Novel method to study pericyte contractility and responses to ischaemia in vitro using electrical impedance

Abstract

Pericytes are contractile vascular mural cells overlying capillary endothelium, and they have been implicated in a variety of functions including regulation of cerebral blood flow. Recent work has suggested that both in vivo and ex vivo, ischaemia causes pericytes to constrict and die, which has implications for microvascular reperfusion. Assessing pericyte contractility in tissue slices and in vivo is technically challenging, while in vitro techniques remain unreliable. Here, we used isolated cultures of human brain vascular pericytes to examine their contractile potential in vitro using the iCelligence electrical impedance system. Contraction was induced using the vasoactive peptide endothelin-1, and relaxation was demonstrated using adenosine and sodium nitroprusside. Endothelin-1 treatment also resulted in increased proliferation, which we were able to monitor in the same cell population from which we recorded contractile responses. Finally, the observation of pericyte contraction in stroke was reproduced using chemical ischaemia, which caused a profound and irreversible contraction clearly preceding cell death. These data demonstrate that isolated pericytes retain a contractile phenotype in vitro, and that it is possible to quantify this contraction using real-time electrical impedance recordings, providing a significant new platform for assessing the effects of vasoactive and vasculoprotective compounds on pericyte contractility.

Introduction

Pericytes are vascular mural cells located within the basal lamina and found at particularly high densities in the capillaries of the central nervous system (CNS).¹ Although first described in the 19th century, their physiological importance remained poorly understood until recently, when they have been shown to induce and maintain the blood–brain barrier (BBB) phenotype of the cerebral endothelium in development, adulthood and aging, in both the brain and spinal cord,^2–4 as well as regulate vascular stabilisation⁵ and angiogenesis.⁶ Given these crucial roles, it is unsurprising that pericyte loss and dysfunction have been described in a number of neurological disorders, including Alzheimer’s disease, mild cognitive impairment and amyotrophic lateral sclerosis, which are also known to feature BBB alterations.^7–9

Due to their location and morphology – featuring elongated processes along vessels with shorter transverse processes enveloping the underlying endothelium – it was hypothesised that pericytes may have contractile functions, potentially contributing to regulation of cerebral blood flow (Figure 1(a)). Although the issue remains contentious,^10,11 there is nevertheless strong evidence that pericytes are contractile and capable of constricting and dilating capillaries both in vitro and in vivo.^12,13 This is particularly relevant to diseases such as acute ischaemic stroke, a leading cause of disability and mortality.¹⁴ Irreversible pericyte constriction has been implicated as a major component of the ‘no-reflow’ phenomenon in stroke, where microvascular patency (and as a consequence nutrient supply to the tissue) is not restored even when the large artery occlusion is removed.

Figure 1.

Schematic of pericyte contractility in vivo, and in vitro using electrical impedance. Pericytes are vascular mural cells that wrap around capillaries and provide contractile tone (a). Plating cells on E-plates ((b)–(d)) allows for measurement of basic adhesion (b), as well as changes in morphology (c) and proliferation (d) by quantifying impedance. Cell index as a measure of impedance is represented here as I.

One substantial limitation of studying pericyte contraction is the methodological difficulty associated with the various models that are available. Clearly, the most reliable models are in vivo, as a naturally pressurised vasculature is paramount to accurate evaluation of vascular physiology. Unsurprisingly, however, in vivo models are also the most technically challenging and require the use of confocal microscopes and, particularly in the case of the cerebral microvasculature, two-photon laser scanning microscopes.^5,11 Even so, they are restricted in terms of imaging depth and resolution and provide very limited throughput. There are excellent resources on imaging capillary constriction in acute brain slices^13,15; however, similarly to in vivo imaging, this model is limited in its throughput by the small number of capillaries that can be imaged per slice. Both in vivo and acute brain slice models also make direct pharmacological studies more difficult. Multiple cell types complicate the capacity to isolate specific pathways and the effects of individual drugs and receptors, without the use of elaborate transgenic constructs.^10,11 Bath or intracranial application of substances could be exerting a direct effect on pericytes, or acting via intermediate cells (e.g. perivascular astrocytes).

