Loss of ErbB3 redirects Integrin β1 from early endosomal recycling to secretion in extracellular vesicles

  1. Dorival Mendes Rodrigues-Junior
  2. Ana Rosa Saez-Ibanez
  3. Takeshi Terabayashi
  4. Nina Daubel
  5. Taija Makinen
  6. Olof Idevall-Hagren
  7. Aristidis Moustakas
  8. Ingvar Ferby

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Receptor tyrosine kinases (RTKs) are important cargo of endocytic trafficking, yet the role of RTKs in endocytic sorting and maturation of multivesicular bodies (MVBs) per se remains unclear. Here we show that the ErbB3 receptor, frequently overexpressed in invasive breast cancers, sorts internalized transferrin receptor (TfR) and Integrin β1 for endocytic recycling, in a manner that does not require ligand-induced ErbB3 signaling in breast epithelial cells. Loss of ErbB3 abrogates recycling of Integrin β1, likely from a Rab4-positive compartment, and redirects it towards lysosomal degradation or secretion as cargo of extracellular vesicles (EVs). Depletion of ErbB3 impairs migration of sheets of breast epithelial cells, concomitant with imposing a reduction of Integrin β1 cell surface level and an increase in release of EVs containing Integrin β1. In contrast, EVs secreted from ErbB3-depleted cells enhanced the motility of wild-type cells, potentially counteracting the reduced migration of ErbB3-deficient cells. Mechanistically, ErbB3 facilitates assembly of the Arf6-GGA3-Rabaptin5 endosomal sorting complex to promote early recycling, which consequently suppresses EV release. Therefore, ErbB3 constitutes part of the endosomal trafficking machinery, provoking the notion that RTKs might play a novel and yet unrecognised role in vesicular trafficking.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    *We are grateful for the overall positive feedback and constructive suggestions. We have been able to experimentally address several of the suggested points and provide here a revision plan addressing all of the reviewers’ additional concerns. *

    *In summary, this study is of fundamental novelty and high impact as it: *

    1. Reveals an unexpected role of ErbB3 in controlling ____Integrin β1 ____trafficking ____and thus epithelial cell motility and extracellular vesicles secretion. This may shed important insights into the role of ErbB3____* in cancer. *
    2. Uncovers the first ligand-independent, non-canonical cellular function* for ErbB3 as a scaffold for the Arf6-Rabaptin5-GGA3 endosomal sorting complex.*
    3. Provoking the notion that pseudo-RTKs may have evolved cellular functions beyond receptor signaling, such as by scaffolding endosomal sorting compartments. *We hope that you share our view that these conceptually ground breaking findings will be of interest to a broad cross-disciplinary audience interested in cell signaling, cancer biology, endocytic trafficking and integrin biology. *

    1. Point-by-point description of the revisions

    Reviewer #1 (Evidence____, reproducibility and clarity (Required)):

    ErbB3 is well-known for its significance in cancer, which is dependent on ligand-binding and heterodimerization with other ErbB family members. In the current work, Rodrigues-Junior et al. identified novel, unexpected functions of ErbB3 in promoting early endocytic recycling and restricting exocytic trafficking (extracellular vesicles secretion) of membrane receptors, such as integrin b1 and transferrin receptor, via stabilizing the Arf6-GGA3-Rabaptin5 endosomal sorting complex. Via ErbB3 siRNA knockdown, they observed an impaired recycling of transferrin receptor and integrin b1 back to the cell membrane. The recycling assay condition (growth factor-deprived) provided a very clean result to support that this ErbB3-dependent endocytic trafficking is ligand-binding independent. The trafficking-dependence on ErbB3 (both the endocytic and the exocytic) was further supported by integrin b1 functional assays (scratch closure assay and Matrigel invasion assay). There are still some details that need to be clarified to fully understand the conclusion.

    Major* points:*

      • The manuscript started with a pathological correlation between high ErbB3 level and poor patient survival rate. In Fig.1, the impaired TfR recycling, and the co-localization between ErbB3 and integrin b1 were also performed in the pathological breast cancer cell line, MCF7. While investigating integrin b1 recycling, the authors suddenly switched to another two non-malignant human breast epithelial cell lines, which led to a difficult correlation of ErbB3-mediated recycling back to the disease situation. The authors should state more clearly this point, rather than data not shown. This inconsistency occurred also in other assays, for example, when addressing the trafficking from TGN to cell surface, MCF7 was utilized; while when addressing extracellular vesicle secretion, MCF10A was utilized.* Response*: we thank the reviewer for the comment. The rationale for using different cell-lines or primary cells is now better explained in the manuscript. We found that depletion of ErbB3 impaired recycling of Integrin β1 in the non-malignant cells, including MCF10A and primary breast epithelial cells, but not in malignant MCF7 cells that overexpress ErbB3 (data not shown). We now speculate in the manuscript that perhaps the dependence on ErbB3 for Integrin b1 recycling is lost at some point during carcinogenesis, although further studies will be needed to address this possibility. **MCF7 cells were used to detect endogenous ErbB3 as normal expression levels of ErbB3 (primary MECs and MCF10A) were not detectable by immunofluorescence microscopy in our hands with a range of antibodies we tested. With regard to the transferrin recycling assay, we first attempted to use MCF10A cells for consistency, however we found that transferrin internalized poorly in these cells and the limited pool of transferrin that internalised was retained in these cells for an extended time (3 h), thus rendering them unsuitable for our transferrin experiments. *

