Eukaryotic cells develop asymmetric shapes suited for specific physiological functions. Morphogenesis of polarized domains and structures requires the amplification of molecular asymmetries by scaffold proteins and regulatory feedback loops. Small monomeric GTPases signal polarity, but how their downstream effectors and targets are spatially co-ordinated to break cell symmetry is poorly understood. Septins comprise a novel family of GTPases that polymerize into non-polar filamentous structures which scaffold and restrict protein localization. Recent studies show that septins demarcate distinct plasma membrane domains and cytoskeletal tracks, enabling the formation of intracellular asymmetries. Here, we review these findings and discuss emerging mechanisms by which septins promote cell asymmetry in fungi and animals.
From unicellular to multicellular organisms, cell asymmetry underlies many biological functions. Cells assume polarized shapes and develop specialized structures that are essential for their survival, propagation and communication. Understanding how cellular asymmetry arises from individual molecular interactions remains a final frontier for cell biology. What are the mechanisms that place individual molecules at discrete positions in the cell? How do molecular complexes generate spatially and functionally distinct structures? Answers to these questions are key in our efforts to reconstitute the cell and regenerate tissues and organs.
The discovery and study of the GTPases have provided key insights into the regulation of membrane- and cytoskeleton-dependent processes that affect cellular asymmetry. The Ras superfamily of small GTPases contains many signaling molecules that initiate symmetry breaking and trigger the assembly of cellular structures. However, the monomeric and diffusible nature of small GTPases does not lend itself to the maintenance of higher-order structures through time and space. Septins are GTPases that assemble into large and persistent structures capable of controlling and maintaining the localization of diverse proteins. Here, we review how septins promote and maintain cell asymmetry from fungi to mammals.
The Septin GTPase Module
Evolved from a common ancestral GTPase, the septin family arose by independent gene duplications in fungi and animals . Presence of septins in certain photosynthetic algae and ciliates suggests that the primordial septin gene evolved in some lineages, while it was lost from others (e.g. terrestrial plants) . Phylogenetic analyses place septins under the class of P-loop GTPases that include the Ras-like small GTPases and the myosin and kinesin motors, the ATP-binding domain of which bears a similar fold to canonical GTPases .
The septin family consists of multiple genes (e.g. 7 in budding yeast, 14 in humans). In fungi, septins are under developmental control with distinct genes expressed for vegetative growth and sporulation . In mammals, alternative splicing, transcription and translation start sites generate a diversity of septin isoforms, some of which have tissue-specific expression . The amino acid sequences of septins vary mainly in the N- and C-terminal extensions that flank the conserved GTP-binding domain and a 53 amino acid sequence termed the septin unique element (SUE) (Figure 1A). On the basis of sequence similarities, septins are classified into different groups that share similar features such as length of their C-terminal alpha helical tails or N-terminal extensions (Figure 1A). Through these domains, different septins may interact with different cellular proteins .
Septins use their GTP-binding and SUE domains as a dimerization/oligomerization interface (Figure 1B), assembling into non-polar filaments, which pair up to form highly ordered structures [9, 13-15]. These filamentous assemblies are posited to represent the biologically active form of septin GTPases . In mammalian cells, turnover of septins within their respective filaments is at least two to three times slower than actin and tubulin, but significantly more rapid than intermediate filament subunits . In fungi, septins are also added and/or exchanged after filaments polymerize [17-19] and septin filaments are capable of reorganizing in orientation after a transient loss of order [14, 20]. Therefore, septin filaments are relatively stable structures and yet dynamic enough to respond to intrinsic and extrinsic signals.
How septins cycle between their filamentous and non-filamentous states is not understood. Septins exchange and hydrolyze GTP intrinsically, but at a slow rate [21-25]. Rates of GTP hydrolysis vary among different septins; for example, members of the SEPT6 group are constitutively bound to GTP . An excess of GDP within septin filaments indicates that GTP hydrolysis does not lead to the disassembly of septin multimers [9, 21, 26]. To date, recycling of septin subunits appears to be mediated mainly by post-translational modifications and binding partners/cofactors (reviewed in ), which could act as surrogate GTPase activating proteins and guanine nucleotide exchange factors.
Although much remains to be known, septins are a novel regulatory module, the biological function of which depends on the assembly of higher-order multimers. The diversification and repeatability of septin subunits impose a level of specificity and co-operativity or avidity in their interaction with cellular proteins. Thus, septins can spatially determine the localization and amplify the concentration of specific proteins, contributing to the generation and maintenance of intracellular asymmetries.
