Caveolae couple mechanical stress to integrin recycling and activation
Abstract
Cells are subjected to multiple mechanical inputs throughout their lives. Their ability to detect these environmental cues is called mechanosensing, a process in which integrins play an important role. During cellular mechanosensing, plasma membrane (PM) tension is adjusted to mechanical stress through the buffering action of caveolae; however, little is known about the role of caveolae in early integrin mechanosensing regulation. Here, we show that Cav1KO fibroblasts increase adhesion to FN-coated beads when pulled with magnetic tweezers, as compared to wild type fibroblasts. This phenotype is Rho-independent and mainly derived from increased active β1-integrin content on the surface of Cav1KO fibroblasts. Florescence recovery after photobleaching analysis and endocytosis/recycling assays revealed that active β1-integrin is mostly endocytosed through the clathrin independent carrier/glycosylphosphatidyl inositol (GPI)-enriched endocytic compartment pathway and is more rapidly recycled to the PM in Cav1KO fibroblasts, in a Rab4 and PM tension-dependent manner. Moreover, the threshold for PM tension-driven β1-integrin activation is lower in Cav1KO mouse embryonic fibroblasts (MEFs) than in wild type MEFs, through a mechanism dependent on talin activity. Our findings suggest that caveolae couple mechanical stress to integrin cycling and activation, thereby regulating the early steps of the cellular mechanosensing response.
Editor's evaluation
This valuable cell biological study uses magnetic tweezers to explore how integrins and caveolae interact to regulate mechanosensing. The authors describe a convincing link between the presence of caveolae and the trafficking of integrins between the cell surface and intracellular compartments to control plasma membrane tension.
https://doi.org/10.7554/eLife.82348.sa0eLife digest
Cells can physically sense their immediate environment by pulling and pushing through integrins, a type of proteins which connects the inside and outside of a cell by being studded through the cellular membrane. This sensing role can only be performed when integrins are in an active state.
Two main mechanisms regulate the relative amount of active integrins: one controls the activation of the proteins already at the cell surface; the other, known as recycling, impacts how many new integrins are delivered to the membrane. Both processes are affected by changes in cell membrane tension, which is itself controlled by dimples (or ‘caveolae’ – little caves in Latin) present in the cell surface. Caveolae limit acute changes in tension by taking in (pinching off the dimples) or releasing (dimples flattening) segments of the membrane. However, it is still unclear how integrins and caveolae mechanically interact to regulate the ability for a cell to read its environment.
To understand this process, Lolo et al. focused on mouse cells genetically manipulated to not build caveolae on their surfaces, and which cannot properly sense mechanical changes in their surroundings. These were exposed to beads covered in an integrin-binding protein and manipulated using magnetic tweezers. The manipulation showed that mutated cells bound to the beads more strongly than non-modified cells, indicating that they had more active integrins on their surface. This change was due to both an accelerated recycling mechanism (which resulted in more integrin being brought at the surface) and an increase in integrin activation (which was triggered by a higher membrane tension). Caveolae therefore couple mechanical inputs to integrin recycling and activation.
Healthy tissues rely on cells correctly sensing physical changes in their environment so they can mount an appropriate response. This ability, for example, is altered in cancerous cells which start to form tumours. The findings by Lolo et al. bring together physics and biology to provide new insights into the potential mechanisms causing such impairments.
Introduction
Cells constantly adjust their plasma membrane (PM) composition in response to changes in extracellular matrix (ECM) stiffness, which regulates many aspects of cell behavior (Wells, 2008). How cells sense and react to ECM stiffness changes is critical to understanding both physiological and pathological processes (Handorf et al., 2015). The ECM is mechanically linked to the cytoskeleton through integrins, which provide a bridge between extracellular cues and downstream cellular events (Schwartz, 2010). The mechanosensing function of integrins is well established, and much progress has been made in defining the molecular details of integrin action; for example, how α5β1 and αvβ3 integrins withstand and detect forces, respectively (Roca-Cusachs et al., 2009), and how differences in integrin bond dynamics contribute to tissue rigidity sensing (Elosegui-Artola et al., 2014). Integrin function requires an activation step commonly achieved by binding to activator molecules, such as talin and kindlins (Moser et al., 2009). Integrins are also activated in response to changes in membrane tension, as recently reported by Ferraris et al., 2014 and Böttcher and Fässler, 2014. However, it is unclear how these events influence mechanosensing.
Tissues experiencing wide variations in PM tension, such as endothelium, muscle, fibroblasts, and adipocytes, have a high-membrane density of caveolae (Sinha et al., 2011; Echarri and Del Pozo, 2015; Parton and del Pozo, 2013). Caveolae are 60–80 nm PM invaginations and are key elements in the sensing and transduction of mechanical forces. Core caveolae protein components are caveolin-1 (Cav1) and cavin-1 (also called Polymerase I and transcript release factor, PTRF), and lack of either results in caveolae loss (Rothberg et al., 1992; Hill et al., 2008). Cav1 is intimately linked to integrins (Parton and del Pozo, 2013 del Pozo et al., 2005; Del Pozo and Schwartz, 2007) and to molecules such as Filamin A, which links integrins to the cytoskeleton (Muriel et al., 2011). To study how integrins and caveolae interact to regulate cell mechanosensing, we used magnetic tweezers (MT; Tanase et al., 2007) to exert mechanical force on mouse embryonic fibroblasts (MEFs) upon their binding to magnetic beads coated with the integrin-binding protein fibronectin (FN). MEFs lacking Cav1 (Cav1KO) showed higher adhesion to FN-coated beads than Cav1WT MEFs, through a process dependent on increased β1-integrin surface availability in Cav1KO MEFs due to both rapid recycling of endocytosed β1-integrin to the PM, in a Rab4 and PM tension-dependent manner, and tension-mediated β1-integrin activation driven by Talin. These results support a role for caveolae in regulating integrin mechanosensing by coupling membrane tension to integrin recycling and activation.
Results
Genetic models for studying the role of caveolae in integrin mechanosensing
To study the role of caveolae in integrin mechanosensing, we generated transgenic fibroblast lines by transducing Cav1KO MEFs with either Cav1- or PTRF-expressing retroviral vectors (Figure 1A–D).

Caveolin-1-based genetic model characterization.
(A–D) Mouse embryonic fibroblast (MEF) caveolae-related phenotypes: PTRF is depicted in dark blue, Cav1 in red, and plasma membrane (PM) in light blue. (E–H) Electron microscopy images of MEF PM regions, showing the presence of caveolae only in (E) wild type MEFs and (H) Cav1-reconstituted Cav1KO MEFs (black arrows). (I–L) Ground State Depletion (GSD)-super-resolution MEF images. Cav1 is shown in red. Scale bar = 1 µm. (M) Biochemical PM fractionation of wild type and Cav1KO MEFs and Cav1KO MEFs reconstituted with PTRF, empty vector, and Cav1. Samples were immunoblotted for Cav1 and transferrin receptor (as both PM marker and loading control). (N) Western blot of total lysates from wild type, Cav1KO, and reconstituted Cav1KO MEFs. Samples were immunoblotted for PTRF, Cav1, and tubulin (loading control).
-
Figure 1—source data 1
Full raw unedited blot corresponding to Figure 1M.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig1-data1-v3.pdf
-
Figure 1—source data 2
Uncropped blot with the relevant bands labeled, corresponding to Figure 1M.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig1-data2-v3.pdf
-
Figure 1—source data 3
Full raw unedited blot corresponding to Figure 1N.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig1-data3-v3.pdf
-
Figure 1—source data 4
Uncropped blot with the relevant bands labeled, corresponding to Figure 1N.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig1-data4-v3.pdf
To validate the transgenic lines for the characterization of mechanosensing properties, we first analyzed the presence or absence of caveolae by electron microscopy (EM), detecting caveolae only in wild type (WT) and Cav1-reconstituted Cav1KO cells (Figure 1E–H, black arrows). Next, super-resolution microscopy analysis of Cav1 topographical distribution revealed that re-expressed Cav1 localizes to the PM and does not form aberrant aggregates (Figure 1I–L). Confirming the super-resolution imaging data, biochemical fractionation indicated that re-expressed Cav1 localizes to the PM (Figure 1M). Western blot analysis confirmed the expected Cav1 and PTRF expression in the different MEF lines and revealed near-endogenous expression levels in Cav1-reconstituted cells (Figure 1N, see also Figure 1—source data 1–4).
Cav1 modulates integrin/cytoskeleton-dependent mechanosensing via Rho-independent mechanisms
To monitor the mechanosensing properties of the different transgenic lines, we used the MT technique. In this method, a pulsed magnetic force (1 Hz, 1nN) is applied to pull magnetic beads attached to the cell surface (Figure 2A and B, and Figure 2—video 1). The magnetic beads oscillate in response to the pulsed force, and the local stiffness of the bead-cell adhesion can be measured as the ratio of the applied force to the bead movement. Cells able to detect the applied force respond through a phenomenon known as reinforcement, by which they progressively strengthen the cell-bead adhesion site, increasing its stiffness and thus reducing oscillation amplitude. Reinforcement can therefore be quantified as the change in adhesion stiffness over a specified time, providing a measure of cellular mechanosensing (Figure 2C and Figure 2—source data 1 [table]; for more details on the technique, see Roca-Cusachs et al., 2009). We studied forces transmitted through integrins (adhesion strength) by coating beads with FN. As a control, we also studied forces transmitted non-specifically through the PM using beads coated with the sugar-binding lectin concanavalin A (ConA; Figure 2D and Figure 2—videos 2 and 3). We first assessed the tethering specificity of the two coatings by mixing FN-coated beads (prepared with non-labeled BSA) with ConA-coated beads labeled with Alexa 546-conjugated BSA (Figure 2E and F). Staining for phalloidin and 9EG7 antibody (which specifically recognizes β1-integrin in its active conformation; Lenter et al., 1993) revealed FN-coated beads surrounded by both signals, indicating engagement of both β1-integrin and the cell cytoskeleton; in contrast, ConA-coated beads were excluded from these stainings (Figure 2E and F), indicating they are only bound bulk PM, as previously reported (Gauthier et al., 2011).

Cav1KO Mouse embryonic fibroblasts (MEFs) show reinforced attachment to magnetic tweezers.
(A) Reinforcement experiment scheme, indicating the fibroblast, the magnetic bead, and the magnetic tweezers apparatus. The red arrow represents the magnetic force exerted on the bead by the magnet. (B) Differential interference contrast (DIC) image showing a MEF, the tip of the magnetic tweezers, and a magnetic bead (white arrows). Scale bar = 3 µm. (C) Examples of bead oscillation as a function of time in two conditions: with and without reinforcement. (D) Representation of the two magnetic beads coatings used: fibronectin (FN), which binds integrins, and ConA, which binds the bulk plasma membrane (PM). (E and F) Confocal microscopy images showing a MEF attached to concanavalin A-coated beads (red) and FN-coated beads (gray). Actin staining is shown in green (phalloidin), active β1-integrin in magenta (9EG7 antibody), and 4′,6-diamidino-2-phenylindole (DAPI) in blue. Note how only FN-coated beads present both phalloidin and 9EG7 staining. Scale bar = 2 µm. (G and H) Reinforcement increment (relative change in reinforcement over the entire experiment, calculated as the difference between the last and initial measurements) of different MEF genotypes for FN-coated beads (G) or ConA-coated beads (H); n≥20 beads per genotype. Statistical comparisons were by t-test, with significance assigned at *p<0.05. N. S., non-significant. Data are presented as mean values +/- SEM.
-
Figure 2—source data 1
Traces of beads movement shown in Figure 2C.
They represent two examples of the data used to generate the reinforcement increment shown in Figure 2G.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig2-data1-v3.xlsx
-
Figure 2—source data 2
Raw data of experiments from Figure 2G and 2H.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig2-data2-v3.xlsx
We next analyzed the integrin/cytoskeleton-dependent mechanical response in different MEF lines. Cells were plated for 10 min on FN-coated coverslips. FN- or ConA-coated beads were then allowed to bind to the cell surface and subjected to magnetic pulses. Reinforcement was significantly higher in Cav1KO MEFs exposed to FN-coated beads than in wild type MEFs (Figure 2G). Reconstitution of Cav1KO MEFs with recombinant Cav1 rescued the WT phenotype, supporting that the Cav1KO phenotype is specific. Reconstitution of Cav1KO MEFs with PTRF induced a partial recovery that did not reach statistical significance. In contrast, MT pulling upon binding to ConA-coated beads yielded no significant differences across genotypes (Figure 2H). These results support an intrinsic role for Cav1/caveolae in determining PM mechanical properties, through integrin-dependent pathways. Integrins link ECM components, notably FN, to the cytoskeleton, transducing external forces into downstream effects and regulating responses such as cell contraction (Schwartz, 2010). We previously showed that Cav1 regulates cell contraction by controlling Rho activity through the localization of p190RhoGAP (Grande-García et al., 2007) within the PM (Grande-García et al., 2007; Goetz et al., 2011). In the absence of Cav1, p190RhoGAP localization to liquid order domains increases, where it can bind and inhibit Rho activity (Goetz et al., 2011), thus attenuating cell contractility and associated ECM remodeling (Grande-García et al., 2007). This reduced cell contractility contrasts with the local stiffening response observed in our MT assays. Because the specific knock down of this p190RhoGAP isoform fully rescues other biomechanical phenotypes in Cav1KO MEFs (Gauthier et al., 2011; Grande-García et al., 2007), we assessed the impact of stably knocking down p190RhoGAP (Goetz et al., 2011) on the increased reinforcement observed in Cav1KO cells (Figure 3A–C and Figure 3—figure supplement 1, see also Figure 3—figure supplement 1—source data 1 and 2). While traction force microscopy confirmed higher overall contractility in p190RhoGAP-depleted Cav1KO MEFs as expected (Goetz et al., 2011; Figure 3D), MT measurements revealed no decreased reinforcement upon p190RhoGAP knockdown (Figure 3C). These observations suggest that Cav1KO MEFs locally respond to FN-coated beads in a Rho-independent manner, as opposed to Rho-dependent whole cell contractility.

Cav1KO Mouse embryonic fibroblasts (MEFs) show Rho-independent reinforcement and increased fibronectin (FN) adhesion.
(A and B) Experimental schemes for (A) local force measurement (magnetic tweezers experiment, reinforcement) and (B) total force measurement (traction force microscopy, total cell contraction). (C) Effect of transfection with scrambled or shp190RhoGAP small hairpin RNA (shRNA) on the reinforcement increment in Cav1KO MEFs (magnetic tweezers assay). Reinforcement increment refers to the relative change in reinforcement over the entire experiment, calculated as the difference between the last and initial measurements; n≥20 beads per condition. (D) Effect of transfection with scrambled or shp190RhoGAP shRNA on mean total force contraction in Cav1KO MEFs (traction force microscopy); n≥12 cells per condition. (E and F) Relative adhesion of the indicated genotypes to plates coated with (E) 5 µg/ml ConA or (F) 5 µg/ml of FN. Measurements (absorbance, optical density, OD, at 595 nm from retained crystal violet dye, see Materials and methods) were normalized to values from adhesion to BSA-coated plates; n≥9 adhesion independent experiments. Data are presented as mean values +/- SEM. Statistical comparisons were by t-test, with significance assigned at *p<0.05. N. S., non-significant.
-
Figure 3—source data 1
Raw data of experiments from Figures 3C–F.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig3-data1-v3.xlsx
Loss of Cav1 increases FN adhesion and active surface β1-integrin
To explore the role of cell adhesion in Cav1KO MEF reinforcement, we first analyzed the ability of the different MEF lines to adhere to FN- or ConA-coated plates. Whereas adhesion to ConA-coated plates did not differ between genotypes (Figure 3E), adhesion to FN was higher in Cav1KO MEFs and PTRF-reconstituted Cav1KO MEFs (Figure 3F). Cav1 re-expression rescued the wild type phenotype, supporting a specific role of Cav1 in cellular adhesion. However, PTRF reconstitution did not have a significant impact on Cav1KO MEFs adhesion and was therefore not considered further in subsequent experiments. In additional experiments with different FN concentrations and other integrin-dependent coatings (collagen and vitronectin), Cav1KO MEFs showed always higher adhesion than wild type MEFs (Figure 3—figure supplement 1B, C and D). The two major FN receptors are integrins α5β1 and αvβ3 (Plow et al., 2000); however, adhesion strength is mainly mediated by the clustering and activation (Friedland et al., 2009; Lin et al., 2013) of integrin α5β1. To evaluate the role of β1-integrin in reinforcement, we imaged the active β1-integrin pool in Cav1KO and wild type MEFs by staining with 9EG7. The active β1-integrin signal was consistently stronger in Cav1KO MEFs than in wild type MEFs over a range of conditions: permeabilized and non-permeabilized cells, around FN-coated beads and at different spreading time points (Figure 3—figure supplement 1E–M). Cav1KO MEFs displayed stronger FN adhesion than wild type MEFs, a phenotype presumably derived from increased active β1-integrin at the cell surface.
