar in a, alpha, and a/alpha cells, additional regulation must exist such that Bud3p restricts the position of bud formation in a and alpha cells but not in a/alpha cells.

Regulation of the polarization of T cells toward antigen-presenting cells by Ras-related GTPase CDC42.

The mechanisms by which cells rapidly polarize in the direction of external signals are not understood. Helper T cells, when contacted by an antigen-presenting cell, polarize their cytoskeletons toward the antigen-presenting cell within minutes. Here we show that, in T cells, the mammalian Ras-related GTPase CDC42 (the homologue of yeast CDC42, a protein involved in budding polarity) can regulate the polarization of both actin and microtubules toward antigen-presenting cells but is not involved in other T-cell signaling processes such as those which culminate in interleukin 2 production. Although T-cell polarization appears dispensable for signaling leading to interleukin 2 production, polarization may direct lymphokine secretion towards the correct antigen-presenting cell in a crowded cellular environment. Inhibitor experiments suggest that phosphatidylinositol 3-kinase is required for cytoskeletal polarization but that calcineurin activity, known to be important for other aspects of signaling, is not. Apparent conservation of CDC42 function between yeast and T cells suggests that this GTPase is a general regulator of cytoskeletal polarity in many cell types.