An optogenetic system to control membrane phospholipid asymmetry through flippase activation in budding yeast

Lipid asymmetry in biological membranes is essential for various cell functions, such as cell polarity, cytokinesis, and apoptosis. P4-ATPases (flippases) are involved in the generation of such asymmetry. In Saccharomyces cerevisiae, the protein kinases Fpk1p/Fpk2p activate the P4-ATPases Dnf1p/Dnf2p by phosphorylation. Previously, we have shown that a blue-light-dependent protein kinase, phototropin from Chlamydomonas reinhardtii (CrPHOT), complements defects in an fpk1Δ fpk2Δ mutant. Herein, we investigated whether CrPHOT optically regulates P4-ATPase activity. First, we demonstrated that the translocation of NBD-labelled phospholipids to the cytoplasmic leaflet via P4-ATPases was promoted by blue-light irradiation in fpk1Δ fpk2Δ cells with CrPHOT. In addition, blue light completely suppressed the defects in membrane functions (such as endocytic recycling, actin depolarization, and apical-isotropic growth switching) caused by fpk1Δ fpk2Δ mutations. All responses required the kinase activity of CrPHOT. Hence, these results indicate the utility of CrPHOT as a powerful and first tool for optogenetic manipulation of P4-ATPase activity.

The only factors regulating the activity of Dnf1p/Dnf2p are the Ser/Thr protein kinase Fpk1p and its paralog, Fpk2p. The FPK1 gene was identified as a gene responsible for a synthetic lethal mutation with cdc50∆, and the fpk1Δ fpk2∆ mutant shows almost the same phenotypes as the dnf1Δ dnf2Δ or the lem3Δ mutant 32 . Moreover, Fpk1p phosphorylates Dnf1p/Dnf2p in vivo and in vitro 32,34,35,36 ; thus, Fpk1p/Fpk2p are Dnf1p/Dnf2p-activating kinases. Fpk1p and Fpk2p share the highest sequence homology in yeast with the AGCVIII kinase domain of phototropins (PHOTs), plant specific blue-light (BL) photoreceptors 32,37 . We have previously introduced a PHOT from Chlamydomonas reinhardtii (CrPHOT) into an fpk1Δ fpk2Δ mutant with a conditional Cdc50p mutant (P GAL1 -CDC50 fpk1Δ fpk2Δ. As a result, we have shown that CrPHOT complements the growth defect of the fpk1Δ fpk2Δ mutant in a BL-dependent manner 37 . This result indicates the possibility that CrPHOT controls P4-ATPases in a light-dependent manner. Therefore, we examined whether this system could be used as a new optogenetic technology.
In this study, we reintroduced CrPHOT into the fpk1Δ fpk2Δ mutant or the P GAL1 -CDC50 fpk1Δ fpk2Δ mutant and biochemically assessed the lipid translocation activity to show that it was indeed regulated by BL irradiation. Furthermore, Dnf1p/Dnf2p-dependent cellular processes, such as vesicle transport, were regulated in these cells by BL. Our results suggest the potential of CrPHOT as an optogenetic tool to regulate membrane functions.

