Calcineurin-dependent regulation of endocytosis by a plasma membrane ubiquitin ligase adaptor, Rcr1

The Rsp5/Nedd4 E3 ubiquitin ligase is responsible for the ubiquitination of membrane proteins, which serves as a sorting signal for endocytosis and lysosomal turnover. PY motifs (with a typical consensus sequence of P/L-P-x-Y) bind the WW domains of Rsp5/Nedd4 family members (Rotin and Kumar, 2009). Many Rsp5/Nedd4 substrates lack PYmotifs and must rely on interactions with adaptor proteins for ubiquitination and vacuolar sorting. Arrestin-related trafficking adaptors (ARTs) recruit Rsp5 to the PM to ubiquitinate PM proteins under certain conditions (Lin et al., 2008). Recently, membrane-anchored Rsp5 adaptors have also been characterized. Ssh4 and Ear1 are membrane-embedded adaptors that recruit Rsp5 to endosome and vacuole membranes, directing the ubiquitination of membrane proteins, which then enter the multivesicular body pathway via the ESCRT machinery for degradation (Léon et al., 2008; Li et al., 2015b; Sardana et al., 2019; Zhu et al., 2017). Overexpression of the membrane-embedded adaptors Ssh4 and Ear1 leads to Rsp5 accumulation at the endosome and vacuole membranes, respectively (Léon et al., 2008; Li et al., 2015b).

Multiple potential Rsp5 adaptor proteins were identified by proteomics (Gupta et al., 2007). After searching for other candidate Rsp5 adaptors, our attention was drawn to a pair of transmembrane proteins, Rcr1 and Rcr2. Based on sequence similarity, Rcr1 and Rcr2 were classified as small type-I transmembrane proteins with PY-motifs. Representative proteins in this family include Comm in Drosophila, PRRG4 in humans, and Rcr1/Rcr2 in yeast (Justice et al., 2017). Comm recruits Nedd4 to down-regulate the Robo receptor at the cell surface of axons, which is required for midline crossing (Myat et al., 2002). RCR1, but not RCR2, was identified as a multicopy suppressor in yeast for congo red or calcofluor white (CFW) sensitivity (Imai et al., 2005). SSH4 and RCR2/SSH5 were found as high copy suppressors for sulfonylurea (SU) sensitivity of shr3Δ mutants, and the SSH gene name is derived as suppressor of SHr3 deletion (Kota et al., 2007b). Later, SSH4 and RCR2 were isolated as high copy suppressors of the Ser/Thr protein phosphatase ptc1Δ mutant at alkaline pH (Tatjer et al., 2016). However, the function and regulation of Rcr proteins remain unknown.

In this study, we found that Rcr1 and Rcr2 are sorted to the PM and vacuole, respectively. We demonstrate that ubiquitin targets Rcr2 to the vacuole, and exomer sorts Rcr1 to the PM using a bipartite sorting motif (LxYYD and LxxPxxxVV) within the Rcr1 tail. Additionally, Rcr2 overexpression results in the down-regulation of some vacuole membrane proteins. We found that Rcr1 is up-regulated (>20 fold) by the calcineurin/Crz1 signaling pathway upon Ca²⁺ treatment, which results in down-regulation of Chs3 by inducing its endocytosis and vacuolar degradation. The calcineurin signaling pathway regulates cell wall remodeling by transcriptional induction of FKS2 (encoding a subunit of the cell wall β[1,3]-D-glucan synthase) upon pheromone, calcium signaling, or loss of FKS1 (Garrett-Engele et al., 1995; Mazur et al., 1995). Here we have uncovered a novel role for the PM adaptor Rcr1 in regulating Chs3 PM abundance via the calcineurin/Crz1 signaling pathway.

Since Rcr1 and Rcr2 have PY-motifs in their cytosolic tails (Fig. 1 A and Fig. S1 A), we first confirmed that Rcr1 and Rcr2 interact with Rsp5. To test this, we performed co-immunoprecipitations (IPs) from cells expressing Rcr1-GFP or Rcr2-GFP and blotted for Rsp5. Indeed, endogenous Rsp5 binds to both Rcr1 and Rcr2. This interaction was abolished in the PY-motif mutants (Fig. 1 B).

We found that Rcr1-GFP localized to the PM (Fig. 1 C). Since rcr1Δ is mildly CFW sensitive (Imai et al., 2005), we confirmed that Rcr1-GFP was as functional as nontagged Rcr1 by testing cell growth on CFW (Fig. S1, B and C). However, Rcr1 was previously characterized as an ER membrane protein (Imai et al., 2005; Smoyer et al., 2016). To confirm that Rcr1 is not localized at the cortical ER, we examined its localization in an ER-PM tether mutant, in which the cortical ER is collapsed into the cytoplasm (Manford et al., 2012). Rcr1 was exclusively localized to PM, even in the ER-PM tether mutant (Fig. S1 D). In contrast to Rcr1, Rcr2-GFP was sorted to the vacuole, and the GFP signal in the vacuolar lumen is the result of the accumulation of free GFP when Rcr2-GFP is internalized into the lumen (Fig. 1 C). Rcr2-Myc was previously shown to be associated with the vacuole membrane (Kota et al., 2007a), but we hypothesized that proteolysis of the Myc tag by vacuolar proteases eliminated its vacuole lumen signal. Consistent with this, Rcr2-pHluorin exhibited a weak vacuole membrane signal in WT cells (the luminal pHluorin signal is quenched by the low pH, Fig. S1 E). We confirmed the functionality of Rcr2-GFP by its SU resistance in a shr3Δ mutant (Fig. S1, F and G). Together, these results establish that Rcr1 and Rcr2 have distinct cellular localizations.

