Skip to main content
Dryad

Piezo1 ion channels are capable of conformational signaling

Cite this dataset

Grandl, Jorg; Lewis, Amanda; Cronin, Marie (2024). Piezo1 ion channels are capable of conformational signaling [Dataset]. Dryad. https://doi.org/10.5061/dryad.c59zw3rfd

Abstract

Piezo1 is a mechanically activated ion channel that senses forces with short latency and high sensitivity. Piezos undergo large conformational changes, induce far-reaching deformation onto the membrane, and modulate the function of two-pore potassium (K2P) channels. Taken together, this led us to hypothesize that Piezos may be able to signal their conformational state to other nearby proteins. Here, we use chemical control to acutely restrict Piezo1 conformational flexibility and show that Piezo1 conformational changes, but not ion permeation through it, are required for modulating the K2P channel TREK1. Super-resolution imaging and stochastic simulations further reveal that both channels do not co-localize, which implies that modulation is not mediated through direct binding interactions; however, at high Piezo1 densities, most TREK1 channels are within the predicted Piezo1 membrane footprint, suggesting the footprint may underlie conformational signaling. We speculate that physiological roles originally attributed to Piezo1 ionotropic function could, alternatively, involve conformational signaling.

README: Piezo1 ion channels are capable of long-range conformational signaling

https://doi.org/10.5061/dryad.c59zw3rfd

Description of the data and file structure

There are two files: One contains data for Neuro2A-Piezo1Myc cells overexpressing TREK1-HA ("Endogenous"); the other contains data for Neuro2a-Piezo1ko cells overexpressing both Piezo1-Myc and TREK1-HA ("Overexpression"). Within each folder are labelled cells ("Cell B", "Cell C", etc.) that correspond to Table 1 (Endogenous) and Table 2 (Overexpression) in the manuscript.

For the endogenous dataset, the PDF for each cell contains a STED image (scale bar: 1 micron) and associated cumulative frequency distribution, as in Figure 5A and 5B of the main manuscript.

For the overexpression dataset, the PDF for each cell contains a STED image and corresponding plot showing NNDT1-P1 as a function of local Piezo density, as in Figure 7A and 7C of the main manuscript.

Code/Software

Python code used to generate random simulations and analyze nearest-neighbor distances is available on GitHub (2_Channel_Spatial_Analysis) https://github.com/GrandlLab?tab=repositories

Methods

Microscopy sample preparation

Cells were transiently transfected with TREK1HA and/or Piezo1Myc plasmids and plated on No 1.5 coverslips (Warner Instruments: CS-12R15, Catalog # 64-0712) 48 hours before staining, fixed in 2% formaldehyde, blocked with 10% normal goat serum, and stained with 1:100 chicken anti-Myc (Novus) primary antibody followed by 1:500 Alexa Fluor plus 594 goat anti-chicken (Thermo Fisher Scientific) secondary antibody to label Piezo1Myc channels and 1:500 rabbit anti-HA (Cell Signaling Technology) primary antibody followed by 1:200 Atto 647N goat anti-rabbit (Rockland Instruments) secondary antibody to label TREK1HA channels. Coverslips were then mounted with ProLong Glass Antifade Mountant (ThermoFisher Scientific) and cured at room temperature overnight before image acquisition. STED images were collected from at least three coverslips per condition, with n=3 independent transfections to generate biological replicates.

STED image acquisition and deconvolution

Two-color STED was performed on co-labeled images of either Neuro2A-Piezo1ko cells overexpressing Piezo1Myc or Neuro2A-Piezo1Myc cells, both transiently transfected with TREK1HA. All image collection was performed on a Leica SP8 instrument equipped with a 100x/1.4 HCX PL APO OIL WD 90 mm objective, pulsed White Light Laser, and HyD detectors, using Leica Application Suite Software (3.5). The 594 red channel, corresponding to Piezo1Myc, was excited using 5% laser power at 591 nm, and emitted light was collected between 603 and 641 nm with 22% gain. The 647 far red channel, corresponding to TREK1HA, was excited using 3% laser power at 641 nm and emitted light was collected between 651 and 779 nm with 42% gain. The 647 channel was collected using 2x frame averaging, to reduce noise. A pulsed 775 nm STED depletion laser at 20% laser power in the 647 channel, with gating (0.7 to 4.2), and 70% laser power in 594 channel, with gating (0.7 to 4.2), was used to improve image resolution to ~80 nm (Supplemental Table 1-2). For STED image collection, Z-stacks were acquired from the middle to the top of the cell in steps of 220 nm with a Märzhäuser linearly encoded piezo Z stage (Supplemental Movie 1).