Results
Optical control of yeast cell growth by Chlamydomonas PHOT. To establish a system to control P4-ATPase (flippase) activity by light (Fig. 1a), we attempted to prove that CrPHOT regulates P4-ATPases at the molecular level. On the basis of the highest sequence homology among the kinase domains of Fpk1p/Fpk2p and CrPHOT 32 , we have shown that CrPHOT complements the synthetic lethality of a P GAL1 -CDC50 fpk1Δ fpk2Δ yeast mutant in a BL-dependent manner 37 . In this study, we introduced CrPHOT and its derivatives (Fig. 1b) into the mutant and examined its growth activity again. Consequently, we reconfirmed that CrPHOT suppresses the growth defect of the P GAL1 -CDC50 fpk1Δ fpk2Δ yeast mutant in glucose-containing medium under BL but not under red light (RL) or in darkness (Fig. 1c, CrPHOT). Furthermore, a K-fragment with constitutive kinase activity suppressed the growth defect regardless of the light condition, and KDm with a kinase-dead mutation, failed to restore the growth even under BL (Fig. 1b, c) 37 . The substitution of a conserved cysteine (Cys) to an alanine (Ala) in the light-oxygen-voltage (LOV) domain is known to decrease the extent of light activation 38 . LOV2m, which has this mutation in the LOV2 domain, was used as a mutant to reduce light sensitivity in previous research 37 . LOV2m certainly reduced but did not completely abolish the ability to restore growth under BL 37 (Fig. 1b, c). Therefore, we newly constructed LOV1/2 m, which has Cys to Ala mutations in two LOV domains (LOV1 and LOV2; Fig. 1b). The growth activity of LOV1/2 m under BL was considerably lower than that of LOV2m (Fig. 1c). Hence, LOV1/2 m was used as a mutant to reduce photosensitivity instead of LOV2m in this study. We then examined the expression level of CrPHOT and derived proteins in yeast with an anti-HA tag antibody (Fig. 1d). The level of CrPHOT protein was not altered with light conditions as in previous analysis 37 . The amount of each CrPHOT derivative was also not changed by BL irradiation, and their levels differed little. These results reconfirmed that complementation by CrPHOT depends on its photoactivation and furthermore indicate that the degree of growth activity by the derivatives depends on their biochemical properties rather than the protein amount.
CrPHOT and Dnf1p/Dnf2p P4-ATPases are localized in similar subcellular compartments. Fpk1p/Fpk2p, which regulate P4-ATPase activity, are mainly distributed throughout the cytoplasm but are partially localized in the early endosome/TGN and plasma membrane 25,32,39 . To clarify the control of P4-ATPases by CrPHOT, we compared the intracellular localization of fluorescent protein-tagged CrPHOT with that of Fpk1p. Before analysis, we confirmed that the tagged proteins were functional by a growth assay of the P GAL1 -CDC50 fpk1Δ fpk2Δ mutant ( Supplementary Fig. S1). An fpk1Δ fpk2Δ mutant expressing GFP-Fpk1p or GFP-CrPHOT cultured in darkness was briefly stained with a lipophilic dye, FM4-64 32,40 , which was employed as a marker for the plasma membrane and endosomal/TGN compartments (Fig. 2a). GFP-CrPHOT was occasionally observed at the plasma membrane but primarily localized to intracellular punctate structures, which mostly merged with the fluorescence of FM4-64 (64.4% of the GFP-CrPHOT speckles were merged, n = 163; Fig. 2a  www.nature.com/scientificreports/ The localization was similar to that of GFP-Fpk1p (66.0% of the GFP-Fpk1p speckles were merged, n = 152; 67.5% in previous analysis 32 ; Fig. 2a). The speckles of GFP-CrPHOT, as with GFP-Fpk1p, were colocalized with another TGN marker, Sec7p-mRFP 32 ( Supplementary Fig. S2). These results suggest that GFP-CrPHOT is primarily localized to endosomal/TGN compartments in yeast cells. We then examined the effect of BL on the subcellular localization of GFP-CrPHOT. This was because BL irradiation changes phototropin intracellular localization in plants 41,42 . The localization analysis by FM4-64 staining in yeast cells was performed under BL irradiation. The results showed that the localization of GFP-CrPHOT under BL was the same as that in the dark and that most of its speckles merged with FM4-64 (62.5%, n = 120; Fig. 2a). To investigate the effect of kinase activity on CrPHOT localization, we observed the localization of GFP-KDm. The localization pattern of GFP-KDm was the same as that of GFP-CrPHOT regardless of the light condition (68.0% in the dark and 57.9% under BL, n = 166 and n = 164, respectively; Fig. 2a). These results suggest that the localization of CrPHOT in yeast cells is not dependent on either BL or its own kinase activity.
Under normal culture conditions, Dnf1p and Dnf2p are primarily localized to early endosomal/TGN compartments and partially to the plasma membrane of the bud and bud neck 22,23,25,29,33 . Fpk1p is known to colocalize with Dnf1p/Dnf2p at endosomal/TGN compartments and the plasma membrane 32 . We then examined the colocalization of mRFP-CrPHOT and Dnf1p-GFP in yeast cells. The results showed that Dnf1p-GFP was localized to punctate structures distributed throughout the cell, and many but not all of them colocalized with mRFP-CrPHOT (Fig. 2b). The influence of BL or the introduction of mRFP-KDm on the localization of Dnf1p-GFP was hardly observed (Fig. 2b). The same results were found for the localization of Dnf2p-GFP (Supplementary

Optical control of phospholipid translocation by CrPHOT in a kinase activity-dependent manner.
