- Cx40ko mice
- D max
- D 0
light dye treatment
N ω-nitro-l-arginine methyl ester
myoendothelial gap junction
smooth muscle cell
- • Regulation of blood flow in microcirculatory networks depends on spread of local vasodilatation to upstream supply arteries.
- • This is mediated by endothelial conduction of hyperpolarization, attenuation of which is expected to occur through current restriction or loss at sites of cell coupling and open ion channels in cell membranes.
- • In an animal model of hypertension, we found that hyperpolarization decays more rapidly when endothelial cell coupling is reduced; however, this could not fully explain the observed attenuation in conducted vasodilatation.
- • We found that increased oxidative stress, due to upregulation of angiotensin II, changed the contribution of L- and T-type calcium channels to resting vessel tone. Inhibition of oxidative stress reversed this change and improved conducted vasodilatation.
- • Our data suggest that cardiovascular disease may impair the ability of microvascular networks to maintain tissue integrity, due to oxidative stress-induced changes in the way blood vessels constrict.
Abstract Regulation of blood flow in microcirculatory networks depends on spread of local vasodilatation to encompass upstream arteries; a process mediated by endothelial conduction of hyperpolarization. Given that endothelial coupling is reduced in hypertension, we used hypertensive Cx40ko mice, in which endothelial coupling is attenuated, to investigate the contribution of the renin–angiotensin system and reduced endothelial cell coupling to conducted vasodilatation of cremaster arterioles in vivo. When the endothelium was disrupted by light dye treatment, conducted vasodilatation, following ionophoresis of acetylcholine, was abolished beyond the site of endothelial damage. In the absence of Cx40, sparse immunohistochemical staining was found for Cx37 in the endothelium, and endothelial, myoendothelial and smooth muscle gap junctions were identified by electron microscopy. Hyperpolarization decayed more rapidly in arterioles from Cx40ko than wild-type mice. This was accompanied by a shift in the threshold potential defining the linear relationship between voltage and diameter, increased T-type calcium channel expression and increased contribution of T-type (3 μmol l−1 NNC 55-0396), relative to L-type (1 μmol l−1 nifedipine), channels to vascular tone. The change in electromechanical coupling was reversed by inhibition of the renin–angiotensin system (candesartan, 1.0 mg kg−1 day−1 for 2 weeks) or by acute treatment with the superoxide scavenger tempol (1 mmol l−1). Candesartan and tempol treatments also significantly improved conducted vasodilatation. We conclude that conducted vasodilatation in Cx40ko mice requires the endothelium, and attenuation results from both a reduction in endothelial coupling and an angiotensin II-induced increase in oxidative stress. We suggest that during cardiovascular disease, the ability of microvascular networks to maintain tissue integrity may be compromised due to oxidative stress-induced changes in electromechanical coupling.
Rapid increases in blood flow to meet the demands of metabolically active tissues require the co-ordinated vasodilatation of both local capillary–arteriolar networks and upstream supply arteries (Bagher & Segal, 2011). Extensive electrophysiological studies of this phenomenon both in vivo and in vitro have convincingly demonstrated that the underlying mechanism is electromechanical (Bagher & Segal, 2011); hyperpolarization is conducted along the vascular wall through the endothelium to proximal supply arteries, due to abundant cell coupling within this layer (Hill et al. 2001a; Sandow, 2004). Transfer of current into the less well-coupled smooth muscle cells (SMCs), via the prominent myoendothelial coupling exhibited by resistance arteries and arterioles (Hill et al. 2001a; Sandow, 2004), produces hyperpolarization, closure of voltage-dependent calcium channels and synchronous vasodilatation along the vessel wall (Bagher & Segal, 2011).
During hypertension, the endothelial gap junctional proteins connexin37 (Cx37) and connexin40 (Cx40) are significantly reduced in an artery-specific fashion (Rummery et al. 2002b, 2005; Goto et al. 2004; Kansui et al. 2004; Rummery & Hill, 2004), and this is reversed by inhibition of the renin–angiotensin system (Kansui et al. 2004; Rummery et al. 2005). A critical role for endothelial Cx40 in the conduction of vasodilatation has been shown by significant impairment of this phenomenon in arterioles of Cx40-deficient (Cx40ko) mice (de Wit et al. 2000). However, these mice are also profoundly hypertensive, due to ectopic localization of renin-secreting cells and dysregulation of the pressure-dependent control of renin secretion (Krattinger et al. 2007; Wagner et al. 2007). Moreover, Cx37 expression is co-ordinately reduced at the sites of Cx40 deletion (Simon & McWhorter, 2003; Krattinger et al. 2007; de Wit, 2010). As a result of the apparent loss of all gap junctional coupling from the endothelium of microcirculatory vessels of Cx40ko mice, it has recently been suggested that conducted vasodilatation is attenuated due to alternative conduction of hyperpolarization through the perpendicularly oriented SMCs (de Wit, 2010). However, no measurements of the underlying electrical events have been made in the cremaster circulation of Cx40ko mice, and the distance over which vasodilatation is conducted exceeds that predicted from computational modelling of electrical responses conducted through the high-resistance smooth muscle cell layer (Diep et al. 2005).
The aims of this study were therefore 3-fold: (i) to determine whether the endothelium is essential to the conduction of vasodilatation in arterioles of Cx40ko mice; (ii) to measure membrane potential and diameter simultaneously, to determine whether the attenuation in conducted vasodilatation in Cx40ko mice can be explained by a steeper decline in the amplitude of the underlying hyperpolarization; and (iii) to determine whether overactivation of the renin–angiotensin system contributes to the attenuation of conducted vasodilatation found in arterioles of hypertensive Cx40ko mice.
All experiments were conducted on male Cx40ko mice (breeding pairs kindly supplied by Dr A. M. Simon; back crossed for 8 generations to C57BL/6 wild-type mice) and age- and strain-matched C57BL/6 wild-type mice, aged 2–3 months, according to protocols approved by the Australian National University Animal Experimentation Ethics Committee, under guidelines of the National Health and Medical Research Council of Australia (NHMRC).
