Pressure‐dependent regulation of Ca2+ signalling in the vascular endothelium

Key points Increased pressure suppresses endothelial control of vascular tone but it remains uncertain (1) how pressure is sensed by the endothelium and (2) how the vascular response is inhibited. This study used a novel imaging method to study large numbers of endothelial cells in arteries that were in a physiological configuration and held at normal blood pressures. Increased pressure suppressed endothelial IP3‐mediated Ca2+ signals. Pressure modulated endothelial cell shape. The changes in cell shape may alter endothelial Ca2+ signals by modulating the diffusive environment for Ca2+ near IP3 receptors. Endothelial pressure‐dependent mechanosensing may occur without a requirement for a conventional molecular mechanoreceptor. Abstract The endothelium is an interconnected network upon which haemodynamic mechanical forces act to control vascular tone and remodelling in disease. Ca2+ signalling is central to the endothelium's mechanotransduction and networked activity. However, challenges in imaging Ca2+ in large numbers of endothelial cells under conditions that preserve the intact physical configuration of pressurized arteries have limited progress in understanding how pressure‐dependent mechanical forces alter networked Ca2+ signalling. We developed a miniature wide‐field, gradient‐index (GRIN) optical probe designed to fit inside an intact pressurized artery that permitted Ca2+ signals to be imaged with subcellular resolution in a large number (∼200) of naturally connected endothelial cells at various pressures. Chemical (acetylcholine) activation triggered spatiotemporally complex, propagating inositol trisphosphate (IP3)‐mediated Ca2+ waves that originated in clusters of cells and progressed from there across the endothelium. Mechanical stimulation of the artery, by increased intraluminal pressure, flattened the endothelial cells and suppressed IP3‐mediated Ca2+ signals in all activated cells. By computationally modelling Ca2+ release, endothelial shape changes were shown to alter the geometry of the Ca2+ diffusive environment near IP3 receptor microdomains to limit IP3‐mediated Ca2+ signals as pressure increased. Changes in cell shape produce a geometric microdomain regulation of IP3‐mediated Ca2+ signalling to explain macroscopic pressure‐dependent, endothelial mechanosensing without the need for a conventional mechanoreceptor. The suppression of IP3‐mediated Ca2+ signalling may explain the decrease in endothelial activity as pressure increases. GRIN imaging provides a convenient method that gives access to hundreds of endothelial cells in intact arteries in physiological configuration.


Introduction
The vascular endothelium is a one-cell-thick layer that directs the formation of new blood vessels (angiogenesis), prevents blood clotting, regulates vascular permeability, controls arterial tone and determines the extent of smooth muscle proliferation. The endothelium's control of each of these functions arises from the cells acting as a sensitive signal processing centre that detects and interprets multiple simultaneous messages such as those derived from mechanical stimuli (hydrostatic pressure, luminal shear stress, circumferential strain) and local and blood borne signals (autocrine, paracrine and electrical signals and neurotransmitters). Endothelial stimuli are transduced to changes in the endothelial Ca 2+ concentration to coordinate the endothelium's control of vascular activity (Behringer & Segal, 2012;Sonkusare et al. 2012;Billaud et al. 2014). Ca 2+ regulates the synthesis and release of various vasoactive agents such as nitric oxide, prostacyclin and endothelium-derived hyperpolarizing factor. Through these Ca 2+ -dependent mediators the endothelium's control of vascular contraction, permeability, cell proliferation and angiogenesis is achieved. Therefore, central to an understanding of endothelial signal processing is an appreciation of the control of Ca 2+ .
Two main sources of endothelial Ca 2+ are recognized, the extracellular fluid and the intracellular stores of the endoplasmic reticulum (Moccia et al. 2012). Ca 2+ entry from the extracellular fluid may occur via a large number of ion channels on the outside (plasma) membrane. The other main cytoplasmic Ca 2+ concentration ([Ca 2+ ] c ) source is the internal endoplasmic reticulum store from which release proceeds mainly via the inositol 1,4,5-trisphosphate (IP 3 ) receptor (IP 3 R). While Ca 2+ release via IP 3 R is well established the contribution of the ryanodine receptor (RyR) to the control of endothelial Ca 2+ (if any) is less clear (Socha et al. 2012a). The sources of Ca 2+ are not independent: Ca 2+ influx regulates Ca 2+ release and Ca 2+ release regulates Ca 2+ influx. For example, Ca 2+ release from the endoplasmic reticulum may alter the activity of ion channels present on the plasma membrane to regulate Ca 2+ entry either via Ca 2+ -gated ion channels (Strotmann et al. 2003) or via membrane potential changes altering passive fluxes of the ion (Behringer & Segal, 2015). Alternatively Ca 2+ influx may alter endoplasmic reticulum Ca 2+ content or activity of IP 3 R in a Ca 2+ -induced Ca 2+ release-like process (Earley & Brayden, 2015). Thus the local change in [Ca 2+ ] c arising from the activity of channels in the plasma membrane or endoplasmic reticulum itself regulates the activity of ion channels to provide a feedback control of Ca 2+ signals and modulate vascular function.
The response to mechanical stimuli involves changes in [Ca 2+ ] c and must be integrated with signals from other sensors to converge on a physiological response. Two major mechanical stimuli are shear stress and pressure (Falcone et al. 1993;Huang et al. 1998;Popp et al. 1998;Muller et al. 1999;Marchenko & Sage, 2000;Paniagua et al. 2000;Sun et al. 2001;Duza & Sarelius, 2004). The endothelial response to shear stress is well characterized and several types of activity may occur. Vascular smooth muscle relaxation may be evoked, cell migration induced and endothelial gene expression changed (Falcone et al. 1993;Muller et al. 1999;Shiu et al. 2004;Chien, 2007). The increases in endothelial Ca 2+ changes underlying these responses may involve several types of mechanically sensitive ion channels forces such as Piezo1 Ranade et al. 2014), ENaC (Kusche-Vihrog et al. 2014), ATP-gated P2X4 (Yamamoto et al. 2006), TREK-1 (Dedman et al. 2009) and various members of the TRP grouping of channels (Corey et al. 2004;Maroto et al. 2005;Spassova et al. 2006;Janssen et al. 2011). Alternatively, shear stress-evoked activation of ion channels may be indirect and force sensed by the cytoskeleton, apical glycocalyx, mechanosensitive or membrane curvature-sensitive protein complexes or G-protein-coupled receptors (Knudsen & Frangos, 1997;Thi et al. 2004;Tzima et al. 2005;Zimmerberg & Kozlov, 2006;Mederos y Schnitzler et al. 2008;Zhao et al. 2011). The diversity of responses highlights the complexity of the response to shear stress.
