Starvation Triggers the Delivery of the Endoplasmic Reticulum to the Vacuole via Autophagy in Yeast


*Yoshinori Ohsumi,


Autophagy is a survival mechanism necessary for eukaryotic cells to overcome nutritionally challenged environments. When autophagy is triggered, cells degrade nonselectively engulfed cytosolic proteins and free ribosomes that are evenly distributed throughout the cytoplasm. The resulting pool of free amino acids is used to sustain processes crucial for survival. Here we characterize an autophagic degradation of the endoplasmic reticulum (ER) under starvation conditions in addition to cytosolic protein degradation. Golgi membrane protein was not engulfed by the autophagosome under the same conditions, indicating that the uptake of ER by autophagosome was the specific event. Although the ER exists in a network structure that is mutually connected and resides predominantly around the nucleus and beneath the plasma membrane, most of autophagosome engulfed ER. The extent of the ER uptake by autophagy was nearly identical to that of the soluble cytosolic proteins. This phenomenon was explained by the appearance of fragmented ER membrane structures in almost all autophagosomes. Furthermore, ER dynamism is required for this process: ER uptake by autophagosomes occurs in an actin-dependent manner.

Unicellular organisms such as yeast cells in nature are constantly challenged by changing external environments which force the cells to adapt their internal systems in order to survive. Nutritional starvation is the most common stress that yeast cells face throughout their lifetime. They are equipped with a multiplicity of mechanisms to overcome such stress by reorganizing their internal system (1,2). A well-studied aspect of such an adjustment is the tight regulation on expressing a specific group of genes. Genes required for protein synthesis and uptake of low affinity to specific amino acids are repressed under such conditions. Such tight regulation can be accomplished by adjusting transcription, down shifting translation and repressing ribosome biosynthesis (3).

Degradation of proteins is another important cell response against starvation stress. The vacuole is a major site for this degradation process. Expression of several groups of genes utilized in catabolism such as hydrolases is induced under such conditions (4). A frequently studied example of regulation by degradation is seen on a set of plasma membrane transporter proteins, tryptophan permease (Tat2p). Tat2p is endocytosed specifically under starvation conditions and degraded in the vacuole (5). In addition to regulated degradation of a specific group of proteins, bulk degradation of cytoplasmic constituents, called autophagy, takes place under starvation conditions. Autophagy is the process that nonselectively engulfs cytoplasm by forming a double membrane structure called autophagosome, which fuses with the vacuole for degradation (6). When autophagy is induced, 1–2% of cytoplasmic contents are degraded hourly and the resulting degradation products, amino acids, are reused to synthesize new proteins necessary for such conditions. Thus, autophagy largely contributes toward reorganizing cellular compositions to adapt to such stress.

In the course of studying the mechanistic aspects of autophagy, we recently found that an active flow of COPII vesicular traffic from the ER is required for autophagosome formation (7). Many studies have followed the changes in expression levels of proteins during the starvation, but what has not been studied is how the protein transport pathway, like the early secretory pathway from the ER, the protein production site of cell surface and secretory proteins, is affected. Here we characterize ER and Golgi resident proteins to see if and how the starvation stress modulates the early secretory pathway. Visualizing these proteins revealed dynamic changes in the early secretory pathway under starvation conditions. Significant amounts of both ER and Golgi proteins were transported to the vacuole during starvation, and ER proteins were degraded by autophagy.