Since Rcr1 and Rcr2 bind Rsp5, we next examined the recruitment of Rsp5 when overexpressing Rcr1 or Rcr2. Overexpression did not affect the sorting of Rcr1-GFP to the PM or Rcr2-GFP to the vacuole (Fig. S1 H). In WT cells, GFP-Rsp5 is diffuse in the cytosol and forms cytoplasmic puncta (Dunn et al., 2004; Fig. 1 D and Fig. S1 I). Upon Rcr2 overexpression, GFP-Rsp5 was recruited to the vacuole, and the cytosolic Rsp5 signal was reduced. Upon overexpressing Rcr1, Rsp5 was recruited to the PM (Fig. 1 D and Fig. S1 I). In addition, the membrane recruitment of Rsp5 was reduced to background levels in Rcr1 or Rcr2 PY-motif mutants (Fig. 1 D and Fig. S1 I). Together, these data demonstrate that Rcr1 and Rcr2 recruit Rsp5 to different cellular locations when they are overexpressed.

Based on the sequence similarity between these two proteins, we considered two possibilities for their distinct localizations. First, Rcr2 was sorted to the PM and then (unlike Rcr1) endocytosed and sorted to the vacuole. To test this, we analyzed Rcr2 localization in an endocytic mutant, end3Δ. The cell wall sensor protein Wsc1, which is continuously endocytosed and sorted to the vacuole (Piao et al., 2007), served as a control. In the end3Δ mutant, Wsc1-GFP was blocked at the PM, but the vacuolar sorting of Rcr2-GFP was unaffected (Fig. 1 E), which ruled out the possibility that Rcr2 was endocytosed. Biosynthetic transport of proteins to the vacuole proceeds via two pathways, the vacuolar protein sorting (VPS) or the AP-3 pathway (Burd et al., 1998). To test which pathway Rcr2 uses, we examined Rcr2-GFP localization in the AP-3 pathway mutant apl5Δ (Cowles et al., 1997a) and the VPS pathway mutant vps45Δ (Cowles et al., 1997b). In comparison with the classic AP-3 cargo protein GFP-ALP, Rcr2-GFP was not affected in apl5Δ. However, Rcr2-GFP was blocked in cytosolic punctate structures in a vps45Δ mutant (Fig. S1 J), suggesting that Rcr2 was sorted to the vacuole via the VPS pathway.

Rcr2 contains four lysines in its cytosolic domain, two of which have acidic residues nearby (KTE, 128–130; DKD, 146–148; Fig. S1 A), which are not present in Rcr1. Lysines with nearby acidic residues are more likely to be ubiquitinated (Radivojac et al., 2010), which may affect Rcr2 sorting. Since the Rcr2 PY-motifs interact with Rsp5, we examined if the Rsp5–Rcr2 interaction is required for vacuolar sorting. Rcr2^py1+2 sorted to both the PM and the vacuole in WT cells, whereas in the end3Δ mutant, it was blocked at the PM (Fig. 1 F), indicating Rcr2 requires Rsp5-mediated ubiquitination to sort to the endosome and vacuole. Consistently, mutating all cytosolic lysines to arginines (Rcr2^K/R) caused PM sorting (Fig. 1 F). In contrast, Rcr1^py1+2 still sorts to the PM (Fig. 1 G). These results suggest that ubiquitin serves as the Golgi-to-endosome sorting signal for Rcr2.

Additionally, adding lysines (EAA^KTE, 130–132; VNS^DKD, 148–150; Fig. S1 A) to Rcr1 induced its vacuolar sorting (Fig. 1 G). To ensure that Rsp5 is the major E3 ligase for Rcr2 ubiquitination, we examined Rcr2-GFP localization in a rsp5^ts mutant (rsp5-1). At semi-permissive temperature, Rcr2 was largely rerouted to the PM (Fig. 1 H). Further, to validate that Rcr2 is ubiquitinated, we expressed Rcr2^WT or Rcr1^+K in a doa4Δ (deubiquitination defective) mutant to stabilize ubiquitinated membrane proteins after multivesicular body sorting. After IP, ubiquitinated species of Rcr2^WT and Rcr1^+K were detected (Fig. 1 I).

