RAB14-dependent tubulovesicular recycling directs MET to invadopodia, promoting TNBC cell invasion

Reviewer #1 (Public review):

Summary:

This study identifies a mechanism responsible for the accumulation of the MET receptor in invadopodia, following stimulation of Triple-negative breast cancer (TNBC) cells with HGF. HGF-driven accumulation and activation of MET in invadopodia causes the degradation of the extracellular matrix, promoting cancer cell invasion, a process here investigated using gelatin-degradation and spheroid invasion assays.

Mechanistically, HGF stimulates the recycling of MET from RAB14-positive endodomes to invadopodia, increasing their formation. At invadopodia, MET induces matrix degradation via direct binding with the metalloprotease MT1-MMP. The delivery of MET from the recycling compartment to invadopodia is mediated by RCP, which facilitates the colocalization of MET to RAB14 endosomes. In this compartment, HGF induces the recruitment of the motor protein KIF16B, promoting the tubulation of the RAB14-MET recycling endosomes to the cell surface. This pathway is critical for the HGF-driven invasive properties of TNBC cells, as it is impaired upon silencing of RAB14.

Strengths:

The study is well-organized and executed using state-of-the-art technology. The effects of MET recycling in the formation of functional invadopodia are carefully studied, taking advantage of mutant forms of the receptor that are degradation-resistant or endocytosis-defective.

Data analyses are rigorous, and appropriate controls are used in most of the assays to assess the specificity of the scored effects. Overall, the quality of the research is high.

The conclusions are well-supported by the results, and the data and methodology are of interest for a wide audience of cell biologists.

Weaknesses:

The role of the MET receptor in invadopodia formation and cancer cell dissemination has been intensively studied in many settings, including triple-negative breast cancer cells. The novelty of the present study mostly consists of the detailed molecular description of the underlying mechanism based on HGF-driven MET recycling. The question of whether the identified pathway is specific for TNBC cells or represents a general mechanism of HGF-mediated invasion detectable in other cancer cells is not addressed or at least discussed.

Reviewer #2 (Public review):

Summary:

In this manuscript, Khamari and colleagues investigate how HGF-MET signaling and the intracellular trafficking of the MET receptor tyrosine kinase influence invadopodia formation and invasion in triple-negative breast cancer (TNBC) cells. They show that HGF stimulation enhances both the number of invadopodia and their proteolytic activity. Mechanistically, the authors demonstrate that HGF-induced, RAB4- and RCP-RAB14-KIF16B-dependent recycling routes deliver MET to the cell surface specifically at sites where invadopodia form. Moreover, they report that MET physically interacts with MT1-MMP - a key transmembrane metalloproteinase required for invadopodia function- and that these two proteins co-traffic to invadopodia upon HGF stimulation.

Although the HGF-MET axis has previously been implicated in invadopodia regulation (e.g., by Rajadurai et al., Journal of Cell Science 2012), studies directly linking ligand-induced MET trafficking with the spatial regulation of MT1-MMP localization and activity have been lacking.

Overall, the manuscript addresses a relevant and timely topic and provides several novel insights. However, some sections require clearer and more concise writing (details below). In addition, the quality, reliability, and robustness of several data sets need to be improved.

Strengths:

A key strength of the study is the novel demonstration that HGF-mediated, RAB4- and RAB14-dependent recycling of MET delivers this receptor, together with MT1-MMP, to invadopodia -highlighting a previously unrecognized mechanism, regulating the formation and proteolytic function of these invasive structures. Another strong point is the breadth of experimental approaches used and the substantial amount of supporting data. The authors also include an appropriate number of biological replicates and analyze a sufficiently large number of cells in their imaging experiments, as clearly described in the figure legends.

Weaknesses:

(1) Inappropriate stimulation times for endocytosis and recycling assays.

The experiments examining MET endocytosis and recycling following HGF stimulation appear to use inappropriate incubation times. After ligand binding, RTKs typically undergo endocytosis within minutes and reach maximal endosomal accumulation within 5-15 minutes. Although continuous stimulation allows repeated rounds of internalization, the temporal dynamics of MET trafficking should be examined across shorter time points, ideally up to 1 hour (e.g., 15, 30, and 60 minutes). The authors used 2-, 3-, or 6-hour HGF stimulation, which, in my opinion, is far too long to study ligand-induced RTK trafficking.