An ER lumen marker protein appeared in the vacuole during starvation

We studied the behavior of fluorescent ER marker proteins under starvation conditions. GFP-HDEL and DsRed-HDEL possessed a signal sequence at the N-termini, and possessed an HDEL sequence, an ER retrieval signal found in ER resident proteins, at the C-terminus (8). Under growing conditions, GFP-HDEL and DsRed-HDEL (not shown) exhibited exclusively the typical yeast ER pattern, staining the nuclear membrane and the peripheral network, as has been reported (Figure 1A) (9). To induce the starvation response, we treated cells with rapamycin, which drives cells into the starvation state by specifically inhibiting Tor kinase, the master regulatory molecule of the growing state (10). In addition to the usual nuclear and peripheral ER localization of the fluorescent proteins, we discovered that these proteins were relatively abundant in the vacuolar lumen when cells were treated with rapamycin (Figure 1B). A similar phenomenon was observed during conditions of nitrogen or carbon starvation, indicating that these results were not rapamycin-specific phenomena (data not shown). We further observed the behavior of DsRed-HDEL in the presence of PMSF under starvation conditions to elucidate the participation of autophagy in this transport. In the presence of PMSF during starvation, cytoplasmic constituents surrounded by membrane structures are no longer degraded within the vacuole, and thus accumulate as visible dot structures called autophagic bodies (11). In PMSF-treated, starving cells, DsRed-HDEL was found as many dot structures in the vacuole (Figure 1C), as was GFP-Atg8p, a well-established marker protein that stains autophagosome and autophagic bodies (12). Interestingly, most of the autophagic bodies appeared to exhibit the DsRed-HDEL signal. Western blot analysis was performed to see whether processed DsRed could be detected during starvation (Figure 1D). An additional band with faster mobility than DsRed-HDEL accumulated during starvation. Armed with the ability to withstand vacuolar proteases due to its packed conformation, this additional band was likely a degradation product of the DsRed fusion protein (13). This band was not detectable in the presence of PMSF, an inhibitor of vacuolar protease B, indicating that this fusion protein was degraded in the vacuole (Figure 1D). The soluble protein, Pho8Δ60p (cytosolic form of alkaline phosphatase (ALP) (14)), which is commonly used to monitor the nonselective autophagic engulfment, was also observed (Figure 1D). Although slightly less for DsRed-HDEL similar trends were observed in both DsRed-HDEL and ALP.

Figure 1.

DsRed-HDEL is trans- ported to the vacuole in an autophagy-dependent manner with rapamycin treatment. The fluorescence and the Nomarski images of wild-type cells (JK9-3 da) expressing DsRed-HDEL (A) under growing conditions and (B) under rapamycin-treated conditions (10 h). The vacuolar limiting membrane was labeled with FM4-64. The vacuolar lumen (‘vac’) became fluorescent during starvation. C) The fluorescence and the Nomarski images of wild-type cells (JK9-3 da) expressing DsRed-HDEL and GFP-Atg8p treated with rapamycin in the presence of PMSF for 10 h. Punctate structures (arrowhead) are seen within the vacuole. D) Wild-type (BY4741) and Δatg16 (Y742) cells were grown to the early logarithmic phase and treated with rapamycin or rapamycin with PMSF (+ P) for 10 h. Total proteins were extracted and subjected to immunoblot analysis using anti-DsRed and anti-ALP antibodies. In addition to DsRed-HDEL (∼ 30 kDa, arrow), the band at ∼ 25 kDa (asterisk) accumulated in the wild-type but not in the Δatg16 cells. Bar: 2 μm.

We next asked which pathway was responsible for transporting DsRed-HDEL to the vacuole. In mutants defective for the vacuolar protein sorting/multivesicular body pathway (Δvps4 and Δpep12), DsRed-HDEL was transported to the vacuole under starvation conditions just as observed in wild-type cells (Figure 2, left panel and not shown). In contrast, the DsRed signal was not observed in the vacuole in an autophagy-deficient mutant (Δatg16) (Figure 2). In addition, the clipped form of DsRed-HDEL was not detected by Western blot analysis in the Δatg16 mutant (Figure 1D).

Figure 2.

EGFP-Rer1p is transported to the vacuole in an autophagy-independent manner. The fluorescence and the Nomarski images of wild-type (BY4741), Δatg16 (Y742) and Δvps4 (Y5588) cells expressing DsRed-HDEL and EGFP-Rer1p under vegetative growth condition (G) and the starvation condition (+ R) with or without PMSF (+ P) for 10 h. Bar: 2 μm.