All channels from the STED images were deconvolved using Huygen’s Professional (Scientific Volume Imaging). The refractive index was corrected to match the immersion oil (1.5) and images were cropped as necessary to isolate single cells. Deconvolution was then performed with an automatically generated theoretical point spread function and the preset Classic Maximum Likelihood Estimation CMLE deconvolution algorithm, with the signal-to-noise ratio set to 5.0.

Confocal microscopy image acquisition

Confocal imaging was used to assess antibody specificity (Figure S5F-I, Figure S6B-C) and quantify membrane expression levels of all Myc-tagged Piezo constructs (Figure S8D-E). Antibody specificity was assessed using widefield confocal images at 100X magnification acquired with the same objective, excitation and emission collection parameters as STED images, as described above, but without depletion. For membrane expression quantification, images were acquired using an HC PL APO CS2 40x/1.30 Oil objective on the above described Leica SP8 scope. The 594 red channel, corresponding to Piezo1Myc, was collected using 5% laser power at 591 nm, and emitted light was collected between 603 and 641 nm with 13% gain. The 488 green channel, corresponding to GFP, was collected using 7% laser power at 490 nm, and emitted light was collected between 501 and 561 nm with 52% gain. Single Z-slices of the central plane, the midpoint of most cells in a field view, were collected over ~12 images, yielding approximately 200 cells per condition.

Image processing

Images were processed in FIJI (version 2)66 as 16-bit TIFF files. A single z-slice for spatial analysis from the cell surface (top) was manually chosen from the Z-stack. Noise thresholds were set for each cell by identifying the intersection of Gaussian distributions fit to the intensity histograms of segmented puncta from the surface z-slice and a separate z-slice chosen from the unlabeled interior of the cell.  Manually drawn ROIs were generated around the xy perimeter of each cell, and signal outside the ROI was cleared. Both 594 (Piezo1Myc) and 647 (TREK1HA) channels were auto-enhanced and filtered with a 2.0 pixel Gaussian Blur Filter. Individual puncta from both overexpressed TREK1 and Piezo1 channel conditions  were segmented using StarDist’s Versatile (fluorescent nuclei) Model with the following settings (Normalized Image Percentile 3-100, Probability/Score Threshold: 0.5, Overlap Threshold: 0.2, Number of Tiles: 1, Boundary Exclusion: 2)67. Automated segmentation parameters were developed based on manual segmentation.  Endogenous Piezo1 puncta from Neuro2a-Piezo1Myc images were manually segmented. We note that the size and intensity of individual segmented puncta in images varies, particularly for TREK1. This variance primarily stems from the variable z position along the axial point spread function of a punctum, especially given that our image acquisition settings were optimized exclusively to improve lateral resolution in x and y. Additional variance may come from two or more channels located closer to each other than the resolution limit of our experiments. The center of mass for each Piezo1 or TREK1 punctum was identified using the ‘Measure’ function in FIJI. Puncta with mean intensity values under the identified threshold were excluded, and corresponding XY coordinates were exported for analysis. Resolution for each cell was quantified from the Full-Width Half Max (FWHM) across n=28-70 puncta from 2-3 2 µm x 2 µm regions in each image. Lineplots were used to measure the intensity profile across individual points, and gaussian curves were fit to these data [see github for Jupyter notebook] using the equation: .

The FWHM was calculated from Gaussian fits to line intensity profiles of each punctum using the equation FWHM = ℴ*2.355 and averaged across the puncta for each image.

Spatial distribution analysis and modeling

Nearest Neighbor Distances (NNDs) between empirical TREK1 and Piezo1 coordinates were calculated using the KDtree function from the SciPy package in Python368. To compare the empirical spatial relationship of TREK1 to Piezo1 puncta, we retained the empirical XY locations of Piezo1 puncta and simulated random populations of TREK1 puncta. Each simulation was performed 1,000 times per image. For random TREK1 distributions, for each image, TREK1 positions were simulated from a random distribution within the image ROI at a density equivalent to the empirical TREK1 density for that respective cell. Local Piezo1 densities were calculated for a 564 nm radius (=1 mm2 area) around each TREK1 punctum using the Ball Query function in SciPy.

Our estimate of 30 nm as the maximum cutoff for TREK1 and Piezo1 binding is based on the central location of the Piezo1 epitope, the predicted radius for Piezo1 in a biological membrane (14 nm)19,21,34, the location of the TREK1 epitope and radius of the TREK1 channel (~5 nm69), and some accounting for linkage error from our labeling strategy. Our estimate of 100 nm as the maximum cutoff for the Piezo1 footprint comes from estimates of the Piezo1 diameter (10-30 nm)20,23,25,26,34 plus 5-fold the predicted decay length of the Piezo1 footprint in a biological membrane (14 nm)24.

Funding

National Institutes of Health, Award: 5R01NS110552