Yeast cells select the position of their new bud at the beginning of each cell cycle. The recruitment of septins to this prospective bud site is one of the critical events in a complex assembly pathway that culminates in the outgrowth of a new daughter cell. During recruitment, septin rods follow the high concentration of Cdc42GTP that is generated by the focused localization of the Cdc42 guanine-nucleotide-exchange factor Cdc24. We show that, shortly before budding, Cdc24 not only activates Cdc42 but also transiently interacts with Cdc11, the septin subunit that caps both ends of the septin rods. Mutations in Cdc24 that reduce affinity to Cdc11 impair septin recruitment and decrease the stability of the polarity patch. The interaction between septins and Cdc24 thus reinforces bud assembly at sites where septin structures are formed. Once the septins polymerize to form the septin ring, Cdc24 is found at the cortex of the bud and directs further outgrowth from this position.
Yeast cells grow and divide by forming a bud at the beginning of each cycle. The bud matures through polarized growth to form the future daughter cell. In a first step, the formation of the new bud requires the focused assembly of a complex set of proteins and lipids at a preselected point of the plasma membrane (Howell et al., 2009; Howell and Lew, 2012; Fairn et al., 2011). The position of this prospective bud site (PBS) is selected by landmark proteins that locally activate the Ras-like Rsr1 (Bender and Pringle, 1989; Chant and Herskowitz, 1991; Kang et al., 2001; Park et al., 1997). Rsr1GTP then recruits and possibly stimulates Cdc24, the guanine-nucleotide-exchange factor (GEF) for the Rho GTPase Cdc42 (Park et al., 1997; Shimada et al., 2004). Cdc42GTP directly binds and activates approximately a dozen effector proteins whose combined activities organize plasma membrane and cell wall outgrowth from this site (Bose et al., 2001; Chiou et al., 2017; Evangelista et al., 1997; Irazoqui et al., 2003; Meca et al., 2019). As a direct or indirect consequence of Cdc42 activation, numerous additional proteins are recruited (Pruyne et al., 2004). In spite of this massive protein and membrane influx, PBS assembly remains remarkably focused. Only one bud, with a base of less than 2 µm in diameter, is formed. The sustained association of the landmark proteins with the previous cell division site ensures that the new bud of haploid cells always forms in a position adjacent to the previous bud (Chant and Herskowitz, 1991).
A critical step during PBS assembly involves recruitment of the septins. The four septin units, Cdc11, Cdc12, Cdc3 and Cdc10, form octameric rods that polymerize into filaments and higher-order structures like rings and collars at the bud neck (Bertin et al., 2008; Marquardt et al., 2019). Cdc11 is the subunit that caps the rod at both ends and thus plays a critical role in filament formation (Brausemann et al., 2016; Garcia et al., 2011). During mitosis the septins split into two rings and enclose the space where cytokinesis and abscission are prone to occur (Dobbelaere and Barral, 2004; Marquardt et al., 2019). After abscission, the septins are disassembled from the site of cytokinesis and transferred to the new PBS in a Cdc42GTP-dependent manner. There they polymerize again into ring-forming filaments (Caviston et al., 2003; Gladfelter et al., 2002). The ring remains at the base of the bud during further membrane and cell wall growth and restricts the free movement of cortical proteins between the mother and daughter cells (Barral et al., 2000). The mechanisms of septin recruitment and assembly are not fully understood (Marquardt et al., 2019). Central roles are played by the kinase Cla4 and the paralogous proteins Gic1 and Gic2 (Kadota et al., 2004). Gic1 and Gic2 are effectors of Cdc42GTP and are proposed to facilitate the transfer and the incorporation of septin rods at the PBS (Iwase et al., 2006; Okada et al., 2013; Sadian et al., 2013). Cla4 phosphorylates certain subunits of the septins and is very likely stimulated by its direct interaction with Cdc42GTP (Lamson et al., 2002; Versele and Thorner, 2004). A loss of Cla4 impairs septin localization and its assembly into rings (Kadota et al., 2004; Weiss et al., 2000). These effects are aggravated by the simultaneous deletion of members of the polarisome, a cluster of proteins that catalyzes the formation of actin filaments at sites of polar growth (Kadota et al., 2004).
The complex structure of the nascent bud requires mechanisms that coordinate and regulate its assembly. As the only source of active Cdc42, the localization and regulation of Cdc24 is especially critical for bud site formation (Sloat and Pringle, 1978; Sloat et al., 1981). Cdc24 is bound to the scaffold protein Bem1 throughout the cell cycle (Woods et al., 2015; Witte et al., 2017; Butty et al., 2002; Grinhagens et al., 2019 preprint). Both proteins interact through their C-terminally located Phox and Bem1 (PB1) domains (Ito et al., 2001; Peterson et al., 1994). Bem1 not only determines the localization of the Cdc24–Bem1 complex but also influences its activity either directly or indirectly through its bound partner proteins (Rapali et al., 2017; Smith et al., 2013; Witte et al., 2017; Woods et al., 2015; Grinhagens et al., 2019 preprint). Positive and negative feedback loops thus convene on the GEF activity of the Cdc24–Bem1 complex to achieve singularity and robustness of budding (Butty et al., 2002; Goryachev and Pokhilko, 2008; Kozubowski et al., 2008; Kuo et al., 2014; Shimada et al., 2004; Smith et al., 2013; Wai et al., 2009; Wedlich-Soldner et al., 2004; Witte et al., 2017; Rapali et al., 2017).
This work characterizes a novel interaction between the septins and the Cdc24–Bem1 complex. We show that this interaction is part of a feedback mechanism that links the continuous production of active Cdc42GTP to successful septin assembly.
