Syncytin-mediated open-ended membrane tubular connections facilitate the intercellular transfer of cargos including Cas9 protein

Article activity feed

1. 

   Version published to 10.7554/elife.84391 on eLife

   Mar 10, 2023
2. 

   eLife

   Feb 7, 2023

   Author Response

   Reviewer #1 (Public Review):

   In this work, the authors investigate a means of cell communication through physical connections they call membrane tubules (similar or identical to the previously reported nanotubes, which they reference extensively). They show that Cas9 transfer between cells is facilitated by these structures rather than exosomes. A novel contribution is that this transfer is dependent on the pair of particular cell types and that the protein syncytin is required to establish a complete syncytial connection, which they show are open ended using electron microscopy.

   The data is convincing because of the multiple readouts for transfer and the ultrastructural verification of the connection. The results support their conclusions. The implications are obvious, since it represents an avenue of cellular …

   Author Response

   Reviewer #1 (Public Review):

   In this work, the authors investigate a means of cell communication through physical connections they call membrane tubules (similar or identical to the previously reported nanotubes, which they reference extensively). They show that Cas9 transfer between cells is facilitated by these structures rather than exosomes. A novel contribution is that this transfer is dependent on the pair of particular cell types and that the protein syncytin is required to establish a complete syncytial connection, which they show are open ended using electron microscopy.

   The data is convincing because of the multiple readouts for transfer and the ultrastructural verification of the connection. The results support their conclusions. The implications are obvious, since it represents an avenue of cellular communication and modifications. It would be exciting if they could show this occurring in vivo, such as in tissue. The implication of this would be that neighboring cells in a tissue could be entrained over time through transfer of material.

   Thank the reviewer for his/her comments and suggestion. It’s possible that the thick tubular connections found in this study also exist in vivo. A previous study reported that TNT-like structures were found in mouse or human primary tumor cells (PMID: 34494703; PMID: 34795441). Our transfer assays could be adopted to evaluate such transfer in primary cultures and in vivo. We anticipate this for future work.

   Reviewer #2 (Public Review):

   There is a lot of interest in how cells transfer materials (proteins, RNA, organelles) by extracellular vesicles (EV) and tunneling nanotubes (TNTs). Here, Zhang and Schekman developed quantitative assays, based on two different reporters, to measure EV and direct contact-dependent mediated transfer. The first assay is based on transfer of Cas9, which then edits a luciferase gene, whose enzymatic activity is then measured. The second assay is based on a split-GFP system. The experiments on EV trafficking convincingly show that purified exosomes, or any other diffusible agent, are unable to transfer functional Cas9 (either EV-tethered or untethered) and induce significant luciferase activity in acceptor cells. The authors suggest a plausible model by which Cas9 (with the gRNA?) gets "stuck" in such vesicles and is thus unable to enter the nucleus to edit the gene.

   To test alternative pathways of transfer, e.g. by direct cell-cell contact, the authors co-cultured donor and acceptor cells and detect significant luciferase activity. The split GFP assay also showed successful transfer. The authors further characterize this process by biochemical, genetic and imaging approaches. They conclude that a small percentage of cells in the population produce open-ended membrane tubules (which are wider and distinct from TNTs) that can transfer material between cells. This process depends on actin polymerization but not endocytosis or trogocytosis. The process also seems to depend on endogenously expressed Syncytin proteins - fusogens which could be responsible for the membrane fusion leading to the open ends of the tubules.

   The paper provides additional solid evidence to what is already known about the inefficiency of EV-mediated protein transport. Importantly, it provides an interesting new mechanism for contact-dependent transport of cellular material and assigns valuable new information about the possible function of Syncytins. However, the evidence that the proteins and vesicles transfer through the tubules is incomplete and a few more experiments are required. In addition, certain inconsistencies within the paper and with previous literature need to be resolved. Finally, some parts of the text, methods and the figures require re-writing or additional information for clarity.

   Major comments

   1. In Figure 1F, the authors compare the function of exosome-transported SBP-Cas9-GFP vs. transient transfection of SBP-Cas9-GFP. It is not clear if the cells in the transiently transfected culture also express the myc-str-CD63 and were treated with biotin. It is important to determine if CD63-tethering itself affects Cas9 function.

   Thank the reviewer for his comments and suggestions. We now show in Figure 1- figure supplement 1D that CD63-tethering itself does not affect Cas9 function.

