The multi-vesicular body (MVB) organelle and the cellular sorting pathway associated with it are very well characterized in budding yeast (Saccharomyces cerevisiae). However, little is known about MVB structure and function in fission yeast (Schizosaccharomyces pombe). In this work, we investigated and characterized the three-dimensional ultrastructure of MVBs in S. pombe using electron tomography. We discovered a positive correlation between MVB size and the number of intralumenal vesicles (ILVs) contained in them. MVBs grew larger with the progression of the cell cycle, hinting at an unknown interaction between the regulation of the two processes. Larger MVBs were also observed in the temperature-sensitive cdc25-22 mutant, which is defective in cell cycle progression into mitosis at the restrictive temperature.

This study offers, to our knowledge, the first electron microscopic observation and tomographic reconstruction of MVBs in S. pombe.

Multi-vesicular bodies (MVBs) are endosomal compartments filled with intraluminal vesicles (ILVs) that form by inward budding of the membrane^[1]. The contents of MVBs can be targeted for degradation, via fusion with a lysosome^[2][3], or exported out of the cell, via fusion with the plasma membrane to release exosomes^[4]. Although most of our knowledge of the MVB pathway and the ESCRT proteins associated with it have come from studies using S. cerevisiae^[5][6], little is known about its equivalent in S. pombe. Previous studies using light microscopy have confirmed that the S. pombe MVB pathway functions with proteins that are homologous to other systems^[7]. Here, we present images of MVB ultrastructure during different cell cycle stages in S. pombe using the 3D reconstruction technique of electron tomography^[8].

Very little is known about the MVB in fission yeast, and no electron microscopic images have been previously published. Here we show why they might have remained elusive to conventional thin-section electron microscopy and reveal their 3D ultrastructure using electron tomography.

We first analyzed wild-type fission yeast cells in interphase. Electron tomography was used to reconstruct approximately 3 to 4 whole-cell volumes, in which we discovered a total of 11 MVBs. Thus, we deduce that MVBs are rare in S. pombe. We then studied wild-type cells during mitosis (anaphase) and observed 8 additional MVBs.

Out of the total 19 MVBs, 16 were found in small clusters of 2 to 4 MVBs located adjacent to each other (as in Panels A and B). The MVB shape was generally ellipsoid.

In contrast to the MVBs of cells in interphase, mitotic MVBs were found to contain significantly more ILVs (7.4 ± 2.8 versus 2.9 ± 1.4; t-test, p <0.005) and were larger (108.8 ± 6.5 versus 82.5 ± 17.6 nm; t-test, p <0.005), as shown in panel C. Some of the MVBs identified were not contained completely within the volume of the reconstruction; therefore, the calculated number of ILVs may be an underestimation of the real value. These organelles have been marked with an asterisk in panel C.

Our results show that there is an increase in the size of MVBs as the cell cycle progresses. This observation is unprecedented and implies that the regulation of the MVB pathway is somehow dependent on the cell cycle. The ESCRT machinery, which is responsible for budding of ILVs, has previously been shown to be involved in the process of cytokinesis^[9]. It is possible that the interaction between the two processes is much more interconnected than it is currently believed.

As the size of MVBs increases, the number of ILVs contained in them also rises (Panel C). A similar result was previously shown in tomographic analysis of MVBs in Arabidopsis thaliana^[10], suggesting that this property is conserved between life kingdoms. This shows that the production of ILVs via inward budding does not reduce the surface area of the endosomal membrane. Hence, we suggest that external membrane is added during MVB formation.

In total, wild-type S. pombe MVBs were between 50 and 120 nm in diameter (average 93.6 ± 19.1 nm), making them smaller than the MVBs found in wild-type S. cerevisiae^[11], which had an average diameter of 150 nm. The convention we adopted for measuring the narrowest diameter of each MVBs (see “Methods”) may affect this result, if the analysis carried out in the other study was performed differently. The ILVs varied from 11 to 38 nm in diameter (average 22.1 ± 5.1 nm; n = 91; Panel D), which is comparable to the 25 nm average diameter of ILVs in wild-type S. cerevisiae^[11].

ILVs in human cells have been previously shown to present morphological variability, consisting in different electron density and in a variable number of vesicles contained in some ILVs themselves (https://doi.org/10.1101/094045). However, we did not observe the same in S. pombe. This suggests that the ILV formation works differently for the two organisms, managing to produce more ILV diversity in human cells.

Two MVBs were also observed in cdc25-22 -mutant mitotic cells, grown at the permissive temperature of 25°C. They were both larger than any MVB observed in the wild type (131 and 155 nm in diameter, respectively) and they contained more ILVs (22 and 14 respectively; Panel C). This might be a direct consequence of the fact that cdc25-22 mutants are approximately 50% larger in size than wild type when grown at 25°C^[12]. A larger cell volume might allow the organelles to grow larger as well. This hypothesis might also explain why MVBs are larger in mitotic cells (Panel C), since it is right before cell division that a cell has the highest volume. Further studies on MVB size and fate during mitosis and cytokinesis would be necessary to explain this observation.

The absence of S. pombe MVBs from the literature can possibly be explained by their small size, which makes them hard to detect in normal thin section electron microscopy (usually 60–80 nm thick sections). To illustrate this, we compared a 1 nm thick slice (Panel E) and a 70 nm thick slice (Panel F) from an electron tomogram, showing how unclear the MVBs would have appeared in a thin section micrograph. The interphase MVBs were mostly found in the cell tip (Panel G).

One big limitation of the technique of electron tomography is small sample size. However, by pooling data from 3 different labs, we believe that the sample size presented in this study is unusually high compared to other tomographic studies, at least for the wild type.

3 different methods for sample preparation were used, one for each data set (wild type in interphase, wild type in mitosis and cdc25-22 mutant). This might be a weakness of our study, since the reproducibility of our data may be affected. On the other hand, the fact that coherent results were presented on the basis of different methods strengthens the results themselves. Furthermore, Ding et al.^[13] compared mitotic spindle organization in plunge frozen samples with high pressure frozen samples and found no difference in morphology. Therefore, we trust that our samples are comparable.

Around 50% of the analyzed MVBs were not fully contained within the reconstructed volume, or they ended between serial electron tomograms. Therefore, our measurements regarding the number of ILVs contained in them may be an underestimation.

Logarithmically growing wild-type or cdc25-22 S. pombe cells were maintained at permissive temperature (30°C and 25°C, respectively) and prepared for electron tomography by high pressure freezing or plunge freezing. 3 different protocols for freeze substitution, plastic infiltration, and image acquisition were used. For details, see Höög et al.^[14] for interphase wild-type MVBs, Ward et al.^[15] for mitotic wild-type MVBs, and Ding et al.^[13] for cdc25-22 cells. Tilt series were collected using a Tecnai TF30 or TF20 using the SerialEM image acquisition software^[16]. All reconstruction and 3D modeling was done using the IMOD software^[17]. The diameter of the MVBs was always measured across its narrowest dimension.

J.L.H. is funded by a VR young investigator grant and the Göran Gustafsson Foundation for Research in Natural Sciences and Medicine.

The authors declare no conflicts of interest.

Not applicable.

No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.