SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

The goal of this investigation was to probe intercellular conduction in skeletal muscle feed arteries and to address why smooth muscle-initiated responses fail to robustly spread like their endothelial counterpart. Using computational and experimental approaches, two interrelated rationales were developed to explain this apparent discrepancy in cell-to-cell communication. The first rationale stressed that smooth muscle electrical responses, if initiated, will be actively dissipated as they spread from cell-to-cell along the arterial wall. Charge dissipation is promoted within arteries by the structural and connectivity properties of vascular cells. The second rationale centred on the idea that when agents other than KCl stimulate a limited number of smooth muscle cells, they fail to generate the currents required to elicit a localized membrane potential (VM) response. This insufficiency results in part from charge loss, via gap junctions, to neighbouring unstimulated cells. Experiments confirmed the latter rationale by showing that focal phenylephrine application: (1) elicited a localized constriction insensitive to L-type Ca2+ channel blockade; and (2) failed to substantially depolarize vascular smooth muscle cells. Further investigation revealed that while focal phenylephrine-induced constriction was VM independent, it was reliant on internal Ca2+ mobilization and the activation of inositol 1,4,5-trisphosphate (IP3) receptors. The preceding findings illustrate that by using computational modelling and experimentation in a complementary manner, one can isolate key cellular properties and rationally examine their role in limiting the conduction of smooth muscle-initiated responses. Functionally, these observations enable investigators to assign the concept of ‘local and global’ blood flow control to the electrical and/or non-electrical behaviour of specific cell types.

Vascular networks dilate and constrict in a coordinated fashion to control blood flow to metabolically active tissue (Kurjiaka & Segal, 1995; Segal & Jacobs, 2001). For resistance arteries to respond in an integrative manner, vascular cells must communicate with one another (Segal & Beny, 1992; Xia et al. 1995). Electrical communication in the arterial wall is enabled by gap junctions, intercellular channels that permit ions and second messengers to flow down their electrochemical gradient (Little et al. 1995a,b). Gap junctions comprise two hemichannels (i.e. connexons), each containing six connexin subunits. The connexin gene family comprises at least 19 members, with the most predominant subtypes in vascular tissue being Cx37, Cx40, Cx43 and Cx45 (Saez et al. 2003). In resistance arteries, functional gap junctions couple adjacent smooth muscle cells, adjacent endothelial cells and the two cell layers (Sandow et al. 2002, 2003).

Cell-to-cell communication is typically studied in small resistance arteries by applying vasoactive agents to a discrete portion of the vessel wall (Segal & Duling, 1986; Kurjiaka et al. 2005; Jantzi et al. 2006). This focal application is thought to elicit a localized change in smooth muscle or endothelial membrane potential (VM) which conducts to neighbouring cells via gap junctions (Xia & Duling, 1995; Welsh & Segal, 1998). The extent to which the electrical or the corresponding vasomotor response conducts provides functional insight into the nature of vascular communication. Using this approach, early studies advanced the notion that, like the myocardium, vascular cells form a functional syncytium (Segal & Beny, 1992; Xia et al. 1995). As such, electrical responses initiated in smooth muscle or the endothelium should spread equally to the adjacent cell type as it longitudinally conducts. While this theory is attractive for its simplicity, it is evident that not all cell-specific responses spread equally along the arterial wall. This divergence has been particularly evident in skeletal muscle arteries where endothelial-initiated events conduct with little decay while those initiated in smooth muscle remain highly localized (Segal et al. 1999; Kurjiaka et al. 2005).

The inability of smooth muscle responses to robustly conduct along skeletal muscle arteries has been a source of active inquiry and analysis. Past studies have presented a variety of theories, each focusing on how a smooth muscle electrical response, once initiated, is dissipated by different cellular properties. For example, early investigations centred on whether specific ionic conductances feed back upon a smooth muscle response to limit longitudinal spread (Yashiro & Duling, 2000, 2003). This work typically highlighted the importance of voltage- and/or Ca2+-dependent K+ channels whose expression is cell type specific (Yashiro & Duling, 2000, 2003). In contrast, recent investigations have alternatively stressed that charge dissipation arises from the structural and connectivity properties of vascular cells (Diep et al. 2005). Irrespective of these or other valid perspectives, it is important to resolve the biophysical basis of this conduction disparity as the resultant knowledge will impact on our conceptual understanding of blood flow control.

The goal of this investigation was to probe the basis of differential communication and to understand why smooth muscle responses fail to conduct robustly along skeletal muscle arteries. To achieve this objective, this study combined the methodological strengths of computational modelling and experimentation. Two rationales were carefully developed to explain why smooth muscle responses poorly conduct unlike their endothelial-initiated counterparts. The first rationale stressed that if initiated, a smooth muscle electrical response will be actively dissipated by the structural and connectivity properties of vascular cells. The second explanation stressed the idea that when agents other than KCl stimulate a limited number of smooth muscle cells, they do not generate sufficient current to initiate a localized VM response. The latter rationale was confirmed experimentally by demonstrating: (1) the insensitivity of focal agonist-induced constriction to L-type Ca2+ channel blockade; and (2) the inability of smooth muscle cells to depolarize to focal agonist application. Further investigation revealed that while focal phenylephrine-induced constriction was VM independent, it was reliant on internal Ca2+ mobilization and IP3 receptor activation. The significance of these observations was discussed, with investigators assigning the concept of ‘local and global’ blood flow control to the electrical and/or non-electrical behaviour of specific cell types.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Animal and tissue preparations

Animal handling procedures were approved by the Animal Care and Use Committee at the University of Calgary. Briefly, each male golden Syrian hamster (10–12 weeks of age) was anaesthetized with sodium pentobarbital (65 mg kg−1, intraperitoneal injection (i.p.)). An incision was made through the skin overlying either the right or left retractor muscle. Superficial connective tissue was removed from the exposed tissue which was continuously superfused with phosphate buffered saline (PBS) containing (in mm): 138 NaCl, 3 KCl, 10 Na2HPO4, 2 NaH2PO4, 5 glucose, 0.1 CaCl2 and 0.1 MgSO4. The retractor muscle was then excised and pinned out in a dissecting dish containing PBS. Fat cells and connective tissue were further removed and the feed arteries cut into 2–3 mm lengths. The hamster was killed with an i.p. overdose of sodium pentobarbital.