Numerous protocols are available for pericyte isolation and culturing, with well-characterised phenotypes.^16,17 Contractility in these cells has been largely studied through the use of deformable collagen or silicone substrates – the pericytes adhere to the substrate and cause wrinkling when vasoactive compounds are applied; the cells are then imaged with videomicroscopy and the degree of contraction/dilation can be evaluated based on the extent of wrinkling. These models have been used in the literature for a number of decades¹⁸ and suffer from notable drawbacks: first, the measurements are, at best, semi-quantitative, and their correspondence to cell size changes is indirect. Second, the need for microscopy makes time course studies considerably time-consuming without providing much advantage over slice experiments in terms of throughput.

Here, we describe the novel use of a commercially available, high-throughput, real-time and label-free assay, based on electrical impedance, to evaluate pericyte contraction in vitro. We demonstrate that pericytes dose-dependently respond to endothelin-1 (ET-1), adenosine and other vasoactive substances, and demonstrate the receptor-specificity of these effects. We also show that this setup can be used to not only quantify contraction in the acute setting but also to monitor effects on proliferation over several days. This methodology allows much more detailed study of pericyte contractility in vitro than with previous techniques, and opens new avenues for high-throughput pharmacological studies of pericyte function.

Materials and methods

Cell culture

Human brain vascular pericytes (HBVP; ScienCell Research Laboratories, USA) were grown in pericyte growth medium (ScienCell) under standard incubation conditions (37℃; 5% CO₂ in 95% humidified air). All experiments were done at passages of <10.

Immunocytochemistry

HBVP cells were grown to ∼60% confluence on sterilised uncoated glass coverslips. Cells were washed with PBS and fixed with 4% paraformaldehyde for 10 min. Fixed cells were washed with PBS and air-dried for 15 min. Non-specific staining was blocked for 1 h at room temperature using 10% donkey serum in PBS. Primary antibodies were applied in 1% serum/PBS as follows; cluster of differentiation 31 (CD31; mouse-monoclonal; Dako, Ely, UK; 1:200), glial fibrillary acidic protein (GFAP; goat polyclonal; AbCam, Cambridge, UK; 1:500), Iba-1 (goat polyclonal; AbCam; 1:500), beta-type platelet-derived growth factor receptor (PDGFRβ; goat polyclonal; R&D Systems, Abingdon, UK; 1:100), desmin (rabbit polyclonal; AbCam; 1:100) and NG2 (rabbit polyclonal; Millipore, USA; 1:200) at 4℃ and left to incubate overnight. Secondary antibodies (anti-goat, anti-rabbit and anti-mouse Alexa488-conjugated; AbCam, UK; 1:400) were applied for 2 h at room temperature in the dark. Cytoskeletal actin was counterstained using Alexa594-conjugated phalloidin (Sigma-Aldrich, Dorset, UK; 1:200) and nuclei were counterstained with DAPI (Sigma-Aldrich, UK). Coverslips were mounted using aqueous, anti-fade mounting medium (Dako, UK). Images were acquired using a Zeiss LSM 510 confocal microscope (Carl Zeiss, Jena, Germany) or a Nikon Eclipse 1000 M (Nikon, Tokyo, Japan) with appropriate filter settings and acquisition parameters.