Some mice were treated for 14–18 days with the angiotensin II receptor antagonist, candesartan (1.0 mg kg−1 day−1 in the drinking water). Blood pressure measurements were made using tail-cuff plethysmography.
Conducted hyperpolarization and vasodilatation in cremaster arterioles in vivo
Conducted vasodilatation was studied in second- or third-order cremaster arterioles of mice anaesthetized intraperitoneally (1 mg kg−1 medetomidine and 10 mg kg−1 midazolam from Pfizer Australia Pty Ltd, West Ryde, NSW, Australia; and 0.1 mg kg−1 fentanyl from Mayne Pharma Ltd, Salisbury, South Australia, Australia) and continuously infused via a jugular vein catheter (medetomidine 0.02 mg h−1; midazolam 0.2 mg h−1; fentanyl 0.002 mg h−1). The right cremaster muscle was spread over a coverslip to enable visualization of arterioles and superfused (3 ml min−1) with Krebs solution (in mmol l−1: NaCl, 118.4; KCl, 3.8; NaHCO3, 25; KH2PO4, 1.2; CaCl2.2H2O, 2.5; and MgSO4.7H2O, 1.2), gassed with 5% CO2 in N2, pH 7.4, at 34°C. l-NAME (10 μmol l−1) and indomethacin (10 μmol l−1) were present in all experiments to block nitric oxide and prostaglandins, respectively. These endogenous vasodilators contribute to the local, but not the conducted, mechanical response, because they do not affect membrane potential (Wölfle et al. 2011).
Conducted dilatations were studied after local stimulation with acetylcholine (ACh; 1 mol l−1) via ionophoresis from a micropipette (tip diameter, 1 μm; 500 nA, 1 s), positioned adjacent to the vessel surface; a retaining current (−200 nA) was applied to prevent continuous release of ACh from the pipette. Membrane potential recordings were made with intracellular microelectrodes (140–200 MΩ; Axoclamp 2B; Molecular Devices, Sunnyvale, CA, USA) from dye-identified SMCs or endothelial cells (ECs; 2% fluorescein in 0.5 mol l−1 KCl), while mechanical responses were measured with a line-detection program (Neild, 1989). Diameter and membrane potential were recorded simultaneously at the site of ACh application. The intracellular microelectrode was then moved to a remote site, located 1000 μm upstream, where additional cells were impaled, and measurements of diameter and membrane potential were repeated after new local stimulation without movement of the ionophoretic pipette, as we have done previously (Wölfle et al. 2011). In a separate series of experiments, conducted dilatations were studied before and after acute inhibition (30 min) of reactive oxygen species with the superoxide dismutase mimetic 4-hydroxy-TEMPOL (tempol; 1 mmol l−1; Sigma). Only vessels without intervening side-branches were studied. Mice were killed at the end of the experiments by cervical dislocation while still deeply anaesthetized.
Role of L- and T-type voltage-dependent calcium channels in arteriolar tone in vivo
The relative importance of L- and T-type voltage-dependent calcium channels was determined using nifedipine (1 μmol l−1) and the T-type antagonist, NNC 55-0396 (3 μmol l−1; Li et al. 2005; Handforth et al. 2010; Kuo et al. 2010; Quesada et al. 2011; Tzeng et al. 2012), respectively, before and after acute inhibition (30 min) of reactive oxygen species with the NADPH oxidase inhibitor 4′-hydroxy-3′-methoxyacetophenone (apocynin; 500 μmol l−1; Sigma) or tempol (1 mmol l−1). Given that we have previously found that NNC 55-0396 can block nifedipine-sensitive and nifedipine-insensitive voltage-dependent calcium currents in vascular smooth muscle cells, we always added NNC 55-0396 after nifedipine to prevent any non-specific action on L-type channels (Kuo et al. 2010). The concentration of 3 μmol l−1 of NNC 55-0396 was chosen because higher concentrations did not provide any additional inhibition of vascular tone (Kuo et al. 2010). Maximal vessel diameter was determined at the end of experimentation by superfusion of Krebs solution containing sodium nitroprusside (SNP), ACh and adenosine (30 μmol l−1 each). Mice were killed by cervical dislocation without regaining consciousness.
Vascular tone was calculated as 100 × (Dmax – D0)/Dmax, where D0 represents resting diameter and Dmax is the maximal diameter. The amplitude of ACh-induced dilatation (D – D0) was normalized to dilator capacity (Dmax – D0), such that the percentage of maximal dilatation = 100 × (D – D0)/(Dmax – D0), in order to account for variability in the diameter and resting tone of individual arterioles and preparations. The percentage contribution of L-type channels to vascular tone was determined as 100 × (Dnifedipine – D0)/(Dmax – D0), while the percentage contribution of T-type channels to vascular tone was determined by 100 × (DNNC+nifedine – Dnifedipine)/(Dmax – D0).
Endothelial interruption with light dye treatment
In order to determine the cellular pathway for the conducted response, the endothelium was destroyed at the 1000 μm site, following i.v. injection of 1% (w/v) sodium fluorescein in saline (100 μl) via the jugular vein catheter and exposure to 488 nm light to generate reactive oxygen species at the 1000 μm site (Martin & Logsdon, 1987; Emerson & Segal, 2000). Success of the procedure was determined in each vessel by loss of endothelium-dependent dilatation in response to ACh, while lack of damage to the smooth muscle was confirmed by an intact dilatation in response to SNP. Selectivity of endothelial destruction was assessed at the end of each experiment by staining nuclei with the cell death marker, propidium iodide (0.1%). Vessel responses to ionophoresis of ACh were measured at the local site and at 500, 1000 (site of endothelial damage) and 1500 μm upstream in the same vessel.
Plasma renin concentration was determined by radioimmunoassay (ProSearch, Malvern, Victoria, Australia) in individual blood samples from mice anaesthetized with isoflurane.