The response of the endothelium to pressure differs significantly from that of shear stress. Rather than being activated, a decrease in activity may occur as mechanical stimulation (pressure) is increased (Hishikawa et al. 1992;Gunduz et al. 2008). The decrease in endothelial activity may suppress smooth muscle relaxation (De Bruyn et al. 1994;Huang et al. 1998;Paniagua et al. 2000;Zhao et al. 2015). In healthy human volunteers, short-term increases in arterial blood pressure cause long-lasting inhibition of endothelium-dependent dilatation (Jurva et al. 2006;Phillips et al. 2011). In isolated arteries, acute exposure to increases in transmural pressure may impair by the endothelium, specifically when the arteries are in their natural 'tubular' configuration.
The biological response of arteries to pressure depends critically on the complex, cylindrical, three-dimensional arrangement of cells and on the interactions with other cell types, and may change with artery configuration. For example, in many intact arteries in their cylindrical configuration, myogenic contraction occurs when the artery is subjected to circumferential stretch by increases in pressure. However, this contraction does not occur in arteries stretched, with equivalent forces, on wires (Dunn et al. 1994). Increased mechanical stimulation by stretch, hypotonic cell swelling or shear stress activates TRPV4 channels (Strotmann et al. 2000;Alessandri-Haber et al. 2003;Loukin et al. 2010), a mechanism that may contribute to flow-induced dilatation (Kohler et al. 2006;Mendoza et al. 2010;Bubolz et al. 2012). However, TRPV4 channels are also deactivated by increased stretch when transmural pressure difference is increased in pressurized arteries (Bagher et al. 2012) in a cylindrical configuration. The cell shape and operative mechanical forces in a pressurized artery are quite different from experiments on isolated or cultured cells.
The carotid artery contributes to the control of cerebrovascular blood flow and cerebral vascular resistance (Mchedlishvili, 1986;Faraci & Heistad, 1990). The endothelium regulates carotid artery tone. Vascular relaxation is evoked by several endothelial activators including flow (shear stress), adrenonedullin and acetylcholine (Faraci et al. 1994;Plane et al. 1998;Chataigneau et al. 1998a,b;Ohashi et al. 2005) and the endothelium may also attenuate the vascular contractile response to vasoconstrictors (Lamping & Faraci, 2001). Many of the most serious forms of cardiovascular diseases (e.g. atherosclerosis) reside in larger arteries like the carotid artery and begin with endothelial dysfunction (Deanfield et al. 2007). However, studying the effects of mechanical forces like pressure on endothelial function (and dysfunction) in larger arteries in a physiological configuration has been exceptionally difficult. Assessment of endothelial function in large arteries has been largely indirect. The majority of papers in the past decade involved only the measurement of endothelium-dependent dilatation (e.g. Craig & Martin, 2012).
To study the function of endothelial cells in intact arteries, some investigations have used either wide-field or point-scanning fluorescence microscopes to visualize the endothelium through the wall of the artery (Bagher et al. 2012;Sonkusare et al. 2012;Tran et al. 2012). However, movement of the artery in pressure myograph systems is almost unavoidable and results in the vessel moving in and out of the focal plane, changing light levels and altering image quality to present a significant challenge to data analysis. Light scattering by the artery wall also reduces contrast and the curvature of the artery limits the number of cells that may be visualized within a single optical plane. To overcome each of these difficulties and allow the endothelium to be studied in intact arteries, we have developed a miniature optical probe to record Ca 2+ signalling from inside pressurized arteries. The probe has a field of view of 0.5 mm diameter, which allowed a large number (ß200) of naturally connected endothelial cells to be imaged with subcellular resolution and has a high depth of field (141 μm) sufficient to maintain good focus across the highly curved, intact endothelial layer of a large artery. To handle the data from such a large number of cells a largely automated image processing routine was also developed. We show in native endothelial cells in their physiological configuration, that acetylcholine-evoked Ca 2+ rises originate as IP 3 -mediated Ca 2+ signals in particular regions of the endothelium from which they progress to other cells as Ca 2+ waves. Detection of pressure-dependent mechanical forces by the endothelium is integrated effortlessly into the same signalling pathway by geometric modulation of IP 3 -evoked Ca 2+ release brought about by changes in endothelial cell shape. We suggest the suppression of IP 3 -mediated Ca 2+ signals may underlie the inhibition of endothelial responses with increased pressure and, significantly, may not require a conventional mechanosensor for mechanotransduction to occur.

Ethical approval
All experiments employed tissue obtained from male Sprague-Dawley rats (10-12 weeks old; 250-350 g). Rats were humanely killed by overdose of pentobarbital sodium with the approval of the University of Strathclyde Local Ethical Review Panel (200 mg kg −1 ; Schedule 1 procedure; Animals (Scientific Procedures) Act 1986, UK), under UK Home Office Project and Personal Licence authority.

Tissue preparation
The left and right common carotid arteries were exposed by blunt dissection. To prevent collapse of the artery upon removal, the rostral and caudal ends of the exposed carotid arteries were ligated with 8-0 suture and the arteries were then rapidly excised. Arteries were then cleaned of connective tissue under a dissection microscope, and visually checked for the presence of side branches. Subsequently, arteries without side branches were mounted onto two blunted and deburred 22-gauge cannula in a custom-designed imaging bath using two lengths of suture thread. Blood was removed from the arteries by flushing the lumen with physiological saline solution (PSS) for 10 min (150 μl min −1 ) before the arteries were pressurized to 60 mmHg, gently heated to 37°C J Physiol 593.24 and allowed to equilibrate at 37 o C for 30 min. Flow was provided by a peristaltic pump (PS-200, Living Systems Instrumentation, At Albans City, VT, USA) connected to the proximal cannula. The endothelium of mounted arteries was then selectively loaded with a fluorescent Ca 2+ indicator by intraluminal perfusion of PSS containing the membrane-permeant form of Oregon Green BAPTA-1 (20 μM, OGB-1/AM) and 0.04% Pluronic acid. Flow was stopped for 30 min to permit sufficient loading. Arteries were continuously superfused with PSS during this time. Following removal of excess dye, the distal cannula was removed from the artery and the artery was mounted on a gradient index (GRIN) microprobe (described below) and secured with suture thread. Transmural pressure was then incrementally increased to 160 mmHg, whilst stretching the artery to remove any buckle. Following this procedure, the pressure was decreased to 60 mmHg and the artery left for a further 30 min to equilibrate. During equilibration arteries were tested for leaks (which may indicate side branches, tears in the vessel wall or insufficiently tied sutures) by switching off the feedback on the pressure servo. Arteries that showed signs of leakage, identified as a reduction in pressure whilst isolated from the servo system, which could not be stopped by re-tying the sutures were discarded at this point.