Vessel myography

An isolated feed artery was cannulated in a customized arteriograph chamber designed to measure both VM and diameter. The arteriograph was positioned on an inverted microscope and the vessel equilibrated at 15 mmHg intravascular pressure (30 min) in a physiological saline solution (PSS: 37°C; 5% CO2; 21% O2) containing (in mm): 119 NaCl; 4.7 KCl; 1.7 KH2PO4; 1.2 MgSO4; 1.6 CaCl2; 5 glucose and 20 NaHCO3. When assessing the ability of phenylephrine, U46619 or KCl to initiate smooth muscle-dependent conduction, intravascular pressure was set to 40 mmHg. At this pressure, arteries maintain a more dilated/hyperpolarized state; thus the range over which a vessel can constrict/depolarize is enhanced. When assessing acetylcholine's ability to elicit endothelial-dependent conduction, intravascular pressure was first elevated to 60 mmHg to enhance myogenic tone. Pressure-sensitive arteries were then exposed to phenylephrine (0.1 μm in the superfusate). Such modest alterations help standardize tone within an experimental set and do not dramatically alter the ability of a smooth muscle- or endothelial-initiated response to conduct (Kurjiaka et al. 2005). All experiments were performed under conditions of no luminal flow. Note that feed arteries that did not robustly respond to superfused KCl (60 mm) were excluded from experimentation.

Diameter and VM measurements

Internal feed artery diameter was monitored using a 10× objective and manual video calipers. Smooth muscle VM was assessed by inserting a glass microelectrode backfilled with 1 m KCl (tip resistance = 120–150 MΩ) into the vessel wall. The criteria for successful cell impalement included: (1) a sharp negative VM deflection upon entry; (2) a stable recording for at least 1 min following entry; and (3) a sharp return to baseline upon electrode removal. To monitor the VM change associated with focal phenylephrine application, a continuous recording was required. In contrast, the VM change arising from global agonist application was derived as the difference between two independent/steady-state recordings.

Experimental protocols

To probe cell-to-cell communication, vasoactive agents must be discretely applied via micropipette to initiate focal changes in smooth muscle or endothelial VM (Fig. 1 inset). In this study, smooth muscle cells were stimulated by: (1) microiontophoresing phenylephrine (1 mm; 5, 7 or 20 s pulse; ejection current, 0.5 μA; retain current, 200 pA); (2) microiontophoresing U46619 (1 μm; 5 s pulse; ejection current, 0.5 μA; retain current, 200 pA); or (3) pressure-ejecting KCl (200 mm; 5 s pulse; ejection pressure 10 p.s.i.) onto the feed artery wall. In contrast, endo-thelial cells were activated by: (1) microiontophoresing acetylcholine (10 mm; 5 s pulse, ejection current, 0.5 μA; retain current, 200 pA); or (2) by lengthening the duration of the KCl stimulus from 5 to 20 s. Note that acetycholine and prolonged KCl application were used to focally hyperpolarize and depolarize endothelial cells, respectively. Changes in arterial diameter were evaluated at 0–1800 μm distal to the application site. Intercellular conduction was assessed under control conditions and following one of the four experimental treatments. These included: (1) superfusion with tetraethylammonium (TEA; 0.3–3 mm); (2) intraluminal perfusion with apamin (50 nm) and TRAM34 (1 μm); (3) superfusion with diltiazem (30 μm) ± thapsigargin (100 nm), 2-APB (50 μm) or xestospongin C (10 μm); and (4) superfusion with ML-7 (60 μm). In a small complement of experiments, the endothelium was removed by passing air bubbles through the vessel's lumen; successful removal was confirmed by documenting the absence of acetylcholine-induced dilatation.

image

Figure 1. Intercellular conduction along skeletal muscle arteries Top: illustration of the experimental protocol. Vasoactive agents are applied via pipette to the vessel wall while vasomotor responses are measured 0–1800 μm distal to the point of agent application. A and B, representative trace and summary data (n= 7: absolute resting, maximum and minimum diameters were 42 ± 5 μm, 72 ± 3 μm and 22 ± 2 μm) highlighting acetylcholine's (5 s pulse) ability to elicit endothelial-initiated conduction. C and D, representative trace and summary data highlighting the inability of smooth muscle-selective agents (phenylephrine, 5 s pulse; U46619, 5 s pulse; KCl, 5 s pulse) to elicit conduction. Absolute resting, maximal and minimum diameters were as follows: phenylephrine (n= 7), 56 ± 3 μm, 76 ± 2 μm, 21 ± 2 μm; U46619 (n= 5), 76 ± 4 μm, 94 ± 4 μm, 18 ± 2 μm; KCl (n= 6), 56 ± 10 μm, 74 ± 4 μm, 24 ± 2 μm. E and F, representative trace and summary data (n= 5: absolute resting, maximum and minimum diameters were 77 ± 3 μm, 87 ± 2 μm and 22 ± 2 μm) highlighting the ability of a prolonged KCl pulse (20 s) to elicit an endothelial-dependent conducted response. Absolute resting diameter following endothelial removal was 73 ± 3 μm.

Download figure to PowerPoint

Ca2+ sensitization was examined using a standard permeabilization approach (Gokina & Osol, 1998). Briefly, feed arteries pressurized to 40 mmHg were equilibrated (15 min) in an activating solution (pCa 6.5) containing (in mm): 63.0 potassium methanesulfonate; 5.8 CaCl2; 2.0 MgCl2; 4.5 MgATP; 10.0 EGTA; 10.0 phosphocreatine; 1.0 phosphocreatine kinase; 30.0 piperazine-N,N′-bis (2-ethanesulfonic acid); 0.001 carbonyl cyanide p-trifluoromethoxyphenyl-hydrazone; 0.001 leupeptin and 0.03 ryanodine; pH = 7.2. Staphylococcus aureus toxin (800 U ml−1) was subsequently added to the superfusate until such time that the feed artery constricted to ∼60% of its maximal resting diameter (∼20 min). Vessel segments were then rinsed 3 times and placed in a relaxing solution (pCa 9.0; 10 min) that was identical to the activating solution except that it contained 2.0 mm EGTA and 0.01 CaCl2. Following this procedure, feed arteries were placed in pCa 7.25 or 7.0 solution to initiate a submaximal Ca2+-dependent constriction. Phenylephrine was then applied via micropipette and vasomotor responses were monitored at the focal site of agent application. Free [Ca2+] was calculated using the WEBMAXCLITE program and the ionic strength was kept constant (200 mm) by adjusting the potassium methanesulfonate concentration.