Septins and asymmetry in fungi
Fungal cells polarize and generate asymmetries for both growth and reproduction. Formation of buds, mating projections and long tubular extensions (hyphae), all involve polarized growth in response to either an intrinsic growth program, or extracellular chemical or pheromone gradients. For this reason, fungi have served as powerful model systems to discover the organizing principles of cell asymmetry.
Septins emerged from the classic cell division cycle mutant screen performed in the early 1970s by Lee Hartwell and colleagues in the budding yeast Saccharomyces cerevisiae. Four genes (CDC3, CDC10, CDC11 and CDC12) that proved to encode for septin proteins were identified in this screen; the mutant cells had multiple nuclei and abnormally elongated buds. By the late 1970s, filamentous electron densities had been noted at the mother-bud neck and later work suggested that these filaments vanished in cdc10 mutants [28-30]. John Pringle localized the gene products of CDC3, CDC10, CDC11 and CDC12 and coined the term septin because of their inter-dependent colocalization at the septum region between the mother cell and the bud . Since their discovery, septins have been shown to localize at the base of round and tubular protrusions in many fungal organisms including Ashbya gossypii, Candida albicans[33, 34], Cryptococcus neoformans, Ustilago maydis and Aspergillus nidulans.
In S. cerevisiae, septins are recruited to incipient bud sites in response to the initial budding polarity cue from Cdc42p and assemble into a ring, which in turn maintains the polarized localization of Cdc42p . The mechanisms of ring assembly remain quite mysterious; however, it is clear that multiple effectors of Cdc42 are required for assembly and Cdc42 must actively cycle between the GTP and GDP bound states [18, 39]. The septin ring has an asymmetry such that some proteins localize specifically to the inside of the ring and other proteins localize to the outside of the ring [40-42]. How an asymmetric ring is generated out of non-polar septin filaments remains unknown. This asymmetry remains after bud emergence and as the ring transitions into an hourglass-shaped collar. Thus, the septin ring and collar maintain the asymmetric distribution of dozens of proteins by acting as scaffolds for the recruitment of cytoplasmic proteins and as diffusion barriers for the compartmentalization of membrane proteins [40, 43]. Through these general scaffold and barrier roles, septins influence diverse processes related to the morphogenesis and division of the budding yeast.
Similarly, septins affect the morphology of the filamentous fungus Ashbya gossypii. In this system, septins assemble into cortical bars that form a ring around the tubular shaped cells . These bars are strikingly similar to the bars seen at the base of the germ tubes of C. albicans and the mating projections (shmoos) of S. cerevisiae. Septin mutations alter the hyphal morphology of A. gossypii and reduce the number of mitotic nuclei at hyphal branchpoints . As mitosis occurs preferentially at branch points that contain septin rings, septins appear to control the location of nuclear division within the growing hyphae. Thus, septins could spatially regulate hyphal growth in response to nutrient gradients. Recent studies suggest that septins similarly regulate the emergence of germ tubes and branches in the multicellular fungus A. nidulans. Overall, septins appear to exert spatial control over fungal morphogenesis.
Shaping morphogenesis in animals
Animals consist of tissues and organs that are highly specialized in function and form. Septins are essential not only for the morphogenesis of specialized cell types, but also for the spatial guidance of their movement and division, which is critical for tissue and organ development.
In early embryogenesis, establishment of the body plan requires collective cell movement and elongation (convergent elongation) along the presumptive anterior–posterior axis. In the frog Xenopus laevis, septin knock-down leads to defective cell elongation and aberrant cell movements . Similar phenotypes have been observed in later stages of embryonic development during the formation of the nervous system. In C. elegans, motor and sensory neurons fail to guide their axons properly , and in mouse embryos, migrating cortical neurons lack leading processes and stall before reaching the cortical plate . Alterations in septin expression also affect the motility of T cells and breast epithelial cells in vitro. Septin-depleted T cells are characterized by abnormally elongated uropods and an excess of protrusions, which diminish the velocity and displacement of movement . Conversely, over-expression of a septin isoform in breast epithelia hinders the directional persistence of movement . Collectively, these findings underscore the role of septins in cell motility and the morphogenetic movements that drive tissue and organ development.