At the plasma membrane, Dnf1p/Dnf2p are involved in flipping phospholipids (mainly PC and PE) and glycosphingolipids 13,21,23,26,28,29,30 . Fpk1p/Fpk2p regulates phospholipid uptake through Dnf1p/Dnf2p across the plasma membrane 32,34,39,43 . Since CrPHOT was able to complement Fpk1p/Fpk2p function in yeast cell growth 37 (Fig. 1b), BL irradiation likely activates CrPHOT to control phospholipid uptake in the same way as Fpk1p/ Fpk2p. We thus examined whether CrPHOT controls the uptake of nitrobenzoxadiazole (NBD)-labelled phospholipids (NBD-PE and NBD-PC) in the fpk1Δ fpk2Δ mutant. NBD-phospholipids taken up by P4-ATPases are transported mainly to the endoplasmic reticulum (ER) in a short time, and strong fluorescence at the ER membrane is observed 30,44 . When NBD-PE was added, fluorescence was observed at the ER membrane in wild-type cells but not in the fpk1Δ fpk2Δ mutant (Fig. 3a, WT + vector and fpk1Δ fpk2Δ + vector). The internalized NBD-PE was quantified by flow cytometry of the cells (Fig. 3b). When the fpk1Δ fpk2Δ mutant harboured empty vector, the amount of NBD-PE was significantly decreased as described previously 32 (53 ± 8% in the dark and 53 ± 8% under BL relative to wild-type levels, Fig. 3b). When the fpk1Δ fpk2Δ mutant harboured CrPHOT plasmids, intracellular fluorescence was hardly observed and showed a low value in the dark ( Fig. 3a; 73 ± 5%, Fig. 3b) but was observed to the same extent as that in the wild type under BL ( Fig. 3a; 99 ± 2%, Fig. 3b). In contrast, no promotion of uptake by CrPHOT under BL was observed in the fpk1Δ fpk2Δ lem3Δ background strain ( Fig. 3a; 31 ± 9% in the dark and 32 ± 7% under BL for NBD-PE, Fig. 3b). This indicates that NBD-PE uptake promoted by CrPHOT depends on Dnf1p/Dnf2p-Lem3p. The same results were obtained when NBD-PC was added (Fig. 3b, Supplementary Fig. S3). These results suggest that P4-ATPase-mediated phospholipid uptake can be photocontrolled via CrPHOT. In addition, when the fpk1Δ fpk2Δ mutant harboured KDm plasmids, intracellular fluorescence was hardly observed even under BL ( Fig. 3a; 50 ± 8% for NBD-PE and 64 ± 13% for NBD-PC under BL, Fig. 3b; Supplementary Fig. S3), suggesting that control by CrPHOT is dependent on its own kinase activity.
These results were consistent with the altered sensitivity of the fpk1Δ fpk2Δ mutant expressing CrPHOT or its derivatives to duramycin (Supplementary Fig. S4) 37 . Duramycin is a peptide toxin that specifically binds to PE in biological membranes 45,46 . The fpk1Δ fpk2Δ mutant is sensitive to duramycin because PE is not enriched in the inner leaflet and is exposed on the outer leaflet of the plasma membrane 32,34,39,43 . CrPHOT suppressed the sensitivity of the fpk1Δ fpk2Δ mutant to duramycin in a BL-dependent manner, and the K-fragment suppressed it regardless of light conditions. In contrast, the fpk1Δ fpk2Δ mutant harbouring KDm showed increased sensitivity to duramycin even under BL ( Supplementary Fig. S4) 37 . Under BL, the resistance activity by LOV1/2 m remained moderate, although it decreased the suppression compared to that of CrPHOT ( Supplementary Fig. S4). These results reconfirm the suggestion of controlling phospholipid flipping by light as described above.