Cremaster arterioles were isolated from mice (n = 3) anaesthetized intraperitoneally (1 mg kg−1 medetomidine; 10 mg kg−1 midazolam; 0.1 mg kg−1 fentanyl) and killed by cervical dislocation while still deeply anaesthetized. Arterioles were processed for electron microscopy, as previously described (Sandow & Hill, 2000). Myoendothelial gap junctions (MEGJs) were identified by the presence of pentalaminar membrane between endothelium and smooth muscle, and quantified after examination of the internal elastic lamina in each section (100 nm) of a transverse series taken over 5μm. Gap junctions between ECs and between SMCs were identified, and general morphological characteristics were measured in individual sections.
Wild-type and Cx40ko mice were anaesthetized with ketamine and xylazine (120 and 16 mg kg−1 i.p., respectively; Troy Laboratories Pty Ltd, Glendenning, NSW, Australia), and the cremaster muscle was removed, pinned flat and fixed in 2% paraformaldehyde. Mice were killed by cervical dislocation while still deeply anaesthetized. After blocking for 2 h in 1% bovine serum albumin–0.2% Triton X-100 in PBS, arteriolar whole mounts were incubated with sheep anti-Cx40 (1:100 dilution), sheep anti-Cx37 (1:100 dilution; Rummery et al. 2002a, 2005) or rabbit anti-Cx43 (Molecular Probes, Invitrogen Life Technologies Australia Pty Ltd, Mulgrave, Victoria, Australia), overnight at room temperature; or with rabbit anti-Cav1.2 (1:300 dilution; Jerusalem, Alomone, Israel), anti-CaV3.1 (1:500 dilution) or anti-CaV3.2 (1:300 dilution; kindly provided by Dr Leanne Cribbs, Loyola University, Chicago, IL, USA) for 48 h at 4°C. Specificity of Cav3.1 and Cav3.2 antibodies has been demonstrated previously by Western blotting, heterologous cell transfection and neutralization with immunizing peptides (Rodman et al. 2005; Kuo et al. 2010). Control preparations were incubated without the primary antibody.
Staining was visualized with Alexa 568-conjugated donkey anti-sheep or anti-rabbit IgG (1:500 dilution; Molecular Probes Invitrogen) or Alexa 488-conjugated donkey anti-rabbit IgG (1:800 dilution; Molecular Probes Invitrogen). Nuclei were stained with 15 μmol l−1 4′,6-Diamidino-2-Phenylindole (DAPI) or 0.01% propidium iodide, and preparations were mounted in buffered glycerol. Image series were collected at 0.3 μm intervals with a Leica SP2 or Zeiss LSM PASCAL 5 confocal microscope and recombined into a single image through the smooth muscle or through the endothelium. To enable comparisons of staining between different samples stained with the same antibody, the same pinhole and gain settings were used for the acquisition of all images. ImageJ (NIH; http://rsb.info.nih.gov/ij) was used to measure EC dimensions or to determine the area of SMCs occupied by Cav staining, using the thresholding function.
Experimental data and morphological parameters obtained from direct ultrastructural or immunohistochemical measurements were used in a computational model of the vessel wall, which consisted of a 10-mm-long segment of longitudinally oriented ECs surrounded by a single layer of SMCs, as described previously (Diep et al. 2005; Wölfle et al. 2011). Focal application of ACh was simulated by applying a voltage clamp to a segment of vessel 104 μm long, at one end of the modelled vessel, because this was a realistic representation of the spread of ACh from the ionophoretic electrode (Goto et al. 2004). The experimentally determined coupling resistance of 90 MΩ was used for SMCs (Yamamoto et al. 2001), while the myoendothelial coupling resistance of 1350 MΩ was calculated as described previously from the data of Yamamoto et al. (2001). Endothelial coupling resistances were adjusted for the best fit to the data (wild-type, 4 MΩ; Wölfle et al. 2011; Cx40ko, 13 MΩ), while membrane capacitance was set to 1 μF cm−2.
When a threshold potential was defined, a linear relationship existed between dilatation and hyperpolarization to this value, after which the dilatation was maximal.
Drugs and solutions
All pharmacological compounds were obtained from Sigma (Sigma-Aldrich Pty Ltd, Castle Hill, NSW, Australia), except for NNC 55-0396, which was from Tocris (Tocris Bioscience, Ellisville, MO, USA). Stock solutions were prepared in distilled water, with further dilution in Krebs solution (minimum, 1:1000), with the exception of indomethacin (DMSO) and nifedipine (ethanol).
Data are expressed as means ± SEM, where n is defined in the text or figure legend. Statistical significance was determined by one-way ANOVA followed by Bonferonni post hoc test for multiple comparisons. Comparisons of two variables were performed with Student's paired or unpaired t tests, as appropriate, with P < 0.05 considered statistically significant.
Membrane potential threshold was estimated by calculating the breakpoint using two-line linear regression and forcing one horizontal line in Genstat ver13.0 (VSN International Ltd, Hemel, Hempstead, UK), as previously (Wölfle et al. 2011).
Cell coupling is reduced in cremaster arterioles of Cx40ko mice
Cremaster arterioles of Cx40ko mice, like those of wild-type mice (Wölfle et al. 2011), comprised an endothelial layer surrounded by a single layer of SMCs (Fig. 1A; diameter, 33 ± 3.5 μm; n = 3). Serial section electron microscopy revealed that the two cell layers were well coupled by gap junctions containing pentalaminar membranes (Fig. 1B and inset). Gap junctions were also found amongst the ECs (Fig. 1D–G) and less frequently amongst the SMCs (Fig. 1C).
As expected, Cx40 was abundantly expressed in the endothelium of cremaster arterioles of wild-type mice, but was absent from vessels of Cx40ko mice (Fig. 1H and I). Connexin37 was also highly expressed between adjacent ECs in arterioles of wild-type mice, but the staining intensity of Cx37 was much less in arterioles from Cx40ko mice (Fig. 1J and K). Connexin37 was also weakly expressed in SMCs of both wild-type and Cx40ko arterioles. Connexin43 was not observed in either wild-type or Cx40ko arterioles (Fig. 1L and M). Endothelial cells were significantly smaller in length and area in arterioles from Cx40ko than wild-type mice (P < 0.05; Cx40ko length, 84 ± 5 μm and area, 458 ± 18 μm2, n = 60 cells from 7 mice: wild-type length, 104 ± 4 μm, and area, 541 ± 17 μm2, n = 61 cells from 7 mice; Wölfle et al. 2011). The number of MEGJs/EC in Cx40ko arterioles was 4.4 ± 0.5 (n = 3), compared with 10.9 ± 0.9 MEGJs/EC previously reported in arterioles from wild-type mice (n = 5; Wölfle et al. 2011).