Cav1KO MEFs recycle β1-integrin faster than wild type MEFs
We first hypothesized that enhanced recruitment of active β1-integrin to FN beads in Cav1KO MEFs could be derived from increased lateral mobility. However, experiments measuring florescence recovery after photobleaching (FRAP) in Cav1KO and wild type MEFs expressing similar levels of GFP-fused β1-integrin (Figure 4A, B) revealed no significant differences in fluorescence recovery between the two cell populations, indicating that lateral mobility of β1-integrin was unaffected (Figure 4C). Total Internal Reflection Fluorescence (TIRF) videos revealed that β1-integrins in wild type MEFs as stable, largely immobile structures; in contrast, β1-integrins in Cav1KO MEFs showed a dynamic behavior, rapidly appearing and disappearing from the PM plane (Figure 4—videos 1 and 2 and Figure 4—figure supplement 1A and 1B). Interestingly, integrins increased their dynamicity after treatment with high hypoosmotic pressure in wild type MEFs, mimicking Cav1KO phenotype (Figure 4—figure supplement 1C–F, and Figure 4—videos 3–6, quantified in Figure 4—figure supplement 1F, see Materials and methods for more details), suggesting that PM tension changes could affect integrin trafficking dynamics. In adherent cells, integrins undergo constant endocytic-exocytic shuttling to facilitate the dynamic regulation of cell adhesion (Bretscher and Aguado-Velasco, 1998). To study the effect of these dynamics on β1-integrin surface availability across our tested genotypes, we performed a series of endocytosis/recycling assays with an ELISA-based protocol (Li et al., 2016; Roberts et al., 2001; Figure 4D and E). We first analyzed the endocytic rates of total and active β1-integrin at early time points (2, 5, and 10 min) during early spreading (2 hr after seeding), to recapitulate the conditions used in MT experiments (see Materials and methods for details). Cav1KO and wild type MEFs showed no significant differences, indicating that β1 endocytosis is Cav1-independent at these early time points (Figure 4F and H). We next studied the recycling rates after allowing β1 endocytosis to proceed for 5 or 10 min; recycling was tested over two time point sets: 2, 5, and 10 min and 1, 3, and 5 min. After 5 min of endocytosis, wild type and Cav1KO MEFs showed no major differences in total β1-integrin recycling rates at either time point set (Figure 4—figure supplement 1G, H). In contrast, after 10 min endocytosis, Cav1KO MEFs recycled total and active β1-integrin faster than wild type MEFs in the 1-3-5 min recycling set (Figure 4G and I). Importantly, reconstitution of Cav1KO MEFs with recombinant Cav1 rescued active β1-integrin recycling but not total β1-integrin, supporting that the Cav1KO phenotype is specific at least for the active pool (Figure 4—figure supplement 1I, J). These results suggest that β1-integrin is stabilized in the presence of Cav1 after 10 min endocytosis, whereas in Cav1KO MEFs its shuttling to PM is accelerated, increasing its surface availability and thus enhancing adhesion to FN-coated beads and reinforcement response. Different mechanisms could account for this stabilization; however, recent work in our lab showed increased exocytosis after loading wild type MEFs with cholesterol, phenocopying Cav1KO MEFs where cholesterol accrued in different endosomal compartments (Albacete-Albacete et al., 2020). To test if cholesterol could play a role in β1-integrin stabilization, we treated wild MEFs with either U18666A (which promotes cholesterol accumulation in endosomal compartments Cenedella, 2009) or low-density lipoproteins (LDLs) (which also increases cholesterol content) and analyzed β1-integrin levels by ELISA. Interestingly, both treatments induced a significant increase in surface active β1-integrin (Figure 4—figure supplement 1K) that was also accompanied by an increased colocalization with EEA-1 positive endosomes in LDL-treated cells (Figure 4—figure supplement 1L–N). These results might be indicative of a Cav1-dependent cholesterol threshold above which β1-integrin trafficking is altered, phenocopying Cav1KO MEFs where both surface and intracellular active β1-integrin levels are increased.

Cav1KO mouse embryonic fibroblasts (MEFs) show faster β1-integrin recycling.
(A and B) Confocal microscopy images of wild type and Cav1KO MEFs transfected with β1-integrin-GFP expression vector. Scale bar = 10 µm. (C) Normalized fluorescence intensity recovery after photo bleaching of wild type MEFs (black line) and Cav1KO MEFs (red line) at the indicated time points (the graph is representative of a minimum of nine independent experiments in which between 8 and 10 cells were bleached per genotype). Statistical significance of between time point differences was estimated by t-test; N. S., non-significant. (D and E) Experimental schemes for the analysis of endocytosis and recycling followed by ELISA (according to Li et al., 2016). (F and H) Net endocytosis of (F) total β1 integrin and (H) active β1-integrin in wild type and Cav1KO MEFs at the time points indicated. Net endocytosis is expressed as internalized biotinylated β1 integrin (cytosolic) at each time point normalized to total biotinylated β1-integrin (internal and surface bound; see Materials and methods); n≥6 endocytosis assays per genotype. (G and I) Net recycling after 10 min endocytosis of (G) total and (I) active β1-integrin in wild type and Cav1KO MEFs at the time points indicated. Net recycling is expressed as internal biotinylated β1 integrin (cytosolic) at each time point normalized to time point 0 (which contains all the biotinylated β1 integrin internalized after 10 min of endocytosis, see Materials and methods); then, decreasing values mean increased recycling; n=10 recycling assays per genotype. Data are presented as mean values +/- SEM. Statistical comparisons were by t-test, with significance of between-group differences denoted *p<0.05, **p<0.01, or ***p<0.001. N. S., non-significant.
-
Figure 4—source data 1
Raw data of experiments from Figure 4.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig4-data1-v3.xlsx
Clathrin independent carrier/GPI enriched endocytic compartment-dependent uptake contributes to β1-integrin endocytosis in Cav1KO MEFs
β1-integrin endocytosis has been suggested to be Cav1-dependent in different systems (Shi and Sottile, 2008; Du et al., 2011); however, we observed no differences between Cav1KO and wild type MEFs in net β1-integrin endocytosis. This discrepancy might reflect assay timings, as we studied early time points (2, 5, and 10 min), whereas previous studies focused on endocytosis over longer time periods. To investigate how Cav1KO MEFs achieve similar levels of β1-integrin endocytosis as wild type MEFs, we decided to analyze endocytosis via clathrin independent carriers (CLICs), especially because this endocytic modality is negatively regulated by caveolar proteins, including Cav1 and PTRF (Chaudhary et al., 2014; Cheng et al., 2010). Furthermore, this endocytic route is highly relevant for fibroblast trafficking dynamics—accounting for ~threefold internalized volume as compared to clathrin-mediated endocytosis—and β1-integrin is a specific cargo (Howes et al., 2010; Lakshminarayan et al., 2014). Interestingly, cells lacking Cav1 increase CLIC endocytosis through the small GTPase Cdc42 activation (Cheng et al., 2010), which is known to regulate this endocytic pathway (Mayor et al., 2014). We therefore asked ourselves whether Cav1KO MEFs might preferentially endocytose β1-integrin via this mechanism. We treated wild type and Cav1KO MEFs with the Cdc42 inhibitor ML141 (Surviladze, 2010), which inhibits the CLIC-GEEC (GPI enriched endocytic compartment) pathway of endocytosis, involved in fluid-phase uptake and the entry of many specific cargoes, including integrins (Howes et al., 2010; Thottacherry et al., 2017). ML141 significantly reduced β1-integrin endocytosis in Cav1KO MEFs, whereas wild type MEFs were unaffected (Figure 5A and B), indicating that β1-integrin is partially endocytosed through the CLIC/GEEC pathway in Cav1KO MEFs. In accordance with this finding, Cav1KO MEFs displayed significantly higher uptake of the fluid-phase endocytosis marker dextran (Sabharanjak et al., 2002), as well as elevated colocalization of dextran and Alexa-488-labeled β1-integrin (Figure 5C–H and Figure 5—figure supplement 1), as compared to wild type cells. Thus, in the absence of Cav1, early endocytosis of β1-integrin occurs at least in part through CLIC uptake, which provides an alternative entry route that would compensate for lack of Cav1-dependent internalization. To further delineate the relative contribution of the different endocytic routes to β1 integrin endocytosis, we performed a series of colocalization studies of active β1-integrin and previously characterized markers for caveolar-dependent (BODIPY-LacCer), CLIC-dependent (CD44) and clathrin-dependent (transferrin, Tnf-568) endocytosis (; Cheng et al., 2006; Singh et al., 2003; Harding et al., 1983). Consistent with previous results, β1-integrin endocytosis was mainly Cav1-dependent in wild type MEFs as it: (i) colocalized with Cav1 and LacCer, (ii) was significantly reduced upon genistein treatment (a caveolar endocytosis inhibitor Rejman et al., 2005), and (iii) was unaffected by ML141 treatment (the CLIC inhibitor; Figure 5—figure supplement 1D–K). On the other hand, β1-integrin endocytosis was mainly CLIC-dependent in Cav1KO MEFs as it: (i) colocalized with CD44 and (ii) was significantly reduced upon ML141 treatment as compared to wild type MEFs (Figure 5—figure supplement 1L–P). Finally, no significant differences were found in clathrin-dependent β1 integrin endocytosis between wild type and Cav1KO MEFs (Figure 5—figure supplement 1Q–S). Altogether, these results further prove that β1-integrin endocytosis is mainly endocytosed by CLIC-dependent mechanisms in Cav1KO MEFs.

Cav1KO mouse embryonic fibroblasts (MEFs) take up β1-integrin by clathrin independent carrier (CLIC) endocytosis.
(A and B) Net endocytosis (normalized to Total biotinylated β1-integrin) at the time points indicated. (A) wild type MEFs treated with ML141 (red line) and untreated controls (black line). (B) Cav1KO MEFs treated with ML141 (green line) and untreated controls (orange line); n≥5 endocytosis assays per genotype. (C–F) Confocal microscopy images of wild type MEFs (C and D) and Cav1KO MEFs (E and F) incubated with anti-active β1-Alexa 488 antibody (green) for 1 hr at 4°C followed by incubation with dextran-Alexa 647 (magenta) for 3 min at 37°C. White arrows in F mark colocalization between β1–488–positive particles and dextran-647–positive particles in Cav1KO MEFs. Insets of both Cav1WT and Cav1KO (right side of each panel) show colocalizing structures. Scale bar = 10 µm. (G and H) Quantification of colocalization between active β1-Alexa 488 and dextran-647, expressed as Pearson’s correlation coefficient (G) or Spearman’s correlation coefficient normalized by the mean of Cav1WT MEFs (H); n≥18 cells per genotype. Data are presented as mean values +/- SEM. Statistical significance of differences across indicated conditions was assessed by t-test: *p<0.05 P**p<0.01; ***p<0.001 N. S., non-significant.
-
Figure 5—source data 1
Raw data of experiments from Figure 5.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig5-data1-v3.xlsx
Cav1 is required for Rab11-dependent recycling of β1-integrin
β1-integrin follows the canonical Rab21-Rab11-dependent endosomal trafficking route—which takes longer times to recycle back to the PM—in wild type MEFs; while other paralogs, such as, for instance, β3-integrin follow a Rab4-dependent ‘short’ loop (Roberts et al., 2004; Roberts et al., 2001). As stated above, this scenario is in accordance with integrin β1 being localized to Rab11-positive endosomal compartments after 10 min endocytosis in wild type MEFs. In contrast, in the absence of Cav1, β1-integrin is partially sorted to a CLIC-dependent endosomal compartment in Cav1KO MEFs, from which it might be recycled to the PM following different dynamics. We have previously reported that integrins are rapidly delivered to nascent focal contacts in the absence of Cav1 (Grande-García et al., 2007). The recycling of CLIC cargo proteins is controlled by a number of factors including several Rabs such as Rab22a (Weigert et al., 2004), which has been shown to collaborate with the microtubule and tethering protein HOOK1 during this process (Maldonado-Báez et al., 2013). Interestingly, knocking down HOOK1 shifts the trafficking of CLIC cargo proteins from recycling to endosomal targeting, such as the surface glicoprotein CD147, which accumulates in the early endosomal antigen-1 positive compartment (EEA-1) (Maldonado-Báez et al., 2013). To analyze whether HOOK1 is required for active β1−integrin recycling, we transfected Cav1KO MEFs with siRNA against HOOK1 and studied the colocalization of EEA1 and 9EG7 (i.e. active β1−integrin) immunolabeling. Whereas HOOK1-deficient Cav1KO MEFs showed increased CD147 and EEA1 colocalization as expected (Maldonado-Báez et al., 2013), no significant differences were observed for β1–integrin (Figure 6A-E). Accordingly, surface active β1−integrin showed similar levels in both HOOK1-deficient and control Cav1KO MEFs, even 72 hr after siRNA treatment (Figure 6—figure supplement 1A). These results indicate that HOOK1 is not required for β1−integrin recycling in Cav1KO MEFs. We then decided to analyze the contribution of canonical Rab11 and Rab4-dependent recycling routes, using dominant-negative (DN) mutants described previously: Rab11 N124I, and Rab4 S22N, respectively (Roberts et al., 2001; White et al., 2007). We first assessed the impact of disrupting ‘long loop’-dependent recycling upon expression of the Rab11 DN mutant (White et al., 2007), as assessed by the degree of colocalization between the endosomal compartment (LBPA, lysobisphosphatidic acid, a late endosomal marker) and active β1−integrin (9EG7 label) in either wild type or Cav1KO cells. While we observed significant differences in the colocalization of 9EG7 and LBPA labels when comparing Cav1KO cells transfected with the Rab11 DN mutant to non-transfected Cav1KO cells, no significant differences were observed on the surface exposure of active β1−integrin in the same cells (Figure 6G, I and J and Figure 6—figure supplement 1B). In contrast, wild type MEFs showed increased colocalization between 9EG7 and LBPA-positive vesicles (Figure 6F, H and J) and reduced surface active β1−integrin levels upon Rab11 DN transfection (Figure 6—figure supplement 1B), suggesting that this is the main recycling pathway in wild type cells. We then assessed the impact of expressing a Rab4 S22N DN mutant (which blocks a ‘short loop’-dependent recycling Roberts et al., 2001) on the trafficking of active β1−integrin in either wild type or Cav1KO cells. Whereas no significant differences in 9EG7-EEA1 colocalization was observed upon disrupting Rab4-dependent trafficking in wild type MEFs, expression of the Rab4 DN mutant increased the colocalization between both labels in Cav1KO cells (Figure 6K–O). This was consistent with a significant reduction in surface active β1−integrin levels in Cav1KO cells expressing the Rab4 DN mutant, as compared to non-transfected Cav1KO cells, while no difference was observed for wild type cells (Figure 6—figure supplement 1C). Taken together, these results suggest that in the absence of Cav1, β1−integrin recycling is partially switched from ‘slow’ Rab11-dependent to ‘fast’ Rab4-dependent, recycling.

Cav1 is required for β1-integrin Rab11-dependent recycling.
(A–D) Confocal microscopy images of Cav1KO mouse embryonic fibroblasts (MEFs) stained for CD147 (green in A and C), active β1-integrin (9EG7 antibody, green in B and D) and EEA1 (magenta), either non-treated (A and B) or treated with siRNA against HOOK1 for 48 hr (C and D). Insets (below each panel) show colocalizing structures. Scale bar = 10 µm. (E) Quantification of colocalization between EEA1 and CD147 or 9EG7 normalized to control, expressed as Pearson’s correlation coefficient normalized by the mean of the corresponding control condition as indicated; n≥20 cells per condition. (F–I) Confocal microscopy images of wild type (F and H) or Cav1KO MEFs (G and I), stained for active β1-integrin (9EG7 antibody, green) and lysobisphosphatidic acid (LBPA) (magenta), either non-transfected (F and G) or transfected with Rab11 N124I dominant negative mutant for 48 hr (H and I). Insets (below each panel) show colocalizing structures. Scale bar = 10 µm. (J) Quantification of colocalization between LBPA and 9EG7 normalized by each control, expressed as Pearson’s correlation coefficient normalized by the mean of the corresponding control condition as indicated; n≥20 cells per condition. (K–N) Confocal microscopy images of wild type (K and M) or Cav1KO MEFs (L and N), stained for active β1-integrin (9EG7 antibody, green) and EEA-1 (magenta), either non-transfected (K and L) or transfected with Rab4 S22N dominant negative mutant for 48 hr (M and N). Insets (below each panel) show colocalizing structures. Scale bar = 10 µm. (O) Quantification of colocalization between EEA-1 and 9EG7 normalized to each control, expressed as Pearson’s correlation coefficient normalized by the mean of the corresponding control condition as indicated; n≥30 cells per condition. Colocalization was analyzed using the plugin intensity correlation analysis (Fiji Li et al., 2004). (P–V) Confocal microscopy images of wild type MEFs (P–R) and Cav1KO MEFs (T–V) incubated with anti-active β1-Alexa 488 antibody (green) for 1 hr at 4°C followed by 3 min endocytosis at 37°C. Remaining surface fluorescence was removed by acid stripping prior to fixation with paraformaldehyde (PFA) at 4%. Scale bar = 10 µm. Quantification of mean fluorescence intensity of endocytosed active β1-integrin in Cav1WT (S) or Cav1KO MEFs (W). Values were normalized to mean fluorescence intensity of control; n≥30 cells per genotype and condition. Data are presented as mean values +/- SEM. Statistical significance of differences across indicated conditions was assessed by t-test: * p<0.05; P**p<0.01; ***p<0.001 N. S., non-significant.