We next investigated the possible functions of Rcr2. Since Rcr2 overexpression recruited Rsp5 to the vacuole, we hypothesized that vacuolar membrane proteins may be down-regulated upon Rcr2 overexpression. To test this, we used Wsc1-acLL-GFP (WaG), an artificial cargo protein routed to the vacuolar membrane via an acidic-dileucine motif that is recognized by AP-3 adaptor protein. WaG is ubiquitinated and sorted to the vacuole lumen in an Ssh4-dependent manner (Sardana et al., 2019). In an ssh4Δ mutant, WaG was sorted to the vacuole lumen when Rcr2 or Rcr1^+K were overexpressed (Fig. 1 J; and Fig. S2, A and B), but not Rcr2^py1+2 (Fig. S2, C and D). Other vacuolar membrane proteins (Ypq1, Avt3, Avt4, and Vba4) were also sorted to the vacuolar lumen upon Rcr2 overexpression (Fig. S2 E). Gap1 was shown to have altered activity at the surface in an Rcr1/Rcr2-dependent manner (Kota et al., 2007b). As shown in the Fig. S2 F, no significant vacuole lumen signal of Gap1 was observed, suggesting that Gap1 is not a direct cargo for Rcr2 or Rcr1. These results indicate that recruitment of Rsp5 to the vacuole membrane by Rcr2 induces vacuole membrane proteins turnover, and suggests that Rcr2 may also function as a Rsp5 adaptor on the vacuole membrane, although its native cargo proteins and the conditions that result in their degradation require further investigation.

Exomer is the only known vesicle coat mediating TGN-to-PM trafficking in yeast (Spang, 2015). Exomer contains five subunits: the core coat protein Chs5 and four paralogous Chs5-Arf1 binding proteins: Chs6, Bch2, Bch1, and Bud7. PM sorting signals in most exomer-dependent cargos remain unidentified (Anton et al., 2017; Ritz et al., 2014; Weiskoff and Fromme, 2014); the motif in Fus1 was identified, but it cannot sort a foreign protein to the PM in an exomer-dependent manner (Barfield et al., 2009). To test if Rcr1 requires exomer for its PM delivery, we expressed Rcr1-GFP in chs5Δ and bud7Δbch1Δ, and found that it mis-sorted to the vacuole membrane (Fig. 2 A and Fig. S3 A). To determine if vacuole membrane Rcr1 resulted from endocytosis, we used chs5Δend3Δ and found that Rcr1 is partially mis-sorted to vacuole (Fig. S3 B). These results show that efficient sorting of Rcr1 to the PM requires exomer function. A portion of Rcr1 is still trafficked to the PM in exomer mutants, indicating that an alternative exomer-independent sorting pathway, likely a general secretion pathway for PM proteins, may traffic some Rcr1.

We next mapped the PM targeting sequence in Rcr1. Rcr1 with a truncated C terminus (Rcr1(ΔC])-GFP) sorts to the vacuole membrane, suggesting that the cytoplasmic tail contains PM targeting signals (Fig. S3 C). We generated a series of deletions in the Rcr1 cytosolic tail and screened for mutants that mis-sorted to the vacuole. We quantified the ratio of each Rcr1 mutant mis-sorted to vacuole membrane relative to the total amount of Rcr1 per cell and then converted it to “PM sorting defect score of Rcr1 by exomer” as described in Materials and methods. Removing residues 110–129 or 130–171 mis-sorted Rcr1 (Fig. 2 B and Fig. S3 D). We then performed alanine scanning within this region (Fig. S3, E and F), and found three Rcr1 regions that had defects in PM sorting: 115–120, 151–155, and 156–160 (Fig. 2 C). We mutated single amino acids to alanine within these patches, and found two motifs required for PM sorting: LxYYD and LxxPxxxVV (Fig. 2 D and Fig. S3 G). Further sequence analysis revealed that this bipartite sorting motif is highly conserved in Rcr1 and Rcr2 in different fungal species (Fig. 2, E and F).

To test if these sorting motifs are sufficient to target non-PM proteins to the PM, we made chimeras linking the N terminus and transmembrane domain of Vps10, a vacuole membrane protein, to the Rcr1 cytosolic tail (Vps10(ΔC)-Rcr1-GFP; Fig. S3 H). Vps10(ΔC)-GFP localized to the vacuole membrane (Fig. S3 I; Arlt et al., 2015). Interestingly, Vps10(ΔC)-Rcr1-GFP was partially sorted to the PM. Consistent with Rcr1 sorting, exomer is required for the PM targeting of this chimeric protein (Fig. 2 G; and Fig. S3, H and I), suggesting that the PM sorting motifs of Rcr1 are sufficient for exomer-dependent PM transport.

To see if the Rcr1 cytosolic tail is required for exomer binding, we examined the interaction between the Rcr1 C-terminal tail mutants and Chs5 using yeast two-hybrid and co-IP. Indeed, mutating 115–120 and 151–155 reduced the interaction between the Rcr1 cytosolic tail and Chs5 (Fig. 2 H and Fig. S3 J), indicating that these two regions are required for the Rcr1-Chs5 interaction, directly or indirectly.

We searched the yeast proteome for other proteins containing the LxxPxxxVV motif, and a weak match was found in the cytosolic tail of Fus1. The IxTPK motif is required for exomer-dependent sorting of Fus1 (Barfield et al., 2009). To make sure the mis-sorted Fus1 is not due to endocytosis, we confirmed that Fus1 is sorted to the shmoo tip in end3Δ and mis-sorted to vacuole in chs5Δend3Δ with α-factor treatment (Fig. S3 K). Mutating single amino acids within the IxTPK patch and its downstream region revealed that Fus1 PM sorting was abolished in I176A and P199A mutants, whereas the Fus1 PM sorting exhibited mild defects in V183A and K184A (Fig. 2 I and Fig. S3 L), which indicates Fus1 has an exomer sorting motif (IxxPxxxVK), similar to the motif in Rcr1.