Measurement of contractility

We used an electrical impedance system (Figure 1) to detect changes in the contact area between pericytes and the culture dish. HBVP were plated on impedance plates (E-Plate L8; ACEA Biosciences, USA) and briefly allowed to adhere. The cells were then placed into the iCelligence platform (ACEA Biosciences) and allowed to proliferate for 48 h prior to testing. Where applicable, drugs were added to wells in situ and measurement continued immediately after addition. For inhibitors, drugs were added 1 h prior to application of agonists. Data are presented as normalised cell index (a unitless parameter, automatically derived from recorded impedance values in iCelligence software and normalised to the last time point pre-treatment), calculated slope and delta (max–min values between baseline and peak of contraction/relaxation), as well as doubling time at selected times post-treatment. For proliferation analysis, cell index values were evaluated from the last time point with no intergroup differences, to account for batch-to-batch variability in overall proliferation. Controls in each case were the vehicle of the designated drug and were applied at the same time as the drugs. Suppliers and concentrations of drugs are reported in the text. An example raw data plot is included in the supplementary material (supplementary Figure S2).

Chemical Ischaemia

HBVP were seeded on E-plates, as above, and treated with three doses of chemical ischaemia. High dose (500 µM iodoacetate (IA); 5 µM antimycin-A (AmA)), medium dose (50 µM IA; 500 nM AmA) or low-dose (5 µM IA; 50 nM AmA) chemical ischaemia were applied after a 48-h adhesion phase and their responses recorded for 24 h. At 24 h, the medium was removed and cells lysed to determine cell death.

Lactate dehydrogenase assay

Cytotoxicity was evaluated by lactate dehydrogenase (LDH) release into the culture medium, which was quantified as a percentage of total LDH (released and intracellular from lysed cells) using a commercial kit (Promega, UK) as per manufacturer’s instructions.

Quantitative PCR

HBVP cells and human vascular cerebral endothelial cells (hcMEC/D3; Millipore) were seeded at 1 × 10⁶ in six-well plates and allowed to adhere overnight. RNA was extracted and purified using the QIAGEN RNEasy Mini kit (Qiagen, UK), according to manufacturer’s instructions; 500 ng whole RNA was converted to cDNA using the High Capacity Reverse Transcription Kit (Applied Biosystems, UK). Quantitative PCR was performed using SYBR green technology (PrimerDesign Ltd, Southampton, UK) in combination with the Roche LightCycler480 (Roche, UK). Standard curves were generated from mixed samples of cDNA and compared using the Pfaffl method,¹⁹ with normalisation to housekeeping gene glyceraldehyde 3-phosphate dehydrogenase. Quantities were further normalised so that hcMEC expression levels represented 1 and HBVP expression levels were shown as relative-fold compared to hcMEC expression.

TUNEL staining

Terminal deoxynucleotidyl transferase (TdT) dUTP nick-end labelling (TUNEL) was performed using a standard TUNEL kit (Millipore, UK). Positive control cells were treated for 10 min with DNAse 1.

Statistics

Statistical analyses were performed using Prism 5.0 and 6.0 (Graphpad, USA) using one- and two-way analysis of variance, and Student’s t-tests, and appropriate post-hoc analysis, as described in the text. Data are presented as mean with standard error.

Results

HBVP express canonical pericyte markers

Primary pericyte populations frequently undergo a passage-dependent phenotypic change, resulting in alterations in morphology and function.²⁰ HBVP cells were grown to 60% confluency on glass coverslips at a low passage (<P10) and labelled for markers of non-pericyte CNS cell types. HBVP did not express markers for endothelial cells (CD31), astrocytes (GFAP) or microglia (Iba-1; Figure 2). PDGFRβ is a receptor known to be expressed by pericytes of the CNS²¹ and was found in high abundance in this cell population (Figure 2). The cells also exhibited labelling for NG2, another pericyte marker, and desmin (supplementary Figure S1). HBVP cells also express a number of other pericyte markers, and mRNA expression of pericyte markers such as PDGFRβ is increased when the cells are compared to human cerebral vascular endothelial cells (Figure 2(b)). Conversely, markers of vascular endothelial cells such as CD31 are decreased in HBVP cells compared to hcMEC/D3 cells (Figure 2(d)).

Figure 2.

