Sep(t)arate or not – how some cells take septin-independent routes through cytokinesis

Eukaryotic cell division involves a highly coordinated separation of the sister chromatids and reassembly of the daughter nuclei, and is essential for the transmission of the genetic program. However, division is not complete and productive until the daughters achieve full cellular integrity. This requires the mechanical separation of the cytoplasm and reestablishment of a continuous outer cell membrane, a process termed cytokinesis, which is achieved by diverse mechanisms in the different kingdoms of living beings.

In the budding yeast Saccharomyces cerevisiae, a septum is formed between the two daughter cells to fulfill the task of cytokinesis. Interestingly, the first yeast screens for genes required for progression of the cell division cycle revealed various CDC genes coding for structurally related GTP-binding proteins that are necessary for septum formation (Cdc3p, Cdc10p, Cdc11p and Cdc12p), which were subsequently designated as septins (Hartwell et al., 1970). Septins have since been identified in filamentous fungi, in metazoan model organisms, such as the nematode Caenorhabditis elegans and the fruit fly Drosophila melanogaster, and in various mammals, including mouse and human (reviewed in Weirich et al., 2008).

In metazoa, cytokinesis is more complex and, to a certain degree, also dependent on the context of the tissues and organs. However, the formation of a midbody (or Flemming body), a dense structure consisting of microtubules and other proteins not dissimilar to the septum in yeast, is crucial in late cytokinesis and for abscission of the daughter cells (Flemming, 1891). Septins are constituents of this midbody (reviewed in Hu et al., 2012) and their role in cytokinesis (El Amine et al., 2013) and chromosome segregation (Spiliotis et al., 2005) has been firmly established in various model organisms. Owing to their unique mode of assembly, which is unlike that found in the microtubule and microfilament networks, septins have been increasingly acknowledged as the fourth component of the animal cell cytoskeleton (Box 1) (reviewed in Mostowy and Cossart, 2012; Fung et al., 2014). Septins colocalize with the microtubule and actin cytoskeleton to a variable degree depending on the cell type and cell cycle stage analyzed (Xie et al., 2007; Spiliotis, 2010; Bowen et al., 2011; Ghossoub et al., 2013). Actin-depolymerizing agents have been shown to affect septin assembly, and depletion of septins has been associated with altered F-actin content in some cell types (Kinoshita et al., 2002; Kim et al., 2011). Recent studies have shown that septins can directly induce curvature of actin bundles, both in vitro and in vivo (Mavrakis et al., 2014). Septins have also been shown to be necessary for microtubule polymerization and vice versa (Nagata et al., 2003; Spiliotis et al., 2008; Bowen et al., 2011; Sellin et al., 2011b). The microtubule cytoskeleton is a prerequisite for spindle formation and mitotic chromosome segregation, and a similar and unquestionable role had been assigned to septins in animal cell cytokinesis. However, recent studies have indicated that the requirement of septins for cytokinesis is context dependent and that there are specific scenarios in which cells can undergo cytokinesis in the absence of septins (Tooley et al., 2009; Founounou et al., 2013; Guillot and Lecuit, 2013; Menon et al., 2014).

In this Commentary, we will focus on the cell-type-specific requirement of the core septin SEPT7 for cytokinesis in Drosophila and in mice. Our aim is to demonstrate differences in the dependence of cytokinesis on septins by focusing on the cellular context of their action in these organisms rather than the still largely unknown molecular mechanisms. We will discuss a role for a complex microtubule–septin interplay in the septin-dependence of mammalian cytokinesis. In light of recent findings that suggest septin-independence of hematopoietic cell proliferation, we will summarize the non-canonical functions of septins in these cells. Finally, we will discuss the possibility of septin targeting as a means of inhibiting cell division and proliferation in a cell-type-specific manner – an approach that potentially could become important for the treatment of solid tumors arising from cells that display septin-dependent proliferation, without adverse effects on septin-independent hematopoiesis.