Conduction of vasodilatation occurs through the endothelium in arterioles of Cx40ko mice
Light dye treatment (LDT) was applied to a section of vessel, located 1000 μm from the site of ACh application, following infusion of fluorescein into the jugular vein (Fig. 2A). Damage to the endothelium was confirmed by failure of the endothelial specific dilator, ACh (10 μmol l−1), to induce relaxation at the treatment site (treat) after light exposure (Fig. 2C), while dilatation was not altered at an upstream control site (con) that was not directly exposed to the excitation wavelength (Fig. 2C). Absence of damage to the smooth muscle layer was confirmed by intact dilatation in response to the smooth muscle specific dilator, SNP, before and after LDT at both control and treatment sites (Fig. 2D). Successful and selective destruction of the endothelium was further shown by uptake of propidium iodide into EC, but not SMC, nuclei (Fig. 2E and F).
Conduction of vasodilatation through the endothelium was demonstrated by the failure of vasodilatation to spread beyond 1000 μm, following selective endothelial damage by LDT at this site (Fig. 2B; n = 5 vessels in 5 mice; see Table 1 for absolute values).
|Figure||Site (μm)||Dilatation (μm)||Resting diameter (μm)||Maximal diameter (μm)|
|Cx40ko: pre-LDT||0||8 ± 0.4||16 ± 3.0||34 ± 2.8|
|(n = 5 vessels in 5 mice)||500||7 ± 0.4||15 ± 2.0||34 ± 2.2|
|1000||5 ± 0.5||16 ± 1.5||34 ± 2.2|
|1500||4 ± 0.4||18 ± 1.0||35 ± 2.3|
|Cx40ko: post-LDT||0||9 ± 0.4||15 ± 2.0||35 ± 2.3|
|(n = 5 vessels in 5 mice)||500||7 ± 0.5||16 ± 1.1||36 ± 2.3|
|1000||0 ± 0.2||14 ± 1.6||35 ± 2.3|
|1500||0 ± 0.1||17 ± 1.1||36 ± 2.0|
|Wild-type||0||8 ± 0.4||13 ± 1.0||31 ± 1.6|
|(n = 6 vessels in 6 mice)||500||8 ± 0.2||15 ± 1.5||31 ± 1.6|
|1000||7 ± 0.3||17 ± 1.2||31 ± 1.6|
|1500||6 ± 0.4||18 ± 0.7||31 ± 1.6|
|Cx40ko||0||8 ± 0.6||14 ± 1.5||31 ± 1.5|
|(n = 7 vessels in 7 mice)||500||6 ± 0.6||16 ± 1.5||31 ± 1.5|
|1000||5 ± 0.5||16 ± 1.6||31 ± 1.5|
|1500||3 ± 0.6||19 ± 1.9||31 ± 1.5|
|Wild-type||0||8 ± 0.5||13 ± 1.1||29 ± 1.2|
|(n = 7 vessels in 7 mice)||1000||7 ± 0.2||14 ± 1.3||29 ± 1.2|
|Cx40ko||0||9 ± 1.9||13 ± 1.3||30 ± 2.1|
|(n = 4 vessels in 4 mice)||1000||5 ± 1.2||14 ± 1.8||30 ± 2.1|
|Wild-type||0||11 ± 0.7||20 ± 2.7||48 ± 5.3|
|(n = 5 vessels in 5 mice)||500||12 ± 1.2||21 ± 2.8||48 ± 5.4|
|1000||12 ± 1.6||20 ± 3.1||48 ± 5.4|
|1500||10 ± 0.5||19 ± 3.4||48 ± 5.4|
|Wild-type + tempol||0||9 ± 1.3||26 ± 5.3||48 ± 5.3|
|(n = 5 vessels in 5 mice)||500||10 ± 1.5||25 ± 5.5||48 ± 5.4|
|1000||10 ± 1.5||25 ± 5.8||48 ± 5.4|
|1500||10 ± 2.1||24 ± 5.7||48 ± 5.4|
|Cx40ko||0||7 ± 0.4||12 ± 1.7||33 ± 2.5|
|(n = 6 vessels in 6 mice)||500||6 ± 0.5||13 ± 1.9||33 ± 2.3|
|1000||4 ± 0.4||13 ± 1.9||33 ± 2.3|
|1500||2 ± 0.4||13 ± 1.7||33 ± 2.5|
|Cx40ko + tempol||0||5 ± 0.9||17 ± 2.6||33 ± 2.5|
|(n = 6 vessels in 6 mice)||500||5 ± 0.8||18 ± 2.2||33 ± 2.3|
|1000||5 ± 0.9||19 ± 2.4||33 ± 2.3|
|1500||4 ± 0.8||19 ± 2.2||33 ± 2.5|
|Cx40ko + candesartan||0||10 ± 1.6||18 ± 1.9||37 ± 2.9|
|(n = 7 vessels in 7 mice)||500||9 ± 1.3||16 ± 1.8||37 ± 2.9|
|1000||8 ± 1.0||16 ± 2.1||37 ± 2.9|
|1500||6 ± 0.7||16 ± 2.0||37 ± 2.9|
|Cx40ko + candesartan + tempol||0||7 ± 0.9||22 ± 2.0||38 ± 3.4|
|(n = 6 vessels in 6 mice)||500||6 ± 0.7||22 ± 2.2||38 ± 3.4|
|1000||7 ± 1.1||22 ± 2.3||38 ± 3.4|
|1500||5 ± 0.7||22 ± 2.9||38 ± 3.4|
Hyperpolarization decays more rapidly in arterioles from Cx40ko mice than wild-type mice
Conducted vasodilatation attenuated significantly with distance in arterioles from Cx40ko mice compared with arterioles from wild-type mice (Fig. 3A; P < 0.05; see Table 1 for absolute values) as described previously (de Wit et al. 2000). At the local site, there was no significant difference between the genotypes in the amplitude of the hyperpolarization or the vasodilatation evoked by ACh (Fig. 3A; Supplementary Fig. 1A), nor was there any difference in the resting tone of the vessels or the resting membrane potential of either SMCs or ECs (Table 2). The amplitude of the hyperpolarization recorded at the local site was also not different between the SMCs and ECs of either genotype (Table 2; Supplementary Fig. 1A).