Immuncytochemistry of human brain vascular pericytes. (a) HBVP cells were fixed in 4% paraformaldehyde and stained for endothelial (CD31), astrocyte (GFAP), microglial (Iba-1) and pericyte (PDGFRβ) markers (all green – fluorescein) to confirm a pericyte phenotype. Cells were counterstained with a cytoskeletal marker (phalloidin – red) and a nuclear marker (DAPI – blue). mRNA expression of canonical pericyte marker (PDGFRβ (b)), alpha-smooth muscle actin (c) and an endothelial cell marker (CD31 (d)) were measured in HBVP cells and human cerebral vascular endothelial cells (hcMEC/D3). Expression levels were normalised to housekeeping gene GAPDH. Scale bar represents 30 µm. Data are mean with SEM; n = 4; *p < 0.05 and ***p < 0.001 compared to hcMEC expression.

Endothelin-1 causes a dose-dependent increase in pericyte contraction and proliferation

The electrical impedance iCelligence system has traditionally been used to study rapid contractility in cardiomyocyte populations.²² Here, we wished to apply this system to study the contractility of pericytes. The ET family of potent vasoconstrictive peptides is produced by the vascular endothelium in response to a variety of stimuli.²³ HBVP grown on the iCelligence system contracted in a dose-dependent manner in response to ET-1 (Sigma-Aldrich, UK; Figure 3(a)). This contraction was due to the addition of drugs and not due to changes in the movement of liquid over the electrodes (supplementary Figure S2). Analysis of the time course shows a significant main effect of both time and ET concentration, as well as an interaction between factors (time p < 0.0001; concentration of ET-1 p < 0.01; time:concentration p < 0.0001). This contraction resulted in a rapid and significant decrease in surface area and subsequent change in the calculated slope of cell index in ET-1-treated cells, when compared to control cells (treated with medium alone; Figure 3(b); 0.5 nM p < 0.05, 5 nM p < 0.01, 50 nM p < 0.001). The calculated delta (change in cell index at max and min points during contraction) was also significantly different in ET-1-treated cells, compared to vehicle-treated cells (Figure 3(c); 0.5 nM p < 0.05, 5 nM p < 0.01, 50 nM p < 0.001). This contraction was transient and cell index was reversed when slope was measured onwards from 60 h post-treatment, when both ET-1 and control cells had positive slope values (Figure 3(d)). Analysis of the time course at this chronic stage post-application shows a significant main effect of both time and ET concentration, as well as an interaction between factors (Figure 3(d); time p < 0.0001; concentration of ET-1 p < 0.0001; time:concentration p < 0.0001). The slope of cell index reflected this, and furthermore was significantly increased in all ET-1-treated groups when compared to controls (Figure 3(e)). This slope also reflects doubling time, and therefore proliferative capacity, over longer periods and is significantly shorter in ET-1-treated cells (Figure 3(f); p < 0.0001 for all concentrations). Reductions in impedance measurements during ET-1 treatment were not due to cell death, based on LDH release (supplementary Figure S3).

Figure 3.

Electrical impedence measurement of contraction and proliferation of HBVP in response to endothelin-1. HBVP were grown on the iCelligence system for 48 h prior to treatment with either control medium, or medium containing ET-1 (0.5, 5 and 50 nM). Measurement during the first 30 min shows a rapid contraction and relaxation phase in response to the drug (a), with the initial slope reflecting this contraction (b). Calculation of delta (max–min difference at peak contractility) also reflects differences in response to ET-1 (c). Full time course demonstrating proliferation of HBVP cells is reflected in both the overall cell index curves ((d) arrow indicates drug application), as well as the slope (e) and doubling time (f) of cells treated with ET-1. Data are mean with SEM; n = 6; *p < 0.05, **p < 0.01 and ***p < 0.001 compared to vehicle controls.

In addition, the proliferative capacity of ET-1 on HBVP cells was tested by lifting the cells at 48 h post-drug and counting using the exclusion dye trypan blue and a standard haemocytometer. All cells were seeded at 10,000 per well. Vehicle or adenosine-treated cells grew to an average of 18,750 per well (±4787 viable, adherent cells) while ET-treated cells grew to an average of 40,000 per well (±4082 viable, adherent cells; data not represented graphically).