Computational modelling

A computational model developed by Diep et al. (2005) was used to foster a quantitative understanding of intercellular conduction in small arteries. Computational theory and base parameters were similar to the original publication (Diep et al. 2005) with Fig. 2C being a notable exception. Specific simulation details are provided below.

image

Figure 2. Computational modelling and the factors limiting the conduction of smooth muscle-initiated depolarization Top: general illustrative diagram of the virtual artery. The virtual artery was 2.0 mm long and comprised one layer of endothelium (red) and one layer of smooth muscle (black). The artery was subdivided into 40 arterial segments, each 50 μm long, and consisting of an outer layer of 30 smooth muscle cells (arranged as 10 consecutive bands each consisting of 3 cells) and an inner layer of 48 endothelial cells (arranged side-by-side parallel to the artery's longitudinal axis). Cells were treated as discrete elements with defined physical dimensions, gap junctional coupling and ionic conductance. Neighbouring smooth muscle cells were electrically coupled to one another as were neighbouring endothelial cells. Every smooth muscle cell was randomly coupled to two endothelial cells (red dot denotes myoendothelial contact site). Simulations in A and B: one arterial segment of endothelium (A) or smooth muscle (B) was voltage clamped (200 ms) −15 mV negative/positive to resting VM (−40 mV) while steady-state electrical responses were monitored along the virtual artery. Simulations in C: smooth muscle cells were oriented circumferential or parallel to the virtual artery's longitudinal axis. Smooth muscle-to-smooth muscle coupling resistance was set to 90 or 3 MΩ; myoendothelial coupling was eliminated by increasing resistance to 500 000 MΩ. One arterial segment of smooth muscle was voltage clamped from −40 mV to −25 mV while steady-state electrical responses were monitored. Simulations in D: one arterial segment of smooth muscle was voltage clamped from −40 mV to −25 mV. Steady-state electrical responses were monitored under conditions in which endothelial-to-endothelial coupling resistance was 3 or 90 MΩ. Arrows point to the endothelial response underneath the focal site of smooth muscle stimulation. Simulations in E and F: one arterial segment of smooth muscle was voltage clamped from −40 mV to −25 mV while steady-state electrical responses were monitored under control conditions and with the smooth muscle (E) or endothelial (F) ionic conduction reduced by 50% to simulate K+ channel inhibition.

Download figure to PowerPoint

  • (a) 
    Figure 2A and B: one arterial segment of smooth muscle or endothelium was voltage clamped 15 mV negative or positive to the resting VM(−40 mV) for 200 ms. Voltage responses were monitored over time along the virtual artery.
  • (b) 
    Figure 2C: smooth muscle cells were oriented circumferential or parallel to the virtual artery's long axis. Smooth muscle-to-smooth muscle coupling resistance was set to 90 or 3 MΩ. Myoendothelial coupling was eliminated by increasing resistance to 500 000 MΩ. One arterial segment of smooth muscle was then voltage clamped 15 mV positive to resting VM(−40 mV) for 200 ms. Smooth muscle VM responses were monitored along the virtual artery.
  • (c) 
    Figure 2D: one arterial segment of smooth muscle was voltage clamped 15 mV positive to resting VM (−40 mV) for 200 ms. Endothelial and smooth muscle VM responses were monitored along the virtual artery under conditions in which endothelial-to-endothelial coupling resistance was 3 or 90 MΩ.
  • (d) 
    Figure 2E and F: one arterial segment of smooth muscle was voltage clamped 15 mV positive to resting VM (−40 mV) for 200 ms. Endothelial and smooth muscle VM responses were monitored along the virtual artery under control conditions and with the smooth muscle (E) or endothelial (F) ionic conductance reduced by 50%. This reduction in ionic conductance was used to simulate K+ channel inhibition.
  • (e) 
    Figure 4: each smooth muscle cell within 1 or 40 arterial segments was injected with 4.7 pA of depolarizing current for 250 ms. Smooth muscle VM responses were monitored at the site of stimulation.
  • (f) 
    Figure 5: one arterial segment of endothelium was voltage clamped from −40 mV to −25 mV for 200 ms. Current and voltage responses were plotted over time in three smooth muscle cells positioned 0, 1000 and 2000 μm distal to the site of endothelial cell stimulation.
  • (g) 
    Figure 8E: each smooth muscle cell within 1 or 40 arterial segments was injected with 4.7, 10, 20, 30, 40 or 43.5 pA of depolarizing current for 250 ms. Smooth muscle VM responses were monitored at the site of stimulation.
image

Figure 4. Computational modelling predicts that focal smooth muscle stimulation does not substantially alter VM Simulations: each smooth muscle cell within 1 or 40 arterial segments was injected with 4.7 pA of depolarizing current for 250 ms. Smooth muscle VM was colour-mapped along the virtual artery (A) or plotted at a single measurement site (B: at 250 ms).

Download figure to PowerPoint

image

Figure 5. Modelling the current–voltage relationship of smooth muscle cells Simulation: one arterial segment of endothelium was voltage clamped from −40 mV to −25 mV for 200 ms. Current and voltage responses were plotted over time in three smooth muscle cells positioned 0,1000 and 2000 μm distal to the site of endothelial cell stimulation. Input resistance was calculated from steady-state current/voltage values with the dotted box.

Download figure to PowerPoint

image

Figure 8. Prolonging focal phenylephrine application does not induce arterial depolarization Phenylephrine was focally (5, 7 or 20 s pulse) or globally (0.1 μm) applied while diameter and VM responses were monitored in the presence or absence of dilitazem (A and B) or ML-7 (C and D). Representative traces (A and C) and summary data (B and D) of vasomotor responses to local and global phenylephrine application. In B, absolute diameter at rest, with diltiazem, at maximum and at minimum were as follows: 5 s pulse (n= 7), 56 ± 3 μm, 64 ± 2 μm, 76 ± 2 μm, 21 ± 2 μm; 7 s pulse (n= 6), 74 ± 5 μm, 85 ± 5 μm, 90 ± 9 μm, 11 ± 3 μm; 20 s pulse (n= 6), 76 ± 6 μm, 87 ± 6 μm, 98 ± 8 μm, 16 ± 2 μm; global (n= 6), 54 ± 5 μm, 64 ± 4 μm, 72 ± 3 μm and 22 ± 2 μm. In D, resting VM prior to focal (5, 7 or 20 s) or global application of phenylephrine were −47 ± 1 mV (n= 6), −48 ± 1 mV (n= 6), −42 ± 1 mV (n= 6) and −47 ± 1 mV (n= 6), respectively. Simulations in E: each smooth muscle cell within 1 or 40 arterial segments was injected with 4.7, 10, 20, 30, 40 or 43.5 pA of depolarizing current for 250 ms. Smooth muscle VM responses were monitored at the site of stimulation.

