Phrenic nerve afferent activation of neurons in the cat SI cerebral cortex

Abstract

Stimulation of respiratory afferents elicits neural activity in the somatosensory region of the cerebral cortex in humans and animals. Respiratory afferents have been stimulated with mechanical loads applied to breathing and electrical stimulation of respiratory nerves and muscles. It was hypothesized that stimulation of the phrenic nerve myelinated afferents will activate neurons in the 3a and 3b region of the somatosensory cortex. This was investigated in cats with electrical stimulation of the intrathoracic phrenic nerve and C₅ root of the phrenic nerve. The somatosensory cortical response to phrenic afferent stimulation was recorded from the cortical surface, contralateral to the phrenic nerve, ispilateral to the phrenic nerve and with microelectrodes inserted into the cortical site of the surface dipole. Short-latency, primary cortical evoked potentials (1° CEP) were recorded with stimulation of myelinated afferents of the intrathoracic phrenic nerve in the contralateral post-cruciate gyrus of all animals (n= 42). The mean onset and peak latencies were 8.5 ± 5.7 ms and 21.8 ± 9.8 ms, respectively. The rostro-caudal surface location of the 1° CEP was found between the rostral edge of the post-cruciate dimple (PCD) and the rostral edge of the ansate sulcus, medio-lateral location was between 2 mm lateral to the sagittal sulcus and the lateral end of the cruciate sulcus. Histological examination revealed that the 1° CEP sites were recorded over areas 3a and 3b of the SI somatosensory cortex. Intracortical activation of 16 neurons with two patterns of neural activity was recorded: (1) short-latency, short-duration activation of neurons and (2) long-latency, long-duration activation of neurons. Short-latency neurons had a mean onset latency of 10.4 ± 3.1 ms and mean burst duration of 10.1 ± 3.2 ms. The short-latency units were recorded at an average depth of 1.7 ± 0.5 mm below the cortical surface. The long-latency neurons had a mean onset latency of 36.0 ± 4.2 ms and mean burst duration of 32.2 ± 8.4 ms. The long-latency units were recorded at an average depth of 2.4 ± 0.2 mm below the cortical surface. The results of the study demonstrated that phrenic nerve afferents have a short-latency central projection to the SI somatosensory cortex. The phrenic afferents activated neurons in lamina III and IV of areas 3a and 3b. The cortical representation of phrenic nerve afferents is medial to the forelimb, lateral to the hindlimb, similar to thoracic loci, hence the phrenic afferent SI site in the cat homunculus is consistent with body position (thoracic region) rather than spinal segment (C₅–C₇). The phrenic afferent activation of the somatosensory cortex is bilateral, with the ipsilateral cortical activation occurring subsequent to the contralateral. These results support the hypothesis that phrenic afferents provide somatosensory information to the cerebral cortex which can be used for diaphragmatic proprioception and somatosensation.

Introduction

Stimulation of respiratory afferents elicits neural activity in the somatosensory region of the cerebral cortex in humans and animals (Frankstein et al. 1979; Davenport & Reep, 1985; Davenport et al. 1986, 1993). Respiratory afferents have been stimulated with mechanical loads applied to breathing (Davenport & Reep, 1985; Davenport et al. 1993; Davenport & Hutchison, 2002) and electrical stimulation of respiratory nerves (Aubert & Legros, 1970; Aubert & Guilhen, 1971; Davenport & Reep, 1985; Davenport et al. 1993; Yates et al. 1994) and muscles (Gandevia & Macefield, 1989). Somatosensory cortical neural activation was shown by recording cortical surface or scalp surface current dipoles generated by respiratory afferent activation of a population of neurons. The latency for phrenic afferent cortical activation is consistent with the response elicited by large myelinated group I and II phrenic afferents (Frankstein et al. 1979; Davenport et al. 1985; Yates et al. 1994). While diaphragmatic muscle spindles are sparse in cats (Duron et al. 1978; Jammes et al. 1986; Jammes & Balzamo, 1992; Balkowiec et al. 1995), muscle spindle, Golgi tendon organ (Corda et al. 1965; Duron et al. 1978; Jammes et al. 1986; Balkowiec et al. 1995) and diaphragmatic pressure mechanoreceptors (Holt et al. 1991) with group I and II conduction velocities are present and active during spontaneous ventilation (Corda et al. 1965; Jammes et al. 1986; Balkowiec et al. 1995). Group III and IV phrenic afferents can modulate phrenic motor output and may mediate EEG activity changes related to diaphragmatic fatigue (Jammes et al. 1986; Speck & Revelette, 1987a,b;). There are, however, no recordings of cortical neurons activated by group I and II respiratory afferent stimulation. The purpose of the present study was to define the region of the somatosensory cortex that is activated by large myelinated diaphragmatic afferents using surface dipole recordings and then to explore the activated region with intracortical neuronal recordings. In this manner we hoped to identify somatosensory cortical neurons activated by diaphragmatic afferents.