We demonstrated that Rcr1 and Rcr2 were sorted to the PM and vacuole using different sorting signals (Fig. 2 J). We next investigated the regulation of Rcr1 expression and function. Previously, genome-wide studies (Marton et al., 1998; Yoshimoto et al., 2002) showed RCR1 transcriptional up-regulation via the calcineurin signaling pathway. Consistently, Rcr1-GFP expression was highly induced by Ca²⁺ treatment (Fig. 3, A and B). After Ca²⁺ induction, Rcr1 is mainly sorted to the PM (Fig. 3 C) with some intracellular puncta. In contrast, Ca²⁺ treatment does not affect the levels of Rcr2-GFP (Fig. 3 A), suggesting that the Rcr2 and Rcr1 have distinct transcriptional regulatory mechanisms.

In yeast, environmental stress increases intracellular Ca²⁺ levels, which activates calmodulin and leads to the activation of calcineurin phosphatase (Cyert, 2003; Rusnak and Mertz, 2000). Activated calcineurin dephosphorylates several targets to promote cell survival, including the transcription factor Crz1 (Stathopoulos and Cyert, 1997), which binds to calcineurin-dependent response elements (Yoshimoto et al., 2002). To see if Rcr1 induction requires the calcineurin signaling pathway, we examined Rcr1 levels in crz1Δ or cnb1Δ mutants (CNB1 encodes the calcineurin regulatory B subunit; Cyert and Thorner, 1992), and found that Rcr1 up-regulation was completely abolished in these mutants (Fig. 3 D). The calcineurin pathway inhibitor FK506 also blocked Rcr1 up-regulation (Fig. 3 E). In Fig. 3 F, we showed that Rcr1 induction does not require the osmoregulatory mitogen-activated protein (MAP) kinase Hog1 signaling pathway. Thus, Rcr1 is up-regulated via the calcineurin/Crz1 signaling pathway.

We set out to test if Ca²⁺ treatment affects Rsp5 recruitment by Rcr1. Co-IPs showed that more Rsp5 is recruited to Rcr1 with Ca²⁺ treatment (Fig. 3 G). To ensure that this enhanced Rsp5 recruitment does not cause aberrant PM protein endocytosis, we examined the endocytosis of Mup1 after Ca²⁺ treatment. Mup1 remains at the PM despite increased Rsp5 on the PM, and the endocytosis rate of the Mup1 induced by methionine does not change (Fig. 3, H and I). In line with this, overexpression of Rcr1 does not cause Mup1 vacuolar sorting (Fig. 3 J). It was reported that adaptor protein overexpression interferes with other noncognate Rsp5-dependent processes (MacDonald et al., 2020). We therefore asked if the Rsp5 pool is depleted due to enhanced Rsp5 recruitment to Rcr1. We measured Rsp5 binding with the adaptor Art1, but the interaction between Rsp5 and Art1 is unaffected by Ca²⁺ treatment (Fig. 3 K). Further, the PM Ca²⁺ channels Mid1, Cch1, and Ecm7 were not degraded upon Ca²⁺ treatment (Fig. 3, L–N).

It was shown that Rcr1 up-regulation gave rise to resistance to CFW (Imai et al., 2005). We observed that the overexpression of either RCR1^WT or RCR2^K/R gives rise to CFW resistance in WT cells (Fig. 4 A). Since Chs3 produces over 90% of the chitin in the yeast cell wall (Bulawa, 1992; Shaw et al., 1991), we hypothesized that Chs3 is degraded with Rcr1 overexpression. As anticipated, Chs3 sorts to the vacuole for degradation with overexpression of RCR1, but not RCR2 or SSH4 (Fig. 4, B–D), and this is blocked with Rcr1^py1+2 overexpression, demonstrating Rsp5 dependence (Fig. 4, E–G).

We next examined the Chs3 sorting in WT and rcr1Δ after Ca²⁺ treatment. In WT cells, Chs3 is initially localized at the bud-neck and the PM then sorted to the vacuole, and this sorting and degradation is reduced in rcr1Δ (Fig. 4, H–J), demonstrating that Chs3 vacuolar sorting upon Ca²⁺ treatment requires Rcr1. To further show that Rcr1 functions as a Rsp5 adaptor, we examined the interaction between Rcr1 and Chs3. A small fraction of Rcr1-GFP co-IPed Chs3, but only after Ca²⁺ treatment (Fig. 4 K), suggesting that the activation of Rcr1 via the calcineurin/Crz1 pathway is required for its adaptor function. Chs3 vacuolar sorting is due to endocytosis, since Chs3 remains at the PM in an end3Δ mutant (Fig. 4, L and M), despite normal Ca²⁺ Rcr1 induction (Fig. 4 N).

