Biomechanical Control of Lysosomal Secretion Via the VAMP7 Hub: A Tug-of-War between VARP and LRRK1.

The rigidity of the cell environment can vary tremendously between tissues and in pathological conditions. How this property may affect intracellular membrane dynamics is still largely unknown. Here, using atomic force microscopy, we show that cells deficient in the secretory lysosome v-SNARE VAMP7 are impaired in adaptation to substrate rigidity. Conversely, VAMP7-mediated secretion is stimulated by more rigid substrate and this regulation depends on the Longin domain of VAMP7. We further find that the Longin domain binds the kinase and retrograde trafficking adaptor LRRK1 and that LRRK1 negatively regulates VAMP7-mediated exocytosis. Conversely, VARP, a VAMP7- and kinesin 1-interacting protein, further controls the availability for secretion of peripheral VAMP7 vesicles and response of cells to mechanical constraints. LRRK1 and VARP interact with VAMP7 in a competitive manner. We propose a mechanism whereby biomechanical constraints regulate VAMP7-dependent lysosomal secretion via LRRK1 and VARP tug-of-war control of the peripheral pool of secretory lysosomes.

Introduction

From the softest tissue like the brain (<1 kPa) to the hardest ones like bones (∼100 kPa), the stiffness of cell environment can greatly vary in the body of mammals. Matrix elasticity was shown to affect the differentiation of stem cells (Engler et al., 2006), cell spreading and morphology, and the capacity to migrate (Tzvetkova-Chevolleau et al., 2008). Cells adhere to the substrate and sense the rigidity of the substrate via integrin-mediated focal adhesions (Chen et al., 2015, Sun et al., 2016). Cell adhesion is able to regulate the cytoskeleton and membrane tension (Sens and Plastino, 2015). Previous work showed that exocytosis and endocytosis are regulated by cell spreading and osmotic pressure (Gauthier et al., 2011) and membrane tension regulates secretory vesicle docking through a mechanism involving Munc18-a (Papadopulos et al., 2015). How substrate rigidity sensing may regulate exocytosis, which in turn regulates membrane tension, is still largely unknown. Secretory mechanisms involve SNAREs, the master actors of intracellular membrane fusion (Südhof and Rothman, 2009). Exocytosis involves the formation of a SNARE complex comprising a vesicular SNARE (v-SNARE) on the vesicle side. Two main v-SNAREs were shown to mediate distinct exocytic mechanisms: clostridial neurotoxin-sensitive VAMP2 (and the closely related VAMP1 and VAMP3) mediates synaptic vesicle and early endosomal exocytosis, whereas clostridial neurotoxin-insensitive VAMP7 mediates Golgi-derived, late endosomal and lysosomal secretory pathways (Proux-Gillardeaux et al., 2005a), which are best defined by the presence of the tetraspanin CD63 (Chiaruttini et al., 2016, Coco et al., 1999). In the recent years, lysosomal exocytosis has appeared as a very general mechanism that can be found in virtually any cell type (Ghosh et al., 2016, Jaiswal et al., 2002, Li et al., 2008, Verderio et al., 2012). Interestingly enough, VAMP2 and VAMP3 were shown to mediate integrin recycling (Hasan and Hu, 2010, Proux-Gillardeaux et al., 2005b, Skalski and Coppolino, 2005, Tayeb et al., 2005), and VAMP7, to play an essential role in cell migration and invasion (Proux-Gillardeaux et al., 2007, Steffen et al., 2008, Williams and Coppolino, 2011). VAMP7 also contributes to the regulation of membrane composition of sphingolipids and glycosylphosphatidylinositol (GPI)-anchored protein (Molino et al., 2015), which in turn may modulate integrin dynamics and adhesion (Eich et al., 2016, van Zanten et al., 2009). VAMP7 mediates the release of ATP (Fader et al., 2012, Verderio et al., 2012) and interleukin-12 (Chiaruttini et al., 2016) and is required for the transport of cold-sensing channel TRPM8- and lysosomal-associated membrane protein 1-containing vesicles in sensory neurons (Ghosh et al., 2016). In neurons, VAMP2 exocytosis was shown to be regulated by an integrin-independent mechanism and VAMP7 exocytosis by an integrin-, FAK-, and Src-dependent mechanism (Gupton and Gertler, 2010).

Here we used atomic force microscopy, micropatterned surfaces, pHluorin live imaging of lysosomal exocytosis (Balaji and Ryan, 2007), and substrate of controlled rigidity and composition to explore the role of lysosomal exocytosis in cell response to biomechanical constraints. Our results suggest that VAMP7-dependent lysosomal secretion responds to rigidity via control by its partners LRRK1 and VARP of the peripheral pool of secretory lysosomes.

Results

VAMP7 Is Required for Fibroblast Mechano-Adaptation

Cells have the ability to adapt their internal stiffness to substrate rigidity (Solon et al., 2007). We cultured COS7 cell on soft gels with a rigidity of 1.5 and 28 kPa, mimicking brain and cartilage rigidity, respectively, and measured the cell elasticity of control and VAMP7 knockout (KO) cells by atomic force microscopy (Figure 1A). Wild-type (WT) COS7 cells had a slightly higher elasticity when plated on 28-kPa than on 1.5-kPa substrate (Figure 1B). Surprisingly, VAMP7 CRISPR/Cas9 KO cells (Figure S1) had a higher elasticity in the 1.5-kPa condition than control cells and reacted differently to the change in substrate rigidity, with a lower elasticity on a more rigid substrate. The observed difference between 1.2 and 0.9 kPa, a one-third-fold change, appeared in the range of the one observed for vimentin KO (Messica et al., 2017), suggesting a strong impact of VAMP7 KO on cell elasticity. We did not observe any visible change in cytoskeletal organization (Figure S2A) or cell spreading area (Figure S2B) between WT and VAMP7 KO COS7 cells in our conditions, data possibly explained by the fact that COS7 cells are devoid of myosin IIa and show some contractility defects (Even-Ram et al., 2007).

