Komagataella phaffii Cue5 Piggybacks on Lipid Droplets for Its Vacuolar Degradation during Stationary Phase Lipophagy

Recently, we developed Komagataella phaffii (formerly Pichia pastoris) as a model for lipophagy, the selective autophagy of lipid droplets (LDs). We found that lipophagy pathways induced by acute nitrogen (N) starvation and in stationary (S) phase have different molecular mechanisms. Moreover, both types of lipophagy are independent of Atg11, the scaffold protein that interacts with most autophagic receptors and, therefore, is essential for most types of selective autophagy in yeast. Since yeast aggrephagy, the selective autophagy of ubiquitinated protein aggregates, is also independent of Atg11 and utilizes the ubiquitin-binding receptor, Cue5, we studied the relationship of K. phaffii Cue5 with differentially induced LDs and lipophagy. While there was no relationship of Cue5 with LDs and lipophagy under N-starvation conditions, Cue5 accumulated on LDs in S-phase and degraded together with LDs via S-phase lipophagy. The accumulation of Cue5 on LDs and its degradation by S-phase lipophagy strongly depended on the ubiquitin-binding CUE domain and Prl1, the positive regulator of lipophagy 1. However, unlike Prl1, which is required for S-phase lipophagy, Cue5 was dispensable for it suggesting that Cue5 is rather a new substrate of this pathway. We propose that a similar mechanism (Prl1-dependent accumulation on LDs) might be employed by Prl1 to recruit another ubiquitin-binding protein that is essential for S-phase lipophagy.


Introduction
The Cue5 protein belongs to a protein family called CUET. This family includes Cue5 in yeast and toll interacting protein (TOLLIP) in humans. The Cue5 and TOLLIP possess at least two common structural features, the ubiquitin-binding CUE (coupling of ubiquitin conjugation to endoplasmic reticulum (ER) degradation) domain and AIM, the Atg8-interacting motif [1,2]. The CUE domain is structurally similar to the ubiquitinassociated (UBA) domain [3] and enables the proteins to bind a single ubiquitin molecule [4] or polyubiquitin chains [1]. The AIM or LIR (for LC3-interacting region; an acronym commonly used in mammals) is a conserved motif that confers proteins the ability to bind the autophagosomal marker proteins of Atg8 family, e.g., yeast Atg8 or human microtubule associated protein 1 light chain 3 beta (MAP1LC3B) [5]. Proteins that possess both the ubiquitin-binding domain and AIM often act as the ubiquitin-binding autophagic receptors or, simply, "ubiquitin-Atg8 adaptors", as they facilitate the interaction of ubiquitinated substrates (cargo) with the Atg8-family proteins on the inner surface of autophagosome (vesicular carrier) leading to cargo sequestration and delivery to the vacuole/lysosome for degradation and recycling. Such a type of autophagic degradation where cargo is selected via ubiquitin (or another protein ligand) is termed "selective autophagy" and the selective autophagy of ubiquitinated protein aggregates is called aggrephagy [6]. The Table 1 shows the K. phaffii wild-type (WT) and mutant strains, as well as plasmids that were used in this study. The pRK2 plasmid with the Erg6-GFP expression cassette was described before [17]. The pRK6 plasmid has the K. phaffii CUE5 (PAS_chr2-2_0292) deletion cassette, which is released by double digestion with KpnI and SacI. It was built in two steps: first, by inserting the 1000 bp 5 -untranslated region of CUE5 as KpnI-SalI fragment into the Zeocin R vector, pAP1, to create an intermediate plasmid, pRK5, and then, by inserting the 934 bp 3 -untranslated region of CUE5 as NotI-SacI fragment into pRK5 to create pRK6. The pRK10 plasmid contains the 347 bp promoter and open reading frame (without STOP codon) of CUE5 cloned as KpnI-SpeI fragment into an intermediate plasmid, pRK1, that was constructed by inserting the 6xGly-GFP (without first Met) as SphI-SphI fragment into the integrative vector, pIB1 [19]. The pRK24 and pRK25 plasmids are the site-directed mutagenesis products of pRK10. The pRK24 and pRK25 encode the Cue5 CUE (F78A, P79A) and Cue5 AIM (W308A, Q309A, P310A, L311A) variants of Cue5-GFP, respectively. All the polymerase chain reaction (PCR) products were verified after cloning by sequencing. The K. phaffii strains were transformed with plasmids by electroporation [20,21]. The cue5 deletion mutant was selected as a Zeocin R -transformant of PPY12h strain with KpnI-and SacI-digested pRK6 plasmid on YPD + Zeocin plates (10 g/L yeast extract, 20 g/L peptone, 20 g/L dextrose, 20 g/L agar and 100 mg/L Zeocin) and verified by PCR. The plasmids with HIS4 marker were linearized in the middle of HIS4 with EcoNI for their integration into his4 locus. The His + -transformants were selected on SD + CSM-His plates and screened for the expression of GFP-fusions by immunoblotting with anti-GFP bodies (11814460001; Roche Diagnostics, Mannheim, Germany), as previously described [17].