(2) Low efficiency of MET silencing in Figure S1I.

The very low MET knockdown efficiency shown in Figure S1I raises concerns. Given the potential off-target effects of a single shRNA and the insufficient silencing level, it is difficult to conclude whether the reduction in invadopodia number in Figure 1F is genuinely MET-dependent. The authors later used siRNA-mediated silencing (Figure S5C), which was more effective. Why was this siRNA not used to generate the data in Figure 1F? Why did the authors rely on the inefficient shRNA C#3?

(3) Missing information on incubation times and inconsistencies in MET protein levels.

The figure legends do not indicate how long the cells were incubated with HGF or the MET inhibitor PHA665752 prior to immunoblotting. This information is crucial, particularly because both HGF and PHA665752 cause a substantial decrease in the total MET protein level. Notably, such a decrease is absent in MDA-MB-231 cells treated with HGF in the presence of cycloheximide (Figure S2F). The authors should comment on these inconsistencies.

Additionally, the MET bands in Figure S1J appear different from those in Figure S1C, and MET phosphorylation seems already high under basal conditions, with no further increase upon stimulation (Figure S1J). The authors should address these issues.

(4) Insufficient representation and randomization of microscopic data.

For microscopy, only single representative cells are shown, rather than full fields containing multiple cells. This is particularly problematic for invadopodia analysis, as only a subset of cells forms these structures. The authors should explain how they ensured that image acquisition and quantification were randomized and unbiased. The graphs should also include the percentage of cells forming invadopodia, a standard metric in the field. Furthermore, some images include altered cells - for example, multinucleated cells - which do not accurately represent the general cell population.

(5) Use of a single siRNA/shRNA per target.

As noted earlier, using only one siRNA or shRNA carries the risk of off-target effects. For every experiment involving gene silencing (MET, RAB4, RAB14, RCP, MT1-MMP), at least two independent siRNAs/shRNAs should be used to validate the phenotype.

(6) Insufficient controls for antibody specificity.

The specificity of MET, p-MET, and MT1-MMP staining should be demonstrated in cells with effective gene silencing. This is an essential control for immunofluorescence assays.

(7) Inadequate demonstration of MET recycling.

MET recycling should be directly demonstrated using the same approaches applied to study MT1-MMP recycling. The current analysis - based solely on vesicles near the plasma membrane - is insufficient to conclude that MET is recycled back to the cell surface.

(8) Insufficient evidence for MET-MT1-MMP interaction.

The interaction between MET and MT1-MMP should be validated by immunoprecipitation of endogenous proteins, particularly since both are endogenously expressed in the studied cell lines.

(9) Inconsistent use of cell lines and lack of justification.

The authors use two TNBC cell lines: MDA-MB-231 and BT-549, without providing a rationale for this choice. Some assays are performed in MDA-MB-231 and shown in the main figures, whereas others use BT-549, creating unnecessary inconsistency. A clearer, more coherent strategy is needed (e.g., present all main findings in MDA-MB-231 and confirm key results in BT-549 in supplementary figures).

(10) Inconsistency in invadopodia numbers under identical conditions.

The number of invadopodia formed in Figure 1E is markedly lower than in Figure 1C, despite identical conditions. The authors should explain this discrepancy.

(11) Questionable colocalization in some images.

In some figures - for example, Figure 2G - the dots indicated by arrows do not convincingly show colocalization. The authors should clarify or reanalyze these data.

(12) Abstract, Introduction, and Discussion require substantial rewriting.

(a) The abstract should be accessible to a broader audience and should avoid using abbreviations and protein names without context.

(b) The introduction should better describe the cellular processes and proteins investigated in this study.