ER membrane protein was also observed in the vacuole during autophagy

As GFP-HDEL and DsRed-HDEL are ER lumenal proteins, we next examined the behavior of ER membrane proteins under starvation conditions. Sec71p is an ER integral membrane protein that does not normally escape from the ER and is therefore not found on COPI or COPII vesicles (15). Sec71p-GFP highlighted the typical ER pattern in cells under growing conditions in wild-type cells (Figure 3A) (16). When the starvation response was induced by rapamycin, Sec71p-GFP stained the vacuolar lumen in addition to peripheral and nuclear ER (Figure 3A). Like DsRed-HDEL, Sec71p-GFP stained vacuoles of Δvps4 and Δpep12 mutant cells but was not observed to stain the vacuoles of Δatg16 cells during starvation (not shown). These results imply that not only ER lumenal proteins but also ER membrane proteins were transported to the vacuole. To exclude the possibility that these phenomena were caused by overexpression of artificial proteins, we used DiOC6, a fluorescent lipophilic dye, to stain living nontransformed cells. A high concentration of this dye selectively stains the nuclear and peripheral ER (Figure 3B) (17). Under starvation conditions, accumulated autophagic bodies in the vacuole were stained with this dye, suggesting the presence of ER in the vacuole (Figure 3B). Taken together, these data demonstrate that starvation triggers autophagy-dependent transport of ER lumenal and membrane proteins to the vacuole.

Figure 3.

Markers for the ER membrane are also found in the vacuole in an autophagy-dependent manner. A) Logarithmic cultures of wild-type cells (BY4741) expressing Sec71p-GFP were treated with rapamycin for 10 h (+ R) then observed by Nomarski and fluorescence microscopy. Vacuolar limiting membrane was labeled with FM4-64. B) Wild-type (BY4741) cells in growing condition (G) or cultured in the presence of rapamycin and PMSF for 10 h (+ P) were stained with DiOC6 as described in Materials and Methods and observed by Nomarski and fluorescence microscopy. DiOC6 stained autophagic bodies (arrowhead) accumulated in the vacuole, as well as in the peripheral and the nuclear ER. Bar: 2 μm.

Golgi proteins are transported to the vacuole via the vacuolar sorting pathway but not via autophagy

Next, we extended our analysis beyond ER resident proteins and examining the localization of two marker proteins that cycle between the ER and cis-Golgi via COPI and COPII vesicles: Rer1p and invertase-Wbp1 chimeric protein. Rer1p is a membrane protein that is primarily located on cis-Golgi and functions in the retrieval of ER membrane proteins, such as Sec12p, from the early Golgi compartment back to the ER via COPI vesicles (18). A functional GFP fusion protein, GFP-Rer1p, showed typical Golgi localization in the growing state (Figure 2) (18). However, a significant amount of GFP-Rer1p localized to the vacuole under starvation conditions (Figure 2). In contrast to the ER marker proteins, transport of GFP-Rer1p to the vacuole was not blocked in autophagy-deficient, Δatg16, mutant cells (Figure 2, middle panel). In addition, GFP-Rer1p was detected in the vacuole as a diffused pattern and not on autophagic bodies in the presence of PMSF (Figure 2, bottom). Vacuolar transport of GFP-Rer1p was blocked in the vacuolar protein sorting pathway-deficient Δvps4 and Δpep12 mutant cells (Figure 2 and not shown).