|Genotype||Diameter (μm)||Resting membrane potential (mV)||ACh-induced hyperpolarization (mV)|
|Wild-type||15 ± 1||32 ± 2||−29 ± 0.3||−28 ± 0.2||−28 ± 0.2||−20 ± 3||−18 ± 3||−19 ± 2|
|Cx40ko||14 ± 2||29 ± 1||−29 ± 1||−29 ± 1||−29 ± 1||−19 ± 4||−23 ± 3||−22 ± 2|
In contrast, at the remote site (1000 μm), the hyperpolarization was significantly smaller in the Cx40ko arterioles than in the wild-type arterioles (P < 0.05; Supplementary Fig. 1B). As there was no significant difference between the amplitude of the hyperpolarizations measured in the SMCs and ECs at the remote site in either genotype (Supplementary Fig. 1B), data for the two cell types were pooled in subsequent paired recording experiments.
In order to confirm that the results did not arise due to variation in the location of the ionophoretic pipette from the vessel surface in different vessels, paired recordings of membrane potential and diameter were made in the same vessels (Fig. 3B–D). This was accomplished without movement of the ionophoretic pipette, by impaling cells sequentially at the local site and then at the remote site, after additional local ACh stimulation. At the local site, hyperpolarization and dilatation were again not significantly different between arterioles from wild-type and Cx40ko mice (Fig. 3B–D; wild-type, local, −20 ± 2 mV, 8 ± 1 μm; and Cx40ko, local, −20 ± 4 mV, 9 ± 2 μm). However, at the remote site, hyperpolarization was significantly smaller in arterioles from Cx40ko mice than wild-type mice (Fig. 3C; wild-type, 1000 μm, −12 ± 1 mV; and Cx40ko, 1000 μm, −7 ± 1 mV), while vasodilatation decayed significantly in the arterioles from only Cx40ko mice (Fig. 3D and Table 1; wild-type, 1000 μm, 7 ± 1 μm; and Cx40ko, 1000 μm, 5 ± 1 μm).
Relationship between membrane potential and vascular tone is non-linear in arterioles from Cx40ko mice
The greater rate of attenuation of hyperpolarization observed in the Cx40ko arterioles compared to wild-type arterioles (Fig. 3C) was consistent with reduction in EC coupling (Fig. 1) and a linear relationship between membrane potential and vascular tone. In contrast, our previous studies, in arterioles of wild-type mice, identified a non-linear relationship between membrane potential and vascular tone, which facilitated the long-distance conduction of vasodilatation (Wölfle et al. 2011). A threshold potential of −35 ± 1 mV, beyond which vasodilatation was maximal, was obtained previously in wild-type mice (Fig. 4B, black lines; Wölfle et al. 2011).
In order to determine the relationship between membrane potential and diameter in arterioles from Cx40ko mice, we plotted all paired values of membrane potential and diameter obtained from simultaneous recordings made in both SMCs and ECs at the local site, before and after ionophoretic application of ACh (Fig. 4A; n = 60 recordings from 30 cells in 7 mice). Using two-line linear regression, membrane potential and vessel tone were linearly related for membrane potentials more depolarized than −45 ± 3 mV, while hyperpolarization beyond −45 mV did not result in any additional relaxation (Fig. 4A and B, grey lines). Given that the hyperpolarization at the local site was large enough (approximately −20 mV) to drive the membrane potential to a value more negative than the threshold potential, the vasodilatation at the local site was maximal.
In order to test whether application of this threshold potential could predict the attenuation of vasodilatation observed in the Cx40ko arterioles, we used a computational model of the vascular wall (Wölfle et al. 2011), applying the anatomical and electrophysiological parameters that we had measured in cremaster arterioles of Cx40ko mice (Supplementary Table 1). More ECs were present along the length of the model segment due to the significant reduction measured in endothelial cell length, and this contributed to the decline in the hyperpolarization with distance, as we have shown previously (Wölfle et al. 2011). Endothelial coupling resistance was estimated at 13 MΩ to provide the best fit for the decline in hyperpolarization measured between the local and distal sites (Fig. 3C, Cx40ko). Application of the −45 mV threshold potential to convert these data to vasodilatation mimicked the observed attenuation of conducted vasodilatation (Fig. 4D, grey dashed line; cf. Fig. 3A and D, Cx40ko). Application of the threshold obtained previously from arterioles of wild-type mice (−35 mV) could not predict the observed decay of vasodilatation (Fig. 4D, black line).
Contribution of T-type voltage-dependent calcium channels to vascular tone is increased in arterioles from Cx40ko mice
The changed threshold potential in arterioles from Cx40ko mice suggested an alteration in electromechanical coupling. To test this possibility, we determined the contribution of L- and T-type calcium channels to vascular tone, because such a change has been described previously in pathophysiological conditions (see Kuo et al. 2011). Although vascular tone was not significantly different (Fig. 5A), the percentage contribution of T-type channels was significantly increased in arterioles of Cx40ko mice compared with wild-type mice (Fig. 5B; L-type, 1 μmol l−1 nifedipine; and T-type, 3 μmol l−1 NNC 55-0396), due to a significant decrease in the non-L-/non-T-type component compared with wild-type mice (Fig. 5B).
Expression of T-type calcium channels is increased in arterioles from Cx40ko mice
Using immunohistochemistry, we found that protein expression of the T-type channels, Cav3.1 and Cav3.2, in the vascular smooth muscle was significantly increased in arterioles of Cx40ko mice compared with wild-type mice (Fig. 6; group data in Fig. 7D). No staining was observed in the absence of primary antibody.