(c) The discussion currently reads more like an extended summary of results. It lacks deeper interpretation, comparison with existing literature, and consideration of the broader implications of the findings.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

This study identifies a mechanism responsible for the accumulation of the MET receptor in invadopodia, following stimulation of Triple-negative breast cancer (TNBC) cells with HGF. HGF-driven accumulation and activation of MET in invadopodia causes the degradation of the extracellular matrix, promoting cancer cell invasion, a process here investigated using gelatin-degradation and spheroid invasion assays.

Mechanistically, HGF stimulates the recycling of MET from RAB14-positive endosomes to invadopodia, increasing their formation. At invadopodia, MET induces matrix degradation via direct binding with the metalloprotease MT1-MMP. The delivery of MET from the recycling compartment to invadopodia is mediated by RCP, which facilitates the colocalization of MET to RAB14 endosomes. In this compartment, HGF induces the recruitment of the motor protein KIF16B, promoting the tubulation of the RAB14-MET recycling endosomes to the cell surface. This pathway is critical for the HGF-driven invasive properties of TNBC cells, as it is impaired upon silencing of RAB14.

Strengths:

The study is well-organized and executed using state-of-the-art technology. The effects of MET recycling in the formation of functional invadopodia are carefully studied, taking advantage of mutant forms of the receptor that are degradation-resistant or endocytosis-defective.

Data analyses are rigorous, and appropriate controls are used in most of the assays to assess the specificity of the scored effects. Overall, the quality of the research is high.

The conclusions are well-supported by the results, and the data and methodology are of interest for a wide audience of cell biologists.

We sincerely thank the reviewer for his/her positive feedback and for considering our study to be well executed and rigorous. The valuable suggestions and comments will certainly improve the understanding of the role of the RAB14-RCP-KIF16B axis in MET trafficking and breast cancer invasion. Below we have addressed each of the concerns and suggestions point to point raised by the reviewer.

Weakness:

The role of the MET receptor in invadopodia formation and cancer cell dissemination has been intensively studied in many settings, including triple-negative breast cancer cells. The novelty of the present study mostly consists of the detailed molecular description of the underlying mechanism based on HGF-driven MET recycling. The question of whether the identified pathway is specific for TNBC cells or represents a general mechanism of HGF-mediated invasion detectable in other cancer cells is not addressed or at least discussed

We thank the reviewer for raising this point. We want to clarify that in TNBCs, the lack of the hormonal receptor progesterone receptor, estrogen receptor, and HER2 makes the overexpression of EGFR and MET crucial in terms of prognosis and treatment (PMID: 27655711, 25368674). Hence study of MET signalling and trafficking is more relevant for TNBCs compared to other cancer cells. We will add an explanation in the discussion section in the revised manuscript.

Reviewer #2 (Public review):

Summary:

In this manuscript, Khamari and colleagues investigate how HGF-MET signaling and the intracellular trafficking of the MET receptor tyrosine kinase influence invadopodia formation and invasion in triple-negative breast cancer (TNBC) cells. They show that HGF stimulation enhances both the number of invadopodia and their proteolytic activity. Mechanistically, the authors demonstrate that HGF-induced, RAB4- and RCP-RAB14-KIF16B-dependent recycling routes deliver MET to the cell surface specifically at sites where invadopodia form. Moreover, they report that MET physically interacts with MT1-MMP - a key transmembrane metalloproteinase required for invadopodia function- and that these two proteins co-traffic to invadopodia upon HGF stimulation.

Although the HGF-MET axis has previously been implicated in invadopodia regulation (e.g., by Rajadurai et al., Journal of Cell Science 2012), studies directly linking ligand-induced MET trafficking with the spatial regulation of MT1-MMP localization and activity have been lacking.

Overall, the manuscript addresses a relevant and timely topic and provides several novel insights. However, some sections require clearer and more concise writing (details below). In addition, the quality, reliability, and robustness of several data sets need to be improved.

Strengths:

A key strength of the study is the novel demonstration that HGF-mediated, RAB4- and RAB14-dependent recycling of MET delivers this receptor, together with MT1MMP, to invadopodia -highlighting a previously unrecognized mechanism, regulating the formation and proteolytic function of these invasive structures. Another strong point is the breadth of experimental approaches used and the substantial amount of supporting data. The authors also include an appropriate number of biological replicates and analyze a sufficiently large number of cells in their imaging experiments, as clearly described in the figure legends.