Micro-endoscope GRIN imaging probe
The GRIN microprobe was developed to fit inside pressurized arteries (Fig. 1A) and consisted of a 0.5 mm diameter, 30.2 mm long, single pitch GRIN relay lens (SRL-050; Nippon Sheet Glass, USA) with a 0.5 mm × 0.5 mm × 0.5 mm aluminium-coated micro-prism (66-771; Edmund Optics, USA) attached to the distal surface with ultraviolet curing optical epoxy (NAO 68; Norland Products, USA). The lens was sheathed in a surgical stainless steel tube (0.71 mm outer diameter) for mechanical protection. In this single lens configuration, the GRIN rod acts to reconjugate the image plane of detection optics through the length of the cylinder (Fig. 1D). By focusing a conventional microscope objective to a sufficient depth inside the GRIN rod, the image plane can be extended beyond the front surface of the prism and into tissue ( Fig. 1C-E). This property renders the probe suitable for the replacement of a cannula in a custom-made pressure myograph, where the vessel must be tied to the probe to maintain pressure (Fig. 1B). The focal plane of the optical system can be varied in response to diameter variations in the vessel (Fig. 1C), without moving the probe, by focusing the microscope objective ( Fig. 1B) further into the GRIN lens (see also Flusberg et al. 2008;Kim et al. 2010).
A 0.1 NA GRIN probe was used to ensure a good depth of focus (141 μm) across the entire endothelial surface of curved artery. However, the use of a low NA GRIN relay lens imposes two opposing constraints on the transmission of light through the system. First, to maximize the delivery of excitation light through the GRIN rod, the NA of the excitation light should not be greater than the NA of the GRIN lens (Fig. 1D). Second, to maximize fluorescence detection, the NA of the collection lens should be higher than that of the GRIN rod. The solution to these opposing constraints is to illuminate the proximal end of the GRIN lens with collimated light, which is re-collimated at the output ( Fig. 1E) (Saunter et al. 2012) evenly illuminating the entire field of view whilst keeping the illumination constant despite probe refocusing.
The assembled GRIN probe was held in place in a custom-designed arteriograph (Fig. 1G) that was mounted on the microscope stage of a standard inverted microscope (Fig. 1H). The fluorescence excitation and delivery system was constructed using 30 mm cage system components (Thorlabs, UK), and was also mounted onto the microscope stage via a three-axis translation stage. Two axes of the translation stage permitted the external optical system to be coupled to the GRIN microprobe, whilst the third enabled the focus of the system to be altered. Mounting both the arteriograph and the external optical system on the stage ensured that movement of the microscope stage did not decouple the probe from the system. Fluorescence excitation was provided by a fibre-coupled diode-pumped solid-state laser operating at 488 nm, which was collimated ( Fig. 1B and H) before being focused by another lens and guided to the back of a 20 × 0.5 NA infinity-corrected microscope objective (Plan Fluor; Nikon, UK) via a dichroic mirror. Optical excitation power density, measured at the output of the GRIN probe was 1 nW μm 2 . The microscope objective and a 65 mm focal length tube lens imaged fluorescence emission, returning through the probe, through an emission filter and onto a sCMOS camera (Zyla 3-Tap; Andor, UK) controlled by μManager (Edelstein et al. 2010), providing an effective pixel size of 1 μm at the object plane and permitting up to 200 endothelial cells to be imaged with subcellular resolution (see below, Optical characterization).

Optical characterization
To demonstrate the fluorescence signal detected by the microendoscope camera, we recorded images of a diffuse fluorescein solution (1 μM; Fig. 2A), and of large 15.45 ± 0.04 μm (mean ± standard deviation; dimensions provided by manufacturer) diameter fluorescent beads (FS07F; Bang Laboratories Inc., USA; Fig. 2B). To image fluorescein fluorescence, the bath chamber was filled with the fluorescent solution. Large-diameter fluorescent beads were diluted in water (100× dilution in water) and left to settle on the microprobe prism surface. Figure 2A shows the normalized intensity across the centre of the circular field of view (raw image of fluorescein fluorescence inset). Due to vignetting of the illumination light and of the fluorescence excitation by the cylindrically shaped GRIN lens, the efficiency drops to 50% at a distance approximately 150 μm from the centre of the probe. Translating the focal plane 500 μm along the z-axis, beyond the distal prism surface resulted in a slight increase in fluorescence signal, relative to maxima. Note that due to the curvature of pressurized arteries, this intensity profile does not necessarily reflect that obtained in imaging experiments performed in intact arteries. Average optical excitation power density (1 nW μm −2 ) was calculated across the full 500 μm field of view from power measurements made at the output of the GRIN microprobe using a photodiode power sensor and power meter (200-1100 nm; S120VC and PM100A; Thorlabs, UK).
To study the resolving power of the micro-endoscope, we imaged subresolution 1 μm fluorescent microspheres (F-8823; Invitrogen, UK). The mean fluorescent microsphere diameter (as provided by the manufacturer) was 1.1 μm with a coefficient of variation (standard deviation/mean) of 4%. A small droplet of a fluorescent microsphere suspension (1 × 10 6 dilution in water) was manually pipetted onto the distal prism surface. Fluorescent beads were left to settle on the distal prism surface before being brought into focus and imaged (Fig. 2C). Line intensity profiles of the fluorescence emitted from individual beads were taken (Fig. 2D).  Gaussian fits of these line intensity profiles yielded a lateral resolution of 4.51 ± 0.04 μm (n = 6 beads). This measured value may be considered to be the actual resolution of our system because it is much larger than the actual size of the beads (Kim et al. 2010). All optical characterization experiments were performed using the same set-up and external optics as that used for intraluminal imaging of the endothelium.

Ca 2+ imaging
Endothelial Ca 2+ signals were recorded (5 Hz) from ß200 endothelial cells in each artery by GRIN microendoscopy.