-
Figure 6—source data 1
Raw data of experiments from Figure 6.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig6-data1-v3.xlsx
Hypoosmotic shock increases β1-integrin recycling and activation in Cav1KO MEFs
Many studies have shown that membrane trafficking and membrane tension are tightly coupled (Gauthier et al., 2012; Apodaca, 2002). For example, increases in membrane tension increase exocytosis from the endocytic recycling compartment (Gauthier et al., 2009). Cav1KO MEFs lacking caveolae cannot buffer membrane tension properly (Sinha et al., 2011). This can lead to increased exocytosis (Osmani et al., 2018), as we observed in the specific case of β1-integrin recycling, which in turn increases adhesion to FN-coated beads. To specifically link integrin recycling to caveolae buffering in response to mechanical stress, we incubated Cav1KO and Cav1WT MEFs with anti-active beta1-Alexa 488 antibody for 1 hr at 4°C followed by 3 min endocytosis at 37°C. Immediately after that, we challenged cells for 1 min with either DMEM diluted 1:10 (which can be buffered by caveolae flattening, as previously described Sinha et al., 2011, and is shown in our own results, see below in both Figure 7A-F and Figure 7—figure supplement 1N-U), or DMEM diluted 1:20 in distilled water, which exceeds the buffering capacity of caveolae (as we show below in Figure 7—figure supplement 1N–U). To remove any remaining surface fluorescence, two quick steps of acid stripping were done prior to fixation. Strikingly, whereas Cav1KO MEFs showed a significant reduction in the endocytosed active beta1-integrin pool both at 1:10 and 1:20 dilutions (Figure 6T–W), Cav1WT MEFs only showed a similar significant reduction at 1:20 dilution (Figure 6P–S). These results indicate that caveolae buffering prevents integrin recycling below a certain force threshold and therefore regulate integrin dynamics in response to mechanical stress. However, PM tension can also affect adhesion more directly through increasing integrin activation independently from the cytoskeleton (Wang et al., 2015) and in a ligand-independent mechanism (Ferraris et al., 2014). To study the role of caveolae in membrane-tension-induced β1-integrin activation, we exposed cells to hypoosmotic conditions. Cav1KO and wild type MEFs were incubated for 10 min in DMEM diluted 1:10 in distilled water, fixed, and stained for 9EG7. Hypoosmotic shock sharply increased β1-integrin activation in Cav1KO MEFs, whereas no significant change was observed in wild type MEFs (Figure 7A–F; similar results were obtained with hypoosmotic shock exposure for 30 s and 1 min; data not shown). We also studied the amount of active β1-integrin around FN-coated beads before and after magnetic twisting (Figure 7—video 1), observing a significant tension-induced increase in Cav1KO MEFs (Figure 7—figure supplement 1A–J). We confirmed that this phenotype is caveolae-dependent, Cav1-independent, as PTRFKO MEFs (that lack caveolae but still express Cav1 Hill et al., 2008) also show increased β1-integrin activation upon hypoosmotic shock (Figure 7—figure supplement 1K–M). Strikingly, β1-integrin activation was also observed in Cav1WT MEFs after both longer hypoosmotic treatments and with higher hypoosmotic pressures (Figure 7—figure supplement 1N–W). To rule out any possible antibody penetration due to paraformaldehyde (PFA)-induced PM disruption, we confirmed these results by incubating MEFs with 9EG7 at 4°C followed by the different hypoosmotic treatments (Figure 7—figure supplement 2A–J). Altogether, these results might indicate that caveolae restrict integrin activation upon changes in PM tension until their buffering capacity is exhausted; consequently, Cav1KO MEFs lack the ability to adapt integrin activation to mechanical stress.

Talin is required for the enhanced adhesion and β1-integrin activation phenotype of Cav1KO mouse embryonic fibroblasts (MEFs).
(A–D) Confocal microscopy images of wild type MEFs (A and B) and Cav1KO MEFs (C and D) stained for active β1-integrin (9EG7 antibody, green) after culture in standard medium (control in A and C) or hypoosmotic medium (diluted 1:10; B and D) for 10 min at 37°C. DAPI is shown in blue. Scale bar = 10 µm. (E and F) Quantification of 9EG7 mean fluorescence intensity in control and hypoosmotic shock-exposed wild type (E) or Cav1KO MEFs (F). Values were normalized by each analyzed area and finally referred to area control; n≥35 cells per genotype. (G–J) Confocal microscopy images of active β1-integrin staining (9EG7 antibody, green) in Cav1KO MEFs subject to indicated RNAi treatments and cultured for 10 min at 37°C in standard medium (control) or hypoosmotic medium (diluted 1:10). Scale bar = 10 µm. (K and L) Active β1-integrin immunostaining in wild type MEFs cultured in standard (control) or hypoosmotic medium. DAPI counterstain is shown in blue. Scale bar = 10 µm. (M, P, and S) Relative adhesion of MEFs of the indicated genotypes (M and P) or HeLa Cells (S, wild type – control -; knocking down Cav1 -Cav1KD-; and knocking down both Cav1 and Talin1 -Cav1KD + siTnl1) to plates coated with 5 µg/ml fibronectin (FN). Values were normalized to control condition; n≥18 cells in three independent adhesion experiments. THWT: Talin head wild type; THMut: Talin head mutant. (N, O, Q, and R) Confocal microscopy images of talin head domain (N and Q) and active β1-integrin (9EG7 antibody; O and R) in Cav1KO MEFs transfected with Tln2 shRNA and Tln1 siRNA plus either WT Talin head (N and O) or mutant Talin head (Q and R). Before immunostaining, cells were cultured for 10 min at 37°C in hypoosmotic medium (diluted 1:10). DAPI is shown in blue. Scale bar = 10 µm. All immunostainings in this figure were performed following the extracellular staining after fixation (see Materials and methods for more details). Data are presented as mean values +/- SEM. Statistical significance of differences across indicated conditions was assessed by t-test: * p<0.05 P**p<0.01; ***p<0.001 N. S., non-significant.
-
Figure 7—source data 1
Raw data of experiments from Figure 7.
- https://cdn.elifesciences.org/articles/82348/elife-82348-fig7-data1-v3.xlsx
Talin supports increased β1-integrin activation and recycling in Cav1KO MEFs
Integrin activation is controlled by a number of intracellular proteins (Moser et al., 2009). One of them, talin, regulates integrin adhesion strength by interacting through its amino-terminal FERM domain (the talin head domain; Tanentzapf and Brown, 2006). This interaction is necessary and sufficient to induce inside-out integrin activation (Tadokoro et al., 2003; Zhang et al., 2008). To study the possible role of talin in β1-integrin activation in Cav1KO MEFs, we first confirmed β1-integrin binding competence after hypoosmotic treatment. Colocalization between β1-integrin and both soluble FN and talin proved its active conformation (Figure 7—figure supplement 2L–O). We then analyzed the effect on adhesion and activation after silencing Tln1 and Tln2. Interestingly, Tln2-silencing did not significantly affect either hypoosmotic-induced β1-integrin activation or adhesion in Cav1KO MEFs (Figure 7G–I and K–M and Figure 7—figure supplement 2K). In contrast, simultaneous silencing of Tln1 and Tln2 in Cav1KO MEFs reduced hypoosmotic-induced β1 activation and reduced adhesion, rescuing the wild type phenotype (Figure 7J and M and Figure 7—figure supplement 2K, see also Figure 7—figure supplement 2—source data 2 and 3). The same result was observed in Cav1-silenced HeLa cells (Figure 7—figure supplement 2—source data 2). Surprisingly, Tln1/2 knockdown also affected integrin trafficking, as surface active β1-integrin was significantly reduced and consistently increased intracellularly in EEA-1 positive endosomes (Figure 7—figure supplement 2P–W). Given that the talin head domain binds and activates integrins (Zhang et al., 2008), we studied the ability of the wild type talin head to rescue adhesion in Tln1- and Tln2-silenced Cav1KO MEFs. As a control, we transfected cells with a DN talin-head mutant (L325R) that does not activate integrins or link them to the cytoskeleton (Wegener et al., 2007). Expression of the WT talin head, but not the data mutant, rescued both the activation and the adhesion ability of Cav1KO MEFs (Figure 7N–P and Q–R). The talin head thus regulates adhesion and β1-integrin activation in Cav1KO MEFs.
Discussion
Cellular mechanosensing is dependent on both integrins and caveolae, but how these two are coupled in this cellular response is poorly understood, especially during early steps of mechanoadaptation. Our MT-pulling experiment results show that Cav1KO MEFs adhere more strongly than wild type MEFs to FN-coated beads. This is consistent with the higher adhesion behavior of Cav1KO MEFs as compared to wild type MEFs in substrate adhesion assays. FN binds to integrins, and β1 is its main receptor. The analysis of β1 distribution revealed an increased surface pool of active β1-integrin in Cav1KO MEFs. TIRF imaging and FRAP measurements indicated that β1-integrin has a more dynamic behavior in Cav1KO MEFs, appearing at and disappearing from the membrane plane faster than in wild type MEFs, through mechanisms distinct from diffusion. This prompted us to study endocytosis/recycling rates. ELISA-based assays revealed no differences between Cav1KO and wild type MEFs in endocytosis rate at the time points analyzed, suggesting that Cav1 does not play a specific role. This finding contrasts with other reports showing that Cav1 is required for proper β1-integrin endocytosis (Li et al., 2016; Du et al., 2011); however, these observations were not made in a Cav1KO background and were performed at later time points than examined in our experiments (as we wanted to analyze specifically early integrin mechanosensing events, studied in the MT experiments). This time difference suggests that net integrin endocytosis at early time points could be compensated by other mechanisms in Cav1KO MEFs. Furthermore, and in accordance with previous reports , Cav1 could be restricting the endocytosis of a pool of β1-integrin in wild type MEFs (where it is mainly endocytosed through caveolae-dependent mechanisms) and becomes freed in Cav1KO MEFs to follow a different entry route. Indeed, our results indicate that β1-integrin is partially taken up in Cav1KO MEFs through CLIC/GEEC endocytosis, which provides a fast entry route, as reported for other cargoes (Chaudhary et al., 2014; Cheng et al., 2010; Paul et al., 2015). Interestingly, net endocytosis of active β1-integrins is higher than that of inactive pools (Arjonen et al., 2012; Valdembri et al., 2009). This is consistent with our data showing that Cav1KO MEFs display higher 9EG7 signal inside the cell and harmonizes with our findings that Cav1 genetic deficiency enhances CLIC endocytosis.
Differences between Cav1KO and wild type MEFs were apparent when we analyzed β1-integrin recycling rates at early time points after 10 min of endocytosis, being faster in Cav1KO MEFs. Interestingly, no differences in recycling were observed after 5 min of endocytosis, suggesting that in the presence of Cav1, β1-integrin becomes stabilized over time. This could potentially depend on cholesterol levels as loading wild type MEFs with cholesterol increases both surface and endosomal active β1-integrin availability, phenocopying Cav1KO MEFs phenotype. Raising endosomal cholesterol levels leads to increased exocytic activity in wild type MEFs, mirroring Cav1KO MEFs cholesterol accumulation, as work in our lab has previously demonstrated (Albacete-Albacete et al., 2020). This suggests that there might be a cholesterol threshold above which β1-integrin trafficking is dysfunctional, as it happens in the absence of Cav1. Furthermore, our results suggest that Cav1 is required for Rab11-dependent recycling of β1-integrin, which takes longer to reach the PM. Interestingly, in the absence of Cav1, β1-integrin accumulates preferentially at EEA-1-positive vesicles, following the Rab4-dependent ‘short’ recycling loop. This is consistent with our previous observations of Cav1 playing a role in determining the migration mode of fibroblasts (Grande-García et al., 2007), alternating between persistent or random migration. Faster recycling in Cav1KO MEFs can account for the elevated β1-integrin surface availability and therefore explain the resulting reinforcement. This interpretation is further supported by our previous findings that Cav1KO MEFs have a higher number of small focal adhesions that are rapidly turned over (Grande-García et al., 2007), since short but frequent integrin-ECM contacts would strengthen adhesion over time (ECM-coated beads in the MT experimental design). In the same report, we showed that Cav1KO MEFs have low Rho activity resulting in impaired actomyosin contraction (Grande-García et al., 2007), a condition that alters the overall cell mechanical response, as revealed in our traction force experiments. However, in the present study, Rho appears to be dispensable for observed differences in reinforcement. Consistently, previous reports have shown initial integrin adhesion in the absence of cytoskeleton connection, suggesting that early mechanosensing could be locally triggered (Elosegui-Artola et al., 2014; Bakker et al., 2012; Changede et al., 2015). This actomyosin-independent adhesion can derive from increased PM tension (Wang et al., 2015), which has also been shown to induce ligand-independent integrin activation (Ferraris et al., 2014). Cav1KO MEFs lack caveolae and are unable to buffer membrane tension upon mechanical stress (Sinha et al., 2011). In this condition, integrin recycling and activation at a lower force threshold might be facilitated by both faster exocytosis and easier switch of β1-integrin to its active conformation, as we observed upon hypoosmotic treatment. This increased sensitivity to membrane tension contributes, together with increased recycling, to the higher surface availability of active β1-integrin we observed in Cav1KO MEFs. Interestingly, we have also observed enhanced integrin recycling and activation in Cav1WT MEFs but after longer treatment times and with increased hypoosmotic pressure. These results suggest that caveolae membrane buffering is limiting both integrin recycling and activation before a PM tension threshold is reached. Previous studies have shown that talin controls inside-out integrin activation (Calderwood, 2004; Roca-Cusachs et al., 2013) through its head domain (Tadokoro et al., 2003; Zhang et al., 2008). Interestingly, our results show that talin expression is required both for adhesion and for integrin activation in Cav1KO MEFs. It also seems to play a role in integrin recycling, as Tln1/2 knockdown altered integrin trafficking. Importantly, depletion of both talin paralogs does not affect initial cell spreading (Zhang et al., 2008). This is consistent with our observation that talin depletion does not affect initial attachment of Cav1KO MEFs to the substrate (Figure 7M and 7P, see Materials and methods for details), as cells were allowed in our adhesion assays to spread for 30 min prior to measurements. In contrast, cells were allowed to spread and form attachments for at least 48 hr after reverse siRNA transfection in experiments were hypoosmotic shock was induced, and—while still attached—Cav1KO/KD cells depleted for Talins were rounder and less spread (compare Figure 7H with Figure 7J) than control cells. Rescue experiments indicated that this situation is mainly supported by the talin head domain. This suggests the intriguing possibility that membrane tension could favor local integrin–talin-head interaction without force transmission in Cav1KO MEFs. These observations suggest that the cellular cytoskeleton might play only a minor role in the early cellular response to mechanical stress. A similar conclusion was recently prompted by the detection of mechanical-stress–induced integrin recruitment in the absence of significant cytoskeletal changes (Elosegui-Artola et al., 2014).
Our results indicate that caveolae impact both integrin surface availability, through adjusting recycling, and integrin activation, through membrane tension regulation and talin activity. Cav1KO MEFs, lacking this control mechanism, show both increased β1-integrin recycling and surface activation. This results in dysregulated early mechanosensing (Figure 8) and subsequent inability to properly sense environmental stiffness, a situation known to impact tumorigenesis (Lin et al., 2015) and stem cell differentiation (Li et al., 2016). Mechanobiology is an emerging field, essential to understand how cells and tissues adapt to their environment in health and disease (Zaidel-Bar, 2017). Our study provides novel insight on the role of caveolae in early integrin mechanosensing, revealing a new layer of complexity at the interface of physics and biology.

Caveolae adjust membrane tension to integrin mechanosensing by regulating integrin cycling and activation.
Wild type mouse embryonic fibroblasts (MEFs) adapt to membrane tension changes through the buffer system of caveolae, driving a physiological integrin mechanosensing (in this case α5β1-integrin). In the absence of caveolae, dysregulation of this response leads to increased plasma membrane (PM) tension, which accelerates integrin recycling and switches integrin from the inactive forms to the active forms (close vs open conformation, respectively). Both increased β1-integrin recycling and activation is supported by increased talin activity in the absence of caveolae.
Materials and methods
Cloning, cells, and reagents
Caveolin-1 flag was excised from pCDNA3.1 Cav1 with BamH1/EcoR1, klenow treated, and ligated into the klenow blunt-ended EcoR1 site of the GFP-expressing retroviral vector MIGR1. PTRF was excised from pIRES2-cavin1 EGFP with BglII/BamH1 and ligated into the BglII site of MIGR1. The C-terminally EGFP-tagged β1-integrin was developed by Prof. Martin Humphries (University of Manchester, UK) and described elsewhere (Parsons et al., 2008) and was requested through the Addgene public repository under number #69767.
All cells were cultured at 37°C and 5% CO2 in DMEM (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin. Cav1KO MEFs were kindly provided by Michael Lisanti (Institute of Cancer Sciences, Manchester). All cell cultures were routinely checked for mycoplasma contamination. To deplete talin 2, cells were transfected with a plasmid encoding talin 2 shRNA and puromycin resistance (Zhang et al., 2008). Puromycin (2 µg/ml) was added to the cells 24 hr after transfection and maintained for 4 days to select transfected cells. To deplete talin 1, talin 2 shRNA-stable cells were transfected with Tln1 esiRNA (EMU083531, Sigma Aldrich) or, for rescue experiments with the talin head, with Tln1 siRNA (Silencer Select, Life technologies), which does not target the talin head domain. For rescue experiments, talin 2 shRNA-stable cells were co-transfected with Tln1 siRNA and EGFP-talin 1 head (Addgene plasmid no. 32856) or EGFP-talin 1 L325R (the mutant version, kindly provided by M. Ginsberg, UC San Diego, USA).