Since the calcineurin pathway triggers Rcr1 up-regulation, we anticipated that Ca²⁺ treatment will cause CFW resistance. However, Ca²⁺ and CFW form a precipitate, preventing us from carrying out this experiment. The calcineurin-docking site of Crz1, PxIxIT, can be altered to PVIVIT (Crz1^PVIVIT) to increase its affinity for calcineurin and resulting in hyperactivity (Roy et al., 2007). As predicted, Crz1^PVIVIT increases Rcr1 levels approximately eightfold (Fig. 4 O). Consistently, Chs3 sorted to the vacuole with Crz1^PVIVIT in a Rcr1-dependent manner (Fig. 4 P), and CFW resistance increased (Fig. 4 Q), suggesting that the calcineurin/Crz1 signaling pathway regulates Chs3 at the PM via Rcr1 up-regulation.

To test if Chs3 undergoes ubiquitination upon Ca²⁺ treatment, we IPed Chs3-GFP in doa4Δ while expressing Myc-Ub. After Ca²⁺ treatment, the ubiquitinated species of Chs3-GFP can be detected (Fig. 5 A).

Since Rcr1 interacts with Rsp5, we confirmed that Chs3 degradation is delayed with Rcr1^py1+2 (Fig. 5, B–D), indicating that the interaction between Rcr1 and Rsp5 is required for Chs3 turnover upon Ca²⁺ treatment. We next asked if Rsp5 is required for Chs3 ubiquitination upon Ca²⁺ treatment. Using rsp5-1, at nonpermissive temperature (37°C) Chs3 remained at the PM, whereas its vacuole sorting and degradation was normal in WT at 37°C (Fig. 5, E–G), indicating that Rsp5 ubiquitinates Chs3, causing its endocytosis and vacuolar sorting.

In this study, we showed that Rcr2 is delivered to the vacuole using ubiquitin as a sorting signal. At the Golgi, the Golgi-located, gamma ear–containing, ARF-binding (GGA) family of coat proteins binds ubiquitinated membrane proteins and facilitate the incorporation of proteins into clathrin-coated vesicles destined for transport from the TGN to the endosome (Lauwers et al., 2009; Scott et al., 2004). Rcr2 contains a similar PM sorting signal as Rcr1, as evidenced by its PM sorting when ubiquitination is impaired, indicating that Rcr2 ubiquitination serves as the dominant targeting signal. One possible model that could explain this is that GGAs may have a higher affinity than the exomer for cargoes, or they may bind cargoes earlier in the Golgi than exomer. Consequently, Golgi-to-endosome targeting directed by ubiquitin overrides PM targeting.

Similar to Rcr1, transcription of FKS2 is highly induced by exogenous Ca²⁺ via the calcineurin signaling pathway (Mazur et al., 1995; Stathopoulos and Cyert, 1997). Why do yeast cells remodel the cell wall composition upon stimulation of the calcineurin signaling pathway? We reason that this is due to the compensatory effect between chitin and β(1,3)-D-glucan biosynthesis—the yin and yang of the yeast cell wall (Levin, 2005; Munro, 2013). β(1,3)-D-glucan is responsible for cell wall rigidity (Cid et al., 1995; Klis, 1994), whereas chitin maintains cell wall plasticity (Tharanathan and Kittur, 2003). Ca²⁺ treatment gives rise to increased β(1,3)-D-glucan content (Luo et al., 2019). Remodeling these two cell wall components has been an adaptive response necessary to counterbalance cell wall stress (Cid et al., 1995; Klis, 1994). After calcineurin activation by Ca²⁺ stress, yeast cells redistribute chitin and β(1,3)-D-glucan in the cell wall by up-regulation of Fks2 (to increase β[1,3]-D-glucan synthesis) and Rcr1 (to down-regulate chitin synthase Chs3).

Rcr1 and Rcr2 exhibit distinct cellular localization and transcriptional regulation. Overexpression of RCR2 results in down-regulation of vacuolar membrane proteins, whereas overexpression (or up-regulation) of RCR1 causes endocytosis and down-regulation of Chs3. These divergent results are in line with the recruitment of Rsp5 to distinct locations by overexpression of these adaptors (Fig. S1 I). How could Rcr1 recognize cargo proteins? We suggest two possible models for the Rcr1-Chs3 interaction upon Ca²⁺ treatment. First, Rcr1-Rsp5 is up-regulated at the PM above a key threshold to ubiquitinate the Chs3 in its vicinity. Second, Chs3 may undergo conformational changes so that Rcr1-Rsp5 can recognize specific motifs of Chs3. In addition to Ca²⁺, there are several cellular stimuli that induce the calcineurin pathway, such as high pH, Na⁺, or Mn²⁺ (Stie and Fox, 2008). We therefore believe that additional Rcr1-dependent cargo proteins at the PM will be identified.

In summary, our findings demonstrate that the Rcr1-Rsp5 complex regulates Chs3 abundance by ubiquitination, which results in its endocytosis. Instead of being recycled from the endosome to TGN by retromer, ubiquitinated Chs3 is delivered to the vacuole lumen upon Ca²⁺ stress (Fig. 5 H). The existence of this ubiquitin-dependent PM protein down-regulation system provides an acute response to degrade unwanted proteins under stress conditions, thereby maintaining cell integrity.