Fluorescence Microscopy
The K. phaffii cells were imaged as previously described [17]. Briefly, they were grown for 1 d in 1 mL of YPD (1% w/v yeast extract, 2% w/v peptone and 2% w/v dextrose) medium with 1 µL of 1 mg/mL FM 4-64 (T3166; Molecular Probes, Eugene, OR, USA) dissolved in DMSO as the vacuolar membrane stain. During the last hour of incubation, we added 1 µL of 0.1 M monodansylpentane (MDH) (SM1000a; Abcepta, San Diego, CA, USA) to stain the LDs. The 20 µL aliquots were removed and kept on ice before imaging "YPD, 1 d" time-point. The rest of YPD culture was incubated for 2 more days to reach the late S-phase for imaging "YPD, 3 d" time-point. In some experiments, we also removed 3 OD 600 of cells from "YPD, 1 d" cultures, washed them twice with 1 mL of 1× YNB (1.7 g/L yeast nitrogen base without amino acids and ammonium sulfate), and resuspended in 3 mL of SD-N medium (1× YNB and 20 g/L dextrose). After removing the 1 mL aliquots of "SD-N, 0 h" time-point, the 2 mL SD-N cultures were incubated for 24 h (the last hour with 2 µL of 0.1 M MDH) to generate "SD-N, 24 h" time-point. At each time-point, we immobilized the cells using 1% low-melt agarose and imaged them as previously described [17]. All fluorescence microscopy experiments were performed at least in duplicate.

Biochemical Studies
The K. phaffii cells were processed as previously described [17]. Briefly, they were grown in 1 mL YPD medium and equal biomass (1 OD 600 ) was taken at time-points 1 d and 3 d for studying S-phase lipophagy. Ponceau S staining was used as a loading control.
In some experiments, we also removed 3 OD 600 of cells from "YPD, 1 d" cultures, washed them twice with 1 mL of 1× YNB, and resuspended in 3 mL of SD-N medium (starting OD 600 = 1) for studying N-starvation lipophagy. Equal volumes of cultures (1 mL) were taken at time-points 0 h and 24 h in SD-N medium to nullify the differential growth (GFPfusion dilution) effects of various strains in this medium (loading control not applicable). The samples of cell cultures at all time-points were trichloroacetic acid precipitated [26] and analyzed by immunoblotting with anti-GFP (commercially available; see above) or anti-Ape1 [27] bodies. All biochemical studies were performed at least in duplicate.

Statistical Analysis
For statistical analysis of fluorescence microscopy, we calculated the percentage of cells with at least one instance of Cue5-GFP/MDH co-localization for 10 non-overlapping fields of view (with around 50-100 cells per image) from two independent experiments. To qualify as a dual Cue5-GFP/MDH punctum, the intensity of Cue5-GFP fluorescence at the MDH punctum had to be above the background level of diffuse cytosolic Cue5-GFP fluorescence. Then, we calculated the average and standard error for percentage of cells with Cue5 on LDs for 10 non-overlapping fields of view. Student's t-test (two-tailed distribution, two-sample unequal variance) was used to calculate p-values.