The following primary antibodies were used: mouse monoclonal anti-transferrin receptor (H68.4, Catalog # 13–6800), rat monoclonal anti-mouse total β1-integrin (clone MB1.2, MAB1997 Millipore); rat monoclonal anti-mouse β1-integrin, activated (clone 9EG7, BD Pharmingen); Alexa 488 conjugated anti-integrin β1, activated (clone HUTS-4, MAB2079-AF488 Millipore); rabbit polyclonal anti-mouse PTRF (Abcam); rabbit monoclonal anti-mouse caveolin-1 (Cell Signaling, #3238); mouse monoclonal anti-tubulin (Abcam, clone DM1A); rabbit monoclonal anti-CD147 (Invitrogen, clone JF1-045), mouse monoclonal anti-EEA-1 (BD transduction, clone 14); mouse monoclonal anti-LBPA (Echelon Z-SLBPA); mouse monoclonal anti-p190RhoGAP (Upstate, clone D2D6, 1:1000); rabbit monoclonal anti-mouse Caveolin-1 (Cell signaling, 1:1000); mouse monoclonal anti-alpha tubulin (ab7291, Abcam, 1:10.000); mouse monoclonal anti-Talin (Sigma Aldrich, clone 8d4, 1:200); and mouse anti-CD44 (clone 5035–41.1D, Novus Biologicals). FN and LDLs were obtained by purification from blood donors and were conjugated with FITC (Thermo Fisher) or used directly, respectively. U18666A was from sigma (U3633). Secondary antibodies were Alexa Fluor–488 goat anti-rat (Thermo Fisher Scientific); Alexa Fluor–647 goat anti-rat (Thermo Fisher Scientific); Alexa Fluor–488 phalloidin (Thermo Fisher Scientific); HRP-linked anti-biotin from Cell Signaling (#7075); and Alexa Fluor–647 phalloidin (Thermo Fisher Scientific). EZ-Link SulfoNHS-SS-biotin was from Thermo Fisher Scientific (D21331), 2-mercaptoethanesulfonic acid (MESNA) and iodoacetamide from Sigma Aldrich (63707 and I1149), the Cdc42 inhibitor ML141 from Tocris Bioscience, and Alexa Fluor 647-Dextran from Thermo Fisher Scientific (D22914). Silencing of p190RhoGAP was as previously described (Grande-García et al., 2007).
PM fractionation and western blot analysis
MEFs were processed for PM isolation as described (Smart et al., 1995). All steps were carried out at 4oC. Cells were first washed with cold-PBS 1× and pelleted by centrifugation at 14000×g for 5 min. Cells were then manually homogenized with 20 strokes of a PTFE head Tissue homogenizer (VWR) and centrifuged at 1000×g for 10 min. The post-nuclear supernatant was collected and layered atop a 30% Percoll column. After centrifugation of the Percoll column at 84,000×g for 30 min, the PM fraction was a visible band around 5.7 cm from the bottom of the centrifuge tube, was collected, further centrifuged at 105,000 g for 1 hr to remove Percoll, separated by SDS-PAGE, and finally analyzed by western blot. Samples were immunoblotted with rabbit monoclonal anti-mouse caveolin-1 and rat monoclonal anti-mouse total β1-integrin (clone MB1.2, MAB1997 Millipore) as a loading control and a marker for PM fraction.
Total cell lysates were separated by SDS-PAGE and analyzed by western blot with rabbit monoclonal anti-mouse Caveolin-1 and rabbit polyclonal anti-mouse PTRF (Abcam), with mouse monoclonal anti tubulin used as the loading control. Secondary antibodies were goat anti-mouse 800 and goat anti-rabbit 680. All membranes were scanned with the Odyssey imaging system (Li-COR).
Electron microscopy
MEFs were processed for EM using standard procedures. Briefly, cells were fixed for 1 hr with 2.5% glutaraldehyde in 100 mM cacodylate buffer, pH 7.4, and then post-fixed for 3 hr with 1% osmium tetroxide in 100 mM cacodylate buffer, pH 7.4. The samples were dehydrated with acetone, embedded in Epon and sectioned. Ruthenium red (1 mg/ml) was added during fixing and post-fixing to stain the PM.
Confocal and ground state depletion microscopy
Confocal images were obtained with an LSM 700 inverted confocal microscope (Carl Zeiss) fitted with a 63×1.4 NA objective and driven by Zen software (Carl Zeiss). Superresolution imaging was performed with a GSD-TIRF microscope (Leica Microsystems). Samples were prepared according to standard procedures indicated by Leica Microsystems. The primary antibody was rabbit monoclonal anti-mouse caveolin-1 (1:100), and Alexa Fluor 647 Fab1 fragment goat anti-rabbit (Jackson Immunoresearch; 1:100) was used as the secondary to further improve spatial resolution.
MTs and reinforcement measurements
Bead coating
Carboxylated magnetic beads (Invitrogen) were mixed in a solution containing 500 µl 0.01 M sodium acetate pH 5, 0.75 mg Avidin (Invitrogen), and 4 mg EDAC (Sigma). Beads were incubated for 2 hr at room temperature (RT) and then washed in PBS and further incubated for 30 min in 1 ml 50 mM ethanolamine (Polysciences). The beads were then washed three times in PBS and left in PBS on a cold room rotator.
Force measurements
MTs experiments were performed as described (Roca-Cusachs et al., 2009). Briefly, carboxylated 3 μm magnetic beads (Invitrogen) were coated with biotinylated pentameric FN7-10 or ConA (Sigma Aldrich) mixed 1:1 with biotinylated BSA. For measurements, cells were first plated on coverslips coated with 10 μg/ml FN (Sigma) in Ringer’s solution (150 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 20 mM HEPES, and 2 g/L glucose, pH 7.4) for 30 min. FN-coated beads were then deposited on the coverslips and allowed to attach to the cells. The tip of the MTs device was then used to apply a force of 1 nN for 2 or 3 min to beads attached to cell lamellipodia. The apparatus used to apply force to the magnetic beads was as previously described (Tanase et al., 2007). The system was then mounted on a motorized 37°C stage on a Nikon fluorescence microscope. DIC images and videos were recorded with a 60× objective linked to a CCD camera at a frequency of 250 frames/s.
Magnetic twisting
Cells seeded on FN-coated coverslips were subjected to magnetic twisting as previously described (Trepat et al., 2004). After twisting, cells were fixed for 20 min in 4% PFA and then stained with 9EG7 antibody without permeabilization.
Silica bead coating and staining
Carboxylated silica beads (3 μm, Kisker Biotech) were prepared as described above except that ConA-coated beads were incubated with biotinylated BSA previously labeled with an Alexa Fluor 555 protein labeling kit (Invitrogen). Unlabeled beads (FN-coated) and labeled beads (ConA-coated) were mixed in the same proportion (1:1). Cells were allowed to spread for 15 min and then fixed with 4% PFA, permeabilized with 0.1% Triton-X 100, and incubated at room temperature for 1 hr with the indicated antibodies. Only beads at the cell periphery were analyzed (excluding cells in the ectoplasm-endoplasm border zone). To quantify fluorescence intensity, a 10-pixel-diameter ring was drawn around each selected bead using ImageJ. Mean fluorescence per area was normalized and plotted.
Traction force microscopy
Traction force was measured as previously described (Elosegui-Artola et al., 2014). Briefly, cells were seeded on polyacrylamide gels incorporating embedded fluorescent nanobeads. Cells were imaged by phase contrast and embedded nanobeads by fluorescence. Cells were then trypsinized, and bead position images were acquired in the relaxed gel state. Comparison of bead positions in gels with and without cells was used to obtain a gel deformation map (Serra-Picamal et al., 2012; Bazellières et al., 2015). Images were obtained with a Nikon Eclipse Ti inverted microscope fitted with a 40× objective (numerical aperture = 0.6).
Adhesion assay
Cell adhesiveness was assessed by seeding MEFs on 96-well plates coated with FN or ConA (both at 5 µg/ml) and incubating at 37°C for 30 min. Wells with no coating were included as negative controls. Cells were then fixed with methanol and stained with crystal violet (Sigma Aldrich). Wells were washed thoroughly to remove excess dye and were finally eluted with a mixture of 50% ethanol and 50% 0.1 M sodium citrate (pH 4.2). The absorbance was read at 595 nm.
Florescence recovery after photobleaching
Cav1KO and wild type MEFs were transfected with the β1-GFP expression vector. Two pre-bleached events were acquired before bleaching by stimulation with the Nikon scanner at 488 nm. Fluorescence recovery was monitored continuously until the intensity plateaued (approximately 1.5 min). Fluorescence during recovery was normalized to the pre-bleach intensity. Cells were cultured in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS, and 1% penicillin, and streptomycin.
Endocytosis/recycling assay
β1-integrin kinetics were analyzed after biotin labeling of cell-surface integrins followed by a capture ELISA-based assay, using a modification of a previously described protocol (Li et al., 2016).
Cell-surface integrin biotinylation
Wild type or Cav1KO MEFs (5×105) were seeded on five (endocytosis) or four (recycling) matrigel-coated plates* (labeled total, 0 min, 2 min, 5 min, and 10 min for endocytosis and 0 min, 1 min, 3 min, and 5 min for recycling). The cells were incubated in complete DMEM (Thermo Fisher Scientific) for 2 hr at 37°C, which was the shortest spreading time for cells to stand the assay conditions and also matches the MTs measurement experimental time frame. The plates were then placed on ice, washed twice with ice-cold PBS and incubated for 40 min at 4°C with 0.25 mg/ml of EZ-Link SulfoNHS-SS-Biotin in Hank’s balanced salt solution (Sigma Aldrich). After two further washes with ice-cold PBS, the plates were labeled for the appropriate time points and processed as described below. Matrigel was used instead of FN to mimic a more physiological environment. While different coatings can clearly affect integrin trafficking, matrigel contains, among other extracellular components, collagen and certain levels of FN, which we have shown to increase Cav1KO adhesion as compared to wild type MEFs (see Figure 3—figure supplement 1B and C). Importantly, cells were allowed to spread for 2 hr before starting the assay in complete medium, thus ensuring an additional supply of FN.
Endocytosis
The 2, 5, and 10 min plates were incubated at 37°C with 2 ml of pre-warmed DMEM (without FBS) for the indicated times. The total and 0 min plates were placed on ice with 2 ml ice-cold DMEM (without FBS). All plates except total were then washed twice with ice-cold PBS and incubated for 40 min at 4°C with MESNA-containing buffer (Sigma Aldrich) to remove remaining surface-associated biotin. All the plates were then washed twice with ice-cold PBS and incubated with iodoacetamide for 10 min at 4°C. After washing again with ice-cold PBS, cells were lysed and processed for ELISA.
Recycling
The 0, 1, 3, and 5 min plates were incubated at 37°C with 2 ml of pre-warmed DMEM (without FBS) for 10 min (to allow time for endocytosis). The plates were then washed twice with ice-cold PBS and incubated for 40 min at 4°C with MESNA-containing buffer to remove remaining surface-associated biotin. The 1, 3, and 5 min plates were incubated at 37°C with 2 ml of pre-warmed DMEM (without FBS) for the indicated times; the 0 min plate was placed on ice with 2 ml ice-cold DMEM (without FBS). At the end of the incubation, all plates were washed twice with ice-cold PBS and incubated again with MESNA-containing buffer to remove biotin-labeled integrins that had recycled to the PM. Plates were finally washed with ice-cold PBS, incubated with iodoacetamide for 10 min at 4°C, lysed, and processed for ELISA.
ELISA-based assay
96-well ELISA plates were coated overnight at 4°C with anti-mouse total β1-integrin (Millipore) or anti-mouse β1-integrin, activated (BD Pharmingen). The plates were then washed three times with solution A (0.02% Tween-20 in PBS), blocked for 1 hr at room temperature with solution B (0.02% Tween-20, 1% BSA in PBS), and incubated with cell lysates from endocytosis or recycling assays for 2 hr at room temperature or overnight at 4°C. After washing three times with solution A, the plates were incubated for 1 hr at room temperature with anti-biotin HRP-linked antibody. Plates were then washed three more times with solution A, and integrin was detected by TMB reaction (Sigma Aldrich). Endocytosis results were normalized by dividing with the signal from Total wells; graphs represent the progressive increase in the amount of total or active β1-integrin. Recycling results were normalized by dividing with the signal from 0 min wells; graphs represent the β1-integrin remaining inside cells, so that negative curves indicate an increase in the recycling rate.
To determine total cell-surface integrin, cells were processed as in the total plates. For Rab DNs and Tln siRNA experiments, cells were plated 48 hr before the experiment avoiding re-plating to prevent losing lesser adherent cells. For cholesterol loading experiments, cells were treated with U18666A 2 µg/ml or LDL 100 µg/ml for 24 hr before surface integrin quantification.
Total internal reflection fluorescent microscopy videos
TIRF microscopy was performed with a Leica AM TIRF MC microscope. TIRFm movies were acquired with a 100 X_1.46 NA oil-immersion objective at 488 nm excitation and an evanescent field with a nominal penetration depth of 100 nm. Images were collected with an ANDOR iXon CCD at 300 ms per frame. Quantification of TIRF videos show normalized fluorescence integrated density (IntDen) over frames (Figure 4—figure supplement 1C–1E). Graph represents the mean of the difference between normalized fluorescence IntDen of adjacent frames (framex–framex-1) (Figure 4—figure supplement 1F). IntDen was calculated as in the following formula: IntDen = Raw IntDen (sum of pixel values in selection) × (area in scaled units)/(area in pixels), which was then normalized by the IntDen mean of all frames analyzed.
Hypoosmotic treatment
Cells were cultured for the indicated time points at 37°C in DMEM diluted as indicated with distilled water. The cells were then immediately fixed for 15 min by adding an equal volume of 8% PFA (yielding a final PFA concentration of 4%) and then stained with 9EG7 antibody without permeabilization.
Endocytosis experiments
For dextran endocytosis: cells were first incubated for 1 hr at 4oC (to prevent endocytosis) with β1-Alexa 488 conjugated antibody, activated. Cells were then washed twice with PBS and incubated for 3 min at 37°C with 1 mg/ml Alexa Fluor 647-Dextran (Invitrogen, REF D22914) without pre-incubation or acid stripping. Cells were then fixed with 4% PFA and analyzed by confocal microscopy. For other endocytosis experiments please see extended Materials and methods. Colocalization was analyzed using the plugin Coloc 2 (Fiji Schindelin et al., 2012).
For caveolar uptake, cells were treated, when indicated, with genistein 200 uM for 2 hr before incubation with BODIPY-LacCer 5 uM for 1 hr at 4°C followed by 3 min endocytosis at 37°C. BODIPY-LacCer (Invitrogen, REF B34402) remaining at the PM was then removed by back exchange at 4°C following a previous protocol (Martin and Pagano, 1994). Cells were then fixed with 4% PFA, stained, and analyzed by confocal microscopy.
For CLIC endocytosis, cells were treated, when indicated, with ML141 inhibitor 10 uM (Tocris, REF 4266) for 30 min before incubation with anti-active 1-Alexa 488 antibody and anti-CD44 antibody for 1 hr at 4°C followed by 3 min endocytosis at 37°C. Cells were then acid stripped to remove surface staining, fixed with 4% PFA, stained, and analyzed by confocal microscopy.
For clathrin endocytosis cells were incubated with anti-active 1-Alexa 488 antibody and Tnf-568 (Invitrogen, REF T23365) for 1 hr at 4°C followed by 3 min endocytosis at 37°C. Cells were then acid stripped to remove surface staining, fixed with 4% PFA, stained, and analyzed by confocal microscopy. Colocalization was analyzed using the plugin Coloc 2 (Fiji ).
Extracellular staining
To analyze cell-surface β1-integrin, cells were fixed for 20 min with 4% PFA and then stained with 9EG7 antibody without permeabilization. Alternatively, to rule out any possible PM disruption due to PFA fixation, cells were also incubated with 9EG7 antibody at 4°C for 1 hr as indicated in Figure 7—figure supplement 2A–J.
Statistical analysis
Data are presented as mean ± SEM unless otherwise indicated. Mean values were compared by two-tailed paired Student t-test unless otherwise indicated. Differences were considered statistically significant at p<0.05 (*), <0.01 (**), and <0.001 (***).
Data availability
Raw data of all figures is included as excel files.
