All plasmids and strains used in this study are listed in Supplementary Tables S1, S2, respectively. All engineered strains were confirmed by PCR and DNA sequencing. S. cerevisiae strain YKA1038 expressing cdc14^hm in W303 background was created using the delitto perfetto approach exactly as described (Storici and Resnick, 2006). The plasmid shuffle cdc14Δ strain HCY109 and the pRS314 plasmid expressing CDC14 from its natural promoter were gifts from Dr. Harry Charbonneau, Purdue University. To replace wild-type CDC14 in HCY109, cells were transformed with pRS314 expressing either cdc14^hm or CDC14 followed by selection on 5-FOA to eliminate the URA3-based pRS316-CDC14 plasmid. C. albicans strain JC2712 was created by integration of CaURA3 at the NEUT5L locus in CAI4. JC2711 was created by deletion of both copies of CDC14 using the URA blaster method (Fonzi and Irwin, 1993) followed by integration of CaURA3 at the NEUT5L locus. C. albicans complementing strains expressing a single-copy of wild-type CDC14-3xHA (JC2721) or cdc14^hm-3xHA (HCAL102) were created by transformation of a SalI and EcoRV fragment of pJC347 or pHLP661, respectively, into JC8 (CAI4 cdc14Δ/Δ) and selecting on medium lacking uridine. To create pJC347, the 3’ UTR of CDC14 (+250 to +536 bp after the STOP codon) was PCR-amplified with the addition of SacI and EcoRV sites to the 5′ and 3′ ends, respectively, and cloned into SacI and EcoRV-digested pFA-URA3 (Gola et al., 2003) to generate pFA-URA3-CDC14utr. Then, a synthetic DNA with SalI and BamHI sites added to the 5′ and 3′ ends, respectively, containing 400 bp of CDC14 promoter, the CDC14-3xHA ORF and a 3-UTR with the first 249 bp after the STOP codon of the CDC14 gene was cloned into the SalI and BamHI sites on pFA-URA3-CDC14utr. To create pHLP661, the cdc14^hm mutation (encoding ⁴¹⁴QPRK➔ ⁴¹⁴AARA substitutions) was generated with the QuikChange Lightning site-directed mutagenesis kit (Agilent) using pJC347 as the template and confirmed by DNA sequencing.

Changes in URA3 marker expression are known to impact C. albicans virulence and hyphal differentiation (Lay et al., 1998; Sundstrom et al., 2002; Sharkey et al., 2005). Our strains all have the identical URA3 marker cassette from pFA-URA3 but integrated at 2 different genomic loci (NEUT5L and CDC14, see Table S2). For each locus we have strains with wild-type CDC14 and mutant cdc14 (either cdc14Δ/Δ or cdc14^hm/Δ) to control for URA3 expression effects on observed phenotypes. In addition, we used qRT-PCR (see below for method details) to measure relative URA3 transcript levels in strains JC2712, JC2711, JC2721, and HCAL102 (Supplementary Figure S1). No statistically significant differences in transcript levels were observed between URA3 inserted at NEUT5L versus CDC14, demonstrating that all four strains can be compared without concern that observed phenotypes are influenced by locus-specific URA3 expression differences.

The yap1Δ and hog1Δ strains were obtained from the S. cerevisiae haploid deletion library (Horizon Discovery). The yen1Δ mus81Δ S. cerevisiae strain was a gift from Dr. Lorraine Symington, Columbia University. S. pombe strains were a gift from Dr. Kathleen Gould, Vanderbilt University. The cdc14-3 strain was a gift from Dr. Michael Weinreich, Van Andel Research Institute.

Escherichia coli expression plasmids for purification of recombinant Cdc14 enzymes were generated using the Gateway cloning system (Invitrogen). S. cerevisiae CDC14 coding sequence was amplified by PCR from an existing plasmid. C. albicans CDC14 coding sequence was amplified from a cDNA preparation to avoid its intron. Both were cloned into the pENTR/D-TOPO entry vector and then transferred to pDEST17 destination vector following the provided instructions for expression with an N-terminal 6xHis tag. Mutations to the QPRK motif of both were generated by QuikChange mutagenesis. Full coding sequences for all plasmids were verified by DNA sequencing.

Liquid S. cerevisiae and C. albicans cultures were grown in YPD medium (10 g/l yeast extract, 20 g/l peptone, 20 g/l glucose). YPD was supplemented with 40 mg/l adenine (YPAD) for S. cerevisiae W303 strains. Cultures were grown at 30°C with shaking at 225 rpm. S. pombe strains were grown in YES (5 g/l yeast extract, 30 g/l glucose, 225 mg/l adenine, histidine, leucine, uracil, and lysine) at 30°C with shaking at 225 rpm. E. coli were grown in 2xYT (16 g/l tryptone, 10 g/l yeast extract, and 5 g/l NaCl) at 37°C with shaking at 225 rpm. Agar was added to 20% (w/v) for growth on solid medium. For agar plate spotting assays, single colonies were grown to saturation. Strains were serially diluted in eightfold steps starting from OD₆₀₀ = 1.0 and 5 μl of four consecutive dilutions were spotted on plates and grown at 30°C for 3–5 days. C. albicans cdc14Δ/Δ cultures were diluted one less time than other strains and spots therefore correspond to 8-fold higher absorbance; this was necessary to normalize colony density on untreated control plates. For hyphal induction on solid medium, C. albicans strains were streaked for single colonies on Spider media plates (10 g/l beef broth, 10 g/l mannitol, 2 g/l K₂HPO₄, 20 g/l agar) or YPD plates supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals) and grown at 37°C for 5 days. For liquid hyphal induction, strains were grown overnight to saturation at 30°C and diluted to an OD₆₀₀ = 0.1 in pre-warmed 37°C YPD containing 10% FBS. Cultures were then incubated at 37°C with shaking at 225 rpm.