    *Concerning the data on trafficking from the TGN to cell surface we mistakenly wrote that they were performed in MCF7 cells although they were in fact done in MCF10A cells. This is now corrected in the new version of this manuscript. *

    Additionally, based on the constructive comment by this reviewer, we have now extended the analysis of EV secretion in ErbB3, Rab4 and Rabaptin5 silenced cells to MCF7 cells. The new data is in line with our findings in MCF10A and prHMEC cells, that absence of ErbB3 significantly increased EV secretion. Moreover, Rab4 and Rabaptin5 knockdown also enhanced the amount of EVs secreted by MCF7 cells. These results were incorporated in the manuscript as new Supplementary Figure S7F-G and new Supplementary Figure S9F-G, as recommended. Furthermore, we also included in this new version that GGA3 and to a lesser extent Rab GTPase-binding effector protein 1 (Rabaptin5 or RABPT5) shared colocalisation with endogenous ErbB3 in MCF7 cells as the new Supplementary Figure 9A, B. Finally, we also attempted to conduct the Arf6 IP in MCF7 cells, but as opposed to MCF10A cells, the yield of Arf6 in pull down experiments was much lower than in MCF10A cells, and interacting proteins were not detectable.

    It was shown before that ErbB3 undergoes constitutive internalization and degradation within several hours that is independent of ligand-binding (ref#13). Can the authors provide experimental evidences to show the correlation of TfR or integrin b1 recycling with this dynamic ErbB3 levels rather than ErbB3 knockdown?

    Response: we have performed colocalization of ErbB3, traced Integrin β1 and the recycling endosome marker EHD1, showing triple colocalization in a subset of endosomes, as shown in the new Supplementary Figure S2H. Experimental limitations prevented us from including EEA1 in triple staining for mCherry-ErbB3 or endogenous ErbB3 protein. Furthermore, ectopically expressed ErbB3 in MCF10A cells did not show convincing co-localisation. We hope that the new EHD1 triple colocalization with ErbB3 and Integrin β1 in endosomal compartments satisfies this specific comment.

    As mentioned above, regarding the transferrin recycling assay, we first attempted to use MCF10A cells for consistency, however we found that transferrin internalized poorly in these cells and the limited pool of transferrin that internalised was retained in these cells for an extended time (3 h), thus preventing their use.

    The efficiency of siRNA knockdown of ErbB3 (both #1 and #2) should support the observed phenotype (Fig. 1I-J, K-L). Is there a correlation between the ErbB3 level with integrin recycling? For example, siRNA#2 led to more efficient knockdown of ErbB3 in MCF10A?

    Response: notably, the immunoblots presented here to assess the efficiency of the two different siRNAs are one example and we noted some variability between different experiments but find that both siRNAs work well and yield comparable effects on recycling of Integrin β1. Importantly, the recycling data represents biological repeats of independently performed experiments, and have yielded reproducible and consistent ErbB3 silencing using both siRNAs. This is noted by the lack of significance between ErbB3 knocked down cells in Fig. 1I-J and K-L. Hence, we consider that both siRNAs against ErbB3 worked efficiently with comparable outcome. Please also note our reply to Rev2 #07.

    ErbB3 loss led to more extracellular vesicles secretion, but also lysosomal degradation of integrin b1. This conclusion is supported by results shown in Fig.4D-E and Fig. S8A-B, while the analysis from the same cell line (MCF10A, Fig. S3A) results in no change of integrin b1 levels upon ErbB3 depletion. Fig. S3B showed also no change in a second non-malignant cell line (prHMEC). How do the authors explain this conflict?

    Response*: we thank the reviewer for this comment. We believe that the increase in EV secretion and lysosomal degradation is compensated by increase in de novo synthesis of Integrin β1 (see data below, from Fig. S3C). In the original manuscript we did not perform the appropriate statistical analysis of the RT-qPCR data. The unpaired two-tail Student’s T-test is only suitable for normally distributed samples, which is not the case here. Instead, we performed the appropriate Mann-Whitney U-test assuming non-normal distribution, yielding an exact p-value of 0.017. The figure S3A and associated text has been modified accordingly. *

    Minor points: 1. Is TfR also colocalizing with endogenous ErbB3?

    *Response: as mentioned in the major comment #02, we attempted to perform the transferrin recycling assay using MCF10A cells to enable direct comparisons with the integrin b1 recycling, but found that transferrin internalized poorly in these cells. *

    Fig. 3J, TSG101, T is masked by 3I

    Response: we apologize for this oversight. We have gone through the manuscript in detail and corrected all pointed errors accordingly.

    Page 10, the description of the EV secretion in prHMEC cells is annotated to the wrong figure. Fig S5Dà S7D; S5Eà S7E

    Response: we apologize for this oversight and have now corrected the mistake.

    Fig. 4M: How was the motility/invasion into Matrigel determined? Images? Only quantifications are shown.