References
-
ECM deposition is driven by caveolin-1-dependent regulation of exosomal biogenesis and cargo sortingThe Journal of Cell Biology 219:e202006178.https://doi.org/10.1083/jcb.202006178
-
Modulation of membrane traffic by mechanical stimuliAmerican Journal of Physiology. Renal Physiology 282:F179–F190.https://doi.org/10.1152/ajprenal.2002.282.2.F179
-
Control of cell-cell forces and collective cell dynamics by the intercellular adhesomeNature Cell Biology 17:409–420.https://doi.org/10.1038/ncb3135
-
Membrane tension drives ligand-independent integrin signalingThe EMBO Journal 33:2439–2441.https://doi.org/10.15252/embj.201489886
-
Membrane traffic during cell locomotionCurrent Opinion in Cell Biology 10:537–541.https://doi.org/10.1016/s0955-0674(98)80070-7
-
Talin controls integrin activationBiochemical Society Transactions 32:434–437.https://doi.org/10.1042/BST0320434
-
Distinct mechanisms of clathrin-independent endocytosis have unique sphingolipid requirementsMolecular Biology of the Cell 17:3197–3210.https://doi.org/10.1091/mbc.e05-12-1101
-
Co-regulation of caveolar and cdc42-dependent fluid phase endocytosis by phosphocaveolin-1The Journal of Biological Chemistry 285:15119–15125.https://doi.org/10.1074/jbc.M109.069427
-
Phospho-caveolin-1 mediates integrin-regulated membrane domain internalizationNature Cell Biology 7:901–908.https://doi.org/10.1038/ncb1293
-
Rac, membrane heterogeneity, caveolin and regulation of growth by integrinsTrends in Cell Biology 17:246–250.https://doi.org/10.1016/j.tcb.2007.03.001
-
Caveolae - mechanosensitive membrane invaginations linked to actin filamentsJournal of Cell Science 128:2747–2758.https://doi.org/10.1242/jcs.153940
-
Rigidity sensing and adaptation through regulation of integrin typesNature Materials 13:631–637.https://doi.org/10.1038/nmat3960
-
Plasma membrane area increases with spread area by exocytosis of a GPI-anchored protein compartmentMolecular Biology of the Cell 20:3261–3272.https://doi.org/10.1091/mbc.E09-01-0071
-
Mechanical feedback between membrane tension and dynamicsTrends in Cell Biology 22:527–535.https://doi.org/10.1016/j.tcb.2012.07.005
-
Caveolin-1 regulates cell polarization and directional migration through Src kinase and Rho GTPasesThe Journal of Cell Biology 177:683–694.https://doi.org/10.1083/jcb.200701006
-
Receptor-Mediated endocytosis of transferrin and recycling of the transferrin receptor in rat reticulocytesThe Journal of Cell Biology 97:329–339.https://doi.org/10.1083/jcb.97.2.329
-
Clathrin-independent carriers form a high capacity endocytic sorting system at the leading edge of migrating cellsThe Journal of Cell Biology 190:675–691.https://doi.org/10.1083/jcb.201002119
-
Galectin-3 drives glycosphingolipid-dependent biogenesis of clathrin-independent carriersNature Cell Biology 16:595–606.https://doi.org/10.1038/ncb2970
-
Integrin endocytosis on elastic substrates mediates mechanosensingJournal of Biomechanics 49:2644–2654.https://doi.org/10.1016/j.jbiomech.2016.05.024
-
Microtubule-Dependent endosomal sorting of clathrin-independent cargo by Hook1The Journal of Cell Biology 201:233–247.https://doi.org/10.1083/jcb.201208172
-
Clathrin-independent pathways of endocytosisCold Spring Harbor Perspectives in Biology 6:a016758.https://doi.org/10.1101/cshperspect.a016758
-
Phosphorylated filamin A regulates actin-linked caveolae dynamicsJournal of Cell Science 124:2763–2776.https://doi.org/10.1242/jcs.080804
-
An arf6- and caveolae-dependent pathway links hemidesmosome remodeling and mechanoresponseMolecular Biology of the Cell 29:435–451.https://doi.org/10.1091/mbc.E17-06-0356
-
Quantification of integrin receptor agonism by fluorescence lifetime imagingJournal of Cell Science 121:265–271.https://doi.org/10.1242/jcs.018440
-
Caveolae as plasma membrane sensors, protectors and organizersNature Reviews. Molecular Cell Biology 14:98–112.https://doi.org/10.1038/nrm3512
-
Endocytic trafficking of integrins in cell migrationCurrent Biology 25:R1092–R1105.https://doi.org/10.1016/j.cub.2015.09.049
-
Ligand binding to integrinsThe Journal of Biological Chemistry 275:21785–21788.https://doi.org/10.1074/jbc.R000003200
-
Protein kinase B/Akt acts via glycogen synthase kinase 3 to regulate recycling of alpha v beta 3 and alpha 5 beta 1 integrinsMolecular and Cellular Biology 24:1505–1515.https://doi.org/10.1128/MCB.24.4.1505-1515.2004
-
Fiji: an open-source platform for biological-image analysisNature Methods 9:676–682.https://doi.org/10.1038/nmeth.2019
-
Integrins and extracellular matrix in mechanotransductionCold Spring Harbor Perspectives in Biology 2:a005066.https://doi.org/10.1101/cshperspect.a005066
-
Mechanical waves during tissue expansionNature Physics 8:628–634.https://doi.org/10.1038/nphys2355
-
Caveolin-1-Dependent beta1 integrin endocytosis is a critical regulator of fibronectin turnoverJournal of Cell Science 121:2360–2371.https://doi.org/10.1242/jcs.014977
-
Selective caveolin-1-dependent endocytosis of glycosphingolipidsMolecular Biology of the Cell 14:3254–3265.https://doi.org/10.1091/mbc.e02-12-0809
-
Reportin Probe Reports from the NIH Molecular Libraries ProgramNational Center for Biotechnology Information.
-
Magnetic tweezers in cell biologyMethods in Cell Biology 83:473–493.https://doi.org/10.1016/S0091-679X(07)83020-2
-
Viscoelasticity of human alveolar epithelial cells subjected to stretchAmerican Journal of Physiology. Lung Cellular and Molecular Physiology 287:L1025–L1034.https://doi.org/10.1152/ajplung.00077.2004
-
Integrin molecular tension within motile focal adhesionsBiophysical Journal 109:2259–2267.https://doi.org/10.1016/j.bpj.2015.10.029
-
Rab22A regulates the recycling of membrane proteins internalized independently of clathrinMolecular Biology of the Cell 15:3758–3770.https://doi.org/10.1091/mbc.e04-04-0342
-
The role of matrix stiffness in regulating cell behaviorHepatology 47:1394–1400.https://doi.org/10.1002/hep.22193
-
Mechanosensing: from proteins to tissuesSeminars in Cell & Developmental Biology 71:1–2.https://doi.org/10.1016/j.semcdb.2017.08.029
Decision letter
-
Suzanne R PfefferSenior and Reviewing Editor; Stanford University, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Decision letter after peer review:
[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]
Thank you for submitting your work entitled "Cav1/caveolae couple mechanical stress to integrin recycling and activation" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.
Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your manuscript will not be considered further for publication in eLife.
There is no question that the role of caveolae in the mechanobiology of the plasma membrane and crosstalk between caveolae and integrins is of broad interest and importance. However, as you will see from the reviewer comments, there was concern that potential trafficking defects had not been connected to integrin activation and that the Rab4 trafficking experiments would require further work. Overall the reviewers felt that the story lacked a mechanistic link and many of the conclusions were not adequately documented to support the conclusions of the study.
We realize that this decision will be disappointing but we hope that the referee comments will be helpful to you as you plan your next steps. If you are able to address all of the concerns with significant additional experiments, a new manuscript could be evaluated in the future.Reviewer #1:
Caveolae are now widely recognized for their roles in the mechanobiology of the plasma membrane of cells submitted to mechanical stress. In the current manuscript, the authors have studied the crosstalk between caveolae and another mechanosensitive group of proteins, the integrins. Using as a model system mouse embryonic fibroblasts (MEFs) that are deleted for the caveolin-1 gene, it is described that cell adhesion to fibronectin-coated surfaces is increased in the absence of caveolae. The underlying reason for this is identified in the increased presence of the active conformation of beta1 integrin at the cell surface, which results from 2 phenomena: increased endocytic recycling, and increased activation upon mechanical stress.
The overall area of research on mechanobiology in relation with membrane trafficking that is addressed in the current study is of high interest to a general readership in the life sciences. The study is based on a rich arsenal of relevant techniques, and represents a substantial amount of work. Some techniques such as cell adhesion stiffness measurements based on magnetic bead oscillation or traction force microscopy are really quite elegant. The manuscript is well written, and pleasant to read.
Possibly the most important limitation of the current study is that it remains descriptive on the molecular mechanism aspects. It is clear that the absence of Cav1/caveolae has a number of effects that are clearly defined in the current study. The molecular wiring that underlies these effects remains unexplored at this stage. A few examples: In the absence of Cav1/caveolae, endocytic recycling of beta1 integrin passes from Rab11-dependent slow recycling to Rab4-dependent fast recycling – by which molecular mechanism does Cav1/caveolae connect to the endosomal machinery? Mechanical stress increases beta1 activation in Cav1KO MEFs – by which molecular mechanism does Cav1/caveolae limit this activation in wild-type conditions? Talin is important in the context of this mechanical stress-controlled activation reaction – by which molecular mechanism does Cav1/caveolae interact with talin? Having said this, I still believe that the manuscript will contribute in a substantial manner to the dynamic field of membrane biology research. The following points should be clarified, though.
Lines 262-269: The faster beta1 integrin recycling phenotype in Cav1KO MEFs is seen only after 10 min of endocytosis, and not when recycling is measured after 5 min of endocytosis. The authors suggest that "beta1-integrin is stabilized in the presence of Cav1 after 10 minutes endocytosis". How would such stabilization work? The authors most likely don't have the experimental response to this question at this stage. However, it would be helpful if some ideas could be discussed on the molecular mechanisms by which this would work.
Lines 315-317: In the paragraphs that precede these lines, the authors present experiments based on the small molecule Cdc42 inhibitor ML141 that lead to the conclusion that in Cav1KO MEFs, part of endocytic uptake occurs by the CLIC/GEEC pathway, while this does not appear to be the case in MEFs that express Cav1. It is concluded that in "wild-type" conditions, part of uptake occurs through caveolae, and that this Cav1-dependent internalization would be compensated by the CLIC/GEEC pathway in the absence of Cav1/caveolae. Can other interpretations be excluded? For example, wouldn't one observe the same phenotype if Cav1/caveolae were to inhibit uptake of a fraction of the beta1 integrin molecules that once liberated in the Cav1KO condition would now be internalized by the CLIC/GEEC pathway? The most direct way to address this point would be electron microscopy to provide ultrastuctural images of the uptake structures in which beta1 integrin is found in the different experimental conditions. To the least, alternative interpretations of the data should be discussed.Reviewer #2:
This manuscript describes the role of Caveolin-1 (Cav1) in integrin recycling using Cav1 knockout MEFs and links integrin endocytosis and recycling to cell mechanosensing and adhesion. I find the subject interesting and, in light of the role of Cav1 and caveolae in the cellular response to mechanical stress, timely and relevant. That being said I find the manuscript to be a compilation of interesting results that lack mechanistic connection between them. We are provided with interesting data on: (1) the role of Cav1 in integrin mechanosensing using an elegant magnetic tweezer approach; (2) Cav1 regulation of integrin endocytosis and recycling but with no direct link to mechanosensing or stretch; (3) the role of talin in regulating CAV1-independent integrin activation. It seems as though Cav1 is altering integrin dynamics and response to mechanical stretch which is quite interesting. While the schematic in figure 8 highlights the Cav1-dependent changes reported in the paper, mechanistic connections between them need to be clarified and I have many questions which I outline below.
1. "Cav1/caveolae" Cav1 and caveolae are not the same thing and it cannot be assumed that effects observed in Cav1 KO cells are necessarily attributed to loss of caveolae. I am concerned about use of this term throughout the paper (even the title) and suggest that it would be important to define the specific role of caveolae in the processes described using cell lines expressing Cav1 but not PTRF. Indeed, the only data linking caveolae flattening to the effects shown is Figure 7 and is interesting in that it suggests that lack of membrane buffering by caveolae induces an integrin response. However, to definitively show that this is caveolae dependent and not Cav1 dependent it would be important to use cells expressing Cav1 but not caveolae.
2. The studies on b-integrin endocytosis switch to a CLIC pathway in Cav1 KO is interesting and supports a role for Cav1 as an inhibitor of CLIC endocytosis. It would be important to provide more evidence for CLIC endocytosis than inhibition of Cdc42. Is b-integrin cointernalized with CD44? Is this pathway CD44- or raft-dependent? Can the effects of Cav1 on integrin mechanosensing be attributed to CLIC endocytosis of integrin? How much of total surface integrin is internalized via this pathway and not clathrin or caveolin pathways? Is CLIC (and fluid phase) endocytosis generally upregulated in CAV1 KO cells? If the endocytosis effects are most clearly seen at early times of cell spreading how does this relate to the mechanosensing experiments done on spread cells?
3. I am also a little confused as to how the authors envisage Cav1 regulating integrin recycling. Is Cav co-internalized with b-integrin and stabilizes it in endosomes, slowing recycling? Is this occurring in caveosomes? If this is the case it would be important to show it. Can the TIRF videos be quantified to support the ELISA data?
4. If the issues related to Cav1 vs caveolae mentioned in 1 are addressed, caveolae membrane buffering may be limiting integrin activation and an integrin-dependent tensive response. This is very interesting. Caveolae membrane buffering is thought to be an early response to mechanical stress so shouldn't WT cells show the same effect at longer times or with increased hypoosomotic pressure? Can this be tested? And what about other mechanical stressors?
5. Finally, we are shown in Figure 7 that Talin regulates adhesion and b-integrin activation in KO cells. Is Talin required for the increased integrin dynamics shown in previous figures?Reviewer #3:
This paper from Lolo et al. described experimentation aimed at determining how caveolin and caveolae influence integrin function. The use mouse embryonic fibroblasts (MEFs), and caveolin null MEFs that have, or have not been rescued with exogenously expressed caveolin. They first use magnetic tweezers to show that knockout of caveolin influences integrin mechanical behaviour. They then used a surface biotinylation assay to look at the endocytosis and endocytic recycling of beta1 integrin. They found that caveolin null MEFs had unaffected levels of integrin endocytosis, but that they recycled bet1 integrin more rapidly than wild-type MEFs. They also show that a small pool of beta1 integrin is internalised through a cdc42-dependent pathway and attribute this to the clathrin independent carrier (CLIC) pathway, but it is unclear from these data whether this is influenced by caveolin. The authors proceed from this to use immunofluorescence colocalisation studies and dominant negative mutants of Rab GTPases to determine that caveolin null MEFs may have altered integrin trafficking, possibly via a Rab4-dependent route. Finally, the authors show that hypoosmotic shock (to put the plasma membrane under tension) leads to increased integrin activation in caveolin null MEFs and that this may be talin-dependent.
In general this paper describes a series of observations which are not sufficiently well-connected to support the synthesis that the authors assemble for how caveolin may regulate integrin trafficking and function. Moreover, much of the experimentation and the way that the results are presented are neither convincing nor of the standard necessary for publication in eLife.
1. Figure 4. The authors have used a surface biotinylation method to show that integrin recycling is different in wild-type and caveolin null MEFs. These assays need to be performed with the rescue cells too.
2. The authors pursue their recycling results using immunofluorescence (Figure 6) to try to infer that differences in recycling, shown in Figure 4 are Rab4-dependent. I do not find any of the data in Figure 6 convincing. It is not clear to me why the various colocalisations tested represent recycling of integrins. Moreover, how can one look at colocalisation with dominant negative mutants of Rabs (which are known to localise aberrantly) and conclude that a particular cargo is in any given compartment?
3. How are the data in Figure 5 consistent with those in Figure 4? The authors show in Figure 4 that integrin endocytosis is not affected by caveolin knockout. In Figure 5, they show that an inhibitor of the CLIC pathway (although they do not validate its efficacy or specificity) slightly inhibits integrin endocytosis. They also show that fluid-phase endocytosis is increased in in caveolin null MEFs and infer that integrins are internalised through this pathway. But, in Figure 4 integrin endocytosis is not different between caveolin wild-type and null fibroblasts. Thus, I am completely at a loss to interpret the authors' concluding statement which reads: 'Thus, in the absence of Cav1/caveolae, early endocytosis of b1-integrin occurs at least in part through fluid-phase uptake, which provides an alternative entry route that would compensate for lack of Cav1-dependent internalization.'
4. There are no attempts, as far as I can see, to link the alterations in integrin trafficking to their activation. I would have thought that, having shown that integrin trafficking, their mechanical properties and their activation-state (9EG7 binding) are different between caveolin null and wild-type cells, the authors would want to attempt to determine which of these events are mechanistically connected. At least, for eLife I would expect some of these questions to be addressed.
5. The 9EG7 staining data presented in Figure 7 are really not convinding. I do not see how the authors can quantify images such as those shown in Figure 7D. The staining just looks like a splodge of fluorescence implying some kind of aggregation, or gross disruption of the cell membrane. Also, some of this staining looks to be at the cell surface (Figure 7L), and some is clearly intracellular (Figure 7I), so how can this necessarily reflect active, talin-bound integrin that is competent to bind ligand?
6. The presentation of much of the data in this manuscript is not of the standard that is necessary for acceptance in eLife. Many of the axes are not labelled (such as the y-axes in Figure 4F – I), or there are labels/designations that are unclear; for instance, what does 'recycling by timepoint 0 after 10 min endocytosis' mean (Figure 4G)? Also, there are metrics which are not described at all. For instance, I cannot seem to find in the manuscript (or in reference 19) what 'reinforcement increment' is and how it is obtained. Is it the same as 'relative stiffening' in reference 19?
