“Real-time” imaging of cortical and subcortical sites of cardiovascular control: concurrent recordings of sympathetic nerve activity and fMRI in awake subjects

Abstract

We review our approach to functionally identifying cortical and subcortical areas involved in the generation of spontaneous fluctuations in sympathetic outflow to muscle or skin. We record muscle sympathetic nerve activity (MSNA) or skin sympathetic nerve activity (SSNA), via a tungsten microelectrode inserted percutaneously into the common peroneal nerve, at the same time as performing functional magnetic resonance imaging (fMRI) of the brain. By taking advantage of the neurovascular coupling delay associated with BOLD (blood oxygen level dependent) fMRI, and the delay associated with conduction of a burst of sympathetic impulses to the peripheral recording site, we can identify structures in which BOLD signal intensity covaries with MSNA or SSNA. Using this approach, we found MSNA-coupled increases in BOLD signal intensity in the mid-insula and dorsomedial hypothalamus on the left side, and in dorsolateral prefrontal cortex, posterior cingulate cortex, precuneus, ventromedial hypothalamus and rostral ventrolateral medulla on both sides. Conversely, spontaneous bursts of SSNA were positively correlated with BOLD signal intensity in the ventromedial thalamus and posterior insula on the left side, and in the anterior insula, orbitofrontal cortex and frontal cortex on the right side, and in the mid-cingulate cortex and precuneus on both sides. Inverse relationships were observed between MSNA and BOLD signal intensity in the right ventral insula, nucleus tractus solitarius and caudal ventrolateral medulla, and between SSNA and signal intensity in the left orbitofrontal cortex. These results emphasize the contributions of cortical regions of the brain to sympathetic outflow in awake human subjects, and the extensive interactions between cortical and subcortical regions in the ongoing regulation of sympathetic nerve activity to muscle and skin in awake human subjects.

until recently, the majority of studies on functional imaging of human autonomic control have been based on correlating brain activity with physiological patterns of activity, measured either as end-organ responses recorded at the same time or by performing brain imaging during well-characterized physiological maneuvers known to evoke predictable autonomic responses. While such approaches continue to be very useful, the sluggish nature of the effector-organ responses means that temporal coupling to changes in brain activity can be limited, although this has certainly not limited the utility of correlating brain activity to the responses of certain effector organs. Changes in heart rate can be brought about very quickly, while changes in sweat release, and especially blood flow and blood pressure, are much slower, so identifying the central processes involved in the generation of these autonomic responses needs to consider neural and neuroeffector coupling delays. And while direct recordings of muscle sympathetic nerve activity (MSNA) or skin sympathetic nerve activity (SSNA) can be made via tungsten microelectrodes inserted into a peripheral nerve (microneurography), for technical reasons it has not, until now, been possible to record this neural activity at the same time as conducting functional magnetic resonance imaging (fMRI) of the brain. However, we recently overcame these technical limitations and succeeded in performing concurrent microneurography and fMRI, using the spontaneous bursts of sympathetic nerve activity, recorded from groups of postganglionic axons traveling in either muscle or cutaneous fascicles of the peroneal nerve, to functionally identify cortical and subcortical sites involved in the generation of the sympathetic outflow in awake human subjects. Importantly, we undertake this work without asking subjects to perform any specific maneuvers, using only the spontaneous bursts of MSNA or SSNA to identify areas in the brain that are temporally coupled to this activity. We examine covariations in the fluctuations in BOLD (blood oxygen level dependent) signal intensity within the brain and fluctuations in burst incidence and amplitude of the peripherally recorded nerve signal.

Methodological Approach

The key to the success of our approach is the use of a high-impedance, low-noise, preamplifier (headstage), with a gain of 100× and a bandpass of 100 Hz to 5 kHz (NeuroAmpEx, ADInstruments, Sydney, NSW, Australia). The headstage, which is electrically isolated and incorporates protection circuitry against high currents, connects to the NeuroAmpEX amplifier via a 6-m cable and was designed in consultation with ADInstruments. Because it is encased in military-grade stainless steel, it is nonmagnetizable in an MRI environment. The tungsten microelectrodes, copper coupling wires and gold pins are intrinsically magnetic resonance compatible. We perform sparse sampling, in which scans are delivered for 4 s, followed by a nonscanning period of 4 s; this cycle is repeated 200 times. The rationale here is that we are taking advantage of the neurovascular coupling delays inherent in BOLD imaging, and the delays associated with conduction of sympathetic traffic along slowly-conducting unmyelinated pathways (see below). We use this approach routinely at 3 Tesla (T) and have also commenced obtaining data at 7 T. There are no issues with heating of the microelectrodes in the MRI environment, even at 7 T.