    Response*: the matrigel invasion assay was described in the Material and Methods section. Accordingly, the data were expressed as the percentage of invasion based on the ratio of the mean number of cells invading through Matrigel matrix per mean number of cells in the uncoated support. For this rebuttal letter, the reviewer can find representative images of invaded MCF10A siCtrl non-treated (Ctrl) or treated with VSF secreted from MCF10A siCtrl or siErbB3. Since this is an established method to measure cell invasion, we hope the reviewer agrees that these images do not add value to the manuscript. *

    Fig. 4M: Exosomes collected from ErbB3-depleted cells promotes the migration in MCF10A-wild type cells, how about the effects on ErbB3-depleted cells? This group should be included for analysis.

    Response*: as proposed, we have treated both control and ErbB3-silenced MCF10A cells with normalized concentrations of EVs secreted from siCtrl and siErbB3 (1 x 109 nanoparticles/ mL) for 48 hours, followed by cell viability and cell invasion assays. The new data show that both EV pools modestly increased cell viability and substantially increased invasiveness of both wild-type and ErbB3-depleted cells through Matrigel (new Figures 4K and L). Together, our results indicate that while ErbB3-silenced MCF10A cells exhibited lower basal motility, ErbB3 is not required for the observed EV induced motility. The new Figures 4K and L were included and further discussed in this manuscript. *

    Quantification of the blots should be provided for Fig. 5A (GGA3), 5B (GGA3, Rabaptin5 and Arf6), 5F (GGA3) and 5G (GGA3, Rabaptin5 and Arf6). What is mock IP in each graph? The mock IP is neither mentioned in methods nor in legends.

    Response*: we have now carried out densitometry analysis in all the requested immunoblots shown in Figure 5. We also changed the mock IP term to IgG IP for clarity. The use of non-immunogenic IgG in control IPs is now specified in the methods and respective figure legend. *

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    Summary: In their manuscript, Rodrigues-Junior and colleagues identify a novel ligand-independent function of the tyrosine kinase receptor (RTK) ErbB3 as a regulator of integrin β1 recycling. In particular, the authors demonstrate that ErbB3 depletion reduce β1 integrin surface expression, triggering its lysosomal degradation and increasing its secretion in extracellular vesicles (EVs). Moreover, the authors show that these EVs enhanced the invasive capacity of ErbB3 wild type breast epithelial cells. In addition, the authors evidence the interaction between ErbB3, GGA3 and Rabaptin5. Loss of any of these proteins destabilizes this interaction, which abrogates integrin β1 recycling and leads to its degradation and secretion. The work is potentially interesting; however, there are some aspects that need to be analyzed in a more robust manner.

    Major comments:

    1. The manuscript is mainly focused on β1 integrin endocytic and post-endocytic fate following ErbB3 silencing, describing also a molecular mechanism underlying these observations. Despite the cited manuscript by Deneka, A. and colleagues indicates a similar mechanism for transferrin receptor (TfR) recycling, the Authors only studied the receptor internalization upon ErbB3 silencing. Therefore, this observation does not add any significance to the main topic of the manuscript and its removal should be considered. Response*: we agree with the reviewer the fate of Integrin β1 is the main focus of this manuscript. We would however favour retaining the TfR data as it implies a wider role of ErbB3, beyond trafficking of Integrin β1. We ask for the reviewer’s understanding of our rationale. *

    2.Data from Figure S1A seems to be not normally distributed. Have the Authors tested the data for normal distribution? If not, please consider it. If the data is not normally distributed, a non parametric Mann-Whitney U-Test would be more suitable.

    Response: we thank the reviewer for the comment. The differential ErbB3 mRNA expression analysis was retrieved from the widely used GEPIA2 portal (to date about 600 manuscripts cite this portal on PubMed), based on the selected datasets (“TCGA tumors vs TCGA normal + GTEx normal” or “TCGA tumors vs TCGA normal”). The method for differential analysis is one-way ANOVA, using disease state (Tumor or Normal) as variable for calculating differential expression, as it considers differential expression among several tumors.

    *Tang, Z., Kang, B., Li, C., Chen, T., and Zhang, Z. (2019). *GEPIA2: an enhanced web server for large-scale expression profiling and interactive analysis. Nucleic Acids Res. 47, W556–W560. https://doi.org/10.1093/nar/gkz430.

    1. The Authors studied the colocalization of ErbB3, Rab4 and Rab11, observing an increased colocalization between ErbB3 and Rab4 10 minutes following primaquine. However, the Authors previously referred to Sönnichsen, B et al. manuscript, in which TfR colocalized with Rab11 at 30min. It would be interesting to see whether ErbB3 and Rab11 colocalize at later time points in the presence or absence of primaquine. This will reinforce the conclusion that ErbB3 is involved in early Rab4-dependent recycling.

    Response: we appreciate the reviewer’s comment. However, we consider that these requested experiments will not add significant value to the novelty of this manuscript and hope that the reviewer accepts that we politely refrain from reproducing them.

    In Figure 4C the Authors observed a reduction in β1 integrin levels in ErbB3 silenced cells compared to the control already at the beginning of tracing (0 min), which might be due to accelerated turnover at the internalization step of their experimental design. To confirm this, immunofluorescence of β1 integrin in control and ErbB3 silenced cells could be performed just right after the 15min integrin internalization.

    Response: this is likely a misunderstanding as the timepoint (0 min) is defined as the point after the 15 min internalization step when the imaging-based tracing begins, which aligns perfectly with the reviewer’s request.

