Molecular motion and tridimensional nanoscale localization of kindlin control integrin activation in focal adhesions

Focal adhesions (FAs) initiate chemical and mechanical signals involved in cell polarity, migration, proliferation and differentiation. Super-resolution microscopy revealed that FAs are organized at the nanoscale into functional layers from the lower plasma membrane to the upper actin cytoskeleton. Yet, how FAs proteins are guided into specific nano-layers to promote interaction with given targets is unknown. Using single protein tracking, super-resolution microscopy and functional assays, we link the molecular behavior and 3D nanoscale localization of kindlin with its function in integrin activation inside FAs. We show that immobilization of integrins in FAs depends on interaction with kindlin. Unlike talin, kindlin displays free diffusion along the plasma membrane outside and inside FAs. We demonstrate that the kindlin Pleckstrin Homology domain promotes membrane diffusion and localization to the membrane-proximal integrin nano-layer, necessary for kindlin enrichment and function in FAs. Using kindlin-deficient cells, we show that kindlin membrane localization and diffusion are crucial for integrin activation, cell spreading and FAs formation. Thus, kindlin uses a different route than talin to reach and activate integrins, providing a possible molecular basis for their complementarity during integrin activation.

1. The Calderwood lab showed in 2014 (Huet-Calderwood et al. J Cell Sci) that ILK and Kindlin-2 interactions appeared to be important for kindlin targeting to FAs. However, this seems to contrast with the results in Fig. S3 in this study where the ILK-binding mutant of kindlin, L357A shows essentially the same behaviors as WT. In the Calderwood study, L357A kindlin-2 seems to have FAlocalization defect. However, in inset of Fig. S3a, the FA localization of L357A kindlin-2 seems to be quite substantial. While the authors may have mentioned these localization differences in passing, given the extensive and rigorous works on ILK, Kindlin-2, and FAs reported earlier by the Calderwood lab, I feel that any differences on ILK/Kindlin-2 interactions observed here should be looked into or discussed at some depth in context of these earlier findings.
Minor points: -It would be helpful to readers to end with a graphical illustration that summarize the findings. -3D is probably a more common and appropriate term that 'tridimensional' used by the authors Reviewer #3 (Remarks to the Author): The manuscript by Orré et al present a set of elegant data using advanced quantitative bioimaging approaches that unravelled the mobility behavior and nanoscale threedimensional position of the protein kindlin-2 in the context of integrin activation at focal adhesions. Mobility within the plasma membrane as well as vertical positioning of kindlin molecules within single focal adhesions are important parameters to understand how kindlin immobilizes integrins and contribute to their activation in a manner different from talin. Since kindlin and talin are known to have a complementary action during integrin activation, this manuscript provides interesting biophysical insights into the role of kindlin: while talin comes to the FAs directly from the cytosol, kindlin is already diffusing within the plasma membrane via its PH domain before reaching the FAs.
The manuscript is very well-written, the figures are rich but very clear, the methodology is elegant and thorough and appeals to a broad readership and the statements provided are sufficiently novel and link mobility to nanoscale 3D localization of an important integrin activator. Therefore, in my view, this study deserves publication. However, there are a number of points the authors must first address in a revised manuscript. 1) the authors show that beta1 and beta3 integrins seem differently sensitive to impairment of the interaction with kindlin in Figure 1. It is not clear how useful and important these data are for the remaining of the story as they are not discussed further. 5) why is beta3 wild-type integrin immobile outside FAs (Fig 1h)? Why is talin so highly immobile outside FAs (Fig 2d)? Why is 70% of paxillin immobile outside FAs (Fig S5c)? 6) the Diff coefficients are calculated from the small fraction that is mobile. which is very small inside FAs, at the same time the changes in D values are also very very modest: they are statistically different, but what do these tiny changes tell biologically? Could the authors strengthen their interpretation of these data? 7) Figure 6: I appreciate the complexity of the imaging approaches used in this study but presenting data in a main figure that are obtained by two or one experiment is not sufficiently sound. In Fig. S6, four experiments are mentioned in which the same mutants used for figure 6 were used. I therefore assume the authors have the possibility to add more experimental data and strengthen the results shown in fig 6. In addition, I am wondering why the information provided in fig6 is actually not shown and discussed earlier in the manuscript, right before the authors embark on sptPALM experiments? What is the reason for not putting these data as supplementary figure?
8) the data provided in Figure 7 are used a conclusive statement to link the biophysical parameters to integrin activation (cell spreading). However, these data are not entirely novel as the eLife paper of the Faessler group already provided the same knowledge. I am therefore wondering: why not using these data as starting point? As motivation to better understand mechanistically what drives these differences? Mobility and nanoscale 3D localization of kindlin would then be the mechanistic explanation. In Fig S8, interesting data are shown for individual FAs formed after expression of two out of the five kindlin-2 mutants used. I think the same parameters should be provided for the QW, L357A and K390A mutants, to link the effects of these mutations to individual FA properties and eventually to cell spreading. Unfortunately, Fig S8 seems not mentioned in the Results section.

REVIEWER #1 (REMARKS TO THE AUTHOR):
In the present manuscript, T. Orré and colleagues use single particle tracking and 3D nanoscopy to address the localization and dynamics of kindlin 2 in focal adhesions. In particular, they reveal that kindlin is important for integrin immobilization at focal adhesions and that, contrarily to talin, it freely diffuses at the plasma membrane. Furthermore, they demonstrate the importance of its PH domain for its localization. This very complete, well-documented and clear manuscript constitutes an important source of information to understand the role of kindlin 2 in cell adhesion. However, some information could be added and clarified to strengthen the impact of the manuscript.
We are grateful to the reviewer for her/his positive evaluation of our work and constructive suggestions. Furthermore, we appreciate the thorough evaluation of the reviewer #1 and we took into account his/her advices very seriously to strengthen the impact of our manuscript.
In addition to the insightful questions and interesting experiments proposed by the reviewer#1, we also performed additional experiments, suggested by the other reviewers, to improve and strengthen our manuscript.
This led to 4 modified main figures (Fig. 2 The main results are summarized below: -As found in mature focal adhesions (FAs), kindlin-2 is immobile and enriched in early nascent adhesions (NAs) ( Supplementary Fig. S3), and displays membrane free diffusion both inside and outside NAs ( Supplementary Fig. S3). Similarly, a kindlin-2 mutant with impaired binding to integrins (kindlin-2-QW) displayed a decreased immobile fraction in NAs compared to kindlin-2 ( Supplementary Fig. S3), suggesting that association between integrin and kindlin contributes to kindlin-2 immobilization in NAs. The results we found in NAs mirrors the ones found in mature FAs. Thus, the molecular mechanisms leading to integrin activation by kindlin-2 in FAs and NAs are probably closely related.
