Cleveland Clinic Journal of Medicine

Interactions between the brain and blood are essential to health. Metabolic supply of the brain is provided through the vasculature, and disruptions of this relationship, in extreme cases such as stroke, is a key characteristic of neurologic disease. Neuro-hemodynamic coupling is also demonstrated in healthy individuals on faster time scales in functional hyperemia, the local increase in blood flow and volume that accompanies neural activity.^1,2

We have recently proposed a further level of interdependence between the two systems —ie, the hemoneural hypothesis—which predicts that hemodynamic events such as functional hyperemia will modulate neural activity.³ An impact of hemodynamics on neurons could occur through a number of mechanisms, including the activation of mechanoreceptors on astrocytes or neurons, a thermal impact of increased blood flow on ion channels and vesicle release, and the local increase and diffusion of blood-borne factors such as nitric oxide.^4,5 Astrocytes are predicted to play a key role in hemo-neural modulation, as they are tightly coupled to the vascular system and participate in a number of neural functions.^6,7 Through these mechanisms and others, hemodynamics could shift the “state” of the local neural circuit, thereby impacting information processing. This regulation of neural dynamics could also provide a homeostatic mechanism for promoting healthy brain function (eg, prevention of kindling).

To study the impact of hyperemia on neural and astrocytic activity in vivo, it is essential to independently control blood flow in the brain with means that do not directly impact neurons or astrocytes. Pinacidil is a sulfonylurea receptor agonist that opens the SUR2B potassium-sensitive ATP channel.⁸ In the telencephalon, SUR1 receptors are localized to neurons and glia.^9,10 In contrast, SUR2 receptors are localized to vasculature, with SUR2A in cardiac and skeletal muscle, and SUR2B in vascular smooth muscle, with primary expression in smaller arteries, arterioles, and capillaries.¹¹ By opening the SUR2B channel, pinacidil hyperpolarizes and relaxes smooth muscle, causing vasodilation. Pinacidil is a potent and selective SUR2B agonist, with a dissociation constant of 135 nM and a half maximal effective concentation (EC₅₀) value of 680 nM.¹² This agonist is approximately 5 times more specific for SUR2B than for SUR2A and shows approximately 5 orders of magnitude lower affinity for SUR1 (in the mM range).^12–15 Previous studies have demonstrated the efficacy of this agent as a vasodilator.^16–18

In the present study, we systematically examined the utility of pinacidil for the selective induction of hyperemia. First, we quantified the vasodilation induced by pinacidil in vivo, and examined local increases in blood volume in the parenchyma. These studies were conducted in anesthetized rats and awake mice. Second, we used in vitro slice recordings to examine whether direct application of relatively high concentrations of pinacidil would have any impact on the physiology of neurons and astrocytes. We found that (1) in vivo, pinacidil induces a level of vasodilation and increased local blood volume consistent with natural functional hyperemia across a variety of preparations, and (2) in vitro, pinacidil has no detectable impact on intrinsic biophysical measures in neurons and astrocytes.

METHODS

Animal preparation in vivo

To probe the impact of pinacidil on arterial diameter and parenchymal blood volume in vivo, we measured the effects of topical application to the primary somatosensory cortex (SI) of rats and mice. Sprague-Dawley rats (250–500 g) and C57BL/6 mice (~25 g) were anaesthetized with pentobarbital (50 mg/kg intraperitoneally initial dose, followed by 5-mg supplements as needed for maintenance). Animals were maintained at approximately 37°C by a heating blanket. Craniotomy (diameter of ~2 mm in rats, ~1 mm in mice) and durotomy were performed over SI, and the cortex was protected with Kwik-Cast silicone elastomer sealant (WPI, Sarasota, FL) while an imaging chamber was attached with dental cement. Kwik-Cast was removed, and the chamber filled with 0.9% saline and sealed with a round cover glass (avoiding bubbles) secured with cyanoacrylate.

Controlling visualization during drug delivery in vivo

To minimize brain motion and flow artifacts during visualization of hemodynamics in the rat preparation, we constructed a customized pressurized chamber with inflow and outflow for constant perfusion. The volume of the chamber was approximately 0.3 mL, and the flow through the system averaged about 2 mL/min. The chamber consisted of a plastic ring 1 cm in diameter and 3 mm high with a flat-top profile and a base shaped to the angle of the lateral skull edge over SI. In the wall of this chamber, three large holes were drilled and patched with pieces cut from rubber NMR septa (VWR International, West Chester, PA) to create resealable ports for drug application and bubble removal. Three additional permanent holes were drilled in the chamber walls, through which blunted 1-cm lengths of 18-gauge stainless steel needles were wedged and affixed with Super glue: one for artificial cerebrospinal fluid (ACSF) inflow, one for combined outflow, and the third for pressure regulation. The overall pressure of the chamber was regulated by a small vertical tube whose height (and thus fluid level) could be adjusted on a manipulator stand, and whose other end was open to the atmosphere. Inflow and outflow were controlled via regulators on a gravity feed system. In the mouse preparation, the need to control visualization was addressed by maintaining a constant rate of wicking in a smaller-profile open chamber, and a microfluidic switch with 12 μL of dead space was added to minimize propulsive impact and delay due to switching between solutions. Drug and ethanol solutions were delivered to rat and mouse chambers after being heated to physiological temperature (37°C).

Optical measurement of hemodynamics in vivo

We used a charge-coupled device camera (the Roper 512B, Princeton Instruments, Trenton, NJ) to image the cortical surface at a frame rate of approximately 4 Hz, with illumination from a voltage-regulated xenon arc lamp. A green band-pass filter (550 nm) was used to maximize imaging near the isosbestic point of hemoglobin, providing optimal vessel contrast and a surrogate measure for blood volume change in the parenchyma. Lenses (50 and 125 mm) were arranged in series to form a macroscope.

Pinacidil is hydrophobic and was therefore dissolved in ethanol at approximately 12 mg/kg and then diluted 1:100 in ACSF to achieve a 400-μM solution in 1% ethanol. Stock solutions were stored at −20°C and diluted in fresh ACSF for each experiment. For each run, the cortex was imaged for 3 minutes to establish baseline. For the pressurized rat chamber, at the end of the baseline period, 0.1 to 0.3 mL of 400 μM pinacidil in 1% ethanol in ACSF or saline would be pumped into the 0.3-mL chamber (taking about 1 second). Simultaneously, an equivalent volume was drawn out to balance pressure by a push-pull pump with access through two of the resealable rubber ports.

Pinacidil administration in vitro

Coronal slices were prepared from Sprague-Dawley rats at postnatal day 14 to 40 and maintained in a submersion chamber at 27°C for recording. Solutions were prepared in ACSF: 125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 1.25 mM NaH₂PO₄, 2 mM CaCl₂, 25 mM NaHCO₃, and 25 mM d-glucose). The applied solutions were 1% ethanol, ~400 μM pinacidil in 1% ethanol, and ACSF, all perfused with 5% CO₂ in 95% O₂ (carbogen).

For astrocyte recordings, slices were incubated immediately after cutting for 20 minutes in ACSF containing 50 μM sulforhodamine 101 (SR101), a water-soluble fluorescent dye specifically taken up by astrocytes.¹⁹ The slices were then allowed to rest for an hour before recording as usual. For fluorescence, the light source was a 100-W mercury arc lamp, with excitation and barrier filters and dichroic mirrors tailored to the spectral characteristics of SR101 (excitation ~586 nm, emission ~605 nm)

Slices were imaged under differential interference contrast optics with infrared illumination. Cells in layers 2/3 and in the same field and plane of view as a blood vessel greater than 20 μm in diameter were targeted, and vessel expansion was monitored during intracellular recording at approximately 4 Hz with a cooled charge-coupled device camera (Retiga EX, QImaging, Surrey, BC, Canada) connected to the microscope via a parfocal C-mount. This configuration also enabled imaging of neurons and astrocytes in the slice.

Drug/control solutions were switched every 90 to 180 seconds. Recording pipettes were filled with 120 mM KGlu, 10 mM NaCl, 20 mM KCl, 10 mM HEPES, 2mM Mg-ATP, 0.3 mM Na-GTP, 0.5 mM EGTA, and 0.3% to 1% biocytin (wt/vol) for subsequent visualization of the neurons.

RESULTS

Pinacidil induces vasodilation in anesthetized rats and awake mice

Figure 1A shows the cortical surface over SI. The green bisection line over the MCA shows the point of sampling for the darkness plot in the rectangular box in Figure 1B. The width of this dark band is the width of the MCA over time, showing expansion after pinacidil addition (gray bar). The point at which dilation is observed corresponds to a darkening in a parenchymal region (red triangle in Figure 1A) and over the MCA and surrounding cortex (purple rectangle in Figure 1A, reflecting expansion). Vasodilation and parenchymal signal increases were consistently observed on the first trial in all experiments (4 first runs from 4 anesthetized rats, Figure 1C and 1D). Dilation began less than 10 seconds after drug arrival, with maximal dilation and parenchymal darkening at an approximately 50- to 60-second latency. Ethanol in a 1:100 solution with ACSF under the same conditions evoked a nonsignificant reduction in vessel diameter and no change in parenchymal darkening. Following the first presentation, subsequent pinacidil effects were less reliable.

As shown in Figure 1E and 1F, the hemodynamic impact of pinacidil in anesthetized rats was replicated in awake, head-posted mice (2 mice, 2 runs each). Presentation of 220 or 440 μM pinacidil evoked comparable mean increases in arterial diameter (peak diameter increase of ~20%) and parenchymal darkening (peak increase of ~2%), effects that were repeatable within subjects in a single session (N = 2).

Pinacidil does not have direct effect on nonvascular tissue

Pinacidil does not impact spiking probability or input resistance in neurons or membrane potential in neurons and astrocytes

In recordings from regular-spiking neurons of pyramidal shape (N = 30), we saw no change in any metric measured. At 50 seconds after application, approximately the time of peak vasodilatory effects in vivo, the resting membrane potential did not change (variation of −0.5 ± 0.9 mV standard deviation; Figure 2E), the spike rate induced by current injection did not change (variation of 0.1 ± 1.1 spikes/stimulation; Figure 2F), and input resistance did not change (variation of −0.7 ± 6.5 MΩ; Figure 2D). Similarly, in a limited subset of recordings from fast-spiking interneurons (N = 3), we did not observe any impact of pinacidil application. All significance tests were paired t tests (P > .10).

Astrocytes (N = 35) also showed no significant effects of pinacidil application, demonstrating only a slow depolarization during application of ethanol (1.3 ± 1.7 mV at 50 seconds after drug application) and pinacidil with ethanol (0.5 ± 2.1 mV) (Figure 2B). When we plotted the change in membrane potential at 10-second intervals since drug application (0 to 80 seconds post-drug), we found no trends in astrocytic response to either ACSF or pinacidil (Figure 2C).