We greatly appreciate the positive assessment we have from the reviewer, who also acknowledged the novelty and relevance of our study. Below, we have carefully addressed the comments/concerns raised regarding this study and will strengthen the reliability and robustness by revisiting the data, providing additional analyses where required, and clarifying methodological details.

Weakness:

(1) Inappropriate stimulation times for endocytosis and recycling assays. The experiments examining MET endocytosis and recycling following HGF stimulation appear to use inappropriate incubation times. After ligand binding, RTKs typically undergo endocytosis within minutes and reach maximal endosomal accumulation within 5-15 minutes. Although continuous stimulation allows repeated rounds of internalization, the temporal dynamics of MET trafficking should be examined across shorter time points, ideally up to 1 hour (e.g., 15, 30, and 60 minutes). The authors used 2-, 3-, or 6-hour HGF stimulation, which, in my opinion, is far too long to study ligandinduced RTK trafficking.

We understand the reviewer’s concern regarding the HGF stimulation time point for endocytosis and recycling. We want to highlight that to study the recycling/surface delivery of MET in response to HGF, we performed TIRF microscopy-based imaging, where images were taken within 1h of HGF addition (Fig. 2I). Additionally, we will incorporate surface biotinylation to show the recycling of MET as suggested in comment -7. Moreover, we have observed the effect of HGF on gelatin degradation and invadopodia formation after 3h of HGF stimulation. We were curious to know where MET resides with prolonged ligand stimulation. Hence, to study the localization of MET to invadopodia or the endocytic markers, the cells were stimulated with HGF for 2-3 hours.

(2) Low efficiency of MET silencing in Figure S1I. The very low MET knockdown efficiency shown in Figure S1I raises concerns. Given the potential off-target effects of a single shRNA and the insufficient silencing level, it is difficult to conclude whether the reduction in invadopodia number in Figure 1F is genuinely MET-dependent. The authors later used siRNA-mediated silencing (Figure S5C), which was more effective. Why was this siRNA not used to generate the data in Figure 1F? Why did the authors rely on the inefficient shRNA C#3?

We understand the concern raised by the reviewer. We want to emphasize that we have employed three different approaches to investigate the effect of MET silencing/inhibition on invadopodia formation. (i) A MET kinase inhibitor, PHA665752, which shows reduced invadopodia formation. (Fig. 1D, E). (ii) Silencing with shRNA: Since the level of silencing of MET with the shRNA was not sufficient, cells were stained with MET as a readout for MET silencing, and images of the cells with depleted MET expression were captured, and invadopodia numbers were quantified (Fig. 1F). (iii) Using the SMARTpool siRNA of MET, we have shown the MT1-MMP containing invadopodia in Fig S5E, which shows another evidence of the role of MET in invadopodia activity. An additional graph showing invadopodia formation derived from the siRNA-mediated MET silencing will be added to the revised figure.

(3) Missing information on incubation times and inconsistencies in MET protein levels. The figure legends do not indicate how long the cells were incubated with HGF or the MET inhibitor PHA665752 before immunoblotting. This information is crucial, particularly because both HGF and PHA665752 cause a substantial decrease in the total MET protein level. Notably, such a decrease is absent in MDA-MB-231 cells treated with HGF in the presence of cycloheximide (Figure S2F). The authors should comment on these inconsistencies. Additionally, the MET bands in Figure S1J appear different from those in Figure S1C, and MET phosphorylation seems already high under basal conditions, with no further increase upon stimulation (Figure S1J). The authors should address these issues.

We apologise for the unintentional omission of experimental detailing about HGF or drug incubation time, which will be incorporated into the figure legend appropriately. The blot will be replaced with a more appropriate representative image.

Regarding the decreased MET level in the drug-treated condition: literature suggests that the MET inhibitor PHA665752 also promotes MET degradation, corroborating our result shown in Fig. S1J (PMID: 15788682, 18327775). Further in Fig. S1J, the relative phosphorylation of MET when compared to the total MET level in the HGF-treated condition is higher. We will add the quantification in the revised manuscript to add more clarity.

