Vps8Dp is concentrated at junctions between CV-related membranes
In budding yeast, the HOPS complex concentrates at the junctions between membranes destined for fusion during vacuole maintenance (Wang et al., 2003).To see whether Vps8Dp in Tetrahymena behaves similarly, we took advantage of the observation that many cells exposed to hypoosmotic stress showed fragmented CVC structures. As shown in Figure 5A, such cells contain multiple Dop1p-labeled vacuole-like compartments, concentrated in the cell posterior and undergoing frequent fusion to form larger compartments, in addition to what appears to be the central contractile bladder. Dop1p itself appears to be irregularly distributed, often as puncta, around the periphery of the labeled compartments. Interestingly, the CVC similarly fragments into multiple Dop1p-decorated large vacuole-like compartments under conditions of hyper osmotic stress, though with the difference that no contractile central bladder is visible (Fig. 5B).
Vacuole-like compartments formed under hypoosmotic conditions are also decorated with Vps8Dp (Fig. 5C). The Dop1p-labeled and Vps8Dp-labeled vacuoles appear to be the same structures, since there is strong co-localization of the two proteins at such vacuoles in cells expressing both Dop1p-mNeon and Vps8Dp-mCherry (Fig. S5). However, a notable difference between the Dop1p and Vps8Dp distributions is that in many cases the latter appears strikingly concentrated in regions between adjoining vacuoles, where the membranes come into close contact (Fig. 5C).
The concentration of Vps8Dp-mNeon between adjoining vacuoles/vesicles could be seen even more strikingly in some osmotically-stressed samples in which cells occasionally suffered localized plasma membrane rupture under the pressure from the microscope cover slip. These breaks allowed leakage of the large Vps8Dp-labeled vesicles into the surrounding buffer. In such samples, Vps8Dp could be seen highly concentrated at junctures between the membranes, consistent with a role in tethering and fusion (Fig. 5D). Moreover, in some cases we could capture membrane fusion occurring at Vps8Dp-enriched junctions (Fig. 5D, arrowheads). Interestingly, we also captured episodes in which two of the leaked vesicles shifted closer to one another, and Vps8Dp accumulated rapidly at their meeting site just when they came into contact (Fig. 5E and 5F).
Discussion
Genes encoding CORVET subunits underwent a marked expansion in the lineage leading to Tetrahymena, leading to the six distinct complexes inT. thermophila called 8A-8F based on their unique Vps8 subunit (Klinger et al., 2013, Sparvoli et al., 2018, Sparvoli et al., 2020). These complexes are distinct from one another in their localization and their subunit makeup (Sparvoli et al., 2020). They are accordingly likely to be individually specialized for distinct functions. We confirmed this previously for the 8A complex, which we found to be essential for the formation of lysosome-related organelles called mucocysts (Sparvoli et al., 2018).
In this manuscript we report analysis of the Vps8 subunit of the 8D CORVET. Vps8Dp was previously shown to be strongly associated with the osmoregulatory CVC (Sparvoli et al., 2020, Cheng et al., 2023). Because the gene was found to be essential for cell viability and therefore difficult to analyze via gene knockout (Sparvoli et al., 2018), in this manuscript we instead used a knockdown approach based on induced expression of a gene-specific RNA hairpin (Howard-Till and Yao, 2006). One striking finding was that even relatively minor depletion of Vps8Dp produced cells that were highly sensitive to osmotic shock (Fig. 1). More informatively, following hairpin induction the cells showed time-dependent changes in the morphology and functioning of the CVC, affecting both the spongiome and bladder.
Within the cohorts of knockdown cells in each experiment there was significant phenotypic heterogeneity, making it difficult to rigorously establish the order in which different defects were manifest. Nonetheless, our results suggest that spongiome structure may be more sensitive to VPS8D dosage compared with the bladder structure, since at early timepoints following hairpin induction only the spongiome was reduced in size relative to the wildtype, a response that became more pronounced at longer time points (Fig. 2J). Interestingly, this shrinkage of the spongiome is very similar to the response seen when cells are briefly exposed to hypoosmotic shock (Cheng et al., 2023). The bladder size in contrast did not detectibly change, but contractions grew less frequent (Fig. 2G). Since the bladder fills with water transported from the spongiome, the slowed contraction rate may reflect less efficient filling from a partially disabled water-collecting reticulum. The bladder in knockdown cells at these early time points also showed reduced labeling by Dop1p, with a corresponding increase in diffuse Dop1p signal in the cytoplasm (Fig. 2A and 2B). The mechanistic role of Dop1p is not yet understood, but in its absence cells show only infrequent contraction of an enormously expanded bladder (Cheng et al., 2016, Cheng et al., 2023).