Total protein extracts were prepared as described (Kushnirov, 2000; von der Haar, 2007) with subtle modifications. Briefly, 8 ml of mid-log phase cells were treated with 10% trichloroacetic acid and pelleted by centrifugation. Cell pellets were washed with 10 ml 70% ethanol and 2x with 1 ml water, then re-suspended in 1 ml 0.2 M NaOH and incubated 10 min on ice. Cells were pelleted, resuspended in 160 x OD₆₀₀ μl loading dye (120 mM Tris–HCl pH 6.8, 4% sodium dodecyl sulfate, 0.02% bromophenol blue, 20% glycerol), heated at 95°C for 10 min, and lysates clarified by centrifugation at 16,000 x g. Proteins were separated on 10% tris-glycine SDS-PAGE gels, transferred to 0.45 μm nitrocellulose membranes (Bio-Rad) and probed overnight at 4°C with mouse anti-HA (1:5,000; Sigma-Aldrich, 12CA5), rabbit anti-PSTAIR (1:5,000; Millipore-Sigma, 06–923), rabbit anti-p44/42 MAPK (1:2,500; Cell Signaling Technology, 9102) or rabbit anti-G6PDH (1:5,000; Sigma-Aldrich, A9521). Secondary anti-mouse and anti-rabbit antibodies conjugated to horseradish peroxidase were from Jackson ImmunoResearch (115–035-003 or 111–035-003) and used at 1:10,000 dilution for 60 min at 4°C. Immunoblots were developed using Clarity Western ECL Substrate (Bio-Rad) and imaged on a ChemiDoc MP multimode imager (Bio-Rad).

6xHis-Cdc14 enzymes were purified from E. coli as previously described (DeMarco et al., 2020).

Activities toward varying concentrations of pNPP and synthetic phosphopeptides were assayed as previously described (DeMarco et al., 2020). Data were fit with the Michaelis-Menten equation in GraphPad Prism to determine steady-state kinetic parameters.

Cells were grown to mid-log phase (OD₆₀₀ ~ 0.8) in YPD and, where indicated, treated with 50 ng/ml micafungin for 1 h prior to RNA extraction. RNA was isolated and purified using acid phenol as described (Serratore et al., 2018; Baker et al., 2022). Reverse transcription was performed using All-in-One 5X RT MasterMix (ABM). Primer Express 3.0 software was used to design primers and qRT-PCR was performed as previously described (Serratore et al., 2018; Baker et al., 2022) with at least three biological replicates. Data were analyzed using the comparative C_T method (2^-ΔΔCT). Internal controls were RDN18 (18S rRNA) for C. albicans and ACT1 for S. cerevisiae. All samples were normalized to an untreated wild-type strain. Primers used for all qRT-PCR analyses are listed in Supplementary Table S3.

Larvae of G. mellonella were purchased from waxworms.net (St. Marys, OH, United States) and infection assays were performed similar to previous reports (Cotter et al., 2000; Ames et al., 2017). Upon arrival, larvae were held at room temperature in the dark, without food for 24 h before injection. C. albicans strains were grown to saturation in YPD and cells washed three times with PBS, sonicated for 10 s, and counted using a Luna-FL automated cell counter (Logos Biosystems). Larvae (n ≥ 40 per strain) of similar size and color were injected with 6 μl of 2 × 10⁵ cells into their lower left proleg using a Hamilton syringe. An additional group was injected with PBS to serve as a negative control. Injected larvae were incubated in petri dishes at 37°C in the dark for 5 days. Death was measured every 24 h based on the ability of the larvae to flip over after being placed on their backs. Survival data were plotted using the Kaplan–Meier survival curve and statistical analysis was performed using a log-rank Mantel–Cox test.