The second marker protein examined was a chimeric protein composed of invertase, a secretory protein, and the C-terminal portion of Wbp1p, an ER type I transmembrane protein which contains a dilysine motif (19,20). This fusion protein is efficiently transported to the cis-Golgi, where it is glycosylated. It is then sent back to the ER by way of the COPI vesicle using its dilysine motif (19,20). Total cell lysates of yeast expressing this reporter protein were treated with endoglycosidase H to remove N-linked oligosaccharide side chains, and subjected to Western blot analysis with an anti-invertase antibody. The ER-Golgi form (∼ 72 kDa) of this fusion protein was identified under growing conditions (Figure 4, lane 1) (19,20). It is known that this chimeric protein localizes to the vacuole if lysine residues within the dilysine motif are replaced with serine. This change leads to loss of retrieval, and the vacuolar form (∼ 62 kDa) can be detected even under growing conditions (Figure 4, lane 3) (20). When cells expressing wild-type invertase-Wbp1p were exposed to nitrogen starvation, the vacuolar form of this chimeric protein was detected (Figure 4, lane 2). The vacuolar form of the invertase-Wbp1p chimeric protein was also detected in autophagy-deficient, Δatg16, mutant cells (Figure 4, lane 5). Taken together, these data indicate that Golgi proteins that actively cycle between the ER and Golgi were also transported to the vacuole during starvation. This transport is conducted via the vacuolar protein sorting pathway. Altogether, starvation induces transport of ER proteins via autophagy and Golgi proteins via the vacuolar protein sorting pathway.

Figure 4.

Invertase/Wbp1p fusion proteins are transported to the vacuole under nitrogen starvation conditions. Wild-type cells (BY4741) and Δatg16 (Y742) cells expressing invertase/Wbp1p fusion protein (denoted as WT or its derivative K- > S mutant) grown to the early log phase (G) were placed in SD(– N) medium for 4 h (– N). Cell lysates were treated overnight with endoH at 37 °C and subjected to immunoblotting analysis using the anti-invertase antibody. ‘ER’ and ‘vac’ indicate the ER/Golgi and the vacuolar form of fusion proteins, respectively.

Electron microscopic studies identified ER fragments within autophagic bodies

The mechanism by which autophagy results in the appearance of ER resident proteins in the vacuole was further examined by immunoelectron microscopy of GFP-HDEL. Under conditions that result in strong staining of the peripheral ER (Figure 5A) and the nuclear ER (Figure 5B), hardly any staining was detected in cytoplasm with anti-GFP antibody. Under such conditions, the GFP-HDEL signal was also detected in the vacuole; furthermore, most of them were found inside of (Figure 5) rather than on the membrane of autophagic bodies. The signals in the autophagic bodies were associated with membranous structures of various sizes, the majority of which were larger than the size of COPI/II vesicles (40–50 nm) (21) (Figure 5). One possible explanation for this phenomenon is that this type of membrane structure exists throughout the cytosol and is nonselectively engulfed during autophagy. In fact, the extent of ER protein uptake into vacuoles via autophagy was similar to that of the soluble cytosolic protein Pho8Δ60p (cytosolic form of ALP) (Figure 1D) (14). To compare the relative amount of GFP-HDEL signal appearing within the cytoplasm and in autophagic bodies, we counted the number of gold particles in each area. The density of gold particles in autophagic bodies was twice as high as the density in the total cell area including the nuclear and peripheral ER (Figure 5C). However, since most gold particles in the cell area were found on the ER, the relative density in autophagic bodies is much higher than the density in the cytosol itself. On the other hand, the densities of gold particles of ADH, an evenly distributed typical cytosolic protein, were identical in the cytoplasm and in autophagic bodies (22). This indicates that the ER marker, GFP-HDEL, was incorporated into autophagic bodies in a more specific manner than non selective cytoplasmic engulfment. To further examine these membrane structures found in the autophagosome and autophagic bodies, we employed TEM with rapid-freezing and freeze-substitution fixation. Serial sections were prepared from carbon starved cells and examined using an electron microscope. Under these conditions, the integrity of membrane structures within autophagic bodies was well preserved even after the fixation process. Analysis of serial sections revealed the existence of membrane structures possessing characteristics of rough ER within autophagic bodies, but these structures were much smaller in size than previously reported yeast ER (Figure 6, arrows). Furthermore, the uptake frequency of ribosome attached small ER fragments by the autophagic process was high. When a complete set of serial sections was examined, almost all autophagic bodies and autophagosomes were found to contain ER fragments (Figure 6, bottom).