Endothelin-1-mediated pericyte contraction is primarily dependent on the ET_A receptor

To determine the mechanism by which ET-1 caused contraction and proliferation, we pre-treated the cells with BQ-123 and/or BQ-788, antagonists of ET-1 receptor subtypes A and B, respectively. While there was still a significant response to ET-1 in the presence of the antagonists (Figure 4(a)), we observed a reduction in slope at the acute time point with both antagonists; however, the diminution was only significant with ET_A blockade when compared to ET-1 alone (Figure 4(b); p < 0.0001). The combination of antagonists did not augment the response over and above the ET_A-alone blockade (Figure 4(b); p < 0.0001). Again, this pattern was reflected in calculation of delta (Figure 4(c); p < 0.0001). Similarly, the chronic effects of ET-1 were inhibited to a greater extent by BQ-123 (Figure 4(d)); although both antagonists significantly reduced chronic slope, the effect of ET_A antagonism was significantly greater than ET_B blockade (Figure 4(e); p < 0.0001). Quantified as doubling time, BQ-788 did not significantly counteract the effects of ET-1 whereas BQ-123 extended doubling time by approximately 25 h (Figure 4(f); p < 0.01). BQ drugs alone did not significantly affect proliferation (supplementary Figure S4).

Figure 4.

Electrical impedence measurement of HBVP contraction in response to endothelin-1 in the presence of specific antagonists. HBVP were grown on the iCelligence system for 48 h prior to treatment with either control medium, or medium containing ET-1 (50 nM). BQ-123 (10 µM) and BQ-788 (10 µM) were applied 1 h prior to ET-1 application and had little independent effect on cell index. Measurement during the first 30 min shows a rapid contraction and relaxation phase in response to ET-1 (a), with the initial slope reflecting this contraction (b). Calculation of delta (max–min difference at peak contractility) also reflects differences in response to ET-1 in the presence of antagonists (c). Full time course showing proliferation of HBVP cells in the presence of antagonists and in response to ET-1, is reflected in both the overall growth ((d) arrow indicates drug application), as well as the slope (e) and doubling time (f) of cells treated with ET-1. Data are mean with SEM; n = 4; *p < 0.05, **p < 0.01 and ***p < 0.001 compared to ET-1 alone.

Adenosine and sodium nitroprusside cause a dose dependent increase in pericyte relaxation

In vivo, contractile vascular mural cells have a basal tone to allow for bidirectional regulation of vessel diameter; therefore, it was important to validate whether we were able to detect relaxation in cultured pericytes. Adenosine, a known vasodilator, evoked a dose-dependent increase in cell index above vehicle baseline, indicative of increased surface area (Figure 5(a)). The difference in slope, when compared to vehicle controls, became apparent at doses above 10 µM (Figure 5(b); p < 0.0001 for 10 µM and 1 mM). In a similar manner to the vasoconstrictor, this was reflected in the delta calculations (Figure 5(c); p < 0.0001 for 10 µM and 1 mM). Similarly, the NO donor sodium nitroprusside (SNP) caused a dose-dependent increase in cell index (Figure 5(d)) where significant changes in slope were found only at the highest dose (Figure 5(d); p < 0.0001 100 µM). However, calculation of delta showed a significant effect at lower doses, and a more pronounced dose response, demonstrating the importance of data analysis when using this technique (Figure 5(f); p < 0.05 1 µM and p < 0.0001 100 µM). It should be noted that there were also differences in the temporal dynamics, as SNP took longer to exert a maximal effect compared to adenosine; as a consequence, the vehicle group had a positive sign in contrast to the negative sign seen with adenosine. However, neither adenosine nor SNP had any effect on the long-term growth and proliferation of HBVP cells (data not shown).

Figure 5.

