We have characterized the morphology of the yeast endocytic pathway leading from the plasma membrane to the vacuole by following the trafficking of positively charged nanogold in combination with compartment identification using immunolocalization of t-SNARE proteins. The first endocytic compartment, termed the early/recycling endosome, contains the t-SNARE, Tlg1p. The next compartment, the prevacuolar compartment, contains Pep12p. After transport to the prevacuolar compartment, where vacuolar enzymes are seen on their way to the vacuole, endocytic content is delivered to the late endosome and on to the vacuole, both of which are devoid of Pep12p immunolabel. Traffic to the prevacuolar compartment is reduced in strains mutant for the Rab5 homologs, Vps21p, Ypt52p, and Ypt53p and in vps27 mutant cells. On the other hand, traffic to the early recycling endosome is less dependent on Rab5 homologs and does not require Vps27p.
The yeast endocytic pathway has been studied using mainly genetic and biochemical techniques. These studies have revealed requirements for actin and proteins controlling actin dynamics, ubiquitination of endocytic receptors and other unidentified target(s), and specific lipids (1). The precise role of clathrin in the endocytic internalization step is unknown because clathrin mutants only affect this step partially and because deletion of clathrin adaptors and proteins of related function have no effect on this process (2,3). Later steps in the pathway are often affected by mutations in VPS genes, which were isolated due to their effect on vacuolar protein sorting. Analysis of VPS and other genes has revealed requirements for several proteins known to be involved in membrane fusion, such a small GTPases, SNAREs (receptors for the NEM-sensitive fusion protein), and proteins that may be involved in vesicle tethering before fusion (4–6). SNARE molecules are preferentially found on the vesicle (donor) or target membrane and have been termed v- and t-SNAREs, respectively. Different t-SNARE molecules have partially overlapping localizations by immunofluorescence techniques, but they are primarily found in distinct compartments (7). Some t-SNAREs, including Tlg1p, Pep12p, and Sed5p have been localized by immunoelectron microscopy and while some of them do have significant overlap by this method also, most of their labeling is separate (8). However, many of the compartments are not easily distinguishable by their morphology, even though they are labeled with distinct antibodies. Specific lipid requirements have also been seen for this pathway (9).
Traffic within the vps/endocytic pathways is complex because there are several distinct but dynamic compartments, and apparently several different pathways that interconnect them. Very little morphological work has been done to examine the pathway that vacuolar proteins follow during their biogenesis, but some initial studies have been performed on the endocytic pathway.
Two morphological studies have followed the endocytic pathway in yeast. Internalization and delivery of the alpha factor receptor to internal compartments have been followed. These studies focused on the role of cortical actin patches in the internalization step and came to the conclusion that cortical patches are not the sites of endocytic internalization (10). Studies following uptake of positively charged nanogold allowed visualization of several intermediates in the endocytic pathway in yeast. The initial endocytic compartment seems to be a vesicle of approximately 50 nm diameter, which has been proposed to fuse with a compartment of tubular–vesicular nature, termed an early endosome. The nanogold was next found in the late endosome, which is a multivesicular structure that is most often found next to the vacuole. Contents of the multivesicular structure are then delivered to the vacuole, most likely by fusion of the two compartments, because the markers found in the interior of this organelle are delivered to the vacuole (11,12).
As it is difficult to distinguish the early compartments of the endocytic pathway by morphological methods, we decided to combine the visualization of the internalization of positively charged nanogold with immunolocalization of t-SNARE proteins as markers of endocytic organelles. Using this technique we have been able to show that endocytic content is first delivered to a Tlg1p-positive compartment and later delivered to a Pep12p-positive compartment. These two compartments have a similar morphology. We could not detect Pep12p in late endosomes in wild-type cells. We conclude that there are two distinct compartments preceding the late endosome in the endocytic pathway, one Tlg1p-positive and the next, Pep12p-positive. Deletion of the yeast homologs of Rab5, encoding a small GTPase of the Sec4/Rab family involved in early endosome biogenesis, delayed, but only slightly reduced, delivery of endocytic content to the Tlg1p-positive endocytic compartment.
Results and Discussion
In order to determine the order and identity of compartments of the endocytic pathway by electron microscopy in yeast, we needed to develop a method to simultaneously follow a time-course of endocytosis and to colocalize proteins known to be localized within the pathway. As an endocytic marker we used positively charged nanogold, which we colocalized with two t-SNARE proteins, Tlg1p and Pep12p. Tlg1p is localized in endosomal structures, as well as Golgi membranes and vesicles (13). It most likely has functions in endocytosis and recycling of endocytic content. Pep12p has been localized to the prevacuolar compartment and plays a role in both the endocytic and vps pathways, probably at or before the intersection of the two pathways (8,14).