Figure 5.

Immunoelectron microscopy of cells expressing GFP-HDEL. A, B) BGY418 cells (JK9-3 da cells expressing GFP-HDEL) were cultured in the presence of rapamycin and PMSF for 14 h. Cells were immunolabeled with anti-GFP antibodies followed by 10 nm colloidal gold-conjugated goat anti-rabbit IgG. Arrow, peripheral ER. Arrowheads, HDEL stained vesicles. AB, autophagic bodies. N, nucleus. V, vacuole. C) The number of gold particles found in the cytoplasm and the autophagic bodies. The number of gold particles detected in the cytoplasm or in the autophagic bodies were counted and divided by their total area. The total cell area excluding the vacuole was used as the cytoplasmic area in this study. All values are mean ± SD (n = 10).

Figure 6.

Autophagic bodies and autophagosomes engulf rough ER. (top and middle) Serial sections were made from BJ926 cells that were incubated in SG medium. Fifteen sections were examined using electron microscopy, and two representative electron microscopic micrographs are shown. Arrows indicate rough ER present within autophagic bodies and autophagosomes. AB, autophagic body. AP, autophagosome. V, vacuole (bottom). Graphic representations were made of the 15 electron microscopic micrographs, and a two-dimensional cumulative image was generated.

Latrunculin A treatment abolishes the uptake of ER by autophagy

We next examined the mechanism by which the autophagosome engulfs the mutually connected mesh-like ER network structure that is located mainly on the nuclear membrane and beneath the plasma membrane. Recent reports have revealed that the ER network in yeast cells exists in a dynamic state and is constantly being rearranged, as is also the case in mammalian cells (23,24). Latrunculin A (Lat-A) is a drug that disrupts actin structures by reducing polymers into monomeric form (25). This drug blocks the dynamics of the ER network while maintaining its overall structure (23). We examined the relationship between the network dynamics of the ER and autophagic engulfment. First, we studied the effects of rapamycin on the structure of the ER network by observing DsRed-HDEL in serial sections. No obvious effect was detected and the network was similar to networks observed during growing conditions (Figure 7A). In these sections, most of the vacuoles were stained by both GFP-Atg8p and DsRed-HDEL in the absence of Lat-A (Figure 7B). Next, we observed the effect of Lat-A treatment, which was done prior to the addition of rapamycin, on cells. Although some fraction of the cells started to lose viability, the peripheral ER network was almost the same as in cells not treated with Lat-A. Only a slight distortion was observed, as seen by the partially diffused pattern of DsRed-HDEL (Figure 7C). Moreover, the vacuole was stained for GFP-Atg8p in the presence of Lat-A, indicating that autophagy can proceed even with disordered actin (Figure 7D). However, vacuoles were never stained by DsRed-HDEL in these cells (Figure 7D). This demonstrates that the autophagosome could no longer engulf the ER in the presence of Lat-A. This result suggests that the dynamics of the ER network lead to transient formation of ER fragments in the cytoplasm and that these fragments can then be engulfed by autophagosomes.

Figure 7.

The effect of the Lat-A treatment on ER uptake by autophagosomes. Wild-type cells (JK9-3 da) expressing both DsRed-HDEL and GFP-Atg8p were treated with rapamycin for 6 h (A, B),c treated first with Lat-A for 30 min then further incubated with rapamycin for 6 h D), and observed by fluorescence microscopy. C) The serial sectioning images of DsRed-HDEL were taken from the periphery of the cell to the central focal plane (0.2 μm = 1 slice) and were deconvolved. Every second image is presented. The Nomarski and the fluorescence images of the same cells. Bar: 2 μm.