Blockade of angiotensin II decreases T-type calcium channel expression and function in Cx40ko mice
Consistent with previous studies showing that hypertension of Cx40ko mice resulted from abnormal development and control of renin secretion (Krattinger et al. 2007; Wagner et al. 2007), plasma renin concentration was significantly higher in Cx40ko mice than in wild-type mice (Cx40ko, 910 ± 117 ng angiotensin I ml−1 h−1, n = 10; and wild-type, 234 ± 22 ng angiotensin I ml−1 h−1, n = 10; P < 0.05), and treatment of Cx40ko mice with the angiotensin AT1 receptor antagonist, candesartan (1.0 mg kg−1 day−1 for 2 weeks) significantly reduced blood pressure (Cx40ko + candesartan, 107 ± 1.5 mmHg, n = 4; and Cx40ko, 129 ± 1.4 mmHg, n = 5; P < 0.05).
Vascular tone was not significantly different amongst arterioles from candesartan-treated wild-type, candesartan-treated Cx40ko (Fig. 5C) and untreated Cx40ko mice, although candesartan treatment did reduce tone of Cx40ko arterioles compared with untreated wild-type vessels (Fig. 5A and C). The relative contribution of T-type channels to vascular tone of Cx40ko arterioles after candesartan treatment was significantly reduced and that of L-type channels significantly increased compared with untreated Cx40ko mice (Fig. 5D). However, this was entirely due to a reduction in the contribution of T-type channels [Cx40ko + candesartan, 10.1 ± 1.3 (100 – %D/Dmax); and Cx40ko, 17.2 ± 1.4 (100 – %D/Dmax); P < 0.05], because the absolute constriction due to L-type channels was not changed [Cx40ko + candesartan, 28.7 ± 3.0 (100 – %D/Dmax), n = 81 vessels in 7 mice; and Cx40ko, 28.1 ± 2.6 (100 – %D/Dmax), n = 122 vessels in 11 mice], nor was the contribution of non-L-/non-T-type channels changed (Fig. 5B and D).
Chronic candesartan treatment of wild-type mice did not alter vascular tone (Fig. 5C), nor the relative contribution of L- and T-type channels (Fig. 5D), compared with untreated wild-type mice (Fig. 5A and B). However, there was a trend to an increase in the L-type component, while the non-L-/non-T-type component was significantly reduced (Fig. 5B and D).
Protein expression of the T-type channels, Cav3.1 and Cav3.2, in the smooth muscle of cremaster arterioles was significantly decreased by chronic candesartan treatment of Cx40ko mice, such that expression was no longer significantly different from that in arterioles of wild-type mice (Fig. 7).
Reduction in reactive oxygen species decreases T-type calcium channel function in Cx40ko mice
Given that angiotensin II is a major activator of NADPH oxidase, we used the superoxide scavenger, tempol, to test the involvement of reactive oxygen species. There was no significant difference in vessel tone between arterioles from wild-type and Cx40ko mice superfused with the superoxide scavenger, tempol (1 mmol l−1; Fig. 5C), although tempol treatment did reduce tone of wild-type, but not Cx40ko, arterioles compared with their corresponding, untreated controls (Fig. 5C, cf. Fig. 5A).
The increased contribution of T-type channels to vascular tone of arterioles from Cx40ko mice (Fig. 5B) was completely abolished by tempol [Fig. 5D; Cx40ko + tempol, 1.3 ± 0.7 (100 – %D/Dmax), n = 46 vessels in 5 mice; P < 0.05]. Vascular tone was maintained due to an increased contribution of L-type channels, without change in the non-L-/non-T-type component (Fig. 5D). In contrast, in wild-type arterioles acutely treated with tempol, the reduced vascular tone resulted from a reduction in the absolute constriction due to L-type channels [wild-type, 35.4 ± 4.1 (100 – %D/Dmax), n = 124 vessels in 12 mice; wild-type + tempol, 24.8 ± 5.8 (100 – %D/Dmax), n = 73 vessels in 7 mice; P < 0.05], without change in the absolute constriction due to T-type channels [wild-type, 9.5 ± 1.2 (100 – %D/Dmax); wild-type + tempol, 10.4 ± 1.2 (100 – %D/Dmax)]; leading to an increase in the relative contribution of T-type channels to vascular tone (Fig. 5D).
Acute treatment of arterioles with the NADPH oxidase inhibitor, apocynin, also reduced the T-type component to vascular tone in Cx40ko mice [Cx40ko + apocynin, 3.1 ± 1.1 (100 – %D/Dmax), n = 79 vessels in 7 mice; P < 0.05 compared with Cx40ko], but did not alter the T-type component to tone in wild-type mice [wild-type + apocynin, 7.9 ± 2.4 (100 – %D/Dmax), n = 40 vessels in 4 mice; P > 0.05 compared with wild-type].
Reduction in reactive oxygen species or blockade of angiotensin II improves spreading vasodilatation in arterioles of Cx40ko mice
In order to determine whether a reduction in the T-type channel contribution to vascular tone had any effect on conducted vasodilatation, we treated cremaster arterioles of both wild-type and Cx40ko mice acutely with tempol. In arterioles of wild-type mice, conducted vasodilatations were not altered by tempol (1 mmol l−1; Fig. 8A and Table 1; P > 0.05). However, in Cx40ko mice, tempol treatment significantly improved conducted vasodilatations over 1.5 mm (Fig. 8B and Table 1; P < 0.05). Chronic candesartan treatment of Cx40ko mice also significantly improved conducted vasodilatation over 1.5 mm (Fig. 8C and Table 1; P < 0.05). Acute treatment of these candesartan-treated vessels with tempol had no further effect (Fig. 8C and Table 1; P > 0.05).
It should be noted that measurements could not be made consistently over longer distances because of the complications imposed by arteriolar branch points. However, the data show that responses do still decay with distance (Fig. 8C and Table 1), due to the prominent effect of the loss of Cx40 in decreasing cell coupling in the endothelial cell layer.