At the same time as the morphogenetic movements, embryonic cells differentiate into specialized cell types with distinct structures and functions. Several studies have implicated septins in the morphogenesis of neurons, epithelia and spermatozoa. In neuronal cells, axons and dendrites differentiate from membrane protrusions termed neurites. Neurite sprouting, axon and dendrite growth, branching of dendrites and the formation of dendritic spines require proper septin expression [49, 52-54]. Like neurons, epithelial cells develop specialized membrane domains (apical and basolateral), which mediate the exchange of ions and solutes between adjacent tissues, and organelles such as the primary cilium, which senses the extracellular environment. Recent work shows that septins are essential for the polarized columnar morphology of kidney epithelia and the formation of primary cilia [55, 56]. Similarly, septins are important for the morphogenesis of the sperm flagellum, a structure that resembles the primary cilium. In septin null spermatozoa, the flagellar cortex is severely compromised and sperm motility is disrupted [57, 58]. In addition, mitochondrial fission defects and retention of cytoplasmic droplets are observed .
From fungal buds and hyphae to mammalian neurites and cilia, septins appear to promote cell asymmetry by affecting the formation of protrusive membrane structures. But how do they do it? Decades of research in budding yeast have shown that septins are required for the localization of many proteins . The assembly and scaffold-like properties of septins have been extensively reviewed elsewhere [41, 59]. Here, we focus on studies that point to an active role for septins in promoting intracellular asymmetry by reorganizing cell membranes and the cytoskeleton.
Barricading, Deforming and Rigidifying Membrane Domains
Cell membranes are intrinsically patchy, containing microdomains of variable size and chemical composition . This asymmetry is amplified through the interaction of membrane lipids and proteins with the underlying cytoskeleton [61, 62]. Larger scale domains are then formed and serve as local platforms for signal transduction, vesicle exocytosis/endocytosis and the assembly of cellular structures . Recent findings indicate that septins affect the structure and shape of membrane macrodomains by three potential and non-mutually exclusive mechanisms: (i) blocking the lateral diffusion of membrane proteins, (ii) deforming the membrane bilayer, and (iii) rigidifying the cell membrane (Figure 2).
Biochemical studies have demonstrated that some yeast and mammalian septins bind directly to phosphoinositides through a conserved N-terminal polybasic domain [63, 64]. Yeast septins associate preferentially with phosphatidylinositol (PtdIns) 4-phosphate and PtdIns 5-phosphate [63, 65], and the mammalian SEPT5 binds strongly to PtdIns 4,5-biphosphate and PtdIns 3,4,5-triphosphate . SEPT2 and SEPT12 also associate with phospatidylserine, which is predominately present in the inner leaflet of cell membranes . On artificial monolayers of PtdIns 4,5-biphosphate, recombinant yeast septins assemble into a meshwork of parallel and orthogonal filaments, resembling the gauze-like structures seen by electron microscopy on the cell membrane of yeast spheroplasts [67, 68]. These findings suggest that septins are bona fide components of the membrane skeleton. Thus, septins could cross-link and corral membrane domains enriched in particular phosphoinositides.
The existence of septin barriers that constrain the lateral diffusion of membrane proteins is supported by work in several systems . Breakthrough studies by Barral and Snyder as well as Takizawa and Vale showed that septins are required for a diffusion barrier at the mother-bud neck of the budding yeast [69, 70]. Positioned between the cytoplasmic leaflets of the plasma membrane and the smooth ER, a filamentous septin ring is required to maintain a polarized localization of certain plasma membrane proteins in the growing bud and impedes the free diffusion of integral ER proteins (e.g. the translocon) [69, 71]. The septin diffusion barrier could help maintain asymmetries that are established by asymmetry in mRNA localization and translation [70-72]. During mitosis, splitting of the septin collar into two rings is posited to demarcate a new membrane compartment, in which exocyst proteins concentrate to direct cytokinesis . In dividing mammalian cells, a similar barrier limits the diffusion of inner leaflet membrane proteins and correlates with the presence of septins at the cleavage plane .
Septin-containing barriers have been shown to maintain the localization of ciliary and sperm membrane proteins. Bearing an uncanny resemblance to the septin rings in budding yeast, septins form a ring between the basal body and the periciliary plasma membrane, restricting the diffusion of ciliary membrane proteins while permitting the passage of cytoplasmic molecules . In SEPT2-depleted epithelia, primary cilia do not form or are small, leaky and defective in sonic hedgehog signaling . In Sept4−/− sperm cells, the cortical barrier between the principal and mid pieces of the sperm is breached, resulting in the mislocalization of a glycoprotein that is critical in spermiogenesis . Taken together, these findings show that septin diffusion barriers affect the formation of specialized structures and cell asymmetry.