Microneurography.

All studies described herein were performed at the Clinical Research Imaging Centre at Neuroscience Research Australia, using a 3-T MRI machine (Achieva, Phillips Medical Systems). All procedures were approved by the Human Research Ethics Committees of the University of New South Wales and the University of Western Sydney and conducted in accordance with the Declaration of Helsinki. All subjects provided informed, written consent prior to commencement of the experiment. Subjects lay supine on an MRI bed inside the laboratory and the optimal site for stimulating the right common peroneal nerve at the fibular head identified by electrical stimulation through a surface probe (3–10 mA, 0.2 ms, 1 Hz; Stimulus Isolator, ADInstruments). An insulated tungsten microelectrode (FHC) was inserted through the skin into nerve and manually guided into a muscle or cutaneous fascicle of the nerve using standard approaches employed in our laboratory (Macefield 2013). A nearby subdermal tungsten microelectrode, with 1-mm insulation removed, served as the reference electrode for the differential preamplifier; a surface Ag/AgCl electrode on the leg acted as the ground electrode.

Neural activity was amplified (gain 10⁴, bandpass 0.3–5.0 kHz) using the electrically isolated headstage referred to above (NeuroAmpEX, ADInstruments). The innervation territory of the fascicle was identified by percussion or stretch of the muscle belly or tendon, or by stroking the skin of the innervation territory, and the position of the microelectrode tip adjusted until spontaneous bursts of MSNA or SSNA were identified. Neural activity was acquired (10-kHz sampling) and a root mean square (RMS)-processed signal (200-ms moving average) generated on a computer-based data-acquisition and analysis system (LabChart 7, PowerLab 16S; ADInstruments). The subject was then wheeled to the MRI facility with the microelectrodes in situ. The subject's head was enclosed in a 32-channel SENSE head coil; earplugs were used to reduce noise and headphones were provided to allow communication with the researchers. Sympathetic burst amplitudes were manually measured from the RMS-processed nerve signal during the 4-s interscan period. This was divided into 1-s epochs, and the presence or absence of a burst was determined, and its amplitude calculated (Peak Parameters, LabChart 7, ADInstruments).

fMRI acquisition and analysis.

Two hundred BOLD contrast gradient echo, echo-planar images were collected consecutively, covering the entire brain. The data were obtained over a period of 27 min, with 46 axial slices being collected in a caudal to rostral direction in the first 4 s of the 8-s scanning sequence [repetition time (TR) = 8 s, echo time (TE) = 4 s, flip angle = 90°, raw voxel size = 1.5 × 1.5 × 2.75 mm]. In addition to these whole brain scans, we also performed high-resolution brain stem-only scans (46 axial slices, TR = 8 s, TE = 4 s, flip angle = 90°, raw voxel size = 1.5 × 1.5 × 1.75 mm). Finally, a three-dimensional T1-weighted anatomical image set of the whole brain was acquired (turbo field echo; TE = 2.5 ms, TR = 5,600 ms, flip angle = 8°, voxel size = 0.8 × 0.8 × 0.8 mm). The functional image sets for each individual subject were realigned and co-registered to their T1-weighted image set, and global signal intensity drifts were removed using a linear detrending method, implemented in SPM8 software (Friston et al. 1995). Manual correction of the images was performed to create an accurate match between the functional and anatomical image sets. The whole brain scans were then spatially normalized into Montreal Neurological Institute (MNI) space and resliced into 2 × 2 × 2-mm voxels. These normalized images were then smoothed with a 5-mm full-width-half-maximum Gaussian filter. The brain stem and cerebellum were isolated using the SUIT toolbox, and the images spatially normalized into MNI space using a spatially unbiased atlas template of the cerebellum and brain stem and resliced into 1.5 × 1.5 × 1.5-mm voxels (Diedrichsen 2006). To maintain fine spatial detail, we did not spatially smooth the brain stem-only images.

