The interactome of the copper transporter ATP7A belongs to a network of neurodevelopmental and neurodegeneration factors

1) Transparency and rigor are key factors at eLife. As such, a compendium of easily accessible and annotated mass spectrometry data from all experiments used for analysis is required to be submitted as supplement. Further, source mass spectrometry (e.g., raw data files, xml, etc.) should be made available in the public domain (see a recent publication on the subject PMID 25721607).

We have now included Supplementary files 1–3 that contain all mass spectrometry data from three independent experiments. Concerning source MS files, we will make them available through http://www.peptideatlas.org/, which is suggested in PMID 25721607. File submission asks for publication PMID or title-journal information. Currently, we are contacting service providers to gain access to these files.

Reviewers would also like the 'interactome' filtered via known contaminants in IPs (see PMID 23921808).

This is a great suggestion that further emphasizes the confidence in our interactome. We tested the ATP7A interactome with a CRAPome dataset that resembles our experimental conditions. The CRAPome dataset is similar to either the BCS or the copper ATP7A interactomes in its number of hits (see revised Figure 2C for Venn diagrams). Importantly, the overlap of our experimentally curated ATP7A interactome with the CRAPome is just 31 proteins, or a 5.7% of the totality of the curated ATP7A interactome. This indicates that our experimental approach to define a curated ATP7A interactome carefully eliminated proteins that spuriously bind to beads-antibody matrices. We attribute this success to the curation of the ATP7A interactome with our dataset generated with ATP7A null cells (revised Figure 1C and Supplementary file 2–3), which in essence is our own ‘CRAPome’.

The 31 proteins overlapping between the CRAPome and the ATP7A interactome may be real interactors. Thus, we decided to leave them in the final ATP7A interactome dataset. Importantly, eliminating these 31 proteins neither affect our bioinformatic analyses nor they compromise our experimental choices for study. We now provide the list of these 33 proteins in Supplementary file 3 for readers to judge. We hope reviewers will agree with our rationale and decision to keep these 31 proteins as part of the ATP7A interactome.

Finally, all raw data and bioinformatic analyses that compose the main figures need to be presented in full as supplemental material and in the statistical analysis section.

We included these data and now they are part of Supplementary files 1–3.

2) All three reviewers felt that the knockout experiments with COG and ATP7A mutation are not sufficient to demonstrate a role for the interaction of the two. In other words, only part of the proposed COG and ATP7A interaction has been studied and the "loop" needs to be closed with additional approaches. Reviewers suggested addressing COG toxicity by delivering copper to flies/cells (e.g., CuGTSM), or using other genetic and biochemical approaches to show the relationship between COG defects and ATP7A loss of function as a primary driver of toxicity.

This is a very important suggestion that we addressed in revised Figures 9 and 10. These data provide genetic evidence of interactions between ATP7A and COG complex subunits in the developing synapse and adult dopaminergic neurons in Drosophila. In particular, revised Figure 10 demonstrates that a genetic interaction between ATP7A and COG is copper-dependent.

We previously demonstrated that feeding copper in the diet of adult flies is lethal in wild type animals of both sexes (PMID: 26199316). We reasoned that ATP7A overexpression in Drosophila dopaminergic neurons should be protective of copper feeding induce lethality because ATP7A overexpression cell-autonomously decreases cellular levels of copper. This effect is due to ATP7A mistargeting to the cell surface as reported by others (see references Lye et al., 2011 and Lin and Goodman, 1994 in revised manuscript). We selected dopaminergic neurons because of their relevance to Parkinson disease, a disease associated by bioinformatics to the ATP7A interactome (revised Figure 3A and Supplementary file 3). We hypothesized that protective phenotypes induced by genetically increasing ATP7A expression in neurons should be modulated by loss-of-function mutations in the COG complex, which decreases ATP7A expression in human cells (revised Figure 6A-B). This model was thoroughly tested and its predictions fulfilled, thus offering a powerful genetic argument for ATP7A and COG participating on the same copper homeostasis pathway.

Related to this, the specificity of the COG complex activity toward copper transporters over other transporters like zinc metal could be strengthened using the tools already developed.

This is an interesting suggestion. The human genome encodes 10 SLC30 type and 14 SLC39 type zinc transporters that have diverse expression patterns and topologies of zinc transport towards or away from the cytoplasm PMID: 24745988. Drosophila has 2 SLC30 and 2 SLC39 zinc transporters PMID: 24063361. Because of this diversity of transporters and zinc transport topologies, it is hard to assess the consequences of COG deficiency on zinc transporters since the selection of candidate(s) for study is so big. Currently, we do not have mammalian or Drosophila tools to assess these transporters. However, our data show that COG genetic defects negligibly affect zinc cellular content (see revised Figure 7B). We would like ask for a waiver for experiments addressing zinc transporters for the reasons above indicated.