    In the discussion, the Authors indicate that "loss of ErbB3 redirects Integrin β1 towards lysosomes for degradation, mimicking loss of GGA3 that similarly redirects both Integrin β1 and c-Met towards lysosomal degradation, or Rabaptin5 depletion that we find similarly redirects trafficking of internalised Integrin β1 towards lysosomal degradation". However, the involvement of lysosomal degradation was only studied for ErbB3 silencing by employing chloroquine. To further support this statement, the use of chloroquine in Rabaptin5- and GGA3-depleted cells is recommended.

    Response: we appreciate the reviewer’s comment, but since these findings have been published earlier, we think that they will not add significant value to the manuscript and hope that the reviewer accepts that we politely refrain from reproducing them.

    Minor comments:

    6.The Authors should consider shortening the following sentences from the Introduction: "GGA proteins contain several functional domains that...thereby regulating sorting of cargo including Integrin β3 and TfR into recycling endosomes".

    Response: we thank the reviewer for the comment. We have now divided this sentence into two for smoother reading.

    The Authors do not show ErbB3 silencing efficiency at the protein level until Figure 3G, which should have been shown in Figure 1 or Supplementary Figure 1, as all the research is based on it. Moreover, GGA3 silencing efficiency was never tested.

    Response*: we thank the reviewer for this comment. We have included a new immunoblot confirming the silencing of ErbB3 by two independent siRNAs in MCF7 cells, as the new Supplementary Figure S2A. Please, note that GGA3 silencing was shown in the main Figure 6J. *

    Figure 1I and Figure 1K may include the representative images for the missing siErbB3 to properly illustrate the associated quantification.

    Response: we thank the reviewer for the comment. We have now included the representative images, as suggested.

    Consider including a Western blot showing the effect of lapatinib in EGFR, ErbB2 and ErbB3 protein expression, including their phosphorylated forms.

    Response: we thank the reviewer for the comment. As requested, we now show that at used concentration, lapatinib efficiently blocked tyrosine phosphorylation of ErbB3 and ERK1/2, without perturbing EGFR or ErbB3 expression levels. We also considered it relevant to show that 1 µM lapatinib used was not cytotoxic to MCF10A and MCF7 cells. We hope that these new results satisfy this specific request.

    Some supplementary figures are mislabelled, such as Supplementary Figure S5D and S5E on page 10, which should be S7D and S7E, respectively. Supplementary Figure S7C on page 15 should be S9C.

    Response: we apologize for this oversight and have performed the corrections.

    The following sentence on page 8 should be revised as a verb is missing: "which corresponds to the reported peak time when colocalization of Rab4 with traced TfR, preceding Rab11 and TfR colocalization that peaks later at 30 minutes".

    Response: we apologize for this oversight. It now reads: "which corresponds to the reported peak time of colocalization of Rab4 with traced TfR, which precedes Rab11 and TfR colocalization that peaks later at 30 minutes".

    The main text indicates that the amount of VSV-G transported to the cell surface after 30min it is not affected by ErbB3 silencing. However, in Figure 3E seems to slightly decrease following the silencing. The Authors may consider employing another Western blot image to match the main text and the quantification in Figure 3F.

    Response: as the reviewer noted the immunoblot showed a slight decrease. It is however a very modest decrease that is also observed in the positive control (MUC1) in the same Streptavidin IP sample. We ask for permission to keep these representative images.

    In the main text, a significant difference in the nanoparticles/cell between ErbB3-depleted cells and wild type or control cells were reported. However, Figure 3I only showed the statistics of each siRNA vs the control and not the wild type condition.

    Response: we apologize for this oversight. We removed from the text the comparison with the wild-type non-transfected cells to avoid misunderstanding.

    The Authors concluded that "chloroquine treatment significantly restored traced Integrin β1 levels". However, this conclusion is not reflected in the statistical analysis reported in Figure 4H, which only showed the differences between control and ErbB3 silenced cells. Thus, the statistics reported for the chloroquine results should be added.

    Response: we appreciate the comment by the reviewer. The requested comparison is now included in the new Figure 4H.

    The Authors concluded that "loss of either GGA3 or Rabaptin5 mimics the effect of loss of ErbB3 on endocytic trafficking of Integrin β1, consistent with the hypothesis that GGA3 and Rabaptin5 are effectors of ErbB3 in promoting endosomal recycling and impeding EV release". To confirm this conclusion, the inclusion of siRabaptin5 results in Figures 6H and 6J is suggested.

    Response*: we thank the reviewer for the comment. We have now included immunoblots of MCF10A cell lysate after silencing ErbB3 or Rabaptin5, as the results shown in the previous Figure 6G. We believe that these new data satisfy the specific request. *

    To be consistent with the results presentation:

    • The inclusion of Modal size is recommended in Figure 6I.

    • Some graphs show the number of cells or biological replicates while other ones no.

    • Figure 4E showed different time points for both siRNAs.

    Response: we appreciate the comment and we have now included as the new main Figure 6H the modal size for the EVs secreted by MCF10A cells upon Rabaptin5 silencing. We will ensure that all respective Figure legends indicate the number of replicates. The intermediate time points showed in the main Figure 4E are different, however since the final read out at 9 h using two independent siRNAs against ErbB3 are directly comparable we ask permission to maintain the time points with respect to the analysis we performed.