A mouse model of invasive C. albicans infection of immune-compromised individuals was based on several previously described experimental models (Abruzzo et al., 2000; Cheng et al., 2013; Segal and Frenkel, 2018). BALB/c mice (n = 8 per strain, 4 males and 4 females, 5-week-old) were acclimatized for 7 days upon arrival. On days 8 and 11, mice were rendered neutropenic by intraperitoneal (IP) doses of cyclophosphamide (Segal and Frenkel, 2018); 150 mg/kg on day 8 and 100 mg/kg on day 11. On day 12, 5 × 10⁵ CFUs of a C. albicans strain (measured using a hemocytometer after staining with 0.4% trypan blue to rule out dead cells) in 0.2 ml PBS were administered via IP injection (Cheng et al., 2013). An additional group received cyclophosphamide but was not infected to serve as a negative control. Mice were monitored for morbidity every 6 h and were euthanized upon showing severe signs of illness, including weight loss (>20%), dyspnea, unresponsiveness, staggered gait, and inability to eat. On day 19, surviving mice were humanely euthanized through CO₂ inhalation. Lungs, liver, spleen, and kidneys were collected from euthanized mice for histopathology. Survival data were plotted using the Kaplan–Meier survival curve and statistical analysis was performed using a log-rank Mantel–Cox test. Selection of 5 × 10⁵ CFUs of each C. albicans strain was based on a pilot study performed using different doses (5 × 10⁶ to 1 × 10⁴ CFUs) of both wild-type and cdc14Δ/Δ strains.

Lungs, kidneys, and livers from euthanized mice were formalin-fixed, and transferred to 70% ethanol on the next day. Tissues were embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E) at the Purdue Histology Research Laboratory for histopathological examination. Tissues were examined by a board-certified veterinary anatomic pathologist, histological features were evaluated, and representative tissue sections were photographed. Fungal pathogens were visualized and evaluated for morphology using periodic acid Schiff (PAS) staining.

Samples were prepared based on a previously published protocol (Wright, 2000) with minor modifications. Day 1: Samples were fixed by adding cell culture to an equal volume of 5% glutaraldehyde in 0.1 M sodium cacodylate buffer. After 5 min cells were pelleted by centrifugation, resuspended in 2.5% glutaraldehyde in the same buffer, and fixed overnight at 4°C. Day 2: Cells were rinsed and resuspended in 2% aqueous potassium permanganate for 1 h, then rinsed until the solution cleared. Cells were then embedded in agarose, en bloc stained in 1% aqueous uranyl acetate for 1 h and rinsed. Day 3: Samples were dehydrated with a graded series of ethanol, then transferred into a 2:1 mixture of ethanol and LR White resin and rotated for 2 h. They were then transferred into a 1:1 mixture of ethanol and LR White resin and rotated overnight. Day 4: Samples were moved into pure LR White resin, rotated for 2 h, then placed under house vacuum. The resin was changed three times throughout the day and the samples remained under vacuum overnight. Day 5: Samples were embedded in LR White resin in gelatin capsules and polymerized at 60°C for 24 h. Thin sections were cut on a Leica UC6 ultramicrotome and stained with 4% uranyl acetate and lead citrate. Images were acquired on a FEI Tecnai T12 electron microscope equipped with a tungsten source operated at 80 kV.

The observation that the isolated catalytic domains of multiple fungal Cdc14 orthologs have reduced enzyme activity raised the possibility of creating hypomorphic mutants to test the importance of full Cdc14 activity. Using S. cerevisiae Cdc14 (ScCdc14), we set out to identify regions of the ~200 amino acid C-terminal tail that specifically contribute to enzyme activity. We purified full length ScCdc14(1–551), the catalytic domain truncation ScCdc14(1–374) and a truncation of intermediate length, ScCdc14(1–449) (Figure 1A; Supplementary Figure S2A) and compared activity under steady-state conditions. Consistent with prior reports (Taylor et al., 1997; Li et al., 2015), ScCdc14(1–374) reaction rate toward pNPP was greatly reduced compared to ScCdc14(1–551) (Figure 1B). ScCdc14(1–449) activity was nearly identical to ScCdc14(1–551), demonstrating that the sequence between amino acids 374 and 449 is required for full activity. We used ScCdc14(1–449) to represent full, or “wild-type,” activity in subsequent experiments. The same analysis with a biologically-relevant phosphopeptide substrate derived from a known ScCdc14 substrate site in Yen1 (Eissler et al., 2014) revealed a similar large reduction in k_cat for ScCdc14(1–374) compared to ScCdc14(1–449), indicating that the difference in activity is not substrate-dependent (Figure 1C).

A multiple sequence alignment of diverse fungal Cdc14 orthologs revealed a single, short invariant sequence within the otherwise poorly conserved C-terminus (Figure 1D) that was not observed in animal Cdc14 sequences (Supplementary Figure S3). Part of this sequence, QPRK, strongly resembles the optimal substrate recognition motif of Cdc14 enzymes, SPxK (“x” is preferentially a basic amino acid) (Bremmer et al., 2012; Eissler et al., 2014; Kobayashi and Matsuura, 2017), suggesting it could influence catalysis by interacting with the active site as a pseudosubstrate. Furthermore, this motif is within residues 374–449 required for full activity. We substituted Ala either for the Pro and Lys residues, analogous to two critical positions for Cdc14 substrate recognition, or for the Gln, since this residue would occupy the pSer binding pocket if the motif were acting as a pseudosubstrate. The resulting ScCdc14(1-449^QARA) and ScCdc14(1-449^APRK) variants exhibited k_cat decreases similar to ScCdc14(1–374) with both pNPP (Figure 1E) and phosphopeptide (Figure 1F) substrates. Mutation of QPRK did not affect substrate specificity as activity was still dependent on the same major substrate features as wild-type Cdc14 enzymes (Supplementary Figure S2B). We conclude the QPRK motif makes a significant contribution to enzyme turnover rate and is responsible for the observed rate difference between catalytic domain truncations and full-length enzymes.