K. phaffii Cue5 Accumulates on LDs Specifically in the Stationary Phase
Since the SARs tag their substrates for autophagic degradation, we tested if Cue5 would tag LDs under lipophagy conditions. To study the localization of K. phaffii Cue5, we expressed the Cue5-GFP fusion protein under K. phaffii CUE5 promoter in WT, atg8 and prA,B (pep4 prb1) strains ( Figure 1). Atg8 is a protein that resides on the isolation membrane and completed autophagosome and acts as a binding partner for all the SARs, including Cue5 [1]. Therefore, Cue5 is not associated with any autophagic membranes in atg8 cells, whereas it is expected to be trapped in the intravacuolar autophagic bodies in protease A and B deficient pep4 prb1 cells. We found that after 1 d in YPD medium, Cue5-GFP was mostly cytosolic and absent on the MDH-stained LDs (except in prA,B, see below), as well as in the FM 4-64-delineated vacuoles (in all strains) (Figure 1a,b, top panels). Transferring the cells to N-starvation conditions led to appearance of Cue5-GFP in the vacuoles of WT and prA,B cells (Figure 1a, middle panels) consistent with its role in aggrephagy pathway [1]. The atg8 cells had fragmented vacuoles in SD-N medium, as expected, which precluded addressing the co-localization of Cue5-GFP with the vacuoles. However, incubating the cells 2 d longer in YPD medium and letting the cultures to enter the late S-phase revealed a similar relocation of Cue5-GFP to the vacuoles in WT and prA,B, but not atg8, cells, as expected. Remarkably, now, the Cue5-GFP co-localized with LDs in the substantial fraction of cells of all strains (Figure 1a,b, bottom panels). Therefore, we concluded that K. phaffii Cue5 accumulates on LDs specifically in the S-phase. The efficiency of this process depends on intact autophagy and vacuolar proteolysis. The reason why prA,B cells start accumulating Cue5 on LDs earlier than other strains is probably because they advance to the late S-phase sooner than other cells due to their inability to recycle nutrients in the vacuole.

Cue5 Is Degraded by the Stationary Phase Lipophagy
The K. phaffii LDs are degraded in S-phase via the process known as S-phase lipophagy [17]. Strikingly, Cue5-GFP localizes to both LDs and vacuoles in S-phase of K. phaffii WT and prA,B cells ( Figure 1a) opening a possibility that Cue5 might be delivered to the vacuole via S-phase lipophagy. To test this hypothesis, we performed the Cue5-GFP processing assay (Figure 2, top panels). In this assay, Cue5-GFP is processed to GFP only after its delivery to the vacuole due to the resistance of GFP to vacuolar proteases. Indeed, acute N-starvation in SD-N medium (control) or reaching the late S-phase in YPD medium induced a robust processing of Cue5-GFP to GFP in WT cells but not in prA,B mutant, consistent with the vacuolar degradation of Cue5 under both conditions. If the delivery of Cue5-GFP to the vacuole in S-phase proceeds via S-phase lipophagy, then it should have the same requirements, as the delivery of LD-associated Erg6-GFP that only partially depends on the core autophagic machinery (e.g., Atg8) [17]. Therefore, we tested the Cue5-GFP processing in atg8 cells. While Cue5-GFP processing fully depended on Atg8 under N-starvation conditions, it only partially relied on Atg8 in S-phase ( Figure 2, long exposure for GFP) mimicking the requirement of Erg6-GFP processing [17]. These results suggest that in addition to common localization of Cue5 ( Figure 1) and Erg6 [17] to the LDs in S-phase, these proteins also follow the same route to the vacuole.