The Cdc24–Bem1 complex binds the septin subunit Cdc11
A more complete understanding of the Cdc24 interactome is likely to reveal how its localization and activity are regulated during the cell cycle. A library-based split-ubiquitin screen identified the septin subunit Cdc11 as novel interaction partner of Cdc24. We confirmed our initial finding by testing the interactions of Cdc24, extended by the Cub–RUra3 module (Cdc24CRU), against an array of 548 yeast strains, each expressing a different Nub fusion. The Nub fusions of this array were enriched in proteins known to play roles in polarity establishment and polar growth, stress response, cytokinesis and further cellular processes (Hruby et al., 2011; Wittke et al., 1999; Johnsson and Varshavsky, 1994; Laser et al., 2000). The array confirmed known, and identified novel, binding partners of Cdc24 (Fig. 1A, Table 1; Fig. S1). The assay also proved the specificity of the Cdc24–Cdc11 interaction, as none of the other four Nub-labeled mitotic septins generated an interaction signal with Cdc24CRU (Fig. 1B). The library screen and systematic Nub-array also confirmed that Cdc24 forms a stable complex with Bem1 (Bose et al., 2001). To determine whether Cdc11 interacts directly with Cdc24, or indirectly through Bem1, we tested Nub–Cdc11 against Cdc24CRU in the presence and absence of Bem1. The deletion of BEM1 is lethal in the strain JD47, but bem1Δ cells can be kept viable through the simultaneous deletion of the Cdc42 GTPase-activating protein (GAP) BEM3 (Dowell et al., 2010; Laan et al., 2015; Grinhagens et al., 2019 preprint). Split-ubiquitin analysis was carried out in this double deletion strain with a subset of the identified interaction partners. It could be clearly demonstrated that Cdc11, like Rga2 and Rsr1, does not require Bem1 in order to interact with Cdc24, whereas the interaction between Cdc24 and Boi1 and Boi2 depended on Bem1 (Fig. 1C).
To localize the binding site for Cdc11 on Cdc24 we screened fragments of Cdc24 as CRU fusions against Nub–Cdc11. A C-terminal fragment (Cdc24428–854) harboring the pleckstrin homology (PH) and PB1 domains of Cdc24 seemed to contain the complete interface to Cdc11 and Bem1 (Fig. 1D). Deletion of the C-terminal PB1 domain abolished the interaction of Cdc24428–759 with Bem1 and with Cdc11. Conversely, Nub–Bem1, but not Nub–Cdc11, generated an interaction signal with Cdc24755–854CRU, indicating that Cdc11 contacts regions of Cdc24428–759 in addition to interaction with the PB1 domain of Cdc24 (Fig. 1D).
To test whether Cdc24 can simultaneously bind Cdc11 and Bem1, we reconstituted the complex in vitro. GST–Cdc11 precipitated the PB1 domain of Bem1 (PB1Bem1) specifically, and only in the presence of Cdc24428–854 (Fig. 2B), thus proving the existence of a trimeric Cdc11–Cdc24–Bem1 complex. In accordance with this observation, a well-described mutation in the PB1 domain of Cdc24 that interferes with the formation of the Cdc24–Bem1 complex (Cdc24428–854 D820A) did not affect the binding to Cdc11 (Fig. 2A) (Yoshinaga et al., 2003). Additional in vivo evidence for the Cdc11–Cdc24–Bem1 complex was recently obtained in the form of a proximity between Bem1 and Cdc11, detected in a split-ubiquitin assay. This proximity is lost when the PB1 domain of Bem1 is deleted (Grinhagens et al., 2019 preprint).
We noticed that GST–Cdc11 also precipitated a degradation product of 6×His–Cdc24428–854 that, according to its apparent molecular weight, lacked the PB1 domain and a large part of the linker region (Fig. 2B). In accordance, 6×His–Cdc24668–854, lacking the PH domain, bound to Cdc11 less strongly than 6×His–Cdc24428–854 (Fig. 1D, upper panel; Fig. S2B). We conclude that the interface between Cdc24 and Cdc11 extends from the PB1 domain to the PH domain of Cdc24.
Cellular Cdc11 is predominantly found in septin rods and filaments. We could show by surface plasmon resonance spectrometry (SPR) that 6×His–Cdc24428–854 binds not only to free Cdc11 but also to octameric septin rods (Fig. 2C).
Cdc24 binds septins in G1 during bud site assembly
Live-cell imaging of cells co-expressing Cdc11–mCherry and Cdc24–GFP showed restricted colocalization of both proteins at the bud neck during cytokinesis (Fig. 3A, 86–92 min) and to the prospective bud site shortly before bud emergence (Fig. 3A, 112 min). To decide when and where the septins interact with Cdc24 we followed the interaction through the cell cycle using SPLIFF analysis. SPLIFF is a modification of the split-ubiquitin technique where the Cub is sandwiched between autofluorescent mCherry and GFP (CCG) (Moreno et al., 2013). Upon interaction-induced reassociation with an Nub fusion, the GFP is cleaved off and rapidly degraded. The subsequent local increase in the ratio of red to green fluorescence indicates where and when the interaction between both proteins took place. Haploid yeast cells co-expressing Cdc11–mCherry–Cub–GFP (Cdc11CCG) together with Nub–Cdc24 were observed by fluorescence microscopy at two-minute intervals. The Nub-induced conversion of Cdc11CCG to Cdc11CC (the GFP-cleaved form) was plotted as ratio of conversion against time (Fig. 3B). A rise in conversion indicated interaction whereas a decreasing or stable ratio indicated no interaction (Moreno et al., 2013). We detected an increase in this ratio at the time when the new bud site is established. This increase was caused by the Nub–Cdc24-induced cleavage of the RGFP moiety from Cdc11CCG, indicating interaction between the two fusion proteins. Complex formation lasted approximately four minutes during the assembly of the septins in the G1 phase of the cell cycle (Fig. 3B). No significant interactions between Cdc11 and Cdc24 were detected at the septin ring, the septin collar or the split-septin structure (Fig. 3B; Fig. S3A, Tables S1, S2). A close inspection and quantification of the still images shortly before and during the assembly of the new septin patch revealed that the labeled Cdc11 appeared at the PBS predominantly in its converted Cdc11CC form (red fluorescence). Cdc11 fusions at the old division sites of mother and daughter cells consisted predominantly of the unconverted Cdc11CCG (Fig. 3C,D). The assay could not resolve whether Cdc11CCG was converted by Nub–Cdc24 on its way to the PBS or immediately upon targeting. The latter is the more likely interpretation because the Cdc24–Bem1 complex arrives at the PBS shortly before Cdc11 (Lai et al., 2018; Okada et al., 2013) (see also Fig. 5).