    Figure 1E represents the squared regions of Figure 1D, but it is not indicated in the figure legend.

    Response: we apologize for this oversight. We have now indicated in Figure 1 legend that Figure 1E represents the squared regions of Figure 1D, as suggested.

    In the legend of Figure 1D-G, 30min of integrin internalization is reported, where it should be 15min according to main text and methods.

    Response: we apologize for this oversight and we thank the reviewer for this comment. We have now indicated the correct time point in Figure 1 legend.

    The addition of representative images in Figure 6A is recommended, as already present in Figure 1I.

    Response: we thank the reviewer for the comment. Representative images of Fig. 6A-D were included as the new panel Fig. 6B.

    As two different siRNAs for ErbB3 were used and not in all experiments, the employed siRNA should be indicated in each experiment. In the cases where both ErbB3 siRNAs were employed, figures should report them either as main results or supplementary.

    Response: we appreciate this meticulous comment. We have now indicated in the figure and in the respective figure legends which siRNA was used in the respective set of experiments (siErbB3 #01 or #02).

    Why do the Authors use EVs enriched in the VSF or by UC to show the same result? What is the criteria to choose one or the other one? For example, in Figures 6G and 6K.

    Response*: based on the guidelines suggested by MISEV 2018 and 2023, there is no gold standard method for EV isolation. Thus, by using at least two independent methods (i.e., tangential flow filtration, followed by immuno-affinity and ultracentrifugation; UC) we validate the enrichment of EVs in our sample preparations, showing reproducible results among the different EV enrichment protocols (Figure 3). *

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    The paper by Dorival Mendes Rodrigues-Junior et al., focuses on a novel ligand-independent role of ErbB3 receptor, modulating Transferrin receptor and integrin beta1 early recycling. Authors perform several in vitro studies where they show how ErbB3 depletion diverts integrin beta1 from recycling towards lysosomal degradation and extracellular vesicle secretion, impairing cell migration. They also provide mechanistic experiments showing the role of ErbB3 on Arf6-GGA3-Rabaptin5 endosomal complex assembly.

    Major comments:

    1. Fig. 1. Authors should co-stain with early endosomal markers (such as EEA1) to clearly show endogenous ErbB3 and Beta1 integrin endosomal co-localization. Including some insets with higher magnifications would also improve visual inspection of such interactions. Response: as requested, we have performed colocalization of ErbB3, traced Integrin β1 and the recycling endosome marker EHD1, showing triple colocalization in a subset of endosomes, as shown in the new Supplementary Figure S2H. Experimental limitations prevented us from including EEA1 in triple staining for mCherry-ErbB3 or endogenous ErbB3 protein. Furthermore, ectopically expressed ErbB3 in MCF10A cells did not show convincing co-localisation with EEA. We believe that the new triple colocalization showing ErbB3 and Integrin β1 in EHD1-positive endosomal compartments satisfies this specific comment.

    Fig. 1H and 1I. Authors need to provide TIRF penetration depth to better evaluate the potential cytosolic contribution. Additionally, plasma membrane purification studies would help to validate their live imaging results.

    Response:* the TIRF penetration depth was 83nm which has now been added to the methods section. Purifications of plasma membrane fractions, following recycling of traced surface-labelled Integrin β1 in control or siErbB3 depleted cells, by cell surface biotinylation and immunoblotting of the recovered proteins is indeed a valuable approach to validate our findings. Nevertheless, we are confident about the results of our confocal imaging results. Thus, including these results might not contribute significantly to the novelty of this manuscript. Hence, we ask permission to publish the paper at this stage, without the plasma membrane purification, as this requires optimizations and will delay the publication of our paper, in addition to exhausting our limited financial resources.*

    Fig. 1J. Authors should explain better how they calculated normalized fluorescence.

    Response*: the normalized fluorescence is explained in the Fig. 1J legend and in the respective method section. *Alexa488 intensity was normalized between 0-1, with the control as reference where Fnorm=((Fmax-Fmin)/(F-Fmin)). All data points were background corrected, followed by normalization to the pre-stimulatory level (F/F0).

    Fig. 2B. Authors should include some plasma membrane markers (such as WGA) to better localize cell surface after beta1 integrin tracing.

    Response: we appreciate the reviewer’s comment, and have attempted the suggested experiment, but in our hands, WGA did not give a clear membrane staining but a diffuse faint signal in MCF10A cells for reasons we do not fully understand.

    Fig. 1J, 1M-1L: beta1 integrin endocytic recycling should be compared across the same time-points to better evaluate kinetic differences.

    Response: the intermediate time points showed in the main Figure 1J, M-L are based on the final read out. We understand that it could be interesting evaluating the kinetic differences but this will generate a substantial number of comparisons that might be difficult for visualization. We ask permission to keep the comparisons among the latest respective time points with respect to the performed analysis.

    Fig. 3. Author should consider adding additional experiments with Rab4 and Rab11 dominant negative forms to validate their results.

    Response: the experiments proposed have been performed, but the ectopic expression of dominant negative Rab4 and Rab11 had detrimental effects to the cells, with the formation of large endosomal blobs and rounding up of the MCF10A cells. Subsequently we do not feel confident with the possible conclusions from these data. We ask the reviewer to understand this technical detail and accept the fact that we are not able to address this point.