Constraining lateral diffusion is a way to gather the molecular tools that sculpt cell membranes into distinct shapes, but recent evidence suggests that septins themselves could make a dent. Addition of pure SEPT2/6/7 complexes to spherical giant liposomes composed of PtdIns and phospatidylcholine triggers the formation of membrane tubules . This observation correlates with the propensity of septins to form kinked and circular filaments [9, 77]. Structurally, a flexible hinge-like region in the dimerization interface of SEPT2 allows SEPT2/6/7 subunits to bend by 25–30° as they assemble into longer filaments . Conceivably, membrane bilayers could cave in under the rigid meshwork of bent septin filaments. It is uncertain if septins can exert such an effect in vivo and on membranes of both negative and positive curvature. In some ways, septins are reminiscent of the Bin-Amphiphysin-Rvs (BAR)-domain proteins, which also oligomerize into structures of intrinsic curvature and form rigid scaffolds . Future work will show whether septins provide an alternative mechanism for membrane bending and deformation, or simply synergize with BAR-containing proteins.
Regardless of their ‘sculpting’ ability, lipid-bound septins boost the rigidity and stiffness of the cell membrane. This is the third mechanism by which septins break symmetry within the plane of the cell membrane. In moving T cells, septins localize to the middle zone of the plasma membrane, where they suppress the formation of blebs and protrusions . Thus, septins limit protrusive activity to the front (lamellipodia) and back (uropod) of the cell, reinforcing the directionality of cell movement. In the absence of septins, T cells exhibit uncontrolled blebbing and squeeze through submicron size pores more efficiently, which is indicative of decreased rigidity . Indeed, biophysical measurements of the effective viscosity and elasticity of the plasma membrane show a dramatic decrease in membrane tethering and rigidity after depletion of septins in HeLa cells . Notably, these effects are accompanied by changes in HeLa shape and morphology . Thus, septins can influence the asymmetry of cell shape by modulating the rigidity of the plasma membrane.
In summary, the local effects of septins on membrane protein composition, curvature and rigidity affect membrane shape and protrusive activity. Next, we discuss how septins interface with actin microfilaments (F-actin) and microtubules, both of which are essential for the establishment and maintenance of cellular asymmetry [80, 81].
Orienting Actin and Microtubule Tracks
Actin microfilaments and microtubules are rigid polymers, the polarized structure and dynamics of which reinforce the formation and expansion of membrane protrusions [81, 82]. Owing to their asymmetric growth, cytoskeletal polymers exert pushing forces, but also support the movement of molecular motors, which transport cargo and pull on cell membranes or the cytoskeleton itself . It is therefore essential that cytoskeletal tracks are properly positioned and motor movement is spatially controlled.
In budding yeast, actin cables polymerize from the tip and base of the growing bud, guiding vesicle transport to sites of membrane growth . At the presumptive bud site, assembly of actin cables takes place under the direction of Cdc42 and the formins Bni1 and Bnr1, which nucleate and elongate actin filaments [84-86]. Localization of Bnr1 requires septins, which are similarly recruited to the presumptive bud site by Cdc42, but independently of actin [87, 88]. Through the localization of Bnr1, septins are required for the organization of the actin cables that terminate at the mother-bud neck . In addition, septins might be indirectly important for the retrograde flow of actin cables, which is driven by the myosin II motor Myo1 . A recent study shows that the septin-binding protein Bni5 mediates recruitment of Myo1 to the bud neck . Thus, in budding yeast, the spatial organization and dynamics of the actomyosin cytoskeleton is in part guided by septins.
The interaction of septins with actomyosin is evolutionarily conserved. Mammalian septins associate with actin indirectly through myosin II and anillin, an actin bundling protein expressed only during mitosis [77, 91]. SEPT2 is posited to provide a molecular scaffold for the phosphorylation of myosin II by the Rho and citron kinases, and consequently, is essential for the ingression of the cleavage furrow during cytokinesis . Interestingly, septins and anillin are required for the asymmetric furrowing observed in early C. elegans embryos . This raises the possibility that septins function similarly in mitotic epithelia, the cleavage furrows of which ingress from the basal to the apical membrane . Thus, septins may ensure that new junctional complexes assemble on the apical side of the two daughter cells, maintaining the polarity of the monolayer.
In mammalian cells, septins have been linked to signaling pathways that control F-actin assembly [94, 95]. However, it is unknown if and how septins function in actin organization and protrusive activity. Septins colocalize with F-actin in filopodia and lamellopodial protrusions as well as in stress fibers and cortical bundles (Figure 3A; reviewed in ). The extent of colocalization depends on septin isoform, cell type and intracellular region. Loss of stress fibers is a common phenotype of septin depletion and is attributed to loss of myosin II phosphorylation . Given that myosin contractility is essential for the maturation and disassembly of focal adhesions, septins could underlie the asymmetry in focal adhesion disassembly, which is necessary for directional cell movement . Alternatively, septins might be involved in the actin-driven formation of filopodial and lamellopodial protrusions at the front of the cell.