The somatosensory cortex is divided into areas based on the histological cytoarchitecture and type of afferents that activate the cortical neurons. In cats, area 3a is caudal to the primary motor region 4γ (Hassler & Muhs-Clement, 1964). Area 3a is distinguished from 4γ by a decrease in the density of lamina V pyramidal cells and an increase in laminas III and IV (Hassler & Muhs-Clement, 1964). This cytoarchitecture is consistent with a decrease in lamina V motor output and an increase in lamina II and IV sensory function (Hassler & Muhs-Clement, 1964). Muscle proprioceptors are known to activate neurons in lamina III and IV in area 3a (Oscarsson & Rosen, 1963, 1966; Oscarsson et al. 1966). Area 3b is immediately caudal to area 3a and is reported to have a tactile information function (Rasmusson et al. 1979; Dykes et al. 1980). Area 3b is distinguished from 3a by the absence of pyramidal cells in lamina V and wide lamina III and IV (Hassler & Muhs-Clement, 1964). Cutaneous afferents are known to activate neurons in area 3b (Friedman & Jones, 1981; Jones & Friedman, 1982). The topographic representations mapped out in areas 3a and 3b revealed that head and upper limb afferents activate neurons in the most lateral extent of these areas, whereas afferents from the hind limb activate neurons in the medial region, near the sagittal sulcus (Felleman et al. 1983; Ito & Craig, 2003). Afferents activating areas 3a and 3b ascend through the dorsal column–medial lemniscal pathway to the ventro-posterior lateral (VPL) region of the thalamus (Andersson et al. 1966; Landgren et al. 1967). This area of the thalamus then projects to neurons in lamina III and IV of areas 3a and 3b (Landgren et al. 1967; Jones & Porter, 1980). Phrenic afferents have been reported to activate neurons in the dorsal columns (Chou & Davenport, 2005), the VPL (Zhang & Davenport, 2003) and elicit a cortical surface recorded dipole at the border of areas 3a and 3b (Frankstein et al. 1979; Davenport & Reep, 1985; Yates et al. 1994). It is unknown, however, if phrenic nerve afferents activate neurons in the 3a/3b region and which lamina contain activated neurons.

The processing of respiratory muscle pump information by the central nervous system requires the transduction of muscle mechanics by mechanoreceptors in the diaphragm and intercostal muscles (Davenport & Vovk, 2009). These afferents must, in turn, provide a definable neural code and project to specific regions of the central nervous system. Diaphragmatic afferents enter the central nervous system via the cervical dorsal roots, primarily C₅–C₆ in the cat (Chou & Davenport, 2005). Stimulation of diaphragmatic afferents and mechanical stimulation of the diaphragm has been reported to elicit neural activation of thalamic neurons (Zhang & Davenport, 2003). Thus, there appears to be a neural substrate for a phrenic afferent pathway to the somatosensory cortex via the dorsal columns and thalamus. There are very few studies, however, of the higher brain centre projections of diaphragmatic afferents, yet diaphragmatic afferents can activate this region of the cerebral cortex (Frankstein et al. 1979; Davenport & Reep, 1985; Davenport et al. 1985; Yates et al. 1994). If respiratory muscle afferents function similar to limb muscles, then it is likely that phrenic afferent activation of the SI cortex may have a SI-to-MI pathway (Lipski et al. 1986; Orem & Netick, 1986). Thus, we hypothesize that diaphragmatic afferents project to neurons in area 3a/3b of the somatosensory region of the cerebral cortex.

The diaphragm has been reported to contain muscle spindles, tendon organs, A-δ and non-myelinated afferents (Corda et al. 1965; Duron et al. 1978; Revelette et al. 1988; Bolser et al. 1991; Holt et al. 1991; Balkowiec et al. 1995). Anatomical studies have shown that phrenic afferents enter the spinal cord by the cervical dorsal roots (Corda et al. 1965; Malakhova & Davenport, 2001; Chou & Davenport, 2005) and project to the external cuneate nucleus (Marlot et al. 1985). Electrical stimulation of the whole phrenic nerve (Frankstein et al. 1979; Davenport et al. 1985; Yates et al. 1994) elicited neural activity on the surface of the sensorimotor region of the cerebral cortex of cats. The stimulus parameters used in these studies activated primarily myelinated afferents (Frankstein et al. 1979). The primary evoked potentials were reported to be in somatosensory cortex areas 3a and 3b (Frankstein et al. 1979; Davenport et al. 1985). It has been further reported that mechanical loads applied to lambs respiring through an endotracheal tube elicited a somatosensory cortical surface dipole (Davenport & Hutchison, 2002). Somatosensory cortical activation has also been reported in humans respiring against mechanical loads (Davenport et al. 1996; Logie et al. 1998). Hence, it is evident that respiratory muscle afferents can elicit short-latency somatosensory cortical neural dipoles in both experimental animals and humans (Frankstein et al. 1979; Davenport & Reep, 1985; Davenport et al. 1985, 1993; Gandevia & Macefield, 1989; Zifko et al. 1995, 1996). However, somatosensory cortical neuronal activity has not been recorded. It was hypothesized that stimulation of the phrenic nerve myelinated afferents will activate neurons in the 3a and 3b region of the somatosensory cortex. This was investigated in cats with intrathoracic phrenic nerve and C₅ root of the phrenic nerve electrical stimulation of large myelinated afferents. The somatosensory cortical response to phrenic afferent stimulation was recorded from the cortical surface, contralateral to the phrenic nerve, ispilateral to the phrenic nerve. In addition, microelectrodes inserted into the cortical site of the surface dipole were used to identify cortical neurons that discharged in response to phrenic nerve stimulation.

Methods

Surgical preparation

Adult cats of either sex (mean weight, 3.15 ± 0.78 kg) were studied. Anaesthesia was induced by inhalation of 4% halothane–40% nitrous oxide–balance oxygen. The femoral artery, femoral vein and trachea were cannulated. The cats were vagotomized bilaterally. Gas anaesthesia was then replaced with α-chloralose (60 mg kg⁻¹)–urethane (400 mg kg⁻¹) by slow i.v. infusion. A deep surgical plane of anaesthesia was maintained throughout each experiment. The animals were placed prone with the head and spine fixed into a stereotaxic apparatus. The body temperature was monitored with a rectal probe and maintained at 38 ± 1°C with the periodic use of a heating pad. Arterial blood pressure, expired CO₂ and tracheal pressure were continuously monitored on a polygraph. The animals were connected to a mechanical ventilator and neuromuscular blockade was performed with gallamine triethiodide (6 mg kg⁻¹, i.v.). The lungs were periodically inflated to prevent atelectasis. Arterial blood gases and pH were periodically measured and maintained within the normal range. The animals were kept on a continuous infusion of lactated Ringer solution. If fluctuation in blood pressure and heart rate were observed, all experimental procedures were suspended and supplemental anaesthesia was administered until a surgical plane of anaesthesia was re-established. All methods and procedures for these studies were reviewed and approved by the University of Florida Institutional Animal Care and Use Committee.