To understand the role of septins in cytokinesis of metazoans, we first have to define the different steps of cytokinesis (see e.g. Green et al., 2012). As depicted in Fig. 1A, cytokinesis has already started by the anaphase of mitosis, when an actomyosin contractile ring is formed. During telophase, which is characterized by completed chromatid segregation, ingression of the contractile ring then leads to the development of an intercellular bridge, in which a midbody containing a midbody ring is observed. Finally, on the flank of the midbody, the abscission site is generated, which causes membrane separation of the sister cells.

So far, acknowledged roles of septins have been assigned to all of the above steps. In Drosophila, the core septin Peanut (Pnut; see Table 1), a homolog of the mammalian SEPT7, is colocalized with the contractile ring and, subsequent to ring ingression, accumulates in the intercellular bridge (Neufeld and Rubin, 1994). Its depletion leads to cytokinetic failure and peanut-shaped multinucleated cells. Furthermore, septin filaments serve as membrane anchors necessary for coupling the contractile ring to the cell cortex, a process that also requires anillin, which crosslinks septins to myosin (Kechad et al., 2012; Field et al., 2005) (Fig. 1C). At membrane-associated septin complexes, anilin binds to Pnut with its C-terminus and, with its N-terminus, anchors the contractile ring to the membrane (Kechad et al., 2012). Interestingly, Pnut, as well as other Drosophila septins (Sep1 and Sep2), are also able to bundle actin filaments into ring-like structures in vitro in the absence of anillin (Mavrakis et al., 2014), indicating that there is an additional role of septins in the structural assembly of the contractile ring. Coordinated interactions between septin and anillin are also crucial for the transition from the contractile ring to the midbody ring (El Amine et al., 2013). Anillin has also been shown to interact with the Drosophila Cindr protein and stabilize the intercellular bridge during cytokinesis (Haglund et al., 2010). CD2-associated protein (CD2AP), a mammalian homolog of Cindr, interacts with septins and has also been shown to localize to the midbody and regulate cytokinesis in HeLa cells (Monzo et al., 2005; Wasik et al., 2012). In addition, in HeLa cells, formation and maturation of the intercellular bridge is dependent on septin filaments and septin–anillin rings; constriction at the abscission sites and the recruitment of the abscission machinery are primed by anillin–septin complexes (Renshaw et al., 2014). In a remarkable study, Trimble and co-workers presented data for distinct roles for specific septin isoforms in mammalian cytokinesis, wherein individual depletion of SEPT2 or SEPT11, or pan-septin depletion by small interfering RNA (siRNA) against SEPT7 led to cleavage furrow ingression defects in HeLa cells, whereas depletion of SEPT9 only caused late stage defects in midbody abscission (Estey et al., 2010). In mouse fibroblasts, a SEPT7-containing midbody ring has been shown to surround the microtubules of the intercellular bridge (Fig. 1B). Interestingly, in SEPT7-deficient fibroblasts, persistent midbody microtubule structures and stabilized intercellular bridges have been detected, suggesting defective abscission (Menon et al., 2014). Taken together, septins are clearly essential players in the complex processes of cytokinesis.

However, in addition to their role in cytokinesis, a role of septins in mitosis in mammals has been described. In mammals, septins 2, 6 and 7 form the core hexameric subunit of higher-order septin structures (see Box 1), and depletion of one of these septins leads to decreased levels of the others owing to destabilization of the core complex. Interestingly, septin 2 and 6 have been detected in the microtubule spindle of HeLa and MDCK cells during metaphase (Spiliotis et al., 2005). Knockdown of SEPT2, SEPT6 or SEPT7 leads to the accumulation of pro-metaphase cells and a significant reduction in the amount of the centromere-associated protein E (CENP-E) at the kinetochores, indicating that the septin core complex might have a more general role as a scaffold at the midplane of the mitotic spindle (Spiliotis et al., 2005; Zhu et al., 2008). Therefore, in addition to their key role in cytokinesis, septins also appear to be essential in mitosis. Thus, the data from diverse vertebrate and invertebrate models have established septins as vital molecular components for successful cell division.