For each of the 200 image volumes, the 4 s during which BOLD signals were recorded were related to the sympathetic burst in the preceding 4-s period. This is possible because 1) neurovascular coupling delays inherent in BOLD-contrast imaging mean that changes in signal intensity follow the actual neuronal events in the brain by ∼5 s (Logothetis et al. 2001); and 2) slow conduction along unmyelinated postganglionic sympathetic axons means it takes ∼1 s for an individual sympathetic burst to travel from the brain to the peripheral recording site (Fagius and Wallin 1980). By taking these two factors into account, one can see that an increase in BOLD signal intensity should appear ∼4 s following the burst of sympathetic nerve activity recorded in the periphery, as indicated in Fig. 1. Moreover, since the scanning sequence was conducted in a caudal to rostral direction, from the upper cervical spinal cord to the vertex, we could identify specific regions in the brain on the basis of the temporal relationship between the scanning slice and scanned structure.

Fig. 1.

Screen shot of recording of spontaneous bursts of muscle sympathetic nerve activity (MSNA) acquired during BOLD (blood oxygen level dependent) contrast imaging of the brain. The raw nerve signal is shown in the top; the RMS-processed signal is shown in the bottom. Note the scanning artifacts, which limited measurement of MSNA from the RMS signal to the 4-s interscan periods. [Reprinted from Macefield (2013) with permission from Elsevier.]

In each subject, the whole brain or brain stem (brain stem-only scans) was divided into four segments from caudal to rostral, corresponding to the 1st-, 2nd-, 3rd- and 4th-s epochs. As noted above, the recording of sympathetic nerve activity was also divided into corresponding 1-s epochs, in which a “1” was entered into an fMRI search model if a sympathetic burst occurred, and a “0” was entered if no burst occurred. This was repeated for the entire 200-volume scanning period.

For both the whole brain and brain stem-only fMRI scans, changes in BOLD signal intensity that matched each subject's sympathetic burst model were then determined in each subject for each of the 1-s periods; the 6-directional movement parameters derived from the realignment step were added as nuisance variables. For both analyses, we also included signal changes derived from a 2-mm sphere placed in the center of the fourth ventricle, as a nuisance variable, to eliminate the effects of the pulsatile waveform, resulting from the heart on BOLD signal. For both the whole brain and brain stem analyses, a mean spatially normalized T1-weighted anatomical image was calculated from all subjects, and this whole brain or brain stem only image was masked so that it contained only the region collected during either the 1st-, 2nd-, 3rd- or 4th-s epochs. These masked images were then used to restrict the BOLD analysis to the relevant brain regions for each epoch.

For the whole brain analyses, signal intensity changes that positively and negatively correlated with spontaneous MSNA bursts were determined using a random effects procedure in 15 healthy control subjects (P < 0.001 uncorrected for multiple comparisons). Since we used an uncorrected threshold, we minimized the chances of a type 1 error by using a minimum cluster threshold of 10 voxels and also extracted signal intensity changes to verify the significance of any signal intensity changes. For the brain stem only analysis, since only 10 healthy control subjects were explored, signal intensity changes that positively and negatively correlated with spontaneous MSNA bursts were determined using a fixed effects procedure (P < 0.05, false discovery rate corrected for multiple comparisons).

In addition to increases and decreases in signal intensity coupled to MSNA, functional connectivity analyses were performed using data from all 1-s epochs collected during the 4-s periods for all 200 brain volumes. This produces a spatial map of the correlation coefficients between each voxel's time series with that of a region-of-interest, a “seed” (Rogers et al. 2007). Using the significant clusters derived from the whole brain MSNA-coupled analyses, the raw signal intensity within a particular “seed” cluster was extracted for the entire 200 volume scan in each subject. Changes in BOLD signal intensity correlated to this input signal were determined for each subject, and a group analysis was then performed (random effects, minimum cluster size 10 voxels). A correlation value greater than r = >0.7 was set as the minimum threshold.

Functional Identification of Cortical and Subcortical Areas Contributing to Sympathetic Outflow to Muscle

Figure 2 shows the cortical and subcortical areas that exhibited significant changes in BOLD signal intensity that were temporally coupled to the corresponding burst of MSNA, represented as mean data obtained from 14 healthy, young adult subjects (35 ± 3 yr). Because the scans were conducted in the caudal to rostral direction over 4 s, and the scan volumes were divided into 1-s epochs, we could correlate BOLD signal intensity with the MSNA burst amplitude during the corresponding 1-s period recorded 4 s previously (Fig. 1). The colors shown in Fig. 2 reflect the corresponding epochs of the recording of sympathetic bursts shown in Fig. 1.

