Rsr1 Palmitoylation and GTPase Activity Status Differentially Coordinate Nuclear, Septin, and Vacuole Dynamics in Candida albicans

Understanding how single eukaryotic cells self-organize to replicate and migrate is relevant to health and disease. In the fungal pathogen, Candida albicans, the small GTPase, Rsr1, guides the directional growth of hyphae that invade human tissue during life-threatening infections. Rsr1 is a Ras-like GTPase and a homolog of the conserved Rap1 subfamily, which directs migration in mammalian cells. Research into how this single GTPase delivers complex intracellular patterning is challenging established views of GTPase regulation, trafficking, and interaction. Here, we show that Rsr1 directly and indirectly coordinates the spatial and temporal development of key intracellular macrostructures, including septum formation and closure, vacuole dynamics, and nuclear division and segregation, as well as whole-cell morphology by determining branching patterns. Furthermore, we categorize these functions by differential Rsr1 localization and activity state and provide evidence to support the emerging view that the cytosolic pool of Ras-like GTPases is functionally active.

C-terminal Cys244 putative palmitoylation site and Cys245 CAAX box farnesylation site, respectively. The phenotypes of rsr1Δ mutants expressing Rsr1 K16N or Rsr1 G12V , which mimicked the GDP-or GTP-locked forms of Rsr1, respectively, were compared with those of strains deleted in Bud5 (Rsr1 GEF) or Bud2 (Rsr1 GAP). We identified five phenotypic groupings, each requiring differential regulation of Rsr1. All the phenotypes exhibited by rsr1Δ and hyperactive Rsr1 mutants were phenocopied by deletion of cla4. As Cla4 acts downstream of the Rsr1 effector, Cdc42, and upstream of cortical Rsr1 via Lte1 (11,39), we conclude that Rsr1 and Cdc42-Cla4 interactions at diverse cell sites couple the spatial organization of the cell with the ordered temporal events of the cell cycle.

RESULTS
Rsr1 localization depends on C-terminal Cys 244 and Cys 245 modification but is independent of activation state or GDP-GTP cycling. Farnesylation of the C-terminal CCAAX motif is irreversible and directs nascent peptides to the endoplasmic reticulum (ER) for proteolysis of AAX and methylation (43,44). Palmitoylation at the Golgi apparatus then promotes stable, but reversible, association with the plasma membrane via the exocytic pathway (42,45,46). The role of the CCAAX domain in Rsr1 localization was investigated by expressing a single copy of YFP-Rsr1, YFP-Rsr1 C244S , or YFP-Rsr1 C245A in the rsr1Δ null background ( Fig. 1A to C). The native promoter generated only a faint signal for YFP-Rsr1 and so was replaced by the ACT1 promoter. ACT1p-YFP-Rsr1 rescued all rsr1Δ phenotypes similarly to reintegration of a single copy of RSR1, demonstrating that N-terminally tagged Rsr1 was functional. YFP-Rsr1 localized throughout the PMs of yeast and hyphal cells, including at septa, in a dynamic manner and was enriched toward the hyphal tip ( Fig. 1D and E; see also Movie S1 in the supplemental material). Mutation of the palmitoylation site, Cys244, to serine abolished the YFP-Rsr1 C244S signal at the PM and localized it to endomembranes and the vacuole (Fig. 1F). Expression of Rsr1 C244S rescued all rsr1Δ phenotypes except bud site selection and agar penetration, suggesting that Rsr1 C244S retains function at both the PM and endomembrane, and undergoes GDP-GTP cycling (see below). Although no YFP-Rsr1 C244S signal was observed at the PM, farnesylation may direct Rsr1 C244S transiently to the PM via the normal trafficking pathway, as reported for S. cerevisiae Ras2 [ScRas2(SCaax)] (47), where it may be activated by its GEF, Bud5. Alternatively, there may be as-yetunidentified GEFs at endomembranes that activated Rsr1 C244S . Mutation of Rsr1 Cys245 to alanine abolished YFP-Rsr1 C245A membrane association, and it became cytosolic (Fig. 1G). This is consistent with loss of the CAAX box farnesylation signal that targets proteins to the ER and secretory pathway, as seen for CaRas1 (42).