1 -To further link the molecular behavior of kindlin with its function in integrin activation inside FAs, we performed additional experiments concerning the formation of FAs in kindlin-1,2 Knock Out fibroblasts (Fig. 7). In the original version of the manuscript we performed rescue experiments in kindlin-1,2 KO MEFs, concerning FAs formation, for kindlin-2-WT, kindlin-2-ΔPH, and kindlin-2-ΔPH-CAAX. In the revised manuscript we performed additional experiments and quantifications for kindlin-2-QW and kindlin-2-CAAX. Importantly, these new results on FAs formation follow a similar trend to the one quantified for these mutants regarding enrichment in FAs (Fig. 6), and cell spreading ( Fig. 7): Kindlin-2 = Kindlin-2-CAAX > Kindlin-2-ΔPH-CAAX = Kindlin-2-QW > Kindlin-2-ΔPH. As demonstrated in our study this trend in cell spreading and FAs formation reflects the ability of kindlin-2-WT and mutants to bind and activate integrins, which is driven by membrane recruitment, membrane free-diffusion, and 3D nanoscale localization in the integrin layer inside FAs. -We performed additional experiments and quantification to increase the number of independent experiments concerning the enrichment in FAs of kindlin-2 mutants (Fig. 6). The results obtained confirmed what we have found in the original manuscript and follow the trend: Kindlin-2 = Kindlin-2-K390A > Kindlin-2-ΔPH-CAAX = Kindlin-2-L357A > Kindlin-2-QW > Kindlin-2-ΔPH > cytosolic mEos2. -Kindlin-2 dwell-time and rearward motion in mature FAs. Finally, we generated super-resolved timelapses and kymographs (Fig. 2) to measure kindlin-2 dwell-time in mature FAs and to test whether kindlin-2 is moving rearward inside FAs as demonstrated for talin. First, we found that immobilization durations were much shorter for kindlin-2 compared to talin-1 (Fig. 2h, i). Analysis of longer kymographs showed that the fraction of immobilized kindlin-2 undergoing retrograde flow with speed above 2 nm.s-1 was around 40 % similar to what was measured for talin-1 (Fig.  2j, k) (Rossier et al., 2012). Our results suggest that the interaction of kindlin-2 with stationary integrins is more prevalent than with F-actin, because the fraction of F-actin moving rearward above 2 nm.s-1 is around 75 % (Rossier et al., 2012). We added these new results in Fig. 2, and we mentioned these results in the results section (p.6) and the discussion section to establish a model with the sequential formation of a transient tripartite integrin-kindlin-talin complex (p.12). -We have performed all the additional quantifications that has been proposed by the reviewers (Fig. 2, 5, 6, 7 and Supplementary Fig. S3, S4, S9).

Main comments:
-The authors chose to explore the function of kindlin in mature focal adhesions only, and not in nascent adhesions. However, it is not specified how mature adhesions, and not nascent adhesions, were selected for the analysis. Furthermore, as kindlin was shown to be important for the formation of nascent adhesions (Bachir et al. 2014;Theodosiou et al. 2016;Böttcher et al. 2017), it would also be interesting to address the dynamics of kindlin in nascent adhesions.
In mesenchymal cells such as fibroblasts, nascent adhesions (NAs) form in protrusive structures such as the lamellipodium. We studied extensively the formation of nascent/early adhesions in the lamellipodium of fibroblasts (Giannone et al., Cell 2004, Cell 2007Dubin-Thaler et al., Biophysical Journal 2004). In those articles, we revealed protrusion/retraction cycles ('periodic contractions') during lamellipodium protrusions during cell spreading but also cell migration. We used the same Mouse Embryonic Fibroblasts (MEFs) in these published articles and in the current submitted manuscript.
In these articles, we performed a precise characterization of lamellipodium dynamics but also of integrin, actin and myosin II regulators involved in this phenomenon (VASP, integrin, paxillin, myosin light chain, MLCK, α-actinin). Thus we know where are located all these proteins with the classical optical resolution of ∼ 250 nm. In these previous studies and in the current submitted manuscript, we can clearly discriminate NAs that are initiated 0.5 μm back from the lamellipodium tip (Giannone et al., Cell 2004, Cell 2007 and mature focal adhesions (FAs) that result from the subsequent maturation of a fraction of NAs outside of the lamellipodium. In addition, in all the experiments we performed we used MEFs spread on fibronectin for more than 3 hours possessing well-defined mature FAs. Region of interests where chosen outside of active lamellipodia where are located NAs.
We added 3 sentences in the method section to clarify this point (p. 15): "Cells were co-transfected with mEos2-fused proteins and GFP-paxillin as a FA reporter. To clearly discriminate NAs, initiated 0.5 μm back from the lamellipodium tip (Giannone et al., Cell 2004, Cell 2007 (Giannone et al., Cell 2004, Cell 2007." Following the advice of reviewer#1, in the revised version of the manuscript, we studied the diffusive behavior of kindlin-2 in NAs of MEFs. We performed sptPALM experiments on NAs in lamellipodia of spreading MEFs using mEos2-kindlin-2 and mEos2-kindlin-2-QW614/615AA (impaired interaction with integrins). As found in mature FAs, mEos2-kindlin-2 is immobile and enriched in NAs ( Supplementary Fig.  S3), and displays membrane free diffusion both inside and outside NAs ( Supplementary Fig. S3). As in FAs, kindlin-2-QW displayed a decreased immobile fraction compared to kindlin-2-WT ( Supplementary  Fig. S3), suggesting that association between integrin and kindlin contributes to kindlin-2 immobilization in NAs. The results we found in NAs mirrored those found in FAs. Thus, the molecular mechanisms leading to integrin activation by kindlin-2 in FAs and NAs are probably closely related.
Those new results about kindlin-2 in NAs were added in the results section (p. 6) in "Kindlin-2 undergoes free diffusion along the plasma membrane, inside and outside FAs": "Furthermore, as kindlin was shown to be important for the formation of NAs (Bachir et al. 2014;Theodosiou et al. 2016;Böttcher et al. 2017), we also studied the diffusive behavior of kindlin-2 in NAs within the lamellipodia of spreading MEFs. Like found in mature FAs, mEos2-kindlin-2 is immobile and enriched in NAs ( Supplementary Fig. S3), and displays membrane free diffusion both inside and outside NAs ( Supplementary Fig. S3). Altogether, these results indicate that membrane free diffusion is a general feature of the kindlin family that could confer distinct capabilities compared to talin during integrin activation in NAs and FAs" and in "Disruption of the integrin-kindlin interaction increases kindlin diffusion inside FAs": "We obtained the same results for kindlin-2-QW in NAs ( Supplementary Fig. S3). The remaining fraction of immobilized kindlin-2-QW could result from binding to other partners, especially..." - Fig. 4d shows that the PH domain alone displays the same fraction of immobilization than K2-WT. This seems inconsistent with Fig. 3 that would argue for a dependency on interaction with integrin for kindlin immobilization. This should be commented.
We would like to thank the reviewer for the detailed evaluation of our study. It is true that a decreased immobilization fraction in conjunction with an increased fraction of membrane free-diffusion often reflects a shift in population from an immobile pool engaged with a binding partner inside FAs to a freely diffusing pool detached from this binding partner. This is the case for instance for kindlin-2 versus kindlin-2-QW ( Fig. 3 in MEFs and Supplementary Fig. S8 in Kind Ko cell), or 31-integrin versus 31-integrin-Y795A (Fig. 1).
However, we want to emphasize that the immobilization fraction is not always correlated with the immobilization density or recruitment within a subcellular structure. For example, it is possible to obtain similar immobilization fractions but very different immobilization densities. For the same area, you can find 10 stationary and 10 free-diffusive trajectories, or 100 stationary and 100 free-diffusive trajectories, the immobilization fraction (50%) will be the same but the density will be an order of magnitude different. This is illustrated by the diffusive behavior of talin-1, which exhibits similar levels of immobilization inside compared to outside FAs (82% versus 73%, Fig. 2d), but with a much higher immobilization density inside than outside FAs (~5 times more detections inside than outside as shown in Rossier et al., Nature Cell Biology 2012 - Fig. 4f). This is also the case, for instance, for paxillin, which possesses large immobile fractions inside and outside FAs but is specifically enriched in FAs ( Supplementary Fig. S6). Likewise, kindlin-2 and kindlin-2-ΔPH exhibit similar immobile fractions within FAs (Fig. 4) despite the reduced enrichment of kindlin-2-ΔPH in FAs compared to kindlin-2 ( Fig. 6f).
As pointed out by the reviewer, this is also the case for the PH-domain of kindlin-2 that display similar immobile fraction within FAs than kindlin-2. However, its recruitment in FAs is reduced, as showed by the quantification of the density of detections inside vs. outside FAs for kindlin-2, and for the PH domain of kindlin-2 (kindlin-2: 5.7 times more detections inside than outside; PH-domain: 2.7 times more detections inside than outside). Thus, the PH domain of kindlin-2 within FAs is much less prone to immobilization than kindlin-2. Nevertheless, these immobilizations of the PH-domain of kindlin-2 inside FAs could be due to a direct interaction with paxillin, as previously described (Theodosiou eLife 2016), or to other unknown binding partners. Kindlin being located at 48.5 nm above the coverslip, it should be considered in the "force transduction layer", rather than in the integrin layer. Since there seems to be some differences with the work of Kanchanawong et al., I suggest to determine the axial localization of other markers of the different layers of focal adhesions (e.g. CAAX, integrins, vinculin and actinin) to be able to conclude about the layer in which kindlin is localized.