The plane of focus of the GRIN imaging system was set to the endothelial layer. Arteries at resting diameter were stimulated by application of ACh (100 μM), delivered by a handheld pipette to the outside of the pressurized artery to confirm endothelial viability. Preliminary experiments established that ß80% of the cells in the field of view responded to ACh (100 μM). Arteries in which the majority of endothelial cells did not exhibit a Ca 2+ response to ACh were discarded. To examine the effect of increases in transmural pressure, endothelial Ca 2+ signalling was recorded at pressures within the physiological range for the artery (60 mmHg, 110 mmHg and 160 mmHg). Following each change in pressure, the imaging system was refocused on the endothelium and arteries were left to equilibrate for 20 min. The Ca 2+ responses evoked by ACh at various transmural pressures were studied in single experiments and expressed relative to the control response (60 mmHg). There was significant overlap in the cells imaged at each pressure, but matching individual cells was not possible.
In other experiments, the extraluminal PSS was exchanged for PSS containing various pharmacological inhibitors or Ca 2+ -free PSS and left for 20 min, before activation again with 100 μM extraluminally applied ACh. The Ca 2+ responses evoked by ACh in arteries treated with pharmacological inhibitors or Ca 2+ -free bath solution were studied (control and treatment) in the same artery and expressed relative to the control response. Following each acquisition period, the bath solution was immediately exchanged and the arteries were left for at least 20 min to re-equilibrate. Smooth muscle cells were not loaded with OGB-1/AM in any of our preparations, as indicated by an absence of fluorescence staining orthogonal to the longitudinal vessel axis.

Signal analysis
Individual endothelial cells were segmented using a custom, semi-automated image processing procedure. Previous successful fully automated image processing of endothelial Ca 2+ signals has been achieved by assigning regions of interest (ROIs), of predetermined size and shape to image sequences based on subcellular activity (Francis et al. 2012 Schindelin et al. 2012) and for manually splitting joined cells, all processing was performed using batch-processing algorithms in ImagePro Plus.
Individual fluorescence signals were extracted for each polygonal region from the raw image stacks using a custom program in the Python language. Fluorescence signals are expressed as baseline corrected values (F/F 0 ), calculated by dividing the raw signals by the average value of using a user-defined period (typically 50 frames) preceding ACh-evoked Ca 2+ activity. There was a significant variation in the time taken for each cell to respond to ACh. To aid analysis, individual F/F 0 traces were aligned with respect to their peak rate of change. The alignment provides a clear illustration of total Ca 2+ activity. Baseline values of F/F 0 , peak amplitudes and the time of peak rate of change for each signal were calculated automatically and stored as data tables within .csv files. These .csv files were then imported into Origin 9.1 for calculation of peak changes in fluorescence intensity ( F/F 0 ), and for plotting using custom analysis scripts. For presentation, an 6-point (1.2 s), third-order polynomial Savitzky-Golay filter was applied within Origin; all measurements were from unsmoothed traces.

Measurement of arterial diameter
At the end of some experiments, we recorded the diameter of arteries as pressure was increased, in 5 mmHg increments, from 0 mmHg to 200 mmHg. For videomicroscopy-based diameter measurements, arteries were illuminated with bright field illumination, which was guided to a CCD camera (Sony XC-77; Sony, Japan) mounted on the side-port of the inverted microscope on which our GRIN imaging system was mounted. Images were captured from the CCD camera using Micromanager software (Edelstein et al. 2010), and a USB video capture device (Dazzle; Pinnacle Systems, USA) and stored on a computer for subsequent analysis. Due to light scattering by the artery wall, the luminal diameter could not be assessed. Thus, outer artery diameter was measured using the Vessel Diameter plugin for ImageJ (Fischer et al. 2010).

Histological analysis
Rat carotid arteries were fixed at pressure after length adjustment (to remove buckle) at 160 mmHg. Following length adjustment, lumenal PSS was replaced with Zenker's fixative. Once the lumen was filled with fixative, the artery was sealed and the pressure was immediately raised to 60 mmHg or 160 mmHg, and the extraluminal PSS was replaced with Zenker's solution. Arteries were left to fix for at least 2 h -a time determined, in preliminarily experiments to be required to prevent a reduction in arterial dimensions upon removal of pressure. Following fixation, arteries were removed from the myography chamber, washed overnight in tap water and stored in 70% ethanol at 4°C until use. Arteries were paired for J Physiol 593.24 analysis (i.e. two arteries from each animal were fixed at the two different pressures (60 and 160 mmHg).
Although a relatively high extraluminal concentration of ACh was required to activate the majority of endothelial cells, it is likely that the endothelium was not exposed to this concentration due to the presence of an adventitial barrier to diffusion. The concentration of ACh in the lumen in the present experiments was estimated to be ß100-fold less than the bath concentration at the time of Ca 2+ measurement. Two experiments support this conclusion. First, in arteries surgically opened, ACh (1 μM) produced approximately equivalent responses (i.e. ß80% cells responding) to ACh (100 μM) applied to the bath in intact artery. Secondly, the concentration of ACh was also estimated from the time course of diffusion of fluorescein across the vascular wall in the same experimental conditions as the pressurized artery. The fractional fluorescence change (relative to the final steady-state value) at the time (10 s after addition) that ACh evoked Ca 2+ responses was used to estimate the fraction of the ACh present in the lumen (ß1 μM). In other studies, the potency of extraluminally applied ACh was reported to be ß1/50 of intraluminally applied ACh in the dog mesenteric artery (Toda et al. 1990) and 50-100 times less potent in femoral artery (Angus et al. 1983;Toda et al. 1988). Additionally bradykinin is unable to evoke relaxant responses in isolated porcine coronary arteries when applied extraluminally, independent of enzymatic degradation and luminal pressure, but is able to evoke responses when applied intraluminally (Tanko et al. 1999).
Drugs were all obtained from Sigma except for OGB-1/AM and Pluronic F-127, which were obtained from Invitrogen. All drugs were dissolved in DMSO and diluted to working concentration in PSS such that the total volume of DMSO was less than or equal to 0.1%.
Harris's haematoxylin, Scotts tap water and Eosin were obtained from and Sigma (UK). Zenker's fixative was obtained from Fisher Scientific.

Statistics
Summarized data are expressed as means ± standard error of the mean (SEM). One-way ANOVA (with Tukey's post hoc test as appropriate) was used for comparisons between groups, and biological replicate (animal) was treated as a random factor. Statistical analyses were performed using Minitab 17 (Minitab Inc., USA). A P-value less that 0.05 was considered significant and n is number of animals.