Download figure to PowerPoint

Statistical analysis

Data are expressed as means ±s.e.m. and n indicates the number of feed arteries. One feed artery was used per animal. Where appropriate, paired t tests were used to compare feed artery responses prior to and following an experimental treatment. P values ≤ 0.05 were considered statistically significant.

Solutions and chemicals

All buffers, chemicals and reagents were purchased from Sigma-Aldrich (St Louis, MO, USA) unless otherwise noted.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Feed arteries isolated from the hamster retractor muscle were used to probe the basis of differential conduction. Consistent with past observations, focal acetylcholine application initiated an endothelial-dependent dilatation that conducted robustly along these resistance arteries (Fig. 1A and B). In contrast, focal smooth muscle stimulation (phenylephrine, U46619 or a 5 s KCl pulse) elicited a localized constriction that failed to substantially spread along the arterial wall (Fig. 1C and D). In these latter experiments, the KCl pulse duration was limited to prevent extracellular [K+] from increasing around the endothelium. Such a change would have shifted the K+ equilibrium potential rightward, initiating an endothelial depolarization that could have robustly conducted. Indeed, if the KCl pulse duration was lengthened (20 s), a robust endothelial-dependent conducted response was observed in retractor muscle feed arteries (Fig. 1E and F).

Cell orientation and coupling resistance probably play an important role in limiting the conduction of smooth muscle-initiated responses (Diep et al. 2005). To fully explore this idea, a range of simulations were performed on a virtual artery designed to represent a small resistance artery. Briefly, the virtual artery was 2000 μm in length, 75 μm in diameter and consisted of 3168 discrete vascular cells each of which retained distinct structural/electrophysiological properties (Fig. 2). Initial simulations predicted the differential behaviour observed in Fig. 1. When one arterial segment of endothelium was voltage clamped 15 mV negative or positive to resting VM (−40 mV), the resulting electrical response conducted robustly along the arterial wall (Fig. 2A). In contrast, a robust conducted response was not observed when one arterial segment of smooth muscle was voltage clamped from −40 to −25 mV (Fig. 2B). Subsequent simulations ascribed the absence of smooth muscle-initiated conduction to specific tissue properties and their ability to dissipate electrical responses as they spread to neighbouring cells. Charge dissipation along the smooth muscle layer is promoted by these cells’ circumferential orientation and elevated coupling resistance (90 MΩ). This was illustrated in Fig. 2C by assigning endothelial-like properties (i.e. parallel orientation and lower coupling resistance) to an isolated layer of smooth muscle cells and documenting enhanced intercellular conduction. Charge flow to the underlying endothelium also fails to elicit a substantial depolarization due to dissipation. Endothelial charge dissipation is facilitated by the robust expression of gap junctions which ensures that this modicum of charge will effectively spread throughout the broader endothelial cell layer. This concept was highlighted in Fig. 2D by elevating endothelial-to-endothelial coupling resistance (3–90 MΩ) and documenting augmented endothelial depolarization underneath the site of smooth muscle stimulation. It is interesting to note that while higher coupling resistance promoted focal endothelial depolarization in these simulations, it also enhanced the electrical decay along the virtual artery.

Early studies noting the inability of smooth muscle-initiated responses to robustly conduct ascribed this absence to the ability of voltage/Ca2+-dependent K+ channels to effectively feedback upon and diminish the locally initiated depolarization (Yashiro & Duling, 2000, 2003). While an interesting concept, computational observations in Fig. 2E and F highlight the implausibility of this scenario. In particular, they show that simulated K+ channel blockade, induced by reducing the smooth muscle or endothelial ionic conductance by 50%, had no sizable effect on enabling smooth muscle-initiated conduction. To reinforce these theoretical observations, a range of experiments were performed in which focal smooth muscle-initiated constriction was monitored along feed arteries perfused: (1) abluminally with TEA (large-conductance Ca2+-activated-(BKCa)/voltage-dependent (KV) K+ channel blocker); or (2) luminally with apamin/TRAM34 (small (SK) and intermediate (IK) conductance Ca2+-activated K+ channel blockers, respectively). At low millimolar concentrations, TEA should completely block BKCa and partially inhibit KV in vascular smooth muscle (Langton et al. 1991; Cox et al. 2001). This application (0.3–3.0 mm) elicited a predictable rise in resting arterial tone but did not enable phenylephrine (Fig. 3A), U46619 (Fig. 3C) or a 5 s pulse of KCl (Fig. 3E) to elicit a conducted response. The luminal introduction of apamin/TRAM34 was equally ineffective at facilitating phenylephrine- (Fig. 3B), U46619- (Fig. 3D) or KCl- (Fig. 3F) induced conduction. The efficacy of SKCa/IKCa blockade was confirmed, a priori, by superfusing arteries with acetylcholine (1 μm) and documenting the absence of endothelial-dependent dilatation. This absence does not preclude a role for nitric oxide in acetylcholine-induced dilatation as an impairment of hyperpolarization could limit the increase in endothelial [Ca2+] required for NOS activation.