    Fig. 4M. To validate authors' claim on the role of integrin Beta1-containing EVs on invasive behaviour, they should repeat the experiment using blocking beta1 antibodies prior to EV addition.

    Response*: we thank the reviewer for this comment. As requested, we performed the experiment using the Integrin β1 blocking monoclonal antibody (mAb; clone P4C10). The new data show that P4C10A treatment alone or in combination with EVs derived from MCF10A cells transfected with siCtrl or siErbB3 significantly reduced invasiveness in comparison to IgG treatments, confirming the mechanistic role of Integrin β1 promoting MCF10A invasive behaviour. The new Figure 4M was included and further discussed in this manuscript. *

    While authors claim that their results could potentially clarify different aspects of tumour dissemination, most of their experiments are done in MCF10A, a non-tumorigenic epithelial cell line. To better support their conclusion, they should reproduce key experiments in MCF7 or other tumorigenic cell line.

    Response*: we thank the reviewer for the comment. As explained in response to reviewer 1, the rational for using different cell-lines or primary cells is now better explained in the manuscript. We found that depletion of ErbB3 impaired recycling of Integrin β1 in the normal non-malignant cells including MCF10A and primary breast epithelial cells, but not in malignant MCF7 cells that overexpress ErbB3 (data not shown), which is now discussed in the paper. Moreover, **MCF7 cells were used to detect endogenous ErbB3 as normal expression levels of ErbB3 (primary MECs and MCF10A) were not detectable by immunofluorescence microscopy with a range of antibodies we tested. Furthermore, we also included in this new version that GGA3 and Rab GTPase-binding effector protein 1 (Rabaptin5 or RABPT5) shared colocalisation with endogenous ErbB3 in MCF7 cells as the new Supplementary Figure 9A, B. Finally, we also attempted to conduct the Arf6 IP in MCF7 cells, but as opposed to MCF10A cells, the yield of Arf6 in pull down experiments was much lower than in MCF10A cells, and interacting proteins were not detectable. *

    Minor comments:

    1. Fig. 1D-1F: please explain better if beta1 integrin surface signal was quenched in these specific set of studies. Response: Beta1 Integrin was quenched on ice with an antibody against Alexa488 as described by Arjonen et al. (Traffic, 2012; DOI: 10.1111/j.1600-0854.2012.01327.x), and further outlined in the methods section and results section (page 6 and schematic Fig4A).

    Suppl. Fig. 3A: last WB lane should read "siErB2" instead of "siErbB3".

    Response: we thank the reviewer and we apologize for this oversight. We corrected the siErbB2 lane in Supplementary Figure 3A, as requested.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    The paper by Dorival Mendes Rodrigues-Junior et al., focuses on a novel ligand-independent role of ErbB3 receptor, modulating Transferrin receptor and integrin beta1 early recycling. Authors perform several in vitro studies where they show how ErbB3 depletion diverts integrin beta1 from recycling towards lysosomal degradation and extracellular vesicle secretion, impairing cell migration. They also provide mechanistic experiments showing the role of ErbB3 on Arf6-GGA3-Rabaptin5 endosomal complex assembly.

    Major comments:

    • Fig. 1. Authors should co-stain with early endosomal markers (such as EEA1) to clearly show endogenous ErbB3 and Beta1 integrin endosomal co-localization. Including some insets with higher magnifications would also improve visual inspection of such interactions.
    • Fig. 1H and 1I. Authors need to provide TIRF penetration depth to better evaluate the potential cytosolic contribution. Additionally, plasma membrane purification studies would help to validate their live imaging results.
    • Fig. 1J. Authors should explain better how they calculated normalized fluorescence
    • Fig. 1J, 1M-1L: beta1 integrin endocytic recycling should be compared across the same time-points to better evaluate kinetic differences.
    • Fig. 2B. Authors should include some plasma membrane markers (such as WGA) to better localize cell surface after beta1 integrin tracing.
    • Fig. 3. Author should consider adding additional experiments with Rab4 and Rab11 dominant negative forms to validate their results.
    • Fig. 4M. To validate authors' claim on the role of integrin Beta1-containing EVs on invasive behaviour, they should repeat the experiment using blocking beta1 antibodies prior to EV addition.
    • While authors claim that their results could potentially clarify different aspects of tumour dissemination, most of their experiments are done in MCF10A, a non-tumorigenic epithelial cell line. To better support their conclusion, they should reproduce key experiments in MCF7 or other tumorigenic cell line.

    Minor comments:

    • Fig. 1D-1F: please explain better if beta1 integrin surface signal was quenched in these specific set of studies
    • Suppl. Fig. 3A: last WB lane should read "siErB2" instead of "siErbB3".

    Significance

    The paper gathers important observations showing a new role of ErbB3 in vesicular trafficking. While these results provide new mechanistic insights that potentially deepen our understanding of tumor dissemination, most of the experiments are done with a non-tumorigenic cell line, and therefore key results should be validated in a tumor cell line context before considering for publication.