The strict conservation of the QPRK motif across the fungal kingdom suggests it is biologically important. To characterize its biological significance, we generated a haploid S. cerevisiae strain expressing the hypomorphic cdc14^QARA allele (hereafter called cdc14^hm) as the sole source of Cdc14 from the natural CDC14 locus using the delitto perfetto approach (Storici and Resnick, 2006). In liquid culture, the mutant strain grew at a similar rate to the wild-type parental strain at low cell density but slowed somewhat as cell density increased and ultimately terminated at a slightly lower optical density (Supplementary Figure S4A). Thus, the essential mitotic exit function was not disrupted by the reduction in Cdc14 catalytic rate, consistent with our prior observations that it requires a small fraction of normal Cdc14 activity (Powers and Hall, 2017). We performed agar plate spotting assays in the presence of a variety of cellular stress conditions to determine if reducing Cdc14 activity sensitizes cells to stress. cdc14^hm colony size was slightly smaller than wild-type on untreated control plates but otherwise grew similarly and cdc14^hm was no more sensitive than wild-type to genotoxic, oxidative, and osmotic stresses (Figure 2A; Supplementary Figures S4B–D). The temperature-sensitive cdc14-3 allele was previously shown to impart some sensitivity to calcofluor white (CFW), a chitin-binding compound that causes cell wall stress (Breitkreutz et al., 2010). We confirmed this observation and found that cdc14^hm exhibited strong sensitivity to CFW (Figure 2B) and another chitin-binding compound, Congo red (Supplementary Figure S4E). To determine if this sensitivity was unique to chitin binding chemicals we performed spotting assays in the presence of echinocandin drugs, which cause cell wall stress by inhibiting synthesis of the major cell wall polymer (1,3)-β-d-glucan (Sucher et al., 2009). cdc14^hm was acutely sensitive to both micafungin and caspofungin compared to wild-type (Figures 2C,D). cdc14^hm was also sensitive to growth at high temperature, another condition that generates cell wall stress (Fuchs and Eleftherios, 2009; Supplementary Figure S4F). Cell wall stress sensitivity was observed when cdc14^hm, but not CDC14, was expressed from a CEN plasmid in a cdc14Δ background (Supplementary Figure S4G). Sensitivity in cdc14^hm was complemented by transformation with a CEN plasmid expressing wild-type CDC14 (Supplementary Figure S4H), suggesting it is a recessive loss of function phenotype caused by the reduced activity. Micafungin sensitivity was rescued in the presence of sorbitol, an osmotic stabilizer known to suppress cell wall integrity defects (Torres et al., 1991), suggesting that reduced Cdc14 activity compromises cell wall integrity (Figure 2E).

Cell wall damage activates the cell wall integrity (CWI) signaling pathway (Levin, 2011). Activation of the CWI pathway can be monitored by immunoblotting for the phosphorylated CWI MAP kinase, Slt2. We compared phospho-Slt2 levels in log phase wild-type and cdc14^hm cultures in the absence and presence of micafungin. cdc14^hm showed a 3-fold increase in phospho-Slt2 in the absence of micafungin, but both wild-type and cdc14^hm strains showed similar Slt2 activation after brief treatment with micafungin (Figure 2F). We also monitored expression of FKS2, which encodes a catalytic subunit of (1,3)-β-d-glucan synthase that is induced by CWI signaling during cell wall stress (Mazur et al., 1995; Zhao et al., 1998; Jung and Levin, 1999). Consistent with the pSlt2 result, FKS2 transcript level was twofold higher in cdc14^hm compared to the isogenic wild-type strain in the absence of micafungin (Supplementary Figure S4I). These results are consistent with a cell wall integrity defect, but mostly normal CWI signaling, in cdc14^hm S. cerevisiae cells.

To test if Cdc14-dependent sensitivity to cell wall stress is conserved in other fungal species, we performed spotting assays with CDC14 deletion strains of S. pombe and C. albicans. S. pombe clp1Δ exhibited clear sensitivity to caspofungin, but not calcofluor white, presumably due to the lack of chitin in S. pombe cell walls (Pérez et al., 2018; Figure 3A). C. albicans cdc14Δ/Δ exhibited strong sensitivity to calcofluor white, caspofungin, and micafungin (Figure 3B). These observations suggest that Cdc14 makes a broadly conserved contribution to fungal cell wall integrity.