The first prerequisite to perform these colocalization experiments is to be able to distinguish the positively charged nanogold, which is seen after enhancement with silver, from the goat anti-rabbit IgG-conjugated colloidal gold (10 nm) that is used to detect the primary antibodies reacting with the different t-SNAREs. To test whether this was possible, we bound positively charged nanogold to the surface of spheroplasts at 4 °C, and then warmed them to room temperature for 35 min The spheroplasts were fixed, embedded and thin sections were prepared. The sections were either treated with silver to enhance the positively charged nanogold, or immunolabeled with antibodies against Tlg1p or Pep12p, followed by IgG colloidal gold. The positively charged nanogold appeared as fine particles of irregular shape, which were not very electron dense, rendering a grayish tone to the particles (Figure 1A). On the other hand, the colloidal gold particles resulting from the immunolabeling were larger, regular in size, and dense black (Figure 1B,C). There was also a small difference in position of the labels. The colloidal gold labeling often appeared just outside of the membranous compartments, consistent with the localization of SNARE proteins on the outer surface of the organelles. On the other hand, the positively charged nanogold was always seen inside the membranous compartments. Next, we tried to colocalize the positively charged nanogold and the individual t-SNAREs. As can be clearly seen (Figure 1D,E), the positively charged nanogold and colloidal gold particles can be seen in distinct and in overlapping compartments for both t-SNAREs. In fact, the compartments where the positively charged nanogold colocalizes with either of the two t-SNAREs resemble each other strikingly. Previously, it was shown that Tlg1p and Pep12p localization partially overlaps in the same compartment by immunoelectron microscopy (8). These compartments resemble the ones seen here. Therefore, it is possible that the compartments showing double labeling with the different t-SNAREs are identical.
In order to test this we performed a time-course of uptake of positively charged nanogold, fixed the spheroplasts at different times and prepared them for immunolocalization experiments. If the two compartments were identical then one would expect them to be labeled with positively charged nanogold with the same kinetics. At 4 min of incubation, internalized positively charged nanogold can be detected in vesicles and tubular-vesicular compartments. Forty-eight percent of the Tlg1p-positive compartments were labeled with positively charged nanogold (Figure 2, Table 1). On the other hand, colabeling with antibodies against Pep12 was rare (Figure 2, Table 1). These data show that the earliest endocytic tubular-vesicular compartment contains Tlg1p. As Tlg1p has been implicated in recycling of endocytic content (8,15), it is likely that this compartment represents an early/recycling endosome.
Table 1. : Percentage of compartments that have positively charged nanogold
After 8 min, more positively charged nanogold was internalized and it still localized mainly to tubular–vesicular compartments. Despite the similar morphology of the positively charged nanogold-labeled compartments, there must be at least two distinct compartments because the frequency of immunolocalization of Pep12p on these structures increased strongly (Figure 2, Table 1), while the frequency of labeling of Tlg1p structures remained similar to the 4-min time point. Images of whole cells at the 8-min time point provide an idea of the extent and localization of the intracellular labeling for Tlg1p and Pep12p in combination with nanogold (Figure 3).
The number of Tlg1p- and Pep12p-compartments that label with positively charged nanogold increased slightly at 15 min to reach a steady state (Table 1). At later times (35 min), positively charged nanogold was found in late endosomes and vacuoles (Figure 4). It is interesting to note that late endosomes, characterized by their internal membranes, juxtaposition to the vacuole and positively charged nanogold content only at late time points, do not label with either Tlg1p or Pep12p antibodies, distinguishing the late endosomal compartment from the previously described prevacuolar compartment (PVC). The prevacuolar compartment, as defined by the presence of Pep12p, is tubular/vesicular in nature and does not seem to contain internal membranes, whereas the late endosome is more spherical and contains internal membranes (Figure 4). The absence of Pep12p from the late endosome is perhaps somewhat surprising because it has been localized to a large compartment near the vacuole in class E vps mutant cells (16).