Figure 1

VAMP7 Is Required for Fibroblast Mechano-Adaptation

(A) Elasticity heatmaps of cells plated on laminin-coated PDMS gels of 1.5 or 28 kPa. Measurements were systematically made in a 20-μm-wide rectangular area whose typical placement is indicated by the black box. Scale bar, 10 μm.

(B) Quantification of cell elasticity (elastic modulus E). Graph shows scatterplot with mean ± 95% confidence interval (CI). Each point represents the median E value of a cell from four independent experiments. n = 51, 62, 57, and 53 cells, respectively. *p < 0.05, **p < 0.01, and ****p < 0.0001; ANOVA with Tukey's post hoc or Welsh's t test was used as indicated.

To understand the potential importance of VAMP7 in cell mechanics and response to mechanical constraints, we first localized VAMP7 in COS7 cells grown on micropatterned glass coverslips coated with laminin. We used cells grown on O pattern, a pattern with homogeneous mechanical constraint as control, and Y pattern, a condition wherein cells are under peripheral traction forces (Albert and Schwarz, 2014). We found that VAMP7 was particularly enriched in actin-rich cell tips (Figures 2A and 2B), where contractile forces are generated, in cells grown on a Y pattern.

Figure 2

VAMP7 Enriches in Tips on Y-Shaped Micropattern

(A) Projection of confocal microscopy optical sections of COS7 cells plated on laminin-coated micropatterns. n: O = 27, Y = 27 cells. Scale bar, 10 μm.

(B) Quantification of RFP-VAMP7 intensity from cell center to cell periphery. Graph shows mean ± 95% confidence interval (CI) (dashed lines).

(C) Projection of confocal microscopy optical sections of control, VAMP7 KO, and VAMP7 KO re-expressing GFP-VAMP7 cells plated on Y micropatterns. n = 43, 59, and 63 cells respectively. Scale bar, 10 μm.

(D) Quantification of CD63 immunofluorescence in cell center area (<10 μm from the geometry center), neck area (between 10 and 20 μm), and tip area (>20 μm). Graph shows scatterplot with mean ± 95% CI. Each point represents the value obtained from cells from two independent experiments. **p < 0.01 and ***p < 0.001; ANOVA with Tukey's post hoc. ns, not significant.

VAMP7 localizes into different post-Golgi compartments, mostly endosomal compartments, and CD63, a tetraspanin of secretory lysosomes, is the best marker colocalizing with VAMP7 described so far (Chiaruttini et al., 2016, Coco et al., 1999). Thus we analyzed the colocalization of endogenous VAMP7 and CD63 on laminin-coated glass coverslips (Figure S3A) and fluorescent protein-tagged VAMP7 and CD63 on rigidity-defined polydimethylsiloxane (PDMS) surface (Figures S3B and S3C). As expected, VAMP7 showed high level of colocalization with CD63 in all conditions. We further characterized the subcellular localization of exogenously expressed VAMP7 with APP, a protein known to traffic through late endosomes/lysosomes (Tam et al., 2014); VAMP7's partner, the adaptor AP3 (Kent et al., 2012, Martinez-Arca et al., 2003); and the endosomal markers Rab5 and Rab7 (Figure S4). Tagged VAMP7 showed similar degree of colocalization with the tested endosomal markers on both soft and rigid substrates. Thus, the subcellular targeting of VAMP7 to late endosomes and lysosomes was not significantly influenced by substrate stiffness. The level of expression of VAMP7 further appeared to have an effect on the peripheral positioning of CD63 in constrained cells. Indeed, whereas KO of VAMP7 did not significantly affect CD63 subcellular localization compared with control cells on Y micropatterns, re-expression of the protein in KO cells modified the distribution of CD63 with an enrichment in cell necks (Figures 2C and 2D) without changing its total expression level (Figure S5A) or cytoskeleton organization (Figure S5B). Altogether, these experiments show that VAMP7 participates in cell response to biomechanical constraints likely via a role in CD63+ secretory lysosome positioning.