Despite the pivotal role of septins in animal cell cytokinesis, no lethal septin mutants have been identified in fission yeast (Berlin et al., 2003). However, these observations were initially overlooked owing to concerns regarding isoform redundancy and considerable differences in the physiology between the budding and fission yeasts (Balasubramanian et al., 2004; Pringle, 2008). Other initial indications that septins might be dispensable for cytokinesis came from the C. elegans and Drosophila model systems, as septin-deficient worms and flies are able to undergo early embryogenesis without visible cellularization defects (Neufeld and Rubin, 1994; Nguyen et al., 2000). These effects are clearly context specific, rather than representing a lack of specific septin functions in these organisms, as the mutant flies and worms do have defects in adult development and morphogenesis.

Interestingly, a detailed study analyzing the asymmetric furrow ingression in cytokinesis in C. elegans embryos has established a role for septins in the generation of contractile ring asymmetry (Maddox et al., 2007). Asymmetric furrowing is thought to serve a mechanical function, making the cytokinesis robust and resistant to alteration in the contractility of the actomyosin ring. It is possible that such a process is dispensable in the early embryo owing to the relative simplicity of the tissue architecture. In more complex tissues, the orientation of the contractile ring and the cytokinetic planes are severely restricted, which necessitates a more efficient molecular mechanism for effective cytokinesis. This could explain the defects observed in the late-stage cell divisions in the adult mutant worms mentioned above (Finger et al., 2003). This is in line with the observation that in Drosophila and C. elegans embryos, septins colocalize with the cytokinetic apparatus even though septin mutants fail to exhibit significant phenotypes (Neufeld and Rubin, 1994; Nguyen et al., 2000).

Elegant in-depth studies of the in vivo development of the Drosophila neuroepithelium have provided clear evidence that the septin requirement in cytokinesis differs greatly depending on the cellular complexity and constraints. In developing epithelia, the plane of cytokinesis is not random. The furrowing can be either parallel (orthogonal division) or perpendicular (planar division) to the plane of the cell layer (Fig. 2A). Whereas planar division repopulates the epithelial layer, orthogonal division leads to the eventual release of one daughter cell perpendicular to the epithelium. This process is crucial to the generation of stratified epithelia. Using the adult Drosophila dorsal thorax model for the observation of orthogonal and planar divisions in the neuroepithelium, Le Borgne and co-workers have been able to demonstrate a role for septins specifically in planar cytokinesis (Fig. 2A) (Founounou et al., 2013). In parallel, evidence has emerged that planar cytokinesis in the epithelium is a multicellular process, where constraints that are enforced by the neighboring cells affect the dividing cell in order to couple cytokinesis to the formation of adherens junctions, which is important for maintaining tissue cohesiveness and epithelial integrity (Herszterg et al., 2013). In this model, cytokinetic furrow ingression is aided by the deformation of the adjacent cell membranes together with the localized accumulation of myosin II in the neighboring cells. The tensile forces generated by the contractile ring and locally accumulated myosin II in the neighboring cells coordinately achieve cytokinesis in intact epithelia. The Herszterg et al. study also suggests that septins have a role in the recruitment of Arp2/3, an F-actin-nucleation complex, to the midbody, which facilitates the expansion of cell–cell contacts and establishment of adherens junctions (Nakahira et al., 2010; Herszterg et al., 2013). In contrast to the observations in C. elegans, a recent study has presented evidence for a septin-independent establishment of contractile ring asymmetry in Drosophila embryos (Guillot and Lecuit, 2013). Although septins and anillin were shown to be crucial for contraction of the actomyosin ring, the asymmetry of the ring was unaffected in embryos in which Pnut or anillin had been depleted.

Two principal conclusions can be drawn from these studies in invertebrate models. First, septins associate with the actomyosin apparatus at the cytokinetic furrow and might have supportive roles in the assembly, orientation and function of the contractile ring. Second, in simple cell systems, cytokinesis can proceed without the assistance of septins, whereas the increased complexity and constraints that are imposed by neighboring cells in tissues necessitates septin function for robust cytokinesis.