Cortical evoked potential (CEP) recording

The sensorimotor areas of the cerebral cortex (pericruciate region) were exposed by craniotomy. The cruciate sulcal lengths and distances between cruciate and ansate sulci were measured and recorded (Fig. 1) to generate a 2 mm × 2 mm grid between these sulci (Fig. 1). A silver–silver chloride ball electrode was placed on the pial surface of the exposed cortex to record cortical activity (ECoG) initiated by stimulation of the phrenic nerve (see below). An epoch of at least 5 ms pre-stimulus and a minimum of 50 ms post-stimulus ECoG for each stimulus pulse was saved to computer. A minimum of 128 epochs were recorded at each cortical site. Upon completion of the ECoG recording, a tungsten electrode was inserted into the cortex at the last recording site and an electrolytic lesion was made at 3 cm below the cortical surface. The animals were then perfused with heparinized saline followed by 4% formalin. The brain was removed and fixed in buffered 4% formalin. The fixed tissue was then cut saggitally into 50 μm thick sections with a freezing microtome. The sections were mounted and stained with haematoxylin and eosin. The stained sections were examined to identify the lesion. The cytoarchitechtonic cortical region for each nerve primary site was identified.

Figure 1. Cat cortical loci activated by phrenic nerve afferents.

The left panel is a photograph of the rostral cat cerebral cortex. The PCD is the location of the post-cruciate dimple. The right panel is a schematic representation of the sensorimotor, pericruciate region of the cat cerebral cortex. Each filled circle represents to cortical surface location of the 1° CEP for an individual animal. The shaded oval is the cortical area that contains more than 75% of the 1° CEP loci.

For surface averaged cortical evoked potential (CEP) analysis, the individual ECoG epochs were recalled for the individual site, each epoch was inspected for artifact and a minimum of 64 artifact-free epochs were computer averaged. The averaged CEP was stored for analysis. Each averaged CEP (Fig. 2) was analysed to determine onset latency, the peak polarity (positive or negative) and peak amplitude. The peak onset latency was defined as the time from the stimulus artifact to the time when the voltage was greater or less than zero (Fig. 2). The peak latency was defined as the time from the stimulus artifact to the individual CEP peak. The zero peak voltage was measured for each individual peak and was defined as the peak amplitude (Fig. 2). The cortical grid point with the shortest onset latency for the greatest positive peak amplitude was defined as the primary (1° CEP) peak site for each individual animal. The cat sensorimotor grid was drawn from photographic images of the exposed cat cortex (Fig. 1). The grid recording locations for the individual experimental animal were produced by calibrating the standard grid with the individual animal measured ansate sulcus, cruciate sulcus and ansate–cruciate lengths. The coordinates of the animal's 1° CEP site was entered on the standardized grid for the cat sensorimotor region (Fig. 1).

Figure 2. Cerebral cortical surface 1° CEP elicited by electrical stimulation of the whole phrenic nerve intrathoracically for an individual animal.

The onset and peak latencies are indicated.

Whole phrenic nerve stimulation

The whole phrenic was isolated intrathoracically in 42 cats. The skin and muscles overlying the ribs were incised on the left side. One intercostal space, usually the 6th or 7th, was incised and widely separated to provide access to the left phrenic caudal to the heart and cranial to its entry into the diaphragm. Approximately 1–2 cm of the left phrenic nerve was gently separated from the caudal vena cave to the diaphragm, keeping the phrenic nerve blood supply intact. The nerve was placed over bipolar platinum stimulating electrodes. The electrodes were embedded in a custom designed plexiglass block to isolate them from the surrounding tissue. The nerves were covered with petroleum jelly to prevent drying of the nerve. The whole left phrenic nerve was stimulated electrically with single pulses: 0.1 ms duration, 0.3 Hz, 5–100 μA. The somatosensory cortical evoked potential response was also elicited in the same animal by electrical stimulation of the: (1) forelimb cutaneous superficial radial nerve, (2) forelimb muscle deep radial nerve, (3) hindlimb cutaneous sural nerve, (4) hindlimb muscle medial gastrocnemius nerve, (5) thoracic intercostal nerve and (6) visceral major splanchnic nerve using the same stimulus parameters as for the phrenic nerve. The stimulator triggered the signal averager and 128 consecutive samples were collected.

Data analysis

Surface evoked potentials elicited by stimulation of each of the nerves were collected at each grid position, to map the site of the 1° CEP for each nerve, respectively. The 1° CEP was determined in each animal for each stimulated nerve from the averaged CEP. The 1° CEPs were defined as the cortical location with the averaged CEP having the shortest latency, largest amplitude and positive peak (Frankstein et al. 1979; Davenport et al. 1985). The individual ECoG epochs were recalled from memory. Any epoch with an artifact was rejected from the average. An artifact was defined as an epoch that did not begin (pre-stimulus pulse) from 0 V baseline, contained a waveform greater than ± 50 mV or did not have a stimulus pulse. A minimum of 64 ECoG epochs were averaged to obtain the averaged CEP for each recording site. The averaged CEP was saved in a separate file. Each averaged CEP was analysed for peak polarity, zero peak voltage, onset latency and peak latency. The cortical surface grid site was determined for the 1° CEP for each nerve. The 1° CEP for each nerve for each animal was entered on the standard brain diagram (Fig. 1). The 1° CEP peak amplitude and latencies were averaged for all animals.