The RCR1, RCR2, and FUS1 genes were cloned from yeast strain SEY6210. The yeast strains bearing N-terminally GFP-tagged AVT3 and AVT4 were made by homologous recombination. The Ypq1-GFP and Vba4-GFP–expressing yeast strains were from laboratory stock (Li et al., 2015a,b). The yeast strains K1619 (W303-1A, ECM7-MYC::TRP), K1643 (W303-1A, CCH1-MYC::TRP), and K1659 (W303-1A, MID1-MYC::KAN) were generous gifts from the laboratory of K. Cunningham (Johns Hopkins University, Baltimore, MD). These tagged Ca²⁺ channels coding genes with selection markers were PCR-amplified and transformed into yeast strain SEY6210.1. When necessary, additional gene deletion and tagging were made using a gene replacement technique with PCR-amplified cassettes (Longtine et al., 1998). All yeast strains and plasmids are described in Table S1 (Strains and plasmids). For fluorescent microscopy experiments, cells were grown overnight to mid-log phase (OD 600∼0.5) in synthetic media at 30°C. For Ca²⁺ stimulation experiments, cells were grown in synthetic media (yeast nitrogen base, dextrose, NH₄Cl, and necessary amino acids) to log phase (OD 600∼0.8), then treated with CaCl₂ (final concentration, 200 mM). For FK506 (LC Laboratories, dissolved in DMSO) with Ca²⁺ treatment, cells were grown in synthetic media to log phase (OD 600∼0.8), then treated with FK506 (final concentration, 1 mM) for 1 h before Ca²⁺ treatment. For α-factor pheromone treatment, cells were grown in synthetic media to log phase (OD 600∼0.4) and treated with 50 ng/ml α-factor pheromone peptide (Zymo Research) for 90 min, fixed with 3.7% formaldehyde (Thermo Fisher Scientific), and washed with water twice before imaging.

Yeast strains were grown in appropriate medium to log phase (OD 600∼0.8). 4 OD of cells were spun down and washed with sterilized water twice, then resuspended with 1ml of water. 10-fold dilutions of cells (starting from 4 OD/ml) were spotted onto synthetic growth medium plate (with selection if necessary) containing 150∼200 µg/ml of CFW (Sigma-Aldrich, Fluorescent Brightener 28, Synonym: Calcofluor White). The plates were incubated at 30°C for 3–4 d. Sensitivity to CFW was determined by comparing the extent of colony density between strains on the control plate and the plates containing CFW.

Yeast mutant shr3Δ harboring different plasmids was cultured in synthetic dextrose medium to log phase (OD 600∼0.8). 4 OD of cells were spun down and washed with sterilized water twice, then resuspended with 1 ml water. 10-fold dilutions of cells (starting from 4 OD/ml) were spotted onto a Yeast Extract–Peptone–Dextrose (YPD) agar plate containing 100 µg/ml of sulfonylurea (2-[[[[(4-Methoxy-6-methyl-1,3,5-triazin-2-liter)amino]carbonyl]amino]sulfonyl] benzoic acid methyl ester, Quali-Pro Incorporation). The plates were incubated at 30°C for 3–4 d. Sensitivity to SU is determined by comparing the extent of colony density on the control plate and the plates containing SU.

For fluorescence microscopy, cells expressing GFP, pHluorin, or mCherry proteins were visualized using a DeltaVision Elite system (GE), equipped with a Photometrics CoolSnap HQ2/sCMOS Camera, a 100× objective, and a DeltaVision Elite Standard Filter Set (“FITC” for GFP/pHluorin fusion protein and “mCherry” for mCherry fusion proteins). Image acquisition and deconvolution were performed using Softworx.

Whole cell extracts were prepared by incubating 6 ODs of cells in 10% trichloroacetic acid on ice for 1 h. Extracts were fully resuspended with ice-cold acetone twice by sonication, then vacuum dried. Dry pellets were mechanically lysed (3×, 5 min) with 100 µl glass beads and 100 µl Urea-Cracking buffer (50 mM Tris-HCl, pH 7.5, 8 M urea, 2% SDS, and 1 mM EDTA). 100 µl protein 2× sample buffer (150 mM Tris-HCl, pH 6.8, 7 M urea, 10% SDS, 24% glycerol, and bromophenol blue) supplemented with 10% 2-mercaptoethanol was added, and samples were vortexed for 5 min. The protein samples were resolved on 11% (m/v) SDS-PAGE gels and then transferred to nitrocellulose blotting membranes (GE Healthcare Life Sciences).

The following antibodies and dilutions were used in this study: rabbit polyclonal anti-G6PDH (1:30,000; SAB2100871; Sigma-Aldrich), mouse monoclonal anti-Pgk1 (1:20,000; 459250; Invitrogen), rabbit polyclonal anti-GFP (1:10,000; TP401; Torrypines), mouse monoclonal anti-GFP (1:1,000; B-2, sc-9996; Santa Cruz), mouse monoclonal anti-Myc (1:5,000, sc-40, Santa Cruz), Chs3 antiserum (generous gift from C. Fromme, Cornell University, Ithaca, NY), IRDye 800CW goat anti-mouse (1:10,000; 926–32210; LI-COR), IRDye 800CW goat anti-rabbit (1:10,000; 926–32211; LI-COR), IRDye 680LT goat anti-rabbit (1:10,000; 926–68021; LI-COR), and IRDye 680LT goat anti-mouse (1:10,000; 925–68070; LI-COR).