During recordings at double our typical application dose (800 μM), we observed 2 pyramidal cells (out of 8) that showed depolarization (peak of 10 to 15 mV) and a loss of spike initiation capability. Following washout, these cells recovered membrane potential but spiking responses to current injection remained impaired. We did not evaluate the impact of this dose on vascular tone or rhythmic vasomotion.

In contrast to the absence of a detectable impact of pinacidil, we found that the membrane potentials of neurons and astrocytes were sensitive to flow rate. Decreasing flow rate caused a consistent depolarization of up to approximately 10 mV that showed an immediate onset, reaching a new baseline within 2 to 5 seconds; increasing flow rate had the opposite effect. In preliminary experiments, we observed two astrocytes that depolarized on switching to the pinacidil solution. These two recordings were obtained prior to placement of an inline pressure meter in the flow pathway that allowed us to monitor and exclude trials that showed flow changes. In the 35 subsequent recordings that did not have flow changes, we never observed a detectable impact of either ethanol or pinacidil on astroctyes or neurons. We also noted that neurons and astrocytes were more likely to die and/or to lose recording quality during a cycle of ethanol or pinacidil presentation, as opposed to ACSF presentation.

DISCUSSION

Pinacidil provides an effective means of inducing vasodilation in vivo. At concentrations less than 400 μM, pinacidil is also selective for cortical vascular smooth muscle, exhibiting no direct effect on intrinsic properties of neurons or astrocytes. As an independent means to induce increased vasodilation and blood volume in a manner analogous to that seen in functional hyperemia, pinacidil provides a viable method for testing the impact of hyperemic events on neural or astrocytic activity. Pinacidil may also be a selective means of emulating other normal hemodynamic phenomena and could have therapeutic applications, such as targeted administration of pinacidil in response to acute vessel obstruction to maintain sufficient perfusion.

The hemodynamic effects induced by pinacidil are similar to natural functional hyperemia. In SI during sensory stimulation in rodents, increases in total oxygenated hemoglobin during sensory stimulation— analogous to our measurement of cortical darkening at 550 nm—peak in a range of 2% to 5%,^20–22 and arteries/arterioles dilate 10% to 20%.²³ The time course of pinacidil’s effects also parallels the sustained response to continued sensory drive. Arterial diameter in rodent SI and the blood oxygen level–dependent response on functional magnetic resonance imaging in humans and rodents remain high when tactile input is sustained for periods lasting tens of seconds,^23,24 as they do under pinacidil application.

Although pinacidil represents an important step forward in our ability to control blood flow while probing the impact of hemodynamics in cortex, it has limitations. The drug is only capable of producing vasodilation; drugs in the same family that block the SUR2B channels to create vasoconstriction (such as diazoxide or glibenclamide) or thromboxane receptor agonists^18,25,26 are unfortunately known to be nonspecific, affecting neurons as well as blood vessels. Pinacidil is also not water-soluble, requiring its dissolution in ethanol or DMSO, agents that can have confounding impacts on the system. Applied in vivo, pinacidil also does not appear to wash out fully, or its impact on smooth muscles persists, so that the first trial in each animal is the most consistent and effective one. These limitations stated, this pharmacological approach nevertheless represents a unique means of selective hyperemia induction in vivo.

Interactions between the brain and blood are essential to health. Metabolic supply of the brain is provided through the vasculature, and disruptions of this relationship, in extreme cases such as stroke, is a key characteristic of neurologic disease. Neuro-hemodynamic coupling is also demonstrated in healthy individuals on faster time scales in functional hyperemia, the local increase in blood flow and volume that accompanies neural activity.^1,2

We have recently proposed a further level of interdependence between the two systems —ie, the hemoneural hypothesis—which predicts that hemodynamic events such as functional hyperemia will modulate neural activity.³ An impact of hemodynamics on neurons could occur through a number of mechanisms, including the activation of mechanoreceptors on astrocytes or neurons, a thermal impact of increased blood flow on ion channels and vesicle release, and the local increase and diffusion of blood-borne factors such as nitric oxide.^4,5 Astrocytes are predicted to play a key role in hemo-neural modulation, as they are tightly coupled to the vascular system and participate in a number of neural functions.^6,7 Through these mechanisms and others, hemodynamics could shift the “state” of the local neural circuit, thereby impacting information processing. This regulation of neural dynamics could also provide a homeostatic mechanism for promoting healthy brain function (eg, prevention of kindling).

To study the impact of hyperemia on neural and astrocytic activity in vivo, it is essential to independently control blood flow in the brain with means that do not directly impact neurons or astrocytes. Pinacidil is a sulfonylurea receptor agonist that opens the SUR2B potassium-sensitive ATP channel.⁸ In the telencephalon, SUR1 receptors are localized to neurons and glia.^9,10 In contrast, SUR2 receptors are localized to vasculature, with SUR2A in cardiac and skeletal muscle, and SUR2B in vascular smooth muscle, with primary expression in smaller arteries, arterioles, and capillaries.¹¹ By opening the SUR2B channel, pinacidil hyperpolarizes and relaxes smooth muscle, causing vasodilation. Pinacidil is a potent and selective SUR2B agonist, with a dissociation constant of 135 nM and a half maximal effective concentation (EC₅₀) value of 680 nM.¹² This agonist is approximately 5 times more specific for SUR2B than for SUR2A and shows approximately 5 orders of magnitude lower affinity for SUR1 (in the mM range).^12–15 Previous studies have demonstrated the efficacy of this agent as a vasodilator.^16–18

In the present study, we systematically examined the utility of pinacidil for the selective induction of hyperemia. First, we quantified the vasodilation induced by pinacidil in vivo, and examined local increases in blood volume in the parenchyma. These studies were conducted in anesthetized rats and awake mice. Second, we used in vitro slice recordings to examine whether direct application of relatively high concentrations of pinacidil would have any impact on the physiology of neurons and astrocytes. We found that (1) in vivo, pinacidil induces a level of vasodilation and increased local blood volume consistent with natural functional hyperemia across a variety of preparations, and (2) in vitro, pinacidil has no detectable impact on intrinsic biophysical measures in neurons and astrocytes.

METHODS

Animal preparation in vivo

To probe the impact of pinacidil on arterial diameter and parenchymal blood volume in vivo, we measured the effects of topical application to the primary somatosensory cortex (SI) of rats and mice. Sprague-Dawley rats (250–500 g) and C57BL/6 mice (~25 g) were anaesthetized with pentobarbital (50 mg/kg intraperitoneally initial dose, followed by 5-mg supplements as needed for maintenance). Animals were maintained at approximately 37°C by a heating blanket. Craniotomy (diameter of ~2 mm in rats, ~1 mm in mice) and durotomy were performed over SI, and the cortex was protected with Kwik-Cast silicone elastomer sealant (WPI, Sarasota, FL) while an imaging chamber was attached with dental cement. Kwik-Cast was removed, and the chamber filled with 0.9% saline and sealed with a round cover glass (avoiding bubbles) secured with cyanoacrylate.

Controlling visualization during drug delivery in vivo

To minimize brain motion and flow artifacts during visualization of hemodynamics in the rat preparation, we constructed a customized pressurized chamber with inflow and outflow for constant perfusion. The volume of the chamber was approximately 0.3 mL, and the flow through the system averaged about 2 mL/min. The chamber consisted of a plastic ring 1 cm in diameter and 3 mm high with a flat-top profile and a base shaped to the angle of the lateral skull edge over SI. In the wall of this chamber, three large holes were drilled and patched with pieces cut from rubber NMR septa (VWR International, West Chester, PA) to create resealable ports for drug application and bubble removal. Three additional permanent holes were drilled in the chamber walls, through which blunted 1-cm lengths of 18-gauge stainless steel needles were wedged and affixed with Super glue: one for artificial cerebrospinal fluid (ACSF) inflow, one for combined outflow, and the third for pressure regulation. The overall pressure of the chamber was regulated by a small vertical tube whose height (and thus fluid level) could be adjusted on a manipulator stand, and whose other end was open to the atmosphere. Inflow and outflow were controlled via regulators on a gravity feed system. In the mouse preparation, the need to control visualization was addressed by maintaining a constant rate of wicking in a smaller-profile open chamber, and a microfluidic switch with 12 μL of dead space was added to minimize propulsive impact and delay due to switching between solutions. Drug and ethanol solutions were delivered to rat and mouse chambers after being heated to physiological temperature (37°C).

Optical measurement of hemodynamics in vivo

We used a charge-coupled device camera (the Roper 512B, Princeton Instruments, Trenton, NJ) to image the cortical surface at a frame rate of approximately 4 Hz, with illumination from a voltage-regulated xenon arc lamp. A green band-pass filter (550 nm) was used to maximize imaging near the isosbestic point of hemoglobin, providing optimal vessel contrast and a surrogate measure for blood volume change in the parenchyma. Lenses (50 and 125 mm) were arranged in series to form a macroscope.

Pinacidil is hydrophobic and was therefore dissolved in ethanol at approximately 12 mg/kg and then diluted 1:100 in ACSF to achieve a 400-μM solution in 1% ethanol. Stock solutions were stored at −20°C and diluted in fresh ACSF for each experiment. For each run, the cortex was imaged for 3 minutes to establish baseline. For the pressurized rat chamber, at the end of the baseline period, 0.1 to 0.3 mL of 400 μM pinacidil in 1% ethanol in ACSF or saline would be pumped into the 0.3-mL chamber (taking about 1 second). Simultaneously, an equivalent volume was drawn out to balance pressure by a push-pull pump with access through two of the resealable rubber ports.

Pinacidil administration in vitro

Coronal slices were prepared from Sprague-Dawley rats at postnatal day 14 to 40 and maintained in a submersion chamber at 27°C for recording. Solutions were prepared in ACSF: 125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 1.25 mM NaH₂PO₄, 2 mM CaCl₂, 25 mM NaHCO₃, and 25 mM d-glucose). The applied solutions were 1% ethanol, ~400 μM pinacidil in 1% ethanol, and ACSF, all perfused with 5% CO₂ in 95% O₂ (carbogen).

For astrocyte recordings, slices were incubated immediately after cutting for 20 minutes in ACSF containing 50 μM sulforhodamine 101 (SR101), a water-soluble fluorescent dye specifically taken up by astrocytes.¹⁹ The slices were then allowed to rest for an hour before recording as usual. For fluorescence, the light source was a 100-W mercury arc lamp, with excitation and barrier filters and dichroic mirrors tailored to the spectral characteristics of SR101 (excitation ~586 nm, emission ~605 nm)

Slices were imaged under differential interference contrast optics with infrared illumination. Cells in layers 2/3 and in the same field and plane of view as a blood vessel greater than 20 μm in diameter were targeted, and vessel expansion was monitored during intracellular recording at approximately 4 Hz with a cooled charge-coupled device camera (Retiga EX, QImaging, Surrey, BC, Canada) connected to the microscope via a parfocal C-mount. This configuration also enabled imaging of neurons and astrocytes in the slice.