Mammalian cells express multiple septin family members (there are 13 genes that encode septins in humans and mouse), and the assignment of specific septin functions is further complicated by the presence of splice variants and tissue-specific expression patterns. Nevertheless, the first mammalian homologs of septins were cloned in the early 1990s (Nottenburg et al., 1990; Nakatsuru et al., 1994) and an expected and evolutionarily conserved role in cytokinesis was established soon after (Kinoshita et al., 1997). Since then, several studies have utilized knockdown approaches against septin isoforms in mammalian cells with consistent reports of defects in cell proliferation and cytokinesis (Surka et al., 2002; Nagata et al., 2003; Spiliotis et al., 2005; Estey et al., 2010). Strong expression of several septin family proteins in the terminally differentiated cells of the nervous system has suggested that this family also have cell-division independent roles, generating great interest early on in the analysis of their neuronal functions (Kinoshita et al., 2000). Further studies conclusively established a role for septins in neuronal morphogenesis and synaptic transmission (Tada et al., 2007; Xie et al., 2007; Yang et al., 2010), and soon after septins were also shown to be crucial mediators of ciliary biogenesis (Hu et al., 2010).

The presence of more than a dozen genes for septins in the mouse genome has hindered the development and analysis of genetic deletion models because of possible redundancy, although the presence of such a complexity also supports the notion that septins could perform diverse non-canonical functions in higher organisms. In line with this idea, available septin-deficient mouse models show a wide variety of isoform-specific phenotypes that range from mild with almost no visible defects in the Sept6^−/− mice, to complete embryonic lethality in the Sept7^−/− mice (Ageta-Ishihara et al., 2013; Menon et al., 2014; Lassen et al., 2013; reviewed in Kinoshita, 2008; Mostowy and Cossart, 2012). Interesting isoform-specific functions of septins have emerged from analysis of the genetic models: Sept4^−/− mice show defects in sperm maturation (Kissel et al., 2005), and SEPT5-deficiency is associated with platelet degranulation defects (Dent et al., 2002; Peng et al., 2002).

Probably owing to the availability of established gene silencing approaches, all initial investigations on cellular functions of mammalian septins in cytokinesis were restricted to adherent epithelial cell lines. One of the first studies investigating septin expression and function in hematopoietic cells proposed a role for septins in phagosome formation (Huang et al., 2008). The first report addressing the requirement of septins in hematopoietic cell division came a year later. Interestingly, Krummel and co-workers analyzed the role of septins in the amoeboid migration of the murine T-lymphocytic D10.G4 cell line and observed that septin depletion had no effect on the proliferation of these cells (Tooley et al., 2009). This was followed by similar reports of septin-independent survival and proliferation of the human myeloid K562 cell line and of human T-lymphoblast jurkat cells (Sellin et al., 2011a). Although both studies demonstrated a significant degree of siRNA-mediated downregulation of the core septins, it could be argued that the residual septins present averted a lethal phenotype. However, in line of the conclusions we drew above based on invertebrate model systems, cytokinesis in these lymphoid and myeloid cell lines could be subject to a lesser degree of mechanical restrictions owing to their growth in suspension and persistent motility. This would make these cells ideal candidates to survive by means of septin-independent proliferation.