Binding to Cdc11 does not influence the GEF activity of Cdc24
Cdc11 and Cdc24 are essential proteins. To analyze the functional significance of the Cdc11–Cdc24–Bem1 complex we searched for mutations in Cdc24 that specifically impair the interaction with Cdc11. The PB1 domain of Cdc24 is required for interaction with Cdc11 (Fig. 1D). Our data demonstrating the existence of a trimeric Cdc11–Cdc24–Bem1 complex imply that the contact areas of Cdc11 and Bem1 on Cdc24 do not overlap (Fig. 2). We replaced a conserved lysine residue with alanine at position 801 within the PB1 domain of 6×His–Cdc24428–854. This residue is located opposite the binding interface for Bem1. SPR analysis of the interaction between Cdc11 and Cdc24428–854 K801A showed a modest reduction in binding affinity (Fig. 4A). Cdc11 also binds to regions of Cdc24428–854 that are to the N-terminal side of its PB1 domain (Figs 1D, 2A; Fig. S2). We thus exchanged the conserved and exposed lysine residue in position 525 of the PH domain with a glutamic acid (Cdc24428–854 K525E). SPR analysis of this mutant revealed a reduced affinity to Cdc11 and a faster dissociation from the immobilized Cdc11 (Fig. 4A). Combining both mutations in one molecule (Cdc24428–854 KK, short for Cdc24428–854 K525E K801A) further decreased the affinity towards Cdc11, and increased the off-rate from the sensor chip compared to the binding of Cdc24428–854 (Fig. 4A).
Full length Cdc24 and Cdc24KK were expressed and enriched as MBP fusions from E. coli cells, or as TAP-tagged fusions from yeast cells (Fig. S3). Wild type and mutated forms of Cdc24 displayed roughly the same GDP/GTP exchange activity towards Cdc42 in both experiments (Fig. 4B; Fig. S4). The addition to the assay of purified 6×His–Cdc11 in excess of the amount of MBP–Cdc24 did not significantly alter the GDP/GTP exchange activity of Cdc24 either (Fig. 4B). We conclude that neither the interaction-impairing mutations nor the binding to Cdc11 seem to affect the catalytic activity of Cdc24.
We introduced the two interaction-impairing mutations into the genomic sequence of CDC24 to create the cdc24KK allele (cdc24K525E K801A) in yeast. In accordance with our in vitro analysis, a CRU fusion to cdc24KK displayed a strongly impaired but not completely abolished interaction with Nub–Cdc11 (Fig. 4C). Although both mutations map into a region of Cdc24 that harbor interaction sites for Bem1 and Rsr1, Cdc24KKCRU still bound to Nub–Bem1 and to Nub–Rsr1 (Fig. 4C, Fig. S4) (Park et al., 1997). Cells containing the cdc24KK allele did not display any obvious growth phenotype (see Fig. 6C).
The Cdc11–Cdc24–Bem1 complex promotes formation of the PBS
Cdc24KK–GFP, or Cdc11–mCherry and Bem1–GFP in cdc24KK cells displayed a wild-type-like distribution (Fig. 5A,B). However, closer inspection of the G1 cell cycle phase revealed that Cdc24KK–GFP did not always remain fixed to a chosen PBS (Fig. 5A,B; Movies 1, 2). The Cdc24KK–GFP patch, indicative of the PBS, occasionally dissociated and reappeared at, or close to the site of previous cell division (Fig. 5A,B; Movie 2). Bem1–GFP and Gic2PBD–RFP, visual reporters of Cdc42GTP, showed a similar retraction from the initially chosen site of bud formation in cdc24KK cells (Fig. 5B) (Atkins et al., 2013; Brown et al., 1997; Orlando et al., 2008).
We next quantified the dynamic distribution of the Cdc24–Bem1 complex and the septins during PBS formation in cdc24KK and wild-type cells (Fig. 5C). Wild-type cells co-expressing Cdc11–mCherry and Bem1–GFP behaved uniformly. Bem1–GFP rapidly accumulated at the PBS and remained at this site for five minutes before decreasing to ∼50% of its peak intensity in the next 4 min (Fig. 5C). A decrease in intensity of Bem1 during PBS formation has already been reported and is also observed for active Cdc42 (Okada et al., 2013; Woods et al., 2016). Starting shortly after the initial rise in Bem1–GFP fluorescence, Cdc11–mCherry accumulated steadily and reached ∼50% of its final intensity at the PBS when the signal of Bem1–GFP had already leveled off to 50% of its maximal intensity (Fig. 5C).