The present study provides the first evidence that reduction in endothelial coupling in the microcirculation of the Cx40ko mouse leads to significant attenuation of the axial conduction of hyperpolarization. Moreover, we show that this decline in hyperpolarization results from transfer through the less well-coupled endothelium, rather than alternative conduction through the high-resistance smooth muscle layer. As we have found previously in wild-type mice, a non-linear relationship exists between membrane potential and diameter; however, in the Cx40ko mice, the linear part of the relationship is expanded to encompass more hyperpolarized membrane potentials; a change which acts to reduce further the capacity of the microcirculation to conduct vasodilatation over distance. In line with these data, we show that a significant change in electromechanical coupling occurs in the Cx40ko mouse, with an increase in the contribution of T-type voltage-dependent calcium channels to vascular tone. This change in electromechanical coupling results from increased oxidative stress due to overactivation of the renin–angiotensin system, because it can be reversed by acute treatment with the superoxide scavenger, tempol, or chronic inhibition of the renin–angiotensin system with candesartan. Importantly, these two treatments improve the conduction of vasodilatation in Cx40ko mice, suggesting that oxidative stress, as occurs during cardiovascular disease, may impair the ability of microcirculatory networks to regulate blood flow dynamically by altering the relative contribution of L- and T-type calcium channels.
Conduction of hyperpolarization occurs through the endothelium in Cx40ko mice
In the present study, we have clearly shown that vasodilatation is conducted solely through the endothelium in cremaster arterioles of Cx40ko mice, because light dye disruption of the endothelium completely abolished the response beyond the site of cell damage (Fig. 2). While this result seems at odds with the absence from the endothelium of Cx40 and Cx43, and only sparse Cx37 expression at the light microscope level (Fig. 1J and K), our ultrastructural studies detected pentalaminar gap junctions between ECs, and between the endothelium and smooth muscle of these arterioles (Fig. 1A–G). The reduced appearance of Cx37 staining (Fig. 1J and K) could therefore result from reduction in the size of the aggregations or plaques of Cx37 and, concomitantly, their appearance at the light microscope level.
Whether the total number of Cx37 channels in the cell membrane is the same in arterioles of Cx40ko mice and wild-type mice is unclear. What we do know is that Cx40 is completely absent from the endothelium (Fig. 1H and I) and that the incidence of MEGJs in arterioles of Cx40ko mice is 58% less than that previously found in wild-type mice (Cx40ko, 4.4 ± 0.5 MEGJs/EC cf. wild-type, 10.5 ± 0.9 MEGJs/EC; Wölfle et al. 2011). Interestingly, the best estimate for the endothelial coupling resistance in our computational model, in order to fit the electrophysiological data obtained from the Cx40ko mice, suggested a 69% reduction in coupling compared with wild-type mice (Cx40ko, 13 MΩ cf. wild-type, 4 MΩ). Together, these data suggest that expression of Cx37 gap junctions may only be slightly reduced in the EC membranes of arterioles from Cx40ko mice. However, the aggregation of these gap junctions into plaques within the membrane may depend on the presence of Cx40 or one of its scaffolding proteins. Nevertheless, in the absence of Cx40 and Cx43, endothelial gap junctions comprising Cx37 clearly play an important ancillary role in the conduction of hyperpolarization and vasodilatation (Fig. 3).
Attenuation of conducted hyperpolarization and the hyperpolarized threshold potential together account for attenuation of vasodilatation in Cx40ko mice
Our paired recordings of membrane potential and diameter, in response to focal stimulation with the vasodilatator ACh, have clearly shown that the efficacy of conduction of electrical signals is reduced in arterioles from Cx40ko mice compared with wild-type mice (Fig. 3), consistent with the reduction in EC coupling due to the loss of Cx40. The patency of the myoendothelial gap junctions found here in Cx40ko arterioles in vivo was supported by the lack of difference in the resting membrane potential of dye-identified SMCs and ECs, as well as the coincidence in the amplitude of the local and conducted responses to ACh in the two cell types (Supplementary Fig. 1).
As we have found previously in cremaster arterioles from wild-type mice (Wölfle et al. 2011), analysis of membrane potential and vessel tone recorded simultaneously in arterioles from Cx40ko mice at the local site, before and after ACh application, showed that the relationship was not linear over the entire voltage range, but could be fitted with two-line regression. However, the threshold potential, beyond which vasodilatation was maximal, was more hyperpolarized than the one that we previously obtained in arterioles of wild-type mice (−45 ± 3 cf. −35 ± 1 mV; Wölfle et al. 2011). Computational modelling of the vessel wall, using anatomical parameters obtained from Cx40ko arterioles (Fig. 4C), demonstrated that the recorded attenuation in conducted vasodilatation could only be estimated by application of the more hyperpolarized threshold potential found in the present study (Fig. 4D). Thus, both the steeper decay of the hyperpolarization and the more hyperpolarized threshold potential together contributed to the observed attenuation of conducted vasodilatation; the threshold potential of −45 mV resulted in entry into the linear part of the voltage–diameter relationship closer to the local stimulus, while the steeper decay of the hyperpolarization (Fig. 3C), due to reduced cell coupling, facilitated the decline in the linearly related vasodilatation.
Contribution of T-type calcium channels to vascular tone is increased in Cx40ko mice
We have previously suggested that the threshold potential might coincide with the activation threshold of voltage-dependent calcium channels (Wölfle et al. 2011), because these channels are integral for the development of myogenic tone (Hill et al. 2001b). The present data show that, in addition to the predominant contribution made by L-type voltage-dependent calcium channels to vascular tone of cremaster arterioles, there is a minor contribution of T-type voltage-dependent calcium channels, which is significantly increased in arterioles of Cx40ko mice compared with wild-type mice (Fig. 5B). These low-voltage-activated T-type channels have increasingly been shown to be expressed in vascular SMCs, where they have been reported to be responsible for vasomotor responses, particularly those of small, resistance-sized vessels (Hansen et al. 2001; Morita et al. 2002; Hayashi et al. 2007; Jensen & Holstein-Rathlou, 2009; Kuo et al. 2011; Poulsen et al. 2011). In line with our pharmacological studies, our immunohistochemical studies have shown that expression of the T-type channels, Cav3.1 and Cav3.2, is significantly increased in cremaster arterioles of Cx40ko mice compared with wild-type mice (Figs 6 and 7D). After inhibition of both L- and T-type channels in arterioles of wild-type mice, there was a small residual constriction, which was significantly reduced in Cx40ko mice (Fig. 5B). Whether this component involves other voltage-dependent calcium channels, such as members of the Cav2 subfamily, as occurs in the renal microcirculation, or activation of a voltage-independent calcium source (Hansen et al. 2001, 2011; Hayashi et al. 2007; Mufti et al. 2010), is currently unknown.