   1. The authors do not rule out that TNTs are a mode of transfer in any of their experiments. Their actin polymerization inhibition experiments are also in-line with a TNT role in transfer. This possibility is not discussed in the discussion section.

   Yes, the results in this study do not rule out a role for TNTs in the transfer. At present, we are not aware of conditions that would functionally distinguish transfer mediated by TNTs and thick tubules. We have now included this in the Discussion section.

   1. Issues with the Split GFP assay:

   a) On page 4, line 176, the authors claim that "A mixture of cells before co-culture should not exhibit a GFP signal". However, this result is not presented.

   The results of mixture experiment are included in Figure 2-figure supplement 1D, E.

   b) The authors show in Figure 2C and F that in MBA/HEK co-culture or only HEK293T co-culture, there are dual-labeled, CFP-mCherry, cells. First - what is the % of this sub-population? Second, the authors dismiss this population as cell adhesion (Page 5, line 192) - but in the methods section they claim they gated for single particles (page 17, line 642), supposedly excluding such events. There is a simple way to resolve this - sort these dual labeled cells and visualize under the microscope. Finally - why do the authors think that the GFP halves can transfer but not the mature CFP or mCherry?

   The plot in the Figure 2C and F are displayed in an all-cell mode, not in singlet mode. The percentage of dual-labeled CFP-mCherry in singlet was 0-0.2%. Thus, most of the signal was from doublet, or cell adhesion. We did not claim that the mature CFP or mCherry cannot be transferred. We suggested that the GFP signal of split-GFP recombination may be a more accurate reflection of cytoplasmic transfer between cells. In contrast, mature CFP or mCherry may simply attach to the cell surface but not enter into the other cells.

   c) In the Cas9 experiments - the authors detect an increase in Nluc activity similar in order of magnitude that that of transient transfection with the Cas9 plasmid - suggesting most acceptor cells now express Nluc. However, only 6% of the cells are GFP positive in the split-GFP assay. Can the authors explain why the rate is so low in the split-GFP assay? One possibility (related to item #2 above) is that the split-GFP is transferred by TNTs.

   The Cas9-based Nluc activity assay is more sensitive as it measures an enzyme with a very high turnover number. The split-GFP assay requires a transfer of GFP fragments to produce intact GFP molecules where the signal is not amplified. We think this explains the dramatic increase in a signal once Cas9 is transferred. Our cell sorting results suggest that at least 6% of the receptor cells are transferred in the co-cultures. Of course, nothing in either analysis rules out a role for TNTs in this transfer.

   1. The membrane tubules, the membrane fusion and the transfer process are not well characterized:

   a) The suggested tubules are distinct from TNTs by diameter and (I presume, based on the images) that they are still attached to the surface - whereas TNTs are detached. However, how are these structures different from filopodia except that they (rarely) fuse?

   We used TIRF microscopy and found that the thick tubules are not attached to the surface (not shown). Filopodia are much closer in diameter to TNTs (0.1-0.4 micron). The thick tubules we observe are much thicker (2-4 micron in diameter).

   b) Figure 5E shows that the acceptor cells send out a tubule of its own to meet and fuse. Is this the case in all 8 open-ended tubules that were imaged? Is this structure absent in the closed-ended tubules (e.g. as seen in Figures 6 & 8)?

   Around half of open-ended tubules appeared to emanate from acceptor cells. Likewise, for closed-ended tubules, for example, in Figure 6E where a recipient HEK293T cell projected a short tubule.

   c) The authors suggest a model for transport of the proteins tethered to vesicles (via CD63 tethering). However, the data is incomplete.

   i) They show only a single example of this type of transport, without quantification. How frequent is this event?

   The transport of the proteins tethered to vesicles (via CD63 tethering) were found in all 8 open-ended tubules that we detected in this study.

   ii) Furthermore, the labeling does not conclusively show that these are vesicles and not protein aggregates. Labeling of the vesicle - by dye or protein marker will be useful to determine if these are indeed vesicles, and which type.

   In Figure 4B, the moving punctum in a tubular connection appears to contain SBP-Cas9-GFP, Streptavidin-CD63-mCherry, and the cell surface WGA conjugate that may have been internalized into a donor cell endosome, which indicates that the moving punctum is vesicle type. Nonetheless, in general we cannot distinguish the forms of Cas9 that are transferred and become localized to the nucleus of target cells and we make no claim other than to suggest this possibility that Cas9 may be transferred as an aggregate.

   iii) The data from Figure 2 suggest (if I understand correctly) transfer of the CD63-tethered half-GFP, further strengthening the idea of vesicular transfer. However, the authors also show efficient transfer of untethered Cas9 protein (Figure 2A and other figures). Does this mean that free protein can diffuse through these tubules? The Cas9 has an NLS so the un-tethered versions should be concentrated in the nucleus of donor cells. How, then, do they transfer? The authors do not provide visual evidence for this and I think it is important they would.