Bud site assembly occurred differently in cdc24KK cells (Fig. 5C). In 70% of the measured mutants the initially chosen PBS was kept, and a new bud was formed next to the previous division site (Fig. 5C, left). Here, Bem1–GFP intensity rapidly rose to similar levels as in wild-type cells but stayed at this intensity for only two minutes. An initial rapid decrease was followed by a much slower decline that finally trailed off to 50% of the intensity of its peak value (Fig. 5C). The accumulation of Cdc11–mCherry occurred indistinguishably from that in wild-type cells.
The remaining 30% of cdc24KK cells aborted bud site assembly (Fig. 5C, right). Here Bem1–GFP initially appeared with similar kinetics at the PBS, but stayed at its peak value for only ∼2 min before rapidly dissolving (Fig. 5C). Cdc11–mCherry accumulation stopped early in these cells, and the small fraction of attached Cdc11 disappeared together with Bem1 from the PBS (Fig. 5C). The cells then started a new round of PBS assembly, either next to or at the site of the previous cell division.
Genetic analysis of the Cdc11–Cdc24 interaction
Bem1 is a major determinant of Cdc24 localization and a stimulator of its catalytic activity (Rapali et al., 2017; Smith et al., 2013; Woods et al., 2015). We probed the genetic interaction between cdc24KK and bem1Δ to test whether the mutations in Cdc24KK interfere with the positive effect of Bem1 on Cdc24 location or activity. Unlike bem3Δ cells carrying the wild-type allele of CDC24, bem3Δ cells carrying the cdc24KK allele did not tolerate the additional deletion of BEM1 (Fig. 6A). This strong negative interaction suggests that the mutations, and consequently binding to Cdc11, affect a different activity of Cdc24 than binding to Bem1. By fusing the PB1 domain of Bem1 to the C-terminus of Cdc11 (Cdc11–PB1Bem1; Fig. 6B) we asked more directly whether the impaired interaction with Cdc11 causes the synthetic lethality between cdc24KK and bem1Δ. Cdc11–PB1Bem1 anchored a large fraction of Cdc24–GFP to the bud neck without visibly affecting the stability of the protein (Fig. S3C). Fig. 6B shows that ectopic expression of Cdc11–PB1Bem1 rescues the growth of cdc24KK bem1Δ cells (in a bem3Δ background).
To map the contribution of the Cdc24–Cdc11 complex within the network of polarity genes, we compared the growth of the cdc24KK strain with strains carrying additional deletions of individual polarity genes (Fig. 6C). Cells lacking RSR1 or subunits of the polarisome were strongly affected by cdc24KK,, whereas cdc24KK cells were not impaired by deletion of CLA4 or STE20 (Fig. 6C). When spa2Δ cdc24KK cells were shifted from 30°C to 37°C for four hours, the majority of cells arrested growth, forming large, unbudded cells (Fig. 6D).
Impairment of the Cdc24–septin interaction increases budding at the site of previous cell divisions
Quantification of fluorescence images of cdc24KK cells co-expressing Bem1–GFP and Cdc11–mCherry suggested that 18% of the cells select a new bud at, or very close to, the old division site (Figs 5A,C, 7B). This observation was confirmed by comparing scanning electron micrographs obtained from wild-type and cdc24KK cells (Fig. 7A,B). For quantitative comparison, we counted buds forming in, or on the rim of, an old bud or birth scar, and defined these as having a ‘bud in scar’ phenotype. Although each single mutation (cdc24K525E and cdc24K801A) had only a minor effect on cellular morphology, cdc24KK cells showed a twofold higher incidence of bud in scar growth (35%) than wild-type cells (18%) (Fig. 7A,B). Cells repeatedly budding centrally through a former bud or birth scar were only observed amongst cdc24KK cells, but never in wild-type cells (Fig. 7A). The increase in incidence of bud in scar growth of 17% was consistent with our observation, made using fluorescence microscopy, that 18% of cdc24KK cells relocate their PBS from its initial axial site back to the old division site (Fig. 7B). A similar phenotype was described in cells that lack NBA1 or RGA1 (Meitinger et al., 2014; Tong et al., 2007). Rga1 is a GAP for Cdc42 whose activity during G1 prevents the accumulation of active Cdc42 in the old division site (Tong et al., 2007). Scanning electron microscopy revealed that the frequency of incorrect budding found in rga1Δ cdc24KK cells (89%) is roughly the sum of frequencies observed in rga1Δ (77.5%) and cdc24KK cells (17%) (Fig. 7B). We conclude that the deletion of RGA1 and the cdc24KK mutation affect budding independently of each other.
The Cdc24–Bem1 complex generates active Cdc42 that ultimately leads to the assembly of septins and other effectors at the PBS. However, the timing and location of the herein discovered interaction between Cdc24 and Cdc11 point to a more direct role of the Cdc24–Bem1 complex in septin recruitment. Indeed, fluorescence microscopy of cdc24KK cells, which express a mutant of Cdc24 with reduced affinity to Cdc11, revealed two distinguishable phenotypes. In 30% of the cells, the assembly of the PBS was aborted before a septin patch was formed. The Cdc24KK–Bem1 complex relocated to a new site and initiated PBS assembly, often at the site of previous cell divisions. In 70% of the cells, bud formation occurred as in wild-type cells, although the Cdc24KK–Bem1 complex was less stably associated with the PBS. The two phenotypes seem to reflect two interdependent roles of the Cdc11–Cdc24 interaction during bud site formation. In a first step, the Cdc24–Bem1 complex helps to attract the septins to the PBS, whereas in the second step the interaction with Cdc11 acts to retain the Cdc24–Bem1 complex at this site (Fig. 7C). The elucidation of this positive feedback loop seems to contradict experiments that demonstrated a negative influence of the septins on polarity patch formation (Okada et al., 2013; Schneider et al., 2013).