Cue5 Is Degraded by the Stationary Phase Lipophagy
The K. phaffii LDs are degraded in S-phase via the process known as S-phase lipophagy [17]. Strikingly, Cue5-GFP localizes to both LDs and vacuoles in S-phase of K. phaffii WT and prA,B cells ( Figure 1a) opening a possibility that Cue5 might be delivered to the vacuole via S-phase lipophagy. To test this hypothesis, we performed the Cue5-GFP processing assay ( Figure 2, top panels). In this assay, Cue5-GFP is processed to GFP only after its delivery to the vacuole due to the resistance of GFP to vacuolar proteases. Indeed, acute The cells of wild-type (WT), atg8 and prA,B (pep4 prb1) strains that express Cue5-GFP were grown for 1 d in YPD with the vacuolar membrane stain, FM 4-64 (red). During the last hour of growth, LDs were stained blue with monodansylpentane (MDH). One aliquot of cell cultures was imaged immediately after 1 d in YPD (top panels), while another was transferred to SD-N and imaged after 24 h (LDs were stained with MDH during the last hour) of nitrogen (N) starvation (middle panels). The rest of YPD cultures was incubated for 2 more days to reach the late S-phase (bottom panels). pends on the core autophagic machinery (e.g., Atg8) [17]. Therefore, we tested th GFP processing in atg8 cells. While Cue5-GFP processing fully depended on Atg N-starvation conditions, it only partially relied on Atg8 in S-phase ( Figure 2, lon sure for GFP) mimicking the requirement of Erg6-GFP processing [17]. These res gest that in addition to common localization of Cue5 ( Figure 1) and Erg6 [17] to in S-phase, these proteins also follow the same route to the vacuole.  Figure S1) was used as a loading c fraction of cells from "YPD, 1 d" cultures was transferred to SD-N medium at OD600 = 1 a volumes of cultures were taken at time-points 0 h and 24 h for testing N-starvation lipopha ing control not applicable). prApe1: precursor form of Ape1; mApe1: mature form of Ape Cue5 is degraded by the S-phase lipophagy. The cells of WT, atg8 and prA,B strains that express Cue5-GFP were grown in YPD medium and equal biomass was taken at time-points 1 d and 3 d for testing S-phase lipophagy. Ponceau S staining ( Figure S1) was used as a loading control. A fraction of cells from "YPD, 1 d" cultures was transferred to SD-N medium at OD 600 = 1 and equal volumes of cultures were taken at time-points 0 h and 24 h for testing N-starvation lipophagy (loading control not applicable). prApe1: precursor form of Ape1; mApe1: mature form of Ape1. To test our hypothesis further and exclude a possibility that Atg8 has a partial requirement for all autophagic pathways in S-phase, we blotted the same lysates with the antibodies to Ape1 (Figure 2, bottom panels). Ape1 is a cargo of the cytoplasm-to-vacuole targeting (Cvt) pathway [27]. After delivery to the vacuole, the precursor form of Ape1 (prApe1) undergoes maturation process, which results in the appearance of two mature forms of the protein (mApe1) [24]. As expected, both N-starvation (control) and late S-phase induced maturation of Ape1 in WT, but not pep4 prb1, cells in agreement with the vacuolar processing of Ape1 under both conditions. Importantly, the maturation of Ape1 was fully blocked in atg8 cells under both conditions (Figure 2, long exposure for Ape1). These results suggest that while being partially required for the selective autophagy of LDs [17], Atg8 is essential for the selective autophagy of Ape1 in S-phase. As such, atg8 mutant can indeed be used to distinguish between the delivery routes to the vacuole in S-phase. Collectively, our results indicate that the LD-associated Cue5, similar to the LD-associated Erg6, is degraded in the vacuole via S-phase lipophagy.
The K. phaffii Cue5-GFP fusion with the WT Cue5 (Cue5 WT ) variant displayed a colocalization with LDs in more than half of WT cells after their incubation for 3 d in YPD medium, as observed earlier. While the Cue5 AIM variant displayed the same pattern of localization as Cue5 WT , the Cue5 CUE variant was severely affected in its ability to accumulate on LDs in S-phase (Figure 3b, top panels). Quantification of images confirmed these observations (Figure 3c, blue columns) suggesting that CUE domain plays an essential role in the recruitment of Cue5 to LDs. We also noticed that Cue5 WT and Cue5 AIM variants of Cue5-GFP had a higher incidence of localization inside the vacuole lumen compared to Cue5 CUE variant. Therefore, we followed up with the Cue5-GFP processing assay that measures vacuolar delivery and degradation (Figure 3d). Indeed, the Cue5 CUE variant of Cue5-GFP generated much less free GFP than Cue5 WT and Cue5 AIM variants in WT background (Figure 3d, left side). Instead, the Cue5 CUE -GFP might have been cleaved by some cytosolic protease that generated an intermediate GFP-fusion band (marked by an asterisk). In conclusion, CUE domain is required for both accumulation of Cue5 on LDs and its subsequent degradation via S-phase lipophagy.