In this study, we characterized the fate of the ER in response to starvation, and demonstrated a significant linkage to autophagy. Fluorescence microscopy showed that the lumenal ER marker, GFP-HDEL, was transported to the vacuole via autophagy during starvation. There is a possibility that translocation of ER proteins across the ER membrane might be inhibited under starvation conditions and such mistranslocated ER proteins are nonselectively engulfed by autophagosomes. We have checked the translocation of HDEL-GFP in sec61-1 mutant cells and the processed form but no precursor form was detected under starvation conditions by pulse-chase experiment (not shown). The possibility of mistranslocation can also be dismissed from the results of electron microscopic pictures; the pattern of gold-particles of HDEL-GFP in the cytoplasm was nowhere near the same as the pattern of ADH, a typical cytosolic marker protein (26). The gold particles of GFP-HDEL were not found as an evenly distributed pattern, with most of them being found on the peripheral and nuclear ER (Figure 5). They were also found inside autophagic bodies, and were detected on small-vesiculated structures. In addition, the ER membrane protein, Sec71p-GFP, was also transported to the vacuole via autophagy during starvation (Figure 3). A high concentration of DiOC6 that stains ER membrane, stained autophagic bodies accumulated in the vacuole. Furthermore, TEM pictures clearly showed the existence of rough ER in nearly all of autophagic bodies in carbon-starved cells (Figure 6). Although the total amount of ER delivered to the vacuole via autophagy is small, the frequency of ER in the autophagic bodies was strikingly high. We concluded that ER itself was engulfed and delivered to the vacuole via autophagy. Golgi proteins were also examined under starvation conditions. Although the delivery of some Golgi localized proteins to the vacuole was observed, such as that found with the ER, the process did not depend on autophagy.

Starvation-specific uptake of ER by autophagosome raises two interesting issues. The first is the existence of a fragmented form of the ER in the cytoplasm in addition to the nuclear and peripheral ER. ER in mammalian cells is transiently fragmented during the cell division cycle (27). In contrast, yeast ER is generally thought to maintain its integrity throughout the cell cycle. However, in addition to such generally accepted ER architecture, we have demonstrated the existence of ER fragments within the autophagosome. The simplest explanation for the localization of these ER fragments is that they exist in the cytoplasm and a small portion of that is nonselectively engulfed and transported to the vacuole by autophagy. However, immunoelectron microscopic observation barely detected ER fragments within the cytoplasm (Figure 5). Furthermore, after Lat-A treatment, which affects ER dynamics, the ER fragments were no longer engulfed by autophagosomes (Figure 7). This suggests that ER fragments are generated transiently as a result of the dynamic state of ER: ER is constantly branching and sliding the interconnected tubules of the mesh-like network (24,27), and such transient structures are captured by autophagosome and transported to the vacuole. The second interesting issue is the high frequency of ER in autophagosomes. Although autophagy is known to be a nonselective degradation system of cytoplasm, autophagic bodies scarcely engulfed glycogen granules, which are formed when cells are nitrogen starved and located rather near the peripheral ER within cells (Figure 5A). This leaves open the possibility that ER might be selectively engulfed by autophagosomes. A different and more plausible explanation is that an autophagosome forms near the ER, where transient ER fragments are constantly available, and it therefore nonselectively but highly efficiently engulfs these fragments. Autophagosomes are thought to form at or near a preautophagosomal structure (PAS), known to exist in the vicinity of the vacuole (12,28). Furthermore, our preliminary observations suggest that the PAS is mostly located at the nuclear vacuolar junction (unpublished data). We have recently reported that active flow from the ER is an important factor in forming the autophagosome (7). However, ER dynamism itself seems dispensable for autophagy since it was able to proceed in the presence of Lat-A. The linkage between ER and autophagosome formation highlighted by this and previous studies may provide a key to understanding the mechanism of autophagosome formation.