To reconcile both observations, we postulate that the transformation from septin rods to filaments and rings might be accompanied by changes in the accessible surfaces of the septin units. Whereas rods bind the Cdc24–Bem1 complex and promote PBS formation, rings or filaments might exclude the Cdc24–Bem1 complex and recruit Cdc42 GAPs instead to terminate PBS formation (Okada et al., 2013). This switch would thus couple the maturation of the PBS to the structural transformations of the septins. Both feedback mechanisms are compatible with the growth of the septin structures before and after the switch. Although the initial septin recruitment depends on Cdc42GTP, the further assembly of the septins during and after the transformation to ring and collar is template-assisted and independent of Cdc42. Similar feedback mechanisms might operate during the assembly of higher-order septin structures in other organisms as well (Nagata and Inagaki, 2005).
Assembly of the PBS is under the control of the cell cycle and requires activation of cyclin-dependent kinase (CDK) in late G1 (Atkins et al., 2013; Gulli et al., 2000; Lai et al., 2018; McCusker et al., 2007; Moran et al., 2019; Witte et al., 2017). CDK also influences the timing of septin recruitment, as the lag between the arrival of the Bem1–Cdc24 complex and the septins can be shortened by pre-activation of CDK (Lai et al., 2018). This delay can be explained by a need to prime the septins, or their receptor, by either CDK or a CDK-activated mediator (Lai et al., 2018). The Cdc11–Cdc24 interaction is a candidate for such a priming event. Phosphorylation of either of the two components might increase their affinity to each other and induce the assembly of septins at the PBS (Wai et al., 2009). Cla4 is known to bind Bem1 and to phosphorylate Cdc24 and the septins (Gulli et al., 2000; Versele and Thorner, 2004). In contrast to other polarity genes, deletion of CLA4 shows no negative genetic interaction with the cdc24KK allele. Cdc11–Cdc24 complex formation and Cla4-induced phosphorylation might thus be successive steps in the same septin assembly and transformation pathway. Whether Cla4 serves as mediator of CDK and directly primes the interaction between Cdc24 and the septins is an open but experimentally testable question.
The Cdc24–Bem1 complex is not found at mature septin rings during later cell cycle stages (Fig. 3A,B). Dissociation of the Cdc24–Bem1 complex from the septins might thus mark the transition between PBS assembly and incorporation of new plasma membrane and cell wall material. We propose that the release of Cdc24–Bem1 from the septins serves as a checkpoint ensuring that tip growth only begins once the septin patch has been successfully transformed into a ring (Fig. 7C).
MATERIALS AND METHODS
Growth conditions and cultivation of yeast strains
All yeast strains were derivatives of JD47, a descendant from a cross of the strains YPH500 and BBY45 (Dohmen et al., 1995). Cultivation of yeast was performed in standard SD or YPD media at 30°C, or at the indicated temperatures, as described by Kustermann et al. (2017). Media for split-ubiquitin interaction assays and selection for the loss of centromeric URA3-containing plasmids comprised 1 mg/ml 5-fluoro-orotic acid (5-FOA; Formedium, Hunstanton, UK).
Construction of plasmids, gene fusions and manipulations
Construction of Nub and Cub gene fusions, as well as GFP-, mCherry- or mCherry–Cub–RGFP (CCG)-fusions were as described previously (Dünkler et al., 2012; Moreno et al., 2013; Neller et al., 2015; Wittke et al., 1999). Cdc24CRU/–GFP, Bem1–GFP, Shs1–mCherry and Cdc11–mCherry/–CCG were constructed by genomic in-frame insertions of the GFP-, mCherry-, CRU-, or CCG-modules behind the coding sequences of the relevant genes. In brief, a PCR-fragment of the C-terminal region of the respective target gene lacking the stop codon was cloned via EagI and SalI restriction sites in front of the CRU-, GFP-, mCherry-, or CCG-module on a pRS303, pRS304 or pRS306 vector (Sikorski and Hieter, 1989). Plasmids were linearized using unique restriction sites within these sequences and transformed into yeast cells for integration into the genomic target ORFs. Colony PCR with diagnostic primer combinations was used to verify the successful integration. CDC24428–854-, CDC24705–854-, CDC24755–854-, and CDC24428–759CRU were obtained by ligation of EagI/SalI-cut PCR fragments spanning the respective ORF between the sequences of the PMET17 promoter and the CRU module on the pRS313 vector (Sikorski and Hieter, 1989). Mutations in the coding region of CDC24 were obtained by overlap-extension PCR using plasmids containing CDC24428–854 as templates.
Constructions of genomic cdc24K525E, cdc24K801A, cdc24K525E K801A (cdc24KK), and CDC24WT alleles were achieved by recombination of PCR fragments generated from plasmids pFA6a-Cdc24428–854 KK+term, and pFA6a-Cdc24428–854+term using a forward primer starting at amino acid 428 and a standard reverse S2 primer annealing to a stretch 3′ of base pair 186 of the CDC24 terminator. C-terminal GFP fusions to cdc24KK and CDC24WT were created by PCR-based in-frame fusion using pYM26 and pYM28 as templates (Janke et al., 2004). PMET17-CDC11-PB1Bem1 pRS313 was made by fusing a PCR product of the sequence encoding the PB1 domain starting from residue 470 in frame to CDC11 in a PMET17 pRS313 vector. Cdc11 and PB1Bem1 were separated by a linker sequence encoding a single HA epitope tag.