Cervical phrenic root stimulation

The C₅ portion of the phrenic nerves was isolated bilaterally in the neck (n= 9). A bipolar stimulating cuff electrode was placed around one C₅ phrenic rootlet and insulated from the surrounding tissue. The intrathoracic phrenic nerve was placed across bipolar platinum recording electrodes. The stimulator triggered the signal averager and 128 consecutive samples were recorded. When the surface CEPs were collected at each grid position, the site of the C₅ phrenic afferent 1° CEP was determined.

The conduction velocity of the stimulated fibres in the whole phrenic nerve was determined in these animals with the C₅ portion of the phrenic nerve exposed. The whole phrenic nerve was isolated and electrically stimulated intrathoracically as described above. The compound action potential recorded in the C₅ phrenic nerve in the neck. The nerve length from the stimulating electrode to the C₅ recording site was measured during post-mortem analysis of the animal.

In three animals, the somatosensory cortex was mapped with intrathoracic whole phrenic nerve stimulation, then with stimulation of the C₅ portion of the phrenic nerve. The C₅ portion of the phrenic nerve was then severed in these animals and the whole phrenic nerve again stimulated.

Data analysis

The averaged CEPs were obtained for each recording site as described above. The 1° CEP site for the C₅ portion of the phrenic nerve was identified and entered on the standard brain grid. In the three animals the 1° CEP site was determined for C₅ phrenic, whole phrenic and C₅ phrenic severed stimulation. The 1° CEP peak amplitude and latencies were compared with repeated measures ANOVA. Significance was P < 0.05.

Inspiratory occlusion elicited cortical evoked potential

In three spontaneously breathing animals without neuromuscular blockade, single-breath inspiratory occlusions were presented. Each occluded breath was separated by three to six unoccluded breaths. The occlusions were presented 100 times and the surface cortical activity recorded. The occlusion presentation was marked by airway pressure (P_AW). The ECoG was recorded from the left somatosensory area of the cortex as described above. The recorded data were analysed by digitizing the P_AW and ECoG at 2 kHz (Cambridge Electronics Design, Model 1401+). For each load presentation, 200 ms of ECoG activity and P_AW (50 ms pre-occlusion and 150 ms post-occlusion) were digitized for computer signal averaging (SIGAVG, Cambridge Electronic Design). The computer stored each individual load presentation. For the occlusion trial, each presentation was recalled from memory and inspected for the presence of a rapid P_AW change greater than −1 cmH₂O, indicative of an occlusion. A minimum of 64 occlusions were included in each averaged occlusion CEP. The averaged CEPs were then stored on computer disk. This was repeated for the no-load control. A minimum of 64 control breaths were included in each averaged control CEP. The occlusion and control CEPs were recorded over a 2mm × 2mm grid between the cruciate and ansate sulci. The primary inspiratory occlusion CEP site was determined.

Neuromuscular blockade was then performed and the cat was mechanically ventilated. The right whole phrenic nerve was isolated intrathoracically as described above. The same cortical grid used for inspiratory occlusion was remapped with right whole phrenic nerve stimulation. When the surface evoked potentials were collected at each grid position, the site of the phrenic afferent 1° CEP was determined.

Microelectrode cortical laminar analysis

Low-impedance, multiunit somatosensory cortical neural activity was recorded in six animals. A glass-insulated tungsten microelectrode (0.5–2 MΩ) was positioned perpendicular to the cortical surface at the whole phrenic afferent-elicited 1° CEP site. The tip of the electrode was placed at the pial surface and the electrical stimulation of the whole phrenic nerve repeated and the evoked activity recorded. The electrode was advanced with a microdrive 25 μm into the cortex and the nerve again stimulated. This was repeated at 25 μm increments to a total depth of 2.5 mm. A minimum of 128 epochs were recorded at each cortical depth. When a whole phrenic nerve-activated cortical neural activity was found, the position was recorded together with the response to stimulation.

Single-unit cortical neuron recording

The somatosensory region of the cortex was systematically probed in nine animals with microelectrodes (5–10 MΩ). A microelectrode was inserted in a 0.5 mm × 0.5 mm grid surrounding the whole phrenic afferent-elicited 1° CEP site. The extracellular neuronal recordings were made with tungsten microelectrodes in a microdrive. The electrode was connected to a high-impedance probe connected to an AC preamplifier (P511, Grass Instruments, Quincy, MA, USA), band-pass filtered at 0.3–3.0 kHz, displayed on an oscilloscope (Tektronix, Model 5111) and recorded on a computer (Model 1401, Cambridge Electronics Design) for off-line analysis. The microelectrode was lowered to the cortical surface and advanced in 25 μm increments to a total depth of 2.5 mm at each grid position tract. At each increment, the whole phrenic nerve was stimulated. When a phrenic afferent-activated neuron was encountered, the neuronal responses to a minimum of 64 nerve stimulations were recorded, and the electrode tip coordinates were then logged.

A marking lesion was made in the last electrode tract at the completion of the recording session. The animals were then perfused with heparinized saline followed by 4% formalin. The brain was removed and fixed in buffered 4% formalin. The fixed tissue was cut coronally into 50 μm thick sections with a freezing microtome. The sections were mounted and stained with haematoxylin and eosin. The stained sections were examined to identify the lesion and the corresponding electrode tract.

Data analysis

The recorded neural activity was digitized with a sampling rate of 6 kHz using a computer analysis system (Model 1401, Cambridge Electronics Design). Relative to the stimulus trigger, neural responses were recorded from 50 ms pre-stimulus to 200 ms following the stimulus, and each was recorded in computer memory. The individual responses were then recalled from memory and the stimulus was marked as zero-time. The neuron onset latency was measured as the time from the stimulus to the first action potential. The duration of the neuron response was measured as the time from the first action potential until the neuron discharge ceased. In a very few recording sites, the activity of two to three neurons could be identified and separated by the amplitude and frequency of the action potentials. Simultaneously recorded neurons from extracellular electrode recordings were sorted by: (1) identification of waveforms that are likely to belong to individual spike groups; (2) differentiation between spike amplitude; (3) determination of the discriminable spike waveforms; (4) spike detection and classification to group spike waveforms; and (5) separation of loosely overlapped spikes.