The reviewer is correct in saying that the axial localizations found using our experimental workflow are shifted upwards compared to the axial localizations found in the study published by Kanchanawong et al. Nature 2010. However, we think that we can explain this apparent discrepancy.
Indeed, the axial localization found for GFP-kindlin-2 in our study was Zpeak = 48.7 nm. The localization we found for paxillin-Nterm was Zpeak = 58.9 nm, compared to Zcenter = 46.2 nm found in the study published by Kanchanawong et al. (Nature 2010). Note that in the same study, the axial localization of paxillin-Cterm (Z center = 43.1) is below the one of paxillin-Nter (Zcenter = 46.2 nm), highlighting that the axial localization also depends on the orientation of proteins, as found for talin (Kanchanawong et al., Nature 2010) and vinculin (Case et al., Nature Cell Biology 2015). Thus, the localization we obtained for paxillin-Nterm is shifted upwards by 12.7 nm (58.9 nm -46.2 nm) from that found in the Kanchanawong article. It should be highlighted that the axial localization found for CAAX-tdEos in Kanchanawong's is Z center = 32.3 nm, while we found an axial localization of 39.0 nm for GFP-kindlin-2-ΔPH-CAAX, which could also be considered as a close marker of the plasma membrane. Importantly, we again obtained a similar upward shift of few nanometers, 6.7 nm (39.0 nm -32.3 nm). Thus it seems that the axial localizations found in our study are upwardly shifted of few nanometers compared to the ones found in Kanchanawong's study.
It is important to note that in the study published by Kanchanawong et al. (Nature 2010), the authors used iPALM to localize the Z position of proteins labelled with a photo-convertible fluorescent proteins (tdEos or mEos2). Thus they measure the axial localization of photo-convertible fluorescent proteins (PA-FPs): target protein + PA-FPs. In our study, we used GFP-tagged proteins that we imaged by dSTORM using anti-GFP nanobodies labelled with AlexaFluor-647. Thus we measure the axial localization of the AlexaFluor: target protein + GFP + Nanobody + AlexaFluor. To note, the size of a GFP is around 4.2 nm x 2.4 nm, while the size of a Nanobody is around 4.8 nm x 2.2 nm. Therefore, the shifted axial localization we obtained could be in part explained by the additional layers/components we used to label the protein of interest.
In the study of Kanchanawong, paxillin is defined as being part of the integrin signaling layer. Thus, in any case, we can use GFP-paxillin-Nterm as the upper limit of the integrin signaling layer. Since, in our DONALD experiments, the axial localization of GFP-kindlin-2 (Nterm labelled, Zpeak = 48.7 nm) is below the one of GFP-paxillin (Nterm labelled, Zpeak = 58.9 nm), we think that we can also define Kindlin-2 as being part of the integrin signaling layer.
-It is not clear whether the 3D nanoscopy analysis was performed only in focal adhesions or in the whole cell edges. As for the other figures, I also suggest to perform these measurements both inside and outside adhesions.
In the initial submission, the results displayed in Fig. 5 related to 3D nanoscopy correspond to axial localizations measured within mature focal adhesions. Following the advice of the reviewer, we also analyzed the axial localizations outside focal adhesions. These results are now presented in Fig. 5 together with the axial localisations inside FAs. The results show that compared to inside FAs, proteins' axial localization outside FAs is generally shifted upwards but conserves the same trends between kindlin-2 and mutants with: kindlin-2-ΔPH-CAAX < kindlin-2 = kindlin-2-QW < kindlin-2-ΔPH. Paxillin displays the widest shift in amplitude between inside and outside FAs. As paxillin outside FAs display no membrane freediffusion and membrane recruitment according to sptPALM results ( Supplementary Fig. S6), paxillin axial localization outside FAs corresponds to the distribution of a cytosolic protein in our experimental conditions. Kindlin-2-ΔPH axial localization is close to the one of paxillin (Zpeak = 81.9 nm) whereas kindlin-nm). This further reinforces the idea that the kindlin-2 PH domain is an important determinant to maintain kindlin-2 at the plasma membrane even outside FAs.

Minor comments:
-The introduction of this manuscript lacks an introduction on the current knowledge on the function of kindlins at focal adhesions.
We added a paragraph to introduce the current knowledge about kindlin functions in integrin adhesions and its potential interplay with talin in p. 3: " Talin We added a sentence in the results section (p. 5) indicating that this point mutation also affects binding to filaminA and also tensin: -I propose to discuss the possible dimerization of kindlin, and in particular how it could affect kindlin diffusion.
We thank the reviewer for this interesting suggestion. As we obtained by sptPALM similar results for kindlin-2 and several mutants in absence or presence of endogenous kindlin-2 (respectively with Kind Ko cells and wild-type MEFs), we decided to mention this possibility in the results section (p. 8) as followed:

"Furthermore, the similarity of diffusive behavior found in wild-type MEFs and Kind Ko cells suggests that sptPALM results obtained in WT MEFs are not biased by potential formation of heterodimers between endogenous kindlins and transfected kindlin-2 mutants as suggested by structural studies (Li et al., 2017)."
-Result/Disruption of the integrin-kindlin interaction increases kindlin diffusion inside FAs: "These results support the idea that kindlin controls integrin immobilization inside mature FAs." Shouldn't it be written: "These results support the idea that integrin controls kindlin immobilization inside mature FAs." ?
We thank the reviewer for this correction. We changed the sentence accordingly on p.6 as followed: "Thus, our results support the idea that integrin are playing a crucial role in kindlin immobilization inside mature FAs."

-O. Rossier et al. have previously shown that talin is partly moving rearward in focal adhesions. It would be interesting to address whether it is also the case for kindlin, in order to evaluate to what extent it is coupled to the actin retrograde flow.
We thank the reviewer for this question and comments. As asked by the reviewer, we tested whether kindlin-2 is moving along the actin retrograde flow in mature focal adhesions (FAs) as demonstrated for talin. We generated super-resolved time-lapses and kymographs (Fig. 2) to measure kindlin-2 rearward flow in FAs, but also kindlin-2 immobilization durations (as done for β-integrins in the original version of the manuscript, Fig. 1i-l).
First, the durations of kindlin-2 immobilizations were much shorter than those measured for talin (Fig. 2).

As kindlin-2 and talin-1 immobilizations in FAs correspond in part to interactions with immobilized integrins, these results suggest that kindlin-integrin interactions are shorter than talin-integrin interactions.
These results suggest a low occurrence of stable tripartite integrin-kindlin-talin complexes, otherwise the immobilization durations would have been similar for kindlin and talin. These results rather support the existence of transient tripartite integrin-kindlin-talin complex, either enabling the initiation of integrin-talin interactions, or extending the duration of integrin/talin interactions.
Second, analysis of kymographs showed that the fraction of kindlin-2 undergoing retrograde flow with speed above 2 nm.s -1 was 40 % which is similar to what was measured for talin-1 (50 % in Fig. 2 "Interestingly, using super-resolved time-lapses and kymographs (Fig. 2f-h) we found that immobilization durations were much shorter for kindlin-2 compared to talin-1 (Fig. 2i) and we used these results in the discussion section to establish a model with the sequential formation of a transient tripartite integrin-kindlin-talin complex (p. 12; Fig. 7) as follows: "As the probability for three proteins to meet simultaneously with the correct orientation is extremely low, we favor a model with the sequential formation of an immobile integrin-kindlin complex followed by the formation of a transient tripartite integrin-kindlin-talin complex where kindlin could be replaced by talin head and reciprocally (Fig. 7). Indeed, as kindlin-2 and talin-1 immobilizations in FAs correspond in part to interactions with immobilized integrins, the shorter duration of kindlin-2 immobilizations suggests that kindlin-integrin interactions are shorter than talin-integrin interactions. This also suggests a low occurrence of stable tripartite integrin-kindlin- This would be a good method to represent the data, but we do not currently have a readily available approach to achieve this type of representation.