Model description
To understand how changes in cell geometry alter IP 3 -evoked Ca 2+ release, local [Ca 2+ ] in the IP 3 receptor (IP 3 R) microdomain was determined computationally in the time period encompassing ion channel opening. The cytosolic concentrations of ionic Ca 2+ , buffer and buffered Ca 2+ are represented by C Ca,cyt , C B,cyt and C CaB,cyt respectively. The partial differential equation governing the concentration, C s , of a species, s, is given by: where C s is the concentration, D s is the diffusion coefficient, φ s is the source term derived from chemical reactions and J s is the source term resulting from trans-membrane flux, which was taken as zero for all species except Ca 2+ , for which it comprises IP 3 R-mediated Ca 2+ currents.
Buffering. We employed the first order mass action reaction kinetic: where K on and K off are the rate constants for the buffer.
Trans-membrane flux. Ca 2+ pumps (SERCAs) and membrane (plasma and ER) leakage currents are typically continuous, low magnitude processes that function to maintain specific concentrations of Ca 2+ in the cytosol (typically < 100 nM) and the ER (typically > 0.5 mM) in the long term (Table 1). This is in contrast to IP 3 Rs, which are reported to open for durations of between 2 ms and 20 ms, with a transient current that is far higher than those of the aforementioned long-term processes. We therefore omitted the slow acting sources and sinks, as their effects over the brief temporal and spatial scales of an individual microdomain, with which we are concerned, are limited. The transport of Ca 2+ from the ER to the cytosol, through an open IP 3 R, is a purely diffusive process, and is therefore driven in linear proportion to the ionic concentration gradient between the two partitions. Therefore individual IP 3 R are represented by the source term: where C Ca,er is the concentration of Ca 2+ in the ER, taken as a constant, J 0 is the experimentally measured maximal ion current of an isolated IP 3 R and α is a conversion factor from a current (in moles per second) to a molar concentration for the voxel to which the current is applied.
Computation. Equation (1) was solved for a regular voxel grid in Cartesian space (Table 1) using a first order finite difference solver. A sufficiently large volume was used such that edge effects are negligible, with the condition being that C Ca,cyt did not exceed a level 1% above baseline at the edges (excluding the edge containing the IP 3 Rs), except where we intentionally simulated compressed cells. The Courant-Freidcrichs-Lewy condition: was obeyed for each species, where dt is the simulation time step and dx is the linear size of a voxel. Simulation was conducted using bespoke C code employing Intel AVX-2 vector extensions. The Python language was used to process and analyse simulation results.
Boundary conditions. Our model was initialized with homogeneous Ca 2+ and buffer concentrations as given in Table 1, with C CaB,cyt being initialized to equilibrium values. A zero flux boundary condition was applied to the edges of the cuboidal simulation volume.
Simulation verification. Our simulation code was compared with analytical expressions for simplified test cases of 3D diffusion from a point source and for the microdomain profile of an isolated Ca 2+ source in a strongly buffered environment.
Analytical diffusion. Equation (6) gives the analytical form of the concentration profile for a slug of mass M released at position r = 0 and time t = 0, where r is scalar radius from the origin (Balluffi et al. 2005).
The Ca 2+ diffusion component of our simulation was initialized with a volume of 1.68 μm on a side, with Ca 2+ parameters as per Table 1, and with zero Ca 2+ concentration except in the central voxel, which was set to 1 M. Figure 3A and B illustrates the good agreement between our simulation and the analytical case.
Analytical microdomain profile. An analytical solution for the equilibrium concentration profile, at distance r from an isolated source of constant current J Ca in an isotropic, exists for an inexhaustible buffer that only forward binds (known as the excess buffer approximation; Smith, 1996): where c Ý is the Ca 2+ concentration far from the sourcethat is the equilibrium level, γ = (Dτ) 0.5 is a space constant and τ = 1/(K on B) is the mean capture time of the buffer, of concentration B. Our simulation was configured to match, as well as possible, this analytical case. Specifically parameters from Table 1 were used to simulate a single source, with the following modifications: K Off was set to zero (forward binding only), φ B,cyt from Eqn (2) was set to zero (inexhaustible buffer) and the concentration gradient was removed from Eqn (3) (constant current source). As a consequence of these changes any background Ca 2+ was rapidly depleted in this model, so c Ý was also set to zero to facilitate direct comparison. After running the model, the total quantity of free (unbuffered) cytosolic Ca 2+ was examined (Fig. 3C). This level rose rapidly, stabilizing to equilibrium in ß0.1 ms. A comparison of Eqn (6) and the concentration profile of the model, once equilibrium is reached, is shown in Fig. 3. The responses were very similar, with the effect of quantized voxel sizes limiting the model concentration at small r and edge effects raising the concentration at high r.
Simulation parameters. Parameters for the simulation were chosen to be physiologically relevant, and are given in Table 1 along with references. Rate constants and diffusion coefficients were obtained from Means et al. (2006) and Bortolozzi et al. (2008) and reported as measured in the cytoplasm of cells.

Results
The GRIN microprobe inserted easily into the lumen of an artery (Fig. 1) delivered a field of view of 0.5 mm diameter and allowed a large number (ß200) of naturally connected endothelial cells to be imaged ( Fig. 1I) with subcellular resolution (ß4.5 μm; Fig. 2D). In arteries pressurized to 60 mmHg, activation of the endothelium by extra-luminal acetylcholine (ACh; 100 μM applied to the chamber, ß1 μM estimated at the vessel lumen; 60 mmHg; online Supporting information, Movies S1 and S2) evoked reproducible rises in cytoplasmic Ca 2+ concentration ([Ca 2+ ] c ) in the majority of cells in the field (Fig. 4A). The response was composed of temporally distinct components. Initially, small spatial groupings of cells (macrodomain) activated (Fig. 4A, Movie S2) in small regions of the endothelium and from which waves expanded (Fig. 4A) to recruit the remaining majority of cells in the field of view (Figs 4 and 5). As a result of the wave progression there was a large spread of times to first Ca 2+ response (latency) in various cells ( Fig. 4B and C, left panels) which complicated summarizing and evaluating responses. Therefore, the temporal latency between individual Ca 2+ traces (F/F 0 ) was removed by aligning the signals to each other (using the initial peak in the first derivative of each signal) ( Fig. 4B and C, middle panels). This procedure removed heterogeneity resulting from the time to onset of individual signals and provided a clearer illustration of total Ca 2+ activity. Upon consecutive stimulation with ACh, the response, as measured from the number of cells activated and the average peak change in F/F 0 , was approximately reproducible (Fig. 4D, n = 5).
