Vesicular transport mediates the uptake of cytoplasmic proteins into mitochondria in Drosophila melanogaster

Mitochondrial aging, which results in mitochondrial dysfunction, is strongly linked to many age-related diseases. Aging is associated with mitochondrial enlargement and transport of cytosolic proteins into mitochondria. The underlying homeostatic mechanisms that regulate mitochondrial morphology and function, and their breakdown during aging, remain unclear. Here, we identify a mitochondrial protein trafficking pathway in Drosophila melanogaster involving the mitochondria-associated protein Dosmit. Dosmit induces mitochondrial enlargement and the formation of double-membraned vesicles containing cytosolic protein within mitochondria. The rate of vesicle formation increases with age. Vesicles originate from the outer mitochondrial membrane as observed by tracking Tom20 localization, and the process is mediated by the mitochondria-associated Rab32 protein. Dosmit expression level is closely linked to the rate of ubiquitinated protein aggregation, which are themselves associated with age-related diseases. The mitochondrial protein trafficking route mediated by Dosmit offers a promising target for future age-related mitochondrial disease therapies. Mitochondrial dynamics change during ageing, with larger mitochondria and altered protein import in older animals. Here the authors show that Dosmit protein mediates mitochondrial morphology with Rab32 by inducing double-membraned vesicles that regulate protein trafficking into mitochondria.