To be able to conclusively verify that hematopoietic cells could undergo septin-independent proliferation, it was necessary to analyze data from genetic deletion models. At the time that the above knockdown experiments were performed, the only septin-knockout mouse models that had been analyzed for effects on the hematopoietic system were Sept5^−/− and Sept6^−/− mice, mainly because chromosomal translocations of these genes are present in myeloid leukemia (Kinoshita, 2008). The only detectable phenotype reported was a mild hypersensitivity of Sept5^−/− platelets to specific agonists (Dent et al., 2002). Now, the recent generation of a conditional knockout Sept7 allele has finally solved the mystery behind the dispensability of septins in adult hematopoiesis (Menon et al., 2014). Consistent with the pivotal role of SEPT7 in cell division, the constitutive deletion resulted in embryonic lethality between day E7 and E10.5 and Sept7^−/− fibroblasts underwent multinucleation in vitro as expected, displaying a parallel depletion of all other septins analyzed (Menon et al., 2014). In contrast to this, Sept7-deleted stem progenitor cell populations from the bone marrow formed large myeloid colonies in vitro, suggesting that these cells undergo robust proliferation and differentiation independently of septins (Menon et al., 2014). Conclusive evidence of septin-independent growth and development of hematopoietic cells in vivo came from our lymphocyte-specific Sept7^−/− model where complete deletion of Sept7 from both B- and T-lymphocytes does not result in any significant defects in the development of lymphoid or myeloid lineages in these mice. Septin-independent cytokinesis of lymphocytes in this genetic deletion model is further supported by efficient in vitro proliferation of these Sept7^−/− cells, which also display strong co-depletion of the other core septins, as observed in Sept7^−/− fibroblasts (Menon et al., 2014). These observations, like many other scientific discoveries, leave us with more questions than answers. If indeed septin-dependent and -independent cytokinesis mechanisms exist in the same organism, as depicted in Fig. 2A, what are the factors that make cells either susceptible or resistant to septin depletion? If septins are merely accessory factors that increase the robustness of cytokinesis, can the same cell lineages differ in their sensitivity to septin depletion depending on their extracellular environment? In line with this, could the expression of such additional cell-type-specific factors compensate for the lack of septin function, e.g. in Sept7^−/− hematopoiesis? Although these questions are not yet fully answered, as discussed below, there are some studies that imply that the interplay between microtubule and septins plays roles in determining the septin-dependence of cytokinesis.

Although, mammalian septins often colocalize with the actin and microtubule cytoskeletons (Xie et al., 2007; Spiliotis, 2010; Bowen et al., 2011; Ghossoub et al., 2013), a direct interplay between septins and microtubules became apparent when Macara and co-workers showed that septin depletion in HeLa cells results in increased stabilization and acetylation of microtubules. The authors identified microtubule-associated protein 4 (MAP4), a microtubule-stabilizing factor, as an interaction partner for SEPT6 and proposed a model whereby septins act as competitive inhibitors of MAP4-mediated microtubule stabilization (Kremer et al., 2005). More interestingly, the cell division defects associated with septin depletion in HeLa cells could be rescued by a parallel knockdown of MAP4, suggesting that the altered microtubule stability is indeed the underlying cause for the cytokinetic defects upon septin depletion. These data are in agreement with the presence of strong tubulin acetylation at the midbody and the requirement for a tightly regulated deacetylation of the midbody microtubules for completion of cytokinesis (Wickström et al., 2010; Zhang et al., 2003). When tubulin dynamics is abrogated with the microtubule-stabilizing drug taxol, cytokinetic furrowing is initiated but there is no productive abscission, a phenotype similar to that seen upon SEPT7 deficiency (Shannon et al., 2005). Septin-deficiency-mediated microtubule stabilization and hyperacetylation have also been shown to contribute to the defects in neuronal morphogenesis observed in Sept7^−/− mice (Ageta-Ishihara et al., 2013). Interestingly, septins are involved in the direct recruitment of histone deacetylase 6 (HDAC6) to microtubules for de-acetylation of the tubulin subunits (Ageta-Ishihara et al., 2013). However, as MAP4 and HDAC6 are ubiquitously expressed, it is less likely that they act as the differential determinants for the septin dependence of cytokinesis. Nevertheless, cell-type-specific differences in septin-deficiency-mediated microtubule hyperstabilization might hold the clue. Consistent with this notion, Sept7^−/− fibroblasts also display abnormalities in the microtubule cytoskeleton (Menon et al., 2014). Although the general microtubule architecture is unaltered in the septin-deficient fibroblasts, a significant increase in microtubule acetylation and the persistence of stable midbody remnants are detected in multinucleated knockout cells, suggesting that there are defects in the dissolution of the midbody microtubules in the absence of septins (Menon et al., 2014). Most importantly, septin-deficient lymphocytes, which undergo septin-independent cytokinesis, do not show any detectable microtubule acetylation (Menon et al., 2014), further pointing to a causal relationship between microtubule hyperacetlyation and sensitivity to septin depletion. However, this theory still needed further evidence from gain-of-function studies and led us to initiate a search for proteins that were differentially expressed between cells sensitive and resistant to septin depletion that are able to induce altered microtubule stability in a Sept7-deficient background (Menon et al., 2014; and our unpublished data). One initial candidate was the shorter isoform of SEPT9, which is predominantly expressed in the cells of the hematopoietic system (Estey et al., 2010; Sellin et al., 2014). However, the fact that this isoform is expressed at considerable levels in some of the cells that undergo septin-dependent cytokinesis and is also destabilized in Sept7^−/− cells suggest it does not serve as a factor that supports SEPT7-independent cytokinesis (Menon et al., 2014; and our unpublished data).