Results

Phrenic afferent conduction velocities

Conduction velocities of the phrenic nerve fibres were calculated from recordings of the C₅ portion of the phrenic following stimulation of the thoracic phrenic nerve. The compound action potential recordings were characterized by three peaks with mean stimulus-to-peak conduction velocities of 60.1 ± 4.1, 41.3 ± 2.9 and 27.7 ± 2.9 m s⁻¹, respectively.

Phrenic afferent cortical evoked potential recording

Short-latency, initially positive 1° CEPs were recorded with stimulation of myelinated afferents of the intrathoracic phrenic nerve in the contralateral post-cruciate gyrus of all animals (n= 42) tested (Fig. 1). The mean onset and peak latencies were 8.5 ± 5.7 ms and 21.8 ± 9.8 ms, respectively (Fig. 2). The rostral–caudal surface location of the 1° CEP was found between the rostral edge of the post-cruciate dimple (PCD) and the rostral edge of the ansate sulcus. The medio-lateral surface location was between 2 mm lateral to the sagittal sulcus and the lateral end of the cruciate sulcus (Fig. 1). The highest densities of 1° CEP locations were near the PCD. Histological examination revealed that the 1° CEP sites were recorded over areas 3a and 3b of the SI cortex (Fig. 3). It was common for the 1° CEP site to be found at the border of areas 3a and 3b. Right phrenic and left phrenic stimulation in the same animal (n= 2) elicited a 1° CEP in their respective contralateral cortices. The CEP was recorded bilaterally (n= 6) in response to left phrenic nerve stimulation; the contralateral 1° CEP onset and peak latencies were significantly shorter than ipsilateral CEP latencies (Fig. 4). The 1° CEP was abolished when the dorsal roots for C₅, C₆ and C₇ were severed (n= 5).

Figure 3. Photograph of an histological saggital section through the rostral cerebral cortex of an individual animal.

The arrows indicate the lesion made at the 1° CEP locus elicited by whole phrenic nerve stimulation intrathoracically. The architechtonic areas of this region of the cerebral cortex are indicated (Hassler & Muhs-Clement, 1964).

Figure 4. Bilateral cat cortical loci activated by phrenic nerve afferents.

The top panels are the rostral cat cerebral cortex schematic representation of the sensorimotor, pericruciate region. The 1° CEP was recorded contralateral and ipsilateral to whole phrenic nerve stimulation intrathoracically. Each filled circle represents the cortical surface location of the 1° CEP for an individual animal. The bottom panel is a 1° CEP from an individual animal with the contralateral and ipsilateral cerebral cortical CEP recorded simultaneously.

Cervical phrenic root stimulation

Stimulation of the C₅ portion of the phrenic nerve (n= 9) in the neck elicited a short-latency 1° CEP in the same location as intrathoracic stimulation of the whole phrenic nerve (Fig. 5). The onset and peak latencies for C₅ phrenic stimulation were 9.5 ± 3.4 ms and 23.4 ± 6.5, respectively. The surface location of the 1° CEP for C₅ phrenic afferent stimulation colocalized with intrathoracic phrenic nerve stimulation (Fig. 5). Stimulation of the intrathoracic whole phrenic nerve with the C₅ portion severed (n= 3) elicited a 1° CEP at the same location (Fig. 6). The latency was not significantly different but the amplitude was reduced.

Figure 5. Cat cortical loci activated by stimulation of C5 rootlet phrenic nerve afferents.

The inset panel is a schematic representation of the sensorimotor, pericruciate region of the cat cerebral cortex with each filled circle representing the cortical surface location of the C5 phrenic 1° CEP for an individual animal.

Figure 6. The 1° CEPs with bilateral phrenic nerve stimulation for an individual animal.

The 1° CEPs recorded and elicited in the cerebral cortex contralateral to the phrenic nerve stimulated. The top trace is the 1° CEP recorded in the left cerebral cortex and elicited by stimulation right whole phrenic nerve intrathoracically. The middle trace was recorded at the same cerebral cortical site as the top trace with the 1° CEP elicited by stimulation of the C5 rootlet of the right phrenic nerve. The bottom trace is the 1° CEP recorded in the right cerebral cortex an elicited by stimulation left whole phrenic nerve intrathoracically.

Inspiratory occlusion elicited cortical evoked potential

Inspiratory occlusions elicited a 1° CEP in the pericruciate region of the cat cortex (Fig. 7). The peak latency of the primary positive CEP peak was 28.9 ± 6.3 ms. The inspiratory occlusion sites (Fig. 7) were within the area of highest density 1° CEP loci evoked by phrenic nerve stimulation (Fig. 1).

Figure 7. Cat cortical loci activated by inspiratory occlusion.

The left panel is the 1° CEP elicited by repeated single breath occlusions in an individual animal. The right panel is a schematic representation of the sensorimotor, pericruciate region of the cat cerebral cortex with each filled circle representing the cortical surface location of the inspiratory occlusion-elicited 1° CEP for an individual animal. The open circles are the 1° CEP for whole phrenic nerve stimulation in the same animals. The shaded oval is the cortical area that contains more than 75% of the phrenic nerve-elicited 1° CEP loci from Fig. 1.