We next investigated the time-course of phospholipids uptake in response to BL. Cells grown in the dark were further incubated for a total of 3 h. During incubation, cells were irradiated with different lengths of BL (0, 0.5, 1, 3 h) towards the end of the period. NBD-PE was added 1 h before the end of incubation time (Fig. 3c). As a result of quantification by flow cytometry, the internalization of NBD-PE reached the same level as that of the wild-type in 1 h BL irradiation (112 ± 7% relative to wild-type levels, Fig. 3c), although the effect of 0.5 h BL was not clear. These data demonstrate that this CrPHOT system results in a superior, light-induced control of flipping activity.
Light regulates both actin depolarization and switching of apical-isotropic growth. In smallbudded cells (at G2/early mitotic phase), cortical actin patches (small assemblage of actin filaments) are polarized at the tip of the bud, and daughter cells exhibit apical growth 47 . In large-budded cells (at a late mitotic phase), the actin patches are randomly distributed, and the cells switch to isotropic growth 47 . Dnf1p/Dnf2p-Lem3p and Fpk1p/Fpk2p are involved in actin depolarization and switching from apical to isotropic growth. Therefore, defects in these genes result in prolonged polarization of actin patches at the tip and extended elongation of the bud even at the late mitotic phase 31,32 . We confirmed this phenotype in the fpk1Δ fpk2Δ mutant. Cells cultured in each light condition were stained with phalloidin-TRITC (tetramethylrhodamine B isothiocyanate peptide) to visualize actin and with DAPI (4′,6-diamidino-2-phenylindole) to confirm the cell cycle stage. As a result, the polarized actin patches and elongated bud shape were most prominent in large-budded cells regardless of light conditions (Fig. 4a, vector), and the ratio of cells with dispersed actin patches was much lower than that in FPK1p-expressing cells (Fig. 4a, vector; 40.8% in the dark and 37.5% under BL, Fig. 4b, vector). If light controls P4-ATPase activity, it will also restore both switching of actin positioning and proper growth direction. We next examined the localization of actin patches and bud morphology in the fpk1Δ fpk2Δ mutant expressing CrPHOT. In the dark, a low number of cells had depolarized actin ( Fig. 4a; 38.5%, Fig. 4b), and the bud morphology remained tapered in large-budded cells. When cells were irradiated with BL, actin patches were distributed throughout the daughter cells, and the buds exhibited a round shape similar to FPK1p-expressing cells ( Fig. 4a; 79.3%, Fig. 4b). These results suggest that CrPHOT is able to control both actin depolarization and switching of cell growth in a BL-dependent manner.
The same results were obtained in the localization analysis of Myo2p-GFP (Fig. 4c, d). Type V myosin Myo2p transports polarity proteins and is mostly localized at the bud tip during bud formation and dispersed throughout the cell at the late mitotic phase 48 . In the lem3Δ or the fpk1Δ fpk2Δ mutant, Myo2p-GFP remains polarized at the bud tip even at the late mitotic phase 31,32 . We thus investigated the photo-regulation of Myo2p-GFP localization by CrPHOT. In the fpk1Δ fpk2Δ mutant expressing CrPHOT, Myo2p-GFP was polarized in the dark, but it was distributed throughout the cells under BL as FPK1 was (Fig. 4c; 23.0% in the dark and 79.6% under BL, ratio of cells with dispersed Myo2p-GFP, Fig. 4d) 49 . mRFP-Snc1p is primarily localized at the plasma membrane of daughter cells during bud formation in partially Cdc50p-depleted fpk2Δ cells (Fig. 5a, FPK1) 32 . A target-SNARE, Tlg1p, is recycled between the TGN and early endosomes, and thus, Tlg1p-GFP was observed in punctate fluorescence, indicating endosome/TGN localization in this cell (Fig. 5a, FPK1) 50 . When the retrieval pathway from early endosomes to the TGN is inhibited, Snc1p and Tlg1p accumulate in abnormal structures in the cell (Fig. 5a, vector) 32,33 . Plasmids encoding CrPHOT or its derivatives were introduced into the Cdc50p-depleted fpk1Δ fpk2Δ mutant expressing mRFP-Snc1p and GFP-Tlg1p and cultured in the dark or under BL. After fixing the cells with formaldehyde, the localization of those marker proteins was observed using a microscope. In CrPHOT-expressing cells, mRFP-Snc1p and GFP-Tlg1p were observed in the abnormal aggregates in the dark; cells in which mRFP-Snc1p was normally localized in the plasma membrane were hardly observed ( Fig. 5a; 13.3%, Fig. 5b, CrPHOT) as in the case of vector (15.0% in the dark, 20.9% under BL). Under BL irradiation, GFP-Tlg1p was localized in the normal punctate structures, and mRFP-Snc1p was localized in the cell periphery ( Fig. 5a; 70.7%, Fig. 5b, CrPHOT), as in the case of the cells expressing Fpk1p (Fig. 5a; 74.5% in the dark, 74.2% under BL, Fig. 5b, FPK1). LOV1/2 m exhibited abnormal aggregation of both marker proteins and failed to restore them to normal localization even under BL irradiation ( Fig. 5a; 15.5% in the dark, 29.2% under BL, Fig. 5b, LOV1/2 m). These results suggest that the endocytic recycling pathway can be photo-controlled by CrPHOT. We then investigated the necessity for kinase activity in the photo-control of endocytic recycling. As a result, both mRFP-Snc1p and GFP-Tlg1p showed normal localization in cells harbouring the K-fragment regardless of light irradiation ( Fig. 5a; 67.1% in the dark and 75.0% under BL, Fig. 5b), but failed for KDm even under BL ( Fig. 5a; 13.9% in the dark and 15.5% under BL, Fig. 5b). These results suggest that the control of endocytic recycling by light requires CrPHOT kinase activity.

Discussion
In this study, we succeeded in photo-controlling phospholipid (such as PC and PE) flipping and biological membrane functions (actin depolarization and endocytic recycling) in yeast. To date, some optogenetic techniques have been established to transiently control the amounts of phosphatidylinositol 4,5-bisphosphate (PI (4,5) P2) and its metabolites in cells using PI-metabolizing enzymes 51 . The tool we proposed is the first technique to optically control the intracellular distribution of non-PI phospholipids.
The basic mechanism of this new tool is that CrPHOT light-controls the activity of select P4-ATPases in S. cerevisiae. The P4-APases localized in the plasma membrane of yeast are mainly Dnf1p/Dnf2p-Lem3p 22,23,24,25 . In the NBD-phospholipid uptake analysis, NBD-PC and NBD-PE added to the medium were taken up into the cell in a light-dependent manner in the presence of CrPHOT, and the lem3Δ mutation impaired this uptake (Fig. 3). These results provide direct evidence that CrPHOT optically regulates the flipping of PE and PC through activation of Dnf1p/Dnf2p-Lem3p in the plasma membrane. Furthermore, we showed that NBD-PE uptake by CrPHOT reached that of the wild-type cells at least within one hour of BL irradiation (Fig. 3c). Therefore, the activation itself of flippase by BL is presumed to occur in a short time. However, the fpk1Δ fpk2Δ mutant harbouring CrPHOT plasmids showed slightly higher NBD-phospholipids uptake even in the dark (Fig. 3b,c). In the case of KDm, the uptake activity was suppressed to the same level as that of empty vector (Fig. 3b), hence the CrPHOT activity within yeast cells is presumed to leak in the dark. Improvements of CrPHOT molecules to strictly inhibit activity in the dark are awaited. Although detailed analysis of photo-reversibility is also needed in the future, we provide the first basis of a tool for optic controlling lipid uptake in this study.