The yeast two-hybrid strain HF7C (Feilotter et al., 1994) was cotransformed with pGADGH (empty vector or pGADGH-RCR1 mutants) and pGBT9 (empty vector or pGBT9-CHS5 [1–261]). Cells were grown to saturation in synthetic growth media (-Leu,-Trp) and 10-fold dilutions were spotted on synthetic growth media (-Leu,-Trp) and synthetic growth media (-Leu,-Trp,-His) supplemented with 2 mM 3-amino-1,2,4-triazole. The plates were incubated at 30°C for 3–4 d.

To examine the interaction between Rsp5 and Rcr1, Rcr2, or Art1, yeast strain GOY24 (lacking vacuolar hydrolase Pep4 and Prb1) bearing vector control, Rcr1-GFP, Rcr2-GFP, or Art1-GFP expression vectors were cultured in minimal media and grown to mid-log stage. 100 ODs of cells were collected and washed with water at 4°C. To examine the interaction between Chs3 and Rcr1-GFP, yeast strains GOY24, LZY1390, and LZY1533 were cultured in minimal media (using NH₄Cl as a nitrogen source) and grown to mid-log phase. 150 ODs of cells were collected before and after 200 mM of CaCl₂ treatment and resuspended with ice-cold water three times. The cells were lysed in 500 µl of IP buffer (20 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, pH 8.0, 0.5 mM EGTA, 0.5 mM NaF, 150 mM NaCl, 10% glycerol, 1 mM PMSF, 10 mM N-ethylmaleimide, and 1× Roche cOmplete Protease Inhibitor Tablet/50 ml). Cell extracts were prepared by glass-bead beating with 0.5-mm zirconia beads for five cycles of 30 s vortexing with 1 min breaks on ice. Membrane proteins were solubilized by adding 500 µl of 1% Triton X-100 in IP buffer. The lysates were incubated at 4°C for 30 min with rotation, then spun at 500 g for 5 min at 4°C. The supernatant was clarified by centrifugation at 16,000 g for 10 min. To detect the interaction between Rsp5 and adaptor proteins, the cleared lysate was incubated with GFP-nanotrap resin for 2 h at 4°C. To examine the interaction between Chs3 and Rcr1, the cleared lysate was prebound with 25 µl of protein-G Sepharose resin (GE Healthcare, 17–0885-01) at 4°C for 1 h with rotation, and the supernatant was incubated with 10 µl of Chs3 antiserum and 25 µl of protein-G Sepharose resin at 4°C for 3 h with rotation. After incubation, the resin was washed five times with 0.1% Triton X-100 in IP buffer, and the bound protein was eluted by 2× sample buffer and resolved on 11% (m/v) SDS-PAGE gels.

To examine the interaction between Chs5 and Rcr1 mutants, yeast strain LZY1506 bearing vector control or Rcr1-GFP mutants expression vectors were cultured in minimal media and grown to mid-log stage. 150 ODs of cells were collected and washed with water at 4°C. The cells were lysed in 500 µl of IP buffer (50 mM Hepes-KOH, pH 7.5, 2 mM MgOAc, 1 mM CaCl₂, 150 mM KOAc, 15% glycerol, and 1× Roche cOmplete Protease Inhibitor Tablet/50 ml). Cell extracts were prepared and membrane proteins were solubilized as above. The lysates were incubated with 2 mM DSP (Dithiobis(succinimidyl propionate), Thermo Fisher Scientific) for 2 h on ice and then mixed with 50 mM Tris-HCl (pH 7.5) for 15 min to stop the cross-linking reaction. To detect the Chs5 and Rcr1 protein interaction, the cleared lysate was incubated with GFP-nanotrap resin for 2 h at 4°C. After incubation, the resin was washed five times with 0.1% Triton X-100 in IP buffer, and the bound protein was eluted by 50 µl of 2× sample buffer and resolved on 11% (m/v) SDS-PAGE gels.

To examine the ubiquitination of Chs3 and Rcr proteins, Cells were grown to early log phase in synthetic media. Yeast strain (doa4Δpep4Δ) bearing Myc-Ub expression vector (Zhu et al., 2017) was induced with 100 µM CuSO₄ for 4 h before Ca²⁺ treatment. 100 ODs of cells were harvested after 2 h Ca²⁺ treatment and washed once with water at 4°C. The pellets were resuspended with 500 µl IP buffer (50 mM Hepes-KOH, pH 6.8, 150 mM KOAc, 2 mM MgOAc, 1 mM CaCl₂, 15% glycerol, and 1× Roche cOmplete Protease Inhibitor Tablet/50 ml). Cell extracts were prepared by adding 500 µl glass beads and beating for 30 s (five cycles) with a 1-min break at 4°C. Membrane proteins were solubilized by adding 500 µl of 2% Triton X-100 in IP buffer. Cell lysates were clarified by spinning at 16,000 g for 10 min at 4°C. The resulting lysate was then incubated with 20 µl GFP-nanotrap resin for 4 h at 4°C. The resin was washed five times with 0.1% Triton X-100 in IP buffer. Bound protein was eluted by 50 µl of 2× sample buffer and resolved on 7% (m/v) SDS-PAGE gels.