To investigate whether activity state affected Rsr1 localization and function, strains expressing YFP-Rsr1 K16N (GDP-bound mimic) or YFP-Rsr1 G12V (GTP-bound mimic) were imaged (48,49). YFP-Rsr1 G12V localized uniformly at the cell periphery in yeast and hyphae and was visible at septa ( Fig. 2A). YFP-Rsr1 K16N localized throughout the PM and at septa but was biased toward the hyphal tip and appeared as small aggregates in yeast and hyphal membranes (Fig. 2B). Signal intensity analysis of the hyphal tip region showed that enrichment of YFP-Rsr1 K16N in the 3-m subapical PM accounted for the overall increase in the YFP-Rsr1 signal (Fig. 2C). That Rsr1 localized to the PM in its constitutively GDP-or GTP-bound state indicated that a specific activity state or GDP-GTP cycling was not required for targeting Rsr1 to the PM. The site of Rsr1 activity in S. cerevisiae is determined by positioning of its GEF, Bud5, and GAP, Bud2 (32,50), and so we imaged YFP-Bud5 and YFP-Bud2 in C. albicans yeast and hyphae. YFP-Bud5 was strictly confined to sites of polarized growth, with a faint signal at hyphal septa ( Fig. 2D) (32). YFP-Bud2 appeared as a subapical collar around the mother bud neck in yeast, in hyphae, and at septa (Fig. 2E). The positioning of Bud5 and Bud2 therefore confines Rsr1 activity to sites of growth at the PM but raises the question as to whether all Rsr1 GDP-GTP exchanges require passage through these highly defined PM domains. directional growth, hyphal penetration depth was determined after growth on hyphainducing Spider agar, and Rsr1-mediated stability of the Spk within the hyphal tip was examined using Mlc1-YFP as a marker (8). None of the Rsr1 variants rescued the rsr1Δ invasion phenotype ( Fig. 3C and D), indicating that hyphal invasion required stable Rsr1 palmitoylation and GDP-GTP cycling. However, the stability of Spk positioning in the apex of hyphae expressing Rsr1 C244S was normal (see Fig. S3). Rsr1 membrane association through farnesylation alone was sufficient to confer proper Spk positioning but not to direct substrate invasion. Bud2 was required for agar penetration and Spk positioning but Bud5 was not, even with Rsr1 GDP-GTP cycling, again suggesting an alternative activation mechanism for Rsr1. Reintegration of RSR1 or YFP-Rsr1 in the rsr1Δ null restored the wild-type phenotypes in both yeast and hyphae. Rsr1 deletion or hyperactivation mutants generate a multinucleate phenotype that is rescued by expression of Rsr1-GDP. Yeast cells that are large and rounded are often multinucleate (51,52). To investigate ploidy in the mutants, the percentage of mutant yeast cell populations containing 1, 2, 3, or Ͼ3 4=,6-diamidino-2-phenylindole (DAPI)-stained nuclei was quantified. In the control, reintegrant, and YFP-Rsr1expressing strains, yeast cells were of normal size and contained a single nucleus (Fig. 3E). In the rsr1Δ, bud2Δ, and Rsr1 G12V strains, ϳ40% of cells contained 1 nucleus and a similar percentage contained 2 nuclei. The remaining ϳ20% contained 3 to 6 distinguishable nuclei, where 6 was the limit of detection of discrete nuclei. Hyphal compartments were similarly multinucleate (Fig. 3F). Deletion or hyperactivation of Rsr1 therefore resulted in multinucleate cells. Yeast cells expressing Rsr1 K16N , Rsr1 C244S , or Rsr1 C245A were large and round with random bud sites but were mononucleate, as was the bud5Δ null. Cells expressing nonpalmitoylated but GDP-GTP cycling Rsr1 (Rsr1 C244S ) or GDP-bound Rsr1 (Rsr1 K16N or Rsr1 C245A ), irrespective of cytosolic or PM localization, were mononucleate. This indicates that Rsr1-GDP plays an isoform-specific role in the negative regulation of nuclear division.
Others have reported multinuclear phenotypes that we did not observe; thus, we can exclude the mechanisms involved. First, nuclei did not divide in cell compartments that were bounded by closed septa unless accompanied by the emergence of a new branch, and so the nuclear division and cell polarization signals downstream of START both initiated as normal. Second, anucleate cells did not arise, which excluded a failure to detect correct spindle alignment along the polarity axis, as seen in the cdc10Δ mutant (53,54). Third, although single daughter nuclei underwent retrograde migration into the mother cell, we did not see both daughter nuclei translocate in this way. This suggests that elongation of the mitotic spindle was normal in the rsr1Δ mutants, as this drives retrograde movement of the daughter nucleus that is retained by the mother cell (12). Finally, septal closure did not occur without the transfer of at least one nucleus into the daughter cell, suggesting that the signal to undergo AMR contraction remained coupled to nuclear migration.