We next tested if the QPRK motif functions similarly in C. albicans Cdc14 (CaCdc14) and if mutating it is sufficient to cause cell wall stress sensitivity, as in S. cerevisiae. We purified recombinant truncated CaCdc14 (residues 1–427, which retains the QPRK motif), as full-length protein was insoluble, and created a variant with Ala substitutions at the Gln, Pro, and Lys positions of the QPRK motif (Supplementary Figure S5A). CaCdc14(1–427) exhibited the same major substrate preferences as ScCdc14 and other Cdc14 orthologs (Supplementary Figure S5B), consistent with its identical active site composition (Supplementary Figure S3). Steady-state kinetic analyses with the Yen1 phosphopeptide substrate revealed a 6-fold reduction in k_cat for CaCdc14(1–427^AARA) compared to CaCdc14(1–427) (Figure 3C). Thus, the QPRK motif makes a similar, albeit somewhat weaker, contribution to CaCdc14 phosphatase activity.

To study the biological significance of the QPRK motif in C. albicans we complemented cdc14Δ/Δ with either cdc14^hm-3xHA (where cdc14^hm encodes the QPRK➔AARA substitutions) or CDC14-3xHA (Supplementary Figure S5C). cdc14^hm growth rate was intermediate between the wild-type and cdc14Δ/Δ strains, consistent with a partial loss of activity and function (Supplementary Figure S5D). In agar plate spotting assays cdc14^hm/Δ displayed hypersensitivity to a range of echinocandin concentrations, similar to cdc14Δ/Δ, whereas CDC14/Δ behaved like the CDC14/CDC14 wild-type parent strain (Figures 3D,E). Thus, echinocandin hypersensitivity requires only a partial reduction in Cdc14 activity. As in S. cerevisiae, hypersensitivity was specific to cell wall stress as cdc14Δ/Δ and cdc14^hm/Δ were mostly insensitive to genotoxic, oxidative, and osmotic stress (Supplementary Figure S5E).

We next tested if CWI signaling was altered in Cdc14-deficient C. albicans strains, similar to S. cerevisiae. We used the same anti-p44/42 MAPK antibody used to detect the phosphorylated, active form of the CWI MAP kinase and Slt2 ortholog, Mkc1, by immunoblotting. In the absence of added stress, liquid cultures of both cdc14Δ/Δ and cdc14^hm/Δ displayed an equal elevation in active Mkc1 compared to CDC14/CDC14 (Figures 4A,B). Interestingly, CDC14/Δ cells displayed an intermediate elevation in phospho–Mkc1 signal, suggesting that even CDC14 haplo-insufficiency can cause some chronic cell wall abnormality. The reason for this is unclear, but it suggests that CWI is particularly sensitive to Cdc14 activity level in C. albicans. In contrast to the basal state, when cultures were briefly challenged with micafungin, activated Mkc1 was significantly lower in both cdc14Δ/Δ and cdc14^hm/Δ compared to CDC14/CDC14 and CDC14/Δ (Figures 4A,B). This suggests that Cdc14 activity may also play a role in normal CWI pathway activation in response to certain cell wall stress signals.

To test if Cdc14 activity affects gene expression downstream of Mkc1 we used qRT-PCR to measure transcript levels of several cell wall stress-responsive genes (Ibe and Munro, 2021) in the absence and presence of micafungin. Consistent with the phospho-Mkc1 immunoblotting results, ECM331, PGA13, and PGA31 transcripts were significantly elevated in cdc14Δ/Δ and/or cdc14^hm/Δ strains compared to CDC14/CDC14 and CDC14/Δ in the absence of micafungin, although CRH11 showed no change (Figures 4C–F). Interestingly, after brief micafungin treatment cdc14Δ/Δ cells failed to induce expression of any of the four genes and the fold-induction in cdc14^hm/Δ was significantly reduced compared to wild-type strains. This effect was not due to the genes already being maximally induced since three out of four transcript levels were lower in micafungin-treated cdc14Δ/Δ and cdc14^hm/Δ compared to CDC14/CDC14 and CDC14/Δ. Thus, Cdc14 activity is important for expression of at least some cell wall stress-responsive genes, and our results collectively imply a complex relationship between Cdc14 activity and cell wall integrity signaling in C. albicans (see section Discussion).

We used TEM to look for overt differences in cell wall structure between wild-type and Cdc14-deficient C. albicans yeast form cells that could explain the observed cell wall integrity phenotypes. The most striking difference was the structure of the septum in dividing cells. In CDC14/CDC14 and CDC14/Δ cells the primary septum was uniformly visible as a bright, straight band bisecting the bud neck, with thin layers of secondary septum on either side (Figures 5A,B), as expected (Meitinger and Pereira, 2016). Relatively normal primary septa were also observed in some cdc14^hm/Δ and cdc14Δ/Δ cells, however we frequently observed primary septa with abnormal morphologies in these strains that were never observed in wild-type cells (Figures 5C,D). The primary septa were often discontinuous, non-linear, forked, or deviated from the bud neck. Moreover, the secondary septa in cdc14^hm/Δ and cdc14Δ/Δ were often thickened, and in some cases appeared to lack primary septa altogether. These secondary septa were reminiscent of previously described remedial septa that form in the absence of primary septum assembly (Walker et al., 2013). From the TEM analysis we conclude that high Cdc14 activity is critical for proper septum assembly during cell division.