Drug/control solutions were switched every 90 to 180 seconds. Recording pipettes were filled with 120 mM KGlu, 10 mM NaCl, 20 mM KCl, 10 mM HEPES, 2mM Mg-ATP, 0.3 mM Na-GTP, 0.5 mM EGTA, and 0.3% to 1% biocytin (wt/vol) for subsequent visualization of the neurons.

RESULTS

Pinacidil induces vasodilation in anesthetized rats and awake mice

Figure 1A shows the cortical surface over SI. The green bisection line over the MCA shows the point of sampling for the darkness plot in the rectangular box in Figure 1B. The width of this dark band is the width of the MCA over time, showing expansion after pinacidil addition (gray bar). The point at which dilation is observed corresponds to a darkening in a parenchymal region (red triangle in Figure 1A) and over the MCA and surrounding cortex (purple rectangle in Figure 1A, reflecting expansion). Vasodilation and parenchymal signal increases were consistently observed on the first trial in all experiments (4 first runs from 4 anesthetized rats, Figure 1C and 1D). Dilation began less than 10 seconds after drug arrival, with maximal dilation and parenchymal darkening at an approximately 50- to 60-second latency. Ethanol in a 1:100 solution with ACSF under the same conditions evoked a nonsignificant reduction in vessel diameter and no change in parenchymal darkening. Following the first presentation, subsequent pinacidil effects were less reliable.

As shown in Figure 1E and 1F, the hemodynamic impact of pinacidil in anesthetized rats was replicated in awake, head-posted mice (2 mice, 2 runs each). Presentation of 220 or 440 μM pinacidil evoked comparable mean increases in arterial diameter (peak diameter increase of ~20%) and parenchymal darkening (peak increase of ~2%), effects that were repeatable within subjects in a single session (N = 2).

Pinacidil does not have direct effect on nonvascular tissue

Pinacidil does not impact spiking probability or input resistance in neurons or membrane potential in neurons and astrocytes

In recordings from regular-spiking neurons of pyramidal shape (N = 30), we saw no change in any metric measured. At 50 seconds after application, approximately the time of peak vasodilatory effects in vivo, the resting membrane potential did not change (variation of −0.5 ± 0.9 mV standard deviation; Figure 2E), the spike rate induced by current injection did not change (variation of 0.1 ± 1.1 spikes/stimulation; Figure 2F), and input resistance did not change (variation of −0.7 ± 6.5 MΩ; Figure 2D). Similarly, in a limited subset of recordings from fast-spiking interneurons (N = 3), we did not observe any impact of pinacidil application. All significance tests were paired t tests (P > .10).

Astrocytes (N = 35) also showed no significant effects of pinacidil application, demonstrating only a slow depolarization during application of ethanol (1.3 ± 1.7 mV at 50 seconds after drug application) and pinacidil with ethanol (0.5 ± 2.1 mV) (Figure 2B). When we plotted the change in membrane potential at 10-second intervals since drug application (0 to 80 seconds post-drug), we found no trends in astrocytic response to either ACSF or pinacidil (Figure 2C).

During recordings at double our typical application dose (800 μM), we observed 2 pyramidal cells (out of 8) that showed depolarization (peak of 10 to 15 mV) and a loss of spike initiation capability. Following washout, these cells recovered membrane potential but spiking responses to current injection remained impaired. We did not evaluate the impact of this dose on vascular tone or rhythmic vasomotion.

In contrast to the absence of a detectable impact of pinacidil, we found that the membrane potentials of neurons and astrocytes were sensitive to flow rate. Decreasing flow rate caused a consistent depolarization of up to approximately 10 mV that showed an immediate onset, reaching a new baseline within 2 to 5 seconds; increasing flow rate had the opposite effect. In preliminary experiments, we observed two astrocytes that depolarized on switching to the pinacidil solution. These two recordings were obtained prior to placement of an inline pressure meter in the flow pathway that allowed us to monitor and exclude trials that showed flow changes. In the 35 subsequent recordings that did not have flow changes, we never observed a detectable impact of either ethanol or pinacidil on astroctyes or neurons. We also noted that neurons and astrocytes were more likely to die and/or to lose recording quality during a cycle of ethanol or pinacidil presentation, as opposed to ACSF presentation.

DISCUSSION

Pinacidil provides an effective means of inducing vasodilation in vivo. At concentrations less than 400 μM, pinacidil is also selective for cortical vascular smooth muscle, exhibiting no direct effect on intrinsic properties of neurons or astrocytes. As an independent means to induce increased vasodilation and blood volume in a manner analogous to that seen in functional hyperemia, pinacidil provides a viable method for testing the impact of hyperemic events on neural or astrocytic activity. Pinacidil may also be a selective means of emulating other normal hemodynamic phenomena and could have therapeutic applications, such as targeted administration of pinacidil in response to acute vessel obstruction to maintain sufficient perfusion.

The hemodynamic effects induced by pinacidil are similar to natural functional hyperemia. In SI during sensory stimulation in rodents, increases in total oxygenated hemoglobin during sensory stimulation— analogous to our measurement of cortical darkening at 550 nm—peak in a range of 2% to 5%,^20–22 and arteries/arterioles dilate 10% to 20%.²³ The time course of pinacidil’s effects also parallels the sustained response to continued sensory drive. Arterial diameter in rodent SI and the blood oxygen level–dependent response on functional magnetic resonance imaging in humans and rodents remain high when tactile input is sustained for periods lasting tens of seconds,^23,24 as they do under pinacidil application.

Although pinacidil represents an important step forward in our ability to control blood flow while probing the impact of hemodynamics in cortex, it has limitations. The drug is only capable of producing vasodilation; drugs in the same family that block the SUR2B channels to create vasoconstriction (such as diazoxide or glibenclamide) or thromboxane receptor agonists^18,25,26 are unfortunately known to be nonspecific, affecting neurons as well as blood vessels. Pinacidil is also not water-soluble, requiring its dissolution in ethanol or DMSO, agents that can have confounding impacts on the system. Applied in vivo, pinacidil also does not appear to wash out fully, or its impact on smooth muscles persists, so that the first trial in each animal is the most consistent and effective one. These limitations stated, this pharmacological approach nevertheless represents a unique means of selective hyperemia induction in vivo.

Interactions between the brain and blood are essential to health. Metabolic supply of the brain is provided through the vasculature, and disruptions of this relationship, in extreme cases such as stroke, is a key characteristic of neurologic disease. Neuro-hemodynamic coupling is also demonstrated in healthy individuals on faster time scales in functional hyperemia, the local increase in blood flow and volume that accompanies neural activity.^1,2

We have recently proposed a further level of interdependence between the two systems —ie, the hemoneural hypothesis—which predicts that hemodynamic events such as functional hyperemia will modulate neural activity.³ An impact of hemodynamics on neurons could occur through a number of mechanisms, including the activation of mechanoreceptors on astrocytes or neurons, a thermal impact of increased blood flow on ion channels and vesicle release, and the local increase and diffusion of blood-borne factors such as nitric oxide.^4,5 Astrocytes are predicted to play a key role in hemo-neural modulation, as they are tightly coupled to the vascular system and participate in a number of neural functions.^6,7 Through these mechanisms and others, hemodynamics could shift the “state” of the local neural circuit, thereby impacting information processing. This regulation of neural dynamics could also provide a homeostatic mechanism for promoting healthy brain function (eg, prevention of kindling).

To study the impact of hyperemia on neural and astrocytic activity in vivo, it is essential to independently control blood flow in the brain with means that do not directly impact neurons or astrocytes. Pinacidil is a sulfonylurea receptor agonist that opens the SUR2B potassium-sensitive ATP channel.⁸ In the telencephalon, SUR1 receptors are localized to neurons and glia.^9,10 In contrast, SUR2 receptors are localized to vasculature, with SUR2A in cardiac and skeletal muscle, and SUR2B in vascular smooth muscle, with primary expression in smaller arteries, arterioles, and capillaries.¹¹ By opening the SUR2B channel, pinacidil hyperpolarizes and relaxes smooth muscle, causing vasodilation. Pinacidil is a potent and selective SUR2B agonist, with a dissociation constant of 135 nM and a half maximal effective concentation (EC₅₀) value of 680 nM.¹² This agonist is approximately 5 times more specific for SUR2B than for SUR2A and shows approximately 5 orders of magnitude lower affinity for SUR1 (in the mM range).^12–15 Previous studies have demonstrated the efficacy of this agent as a vasodilator.^16–18

In the present study, we systematically examined the utility of pinacidil for the selective induction of hyperemia. First, we quantified the vasodilation induced by pinacidil in vivo, and examined local increases in blood volume in the parenchyma. These studies were conducted in anesthetized rats and awake mice. Second, we used in vitro slice recordings to examine whether direct application of relatively high concentrations of pinacidil would have any impact on the physiology of neurons and astrocytes. We found that (1) in vivo, pinacidil induces a level of vasodilation and increased local blood volume consistent with natural functional hyperemia across a variety of preparations, and (2) in vitro, pinacidil has no detectable impact on intrinsic biophysical measures in neurons and astrocytes.

METHODS

Animal preparation in vivo

To probe the impact of pinacidil on arterial diameter and parenchymal blood volume in vivo, we measured the effects of topical application to the primary somatosensory cortex (SI) of rats and mice. Sprague-Dawley rats (250–500 g) and C57BL/6 mice (~25 g) were anaesthetized with pentobarbital (50 mg/kg intraperitoneally initial dose, followed by 5-mg supplements as needed for maintenance). Animals were maintained at approximately 37°C by a heating blanket. Craniotomy (diameter of ~2 mm in rats, ~1 mm in mice) and durotomy were performed over SI, and the cortex was protected with Kwik-Cast silicone elastomer sealant (WPI, Sarasota, FL) while an imaging chamber was attached with dental cement. Kwik-Cast was removed, and the chamber filled with 0.9% saline and sealed with a round cover glass (avoiding bubbles) secured with cyanoacrylate.

Controlling visualization during drug delivery in vivo

To minimize brain motion and flow artifacts during visualization of hemodynamics in the rat preparation, we constructed a customized pressurized chamber with inflow and outflow for constant perfusion. The volume of the chamber was approximately 0.3 mL, and the flow through the system averaged about 2 mL/min. The chamber consisted of a plastic ring 1 cm in diameter and 3 mm high with a flat-top profile and a base shaped to the angle of the lateral skull edge over SI. In the wall of this chamber, three large holes were drilled and patched with pieces cut from rubber NMR septa (VWR International, West Chester, PA) to create resealable ports for drug application and bubble removal. Three additional permanent holes were drilled in the chamber walls, through which blunted 1-cm lengths of 18-gauge stainless steel needles were wedged and affixed with Super glue: one for artificial cerebrospinal fluid (ACSF) inflow, one for combined outflow, and the third for pressure regulation. The overall pressure of the chamber was regulated by a small vertical tube whose height (and thus fluid level) could be adjusted on a manipulator stand, and whose other end was open to the atmosphere. Inflow and outflow were controlled via regulators on a gravity feed system. In the mouse preparation, the need to control visualization was addressed by maintaining a constant rate of wicking in a smaller-profile open chamber, and a microfluidic switch with 12 μL of dead space was added to minimize propulsive impact and delay due to switching between solutions. Drug and ethanol solutions were delivered to rat and mouse chambers after being heated to physiological temperature (37°C).