Endothelial cells are stimulated almost constantly by extrinsic blood-borne bioactive molecules amid a background of haemodynamic mechanical forces. The combined effect of chemical activators and mechanical force is presumably incorporated into the endothelium's biological response, but details are largely unknown. We addressed whether a change in endothelium-dependent Ca 2+ signalling, evoked by a bioactive molecule (ACh), occurs with variation in transmural pressure, and hence mechanical force. At 60 mmHg, ACh (100 μM) evoked asynchronous Ca 2+ waves and oscillations that originated in macrodomains ( Fig. 7A; Movie S2). As transmural pressure increased (60 to 110 to 160 mmHg), both the amplitude of the Ca 2+ rise that occurred in each cell and the temporal spread of total activation decreased ( Fig. 7A-D; Movie S3). The suppression of the amplitude of the Ca 2+ signals reversed as pressure decreased (not shown). These results suggest that increased pressure in the intact artery suppressed IP 3 -evoked Ca 2+ signals. IP 3 R itself is not reported to be stretch sensitive, and mechanical activation of IP 3 -generating processes is unlikely to explain our observations as transient increases in IP 3 production and IP 3 -evoked Ca 2+ release occur typically (Jena et al. 1997).
The question arises as to how increased pressure suppressed IP 3 -evoked Ca 2+ signals. Significant changes in endothelial cell shape occurred with increased pressure (Fig. 8). Increased circumferential stretch, with increased pressure, elongated cells around the artery, and cell thickness was significantly decreased to maintain volume ( Fig. 8A and B). Increased transmural pressure may have also compressed the endothelial cells. The change in cell geometry is likely to alter Ca 2+ dynamics in the endothelial cells at the microdomain level near IP 3 R. The effect of changes in cell geometry on IP 3 -evoked Ca 2+ signals (moles) 10 1 10 2 10 3 r (nm) Figure 3. IP 3 -evoked Ca 2+ release simulation A, shell density plotted against distance from source (i.e. IP 3 R). Theoretical (analytical test case; lines) plots of shell density for a slug of mass released at the origin shows good agreement with our simulated results (circles). B, Ca 2+ concentration as a function of radial distance from IP 3 Rs. When equilibrated, the Ca 2+ profile may be approximated by an analytical case (black line), which compares accurately to an examination of our simulation (yellow line). C, free Ca 2+ concentration at an IP 3 R as a function of time from receptor opening. The total cytosolic Ca 2+ rises immediately after IP 3 R opening then plateaus as the current flowing through IP 3 R and cytosolic Ca 2+ buffering reaches equilibrium (ß0.1 ms). See Methods and Table 1 for equations, source references and parameters used to generate the figures. J Physiol 593.24 was therefore considered further. To explore the effects of changes in the cells' geometry on IP 3 -mediated Ca 2+ release at the relevant scale, which exists below functional optical microscopy, we computationally simulated events at an IP 3 R cluster.
IP 3 Rs are arranged in clusters and the sensitivity of IP 3 Rs to Ca 2+ is such that the [Ca 2+ ] c in the microdomain of a single open IP 3 R (or cluster) may be high enough to activate nearby IP 3 Rs (or clusters) (Callamaras et al. 1998). This process of Ca 2+ activation may become regenerative and carry a wave of activity across cells (Figs 4 and 5). In the transmission process, IP 3 R activity is exquisitely sensitive to distance between channels on the ER and to the cytosolic space between the ER and membrane structures, such as mitochondria and the plasma membrane (McGeown et al. 1996;Rizzuto et al. 1998;Mazel et al. 2009;Olson et al. 2010Olson et al. , 2012. These features (clustering, cytosolic space) create a local diffusive environment (i.e. microdomain) in which Ca 2+ and Ca 2+ buffers regulate the activity of IP 3 Rs and the release of Ca 2+ from the ER due to the local [Ca 2+ ] c .
The opening of a single IP 3 R introduced a point source of Ca 2+ that equilibrated rapidly by diffusion and aggressive cytosolic buffering, to produce a localized, quasi-static region of elevated [Ca 2+ ] c (microdomain; Fig. 9A and B) around the channel. The Ca 2+ concentration decreased sharply with distance from the IP 3 R with a characteristic scale (the distance at which [Ca 2+ ] remains  response evoked by ACh (100 μM; bath applied; 1 μM estimated at the vessel lumen) recorded at 60 mmHg. Images are composed of instantaneous Ca 2+ activity (green) overlaid on standard deviation images (greyscale) indicative of total Ca 2+ activity. The rise began in a distinct cluster of cells (see 1 s) and progressed from there. Each bright spot is an individual cell. Scale bar: 100 μm. B, ACh-evoked Ca 2+ signals from the same cells shown in A. In the left panel (unaligned data) the range of times for Ca 2+ to increase for each of the ß200 cells is approximately 4 s. The spread of responses resulted in mean data representing the data poorly. The position of the first peak of derivative signals was used to align Ca 2+ signals. This procedure synchronizes the Ca 2+ rises occurring in each due to the action of ACh (middle panel 'aligned') and illustrates total Ca 2+ activity (thick red line) with increased clarity. On successive application of ACh (20 min reequilibration following wash), the response to ACh was reproducible (right panel). C, select examples of unaligned (left) and aligned (right) cellular Ca 2+ responses from ROIs shown in A. The grey box illustrates the time point of measurements from aligned signals. D, summary data illustrating the reproducibility of Ca 2+ responses upon repeat application (wash + 20 min reequilibration) with ACh. Data are presented as means ± SEM and normalized to control (n = 5; first ACh application, 1; not shown), P < 0.05 was considered significant. elevated by at least one order of magnitude) on the order of 100 nm (Fig. 9B). The presence of many such point sources resulted in a greater [Ca 2+ ] c in the vicinity of IP 3 R clusters than isolated receptors (Fig. 9C). The raised [Ca 2+ ] c that occurred as a result of IP 3 R clustering reduced the local ion gradient between the store ([Ca 2+ ] ER ) and the cytosol, and so reduced the entropic force driving Ca 2+ release through the channel. In addition, the elevated [Ca 2+ ] c merged the microdomain of each IP 3 R into a unified, larger domain with scales approaching the 500 nm thickness typical of endothelial cells.
In wide, thin cells (e.g. endothelium) IP 3 Rs release Ca 2+ into the cytosolic space between the ER and the cellular plasma membrane, where proximity of the IP 3 R to the cellular plasma membrane may create a restricted environment limiting the diffusion of Ca 2+ . The decreased cell height as a result of the change in cell shape (Fig. 8) reduces the distance between the IP 3 Rs and the opposing plasma membrane, restricting Ca 2+ diffusion away from IP 3 R and, as a result, increasing the [Ca 2+ ] c at the receptors ( Fig. 9A and C). Thus, the decreased height reduced the total Ca 2+ liberated from the ER by reducing the entropic force driving Ca 2+ release (Fig. 9D). In this way, a mechanical force (circumferential stretch) can reduce Ca 2+ release via IP 3 R, in the absence of any forceor stretch-sensitive molecular machinery. Notably, the clustering of IP 3 Rs extends this effect, when compared to isolated receptors, from sub-100 nm thick cells to those with a 100-500 nm thickness typical of the endothelium (Fig. 9).