-The authors propose in the discussion that kindlin could stabilize integrin-talin interactions. This is an important question that could be addressed by determining the effect of kindlin KO on talin diffusion. This is a very interesting suggestion. However, it is impossible to obtain kindlin KO cells that will form mature FAs. Indeed, kindlin-1,2 KO cells do not spread unless the cells are treated with Mn 2+ to activate integrins. Even under these conditions, the cells will not form mature FAs but only nascent adhesions (Theodosiou et al., eLife 2016). One alternative way to answer this question would be perhaps to perform dual color Single Protein Tracking to track simultaneously kindlin and talin and study the interactions between these two proteins. We are planning to perform such type of experiments in future studies.
-It would help the reader to add a cartoon that summarizes the findings of this study.
We agree with the reviewer, we have added a cartoon summarizing our findings in Fig. 7.

REVIEWER #2 (REMARKS TO THE AUTHOR):
In this study, live-cell single-particle tracking (SPT) and super-resolution microscopy techniques were applied to study the motion characteristics of kindlin in relation to its localization to focal adhesions (FA). Based on site-specific mutations, integrin-b1 and b3 immobilization in FAs is shown to be dependent on the binding of their cytoplasmic domains to kindlin. Kindlin-1 and -2 were shown to undergo primarily lateral diffusion in the membrane plane, in contrast to Talin which enter FA largely via 3D diffusion. Kindlin-2 immobilization in FA depends on its interaction with activated integrin, but not paxillin or ILK. Kindlin-2 membrane diffusion is mediated by PH domain and interaction with phosphoinositides, and is important for nanoscale targeting to the membrane-proximal compartment. Kindlin association with the membrane is shown to be important for FA localization and enrichment, and promote cell spreading.
Overall this is a rigorous study using state-of-the-art techniques and KO cell lines established earlier by the authors. The approach gives mechanistic insight that is highly informative on the function of kindlin in cell adhesions. Data is well-presented and the experiments are well-controlled. I am supportive of its eventual acceptance to Nature Communications. However there is one important point that should be addressed.
We are grateful to the reviewer for her/his positive evaluation of our work and constructive suggestions. Furthermore, we appreciate the thorough evaluation of the reviewer #2 and we took into account his/her advices very seriously to strengthen the impact of our manuscript.
In addition to the insightful questions and interesting experiments proposed by the reviewer#2, we also performed additional experiments, suggested by the other reviewers, to improve and strengthen our manuscript.
This led to 4 modified main figures (Fig. 2 Supplementary Fig. S3), and displays membrane free diffusion both inside and outside NAs (Supplementary Fig. S3). Similarly, a kindlin-2 mutant with impaired binding to integrins (kindlin-2-QW) displayed a decreased immobile fraction in NAs compared to kindlin-2 ( Supplementary Fig. S3), suggesting that association between integrin and kindlin contributes to kindlin-2 immobilization in NAs. The results we found in NAs mirrors the ones found in mature FAs. Thus, the molecular mechanisms leading to integrin activation by kindlin-2 in FAs and NAs are probably closely related. -To further link the molecular behavior of kindlin with its function in integrin activation inside FAs, we performed additional experiments concerning the formation of FAs in kindlin-1,2 Knock Out fibroblasts (modified Fig. 7). In the original version of the manuscript we performed rescue experiments in kindlin-1,2 KO MEFs, concerning FAs formation, for kindlin-2-WT, kindlin-2-ΔPH, and kindlin-2-ΔPH-CAAX. In the revised manuscript we performed additional experiments and quantifications for kindlin-2-QW and kindlin-2-CAAX. Importantly, these new results on FAs formation follow a similar trend to the one quantified for these mutants regarding enrichment in FAs (Fig. 6), and cell spreading (Fig. 7): Kindlin-2 = Kindlin-2-CAAX > Kindlin-2-ΔPH-CAAX = Kindlin-2-QW > Kindlin-2-ΔPH. As demonstrated in our study this trend in cell spreading and FAs formation reflects the ability of kindlin-2-WT and mutants to bind and activate integrins, which is driven by membrane recruitment, membrane free-diffusion, and 3D nanoscale localization in the integrin layer inside FAs. -We performed additional experiments and quantification to increase the number of independent experiments concerning the enrichment in FAs of kindlin-2 mutants (modified Fig. 6). The results obtained confirmed what we have found in the original manuscript and follow the trend: Kindlin-2 = Kindlin-2-K390A > Kindlin-2-ΔPH-CAAX = Kindlin-2-L357A > Kindlin-2-QW > Kindlin-2-ΔPH > cytosolic mEos2. -Kindlin-2 dwell-time and rearward motion in mature FAs. Finally, we generated super-resolved timelapses and kymographs (Fig. 2) to measure kindlin-2 dwell-time in mature FAs and to test whether kindlin-2 is moving rearward inside FAs as demonstrated for talin. First, we found that immobilization durations were much shorter for kindlin-2 compared to talin-1 (Fig. 2h, i). Analysis of longer kymographs showed that the fraction of immobilized kindlin-2 undergoing retrograde flow with speed above 2 nm.s-1 was around 40 % similar to what was measured for talin-1 (Fig.  2j, k)(Rossier et al., 2012). Our results suggest that the interaction of kindlin-2 with stationary integrins is more prevalent than with F-actin, because the fraction of F-actin moving rearward above 2 nm.s-1 is around 75 %(Rossier et al., 2012). We added these new results in Fig. 2, and we mentioned these results in the results section (p.6) and the discussion section to establish a model with the sequential formation of a transient tripartite integrin-kindlin-talin complex (p.12). -We have performed all the additional quantifications that has been proposed by the reviewers (Fig. 2, 5, 6, 7 and Supplementary Fig. S3, S4, S9).
1. The Calderwood lab showed in 2014 (Huet-Calderwood et al. J Cell Sci) that ILK and Kindlin-2 interactions appeared to be important for kindlin targeting to FAs. However, this seems to contrast with the results in Fig. S3 in this study where the ILK-binding mutant of kindlin, L357A shows essentially the same behaviors as WT. In the Calderwood study, L357A kindlin-2 seems to have FA-localization defect. However, in inset of Fig. S3a, the FA localization of L357A kindlin-2 seems to be quite substantial. While the authors may have mentioned these localization differences in passing, given the extensive and rigorous works on ILK, Kindlin-2, and FAs reported earlier by the Calderwood lab, I feel that any differences on ILK/Kindlin-2 interactions observed here should be looked into or discussed at some depth in context of these earlier findings.
We thank the reviewer for this wise comment and for pointing out those differences.
In the article of Huet-Calderwood et al, JCS 2014 it was demonstrated that kindlin-2 recruitment in FAs depends on an interaction with ILK. Indeed, a point mutation in kindlin-2 (L357A) inhibits the interaction between ILK and kindlin-2 in biochemistry pull-down assays (Fig. 6D in cited reference) and also inhibits kindlin-2 recruitment in FAs, quantified using epifluorescence (Fig. 7A in cited reference). Those experiments were mainly performed in CHO cells. Note that in the Huet-Calderwood article, the authors also found that kindlin-2 could be weakly recruited in FAs in ILK Knock Out fibroblasts (Fig. 7C in cited reference). They also observed, in contrast with what they obtained in CHO cells, that GFP-kindlin-2-L357A displayed recruitment to FAs in ILK KO similar to that of GFP-kindlin-2 ( Fig. 7D in cited reference). These results suggest that kindlin-2 could also be recruited in FAs independently from its binding to ILK.