Previous characterization of cdc14Δ/Δ C. albicans revealed defects in separation of yeast form cells following cell division, resulting in formation of large cell clumps (Clemente-Blanco et al., 2006). This phenotype was linked to regulation of the Ace2 transcription factor that controls expression of cell separation genes following cytokinesis and septation. In liquid cultures cdc14^hm/Δ also formed large cell clumps that were indistinguishable from cdc14Δ/Δ, whereas CDC14/Δ did not (Figure 5E). The clumps were dispersed by sonication, consistent with normal completion of cytokinesis, but incomplete dissolution of septa. In asynchronously growing cultures we detected similar reductions in the transcript levels of Ace2 target genes in cdc14^hm/Δ and cdc14Δ/Δ by qRT-PCR (Supplementary Figure S6). This suggests that high Cdc14 activity is also needed to fully activate Ace2. Thus, CaCdc14 appears to play critical roles in both the assembly and dissolution of the septum during yeast form cell division. Septum defects could, in principle, contribute to the impaired cell wall integrity of Cdc14-deficient cells.

Candida albicans cdc14Δ/Δ was also previously reported to be defective in hyphal differentiation and invasive growth on agar plates (Clemente-Blanco et al., 2006). We therefore tested if cdc14^hm/Δ also exhibits these phenotypes. On Spider medium plates, cdc14Δ/Δ and cdc14^hm/Δ were completely unable to develop filamentous colonies and exhibited very limited agar invasion (Figure 6A). Microscopic analysis of cells extracted from the Spider agar after washing of the plate surface revealed that even invasive cells exhibited a severe reduction in the number and length of hyphae compared to strains with wild-type CDC14 (Figure 6B). Similar lack of invasion and formation of hyphae by cdc14Δ/Δ and cdc14^hm/Δ were also observed on YPD-serum agar plates (Supplementary Figures S7A,B). In liquid cultures, hyphal differentiation differences between the strains were more subtle. Both cdc14Δ/Δ and cdc14^hm/Δ exhibited hyper-polarized growth consistent with hyphae, but at 2 h after serum induction their elongated cells were shorter on average and displayed more pseudohyphal character than those of the wild-type CDC14 strains (Supplementary Figure S7C). We conclude that normal hyphal development also requires high CDC14 activity.

Hyphal differentiation is thought to be important for C. albicans virulence (Lo et al., 1997; Kadosh, 2019), suggesting that reduced Cdc14 activity might compromise pathogenesis. We tested this idea in two established animal models of invasive candidiasis, the larvae of the wax moth Galleria mellonella (Cotter et al., 2000; Fuchs et al., 2010; MacCallum, 2011; Segal and Frenkel, 2018), and immunosuppressed mice (Hohl, 2014). In G. mellonella, cdc14Δ/Δ and cdc14^hm/Δ exhibited dramatically reduced virulence compared to CDC14/CDC14 and CDC14/Δ (Figure 7A). In mice the virulence defect associated with reduced Cdc14 was even more dramatic. An initial titration of fungal dose revealed cdc14Δ/Δ to be almost completely avirulent even at the highest fungal titer (Supplementary Figure S8A,B). At the minimum CDC14/CDC14 dose required for complete mortality, cdc14^hm/Δ failed to kill any mice and, surprisingly, CDC14/Δ resulted in 50% mortality, killing only the female mice, suggesting CDC14 heterozygosity is sufficient to impact virulence and that there may be a gender bias in sensitivity to Cdc14 deficiency (Figure 7B).

Histological analysis of mouse organs following infection revealed extensive colonization by predominantly hyphal cells of CDC14/CDC14 and CDC14/Δ strains, resulting in inflammation and necrosis, as expected (Figures 7C–E; Supplementary Figure S8C). In contrast, minimal colonization and necrosis were observed after infection by cdc14^hm/Δ and cdc14Δ/Δ strains and the rare fungal cells that were detected were almost completely yeast form, consistent with the hyphal differentiation defects observed on agar plates. In mice infected with wild-type C. albicans, organs like the liver and lungs showed evidence of shock prior to appearance of fungal cells. Lesions consistent with shock included endothelial leakage characterized by edema fluid, multifocal hemorrhage and congestion, and occasional intravascular microthrombi (Figures 7D,E). These were not observed in the organs of mice infected with cdc14^hm/Δ or cdc14Δ/Δ, raising the possibility that Cdc14 promotes secretion of soluble factors that contribute to vascular damage. These results show that high Cdc14 activity is critically important for C. albicans pathogenesis.