Discussion
The endothelium's control over cardiovascular structure and function requires cooperation among endothelial cells to both sense and communicate stimuli efficiently with itself and other cell types (Ledoux et al. 2008;Behringer & Segal, 2012;Tran et al. 2012;   In each artery, a time series of Ca 2+ images from the endothelium illustrates that large-scale, ACh-evoked (100 μM; bath application) Ca 2+ waves originate in distinct clusters of cells and propagate from there across the endothelium. The images are composed of instantaneous Ca 2+ activity (green) overlaid on standard deviation images (grayscale), which are indicative of total Ca 2+ activity. Each bright spot is a single endothelial cell. Scale bars: 100 μm. J Physiol 593.24 2012b; Qian et al. 2013). The cooperative behaviour is not well understood because of the difficulties in studying large numbers of endothelial cells in a physiological configuration. The miniature, wide-field GRIN imaging probe described permits endothelial Ca 2+ activity from within an intact, pressurized artery to be observed. The microprobe bypasses the difficulties created by the artery wall in imaging blood vessels in a physiological configuration. Furthermore, the absence of biological tissue between the end of the probe and the endothelium minimized the light intensity required. The microprobe has a large depth-of-focus (141 μm) and delivers a 0.5 mm field of view to allow 200 cells to be viewed simultaneously.
Control (93 Cells The detection of chemical stimuli appears to be distributed among endothelial cells. In response to ACh, Ca 2+ rises initiate in small regions of the endothelium and expand from these regions to recruit most of the cells in the field. The initiator regions (macrodomains) appear to be more sensitive to ACh. The path of the Ca 2+ signal from the sensitive regions at times appeared poorly coordinated among endothelial cells -even chaotic. Nonetheless, the path and pattern were repeatable on successive application of ACh suggesting an encoded signal communicated among cells. The response to ACh was derived from an IP 3 -mediated release of Ca 2+ from the internal store. The Ca 2+ rise persisted in a Ca 2+ -free bath solution and was abolished by the SERCA pump inhibitor cyclopiazonic acid. The ACh-evoked Ca 2+ rise persisted in ryanodine and caffeine failed to evoke a Ca 2+ rise. However, release was inhibited by the IP 3 R blocker 2-APB. The transmission of the Ca 2+ signal among endothelial cells may involve Ca 2+ or IP 3 or both.
In addition to receiving and communicating chemical stimuli, arteries are under permanent mechanical activation from shear stress and blood pressure. While cells often respond to the activation of mechanically sensitive components (e.g. via shear stress) with increased ion channel and cell activity, the mechanical response to pressure differs and endothelial activity is suppressed. . Ca 2+ activity decreased as pressure increased in the three repeated activations of the same artery (20 min wash/re-equilibration period between each ACh application). The data are shown as described in Fig. 2A. Scale bar 100 μm. B, baseline corrected and aligned Ca 2+ signals (average overlaid in black and shown in colour in the bottom panel) from the Ca 2+ images shown in A. C, summary data illustrating the pressure-dependent decrease in peak F/F 0 (n = 8, ±SEM). D, the temporal spread of cellular activation decreases with increased pressure. Note the fourth colour in the histogram (purple) arises from overlap of colours from the data in 110 mmHg (blue) and 160 mmHg (red). E, summary data illustrating a logistic change in vessel diameter as pressure is increased (from 0 to 200 mmHg; n = 3, ±SEM), measured using the Vessel Diameter plugin for ImageJ. Artery diameter was 1135 ± 110 μm at 60 mmHg, 1429 ± 92 μm at diameter at 110 mmHg and 1460 ± 83 μm at 160 mmHg (n = 3). F, summary data illustrating an increase in peak F/F 0 as pressure is decreased (n = 6, ±SEM). J Physiol 593.24 (Hishikawa et al. 1992;De Bruyn et al. 1994;Huang et al. 1998;Paniagua et al. 2000;Jurva et al. 2006;Phillips et al. 2011). How pressure may suppress endothelial activity is unresolved but increased reactive oxygen species may contribute (Huang et al. 1998;Vecchione et al. 2009;Zhao et al. 2015). Here we report that endothelial Ca 2+ signalling is suppressed when pressure is elevated. The suppressed Ca 2+ signalling will presumably result in reduced production of nitric oxide and reduced endothelium-dependent dilatation (Hishikawa et al. 1992;Sun et al. 2001).
The sensing system(s) responsible for detecting changes in pressure in the endothelium are poorly understood. Pressure-evoked mechanical deactivation of ion channels has been proposed to account for some physiological responses (Hoffman et al. 2011;Bagher et al. 2012). We propose that changes in cell shape may be sufficient to underlie the detection of pressure-dependent mechanical forces by the endothelium. The changes in geometric shape of the cell in response to pressure changes are substantial: endothelial cells may be flattened either by circumferential stretch, as the height of endothelial cells is reduced to maintain volume, or by compression due to the radial force of pressure or both. Such changes in the geometric shape of the cell will bring the endoplasmic reticulum and plasma membrane closer together and create a region of restricted diffusion (see also Hong et al. 2008;Qi et al. 2015). As shown in Fig. 9, the region of restricted diffusion changes the local diffusive environment surrounding IP 3 R clusters and results in a higher concentration of Ca 2+ near the IP 3 R microdomain. This higher concentration results in a reduced Ca 2+ gradient between the store and the cytosol. As a result, the entropic force driving Ca 2+ release is reduced and so the amplitude of IP 3 -evoked Ca 2+ signals is also reduced (Fig. 10). Reduced IP 3 -evoked Ca 2+ release may explain the suppression of artery level endothelium-derived responses at high transmural pressure (De Bruyn et al. 1994;Huang et al. 1998;Paniagua et al. 2000;Zhao et al. 2015). A specific force-or stretch-sensitive component may not be required, but rather changes in cell shape may alter the local concentrations of second messengers (e.g. Ca 2+ ) within the cell to modulate the overall response of the cell.  A and B) show the yellow box expanded to illustrate the change in shape of single endothelial cells at each pressure. SM, smooth muscle; ADV, adventitia. Scale bars 10 μm main picture, inset 5 μm. C, average nuclei thickness (a measure of cell depth) at 60 mmHg and 160 mmHg. At 60 mmHg, n = 37 cells from 3 animals; 160 mmHg n = 35 cells from 3 animals ( * P < 0.05).