Therefore, to confirm our assertion that the multivesicular body near the vacuole is the late endosome and is not enriched in Pep12p, we performed a double immunolabeling experiment on sections from whole yeast cells embedded in LR Gold using antibodies against a vacuolar hydrolase, carboxypeptidase Y, and against Pep12p. The large (10 nm) colloidal gold particles were used to localize Pep12p, which was again found on tubular structures. Some of these tubular structures were also labeled with the small (5 nm) colloidal gold particles used to detect carboxypeptidase Y (Figure 5, upper panel). These data demonstrate a colocalization of carboxypeptidase Y, on its biosynthetic pathway with Pep12p at the ultrastructural level in wild-type cells. Carboxypeptidase Y was also detected in late endosomal structures near the vacuole. However, in this organelle, no Pep12p was detected (Figure 5, lower panel). These results show that Pep12p is not concentrated in late endosomes and confirm the passage of carboxypeptidase Y through the PVC and late endosomes in wild-type cells.
The above results show that there are at least three distinct compartments that precede the vacuole in the yeast endocytic pathway. Two compartments are tubular in nature, but are distinguishable because they label with different kinetics with nanogold. The first compartment is enriched in Tlg1p and the second is enriched in Pep12p. We do not know if some subset of these compartments are also compartments that have been previously shown to label with both Pep12p and Tlg1p. The third compartment is the late endosome, which lacks Tlg1p and Pep12p, but contains internal membranes. The order of passage of endocytic content seems to be from the cell surface to the Tlg1p compartment, to the Pep12p compartment and then to the late endosome. Therefore, we will refer to the Tlg1p-positive, positively charged nanogold positive compartment as the early/recycling endosome, the Pep12p-positive, positively charged nanogold positive compartment as the PVC, and the multivesicular endosome that is Pep12p- and Tlg1p-negative as the late endosome.
As mentioned above, many mutant cells accumulate Pep12p in a class E compartment that resembles the late endosome in its size and position in the cells (16–18) and this compartment accumulates endocytic and vacuolar proteins, yet wild-type late endosomes contain no detectable Pep12p. Therefore, we sought to resolve this apparent contradiction. Even though Pep12p could not be detected on late endosomes in wild-type cells, it could be detected in rcy1 mutant cells, which are defective in recycling of endocytic content (15) (Figure 6). Therefore, it seems that under normal conditions Pep12p remains at early locations of the endocytic pathway, but when the pathway is disturbed by mutation, then Pep12p can be found in the late endosome. It is therefore likely that Pep12p is delivered to late endosomes in wild-type cells, but that it is rapidly recycled from this organelle back to the Golgi or other intermediates in the vps pathway. Rcy1p may play a role in one of these recycling pathways. Pep12 is required for many pathways that deliver content to the PVC, including biosynthetic delivery of vacuolar enzymes, delivery of endocytic content from the cell surface, and retrograde transport from the vacuole (19). In order to ensure that delivery from these diverse pathways into the PVC occurs with the required fidelity mechanisms must exist to tightly localize Pep12p to the PVC. This is what we observe because levels of Pep12p in the late endosome are below detection using our methods.
Next, we analyzed a class E mutant cell, vps4, which has been shown to accumulate a large structure juxtaposed to the vacuole (18). We labeled vps4 mutant cells by internalization of positively charged nanogold for 40 min and found that the nanogold labeling was mainly in tubular membranes or stacks of cisternal membranes, and only rarely in an organelle containing internal membrane structures (Figure 6). This structure was localized near to the vacuole. These data show that even though the compartment that labels with nanogold in vps4 cells is localized next to the vacuole, it does not have the same morphology as a late endosome, but usually looks more like stacks of membranes resembling early/recycling endosomes or PVC membranes. Cisternal membranes have previously been shown to accumulate in this mutant (20). These data emphasize the importance of studying the morphology of the endocytic pathway in wild-type cells and show that the compartments accumulating in some mutants, such as vps4 mutant cells, are not the equivalent of late endosomes, but probably represent an exaggerated Pep12p-positive PVC.
In order to confirm the order of the pathway and to begin to analyze genetic requirements for the individual steps, we investigated the early stages of endocytosis using the same techniques in vps mutants. VPS21 encodes a small GTPase that has high homology and functional similarities to Rab5 (21,22). There are two additional yeast homologs, encoded by YPT52 and YPT53. We utilized a triple mutant cell (vps21Δ ypt52Δ ypt53Δ) that has been shown to secrete carboxypeptidase Y into the medium and to have a transport defect in the endocytic pathway (22). We also tested a vps27 mutant. VPS27 has been proposed to control exit from the prevacuolar compartment, towards both the vacuole and the Golgi compartment and is a class E vps mutant (17,23).