VAMP7-Mediated Exocytosis Is Regulated by Mechanosensing

Previous studies suggested that the peripheral positioning of lysosomes is important for their secretion (Encarnação et al., 2016, Guardia et al., 2016, Hämälistö and Jäättelä, 2016, Pu et al., 2016). As lysosome position is a pre-requisite for VAMP7-mediated lysosome fusion at the plasma membrane, we wondered if mechanical cues such as substrate rigidity could directly regulate lysosomal exocytosis. We measured the individual VAMP7 and VAMP2 exocytic events using pHluorin-tagged molecules (Figure 3A, Videos S1 and S2) expressed in COS7 cells grown on surfaces of controlled stiffness generated using PDMS gels of 1.5 and 28 kPa. Here the pHluorin was attached to the luminal terminal of VAMPs as mentioned previously (Burgo et al., 2012). The fluorescence signal from pHluorin is quenched in acidic medium, as inside secretory lysosomes. After the exocytic membrane structure has undergone fusion with the plasma membrane, pHluorin appears in neutral pH extracellular medium and the fluorescence signal suddenly appears as a flash of light. Thus, the pHluorin signals are the direct consequence of SNARE complex formation and membrane fusion corresponding to secretory events. We found that the frequency of exocytosis of VAMP7 had an up to ∼1.5-fold increase on 28 kPa in the presence of laminin compared with on 1.5 kPa in the absence of laminin, whereas VAMP2 exocytosis was insensitive to both substrate stiffness and chemistry (Figures 3B and 3C). This finding was confirmed using polyacrylamide gels coated with polylysine or laminin with increased rigidity, as the frequency of VAMP7 exocytosis doubled from 1.5 to 28 kPa, with already a noticeable intermediate increase at 11 kPa (Figures 3D and 3E). Previous studies have reported an ∼2-fold change in exocytosis under the effect of laminin versus polylysine on glass coverslip, or under pharmaceutical disruption of cytoskeleton and integrin signaling (Gupton and Gertler, 2010). Therefore, the doubling in exocytic frequency that we observed appears to be a strong and significant effect. COS7 cells endogenously express laminin receptors integrin α3, α6, and β1 (Niessen et al., 1997). Integrin-dependent adhesion transduces intracellular signals, which affects actomyosin contraction (Parsons et al., 2010). Here we found that blebbistatin affected VAMP7-dependent exocytosis in a dose-dependent manner (Figure S6), further suggesting biomechanical control of VAMP7 exocytosis, likely downstream of integrin-dependent adhesion, as previously found in neurons (Gupton and Gertler, 2010).

Figure 3

VAMP7-Mediated Exocytosis Is Regulated by Mechanosensing

(A) Principle of exocytic membrane fusion imaging with pHluorin. Upper panel: pHluorin was attached to the luminal terminal of VAMP2 and VAMP7. The fluorescence signal from pHluorin is quenched in acidic medium, as inside secretory vesicles. After the vesicles undergo fusion with the plasma membrane, pHluorin is exposed to the neutral pH of extracellular medium and the fluorescence signal suddenly appears as a flash of light. Lower panel: representative snapshots of two typical individual exocytic events of short (1) and long (2) life time as shown by persistence of the fluorescence signal. Scale bar: 1 μm, 0.5 s per image.

(B and C) Quantification of exocytic events in COS7 cells expressing pHluorin-tagged VAMP2 or VAMP7. Cells were plated on polylysine- or laminin-coated PDMS gel of 1.5 or 28 kPa for 18–24 hr. Graph shows scatterplot with mean ± 95% confidence interval (CI). Each point represents the exocytic rate of cells from two or more independent experiments. (B) n = 51, 48, 54, and 26; (C) n = 54, 62, 49, and 55 cells, respectively. **p < 0.01, Welsh's t test. ns, not significant.

(D and E) Quantification of exocytic events in COS7 cells expressing pHluorin-tagged VAMP2 and VAMP7. Cells were plated on polylysine- or laminin-coated polyacrylamide gel of 1.5, 11, or 28 kPa for 18–24 hr. Graph shows scatterplot with mean ± 95% CI. Each point represents the exocytic rate of cells from two or more independent experiments. VAMP2: n = 21, 25, 16, 25, 28, and 23; VAMP7, n = 40, 41, 47, 19, 19, and 18 cells, respectively. **p < 0.01, Welsh's t test.

Longin-Dependent Regulation of VAMP7 Exocytosis by Mechanosensing

Then, we asked whether the regulation of VAMP7 exocytosis could be due to the presence of the Longin domain (LD), the main regulator of VAMP7 (Daste et al., 2015). Indeed, we found that a mutant of VAMP7 lacking the LD (Δ[1-125]-VAMP7) showed increased exocytosis as previously described (Burgo et al., 2013), but its exocytic frequency was not affected by the substrate stiffness and chemistry and was already maximal on soft substrate (Figure 4A, to be compared with WT VAMP7, Figure 3C). We further analyzed the half-life of pHluorin signals, which represents the kinetics of fusion pore opening and spreading followed by endocytosis and re-acidification (Figure S7A). VAMP2, VAMP7, and ΔLD-VAMP7 showed no significant difference in signal persistence depending on stiffness and chemistry. Altogether, these data suggest that VAMP7 exocytosis is modulated by substrate stiffness and composition in an LD-dependent manner. This mode of regulation did not appear to affect the mode of fusion (i.e., transient fusion versus full fusion [Yudowski et al., 2006]) and thus most likely affects the pool size and/or release probability of secretory VAMP7+ vesicles. VAMP7 and VAMP2 interact with the same plasma membrane target-SNAREs (t-SNAREs) to mediate exocytosis in non-neuronal cells, i.e., syntaxin 4 and SNAP23 (Martinez-Arca et al., 2001, Rao et al., 2004, Sander et al., 2008). Our results showing that VAMP2 exocytosis was insensitive to rigidity thus further suggested that t-SNAREs are likely not regulated by substrate rigidity. The previous results suggested that substrate stiffness could have a specific role in VAMP7 regulation. To more directly test the hypothesis of the role of membrane tension in VAMP7 exocytosis, we used hyper-osmotic changes and pHluorin imaging as mentioned previously. We found that high hyper-osmotic pressure (2x osmolarity) could instantaneously and reversibly reduce the exocytosis frequency of VAMP7 independently of its LD, suggesting different mechanisms of action of membrane tension modulated by osmotic changes and substrate stiffness (Figures 4B and 4C, Video S3). We also found that the half-life of pHluorin signals was moderately decreased following hyperosmotic shocks and then spontaneously restored to a normal level. ΔLD-VAMP7 colocalized with full-length VAMP7 in the cell periphery but was absent in some perinuclear endosomes (Figure 4D), likely corresponding to late endosomes and lysosomes where VAMP7 is targeted in an LD/AP3-dependent manner (Kent et al., 2012, Martinez-Arca et al., 2003). Therefore, these experiments suggest that substrate rigidity specifically affects lysosomal secretion (VAMP7) and not early endosomal recycling (VAMP2, ΔLD-VAMP7).