   Based on the results using the Cas9-based luciferase assay (His- or SBP-tagged Cas9) (Figure 2A) and split-GFP assay (free GFP1-10) (Figure 2G), we suggest that free protein could be transferred between cells. Our current imaging approach is not designed to quantify protein diffusion. However, we are able to detect from images that Cas9-GFP does not colocalize exclusively with CD63 or concentrate in the nucleus, but also appears in the cytoplasm. These data indicate that both vesicle association and free diffusion may mediate the transfer through tubules. We thank the referee for emphasizing this issue which we will consider for future work to distinguish the transfer types through tubules.

   iv) In Figures 6 & 8, where transfer is diminished, there are still red granules in acceptors cells (representing CD63-mcherry). Does this mean that vesicles do transfer, just not those with Cas9-GFP? Is this background of the imaging? The latter case would suggest that the red granule moving from donor to acceptor cells in figure 4 could also be "background". This matter needs to be resolved.

   There are a few red puncta in the acceptor cell in Figure 6B. Since the acceptor cell is close to and overlapped with other donor cells containing CD63-mCherry, the red signal may, as the reviewer suggests, be from donor cells and not as a result of transfer through tubular connections. However, donor-acceptor cultures of HEK293T where transfer is not observed, little CD63-mCherry signal, for example, in Figure 6a, was seen in acceptor cells, even during several hours of observation (Figure 6- figure supplement video). A minor red signal could arise from exosomes secreted by donor cells that are internalized by acceptor cells. Images of single-culture receptor cells were added in Figure 4- figure supplement 1.

   For Figure 8, we used MDA-MB-231 syncytin-2 knock-down cells containing Fluc:Nluc:mCherry as the receptor cell, thus in these experiments the red signal most likely represents mCherry expressed in the acceptor cells.

   In Figure 4, we observed moving punctum in a tubular connection which contained co-localized green, red, and purple signals, corresponding to SBP-Cas9-GFP, streptavidin-CD63-mCherry, and the WGA conjugate, respectively. The video of punctum transport (Figure 4-figure supplement video) suggests that the red signal is not “background”.

   1. Why do HEK293T do not transfer to HEK293T?

   a) A major inexplicable result is that HEK293T express high levels of both Syncytin proteins (Figure 7 - supp figure 1A) yet ectopic expression of mouse Syncytin increases transfer (Figure 7E). Why would that be? In addition, Fig 3A shows high transfer rates to A549 cells - which express the least amount of Syncytin. The authors suggest in the discussion that Syncytin in HEK293T might not be functional without real evidence.

   We cannot yet explain why the basal level of syncytin expressed in HEK293 cells is insufficient to promote open-ended tubular connections between these cells. It could be that the proteins are not well represented in a processed form at the cell surface. Nonetheless, ectopic expression of mouse syncytin-A in HEK293T produced some increased transfer but less than when syncytin-A is ectopically expressed in MDA-MB-231 cells (up to 4-fold vs. 30-fold change of Nluc/Fluc signal) (Figure 7E). Furthermore, we have added new results which show that apparent furin-processed forms of syncytin-A, -1 and -2 can be detected by cell surface biotinylation in transfected MDA-MB-231 cells (Figure 8-figure supplement 1D). All we demonstrate is that syncytin in the acceptor cell is required for fusion and we make no claim that it is the only protein or lipid at the cell surface in the acceptor cell required for fusion. Clearly, more work is essential to establish the complexity of this fusion reaction.

   For A549 cells, syncytin-1 is highly expressed in A549 cells, thus it is possible that syncytin-1 in A549 plays crucial roles in the process.

   b) In addition - previous publications (e.g. PMID: 35596004; 31735710) show that over expression of syncytin-1 or -2 in HEK293T cells causes massive cell-cell fusion. The authors do not provide images of the cells, to rule out cell-cell fusion in this particular case.

