Author response:
The following is the authors’ response to the original reviews.
Response to Reviewer #1 comments:
(1) SY1 aggregation enhances (in terms of number of aggregates) when Sphingolipid biosynthesis is blocked.
a. Line no 132-133: I agree that there is circumstantial evidence that the maturation pathway of SY1 IB is perturbed by knocking down sphingolipid biosynthesis. However, to prove this formally, a time course of IB maturation needs to be reported in the knock-down strains.
Please see Figure 2-figure supplement 1 for the time course of SY1 IB maturation in the knock-down strains. We have added the result to the manuscript, please see lines 129-131on page 5 in the revised version.
b. It will be good to have formal evidence that sphingolipids are indeed downregulated when these genes are downregulated (knocked down).
This issue has been clearly evidenced in previous reports, and we have added the appropriate references in the main text. For example, down-regulation of LCB1 or SPT in yeast decreased sphingolipid levels by Huang et al (https://doi.org/10.1371/journal.pgen.1002493). According to the report from Tafesse FG, et al (https://doi.org/10.1371/journal.ppat.1005188), in mammalian cells in which Sptlc2 was knocked down by CRISPR/Cas9, sphingolipid and glucosylceramide production is almost completely blocked. In addition, the levels of sphingosine, sphingomyelin, and ceramide were significantly lower compared to control cells. Please see lines 143-144 on pages 6 and lines 232-233 on pages 9 in the revised version.
(2) In a normal cell (where sphingolipid biosynthesis is not hampered), the aggregate of SY1 (primarily the Class I aggregate) is localized only on the mitochondrial endomembrane system. These results have been published for other aggregation-prone proteins and are partly explained in the literature. However, their role in the context of maturation is relatively unclear. The authors however provide no strong evidence to show if mitochondria are preferentially involved in any of the stages of IB maturation. Specifically:
a. Line 166-167: It is not clear from Figure 4B that this is indeed the case. Only the large IB seems to colocalize in all three panels (Class I, 2, 3) with Mitotracker. The smaller IBs in 2 and 3 do not show any obvious co-localization. It is also possible that they do co-localize, but it is not clear from the images. I would appreciate it if the authors either provide stronger evidence (better image) or revise this statement. This point is crucial in some claims made later in the manuscript. (pls see comment #5A).
Based on the reviewer's suggestion, we replaced the images in Figure 4B. In addition, we added the 3D reconstruction results of the interrelationship between Class 3 and Mitotracker in Figure 4-figure supplement 1B, to further show their relationship.
(3) The localization is due to the association of SY1 (aggregates) with mitochondrial proteins like Tom70, Tim44 etc. There are some critical points (that can strengthen the manuscript) that are not addressed here. Primarily, the important role of mitochondria in the context of toxicity is neglected. Although the authors have mentioned in the discussion that it was not their main focus, I believe that this is the novel part of the manuscript and this part is potentially a beautiful addition to literature. The questions I found unanswered are:
a. Is the localization completely lost upon deleting these genes? I see only a partial loss in shape/localization. This is not properly explained in the manuscript. The shape of the IB seems to remain intact while the localization is slightly altered. This indicates that even when sphingolipid is present, SY1 localization is dictated by the (lipid-raft embedded) proteins. Interestingly, it shows that even in the absence of mitochondrial localization the shape of the aggregates is not altered in these deletion strains! How do the authors explain this if mitochondrial surface sphingolipids are important for IB maturation? (the primary screen found that sphingolipid biosynthesis promotes the formation of Class I IBs).
We agree that mutation in one mitochondrial binding protein only a partial loss in shape/localization, and we have replaced “association” with “surrounding” in the manuscript. Please see lines 163-166 on page 6 in the revised version. In mutants that interact with SY1, we counted the proportion of Class 3 aggregates formed by SY1 and found an increase in the proportion of SY1 Class 3 aggregates in the deletion mutants compared to controls, partially lost interaction of SY1 with mitochondria has effect on shape of aggregates, as detailed in line 184 on page 7 and Figure 4-figure supplement 1D. We think that SY1 interactions with mitochondrial proteins are important for the localization of SY1 IB in mitochondria, whereas sphingolipids play an important role in facilitating the formation of Class 1 IBs from Class 3 aggregates.
b. What happens to the toxicity when the aggregates are not localized on mitochondria?
We thank the reviewer for the comments, however to investigate this issue, since a single mutant can only partially affect the phenotype, it may be necessary to construct groups of mutants of different genes to observe the effect, which we will further elucidate in our future studies. What we want to show in this work is that SY1 achieves binding to mitochondria by interacting with these mitochondrial proteins.
c. It is important to note that sphingolipids may affect the whole process indirectly by altering pathways involved in protein quality control or UPR. UPR may regulate the maturation of IBs. It is therefore important to test if any of the effects seen could be of direct consequence.
We agree with the reviewer's comments, but there was no significant enrichment for protein quality control or UPR-related pathways in our genome-wide screen, so it is unlikely that sphingolipids indirectly cause maturation of IBs by affecting these two pathways. We addressed this issue in our discussion. Please see lines 325-328 on page 12 in the revised version.
d. In Figure 4D, the authors find SY1 when they pull down Tom70, Tom37 or Tim44. Tim44 is a protein found in the mitochondrial matrix, how do the authors explain that this protein is interacting with a protein outside the mitochondrial outer membrane?
This interaction could be potentially due to that some of the soluble SY1 enter the mitochondrial matrix and interact with Tim44.
e. Is it possible that the authors are immunoprecipitating SY1 since IBs have some amount of unimported mitochondrial proteins in aggregates formed during proteotoxic stress (https://doi.org/10.1073/pnas.2300475120) (Liu et al. 2023).