Fig. 2.

Group data from 14 subjects. Signal intensity (SI) increases and decreases correlated with muscle sympathetic nerve activity (MSNA) during the 2nd- (red), 3rd- (light blue), and 4th-s (light green) of the 4-s image collection period. The hot color scale indicates regions in which SI was high during periods of high MSNA, and low during low MSNA. Conversely, the cool color scale indicates regions where SI was high during low MSNA, and low during high MSNA. In the coronal section, significant clusters were assessed using MSNA during the 2nd-s period, and hence the rest of the brain is covered by a dark bar. A similar display is shown for the sagittal sections in which the 3rd- and 4th-s periods are shown. Slice locations in Montreal Neurological Space are shown on the bottom left of each section and are color coded for collection period. OFC, orbitofrontal cortex; dmHypo, dorsomedial hypothalamus; vmHypo, ventromedial hypothalamus; DlPFC, dorsolateral prefrontal cortex. [Reprinted from James et al. (2013a) with permission from Elsevier.]

It can be seen that BOLD signal intensity was positively coupled with MSNA in several discrete, yet widespread, areas: the left insula, and bilateral dorsolateral prefrontal cortex, posterior cingulate cortex and precuneus all showed an increase in BOLD signal intensity when MSNA was high and reduced signal intensity when MSNA was low (James et al. 2013a). Interestingly, an inverse relationship was found between MSNA and BOLD signal intensity in the right ventral insula. We were perhaps not surprised that activity in the insula covaried with MSNA, given that it is one of the usual suspects in cardiovascular control: human fMRI studies have shown that it is activated during physiological maneuvers that increase MSNA, such as the Valsalva maneuver (Henderson et al. 2002), inspiratory-capacity apnea (Macefield et al. 2006), end-expiratory apnea (Kimmerly et al. 2013), Mueller maneuver (Kimmerly et al. 2013), cold-pressor test (Harper et al. 2003), hand-grip exercise (Sander et al. 2010; Wong et al. 2007) and lower body negative pressure (Kimmerly et al. 2005). The insula does not project directly to the rostal ventrolateral medulla (RVLM), the primary output nucleus for MSNA (Dampney 1994; Macefield and Henderson 2010), but may act via the hypothalamus (Cechetto and Chen 1990). This is interesting, because we did indeed see a positive relationship between MSNA and BOLD signal intensity within the hypothalamus; curiously, while this covariation occurred bilaterally in the ventromedial hypothalamus (VMH), it occurred in the dorsomedial hypothalamus (DMH) only on the left side (James et al. 2013a). We were somewhat surprised that these two hypothalamic nuclei were involved, given what we know about their function in experimental animals. For instance, we know that DMH forms part of the classical hypothalamic defense area (Coote et al. 1979) and plays a role in activation of the hypothalamic-pituitary-adrenal pathway in response to external stressors (DiMicco et al. 2002); the cardiovascular responses to stress are mediated via direct and indirect connections to the RVLM (DiMicco et al. 2002). However, it would appear from the current data that the DMH also contributes to MSNA at rest, in the absence of any overt stress: no subject reported feeling stressed in the scanner environment. DMH and VMH have also been implicated in generating the elevated muscle vasoconstrictor drive in obesity: direct application of leptin, a circulating hormone produced by adipose tissue, into the VMH or DMH evokes significant increases in arterial pressure, heart rate and sympathetic nerve activity (Marsh et al. 2003; Montanaro et al. 2005). Interestingly, VMH, unlike DMH, does not project directly to RVLM, but does project to many forebrain and brain stem sites known to regulate MSNA, such as the nucleus of the solitary tract (NTS) and parabrachial nucleus, the midbrain periaqueductal gray and DMH (Canteras et al. 1994; Jansen et al. 1995; Schramm et al. 1993; Ter Horst and Luiten 1987).