image

Figure 3. K+ channel inhibition does not enable the conduction of smooth muscle-initiated responses KCl (5 s pulse), phenylephrine (5 s pulse) or U46619 (5 s pulse) were focally applied to feed arteries while diameter responses were monitored 0–1800 μm distal to the point of agent application. Measurements were performed in the absence or presence of tetraethylammonium (TEA: A, C and E) or luminal apamin/TRAM34 (B, D and F). In A, C and E, absolute diameter at rest, with TEA (0.3, 1 and 3 mm), at maximum and at minimum, were as follows: A (n= 7), 84 ± 4 μm, 75 ± 5 μm, 60 ± 4 μm, 43 ± 3 μm, 96 ± 6 μm, and 21 ± 1 μm; C (n= 5) 84 ± 5 μm, 74 ± 4 μm, 64 ± 5 μm, 50 ± 3 μm, 103 ± 8 μm and 21 ± 6 μm; E (n= 5), 95 ± 5 μm, 88 ± 5 μm, 69 ± 5 μm, 51 ± 4 μm, 111 ± 9 μm and 22 ± 1 μm. In B, D and F, absolute diameter at rest, with apamin/TRAM34, at maximum and at minimum were as follows: B (n= 6), 70 ± 9 μm, 68 ± 4 μm, 114 ± 6 μm and 12 ± 9 μm; D (n= 5), 98 ± 7 μm, 94 ± 7 μm, 116 ± 7 μm and 22 ± 1 μm; F (n= 5), 100 ± 6 μm, 100 ± 5 μm, 116 ± 7 μm and 22 ± 1 μm. TEA, apamin and TRAM34 were used to block BKCa/KV, SKCa and IKCa channels, respectively.

Download figure to PowerPoint

Conduction studies typically assume that all agents used to stimulate smooth muscle elicit an initial electrical response. While KCl will selectively depolarize smooth muscle when applied as a short discrete pulse, agonists-induced responses may not occur if the stimulated cells fail to produce adequate current. Generating sufficient current could be a problem when the number of stimulated smooth muscle cells is limited and charge is lost to neighbouring unstimulated cells. This idea was conceptualized in Fig. 4 by injecting 4.7 pA of depolarizing current into each of 30 (one arterial segment) or 1200 (40 arterial segments) smooth muscle cells and documenting the resultant 1.4 or 15.0 mV depolarization. This depolarizing current was set, a priori, to produce a standardized 15 mV depolarization under conditions in which all smooth muscle cells are equally stimulated. Its magnitude corresponds with past electrical observations showing that vasoconstrictor-induced changes in whole cell current are in the low picoamp range at physiological voltages (Welsh & Brayden, 2001; Hayabuchi et al. 2001; Luykenaar et al. 2004). Its appropriateness was further supported by computational modelling where the current–voltage relationship of three smooth muscle cells was assessed during the conduction of an endothelial-initiated depolarization (Fig. 5). Clearly, whole cell currents in the low picoamp range rapidly and sustainably alter smooth muscle VM, by 7–13 mV. Smooth muscle input resistance was also calculated and shown to range between 2.1 and 2.3 GΩ. Building on these conceptual findings, experiments subsequently examined the effects of focal agonist application on smooth muscle VM. Consistent with no substantial depolarization, focal phenylephrine or U46619 application induced a constriction that was unaltered by L-type Ca2+ channel blockade (Fig. 6A and B). Control experiments did confirm that diltiazem blocked focal KCl-induced constriction (Fig. 6C). Likewise, diltiazem attenuated the constriction induced by superfused phenylephrine, a finding consistent with modelling observations (Fig. 4). Subsequent experiments used microelectrodes to directly assess the effects of focal phenylephrine application on smooth muscle VM. Initial findings documented a pronounced depolarization (24 ± 9 mV) to the focal phenylephrine application (Fig. 7A and B). While dramatic in magnitude and duration, these continuous recordings were disconcerting in two important aspects. First, the focal VM change was unexpectedly large and greater than the electrical response induced by superfused phenylephrine. Second, the constriction initiated by focal phenylephrine application often preceded the electrical response. Consequently, it was theorized that the local VM response was artifactual, the result of a constricting vessel physically influencing the recording electrode. To address this issue, the preceding experiments were repeated in the presence of ML-7, a myosin light chain kinase inhibitor. With arteries unable to constrict, focal phenylephrine application no longer elicited a sizeable depolarization (Fig. 7C and D). While focal agonist responses were abolished, the depolarization to focal KCl and superfused phenylephrine was maintained; the latter control eliminates the possibility that ML-7 interferes with ion channel modulation. In a small complement of experiments, VM was also measured with electrodes backfilled with Lucifer yellow dye. In three faintly labelled but identifiable smooth muscle cells, focal phenylphrine application failed to elicit a substantial depolarization (control, −45 ± 2 mV; focal phenylephrine, −45 ± 3 mV).

image

Figure 6. L-type Ca2+ channel blockade does not attenuate focal phenylephrine or U46619-induced constriction Representative traces (AC) and summary (D) data highlighting the ability of diltiazem (30 μm; L-type Ca2+ channel inhibitor) to attenuate focal phenylephrine- (5 s pulse), U46619- (5 s pulse) or KCl- (5 s pulse) induced constriction. Absolute diameters at rest, with diltiazem, at maximum and minimum were as follows: phenylephrine (n= 7), 56 ± 3 μm, 64 ± 2 μm, 76 ± 2 μm, 21 ± 2 μm; U46619 (n= 5), 76 ± 4 μm, 84 ± 2 μm, 94 ± 4 μm, 18 ± 2 μm; 5 s KCl pulse (n= 6), 56 ± 4 μm, 66 ± 3 μm, 74 ± 4 μm, 24 ± 2 μm. In E, phenylephrine was globally (0.1 μm) applied to a feed artery while diameter responses were monitored in the absence or presence of diltiazem (30 μm). Absolute diameters (n= 6) at rest, with diltiazem, at maximum and minimum were 54 ± 5 μm, 64 ± 4 μm, 72 ± 3 μm and 22 ± 2 μm, respectively. *Significant difference from control.

Download figure to PowerPoint

image

Figure 7. Focal phenylephrine application does not induce arterial depolarization Phenylephrine was focally (5 s pulse) or globally (0.1 μm) applied to a feed artery while diameter and VM were monitored. Representative tracing (A and C) and summary data (B and D; n= 6) of the responses to local and global phenylephrine. Measurements were taken in the absence and presence of ML-7 (60 μm), a myosin light chain kinase inhibitor. In B, resting VM values prior to focal (n= 6) or global (n= 6) phenylephrine application were −45 ± 1 mV and −45 ± 1 mV, respectively. In D, resting VM values prior to focal phenylephrine (n= 6, 5 s pulse), focal KCl (n= 6, 5 s pulse) or global phenylephrine (n= 6) application were −47 ± 1 mV, −43 ± 1 mV and −47 ± 1 mV, respectively.