Electrical impedance measurement of relaxation of HBVP in response to adenosine and sodium nitroprusside. HBVP were grown on the iCelligence system for 48 h prior to treatment with either control medium, or medium containing adenosine (100 nM, 10 µM and 1 mM) or sodium nitroprusside (10 nM, 1 µM and 100 µM). Measurement during the first 30 min shows a prolonged relaxation in response to adenosine (a), with the initial slope reflecting this relaxation (b). Calculation of delta (max–min difference at peak relaxation) also reflects differences in response to adenosine (c). Similarly, measurement during the first 30 min shows a prolonged relaxation in response to the SNP (d), with the initial slope reflecting this relaxation (e). Delta also reflects differences in response to SNP (f). Data are mean with SEM; n = 4–6; *p < 0.05, **p < 0.01 and ***p < 0.001 compared to vehicle controls.

Adenosine-mediated pericyte relaxation is dependent on the A₁ and A_2A receptors

We then aimed to confirm whether the effect of adenosine was dependent on A₁ and A_2A receptors, both of which are known to cause vasodilation in other contexts.²⁴ Pre-treatment with the A₁/A_2A antagonist CGS-15943 significantly reduced the effects of adenosine (Figure 6(a)), which was reflected in both calculations of slope (Figure 6(c)) and delta (Figure 6(c)): cells receiving adenosine had a significantly different response from vehicle-treated cells (p < 0.0001), while cells pre-treated with CGS-15943 were also significantly different from cells receiving adenosine alone (p < 0.01 for slope and p < 0.0001 for delta). Both contraction and relaxation in response to endothelin and adenosine were confirmed in live imaging experiments (supplementary Figure S5).

Figure 6.

Electrical impedence measurement of relaxation of HBVP in response to adenosine in the presence of an antagonist. HBVP cells were grown on the iCelligence system for 48 h prior to treatment with either control medium, or medium containing adenosine (1 mM). CGS-15493 (10 µM), an A₁/A_2A antagonist, was applied 1 h prior to adenosine application and had little independent effect on cell index. Measurement during the first 30 min shows a rapid relaxation in response to adenosine (a), with the initial slope reflecting this (b). Both the slope and cell index are significantly reduced in the presence of CGS-15943. Calculation of delta (max–min difference at peak contractility) also reflects differences in response to adenosine in the presence of antagonist (c). Data are mean with SEM; n = 4; *p < 0.05, **p < 0.01 and ***p < 0.001 compared to vehicle controls.

Chemical ischaemia results in significant pericyte contraction in the absence of cell death

Previous studies have shown that pericytes constrict following an ischaemic insult, and die in a state of rigour.¹³ To validate the physiological relevance of our model, we aimed to reproduce these results with iCelligence recordings. Application of a chemical ischaemia solution at high and medium doses resulted in a rapid decline in cell index within the first hour of treatment (Figure 7(a)), resulting in significant differences in both slope (Figure 7(b), p < 0.001 for both doses) and delta (Figure 7(c); p < 0.001 for both doses). The low-dose group showed no difference compared to vehicle during the first hour and, following a period of fluctuating values that exceeded vehicle values, underwent a decline to levels equivalent to high- and medium-dose groups. To confirm that this was constriction and not a consequence of cell death and/or detachment, we performed a LDH assay at 2 and 24 h, corresponding to time points where the high/medium doses but not the low dose had reached minimal cell index values (2 h), and where all three had reached minima (24 h). There was no detectable increase in cell death at 2 h, suggesting that the reduction in cell index reflects contraction. At 24 h, cell death had reached approximately 70% in the high- and medium-dose groups, the low-dose group had a non-significant increase (Figure 7(d); p < 0.001 for high and p < 0.01 for medium dose). DNA degradation, as a marker of apoptosis, was confirmed using TUNEL staining at 2 h post-treatment and showed no staining at low dose, minor staining at medium dose and significant staining at the high dose of chemical ischemia, indicating the beginning of cell death via apoptosis. The presence of well-stained nuclei (DAPI) suggests these cells were beginning to apoptose but had yet to fully die (supplementary Figure S6).