Similar to actin treadmilling, dynamic instability is also a hallmark of the microtubule cytoskeleton, and MAPs, including MAP4, perform a balancing act to achieve perpetual tubulin assembly and disassembly. Cell-type-specific MAPs could therefore be excellent candidates for being septin-sensitive factors. One of the microtubule-destabilizing proteins that is known to counteract the function of MAP4 in mammalian cells is stathmin-1 [STMN1, also known as oncoprotein-18 (OP18)] (reviewed in Holmfeldt et al., 2009). STMN1 is hyperphosphorylated and inactivated during mitosis, which faciliates spindle assembly. In late telophase, STMN1 is dephosphorylated and activated again, suggesting that it has a role in destabilization of the midbody microtubules. More interestingly, the ratio between the amounts of STMN1 and tubulin appears to be much lower in fibroblasts compared to that in hematopoietic cell lines, suggesting that it is a more potent microtubule destabilizer in cells of the blood lineage (Larsson et al., 1999; Ringhoff and Cassimeris, 2009). Inducible expression of STMN1 in fibroblasts can rescue the cytokinetic and proliferation defects that result from septin deficiency (Menon et al., 2014). In a complementary approach, siRNA-mediated downregulation of STMN1 was found to be sufficient to make a normally septin-independent human lymphocytic cell line susceptible to the septin inhibitor forchlorfenuron (FCF) (Menon et al., 2014). This indicates that, at least in some cell-types, high expression levels of STMN1 might lead to altered microtubule dynamics that result in reduced microtubule stability and thus a resistance to septin depletion.

Cleavage of the spindle microtubule remnants and establishment of the membrane integrity of the daughter cells is the final step in cytokinesis (Neto and Gould, 2011). This requires the recruitment of microtubule-severing proteins and membrane components to the intercellular bridge. A primary component of this abscission machinery is the endosomal sorting complexes required for transport (ESCRT)-III complex, which in turn is involved in the recruitment of the microtubule-severing protein spastin to the abscission site for a localized microtubule disassembly (Elia et al., 2011; Guizetti et al., 2011). Indeed, spastin-mediated scission of microtubules appears to be the rate-limiting step in abscission (Guizetti et al., 2011; Neto and Gould, 2011). The recruitment of exocyst component Sec8 (also known as EXOC4) to the midbody is abrogated in SEPT9-depleted HeLa cells, resulting in abscission failure (Estey et al., 2010). In addition, the recruitment of the ESCRT-III component CHMP4B to the abscission site requires anillin-dependent septin localization to the intercellular bridge (Renshaw et al., 2014). It is worth noting that all of the studies analyzing the role of the ESCRT-III complex and spastin in abscission have been performed on adherent cells. Several questions remain to be answered regarding the exact mechanism of abscission in mammalian hematopoietic cells. Could it be possible that the planktonic growth, unrestricted motility and altered membrane dynamics in these cells make abscission more dependent on mechanical forces rather than specific septin-dependent recruitment of the abscission machinery? Or do they possess alternative mechanisms for the recruitment of specific factors to the abscission site? Interestingly, early electron microscopic studies have reported clear differences in the organization of abscission sites between HeLa cells and non-adherent hematopoietic cells. For instance, in the case of rat erythroblasts and lymphocytes, a vesiculating equatorial plate is observed, which probably leads to the formation of new plasma membrane between the daughter cells, whereas in adherent HeLa cells, wedge-shaped constriction at the furrow region leads to abscission (Buck and Tisdale, 1962; Robbins and Gonatas, 1964; Murray et al., 1965). Thus, the process of abscission observed in these hematopoietic cells closely resembles that of plant cells, which lack a septin cytoskeleton (Murray et al., 1965). Detailed comparative studies between septin-dependent and independent cell divisions are required for deciphering this mechanism completely (Fig. 2B).