Download figure to PowerPoint

Xia & Duling (1998) have previously noted that focally applied agonists need not depolarize vascular smooth muscle. In this earlier work, it was suggested that this absence could be overcome by lengthening stimulus duration (Xia & Duling, 1998). Our study found no evidence to support this supposition. In particular, findings in Fig. 8AD note that lengthening the focal application of phenylephrine did not induce: (1) a diltiazem-sensitive constriction, or (2) a sizable depolarization. Computational modelling in Fig. 8E further highlighted this scenario's implausibility by showing that for one arterial segment of smooth muscle to depolarize 15 mV, each stimulated cell would have to generate 43.6 pA of net inward current. Currents of this magnitude are not typically observed in vascular smooth muscle. Indeed, if such currents could be generated and every smooth muscle cell produced them simultaneously, modelling predicts that the smooth muscle layer would depolarize to the unrealistic level of +112 mV.

Voltage-independent constriction

The preceding observations illustrate that focal phenylephrine-induced constriction is mediated by a voltage-independent mechanism such as Ca2+ sensitization or perhaps SR-Ca2+ mobilization. To assess Ca2+ sensitization, feed arteries were exposed to α-toxin, an agent that permeabilizes membranes and which enables investigators to set free cytosolic [Ca2+]. Once permeabilized, treated arteries were exposed to an activating solution (pCa 6.5) to confirm contractile viability (Fig. 9A and B). Free cytosolic [Ca2+] (pCa 7.25 or 7.0) was then set to elicit a submaximal response and phenylephrine focally applied. Under these specific conditions, focal phenylephrine application had no further constrictor effect, a finding consistent with a limited role for Ca2+ sensitization. Subsequent experiments showed that localized constriction was blocked by agents that alter SR-Ca2+ mobilization. More specifically, findings in Fig. 10 illustrated that focal phenylephrine constriction was abolished in non-permeabilized feed arteries exposed to thapsigargin (Ca2+-ATPase inhibitor), 2-APB (IP3 receptor blocker) or xestospongin C (IP3 receptor blocker).

image

Figure 9. Focal phenylephrine application does not induce Ca2+ sensitization Feed arteries were cannulated and permeabilized with α-toxin. Diameter responses to fixed concentrations of free Ca2+ and to focal phenylephrine (5 s pulse) application were monitored. Representative trace (A) and summary data (B: n= 6; absolute resting and minimum diameter were 79 ± 14 μm and 11 ± 1 μm) highlight arterial responses.

Download figure to PowerPoint

image

Figure 10. Focal phenylephrine-induced constriction is dependent on SR-Ca2+ mobilization Phenylephrine was focally applied (5 s pulse) while arterial diameter was initially monitored in the absence or presence of diltiazem (30 μm). Diltiazem-treated arteries were then exposed to thapsigargin (A and B: 100 nm; Thaps), 2-APB (C and D; 50 μm) or xestospongin C (E and F: 10 μm; Xest-C) and measurements were repeated. Absolute diameter at rest, with diltiazem, with diltiazem + thapsigargin, 2-APB or xestospongin C, at maximum and at minimum were as follows. B (n= 8): 60 ± 13 μm, 78 ± 3 μm, 77 ± 3 μm, 88 ± 3 μm, 21 ± 1; D (n= 6): 72 ± 4 μm, 79 ± 4 μm, 79 ± 4 μm, 90 ± 4 13 ± 2; F (n= 6): 90 ± 7 μm, 97 ± 6 μm, 96 ± 5 μm, 103 ± 4 and 10 ± 1 μm. *Significant difference from control.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

This study explored differential communication and why responses initiated in smooth muscle fail to robustly conduct along skeletal muscle arteries. Two fundamental explanations were carefully developed using a combination of computational and experimental approaches. The first stressed that if a smooth muscle electrical response is initiated by agents like KCl, the structural and connectivity properties of vascular cells would cooperatively limit conduction by promoting charge dissipation. The second explanation focused on how a small number of smooth muscle cells when stimulated with agonists fail to generate the currents required to change arterial VM. Experiments related to the latter explanation subsequently revealed the importance of voltage-independent mechanisms in enabling focal constriction. Such mechanisms were shown to include the mobilization of SR Ca2+ and the activation of IP3 receptors.

Background

Segal & Duling (1986) were the first to describe a functional approach to assess cell-to-cell communication in small resistance arteries. Briefly, their approach began by focally applying vasoactive agents to an arterial wall so as to discretely alter smooth muscle or endothelial VM (Segal & Duling, 1986; Xia & Duling, 1995). With the aid of gap junctions, these electrical responses would spread to neighbouring vascular cells and it was through the monitoring of VM/diameter at sites distal to the stimulus that vascular communication was assessed (Xia et al. 1995; Xia & Duling, 1995). With this approach, early studies probed vascular behaviour and developed the first concrete theories on cell-to-cell communication. These initial theories highlighted the idea that smooth muscle and endothelial cells form a functional syncytium and that irrespective of the stimulated cell type, electrical information would spread equally to all constitutive cells (Segal & Beny, 1992; Xia et al. 1995). While an interesting perspective, it is one that predicts uniformity to the conduction of electrical and vasomotor responses. This expectation is inconsistent with the experimentation which has documented considerable conduction diversity among vessel types and to different stimuli (Delashaw & Duling, 1991; Segal et al. 1999; Kurjiaka et al. 2005).

Smooth muscle- and endothelial-initiated responses conduct differentially along skeletal muscle arteries. In particular, investigations have shown that while endothelial-initiated responses conduct robustly with limited decay, smooth muscle responses spread poorly beyond the site of agent application (Segal et al. 1999; Kurjiaka et al. 2005; Jantzi et al. 2006). Past studies have attributed the absence of smooth muscle-initiated conduction to a varying range of cellular properties which presumably dissipate the initial electrical event (Yashiro & Duling, 2000; Diep et al. 2005). While a reasonable assertion, one should not overlook other means to explain the absence of smooth muscle-initiated conduction. This includes the possibility that smooth muscle agonists, when focally applied, never elicit an initial electrical response. We carefully consider each possibility in the subsequent sections using, when possible, the synergistic strengths of computational modelling and experimentation.