    The evidence gathered could be of interest for experts across different biomedical fields, specially within cellular and molecular oncology

    My expertise: cell competition, cancer, mechanobiology, integrin recycling

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Summary:

    In their manuscript, Rodrigues-Junior and colleagues identify a novel ligand-independent function of the tyrosine kinase receptor (RTK) ErbB3 as a regulator of integrin β1 recycling. In particular, the authors demonstrate that ErbB3 depletion reduce β1 integrin surface expression, triggering its lysosomal degradation and increasing its secretion in extracellular vesicles (EVs). Moreover, the authors show that these EVs enhanced the invasive capacity of ErbB3 wild type breast epithelial cells. In addition, the authors evidence the interaction between ErbB3, GGA3 and Rabaptin5. Loss of any of these proteins destabilizes this interaction, which abrogates integrin β1 recycling and leads to its degradation and secretion. The work is potentially interesting; however, there are some aspects that need to be analyzed in a more robust manner.

    Major comments:

    1. The manuscript is mainly focused on β1 integrin endocytic and post-endocytic fate following ErbB3 silencing, describing also a molecular mechanism underlying these observations. Despite the cited manuscript by Deneka, A. and colleagues indicates a similar mechanism for transferrin receptor (TfR) recycling, the Authors only studied the receptor internalization upon ErbB3 silencing. Therefore, this observation does not add any significance to the main topic of the manuscript and its removal should be considered.
    2. Data from Figure S1A seems to be not normally distributed. Have the Authors tested the data for normal distribution? If not, please consider it. If the data is not normally distributed, a non parametric Mann-Whitney U-Test would be more suitable.
    3. The Authors studied the colocalization of ErbB3, Rab4 and Rab11, observing an increased colocalization between ErbB3 and Rab4 10 minutes following primaquine. However, the Authors previously referred to Sönnichsen, B et al. manuscript, in which TfR colocalized with Rab11 at 30min. It would be interesting to see whether ErbB3 and Rab11 colocalize at later time points in the presence or absence of primaquine. This will reinforce the conclusion that ErbB3 is involved in early Rab4-dependent recycling.
    4. In Figure 4C the Authors observed a reduction in β1 integrin levels in ErbB3 silenced cells compared to the control already at the beginning of tracing (0 min), which might be due to accelerated turnover at the internalization step of their experimental design. To confirm this, immunofluorescence of β1 integrin in control and ErbB3 silenced cells could be performed just right after the 15min integrin internalization.
    5. In the discussion, the Authors indicate that "loss of ErbB3 redirects Integrin β1 towards lysosomes for degradation, mimicking loss of GGA3 that similarly redirects both Integrin β123 and c-Met towards lysosomal degradation21, or Rabaptin5 depletion that we find similarly redirects trafficking of internalised Integrin β1 towards lysosomal degradation". However, the involvement of lysosomal degradation was only studied for ErbB3 silencing by employing chloroquine. To further support this statement, the use of chloroquine in Rabaptin5- and GGA3-depleted cells is recommended.

    Minor comments:

    6.The Authors should consider shortening the following sentences from the Introduction: "GGA proteins contain several functional domains that...thereby regulating sorting of cargo including Integrin β3 and TfR into recycling endosomes".

    1. The Authors do not show ErbB3 silencing efficiency at the protein level until Figure 3G, which should have been shown in Figure 1 or Supplementary Figure 1, as all the research is based on it. Moreover, GGA3 silencing efficiency was never tested.
    2. Figure 1I and Figure 1K may include the representative images for the missing siErbB3 to properly illustrate the associated quantifications
    3. Consider including a Western blot showing the effect of lapatinib in EGFR, ErbB2 and ErbB3 protein expression, including their phosphorylated forms.
    4. Some supplementary figures are mislabeled, such as Supplementary Figure S5D and S5E on page 10, which should be S7D and S7E, respectively. Supplementary Figure S7C on page 15 should be S9C.
    5. The following sentence on page 8 should be revised as a verb is missing: "which corresponds to the reported peak time when colocalization of Rab4 with traced TfR, preceding Rab11 and TfR colocalization that peaks later at 30 minutes".
    6. The main text indicates that the amount of VSV-G transported to the cell surface after 30min it is not affected by ErbB3 silencing. However, in Figure 3E seems to slightly decrease following the silencing. The Authors may consider employing another Western blot image to match the main text and the quantification in Figure 3F.
    7. In the main text, a significant difference in the nanoparticles/cell between ErbB3-depleted cells and wild type or control cells were reported. However, Figure 3I only showed the statistics of each siRNA vs the control and not the wild type condition.
    8. The Authors concluded that "chloroquine treatment significantly restored traced Integrin β1 levels". However, this conclusion is not reflected in the statistical analysis reported in Figure 4H, which only showed the differences between control and ErbB3 silenced cells. Thus, the statistics reported for the chloroquine results should be added.
    9. The Authors concluded that "loss of either GGA3 or Rabaptin5 mimics the effect of loss of ErbB3 on endocytic trafficking of Integrin β1, consistent with the hypothesis that GGA3 and Rabaptin5 are effectors of ErbB3 in promoting endosomal recycling and impeding EV release". To confirm this conclusion, the inclusion of siRabaptin5 results in Figures 6H and 6J is suggested.
    10. To be consistent with the results presentation:
      • The inclusion of Modal size is recommended in Figure 6I.
      • Some graphs show the number of cells or biological replicates while other ones no.
      • Figure 4E showed different time points for both siRNAs.
    11. Figure 1E represents the squared regions of Figure 1D, but it is not indicated in the figure legend.
    12. In the legend of Figure 1D-G, 30min of integrin internalization is reported, where it should be 15min according to main text and methods.
    13. The addition of representative images in Figure 6A is recommended, as already present in Figure 1I.
    14. As two different siRNAs for ErbB3 were used and not in all experiments, the employed siRNA should be indicated in each experiment. In the cases where both ErbB3 siRNAs were employed, figures should report them either as main results or supplementary.
    15. Why do the Authors use EVs enriched in the VSF or by UC to show the same result? What is the criteria to choose one or the other one? For example in Figures 6G and 6K.