Figure 4

Mechanoregulation of VAMP7 Exocytosis by Substrate Rigidity Specifically Requires Longin Domain

(A) Quantification of exocytic events in COS7 cells expressing pHluorin-tagged ΔLD(Δ[1–125]) VAMP7. Cells were plated on polylysine- or laminin-coated PDMS gel of 1.5 or 28 kPa for 18–24 hr. Graph shows scatterplot with mean ± 95% confidence interval (CI). Each point represents the exocytic rate of cells from two or more independent experiments. n = 72, 72, 80, and 67 cells, respectively. Welsh's t test.

(B and C) Quantification of exocytic rate and pHluorin signals' half-life in COS7 cells expressing pHluorin-tagged VAMP7 or ΔLD-VAMP7. Cells were plated on laminin-coated 28-kPa PDMS gels for 18–24hr. Hyper-osmotic shocks were performed by perfusing the 2x osmolality buffer and then washed out by 1x buffer. At each time point, the exocytic rate in the following minute was calculated. Graph shows mean ± 95% CI (dashed lines). B, n = 13; C, n = 11, pooled from two or more independent experiments.

(D) Representative COS7 cell co-expressing RFP-tagged ΔLD(Δ[1–120]) VAMP7 and GFP-tagged full-length VAMP7. Filled arrowheads show the colocalization. Empty arrowheads indicate structures containing only GFP-VAMP7. Scale bar, 10 μm.

Altogether, pHluorin imaging experiments led us to propose that membrane tension (such as modulated by osmotic shocks) is a master regulator of exocytosis independent of vesicle origin (both endosomal and lysosomal). On the contrary, the regulation of VAMP7 by substrate stiffness appeared to not depend on a pure biomechanical effect via plasma membrane tension but rather required proper sensing of the environment rigidity, such as in the presence of laminin.

Role of VAMP7 Hub in Mechanosensing

VAMP7 interactome includes two proteins connected to molecular motors. LRRK1 interacts with VAMP7 through its ankyrin repeat and leucine-rich repeat domain and also interacts with dynein (Kedashiro et al., 2015a, Toyofuku et al., 2015). The Rab21 guanine nucleotide exchange factor VARP interacts with VAMP7 through a small domain in its ankyrin repeat domains and also interacts with kinesin 1 (Burgo et al., 2009, Burgo et al., 2012, Schäfer et al., 2012). Interestingly, sequence analysis showed that the ankyrin repeat of VARP, which interacts with VAMP7, includes a 10-amino acid (aa) sequence fully conserved in LRRK1 (Figure 5A). This led us to wonder whether or not LRRK1 and VARP may participate in the regulation of VAMP7 by substrate stiffness via its LD, in a potentially competitive manner. First, to determine whether the interaction between VAMP7 and LRRK1 was through the LD, we carried out in vitro binding assay with GST-tagged cytosolic domain (Cyto) and LD of VAMP7 protein. We found that LRRK1 had an ∼10-fold stronger interaction with LD than with the cytosolic portion of the protein (Figures S8A and S8B). Next, we immunoprecipitated GFP-tagged LRRK1 or GFP-tagged VARP and assayed for coprecipitation of red fluorescent protein (RFP)-tagged full length and various deleted forms of VAMP7 (Figure 5B) from transfected COS7 cells. We found that LRRK1 interacted with full length, LD, and SNARE domain, whereas the interaction of VARP was preferentially with full length and SNARE domain, with weak binding to the LD alone (Figures 5C and 5D, Tables S1 and S2). The spacer between LD and SNARE domain alone did not bind to either LRRK1 or VARP, but appeared to increase the binding of SNARE domain to both LRRK1 and VARP. This likely indicates that the spacer could help the folding of the SNARE domain required for interaction with both LRRK1 and VARP. Nevertheless, the spacer could be replaced by GGGGS motifs of similar length rather than the original spacer (20 aa) without affecting neither LRRK1 nor VARP binding, indicating that its role is not sequence specific but only related to its length. We conclude that LRRK1 interacts with VAMP7 via the LD and that its binding to VAMP7 is more sensitive than that to VARP to the presence of the LD. The loss of mechano-sensing of exocytosis when the LD is removed thus likely results from the loss of a competition between LRRK1 and VARP. Furthermore, co-immunoprecipitation experiment showed that expression of the interaction domain (ID) of VARP, which mediates binding to VAMP7, competes with the binding of VAMP7 to VARP as expected and also the binding to LRRK1 (Figures 5E and 5F) to a similar extent (Tables S3 and S4). These data suggest that LRRK1 and VARP bind to VAMP7 via similar regions in ankyrin domains and likely compete for VAMP7 binding and/or generate mutually exclusive conformations of VAMP7. In good agreement with our hypothesis, triple labeling of exogenously expressed VAMP7, LRRK1, and VARP showed striking colocalization spots of VAMP7 and VARP in cell tips and colocalization spots of VAMP7 and LRRK1, without VARP, in the cell center (Figure 5G). GFP-LRRK1 and GFP-VARP but not soluble GFP showed significant colocalization with RFP-VAMP7 on Y patterns with enrichment of LRRK1 in cell center and VARP on cell tips (Figure S9). Altogether these data suggest that LRRK1 and VARP could compete for binding to VAMP7 and may have antagonistic functions in the intracellular distribution of VAMP7+ vesicles.