Multinucleate cells therefore arose through at least three mechanisms: nuclear division with failed/mislocalized septum formation, multiple nuclear division events within single hyphal compartments (prior to septal closure) without daughter nucleus segregation, and aberrant migration through the presumptum of either undivided nuclei or both daughter nuclei, followed by septal closure.
Deletion of Rsr1 caused defective septin ring localization, structure, and separation. Septin ring assembly requires the iterative activity of Cdc42-Cla4 (21,37). As septin ring behavior was compromised in the rsr1Δ mutant, septin ring morphology, localization, and behavior were examined by imaging Cdc3-GFP or Cdc11-GFP in the rsr1Δ mutant and wild-type cells during hyphal growth (Cdc11-GFP and Cdc3 are functional in C. albicans [12,55]). In wild-type cells, a single septin ring formed within hyphae and, after nuclear migration, split in two to provide space for AMR contraction and synthesis of the primary septum (visualized using CFW) (Fig. 5A). In rsr1Δ hyphae, septins occasionally formed an ectopic primary septum at the mother cell cortex. This suggested that septin ring formation had been initiated by START signaling, but lack of Differential Rsr1 Functions in C. albicans Hyphae ® the Rsr1 positioning signal allowed septins to assemble at random cell sites instead of within the hyphal tip ( Fig. 4E and 5B). Septin rings also coalesced into an immobile cortical patch at the hyphal periphery, suggesting the hetero-octamers were unable to maintain higher order ring structure ( Fig. 4F and 5C). In the rsr1Δ-Cdc3-GFP strain, septins appeared as short spirals or incomplete rings, which formed 3 diffuse and incomplete rings on separation and generated 2 parallel primary septa ( Fig. 5D and E). This is consistent with the observation that two START signals were initiated consecutively in one compartment (Fig. 4A) to produce asynchronous division of two nuclei and two septa.
The ultrastructure of septa was examined in all the Rsr1 mutant strains by using transmission electron microscopy. In yeast and hyphae of the control strains and the bud5Δ mutant, the localization and morphology of septa were normal, with a thin, white, chitinous primary septum and the darker ␤-glucan layer of the secondary septum on either side ( Fig. 6A and C). Septa of the bud2Δ and rsr1Δ mutants showed severe structural deformities, including incomplete or distorted septa and double septa ( Fig. 6B and D) and septa that were mislocalized, appearing as in-growths of cell wall material or as "blisters" at the cell cortex, consistent with the patterns formed by FP-tagged septins (see Fig. S1). Septum morphology was rescued by expression of endomembrane-localized Rsr1 C244S but not by expression of Rsr1 G12V , Rsr1 K16N , or cytoplasmic Rsr1 C245A . Together, these results indicate that normal septum formation and localization required Rsr1 to undergo GTP-GDP cycling, although the Rsr1 GEF, Bud5, was not required. As expression of Rsr1 C244S rescued the septin ring phenotype, unpalmitoylated Rsr1 appears to retain membrane-associated function and undergo GDP-GTP cycling, perhaps by transient association with the plasma membrane or by activation at internal membranes by an unknown GEF. Taken together, these results show that Rsr1 plays a role in determining the site and higher order structure of Cdc42-Cla4-mediated septin ring assembly.
To confirm the role of Rsr1 in septin ring organization, glutathione transferase (GST)-6ϫHis-Rsr1 1-243 , a truncated version of Rsr1 lacking the C-terminal membrane localization domain, was heterologously expressed in Escherichia coli and used in pulldown assays with C. albicans lysates from yeast and hyphal cells. Cdc42 GTPase and its GEF, Cdc24, were identified, as expected. The 5 core septins-Cdc3, Cdc11, Cdc12, Cdc10, and Sep7/Shs1-were detected at levels above that for the GST control in hyphae, although Cdc10 was not seen at a level greater than that of the negative control in yeast (see Fig. S2A), confirming the association of Rsr1 with these septins through direct or indirect binding. Bud5 and Bud2 were not detectable in pulldown assays, but association of these regulators with Rsr1 was confirmed by far-Western blot (Fig. S2B).