   Overexpression of syncytin-1 or -2 in cells indeed causes massive cell-cell fusion, while overexpression of syncytin-A induced much less cell fusion than syncytin-1, or -2. We have now added new images shown in Figure 8-figure supplement 1A-C to document these observations. It may be that overexpressed human syncytins are better represented in a furin-processed form in both cell types. In contrast, we did not observe donor-acceptor cell fusion at basal levels of expression of syncytin in HEK293T and MDA-MB-231. For example, the Figure 4-figure supplement video shows that tubular structures were seen to form and break during the course of visualization with a tubule fusion event but no cell fusion to form heterokaryons.

   Reviewer #3 (Public Review):

   In this manuscript, Zhang and Schekman investigated the mechanisms underlying intercellular cargo transfer. It has been proposed that cargo transfer between cells could be mediated by exosomes, tunneling nanotubes or thicker tubules. To determine which process is efficient in delivering cargos, the authors developed two quantitative approaches to study cargo transfer between cells. Their reporter assays showed clearly that the transfer of Cas9/gRNA is mediated by cell-cell contact, but not by exosome internalization and fusion. They showed that actin polymerization is required for the intercellular transfer of Cas9/gRNA, the latter of which is observed in the projected membrane tubule connections. The authors visualized the fine structure of the tubular connections by electron microscopy and observed organelles and vesicles in the open-ended tubular structure. The formation of the open-ended tubule connections depends on a plasma membrane fusion process. Moreover, they found that the endogenous trophoblast fusogens, syncytins, are required for the formation of open-ended tubular connections, and that syncytin depletion significantly reduced cargo Cas9 protein transfer.

   Overall, this is a very nice study providing much clarity on the modes of intercellular cargo transfer. Using two quantitative approaches, the authors demonstrated convincingly that exosomes do not mediate efficient transfer via endocytosis, but that the open-ended membrane tubular connections are required for efficient cargo transfer. Furthermore, the authors pinpointed syncytins as the plasma membrane fusogenic proteins involved in this process. Experiments were well designed and conducted, and the conclusions are mostly supported by the data. My specific comments are as follows.

   1. The authors showed that knocking down actin (which isoform?) in both donor and acceptor cells blocked transfer, and more so in the acceptor cells perhaps due to the greater knockdown efficiency in these cells. However, Arp2/3 complex knockdown in donor cells, but not recipient cell, reduced Cas9 transfer. It would be good to clarify whether the latter result suggests that the recipient cells use other actin nucleators rather than Arp2/3 to promote actin polymerization in the cargo transfer process. Are formins involved in the formation of these tubular connections?

   We thank the reviewer for his/her comments and suggestions. Beta-actin was knocked down in this study. We tried a formin inhibitor, SMIFH2 which resulted in a decrease the Cas9 transfer between cells (Figure 3F).

   1. The authors provided convincing evidence to show that the tubular connections are involved in cargo transfer. Intriguingly, in Figure 4-figure supplement video (upper right), protein transfer appeared to occur along a broad cell-cell contact region instead of a single tubular connection. How often does the former scenario occur? Is it possible that transfer can happen as long as cells are contacting each other and making protrusions that can fuse with the target cell?

   In the Figure 4-figure supplement video (upper right), it may be that several membrane tubes from several different donor cells contact at sites close to one another on the recipient cell resulting in the appearance a broad cell-cell contact. This was a rare observation. In our quantification, only 8 connections were open-ended in 120 cell-cell contact junctions. Once open-ended, or plasma membrane fused, cargo transfer is observed.

   1. The requirement of MFSD2A in both donor (HEK293T) and recipient (MDA-MB-231) cells is consistent with a role for syncytin-1 or 2 in both types of cells. Since HEK293T cells contain both syncytins and MFSD2A but cargo transfer does not occur among these cells, does this suggest that syncytins and/or MFSD2A are only trafficked to the HEK293T cell membrane in the presence of MDA-MB-231 cells?

   A proper answer to this question requires the visualization of syncytins and MFSD2A. The commercial syncytin antibodies were inadequate for immunofluorescence. In advance of the more detailed effort required to tag the genes for endogenous syncytin 1 and 2, we performed live cell imaging and surface biotin labeling of cells transiently transfected to express fluorescently-tagged forms of syncytin-1, -2 and -A. We now show that syncytin-A, -1, and -2 partially localize to the plasma membrane or the cell surface of MDA-MB-231 and at points of cell-cell contact. In fact, overexpression of codon-optimized human syncytin-1, and -2 induced dramatic HEK293T cell-cell fusion. However, at basal levels of syncytin expression, HEK293T could not form open-ended tubular connections, which may be because the basal level of syncytins are not well represented in a processed form at the cell surface or their activity is limited by unknown factors.