In order to examine the earliest stages of the endocytic pathway, positively charged nanogold was internalized for 5 or 10 min by wild-type, triple mutant, and vps27 spheroplasts. Tlg1p and Pep12p were then immunolocalized on thin sections ( (Figures 7 and 8). Positively charged nanogold was found in the Tlg1p-positive compartments with a delay in the vps21Δ ypt52Δ ypt53Δ strain (Table 2), suggesting that the yeast Rab5 homologs affect the kinetics of delivery of endocytic content to the early/recycling endosome. This is consistent with the GDP/GTP exchange activity on Rab5 found on clathrin-coated vesicles and the proposed similar localization and functions (24). Interestingly, delivery to the Tlg1p-positive compartment does occur later in the time course and a similar percentage of Tlg1p-positive compartments are eventually found in the triple mutant cells. On the other hand, delivery of positively charged nanogold to the vacuole is almost completely blocked, even after 40 min incubation (Figure 9). These data show that the yeast Rab5 homologs are not absolutely required for delivery of endocytic content to the early/recycling endosome. This could be because there are multiple pathways leading from the cell surface to this organelle, or because the Rab5 homologs are not directly involved in this step in yeast. In the latter case, the delay could be caused by an alteration of the early/recycling endosomal compartment itself due to the strong defect in the vacuole biogenesis pathway in the triple mutant.
Table 2. : Percentage of compartments that have positively charged nanogold
vps21 ypt52 ypt53
vps21 ypt52 ypt53
The vps27 mutation did not affect trafficking from the cell surface to the early endosome because positively charged nanogold was found with a similar frequency as in wild-type spheroplasts in early/recycling endosomes after the short labeling period. This is consistent with previous experiments on vps27 mutants because they can internalize the a-factor receptor, Ste3p, and accumulate it in the class E compartment (23). The morphology of the Pep12p-positive organelles was more expanded with a larger lumenal compartment. This aberrant morphology is not surprising given the previously characterized defects in delivery of both endocytic and vacuolar content to this organelle. After 10 min of internalization, colocalization of positively charged nanogold with Tlg1p increased in all strains, but colocalization with Pep12p increased only in the wild-type strain (Table 2). These results show that absence of Vps21p or Vps27p causes a defect in transport from early/recycling endosomes to the PVC. These results also confirm the order of the endocytic pathway, with the Tlg1p-positive early endosome preceding the PVC. They are also consistent with published observations that the rab5 homologs affect traffic through early endosomes (24) and the vps27 mutant affects delivery of endocytic content to the PVC (23).
One possible explanation for the defect in delivery of endocytic content to the early/recycling endosome and the PVC in the mutants could be that the destination compartments themselves are disturbed or mixed. These possibilities are difficult to address thoroughly, but we did probe the integrity of these compartments by examining the extent of colocalization of Tlg1p and Pep12p, compared to wild-type cells, where we have previously shown that Tlg1p and Pep12p are partially localized to the same tubular compartment (8). Pep12p and Tlg1p were localized on thin sections by double immunolabeling in wild-type and mutants cells. In wild-type cells we found that 36% of the compartments that were labeled for Tlg1p were also labeled for Pep12p. This is similar to what we found previously. The extent of colocalization in the triple mutant was 30% and in the vps27 mutant 35%. These numbers are very similar and suggest that there is no major mixing of compartments of the early endocytic/vacuolar pathways in the mutants. Therefore, the most likely explanation for the change in kinetics of delivery of nanogold to these compartments in the mutants is a defect in the delivery pathway in the mutants. More detailed and perhaps biochemical studies would be needed to establish a direct role of these proteins in the endocytic pathway.
Many studies have looked at colocalization of various markers of the vacuolar biosynthetic pathways with Pep12p, Tlg1p and other markers. Our studies can help situate these markers along the endocytic and vacuole biosynthetic pathways, but they also suggest that more caution should be taken in the interpretation of these results. First, even though we can situate Tlg1p and Pep12p on the endocytic and for the latter vacuole biosynthetic pathways, our results also suggest that these two SNAREs are also in other locations. Not all structures labeled with either marker can be labeled with positively charged nanogold. Second, our results also show that localization of a marker near to the vacuole by immunofluorescence does not necessarily imply that it is a late endosome, especially if it is seen in a mutant cell. In fact, here we show two structures with very different morphologies next to the vacuole that are found in the rcy1 and vps4 mutants. Third, mutations in the endocytic and vacuole biosynthetic pathways can lead to a redistribution of markers, changing the interpretation of colocalization data.