Diffuse Dbf2 association with septin rings delayed AMR contraction in the rsr1⌬ mutant. In wild-type cells, nuclear migration through the presumptum is immediately followed by AMR contraction and septum formation. The observation that nuclei cross the septal plane bidirectionally in the rsr1Δ mutant suggested that the signal coupling septal closure with nuclear segregation was compromised or delayed in this mutant. The septin collar is a scaffold for 83 septin-associated proteins that integrate multiple signals to coordinate downstream events in cytokinesis (56). A key mediator of AMR contraction and primary septum formation is the essential Lats/ nuclear Dbf2-related (NDR) kinase, Dbf2. Dbf2 translocates from the spindle pole bodies (SPBs) in late mitosis to associate with the separated septin rings after anaphase, where it phosphorylates Hof1, an F-BAR protein, and Chs2 to effect compartment closure (57)(58)(59)(60). We hypothesized that Dbf2 localization might be compromised by the disorganization of septins in the rsr1Δ mutant. We therefore expressed C-terminally tagged Dbf2-GFP in wild-type cells and the rsr1Δ mutant (Dbf2-GFP functionality was demonstrated in reference 57). In wild-type cells, Dbf2-GFP localized to the SPBs of dividing nuclei, appearing as 2 bright dots aligned along the hyphal axis (Fig. 7A). Over a period of ϳ12 min, the spots faded and Dbf2-GFP appeared transiently at the septum. In the rsr1Δ mutant, Dbf2-GFP appeared as several pairs of bright dots aligned along the hyphal axis, consistent with its association with multiple SPBs in this multinucleate mutant, but formed a diffuse cloud around the presumptum (Fig. 7B). Although disorganized, the septins could therefore recruit Dbf2 to some extent, suggesting that effector association with septin rings was not completely abrogated. Nevertheless, when we visualized Mlc1-GFP, myosin light chain 1, across the septum during AMR contraction, it was visible for 14 Ϯ 3 min in the control strain, but for 21 Ϯ 6 min in rsr1Δ hyphae, before contracting to a central spot at septal closure (Fig. 7E). AMR contraction was therefore delayed at disorganized septa, which remained open for ϳ50% longer than in wild-type cells (Fig. 7C to E) and suggests that the coordination of the activity of effectors by malformed septin rings was compromised.
Fragmented vacuoles failed to fuse and enlarge in rsr1⌬ hyphae, leading to retention of cytoplasm and mitochondria in subapical cells. In S. cerevisiae, Cdc42-Cla4 activity is required for the tethering and docking stages of vacuolar fusion (16,61,62). In addition to Cdc42, our pulldown assays detected 5 proteins involved in vacuole priming, docking, and fusion-Sec18, Ykt6, Vac8, Vtc3 and Vtc4 (Fig. S2A)-suggesting that Rsr1 also directs Cdc42 activity during vacuole fusion in C. albicans. To investigate this further, we imaged mutant hyphae stained with the fluorescent lipid membrane dye, FM4-64. In wild-type hyphae, YFP-Rsr1 and the Rsr1 reintegrant, vacuoles in subapical G 1 cells were enlarged and elongated, occupying ϳ56% and 46% of the compartmental space, respectively (Fig. 8A), while in apical cells, newly inherited vacuoles were fragmented and had not yet undergone fusion or enlargement. In the rsr1Δ null, vacuoles were fragmented in apical and subapical cells alike, regardless of cell cycle stage, suggesting that vacuolar inheritance was normal but none had undergone fusion (Fig. 8C). Comparison of the contents of subapical compartments showed that vacuoles occupied only ϳ36% of the cell compartment, leading to a 45% increase in cytosolic volume in rsr1Δ cells compared to that in the wild type (P Յ 0.0002) (Fig. 8E). The rsr1Δ phenotype was shared by the bud2Δ mutant and was not rescued by expression of Rsr1 K16N , Rsr1 G12V , or cytoplasmic Rsr1 C245A (Fig. 8B and C). However, expression of Rsr1 C244S restored normal vacuole morphology, providing supporting evidence that this Rsr1 isoform can undergo GDP-GTP cycling and was functional at endomembranes. As Cdc42-Cla4 activity is required for vacuole fusion in S. cerevisiae, we imaged vacuoles in the cla4Δ mutant and found the same fragmented phenotype (Fig. 8C). Together, these results suggest that normal vacuole morphology The retention of cytoplasm in subapical cells suggested they may also retain metabolic activity. In the growing apical cells of wild-type hyphae, mitochondrial structure is branched and complex, but in older G 1 -phase compartments, mitochondria are small and peripherally localized until sufficient cytoplasm is synthesized to initiate START (63). The complexity of mitochondrial morphology may therefore be an indicator of the metabolic state of the cell. We investigated mitochondrial morphology by expressing and imaging mitoGFP (mtGFP) in the rsr1Δ mutant (64). In G 1 compartments of wild-type hyphae, mitochondria took the form of a thin filament at the periphery of each compartment between the cell cortex and the large vacuoles, visualized by FM4-64 staining, while mitochondria in growing apical cells were reticulate and densely packed around the fragments of newly inherited vacuole (Fig. 8D). In contrast, mitochondria in all compartments of the rsr1Δ mutant had the dense reticulate appearance of those in the apical cell. The quantified signal intensity in rsr1Δ subapical compartments was double that of wild-type cells, indicating that mitochondria occupied significantly more of the cytoplasmic volume ( Fig. 8D and F). Together, these results suggest that mitochondria in the subapical cells of the rsr1Δ mutant were metabolically active and therefore unlikely to be in early G 1 phase.