Importantly, the authors also found that there could be competition between different kindlin isoforms for their recruitment inside FAs. For instance, kindlin-3, the hematopoietic-specific kindlin, is not strongly recruited in FAs in fibroblast since it might be outcompeted by endogenous kindlin-2. However, in kindlin-2 Knock Down fibroblasts, the authors could observe recruitment of kindlin-3-GFP in FAs (Fig. 8C in cited reference). Furthermore, in Kindlin-2 Knock Down fibroblasts GFP-kindlin-2-L357A is also significantly recruited in FAs. In fact, the authors also observed a reduced expression of kindlin-2 in ILK KO, which might explain why GFP-kindlin-2-L357A could be weakly recruited in FAs in these cells.
Altogether, these results suggest that ILK binding is not indispensable for FAs targeting in the absence of competing endogenous kindlin-2. These results also show that kindlin-2 could be recruited in FAs independently from its binding to ILK, as explained by the authors in their article.
Based on the results obtained in Huet-Calderwood et al, JCS 2014, one possible explanation is that endogenous kindlin-2 expression level is low in the WT MEFs we used in our study compared to CHO cells. If this is the case, expression of exogenous kindlin-2-L357A could outcompete endogenous kindlin-2, as proposed by Huet-Calderwood and colleagues for ILK KO and in Kindlin-2 Knock Down fibroblasts.
To test this hypothesis, we performed western blots to compare the expression levels of endogenous kindlin-2 in the MEFs we used and in CHO cells. However, the expression level of endogenous kindlin-2 in our MEFs was about 2 fold the level found in CHO. Thus similar FAs recruitment and diffusive behavior found for kindlin-2-L357A and kindlin-2-WT in our experiments could not be explained solely by a low expression of endogenous kindlin-2.
Transfected kindlin-2-L357A might also outcompete endogenous kindlin-2. We also performed western blots to compare the expression levels of endogenous kindlin-2 and transfected exogenous mEos2kindlin-2 and mEos2-kindlin-2-L357A in MEFs. We used exactly the same condition of transfection than the one used for sptPALM experiments and FAs recruitment experiments. We found a 1.5-fold higher expression of transfected mEos2-kindlin-2 and mEos2-kindlin-2-L357A compared to the endogenous kindlin-2. The higher level of mEos2-kindlin-2-L357A compared to endogenous kindlin-2 might thus partly explain why kindlin-2-L357A behaves similarly than kindlin-2. However, the level of overexpression of exogenous mEos2-kindlin-2-L357A is not much higher than that of endogenous kindlin-2, so the transfected mEos2-kindlin-2-L357A could not compete with endogenous kindlin-2 to a level that accounts for the small effects found in our experiments.
If ILK binding was indispensable for kindlin-2 recruitment in FAs and immobilization in FAs, we would have expected stronger effects in sptPALM experiments and FAs recruitment experiments with mEos2-kindlin-2-L357A as found for example for the kindlin-2 mutant with decreased binding to integrins kindlin-2-QW 12 (Fig. 3, 6 and Supplementary Fig. S8f,g). Thus as the authors of the Huet-Calderwood's article explained in their discussion, the recruitment and function of kindlin-2 in FAs could also be independent from ILK binding, for instance via integrins, actin (Bledzka JCB 2016), paxillin (Theodosiou eLife 2016) or other unknown binding partners. Furthermore, the ILK-dependent recruitment of kindlin-2 in FAs seems to depend on the cellular context. Here is a passage taken from the discussion of the Huet-Calderwood's article commenting these seemingly contradictory results: "Nonetheless, at least in some cells, ILK binding is not absolutely required and the residual FA localization of kindlin-2 in ILK-knockout cells is likely due to the integrin binding site in the F3 subdomain. Indeed, in the absence of competing endogenous kindlin-2, ILK-binding-defective kindlins can target to FAs, providing that their integrin-binding site is intact." In the results section (p. 6) of the our revised manuscript, we now comment further the results obtained with the ILK binding defective kindlin-2 mutant kindlin-2-L357A as followed: Fig. S4), which indicates that in our MEFs ILK-independent recruitment of kindlin is predominant. Thus, our results support the idea that integrin are playing a crucial role in kindlin immobilization inside mature FAs."

Minor points:
-It would be helpful to readers to end with a graphical illustration that summarize the findings.
We agree with the reviewer and have added a graphical illustration summarizing our findings in Fig. 7.
-3D is probably a more common and appropriate term that 'tridimensional' used by the authors We replaced "tridimensional" by 3D trough out the manuscript, but not in the title.

REVIEWER #3 (REMARKS TO THE AUTHOR):
The manuscript by Orré et al present a set of elegant data using advanced quantitative bioimaging approaches that unravelled the mobility behavior and nanoscale threedimensional position of the protein kindlin-2 in the context of integrin activation at focal adhesions. Mobility within the plasma membrane as well as vertical positioning of kindlin molecules within single focal adhesions are important parameters to understand how kindlin immobilizes integrins and contribute to their activation in a manner different from talin. Since kindlin and talin are known to have a complementary action during integrin activation, this manuscript provides interesting biophysical insights into the role of kindlin: while talin comes to the FAs directly from the cytosol, kindlin is already diffusing within the plasma membrane via its PH domain before reaching the FAs.
The manuscript is very well-written, the figures are rich but very clear, the methodology is elegant and thorough and appeals to a broad readership and the statements provided are sufficiently novel and link mobility to nanoscale 3D localization of an important integrin activator. Therefore, in my view, this study deserves publication. However, there are a number of points the authors must first address in a revised manuscript.
We are grateful to the reviewer #3 (Alessandra Cambi) for her positive evaluation of our work and constructive suggestions. Furthermore, we appreciate the thorough evaluation of Alessandra Cambi and we took into account her advices very seriously to strengthen the impact of our manuscript.
In addition to the insightful questions and interesting experiments proposed by Alessandra Cambi, we also performed additional experiments, suggested by the other reviewers, to improve and strengthen our manuscript.
This led to 4 modified main figures (Fig. 2 The main results are summarized below: -As found in mature focal adhesions (FAs), kindlin-2 is immobile and enriched in early nascent adhesions (NAs) ( Supplementary Fig. S3), and displays membrane free diffusion both inside and outside NAs ( Supplementary Fig. S3). Similarly, a kindlin-2 mutant with impaired binding to integrins (kindlin-2-QW) displayed a decreased immobile fraction in NAs compared to kindlin-2 ( Supplementary Fig. S3), suggesting that association between integrin and kindlin contributes to kindlin-2 immobilization in NAs. The results we found in NAs mirrors the ones found in mature FAs. Thus, the molecular mechanisms leading to integrin activation by kindlin-2 in FAs and NAs are probably closely related. -To further link the molecular behavior of kindlin with its function in integrin activation inside FAs, we performed additional experiments concerning the formation of FAs in kindlin-1,2 Knock Out fibroblasts (modified Fig. 7). In the original version of the manuscript we performed rescue experiments in kindlin-1,2 KO MEFs, concerning FAs formation, for kindlin-2-WT, kindlin-2-ΔPH, and kindlin-2-ΔPH-CAAX. In the revised manuscript we performed additional experiments and quantifications for kindlin-2-QW and kindlin-2-CAAX. Importantly, these new results on FAs formation follow a similar trend to the one quantified for these mutants regarding enrichment in FAs (Fig. 6), and cell spreading (Fig. 7): Kindlin-2 = Kindlin-2-CAAX > Kindlin-2-ΔPH-CAAX = Kindlin-2-QW > Kindlin-2-ΔPH. As demonstrated in our study this trend in cell spreading and FAs formation reflects the ability of kindlin-2-WT and mutants to bind and activate integrins, which is driven by membrane recruitment, membrane free-diffusion, and 3D nanoscale localization in the integrin layer inside FAs.