Figure 5

LRRK1 and VARP Compete for VAMP7 Binding

(A) Sequence alignment showing that LRRK1 shares a conserved ankyrin repeat domain with VARP in its interaction domain with VAMP7.

(B) Domain organization of rat VAMP7. Sp, spacer; TM, transmembrane. The constructs used for co-immunoprecipitation assay are shown below.

(C and D) Assays of binding of LRRK1 and VARP to VAMP7. Lysates from COS7 cells co-expressing GFP-LRRK1 or GFP-VARP with indicated RFP-tagged construction of VAMP7 were immunoprecipitated (IP) with GFP-binding protein (GBP) fixed on sepharose beads. Precipitated proteins were subjected to SDS-PAGE, and the blots were stained with antibodies against indicated target proteins. EGFP and monomeric RFP (mRFP) protein were used as control for nonspecific binding. The experiment has been independently repeated three times with similar results.

(E and F) Competition of VAMP7 interaction with LRRK1 and VARP by the VAMP7 interaction domain (ID) of VARP. Lysates from COS7 cells co-expressing indicated constructs were processed as described in (C) and (D). The expression of myc-ID-FLAG was detected by TG40, a rabbit polyclonal antibody raised against ID (Burgo et al., 2009).* previously revealed RFP signal. FLAG, epitope tag.

(G) Representative confocal microscopy sections of COS7 cell co-expressing RFP-VAMP7, FLAG-VARP, and GFP-LRRK1 grown on laminin-coated glass. Filled arrowheads indicate triple colocalization. Empty arrowheads indicate structures where either FLAG-VARP or GFP-LRRK1 is missing or dominant. Scale bar, 10 μm.

To further decipher the role of LRRK1, we silenced its expression by short hairpin RNA (shRNA) and assayed for VAMP7 exocytosis on soft and rigid substrate. We found that the exocytosis frequency of VAMP7 on soft substrate was increased to the same level as on rigid substrate in cells in which the expression of LRRK1 was knocked down (Figure 6A). Silencing of LRRK1 did not affect the persistence of VAMP7 pHluorin signal in the different stiffness conditions tested (Figure S7B), in good agreement with previous observations discussed earlier (Figure S7A). According to previous work on LRRK1, VAMP7-LRRK1 interaction should recruit CLIP-170 and dynein, allowing for retrograde transport on microtubules (Kedashiro et al., 2015a). To further understand the potential role of LRRK1 in VAMP7 trafficking, we carried out live imaging of cells expressing GFP-LRRK1 and RFP-VAMP7 and found that VAMP7 and LRRK1 accumulated together in the cell center upon epidermal growth factor (EGF) stimulation (Figures 6B and 6C), a condition that promotes perinuclear localization of LRRK1-containing endosomes (Hanafusa et al., 2011, Ishikawa et al., 2012). Analysis of confocal images taken from cells expressing GFP-tagged WT LRRK1, Y944F, or K1243M mutants (constitutively active and inactive kinase forms of LRRK1, respectively) and RFP-tagged VAMP7 showed that VAMP7 accumulated more in the perinuclear region in LRRK1 Y944F-expressing cells, and more toward the cell periphery in LRRK1 K1243M-expressing cells (Figures 7A and 7B), suggesting that LRRK1 kinase activity enhanced the retrograde transport of VAMP7 vesicles into the perinuclear region. We further immunoprecipitated GFP-tagged LRRK1 WT, Y944F, and K1243M mutants and assayed for coprecipitation of RFP-tagged VAMP7. We found that the kinase activity of LRRK1 is dispensable for its interaction with VAMP7 (Figure 7C and Table S5), in good agreement with the identification of the ID in the amino-terminal domain of LRRK1 as previously shown (Toyofuku et al., 2015). LRRK1 was previously found to play a role in autophagy (Toyofuku et al., 2015), but we did not find significant autophagy induction as seen by LC3-II imaging in cells on soft versus rigid substrates and western blotting (Figure S10). We conclude that LRRK1 mediates retrograde transport of VAMP7 in a kinase-dependent activity and that LRRK1 is required for the control of VAMP7 exocytosis in response to substrate rigidity.

Figure 6

LRRK1 Regulates VAMP7 Trafficking and Mechanosensing

(A) Quantification of exocytic events in COS7 cells co-expressing VAMP7-pHluorin with control shRNA or LRRK1-shRNA, grown on laminin-coated PDMS gels for 18–24 hr. Graph shows scatterplot with mean ± 95% confidence interval (CI). Each point represents the exocytic rate of cells from two independent experiments. n = 51, 62, 57, and 53 cells, respectively. **p < 0.01, Welsh's t test.