    Significance

    Various studies highlight the involvement of the RTK ErbB3 in cancer development, as well as its potential use a biomarker for prognosis and therapy resistance. It is also known that ErbB3 is constitutively internalized and degraded, in a process controlled by PKC (Dietrich, M. et al. 2019, Exp Cell Res). However, the novelty of this manuscript resides in the idea that ErbB3, as other transmembrane receptors, may regulate the endocytosis and post-endocytic fate of different cargoes, such as integrin β1. The discovery and understanding of new molecular mechanisms might help in the identification of new potential targets for cancer treatment, as well as other diseases in which the endocytic pathways are altered.

    Field of expertise: integrin-mediated cell adhesion and migration, integrin endocytosis and recylcing

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    ErbB3 is well-known for its significance in cancer, which is dependent on ligand-binding and heterodimerization with other ErbB family members. In the current work, Rodrigues-Junior et al. identified novel, unexpected functions of ErbB3 in promoting early endocytic recycling and restricting exocytic trafficking (extracellular vesicles secretion) of membrane receptors, such as integrin b1 and transferrin receptor, via stabilizing the Arf6-GGA3-Rabaptin5 endosomal sorting complex.

    Via ErbB3 siRNA knockdown, they observed an impaired recycling of transferrin receptor and integrin b1 back to the cell membrane. The recycling assay condition (growth factor-deprived) provided a very clean result to support that this ErbB3-dependent endocytic trafficking is ligand-binding independent. The trafficking-dependence on ErbB3 (both the endocytic and the exocytic) was further supported by integrin b1 functional assays (scratch closure assay and Matrigel invasion assay). There are still some details that need to be clarified to fully understand the conclusion.

    Major points:

    1. The manuscript started with a pathological correlation between high ErbB3 level and poor patient survival rate. In Fig.1, the impaired TfR recycling, and the co-localization between ErbB3 and integrin b1 were also performed in the pathological breast cancer cell line, MCF7. While investigating integrin b1 recycling, the authors suddenly switched to another two non-malignant human breast epithelial cell lines, which led to a difficult correlation of ErbB3-mediated recycling back to the disease situation. The authors should state more clearly this point, rather than data not shown. This inconsistency occurred also in other assays, for example, when addressing the trafficking from TGN to cell surface, MCF7 was utilized; while when addressing extracellular vesicle secretion, MCF10A was utilized.
    2. It was shown before that ErbB3 undergoes constitutive internalization and degradation within several hours that is independent of ligand-binding (ref#13). Can the authors provide experimental evidences to show the correlation of TfR or integrin b1 recycling with this dynamic ErbB3 levels rather than ErbB3 knockdown?
    3. The efficiency of siRNA knockdown of ErbB3 (both #1 and #2) should support the observed phenotype (Fig. 1I-J, K-L). Is there a correlation between the ErbB3 level with integrin recycling? For example, siRNA#2 led to more efficient knockdown of ErbB3 in MCF10A?
    4. ErbB3 loss led to more extracellular vesicles secretion, but also lysosomal degradation of integrin b1. This conclusion is supported by results shown in Fig.4D-E and Fig. S8A-B, while the analysis from the same cell line (MCF10A, Fig. S3A) results in no change of integrin b1 levels upon ErbB3 depletion. Fig. S3B showed also no change in a second non-malignant cell line (prHMEC). How do the authors explain this conflict?

    Minor points:

    1. Is TfR also colocalizing with endogenous ErbB3?
    2. Fig. 3J, TSG101, T is masked by 3I
    3. Page 10, the description of the EV secretion in prHMEC cells is annotated to the wrong figure. Fig S5D S7D; S5E S7E
    4. Fig. 4M: How was the motility/invasion into Matrigel determined? Images? Only quantifications are shown.
    5. Fig. 4M: Exosomes collected from ErbB3-depleted cells promotes the migration in MCF10A-wild type cells, how about the effects on ErbB3-depleted cells? This group should be included for analysis.
    6. Quantification of the blots should be provided for Fig. 5A (GGA3), 5B (GGA3, Rabaptin5 and Arf6), 5F (GGA3) and 5G (GGA3, Rabaptin5 and Arf6). What is mock IP in each graph? The mock IP is neither mentioned in methods nor in legends.

    Significance

    Strength: The recycling assay condition (growth factor-deprived) provided a very clean result to support that this ErbB3-dependent endocytic trafficking is ligand-binding independent.

    Limitations: Constantly change cell lines when addressing different questions

  5. Version published to 10.1101/575449v3 on bioRxiv
  6. Version published to 10.1101/575449v2 on bioRxiv
  7. Version published to 10.1101/575449v1 on bioRxiv