Analysis of the SEPT7 conditional knockout mouse has provided conclusive evidence for the presence of septin-independent cytokinesis in the mammalian hematopoietic system. However, several epithelial cancer cell lines and immortalized fibroblasts fail to proliferate in the absence of septins (Menon et al., 2014; Estey et al., 2010; our unpublished data). In our preliminary studies, even cells transformed with oncogenic Ras were sensitive to SEPT7 depletion (our unpublished observations). Could this mean that septin targeting is a feasible approach for anti-tumor therapy with only minimal off-target effects on hematopoiesis? The major question that needs to be answered is whether septins are completely dispensable for hematopoiesis and hematopoietic cell functions (Fig. 3A).

Despite cloning of the first mammalian septin gene from lymphocyte cDNA and clinical cases of translocations of septin genes in hematological malignancies, research into septin functions in the immune system started relatively late (Nottenburg et al., 1990; Osaka et al., 1999). Over the past decade, several immune-relevant roles of septins have been identified, which tie in with the evolutionarily conserved interplay between septin, actin and microtubules, as well as the strong association of septins with cellular membranes. A role for septins in vesicular trafficking emerged initially from studies in the neuronal system where SEPT5 and SEPT2 were found to control synaptic vesicle exocytosis (Beites et al., 1999). However, subsequent studies revealed that, although neurotransmitter release is unaltered in Sept5^−/− mice, platelet degranulation is deregulated (Dent et al., 2002; Peng et al., 2002). In addition, clinically, Sept5 deletion is associated with Bernard–Soulier syndrome (BSS), a bleeding disorder arising from platelet dysfunction (Bartsch et al., 2011). Another septin function emerged from work showing that the interaction between SEPT7 and CD2AP facilitates the exocytosis of glucose transporter GLUT4 storage granules in kidney epithelial cells (Wasik et al., 2012). Because CD2AP was originally isolated as a protein associated with the cytoplasmic domain of the T-cell-specific antigen CD2, this interaction could also have relevance in immune cells (Dustin et al., 1998). The endocytic machinery is often utilized by pathogenic microbes to gain entry into host cells. Although it has been known for some time that local actin polymerization and clathrin recruitment are needed to create membrane pits for pathogen entry, recent evidence suggests that septins also have a role in bacterial and yeast invasion of epithelial and endothelial cells (Mostowy and Cossart, 2009; Mostowy et al., 2009; Phan et al., 2013). More interestingly, intracytosolic Shigella is entrapped by cage-like septin structures and delivered to autophagosomes for degradation (Mostowy et al., 2010). Although these initial findings originate from epithelial cells, similar septin-cage formation has also been observed in phagocytic cells, suggesting a common role whereby septins deliver phagocytic cargo to the autophagy pathway (Gong et al., 2012; Mostowy et al., 2013). Moreover, septins colocalize with phagosomes and are crucial for FcγR-mediated phagocytosis in macrophages (Huang et al., 2008).

Even though the Sept7^−/− model conclusively showed evidence of septin-independent cytokinesis in lymphocytes, several lines of evidence suggest that there are important roles for septins in T-cell biology. Interestingly, thymocyte development is defective in a T-cell-specific Sept9^−/− mice leading to the accumulation of CD4 and CD8 double-negative cells, indicating a crucial role for septin cytoskeleton in the specific stages of T-cell development (Lassen et al., 2013). This is not unexpected in light of a recent genome-wide screen for regulators of Ca²⁺ signaling and nuclear factor of activated T-cells (NFAT) activation that identified septins as important modulators of store-operated Ca²⁺ release, a pathway important for T-cell development and activation (Sharma et al., 2013; Fracchia et al., 2013). Moreover, septins are crucial for the maintenance of membrane dynamics, cell shape and motility in lymphocytes (Tooley et al., 2009; Gilden et al., 2012). Although septins play an important part in immune cell functions that are independent of cell division (summarized in Fig. 3A), septin depletion appears to be well tolerated by the hematopoietic system. This provides the opportunity to therapeutically target septins to combat malignancies, such as carcinomas and sarcomas, without impairing hematopoiesis.