Our Co-IP experiments were performed in the soluble state supernatant, so mitochondrial proteins in aggregates were not detected.
f. Line 261 (Discussion): Does deletion of Tom70 or one of the anchors increase Class III aggregation and increase toxicity? Without this, it is hard to say if mitochondria are involved in detoxification.
We thank the reviewer for the comments, please see our response to comment 3b.
(4) This fuels the loss of mitochondrial function.
a. Line 218-219: Although the change is significant, the percentage change is very slight. Is this difference enough to be of physiological relevance in mitochondrial function? In our hands, the DCF fluorescence is much more variable.
We agree with the reviewer that there is a small difference (but significant). To which extend such a difference be of physiological relevance in mitochondrial function need to be further investigated.
b. Is SY1-induced loss of mitochondrial function less in knockouts of Tom70 or the other ones found to be important for localizing the SY1 aggregate to mitochondria?
We examined mitochondrial membrane potential (indicated by Rho 123 fluor intensity) in tom70Δ, tom37Δ and control his3Δ strains and found that the knocking out of Tom70 or Tom37 reduced the mitochondrial toxicity caused by SY1 expression. Please see lines 212-214 on page 8 in the revised version, and Figure 5-figure supplement 2.
(5) Mitochondrial function is further abrogated when there is a block in sphingolipid biosynthesis.
a. Myriosin acted like the deletion strains that showed less structured aggregates. There were more aggregates (Class 3) but visually they seemed to be spread apart. The first comment (#2A) on aggregate classes and their interaction with mitochondria may become relevant here.
According to a recent review article (https://doi.org/10.3389/fcell.2023.1302472), sphingolipids are present in the mitochondrial membrane, bind to many mitochondrial proteins and have emerged as key regulators of mitochondrial morphology, distribution and function. Dysregulation of sphingolipid metabolism in mitochondria disrupts many mitochondrial processes, leading to mitochondrial fragmentation, impaired bioenergetics and impaired cellular function. Myriocin treatment, which affects sphingolipid metabolism, causes mitochondria to become more fragmented, which may explain why the aggregates appear visually spread apart. Regarding the interaction with mitochondria, we counted the proportion of SY1 aggregates surrounded by mitochondria after treatment with myriocin, and the results were not significantly different compared to the control. Please see lines 168-169 on page 6 in the revised version, and Figure 4-figure supplement 1C.
(6) A similar phenomenon is conserved in mammalian cell lines.
a. Line 225-226: Did the authors confirm that this was the only alteration in the genome? Or did they complement the phenotype, genetically?
We performed SPTLC2 gene complementation experiments in knockout cell lines and found that SPTLC2 gene complementation was able to reduce the number of cells forming IBs and the percentage of dispersed irregular IBs compared to controls. Please see lines 240-242 on page 9 in the revised version, and Figure 6-figure supplement 2B.
b. Line 241-245: One of the significant phenotypes observed by downregulating sphingolipid biosynthesis in yeast and mammalian cells, was the increase in the number of aggregates. This is not shown in myriocin treatment in mammalian cells. This needs to be shown to the main concordance with the original screen and the data presented with the KO mammalian cell line.
Please see Figure 7-figure supplement 1A for the data on the proportion of cells forming SY1 IBs after myriocin treatment in mammalian cells, and myriocin treatment in mammalian cells was the same as in the KO mammalian cell line.
Minor Comments:
Line 273-275: How is this statement connected to the previous statement? Was it observed that aggregate fusion was advantageous to the cells?
Yes, aggregate/oligomer fusion is advantageous to the cells, and we have modified the previous statement. Please see line 280 on page 10 in the revised version.
Line 293-294: I am not sure I understand this statement.
We have modified this statement. Please see lines 302-303 on page 11 in the revised version.
Line 295-296: But the authors have commented at multiple places that mitochondria detoxify the cell from SY1 aggregates. I find this link fascinating and worth investigating. Most of the current work has some known links in literature (not everything). The mitochondrial connection being the most fascinating one.
We have removed this sentence. We have added a validation experiment for the role of mitochondrial activity in SY1 IB maturation in the revised version.
Line 318: Do the authors mean: The open question is...
Thanks to the reviewer, we have corrected it.
Response to Reviewer #2 comments:
I recommend considering live cell microscopy to analyze whether sphingolipid-dependent formation of SY1 IB takes place at the mitochondrial outer membrane. The IBs could also be produced at other membranes and then transported to the mitochondrial outer membrane for storage.
As shown in Figure 4A, SY1 IB primarily interacts with mitochondria.
I recommend analyzing whether mitochondrial activity is needed for sphingolipid-dependent SY1 IB formation. Are these IBs localized to mitochondrial membrane solely as scaffold or are these organelles needed to provide the energy for driving IB formation in concert with sphingolipids? This point could be addressed with rho0 strains lacking mitochondrial DNA.
We thank the reviewer for this recommendation. We expressed SY1 protein in BY4741 rho0 strain as suggested and found that the maturation and mitochondrial surrounding state of SY1 IB was not affected by mitochondrial activity. Please see lines 185-187 on page 7 in the revised version, and Figure 4-figure supplement 1E and 1F.
The authors should be more precise in the statistical methods used in their study (method, pre-/post-tests, number of replicates...).
We thank the reviewer for the comment and we have provided a more precise description of the statistical methods. Please see lines 531-534 on page 19 and figure legends in the revised version.