7. The hugely increased 'reinforcement increment' of ConA beads in Figure 2H over fibronectin beads in Figure 2G doesn't seem to square with the authors' assertion that integrins couple the beads to the cytoskeleton to restrict their movement. It looks to me that these data indicate that the ConA beads are much more tightly associated with the cytoskeletal machinery than the fibronectin-coated ones.
8. To support their conclusion that Rho is not involved in Cav1 knockout mediated reinforcement (Figure 3C) don't the authors need to show that the reinforcement increment still increases when one compares Cav1+/+;p190KD and Cav1-/-; p190KD cells? Curiously, the authors have not shown this.
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Thank you for submitting the paper "Caveolae couple mechanical stress to integrin recycling and activation" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor.
Comments to the Authors:
We realize that you went to great effort to try to improve a prior manuscript but sadly must write to say that, after consultation with the reviewers, we have decided that this manuscript will not be considered further for publication by Life.
Reviewer #1 (Recommendations for the authors):
This paper proposes to link caveolin-1 regulation of the response to mechanical stress to beta1-integrin recycling. While some of the data are interesting, in particular, the Cav1-dependent response of cells to magnetic beads, other aspects of the paper are less than convincing. B1-integrin has been shown to be internalized by the CLIC pathway that is regulated by Cav1 and this paper extends that to argue that this pathway is regulated in response to mechanical stress. It is however not clear how Cav1 regulation of integrin recycling relates to caveolae buffering in response to mechanical stress.
1. While the magnetic bead approach provides a very nice approach to studying local b1-integrin internalization, studies of b1-integrin internalization are for the most part whole cell fixed cell analyses. Some TIRF videos are provided but are not very convincing and are not quantified. Extension of the magnetic bead approach to show local b1-integrin dynamics would be more convincing.
2. There are a number of issues with confocal microscopy. Non-permeabilized paraformaldehyde-fixed cells are presumed to report on surface labeling of the 9GE7 antibody specific for activated integrin. It is not always clear from the text which images were permeabilized or not and it should be noted that paraformaldehyde alone can disrupt the plasma membrane and enable antibody penetration to label cytoplasmic antigen, which often appears as bright localized puncta. The very bright labeling in the hypoosmotic shock treated CAV1 KO cells in figure 7 is suspect; also 1/20 hypoosmotic shock has a larger effect on activated integrin than 1/10 hypoosmotic shock? Some images seem to indicate an ER labeling including the nuclear membrane (i.e. Figure 6D). The only sure way to limit antibody labeling to the surface exposed antigens is to label viable cells at 4C. Colocalization in images that present very diffuse labeling on one channel and a small number of puncta in another is quantified using Pearson's colocalization and used to report on endocytosis. Small changes in normalized colocalization are reported questioning the extent of the reported effect. More appropriate fluorescent-based endocytosis assays are available and should be employed.
Reviewer #2 (Recommendations for the authors):
1. Many of the fluorescence images are difficult for the reader to interpret as presented. This is a critical issue given that many of the conclusions of this work rely on quantitative imaging. For example, it is very difficult to assess from overlayed images shown in many panels what structures actually co-localize. A general suggestion would be to provide in the main figures both the merged image and individual images of the two channels being examined, preferably in grayscale, so that the reader can better see the structures in question. Correspondingly, the graphs could easily be made much smaller than shown here.
2. Overall, the quality of the images is not as high as one would expect for an eLife paper. For example, in some panels, the fluorescence signal appears to be saturated. This is the case for example in Supplementary Figure 1 F and multiple panels in Supplementary Figure 5. In others, large blotchy structures are present, such as in Figure 7.
3. Much of the quantification of imaging data is normalized, and how it is normalized varies from panel to panel. While in principle there is nothing wrong with this, it makes it difficult for the reader to get a sense of the absolute magnitudes of quantities being measured, such as the degree of colocalization of various markers. This makes it difficult to get an overall sense of flux through various pathways.
4. As a reader I would like to have a better sense of how the numbers add up here to explain the phenotype reported in the first figure.
Reviewer #3. (Recommendations for the authors):
A major claim of the manuscript is that the threshold for PM-tension-driven β-1 integrin activation is lower in Cav1 KO cells. However, the experiments as presented are more simply explained by an inability of Cav1 KO cells to buffer membrane tension (already shown by Sinha et al. 2011 Cell), and it is not clear if membrane tension is altered as a consequence of Cav1 KO. This could be addressed by tether pulling experiments or the use of membrane tension probes (e.g. FlipperTR).
In addition, knockdown of talin 1 and 2 is broadly understood to decrease integrin activity across a variety of adherent cell types, and the claim that talin is required for recycling to support active integrin at the cell surface is therefore not well founded.
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Thank you for resubmitting your work entitled "Caveolae couple mechanical stress to integrin recycling and activation" for further consideration by eLife. Your revised article has been evaluated under the supervision of Suzanne Pfeffer (Senior Editor).
You will be pleased to learn that one reviewer is now fully satisfied and the second believes that the manuscript can be published in eLife if you are able to address the following minor issues:
1. The Blot in Figure 1M is still of very low quality and the authors have still not included any compartment markers to determine that their plasma membrane fraction is free of contamination from other compartments. The authors' argument that β 1 integrin can be used as a marker of the plasma membrane obviously falls down in the light of the fact that the authors are studying its trafficking through endosomes and, in doing so, show that an appreciable fraction of this integrin is NOT present at the plasma membrane. Also, there are still no molecular weight markers on this blot or any of the other blots shown in the manuscript. On another note, the blot in Figure 1M is interesting because, if the molecular weight markers run where I think they should, it appears to show that re-expression of Cav1 increases the quantity of mature (120kDa) beta1 integrin at the plasma membrane whereas, in the Cav1KO cells, it is primarily the immature (100kDa) pro-form of beta1 that is at the plasma membrane.
2. There are no scale bars on any of the micrographs
3. I think that reference 46, not 26, is the one for the ELISA-based endocytosis/recycling protocol. Also, I don't think that reference 47 is correct for supporting that beta3 integrin follows a Rab4-dependent "short" loop pathway.
4. The authors have now included, as requested, data showing that re-expression of Cav1 slows-down recycling in Cav1KO cells. But why have the authors shown only active beta1? Surely, they should be showing the recycling data for both total and active beta1 – Indeed, they must have these data as, presumably, they would have split their lysate into ELISA plates coated with antibodies recognising total b1 as well as active beta1. Also, the levels of recycling demonstrated for Cav1KO cells in supplementary Figure 2I differs quite markedly from that shown in Figure 4I.
https://doi.org/10.7554/eLife.82348.sa1Author response
[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]
Reviewer #1:
Caveolae are now widely recognized for their roles in the mechanobiology of the plasma membrane of cells submitted to mechanical stress. In the current manuscript, the authors have studied the crosstalk between caveolae and another mechanosensitive group of proteins, the integrins. Using as a model system mouse embryonic fibroblasts (MEFs) that are deleted for the caveolin-1 gene, it is described that cell adhesion to fibronectin-coated surfaces is increased in the absence of caveolae. The underlying reason for this is identified in the increased presence of the active conformation of beta1 integrin at the cell surface, which results from 2 phenomena: increased endocytic recycling, and increased activation upon mechanical stress.
The overall area of research on mechanobiology in relation with membrane trafficking that is addressed in the current study is of high interest to a general readership in the life sciences. The study is based on a rich arsenal of relevant techniques, and represents a substantial amount of work. Some techniques such as cell adhesion stiffness measurements based on magnetic bead oscillation or traction force microscopy are really quite elegant. The manuscript is well written, and pleasant to read.
Possibly the most important limitation of the current study is that it remains descriptive on the molecular mechanism aspects. It is clear that the absence of Cav1/caveolae has a number of effects that are clearly defined in the current study. The molecular wiring that underlies these effects remains unexplored at this stage. A few examples: In the absence of Cav1/caveolae, endocytic recycling of beta1 integrin passes from Rab11-dependent slow recycling to Rab4-dependent fast recycling – by which molecular mechanism does Cav1/caveolae connect to the endosomal machinery? Mechanical stress increases beta1 activation in Cav1KO MEFs – by which molecular mechanism does Cav1/caveolae limit this activation in wild-type conditions? Talin is important in the context of this mechanical stress-controlled activation reaction – by which molecular mechanism does Cav1/caveolae interact with talin? Having said this, I still believe that the manuscript will contribute in a substantial manner to the dynamic field of membrane biology research. The following points should be clarified, though.
Lines 262-269: The faster beta1 integrin recycling phenotype in Cav1KO MEFs is seen only after 10 min of endocytosis, and not when recycling is measured after 5 min of endocytosis. The authors suggest that "beta1-integrin is stabilized in the presence of Cav1 after 10 minutes endocytosis". How would such stabilization work? The authors most likely don't have the experimental response to this question at this stage. However, it would be helpful if some ideas could be discussed on the molecular mechanisms by which this would work.
We thank the reviewer for these comments. Although we still do not have a complete account as to how Cav1 could affect β 1 integrin stabilization, we have now gathered some evidence showing both increased surface and endosomal (EEA-1 positive) active β 1 integrin in Cav1WT MEFs loaded with cholesterol either by LDL or U18666A treatment, phenocopying Cav1KO phenotype (Suppl. Figure 2J-M). This could be indicative of a Cav1-dependent cholesterol threshold above which beta1 integrin trafficking would be altered. Consistently, previous work in our lab has demonstrated increased exocytic activity upon cholesterol loading in Cav1WT MEFs.
(1). We have now included these ideas both in Suppl. Figure 2 and in the Discussion section.
Lines 315-317: In the paragraphs that precede these lines, the authors present experiments based on the small molecule Cdc42 inhibitor ML141 that lead to the conclusion that in Cav1KO MEFs, part of endocytic uptake occurs by the CLIC/GEEC pathway, while this does not appear to be the case in MEFs that express Cav1. It is concluded that in "wild-type" conditions, part of uptake occurs through caveolae, and that this Cav1-dependent internalization would be compensated by the CLIC/GEEC pathway in the absence of Cav1/caveolae. Can other interpretations be excluded? For example, wouldn't one observe the same phenotype if Cav1/caveolae were to inhibit uptake of a fraction of the beta1 integrin molecules that once liberated in the Cav1KO condition would now be internalized by the CLIC/GEEC pathway? The most direct way to address this point would be electron microscopy to provide ultrastuctural images of the uptake structures in which beta1 integrin is found in the different experimental conditions. To the least, alternative interpretations of the data should be discussed.
We have now included a set of new experiments in Suppl. Figure 3, delineating the relative contribution of different endocytic pathways comparing wild type and Cav1KO MEFs. We now show that active β 1 integrin is mainly endocytosed by Cav1-dependent mechanisms in wild type MEFs as it: (i) co-localized with Cav1 and LacCer, (ii) was significantly reduced upon genistein treatment (a caveolar endocytosis inhibitor (2)) and (iii) was unaffected by ML141 treatment (the CLIC inhibitor), On the other hand, it mainly follows the CLIC-dependent endocytosis in Cav1KO MEFs as it: (i) co-localized with CD44 (a CLIC endocytic marker (5)) and (ii) was significantly reduced upon ML141 treatment as compared to wild type MEFs. Finally, no significant differences were observed in clathrin-dependent β 1 integrin endocytosis between wild type and Cav1KO MEFs. These results might indicate as suggested by the reviewer, that Cav1 is limiting the CLICdependent endocytosis of a pool of active β 1 integrin in wild type MEFs, which becomes available in Cav1KO MEFs. We have now included these ideas in the Discussion section.
Reviewer #2:
This manuscript describes the role of Caveolin-1 (Cav1) in integrin recycling using Cav1 knockout MEFs and links integrin endocytosis and recycling to cell mechanosensing and adhesion. I find the subject interesting and, in light of the role of Cav1 and caveolae in the cellular response to mechanical stress, timely and relevant. That being said I find the manuscript to be a compilation of interesting results that lack mechanistic connection between them. We are provided with interesting data on: (1) the role of Cav1 in integrin mechanosensing using an elegant magnetic tweezer approach; (2) Cav1 regulation of integrin endocytosis and recycling but with no direct link to mechanosensing or stretch; (3) the role of talin in regulating CAV1-independent integrin activation. It seems as though Cav1 is altering integrin dynamics and response to mechanical stretch which is quite interesting. While the schematic in figure 8 highlights the Cav1-dependent changes reported in the paper, mechanistic connections between them need to be clarified and I have many questions which I outline below.
1. "Cav1/caveolae" Cav1 and caveolae are not the same thing and it cannot be assumed that effects observed in Cav1 KO cells are necessarily attributed to loss of caveolae. I am concerned about use of this term throughout the paper (even the title) and suggest that it would be important to define the specific role of caveolae in the processes described using cell lines expressing Cav1 but not PTRF. Indeed, the only data linking caveolae flattening to the effects shown is Figure 7 and is interesting in that it suggests that lack of membrane buffering by caveolae induces an integrin response. However, to definitively show that this is caveolae dependent and not Cav1 dependent it would be important to use cells expressing Cav1 but not caveolae.
We thank the reviewer for these comments. We have now performed a series of new experiments included in Suppl. Figure 5K-5M where we show that β 1 integrin is still activated in PTRFKO MEFs (that lack caveolae but express some levels of Cav1) after treatment with hypoosmotic shock. These results indicate that lacking the membrane buffering provided by caveolae leads to β 1 integrin activation in Cav1KO MEFs. Accordingly, we have now changed Cav1/caveolae for caveolae throughout the text, including the title.
2. The studies on b-integrin endocytosis switch to a CLIC pathway in Cav1 KO is interesting and supports a role for Cav1 as an inhibitor of CLIC endocytosis. It would be important to provide more evidence for CLIC endocytosis than inhibition of Cdc42. Is b-integrin cointernalized with CD44? Is this pathway CD44- or raft-dependent? Can the effects of Cav1 on integrin mechanosensing be attributed to CLIC endocytosis of integrin? How much of total surface integrin is internalized via this pathway and not clathrin or caveolin pathways? Is CLIC (and fluid phase) endocytosis generally upregulated in CAV1 KO cells? If the endocytosis effects are most clearly seen at early times of cell spreading how does this relate to the mechanosensing experiments done on spread cells?
We have now included a series of new co-localization studies in Suppl. Figure 3 delineating the relative contribution of different endocytic routes to β 1 integrin endocytosis. Specifically, we now show that anti-active β 1 integrin antibody colocalizes with Cav1 after 3 minutes of endocytosis at 37ºC, and this entry is significantly reduced by genistein treatment (a common caveolar endocytosis inhibitor (2)) but it is unaffected by ML141 treatment (CLIC inhibitor). Likewise, BODIPY-LacCer (a caveolar endocytic marker (3, 4)) co-localizes with endogenous active β 1 integrin (9EG7) after 3 minutes of endocytosis at 37ºC, which is also significantly reduced by genistein treatment. On the other hand, anti-active β 1 integrin antibody colocalizes with anti-CD44 antibody (a CLIC endocytic marker (5)) after 3 minutes of endocytosis at 37ºC, and this entry is significantly blocked by ML141 in Cav1KO MEFs but not in wild type MEFs. Finally, we observed no significant differences in the co-localization of anti-active β 1 integrin and transferrin (a clathrin endocytic marker) after 3 minutes of endocytosis at 37ºC in wild type and Cav1KO MEFs. Altogether, these results suggest that CLIC-dependent β 1 integrin endocytosis is blocked in wild type MEFs, where β 1 integrin is mainly endocytosed by caveolae, and becomes free in Cav1KO MEFs, where it is mainly endocytosed by CLICs.
Regarding cell spreading: as stated in material and methods section where we described the endocytosis/recycling assay, “cells were incubated in complete DMEM for 2 hours at 37oC”, this was the shortest time-point require for cells to stand assay conditions, and also matches the magnetic tweezers measurements experimental settings, where cells were allowed to spread for a short period of time (30 minutes). We have now included this information in the corresponding material and method section.
3. I am also a little confused as to how the authors envisage Cav1 regulating integrin recycling. Is Cav co-internalized with b-integrin and stabilizes it in endosomes, slowing recycling? Is this occurring in caveosomes? If this is the case it would be important to show it. Can the TIRF videos be quantified to support the ELISA data?
We still do not know the exact mechanism by which Cav1 is regulating integrin recycling, however we suspect it depends on Cav1-dependent cholesterol organization as we have obtained some results showing both increased surface and endosomal (EEA-1 positive) active-β 1 integrin upon loading Cav1WT MEFs with cholesterol either by LDL or U18666A treatment, phenocopying Cav1KO phenotype (Suppl. Figure 2J-M). This might indicate that Cav1 determines a cholesterol threshold above which beta1 integrin trafficking is altered. Accordingly, previous work in our lab has demonstrated increased exocytic activity upon cholesterol loading in Cav1WT MEFs (1), which in our conditions would affect β integrin trafficking dynamics. We have now included these ideas in Suppl. Figure 2J-M and in the Discussion section.
Regarding TIRF videos, we have now included new experiments and quantifications (Suppl. Figure 2C-F and Suppl. Videos 6 and Video 7) showing that integrin trafficking in Cav1WT MEFs resembles that observed in Cav1KO MEFs after treatment with high hypoosmotic pressure conditions. These results might indicate that increasing plasma membrane tension beyond caveolae buffering capacity changes integrin trafficking dynamics.