Recently, it was reported that S. cerevisiae Cue5 can self-interact via its CUE domain [28]. Since oligomerization of the K. phaffii Cue5 AIM -GFP with endogenous Cue5 in WT strain could potentially explain normal processing of Cue5 AIM -GFP via S-phase lipophagy (Figure 3d, left side), we created the cue5 deletion mutant and used it as an alternative background strain (Figure 3b-d). The Cue5 AIM -GFP localized to the S-phase LDs in cue5 cells (even slightly better than in WT background), while the Cue5 CUE -GFP was unable to do so (Figure 3b, bottom panels; Figure 3c, green columns) confirming the essential role of CUE domain in the association of Cue5 with LDs. Surprisingly, the Cue5 AIM -GFP was still efficiently processed to GFP in the absence of endogenous Cue5, unlike Cue5 CUE -GFP (Figure 3d, right side), suggesting that Cue5 AIM motif is dispensable for the S-phase lipophagy. Together, we found that CUE domain is important for Cue5 accumulation on LDs and, consequently, its degradation by the S-phase lipophagy. In contrast, AIM motif is unnecessary not only for Cue5 enrichment on LDs (as expected), but also for S-phase lipophagy, raising the question as to whether Cue5 plays any role in this process (see Section 3.5 below). The K. phaffii Cue5-GFP fusion with the WT Cue5 (Cue5 WT ) variant displayed a colocalization with LDs in more than half of WT cells after their incubation for 3 d in YPD medium, as observed earlier. While the Cue5 AIM variant displayed the same pattern of localization as Cue5 WT , the Cue5 CUE variant was severely affected in its ability to accumulate on LDs in S-phase (Figure 3b, top panels). Quantification of images confirmed these observations (Figure 3c, blue columns) suggesting that CUE domain plays an essential role in the recruitment of Cue5 to LDs. We also noticed that Cue5 WT and Cue5 AIM variants of Cue5-GFP had a higher incidence of localization inside the vacuole lumen compared to Cue5 CUE variant. Therefore, we followed up with the Cue5-GFP processing assay that measures vacuolar delivery and degradation (Figure 3d). Indeed, the Cue5 CUE variant of Cue5-GFP generated much less free GFP than Cue5 WT and Cue5 AIM variants in WT background (Figure 3d, left side). Instead, the Cue5 CUE -GFP might have been cleaved by some cytosolic protease that generated an intermediate GFP-fusion band (marked by an asterisk). In conclusion, CUE domain is required for both accumulation of Cue5 on LDs and its subsequent degradation via S-phase lipophagy.

Cue5 Accumulation on LDs and Degradation by S-Phase Lipophagy Depend on Prl1
Previously, we isolated the K. phaffii prl1 mutant that was specifically affected in lipophagy in the S-phase [17]. Since the accumulation of K. phaffii Cue5 on LDs is also specific for S-phase (Figure 1), we studied the relationship between the two proteins. For this, we transformed the Cue5-GFP variants into prl1 strain and tested their recruitment to LDs and degradation with LDs by lipophagy in the S-phase (Figure 4). The control part of the experiment with the Cue5-GFP variants in WT strain worked as earlier, reiterating a strong requirement of CUE domain for both localization of Cue5 to LDs and its degradation in the vacuole via S-phase lipophagy (Figure 4, WT strain). Interestingly, neither of the Cue5-GFP variants accumulated on LDs in the prl1 mutant (Figure 4a, bottom panels). This was confirmed by quantification of images (Figure 4b, golden columns). As a result, the processing of Cue5-GFP variants to GFP was also severely affected in prl1 cells with prl1 and cue5 CUE mutations having an additive effect (Figure 4c, right side). These results suggest that similar to CUE domain, Prl1 is required for the recruitment of Cue5 to LDs and its degradation via S-phase lipophagy.

Cue5 Is Dispensable for S-Phase Lipophagy
Since our results indicated that AIM motif of K. phaffii Cue5 is not required for Sphase lipophagy (Figure 3d, right side), we wanted to address the question if K. phaffii Cue5 plays a role in this process. Recently, it was reported that S. cerevisiae Cue5 is not required for S-phase lipophagy [29] making such a possibility in K. phaffii even more likely. To test lipophagy in K. phaffii cue5 mutant, we used a previously described Erg6-GFP processing assay [17]. As predicted by the non-LD localization of Cue5 under Nstarvation conditions (Figure 1a, middle panels), the cue5 mutant was proficient in the Erg6-GFP processing to GFP relative to WT and prA,B strains in SD-N medium ( Figure 5, left side). However, the same was also true for cue5 and Erg6-GFP processing in S-phase ( Figure 5, right side) suggesting that K. phaffii Cue5 is dispensable for the S-phase lipophagy despite accumulation of the protein on the LDs in S-phase.