(B) VAMP7 and LRRK1 accumulated together in perinuclear endosomes upon EGF stimulation. COS7 cells co-expressing RFP-VAMP7 and GFP-LRRK1 were imaged by wide field microscopy 24 hr after plating on laminin-coated 1.5- or 28-kPa PDMS gel. EGF was injected into the imaging chamber between minutes −1 and 0 to a final concentration of 100 ng/mL. Scale bar: 10 μm.

(C) Quantification of the accumulation of fluorescence signal at the cell center. The cell center was defined by calculating the local center of mass using an ImageJ script. The total fluorescence intensity was measured inside a 16-μm-diameter oval around the cell center (dashed circle in Figure 6B) and compared with the total fluorescence intensity outside the circle. 1.5 kPa - EGF n = 54, 1.5 kPa + EGF n = 39, 28 kPa + EGF n = 51. Data were pooled from two independent experiments. Graph shows mean with 95% CI (dotted lines). **p < 0.01, ****p < 0.0001, ANOVA with Tukey's post hoc.

Figure 7

LRRK1 Kinase Activity Regulates VAMP7 Retrograde Transport

(A) Representative projection of confocal microscopy optical sections of COS7 cells co-expressing RFP-VAMP7 with GFP-tagged WT LRRK1, Y944F (kinase constitutively active), and K1243M (kinase dead) mutants grown on laminin-coated glass coverslips for 18–24 hr. Images show z-projection of confocal stack. Scale bar: 10 μm.

(B) Quantification of the accumulation of RFP-VAMP7 fluorescence signal at the cell center. The cell center was defined by calculating the local center of mass using an ImageJ script. The fluorescence intensity was measured inside a 10-μm-diameter oval around the cell center and reported to the total fluorescence intensity. Each point represents the data measured from cells from two independent experiments. Graph shows mean with 95% confidence interval (CI). n = 33, 32, and 29 cells, respectively. *p < 0.05, **p < 0.01; ANOVA with Tukey's post hoc. ns, not significant.

(C) Assays of binding of VAMP7 to WT, Y944F, and K1243M LRRK1. Lysates from COS7 cells co-expressing RFP-VAMP7 with indicated GFP-tagged construction of LRRK1 were immunoprecipitated (IP) with GFP-binding protein (GBP) fixed on sepharose beads. Precipitated proteins were subjected to SDS-PAGE, and the blots were stained with antibodies against indicated target proteins. EGFP and mRFP protein were used as control for nonspecific binding. The experiment has been independently repeated three times with similar results.

Opposite Roles of LRRK1 and VARP in Mechanosensing

A prediction from our previous results showing that VAMP7 exocytosis is involved in mechanosensing (Figure 1), and that LRRK1 and VARP compete for binding to VAMP7 (Figure 5), would be that LRRK1 and VARP could generate a tug-of-war mechanism for the cell positioning of secretory lysosomes in the context of mechanosensing. To test this hypothesis, we again used the previous assay with cells grown on substrates of different rigidities. We found that soft substrate promoted more perinuclear accumulation of VAMP7 than rigid substrate (Figures 8A and 8B), similar to the effect of LRRK1 Y944F mutant (Figures 7A and 7B). We found that VAMP7 was localized more to the center in LRRK1-overexpressing cells. The opposite result was found in VARP-overexpressing cells, which showed decreased center-localized VAMP7. VARP-overexpressing cells further display striking concentration of VAMP7 at the tips of cell protrusions. The effects of WT LRRK1 and VARP overexpression were not sensitive to substrate rigidity. These later data suggest that the effect of overexpression of these proteins dominated over the regulation that occurs between soft and rigid environment when they are expressed at physiological levels. To further decipher the role of VARP and LRRK1, we then used the Crispr/Cas9 approach to knock out the expression of the proteins (Figure S1A), cultured the KO cells on substrate of 1.5 and 28 kPa, and assayed for perinuclear accumulation of RFP-VAMP7 (Figures 8C and 8D). We again reproduced the decreased perinuclear concentration of VAMP7 on more rigid substrate in control cells. The effect of rigidity was lost in LRRK1 KO cells. On the contrary, re-expression of LRRK1 in KO cells exacerbated central concentration of VAMP7. Conversely, VARP KO showed strong perinuclear accumulation of VAMP7 on rigid substrate, and this effect was rescued by the re-expression of VARP. In this later case, the effect of substrate rigidity was clearly visible after VARP re-expression in VARP KO cells. The KOs of VARP and LRRK1 did not lead to strong variations of microtubules (Figure S11) and MLC2 subcellular localization (Figure S5B).

Figure 8

LRRK1 and VARP Have Opposite Roles in Rigidity-Dependent VAMP7 Positioning

(A) Representative WT COS7 cells co-expressing RFP-tagged VAMP7 with GFP-tagged LRRK1 or VARP, grown on laminin-coated PDMS gels. Images show z-projection of confocal stack. Arrowheads show the colocalization in cell protrusions. Scale bar: 10 μm.

(C) Representative control, LRRK1 KO, and VARP KO COS7 cells grown on laminin-coated PDMS gels. Control and KO cells were transfected with RFP-VAMP7 and EGFP as indicated, and with GFP-LRRK1 and VARP in rescue conditions. Images show z-projection of confocal stack. Arrowheads show the colocalization in cell protrusions. Scale bar: 10 μm.