   As an independent test of cell surface localization, we used surface biotinylation to show that a fraction of the syncytins can be labeled externally (Figure 8-figure supplement 1D). This fraction shows evidence of proteolytic processing consistent with furin cleavage whereas the overwhelming majority of transfected syncytins detected in a blot of lysates suggests that most remain in the unprocessed precursor form, consistent with the punctate and reticular fluorescence images (Figure 8-figure supplement 1A-C).

   We used IF and GFP-tagged MFSD2A and found this protein partially localized to the plasma membrane of HEK293T cells (Figure 9E, F). Given the results reveal that cargos could be transferred among MDA-MB-231 cells (Figure 2G), syncytin and its receptor appear to function in transfer among these cells.

   Read the original source
3. 

   eLife

   Feb 7, 2023

   eLife assessment

   This convincing demonstration of intercellular transfer of material and its subsequent function in the acceptor cell lends important evidence for the membrane tubular systems' role in cell communication. Importantly a fusogenic protein, syncytin, is shown to play a role in providing an open connection between the cytoplasms of both the acceptor and donor cells. The work has implications for how cells can influence each other's functions.

   Read the original source
4. 

   eLife

   Feb 7, 2023

   Reviewer #1 (Public Review):

   In this work, the authors investigate a means of cell communication through physical connections they call membrane tubules (similar or identical to the previously reported nanotubes, which they reference extensively). They show that Cas9 transfer between cells is facilitated by these structures rather than exosomes. A novel contribution is that this transfer is dependent on the pair of particular cell types and that the protein syncytin is required to establish a complete syncytial connection, which they show are open ended using electron microscopy.

   The data is convincing because of the multiple readouts for transfer and the ultrastructural verification of the connection. The results support their conclusions. The implications are obvious, since it represents an avenue of cellular communication and …

   Reviewer #1 (Public Review):

   In this work, the authors investigate a means of cell communication through physical connections they call membrane tubules (similar or identical to the previously reported nanotubes, which they reference extensively). They show that Cas9 transfer between cells is facilitated by these structures rather than exosomes. A novel contribution is that this transfer is dependent on the pair of particular cell types and that the protein syncytin is required to establish a complete syncytial connection, which they show are open ended using electron microscopy.

   The data is convincing because of the multiple readouts for transfer and the ultrastructural verification of the connection. The results support their conclusions. The implications are obvious, since it represents an avenue of cellular communication and modifications. It would be exciting if they could show this occurring in vivo, such as in tissue. The implication of this would be that neighboring cells in a tissue could be entrained over time through transfer of material.

   Read the original source
5. 

   eLife

   Feb 7, 2023

   Reviewer #2 (Public Review):

   There is a lot of interest in how cells transfer materials (proteins, RNA, organelles) by extracellular vesicles (EV) and tunneling nanotubes (TNTs). Here, Zhang and Schekman developed quantitative assays, based on two different reporters, to measure EV and direct contact-dependent mediated transfer. The first assay is based on transfer of Cas9, which then edits a luciferase gene, whose enzymatic activity is then measured. The second assay is based on a split-GFP system. The experiments on EV trafficking convincingly show that purified exosomes, or any other diffusible agent, are unable to transfer functional Cas9 (either EV-tethered or untethered) and induce significant luciferase activity in acceptor cells. The authors suggest a plausible model by which Cas9 (with the gRNA?) gets "stuck" in such vesicles …

   Reviewer #2 (Public Review):

   There is a lot of interest in how cells transfer materials (proteins, RNA, organelles) by extracellular vesicles (EV) and tunneling nanotubes (TNTs). Here, Zhang and Schekman developed quantitative assays, based on two different reporters, to measure EV and direct contact-dependent mediated transfer. The first assay is based on transfer of Cas9, which then edits a luciferase gene, whose enzymatic activity is then measured. The second assay is based on a split-GFP system. The experiments on EV trafficking convincingly show that purified exosomes, or any other diffusible agent, are unable to transfer functional Cas9 (either EV-tethered or untethered) and induce significant luciferase activity in acceptor cells. The authors suggest a plausible model by which Cas9 (with the gRNA?) gets "stuck" in such vesicles and is thus unable to enter the nucleus to edit the gene.