Multiple secondary growth axes indicate loss of apical dominance in the rsr1⌬ and hyperactive Rsr1 mutants, which is rescued by expression of Rsr1-GDP. C. albicans hyphae exhibit apical dominance, where new lateral growth (branching) is confined to older cells after sufficient uptake of nutrients for re-entry into the cell cycle, i.e., in cells that are furthest from the growing tip (25). To further probe the link between subapical metabolic activity and the prevalence of multinucleate cells, we quantified the emergence of new secondary growth axes as an indicator of entry into START. In hyphae of the 3 control strains, and in bud5Δ after 5 h of growth few secondary growth axes had formed (range, 0 to 2), and they emerged from the oldest cell ( Fig. 9A and B). In bud2Δ and rsr1Δ mutants, multiple polarity axes (range, 1 to 9) were formed within the same time period, primarily from the oldest cell but some from younger compartments. The same values were determined in the cla4Δ mutant. Thus, multiple distal compartments in these mutants initiated START compared to that in the control strain, a hallmark of loss of apical dominance, and confirmed that these cells retained metabolic activity. Interestingly, the total length of hyphae produced per mother yeast cell by these mutants was greater than the wild-type strain (see Fig. S4). However, this rate of growth was unsustainable at the population level, as 40% to 55% of hyphal germination events resulted in cell lysis (P Յ 0.0001). Hence, there was no net increase in overall growth, as increased hyphal extension in one subpopulation of cells was countered by a high rate of bursting and loss of viability in another ( Fig. 9C and D  andS4). The lysis phenotype was hypha specific, as only 4% of yeast cells burst during bud emergence. Hyphal cell bursting was rescued by the presence of 200 mM sorbitol in the medium (Fig. 9C). This implies that cell wall integrity was compromised during hyphal morphogenesis, in which initiation of true hyphal growth was confirmed by visualization of Mlc1-GFP, a reporter for Spk formation and positioning ( Fig. 9D and S3) (26). As Rsr1 focuses the delivery of exocytic vesicles containing cell wall biosynthesis machinery and structural components to the hyphal tip, which are strongly associated with septin organization, the diffuse dispersal of these elements may compromise cell wall assembly (32,65,66).
Loss of apical dominance was not rescued by expression of Rsr1 G12V , consistent with the findings for the bud2Δ mutant, but was instead rescued by expression of Rsr1 K16N , Rsr1 C244S , or Rsr1 C245A . Expression of Rsr1 C244S had already been shown to restore normal vacuole morphology, cytoplasm distribution, and septin ring organization, which are prerequisites for supporting apical dominance. However, expression of GDP-locked Rsr1 K16N or cytoplasmic Rsr1 C245A was specifically able to restore apical dominance, even though the cells retained fragmented vacuoles and increased cyto-plasm. This strongly suggests that the Rsr1-GDP isoform acts as a negative regulator of START.

DISCUSSION
This study reveals Rsr1 involvement in diverse spatial and temporal functions that are crucial for growth and development in the polymorphic fungus, C. albicans (Fig. 10), all of which likely operate through the Cdc42-Cla4 signaling module, as the observed rsr1Δ phenotypes were phenocopied by deletion of Cla4 and/or are linked to known Cla4 activity (Table 1). Furthermore, the multipurpose functions of Rsr1 signaling are Differential Rsr1 Functions in C. albicans Hyphae ® separable by dependency on the posttranslationally modified state. Unexpectedly, Rsr1 exerted function-specific activity when localized to the cytoplasm, raising the possibility that this is a legitimate site of Rsr1 signaling, as is seen for Ras1 in C. albicans and Drosophila melanogaster (42,67). We classified the phenotypes caused by deletion or mutation of the components of the Bud5-Rsr1-Bud2 signaling module into 5 groups, depending on their rescue by specific isoforms, regulators, or combinations thereof ( Table 1).
The canonical Bud5-Rsr1-Bud2 signaling module is required only for proper bud site selection and yeast cell morphology. Group 1 consisted of a single rsr1Δ phenotype-random bud site selection in yeast-which could not be rescued by expression of any of the Rsr1 mutant constructs. Additionally, all components of the well-characterized Bud5(GEF)-Rsr1-Bud2(GAP) signaling module (50) and Cla4 were required for the wild-type phenotype. Surprisingly, no phenotypes in this class were seen in hyphae.