(E) Representative VARP KO cells expressing FLAG tag, FLAG-tagged wild-type VARP, and FLAG-tagged VARP with DADA mutation and grown on laminin-coated 28-kPa PDMS gels. Images show z-projection of confocal stack. Arrowheads show the colocalization in cell protrusions. Scale bar: 10 μm. Note the rescue by wild-type but not DADA mutant of VARP.

(B, D, and F) Quantification of RFP-VAMP7 fluorescence in the perinuclear region. Graph shows scatterplot with mean ± 95% confidence interval (CI). Each point represents the value obtained from a cell pooled from two independent experiments. (B) n = 55, 53, 52, 55, 56, and 58; (D) n = 57, 57, 39, 44, 45, 45, 50, 51, 51, and 44; (F) n = 53, 44, and 47 cells, respectively. *p < 0.05, **p < 0.01, and ****p < 0.0001, ANOVA with Tukey's post hoc or Welsh's t test was used as indicated.

Interestingly, expression of a point mutant of VARP (the so-called DADA mutant) unable to bind VAMP7 (Schäfer et al., 2012) was unable to rescue the phenotype of VARP KO cells on 28-kPa rigid substrate, whereas, as expected, the WT form of VARP fully rescued the subcellular localization of VAMP7 in the cell periphery on rigid substrate (Figures 8E and 8F). In the rescue conditions, exogenous VARP and LRRK1 partially colocalized with VAMP7 (Figure S9). Altogether, these experiments using KO, rescue, and overexpression approaches and culture on soft and rigid substrate suggest that LRRK1 and VARP provides a tug-of-war mechanism, which mediates the fine-tuning of VAMP7 subcellular localization regulated by substrate rigidity (Figure 9). In this regulatory mechanism, the precise expression level of LRRK1 and VARP appeared to be a critical parameter, further reinforcing the notion of a competitive mechanism strongly dependent on the concentration and activity of LRRK1 and VARP.

Figure 9

Hypothetical Working Model

Working hypothesis describing a tug-of-war mechanism with LRRK1 on the retrograde end and Varp on the anterograde end of VAMP7 transport. On the soft substrate, high activity of LRRK1 and low activity of VARP favors the retrograde transport of VAMP7 vesicles, and as a result, depletes the availability of peripheral VAMP7 pool. On the rigid substrate, weaker activity of LRRK1 and stronger activity of VARP increase the amount of VAMP7 vesicles in the periphery.

Discussion

In this study, we found that VAMP7-dependent lysosomal exocytosis was required for cells to sense substrate rigidity and that the latter redistributed VAMP7 to the cell periphery in an LD-, VARP-, and LRRK1-dependent manner. LRRK1 and VARP appeared to compete for VAMP7 binding via a conserved domain in one of their ankyrin repeats and operate via opposite control of the availability for secretion of peripheral VAMP7 vesicles in response to mechanical constraints, thus suggesting a tug-of-war mechanism.

VAMP7 KO COS7 cells showed remarkably increased cell elasticity on soft substrate and decreased elasticity on more rigid substrate compared with control cells. This likely suggests that the lack of VAMP7 may prevent cells from properly responding to mechanical constraints. Conversely, substrate rigidity increased exocytosis of VAMP7, but not VAMP2. This likely indicates the need for different types of membranes being transported to the cell surface depending on the biophysical properties of cell environment, particularly its rigidity. VAMP7 was shown to be important for phagophore formation and autophagosome secretion (Fader et al., 2012, Moreau et al., 2011) and rigidity was shown to increase autophagy (Ulbricht et al., 2013), but we did not find significant LC3-II induction in the different conditions tested, so we do not think that substrate stiffness significantly activated autophagy in our experimental conditions.

Cell tension is the combination of two factors: the cortical tension of actomyosin cytoskeleton and the in-plane tension of plasma membrane (Sens and Plastino, 2015). We found that hyper-osmotic shock inhibited the exocytosis frequency of VAMP7 following a remarkably quick adaptation of exocytosis frequency to strong changes in membrane tension. The effect of hyper-osmotic shock on persistence of the signal at plasma membrane would be best explained by decreased fusion pore flattening because fusion pore growth is promoted or even driven by the membrane tension (Bretou et al., 2014) and potential increased recovery of plasma membrane by endocytosis upon the osmotic shock. Our data thus fit well with the notion that exocytosis increases the surface area and therefore decreases membrane tension, and thus needs to be shut down to compensate for decreased membrane tension following hyper-osmotic shock (Gauthier et al., 2011, Keren, 2011, Sens and Plastino, 2015). Nevertheless, we found similar effects of hyper-osmotic shock on VAMP7 deleted of its LD, whereas this was not the case for increased substrate stiffness. This indicates that acute changes of cell tension, such as osmotic shocks, acting likely via a direct effect on membrane tension, and secretory vesicles in close proximity with the plasma membrane, proceed through mechanisms different from substrate stiffness. Altogether, our results suggest that of the two components of cell tension (membrane tension and cortical tension), the first may operate in the context of hyper-osmotic shock and not be specific of VAMP7 because it was Longin independent (Figures 4C and 4D), whereas the latter may be the main regulation that is activated by substrate rigidity in the presence of laminin (Figure 3).