   To test alternative pathways of transfer, e.g. by direct cell-cell contact, the authors co-cultured donor and acceptor cells and detect significant luciferase activity. The split GFP assay also showed successful transfer. The authors further characterize this process by biochemical, genetic and imaging approaches. They conclude that a small percentage of cells in the population produce open-ended membrane tubules (which are wider and distinct from TNTs) that can transfer material between cells. This process depends on actin polymerization but not endocytosis or trogocytosis. The process also seems to depend on endogenously expressed Syncytin proteins - fusogens which could be responsible for the membrane fusion leading to the open ends of the tubules.

   The paper provides additional solid evidence to what is already known about the inefficiency of EV-mediated protein transport. Importantly, it provides an interesting new mechanism for contact-dependent transport of cellular material and assigns valuable new information about the possible function of Syncytins. However, the evidence that the proteins and vesicles transfer through the tubules is incomplete and a few more experiments are required. In addition, certain inconsistencies within the paper and with previous literature need to be resolved. Finally, some parts of the text, methods and the figures require re-writing or additional information for clarity.

   Major comments
   1. In Figure 1F, the authors compare the function of exosome-transported SBP-Cas9-GFP vs. transient transfection of SBP-Cas9-GFP. It is not clear if the cells in the transiently transfected culture also express the myc-str-CD63 and were treated with biotin. It is important to determine if CD63-tethering itself affects Cas9 function.
   2. The authors do not rule out that TNTs are a mode of transfer in any of their experiments. Their actin polymerization inhibition experiments are also in-line with a TNT role in transfer. This possibility is not discussed in the discussion section.
   3. Issues with the Split GFP assay:
   a. On page 4, line 176, the authors claim that "A mixture of cells before co-culture should not exhibit a GFP signal". However, this result is not presented.
   b. The authors show in Figure 2C and F that in MBA/HEK co-culture or only HEK293T co-culture, there are dual-labeled, CFP-mCherry, cells. First - what is the % of this sub-population? Second, the authors dismiss this population as cell adhesion (Page 5, line 192) - but in the methods section they claim they gated for single particles (page 17, line 642), supposedly excluding such events. There is a simple way to resolve this - sort these dual labeled cells and visualize under the microscope. Finally - why do the authors think that the GFP halves can transfer but not the mature CFP or mCherry?
   c. In the Cas9 experiments - the authors detect an increase in Nluc activity similar in order of magnitude that that of transient transfection with the Cas9 plasmid - suggesting most acceptor cells now express Nluc. However, only 6% of the cells are GFP positive in the split-GFP assay. Can the authors explain why the rate is so low in the split-GFP assay? One possibility (related to item #2 above) is that the split-GFP is transferred by TNTs.
   4. The membrane tubules, the membrane fusion and the transfer process are not well characterized:
   a. The suggested tubules are distinct from TNTs by diameter and (I presume, based on the images) that they are still attached to the surface - whereas TNTs are detached. However, how are these structures different from filopodia except that they (rarely) fuse?
   b. Figure 5E shows that the acceptor cells send out a tubule of its own to meet and fuse. Is this the case in all 8 open-ended tubules that were imaged? Is this structure absent in the closed-ended tubules (e.g. as seen in Figures 6 & 8)?
   c. The authors suggest a model for transport of the proteins tethered to vesicles (via CD63 tethering). However, the data is incomplete.
   i. They show only a single example of this type of transport, without quantification. How frequent is this event?
   ii. Furthermore, the labeling does not conclusively show that these are vesicles and not protein aggregates. Labeling of the vesicle - by dye or protein marker will be useful to determine if these are indeed vesicles, and which type.
   iii. The data from Figure 2 suggest (if I understand correctly) transfer of the CD63-tethered half-GFP, further strengthening the idea of vesicular transfer. However, the authors also show efficient transfer of untethered Cas9 protein (Figure 2A and other figures). Does this mean that free protein can diffuse through these tubules? The Cas9 has an NLS so the un-tethered versions should be concentrated in the nucleus of donor cells. How, then, do they transfer? The authors do not provide visual evidence for this and I think it is important they would.
   iv. In Figures 6 & 8, where transfer is diminished, there are still red granules in acceptors cells (representing CD63-mcherry). Does this mean that vesicles do transfer, just not those with Cas9-GFP? Is this background of the imaging? The latter case would suggest that the red granule moving from donor to acceptor cells in figure 4 could also be "background". This matter needs to be resolved.
   5. Why do HEK293T do not transfer to HEK293T?
   a. A major inexplicable result is that HEK293T express high levels of both Syncytin proteins (Figure 7 - supp figure 1A) yet ectopic expression of mouse Syncytin increases transfer (Figure 7E). Why would that be? In addition, Fig 3A shows high transfer rates to A549 cells - which express the least amount of Syncytin. The authors suggest in the discussion that Syncytin in HEK293T might not be functional without real evidence.
   b. In addition - previous publications (e.g. PMID: 35596004; 31735710) show that over expression of syncytin-1 or -2 in HEK293T cells causes massive cell-cell fusion. The authors do not provide images of the cells, to rule out cell-cell fusion in this particular case.