Figure 3 shows the results of a functional connectivity analysis with a “seed” placed into right VMH. We used this nucleus for the seed because it had a larger cluster size on the right than on the left (62 vs. 27). It can be seen that the VMH is functionally coupled to the dorsolateral prefrontal cortex and insula, and to the RVLM, suggesting that these supraspinal sites could contribute to the generation of MSNA at rest. VMH was also functionally coupled to the precuneus, part of the default mode network (Fransson and Marrelec 2008) that is active at rest and is thought to reflect “self-reflection” and “self-consciousness” (Cavanna and Trimble 2006). Given the robust coupling between precuneus and MSNA, and the fact that signal intensity within precuneus declines during deep sleep (Maquet 2000), as does MSNA and blood pressure (Hornyak et al. 1991), it is possible that the precuneus provides a top-down drive to MSNA, in other words, a “wakefulness drive” to resting MSNA. Indeed, this is supported by the fact that functional connectivity analysis showed that precuneus, dorsolateral prefrontal cortex, insula and VMH were all functionally coupled to RVLM.

Fig. 3.

Top: significant signal intensity (SI) changes within the ventromedial hypothalamus (VMH). The hot color scale indicates regions in which SI was high during periods of high muscle sympathetic nerve activity (MSNA), and low during low MSNA. Bottom: brain regions in which SI covaries significantly with SI within the VMH. Note that activity within the VMH covaries with activity in the anterior insula, dorsolateral prefrontal cortex (dlPFC), precuneus and also in the region encompassing the rostral ventrolateral medulla (RVLM). Slice locations in Montreal Neurological Space are shown on the bottom left of each section. ACC, anterior cingulate cortex. [Modified from James et al. (2013a) with permission from Elsevier; figure of angiotensin 2 imaging reproduced from Allen et al. (1998) with permission from Elsevier.]

In addition to this coupling, we found a strong positive relationship between ongoing fluctuations in MSNA and BOLD signal intensity within the deep cerebellar nuclei, including the region of the fastigial nucleus. In the cat and rabbit, electrical stimulation of the fastigial nucleus causes increases in blood pressure (Bradley et al. 1987; Miura and Reis 1969), while the baroreceptors can inhibit its activity (Lutherer et al. 1989; Rector et al. 2006). Moreover, fMRI studies in humans have shown that maneuvers that increase blood pressure lead to an increase in BOLD signal intensity within the cerebellar dentate nucleus (Harper et al. 2003), again emphasizing the contributions of the cerebellum to cardiovascular control.

We have also performed high-resolution imaging of the brain stem. Significant correlations were found to occur within specific nuclei within the medulla: RVLM, caudal ventrolateral medulla (CVLM) and NTS, with increases in MSNA being associated with increases in BOLD signal intensity in RVLM and decreases in signal intensity in CVLM and NTS (Macefield and Henderson 2010). This is shown in Fig. 4. Given that increases in MSNA occur during spontaneous falls in arterial pressure, when arterial baroreceptor input to NTS decreases, our results essentially show the operation of the arterial baroreflex. For a reduction in baroreceptor input to NTS, we would expect BOLD signal intensity within the NTS to decrease, and, given that NTS sends excitatory projections to CVLM, we would expect that it too would show a decrease. This in turn would lead to an increase in signal intensity in RVLM, and hence an increase in MSNA, because the projection from CVLM to RVLM is inhibitory. This matches the serial pathway established in experimental animals: NTS-CVLM-RVLM (Dampney 1994). Of course this interpretation, along with most BOLD signal interpretations, needs to be considered with some caution given that BOLD signal is dependent on coupling between blood flow, blood volume and oxygen metabolism and, more importantly, evidence that the spatial accuracy of BOLD imaging is in the order of 4–5 mm (Uğurbil et al. 2003; Zarahn 2001). Combined with the potential effects of large drainage vessels on BOLD signal (Kennerley et al. 2010), the spatial accuracy, particularly within small structures such as brain stem nuclei, needs to be viewed with some caution.

Fig. 4.

Signal intensity (SI) changes correlated to spontaneous fluctuations in MSNA in 8 subjects. Increases in SI with increases in MSNA are coded by the hot color scale, and signal decreases with the cool color scale and are overlaid onto a series of sagittal and axial fMRI slices from an individual subject. Myelin-stained sections through the medulla are shown to the right. CVLM, caudal ventrolateral medulla; NTS, nucleus tractus solitaries; RVLM, rostral ventrolateral medulla. [Modified from Macefield and Henderson (2010) with permission.]