It is intriguing that substrate rigidity and composition affected exocytosis of VAMP7, i.e., late endosomal and lysosomal CD63+ exocytosis but not VAMP2, i.e., fast endosomal recycling. This likely indicates the need for different types of membranes being transported to the cell surface depending on the cell environment. We think that our findings are related to the previous demonstration that VAMP7 mediates the transport of GPI-anchored proteins and lipid microdomains to the plasma membrane (Lafont et al., 1999, Molino et al., 2015, Pocard et al., 2007), which in turn modulates integrin dynamics and adhesion (Eich et al., 2016, van Zanten et al., 2009), epidermal growth factor (EGFR)-containing microdomains, and EGF signaling (Danglot et al., 2010). Accordingly, increased exocytosis of GPI-anchored proteins was found in the secondary contractile phase during cell spreading (Gauthier et al., 2011). An attractive hypothesis would be that VAMP7+/CD63+ secretory late endosomes and lysosomes bring to the plasma membrane lipids that best fit a membrane under tension such as on more rigid substrates and more general transport membrane microdomains, which are important for cells to regulate their biomechanical properties and sense substrate rigidity. Further studies are now required to decipher the precise signaling mechanism from rigidity sensing and cortical tension all the way down to VAMP7 exocytosis.

The mechanism unraveled here further suggests the involvement of two members of VAMP7's hub in mechanosensing-dependent regulation of transport and exocytosis (see working hypothesis model in Figure 8). Here we found that LRRK1 strongly interacts with LD and SNARE domain of VAMP7 with a particularly strong interaction with LD in vitro. LRRK1 and VAMP7 were co-transported to the cell center upon EGF addition. Silencing LRRK1 removed the regulation of VAMP7 exocytosis by substrate rigidity. LRRK1 overexpression concentrated VAMP7 in the cell center. This effect dominated over substrate rigidity and was further emphasized by kinase activity, as it was previously shown in the case of the EGFR (Ishikawa et al., 2012). VARP mediates transport of VAMP7 to the cell periphery (Burgo et al., 2009, Burgo et al., 2012, Hesketh et al., 2014). Here we found that VARP bound efficiently ΔLD VAMP7 and its overexpression decreased the perinuclear pool of VAMP7 while increasing the peripheral one. Our data thus give a reasonable explanation for the increased exocytosis frequency of ΔLD VAMP7, as the latter would still efficiently bind to VARP and less to LRRK1. VARP interacts with the retromer and kinesin 1, and these interactions are important for the transport of VAMP7 toward the cell periphery on microtubules (Burgo et al., 2009, Burgo et al., 2012, Hesketh et al., 2014, Schäfer et al., 2012). Conversely, LRRK1 interacts with the dynein complex, which mediates retrograde transport on microtubules toward the cell center (Kedashiro et al., 2015b). In addition, the movement of lysosomes toward the peripheral cytoplasm depends on the BORC/BLOC-mediated coupling to microtubule plus end-directed kinesin motors (Dennis et al., 2016, Guardia et al., 2016, Pu et al., 2015). In addition, a recent article found that Rab3 and its effector myosin IIA play an important role in the transport to the periphery and secretion of lysosomes (Encarnação et al., 2016). Here we have used COS7 cells, which were shown to lack myosin IIA (Even-Ram et al., 2007). Thus future studies should address the potential cross talk between the VARP-LRKK1 tug-of-war mechanism unraveled here, likely connected to motility on microtubules, and the one involving myosin IIA (Encarnação et al., 2016), and possibly the retromer-associated WASH complex recruited by VARP (Hesketh et al., 2014, McGough et al., 2014) and motility of lysosomes on actin microfilaments. Future studies should certainly address the potential regulation of these interactions by substrate rigidity sensing and signaling.

Altogether, the present data lead us to propose a tug-of-war molecular mechanism with LRRK1 and VARP competing for VAMP7 binding, LRRK1 acting on the retrograde end and VARP on the anterograde end of VAMP7 trafficking. Our results further suggest that substrate stiffness would be able to regulate the tug-of-war between LRRK1 and VARP for lysosome positioning in the cell periphery and exocytosis. The effects of LRRK1 and VARP suggest that their concentration in the cell is important for VAMP7 center to periphery distribution, fitting well with the notion of a tug-of-war mechanism. The signaling pathways that may regulate the expression of LRRK1 and VARP will also require further investigation as our data suggest that the relative amounts of these proteins are important for secretory lysosome positioning. Interestingly, VARP binds Rab40C and this protein promotes proteasomal degradation of VARP (Yatsu et al., 2015), and RACK1 inhibits the interaction between Varp and Rab40C, thus counteracting the negative effect of Rab40C (Marubashi et al., 2016). RACK1 is downstream of β1 integrin (Lee et al., 2002, Liliental and Chang, 1998) and IGFR1 (Kiely et al., 2002, Zhang et al., 2006), which has been shown to activate VAMP7 exocytosis (Burgo et al., 2013). Thus it is tempting to speculate that the turnover of VARP may be regulated by integrin-dependent mechanosensing, a hypothesis that will now require investigation.

In conclusion, we suggest that VAMP7 lysosomal secretion is regulated by biomechanical constraints relayed by LRRK1 and VARP (Figure 9), a mechanism with potential broad relevance for plasma membrane dynamics in normal conditions (Koseoglu et al., 2015), infection (Chiaruttini et al., 2016, Larghi et al., 2013), stem cell differentiation (Engler et al., 2006), and cancer (Kostic et al., 2009, Steffen et al., 2008), which all were shown to be linked to mechanical constraints.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.