Microelectrode cortical laminar analysis

The phrenic afferent 1° CEP position was further investigated with low impedence microelectrodes (n= 6). Stimulation of the intrathoracic phrenic nerve elicited a positive voltage 1° CEP at the surface. The amplitude of the CEP decreased as the electrode was advanced into the cortex. The polarity of the CEP reversed at a depth of 1.4 ± 0.1 mm in 5 of 6 animals and 0.44 mm in 1 of 6 animals. The laminar analyses were in area 3b (Fig. 3) in these animals (n= 6).

Single-unit cortical neuron recording

Intracortical activation of 16 neurons was investigated in nine animals. Two patterns of neural activity (Fig. 8) were recorded: (1) short-latency, short-duration activation of neurons (n= 8) and (2) long-latency, long-duration activation of neurons (n= 8). Short-latency neurons had a mean onset latency of 10.4 ± 3.1 ms and mean burst duration of 10.1 ± 3.2 ms (Fig. 8A). The short-latency units were recorded at an average depth of 1.7 ± 0.5 mm below the cortical surface. The long-latency neurons had a mean onset latency of 36.0 ± 4.2 ms and mean burst duration of 32.2 ± 8.4 ms (Fig. 8B), significantly greater (P < 0.001) than the short-latency units. The long-latency units were recorded at an average depth of 2.4 ± 0.2 mm below the cortical surface, significantly greater (P < 0.01) than the short-latency units.

Figure 8. Somatosensory cortical neuronal response to contralateral whole phrenic nerve stimulation.

A is a short-latency cortical neuron activated by intrathoracic stimulation of the contralateral whole phrenic nerve. B is a recording from a different animal of 2 neurons, short- and long-latency activated by intrathoracic stimulation of the contralateral whole phrenic nerve.

Somatosensory and phrenic afferent cortical evoked potential recording

The 1° CEPs for phrenic afferents were recorded in the same animals (n= 9) that received forelimb, hindlimb, trunk and visceral nerve stimulation (Fig. 9). Stimulation of the superficial and deep radial nerves (forelimb) elicited 1° CEPs in the lateral post-cruciate gyrus, 1–3 mm lateral to the cruciate sulcus. Stimulation of the sural and medial gastrocnemius nerves (hindlimb) elicited 1° CEPs in the post-cruciate gyrus, 1–3 mm lateral to the saggital sulcus. Intercostal nerve (trunk) stimulation elicited a 1° CEP close to the phrenic nerve location in a medio-lateral plane and 1–2 mm rostral. Stimulation of the major splanchnic nerve (visceral) elicited a 1° CEP on the rostral edge of the PCD. The 1° CEP projections of phrenic nerve afferents were lateral to the hindlimb, medial to the forelimb and similar in a medio-lateral plane for thoracic and visceral nerves of the cat sensorimotor cortex.

Figure 9. Cat cortical 1° CEP loci activated by stimulation of forelimb, hindlimb and visceral nerve afferents.

This is a schematic representation of the sensorimotor, pericruciate region of the cat cerebral cortex. Each number represents a cortical surface location of the 1° CEP for an individual animal for each nerve stimulated. Forelimb nerves were the cutaneous superficial radial (2) and muscle deep radial (3) nerves. The hindlimb nerves were the cutaneous sural (6) and muscle medial gastrocnemius (7) nerves. The thoracic mixed cutaneous and muscle intercostal nerve is indicated as 4. The visceral major splanchnic nerve is indicated as 5. The shaded oval is the cortical area that contains more than 75% of the 1° CEP loci from Fig. 1.

Discussion

Phrenic nerve afferents have a short-latency, myelinated afferent projection pathway to the primary somatosensory cortex (SI) in cats. The specific afferents stimulated in this study are unknown; however, the electrical stimulus parameters were set at levels known to depolarize low-threshold, large myelinated group I and II afferents. In addition, stimulation of the phrenic nerve close to the diaphragm prevented contributions from non-diaphragmatic afferents (mediastinal and cardiac) which join the phrenic nerve cranial to the stimulation site. Thus, the afferents eliciting the central neural activity reported in the present study are primarily diaphragmatic afferents.

The cat diaphragm is known to have muscle spindles and Golgi tendon organs (Corda et al. 1965; Jammes et al. 1986; Holt et al. 1991; Jammes & Balzamo, 1992; Balkowiec et al. 1995). These mechanoreceptors have been identified histologically (Duron et al. 1978) and neurophysiologically (Corda et al. 1965; Holt et al. 1991; Balkowiec et al. 1995). They spontaneously discharge in response to changes in diaphragm mechanics. The C₅ phrenic root compound action potential recorded in the present study had conduction velocities of 28–60 m s⁻¹. These conduction velocities are consistent with group I and II myelinated nerve fibres including diaphragm muscle spindles, Golgi tendon organs and pressure receptors (Corda et al. 1965; Holt et al. 1991; Balkowiec et al. 1995). The phrenic afferent SI CEP latency probably includes at least four synaptic sites: afferent activation of dorsal column neuron (Larnicol et al. 1984; Chou & Davenport, 2005), activation of external cuneate (Larnicol et al. 1985), activation of thalamic neurons (Yates et al. 1994; Zhang & Davenport, 2003) and SI cortical neurons (Frankstein et al. 1979; Davenport et al. 1985; Yates et al. 1994) suggesting a 1.2–2.0 ms combined synaptic delay. The mean phrenic 1° CEP onset latency (10.1 ms) is consistent with group I afferent conduction velocities while the peak latency is consistent with group I and II afferent conduction velocities. The early onset cortical neuron onset latency (mean = 10.1 ms) is also consistent with group I afferent conduction velocities (Frankstein et al. 1979) while the cortical neuron late onset latency (34.0 ms) is consistent with group I and II afferent conduction velocities (Frankstein et al. 1979). Group III and IV afferents have conduction velocities that would require onset latencies greater than 100 ms to mediate the 1° CEP recorded in the present study. Thus, it is most likely that the contralateral phrenic afferent activation of the SI cortex in the present study was mediated by group I and/or II myelinated afferents.