Figure 7.

Chemical ischaemia causes pericyte constriction in a dose-dependent manner followed by cell death. HBVP cells were grown on the iCelligence system for 48 h prior to treatment with either control medium, or medium containing various concentrations of sodium iodoacetate and antimycin A (high being 500 and 5 µM, medium being 50 and 0.5 µM, and low being 5 and 0.05 µM, respectively). High- and medium-dose chemical ischaemia resulted in a rapid decrease in cell index within 1 h, whereas the low dose showed a transient increase and then a decrease from approximately 8 h onwards (a). At 1 h, there were differences in both slope (b) and delta (c). Cell death did not differ between groups at 2 h, but by 24 h high- and medium-dose cell death had reached approximately 70% (d). Data are mean with SEM; n = 4; **p<0.01, ***p < 0.001 compared to vehicle controls.

Discussion

Measuring active cellular processes such as contractility in vitro is very challenging. It either requires very specialised equipment, or only gives discrete, quantal data, rather than an overview of what is, in essence, a continual fluid process. In this study, we have used iCelligence, an electrical impedance measurement system, to show that it is possible to measure both contractility and relaxation in response to a number of vasoactive compounds in HBVP. Our data demonstrate, for the first time, that it is feasible to use electrical impedance to measure both rapid pericyte responses to drugs, as well as more prolonged changes in pericyte growth and proliferation. These data provide considerable and significant scope for the study of pericytes in vitro.

The use of electrical impedance to quantify contractility in a label-free way in rapidly contracting cells such as cardiomyocytes, has enabled high throughput screening of cardiotoxic compounds in vitro.^22,25 By plating cells on a field of microelectrodes, they act as resistors in a circuit, providing feedback information on changes in both growth and morphology, as coverage of the field changes. This allows for dynamic monitoring of cells in response to changes in the cellular microenvironment. In this study, we aimed to determine whether this was a feasible system to study pericyte contractility in vitro. The rapid constriction and death of pericytes in response to CNS injuries such as stroke,^10,13 renders them a viable target for modulation of blood flow post-infarct.

Our data clearly demonstrate that pericytes both contract and relax in response to vasoactive compounds. By measuring cell index, as a surrogate of morphology and growth, we have effectively demonstrated rapid contraction in response to the vasoactive peptide ET-1. Previous studies have shown that pericytes are capable of responding to ET-1^26,27 by directly measuring cell size using either light microscopy or using basic membrane dyes. However, as the responses in control cells here demonstrate, pericytes are sensitive cells and can contract rapidly to changes in their microenvironment, even small changes such as the addition of medium. Planimetric studies taking photos directly of live cells would need to take this into account. Cells grown on silicone substrata for ‘wrinkle assays’²⁸ suffer from similar issues. Even at micron thickness, a silicone matrix is likely to prove a considerable barrier to active contraction, and may affect cell physiology. The system used in this study therefore provides superior feedback on contractile status both in living cells, and in a rapid and physiologically relevant manner.

Furthermore, the long-term effects of ET-1 on cell growth have been observed in the same cells in which contractility was observed. This is a strategy previously unavailable to those studying pericytes, where contractility was measured using one of the options from above, and cell proliferation was measured using a separate generic assay format. The choice of ET-1 as the main agent used for validation here was based not only on its potent vasoconstrictive actions in vivo but also the prominent positive effect it has on pericyte proliferation.²⁹ It is known that pericytes express both ET_A and ET_B receptors,³⁰ but ET_A receptors are more prominently associated with vasoconstriction,³¹ whereas ET_B receptors have a more modulatory influence.³⁰ Our findings are consistent with these reports, showing that while both receptors contribute to the effects of ET-1 on brain pericytes, ET_A antagonism with BQ-123 inhibits both the constrictive and proliferative aspects more potently than the ET_B antagonist BQ-788.