Optical measurement of hemodynamics in vivo

We used a charge-coupled device camera (the Roper 512B, Princeton Instruments, Trenton, NJ) to image the cortical surface at a frame rate of approximately 4 Hz, with illumination from a voltage-regulated xenon arc lamp. A green band-pass filter (550 nm) was used to maximize imaging near the isosbestic point of hemoglobin, providing optimal vessel contrast and a surrogate measure for blood volume change in the parenchyma. Lenses (50 and 125 mm) were arranged in series to form a macroscope.

Pinacidil is hydrophobic and was therefore dissolved in ethanol at approximately 12 mg/kg and then diluted 1:100 in ACSF to achieve a 400-μM solution in 1% ethanol. Stock solutions were stored at −20°C and diluted in fresh ACSF for each experiment. For each run, the cortex was imaged for 3 minutes to establish baseline. For the pressurized rat chamber, at the end of the baseline period, 0.1 to 0.3 mL of 400 μM pinacidil in 1% ethanol in ACSF or saline would be pumped into the 0.3-mL chamber (taking about 1 second). Simultaneously, an equivalent volume was drawn out to balance pressure by a push-pull pump with access through two of the resealable rubber ports.

Pinacidil administration in vitro

Coronal slices were prepared from Sprague-Dawley rats at postnatal day 14 to 40 and maintained in a submersion chamber at 27°C for recording. Solutions were prepared in ACSF: 125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 1.25 mM NaH₂PO₄, 2 mM CaCl₂, 25 mM NaHCO₃, and 25 mM d-glucose). The applied solutions were 1% ethanol, ~400 μM pinacidil in 1% ethanol, and ACSF, all perfused with 5% CO₂ in 95% O₂ (carbogen).

For astrocyte recordings, slices were incubated immediately after cutting for 20 minutes in ACSF containing 50 μM sulforhodamine 101 (SR101), a water-soluble fluorescent dye specifically taken up by astrocytes.¹⁹ The slices were then allowed to rest for an hour before recording as usual. For fluorescence, the light source was a 100-W mercury arc lamp, with excitation and barrier filters and dichroic mirrors tailored to the spectral characteristics of SR101 (excitation ~586 nm, emission ~605 nm)

Slices were imaged under differential interference contrast optics with infrared illumination. Cells in layers 2/3 and in the same field and plane of view as a blood vessel greater than 20 μm in diameter were targeted, and vessel expansion was monitored during intracellular recording at approximately 4 Hz with a cooled charge-coupled device camera (Retiga EX, QImaging, Surrey, BC, Canada) connected to the microscope via a parfocal C-mount. This configuration also enabled imaging of neurons and astrocytes in the slice.

Drug/control solutions were switched every 90 to 180 seconds. Recording pipettes were filled with 120 mM KGlu, 10 mM NaCl, 20 mM KCl, 10 mM HEPES, 2mM Mg-ATP, 0.3 mM Na-GTP, 0.5 mM EGTA, and 0.3% to 1% biocytin (wt/vol) for subsequent visualization of the neurons.

RESULTS

Pinacidil induces vasodilation in anesthetized rats and awake mice

Figure 1A shows the cortical surface over SI. The green bisection line over the MCA shows the point of sampling for the darkness plot in the rectangular box in Figure 1B. The width of this dark band is the width of the MCA over time, showing expansion after pinacidil addition (gray bar). The point at which dilation is observed corresponds to a darkening in a parenchymal region (red triangle in Figure 1A) and over the MCA and surrounding cortex (purple rectangle in Figure 1A, reflecting expansion). Vasodilation and parenchymal signal increases were consistently observed on the first trial in all experiments (4 first runs from 4 anesthetized rats, Figure 1C and 1D). Dilation began less than 10 seconds after drug arrival, with maximal dilation and parenchymal darkening at an approximately 50- to 60-second latency. Ethanol in a 1:100 solution with ACSF under the same conditions evoked a nonsignificant reduction in vessel diameter and no change in parenchymal darkening. Following the first presentation, subsequent pinacidil effects were less reliable.

As shown in Figure 1E and 1F, the hemodynamic impact of pinacidil in anesthetized rats was replicated in awake, head-posted mice (2 mice, 2 runs each). Presentation of 220 or 440 μM pinacidil evoked comparable mean increases in arterial diameter (peak diameter increase of ~20%) and parenchymal darkening (peak increase of ~2%), effects that were repeatable within subjects in a single session (N = 2).

Pinacidil does not have direct effect on nonvascular tissue

Pinacidil does not impact spiking probability or input resistance in neurons or membrane potential in neurons and astrocytes

In recordings from regular-spiking neurons of pyramidal shape (N = 30), we saw no change in any metric measured. At 50 seconds after application, approximately the time of peak vasodilatory effects in vivo, the resting membrane potential did not change (variation of −0.5 ± 0.9 mV standard deviation; Figure 2E), the spike rate induced by current injection did not change (variation of 0.1 ± 1.1 spikes/stimulation; Figure 2F), and input resistance did not change (variation of −0.7 ± 6.5 MΩ; Figure 2D). Similarly, in a limited subset of recordings from fast-spiking interneurons (N = 3), we did not observe any impact of pinacidil application. All significance tests were paired t tests (P > .10).

Astrocytes (N = 35) also showed no significant effects of pinacidil application, demonstrating only a slow depolarization during application of ethanol (1.3 ± 1.7 mV at 50 seconds after drug application) and pinacidil with ethanol (0.5 ± 2.1 mV) (Figure 2B). When we plotted the change in membrane potential at 10-second intervals since drug application (0 to 80 seconds post-drug), we found no trends in astrocytic response to either ACSF or pinacidil (Figure 2C).

During recordings at double our typical application dose (800 μM), we observed 2 pyramidal cells (out of 8) that showed depolarization (peak of 10 to 15 mV) and a loss of spike initiation capability. Following washout, these cells recovered membrane potential but spiking responses to current injection remained impaired. We did not evaluate the impact of this dose on vascular tone or rhythmic vasomotion.

In contrast to the absence of a detectable impact of pinacidil, we found that the membrane potentials of neurons and astrocytes were sensitive to flow rate. Decreasing flow rate caused a consistent depolarization of up to approximately 10 mV that showed an immediate onset, reaching a new baseline within 2 to 5 seconds; increasing flow rate had the opposite effect. In preliminary experiments, we observed two astrocytes that depolarized on switching to the pinacidil solution. These two recordings were obtained prior to placement of an inline pressure meter in the flow pathway that allowed us to monitor and exclude trials that showed flow changes. In the 35 subsequent recordings that did not have flow changes, we never observed a detectable impact of either ethanol or pinacidil on astroctyes or neurons. We also noted that neurons and astrocytes were more likely to die and/or to lose recording quality during a cycle of ethanol or pinacidil presentation, as opposed to ACSF presentation.

DISCUSSION

Pinacidil provides an effective means of inducing vasodilation in vivo. At concentrations less than 400 μM, pinacidil is also selective for cortical vascular smooth muscle, exhibiting no direct effect on intrinsic properties of neurons or astrocytes. As an independent means to induce increased vasodilation and blood volume in a manner analogous to that seen in functional hyperemia, pinacidil provides a viable method for testing the impact of hyperemic events on neural or astrocytic activity. Pinacidil may also be a selective means of emulating other normal hemodynamic phenomena and could have therapeutic applications, such as targeted administration of pinacidil in response to acute vessel obstruction to maintain sufficient perfusion.

The hemodynamic effects induced by pinacidil are similar to natural functional hyperemia. In SI during sensory stimulation in rodents, increases in total oxygenated hemoglobin during sensory stimulation— analogous to our measurement of cortical darkening at 550 nm—peak in a range of 2% to 5%,^20–22 and arteries/arterioles dilate 10% to 20%.²³ The time course of pinacidil’s effects also parallels the sustained response to continued sensory drive. Arterial diameter in rodent SI and the blood oxygen level–dependent response on functional magnetic resonance imaging in humans and rodents remain high when tactile input is sustained for periods lasting tens of seconds,^23,24 as they do under pinacidil application.

Although pinacidil represents an important step forward in our ability to control blood flow while probing the impact of hemodynamics in cortex, it has limitations. The drug is only capable of producing vasodilation; drugs in the same family that block the SUR2B channels to create vasoconstriction (such as diazoxide or glibenclamide) or thromboxane receptor agonists^18,25,26 are unfortunately known to be nonspecific, affecting neurons as well as blood vessels. Pinacidil is also not water-soluble, requiring its dissolution in ethanol or DMSO, agents that can have confounding impacts on the system. Applied in vivo, pinacidil also does not appear to wash out fully, or its impact on smooth muscles persists, so that the first trial in each animal is the most consistent and effective one. These limitations stated, this pharmacological approach nevertheless represents a unique means of selective hyperemia induction in vivo.

HISTORICAL PERSPECTIVES ON THE AUTONOMIC NERVOUS SYSTEM

Central nervous system regulation of visceral organs is the focus of several historic publications that have shaped the texture of physiological inquiry. For example, in 1872 Darwin acknowledged the dynamic neural relationship between the heart and the brain:

. . .when the heart is affected it reacts on the brain; and the state of the brain again reacts through the pneumo-gastric [vagus] nerve on the heart; so that under any excitement there will be much mutual action and reaction between these, the two most important organs of the body.¹

Although Darwin acknowledged the bidirectional communication between the viscera and the brain, subsequent formal description of the autonomic nervous system (eg, by Langley²) minimized the importance of central regulatory structures and afferents. Following Langley, medical and physiological research tended to focus on the peripheral motor nerves of the autonomic nervous sytem, with a conceptual emphasis on the paired antagonism between sympathetic and parasympathetic efferent pathways on the target visceral organs. This focus minimized interest in both afferent pathways and the brainstem areas that regulate specific efferent pathways.

The early conceptualization of the vagus focused on an undifferentiated efferent pathway that was assumed to modulate “tone” concurrently to several target organs. Thus, brainstem areas regulating the supradiaphragmatic (eg, myelinated vagal pathways originating in the nucleus ambiguus and terminating primarily above the diaphragm) were not functionally distinguished from those regulating the subdiaphragmatic (eg, unmyelinated vagal pathways originating in the dorsal motor nucleus of the vagus and terminating primarily below the diaphragm). Without this distinction, research and theory focused on the paired antagonism between the parasympathetic and sympathetic innervation to target organs. The consequence of an emphasis on paired antagonism was an acceptance in physiology and medicine of global constructs such as autonomic balance, sympathetic tone, and vagal tone.