Mechanistic exploration of smooth muscle-initiated conduction

Our investigation of cell-to-cell communication started by documenting the ability of known vasoactive agents to elicit a conducted response (Fig. 1). Of particular importance was the inability of focal smooth muscle agonists (phenylephrine, U46619 or a 5 s KCl pulse) to elicit a robust conducted response. To address the basis of this failure, a series of simulations were subsequently run to probe the relative importance of key biophysical parameters in signal dissipation. Work began in a straightforward manner by redocumenting the inability of a smooth muscle-initiated response to robustly conduct like those initiated in the endothelium (Fig. 2A and B). Electrical dissipation along the smooth muscle layer is facilitated by these cells’ circumferential orientation and elevated coupling resistance. This was illustrated in Fig. 2C assigning endothelial-like properties (i.e. parallel orientation and lower coupling resistance) to an isolated layer of smooth muscle cells and documenting enhanced intercellular conduction. High intercellular resistance along the smooth muscle cell layer promotes charge loss through membranes and myoendothelial gap junctions. While a modicum of charge does spread to the underlying endothelium, it too will be efficiently dissipated. This is due to the robust expression of endothelial gap junctions, which prevent charge from effectively localizing (Sandow et al. 2003; Looft-Wilson et al. 2004). This was nicely illustrated in Fig. 2D by increasing endothelial-to-endothelial coupling resistance and subsequently demonstrating the augmented endothelial depolarization underneath the focal site of smooth muscle stimulation.

Contrary to the preceding simulations, past investigations have suggested that other cellular properties play a more important role in limiting smooth muscle-initiated conduction. For example, Duling and colleagues have previously ascribed the absence of smooth muscle-initiated conduction to the ability of voltage and/or Ca2+-activated K+ conductances to effectively feed back upon and dissipate the initial depolarization (Yashiro & Duling, 2000, 2003). Such work specifically focused on a complex mechanism whereby myoendothelial Ca2+ flux elevates endothelial [Ca2+], activates endothelial Ca2+-activated K+ channels, and initiates a reciprocal hyperpolarization that dissipates smooth muscle depolarization (Yashiro & Duling, 2000, 2003). Although interesting, this study found no evidence to support this line of logic. First, from a theoretical perspective, simulations (Fig. 2E and F) revealed that reducing the smooth muscle- or endothelial-ionic conductance by 50%, an alteration akin to K+ channel blockade, had little effect on promoting the conduction of a smooth muscle-initiated depolarization. Second, from a functional perspective, a full range of experiments illustrated that perfusing tissues (1) abluminally with a BKCa/KV inhibitor or (2) luminally with SK/IK channel blockers had no effect on enabling phenylephrine, U46619 or KCl to elicit a conducted response. Both blocker sets appeared to affect their designated targets as TEA increased resting arterial tone and apamin/TRAM34 eliminated acetylcholine-induced vasodilatation (Doughty et al. 1999; Crane et al. 2003; McNeish et al. 2006).

As evident from the preceding discussion, the absence of smooth muscle-initiated conduction is typically framed through the lens of certain cellular properties (i.e. cell orientation and coupling resistance) dissipating an initial electrical event. Although a valid perspective, such arguments fundamentally assume that all smooth muscle agents elicit an initial electrical response. While stimuli like KCl will depolarize smooth muscle by shifting the K+ equilibrium potential, it is not entirely certain whether agonists like phenylephrine and U46619 induce a comparable electrical event. Indeed, given the biophysical properties of vascular cells, it is quite conceivable that focal agonist application may completely fail to elicit a localized depolarization. This is because a small number of stimulated smooth muscle cells may not produce sufficient current to effectively compensate for the charge lost to neighbouring unstimulated cells functioning as charge sinks. This principle was illustrated in Fig. 4 by injecting a small quanta of current into a variable number of smooth muscle cells while monitoring the subsequent VM responses. When the number of stimulated smooth muscle cells was small relative to number of cellular charge sinks, current injection elicited a limited 1.4 mV depolarization. In contrast, when the number of stimulated smooth muscle cells was maximized, current injection elicited a sizable 15.0 mV depolarization. As noted in the Results, the magnitude of the injected current was set to produce a standard 15 mV depolarization under conditions in which all smooth muscle cells were stimulated. The magnitude of this current appears appropriate as past observations have shown that vasoconstrictor-induced changes in whole cell current are typical in the low picoamp range between the physiological voltages of −30 to −50 mV (Welsh & Brayden, 2001; Hayabuchi et al. 2001; Luykenaar et al. 2004). This value also corresponds well with the simulations in Fig. 5 showing that whole cell currents in the low picoamp range rapidly and sustainably alter smooth muscle VM by 7–13 mV. The innate ability of small currents to elicit substantial VM responses is the result of a smooth muscle cell's high input resistance. Electrophysiological studies generally have reported input resistance values ranging from 5 to 10 GΩ for isolated smooth muscle cells (Welsh & Brayden, 2001; Hayabuchi et al. 2001; Luykenaar et al. 2004). In our virtual model, with smooth muscle and endothelial cells interconnected with gap junctions, the calculated input resistance of the smooth muscle cell varied tightly between 2.1 and 2.3 GΩ.

To functionally address whether focal agonist application elicits depolarization, a range of straightforward experiments were performed on feed arteries. First, consistent with the idea that there are too few charge sources to elicit depolarization, L-type Ca2+ channel blockade had little effect on the focal constriction induced by phenylephrine or U46619 (Fig. 6). Diltiazem did attenuate focal KCl-induced constriction, a stimulus that depolarizes smooth muscle by forcing the K+ equilibrium potential rightward. Diltiazem also limited the constriction induced by superfused phenylephrine. This latter finding is consistent with simulations (Fig. 4) and illustrates that electromechanical coupling becomes more prominent as the number of charge sources increases. Given these findings, microelectrodes were subsequently used to measure VM and contrary to expectations focal phenylephrine application was initially shown to elicit considerable depolarization (Fig. 7). Recording inconsistencies, however, raised the possibility that these responses were artifactual, the result of a constricting vessel physically influencing the recording electrode. This concern was addressed by repeating experiments in the presence of a myosin light chain kinase inhibitor (i.e. ML-7). Consistent with physical artifact, focal phenylephrine application failed to elicit depolarization under these conditions. Control experiments confirmed that ML-7 application did not interfere with the depolarization induced by focal KCl application or global phenylephrine. The latter measure is particularly important in highlighting that ML-7 does not impair phenylephrine's ability to modulate important ionic conductances. Cumulatively, our computational and experimental data consistently illustrate that when the number of activated smooth muscle cells is limited, they are unlikely to generate the current needed to elicit the VM response required for conduction.