relative intergration etc. For Fig. 8 d-g, I would appreciate seeing the split channels of the high mag images to better appreciate the distribution of Rab32 and Dosmit in the different conditions.
Overall, the writing is reasonable but could do with a thorough review to improve the flow and accuracy of the text and data, and the current Abstract and Discussion are rather disjointed and rambling.
Minor: In general, the labels on the immunofluorescence micrographs are hard to read and should be changed. Also, the indexing to the respective data figures could be clearer. And the order of Supplementary data seems to be quite random, e.g. Supp 4 follows Supp 5 in the text. Fig. 1c. these images are too small and should be enlarged (also the corresponding chart). Besides the scale bars are similar in size but one reads 500 µm and the other 500 nm. One must be wrong.
It isn't clear if a HA knock-in was made for Fis1. The text suggests so but there no schematic in Supp. Fig. 1, and the Fig. 2d suggests that a Fis1 antibody was used (and not anti-HA) but none is listed in Methods. There is also a lack of consistency in the Supp figures whether the gene names are capitalised or not. (Follow FlyBase nomenclature).
Lines 186-7: "a Dosmit-EP mutant... suggesting that Dosmit may use Cys-Cys-linked dimer formation". This needs more explanation.  The data showing the presence of these vesicles in Atg8 knockdown should be shown and explained in details, e.g. which gene (Atg8a or b) was targeted? This also leads to a potential issue with the experiment since both genes would need to be targeted in order to block autophagy.
Line 267: "electronically dense" would be better expressed as 'electron-dense'. Lines 267-8: I do not understand the sentence "we found a reduction in electronically dense material within these vesicles compared to control muscle tissue". What are the authors comparing exactly? There are no/few vesicles in the control tissue. Are they trying to compare with cytosol?
Line 315: "Small-membrane vesicles". What is a 'small-membrane'? Do the authors simply mean small, membrane-bound vesicles? If so, they could just say 'small vesicles' (which, by definition, are membrane-bound).
Line 391: "Rab32 was found in abundance in the mitochondria". This claim seems like an overstatement, and should be revised, given the rather feint signal shown.
Lies 388-9. The authors seem to suggest that they did a screen of Rabs but don't show this, which seems unnecessary and/or incomplete.
Lines 408-9: What do the authors mean by "homozygous knockdown mutants". Do they mean simply 'homozygous mutants'? They should also standardise their genetic nomenclature. Here description of genes/transgene/mutants should be Rab32.
Lines 428-39: This section is rather incongruous. First, it disrupts the flow placed here, but moreover, it doesn't specifically address the idea that the authors claim. It is interesting to know that Drp1 overexpression can reverse the Dosmit overexpression mitochondrial enlargement but this in no way implicates mitophagy in the process or phenotype.
Similarly, the next section of Dosmit structure-function disrupts the flow and would be better placed in the initial description of Dosmit overexpression phenotypes.
Lines 451-454: "we found that the N-terminus of Dosmit was also required for the enlargedmitochondria phenotype ( Supplementary Fig. 7d-f), suggesting that iron-sulfur binding may play an important role in regulating mitochondrial dynamics." The first part of this sentence does not justify the second part.
Line 455-6: "our data suggests that Dosmit-induced transport of vesicles into mitochondria which is dependent upon both Rab32 and Drp-1". They haven't shown that vesicle transport is dependent on Drp1, only the enlargement phenotype.
Lines 465-467: "Using an RNAi screen, we showed that the major mitochondrial dynamic components Opa1, Drp1, Marf, and Fis1 are not involved in vesicle formation". This was not shown, only mentioned in the text. Edit.
Lines 479-80: "are involved in subcellular localization". What does this mean?
Line 498: "(MDVs) have been reported to shuttle proteins between the cytosol and mitochondria". Not exactly as stated. They have been shown to transport mitochondrial components to other parts of the cell. This statement implies a reverse transport direction which is not currently claimed.
Reviewer #3 (Remarks to the Author): In this paper, Chen et al. describes the phenotypes of Dosmit (aka cisd2) expression and knockout in Drosophila. Dosmit is an iron-sulfur binding protein and was identified to reduce the age-related size increase of mitochondria when knocked down. Conversely, upon expression of Dosmit, mitochondria contained double-membraned vesicles that appeared to transport cytosolic cargo and were positive for the outer mitochondrial membrane protein TOM20. The mitochondrial localization of Dosmit is dependent on Rab32 and Rab32 knock-out also leads to smaller mitochondria similar to Dosmit knockout. Protein levels of Dosmit and intramitochondrial vesicle formation increase with age of the flies. The authors conclude that Dosmit induced transport of vesicles might represent a therapeutically interesting novel route for mitochondrial protein trafficking and that this pathway might explain agerelated mitochondrial changes.
While the appearance of double-membraned vesicles importing cytosolic components into mitochondria is very interesting and certainly novel, the manuscript seems largely descriptive. It is unclear how Dosmit and Rab32 lead to the formation of intramitochondrial vesicles, what their functions are, if any, and what functional consequences the vesicles have. In addition, it is unclear if the presence of double membrane intramitochondrial vesicles is fly specific or a universal mechanism that also applies to higher organisms. The authors point out that Dosmit knockout has a beneficial effect on mitochondrial morphology and longevity in Drosophila, but in mouse models cisd2 knockout shows quite the opposite effect and recessive mutations in this gene are associated with a severe neurodegenerative disease (Wolfram syndrome) in humans. This leads to the question if the function of Dosmit is conserved in higher organisms. If not this narrows the scope of the presented work and it might be suited better for a more specialized audience. It is ultimately up to the editor to decide whether publication in Nature communications is justified.
Major points: • As stated above, a major issue is that this phenotype seems to be limited to Drosophila • Can the dosmit knockout phentotype(s) be rescued by expression of human cisd1 or 2? • Adequate description of mitochondrial morphology quantification is lacking. The authors should provide a detailed description for each of the assessed parameter and the samples size(s). • Figure 3: Does the increase of Dosmit reflect the "increase in mitochondrial size within muscles as the flies aged" (line 129) and do other mitochondrial proteins increase with age, too? A mitochondrial loading control (ATP5A or Porin) should be added to exclude this possibility. • Supplementary Fig.4: Why is the analysis of longevity only performed in male flies? Is there a statistical assessment? • Figure 6: the depicted model for vesicle formation shows that the inner mitochondrial membrane forms the outer membrane of the vesicles. The authors should support this model experimentally by co-localizing inner mitochondrial markers to the vesicles. • The authors mention that a small scale RNAscreen led to the identification of Dosmit and disqualified other candidates but the data is not shown. • Line 282: The authors state that the uptake of cytosolic components constitutes a novel route of communication between the cytosol and the mitochondria, but it is does not seem that there is any specificity for the uptake nor that there is a release of the vesicular components. As such, it is not clear how this communication works and what it controls. • Line 388: " Rab protein screening (using Rab32 as a target)". Please explain.
• The dependence of Dosmit and Rab32 on each other regarding their mitochondrial localization is intriguing. It would be good to strengthen the data using biochemical methods and interesting to know if Dosmit and Rab32 (directly) interact with each other. • Figure 9: The authors state that the effect of Dosmit expression was partially suppressed in flies coexpressing the dominant negative Rba32 or in rab32 null mutant flies but this is not really the case. To suppress the phenotype would mean that the morphology should go back to normal however there are gross abnormalities visible in Fig. 9i