Next, in the fig. S1C, the rabbit anti-MET (CST, D1C2 XP) antibody has been used, which binds to a c-terminal motif of MET and identifies both the 170kDa as well as 140kDa protein representing the uncleaved and cleaved form of MET. In Fig. S1J, the mouse antiMET (CST, L6E7) antibody has been used, which binds to an N-terminal motif of MET and recognizes only the 140kDa protein.

(4) Insufficient representation and randomization of microscopic data. For microscopy, only single representative cells are shown, rather than full fields containing multiple cells. This is particularly problematic for invadopodia analysis, as only a subset of cells forms these structures. The authors should explain how they ensured that image acquisition and quantification were randomized and unbiased. The graphs should also include the percentage of cells forming invadopodia, a standard metric in the field. Furthermore, some images include altered cells - for example, multinucleated cells - which do not accurately represent the general cell population.

We thank the reviewer for raising this point. The single-cell images are shown for clarity and to visualize the subcellular features; however, the conclusions are made based on the quantitative analysis of multiple cells collected from multiple frames (at least 30 frames per condition). Here, we would like to highlight that the image acquisition has been done over random fields in a coverslip. In the graphs shown in Fig. 1B, 1C, 4F, S1F, S1H, S5J’ it can be seen that there are frames where there is no degradation or invadopodia formed, which has also been taken into account. For a better representation of the population of cellforming invadopodia, a graph showing the percentage of cells forming invadopodia will be added to the figure.

(5) Use of a single siRNA/shRNA per target. As noted earlier, using only one siRNA or shRNA carries the risk of off-target effects. For every experiment involving gene silencing (MET, RAB4, RAB14, RCP, MT1-MMP), at least two independent siRNAs/shRNAs should be used to validate the phenotype.

We would like to clarify that we are using SMARTPool siRNA, which contains 4 individual siRNAs for the target gene. Literature suggests that using a pool of siRNA has reduced offtarget effects compared to using single oligos for gene silencing (PMID: 14681580, 33584737, 24875475).

While SMARTpool siRNA minimizes the off-target effect, it does not eliminate the possibility of it. To confirm that the observed phenotypes are specifically attributable to the genes investigated in this study, we will perform additional experiments using two independent siRNAs targeting RCP and RAB14. RAB4 is known to be associated with MET trafficking (PMID: 21664574, 30537020), and we have taken RAB4 as a positive control. Hence, we feel the suggested experiment is not required to support the conclusion made regarding RAB4.

For MET, we have used shRNA and an inhibitor to show the effect of MET inhibition/perturbation in the invadopodia-associated activity, which validates the observations of siRNA-mediated gene silencing.

We have shown the effect of MT1-MMP depletion on invadopodia formation using a CRISPR-based gene knock-out study, and another study from our group has shown the effect using siRNA (PMID: 31820782), which supports our MT1-MMP KO cell observation.

(6) Insufficient controls for antibody specificity. The specificity of MET, p-MET, and MT1-MMP staining should be demonstrated in cells with effective gene silencing. This is an essential control for immunofluorescence assays.

MET immunofluorescence staining in the MET-depleted condition has been provided in Fig. 1F, and an immunoblot for the siRNA-mediated gene silencing has been provided in Fig. S5C. We will add the entire field of view to show the MET silencing in Fig. 1F.

The inhibition of MET kinase activity using PHA665752 abolished the MET phosphorylation, as shown in Fig S1J. In line with Joffre et.al. Fig 3C, S2I shows increased Tyr 1234/1235 phosphorylation of M1250T MET mutant (PMID: 21642981). Further, studies have shown the specificity of the antibody by immunoblotting and immunofluorescence using MET inhibitors (PMID: 21973114, 41009793).

For the MT1-MMP immunoblot showing significant depletion in MT1-MMP protein level by the SMARTpool siRNA has been provided in Fig. S5L. Further MT1-MMP silencing has been validated by immunofluorescence in the following studies. PMID: 22291036, 21571860, 20505159.

