Initiation of BMP2 signaling in domains on the plasma membrane†
Jeremy Bonor and Elizabeth L. Adams are contributed equally to this study.
Abstract
Bone morphogenetic protein 2 (BMP2) is a potent growth factor crucial for cell fate determination. It directs the differentiation of mesenchymal stem cells into osteoblasts, chondrocytes, adipocytes, and myocytes. Initiation of BMP2 signaling pathways occurs at the cell surface through type I and type II serine/threonine kinases housed in specific membrane domains such as caveolae enriched in the caveolin-1 beta isoform (CAV1β, caveolae) and clathrin-coated pits (CCPs). In order for BMP2 to initiate Smad signaling it must bind to its receptors on the plasma membrane resulting in the phosphorylation of the BMP type Ia receptor (BMPRIa) followed by activation of Smad signaling. The current model suggests that the canonical BMP signaling pathway, Smad, occurs in CCPs. However, several recent studies suggested Smad signaling may occur outside of CCPs. Here, we determined; (i) The location of BMP2 binding to receptors localized in caveolae, CCPs, or outside of these domains using AFM and confocal microscopy. (ii) The location of phosphorylation of BMPRIa on the plasma membrane using membrane fractionation, and (iii) the effect of down regulation of caveolae on Smad signaling. Our data indicate that BMP2 binds with highest force to BMP receptors (BMPRs) localized in caveolae. BMPRIa is phosphorylated in caveolae and the disruption of caveolae-inhibited Smad signaling in the presence of BMP2. This suggests caveolae are necessary for the initiation of Smad signaling. We propose an extension of the current model of BMP2 signaling, in which the initiation of Smad signaling is mediated by BMPRs in caveolae. J. Cell. Physiol. 227: 2880–2888, 2012. © 2011 Wiley Periodicals, Inc.
Embryonic development and tissue homeostasis are dependent upon stem cell fate and differentiation. For stem cells to properly differentiate they must receive both extracellular and intracellular signals. BMP2 is one of the earliest and most effective growth factors to designate cellular fate during embryogenesis. BMP2 signals through BMP receptors (BMPRs), type I and type II serine/threonine kinase receptors. In particular, the BMPR type Ia (BMPRIa) and BMPR type II (BMPRII) are expressed in various tissues (Ide et al., 1998; Yanai et al., 2005; Bragdon et al., 2011). Once BMP2 binds to its receptors, BMPRII phosphorylates and activates BMPRIa. Activated BMPRIa initiates several downstream signaling pathways including the BMP2 canonical pathway Smad, and others such as p38 Map kinase, NFkB, JNK, PI-3K, and ERK 1/2 (Yamashita et al., 1996; Itoh et al., 2000; Reilly et al., 2005; Mukai et al., 2007; Gamell et al., 2008; Sieber et al., 2009; Pachori et al., 2010).
Although BMPRIa and BMPRII are segregated and colocalize to caveolae, clathrin-coated pits (CCPs) and outside of these domains on the plasma membrane (Nohe et al., 2003, 2004, 2005; Hartung et al., 2006; Ramos et al., 2006; Wertz and Bauer, 2008; Bragdon et al., 2009; Jiang et al., 2011), the current model proposes that Smad signaling is activated in CCPs (Hartung et al., 2006). Caveolae are flask-shaped invaginations of the cell membrane enriched with cholesterol and Caveolin-1 (CAV1). There are two isoforms of CAV1, CAV1α, and CAV1β (Fujimoto et al., 2000; Kogo and Fujimoto, 2000; Ramirez et al., 2002). CCPs are sites of classical coated pit endocytosis and receptor recycling where the membrane is coated with Clathrin, Dynamin, and adapter proteins including Adaptor protein 2 (AP2) (Mellman, 1996; Schmid, 1997; McNiven, 1998). However, this signaling model does not include recent findings. Several recent studies suggested Smad signaling may occur outside of CCPs (Rauch et al., 2002; Chen et al., 2007; Murphy, 2007; Shi et al., 2007; Wertz and Bauer, 2008; Bragdon et al., 2009). Using atomic force microscopy (AFM) combined with confocal microscopy, we showed that BMP2 binds with highest force to BMPRs localized in caveolae. Membrane fractionation showed BMPRIa is phosphorylated in caveolae. Additionally, reporter gene assays showed that disruption of caveolae did not induce Smad signaling in the presence of BMP2, suggesting caveolae are necessary for the initiation of Smad signaling. We propose an extension of the current model of BMP2 signaling that the initiation of Smad signaling is mediated by BMPRs in caveolae.
Materials and Methods
Materials
Recombinant BMP2 was obtained from GenScript (Piscataway, NJ). The cell line C2C12 was purchased from American Type Culture Collection (Manassas, VA). Monoclonal mouse anti-phospho-serine (p-Ser) antibody and G-sepharose were purchased from Invitrogen (Camarillo, CA). Polyclonal goat anti-sera against BMPRIa, horseradish peroxidase (HRP)-conjugated goat anti-donkey IgG, goat anti-mouse IgG-HRP, siRNA against CAV1, and BMPRIa were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal mouse anti-CAV1 was purchased from BD Transduction Laboratories (San Jose, CA).
Cell culture
Murine myoblast cells (C2C12) were grown in Dulbecco's Modified Eagle's Medium (Hy-Clone, Pittsburgh, PA) supplemented with 10% (v/v) fetal bovine serum (FBS) (Gemini Bio-Products, West Sacramento, CA), 0.5% (v/v) L-Glutamine (Cellgro, Manassas, VA), and 1% (v/v) penicillin/streptomycin (Hy-Clone).
Transfection of C2C12 cells
Cells were transfected with Turbofect from Fermentas Inc. (Glen Burnie, MD) using manufacturer's protocol. One microgram of plasmid for imaging, 2 µg of plasmid for reporter gene assays, and 20 or 80 ρmole of the siRNA (Santa Cruz Biotechnology) against BMPRIa or CAV1 was used for transfections.
Functionalizing AFM Tip
The AFM probes (Bruker Probes and NanoWorld, nominal spring constant of 0.06–0.08 N/m, August-Schanz-Straße) were silanized [4% 3-amino-propyltrimethylethoxysilane (3-APTES)] in 95% ethanol solution for 1 h at room temperature. Surfaces were activated using 2.5% glutaraldehyde solution, followed by incubation with BMP2 (500 nM; GenScript). Functionalized AFM probes stored in PBS at 4°C until use.
Confocal-AFM setup
Images were collected by incorporating an inverted laser scanning confocal microscope (Zeiss LSM 510 NLO, Thornwood, NY) with an AFM (Bruker Inc, Bioscope II AFM, August-Schanz-Straße). The process to set up the instrument and collect the data involved alignment and calibration of the instruments following integration. The maximum load was set below Bueckle's indentation depth limit of 10% of the specimen thickness (Buckle, 1973; Persch et al., 1994). For the C2C12 cells, the maximum load was set to 21 nN or 230 nm. When all the settings were aligned and tested, the data for a specific region of the cell was collected by using the Force volume setting at ramp speed (4 µm/s tip velocity), scan rate of 14.0 Hz, samples/line 16 at 16 lines resulting in 256 data points collected. These data were then extracted and tested for binding evens, which gave us the binding interaction that have occurred out of the 256 events collected. Image registration was performed using microscopy image registration overlay (MIRO) canvas software on the Bruker Bioscope II AFM in order to align the movements of the AFM stage with the Zeiss LSM 510 confocal image so that any movement in tip position matched the confocal field of view. The number of binding events typically ranged from 80 to 190 with associated binding forces of 0.1–1 nN. The cell membrane was expected to be relatively flat, so it would be anticipated that there would be an absence of large peaks or valleys within the scanned AFM area. After registration, fine tuning of tip position was performed.
Data were collected from regions of the membrane, away from the cell nucleus, using the AFM force volume parameters (16 × 16 image with a scan size of 4.49 µm) immediately followed by confocal microscope imaging. Confocal images (512 × 512) were acquired using a Zeiss 100xPlan-Apochromat objective lens (numerical aperture 1.4) and 488 nm argon laser excitation with 500–550 nm band pass emission filter. The closed loop scanner made it possible to collect data from the area defined by the MIRO software. Force and topographical data were also simultaneous collected.
The data were screened during and after data collection, using the following criteria: data were rejected if; (i) the AFM tip drifted off the cell during data collection, (ii) the cell detached from the substrate causing movement of the cell under the scanning AFM tip (ranging from slight movement of membrane regions to complete detachment of the cell), (iii) Tip contamination resulting in very high or very low number of binding events, (iv) If the agreement was not good between the AFM height image and the confocal image of the cell membrane, and (v) Overexpression of caveolin-1 beta was shown to inhibit BMP2 signaling (Nohe et al., 2005). Therefore, we ensured to select cells not significantly overexpressing caveolin-1 beta by choosing cells with low intensity.
After data passed the validation steps the images from both the AFM and the confocal (collected after the AFM image) were both imported into the Reconstruct software (Release: 1.1.0.0). To take into account any cellular movement the AFM images were then warped to match the confocal image as closely as possible. These aligned images were then imported into Corel Photo-Paint. After the overlay alignment of the image sets, the resulting stack of images was then cropped to the AFM data region using Corel Photo-Paint. The cropped region was then enlarged 300%, using the resample tool, and the dpi were increased by the same 300% resulting in an image of workable size.
Confocal-AFM data analysis
Once equipment setup and calibrations were completed, regions of the cell were then selected for force volume analysis. In each case, a threshold force value was used to help determine the presence or absence of the specific regions in plasma membranes. Analysis of the force volume data sets revealed approximately 80–190 force curves containing detectable binding forces out of the total 256 force curves collected per data set (forces ranged from 0.1 to 1.0 nN). An overlay was then created and used as a reference to map the confocal data from the live cell. The binding forces were matched with the confocal image collected using this overlay. For the confocal image, a z-stack was collected after the force volume mapping and the optical slice representing the upper membrane was used in the analysis. Only a limited number of caveolae vesicles are known to be present in the cytoplasm at room temperature and movement of the domains on the cell surface are very slow (≈0.9 ± 0.2 × 10−12 cm2/sec−1) About 5–10% of caveolae are immobile (Keating et al., 2008).
Sucrose density gradients
Sucrose density gradients were performed by modifying the procedure (Song et al., 1996). C2C12 cells were grown to 90% confluency in large flasks and serum starved for 24 h before stimulation. Cells were then stimulated or not stimulated with 40 nM BMP2 for 0 or 30 min and lysed using 3 ml of 1% Triton X-100 lysis buffer (1% Triton X-100, 1 mM EDTA pH 7.4, 10 mM Tris pH 7.4) in the presence of 1 mM PMSF, 1 mM DTT, 10 mM tetrasodium pyrophosphate, 17.5 mM β-glucrophosphate, and protease inhibitor mix (1 mg/ml of each of Leupeptin, Aprotinin, Soybean Trypsin inhibitor, Benzamidine/HCl, Pepstatin, and Antipain), followed by pulse sonication. Three milliliters of sample were then mixed with 3 ml of 90% sucrose solution in Mes-Buffered saline (25 mM Mes pH 6.5, 0.15 M NaCl), placed at the bottom of the ultracentrifuge tubes, and overlaid with 3 ml 35% Sucrose Mes-Buffered Saline (0.25 M Na2CO3). Three milliliters of 5% sucrose in Mes-Buffered saline containing 0.25 M Na2CO3 was then placed on the top of the tube discontinuous sucrose gradient. Tubes were placed in Beckman swinging bucket rotor (SW41Ti) and spun at 39,000 rpm for 18 h at 4°C in Beckman L8-55 M ultracentrifuge. One milliliter fractions were collected from the top of the gradient followed by trichloroacetic acid (TCA) precipitation (precipitated with 20% v/v TCA, centrifuged, and washed with acetone). BMPRIa was immunoprecipitated as described in (Bragdon et al., 2010). Samples were put in 5× SDS running buffer and run on SDS–PAGE gel. Western blotting was performed using a CAV1 antibody, BMPRIa antibody, or p-Ser antibody. Western blot was detected by enhanced chemiluminescence (ECL) (Pierce, Thermo Scientific, Rockford, IL).
Reporter gene assays
Reporter gene assays were performed as described in (Bragdon et al., 2010) with the exception CAV1 was knocked down. Cells were treated with 5 nM Nystatin to disrupt caveolae (Matveev et al., 2001). Luciferase activity was measured using a Firefly & Renilla Luciferase Assay Kit from Biotium (Hayward, CA).
Statistical analysis
Experiments were performed three independent times or five dependent cells for the confocal-AFM experiments. Values of the standard error of the mean (SEM) were calculated from the raw data at the 95% confidence level and was used for error bars on graphs. Significance was established with ANOVA followed by Tukey test or Student t-test as indicated. Data for reporter gene assays are presented as a normalized value which is the value divided by control values.
Results
Development of a biologically active BMP2-coated AFM tip
AFM combined with confocal microscopy was applied to determine the location of BMP2 binding to its receptors on the plasma membrane in live cells. AFM is a powerful tool that can be used to study ligand–receptor interactions. By combining AFM with confocal microscopy ligand–receptor interactions on the plasma membrane of live cells can be quantified (Haupt et al., 2006; Doak et al., 2008; Muller et al., 2009).
First, a functional BMP2-coated AFM tip was developed by covalently linking BMP2 to the AFM tip as described in Materials and Methods section. This tip was used to quantify the binding force and location of BMP2 interactions with the plasma membrane. Force volume measurements of BMP2 interactions with the cell surface were determined on at least three different cells using an AFM. The specificity of BMP2 binding was verified by measuring the binding forces and events of an uncoated and a BMP2-coated AFM tip.
Figure 1A shows 23 binding events per cell. This was measured using the non-coated AFM tip. The binding events were significantly increased to 78 using the BMP2 functionalized AFM tip. When soluble BMP2 was added to block BMP2-binding sites on the plasma membrane, the number of binding events with the BMP2-coated AFM tips decreased to 60 (Fig. 1B). The binding curves were collected at one location on five cells, resulted in a range of binding forces. The non-coated AFM tip binding force with the cell surface of live C2C12 cells was 0.013 nN. The binding forces for the BMP2-coated AFM tips significantly increased to 0.157 nN (Fig. 1C). Addition of soluble BMP2 (133 nM) to the dish reduced the average binding force of the BMP2-coated tip to 0.09 nN (Fig. 1D).
BMP2 is biologically active when coated to AFM tip. AFM tips were either non-coated (uncoated), BMP2, or BSA-coated and used to probe the cell surface (A and C). As an additional control, soluble BMP2 was added prior to using BMP2-coated tips (soluble BMP2) (B and D). A: Number of binding events from the BMP2-coated tips, and BSA-coated tips were increased compared to uncoated AFM tips. B: Addition of soluble BMP2 reduced the number of binding events significantly. C: The average binding force increased with the BMP2-coated AFM tips compared to uncoated AFM tips. The binding forces for the BSA-coated AFM tips were similar to the uncoated-AFM tips. D: The addition of soluble BMP2 decreased the binding force of the BMP2-coated AFM tips. *Significance where P < 0.05 as determined by ANOVA followed by Tukey's test.
As another control BSA was coated to the AFM tip. Binding of BSA to the cell surface revealed a random binding pattern with an average force of 0.2 nN and 139 binding events per cell, demonstrating unspecific binding at the cell surface (Fig. 1A,C). These results verified that BMP2 is able to interact with the cell surface when functionalized to an AFM tip.
BMP2 binding and force volume is increased in caveolae
With our newly functionalized AFM tip we investigated the location of BMP2 interaction with the plasma membrane using combined confocal-AFM. C2C12 cells were transfected with plasmids encoding CAV1β, provided by Hiroshi Kogo, (Department of Anatomy, Nagoya University School of Medicine, Nagoya, Japan) and AP2, obtained from Dr. Steve Furguson, (Roberts Research Institute, London, ON, Canada) fused to green fluorescent protein (GFP). These proteins were used to visualize caveolae and CCPs, respectively. Additionally, to validate the results of BMP2-binding forces, cells were co-transfected with siRNA against CAV1 or BMPRIa resulting in protein knockdown as described by manufacturing protocol. Combined confocal-AFM was applied to determine the binding forces for BMP2–BMPRIa interactions mapped to caveolae, CCPs, and regions independent of these domains (areas not fluorescently marked) in live cells. Briefly, the alignment procedure for confocal and AFM used three points as shown in Figure 2A–C. The AFM tip position was then fine-tuned (Fig. 2D), data were collected from the area of the cell surface and exported (Fig. 2E,F). The number of binding events typically ranged from 80 to 90 with associated binding forces of 0.1–1 nN. The cell membrane was expected to be relatively flat (Fig. 2F), so it would be anticipated that there would be an absence of large peaks or valleys within the scanned AFM area. After registration, fine tuning of tip position was performed AFM and confocal images were collected. Representative images and binding curves that were collected are shown in Figure 3. An overlay was then created to be used as a reference to map the confocal data from the live cell (Fig. 4). The binding forces were matched with the confocal image collected using the overlay (Table 1).
Alignment of the confocal and closed loop AFM. A–C: Sample images depicting the three points used by the MIRO software to calculate AFM stage position relative to the confocal field-of view. Circles indicate the same region in each image. D: Sample image used to fine-tune the AFM tip position. E: Area of contact of AFM tip shown in canvas overlay exported from MIRO AFM software. F: Three dimensional AFM topographical image of the live cell depicted in E with a height scale of 500 nm show by the square.
Representative images of the combined AFM and confocal. Sample confocal-AFM images depicting the BMP2 interaction sites with the plasma membrane of live C2C12 cells transfected with a plasmid encoding CAV1β fused to GFP. A: Low magnification image showing the DIC image of the cell overlaid with the confocal image obtained in the green channel. B: High magnification confocal images of the green channel. C: Corresponding AFM image of (B). Gray dotted, gray, and black circles show the different binding strengths at a specific location on the plasma membrane. Gray dotted representing the lowest range of binding strength <0.1 nN, gray is intermediate >0.1 nN, and black is the highest binding force >0.3 nN. D,E: Representative binding curves collected with confocal-AFM. D: High affinity binding event. E: Medium-binding event in nN. Please note significant scale difference between graphs D and E.
Force alignment with corresponding confocal images. A: Numbered grid referencing the location of AFM binding sites. B: Overlay of the grid and the collected confocal image. The grid was rotated clockwise 90° to match and then align the confocal and AFM images. Rotation was required due to the configuration of the confocal and its scan direction. C: Overlay from B with mapped plasma membrane binding sites. Gray dotted (84), gray (51), and black (48) circles depict the binding strengths at a specific locations on the membrane, with gray dotted being within the lowest <0.1 nN, gray intermediate >0.1 nN and black highest binding force >0.3 nN. D: Grid with overlaid confocal image depicting the mapping of the binding events. On each square within the grid, the presence of a binding event is mapped to the corresponding grid reference with white circles denotes outside of CAV1β and gray circles shows the interactions within CAV1β. Of the white circles, 67, 36, and 28 corresponded to the lowest, intermediate, and highest binding forces, respectively. For the gray circles or interactions within CAV1β, there were 16 sites each that showed for the lowest and intermediate binding forces while there were 24 binding sites that had the highest binding forces.
| Binding event | CAV β | Binding force (nN) | Binding event | CAV β | Binding force (nN) | Binding event | CAV β | Binding force (nN) |
|---|---|---|---|---|---|---|---|---|
| 1 | 0.044 | 92 | 0.346 | 178 | + | 0.128 | ||
| 2 | + | 0.089 | 93 | 0.069 | 179 | 0.479 | ||
| 4 | 0.059 | 94 | + | 0.151 | 180 | 0.099 | ||
| 5 | + | 0.056 | 95 | + | 0.439 | 181 | 0.005 | |
| 6 | + | 0.104 | 96 | + | 0.274 | 182 | + | 0.377 |
| 7 | + | 0.259 | 97 | 0.089 | 184 | 0.037 | ||
| 8 | + | 0.103 | 98 | 0.286 | 185 | 0.076 | ||
| 9 | + | 0.185 | 100 | + | 0.382 | 186 | 0.101 | |
| 10 | + | 0.434 | 101 | + | 0.158 | 187 | 0.131 | |
| 14 | 0.067 | 104 | + | 0.094 | 188 | 0.02 | ||
| 15 | + | 0.062 | 105 | 0.081 | 189 | 0.093 | ||
| 18 | 0.098 | 106 | 0.204 | 190 | 0.027 | |||
| 19 | + | 0.034 | 108 | 0.042 | 191 | + | 0.447 | |
| 20 | + | 0.113 | 109 | + | 0.543 | 192 | 0.034 | |
| 22 | 0.362 | 110 | 0.034 | 193 | 0.077 | |||
| 23 | + | 0.452 | 112 | + | 0.108 | 194 | + | 0.155 |
| 24 | 0.045 | 114 | + | 0.272 | 195 | + | 0.099 | |
| 25 | 0.081 | 115 | + | 0.239 | 196 | 0.145 | ||
| 26 | + | 0.02 | 116 | + | 0.365 | 197 | 0.027 | |
| 28 | + | 0.463 | 117 | 0.126 | 198 | + | 0.437 | |
| 29 | + | 0.047 | 118 | 0.076 | 200 | + | 0.124 | |
| 30 | 0.355 | 121 | 0.082 | 201 | + | 0.076 | ||
| 31 | 0.457 | 122 | 0.452 | 202 | + | 0.042 | ||
| 33 | + | 0.338 | 123 | 0.116 | 205 | 0.387 | ||
| 35 | + | 0.444 | 124 | 0.256 | 207 | 0.027 | ||
| 36 | + | 0.025 | 125 | 0.035 | 208 | 0.084 | ||
| 37 | + | 0.062 | 127 | 0.076 | 209 | 0.032 | ||
| 38 | + | 0.257 | 128 | 0.336 | 210 | + | 0.131 | |
| 40 | + | 0.37 | 129 | 0.14 | 211 | 0.022 | ||
| 44 | 0.079 | 130 | 0.153 | 212 | + | 0.185 | ||
| 45 | + | 0.392 | 132 | 0.04 | 213 | 0.072 | ||
| 46 | 0.093 | 134 | 0.037 | 214 | 0.128 | |||
| 47 | 0.14 | 136 | 0.252 | 219 | 0.091 | |||
| 48 | + | 0.399 | 137 | 0.079 | 220 | 0.104 | ||
| 49 | + | 0.084 | 138 | 0.045 | 222 | 0.091 | ||
| 50 | + | 0.363 | 142 | 0.045 | 223 | 0.044 | ||
| 51 | + | 0.542 | 143 | 0.002 | 224 | + | 0.13 | |
| 52 | + | 0.111 | 144 | + | 0.118 | 225 | 0.103 | |
| 53 | + | 0.091 | 145 | 0.084 | 228 | 0.404 | ||
| 55 | 0.071 | 146 | 0.365 | 229 | + | 0.35 | ||
| 56 | 0.025 | 148 | 0.138 | 230 | + | 0.345 | ||
| 59 | 0.089 | 149 | 0.175 | 231 | 0.126 | |||
| 60 | + | 0.14 | 150 | 0.146 | 232 | 0.027 | ||
| 61 | 0.167 | 151 | + | 0.37 | 233 | + | 0.483 | |
| 62 | 0.091 | 152 | + | 0.037 | 234 | 0.486 | ||
| 63 | 0.049 | 158 | 0.15 | 235 | 0.419 | |||
| 64 | 0.084 | 160 | + | 0.054 | 236 | 0.005 | ||
| 65 | 0.056 | 161 | + | 0.099 | 237 | 0.385 | ||
| 67 | + | 0.494 | 162 | + | 0.108 | 238 | 0.362 | |
| 68 | + | 0.333 | 163 | 0.069 | 240 | + | 0.328 | |
| 75 | + | 0.309 | 164 | + | 0.197 | 241 | + | 0.093 |
| 76 | + | 0.35 | 165 | + | 0.457 | 242 | 0.18 | |
| 77 | 0.067 | 166 | 0.135 | 243 | 0.042 | |||
| 78 | + | 0.18 | 167 | 0.116 | 244 | 0.089 | ||
| 81 | + | 0.066 | 168 | 0.283 | 245 | + | 0.604 | |
| 83 | 0.05 | 169 | 0.101 | 246 | + | 0.703 | ||
| 84 | 0.012 | 170 | 0.441 | 247 | + | 0.451 | ||
| 85 | + | 0.479 | 171 | 0.207 | 248 | 0.436 | ||
| 88 | 0.303 | 172 | 0.025 | 249 | 0.533 | |||
| 89 | 0.084 | 173 | 0.022 | 250 | 0.523 | |||
| 90 | 0.044 | 174 | + | 0.419 | 252 | 0.367 | ||
| 91 | + | 0.156 | 175 | 0.511 | 253 | + | 0.299 | |
| 254 | + | 0.074 |
- Binding event column represents the grid location number. +: denotes presence of CAV β-GFP detected in confocal images. We defined the interactions that had at least 50% green in that binding event as being in CAV1 β.
The data shown in Figure 5A, the binding force of BMP2 (0.335 nN), was significantly increased in caveolae enriched CAV1β regions compared to the CAV1β independent regions (0.2 nN). Knockdown of CAV1β using siRNA showed a significant reduction in the binding force of BMP2 (0.21 nN) and binding events (60) compared to total binding events (130). Knockdown was achieved as shown in Figure 5E. The knockdown of BMPRIa reduced the total number of binding events from 130 to 78 (20 ρmoles siRNA BMPRIa) and 61 (80 ρmoles siRNA BMPRIa) (Fig. 5B).
BMP2 binds predominately to regions expressing Caveolin-1 β. The binding forces and events were quantified on live C2C12 cells that were transfected with plasmids encoding CAV1β GFP (A–B) or AP2 GFP (C). In addition to the plasmid transfection, other treatments included the knockdown of (A,B,E) CAV1 or (C) BMPRIa (at 20 and 80 ρmoles) by co-transfection with siRNA against CAV1 or BMPRIa. The outside of regions of CAV1β (outside Cavβ) or AP2 (outside AP2) is defined by all other areas not in the specific domain as determined by fluorescence. A. The average binding force of BMP2 to the outside and knockdown of CAV1 (siRNA CAVβ) were significantly decreased compared to regions enriched in CAV1β. B. Knockdown of CAV1β or BMPRIa (20 or 80 ρmoles) significantly decreased the number of binding events compared to total binding events. C: There significantly increased binding forces associated with the outside AP2 compared to the AP2 domains (*Significance where P < 0.05 as determined by ANOVA followed by Tukey's test). D: A sample histogram of the binding strengths. E: The normalized fluorescent intensity of overexpressed CAV1β. Knockdown of CAV1 with siRNA significantly decreased the expression (fluorescence) of CAV1β (*Significance where P < 0.05 as determined by t-test).