For many years, targeting of mis-regulated cell proliferation in cancer has been performed using DNA-binding and -modifying agents that interfere with DNA replication, and tubulin-binding agents that interfere with mitotic spindle formation (Dumontet and Jordan, 2010) (Fig. 3B). The absolute requirement for the microtubule cytoskeleton in mitosis has led to the use of tubulin modulators as anti-cancer drugs (reviewed in Jordan and Wilson, 2004; Dumontet and Jordan, 2010). Microtubule-targeting agents (MTAs), which include stabilizers (such as taxanes and epothilones) and destabilizers (vinca alkaloids and colchicins), can suppress spindle microtubule dynamics, resulting in mitotic arrest and/or cell death (Perez, 2009). Selectivity of MTAs is attributed to the active dividing status and intense microtubule turnover in cancer cells. Unfortunately, undesired side effects are observed owing to adverse effects on fast-diving cells of the hematopoietic lineages and the effects of these drugs on non-mitotic cytoskeletal functions in non-dividing cells (Chan et al., 2012). Despite their systemic toxicity, microtubule modulators are still among the most successful clinical agents against solid tumors and are subject to ongoing research to better understand their mode of action and selectivity (Von Hoff et al., 2013; Alushin et al., 2014). Now, the fourth component of the cytoskeleton, the septins, might enter the stage as targets of anti-mitotic cancer therapy. In particular, the fact that the core septin SEPT7 plays a key role in completion of mitosis in various cells, such as fibroblasts, whereas it is dispensable for proliferation of hematopoietic cells, opens the possibility of differentially targeting cell proliferation with septin inhibitors (Fig. 3B). To date the only known septin inhibitor is FCF (Hu et al., 2008), but its specificity is questionable and off-target effects have recently been described (Heasley et al., 2014). It will therefore be necessary to develop systems to screen for novel cell-permeable septin inhibitors that alter septin filament structure and inhibit cytokinesis of only the targeted cell types. In the meantime, the SEPT7-conditional knockout mice (Menon et al., 2014) should be also analyzed in the context of solid tumor models. Based on the data discussed here, it could be expected that the growth of solid fibrosarcomas or carcinomas will be inhibited in these mice once Sept7 deletion has been induced, whereas proliferation of hematopoietic cells should proceed normally.

Nevertheless, a number of questions need to be addressed before such a strategy could be employed. For instance, is it actually possible to identify potent and specific inhibitors of septins and, if so, would they have any adverse side-effects that arise from interfering with the non-canonical functions of septins? Furthermore, could tumors develop resistance to septin inhibitors as is the case for many other clinically used inhibitors? There is reason for optimism regarding this latter concern; as microtubule stability is crucial for septin-dependence of cytokinesis, any resistance to septin inhibitors could be modulated by a parallel treatment with tubulin-stabilizing or -destabilizing agents, respectively. Thus, a combination of targeting septin and treatment with established microtubule inhibitors could present a promising avenue to advance the therapeutic efficacy and selectivity of anti-proliferative cancer treatment.

Most of our understanding regarding the process of cytokinesis is based on experiments in invertebrate model organisms and mammalian epithelial cell lines. To completely understand the mechanistic differences between septin-dependent and -independent cytokinesis in mammalian cells, there is a need to analyze Sept7^−/− hematopoietic cells in greater detail. In addition to expanding our basic understanding of mammalian cell cytokinesis, these investigations also will open the promising possibility of targeting septin-dependent cytokinesis as an anti-tumor therapy. To bring septin biology from bench to bedside, we need a two-fold approach that gives equal importance to the basic understanding of cytokinesis, as well as development of specific septin inhibitors.

We thank Alexey Kotlyarov for critical reading of the manuscript and Stefanie Hall for proof reading.