Refining the sensory and motor ratunculus of the rat upper extremity using fMRI and direct nerve stimulation



It is well understood that the different regions of the body have cortical representations in proportion to the degree of innervation. Our current understanding of the rat upper extremity has been enhanced using functional MRI (fMRI), but these studies are often limited to the rat forepaw. The purpose of this study is to describe a new technique that allows us to refine the sensory and motor representations in the cerebral cortex by surgically implanting electrodes on the major nerves of the rat upper extremity and providing direct electrical nerve stimulation while acquiring fMRI images. This technique was used to stimulate the ulnar, median, radial, and musculocutaneous nerves in the rat upper extremity using four different stimulation sequences that varied in frequency (5 Hz vs. 10 Hz) and current (0.5 mA vs. 1.0 mA). A distinct pattern of cortical activation was found for each nerve. The higher stimulation current resulted in a dramatic increase in the level of cortical activation. The higher stimulation frequency resulted in both increases and attenuation of cortical activation in different regions of the brain, depending on which nerve was stimulated. Magn Reson Med 58:901–909, 2007. © 2007 Wiley-Liss, Inc.

The cortical representation of the rat forepaw has been well-studied using functional MRI (fMRI) (1–4). Most fMRI studies of the rat upper extremity have focused on the forepaw because it elicits reproducible activation in the sensory region of the cerebral cortex (S1FL) (5). Recent forepaw studies have varied the stimulation parameters (6), while others have focused on understanding the impact of cortical activity under different physiological conditions such as depressed blood pressure (7), cerebral ischemia (8, 9), and hypoglycemia (10). A literature review did not reveal any studies beyond forepaw stimulation that have focused on defining with greater precision the motor and sensory regions of the rodent upper extremity using fMRI.

Brachial plexus and peripheral nerve injuries are extremely debilitating in humans. In infants such injuries are obstetrical-related, while in adults the injuries are more often the result of trauma. Injuries result in varying loss of sensory and motor function in the affected extremity. The most serious of injuries may require surgical neurotization, nerve grafting, or nerve transfer to restore essential function such as elbow and digit flexion. Many of the procedures used clinically have previously been studied using animal models—the rat in particular (11–14).

The four nerves in this study are among five of the major nerves to the upper extremity. The musculocutaneous nerve provides motor innervation to the coracobrachialis, the biceps brachii, brachialis, pronator teres, and flexor carpi radialis. These muscles contribute to the movement of different joints including anterior flexion and adduction of the shoulder joint, elbow flexion, forearm pronation, and wrist flexion. Beyond the elbow, the musculocutaneous nerve becomes the lateral cutaneous nerve of the antibrachium, a sensory nerve that supplies the forelimb and forepaw. The median nerve provides the motor innervation to most of the flexor muscles of the digits, but also the pronator quadratus, which also contributes to pronation of the forearm. The median nerve divides into three common volar digital nerves, each of which divides further to form two proper volar digital nerves supplying the sensory input to the first through fourth digits. The ulnar nerve provides some contribution to wrist flexion and flexion for digits four and five, as well as the intrinsic muscles of the forepaw. The ulnar nerve also provides the sensory input to the fourth and fifth digits. The radial nerve is largely responsible for elbow, wrist, and digital extension; forearm supination; and contributes to elbow flexion. In addition, the radial nerve supplies the sensory innervation to the dorsum of the forearm and dorsal aspect of the second, third, and a portion of the fourth digits. Table 1 shows the muscular and sensory innervations of each nerve (15).

Table 1. Motor and Sensory Distribution of the Ulnar, Median, Radial, and Musculocutaneous Nerves (15)
Muscle innervationFlexor carpi ulnaris Flexor digitorum profundus Interossei Lumbricales Adductor pollicis brevis Flexor pollicis brevis Adductor digiti quinti Flexor digiti quinti brevis Opponens digiti quinti EpitrochleoanconeusFlexor digitorum sublimis Flexor digitorum profundus Flexor pollicis brevis Palmaris longus Pronator quadratusLong, medial, lateral head of triceps Short head of biceps Coracobrachialis Anconeus Brachialis Extensor carpi radialis brevis Extensor digitorum communis Extensor digiti quinti Extensor digiti quinti propius Extensor carpi ulnaris Supinator Abductor pollicis Extensor pollicis brevis Extensor indicis propiusCoracobrachialis Short head of biceps Brachialis Pronator teres Flexor carpi ulnaris
Sensory innervationVolar branch—gives rise to a cutaneous branch that forms a common volar digital nerve that further branches to form two proper digital nerves to adjacent sides of digits four and five and a proper digital nerve to the ulnar side of the fifth digit. Dorsal branch gives rise to two dorsal digital nerves that supplies digit five and adjacent sides of digits four and five.Forms three common volar digital nerves to the first, second, third interdigital space. Further branches to form two proper digital nerves to adjacent sides of the first and second, second and third, third and fourth digits.Two dorsal digital nerves supplies adjacent sides of second and third, third and fourth digits. Communicates with musculocutaneous nerve to form the dorsal digital nerve of first webspace.Lateral cutaneous nerve of the antibrachium Base of pollex Communicates with radial nerve to form the dorsal digital nerve of first webspace.