Cue5 Is Dispensable for S-Phase Lipophagy
Since our results indicated that AIM motif of K. phaffii Cue5 is not required for Sphase lipophagy (Figure 3d, right side), we wanted to address the question if K. phaffii Cue5 plays a role in this process. Recently, it was reported that S. cerevisiae Cue5 is not required for S-phase lipophagy [29] making such a possibility in K. phaffii even more likely. To test lipophagy in K. phaffii cue5 mutant, we used a previously described Erg6-GFP processing assay [17]. As predicted by the non-LD localization of Cue5 under N-starvation conditions (Figure 1a, middle panels), the cue5 mutant was proficient in the Erg6-GFP processing to GFP relative to WT and prA,B strains in SD-N medium ( Figure 5, left side). However, the same was also true for cue5 and Erg6-GFP processing in S-phase ( Figure 5, right side) suggesting that K. phaffii Cue5 is dispensable for the S-phase lipophagy despite accumulation of the protein on the LDs in S-phase. Cells 2022, 11, x FOR PEER REVIEW 11 of 14 Figure 5. Cue5 is dispensable for S-phase lipophagy. The WT, cue5 and prA,B cells expressing Erg6-GFP were grown in YPD medium and equal biomass was taken at 1 d and 3 d for testing S-phase lipophagy. Ponceau S staining ( Figure S4) was used as a loading control. A fraction of cells from "YPD, 1 d" cultures was transferred to SD-N medium at OD600 = 1 and equal volumes of cultures were taken at 0 h and 24 h for testing N-starvation lipophagy (loading control not applicable).

Discussion
In this study, we explored the relationship between the K. phaffii Cue5 protein and LDs, especially their autophagic turnover by lipophagy (reviewed in [18]). Since S. cerevisiae Cue5 is the established SAR for aggrephagy and proteaphagy [1,7], and both lipophagy and aggrephagy are independent of the key selectivity factor, Atg11, in yeast [14][15][16][17], we entertained the possibility of Cue5 playing a role of the SAR for lipophagy. A careful analysis of Cue5 localization showed that Cue5 can indeed tag LDs under certain conditions, such as the S-phase of growth ( Figure 1). Moreover, Cue5 follows LDs to the vacuole for autophagic degradation during the S-phase lipophagy ( Figure 2). The structure-function analysis of Cue5 indicated that the ubiquitin-binding CUE domain (but not the Atg8interacting AIM motif) is essential for the correct localization of Cue5 to LDs and its subsequent degradation by lipophagy in S-phase ( Figure 3). Interestingly, the accumulation of Cue5 on LDs and its vacuolar degradation also strongly depend on the S-phase lipophagy-specific factor, Prl1 [17] (Figure 4). Despite Prl1 promoting the localization of Cue5 to LDs in S-phase, Cue5 is not its lipophagy effector, because Cue5 is dispensable for the Sphase lipophagy ( Figure 5). Instead, Cue5 is a new "passenger" of the S-phase LDs with an as yet unknown "driver" that takes them to the vacuole for degradation and recycling ( Figure 6). Figure 6. Relationship of K. phaffii Cue5 with LDs and lipophagy. In S-phase, Cue5 accumulates on LDs and is degraded together with LDs via S-phase lipophagy. The accumulation of Cue5 on LDs and its vacuolar degradation strongly depend on the ubiquitin-binding CUE domain and Prl1, the positive regulator of lipophagy 1. Since Cue5 is dispensable for the S-phase lipophagy, it is rather a new "passenger" of this trafficking route. However, a similar mechanism might be employed by Prl1 to recruit another ubiquitin-binding protein (AtgX) that will "drive" S-phase lipophagy.  Figure S4) was used as a loading control. A fraction of cells from "YPD, 1 d" cultures was transferred to SD-N medium at OD 600 = 1 and equal volumes of cultures were taken at 0 h and 24 h for testing N-starvation lipophagy (loading control not applicable).