The yeast endocytic pathway is complicated because there are apparently multiple routes that can be taken to reach the vacuole. This is evident from the analysis of targeting information on proteins that traverse the pathway (8), but also from consistent findings that mutants affecting the pathway do not usually completely block delivery of endocytic content to the vacuole. Transport to the vacuole is only reduced, but takes place as witnessed often by accumulation of FM4-64 in vacuolar membranes. Our results with the vps21 ypt52 ypt53 mutant could also be interpreted in this manner. The same is true for the vps mutants, in particular the class E mutants, which do not show strong secretion of vacuolar hydrolases (17). Our mapping of the pathway here should help situate other proteins along the route and help us to unravel the multiple pathways from the Golgi and the cell surface to the vacuole.
Materials and Methods
Yeast strains and growth conditions
Wild-type yeast (RH1800, MATa his4 leu2 ura3 bar1–1), vps27 yeast (RH2383 MATa vps27 his4 leu2 ura3 bar1–1), Rab5 homolog defective yeast (BS63 MATα vps21::LYS2 ypt52::URA3 ypt53:: LEU2 his4 leu2 ura3 lys2 bar1–1) (22), rcy1 defective yeast (RH4344 MATa rcy1Δ::kanMx his4 leu2 ura3 lys2 bar1–1) (15), and vps4 yeast (RH2906 MATa vps4::URA3 his4 leu2 lys2 ura3 bar1–1) were grown overnight in YPUAD medium to early logarithmic phase.
Endocytic uptake of positively charged nanogold
Yeast cells were converted to spheroplasts as described (11), then incubated at room temperature with 5 nmol/ml of positively charged nanogold (Nanoprobes, Inc., Yapank, NY, USA). Aliquots were fixed at various time points, overnight at 4 °C, by direct addition of glutaraldehyde (0.2% final concentration) and formaldehyde (3% final concentration). The fixed spheroplasts were washed in 50 mm HEPES pH 7.0/3 mm KCl, incubated in 1% NaIO4, and free aldehyde groups were quenched for 30 min as described (25). Dehydration, infiltration and polymerization in LRGold resin (London Resin Company, Ltd, Berkshire, UK) were done according to the supplier's instructions. Thin sections of about 60 nm were cut and mounted on nickel grids. The positively charged nanogold was enhanced with HQ SilverTM (Nanoprobes, Inc.) as described previously when no immunolabeling followed.
Thin sections of about 60 nm were cut and mounted on nickel grids. The positively charged nanogold was enhanced with HQ SilverTM (Nanoprobes, Inc.) for only 3 min prior to immunolabeling of proteins. Antibodies against Tlg1p and Pep12p were kindly provided by H. Pelham (LMB, MRC, Cambridge). As secondary antibodies, 10 nm goat anti-rabbit IgG-colloidal gold conjugates (BIO Cell, Cardiff, UK) were routinely used. 5 nm IgG conjugates were used to detect carboxypeptidase Y in Figure 5 and for the localization of Pep12p in Figure 6. 10 nm IgG conjugates were used to detect Pep12p in Figure 5. The protein immunolabeling procedure was essentially performed as described (15). Double immunolabeling experiments were performed and quantified as described previously (8). Quantitation of the colocalization of positively charged nanogold with Tlg1p or Pep12p-labeled structures was done as described below on negatives.
An immunolabeled structure was counted when it was labeled with more than one IgG colloidal gold particle. The limits of that structure were usually easy to determine. However, in general, if the space between two structures was less than the diameter of one of the structures it was counted as one structure. As examples of the above criteria, Figure 1(D) has one Tlg1p-positive structure and Figure 1(E) has two Pep12p-positive structures. All three structures had colocalized nanogold (see criteria next). Positively charged nanogold was distinguished from IgG colloidal gold by two criteria. It was less dense and usually smaller in size. The difference in intensity with IgG colloidal gold was almost always detectable, even when the silver-enhanced nanogold was of the same size as the IgG colloidal gold. An example of this is seen in Figure 1(E). The colloidal gold (filled arrows) is regular in shape, large and dense. The nanogold (open arrow, on right) is less regular in shape, but most importantly less dense than the nanogold. When in doubt, the cells containing those structures were not analyzed. For each time point and immunodetection, 33 cell sections containing one or multiple structures were counted. The number of structures containing nanogold and IgG colloidal gold was divided by the total number of structures labeled with IgG gold indicating labeling for a particular antigen and multiplied by 100 to yield the percentage of labeled structures.
We would like to thank R. Lombardi, P. Morsomme, B. Vallèe, and K. Meier for critical reading of the manuscript B. Singer for a yeast strain and J. Holenstein for technical assistance. This work was supported by the University of Basel and a grant from the Swiss National Science Foundation (to HR).