Consistent with this investigation, Xia & Duling (1998) have previously noted that focal phenylephrine application need not elicit arterial depolarization. Interestingly, this earlier work found that this absence could be overcome by lengthening pulse duration (Xia & Duling, 1998). While an intriguing perspective, this study could not replicate these previous observations. Findings in Fig. 8AD clearly indicate that lengthening pulse duration, from 5 to 20 s, does not elicit a diltiazem-sensitive constriction or induce a measurable depolarization. This lack of experimental support is perhaps unsurprising given the conceptual problems underlying the preceding study. For example, for lengthened pulse duration to induce focal depolarization, a large pool of ion channels would have to be activated within a restricted number of smooth muscle cells. This pool must be large to compensate for charge lost to neighbouring unstimulated cells acting as charge sinks. Figure 8E indicates that under such conditions each smooth muscle cell would have to generate ∼43.5 pA of net inward current to induce a focal 15 mV depolarization. This is exceedingly large and electrical studies rarely observe agonist-induced currents of this magnitude under quasi-physiological conditions. Furthermore, if one argued that such currents could be generated, then one must accept the possibility of a resistance artery depolarizing to +112 mV when all smooth muscle cells are simultaneously stimulated (i.e. a superfusion experiment). Such arguments are unrealistic and consequently highlight the need to re-interpret the electrical findings of Xia & Duling (1998). Perhaps, like this investigation, arterial constriction physically influenced the recording electrode, introducing an artifactual element into the electrical measurements.

Voltage-independent constriction

The preceding findings clearly stress that focal agonist-induced constriction is VM independent. These results raise intriguing questions with respect to the mechanisms underlying discrete vasomotor activity. Agonists could, in theory, induce focal VM-independent constriction by increasing the Ca2+ sensitivity of the contractile apparatus (Somlyo & Somlyo, 2000; Somlyo, 2002). To assess Ca2+ sensitization, we used a classic permeabilization approach whereby the ability of phenylephrine to mediate focal constriction was assessed under conditions in which cytosolic [Ca2+] was fixed. Consistent with a limited role for Ca2+ sensitization, focal phenylephrine application failed to elicit constriction when free cytosolic [Ca2+] was maintained at pCa 7.25 or 7.0. Although apparent, some interpretational caution is warranted since signalling components could have diffused during permeabilization thereby compromising our ability to monitor this biochemical process. There are, however, other contractile mechanisms in vascular smooth muscle that appear to be poorly or loosely coupled to VM. For example, agonists can mobilize SR-Ca2+ in a VM-independent manner and this release often takes the form of a Ca2+ wave (Kuo et al. 2003; Dai et al. 2006). Asynchronous Ca2+ waves arise from the activation of IP3 receptors and these events have been linked to arterial constriction (Kuo et al. 2003; Dai et al. 2006). Consistent with SR-Ca2+ mobilization contributing to discrete vasomotor activity, focal phenylephrine constriction was abolished in feed arteries exposed to agents that deplete SR-Ca2+ (i.e. thapsigargin) or which block IP3 receptors (2−APD and xestospongin C).

Functional implications

Vascular biologists have long recognized that in order to match blood flow delivery with metabolic demand, segments of an arterial tree must respond in isolation and in conjunction with one another (Duling et al. 1987; Segal, 1994; Segal & Kurjiaka, 1995). When reacting in isolation, an arterial segment can subtly ‘tune’ flow to a specified region without over- or under-perfusing neighbouring areas. In contrast, more encompassing changes in segmental tone enable a vascular network to dramatically increase blood flow to a greater tissue region, perhaps during periods of high metabolic activity (Kurjiaka & Segal, 1995; Segal & Jacobs, 2001). While the concept of ‘local and global’ control is integral to the vascular field, studies have yet to resolve how this differential behaviour is achieved in a syncytium of passively coupled cells. Based on our results, we suggest that local vasomotor responses occur when discrete stimuli activate a limited number of smooth muscle cells. Such activation would result in SR-Ca2+ mobilization but not a VM response. In comparison, we propose that global vasomotor responses occur when discrete stimuli directly activate the endothelium. These cells appear capable of generating the substantial currents required to elicit a focal VM response in a cell type retaining a low input resistance (Ledoux et al. 2008). Once initiated, this electrical response would conduct along the endothelium to the overlying smooth muscle eliciting broad changes in arterial tone. In closing, these arguments are important to blood flow regulation in that they assign the concept of global and local control to the electrical and/or non-electrical behaviour of specific cell types.

Summary

Using a combination of computational and experimental approaches, this study carefully developed two rationales to explain why smooth muscle-initiated responses fail to robustly conduct along skeletal muscle arteries. The first rationale stressed that smooth muscle electrical responses, if initiated, will actively dissipate as they spread from cell-to-cell along the arterial wall. This dissipation is promoted within the arterial wall by structural and connectivity properties of vascular cells. The second rationale centred on the idea that when agents other than KCl stimulate a limited number of smooth muscle cells, they are unable to generate the current required to initiate a localized membrane potential (VM) response. This insufficiency results in part from the charge lost to neighbouring unstimulated cells via gap junctions. By focusing on the intrinsic biophysical properties of vascular cells, this work provides new insight into the mechanistic basis of cell-to-cell communication and the physiological foundation of ‘local and global’ blood flow control. This quantitative knowledge is applicable to other vascular beds and can account for most but not all conducted behaviour. One noteworthy exception is hamster cheek pouch arteries where constrictor responses appear to spread along the smooth muscle layer (Barlett & Segal, 2000; Budel et al. 2003). Perhaps, this conducted response is supported in cheek pouch arterioles because: (1) coupling resistance is lower among neighbouring smooth muscle cells; and (2) smooth muscle cells express an ion channel capable of regenerating depolarizing current. Clearly, further exploration of this unique conducted response is warranted.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Appendix

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Acknowledgements

The authors would like to thank Suzanne Brett Welsh, Hai Kim Diep, Jaya Deep Tunuguntla and Kevin Luykenaar for their general contributions. This work was supported by an operating grant from the Heart and Stroke Foundation of Canada. D. G. Welsh is a Senior Scholar with the Alberta Heritage Foundation for Medical Research and a recipient of a Canada Research Chair (Tier II).