Corda et al. (1965) reported a fast-adapting afferent in the cat diaphragm that could not be classified as a muscle spindle or tendon organ. The properties of these afferents suggest that they are low-threshold myelinated fibres. This population of afferents would be stimulated with the parameters used in the present study and may be the probe-sensitive mechanoreceptor recorded in vitro from the rat diaphragm described by Holt et al. (1991). However, it is unlikely that group III afferents were activated with the stimulus parameters used in the present study. These afferents are high-threshold, small-diameter myelinated fibres that require a wider pulse width (or pulse trains) and higher stimulus currents for excitation (Duron & Marlot, 1980; Balkowiec et al. 1995).

The cat phrenic nerve originates from fibres of the C₅, C₆ and C₇ spinal nerves (Duron et al. 1979; Chou & Davenport, 2005). Phrenic nerve afferents have been recorded primarily from the phrenic nerve and C₅ and C₆ dorsal rootlets of the cat (Corda et al. 1965; Duron et al. 1978; Balkowiec et al. 1995; Chou & Davenport, 2005). Excitation of diaphragmatic afferents by intrathoracic phrenic nerve stimulation elicited potentials at all recording sites on C₅ and C₆ dorsal roots (Chou & Davenport, 2005). Cord dorsum potentials were not recorded from C₄, indicating that diaphragmatic afferents do not enter thorough this spinal segment. Dorsal root potentials were, however, recorded in rostral dorsal C₇. This suggests that diaphragmatic afferents enter the C₇ spinal segment but there is a segregation of entry to the rostral dorsal rootlets (Chou & Davenport, 2005). The number and amplitude of the cord dorsum peaks recorded in the C₅–C₇ spinal segments with thoracic phrenic nerve stimulation similar to the present study suggests that predominant phrenic afferent input into the spinal dorsal column pathway is from the C₅ and C₆ spinal segments.

Intrathoracic stimulation of phrenic afferents elicit primary CEPs in the SI cortex of cats (Frankstein et al. 1979; Davenport & Reep, 1985; Davenport et al. 1985; Yates et al. 1994). The regions predominantly activated in the present study were areas 3a and 3b on the rostro-medial edge of the post-cruciate dimple (Hassler & Muhs-Clement, 1964). The primary foci were often at the 3a–3b border (Fig. 3). Stimulation of the C₅ portion of the phrenic nerve in the neck elicited a primary CEP that was colocalized with the intrathoracic phrenic loci. Similar results were obtained when the C₅ portion of the phrenic was severed and the intrathoracic phrenic nerve stimulated. This preparation would remove phrenic afferents entering the C₅ dorsal roots, leaving the primarily C₆ portion of the phrenic nerve intact with a small contribution from C₇. The primary CEP was again colocalized with the loci for whole phrenic stimulation. The reduction in amplitude of the CEP is probably due to a reduction in the number of activated afferents projecting to the cortex. These results also suggest that the phrenic afferents from all the cervical spinal segments contribute to the phrenic afferent activation at the same SI cortical locus.

Area 3a is known to receive short-latency contralateral muscle spindle input from many other muscle systems (Ruch et al. 1952; Oscarsson & Rosen, 1963). Tendon organ activation of 3a neurons has also been reported (Oscarsson & Rosen, 1966; Oscarsson et al. 1966). The diaphragmatic afferents stimulated in the present study included muscle spindles and tendon organs (Corda et al. 1965; Balkowiec et al. 1995), and the cortical projection of these afferents is therefore similar to other muscles. Diaphragmatic afferent excitation of area 3b neurons may be from these muscle spindles and tendon organs as well as the probe-sensitive or fast-adapting receptors (Corda et al. 1965; Holt et al. 1991; Balkowiec et al. 1995). Area 3b neurons have been shown to receive contralateral cutaneous afferent input and to a lesser extent muscle afferent activation (Dykes et al. 1980; Felleman et al. 1983). The projection of diaphragmatic afferents to area 3b may also be the result of convergence onto higher order neurons that also have cutaneous input (Bolser et al. 1991) and contribute to the interaction between cutaneous and respiratory somatosensation (Balzamo et al. 1995; Balzamo et al. 1999).

Phrenic afferent-activated SI neurons were found in laminas III and IV of areas 3a and 3b. These lamina are known to receive thalamic afferent input suggesting a thalamic relay for group I and II phrenic afferents (Andersson et al. 1966; Friedman & Jones, 1981; Yates et al. 1994). Larnicol et al. (1984, 1985) reported phrenic afferent-mediated fluorescent labelling in the dorsal column and external cuneate nucleus. The dorsal column–external cuneate–thalamic cortical pathway is also supported by the recording of thalamic neurons that are excited by low-threshold, fast-conducting group I and II myelinated phrenic afferent stimulation (Zhang & Davenport, 2003). Mechanical probing of the shoulder also elicited activity in the same thalamic neurons (Zhang & Davenport, 2003). Yates et al. (1994) reported that retrograde fluorescent labelling of the primary SI phrenic afferent-activated cortical region resulted in labelling neurons in the VPL region of the thalamus, consistent with electrophysiological recordings (Zhang & Davenport, 2003). Thus, group I and II phrenic afferents activate neurons in the VPL region of the thalamus, and neurons from this region of the thalamus are known to project to the SI phrenic cortex (Yates et al. 1994). It appears, therefore, that diaphragmatic afferents may converge with cutaneous afferents on higher order neurons that then project to the SI cortex via a VPL thalamic relay. These observations, coupled with the results of the present study, support the hypothesis that low-threshold, fast-conduction velocity myelinated phrenic afferents project to the SI cortex via the dorsal columns, external cuneate nucleus, VP thalamus and then into area 3a/3b of the SI cortex. It should be noted that identification of phrenic afferent-activated neurons in this area was biased by the prior identification of the cortical surface CEP dipole. Hence, phrenic afferent-activated neurons are present but may not be limited to area 3a/3b of the SI cortex.