The ER and Golgi localized protein, invertase-Wbp1p, and GFP-HDEL behave differently in response to starvation. A fraction of GFP-HDEL was transported to the vacuole via autophagy, whereas transport of invertase-Wbp1p was dependent on the vacuolar protein sorting pathway. A previous report demonstrated that more than 90% of invertase-Wbp1p exists in the glycosylated Golgi form (19), meaning that it is efficiently packaged into COPII vesicles and leaves the ER. In contrast, the majority of DsRed-HDEL was localized in the ER and was not transported to the Golgi (Figure 1). A similar phenotype was observed with Kar2p: most of Kar2p remains in the ER and is not packaged into COPII vesicles, regardless of the presence of the HDEL retrieval signal (8). An idea that is becoming generally accepted is that the packaging of cargo proteins into COPII vesicles is an active process (29). Thus, we reasoned that the difference found between the two retrieval signals might be explained by differences in the efficiency of packing into COPII vesicles.

The finding that significant amounts of Rer1p and invertase-Wbp1p were transported to the vacuole upon induction of starvation (Figures 2 and 4) can be explained by regulation within the Golgi stacks. The Golgi apparatus consists of several compartments (cis, medial and trans) and is heavily regulated to maintain its dynamic compartments by a balance between anterograde and retrograde transports (30,31). This balance also determines the proper localization of each Golgi protein. It is possible that the retrieval system within the Golgi complex becomes less efficient during starvation; thus, the flow within the Golgi complex is no longer balanced and Golgi proteins are transported to the vacuole. However, the retrieval system was not completely abolished, as the Golgi-like distribution of Rer1p-GFP still existed even after a long starvation period. An alteration in COPI coats or COPI-related proteins might contribute to these changes as the localization of Rer1p and Wbp1p fusion protein is dependent on COPI vesicles (18,19). The further analysis on the efficiency of retrograde transport under starvation conditions remains an interesting proposition. We visualized Sec21p-GFP (γ-cop), one of the subunits of the COPI coat, but no significant changes in its distribution were found during starvation (not shown).

The changes observed in the secretory pathway could hold the following physiological significance. The delivery of ER and Golgi proteins to the vacuole may contribute to slowing down the secretory pathway by reducing levels of ER and Golgi. Another possibility is that this mechanism allows the cell to degrade and reuse proteins from the secretory pathway in addition to cytosolic proteins to generate the proteins required for survival. The starvation-specific ER and Golgi proteins transported to the vacuole were discovered in this study. Importantly, not all of the ER and Golgi was transported to the vacuole under these conditions. This is reasonable; the cells are only trying to survive starvation and will have to function normally as soon as they are no longer being starved. When yeast cells are placed in nutritionally challenging environments, many cellular functions change, including the cessation of growth and proliferation. It is likely that the transport of proteins to the plasma membrane is also down-regulated, as cells no longer need to expand their surface area. Further studies on the molecular mechanisms involved in these phenomena will elucidate the regulation of the secretory pathway during starvation.

Materials and Methods

Strains, media and conditions of growth

Saccharomyces cerevisiae strains used in this study were constructed using standard yeast genetic methods for gene disruption and transformation (32). Yeast strains constructed in this study were derived from BGY418 (JK9-3 da (MATa leu2–3112 ura3–52 rme1 trp1 his4 GAL + HMLa) TRP1::GFP-HDEL (9)) and JK9-3 da TRP1::DsRed-HDEL, which were the kind gift of Dr. B. S. Glick (University of Chicago, Chicago, IL). MHY205 (JK9-3 da TRP1::DsRed-HDEL atg16::KAN) and MHY207 (JK9-3 da TRP1::DsRed-HDEL vps4::KAN) were constructed by replacing targeted genes with the amplified product of ∼ 1000 bp upstream and downstream of their ORF obtained from Y742 and Y1813 (33). Cells were grown either in YPD medium (1% yeast extract, 2% peptone and 2% glucose) or in SD (0.17% yeast nitrogen base with 2% glucose without amino acid and ammonium sulfate) + CA (0.5% casamino acid) medium containing nutritional supplements. For nitrogen starvation, SD(-N) medium or YPD containing 0.5 μg/mL rapamycin was used (34). Phenylmethylsulfonyl fluoride (PMSF) was used to prevent the degradation of autophagic bodies at a final concentration of 1 mm from a 100 mm stock dissolved in EtOH. Latrunculin-A (WAKO, Osaka, Japan) was used at the final concentration of 100 μm from a 10 mm stock in DMSO and incubated for 30 min.