-Kindlin-2 dwell-time and rearward motion in mature FAs. Finally, we generated super-resolved timelapses and kymographs (Fig. 2) to measure kindlin-2 dwell-time in mature FAs and to test whether kindlin-2 is moving rearward inside FAs as demonstrated for talin. First, we found that immobilization durations were much shorter for kindlin-2 compared to talin-1 (Fig. 2h, i). Analysis of longer kymographs showed that the fraction of immobilized kindlin-2 undergoing retrograde flow with speed above 2 nm.s-1 was around 40 % similar to what was measured for talin-1 (Fig.  2j, k) (Rossier et al., 2012). Our results suggest that the interaction of kindlin-2 with stationary integrins is more prevalent than with F-actin, because the fraction of F-actin moving rearward above 2 nm.s-1 is around 75 % (Rossier et al., 2012). We added these new results in Fig. 2, and we mentioned these results in the results section (p.6) and the discussion section to establish a model with the sequential formation of a transient tripartite integrin-kindlin-talin complex (p.12). -We have performed all the additional quantifications that has been proposed by the reviewers (Fig. 2, 5, 6, 7 and Supplementary Fig. S3, S4, S9).
1) the authors show that beta1 and beta3 integrins seem differently sensitive to impairment of the interaction with kindlin in Figure 1. It is not clear how useful and important these data are for the remaining of the story as they are not discussed further.
We agree with the reviewer, we should have included in the discussion a section about the differences we found between 131-and 133-integrins. This is particularly interesting since different integrin classes are present in the same mature FAs and use distinct mechano-transduction and signaling pathways that cooperate to control FAs structure and functions, such as migration, rigidity sensing and signaling.
In a previous study, we demonstrated a horizontal nano-partitioning and nanoscopic dynamics specific to each integrin α5131 and αv133 within the FAs; 133-class integrins are stationary and enriched within FAs, whereas 131-class integrins are less enriched and display rearward movements (Rossier et al., NCB 2012). Our results indicated that specific classes of α/13 integrins (α5131 and αv133) act as distinct 'nanoscale adhesion units' within an individual FA with specific dynamics, organization and force transmission of F-actin motion to the ECM (fibronectin). Thus, nano-scale partitioning of proteins inside could induce functional partitioning that will control the assembly and the mechanical functions of integrin adhesions.
In line with this hypothesis, the Fässler group demonstrated using genetically engineered cells and quantitative proteomics that specific αv-and 131-class integrins use distinct mechano-transduction and signaling pathways that cooperate to control adhesion site assembly, F-actin organization and ECM rigidity sensing (Schiller et al., NCB 2013). Importantly, they showed that αv-class integrins link GEF-H1/RhoA/mDia to stress fiber formation, while 131-class integrins link kindlin-2 and the IPP complex to myosin II activation. These results are in line with the stronger dependence on kindlin-2 we found for 131-integrin activation (α5131 in our MEFs) compared to the activation of 133-integrin (αv133 in our MEFs).
Furthermore, the use of fibronectin micropatterned substrates emphasized the distinct localization of αv-or 131-class integrins in FAs, probably induced by different responses to forces (Schiller et al., NCB 2013). Inhomogeneous localization of αv-or 131-class integrins implies that also their associated signaling proteins are probably segregated. Altogether these findings suggest that nanoscale-partitioning of integrins in FAs reflects a segregation of specific signaling properties and signaling tasks that cooperate to determine the functions of FAs.
In addition, despite the fact that αv133 and α5131 are both connecting the ECM (i.e. fibronectin) to actin, their abilities to sense and transmit forces are distinct. Several reports have shown that increased generation of extracellular forces causes strengthening of the αv133 integrin/F-actin linkages and thus rigidity sensing, while the strength of the α5131 integrin/F-actin connection seems unaffected by force generation (Giannone, JCB 2003;Choquet Cell 1997;Roca-Cusachs, PNAS 2009 To highlight the differences between 131-and 133-integrins in mature FAs and the potential selectivity of kindlin-2 action on 131-integrins we added the following paragraph in the discussion:  (Giannone, JCB 2003;Choquet Cell 1997;Roca-Cusachs, PNAS 2009). On the extracellular side, α581 integrin binding to FN is stabilized in a force-dependent conformational transition (Friedland, Science 2009;Kong, JCB 2009). This could explain why α581 integrin/ECM bonds generate and resist to higher forces compared to αv83 integrin/ECM bonds (Schiller, NCB 2013;Roca-Cusachs, PNAS 2009). An interesting model could be that the activation of 81-integrin is mainly triggered by biochemical reactions such as kindlin-2 binding while the activation of αv83-integrin could be mainly promoted by force generation via talin."

"Integrin activation by kindlin-2 in FAs is integrin selective
2) a number Just as a reminder, in our manuscript, using single protein tracking, super-resolution microscopy and functional assays, we established the link between the function of a critical integrin activator, namely kindlin-2, and its molecular behavior and 3D nanoscale localization. In particular, we showed that kindlin efficient interaction with integrins inside FAs could not result from its cytosolic diffusion alone, but requires kindlin membrane diffusion to drive kindlin in the proper FAs functional layer. Thus our study reveals that the molecular dynamical behavior of a protein in a subcellular compartment is directly linked to its function.
We thank the reviewer to have asked for these additional experiments and quantifications, since these new additional results obtained strengthen the message concerning the link between the function of kindlin-2 and its molecular behavior inside and outside FAs.
The reviewer is correct, at the exception of Fig. 7 and Fig. S8 and S9, we performed experiments in wildtype MEFs transfected with kindlin-2 mutants. The reason for this is, as pointed out by the reviewer, to study the diffusive behavior of kindlin-2 mutants in the context of "normal" FAs. The other reason was to perform experiments in the same MEFs for β1-integrin and β3-integrin mutants and kindlin-2 mutants. Note that the MEFs used in the current manuscript, are the same than the one used in Rossier et al., Nature Cell Biology 2012.
In experiments performed in WT MEFs expressing the kindlin-2 mutants (QW, L357A, K390A, ΔPH), we did not observe obvious effects on FAs number and morphology. For example, we observed no obvious increased difficulty in finding MEFs with mature FAs when expressing kindlin-2-ΔPH, which is the kindlin-2 mutant with the largest deficit in FAs formation in rescue experiments using Kindlin-1,2 KO MEFs. This suggests that the function of endogenous kindlin-1 and kindlin-2 on FAs formation is not altered by dominant negative effects triggered by the expression of exogenous kindlin-2 mutants. Note that the level of expression of transfected kindlin-2 mutants is about 1.5 fold higher than the endogenous kindlin-2 ( Supplementary Fig. S4). We did not quantify FAs number or morphology in WT MEFs transfected with kindlin-2 mutants, to avoid measurements that would be biased by the presence of the endogenous kindlin-2 and kindlin-1. Instead, we performed those experiments and quantifications in kindlin-1,2 KO MEFs. To link the diffusive behavior of kindlin-2 wild-type and mutants (kindlin-2-WT, kindlin-2-ΔPH, kindlin-2-QW, kindlin-2-L357A) with their functions, we confirmed that their diffusive behaviors were identical in WT MEFs and kindlin-1,2 KO MEFs ( Supplementary Fig. S8). Then we quantified cell spreading, FAs numbers and FAs morphology ( Fig. 7 and Supplementary Fig. S9).
In the original version of the manuscript we performed rescue experiments in kindlin-1,2 KO MEFs concerning FAs formation for kindlin-2-WT, kindlin-2-ΔPH, and kindlin-2-ΔPH-CAAX. We found that expression of kindlin-2-WT rescue the formation of FAs. The stronger impairment in rescuing FAs formation was found for kindlin-2-ΔPH, which displays impaired membrane free diffusion. Supporting our hypothesis that kindlin-2 membrane diffusion is crucial for kindlin-2 FAs recruitment and functions in FAs, we found a partial rescue of FAs formation by restoring membrane free-diffusion with kindlin-2-ΔPH-CAAX.
In the revised manuscript we performed those additional experiments on FAs formations and the associated quantifications for kindlin-2-QW and kindlin-2-CAAX.