In group 2, enlarged yeast cell morphology was the only other phenotype we identified for the bud5Δ null (see below "The GEF Bud5 is largely dispensable for Rsr1 Retention of cytoplasm correlated with the presence of complex, likely active, mitochondria and the early onset of START in subapical cells. Disorganized and mislocalized septin ring assembly allowed the formation of a cytoplasmic continuum with aberrant nuclear migration and segregation patterns. Combined, these abnormalities resulted in multinucleate cells and hyperbranching, a defining characteristic of loss of apical dominance in tip-growing cells. Expression of endomembrane-localized Rsr1 C244S rescued these phenotypes. Rescue of inappropriate activation of START by Rsr1 GDP-bound isoforms, Rsr1 K16N or Rsr1 C245A , was sufficient to restore apical dominance, despite the presence abnormal vacuolar and septin ring phenotypes. Bedekovic et al. ® function"). Although the group 2 phenotype was Bud5 dependent, it was rescued in the rsr1Δ mutant by expression of unpalmitoylated Rsr1 C244S . All other components of the Rsr1 signaling module were required for the wild-type morphology, implying that, although visualized only at endomembranes, Rsr1 C244S must be able to under GDP-GTP cycling. As Bud5-GFP resided at the cell periphery and Bud5 was required for normal yeast cell morphology, it is plausible that farnesylated Rsr1 C244S was trafficked to the PM where it was activated. The similarly modified Ras1 C288/7S isoform was reported at the cell periphery in C. albicans, indicating that this trafficking route is possible in farnesylated but unpalmitoylated GTPases (42). The current model for N-Ras cycling is that depalmitoylated N-Ras is trafficked to endomembranes from the PM by cytoplasmic chaperones and is not delivered directly from the Golgi apparatus (68). Without further investigation of the Rsr1 C244S trafficking pathway to endomembranes, it is not clear whether Rsr1 C244S rescued normal cell morphology through transient association with the PM or whether rescue was effected from endomembranes. cla4Δ yeast cells did not exhibit an increase in cell diameter but were misshapen, with elongated bud necks and mislocalized, sometimes incomplete, septa (see Fig. S4C in the supplemental material), consistent with the involvement of Cla4 in septin ring assembly and positioning.
The GEF Bud5 is largely dispensable for Rsr1 function. Group 3 again consisted of a single phenotype, loss of agar penetration by hyphae, where normal invasive growth required wild-type (WT) Rsr1, Bud2, GDP-GTP cycling, and the Cla4 kinase (11). It was independent of Bud5 and was not rescued by any Rsr1 isoform, including Rsr1 C244S , the isoform that restored normal yeast cell morphology and 6 other phenotypes (discussed below), including septin ring organization. Septin mutants form unstable septin rings, are multinucleate, and are defective for invasive growth (69). Rsr1 C244S expression was able to rescue two of these phenotypes but not agar penetration, implying that hyphal invasion is governed by a separable Rsr1-mediated pathway that requires stable association of Rsr1 with the PM and/or palmitoylationdependent targeting to specific lipid domains. We only detected 2 phenotypes on deletion of the Rsr1 GEF, Bud5, and these were yeast specific, raising the possibility that Rsr1 is regulated via an alternative mechanism in hyphae. As Bud2 was required and its deletion mutant phenocopied that of Rsr1 G12V , Bud2 appears to be the only GAP for Rsr1 in yeast and hyphae. Our pulldown assays did not identify Bud5 and Bud2 as Rsr1 interactors, but we were able to show this association by far-Western blot.
Unpalmitoylated Rsr1 C244S rescues septin ring organization, vacuole fusion, and Spk stability. Group 4 phenotypes-septin ring organization, vacuole fusion, and Spk stability-were restored when Rsr1 was able to undergo GDP-GTP cycling and to associate with membranes, as cells expressing GDP-or GDP-locked Rsr1 isoforms or cytoplasmic Rsr1 C245A exhibited the rsr1Δ null phenotype. This group was defined by Bud5 independence and rescue by expression of Rsr1 C244S (Fig. 10), supporting the idea that this farnesylated isoform underwent GDP-GTP cycling and was functional at vacuolar membranes to direct fusion. The rescue of septin ring assembly and Spk stability at the cell periphery further supports the idea that it Rsr1 C244S is trafficked to the PM, despite its lack of a secondary targeting motif, before recycling to endomembranes. Retrograde cycling of depalmitoylated Ras-family GTPases (Ras and Rap1) requires association with cytoplasmic chaperones (Vps35 [vacuolar protein sorting 35] and Cap1/Srv2, respectively), which sequester the hydrophobic acyl group (68,(70)(71)(72)(73). The C. albicans genome contains homologs of Vps35 and Srv2, but the absence of the C-terminal CAAX box in our GST-6ϫHis-Rsr1 1-143 bait prohibited further investigation of this pathway. However, in S. cerevisiae, ϳ10% of Rsr1 was cytosolic, presumably resulting from retrograde trafficking (36).