4. If the issues related to Cav1 vs caveolae mentioned in 1 are addressed, caveolae membrane buffering may be limiting integrin activation and an integrin-dependent tensive response. This is very interesting. Caveolae membrane buffering is thought to be an early response to mechanical stress so shouldn't WT cells show the same effect at longer times or with increased hypoosomotic pressure? Can this be tested? And what about other mechanical stressors?
We have now included a set of new experiments pertaining caveolae buffering and integrin activation after hypoosmotic treatment in suppl. Figure 5N-5W, where we show that integrin activation also occurs in Cav1WT MEFs after both longer treatment times and with increased hypoosmotic pressure, as compared to Cav1KO MEFs. These results suggest that caveolae membrane buffering is limiting integrin activation before a plasma membrane tension threshold is reached.
5. Finally, we are shown in Figure 7 that Talin regulates adhesion and b-integrin activation in KO cells. Is Talin required for the increased integrin dynamics shown in previous figures?
To answer this question, we have performed two complementary experiments now included in Supp. Figure 6F-6M: (1) co-localization studies of 9EG7 (for testing endogenous active β 1 integrin) or anti-active beta1-488 antibody (for testing integrin endocytosis) and EEA-1 (early endosomal marker), and (2) biotinylation assays of surface-active β 1 integrin in Talin-1 and 2silenced Cav1KO MEFs. Interestingly, surface active β 1 integrin is reduced and consistently increased intracellularly in EEA-1 positive endosomes, upon talin1/2 knockdown. Collectively, these new results suggest that talin is modulating not only adhesion and integrin activation but also integrin trafficking in Cav1KO MEFs.
Reviewer #3:
This paper from Lolo et al. described experimentation aimed at determining how caveolin and caveolae influence integrin function. The use mouse embryonic fibroblasts (MEFs), and caveolin null MEFs that have, or have not been rescued with exogenously expressed caveolin. They first use magnetic tweezers to show that knockout of caveolin influences integrin mechanical behaviour. They then used a surface biotinylation assay to look at the endocytosis and endocytic recycling of beta1 integrin. They found that caveolin null MEFs had unaffected levels of integrin endocytosis, but that they recycled bet1 integrin more rapidly than wild-type MEFs. They also show that a small pool of beta1 integrin is internalised through a cdc42-dependent pathway and attribute this to the clathrin independent carrier (CLIC) pathway, but it is unclear from these data whether this is influenced by caveolin. The authors proceed from this to use immunofluorescence colocalisation studies and dominant negative mutants of Rab GTPases to determine that caveolin null MEFs may have altered integrin trafficking, possibly via a Rab4-dependent route. Finally, the authors show that hypoosmotic shock (to put the plasma membrane under tension) leads to increased integrin activation in caveolin null MEFs and that this may be talin-dependent.
In general this paper describes a series of observations which are not sufficiently well-connected to support the synthesis that the authors assemble for how caveolin may regulate integrin trafficking and function. Moreover, much of the experimentation and the way that the results are presented are neither convincing nor of the standard necessary for publication in eLife.
1. Figure 4. The authors have used a surface biotinylation method to show that integrin recycling is different in wild-type and caveolin null MEFs. These assays need to be performed with the rescue cells too.
We thank the reviewer for all these comments. We have now included a set of new recycling experiments in Supp. Figure 2I, comparing Cav1KO and Cav1KO+Cav1 reconstituted MEFs. The results now show that recycling is slow down upon Cav1 reconstitution indicating that the phenotype is Cav1 specific.
2. The authors pursue their recycling results using immunofluorescence (Figure 6) to try to infer that differences in recycling, shown in Figure 4 are Rab4-dependent. I do not find any of the data in Figure 6 convincing. It is not clear to me why the various colocalisations tested represent recycling of integrins. Moreover, how can one look at colocalisation with dominant negative mutants of Rabs (which are known to localise aberrantly) and conclude that a particular cargo is in any given compartment?
We thank the reviewer for pointing out the mistake with 9EG7 and Rab11 co-localization under Rab11 DN transfection. We have now repeated the experiment and performed a co-localization between 9EG7 and LBPA (lysobisphosphatidic acid, a late endosomal marker) instead of Rab11. We now show that blocking Rab11-dependent recycling affect β 1 trafficking both in wild type and Cav1KO MEFs, whereas blocking Rab4-dependent recycling only affect β 1 trafficking in Cav1KO MEFs. Regarding the quality of proof of co-localization studies, it is important to stress that Figure 4 (co-localization studies to analyze endosomal β 1 integrin) has to be read in conjunction with Suppl. Figure 4 (surface biotinylation assays to analyze changes in surface β 1 integrin), as referred to in the text. The former (Figure 4) shows how blocking two well-known recycling pathways (“long-loop”, Rab11-dependent and “short-loop”, Rab4-dependent) leads to a differential endosomal accumulation of active β 1 integrin in wild type and Cav1KO MEFs; the latter (Suppl. Figure 4), on the other hand, reveals a differential reduction in active β 1 integrin surface levels under the same experimental conditions. Combining both results led us to propose the most likely conclusion, i.e. that β 1 integrin preferentially follows a Rab11-dependent recycling pathway in wild type MEFs, whereas it is partially switched to a fast, Rab4-dependent recycling in Cav1KO MEFs.
3. How are the data in Figure 5 consistent with those in Figure 4? The authors show in Figure 4 that integrin endocytosis is not affected by caveolin knockout. In Figure 5, they show that an inhibitor of the CLIC pathway (although they do not validate its efficacy or specificity) slightly inhibits integrin endocytosis. They also show that fluid-phase endocytosis is increased in in caveolin null MEFs and infer that integrins are internalised through this pathway. But, in Figure 4 integrin endocytosis is not different between caveolin wild-type and null fibroblasts. Thus, I am completely at a loss to interpret the authors' concluding statement which reads: 'Thus, in the absence of Cav1/caveolae, early endocytosis of b1-integrin occurs at least in part through fluid-phase uptake, which provides an alternative entry route that would compensate for lack of Cav1-dependent internalization.'
We have now clarified this point by including a series of new co-localization studies in Suppl. Figure 3, delineating the relative contribution of different endocytic routes to β 1 integrin endocytosis. Specifically, we now show that anti-active β 1 integrin antibody colocalizes with Cav1 after 3 minutes of endocytosis at 37ºC, and this entry is significantly reduced by genistein treatment (a common caveolar endocytosis inhibitor (2)) but it is unaffected by ML141 treatment (CLIC inhibitor). Likewise, BODIPY-LacCer (a caveolar endocytic marker (3, 4)) co-localizes with endogenous active β 1 integrin (9EG7) after 3 minutes of endocytosis at 37ºC, which is also significantly reduced by genistein treatment. On the other hand, anti-active β 1 integrin antibody colocalizes with anti-CD44 antibody (a CLIC endocytic marker (5)) after 3 minutes of endocytosis at 37ºC, and this entry is significantly blocked by ML141 in Cav1KO MEFs (which proves ML141 specificity, blocking CLIC endocytosis) but not in wild type MEFs. Finally, we observed no significant differences in the co-localization of anti-active β 1 integrin and transferrin (a clathrin endocytic marker) after 3 minutes of endocytosis at 37ºC in wild type and Cav1KO MEFs. Altogether, these results suggest that CLIC-dependent β 1 integrin endocytosis is blocked by Cav1 in wild type MEFs, where it is mainly endocytosed by caveolae, and becomes free in Cav1KO MEFs, where it is mainly endocytosed by CLICs.
4. There are no attempts, as far as I can see, to link the alterations in integrin trafficking to their activation. I would have thought that, having shown that integrin trafficking, their mechanical properties and their activation-state (9EG7 binding) are different between caveolin null and wild-type cells, the authors would want to attempt to determine which of these events are mechanistically connected. At least, for eLife I would expect some of these questions to be addressed.
We have now included new experiments aiming at mechanistically connect integrin activation and trafficking dynamics:
1. We have performed new TIRF experiments and quantifications (Suppl. Figure 2C-F) showing that integrin trafficking in Cav1WT MEFs resembles that observed in Cav1KO MEFs after treatment with high hypoosmotic pressure conditions. These results might indicate that increasing plasma membrane tension beyond caveolae buffering capacity changes integrin trafficking dynamics.
2. Regarding talin activity, we have performed two complementary experiments now included in Supp. Figure 6F-6M: (1) co-localization studies of 9EG7 (for testing endogenous active β 1 integrin) or anti-active beta1-488 antibody (for testing integrin endocytosis) and EEA-1 (early endosomal marker), and (2) biotinylation assays of surface-active β 1 integrin in Talin-1 and 2silenced Cav1KO MEFs. Interestingly, surface active β 1 integrin is reduced and consistently increased intracellularly in EEA-1 positive endosomes, upon talin1/2 knockdown. Collectively, these new results suggest that talin is modulating not only adhesion and integrin activation but also integrin trafficking in Cav1KO MEFs.
5. The 9EG7 staining data presented in Figure 7 are really not convinding. I do not see how the authors can quantify images such as those shown in Figure 7D. The staining just looks like a splodge of fluorescence implying some kind of aggregation, or gross disruption of the cell membrane. Also, some of this staining looks to be at the cell surface (Figure 7L), and some is clearly intracellular (Figure 7I), so how can this necessarily reflect active, talin-bound integrin that is competent to bind ligand?
All the immunofluorescences shown in Figure 7 were done following the extracellular staining described in material and methods, i.e., after fixation, cells were immediately incubated with the primary antibody without permeabilization, so the majority of the signal comes from surface active β 1 integrin (9EG7 antibody). We show the difference between permeabilized and nonpermeabilized samples in Supp. Figure 1 (please compare 1H and 1I with 1H and 1I). However, to definitively show that β 1 integrin is still active and capable of ligand binding, we have now included new immunostaining of talin and soluble Fibronectin in Cav1KO MEFs after hypoosmotic treatment. Suppl. Figure 6B-6E shows co-localization between β 1 integrin (9EG7 antibody) and both fibronectin-FITC (6B and 6C) and talin (6D and 6E) which proves its active conformation.
6. The presentation of much of the data in this manuscript is not of the standard that is necessary for acceptance in eLife. Many of the axes are not labelled (such as the y-axes in Figure 4F – I), or there are labels/designations that are unclear; for instance, what does 'recycling by timepoint 0 after 10 min endocytosis' mean (Figure 4G)? Also, there are metrics which are not described at all. For instance, I cannot seem to find in the manuscript (or in reference 19) what 'reinforcement increment' is and how it is obtained. Is it the same as 'relative stiffening' in reference 19?
We apologize for these shortcomings. We have now included axes labels where necessary and better explained how recycling is represented. Reinforcement increment refers to the relative change in reinforcement, calculated as the difference between the last and initial measurements; we have now clarified this point in figure legend.
7. The hugely increased 'reinforcement increment' of ConA beads in Figure 2H over fibronectin beads in Figure 2G doesn't seem to square with the authors' assertion that integrins couple the beads to the cytoskeleton to restrict their movement. It looks to me that these data indicate that the ConA beads are much more tightly associated with the cytoskeletal machinery than the fibronectin-coated ones.
Concanavalin A is a lectin that binds sugars, as those presented in glycosylated proteins and lipids within the cellular surface; therefore, conA-coated beads provide a non-specific binding as compared to FN-coated beads, which specifically bind to integrins. Assuming that the average density of surface sugars is higher than integrins, ConA-coated beads surface adhesion will be higher than FN-coated ones, resulting in higher absolute reinforcements.
8. To support their conclusion that Rho is not involved in Cav1 knockout mediated reinforcement (Figure 3C) don't the authors need to show that the reinforcement increment still increases when one compares Cav1+/+;p190KD and Cav1-/-; p190KD cells? Curiously, the authors have not shown this.
The same Cav1KO MEFs used in this manuscript have been extensively studied in our laboratory (10-12). We have previously showed that Cav1 modulates cell contraction by regulating specifically Rho activity through p190RhoGAP localization to the PM. In the absence of Cav1, PM-localized p190RhoGAP increases, leading to a significant reduction in Rho activity, dampening cellular contractility (as we show by traction force microscopy, Figure 3D) and related processes. Accordingly, knocking down p190RhoGAP in Cav1KO MEFs completely rescues all phenotypes we have evaluated so far (10, 12). The fact that we observed no significant differences in magnetic tweezers measurements as compared to traction force (Figure 3C and 3D), speaks of a minor role of Rho in early cellular reinforcement. Consistently, different reports have previously shown initial integrin adhesion in the absence of cytoskeleton connection, suggesting that early mechanosensing could be locally triggered (13-15). We have now included these ideas and references in the discussion.
References
I. N.-L. Lucas Albacete-Albacete, Juan Antonio López , Inés Martín-Padura, Alma M. Astudillo, Alessia Ferrarini, Michael Van-Der-Heyden, Jesús Balsinde , Gertraud Orend , Jesús Vázquez , Miguel Ángel del Pozo, ECM deposition is driven by caveolin-1–dependent regulation of exosomal biogenesis and cargo sorting. Journal of cell biology 11, (2020).
[Editors’ note: what follows is the authors’ response to the second round of review.]
Comments to the Authors:
We realize that you went to great effort to try to improve a prior manuscript but sadly must write to say that, after consultation with the reviewers, we have decided that this manuscript will not be considered further for publication by Life.
Reviewer #1 (Recommendations for the authors):
This paper proposes to link caveolin-1 regulation of the response to mechanical stress to beta1-integrin recycling. While some of the data are interesting, in particular, the Cav1-dependent response of cells to magnetic beads, other aspects of the paper are less than convincing. B1-integrin has been shown to be internalized by the CLIC pathway that is regulated by Cav1 and this paper extends that to argue that this pathway is regulated in response to mechanical stress. It is however not clear how Cav1 regulation of integrin recycling relates to caveolae buffering in response to mechanical stress.
We thank the reviewer for her/his positive appreciation of the magnetic tweezers studies. We have now performed new experiments to specifically connect integrin recycling with the buffering ability of caveolae (new panels in Figure 6P-6V).
1. While the magnetic bead approach provides a very nice approach to studying local b1-integrin internalization, studies of b1-integrin internalization are for the most part whole cell fixed cell analyses. Some TIRF videos are provided but are not very convincing and are not quantified. Extension of the magnetic bead approach to show local b1-integrin dynamics would be more convincing.
We thank the reviewer for these comments. Although an important part of the beta1-integrin internalization studies is based on whole cell stainings after fixation, we have also performed an extensive array of in vivo experiments, examples can be found in: (i) Figure 4D-4I and Suppl. Figure 2G2I, with ELISA-based endocytosis/recycling assays, and (ii) Suppl. Figure 2J, Suppl. Figure 4A-4C and Suppl. Figure 6V, where surface active beta1-integrin is analyzed after altering trafficking dynamics by changing cholesterol homeostasis (Suppl. Figure 2J), or different components of the recycling machinery (Suppl. Figure 4A-4C and Suppl. Figure 6V).
We sincerely apologize if TIRF-related information was not clearly conveyed, as we consider it an important addition to support our main conclusions. We have indeed provided quantifications of TIRF videos, as shown in Suppl. Figure 2C-2F and explained in Supplementary information: “Quantification of TIRF videos show normalized fluorescence integrated density (IntDen) over frames (Suppl. Figure 2C-2E). Graph represents the mean of the difference between normalized fluorescence integrated density of adjacent frames (framex-framex-1) (Suppl. Figure 2F). IntDen was calculated as in the following formula: IntDen= Raw Integrated density (sum of pixel values in selection) x (Area in scaled units)/ (Area in pixels), which was then normalized by the IntDen mean of all frames analyzed”. We have now clarified this in the main text. We also provide two additional TIRF videos comparing Cav1WT MEFs transfected with beta1-GFP before and after hypoosmotic treatment (with DMEM diluted 1:20 in distilled water). We hope differences across conditions are clearer now.
Finally, as an extension of the magnetic bead experiments shown in Figure 2G and 2H, we also performed some magnetic twisting experiments (as shown in Suppl. Figure 5A-5J), which also allow for probing cell micromechanics. Results showed a significant tension-induced increase in beta1-integrin activation in Cav1KO MEFs as compared to Cav1WT MEFs, supporting results from other experiments and reinforcing the main conclusion of the manuscript, i.e., that caveolae regulate both integrin activation and recycling. However, to further support our claims, we have now performed a new set of experiments to specifically link integrin recycling to caveolae buffering in response to mechanical stress. To do so, we incubated Cav1KO and Cav1WT MEFs with anti-active beta1-Alexa 488 antibody for 1 hour at 4oC followed by 3 minutes endocytosis at 37oC. Immediately after that, we challenged cells for 1 minute with either DMEM diluted 1:10 (which can be buffered by caveolae flattening, as previously described, Sinha et al. Cell 2011, and is shown in our own results in both Figure 7A-7F and Suppl. Figure 5N-5W), or DMEM diluted 1:20 in distilled water, which exceeds the buffering capacity of caveolae (as we have shown in Suppl. Figure 5N-5W). To remove any remaining surface fluorescence, two quick steps of acid stripping were done prior to fixation with PFA. Strikingly, whereas Cav1KO MEFs showed a significant reduction in the endocytosed active beta1-integrin pool both at 1:10 and 1:20 dilutions, Cav1WT MEFs only showed a similar significant reduction at 1:20 dilution (new panels in Figure 6P-6V). These results indicate that caveolae buffering prevents integrin recycling below a certain force threshold, and therefore regulate integrin dynamics in response to mechanical stress.