Overactivation of the renin–angiotensin system is responsible for the increased contribution of T-type channels in Cx40ko mice
Angiotensin II is a major activator of NADPH oxidase, upregulation of which is now considered to be responsible for the increased oxidative stress associated with cardiovascular disease (Chrissobolis et al. 2011; Drummond et al. 2011). We therefore hypothesized that the increased contribution of T-type channels in Cx40ko mice might result from overactivation of the renin–angiotensin system. To this end, we confirmed that a significant, 4-fold increase in renin activity occurred in Cx40ko mice compared with wild-type mice, and that chronic blockade of the action of angiotensin II with the AT1 receptor antagonist, candesartan, resulted in a significant decrease in blood pressure. In line with our hypothesis, chronic candesartan treatment of Cx4ko mice led to a significant reduction in protein expression of the T-type channels, Cav3.1 and Cav3.2, in cremaster arterioles (Fig. 7) and a significant decrease in the contribution of T-type channels to vascular tone compared with arterioles of untreated Cx40ko mice, without change in the non-L-/non-T-type component (Fig. 5D).
Inhibition of oxidative stress eliminates T-type channel involvement in Cx40ko mice
The downstream involvement of reactive oxygen species in the action of angiotensin II was tested by acute treatment with the superoxide scavenger, tempol. Remarkably, tempol completely eliminated the contribution of T-type channels to vascular tone in arterioles of Cx40ko mice (Fig. 5D), as did the NADPH oxidase inhibitor, apocynin. The failure of tempol and apocynin to reduce the small contribution of T-type channels to vascular tone of normotensive mice (Fig. 5D) suggests that the control of T-type calcium channel activity may differ in conditions of chronic oxidative stress.
Exogenous application of reactive oxygen species or angiotensin II has been reported to increase L-type channel activity in pressurized cerebral arteries and isolated SMCs taken from normotensive rats (Amberg et al. 2010). The lack of effect of endogenous reactive oxygen species on L-type channel function in our experiments may indicate differences between the cerebral and systemic circulations in the subcellular association of channel subtypes with NADPH oxidases. While future studies are needed to probe the role and location of NADPH oxidases further in physiological and pathophysiological conditions, our previous demonstration that L- and T-type channels do not colocalize (Kuo et al. 2010) adds support to the idea that selective control of L- and T-type channel function may be exerted through association with different messenger systems in spatially separate microdomains.
Inhibition of angiotensin II or oxidative stress improves the conduction of vasodilatation in Cx40ko mice
In order to determine the impact of the change in electromechanical coupling on the conduction of vasodilatation, we studied this response after chronic candesartan or acute tempol treatment; the same conditions which reduced the contribution of T-type channels to vascular tone in Cx40ko mice. We found that both treatments significantly improved the impaired spreading of vasodilatation found in arterioles from untreated Cx40ko mice (Fig. 8), suggesting that the increased involvement of T-type channels to the control of vascular tone, following overactivation of the renin–angiotensin system and increase in oxidative stress, may play a critical role in this phenomenon. However, it was not possible directly to test the role of T-type channels in the conduction of vasodilatation, because our previous studies have shown that NNC 55-0396 can also inhibit L-type channels, leading to near-complete loss of vascular tone (Kuo et al. 2010). Nevertheless, our studies show conclusively that oxidative stress, as occurs during cardiovascular disease, can significantly impair conducted vasodilatation in microcirculatory networks.
Cardiovascular disease is characterized by endothelial dysfunction, a condition of increased oxidative stress. Changes also occur to the coupling of the EC layer, with reduction in the expression of the connexin constituents of gap junctions. Although stated to have dramatic effects on the conduction of electrical signals along the endothelium, the effect of changes in endothelial coupling on electrical conduction has not been tested experimentally.
This study used the hypertensive animal model of global Cx40 deletion to determine the effect of reduction in EC coupling on the conduction of hyperpolarization. Our findings show, for the first time, that hyperpolarization does attenuate more rapidly with distance when endothelial coupling is significantly reduced; however, this reduction alone cannot account for attenuation in the accompanying vasodilatation. Surprisingly, the threshold for entry into the linear relationship between voltage and vascular tone is also altered. This change correlated with a change in the contribution of the voltage-dependent calcium channel subtypes underlying vascular tone; a process governed by oxidative stress resulting from upregulation of the renin–angiotensin system. Importantly, we found that both acute and chronic reduction in oxidative stress in Cx40ko mice improved conducted vasodilatation. We suggest that similar changes in electromechanical coupling could also occur during other cardiovascular conditions, as a result of increased activation of NADPH oxidases through mechanisms other than the renin–angiotensin system (Chrissobolis et al. 2011; Drummond et al. 2011). Understanding this novel link between the control of electromechanical coupling and oxidative stress during hypertension may assist in uncovering new ways to reduce tissue damage by improving the spreading of vasodilatation.
Experiments were performed in the Department of Neuroscience, John Curtin School of Medical Research, Australian National University and the Department of Physiology, School of Medical Sciences, University of New South Wales. C.E.H. designed the study and drafted the article. All authors contributed to the collection, analysis and interpretation of data and revision of the manuscript for intellectual content. All authors approved the final version of the manuscript.
We thank Dr Stephanie Wölfle for electrophysiological studies and immunohistochemical data for Fig. 6 and Dr Terry Neeman for linear regression analysis. This work was supported by grants from the National Health and Medical Research Council of Australia (471420, 471421).