Functional Identification of Cortical and Subcortical Areas Contributing to Sympathetic Outflow to Skin

We have also used this same approach to examine areas in the brain that covary with spontaneous bursts of SSNA in 13 healthy, young adult subjects (24 ± 2 yr). This was somewhat more difficult than recording MSNA, given that the level of SSNA depends on state of arousal of the subject. Indeed, fMRI data from quite a few experiments had to be excluded because the overall level of SSNA showed a progressive decline during the course of the scanning, as subjects relaxed in the scanner environment. Nevertheless, complete data sets were obtained from 13 subjects, with mean data shown in Fig. 5. As with the concurrent recordings of MSNA shown in Fig. 2, the observed changes in BOLD signal intensity reflect covariation in the fluctuations in central neuronal activity with the peripherally recorded sympathetic neural activity. It can be seen that significant increases in SSNA-coupled BOLD signal intensity occurred in several areas, with evidence of lateralization: right orbitofrontal cortex, left ventromedial thalamus, right frontal cortex, right anterior and left posterior insula, while signal intensity in the left orbitofrontal cortex was inversely related to SSNA (James et al. 2013b). Increases in signal intensity also occurred bilaterally in the mid-cingulate cortex and precuneus. There were no significant changes in signal intensity within the hippocampus, amygdala, hypothalamus or brain stem.

Fig. 5.

Group data from 13 subjects. Signal intensity (SI) increases and decreases correlated with skin sympathetic nerve activity (SSNA) during the 2nd- (red), 3rd- (light blue), and 4th-s (light green) of the 4-s image collection period. Note that increases in SSNA are associated with SI increases in the thalamus, left posterior and right anterior insula, right orbitofrontal cortex (OFC), right frontal cortex, precuneus and mid-cingulate cortex. SI decreases during SSNA increases occurred in the right OFC and left operculum. Data were obtained from epochs 2–4 s over all 200 scan volumes. Slice location in Montreal Neurological Institute space are indicated at the bottom left of each image. These values are color coded to indicate which portion of the 4-s period corresponds to that particular slice (orange: second 2–3, blue: second 3–4; green: second 3–4). [Reproduced from James et al. (2013b) with permission from Elsevier.]

Figure 6 shows the results of a functional connectivity analysis, based on seeding areas in two discrete regions. Seeding the right orbitofrontal cortex revealed a strong positive coupling to the right anterior insula. Seeding the precuneus (bilateral cluster) demonstrated functional coupling to the mid-cingulate and posterior cingulate cortices, the cerebellar cortex and the left thalamus, and uncovered an inverse relationship to the anterior and posterior left insula and the vermis of the cerebellum.

Fig. 6.

Functional connectivity of two significant clusters, the right orbitofrontal cortex (OFC) and bilateral precuneus, identified in the whole brain voxel-by-voxel analysis. Note that signal intensity (SI) in the right OFC covaries positively (hot color scale) with the right anterior insular cortex. Conversely, SI within the precuneus covaries positively with SI in the thalamus, cerebellar, and mid- and posterior cingulate cortices. Furthermore, it covaries in a negative fashion (cool color scale) with SI in the cerebellar vermis and the left posterior and anterior insula. Slice locations in Montreal Neurological Institute space are indicated at the top right of each image. [Reproduced from James et al. (2013b) with permission from Elsevier.]

These experiments were performed in an essentially thermoneutral environment, in which subjects reported being at a comfortable temperature, such that the spontaneous bursts of SSNA cannot be seen to be driving thermoregulatory changes in skin blood flow or sweat release. Indeed, if the spontaneous SSNA was related to thermoregulation, we would expect it to continue even as a subject relaxed (and certainly not to disappear), and we would also expect to see activity within the anterior hypothalamus related to thermoregulatory drive, but, as noted above, we saw no SSNA-coupled activity in the hypothalamus. Indeed, Farrell and colleagues (2015) recently showed that psychogenic sweating was not related to any increase in BOLD signal intensity in the anterior hypothalamus, whereas activity in this nucleus did increase during thermal sweating (Farrell et al. 2014, 2015). Accordingly, we conclude that the spontaneous bursts of SSNA were related to internal thought processes that were not influenced by any external stimuli (they would have rapidly habituated to the scanner noise).