Cdc14 has been implicated in regulation of septation and cell separation previously. In S. cerevisiae Cdc14 promotes delivery of the chitin synthase Chs2, which synthesizes the primary septum, to the bud neck after completion of mitosis (Chin et al., 2012). It also regulates actomyosin ring assembly (Miller et al., 2015) and the ingression progression complex that coordinates actomyosin ring contraction and membrane ingression with Chs2-dependent septum assembly (Meitinger et al., 2010; Palani et al., 2012; Foltman et al., 2016). Mutations in the Mitotic Exit Network, which activates ScCdc14, lead to septum structural defects (Meitinger et al., 2010). In the phytopathogenic fungi Fusarium graminearum, Magnaporthe oryzae, and Aspergillus flavus, CDC14 deletion reduces the number of septa in conidia and vegetative hyphae and in M. oryzae perturbs correct localization of appressoria-delimiting septa (Li et al., 2015, 2018; Yang et al., 2018). Despite the evidence that Cdc14 is important for normal septation, our results appear to be the first to show that Cdc14 deficiency causes structural defects in primary septa. Cdc14 is also important for dissolving septa to allow cell separation. ScCdc14 dephosphorylates and activates Ace2 to induce cell separation genes (Brace et al., 2011; Powers et al., 2017), and preventing ScCdc14 nuclear export or reducing Cdc14 level via inducible degradation resulted in cell separation failure (Bembenek et al., 2005; Powers and Hall, 2017). Cdc14 is also needed to activate Ace2 to promote cell separation in C. albicans (Clemente-Blanco et al., 2006). Septation and cell separation defects could be a primary cause of the cell wall stress sensitivity observed in Cdc14-deficient C. albicans, although direct evidence of such a causal relationship is still missing.

In addition to causing a cell wall structural defect, our results suggest that Cdc14 deficiency impacts CWI signaling in C. albicans in a complex way. Despite the elevated CWI pathway activation in unstressed Cdc14-deficient strains (including heterozygous CDC14/Δ), cdc14Δ/Δ and cdc14^hm/Δ exhibited reduced stimulation of phospho-Mkc1 following micafungin treatment. This suggests that Cdc14 deficiency both causes cell wall damage and also plays a role in activating the response to this damage. Consistent with this idea, Cdc14-deficient strains displayed modest up-regulation of a subset of known cell wall stress-responsive genes in the absence of added stress (in line with elevated phospho-Mkc1) but were defective in stimulating expression of these genes in response to micafungin exposure. The extent of this defect correlated with the severity of Cdc14 deficiency; cdc14Δ/Δ showed no micafungin-stimulated expression of the four cell wall stress-responsive genes we tested. These results are somewhat difficult to reconcile but are most consistent with Cdc14 influencing CWI signaling at steps upstream and downstream of Mkc1. Consequently, the cell wall stress sensitivity phenotype could be explained by a Cdc14-dependent cell wall structural defect, a deficiency in the CWI signaling response to the cell wall defect, or both. The fact that CDC14/Δ cells exhibited increased basal phospho-Mkc1 but normal CWI signaling and transcriptional response after micafungin treatment reinforces the idea that Cdc14 functions in multiple capacities to promote cell wall integrity that are differentially sensitive to its activity level. We surveyed a small subset of cell wall stress-responsive genes, and a complete picture of how Cdc14 impacts the response to cell wall stress will ultimately require global approaches like RNA-seq.

Cdc14 was previously linked to several fungal stress pathways. S. pombe Clp1 and ScCdc14 are released from the nucleolus and activated in response to genotoxic stress (Diaz-Cuervo and Bueno, 2008; Broadus and Gould, 2012; Villoria et al., 2017). A global phosphatase and kinase interaction network analysis revealed multiple interactions of ScCdc14 with stress signaling MAP kinases (Breitkreutz et al., 2010) and other studies have reported connections between Cdc14 and the osmotic stress response mediated by the Hog1 MAP kinase (Chasman et al., 2014; Tognetti et al., 2020). In the phytopathogen A. flavus, a cdc14Δ strain exhibited growth delays in the presence of chitin-binding compounds and osmotic stress (Yang et al., 2018). In the entomopathogen Beauveria bassiana, cdc14Δ displayed modest sensitivity to multiple stresses, including the chitin-binding Congo red (Wang et al., 2013). These studies imply that fungal Cdc14 may play diverse roles in facilitating fungal stress responses, although we emphasize that cdc14^hm/Δ C. albicans with partial loss of Cdc14 activity exhibited a very specific sensitivity to cell wall stress. An area for future investigation will be determining if the cell wall stress hypersensitivity of Cdc14-deficient strains is related to the virulence defect. Specific mutations that affect Cdc14 function in cell wall integrity would facilitate this and could be enabled by a thorough understanding of Cdc14 regulation during cell division, in response to cell wall stress, and during hyphal development.