As shown in the original version of the manuscript, kindlin-2-QW displayed a decreased immobile fraction and increased free-diffusive fraction in FAs as compared to kindlin-2-WT (Fig. 3b,d), and kindlin-2-QW enrichment in FAs was decreased compared to kindlin-2-WT (Fig. 6b, f). Thus, it was important to include this mutant in the analysis of FAs formation, to link its diffusive behavior to its function in FAs. As for kindlin-2-CAAX, it is a control which shows that adding a CAAX membrane anchor to WT kindlin-2 do not trigger an enhanced activity of kindlin-2-WT during cell spreading. Using the same CAAX membrane anchor to restore kindlin-2-ΔPH membrane free-diffusion and localization in the integrin layer increases its recruitment to FAs, and its function during cell spreading and FAs formation. However, we did not perform these experiments with kindlin-2-L357A and kindlin-2-K390A which have no obvious effects on kindlin-2 diffusive behavior (kindlin-2-L357A: Supplementary Fig. S4 (MEFs) and S8 (KindKO cells); kindlin-2-K390A: Supplementary Fig. S5), membrane recruitment (Fig. 4f) and FAs enrichment (Fig. 6) compared to kindlin-2-ΔPH and kindlin-2-QW.
Our new results show that expression of kindlin-2-QW rescue the formation of FAs at the same level as kindlin-2-ΔPH-CAAX ( Fig. 7 and Supplementary Fig. S9). These results confirm that the kindlin-2-QW mutant can still bind integrins (as found in sptPALM experiments; Fig. 3, as discussed in the manuscript) but also show that kindlin-2-QW mutant could also partially activate integrins. On its hand, kindlin-2-CAAX rescue the formation of FAs to the same level as kindlin-2, demonstrating that the addition of a CAAX membrane anchor to WT kindlin-2 do not trigger an enhanced activity of kindlin-2-WT during FA formation ( Fig. 7 and Supplementary Fig. S9).
3) could the authors explain why K2-DeltaPH is immobile outside FAs?
Deletion of the entire PH domain of kindlin-2 (kindlin-2-ΔPH) strongly inhibited membrane free diffusion, both inside and outside FAs (Fig. 4a,c,d, Supplementary Movie S2), confining most of kindlin-2-ΔPH freediffusion into the cytosol (Fig. 4f). A protein freely diffusing in the cytosol is characterized by coefficient of diffusion (D) around 100 times faster than the one of a protein freely diffusing on the membrane. With our acquisition parameters (50 Hz) we are not able to reconnect trajectories of proteins freely diffusing in the cytosol with very fast D. Thus with our method we are not able to reconnect trajectories of a protein freely diffusing in the cytosol and thus to measure its fast D.
Instead, our method allows us to record the trajectories of a cytosolic protein that freely diffuses on the membrane if this cytosolic protein interacts with membrane components that display free-diffusion (for instance, kindlin-2 ( Fig. 2), PH-domain (Fig. 4), CAAX (Rossier et al., NCB 2012), kindlin-2-CAAX (Fig. S7)). Similarly, our method allows us to reconnect the trajectories of a cytosolic protein if this protein is associated with immobilized membrane components, such as proteins binding to immobile integrins inside and outside FAs, like for instance paxillin (Fig. S6) or talin (Fig. 2).
In the case of kindlin-2-ΔPH, the fraction of kindlin-2-ΔPH freely diffusing in the cytosol with very fast D is missing in the distribution of D both inside and outside FAs for the reason mentioned above. However, since kindlin-2-ΔPH lacks membrane free-diffusion, kindlin-2-ΔPH interacting transiently with immobile binding partners outside or inside FAs represents the largest fraction at the membrane level. This explains why kindlin-2-ΔPH appears mainly immobile outside FAs (Fig. 4).
We demonstrated that the PH domain of kindlin-2 is not only driving kindlin-2 membrane recruitment and free-diffusion (Fig. 4), but is also leading to the subsequent interaction with integrins ( Fig. 6, 7). Perhaps the immobile fraction of kindlin-2-ΔPH corresponds to residual interactions with integrins occurring both inside and outside FAs. Alternatively, kindlin-2-ΔPH could bind to other proteins, actin, ILK or still unknown binding partners (Bledzka JCB 2016;Huet-Calderwood JCS 2014).
To clarify this point, we added in the method section a paragraph explaining this bias of the SPT method, page 17: "With the acquisition frequency and TIRF illumination used in our experiments, it is impossible to reconnect trajectories lasting at least 260 ms (≥13 points) for a protein freely diffusing in 3D within the cytosol. Therefore, the very low occurrence of free-diffusing trajectories inside or outside FAs for kindlin-2-ΔPH and paxillin indicates that these proteins are mostly moving in the cytosol." Kindlin-2 and kindlin-2-ΔPH displays similar immobile fractions inside FAs (Fig. 4) but the density of immobilization is clearly decreased for kindlin-2-ΔPH compared to kindlin-2 ( Fig. 2a vs Fig. 4a) reflecting a decreased enrichment in FAs for kindlin-2-ΔPH (Fig. 6f). Outside FAs, since kindlin-2-ΔPH is mainly freely diffusing in the cytosol, the dominant apparent diffusive behavior is immobilization as explained above, but the density of immobilization outside FAs is very low.
This answer is also related to the question asked in the point#5 (see below). We added these distributions of coefficient of diffusion again in the Supplementary Fig. S1, to help appreciate the differential effects of Mn 2+ treatment on 131-integrin and 133-integrin activation. These data could be removed if the reviewer thinks that they are not necessary. 5) why is beta3 wild-type integrin immobile outside FAs (Fig 1h)? Why is talin so highly immobile outside FAs (Fig 2d)? Why is 70% of paxillin immobile outside FAs (Fig S5c)?

4) data in
Those 3 queries are all related to the point #3.
Altogether, these results indicate that these immobilizations outside FAs are associated with specific integrin immobilizations and not aspecific immobilizations. They might be triggered by transient activation of integrins occurring outside mature FAs that are required for the initiation of early integrin adhesions, leading to the formation of nascent adhesions (NAs). Actually, reviewer#1 asked us to perform sptPALM experiments on kindlin-2 during the formation of early integrin-adhesions, which is occurring in the lamellipodium. We performed these experiments with kindlin-2-WT and kindlin-2-QW with decreased binding to integrins. These results are presented in Supplementary Fig. S3. They show a significant fraction of immobilized kindlin-2, inside and outside the lamellipodium. Again kindlin immobilizations are also decreased by mutations which reduce binding to integrin. Thus transient immobilizations of integrin adhesion proteins outside mature FAs could correspond to transient activation of integrins, reflecting a basal level of integrin activation. These transient activations could lead to initiation of nascent adhesion in the lamellipodium.
Second, concerning the large fraction of immobilization for talin and paxillin outside FAs: We want to emphasize that the immobilization fraction is not always correlated with the immobilization density or recruitment within a subcellular structure. For example, it is possible to obtain similar immobilization fractions but with very different immobilization densities. For the same area, it is possible to obtain 10 stationary and 10 free-diffusive trajectories, or 100 stationary and 100 free-diffusive trajectories, the immobilization fraction (50%) will be the same but the density will be an order of magnitude different.
As noticed by the reviewer, this is the case for talin-1, which display similar levels of immobilization inside versus outside FAs (82% vs 73%, Fig. 2), but with much higher density of immobilization inside versus outside FAs (~ 5 time more detections inside versus outside FAs, Fig. 4f in Rossier et al., Nature Cell Biology 2012). This is also the case for paxillin which possesses large immobile fractions inside and outside FAs but is specifically enriched in FAs ( Supplementary Fig. S6). Related to the first part of the question, we think that paxillin and talin immobilizations outside mature FAs could be associated with transient activation of integrins that could occur outside mature FAs, as a basal level of integrin activation that could lead to the initiation of nascent adhesions.
There is a clear difference in SPT experiments between proteins that a freely diffusing on the plasma membrane (transmembrane proteins, and cytosolic proteins which interact with diffusive components of the membrane): integrins, kindlins, CAAX-tagged proteins, Rho GTPases (Mehidi Current Biology 2019); and cytosolic proteins that do not associate with diffusive components of the membrane: talin, paxillin, kindlin-2-ΔPH...