In S. cerevisiae, the roles of Cla4 in Cdc42-mediated septin ring assembly and vacuole tethering and docking are well characterized (16,21,22,37,62). Here, we showed that Cla4 or Rsr1 deletion in C. albicans produced similar vacuole fragmentation and mislocalized septin phenotypes. We also identified 5 vacuolar proteins and 5 septins in Rsr1 pulldown assays, strongly linking Rsr1 function to Cdc42-Cla4-mediated activity. The results also point to a distinction between phenotypes that are rescued through farnesyl-mediated Rsr1 association with the PM (yeast morphology, septin ring assembly, and Spk stability) and those that additionally require palmitoylation for more stable PM association and/or raft/domain targeting (bud site selection and substrate penetration) (Fig. 10). None of the phenotypes described above was rescued by expression of cytoplasmic Rsr1 C245A .
Rsr1-GDP rescues precocious nuclear division, apical dominance, and hyphal bursting. The phenotypes in group 5 encompassed multinucleate cells and hyperbranched hyphae, which together indicated precocious cell cycle activation in cells distal to the hyphal tip and loss of apical dominance. Rescue was achieved by expression of either GDP-locked Rsr1 K16N or Rsr1 C245A (Fig. 10). As the Rsr1 C245A construct was cytosolic and failed to rescue 6 rsr1Δ phenotypes, it was initially thought to be nonfunctional. However, its ability to phenocopy rsr1 K16N implied a functional relationship between the two isoforms that was likely to be their constitutive GDPbound state; Rsr1 K16N through nucleotide bind-site mutation and Rsr1 C245A through its own intrinsic GTPase activity and no access to membrane-associated GEFs. The inability of Rsr1 C245A to completely rescue hyphal tip lysis, which likely occurred through disorganized cell wall delivery, suggests separable site-specific functions for Rsr1: cytosolic Rsr1-GDP was able to rescue cell cycle but not PM-associated phenotypes, while PM-associated Rsr1-GDP completely rescued both. Fungal Ras1 drives signaling from differential membrane platforms (74,75), and our findings for Rsr1 support emerging examples of GTPases signaling from the cytosol. In D. melanogaster, eye development was supported by cytosolic Ras1 even though its other functions required CAAX box prenylation (67), and in C. albicans, cytosolic Ras1 C288S rescued specific cAMP-dependent growth and filamentation phenotypes (42). Taken together, there is mounting evidence that membrane association is not strictly required for some functions of Ras family GTPases.
How might expression of Rsr1-GDP rescue group 5 rsr1Δ phenotypes? In multinucleate cells, asynchronous nuclear division prior to septal closure indicated that the cell size checkpoint initiated START iteratively within the same body of cytoplasm. The observation that hyphal compartments had double septa is consistent with this, as START also signals the assembly of a new septin ring at the growing tip (76). Although aberrant nuclear inheritance contributed to the phenotype, precocious re-entry into START appeared to be the primary mechanism underlying the generation of multinucleate compartments and led to a 3-fold increase in the formation of new hyphal branches. In a phenotypic background where cytoplasm and mitochondria were retained subapically, Rsr1-GDP played a novel isoform-specific role in suppressing START in cells that are otherwise primed to transit the cell size checkpoint. Furthermore, this was independent of Rsr1-GDP localization. Rsr1-GDP cannot simply inhibit START, as its constitutive expression would be lethal. Instead, Rsr1-GDP may interfere with signaling at a specific stage of the cell cycle by interacting with, or buffering, drivers of START. This mechanism has been suggested for Rap1, the mammalian Rsr1 homolog, which suppresses oncogenic Ras by "diluting" Ras in signaling nanoclusters and obstructing interaction with its effectors, thereby dampening cell proliferation (77).
