Write a comment

Preprint review 05/10/2026

Selectivity and dynamics of VAP family tether exchange across membrane contact sites – Jongsma et al. – https://doi.org/10.64898/2026.09.16.752049

Summary:

Jongsma et al. investigated the localisation of VAPs (VAPA, VAPB & MOSPD2) by endogenously tagging the proteins with GFP, rather than using ectopically expressed constructs. They demonstrated that individual VAPs localise to specific membrane contact sites (MCSs) between the ER and other organelles, and that this specificity is lost when the proteins are ectopically expressed. Using their endogenously tagged proteins they show that VAPA localises to mitochondria and Golgi MCSs, VAPB localises to mitochondria and peroxisomes, and MOSPD2 localises to endosomes and lipid droplets. Furthermore, they demonstrated that a loss of VAPA and VAPB causes MOSPD2 to now localise to mitochondria instead of endosomes.

Strengths:

It is a well written manuscript that was easy to follow and understand. The manuscript contains clear and concise subheaders. The schematic overviews help summarise the findings, and make for a clear story. The authors have demonstrated the benefits of using endogenously tagged proteins for proteomic studies, rather than relying on ectopic expression.

Major points:

A large part of the discussion is based on binding affinities of the individual VAPs for their substrates. This is in part based on the binding assay in figure 2C. In this binding assay the authors used 3 different FFAT motif containing peptides and the MSP domains of the VAPs. Two out of three peptides chosen were of proteins not shown in the STRING network analysis, and were not pulled down with the VAPs; could the authors comment on their choice of peptides? For these binding assays, only one peptide per organelle was used. If the authors explain the distinct localisation of different VAPs by their binding affinities for organelle specific proteins, they could strengthen their point by repeating the binding assay for a larger array of peptides, including peptides from the pulldown.

In the figure legends of many experiments, it was unclear what was meant by n; is it the sample size used or the number of biological replicates? Since both are important for interpreting the validity of the results, the authors should clearly state how large their sample size is and how many replicates were used.

Minor points:

- Figure S2A has “replace” on the last column of images. Also, this column contains 3 blurry images.

- Abbreviations of protein names are used in images that are not always described in the text. E.g. that endogenous VAPA is eVA and endogenous MOSPD2 is eM2. While it is not hard to figure out from the context, it would aid in readability if either the full name is used, or the abbreviations are mentioned in the figure legend.

- Similarly, STARD3 is labelled as MLN64 in figure 5C & 5D, while it is not mentioned anywhere that it is the same protein.

- The discussion mentions Knorr et al. twice without a reference or if it is a preprint or unpublished. Do the authors refer to: Selective MOSPD2-STARD3 interaction at ER contact sites governs late endosome/lysosome dynamics and cholesterol homeostasis – Knorr et al. (2026)? If it is on bioRxiv it can be put in the reference list.

- Figure 1B shows the western blots of the tagged constructs. Some of these constructs are heterozygous and some homozygous. While this shouldn’t be a problem for studying localisation, the authors could comment on the differences. Also, can the authors comment on the fact that GFP-MOSPD2 in U2OS and GFP-VAPB in MeLJuSo have a more intense band than the untagged version? Especially because the authors emphasise on endogenous protein expression in their work.

- The co-localisation images only show merges. The authors could add single channels in black and white to more clearly show the localisation of the individual VAPs. Also, the authors could use magenta instead of red, as it is more colour-blind friendly.

- For the pulldown with GFP-eVAPs (displayed in figure 2B), it is unclear which cell line is used, it just states “MeLJuSo and/or U2OS cells”, could the authors clarify this? Secondly, have these pulldowns been compared to a pulldown with just a GFP as control? Thirdly, can the authors comment on whether there are differences between the two cell lines?

- Based on the images in figure 4D, it is unclear how the quantifications in figure 4E are so different between the treatments. E.g. siVAPB and siC look very similar, yet they are significantly different in 3 out of 5 criteria. Secondly, figure S4D shows practically no significant differences between treatments and control. Could the authors comment on the large difference between the two cell lines?

- We think the manuscript could benefit from a further discussion of the lack of effect on mitochondria or VAPA/MOSPD2 re-localisation.

- A rescue experiment would strengthen the authors’ claim, e.g. whether MOSPD2 overexpression can rescue the mitochondrial phenotype.

- While it is known that VAPs can form heterodimers, nothing about this was mentioned in the paper, except in the proposed model. We think the ability of VAPs to form heterodimers should be discussed more extensively prior to including it in the model.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.