Plasmids pQR316SEC71-GFP and pSKY5RER1 were a gift from Dr. Nakano (The Institute of Physical and Chemical Research (RIKEN), Wako, Japan) (16). pWWW (invertase-Wbp1p fusion protein) and pWWS (K→S within dilysine motif) were a gift from Dr. N. Nakamura (Kanazawa University, Japan) (20). pRS316 GFP-ATG8 used in this study was described previously (12).

Fluorescence microscopy

Cells expressing GFP and DsRed were incubated in growth or starvation medium and examined under the appropriate conditions. All images were acquired using a DeltaVision microscope (Applied Precision, Issaque, WA) as described previously (12). The sectioning images shown in Figure 4 were deconvolved using DeltaVision software. ER was stained with DiOC6 by adding 1 μL of a 10 μg/mL stock solution in 100% ethanol to 1 mL culture for 2 min as previously described (17).

Immunoelectron microscopy

Cells grown in YPD were transferred to YPD containing 0.5 μg/mL rapamycin and 1 mm PMSF and further incubated for 14 h. Cells were prepared for immunoelectron microscopy as previously described (35). Rabbit polyclonal anti-GFP, the kind gift of Dr. Sakai, was used at a dilution of 1 : 300.

Electron microscopy

Harvested cells were fixed by a freeze-substitution fixation method as described previously (26). Ultrathin sections were examined with a Hitachi H-500H electron microscope at 100 kV.

Immunoblot analysis

Cells grown in the appropriate growth medium were alkali-lysed with 0.2 m NaOH and a 1/100 volume of β-mercaptoethanol, precipitated by trichloroacetic acid, washed with acetone and boiled in SDS sample buffer for SDS-PAGE analysis as previously described (35). Appropriate amounts of prepared lysates were subjected to electrophoresis, transferred to PVDF membrane (Millipore, Billerica, MA) and immunoblotted using the following antibodies. Rabbit polyclonal anti-DsRed (Clontech, Palo Alto, CA) was used at a dilution of 1 : 1000 to detect DsRed-HDEL. Rabbit polyclonal anti-invertase, a gift from Dr. Y. Nakano, was used at a dilution of 1 : 250 to detect invertase-Wbp1p fusion proteins. Rabbit polyclonal anti-CPY and anti-ALP were used at a dilution of 1 : 5000.

Analysis on invertase-Wbp1p fusion protein

Cells expressing pWWW or pWWS grown to an OD600 of 0.8–1.0 in YPD were transferred to SD(-N) and further incubated for 4 h. Two OD600 worth of cells were centrifuged, boiled in 30 μL SDS sample buffer (2% SDS, 62.5 mm Tris-HCl (pH 6.8), 10% glycerol and dithiothreitol for 2 min, disrupted with glass beads, brought to a volume of 0.1 mL by adding SDS sample buffer, boiled again for 2 min and centrifuged at 1000 × g for 2 min. A volume of 30 μL of supernatant was treated overnight with 100 U of EndoHf (New England Biolabs, Beverly, MA) in 50 mm sodium citrate (pH 5.5) at 37 °C. A 4× SDS sample buffer was added, boiled for 5 min and whole samples were analyzed by immunoblotting as described above.


We wish to thank Dr. B. S. Glick (University of Chicago, Chicago, IL) for yeast strains; Dr. A. Nakano (The Institute of Physical and Chemical Research (RIKEN), Wako) and Dr. N. Nakamura (Kanazawa University, Japan) for plasmids, and members of the Ohsumi lab for constructive comments. This work was supported in part by Grants-in-Aids for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan. M. Hamasaki was supported by the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists.