Here, we establish a protocol to simultaneously visualize and analyze multiple SMAD complexes using proximity ligation assay (PLA) in endothelial cells exposed to pathological and physiological fluid shear stress conditions.
Transforming Growth Factor β (TGFβ)/Bone Morphogenetic Protein (BMP) signaling is tightly regulated and balanced during the development and homeostasis of the vasculature system Therefore, deregulation in this signaling pathway results in severe vascular pathologies, such as pulmonary artery hypertension, hereditary hemorrhagic telangiectasia, and atherosclerosis. Endothelial cells (ECs), as the innermost layer of blood vessels, are constantly exposed to fluid shear stress (SS). Abnormal patterns of fluid SS have been shown to enhance TGFβ/BMP signaling, which, together with other stimuli, induce atherogenesis. In relation to this, atheroprone, low laminar SS was found to enhance TGFβ/BMP signaling while atheroprotective, high laminar SS, diminishes this signaling. To efficiently analyze the activation of these pathways, we designed a workflow to investigate the formation of transcription factor complexes under low laminar SS and high laminar SS conditions using a commercially available pneumatic pump system and proximity ligation assay (PLA).
Active TGFβ/BMP-signaling requires the formation of trimeric SMAD complexes consisting of two regulatory SMADs (R-SMAD); SMAD2/3 and SMAD1/5/8 for TGFβ and BMP signaling, respectively) with a common mediator SMAD (co-SMAD; SMAD4). Using PLA targeting different subunits of the trimeric SMAD-complex, i.e., either R-SMAD/co-SMAD or R-SMAD/R-SMAD, the formation of active SMAD transcription factor complexes can be measured quantitatively and spatially using fluorescence microscopy.
The usage of flow slides with 6 small parallel channels, that can be connected in series, allows for the investigation of the transcription factor complex formation and inclusion of necessary controls.
The workflow explained here can be easily adapted for studies targeting the proximity of SMADs to other transcription factors or to transcription factor complexes other than SMADs, in different fluid SS conditions. The workflow presented here shows a quick and effective way to study the fluid SS induced TGFβ/BMP signaling in ECs, both quantitatively and spatially.
Proteins of the transforming growth factors beta (TGFβ) superfamily are pleiotropic cytokines with a variety of members, including TGFβs, bone morphogenetic proteins (BMPs), and Activins1,2. Ligand binding induces the formation of receptor oligomers leading to the phosphorylation and, thereby, activation of cytosolic regulatory SMAD (R-SMAD). Depending on the sub-family of ligands, different R-SMADs are activated1,2. While TGFβs and Activins mainly induce phosphorylation of SMAD2/3, BMPs induce SMAD1/5/8 phosphorylation. However, there are accumulating evidences that BMPs and TGFβs/Activins also activate R-SMADs of the respective other sub-family, in a process termed as 'lateral signaling'3,4,5,6,7,8 and that there are mixed SMAD complexes consisting of both, SMAD1/5 and SMAD2/3, members3,9. Two activated R-SMADs subsequently form trimeric complexes with the common mediator SMAD4. These transcription factor complexes are then able to translocate into the nucleus and regulate the transcription of target genes. SMADs can interact with a variety of different transcriptional co-activators and co-repressors, leading to the diversification of the possibilities to regulate target genes10. Deregulation of SMAD signaling has severe implications in a variety of diseases. In line with this, unbalanced TGFβ/BMP signaling may lead to severe vascular pathologies, such as pulmonary artery hypertension, hereditary hemorrhagic telangiectasia, or atherosclerosis3,11,12,13,14.
Endothelial cells (ECs) form the innermost layer of blood vessels and are, therefore, exposed to shear stress (SS), a frictional force exerted by the viscous flow of the blood. Interestingly, ECs residing at the parts of the vasculature, which are exposed to high levels of uniform, laminar SS, are kept in a homeostatic and quiescent state. In contrast, ECs that experience low, non-uniform SS, e.g., at bifurcations or the lesser curvature of the aortic arch, are proliferative and activate inflammatory pathways15. In turn, sites of dysfunctional ECs are prone to develop atherosclerosis. Interestingly, ECs in these atheroprone areas display aberrantly high levels of activated SMAD2/3 and SMAD1/516,17,18. In this context, enhanced TGFβ/BMP signaling was found to be an early event in the development of atherosclerotic lesions19 and interference with BMP signaling was found to markedly reduce vascular inflammation, atheroma formation, and associated calcification20.
Proximity Ligation Assay (PLA) is a biochemical technique to study protein-protein interactions in situ21,22. It relies on the specificity of antibodies of different species that can bind target proteins of interest, allowing highly specific detection of endogenous protein interactions at a single-cell level. Here, primary antibodies have to bind to their target epitope at a distance of less than 40 nm to allow for the detection23. Therefore, PLA is greatly beneficial over traditional co-immunoprecipitation approaches, where several million cells are needed to detect endogenous protein interactions. In PLA, species-specific secondary antibodies, covalently linked to DNA fragments (termed Plus and Minus probes), bind the primary antibodies and if the proteins of interest interact, Plus and Minus probes come in close proximity. The DNA gets ligated in the following step and the rolling circle amplification of the circular DNA is made possible. During amplification, fluorescently labeled complementary oligonucleotides bind to the synthesized DNA, allowing these protein interactions to be visualized by conventional fluorescence microscopy.
The protocol described here enables scientists to quantitatively compare the number of active SMAD transcription complexes at atheroprotective and atheroprone SS conditions in vitro using PLA. SS is generated via a programmable pneumatic pump system that is able to generate laminar unidirectional flow of defined levels and allows stepwise increases of flow rates. This method allows for the detection of interactions between SMAD1/5 or SMAD2/3 with SMAD4, as well as mixed-R-SMAD complexes. It can easily be expanded to analyze interactions of SMADs with transcriptional co-regulators or to transcription factor complexes other than SMADs. Figure 1 shows the major steps of the protocol presented below.
Figure 1: Schematic representation of the protocol described. (A) Cells seeded in 6-channel slides are exposed to shear stress with a pneumatic pump system. (B) Fixed cells are used for PLA experiment or for control conditions. (C) Images of PLA experiments are acquired with a fluorescence microscope and are analyzed using ImageJ analysis software. Please click here to view a larger version of this figure.
1. Cell culture and fluid shear stress exposure
NOTE: Human umbilical vein ECs (HUVECs) were used as an example to study SS induced interaction of SMADs. The protocol described below can be applied to every SS responsive cell type.
2. Fixation
3. Blocking and primary antibody incubation
4. PLA probe incubation
NOTE: For all steps in section 4.1-7.3, the washing buffers A and B are stored at 4 °C and need to be warmed to RT prior to the use.
5. Ligation
6. Amplification
7. Mounting
8. Image acquisition
9. Image analysis and quantification using ImageJ/FIJI
We have previously used PLA to detect interactions of different SMAD proteins3 and analyzed shear stress induced changes in SMAD phosphorylation28.
Here, both methods were combined with the protocol described above. HUVECs were subjected to shear stress of 1 dyn/cm2 and 30 dyn/cm2 and analyzed for interactions of SMAD transcription factors. We show that, when compared to the high shear stress (30 dyn/cm2), the low shear stress (1 dyn/cm2) leads to a significant increase in SMAD1-SMAD2/3 interactions, the so called mixed-SMAD complexes in both, the cytosol, and nuclei of analyzed cells (Figure 2A, lower panel). PLA events are visible as distinct spots in both samples, and one can distinguish between cytosolic and nuclear events with reference to DAPI staining (Figure 2A, upper panel). In contrast, antibody controls, where only one of both primary antibodies but still both PLA probes were incubated, showed negligible numbers of PLA signals (Figure 2B). Thus, it can be concluded that the experiment was successful.
Figure 2: SMAD2/3-SMAD1 PLA comparing different antibody concentrations. (A) Confocal fluorescence images and quantification of SMAD2/3-SMAD1-PLA in HUVECs exposed to 24 h of indicated SS levels. Antibodies dilution ratio: 1:100. (B) Confocal images of single antibody controls for PLA in A. (C) Confocal fluorescence images and quantification of SMAD2/3-SMAD1-PLA in HUVECs exposed to 24 h of indicated SS levels. Antibodies dilution ratio: 1:50. Scale bars for A–C: 20 µm and 10 µm in zoom in. Dyne equals dyn/cm2. Antibodies used were rabbit anti-SMAD2/3 and mouse anti-SMAD1 (see Table of Materials). For each condition 5 random areas along the center of the flow channel were used for image acquisition. N=1 biological replicate. Please click here to view a larger version of this figure.
To show how different concentrations of antibodies change PLA results, the same experiment was performed with a 1:50 instead of a 1:100 dilution of antibodies. Under these conditions, twice the amount of antibody result in more than four-fold higher PLA signals per cell (compare Figure 2A and Figure 2C). The difference in signals between 1 dyn/cm2 and 30 dyn/cm2 decreases when a higher antibody concentration is used and statistical significance is lost for the total and cytosolic PLA events (Figure 2A, lower panel; Figure 2C, lower panel). This might be due to the signal coalescence and problems of distinguishing PLA events. If such accumulation of signals occurs, antibody concentrations should be decreased.
Furthermore, we showed that self-made buffers for blocking and antibody dilution can be used as an alternative for commercial buffers included in in situ PLA kits. PLA events per cell were compared for SMAD1-SMAD2/3 (Figure 3A versus Figure 3D), SMAD2/3-SMAD4 (Figure 3B versus Figure 3E) and pSMAD1/5-SMAD4 (Figure 3C versus Figure 3F) complexes under 1 dyn/cm2 and 30 dyn/cm2 using either commercial solutions (Figure 3A-C) or self-made BSA/PBS based solutions (Figure 3D-F). Quantifications for both commercial and self-made diluent/ blocking solutions show the same trend of PLA signals in cytosolic and nuclear areas. However, the total number of PLA events per cell is higher when using commercial solutions (Figure 3A-C, lower panels versus Figure 3D-F, lower panels).
Figure 3: PLA experiment comparing commercial and self-made antibody buffers. Confocal images (upper part of each panel) and quantification (lower part of each panel) of different SMAD-SMAD PLAs in HUVECs exposed to 24 hours of indicated shear stress levels. (A-C) Commercial buffers (see Table of Materials). (D-F) Self-made buffers (3% BSA in PBS for blocking, 1 % BSA in PBS for antibody dilution). All primary antibodies were diluted 1:100. Scale bar, 20 µM (10 µM for zoom in). Statistical significance was calculated with two-sided t-Test. ns – non-significant, *P<0.05, **P<0.01, ***P<0.001. Values are depicted as mean + SEM. Dyne equals dyn/cm2. Antibodies used were rabbit anti-SMAD2/3, mouse anti-SMAD1, rabbit anti-phospho SMAD1/5 and mouse anti-SMAD4 (see Table of Materials). For each condition 5 random areas along the center of the flow channel were used for image acquisition. N=1 biological replicate. Please click here to view a larger version of this figure.
We also included an example for a failed PLA experiment. Here, a combination of SMAD2/3-SMAD4 antibodies was used where the SMAD4 antibody was not suited for performing immunofluorescence experiments. When compared to single antibody controls, no increase in spots in either the 1 dyn/cm2 or 30 dyn/cm2 samples was observed (Figure 4A versus Figure 4B). As the formation of SMAD2/3-SMAD4 complexes is induced by shear stress (see Figure 3B,E), it can be concluded that this PLA experiment was unsuccessful. This highlights the importance of choosing the correct antibody combinations to detect PLA events, as orientation and distance of bound antibodies might be crucial for successful annealing of the oligonucleotide probes.
Figure 4: Example for failed PLA experiment. Confocal images, scale bar: 20 µm and 10 µm in zoom in. (A) SMAD2/3-SMAD4 PLA. (B) Single antibody controls. Antibodies used were mouse anti-SMAD2/3 and rabbit anti-SMAD4 (see Table of Materials). Dyne equals dyn/cm2. For each condition 5 random areas along the center of the flow channel were used for image acquisition. N=1 biological replicate. Please click here to view a larger version of this figure.
The PLA based protocol described here offers an efficient way to determine close proximity of two proteins (e.g., their direct interaction) in ECs exposed to shear stress with quantitative and spatial resolution. By using flow slides with multiple parallel channels, several different protein interactions can be examined at the same time in cells under identical mechanical conditions. In contrast, custom-build flow chamber systems often make use of a single channel that is built around a glass coverslip, which would allow only a single PLA experiment without the necessary controls per slide and pump. Although this protocol focuses on the detection of SMAD interactions, it can be adapted to detect any other protein interactions. However, analysis of results must be done carefully as two proteins may also reside in close proximity without forming complexes. If definite statements on interactions of proteins are desired, PLA experiments should be complemented with additional methods such as co-immunoprecipitations. Additionally, PLA cannot be used to detect the protein-protein interaction in live cells since samples need to be fixed for subsequent antibody binding and DNA amplification steps.
For successful detection of protein interactions by PLA, the most critical step is to choose a combination of primary antibodies that fulfil several criteria: (1) The antibodies detecting the individual protein-partners must be generated in different species (e.g., mouse or rabbit) as the secondary antibodies are species-specific; ideally, antibodies were already successfully tested by conventional immunofluorescence microscopy; (2) the distance between epitope bound antibodies should be <40 nm23; (3) as the affinities for antibody-epitope binding might differ, the final concentration of used antibodies has to be adjusted for each experimental setting, as shown here (Figure 2A, lower panel versus Figure 2C, lower panel). Therefore, if very few but specific PLA events are detected, it may be worth increasing the amount of antibody used. However, this must be carefully titrated to avoid oversaturation and unspecific binding events. Also, antibody concentration used in control samples must match the concentration used in the PLA samples.
As for any other biochemical assay, suitable controls are indispensable in PLA. Unspecific binding of the used antibodies might, for example, lead to PLA signals originating from just one primary antibody. Therefore, essential antibody controls for PLA experiments should include addition of only one of the two primary antibodies but both PLA Probes (Plus and Minus). Furthermore, controls with no antibodies added can be used to determine unspecific binding of the PLA probes to the sample. In general, those technical controls should yield no to a very few PLA signals. If several signals are observed, concentration of the primary antibody used, and its specificity should be reconsidered. In addition, it is useful to include a positive biological control, if possible. In the protocol described above, this could be stimulation with a BMP ligand that is known to induce phosphorylation of SMADs and, therefore, trimeric complex formation with SMAD4.
For PLA experiments, it is normally recommended to use cells that are 50-70% confluent, since this simplifies the penetration of reagents. However, when performing experiments with ECs, we would strictly argue against this, except if semi-confluency is part of the experimental set-up. In vivo ECs form monolayers with tight cell-to-cell junctions, which are essential for EC homeostasis and mechano-transduction29. Therefore, experiments on non-confluent ECs could give rise to false results. Furthermore, non-confluent cells are more prone to detach from flow channel slides if high levels of shear stress are used during the experiment. We advise to seed cells (2-2.5 x 106 cells/ mL, see protocol step 1.2) two to three days in advance of experimental start as EC monolayers need time to form and develop mature junctions. Therefore, we would not recommend seeding a higher number of cells (>2.5 x 106 cells/mL) in flow channels just one day before the experiment to achieve a confluent monolayer.
Although various liquid mounting media containing DAPI exist it is worth to include a separate DAPI staining step and mount the cells with liquid mounting medium without DAPI, at least if polymer-based flow slides are used. This prevents heavy background signals during image acquisition. Images should be acquired from positions in the channel center rather than the edges since shear stress is inhomogeneous at the channel walls. We recommend to take at least 5-10 images per biological replicate and condition of random areas along the center region. 3 or more biological replicates are normally used to gain statistical relevance. For image analysis we advise to use the ImageJ/FIJI macro function. In the protocol above, we mentioned an ImageJ macro that is suitable to count cytosolic and nuclear PLA events based on DAPI staining. Users should be aware that parameters like particle size need to be adjusted depending on nuclear size or bigger/smaller PLA dots. The macro saves the threshold PLA images and masks that should be compared to raw images to evaluate specificity of PLA signal detection.
In conclusion, the method presented here allows rapid and efficient spatial and quantitative analysis of transcription factor complex formation at atheroprotective and atheroprone SS conditions. It will allow scientists to further decipher the impact of SMAD complex formation in atherogenesis and vascular disease in general. It can, furthermore, be adapted to investigate the consequences of proximity of different proteins in those vascular pathologies.
The authors have nothing to disclose.
We thank Dr. Maria Reichenbach and Dr. Christian Hiepen for their support on the flow-set up system and Eleanor Fox and Yunyun Xiao for critically reading the manuscript. P-L.M. was funded by the international Max Planck Research School IMPRS-Biology and Computation (IMPRS-BAC). PK received funding by the DFG-SFB1444. Figure 1 was created using BioRender.
µ-Slide VI 0.4 | ibidi | 80606 | 6-channel slide |
Ammonium Chloride | Carl Roth | K298.1 | Quenching |
Bovine Serum Albumin | Carl Roth | 8076.4 | Blocking |
DAPI | Sigma Aldrich/ Merck | D9542 | Stain DNA/Nuclei |
DPBS | PAN Biotech | P04-53500 | PBS |
Duolink In Situ Detection Reagents Green | Sigma Aldrich/ Merck | DUO92014 | PLA kit containing Ligase, ligation buffer, polymerase and amplification buffer (with green labeled oligonucleotides) |
Duolink In Situ PLA Probe Anti-Mouse MINUS | Sigma Aldrich/ Merck | DUO92004 | MINUS probe |
Duolink In Situ PLA Probe Anti-Rabbit PLUS | Sigma Aldrich/ Merck | DUO92002 | PLUS probe |
Duolink In Situ Wash Buffers, Fluorescence | Sigma Aldrich/ Merck | DUO82049 | PLA wash buffers A and B |
Endothelial Cell Growth Supplement | Corning | supplement for medium (ECGS) | |
Fetal calf Serum | supplement for medium | ||
FIJI | Image Analysis software | ||
Formaldehyde solution 4% buffered | KLINIPATH/VWR | VWRK4186.BO1 | PFA |
Full medium | M199 basal medium +20 % FCS +1 % P/S + 2 nM L-Glu + 25 µg/mL Hep + 50 µg/mL ECGS | ||
Gelatin from porcine skin, Type A | Sigma Aldrich | G2500 | Use 0.1% in PBS for coating of flow channels |
GraphPad Prism v.7 | GarphPad | Statistical Program used for the Plots and statistical calculations | |
Heparin sodium salt from porcine intestinal mucosa | Sigma Aldrich | H4784-250MG | supplement for medium (Hep) |
HUVECs | |||
ibidi Mounting Medium | ibidi | 50001 | Liquid mounting medium |
ibidi Pump System | ibidi | 10902 | pneumatic pump |
Leica TCS SP8 | Leica | confocal microscope | |
L-Glutamin 200mM | PAN Biotech | P04-80100 | supplement for medium (L-Glu) |
Medium 199 | Sigma Aldrich | M2154 | Base medium |
mouse anti- SMAD1 Antibody | Abcam | ab53745 | Suited for PLA |
mouse anti- SMAD2/3 Antibody | BD Bioscience | 610843 | Not suited for PLA in combination with CST 9515 |
mousee anti- SMAD4 Antibody | Sanata Cruz Biotechnology | sc-7966 | Suited for PLA |
Penicillin 10.000U/ml /Streptomycin 10mg/ml | PAN Biotech | P06-07100 | supplement for medium (P/S) |
Perfusion Set WHITE | ibidi | 10963 | Tubings used for 1 dyn/cm2 |
Perfusion Set YELLOW and GREEN | ibidi | 10964 | Tubings used for 30 dyn/cm2 |
rabbit anti- phospho SMAD1/5 Antibody | Cell Signaling Technologies | 9516 | Suited for PLA |
rabbit anti- SMAD2/3 XP Antibody | Cell Signaling Technologies | 8685 | Suited for PLA |
rabbit anti- SMAD4 Antibody | Cell Signaling Technologies | 9515 | Not suited for PLA in combination with BD 610843 |
Serial Connector for µ-Slides | ibidi | 10830 | serial connection tubes |
Triton X-100 | Carl Roth | 6683.1 | Permeabilization |