Relaxation is considerably more challenging to study in pericytes, but vital if they are to be targeted therapeutically for phenomena such as no-reflow post-stroke.³² Compounds such as adenosine³³ and nitric oxide donors³⁴ are known to relax pericytes. However, here the only options thus far have been either time lapse imaging or planimetric analysis of static images, once again providing only quantal data with resolution only as rapid as the image acquisition interval. Our electrical impedance data demonstrate that pericytes relax in response to both adenosine and sodium nitroprusside. The overall differences in acute morphological changes to vasodilators and vasoconstrictors are evident in these experiments. The immediate response to ET-1 is rapid and transient, lasting no more than an hour, whereas the dilatory effects of adenosine have a similar time of onset, but last for considerably longer. These acute effects of adenosine were inhibited by CGS-15943, an A₁/A_2A antagonist. While the use of this promiscuous antagonist does not allow us to definitively state which receptor subtypes are key for adenosine-mediated relaxation, it nonetheless validates our use of adenosine as a vasoactive agent causing receptor-dependent changes in pericyte tone. Nitric oxide (NO) is a potent vasodilator, synthesised in the endothelium in response to a number of vasoactive signals such as bradykinin and shear stress.³⁵ Intracellularly, NO activates soluble guanylyl cyclase, resulting in cyclic GMP production and a variety of protein kinase-dependent downstream effects, the ultimate effect of which in smooth muscle cells and pericytes is hyperpolarisation and a reduction in calcium influx.³⁶ SNP, an NO donor, was able to dose-dependently evoke a relaxation in the cultured pericytes in a similar manner to adenosine. Surprisingly, this occurred on a slightly longer time scale than the effects of adenosine or ET-1, which might be attributable to differences in direct receptor signalling vs downstream second messenger signalling. Furthermore, while ET had significant proliferative effects on pericytes in the long term, there was no overall change after the addition of vasodilators on chronic measurements of cell index. However, little data exist to support the mechanistic basis of this, and it lies beyond the scope of this methodological study.

Finally, we have used chemical ischaemia to demonstrate that in culture, pericytes contract, but do not die in significant numbers, under conditions of oxidative stress. Both ex vivo¹³ and in vivo³⁷ studies have shown that pericytes constrict in response to ischaemia. Here, we have demonstrated contraction in response to chemical ischaemia that is both time- and dose-dependent. High-dose iodoacetate and antimycin-A cause rapid constriction, but with little cell death, followed (with a slight delay) by constriction in medium dose-treated cells, followed finally by constriction between 12 and 18 h in cells receiving low-dose chemical ischaemia. However, it is important to note that these cells remain viable, as there is comparatively less cell death in the low-dose group even at 24 h, when contraction is at peak values, suggesting that pericytes constrict and then die in a state of rigour as previously seen in acute brain slices.¹³ To our knowledge, this is the first time this has been demonstrated in cultured pericytes and provides a significant tool for the investigation of pericyte contractility during stroke.

In conclusion, we have demonstrated for the first time the use of a rapid, real-time system to study pericyte contractility in vitro. Our data show that HBVP are capable of responding to vasoactive compounds in both a contractile and dilatory capacity, and that it is possible to measure these responses in concert with chronic measurements of growth and proliferation. This study has the potential to significantly affect the manner in which pericytes are studied in vitro in the future and provides an excellent platform for investigating their response to multiple changes in their cellular microenvironment.

Funding

The author(s) were supported by funding from the Medical Research Council (YC, BAS, AMB), the Medical Research Fund of the University of Oxford (YC), Oxford Biomedical Research Centre (AMB), and a Radcliffe Department of Medicine Scholarship (AAN).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

Conceived and designed experiments: AAN, YC. Performed experiments: AAN, YC. Analysed Data: AAN, YC. Contributed reagents/materials/analysis tools: YC, AMB. Wrote and edited the manuscript: AAN, YC, BAS, AMB.

Supplementary material

Supplementary material for this paper can be found at http://jcbfm.sagepub.com/content/by/supplemental-data