More than 50 years ago, Hess proposed that the autonomic nervous system was not solely vegetative and automatic but was instead an integrated system with both peripheral and central neurons.³ By emphasizing the central mechanisms that mediate the dynamic regulation of peripheral organs, Hess anticipated the need for technologies to continuously monitor peripheral and central neural circuits involved in the regulation of visceral function.

THE VAGAL PARADOX

In 1992, I proposed that an estimate of vagal tone, derived from measuring respiratory sinus arrhythmia, could be used in clinical medicine as an index of stress vulnerability.⁴ Rather than using the descriptive measures of heart rate variability (ie, beat-to-beat variability) frequently used in obstetrics and pediatrics, the paper emphasized that respiratory sinus arrhythmia has a neural origin and represents the tonic functional outflow from the vagus to the heart (ie, cardiac vagal tone). Thus, it was proposed that respiratory sinus arrhythmia would provide a more sensitive index of health status than a more global measure of beat-to-beat heart rate variability reflecting undetermined neural and nonneural mechanisms. The paper presented a quantitative approach that applied time-series analyses to extract the amplitude of respiratory sinus arrhythmia as a more accurate index of vagal activity. The article provided data demonstrating that healthy full-term infants had respiratory sinus arrhythmia of significantly greater amplitude than did preterm infants. This idea of using heart rate patterns to index vagal activity was not new, having been reported as early as 1910 by Hering.⁵ Moreover, contemporary studies have reliably reported that vagal blockade via atropine depresses respiratory sinus arrhythmia in mammals.^6,7

In response to this article,⁴ I received a letter from a neonatologist who wrote that, as a medical student, he learned that vagal tone could be lethal. He argued that perhaps too much of a good thing (ie, vagal tone) could be bad. He was referring, of course, to the clinical risk of neurogenic bradycardia. Bradycardia, when observed during delivery, may be an indicator of fetal distress. Similarly, bradycardia and apnea are important indicators of risk for the newborn.

My colleagues and I further investigated this perplexing observation by studying the human fetus during delivery. We observed that fetal bradycardia occurred only when respiratory sinus arrhythmia was depressed (ie, a respiratory rhythm in fetal heart rate is observable even in the absence of the large chest wall movements associated with breathing that occur postpartum).⁸ This raised the question of how vagal mechanisms could mediate both respiratory sinus arrhythmia and bradycardia, as one is protective and the other is potentially lethal. This inconsistency became the “vagal paradox” and served as the motivation behind the polyvagal theory.

With regard to the mechanisms mediating bradycardia and heart rate variability, there is an obvious inconsistency between data and physiological assumptions. Physiological models assume vagal regulation of both chronotropic control of the heart (ie, heart rate) and the amplitude of respiratory sinus arrhythmia.^9,10 For example, it has been reliably reported that vagal cardio-inhibitory fibers to the heart have consistent functional properties characterized by bradycardia to neural stimulation and a respiratory rhythm.⁹ However, although there are situations in which both measures covary (eg, during exercise and cholinergic blockade), there are other situations in which the measures appear to reflect independent sources of neural control (eg, bradycardic episodes associated with hypoxia, vasovagal syncope, and fetal distress). In contrast to these observable phenomena, researchers continue to argue for a covariation between these two parameters. This inconsistency, based on an assumption of a single central vagal source, is what I have labeled the vagal paradox.

THE POLYVAGAL THEORY: THREE PHYLOGENETIC RESPONSE SYSTEMS

Investigation of the phylogeny of the vertebrate autonomic nervous system provides an answer to the vagal paradox. Research in comparative neuroanatomy and neurophysiology has identified two branches of the vagus, with each branch supporting different adaptive functions and behavioral strategies. The vagal output to the heart from one branch is manifested in respiratory sinus arrhythmia, and the output from the other branch is manifested in bradycardia and possibly the slower rhythms in heart rate variability. Although the slower rhythms have been assumed to have a sympathetic influence, they are blocked by atropine.⁷

The polyvagal theory^7,11–15 articulates how each of three phylogenetic stages in the development of the vertebrate autonomic nervous system is associated with a distinct autonomic subsystem that is retained and expressed in mammals. These autonomic subsystems are phylogenetically ordered and behaviorally linked to social communication (eg, facial expression, vocalization, listening), mobilization (eg, fight–flight behaviors), and immobilization (eg, feigning death, vasovagal syncope, and behavioral shutdown).

The social communication system (ie, social engagement system; see below) involves the myelinated vagus, which serves to foster calm behavioral states by inhibiting sympathetic influences to the heart and dampening the hypothalamic-pituitary-adrenal (HPA) axis.¹⁶ The mobilization system is dependent on the functioning of the sympathetic nervous system. The most phylogenetically primitive component, the immobilization system, is dependent on the unmyelinated vagus, which is shared with most vertebrates. With increased neural complexity resulting from phylogenetic development, the organism’s behavioral and affective repertoire is enriched. The three circuits can be conceptualized as dynamic, providing adaptive responses to safe, dangerous, and life-threatening events and contexts.

Only mammals have a myelinated vagus. Unlike the unmyelinated vagus, originating in the dorsal motor nucleus of the vagus with pre- and postganglionic muscarinic receptors, the mammalian myelinated vagus originates in the nucleus ambiguus and has preganglionic nicotinic receptors and postganglionic muscarinic receptors. The unmyelinated vagus is shared with other vertebrates, including reptiles, amphibians, teleosts, and elasmobranchs.

We are now investigating the possibility of extracting different features of the heart rate pattern to dynamically monitor the two vagal systems. Preliminary studies in our laboratory support this possibility. In these studies we have blocked the nicotinic preganglionic receptors with hexamethonium and the muscarinic receptors with atropine. The data were collected from the prairie vole,¹⁷ which has a very high ambient vagal tone. These preliminary data demonstrated that, in several animals, nicotinic blockade selectively removes respiratory sinus arrhythmia without dampening the amplitude of the lower frequencies in heart rate variability. In contrast, blocking the muscarinic receptors with atropine removes both the low and respiratory frequencies.

CONSISTENCY WITH JACKSONIAN DISSOLUTION

The three circuits are organized and respond to challenges in a phylogenetically determined hierarchy consistent with the Jacksonian principle of dissolution. Jackson proposed that in the brain, higher (ie, phylogenetically newer) neural circuits inhibit lower (ie, phylogenetically older) neural circuits and “when the higher are suddenly rendered functionless, the lower rise in activity.” ¹⁸ Although Jackson proposed dissolution to explain changes in brain function due to damage and illness, the polyvagal theory proposes a similar phylogenetically ordered hierarchical model to describe the sequence of autonomic response strategies to challenges.

The human nervous system, similar to that of other mammals, evolved not solely to survive in safe environments but also to promote survival in dangerous and life-threatening contexts. To accomplish this adaptive flexibility, the human nervous system retained two more primitive neural circuits to regulate defensive strategies (ie, fight–flight and death-feigning behaviors). It is important to note that social behavior, social communication, and visceral homeostasis are incompatible with the neurophysiological states and behaviors promoted by the two neural circuits that support defense strategies. Thus, via evolution, the human nervous system retains three neural circuits, which are in a phylogenetically organized hierarchy. In this hierarchy of adaptive responses, the newest circuit is used first; if that circuit fails to provide safety, the older circuits are recruited sequentially.

Investigation of the phylogeny of regulation of the vertebrate heart^11,12,19,20 has led to extraction of four principles that provide a basis for testing of hypotheses relating specific neural mechanisms to social engagement, fight–flight, and death-feigning behaviors:

• There is a phylogenetic shift in the regulation of the heart from endocrine communication to unmyelinated nerves and finally to myelinated nerves.
• There is a development of opposing neural mechanisms of excitation and inhibition to provide rapid regulation of graded metabolic output.
• A face–heart connection evolved as source nuclei of vagal pathways shifted ventrally from the older dorsal motor nucleus to the nucleus ambiguus. This resulted in an anatomical and neurophysiological linkage between neural regulation of the heart via the myelinated vagus and the special visceral efferent pathways that regulate the striated muscles of the face and head, forming an integrated social engagement system (Figure 1; for more details, see Porges^7,15).
• With increased cortical development, the cortex exhibits greater control over the brainstem via direct (eg, corticobulbar) and indirect (eg, corticoreticular) neural pathways originating in motor cortex and terminating in the source nuclei of the myelinated motor nerves emerging from the brainstem (eg, specific neural pathways embedded within cranial nerves V, VII, IX, X, and XI), controlling visceromotor structures (ie, heart, bronchi) as well as somatomotor structures (muscles of the face and head).

NEUROCEPTION: CONTEXTUAL CUEING OF ADAPTIVE, MALADAPTIVE PHYSIOLOGICAL STATES

To effectively switch from defensive to social engagement strategies, the mammalian nervous system needs to perform two important adaptive tasks: (1) assess risk, and (2) if the environment is perceived as safe, inhibit the more primitive limbic structures that control fight, flight, or freeze behaviors.

Any stimulus that has the potential for increasing an organism’s experience of safety has the potential of recruiting the evolutionarily more advanced neural circuits that support the prosocial behaviors of the social engagement system.

The nervous system, through the processing of sensory information from the environment and from the viscera, continuously evaluates risk. Since the neural evaluation of risk does not require conscious awareness and may involve subcortical limbic structures,²¹ the term neuroception²² was introduced to emphasize a neural process, distinct from perception, that is capable of distinguishing environmental (and visceral) features that are safe, dangerous, or life-threatening. In safe environments, autonomic state is adaptively regulated to dampen sympathetic activation and to protect the oxygen-dependent central nervous system, especially the cortex, from the metabolically conservative reactions of the dorsal vagal complex. However, how does the nervous system know when the environment is safe, dangerous, or life-threatening, and which neural mechanisms evaluate this risk?

Environmental components of neuroception

Neuroception represents a neural process that enables humans and other mammals to engage in social behaviors by distinguishing safe from dangerous contexts. Neuroception is proposed as a plausible mechanism mediating both the expression and the disruption of positive social behavior, emotion regulation, and visceral homeostasis.^7,22 Neuroception might be triggered by feature detectors involving areas of temporal cortex that communicate with the central nucleus of the amygdala and the periaqueductal gray, since limbic reactivity is modulated by temporal cortex responses to the intention of voices, faces, and hand movements. Thus, the neuroception of familiar individuals and individuals with appropriately prosodic voices and warm, expressive faces translates into a social interaction promoting a sense of safety.

In most individuals (ie, those without a psychiatric disorder or neuropathology), the nervous system evaluates risk and matches neurophysiological state with the actual risk of the environment. When the environment is appraised as being safe, the defensive limbic structures are inhibited, enabling social engagement and calm visceral states to emerge. In contrast, some individuals experience a mismatch and the nervous system appraises the environment as being dangerous even when it is safe. This mismatch results in physiological states that support fight, flight, or freeze behaviors, but not social engagement behaviors. According to the theory, social communication can be expressed efficiently through the social engagement system only when these defensive circuits are inhibited.