At later time points following hairpin induction, neither the bladder nor spongiome was detectable, indicating that VPS8D is essential for the maintenance of this organelle (Fig. 3). Proteins that in wildtype cells localized strongly to the CVC were instead found in dispersed cytoplasmic puncta. This process may be reversible, based on our observation that some cells in knockdown cultures recover when hairpin expression is de-induced. Studying the mechanisms underlying such recovery may help to shed light on trafficking pathways involved in CVC biosynthesis, which are poorly understood in any lineage and particularly in Ciliates. The disappearance of both bladder and spongiome compartments after prolonged VPS8D knockdown is similar to a phenotype reported in D. discoideum that is linked to a protein in the BEACH family, called LvsA (Gerald et al., 2002). Interestingly, pulldown experiments in the Apicomplexan Toxoplasma gondii suggest physical interactions between CORVET/HOPS and a BEACH-domain containing protein, but in the context of secretory organelle biogenesis (Morlon-Guyot et al., 2018).
The established role of CORVETs, based on studies in yeast and animals, is in tethering and fusion of membrane compartments (Spang, 2016, van der Beek et al., 2019). Our imaging data are consistent with such a role for the Vps8Dp-containing CORVET at the CVC. For our imaging, we exploited conditions in which the CVC bladder, rather than existing as a single large vacuole, is replaced by a set of smaller vacuoles that can still be recognized due to their labeling by Dop1p as well as Vps8Dp. We found that Dop1p has a relatively uniform distribution at the periphery of such vacuoles, while in contrast Vps8Dp is concentrated at sites where neighboring vacuoles come into close contact (Fig. 5). In some cases, these were seen to be sites of subsequent fusion. A similar phenomenon has been reported of specific protein and lipid accumulation at the interface between vacuoles undergoing homotypic fusion in yeast, termed the vertex ring domain, and it includes the HOPS complex (Wang et al., 2003, Fratti et al., 2004). Interestingly, we witnessed the same highly polarized Vps8Dp distribution on small Vps8Dp-decorated vacuoles that were extruded through breaches in the plasma membrane of perforated cells. In one case, we witnessed rapid Vps8Dp redistribution at the perimeters of two initially widely-spaced vacuoles when they came into contact, leading to accumulation at the junction (Fig. 5E). This may be explained if the 8D-CORVET complexes on each surface are bound to integral membrane proteins undergoing rapid 2-dimensional diffusion, but any complex that simultaneously engages with integral membrane proteins on an adjacent membrane diffuses more slowly. Based on analysis of the vertex ring complex in yeast, this partitioning is also likely to involve lipid subdomains (Fratti et al., 2004).
Our images from osmotically-stressed cells are consistent with a role for Vps8Dp in homotypic tethering and fusion at the CVC, and therefore support the paradigm for CORVET and HOPS established in other eukaryotes. However, no Vps8Dp-labeled vertex ring domains are apparent in cells grown under non-stress conditions, and moreover there is no step clearly involving homotypic fusion in the Tetrahymenacontractile cycle (Cheng et al., 2023). In Paramecium, electrophysiological data support a model in which the junctions between the bladder and emanating spongiome arms undergo fission and subsequent fusion with each contractile cycle (Tominaga et al., 1998b). Whether similar fission/fusion cycles occur in Tetrahymena, where the structure of the bladder/spongiome junction appears quite different, is unknown (Elliott and Bak, 1964). The role of Vps8Dp is unlikely to be restricted to either homotypic fusion between bladder-derived vesicles or spongiome-bladder fusion, since Vps8Dp localizes to puncta throughout the spongiome (Cheng et al., 2023). Spongiome tubules show dynamic extension and branching (Cheng et al., 2023), and one possibility is that the Vps8Dp-containing CORVET complex facilitates homotypic fusion between branches in a way that is required to maintain a cohesive reticulum. The retraction of the spongiome at early time points following VPS8D knockdown may be consistent with this scenario.
However, another broad possibility is that Vps8Dp has activities independent of a conventional CORVET complex, as a monomer or a smaller sub-complex, for which examples exist in other eukaryotes for CORVET and HOPS (Asensio et al., 2013, Lorincz et al., 2016). In pulldowns, Vps8Dp is associated with the 5 other expected subunits of CORVET (Sparvoli et al., 2020). However, some details suggest it may be an outlier among theTetrahymena CORVET complexes. First, it is the most biochemically differentiated, including unique subunit variants for Vps8, Vps16, Vps18, and Vps33 (Sparvoli et al., 2020). Secondly, Vps8Dp is roughly 200 amino acids longer than the other Vps8 paralogs in Tetrahymena, with multiple insertions suggesting the potential for additional interactions and activities. Lastly, Vps8Dp in pulldowns appeared by silver staining to be significantly more abundant than the other subunits with which it associates; significantly, no similar imbalance was present for the Vps8 subunits of the other five CORVETs (Sparvoli et al., 2020). Since the affinity tag for these pulldowns was attached to Vps8Dp, the greater relative abundance of Vps8Dp could be explained if that particular CORVET were comparatively unstable. Alternatively or in addition, it could reflect a super-stoichiometric pool of Vps8Dp in cells. On this basis, we speculate that Vps8Dp could potentially be playing distinct roles at the bladder vs. spongiome, as part of different complexes.