Common Regions Contributing to Sympathetic Drive to Muscle and Skin

Whilst we did not compare directly differences in signal intensity changes coupled to MSNA vs. SSNA in each individual subject, our separate group analyses do suggest that there are different brain activation patterns associated with spontaneous MSNA compared with SSNA bursts. Whereas in certain cortical areas BOLD signal intensity covaried with sympathetic outflow to both muscle and skin, such as the insula and precuneus, there were various regions that appear unique to MSNA or SSNA. For example, unlike MSNA, which was coupled to the dorsolateral prefrontal cortex bilaterally, SSNA showed no such relationship. Unlike MSNA, SSNA showed no such relationship. Rather, SSNA covaried with BOLD signal intensity in the right orbitofrontal cortex, which also covaried with activity within the right anterior insula. Conversely, MSNA was inversely related to the right orbitofrontal cortex, and inversely related to the right ventral insula. It has been suggested that the right orbitofrontal cortex is responsible for the sweating responses to emotional stimuli (Critchley et al. 2000), with damage to the orbitofrontal cortex resulting in blunted responses (Tranel and Damasio 1994). It is also well known that the right insular cortex is activated in several emotional states (Damasio 1996; Singer et al. 2009): internal awareness and subjective feelings have been shown to be correlated with the degree of activation and the size of the right anterior insula, while interoceptive awareness is correlated with the size of the right orbitofrontal cortex (Critchley et al. 2004). As with MSNA, we found strong covariation between sympathetic outflow to the skin and BOLD signal intensity in the precuneus, although the specific region of precuneus was more posterior to the area coupled to MSNA. And, as with MSNA, that SSNA was strongly coupled to activity in the precuneus suggests that the precuneus may contribute to a “wakefulness drive” to sympathetic outflow to both muscle and skin.

Conclusions

We have shown that concurrent recordings of sympathetic nerve activity and fMRI can be used to identify areas of the brain involved in the generation of sympathetic outflow to muscle and skin. This can be achieved without using any particular maneuver, as the bursts of MSNA or SSNA serve as the input model in the analysis of BOLD signal intensity, allowing us to identify areas of the brain in which spontaneous fluctuations in central neuronal activity (BOLD signal intensity) matched the spontaneous fluctuations in directly recorded sympathetic nerve activity. We should also point out that these results did not involve a “region-of-interest” analysis, but rather an analysis of areas showing temporal coupling to the spontaneous bursts of sympathetic nerve activity. It is for this reason that we refer to our approach as providing “real-time imaging,” although it is only through the requisite and extensive off-line analyses do we find areas that “pop out” as being temporally coupled to the bursts of sympathetic outflow to muscle or skin.

Future Directions

The approach we have developed opens up opportunities to examine disease states associated with disturbances in sympathetic control. Indeed, we have recently completed a series of experiments in which we performed concurrent fMRI and microneurography in patients with obstructive sleep apnea (OSA), in whom MSNA and blood pressure are greatly elevated, and a group of older, age-matched normotensive controls: in both groups there was a positive correlation between MSNA and BOLD signal intensity in the insula, hypothalamus, precuneus, cingulate and prefrontal cortices, but in the OSA patients signal intensity was higher in the dorsolateral and medial prefrontal cortex, anterior cingulate cortex, dorsal precuneus and the left thalamus (Fatouleh et al. 2014). Curiously, signal intensity was actually lower in the midbrain, dorsolateral pons, medullary raphe and RVLM (Lundblad et al. 2014). The latter observations were rather counterintuitive, given that one would expect an increase in MSNA to be associated with an increase in signal intensity within RVLM. However, given that BOLD signal intensity is considered to reflect synaptic activity (Logothetis et al. 2001), the decrease in signal intensity within RVLM in patients with OSA may suggest that active inhibition of RVLM is lower in OSA. We also showed that these cortical and subcortical changes were largely reversed (Fatouleh et al. 2015; Lundblad et al. 2015) following successful treatment with continuous positive airway pressure (CPAP), so the increase in MSNA-coupled signal intensity within RVLM (and other brain stem sites) following CPAP must reflect an increase in active inhibition of the RVLM, bringing its total activity and hence MSNA down toward control levels (Lundblad et al. 2014, 2015). We are currently examining functional changes in the brain in patients with congestive heart failure, as well as undertaking studies in healthy subjects at high field strength (7 T).

GRANTS

This work was supported by the National Health and Medical Research Council of Australia (Project Grant 1007557).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

V.G.M. and L.A.H. conception and design of research; V.G.M. performed experiments; V.G.M. and L.A.H. interpreted results of experiments; V.G.M. and L.A.H. prepared figures; V.G.M. drafted manuscript; V.G.M. and L.A.H. edited and revised manuscript; V.G.M. and L.A.H. approved final version of manuscript; L.A.H. analyzed data.