How does CrPHOT regulate P4-ATPases? All P4-ATPase-related responses analysed in this study were dependent on the kinase activity of CrPHOT (Figs. 1, 3, 4, 5, Supplementary Fig. S3 and S4). Yeast Fpk1p phosphorylates Dnf1p/Dnf2p P4-ATPases in vivo and in vitro 32,35,36 . The kinase activity of Fpk1p and phosphorylation of Dnf1p are required for the optimal function of Dnf1p 32,35 . The control of P4-ATPase activity by Fpk1p/Fpk2p and control of P4-ATPase activity by CrPHOT under BL irradiation were remarkably consistent; therefore, we conclude that CrPHOT also regulates Dnf1p/Dnf2p by phosphorylation. Some phosphorylation sites of Dnf1p by Fpk1p have been identified, and the combination of mutations at those sites within Dnf1p reduces PE flipping in yeast cells 35 . The molecular mechanism underlying P4-ATPase regulation by kinase-related phosphorylation is still unknown, but this system that switches the activity of P4-ATPases by light will be quite useful for elucidating it in the future.   www.nature.com/scientificreports/ Next, it is necessary to consider the effect of CrPHOT on factors other than Dnf1p/Dnf2p, e.g., other P4-ATPases. In microscopic observation, localization of CrPHOT to TGN/endosomes was mainly observed (Fig. 2, Supplementary Fig. S2). Drs2p and Dnf3p are P4-ATPases mainly localized to the TGN/endosomes. Furthermore, their function in the endocytic recycling pathway is partially redundant with that of Dnf1p/ Dnf2p 22,23,24,33 . Drs2p and Dnf3p are also phosphorylated by Fpk1p in vitro 32,36 , and Dnf3p is isolated as a phosphorylation substrate for Fpk1p/Fpk2p in vivo 34,35 . However, the necessity of Fpk1p/Fpk2p activity to the functions of Drs2p and Dnf3p is currently unknown. Further analysis is needed to determine whether CrPHOT regulates the functions of other P4-ATPases, such as Drs2p and Dnf3p.
Another is the effects of Fpk1p/Fpk2p on factors other than P4-ATPases. The protein kinases Ypk1p and Akl1p are known as phosphorylation substrates for Fpk1 34,35,36,43 . Target of rapamycin complex 2 (TORC2) serves as a sensor and regulator for plasma membrane status and is involved in actin-cytoskeleton regulation, sphingolipid synthesis, and endocytosis in S. cerevisiae 52,53 . Under plasma membrane stresses, TORC2 phosphorylates and negatively regulates Fpk1p through activation of Ypk1p 34 . Fpk1p activated under normal conditions suppresses the upstream inhibitor Ypk1p by phosphorylation and promotes the endocytic pathway from the plasma membrane through inhibition of Akl1p by phosphorylation, independent of Dnf1p/Dnf2p activation 35,36 .
To investigate the involvement of CrPHOT in this endocytic pathway, CrPHOT or FPK1 was introduced into the P GAL1 -CDC50 fpk1Δ fpk2Δ strain, and endocytosis from the plasma membrane was observed by FM4-64 staining. Unfortunately, the P GAL1 -CDC50 fpk1Δ fpk2Δ cells failed to show any FM4-64 uptake delay even when the corresponding empty vector was introduced, despite experiments at various temperatures (4-24 °C) and depression conditions (data not shown). This observation indicated that the endocytic pathway functioned properly even in the absence of Dnf1p/Dnf2p-Lem3p and Drs2p-Cdc50p in our experimental conditions. Therefore, the involvement of CrPHOT in endocytosis could not be investigated. We would need to investigate whether CrPHOT regulates this pathway, including the phosphorylation of Akl1p and Ypk1p.
RSK3 (belonging to the p90-S6K subfamily) and Ca 2+ -dependent protein kinase C (PKC) are known as kinases that would be involved in phosphorylation of P4-ATPases in mammals. RSK3 has been identified as a functional counterpart of Fpk1p/Fpk2p in yeast screening 43 , but P4-ATPase regulation and phosphorylation in mammalian cells are not understood. PKC controls the endocytosis of the P4-ATPase ATP11C (relates to B-cell maturation, www.nature.com/scientificreports/ erythrocyte shape, anaemia and hyperbilirubinemia), and phosphorylation of the ATP11C C-terminal region is required for endocytosis 54 . Hence, P4-ATPases are likely regulated by kinases in many eukaryotes, and CrPHOT may also function in mammalian cells. Although an analysis of photo-reversibility and improvement in controllability are required, this study is able to propose the basis of a P4-ATPase light control system.