(7) Inadequate demonstration of MET recycling. MET recycling should be directly demonstrated using the same approaches applied to study MT1-MMP recycling. The current analysis - based solely on vesicles near the plasma membrane - is insufficient to conclude that MET is recycled back to the cell surface.

We appreciate the reviewer’s suggestion for an alternative approach to show MET trafficking. We aim to demonstrate MET trafficking using a biochemical approach, which will be included in the revised version.

(8) Insufficient evidence for MET-MT1-MMP interaction. The interaction between MET and MT1-MMP should be validated by immunoprecipitation of endogenous proteins, particularly since both are endogenously expressed in the studied cell lines.

We thank the reviewer for pointing out the lack of MET-MT1-MMP interaction at the endogenous level. We have carried out the immunoprecipitation of endogenous MET to validate the interaction with MT1-MMP. However, we could not capture the interaction of these proteins at endogenous levels. We hypothesize that the interaction between MT1MMP and MET may be weak in nature, with a high K_d value, and accordingly, it was difficult to precipitate the endogenous MT1-MMP by MET. The immunoblot will be added to the revised manuscript and discussed.

(9) Inconsistent use of cell lines and lack of justification. The authors use two TNBC cell lines: MDA-MB-231 and BT-549, without providing a rationale for this choice. Some assays are performed in MDA-MB-231 and shown in the main figures, whereas others use BT-549, creating unnecessary inconsistency. A clearer, more coherent strategy is needed (e.g., present all main findings in MDA-MB-231 and confirm key results in BT549 in supplementary figures).

MDA-MB-231 and BT-549 are two well-characterized TNBC cell lines, which are being used extensively to study breast cancer cell invasion. These two cell lines also show overexpression of MET, making them suitable model cell lines for our study.

MDA-MB-231 has less transfection efficiency compared to BT-549. Additionally, MET is also a difficult gene to transfect, making it hard to perform experiments in MDA-MB-231 with MET overexpression. Though most of the experiments have been performed in both cell lines, a few of the studies have been performed only in the BT-549 cells. Further, we have focused on displaying the different approaches taken to validate an observation in the main figure, which led to showing the data in distinct cell lines.

Also, showing observations in different cell lines is a practice that has been followed by multiple authors in the past. (PMID: 39751400, 41079612, 25049275, 22366451)

(10) Inconsistency in invadopodia numbers under identical conditions. The number of invadopodia formed in Figure 1E is markedly lower than in Figure 1C, despite identical conditions. The authors should explain this discrepancy.

We sincerely thank the reviewer for pointing out the inconsistency in invadopodia numbers across 2 experiments. Fig. 1C has 2 conditions: UT and the HGF-treated condition. The Untreated condition has the serum-free media without any stimulation. Whereas we have added vehicle (DMSO) in Fig. 1D, E, since the drug is resuspended in DMSO. This difference in the treatment is likely to be responsible for the decreased numbers of invadopodia in Fig. 1E.

(11) Questionable colocalization in some images. In some figures - for example, Figure 2G - the dots indicated by arrows do not convincingly show colocalization. The authors should clarify or reanalyze these data.

We thank the reviewer for the valuable comment. The apparent lack of convincing colocalization is likely due to the relatively lower fluorescence intensity of MET at these structures. We will add the line intensity plots for the indicated puncta to show the intensity of both channels in the figure.

To quantify the colocalization of two channels, we have used the automated image analysis software motiontracking (motiontracking.mpi-cbg.de), which has been detailed in the method section. Motiontracking considers only those objects to be colocalized if there is an overlapping area of more than 35% between the two channels. Lastly, the apparent colocalization is corrected for random colocalization, which is the random permutation of object colocalization. This makes object-based colocalization more reliable than intensitybased colocalization.

(12) Abstract, Introduction, and Discussion require substantial rewriting. a) The abstract should be accessible to a broader audience and should avoid using abbreviations and protein names without context. b) The introduction should better describe the cellular processes and proteins investigated in this study. c) The discussion currently reads more like an extended summary of results. It lacks deeper interpretation, comparison with existing literature, and consideration of the broader implications of the findings.

We thank the reviewer for this suggestion. We will modify the abstract, introduction, and discussion as per the suggestion.