The occasional observation of phrenic afferent excitation of area 4-γ suggests a multiple representation in the cat cortex similar to reports from other sensory systems (Landgren et al. 1967; Rasmusson et al. 1979; Dykes et al. 1980; Felleman et al. 1983; Lipski et al. 1986). Area 4-γ has been shown to have motor projections when the lamina V pyramidal tract neurons are stimulated. Stimulation of the cat cortex has been shown to elicit efferent activation of the diaphragm (Lipski et al. 1986). It has also been reported that the diaphragm is represented in the human precentral motor region (Foerster, 1936). While localization of cat motor sites has been reported (Lipski et al. 1986), there was no colocalization of diaphragmatic afferent and efferent loci. This suggests a cortio-cortical connection between SI cortex and the motor cortex mediating a transcortical diaphragmatic sensory–motor reflex pathway.

Low-threshold, fast-conducting phrenic afferents project to the trunk region (Fig. 9) of the cat homunculus (Frankstein et al. 1979; Davenport et al. 1985; Yates et al. 1994). As mentioned above, phrenic afferents enter the spinal cord through the C₅, C₆ and C₇ segments. Upper neck and shoulder forelimb afferents that enter C₅, C₆ and C₇, have a SI representation on the lateral post-cruciate gyrus in the ‘forelimb’ area (Dykes et al. 1980). In the present study, large myelinated phrenic afferents entering the same spinal segments activate the SI cortex medial to the forelimb, in the trunk region of the cat homunculus (Fig. 9). The phrenic afferent primary loci are also medial to intercostal nerve representations (Davenport et al. 1993). It would therefore appear that phrenic afferents are represented in the SI cortex based on the body position of the diaphragm rather than spinal segment. This would suggest that higher order relay neurons receiving myelinated diaphragmatic afferent activation sort the input from the forelimb spinal segment to the lateral area of the SI cortex and diaphragmatic afferents (entering the same dorsal spinal roots) to a medial representation consistent with the diaphragm's location in the middle of the trunk. This may be related to the interaction between respiratory and somatosensory sensations (Balzamo et al. 1995; Balzamo et al. 1999) and the result of convergence of limb and respiratory muscle afferent in the SI cortex. Future investigation of the neural pathway relaying diaphragmatic input to the SI cortex will be necessary to determine where this sorting occurs.

The phrenic myelinated afferent primary projection area of the SI cortex also receives visceral input from vagal afferents (Ito & Craig, 2003), venous afferents (Thompson et al. 1980) and the major splanchnic nerve (Frankstein et al. 1979). The region around the post-cruciate dimple has been reported to be activated by hindlimb venous afferents that enter lumbar spinal segments (Thompson et al. 1980). In the present study, stimulation of major splanchnic afferents elicited CEPs in the PCD region that were not colocalized with phrenic afferents. We speculate that the SI cortex near the PCD is a SI cortical centre for integration of trunk visceral afferent somatosensation.

Finally, it has been reported that inspiratory occlusion elicits somatosensory cortical neural activity (Logie et al. 1998; Davenport & Hutchison, 2002; Davenport & Vovk, 2009); however, the mechanoreceptors mediating this cortical response are unknown. In humans and lambs (Davenport & Hutchison, 2002; Davenport et al. 2006), inspiratory occlusions delivered via an endotracheal tube or facemask elicit somatosensory CEPs in the somatosensory area. In the present study, inspiratory occlusion delivered to a limited number of spontaneously breathing, tracheostomized cats similarly elicited a SI CEP (Fig. 7). The SI loci for the primary occlusion CEP were in the same cortical areas as phrenic (Fig. 1) and intercostal (Fig. 9) muscle afferents elicited electrical and mechanical stimulation (Davenport et al. 1993). While it remains unknown if inspiratory occlusion somatosensation is mediated by respiratory muscle mechanoreceptors, the results of previous reports (Gandevia & Macefield, 1989; Davenport et al. 1993; Zhang & Davenport, 2003) and the present study are consistent with this hypothesis.

Summary

The results of the study demonstrated that myelinated phrenic nerve afferents have a short-latency central projection to the SI somatosensory cortex. The neural pathway is hypothesized to be similar to limb somatosensory projections (Ruch et al. 1952; Dykes et al. 1980). Phrenic myelinated afferents enter the spinal cord through the dorsal roots, suggesting a dorsal column projection to the somatosensory cortex (Chou & Davenport, 2005). Phrenic myelinated afferents are also known to activate the thalamic ventro-posterio-lateral nucleus (Yates et al. 1994; Zhang & Davenport, 2003; Chou & Davenport, 2005). The phrenic myelinated afferent pathway was shown in the present study to activate neurons in lamina III and IV of areas 3a and 3b in the SI somatosensory cortex. The cortical representation of myelinated phrenic nerve afferents is medial to the forelimb, lateral to the hindlimb, similar to thoracic loci, hence the phrenic myelinated afferent SI site in the cat homunculus is consistent with body position (thoracic region) rather than spinal segment (C₅–C₇). There is a biphasic response of phrenic afferent-activated SI neurons with a short-latency activation followed by long-latency activation. The phrenic afferent activation of the somatosensory cortex is bilateral, with the ipsilateral cortical activation occurring subsequent to the contralateral, suggesting a cortico-cortical (perhaps via the corpus collusum) connection. These results support the hypothesis that phrenic afferents provide somatosensory information to the cerebral cortex which can be used for diaphragmatic and respiratory muscle proprioception and somatosensation.

Glossary

Abbreviations

1° CEP

primary cortical evoked potentials

PCD

post-cruciate dimple