Yeast strains (WT, end3Δ mutant, or chs5Δend3Δ mutant) harboring Rcr1-GFP or Fus1-GFP mutants vector were cultured in selection synthetic medium to mid-log phase (OD 600∼0.5) at 30°C. Images of Rcr1-GFP and Fus1-GFP were taken by fluorescence microscopy. The fluorescence densities of Rcr1-GFP and Fus1-GFP sorted to PM or vacuole were selected and quantified using ImageJ. The corrected total fluorescence of each selection = selected density − (selected area × mean fluorescence of background readings). The ratio of Rcr1-GFP or Fus1-GFP mis-sorted to vacuole = (the corrected fluorescence density of Rcr1-GFP or Fus1-GFP localized at vacuole) / (the corrected fluorescence density of total Rcr1-GFP or Fus1-GFP). For each set of experiments, the ratio of WT Rcr1-GFP or Fus1-GFP mis-sorted to vacuole membrane in the chs5Δend3Δ was used as a positive control, as shown in Fig. S3, B and K. For each Rcr1 and Fus1 mutant, the PM sorting defect score of Rcr1 by exomer (%) = (the ratio of Rcr1-GFP or Fus1-GFP mutant mis-sorted to vacuole in end3Δ mutant) / (the ratio of WT Rcr1-GFP or Fus1-GFP mis-sorted to vacuole in chs5Δend3Δ mutant). In all figures, the average PM sorting defect score of Rcr1 and Fus1 by exomer from n = 20 cells is shown for each mutant.

For each Vps10(ΔC)-Rcr1 fusion protein, yeast strains (end3Δ mutant or chs5Δend3Δ mutant) harboring Vps10(ΔC)-Rcr1 mutants vector were cultured in selection synthetic medium to mid-log phase (OD 600∼0.5) at 30°C. Image of Vps10(ΔC)-Rcr1 were taken by fluorescence microscopy. The Vps10(ΔC)-Rcr1 sorted to plasma membrane and the whole Vps10(ΔC)-Rcr1 per cell were selected and measured using ImageJ. The corrected total fluorescence of Vps10(ΔC)-Rcr1 in each selection = selected density − (selected area × mean fluorescence of background readings). The efficiency of Vps10(ΔC)-Rcr1 sorted to PM = (the corrected fluorescence density of Vps10(ΔC)-Rcr1 localized at PM) / (the corrected fluorescence density of total Vps10(ΔC)-Rcr1). In Fig. 2 G, the average efficiency of Vps10(ΔC)-Rcr1 fusion protein sorted to PM is shown for each mutant. The significance was determined using a two-tailed t test, P value < 0.0001, α = 0.05 (Bonferroni correction), n = 20 cells for each fusion protein.

Yeast strain (leu2::pRS305-GFP-Rsp5, Vph1-mCherry::TRP) harboring empty vector or vectors for overexpression of RCR2^WT, RCR2^py1+2, RCR1^WT, or RCR1^py1+2 was cultured in selection synthetic medium to mid-log phase (OD 600∼0.5) at 30°C. Images of GFP-Rsp5 were taken by fluorescence microscopy. The GFP-Rsp5 signal at PM and vacuole were selected and measured by ImageJ. The corrected total fluorescence of each selection = selected density − (selected area × mean fluorescence of background readings). The ratio of GFP-Rsp5 recruitment to PM or vacuole = (the corrected fluorescence density of GFP-Rsp5 localized at PM or vacuole) / (the corrected fluorescence density of total GFP-Rsp5). In Fig. 1 D, the ratios of Rsp5 recruitment were measured from n = 20 cells.

All the Western blots in Fig. 3 A; Fig. 4, C, F, and I; Fig. 5, C and F; and Fig. S2, A and C, were quantified using ImageJ software. The significance for Fig. 4, D and G; and Fig. S2, B and D, was determined by two-tailed t test, α = 0.05 (Bonferroni correction), n = 3. For all figures, n.s. indicates not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Fig. S1 shows that recruitment of Rsp5 to different subcellular localizations by Rcr1 and Rcr2. Fig. S2 shows down-regulation of vacuole membrane proteins upon Rcr2 overexpression. Fig. S3 shows the plasma membrane sorting signal within the Rcr1 cytosolic tail. Table S1 is a list of yeast strains and yeast expression plasmids used in this study.

We thank Dr. Chris Fromme (Cornell University, Ithaca, NY) for reagents, strains, plasmids and helpful advice. We appreciate Dr. Kyle Cunningham (Johns Hopkins University, Baltimore, MD) for providing us yeast strains. We are grateful to Dr. Matthew G. Baile, Dr. Mike Henne, Dr. Jeff Jorgensen, Dr. Sudeep Banjade, and Dr. Sho Suzuki for critical reading of the manuscript. We also thank other members of the Emr laboratory for helpful discussions.

The authors declare no competing interests.

Author contributions: L. Zhu designed and performed most the experiments, created the figures, and wrote the manuscript. R. Sardana developed methods and provided yeast strains. D.K. Jin contributed at an early stage of the project by performing some experiments. S.D. Emr supervised the project and edited the manuscript.