Similar to the actin cables in budding yeast, microtubules in mammalian cells provide a structural framework for the long-range transport of membrane vesicles and organelles. The microtubule network consists of tracks with different dynamics and post-translational modifications . This structural heterogeneity affects the attachment and motility of microtubule motors, and enables spatial decisions such as determining the axes of cell migration and cell division. A growing body of evidence suggests that septins play a central role in the spatial organization of the microtubule network.
Septins colocalize with microtubules in a diversity of cell types and organisms . It is uncertain if every septin in a given complex interacts directly with microtubules, but SEPT9 has been shown to bind microtubules in vitro. In epithelial cells, septins decorate a subset of microtubules, which underlie Golgi sites of vesicle export, and during mitosis are positioned near the kinetochores of chromosomes aligned at the metaphase plate [56, 101]. Strikingly, septin-coated microtubules are predominately bundled and modified with polyglutamyl residues . Septin depletion results in loss of bundled and polyglutamylated microtubules, and consequently in defects in vesicle transport and chromosome alignment and segregation [56, 101, 102]. Thus, septins demarcate and modify microtubules at specific intracellular locales, and affect motor-dependent transport.
It is still too early to know whether septins spatially affect motor movement, but several studies have linked septins to three regulators of motor-cargo movement: (i) microtubule-associated proteins (MAPs), which affect the directionality and persistence of cargo movement by interfering with motor-binding ; (ii) post-translational modifications that alter the affinity of different kinesins and MAPs for microtubules ; and (iii) the movement and spatial orientation of microtubule tracks that influence the directionality of cargo movement . Regarding MAPs, two independent studies have demonstrated that septins interfere with MAP4 binding to microtubules [56, 105]. In addition, septins are reported to associate with motors . With respect to post-translational modifications, both acetylation and polyglutamylation are affected by septins, which could serve as scaffolds for tubulin-modifying enzymes [56, 105]. Finally, septins appear to suppress microtubule catastrophe in perinuclear and peripheral regions of the cytoplasm, providing spatial cues for the bundling and positioning of microtubules in polarizing epithelia . Similarly, in budding yeast, septins are required for proper positioning of the mitotic spindle relative to the plane of cytokinesis . Therefore, septins have conserved roles in the spatial positioning of the microtubule network.
In summary, septins provide a level of spatial regulation over the structural and functional heterogeneity of the actin and microtubule cytoskeletons. Although many of the mechanistic details are missing, we have now a rough picture for how septins control the spatial organization of cell membranes and the cytoskeleton in the establishment and maintenance of cellular asymmetry.
The diversification of septins suggests that this family of GTPases interfaces with multiple signaling networks and mechanisms of membrane and cytoskeletal organization. It is only recently that we have begun to scratch the surface of an intricate network of septin interactions and functions. Although some fundamental principles of assembly and function have emerged, we are still far from understanding the signaling pathways that instruct septin assembly and the mechanisms by which septins affect cell asymmetry. Moreover, the regulatory loops that control septin assembly and function remain unknown. Association of septins with cytoskeletal tracks of distinct post-translational modifications, which in turn depend on the presence of septins, hints at positive feedback loops. Septin binding to Cdc42 effectors and a Rho GEF suggest that septins are positioned at the nexus of signaling pathways, which feed back into one another and may modulate septin assembly accordingly. Future work will elucidate how septins are linked to the regulatory networks of cell polarity.
In conclusion, septins represent a specialized branch of the GTPase machinery. Amidst a sea of diffusive interactions, the relatively stable and polymeric septins arise to physically contain, nucleate and amplify the assembly of molecules into distinct cellular structures. For cell biologists, the study of septin GTPases provides new insights into the fundamentals of molecular morphogenesis. At the dawn of synthetic biology, septin GTPases may hold key insights into how cells and tissues are built.
We thank Jonathan Bowen (Spiliotis lab) for his help in designing the figures of this review. We apologize to any of our colleagues whose work was not referenced owing to space limitations. E. T. S. is supported by NIH grant GM097664 and the Antelo Devereux Award for Young Faculty. A. S. G. is supported by the National Science Foundation under grant MCB-0719126 and NIH under grant GM081506.