To investigate the binding of BMP2 to CCPs the binding force associated with CCPs was collected from C2C12 cells transfected with a plasmid encoding AP2 GFP as a marker. The results in Figure 5C showed a significant decrease in the BMP2-binding force in CCPs (0.15 nN) compared to outside AP2 (0.23 nN).
Phosphorylated BMPRIa is associated with caveolae
Once BMP2 binds to BMPRIa and BMPRII, the BMPRII activates BMPRIa by phosphorylation. Since BMP2 interacts with BMPRIa localized in caveolae, the phosphorylation state of BMPRIa in caveolae was determined using membrane fractionation. C2C12 cells were stimulated with 40 nM BMP2 for 0 or 30 min. Lysates were subjected to continuous and discontinuous sucrose density gradient as described in the Materials and Methods section. After ultracentrifugation fractions were collected from the top of the gradient and were immunoprecipitated for BMPRIa. Samples were subjected to SDS–PAGE followed by detection for p-Ser as a marker of phosphorylated BMPRIa (p-BMPRIa). To determine the fractions containing CAV1, marker for caveolae, cells were lysed at 0 min BMP2 stimulation and subjected to sucrose gradient. After fractions were applied to SDS–PAGE followed by Western blotting for CAV1, caveolae were detected in fractions 3–5 (Fig. 6). Under non-stimulated conditions, the BMPRIa bands were detected in fractions 3–8 with trace amounts of p-BMPRIa identified in the CAV1 associated fraction 4. Addition of BMP2 increased p-BMPRIa in fractions 4 and 5; both fractions are associated with CAV1. Markers for CCPs, Golgi, endoplasmic reticulum (ER), and early endosomes are detected in dense fractions (Shin et al., 2000; Cai et al., 2008). For here it would be fractions 8, 9, and 10. BMPRIa phosphorylation was also detected in other fractions. There was no change with the total BMPRIa distribution after BMP2 stimulation; bands were located in fractions 3–10.
BMPRIa is phosphorylated in caveolae. C2C12 cells were stimulated with BMP2 for 0 or 30 min and lysed. Lysates were subjected to sucrose density gradients and fractions were collected from the top of the gradient (labeled as 1). BMPRIa was immunoprecipitated from the fractions and Western blotting was performed. To determine the fractions containing CAV1, non-stimulated samples were not immunoprecipitated. Western blotting was performed with samples from each fraction showing the presence of CAV1 where caveolae fractions (3–5) are located. It showed BMPRIa is phosphorylated at 30 min of BMP2 stimulation and predominately localized to the caveolae fractions.
Caveolae regulate BMP2-induced Smad signaling
Downstream of the phosphorylation of BMPRIa is the activation of the Smad-1, -5, and -8 pathway. These Smads complex with Smad 4 and translocate to the nucleus where gene transcription is regulated (Itoh et al., 2000). Since our results obtained from the combined confocal AFM experiments and BMPRIa immunoprecipitation and phosphorylation experiment suggested caveolae were the centers for the initiation of BMP2 signaling we defined the role caveolae has on the genetic level of Smad signaling. C2C12 cells were transfected with plasmids encoding pSBE (Smad-binding element fused with firefly luciferase), and pRLU (renilla luciferase used for normalization). Caveolae were disrupted by transfection with siRNA against CAV1 or treated with Nystatin (5 nM). Cells were stimulated with 40 nM BMP2 or non-stimulated (control). Treatment with BMP2 induced a significant 2.6-fold increase in Smad signaling compared to the control (Fig. 7). When caveolae were disrupted by the knockdown of CAV1 or Nystatin treatment, the BMP2 stimulation did not result in a significant Smad signal. The reporter gene assay using the pSBE was recently shown to correlate with Smad phosphorylation in C2C12 cells (Jiang et al., 2011).
BMP2 initiated Smad signaling. C2C12 cells were transfected with plasmids encoding pSBE and pRLuc, in addition the experimental cells were transfected with siRNA against CAV1 in order to knockdown CAV1 protein or nystatin to disrupt caveolae. Control cells were either non-stimulated or stimulated with BMP2 (40 nM). The results showed a significant increase in the luciferase activity with the BMP2 stimulation. Addition of BMP2 and disruption of caveolae did not significantly increase Smad signaling (*Significance where P < 0.05, determined by ANOVA followed by Tukey's test).
Discussion
Our data showed that caveolae are the centers for the initiation of BMP2 signaling, BMP2 interacts with BMPRIa, and BMPRIa is phosphorylated. Further caveolae regulate Smad signaling. Therefore, the current model of Smad signaling must be extended. Combination of confocal microscopy and AFM allowed us to measure the interactions between BMP2 and its receptors by quantifying the number of binding events and forces on live cells. We determined that BMP2 was able to interact with and bind to its receptors at the cell surface when attached to an AFM tip by measuring a range of different binding forces. Additionally, our results using correlative approach demonstrated that BMP2 bound with a higher affinity to BMPRIa localized to domains enriched with CAV1β compared to the other experimental conditions tested (CCPs or outside). Reduction of caveolae by downregulation via siRNA for CAV1 led to a significant reduction in the force of BMP2 binding and number of binding events demonstrating the importance of these regions in BMP2 binding. Downregulation of the BMP2 receptor BMPRIa by siRNA resulted in a significant decrease in the number of binding events demonstrating that our data were dependent upon BMP2 binding specifically to BMPRIa, not non-specific binding with the cell surface or other receptors. BMP2 does binds with high affinity to BMP type I receptors compared to BMP type II receptors (Koenig et al., 1994; Liu et al., 1995). Further the BMP type I receptor, BMBRIb, mRNA was not found in C2C12 cells (Mukai et al., 2007; Zilberberg et al., 2007). The data demonstrated for the first time that BMP2 receptors were housed in distinct environments on the plasma membrane and capable of binding to BMP2 with varying forces.
Downstream of BMP2 binding is BMPRIa phosphorylation by BMPRII and subsequently activation of BMP2 signaling. Downstream is the Smad pathway which regulates the genetic response. The sucrose density gradients showed the appearance of phosphorylated BMPRIa in caveolae therefore the initiation for BMP2 signaling is transduced in caveolae. Although phosphorylated BMPRIa was found to be in more dense fractions these include the Golgi, ER, CCPs, and early endosomes. Interestingly, BMPRIa was slightly phosphorylated in the absence of BMP2. This could be due to low levels of activated BMPRIa resulting in basal signaling or inhibitory phosphorylation, since these cells were serum starved resulting in cell cycle arrest. This phosphorylation also may be due to CK2 which is known to interact with BMPRIa in the absence of BMP2 (Bragdon et al., 2010). Further, the genetic response of BMP2-induced Smad pathway was decreased when caveolae were knocked down by siRNA against CAV1 or disrupted by Nystatin, demonstrating importance of caveolae in Smad signaling. Fascinatingly, treatment with Nystatin decreased Smad signaling to a different extent than the knockdown of CAV1. The Smad signal with the nystatin treatment was approximately a third of the BMP2-induced Smad signal. Work by Shi et al. (2007) showed that 20–30% of the Smad pathway is decreased when the clathrin-coated pit contribution to Smad signaling was decreased by knockdown of Endofin (an anchoring protein for Smad to clathrin-coated pit endocytosis) (Shi et al., 2007). This indicates that CCPs are indeed areas for Smad signaling, but approximately 30%. This is similar to the response seen with the nystatin treatment, and within the error of the experiment.
The current model for BMP2 signaling must be extended to include the specific role of membrane domains in the initiation of Smad signaling (Rauch et al., 2002; Nohe et al., 2005; Hartung et al., 2006; Bragdon et al., 2009). Based on our data and previously published work, we suggest a new model for BMP signaling. Figure 8 shows the previous model for the activation of BMP2-induced Smad signaling. In this model, BMPRIa and BMPRII are localized in caveolae and CCPs. BMP2 binding initiates Smad signaling in CCPs and Smad dependent gene activation is dependent on endocytosis through CCPs (Hartung et al., 2006). On the other hand BMP2 binding to receptors housed in caveolae are responsible solely for the Smad independent signaling. Using our data and published recent literature, we suggest an extension of this model indicating that Smad signaling is additionally activated in caveolae and that BMP2 binds with a higher force and higher frequency to receptors localized in these domains.
Model of caveolae regulation of BMP2 induce Smad signaling. A: Recent model of BMP signaling. BMP receptor type Ia and type II are localized in caveolae and CCPs. BMP2 binding initiates Smad signaling in CCPs and Smad dependent gene activation is dependent on endocytosis through CCPs. On the other hand BMP2 binding to receptors housed in caveolae are responsible solely for the Smad independent signaling. B: New model for BMP2 initated Smad signaling. Using our data and previously published recent literature, suggest an extension of this model indicating that Smad signaling is additionally activated in caveolae and that BMP2 binds with a higher force and higher frequency to receptors localized in these domains.
Acknowledgements
We would like to express our sincere appreciation to Dr. Deni Galileo for the use of his ultracentrifuge, and Dr. Robert Gundersen for his helpful suggestions with sucrose gradients. We would like to acknowledge Joyita Dutta, Hemanth Akkiraju, and Rachel Schaefer for their support in the lab and suggestions.