21
Thank you for your comments, we hope to address them in our detailed   Wild-type flies exhibited mitochondrial enlargement with aging. c-f) One-week-old opa1 s3475 and marf E flies exhibited more fragmented mitochondria than wild-type flies did, but their mitochondrial size had increased by the eight-week-old stage. g, h) One-week-old drp1 T26 exhibited more fused mitochondria than wild-type flies did, and their mitochondrial size had increased by the age of eight weeks old. Scale bars: 10 μm. i) Categorization of mitochondrial size for one-and eight-week-old wild-type, opa1 S3475 ,marf E , and drp1 T26 flies. n=80-100 from 5-7 flies

108
A: Knowing the content of these vesicles would certainly provide clues 109 to understanding their function. However, it has taken researchers 110 decades to understand the content of transport vesicles like exosomes.

112
Atg8, and Ref(2)p, and they exhibit varying capacities for mitochondrial 113 entry via these intramitochondrial vesicles. We agree that there is 114 probably a rule for cargo selection, which would indeed be a good topic

120
There is a report illustrating that mitochondria are involved in

131
In general, the TOM/TIM pathway is responsible for the transport of

138
We also do not yet know whether mitochondrial content is affected 139 if vesicle formation is blocked. We attempted to isolate mitochondrial 140 proteins to conduct a proteomics-based study, in order to compare the  reduced the phenotype. We also showed that Rab32 is associated with 178 mitochondria and mutually dependent on Dosmit for maintenance.

179
To address the question of how Dosmit and Rab32 cooperate to

396
. We also checked the other aging marker-protein ubiquitination- Percentage of wild-type, Dosmit-overexpressing, Dosmit-EP, and Dosmit-2-1 flies categorized as exhibiting low, medium, and high activity. Dosmit-EP and Dosmit-2-1 flies showed significantly more climbing activity than wild-type flies in middle age (three weeks old). n = 100 flies. Statistical test: chi-square test.   Fig. 5c shows only a small   481

502
Another quantification showed that ~70% of gold particles were 503 localized inside intramitochondrial vesicles (Fig. 5f), which means that 504 ~70% of intramitochondrial GFP particles are located within these 505 vesicles. To investigate this intra-vesicle GFP we focused on the 506 mitochondria ( Fig. 5e), but most cytosolic GFP is localized in muscle 507 fibers.

508
Additionally, we created a 3D video of mitochondria in which Dosmit 509 and cytosolic GFP were ectopically coexpressed. In this video, cytosolic

513
Throughout the study we used a section thickness of 0.772 μm. 514 515 516 517 Fig Reviewer#2 Q3 / Fig 5c, d  c) Western blot showing GFP in the mitochondrial fraction of flight-muscle homogenates from flies ectopically expressing Dosmit (Mef2>GFP+Dosmit), but not in the mitochondrial fractions of flight-muscle expressing GFP without ectopic expression of Dosmit (Mef2>GFP) or control flies (wildtype). d) Quantification of band intensity showed that approximately 15% of the cytosolic GFP was internalized in mitochondria when Dosmit (Dsm) was ectopically expressed. (Fig. 7),