Discussion
In this study, we explored the relationship between the K. phaffii Cue5 protein and LDs, especially their autophagic turnover by lipophagy (reviewed in [18]). Since S. cerevisiae Cue5 is the established SAR for aggrephagy and proteaphagy [1,7], and both lipophagy and aggrephagy are independent of the key selectivity factor, Atg11, in yeast [14][15][16][17], we entertained the possibility of Cue5 playing a role of the SAR for lipophagy. A careful analysis of Cue5 localization showed that Cue5 can indeed tag LDs under certain conditions, such as the S-phase of growth ( Figure 1). Moreover, Cue5 follows LDs to the vacuole for autophagic degradation during the S-phase lipophagy ( Figure 2). The structure-function analysis of Cue5 indicated that the ubiquitin-binding CUE domain (but not the Atg8interacting AIM motif) is essential for the correct localization of Cue5 to LDs and its subsequent degradation by lipophagy in S-phase ( Figure 3). Interestingly, the accumulation of Cue5 on LDs and its vacuolar degradation also strongly depend on the S-phase lipophagyspecific factor, Prl1 [17] (Figure 4). Despite Prl1 promoting the localization of Cue5 to LDs in S-phase, Cue5 is not its lipophagy effector, because Cue5 is dispensable for the S-phase lipophagy ( Figure 5). Instead, Cue5 is a new "passenger" of the S-phase LDs with an as yet unknown "driver" that takes them to the vacuole for degradation and recycling ( Figure 6).  Figure S4) was used as a loading control. A fraction of cells from "YPD, 1 d" cultures was transferred to SD-N medium at OD600 = 1 and equal volumes of cultures were taken at 0 h and 24 h for testing N-starvation lipophagy (loading control not applicable).

Discussion
In this study, we explored the relationship between the K. phaffii Cue5 protein and LDs, especially their autophagic turnover by lipophagy (reviewed in [18]). Since S. cerevisiae Cue5 is the established SAR for aggrephagy and proteaphagy [1,7], and both lipophagy and aggrephagy are independent of the key selectivity factor, Atg11, in yeast [14][15][16][17], we entertained the possibility of Cue5 playing a role of the SAR for lipophagy. A careful analysis of Cue5 localization showed that Cue5 can indeed tag LDs under certain conditions, such as the S-phase of growth ( Figure 1). Moreover, Cue5 follows LDs to the vacuole for autophagic degradation during the S-phase lipophagy ( Figure 2). The structure-function analysis of Cue5 indicated that the ubiquitin-binding CUE domain (but not the Atg8interacting AIM motif) is essential for the correct localization of Cue5 to LDs and its subsequent degradation by lipophagy in S-phase ( Figure 3). Interestingly, the accumulation of Cue5 on LDs and its vacuolar degradation also strongly depend on the S-phase lipophagy-specific factor, Prl1 [17] (Figure 4). Despite Prl1 promoting the localization of Cue5 to LDs in S-phase, Cue5 is not its lipophagy effector, because Cue5 is dispensable for the Sphase lipophagy ( Figure 5). Instead, Cue5 is a new "passenger" of the S-phase LDs with an as yet unknown "driver" that takes them to the vacuole for degradation and recycling ( Figure 6). Figure 6. Relationship of K. phaffii Cue5 with LDs and lipophagy. In S-phase, Cue5 accumulates on LDs and is degraded together with LDs via S-phase lipophagy. The accumulation of Cue5 on LDs and its vacuolar degradation strongly depend on the ubiquitin-binding CUE domain and Prl1, the positive regulator of lipophagy 1. Since Cue5 is dispensable for the S-phase lipophagy, it is rather a new "passenger" of this trafficking route. However, a similar mechanism might be employed by Prl1 to recruit another ubiquitin-binding protein (AtgX) that will "drive" S-phase lipophagy. Figure 6. Relationship of K. phaffii Cue5 with LDs and lipophagy. In S-phase, Cue5 accumulates on LDs and is degraded together with LDs via S-phase lipophagy. The accumulation of Cue5 on LDs and its vacuolar degradation strongly depend on the ubiquitin-binding CUE domain and Prl1, the positive regulator of lipophagy 1. Since Cue5 is dispensable for the S-phase lipophagy, it is rather a new "passenger" of this trafficking route. However, a similar mechanism might be employed by Prl1 to recruit another ubiquitin-binding protein (AtgX) that will "drive" S-phase lipophagy.