For proteins freely diffusing on the plasma membrane, a decreased immobilization fraction in conjunction with an increased fraction of membrane free-diffusion reflects a transfer in population from an immobile pool engaged with binding partners to a freely diffusing pool detached from this binding partner. Thus, there is a transfer between immobile and freely diffusive fractions in conjunction with a decreased density of immobilization. This is the case in FAs, for instance for kindlin-2 versus kindlin-2-QW ( Fig. 3 in MEFs and Supplementary Fig. S8 in Kind Ko cell), or 31-integrin versus 31-integrin-Y795A (Fig. 1).
For cytosolic proteins that do not associate with diffusive components of the membrane, a decreased immobilization in a specific sub-cellular compartment such as FAs, mainly lead to a decreased density of immobilization. This does not automatically change the fraction of immobilization, since there is no transfer from immobilization to free diffusion on the membrane, but transfer to free-diffusion in the cytosol that as explained earlier cannot be detected in SPT experiments.
6) the Diff coefficients are calculated from the small fraction that is mobile, which is very small inside FAs, at the same time the changes in D values are also very very modest: they are statistically different, but what do these tiny changes tell biologically? Could the authors strengthen their interpretation of these data?
Indeed, the coefficient of diffusion (D) for free-diffusive events (Ddiff) are calculated from the fraction of freely diffusive molecules. These fractions could be small especially in mature FAs.
First, we want to emphasize that in most cases there are huge differences in Ddiff when compared between inside and outside FAs. Although, changes in Ddiff in mature FAs might be small when comparing 20 different conditions, they are statistically different as pointed out by the reviewer. For instance, the Ddiff in FAs for 33-integrin and 31-integrin are significantly slower compared to integrin mutants defective in binding to kindlin/talin (33-integrin-Y759A/33-integrin-Y747A; 31-integrin-Y795A/31-integrin-Y783A) ( Supplementary Fig. S1). We found similar results in a previous study for integrin mutants defective in binding to fibronectin/talin (Rossier et al., NCB 2012: Fig. 2g). Similarly, the Ddiff in FAs for kindlin-2 is slightly but significantly slower compared to the kindlin-2-QW mutant defective in binding to integrins (Fig. 3e). Similarly, the Ddiff in FAs for kindlin-2 is clearly slower compared to kindlin-ΔPH-CAAX ( Supplementary Fig. S7d), in line with the faster D diff found outside FAs for kindlin-ΔPH-CAAX compared to kindlin-2 ( Supplementary Fig. S7d).
Importantly, we also found significant differences in the Ddiff inside and outside FAs for integrins and kindlins ( Supplementary Fig. S1 and Fig. 2), Ddiff for kindlin being slower than for integrins (Fig. 2). These results suggest that kindlin-2 and integrins are not diffusing at the plasma membrane when bound to each other. This is consistent with a model where kindlin-2 is entering FAs without being co-associated with integrins outside FAs, moving within FAs using membrane free diffusion to reach integrins and trigger their immobilization inside FAs.
We think that Ddiff correspond to the genuine D of free-diffusion for a specific protein, but also conceal transient interactions of this protein with surrounding binding partners that could not be captured because of limited acquisition frequencies inherent to any experiments. We explain this below: Since it is impossible to obtain an infinitely fast acquisition frequency, we are most of the time in SPT quantifying apparent coefficient of free-diffusion (D) and not absolute D. If we take integrin as an example: in principle integrin-WT in its activated or inactivated states, when not interacting with other proteins (fibronectin, talin, kindlin) or specific membrane domains, should diffuse at the same speed (identical D). In other words, the diffusive behavior will be determined by the physical parameters of the protein and the environment in which the protein is evolving (hydrodynamic radius of the protein, viscous drag of the cytosol and the membrane, collision within the membrane...). Now, if we introduce specific binding events in between periods of free-diffusion, but if those binding events are most of the time lasting less than what could be temporally resolved in the experimental and analysis framework, the studied protein will still be identified as a freely diffusive but will be characterized by a decreased rate of free-diffusion, i.e. Dapp will be inferior compared to D absolute. The Mean Squared Displacement (MSD) will remain linear but with a lower slope. The disparity in the Ddiff values between integrin-WT and integrin-33-D119Y (Fig. 2 NCB 2012) or integrin-33-Y747A ( Fig. 2 NCB 2012, or Supplementary Fig. S1 in the present manuscript) suggests that integrin-WT undergoes additional transient interactions.
In our experimental conditions, since we use a minimum of 10 points of the MSD to analysis the diffusive behavior of proteins, we are able to capture immobilization events lasting more than 200 ms at 50 Hz (10 x 20 ms) (see methods). In future studies, it will be interesting to increase the frequency of acquisitions of sptPALM experiments to investigate these much transient interactions between integrins and regulators. For instance, we could perform sptPALM experiments at 333 Hz, enabling to capture immobilization events of 30 ms, as we did in a study focused on Rac1 immobilizations at the tip of the lamellipodium of migrating cells (Mehidi et al., Current Biology 2019). 21 7) Figure 6: I appreciate the complexity of the imaging approaches used in this study but presenting data in a main figure that are obtained by two or one experiment is not sufficiently sound. In Fig. S6, four experiments are mentioned in which the same mutants used for figure 6 were used. I therefore assume the authors have the possibility to add more experimental data and strengthen the results shown in Fig.  6. In addition, I am wondering why the information provided in Fig. 6 is actually not shown and discussed earlier in the manuscript, right before the authors embark on sptPALM experiments? What is the reason for not putting these data as supplementary figure?

From point #8:
Organizing the manuscript as proposed by the reviewer (functional effects on the formation of mature FAs followed by a molecular understanding of what drive these differences) is an interesting alternative to what we chose in the original version of the manuscript (mechanistic to functional). However, we think that both organizations are valid, and we would like to keep the original organization of the manuscript.
Moreover we think that the data displayed in Fig. 6 constitutes a strong argument for the role of membrane diffusion on the function of kindlin-2 as it establishes a clear correlation between kindlin-2 membrane recruitment and accumulation inside FAs.
8) the data provided in Figure 7 are used a conclusive statement to link the biophysical parameters to integrin activation (cell spreading). However, these data are not entirely novel as the eLife paper of the Faessler group already provided the same knowledge. I am therefore wondering: why not using these data as starting point? As motivation to better understand mechanistically what drives these differences? Mobility and nanoscale 3D localization of kindlin would then be the mechanistic explanation. In Fig S8, interesting data are shown for individual FAs formed after expression of two out of the five kindlin-2 mutants used. I think the same parameters should be provided for the QW, L357A and K390A mutants, to link the effects of these mutations to individual FA properties and eventually to cell spreading. Unfortunately, Fig S8 seems not mentioned in the Results section.
The 2016 eLife article from the Fässler group "Theodosiou et al., Kindlin-2 cooperates with talin to activate integrins and induces cell spreading by directly binding paxillin", is focused mainly on the ability of kindlin to activate integrins during cell spreading and thus the formation of early adhesive structures including Nascent Adhesions (NAs). In contrast, in our current manuscript, we are studying the function of kindlin during integrin activation in mature FAs. Nevertheless, we agree with the reviewer that part of our results concerning cell spreading in Kindlin1,2 KO MEFs rescued with expression of wild-type or mutant forms of kindlin-2 may overlap with some results obtained in this eLife article: this mainly concern experiments performed using kindlin-2-WT and kindlin-2-ΔPH ( Fig. 4F in Theodosiou et al.).
However, the experiments performed and the results found are not exactly the same. In the eLife article the authors focused on cells displaying isotropic spreading rapidly after plating (10-30 min.). This isotropic spreading behavior is mainly triggered in Talin-KO fibroblasts after Mn 2+ integrin activation. In the case of isotropic spreading, the shape of cells is not dependent on the formation of mature FAs but depends on