The pressure-sensitive change in IP 3 -evoked Ca 2+ signalling provides an additional level of complexity and subtly to mechanosensitive signalling, which may induce further signals to generate a functional response. For example, the changes in [Ca 2+ ] c as a result of geometric modulation of IP 3 -evoked Ca 2+ release may modulate the activity of plasma membrane-located Ca 2+ -sensitive ion channels to explain, at least in part, some contradictory observations of apparent mechanical activation (shear stress, Strotmann et al. 2000;Alessandri-Haber et al. 2003;Loukin et al. 2010) and inhibition (pressure increases, Bagher et al. 2012) of ion channels. In the latter study, when pressure decreased from 80 mmHg to 5 mmHg, the frequency of transient localized endothelial Ca 2+ rises increased (Bagher et al. 2012). The authors attributed the Ca 2+ changes to influx via TRPV4. The mechanosensor responsible for the pressure-induced change in TRPV4 activity was unclear but the authors suggested reduced radial compression of endothelial cells may activate the channel (Bagher et al. 2012). TRPV4 is Ca 2+ sensitive and activated by increases in [Ca 2+ ] c (Strotmann et al. 2003). Interestingly, in their study Bagher et al. (2012) found the local endothelial Ca 2+ events studied were blocked fully by a phospholipase C inhibitor (U-73122), and an IP 3 R blocker (xestospongin C) almost abolished Ca 2+ events at 80 mmHg and significantly inhibited responses at 5 mmHg. These results suggest that IP 3 -mediated Ca 2+ release may contribute significantly to the spontaneous  Ca 2+ events. Geometric regulation of IP 3 -evoked Ca 2+ release may provide the pressure sensor for the change in TRPV4 activity. Endothelial cells are acknowledged to be sensitive to their geometry and changes are known to regulate proliferation and cell death (Chen et al. 1997). Changes in cell shape alter tension in the cell membrane to modify components within membranes (Sheetz, 1993;Mederos y Schnitzler et al. 2008) with consequences for cell behaviour (Mossman et al. 2005). Alternatively, changes in links between the plasma membrane and cytoskeleton or mechanosensitive ion channels may detect changes in membrane tension (Patel et al. 2001). Several proteins also have domains (BAR or ALPS) found in membranes that are curved and may contribute to detecting cell shape changes. The extent of membrane binding of these protein domains changes with membrane curvature (Zhao et al. 2011). The protein domains may either induce or are sensitive to membrane curvature (Zimmerberg & Kozlov, 2006). Guanine nucleotide exchange factors and GTPase activating proteins with curvature sensing protein domains may alter the activity of GTPases in a tension-dependent manner (de Kreuk et al. 2011) to explain geometry sensitivity of cells. We propose changes in cell geometry may also alter ion diffusional gradients, which act then to sense stimuli and evoke vascular responses.
For normal function, cells in the vascular wall must act in a coordinated manner. The spatial relationships among extracellular chemical stimuli, cellular architecture and Ca 2+ signals are central to signal processing and the coordinated behaviour. Indeed, although 'sensing' may occur on the microdomain scale, the macroscopic configuration of the intact endothelium is essential for physiological mechanotransduction. The detection of mechanical forces like pressure is critically dependent on the organization of the sensing proteins and the contact of cells with other cells. In other investigations on mechanotransduction, a single type of protein was found to evoke different responses when either organized in 3D filaments or arranged on a flat surface (Cukierman et al. 2001). Some types of mechanosensitivity seen in isolated single cells were lost when cells are in contact with one another (Yeung et al. 2005). Our results in intact arteries in a physiological configuration suggest that mechanical activation by pressure may be merged into the agonist-evoked response in the endothelium. The pressure dependence of Ca 2+ release provides a mechanism for transmural pressure to   Figure 10. Model of pressure-dependent, geometric suppression of Ca 2+ release through IP 3 Rs As pressure is increased, from low (A) to high (B), arterial diameter increases and endothelial cells are stretched along the circumferential direction. In addition, the radial force of pressure acts upon the cells. To conserve volume, the height of endothelial cells decrease (B). The insets depict the processes leading to changes in ACh-induced Ca 2+ release in an endothelial cell at low (A, top) and high (B, bottom) pressure. At low pressure, activation of IP 3 R, after ACh application, results in Ca 2+ release from the endoplasmic reticulum (ER). Ca 2+ diffuses rapidly from the release site to maintain a gradient for continued release. The microdomains of Ca 2+ operate over sizes (300 nm) comparable to the height of an endothelial cell (see text). When pressure is increased (B), the height of the cell decreases and so the distance between the ER and the plasma membrane (PM) also decreases. The decrease in ER-PM distance restricts the diffusion of Ca 2+ from the release site (IP 3 Rs), effectively increasing the local Ca 2+ microdomain concentration and decreasing the entropic force (indicated by size of yellow arrows) driving Ca 2+ release through IP 3 Rs. The resultant effect is a decrease in Ca 2+ release from the ER as pressure is increased. SM: smooth muscle. be encoded and integrated to a common, coordinated output without imposing a requirement for mechanosensitivity on any of the molecular machinery in the signalling pathway. The time course of this response would be rapid and may form an early event in an altered mechanical environment. Furthermore, our results suggest how hypertensive conditions may limit the ability of cells to engage in Ca 2+ signalling, offering insight into the associated degeneration of endothelial response (Luscher & Vanhoutte, 1986;Wallace et al. 2007) and the role that IP 3 R clustering may play. temporally smoothed Ca 2+ activity (green) overlaid on standard deviation image (grayscale). Data was acquired at 5 Hz and the scale bar is 100 μm. Movie S3. ACh activated IP 3 -evoked Ca 2+ signals were significantly attenuated as the artery transmural pressure increased from 60 mmHg to 110 mmHg to 160 mmHg. After each application of ACh the bath solution was washed with >20 times the bath volume, and the artery was allowed to re-equilibrate for 20 minutes before the next pressure change and ACh addition. The movie corresponds to Ca 2+ traces and time-series data shown in Figure 7A and Figure 7B respectively. The movie is a time series of Ca 2+ wave activity (green) overlaid on standard deviation images (STDev) (grayscale). Note that STDev images only show cells that exhibit Ca 2+ activity. Data was acquired at 5 Hz and the scale bar corresponds to 100 μm.