570
A: One possible mitochondrial localization marker is Proteinase K.

571
However, very few reports indicate that Proteinase K treatment is 572 successful with Drosophila tissues ref. We tried for several months, but 573 were unable to obtain intact mitochondria after purification and 574 Proteinase K digestion. As an alternative, we used Tom20-HA as an 575 outer membrane marker, and found that both Dosmit and Rab32

593
We have now added the image split into the different channels in Fig.   594

626
We have edited this revised version, and have modified the abstract 627 and discussion.  Supp. Fig. 1, and the Fig. 2d   642

667
A: We have now tested for differences between the groups using the median, 668 which is the number of days taken to reach 50% mortality. The median 669 longevity for da/+ = 41 d; da>Dosmit = 24 d; wild-type = 42 d; and Dosmit-EP 670 = 55 d. This indicates that the median longevity of Dosmit-driven flies was 671 reduced by 41%, and that of the Dosmit-EP line was increased by 31%, 672 relative to the control (Supplementary Fig. 4).

688
A: We only knocked down Atg8a (VDRC109654). However, we 689 confirmed, using a western blot analysis, that the level of lipidation 690 during autophagy decreased following this knockdown (Fig. a, below).

699
A: This has now been changed accordingly.

705
Are they trying to compare with cytosol?

707
A: Electron-dense areas usually indicate the location of membrane 708 compartments. We observed that the content of intramitochondrial 709 vesicles had a lower electron density, similar to that of cytoplasm, than 710 membrane-bound organelles, meaning that these vesicles were unlikely 711 to be organelles. Therefore, we proposed that the vesicles may be

719
A: Thank you for pointing this out. We did mean small, membrane-720 bound vesicles, and have now changed this to "small vesicles".

726
A: Agreed; we have now modified this statement.

728
Q19 Lies 388-9. The authors seem to suggest that they did a 729 screen of Rabs but don't show this, which seems unnecessary 730 and/or incomplete.

734
Rab32 was able to partially suppress the Dosmit-induced mitochondria

776
The structure and function analysis will link to the mammalian 777 counterpart experiment, which was suggested by the comments of  Fig. 16, 17). reduced the phenotype. We also showed that Rab32 is associated with 889 mitochondria and mutually dependent on Dosmit for maintenance.

890
To address the question of how Dosmit and Rab32 cooperate to 891 mediate the formation of intramitochondrial vesicles in more detail, we 892 performed TEM to examine the correlations in previous genetic assay 893 data in detail (line # 494-497) (Fig Reviewer #3 Q1-1 /Sup Fig. 10 l-s )

894
indicates that there is a strong correlation between the activity of Rab32 895 enzyme and Dosmit in connection with intramitochondrial vesicles.

939
The double-membrane intramitochondrial vesicles are driven by the

1130
A: We cannot currently make conclusive statements about the 1131 specificity of the vesicles, but we can say that they are not completely 1132 unspecific. We found that the cytosolic ribosome subunit Rps2 proteins 1133 could be delivered into mitochondria, but Ref(2)p and Atg8a 1134 accumulated next to mitochondria, but not inside vesicles (Fig Reviewer   1135 3 #Q11 / not shown in text). This suggests that there might be selection 1136 rules for cargo uptake.

1137
We do not have enough data to prove whether the vesicular cargo 1138 is released into the mitochondria, but an immunogold EM assay showed 1139 that only a low percentage of GFP gold particles was localized inside 1140 mitochondria but not inside vesicles (Fig. 5e). This indicates that the 1141 cargo of intramitochondrial vesicles might be released into the 1142 mitochondria. However, we need more evidence before we can be 1143 certain about this.

1144
The exact processes remain unclear, but we here provide a novel 1145 perspective on mitochondrial transport and have identified the relevant 1146 components (Dosmit/Rab32). We also speculate that Dosmit is a