Other contributors to neuroception

The features of risk in the environment do not solely drive neuroception. Afferent feedback from the viscera provides a major mediator of the accessibility of prosocial circuits associated with social engagement behaviors. For example, the polyvagal theory predicts that states of mobilization would compromise our ability to detect positive social cues. Functionally, visceral states color our perception of objects and others. Thus, the same features of one person engaging another may result in a range of outcomes, depending on the physiological state of the target individual. If the person being engaged is in a state in which the social engagement system is easily accessible, the reciprocal prosocial interactions are likely to occur. However, if the individual is in a state of mobilization, the same engaging response might be responded to with the asocial features of withdrawal or aggression. In such a state, it might be very difficult to dampen the mobilization circuit and enable the social engagement system to come back on line.

The insula may be involved in the mediation of neuroception, since it has been proposed as a brain structure involved in conveying the diffuse feedback from the viscera into cognitive awareness. Functional imaging experiments have demonstrated that the insula plays an important role in the experience of pain and the experience of several emotions, including anger, fear, disgust, happiness, and sadness. Critchley proposes that internal body states are represented in the insula and contribute to states of subjective feeling, and he has demonstrated that activity in the insula correlates with interoceptive accuracy.²³

SUMMARY

The polyvagal theory proposes that the evolution of the mammalian autonomic nervous system provides the neurophysiological substrates for adaptive behavioral strategies. It further proposes that physiological state limits the range of behavior and psychological experience. The theory links the evolution of the autonomic nervous system to affective experience, emotional expression, facial gestures, vocal communication, and contingent social behavior. In this way, the theory provides a plausible explanation for the reported covariation between atypical autonomic regulation (eg, reduced vagal and increased sympathetic influences to the heart) and psychiatric and behavioral disorders that involve difficulties in regulating appropriate social, emotional, and communication behaviors.

The polyvagal theory provides several insights into the adaptive nature of physiological state. First, the theory emphasizes that physiological states support different classes of behavior. For example, a physiological state characterized by a vagal withdrawal would support the mobilization behaviors of fight and flight. In contrast, a physiological state characterized by increased vagal influence on the heart (via myelinated vagal pathways originating in the nucleus ambiguus) would support spontaneous social engagement behaviors. Second, the theory emphasizes the formation of an integrated social engagement system through functional and structural links between neural control of the striated muscles of the face and the smooth muscles of the viscera. Third, the polyvagal theory proposes a mechanism—neuroception— to trigger or to inhibit defense strategies.

HISTORICAL PERSPECTIVES ON THE AUTONOMIC NERVOUS SYSTEM

Central nervous system regulation of visceral organs is the focus of several historic publications that have shaped the texture of physiological inquiry. For example, in 1872 Darwin acknowledged the dynamic neural relationship between the heart and the brain:

. . .when the heart is affected it reacts on the brain; and the state of the brain again reacts through the pneumo-gastric [vagus] nerve on the heart; so that under any excitement there will be much mutual action and reaction between these, the two most important organs of the body.¹

Although Darwin acknowledged the bidirectional communication between the viscera and the brain, subsequent formal description of the autonomic nervous system (eg, by Langley²) minimized the importance of central regulatory structures and afferents. Following Langley, medical and physiological research tended to focus on the peripheral motor nerves of the autonomic nervous sytem, with a conceptual emphasis on the paired antagonism between sympathetic and parasympathetic efferent pathways on the target visceral organs. This focus minimized interest in both afferent pathways and the brainstem areas that regulate specific efferent pathways.

The early conceptualization of the vagus focused on an undifferentiated efferent pathway that was assumed to modulate “tone” concurrently to several target organs. Thus, brainstem areas regulating the supradiaphragmatic (eg, myelinated vagal pathways originating in the nucleus ambiguus and terminating primarily above the diaphragm) were not functionally distinguished from those regulating the subdiaphragmatic (eg, unmyelinated vagal pathways originating in the dorsal motor nucleus of the vagus and terminating primarily below the diaphragm). Without this distinction, research and theory focused on the paired antagonism between the parasympathetic and sympathetic innervation to target organs. The consequence of an emphasis on paired antagonism was an acceptance in physiology and medicine of global constructs such as autonomic balance, sympathetic tone, and vagal tone.

More than 50 years ago, Hess proposed that the autonomic nervous system was not solely vegetative and automatic but was instead an integrated system with both peripheral and central neurons.³ By emphasizing the central mechanisms that mediate the dynamic regulation of peripheral organs, Hess anticipated the need for technologies to continuously monitor peripheral and central neural circuits involved in the regulation of visceral function.

THE VAGAL PARADOX

In 1992, I proposed that an estimate of vagal tone, derived from measuring respiratory sinus arrhythmia, could be used in clinical medicine as an index of stress vulnerability.⁴ Rather than using the descriptive measures of heart rate variability (ie, beat-to-beat variability) frequently used in obstetrics and pediatrics, the paper emphasized that respiratory sinus arrhythmia has a neural origin and represents the tonic functional outflow from the vagus to the heart (ie, cardiac vagal tone). Thus, it was proposed that respiratory sinus arrhythmia would provide a more sensitive index of health status than a more global measure of beat-to-beat heart rate variability reflecting undetermined neural and nonneural mechanisms. The paper presented a quantitative approach that applied time-series analyses to extract the amplitude of respiratory sinus arrhythmia as a more accurate index of vagal activity. The article provided data demonstrating that healthy full-term infants had respiratory sinus arrhythmia of significantly greater amplitude than did preterm infants. This idea of using heart rate patterns to index vagal activity was not new, having been reported as early as 1910 by Hering.⁵ Moreover, contemporary studies have reliably reported that vagal blockade via atropine depresses respiratory sinus arrhythmia in mammals.^6,7

In response to this article,⁴ I received a letter from a neonatologist who wrote that, as a medical student, he learned that vagal tone could be lethal. He argued that perhaps too much of a good thing (ie, vagal tone) could be bad. He was referring, of course, to the clinical risk of neurogenic bradycardia. Bradycardia, when observed during delivery, may be an indicator of fetal distress. Similarly, bradycardia and apnea are important indicators of risk for the newborn.

My colleagues and I further investigated this perplexing observation by studying the human fetus during delivery. We observed that fetal bradycardia occurred only when respiratory sinus arrhythmia was depressed (ie, a respiratory rhythm in fetal heart rate is observable even in the absence of the large chest wall movements associated with breathing that occur postpartum).⁸ This raised the question of how vagal mechanisms could mediate both respiratory sinus arrhythmia and bradycardia, as one is protective and the other is potentially lethal. This inconsistency became the “vagal paradox” and served as the motivation behind the polyvagal theory.

With regard to the mechanisms mediating bradycardia and heart rate variability, there is an obvious inconsistency between data and physiological assumptions. Physiological models assume vagal regulation of both chronotropic control of the heart (ie, heart rate) and the amplitude of respiratory sinus arrhythmia.^9,10 For example, it has been reliably reported that vagal cardio-inhibitory fibers to the heart have consistent functional properties characterized by bradycardia to neural stimulation and a respiratory rhythm.⁹ However, although there are situations in which both measures covary (eg, during exercise and cholinergic blockade), there are other situations in which the measures appear to reflect independent sources of neural control (eg, bradycardic episodes associated with hypoxia, vasovagal syncope, and fetal distress). In contrast to these observable phenomena, researchers continue to argue for a covariation between these two parameters. This inconsistency, based on an assumption of a single central vagal source, is what I have labeled the vagal paradox.

THE POLYVAGAL THEORY: THREE PHYLOGENETIC RESPONSE SYSTEMS

Investigation of the phylogeny of the vertebrate autonomic nervous system provides an answer to the vagal paradox. Research in comparative neuroanatomy and neurophysiology has identified two branches of the vagus, with each branch supporting different adaptive functions and behavioral strategies. The vagal output to the heart from one branch is manifested in respiratory sinus arrhythmia, and the output from the other branch is manifested in bradycardia and possibly the slower rhythms in heart rate variability. Although the slower rhythms have been assumed to have a sympathetic influence, they are blocked by atropine.⁷

The polyvagal theory^7,11–15 articulates how each of three phylogenetic stages in the development of the vertebrate autonomic nervous system is associated with a distinct autonomic subsystem that is retained and expressed in mammals. These autonomic subsystems are phylogenetically ordered and behaviorally linked to social communication (eg, facial expression, vocalization, listening), mobilization (eg, fight–flight behaviors), and immobilization (eg, feigning death, vasovagal syncope, and behavioral shutdown).

The social communication system (ie, social engagement system; see below) involves the myelinated vagus, which serves to foster calm behavioral states by inhibiting sympathetic influences to the heart and dampening the hypothalamic-pituitary-adrenal (HPA) axis.¹⁶ The mobilization system is dependent on the functioning of the sympathetic nervous system. The most phylogenetically primitive component, the immobilization system, is dependent on the unmyelinated vagus, which is shared with most vertebrates. With increased neural complexity resulting from phylogenetic development, the organism’s behavioral and affective repertoire is enriched. The three circuits can be conceptualized as dynamic, providing adaptive responses to safe, dangerous, and life-threatening events and contexts.

Only mammals have a myelinated vagus. Unlike the unmyelinated vagus, originating in the dorsal motor nucleus of the vagus with pre- and postganglionic muscarinic receptors, the mammalian myelinated vagus originates in the nucleus ambiguus and has preganglionic nicotinic receptors and postganglionic muscarinic receptors. The unmyelinated vagus is shared with other vertebrates, including reptiles, amphibians, teleosts, and elasmobranchs.

We are now investigating the possibility of extracting different features of the heart rate pattern to dynamically monitor the two vagal systems. Preliminary studies in our laboratory support this possibility. In these studies we have blocked the nicotinic preganglionic receptors with hexamethonium and the muscarinic receptors with atropine. The data were collected from the prairie vole,¹⁷ which has a very high ambient vagal tone. These preliminary data demonstrated that, in several animals, nicotinic blockade selectively removes respiratory sinus arrhythmia without dampening the amplitude of the lower frequencies in heart rate variability. In contrast, blocking the muscarinic receptors with atropine removes both the low and respiratory frequencies.

CONSISTENCY WITH JACKSONIAN DISSOLUTION

The three circuits are organized and respond to challenges in a phylogenetically determined hierarchy consistent with the Jacksonian principle of dissolution. Jackson proposed that in the brain, higher (ie, phylogenetically newer) neural circuits inhibit lower (ie, phylogenetically older) neural circuits and “when the higher are suddenly rendered functionless, the lower rise in activity.” ¹⁸ Although Jackson proposed dissolution to explain changes in brain function due to damage and illness, the polyvagal theory proposes a similar phylogenetically ordered hierarchical model to describe the sequence of autonomic response strategies to challenges.