We also found that the efficiency of Cue5 recruitment to LDs in S-phase depends on Atg8, the protein marker of autophagic membranes that binds all the SARs, including Cue5 [1] (Figure 1b, bottom panel). However, we can exclude a direct role of Atg8 in the localization of Cue5 to the S-phase LDs, because the Cue5 AIM -GFP variant with mutated AIM motif was characterized by normal or even slightly elevated accumulation on LDs compared to Cue5 WT -GFP variant in the WT or cue5 background, respectively (Figure 3c). Since (in addition to atg8) a similar partial defect of Cue5 recruitment to LDs was also observed in prA,B cells (Figure 1b, bottom panel), we favor a scenario where intact autophagy and vacuolar proteolysis have an indirect positive effect on this process (e.g., by supporting the energy demanding ubiquitination of LD proteins in S-phase via degradation of intracellular macromolecules).
We present two independent lines of evidence that Cue5 is degraded together with LDs in S-phase. First, the requirement of the vacuolar degradation of Cue5 mimics the specific requirement of the vacuolar turnover of LD-associated Erg6 [17]. Both protein degradation pathways only partially rely on the core autophagic machinery represented by Atg8, while Atg8 is essential for the vacuolar processing of non-LD substrates, such as prApe1 (Figure 2). This suggests that Cue5 is co-degrading with the LDs in S-phase. The second line of evidence comes from our mutational analysis. The point mutations in CUE domain that disrupt the accumulation of Cue5 CUE -GFP variant on LDs in S-phase also disrupt its co-degradation with LDs during lipophagy ( Figure 3). Therefore, Cue5 piggybacks on LDs for its vacuolar degradation during S-phase lipophagy. The reason why Cue5 follows this route is not clear but a similar phenomenon was described for several ER proteins during the ER stress-induced lipophagy [30,31]. Therefore, we hypothesize that Cue5 might aggregate the ubiquitinated ER proteins at the surface of LDs. To our knowledge, this is the first report of the ubiquitin-binding SAR using LDs as its carriers for vacuolar delivery, and not contributing anything to the delivery process. Since the ubiquitin-Atg8 adaptors are more common in mammals than in yeast (and not limited to the Cue5 homolog, TOLLIP), care must be taken in evaluating their roles in lipophagy, because some of them might also "stick" to the LDs as "passengers".
To qualify as a SAR of a given selective autophagy pathway, the protein must not only tag the cargo to be degraded and degrade together with it in the vacuole/lysosome, it must also bridge the cargo with the autophagic membrane via the SAR-Atg8 interaction and both the SAR and its interaction with Atg8 must be essential for degradation [8]. The S. cerevisiae Cue5 meets all the above as a SAR for aggrephagy [1]. However, the K. phaffii Cue5 satisfies only the first two criteria (cargo tagging and vacuolar turnover with the cargo) of a SAR for lipophagy (Figures 1 and 2), the criteria that are common for the SARs and other cargo-associated proteins. Importantly, the Cue5 AIM -GFP variant has a normal vacuolar degradation relative to Cue5 WT -GFP in both the WT and cue5 backgrounds (Figure 3d) suggesting that the Cue5 s AIM motif is dispensable for the S-phase lipophagy. Moreover, the entire Cue5 is dispensable for the S-phase lipophagy, because cue5 mutant has normal degradation of the LD-associated Erg6 ( Figure 5). Therefore, we conclude that K. phaffii Cue5 is a new LD-associated protein and lipophagy substrate in S-phase and not the S-phase lipophagy SAR (denoted as AtgX in Figure 6).
The key to finding the S-phase lipophagy SAR might lie in prl1, the S-phase lipophagyspecific mutant [17] that is blocked in the recruitment of Cue5-GFP variants to LDs (Figure 4). We speculate that Prl1 promotes the ubiquitination of LDs in S-phase attracting the ubiquitin-binding proteins, such as Cue5, to their surface ( Figure 6). One such protein (AtgX) might prove to be essential for lipophagy, as Prl1.