2. There are a number of issues with confocal microscopy. Non-permeabilized paraformaldehyde-fixed cells are presumed to report on surface labeling of the 9GE7 antibody specific for activated integrin. It is not always clear from the text which images were permeabilized or not and it should be noted that paraformaldehyde alone can disrupt the plasma membrane and enable antibody penetration to label cytoplasmic antigen, which often appears as bright localized puncta. The very bright labeling in the hypoosmotic shock treated CAV1 KO cells in figure 7 is suspect; also 1/20 hypoosmotic shock has a larger effect on activated integrin than 1/10 hypoosmotic shock? Some images seem to indicate an ER labeling including the nuclear membrane (i.e. Figure 6D). The only sure way to limit antibody labeling to the surface exposed antigens is to label viable cells at 4C. Colocalization in images that present very diffuse labeling on one channel and a small number of puncta in another is quantified using Pearson's colocalization and used to report on endocytosis. Small changes in normalized colocalization are reported questioning the extent of the reported effect. More appropriate fluorescent-based endocytosis assays are available and should be employed.
Reviewer 1 raises an important point regarding surface staining, for which we are grateful. Following this recommendation we have repeated the analysis by incubating cells with 9EG7 antibody at 4oC for 1h. Although some of the previous cytoplasmic integrin staining was lost, results are analogous to those previously observed (new panels in Suppl. Figure 6A-6J). Additionally, we have clarified in figures legends whether extracellular staining was done before or after fixation. According to previous observations in the lab (unpublished data, currently under review in another journal), the extension in osmolarity reduction correlates with plasma membrane tension increase. This implies that 1/20 dilution leads to higher plasma membrane tension increase than 1/10 dilution, and therefore could induce a larger effect on integrin activation, which is what we have observed. Altogether, these results suggest that caveolae membrane buffering is limiting integrin activation before a plasma membrane tension threshold is reached.
As rightly indicated by the referee, CLIC-dependent beta1-integrin endocytosis has previously been connected to Cav1 regulation, and our co-localizations studies were meant to confirm these results. However, to better show the co-localizing structures, we now provide both the merged image and the individual images of the two channels in all figures. Additionally, to avoid possible bias of normalized Pearson’s correlations, we have quantified mean fluorescence intensity of endocytosed beta1-integrin (followed by anti-active beta1-Alexa 488 antibody, as now shown in Suppl. Figure 3D´ and 3L´) after treatment with either genistein (a Cav1-dependent endocytosis inhibitor) or ML-141 (a CLIC-dependent endocytosis inhibitor). In line with our previous observations, beta1-integrin endocytosis was significantly reduced in Cav1WT MEFs after treatment with genistein, whereas it was unaffected after treatment with ML-141. In contrast, beta1-integrin endocytosis was significantly reduced in Cav1KO MEFs treated with the CLIC inhibitor, ML-141. Again, these results indicate that beta1-integrin endocytosis is mainly Cav1-dependent in Cav1WT MEFs and mainly CLIC-dependent in Cav1KO MEFs. We have now included these new analyses in Suppl. Figure 3D´and 3L´.
Reviewer #2 (Recommendations for the authors):
1. Many of the fluorescence images are difficult for the reader to interpret as presented. This is a critical issue given that many of the conclusions of this work rely on quantitative imaging. For example, it is very difficult to assess from overlayed images shown in many panels what structures actually co-localize. A general suggestion would be to provide in the main figures both the merged image and individual images of the two channels being examined, preferably in grayscale, so that the reader can better see the structures in question. Correspondingly, the graphs could easily be made much smaller than shown here.
We thank the reviewer for these comments. We have now modified all figures to show both merged as well as separated channels to better see how the different structures co-localize.
2. Overall, the quality of the images is not as high as one would expect for an eLife paper. For example, in some panels, the fluorescence signal appears to be saturated. This is the case for example in Supplementary Figure 1 F and multiple panels in Supplementary Figure 5. In others, large blotchy structures are present, such as in Figure 7.
We apologize for this, as embedding figures throughout the main text affected their overall resolution. We have now removed them and included as separated files to improve their quality. It is true, as fairly indicated by the reviewer, that some images show fluorescence saturation and the presence of large blotchy structures within some integrin stainings; however, in order to compare signals across conditions we were forced to use the same laser settings, that together with the large dynamic range of intensities within the integrin signal, led to such unavoidable problems. We have now repeated the analysis by incubating cells with 9EG7 antibody at 4oC for 1h prior to fixation, which reduced any potential antibody penetration and cytoplasmic signal accumulation. Of note, resulting images, now free of any aggregated structures, led to analogous results as those previously observed (new panels in Suppl. Figure 6A-6J). Additionally, we have also improved the quality of diagrams in Figures 8 and Suppl. Figure 6W.
3. Much of the quantification of imaging data is normalized, and how it is normalized varies from panel to panel. While in principle there is nothing wrong with this, it makes it difficult for the reader to get a sense of the absolute magnitudes of quantities being measured, such as the degree of colocalization of various markers. This makes it difficult to get an overall sense of flux through various pathways.
We thank the reviewer for these comments and apologize for having made such confusion. We have now better clarified both in figures and captions the corresponding normalization procedure used. We hope it is now easier to get the overall flow of results. Generally, in co-localization studies we have normalized each Pearson’s correlation coefficient by the mean of the control condition. In those experiments where we quantified fluorescence intensity, values were normalized to mean fluorescence intensity of the control condition when analyzing anti-active beta1-Alexa 488 antibody signal, or to each analyzed area and finally referred to area control when analyzing 9EG7 antibody signal (for more details, please have a look at accompanying raw data excel files).
4. As a reader I would like to have a better sense of how the numbers add up here to explain the phenotype reported in the first figure.
We are sorry but figure 1 does not contain any quantification, we please ask the reviewer if she/he can better specify the figure he/she is referring to.
[Editors’ note: what follows is the authors’ response to the third round of review.]
You will be pleased to learn that one reviewer is now fully satisfied and the second believes that the manuscript can be published in eLife if you are able to address the following minor issues:
1. The Blot in Figure 1M is still of very low quality and the authors have still not included any compartment markers to determine that their plasma membrane fraction is free of contamination from other compartments. The authors' argument that β 1 integrin can be used as a marker of the plasma membrane obviously falls down in the light of the fact that the authors are studying its trafficking through endosomes and, in doing so, show that an appreciable fraction of this integrin is NOT present at the plasma membrane. Also, there are still no molecular weight markers on this blot or any of the other blots shown in the manuscript. On another note, the blot in Figure 1M is interesting because, if the molecular weight markers run where I think they should, it appears to show that re-expression of Cav1 increases the quantity of mature (120kDa) beta1 integrin at the plasma membrane whereas, in the Cav1KO cells, it is primarily the immature (100kDa) pro-form of beta1 that is at the plasma membrane.
We apologize for the low quality of the blot shown in Figure 1M. We have now repeated plasma membrane purification, substituting β 1 integrin with Transferrin Receptor as a plasma membrane (PM) marker (now in revised Figure 1M). We have now included molecular markers in all blots, additionally, whole unprocessed gels are included as source data for visual inspection. We thank the reviewer for his/her interesting observation on integrin maturation, as it is quite certain that Cav1 re-expression in Cav1KO MEFs seems to increase mature β 1 integrin (molecular markers were indeed running where the reviewer was thinking). Although we have now replaced beta1 integrin with Transferrin Receptor as a PM marker, as aforementioned, the difference in maturation levels may well correlate, according to previous reports1, with the random-like migration we have previously reported for Cav1KO MEFs2, but this will be the scope of future studies.
2. There are no scale bars on any of the micrographs
We apologize for having omitted this information. We have now included scale bars when appropriate in figure micrographs, with the corresponding indications in figure legends.
3. I think that reference 46, not 26, is the one for the ELISA-based endocytosis/recycling protocol. Also, I don't think that reference 47 is correct for supporting that beta3 integrin follows a Rab4-dependent "short" loop pathway.
We thank the reviewer for these comments. We have followed the capture ELISA for determination of biotinylated integrins as indicated in reference 26, but it is true that a similar protocol was previously described by Prof. Jim Norman in reference 46, we have then included both references. We have eliminated reference 47 when referring to beta3 short-loop recycling, leaving only 45 and 46.
4. The authors have now included, as requested, data showing that re-expression of Cav1 slows-down recycling in Cav1KO cells. But why have the authors shown only active beta1? Surely, they should be showing the recycling data for both total and active beta1 – Indeed, they must have these data as, presumably, they would have split their lysate into ELISA plates coated with antibodies recognising total b1 as well as active beta1. Also, the levels of recycling demonstrated for Cav1KO cells in supplementary Figure 2I differs quite markedly from that shown in Figure 4I.
We thank the reviewer for these comments. We have indeed split the lysate but we only analyzed active β 1 integrin as it was the most relevant pool for explaining the increased reinforcement observed in our magnetic tweezers studies. However, we have now performed the corresponding ELISA for total β 1 integrin. To our surprise no significant differences between Cav1KO and Cav1KO+Cav1 MEFs are observed. This might indicate, as it usually happens with rescue experiments, that Cav1 re-expression only restores the wild type situation partially, rescuing the recycling of active β 1 integrin pool, but being unable to restore the inactive one. We have now added these results in Suppl. Figure 2J (now Figure 4—figure supplement 1J). Finally, although we have always performed the recycling assay under the same experimental conditions, there is a certain level of variability among experimental replicates: different cell passages, variations in total surface biotinylation, etc., which might explain the differences in recycling levels between Suppl. Figure 2I (now Figure 4—figure supplement 2J) and Figure 4I indicated by the reviewer.
References
1. Guo, M. et al. Altered processing of integrin receptors during keratinocyte activation. Exp Cell Res 195, 315-322, doi:10.1016/0014-4827(91)90379-9 (1991).
2. Grande-Garcia, A. et al. Caveolin-1 regulates cell polarization and directional migration through Src kinase and Rho GTPases. J Cell Biol 177, 683-694, doi:10.1083/jcb.200701006 (2007).
https://doi.org/10.7554/eLife.82348.sa2Article and author information
Author details
Funding
European Union Horizon 2020 Research and Innovation Programme (Marie Sklodowska-Curie grant 641639)
- Miguel A del Pozo
Asociación Española Contra el Cáncer Foundation (PROYE20089DELP)
- Miguel A del Pozo
Spanish Ministry of Economy, Industry and Competitivenes (SAF2014-51876-R)
- Miguel A del Pozo
Spanish Ministry of Economy, Industry and Competitiveness (SAF2017-83130-R)
- Miguel A del Pozo
Spanish Ministry of Economy, Industry and Competitiveness (CSD2009-0016)
- Miguel A del Pozo
Spanish Ministry of Science and Innovation (PID2020-118658RB-I00)
- Miguel A del Pozo
Spanish Ministry of Science and Innovation (PDC2021-121572-100)
- Miguel A del Pozo
Comunidad Autónoma de Madrid (S2018/NMT¬4443)
- Miguel A del Pozo
Fundació la Marató de TV3 (201936-30-31)
- Pere Roca-Cusachs
Ministerio de Ciencia e Innovación (CEX2020-001041-S)
- Miguel A del Pozo
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We thank Dr. Miguel Sánchez for critical reading of the manuscript and Simon Bartlett for scientific editing. We also thank Verónica Labrador Cantarero and Antonio M Santos Beneit from Microscopy Unit (CNIC) for macro development and video editing and Dr. Martin Humphries (The University of Manchester), Dr. Michael Lisanti (Institute of Cancer Sciences, Manchester), Dr. M Ginsberg (UC San Diego, USA), and Dr. Cristina Clemente Toribio for kindly providing reagents and cells.
This project received funding from the European Union Horizon 2020 Research and Innovation Programme through Marie Sklodowska-Curie grant 641639; grants from the Spanish Ministry of Science and Innovation (MCIN/AEI/10.13039/501100011033): SAF2014-51876-R, SAF2017-83130-R cofunded by “ERDF A way of making Europe”, PID2020-118658RB-I00, PDC2021-121572-100 cofunded by “European Union NextGenerationEU/PRTR”, CSD2009-0016, and BFU2016-81912-REDC; and the AECC (Asociación Española Contra el Cáncer) foundation (PROYE20089DELP) all to MAdP. MAdP is member of the Tec4Bio consortium (ref. S2018/NMT¬4443; CAM/FEDER, Spain), co-recipient with PR-C of grants from Fundació La Marató de TV3 (674 /C/2013 and 201936-30-31), and coordinator of a Health Research consortium grant from Fundación Obra Social La Caixa (AtheroConvergence, HR20-00075). The CNIC Unit of Microscopy and Dynamic Imaging is supported by FEDER "Una manera de hacer Europa" (ReDIB ICTS infrastructure TRIMA@CNIC, MCIN). The CNIC is supported by the Instituto de Salud Carlos III (ISCIII), the MCIN and the Pro CNIC Foundation, and is a Severo Ochoa Center of Excellence (grant CEX2020-001041-S funded by MICIN/AEI/10.13039/501100011033).
Senior and Reviewing Editor
- Suzanne R Pfeffer, Stanford University, United States
Publication history
- Preprint posted: April 29, 2022 (view preprint)
- Received: August 2, 2022
- Accepted: October 19, 2022
- Accepted Manuscript published: October 20, 2022 (version 1)
- Accepted Manuscript updated: October 21, 2022 (version 2)
- Version of Record published: December 13, 2022 (version 3)
Copyright
© 2022, Lolo et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 1,745
- Page views
-
- 297
- Downloads
-
- 0
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Cell Biology
- Neuroscience
Mitochondria influence cellular function through both cell-autonomous and non-cell autonomous mechanisms, such as production of paracrine and endocrine factors. Here, we demonstrate that mitochondrial regulation of the secretome is more extensive than previously appreciated, as both genetic and pharmacological disruption of the electron transport chain caused upregulation of the Alzheimer’s disease risk factor apolipoprotein E (APOE) and other secretome components. Indirect disruption of the electron transport chain by gene editing of SLC25A mitochondrial membrane transporters as well as direct genetic and pharmacological disruption of either complexes I, III, or the copper-containing complex IV of the electron transport chain elicited upregulation of APOE transcript, protein, and secretion, up to 49-fold. These APOE phenotypes were robustly expressed in diverse cell types and iPSC-derived human astrocytes as part of an inflammatory gene expression program. Moreover, age- and genotype-dependent decline in brain levels of respiratory complex I preceded an increase in APOE in the 5xFAD mouse model. We propose that mitochondria act as novel upstream regulators of APOE-dependent cellular processes in health and disease.
-
- Cell Biology
Uso1/p115 and RAB1 tether ER-derived vesicles to the Golgi. Uso1/p115 contains a globular-head-domain (GHD), a coiled-coil (CC) mediating dimerization/tethering and a C-terminal region (CTR) interacting with golgins. Uso1/p115 is recruited to vesicles by RAB1. Genetic studies placed Uso1 paradoxically acting upstream of, or in conjunction with RAB1 (Sapperstein et al., 1996). We selected two missense mutations in uso1 resulting in E6K and G540S in the GHD that rescued lethality of rab1-deficient Aspergillus nidulans. The mutations are phenotypically additive, their combination suppressing the complete absence of RAB1, which emphasizes the key physiological role of the GHD. In living hyphae Uso1 recurs on puncta (60 sec half-life) colocalizing partially with the Golgi markers RAB1, Sed5 and GeaA/Gea1/Gea2, and totally with the retrograde cargo receptor Rer1, consistent with Uso1 dwelling in a very early Golgi compartment from which ER residents reaching the Golgi recycled back to the ER. Localization of Uso1, but not of Uso1E6K/G540S, to puncta is abolished by compromising RAB1 function, indicating that E6K/G540S creates interactions bypassing RAB1. That Uso1 delocalization correlates with a decrease in the number of Gea1 cisternae supports that Uso1-and-Rer1-containing puncta are where the protein exerts its physiological role. In S-tag-coprecipitation experiments Uso1 is an associate of the Sed5/Bos1/Bet1/Sec22 SNARE complex zippering vesicles with the Golgi, with Uso1E6K/G540S showing stronger association. Using purified proteins, we show that Bos1 and Bet1 bind the Uso1 GHD directly. However, Bet1 is a strong E6K/G540S-independent binder, whereas Bos1 is weaker but becomes as strong as Bet1 when the GHD carries E6K/G540S. G540S alone markedly increases GHD binding to Bos1, whereas E6K causes a weaker effect, correlating with their phenotypic contributions. AlphaFold2 predicts that G540S increases binding of the GHD to the Bos1 Habc domain. In contrast, E6K lies in an N-terminal, potentially alpha-helical, region that sensitive genetic tests indicate as required for full Uso1 function. Remarkably, this region is at the end of the GHD basket opposite to the end predicted to interact with Bos1. We show that unlike dimeric full-length and CTR∆ Uso1 proteins, the GHD lacking the CC/CTR dimerization domain, whether originating from bacteria or Aspergillus extracts and irrespective of whether it carries or not E6K/G540S, would appear to be monomeric. With the finding that overexpression of E6K/G540S and wild-type GHD complement uso1∆, our data indicate that the GHD monomer is capable of providing, at least partially, the essential Uso1 functions, and that long-range tethering activity is dispensable. Rather, these findings strongly suggest that the essential role of Uso1 involves the regulation of SNAREs.