Methods
Media and growth conditions. Yeast  In all analyses, BL and RL were used at an intensity of 10 μmol m −2 s −1 unless otherwise noted. When grown on the agar medium, the yeast cells were irradiated with light vertically from a height of about 10 cm above the plate. When cultured in a liquid medium, the entire test tube was irradiated with light from a distance of about 10 cm from the side of the tube. Escherichia coli strains were cultured in LB medium [Nacalai] containing appropriate antibiotics as needed. The lithium acetate method was used to introduce plasmids into yeast cells 56,57 .
Strains and plasmids. The S. cerevisiae strains used in this study are listed in Table 1. Yeast strains carrying monomeric red fluorescent protein (mRFP)-tagged SNC1 were constructed by integrating linearized pRS306-mRFP-SNC1 into the URA3 locus, followed by a marker change from URA3 to TRP1. Strain carrying SEC7-mRFP was constructed by PCR-based procedures as described 58,59 . E. coli strain DH5α was used for the construction and amplification of plasmids.
The plasmids used in this study are listed in Table 2. pRS416-GFP-CrPHOT, pRS416-mRFP-CrPHOT, pRS416-GFP-KDm and pRS416-mRFP-KDm were constructed as follows. GFP-or mRFP-tagged CrPHOT or CrPHOT(D549N) 37 was constructed by megaprimer PCR-based procedures 60 and cloned into the BamHI/SalI site of pRS416 61 . pRS416-LOV1/2 m was generated using a QuikChange site-directed mutagenesis kit [Agilent Technologies, Santa Clara, CA] with pRS416-LOV2m 37 . The genes inserted into pRS416 were constitutively expressed in a form tagged with two N-terminal tandem repeats of the influenza virus haemagglutinin epitope (2HA) under control of the TPI1 promoter. All regions constructed by PCR-based procedures were verified by DNA sequencing.
Yeast growth assay. The yeast cells were grown at 28 °C in liquid SGA-Ura medium to an A 600 of 0.6-0.8 and diluted with sterile water to an A 600 of 0.1. Ten-fold serial dilutions (10 -1 , 10 -2 , 10 -3 , 10 -4 ) were made in sterile water. A 10-μl aliquot of each of the diluted cell suspensions was then spotted on a plate with SGA-Ura or SDA-Ura medium. The plates were then placed at 28 °C under different light conditions for 3 days. For growth sensitivity to duramycin, the yeast cells grown in liquid SDA-Ura were diluted as above, and 2 μl of each of the diluted cell suspensions were examined on YPDA plates containing 20 μM duramycin [Sigma-Aldrich, St. Louis, MO] at 28 °C.
Immunoblot analysis. Yeast cells were grown to logarithmic phase in YPGA medium at 28 °C under different light conditions. Protein extraction from yeast cells was performed as described 62 Supplementary Fig. S5, which includes an unprocessed original data that was used to prepare Fig. 1d.

Microscopic observations.
Microscopic observations were performed as previously described 32

Internalization of fluorescence-labelled phospholipids into yeast cells. Large unilamellar vesicles
containing NBD-phospholipids were prepared as described 29 . 1-palmitoyl-2-(6-NBD-aminocaproyl)-PE (NBD-PE), 1-palmitoyl-2-(6-NBD-aminocaproyl)-PC (NBD-PC), and dioleoylphosphatidylcholine (DOPC) were obtained from Avanti Polar Lipids (Alabaster, AL). Fluorescently labelled phospholipid internalization experiments were performed as described 29,30 . Briefly, cells were grown to early logarithmic phase in SDA-U medium at 30 °C in the dark. After dilution to 0.35 A600 ml -1 , cells were incubated for 60 min at 30 °C with liposomes containing 40% NBD-phospholipid and 60% DOPC at a final concentration of 20 μM in the dark or under BL. Cells were then suspended in cold SD containing 20 mM sodium azide and 2.5% bovine serum albumin (BSA), incubated for 20 min, and washed with PBS. Flow cytometry of fluorescently labelled cells was performed on a FACSCanto II cytometer [BD]. For investigation of time response, overnight cultures were diluted to 0.2 A600 ml -1 and incubated for a total of 3 h in the indicated condition (dark, BL 0.5 h, BL 1.0 h, or BL 3.0 h), in which