   Read the original source
6. 

   eLife

   Feb 7, 2023

   Reviewer #3 (Public Review):

   In this manuscript, Zhang and Schekman investigated the mechanisms underlying intercellular cargo transfer. It has been proposed that cargo transfer between cells could be mediated by exosomes, tunneling nanotubes or thicker tubules. To determine which process is efficient in delivering cargos, the authors developed two quantitative approaches to study cargo transfer between cells. Their reporter assays showed clearly that the transfer of Cas9/gRNA is mediated by cell-cell contact, but not by exosome internalization and fusion. They showed that actin polymerization is required for the intercellular transfer of Cas9/gRNA, the latter of which is observed in the projected membrane tubule connections. The authors visualized the fine structure of the tubular connections by electron microscopy and observed …

   Reviewer #3 (Public Review):

   In this manuscript, Zhang and Schekman investigated the mechanisms underlying intercellular cargo transfer. It has been proposed that cargo transfer between cells could be mediated by exosomes, tunneling nanotubes or thicker tubules. To determine which process is efficient in delivering cargos, the authors developed two quantitative approaches to study cargo transfer between cells. Their reporter assays showed clearly that the transfer of Cas9/gRNA is mediated by cell-cell contact, but not by exosome internalization and fusion. They showed that actin polymerization is required for the intercellular transfer of Cas9/gRNA, the latter of which is observed in the projected membrane tubule connections. The authors visualized the fine structure of the tubular connections by electron microscopy and observed organelles and vesicles in the open-ended tubular structure. The formation of the open-ended tubule connections depends on a plasma membrane fusion process. Moreover, they found that the endogenous trophoblast fusogens, syncytins, are required for the formation of open-ended tubular connections, and that syncytin depletion significantly reduced cargo Cas9 protein transfer.

   Overall, this is a very nice study providing much clarity on the modes of intercellular cargo transfer. Using two quantitative approaches, the authors demonstrated convincingly that exosomes do not mediate efficient transfer via endocytosis, but that the open-ended membrane tubular connections are required for efficient cargo transfer. Furthermore, the authors pinpointed syncytins as the plasma membrane fusogenic proteins involved in this process. Experiments were well designed and conducted, and the conclusions are mostly supported by the data. My specific comments are as follows.

   1. The authors showed that knocking down actin (which isoform?) in both donor and acceptor cells blocked transfer, and more so in the acceptor cells perhaps due to the greater knockdown efficiency in these cells. However, Arp2/3 complex knockdown in donor cells, but not recipient cell, reduced Cas9 transfer. It would be good to clarify whether the latter result suggests that the recipient cells use other actin nucleators rather than Arp2/3 to promote actin polymerization in the cargo transfer process. Are formins involved in the formation of these tubular connections?
   2. The authors provided convincing evidence to show that the tubular connections are involved in cargo transfer. Intriguingly, in Figure 4-figure supplement video (upper right), protein transfer appeared to occur along a broad cell-cell contact region instead of a single tubular connection. How often does the former scenario occur? Is it possible that transfer can happen as long as cells are contacting each other and making protrusions that can fuse with the target cell?
   3. The requirement of MFSD2A in both donor (HEK293T) and recipient (MDA-MB-231) cells is consistent with a role for syncytin-1 or 2 in both types of cells. Since HEK293T cells contain both syncytins and MFSD2A but cargo transfer does not occur among these cells, does this suggest that syncytins and/or MFSD2A are only trafficked to the HEK293T cell membrane in the presence of MDA-MB-231 cells?

   Read the original source
7. 

   Version published to 10.1101/2022.11.01.514662v1 on bioRxiv

   Nov 1, 2022