Our results reinforce the fact that different functions of Cdc14 require different levels of activity. In S. cerevisiae the essential mitotic exit function requires a small fraction of normal Cdc14 activity (Powers and Hall, 2017). Why would functions in cytokinesis and cell wall integrity require much higher Cdc14 activity? In S. cerevisiae the essential function is fulfilled by nuclear Cdc14 (Bembenek et al., 2005) whereas cytokinesis functions require cytoplasmic Cdc14 generated by the Mitotic Exit Network kinases (Sanchez-Diaz et al., 2012). One possibility is that cytoplasmic functions require more enzyme because of the greater volume relative to the nucleus, which effectively dilutes Cdc14 concentration. In C. albicans Cdc14 localization is also cell-cycle regulated. It is primarily nuclear prior to mitosis, then appears at the spindle pole bodies during early mitosis, and finally becomes cytoplasmic and localizes to the bud neck in late mitosis (Clemente-Blanco et al., 2006). A second possibility is that the mitotic exit function requires minimal activity to trigger a positive feedback system (López-Avilés et al., 2009), whereas the cytokinesis, septation, and cell wall integrity functions depend on bulk substrate dephosphorylation that necessitates higher Cdc14 concentration. A third possibility is that Cdc14 has high affinity for mitotic exit substrates that must be dephosphorylated relatively early after Cdc14 activation and lower affinity for substrates involved in later processes like cytokinesis and septation, resulting in a need for higher Cdc14 activity to achieve their dephosphorylation. There is evidence that substrate affinity does influence the order of Cdc14 substrate dephosphorylation during mitotic exit (Kataria et al., 2018), providing some support for this possibility.

Regardless of the true explanation, it appears clear that Cdc14 activity levels in C. albicans and S. cerevisiae are physiologically important. Even heterozygous CDC14/cdc14Δ cells exhibited some phenotypes, most notably the reduction in virulence in the mouse infection assay. We speculate that the invariant QPRK motif in the C-terminus of fungal Cdc14 enzymes exists to provide a mechanism for regulating Cdc14 activity level around a critical threshold. Although the need for this mechanism is still unclear, results from both S. cerevisiae and S. pombe support its existence. A C-terminal Cdk phosphorylation site in ScCdc14 was previously shown to inhibit enzyme activity by an unknown mechanism (Li et al., 2014). This site happens to be adjacent to the QPRK motif and also appears to be essentially invariant in fungi (although its exact position relative to QPRK varies, see alignment in Figure 1D). In S. pombe Clp1 phosphorylation by Cdk1 at C-terminal sites (including the one adjacent to its QPRK motif) also reduces enzymatic activity and is required for normal kinetics of mitotic progression (Wolfe et al., 2006). Thus, it appears likely that phosphoregulation of QPRK motif function is a mechanism for controlling overall Cdc14 activity level in fungal species. It will be interesting to test if phosphorylation adjacent to the QPRK motif of CaCdc14 also regulates enzyme activity and if dephosphorylation is required to promote CaCdc14 function in septation, hyphal development, and the response to cell wall stress.

While a number of substrates of ScCdc14 and SpCdc14 involved in regulation of cytokinesis and/or septation have been identified (Mocciaro and Schiebel, 2010; Meitinger et al., 2012; Chen et al., 2013; Kuilman et al., 2015), it remains unknown what substrates of CaCdc14 must be dephosphorylated to promote normal septum assembly, cell wall integrity, and hyphal differentiation. Understanding how Cdc14 influences these processes will require identification of the relevant direct substrates. A few candidate substrates that may be relevant to our observed phenotypes exist. Evidence suggests that Ace2 is a CaCdc14 substrate likely related to the observed cell separation defects in Cdc14-deficient strains (Clemente-Blanco et al., 2006). A previous proteomic study of CaCdc14 interacting proteins and candidate substrates did not identify proteins with obvious connections to septum assembly but did identify Ace2, the chitinase Cht4, and a couple components of the actomyosin contractile ring (Kaneva et al., 2019), although these were not validated as substrates. The Nim1 kinase, Gin4, which helps organize the septin scaffold at the bud neck that is crucial for proper septation and cytokinesis is dephosphorylated by CaCdc14 (Au Yong et al., 2016) and this could potentially be related to the septum structure and cell wall integrity defects observed here. Still, our current knowledge of Cdc14 substrates in C. albicans is very limited and it is likely that CaCdc14 dephosphorylates multiple targets that influence cell wall integrity, septum assembly, and pathogenesis.

The observation that partial reduction in Cdc14 activity severely impairs pathogenesis in animal models of invasive candidiasis suggests that Cdc14 might be a useful antifungal drug target where even modest therapeutic inhibition of activity could successfully combat infections. Moreover, the fact that reduced Cdc14 activity sensitizes C. albicans to echinocandin drugs raises the possibility that Cdc14 inhibitors could be valuable for combating clinical echinocandin resistance. We argued previously that the Cdc14 active site should be amenable to specific inhibitor development (DeMarco et al., 2020). High resolution structural information on substrate recognition by the fungal Cdc14 active site is available (Kobayashi and Matsuura, 2017), and despite sharing a common catalytic mechanism with PTPs, Cdc14 possesses a strict and unusual substrate specificity that could be mimicked for effective and highly selective inhibitor development. In animals, deletion of CDC14 has little impact on cell division and development (Saito et al., 2004; Berdougo et al., 2008; Mocciaro et al., 2010) and the possibility of successful therapeutic intervention by partial inhibition of Cdc14 activity further minimizes the likelihood of deleterious effects on human cells. The Cdc14 active site region is invariant across the fungal kingdom (DeMarco et al., 2020) and therefore inhibitors developed to target CaCdc14 would likely be effective at inhibiting Cdc14 orthologs from many other pathogens. For this reason, it will be useful in the future to determine if Cdc14 is similarly required for infection by other human pathogens.