In certain strains the native promoter sequence was replaced by PMET17 through recombination with a PCR fragment generated from pYM-N35 and primers containing sequences identical to the respective genomic locations at their 5′ ends (Janke et al., 2004). GST fusions were obtained by placing the ORFs of the respective genes or gene fragments in frame behind the E. coli GST sequence on the pGEX-2T plasmid (GE Healthcare, Freiburg, Germany) using BamHI and EcoRI restriction sites. Fusions to the SNAP tag (New England Biolabs, Beverly, MA) were expressed from plasmid pAGT-Xpress, a pET-15b derivative (Schneider et al., 2013). Gene fragments were inserted in frame into a multi-cloning site located between the upstream 6×HIS-tag coding sequence and the downstream SNAP tag coding sequence. The 6×HIS tag fusions were obtained by placing the ORFs of the respective genes or gene fragments behind the E. coli 6×HIS tag sequence on the previously constructed pAC plasmid (Schneider et al., 2013). Fusions to MBP were obtained by inserting the ORF of CDC24 or its mutants in the multi-cloning site of vector pMAL-c5X (New England Biolabs, Beverly, MA).
Gene deletions were performed by one step PCR-based homologous recombination using pFA6a natNT2, pFA6a hphNT1, pFA6a kanMX6, pFA6a CmLEU2 and pFA6a HISMX6 as templates (Bähler et al., 1998; Janke et al., 2004; Longtine et al., 1998; Schaub et al., 2006). Lists of plasmids and yeast strains used in this study can be found in Tables S3 and S4.
Split-ubiquitin library screen
JD47 yeast cells expressing a genomic CRU fusion to CDC24 were transformed with a multi-copy library expressing random fragments of the yeast genome inserted behind the Nub-module and driven by the PADH1 promoter (Laser et al., 2000). Library transformation, screening and the analysis of the FOA-resistant clones were as described previously (Reichel and Johnsson, 2005). A plasmid carrying an Nub fusion to Cdc11 lacking the N-terminal five residues was independently isolated twice.
Split-ubiquitin interaction analysis
For split-ubiquitin array analysis, a library of 548 different α-strains each expressing a different Nub fusion were mated with a CDC24-Cub-R-URA3 (CRU) expressing a-strain. Diploids were transferred as independent quadruplets on SD medium containing 1 mg/ml 5-FOA and different concentrations of copper to adjust the expression of the Nub fusions (Fig. S1) (Dünkler et al., 2012). For individual split-ubiquitin interaction analysis, CRU and Nub expressing strains were mated or co-expressed in haploid cells and spotted onto medium containing 1 mg/ml 5-FOA and different concentrations of copper in four 10-fold serial dilutions starting from OD600=1. Growth at 30°C was recorded every day for 3–5 d.
Preparation of PB1Bem1 affinity matrix
A 200 µl volume of glutathione Sepharose 4 Fast Flow suspension (GE Healthcare, Freiburg, Germany) was diluted in 500 µl of phosphate-buffered saline and the beads were pelleted by centrifugation. After two additional washing steps with phosphate-buffered saline, 2 ml of BL21 cell lysate containing the fusion protein GST–PB1Bem1 were added and rotated for 15 min at 4°C. Finally, the beads were washed three times in phosphate-buffered saline.
Enrichment of Cdc24–TAP fusion proteins
Yeast cells expressing Cdc24–TAP or Cdc24KK–TAP were grown in 500 ml of SD –Trp medium at 30°C to an OD600 of 1.5, harvested by centrifugation, washed with H2O and shock frozen in liquid nitrogen. The frozen pellets were transferred to a pre-cooled mortar and ground for at least 10 min to a fine powder under constant presence of liquid nitrogen. The powder was transferred to a new vessel, dissolved in 1 ml of ice cold lysis buffer (20 mM Tris-HCl pH 7.5, 80 mM NaCl and 1 mM DTT) containing a protease inhibitor cocktail (Roche Diagnostics, Penzberg, Germany). The solution was centrifuged at 1600 g for 10 min at 4°C. A 40 µl volume of magnetic hsIgG beads (Thermo Fisher Scientific, Waltham, MA) were washed three times in lysis buffer and resuspended in the cleared yeast extract. After 4 h of overhead rotation at 4°C, the beads were thoroughly washed six times in 500 µl of 20 mM Tris-HCl pH 8.0, 100 mM NaCl and 5 mM MgCl2, before being used immediately for the GEF assay (without additional buffer). SDS–PAGE of the proteins released from the beads followed by anti-TAP western blotting confirmed the presence of roughly equal amounts of Cdc24–TAP or Cdc24KK–TAP in the assays.