The human nervous system, similar to that of other mammals, evolved not solely to survive in safe environments but also to promote survival in dangerous and life-threatening contexts. To accomplish this adaptive flexibility, the human nervous system retained two more primitive neural circuits to regulate defensive strategies (ie, fight–flight and death-feigning behaviors). It is important to note that social behavior, social communication, and visceral homeostasis are incompatible with the neurophysiological states and behaviors promoted by the two neural circuits that support defense strategies. Thus, via evolution, the human nervous system retains three neural circuits, which are in a phylogenetically organized hierarchy. In this hierarchy of adaptive responses, the newest circuit is used first; if that circuit fails to provide safety, the older circuits are recruited sequentially.

Investigation of the phylogeny of regulation of the vertebrate heart^11,12,19,20 has led to extraction of four principles that provide a basis for testing of hypotheses relating specific neural mechanisms to social engagement, fight–flight, and death-feigning behaviors:

• There is a phylogenetic shift in the regulation of the heart from endocrine communication to unmyelinated nerves and finally to myelinated nerves.
• There is a development of opposing neural mechanisms of excitation and inhibition to provide rapid regulation of graded metabolic output.
• A face–heart connection evolved as source nuclei of vagal pathways shifted ventrally from the older dorsal motor nucleus to the nucleus ambiguus. This resulted in an anatomical and neurophysiological linkage between neural regulation of the heart via the myelinated vagus and the special visceral efferent pathways that regulate the striated muscles of the face and head, forming an integrated social engagement system (Figure 1; for more details, see Porges^7,15).
• With increased cortical development, the cortex exhibits greater control over the brainstem via direct (eg, corticobulbar) and indirect (eg, corticoreticular) neural pathways originating in motor cortex and terminating in the source nuclei of the myelinated motor nerves emerging from the brainstem (eg, specific neural pathways embedded within cranial nerves V, VII, IX, X, and XI), controlling visceromotor structures (ie, heart, bronchi) as well as somatomotor structures (muscles of the face and head).

NEUROCEPTION: CONTEXTUAL CUEING OF ADAPTIVE, MALADAPTIVE PHYSIOLOGICAL STATES

To effectively switch from defensive to social engagement strategies, the mammalian nervous system needs to perform two important adaptive tasks: (1) assess risk, and (2) if the environment is perceived as safe, inhibit the more primitive limbic structures that control fight, flight, or freeze behaviors.

Any stimulus that has the potential for increasing an organism’s experience of safety has the potential of recruiting the evolutionarily more advanced neural circuits that support the prosocial behaviors of the social engagement system.

The nervous system, through the processing of sensory information from the environment and from the viscera, continuously evaluates risk. Since the neural evaluation of risk does not require conscious awareness and may involve subcortical limbic structures,²¹ the term neuroception²² was introduced to emphasize a neural process, distinct from perception, that is capable of distinguishing environmental (and visceral) features that are safe, dangerous, or life-threatening. In safe environments, autonomic state is adaptively regulated to dampen sympathetic activation and to protect the oxygen-dependent central nervous system, especially the cortex, from the metabolically conservative reactions of the dorsal vagal complex. However, how does the nervous system know when the environment is safe, dangerous, or life-threatening, and which neural mechanisms evaluate this risk?

Environmental components of neuroception

Neuroception represents a neural process that enables humans and other mammals to engage in social behaviors by distinguishing safe from dangerous contexts. Neuroception is proposed as a plausible mechanism mediating both the expression and the disruption of positive social behavior, emotion regulation, and visceral homeostasis.^7,22 Neuroception might be triggered by feature detectors involving areas of temporal cortex that communicate with the central nucleus of the amygdala and the periaqueductal gray, since limbic reactivity is modulated by temporal cortex responses to the intention of voices, faces, and hand movements. Thus, the neuroception of familiar individuals and individuals with appropriately prosodic voices and warm, expressive faces translates into a social interaction promoting a sense of safety.

In most individuals (ie, those without a psychiatric disorder or neuropathology), the nervous system evaluates risk and matches neurophysiological state with the actual risk of the environment. When the environment is appraised as being safe, the defensive limbic structures are inhibited, enabling social engagement and calm visceral states to emerge. In contrast, some individuals experience a mismatch and the nervous system appraises the environment as being dangerous even when it is safe. This mismatch results in physiological states that support fight, flight, or freeze behaviors, but not social engagement behaviors. According to the theory, social communication can be expressed efficiently through the social engagement system only when these defensive circuits are inhibited.

Other contributors to neuroception

The features of risk in the environment do not solely drive neuroception. Afferent feedback from the viscera provides a major mediator of the accessibility of prosocial circuits associated with social engagement behaviors. For example, the polyvagal theory predicts that states of mobilization would compromise our ability to detect positive social cues. Functionally, visceral states color our perception of objects and others. Thus, the same features of one person engaging another may result in a range of outcomes, depending on the physiological state of the target individual. If the person being engaged is in a state in which the social engagement system is easily accessible, the reciprocal prosocial interactions are likely to occur. However, if the individual is in a state of mobilization, the same engaging response might be responded to with the asocial features of withdrawal or aggression. In such a state, it might be very difficult to dampen the mobilization circuit and enable the social engagement system to come back on line.

The insula may be involved in the mediation of neuroception, since it has been proposed as a brain structure involved in conveying the diffuse feedback from the viscera into cognitive awareness. Functional imaging experiments have demonstrated that the insula plays an important role in the experience of pain and the experience of several emotions, including anger, fear, disgust, happiness, and sadness. Critchley proposes that internal body states are represented in the insula and contribute to states of subjective feeling, and he has demonstrated that activity in the insula correlates with interoceptive accuracy.²³

SUMMARY

The polyvagal theory proposes that the evolution of the mammalian autonomic nervous system provides the neurophysiological substrates for adaptive behavioral strategies. It further proposes that physiological state limits the range of behavior and psychological experience. The theory links the evolution of the autonomic nervous system to affective experience, emotional expression, facial gestures, vocal communication, and contingent social behavior. In this way, the theory provides a plausible explanation for the reported covariation between atypical autonomic regulation (eg, reduced vagal and increased sympathetic influences to the heart) and psychiatric and behavioral disorders that involve difficulties in regulating appropriate social, emotional, and communication behaviors.

The polyvagal theory provides several insights into the adaptive nature of physiological state. First, the theory emphasizes that physiological states support different classes of behavior. For example, a physiological state characterized by a vagal withdrawal would support the mobilization behaviors of fight and flight. In contrast, a physiological state characterized by increased vagal influence on the heart (via myelinated vagal pathways originating in the nucleus ambiguus) would support spontaneous social engagement behaviors. Second, the theory emphasizes the formation of an integrated social engagement system through functional and structural links between neural control of the striated muscles of the face and the smooth muscles of the viscera. Third, the polyvagal theory proposes a mechanism—neuroception— to trigger or to inhibit defense strategies.

HISTORICAL PERSPECTIVES ON THE AUTONOMIC NERVOUS SYSTEM

Central nervous system regulation of visceral organs is the focus of several historic publications that have shaped the texture of physiological inquiry. For example, in 1872 Darwin acknowledged the dynamic neural relationship between the heart and the brain:

. . .when the heart is affected it reacts on the brain; and the state of the brain again reacts through the pneumo-gastric [vagus] nerve on the heart; so that under any excitement there will be much mutual action and reaction between these, the two most important organs of the body.¹

Although Darwin acknowledged the bidirectional communication between the viscera and the brain, subsequent formal description of the autonomic nervous system (eg, by Langley²) minimized the importance of central regulatory structures and afferents. Following Langley, medical and physiological research tended to focus on the peripheral motor nerves of the autonomic nervous sytem, with a conceptual emphasis on the paired antagonism between sympathetic and parasympathetic efferent pathways on the target visceral organs. This focus minimized interest in both afferent pathways and the brainstem areas that regulate specific efferent pathways.

The early conceptualization of the vagus focused on an undifferentiated efferent pathway that was assumed to modulate “tone” concurrently to several target organs. Thus, brainstem areas regulating the supradiaphragmatic (eg, myelinated vagal pathways originating in the nucleus ambiguus and terminating primarily above the diaphragm) were not functionally distinguished from those regulating the subdiaphragmatic (eg, unmyelinated vagal pathways originating in the dorsal motor nucleus of the vagus and terminating primarily below the diaphragm). Without this distinction, research and theory focused on the paired antagonism between the parasympathetic and sympathetic innervation to target organs. The consequence of an emphasis on paired antagonism was an acceptance in physiology and medicine of global constructs such as autonomic balance, sympathetic tone, and vagal tone.

More than 50 years ago, Hess proposed that the autonomic nervous system was not solely vegetative and automatic but was instead an integrated system with both peripheral and central neurons.³ By emphasizing the central mechanisms that mediate the dynamic regulation of peripheral organs, Hess anticipated the need for technologies to continuously monitor peripheral and central neural circuits involved in the regulation of visceral function.

THE VAGAL PARADOX

In 1992, I proposed that an estimate of vagal tone, derived from measuring respiratory sinus arrhythmia, could be used in clinical medicine as an index of stress vulnerability.⁴ Rather than using the descriptive measures of heart rate variability (ie, beat-to-beat variability) frequently used in obstetrics and pediatrics, the paper emphasized that respiratory sinus arrhythmia has a neural origin and represents the tonic functional outflow from the vagus to the heart (ie, cardiac vagal tone). Thus, it was proposed that respiratory sinus arrhythmia would provide a more sensitive index of health status than a more global measure of beat-to-beat heart rate variability reflecting undetermined neural and nonneural mechanisms. The paper presented a quantitative approach that applied time-series analyses to extract the amplitude of respiratory sinus arrhythmia as a more accurate index of vagal activity. The article provided data demonstrating that healthy full-term infants had respiratory sinus arrhythmia of significantly greater amplitude than did preterm infants. This idea of using heart rate patterns to index vagal activity was not new, having been reported as early as 1910 by Hering.⁵ Moreover, contemporary studies have reliably reported that vagal blockade via atropine depresses respiratory sinus arrhythmia in mammals.^6,7

In response to this article,⁴ I received a letter from a neonatologist who wrote that, as a medical student, he learned that vagal tone could be lethal. He argued that perhaps too much of a good thing (ie, vagal tone) could be bad. He was referring, of course, to the clinical risk of neurogenic bradycardia. Bradycardia, when observed during delivery, may be an indicator of fetal distress. Similarly, bradycardia and apnea are important indicators of risk for the newborn.

My colleagues and I further investigated this perplexing observation by studying the human fetus during delivery. We observed that fetal bradycardia occurred only when respiratory sinus arrhythmia was depressed (ie, a respiratory rhythm in fetal heart rate is observable even in the absence of the large chest wall movements associated with breathing that occur postpartum).⁸ This raised the question of how vagal mechanisms could mediate both respiratory sinus arrhythmia and bradycardia, as one is protective and the other is potentially lethal. This inconsistency became the “vagal paradox” and served as the motivation behind the polyvagal theory.