The observation that the cla4Δ mutant is also multinucleate (11) suggests Rsr1 signaling is epistatic to Cla4 activity. The paradigm is that Rsr1 lies upstream of Cdc42-Cla4 signaling, but studies in S. cerevisiae suggest Rsr1 also lies downstream of Cdc42-Cla4 from late G 1 through mitosis, when Cla4 phosphorylates the GEF domain protein, Lte1. Cla4 and Lte1 localize at the daughter cell cortex, where Lte1 interacts with both Rsr1 and Ras to suppress activation (39,40,78). Our observation that PM-associated Rsr1-GDP was biased toward the hyphal apex is consistent with this. In S. cerevisiae, Cla4-Lte1 activity was viewed as a bifurcated signaling system that inhibits both inappropriate Rsr1-mediated polarization and Ras-mediated START. However, our observation that expression of Rsr1-GDP in a WT Ras1 background can suppress hyperactivation of START implies that Rsr1-GDP inhibits or interferes with Ras1 signal-ing, perhaps by direct binding, as Ras1 was pulled down in our Rsr1 interactor assays (Fig. S2). A further player in this interaction may be Bem1, which specifically binds Rsr1-GDP during G 1 in S. cerevisiae (35,36). However, whether Bem1 acts as a scaffold for Cla4-Lte1-mediated Rsr1-Ras1 inhibition or is sequestered by Rsr1-GDP to avoid inappropriate polarity complex formation remains to be determined.

MATERIALS AND METHODS
Growth conditions and media. The C. albicans strains used in this study are listed in Table 2 Tables 3 and 4, respectively.
(ii) FP-tagging and SDM of YFP-RSR1. The codon-optimized YFP sequence with a Start codon (79) was integrated upstream of the RSR1 ORF (CR_02140W_A including the intron between nucleotides [nt] 8 to 81, inclusive) with a 6-amino-acid (aa) linker (CCGCCATGGCTCGAGGGT). The fragment encoding YFP-RSR1 and flanked by MluI (5=) and SalI (3=) sites was synthesized by GeneArt (Life Technologies) and  (64) was linearized by StuI and integrated at the RPS1 locus of CA8832. Cells were selected and integration was confirmed as described above.
(v) Heterologous expression of GST-Rsr1 1-243 . RSR1 was amplified from plasmid pAB158 using primer 17 (containing the first two RSR1 codons and annealing to RSR1 after the intron) and 18 (annealing upstream of 243-Cys), digested with ApaI and XhoI, and ligated into pET41b-FT, a derivate of commercial plasmid pET41b (Invitrogen) (83), to generate pAB280 (Table 3). A GST-only control was generated by amplifying pET41b-FT with primer pair 15-16 to introduce an EcoRV cut site on each side of the fragment. EcoRV digestion followed by ligation generated plasmid pAB285. Constructs were sequenced and transformed into E. coli BL21(DE3).
(ii) YFP-Rsr1, YFP-Rsr1 G12V , and YFP-Rsr1 K16N fluorescence intensity analysis. Hyphae were grown on a poly-L-lysine slide for 5 h. Images were acquired as described, with 2 ϫ 2 binning and a 3-s exposure. YFP was imaged as before. Images were deconvolved in AxioVision (Rel. 4.8) software, and YFP intensity was measured at the plasma membrane along the 10-m region subapical to the hyphal tip using ImageJ (v. 1.45s).
(iv) Live-cell imaging of FP-tagged proteins. Hyphae were grown in an 8-well ibidi -slide (Ibidi) and imaged every 3 min, or as indicated, on a spinning disk confocal microscope. Optical sections were acquired, and images are maximum intensity projections, unless otherwise stated. Where indicated, the growth medium was supplemented with 2 g/ml CFW. For mitoGFP, hyphae were incubated for 5 h at 37°C and stained with 5 g/ml CFW. Optical sections at 0.4-m increments were captured on a spinning disk confocal microscope, using 488-and 405-nm lasers. The mean fluorescence intensity was analyzed for 5-m 2 areas in the middle of the nonbranched subapical hyphal compartments in ImageJ (v. 1.52p).
(v) Quantification of nuclei and cell diameter. Hyphae were grown on a poly-L-lysine slide for 5 h and fixed in 4% formaldehyde. Nuclei were stained with 1.2 mM DAPI (Sigma) or with Vectashield with 1.5 mg/ml DAPI (Vector Laboratories). Hyphae were co-stained with 0.1 g/ml CFW where indicated. Optical sections were captured on the Zeiss Axio Observer (40ϫ/1.3 NA), using a DAPI filter. Nondividing yeast cells were analyzed for nuclear number and cell diameter.
(vi) CFW staining of bud scars. Hyphae were grown as before and stained with 10 g/ml CFW. Bud scars were visualized after 16 h of incubation at 30°C. Yeast cells were washed with ddH 2 O and stained with 15 g/ml CFW. Z-stacks were captured on a Zeiss LSM880 with Airyscan running Zen 2.3 SP1, with maximum intensity projections generated in Zen.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. MOVIE S1, MOV file, 0.6 MB.