Enrichment of MBP–Cdc24 fusion proteins
E. coli cells expressing either MBP–Cdc24 or MBP–Cdc24KK were grown in 1 l SB-Amp to an OD600 of 0.8, adjusted to 1 mM IPTG and cultivated at 18°C overnight. The cells were harvested by centrifugation, washed with H2O and partitioned into four aliquots. Cell pellets were resuspended in 25 ml phosphate-buffered saline and incubated for 30 min at 4°C in the presence of 1 mg/ml lysozyme and protease inhibitors (Roche Diagnostics, Penzberg, Germany) The cell lysate was cleared by centrifugation at 40,000 g for 10 min after sonification using a Bandelin Sonapuls HD 2070 (Reichmann Industrieservice, Hagen, Germany). A 25 ml volume of the MBP–Cdc24 extract was added to 200 µl PB1Bem1 affinity matrix and the suspension was rotated for 2 h at 8°C. The supernatant was removed, the beads were resuspended in phosphate-buffered saline, transferred to a Mobicol ‘F’ column (Mobitec GmbH Göttingen, Germany) and washed three times with phosphate-buffered saline. To elute the bound protein, 250 µl of 10 mM reduced glutathione in 50 mM Tris-HCl pH 8.0 was added and the suspension was incubated for 10 min at 10°C. After centrifugation, the flow through was transferred in 25 mM storage buffer (Tris-HCl pH 8.0, 100 mM NaCl, 5 mM MgCl2 and 20% w/v glycerol) using a NAP5 column (GE Healthcare, Freiburg, Germany), partitioned into 10 µl aliquots and stored at −20°C for a maximum of 2 weeks. SDS–PAGE and Coomassie staining of the gels allowed estimation of the relative concentration of MBP–Cdc24 and MBP–Cdc24KK in the samples. Equal amounts of the fusion proteins were used for the subsequent GEF assays.
Cdc24 GEF assays
GEF activity was quantified by recording the amount of [α32P]GTP (Hartmann Analytic, Braunschweig, Germany) taken up by 6×His–Cdc42 in the presence of MBP–Cdc24, TAP–Cdc24 or 6×His–Cdc24760–854 as control (Mionnet et al., 2008). The 6×His–Cdc42 protein was purified from crude E. coli extracts by IMAC and size exclusion chromatography. For each time-point, 40 pmol purified Cdc42 and 3 µCi [α32P]GTP were added to the assay buffer (25 mM Tris-HCl pH 8.0, 100 mM NaCl, 5 mM MgCl2, 1 mM DTT, 1 mM EDTA and 5 µM GTP). After the addition of MBP–Cdc24 or MBP–Cdc24KK the mixture was incubated under constant shaking at 22°C. A 25 µl aliquot of the assay volume was removed at each time-point and quenched with 500 µl of ice-cold assay buffer. The samples were immediately filtered on pre-wetted 0.45 µm nitrocellulose filters (WhatmanTM, GE Healthcare, Freiburg, Germany). Each filter was washed with 5 ml of 25 mM Tris-HCl pH 8.0, 100 mM NaCl and 5 mM MgCl2, then suspended in 5 ml of Ultima Gold (Perkin Elmer, Waltham, MA, USA) solution and measured in a Tri-Carb 2810 TR scintillation analyzer (Perkin Elmer, Waltham, MA, USA). The assay was adapted to the matrix-bound TAP–Cdc24 by mixing the magnetic beads with assay buffer, premixed with Cdc42 and [α32P]GTP. For each time-point, 25 µl of the mixture was removed and placed in a magnetic rack to precipitate the beads. The supernatant was subsequently quenched in 500 µl of ice-cold assay buffer and processed as described above.
Sequences encoding 6×His–Cdc24428–854 and its mutants were cloned into a pET15b-derived expression plasmid and expressed in E.coli BL21 DE3 in super broth medium (32 g/l tryptone, 20 g/l yeast extract, 5 g/l NaCl and 5 mM NaOH) at 18°C. The recombinant proteins were purified from the crude extract by IMAC and size exclusion chromatography and transferred into HBSEP buffer (15 mM HEPES pH7.4, 150 mM NaCl, 3 mM EDTA and 0.05% Tween) (Renz et al., 2013). Septin rods with a SNAP-tag attached to Cdc10 were expressed in E.coli BL21 and purified as hetero-octameric rods in high salt buffer by IMAC followed by ion exchange chromatography as described elsewhere (Renz et al., 2013). The 6×His–Cdc11–SNAP protein was purified from E. coli BL21 cells by IMAC, followed by size exclusion chromatography as described previously (Schneider et al., 2013).
Binding of purified 6×His–Cdc24428–854 or its mutants (analyte) to Cdc11 and septin rods (ligands) was recorded on a Biacore X-100 instrument (GE Healthcare, Freiburg, Germany) as described previously (Gronemeyer et al., 2016; Renz et al., 2013). Briefly, the SNAP-tagged proteins were biotinylated with BG-Biotin (New England Biolabs, Beverly, MA) and subsequently immobilized on a CM5 chip (GE Healthcare, Freiburg, Germany) that displayed a covalently linked anti-biotin antibody (US Biologicals, Salem, MA, USA) as capture molecule. Capture levels were typically in the range of 100 RU for both ligands. Purified 6×His–Cdc24428–854 was prepared in suitable concentrations in HBSEP buffer. The contact time with the ligand was set to 120 or 180 s followed by a 600 s dissociation period. Background correction (i.e. subtraction of the signal from the reference cell without captured ligand) was performed by default within the experimental workflow of the instrument. The SPR response units were plotted against the concentrations of the respective analytes and fitted into a titration curve by the Biacore evaluation software. The KD was estimated from the analyte concentrations at half-maximal saturation. All measurements were performed at least as triplicates.
For comparing the binding of 6×His–Cdc24428–854, Cdc24428–854 K801A, Cdc24428–854 K525E and Cdc24428–854 K525E K801A to Cdc11, 6×His–Cdc11–SNAP was biotinylated with BG-Biotin and subsequently immobilized on a CAP chip (Biotin Capture Kit; GE Healthcare, Freiburg, Germany). Preparation of the chip, capture of the ligand, and regeneration of the chip surface was carried out according to the manufacturer's recommendations.