With regard to the mechanisms mediating bradycardia and heart rate variability, there is an obvious inconsistency between data and physiological assumptions. Physiological models assume vagal regulation of both chronotropic control of the heart (ie, heart rate) and the amplitude of respiratory sinus arrhythmia.^9,10 For example, it has been reliably reported that vagal cardio-inhibitory fibers to the heart have consistent functional properties characterized by bradycardia to neural stimulation and a respiratory rhythm.⁹ However, although there are situations in which both measures covary (eg, during exercise and cholinergic blockade), there are other situations in which the measures appear to reflect independent sources of neural control (eg, bradycardic episodes associated with hypoxia, vasovagal syncope, and fetal distress). In contrast to these observable phenomena, researchers continue to argue for a covariation between these two parameters. This inconsistency, based on an assumption of a single central vagal source, is what I have labeled the vagal paradox.

THE POLYVAGAL THEORY: THREE PHYLOGENETIC RESPONSE SYSTEMS

Investigation of the phylogeny of the vertebrate autonomic nervous system provides an answer to the vagal paradox. Research in comparative neuroanatomy and neurophysiology has identified two branches of the vagus, with each branch supporting different adaptive functions and behavioral strategies. The vagal output to the heart from one branch is manifested in respiratory sinus arrhythmia, and the output from the other branch is manifested in bradycardia and possibly the slower rhythms in heart rate variability. Although the slower rhythms have been assumed to have a sympathetic influence, they are blocked by atropine.⁷

The polyvagal theory^7,11–15 articulates how each of three phylogenetic stages in the development of the vertebrate autonomic nervous system is associated with a distinct autonomic subsystem that is retained and expressed in mammals. These autonomic subsystems are phylogenetically ordered and behaviorally linked to social communication (eg, facial expression, vocalization, listening), mobilization (eg, fight–flight behaviors), and immobilization (eg, feigning death, vasovagal syncope, and behavioral shutdown).

The social communication system (ie, social engagement system; see below) involves the myelinated vagus, which serves to foster calm behavioral states by inhibiting sympathetic influences to the heart and dampening the hypothalamic-pituitary-adrenal (HPA) axis.¹⁶ The mobilization system is dependent on the functioning of the sympathetic nervous system. The most phylogenetically primitive component, the immobilization system, is dependent on the unmyelinated vagus, which is shared with most vertebrates. With increased neural complexity resulting from phylogenetic development, the organism’s behavioral and affective repertoire is enriched. The three circuits can be conceptualized as dynamic, providing adaptive responses to safe, dangerous, and life-threatening events and contexts.

Only mammals have a myelinated vagus. Unlike the unmyelinated vagus, originating in the dorsal motor nucleus of the vagus with pre- and postganglionic muscarinic receptors, the mammalian myelinated vagus originates in the nucleus ambiguus and has preganglionic nicotinic receptors and postganglionic muscarinic receptors. The unmyelinated vagus is shared with other vertebrates, including reptiles, amphibians, teleosts, and elasmobranchs.

We are now investigating the possibility of extracting different features of the heart rate pattern to dynamically monitor the two vagal systems. Preliminary studies in our laboratory support this possibility. In these studies we have blocked the nicotinic preganglionic receptors with hexamethonium and the muscarinic receptors with atropine. The data were collected from the prairie vole,¹⁷ which has a very high ambient vagal tone. These preliminary data demonstrated that, in several animals, nicotinic blockade selectively removes respiratory sinus arrhythmia without dampening the amplitude of the lower frequencies in heart rate variability. In contrast, blocking the muscarinic receptors with atropine removes both the low and respiratory frequencies.

CONSISTENCY WITH JACKSONIAN DISSOLUTION

The three circuits are organized and respond to challenges in a phylogenetically determined hierarchy consistent with the Jacksonian principle of dissolution. Jackson proposed that in the brain, higher (ie, phylogenetically newer) neural circuits inhibit lower (ie, phylogenetically older) neural circuits and “when the higher are suddenly rendered functionless, the lower rise in activity.” ¹⁸ Although Jackson proposed dissolution to explain changes in brain function due to damage and illness, the polyvagal theory proposes a similar phylogenetically ordered hierarchical model to describe the sequence of autonomic response strategies to challenges.

The human nervous system, similar to that of other mammals, evolved not solely to survive in safe environments but also to promote survival in dangerous and life-threatening contexts. To accomplish this adaptive flexibility, the human nervous system retained two more primitive neural circuits to regulate defensive strategies (ie, fight–flight and death-feigning behaviors). It is important to note that social behavior, social communication, and visceral homeostasis are incompatible with the neurophysiological states and behaviors promoted by the two neural circuits that support defense strategies. Thus, via evolution, the human nervous system retains three neural circuits, which are in a phylogenetically organized hierarchy. In this hierarchy of adaptive responses, the newest circuit is used first; if that circuit fails to provide safety, the older circuits are recruited sequentially.

Investigation of the phylogeny of regulation of the vertebrate heart^11,12,19,20 has led to extraction of four principles that provide a basis for testing of hypotheses relating specific neural mechanisms to social engagement, fight–flight, and death-feigning behaviors:

• There is a phylogenetic shift in the regulation of the heart from endocrine communication to unmyelinated nerves and finally to myelinated nerves.
• There is a development of opposing neural mechanisms of excitation and inhibition to provide rapid regulation of graded metabolic output.
• A face–heart connection evolved as source nuclei of vagal pathways shifted ventrally from the older dorsal motor nucleus to the nucleus ambiguus. This resulted in an anatomical and neurophysiological linkage between neural regulation of the heart via the myelinated vagus and the special visceral efferent pathways that regulate the striated muscles of the face and head, forming an integrated social engagement system (Figure 1; for more details, see Porges^7,15).
• With increased cortical development, the cortex exhibits greater control over the brainstem via direct (eg, corticobulbar) and indirect (eg, corticoreticular) neural pathways originating in motor cortex and terminating in the source nuclei of the myelinated motor nerves emerging from the brainstem (eg, specific neural pathways embedded within cranial nerves V, VII, IX, X, and XI), controlling visceromotor structures (ie, heart, bronchi) as well as somatomotor structures (muscles of the face and head).

NEUROCEPTION: CONTEXTUAL CUEING OF ADAPTIVE, MALADAPTIVE PHYSIOLOGICAL STATES

To effectively switch from defensive to social engagement strategies, the mammalian nervous system needs to perform two important adaptive tasks: (1) assess risk, and (2) if the environment is perceived as safe, inhibit the more primitive limbic structures that control fight, flight, or freeze behaviors.

Any stimulus that has the potential for increasing an organism’s experience of safety has the potential of recruiting the evolutionarily more advanced neural circuits that support the prosocial behaviors of the social engagement system.

The nervous system, through the processing of sensory information from the environment and from the viscera, continuously evaluates risk. Since the neural evaluation of risk does not require conscious awareness and may involve subcortical limbic structures,²¹ the term neuroception²² was introduced to emphasize a neural process, distinct from perception, that is capable of distinguishing environmental (and visceral) features that are safe, dangerous, or life-threatening. In safe environments, autonomic state is adaptively regulated to dampen sympathetic activation and to protect the oxygen-dependent central nervous system, especially the cortex, from the metabolically conservative reactions of the dorsal vagal complex. However, how does the nervous system know when the environment is safe, dangerous, or life-threatening, and which neural mechanisms evaluate this risk?

Environmental components of neuroception

Neuroception represents a neural process that enables humans and other mammals to engage in social behaviors by distinguishing safe from dangerous contexts. Neuroception is proposed as a plausible mechanism mediating both the expression and the disruption of positive social behavior, emotion regulation, and visceral homeostasis.^7,22 Neuroception might be triggered by feature detectors involving areas of temporal cortex that communicate with the central nucleus of the amygdala and the periaqueductal gray, since limbic reactivity is modulated by temporal cortex responses to the intention of voices, faces, and hand movements. Thus, the neuroception of familiar individuals and individuals with appropriately prosodic voices and warm, expressive faces translates into a social interaction promoting a sense of safety.

In most individuals (ie, those without a psychiatric disorder or neuropathology), the nervous system evaluates risk and matches neurophysiological state with the actual risk of the environment. When the environment is appraised as being safe, the defensive limbic structures are inhibited, enabling social engagement and calm visceral states to emerge. In contrast, some individuals experience a mismatch and the nervous system appraises the environment as being dangerous even when it is safe. This mismatch results in physiological states that support fight, flight, or freeze behaviors, but not social engagement behaviors. According to the theory, social communication can be expressed efficiently through the social engagement system only when these defensive circuits are inhibited.

Other contributors to neuroception

The features of risk in the environment do not solely drive neuroception. Afferent feedback from the viscera provides a major mediator of the accessibility of prosocial circuits associated with social engagement behaviors. For example, the polyvagal theory predicts that states of mobilization would compromise our ability to detect positive social cues. Functionally, visceral states color our perception of objects and others. Thus, the same features of one person engaging another may result in a range of outcomes, depending on the physiological state of the target individual. If the person being engaged is in a state in which the social engagement system is easily accessible, the reciprocal prosocial interactions are likely to occur. However, if the individual is in a state of mobilization, the same engaging response might be responded to with the asocial features of withdrawal or aggression. In such a state, it might be very difficult to dampen the mobilization circuit and enable the social engagement system to come back on line.

The insula may be involved in the mediation of neuroception, since it has been proposed as a brain structure involved in conveying the diffuse feedback from the viscera into cognitive awareness. Functional imaging experiments have demonstrated that the insula plays an important role in the experience of pain and the experience of several emotions, including anger, fear, disgust, happiness, and sadness. Critchley proposes that internal body states are represented in the insula and contribute to states of subjective feeling, and he has demonstrated that activity in the insula correlates with interoceptive accuracy.²³

SUMMARY

The polyvagal theory proposes that the evolution of the mammalian autonomic nervous system provides the neurophysiological substrates for adaptive behavioral strategies. It further proposes that physiological state limits the range of behavior and psychological experience. The theory links the evolution of the autonomic nervous system to affective experience, emotional expression, facial gestures, vocal communication, and contingent social behavior. In this way, the theory provides a plausible explanation for the reported covariation between atypical autonomic regulation (eg, reduced vagal and increased sympathetic influences to the heart) and psychiatric and behavioral disorders that involve difficulties in regulating appropriate social, emotional, and communication behaviors.

The polyvagal theory provides several insights into the adaptive nature of physiological state. First, the theory emphasizes that physiological states support different classes of behavior. For example, a physiological state characterized by a vagal withdrawal would support the mobilization behaviors of fight and flight. In contrast, a physiological state characterized by increased vagal influence on the heart (via myelinated vagal pathways originating in the nucleus ambiguus) would support spontaneous social engagement behaviors. Second, the theory emphasizes the formation of an integrated social engagement system through functional and structural links between neural control of the striated muscles of the face and the smooth muscles of the viscera. Third, the polyvagal theory proposes a mechanism—neuroception— to trigger or to inhibit defense strategies.