Purified 6×His–Cdc24428–854 and its mutants were adjusted to the same concentrations in HBSEP buffer, and their binding to immobilized 6×His–Cdc11–SNAP was recorded on the Biacore X-100. Measurements were performed as triplicate with comparable outcomes.
CDC11 was cloned in frame with GST into the pGEX-2T plasmid. The protein was expressed in LB medium at 37°C after addition of 0.1 mM IPTG. GST–Cdc11 and GST alone were immobilized on 100 µl equilibrated glutathione Sepharose slurry (GE Healthcare, Freiburg, Germany) directly from crude extract. 1 µM of 6×His–Cdc24428–854, or 6×His–PB1Bem1–SNAP in phosphate-buffered saline were added, the beads were washed and bound protein was eluted with an excess of glutathione after 1 h incubation at 8°C under gentle agitation. The eluates were separated by SDS–PAGE, blotted onto nitrocellulose and the proteins were detected with an anti-His antibody (Sigma-Aldrich, Steinheim, Germany; 1:5000). To test whether 6×His–PB1Bem1 and Cdc11 can bind simultaneously to Cdc24428–854, 2 µM or 4 µM of purified 6×His–PB1Bem1–SNAP were added to 6×His–Cdc24428–854 pre-incubated GST–Cdc11-coated beads.
For microscopic inspection yeast cells were grown overnight in SD medium, diluted 1:8 in 3–4 ml fresh SD medium, and grown for 3–6 h at 30°C to mid-log phase. About 1 ml of culture was spun down and the cell pellet resuspended in 20–50 μl residual medium. A volume of 3 μl was spotted onto a microscope slide, the cells were immobilized with a coverslip and inspected under the microscope. For time-resolved imaging, 3 μl of prepared cell suspension was mounted on a SD-agarose pad (1.7% agarose), embedded in a customized glass slide and sealed by a cover slip fixed by parafilm stripes. Imaging was started after 15 to 30 min recovery at 30°C. SPLIFF and other time-lapse experiments were observed using a DeltaVision wide-field fluorescence microscope system (GE Healthcare, Freiburg, Germany) provided with a Olympus IX71 microscope, a steady-state heating chamber, a CoolSNAP HQ2 and CascadeII512-CCD camera (both by Photometrics, Tucson, AZ, USA), a U Plan S Apochromat 100× 1.4 NA oil ∞/0.17/FN26.5 objective and a Photofluor LM-75 halogen lamp (Burlington, VT, USA). Images were visualized using softWoRx software (GE Healthcare, Freiburg, Germany) and adapted z series at 30°C. Exposure time was adapted to the intensity of GFP and mCherry signal for every fluorescently labeled protein to reduce bleaching and phototoxicity. For further analyses an Axio Observer spinning-disc confocal microscope (Zeiss, Göttingen, Germany), equipped with an Evolve512 EMCCD camera (Photometrics, Tucson, AZ, USA), a Plan-Apochromat 63× 1.4 NA oil DIC objective, and 488 nm and 561 nm diode lasers (Zeiss, Göttingen, Germany) were used. Images were analyzed with the ZEN2 software (Zeiss, Göttingen, Germany).
Quantitative analysis of microscopy data and SPLIFF measurements
Regression and slope estimation of SPLIFF analysis
To test significance of changes in percent conversion over time, the non-parametric local regression (loess) (R Core Team, 2019) was first fitted by taking all biological replicates across the entire time window. The fitted line was then used for the generalized additive model (GAM) (Hastie, 2019). For the GAM model, two time intervals with a one time-point sliding window were used to estimate the slope of conversion over time. Positive slopes with a P-value cutoff of 0.05 were considered as statistically significant. The values of the calculated slopes and the significances of their deviations from the null hypothesis are listed in Table S2.
GraphPad Prism7 was used for statistical data evaluation. Student's t-tests were used to compare the percentage of cells of a certain phenotype. If the data sets did not follow a normal distribution, the significance of differences between data were evaluated by Mann–Whitney U tests. Statistical analysis of the SPLIFF experiment in Fig. 3D was performed using a one-way ANOVA test followed by Tukey's post-test for multiple comparisons.
Saturated cultures of the cells were diluted in fresh medium and grown at 30°C until an OD600 of 0.9–1.0 was reached. Cells were washed in phosphate-buffered saline, attached to poly-lysine treated silicon wafers, and fixed with 2.5% glutaraldehyde (in phosphate-buffered saline containing 1% saccharose) for 1 h at room temperature. Cells were post-fixed with Os4O4 (2% in phosphate-buffered saline) and, after washing in phosphate-buffered saline, gradually dehydrated in 30%, 50%, 70%, 90% and 100% propanol. Fixed cells were critically point dried using carbon dioxide. The samples were finally rotary-coated in a BAF 300 freeze-etching device (Bal-tec, Liechtenstein) by electron beam evaporation with 3 nm of platinium carbon from an angle of 45°. Samples were inspected with a Hitachi S-5200 in-lens field emission scanning electron microscope (Hitachi, Tokio, Japan) at an acceleration voltage of 2 kV.
We thank Steffi Timmermann and Ute Nussbaumer for technical assistance. We thank Judith Müller for advice during Cdc24 purification.
Conceptualization: N.J., J.C., A.D.; Methodology: A.D., T.G.; Software: M.A.M.; Formal analysis: M.A.M.; Investigation: N.J., J.C., A.D., A.B., L.V.-P., T.G.; Writing - original draft: N.J.; Writing - review & editing: A.D., N.J.; Supervision: N.J., T.G.; Funding acquisition: N.J.
This work was funded by grants from the Deutsche Forschungsgemeinschaft to N.J. (Jo 187/5-2).
The authors declare no competing or financial interests.