This study describes a surgical approach to nerves of the rat brachial plexus to allow for direct nerve stimulation using implanted electrodes. To our knowledge, there have been no other fMRI studies involving direct nerve stimulation in the rat upper extremity. In the more proximal regions of the upper extremity, the nerves carry both sensory and motor nerve fibers before dividing into pure motor or sensory branches as the nerve travels more distally. Stimulating the mixed nerves gives us an opportunity to study both motor and sensory components of the rat upper extremity and examine cortical regions beyond the sensory forepaw region. Specifically, we examine direct nerve stimulation of the ulnar, median, radial, and musculocutaneous nerves of the rodent upper extremity. By refining the motor and sensory regions of each nerve of the rat upper extremity, we hope to develop an animal model to study and characterize peripheral nerve injuries and treatments.

Anesthesia plays an important role in the success of animal fMRI experiments. It is well known that anesthesia can inhibit neuronal activity and functional hyperemic response (16–20). Proper dosage along with robust physiological monitoring can promote proper homeostasis during the hours the rodent subject spends in the scanner (21). Medetomidine hydrochloride (Domitor, Pfizer, New York, NY) is an injectable anesthetic used in clinical veterinary practice and is also gaining acceptance in the fMRI literature (22–24). The drug is a reversible adrenoreceptor 2 agonist with a short half-life and an ability to promote a steady state of sedation. This study utilizes Domitor combined with the clinically used muscle relaxant pancuronium bromide. The anesthesia regimen supports overall animal well-being, along with preventing motion interference from the animal.


Direct Nerve Stimulation

Twenty-one Sprague-Dawley rats were used to study the musculocutaneous, ulnar, median, and radial nerves (n = 5 for each nerve except ulnar, where n = 6). The rat was placed supine and anesthesia was provided by 1% isoflurane vaporized into 30–70% O2/N2. An incision was made along the medial aspect of the upper extremity and extended into the axilla and a portion of the flank (Fig. 1a). Sutures were placed to retract the subcutaneous tissues and the pectoralis major muscle was identified. The pectoralis was retracted medially to expose the brachial plexus and artery (Fig. 1b). The largest and most visible nerve is the median nerve (Fig. 1c). The ulnar nerve runs parallel and medial to the median nerve, while the radial nerve lies deep to the brachial artery. The long and short heads of the biceps brachii muscles are lateral to this neurovascular bundle. The long head of the biceps brachii was reflected laterally exposing the musculocutaneous nerve between the retracted biceps muscles (Fig. 1c). Care was taken to dissect the desired nerve out circumferentially, handling the nerve solely by the epineurium. A plastic sheet was used to isolate the nerve, the electrode was then hooked around the nerve, and the distal silicone sheet was used to secure the electrode in position (Fig. 1d). The proximal silicone sheet was secured to the surrounding tissue to add further stabilization to the electrode. When placing electrodes on the ulnar, median, or radial nerve, care was taken to ensure the electrode contacted only the nerve to be stimulated. In addition to implanting the electrodes, the right femoral artery and vein were used for invasive blood pressure monitoring and for continuous IV drug administration. The trachea was cannulated to allow for mechanical ventilation during the fMRI acquisition. The total surgical times varied from 1.5 to 2.0 hr.

Figure 1.

a: The rat is positioned supine. The black marker shows the location and size of the incision. b: The double black arrow shows the two heads of the biceps brachii muscle. The blue arrow points to the median nerve. Adjacent to the median nerve is a pure sensory nerve that innervates the upper arm. c: The biceps brachii muscles are retracted with the aid of suture and a retractor. The musculocutaneous nerve and couple branches are seen between the biceps brachii. d: Shows placement of the electrode on the musculocutaneous nerve just prior to securing the electrode in place. e: Picture of the electrode modified with a piece of silicone to ensure the two ends do not touch. The separation between lines of the ruler = 1 mm.

Electrode Preparation

A 150-μm diameter stainless-steel bipolar electrode (AISI 304, Plastics1, Roanoke, VA) was used for direct nerve stimulation during the fMRI experiments. Two small pieces of silicone were attached to the electrode (Fig. 1e, only one piece of silicone shown). The silicone ensured that the bipolar electrodes did not touch and allowed a second point of fixation to the surrounding tissue. Brain image distortion associated with the electrode and silicone was not apparent at 9.4T.

fMRI Protocol

Isoflurane was gradually tapered as a continuous infusion (PHD2000 MRI-Pump, Harvard Apparatus, Holliston, MA) of medetomidine hydrochloride (Domitor, 0.1 mg/kg/hr) for the maintenance of anesthesia and pancuronium bromide (2 mg/kg/hr) for muscle paralysis was started. The rat was placed on a mechanical ventilator (MRI-1 Ventilator, CWE, Ardmore, PA) using 30–70% O2/N2. The rat was loaded onto a custom-built cradle fabricated using G-10 fiberglass material, which has a magnetic susceptibility similar to air. The cradle was built with a warming surface controlled by a waterpump-driven temperature regulator (Medi-therm III, Gaymar Industries, Orchard Park, NY). A bipolar beryllium copper electrode was placed in the left forepaw to serve as a control.

Four separate electrical stimulation (S88 Square Pulse Stimulator, Grass Telefactor, West Warwick, RI) protocols that differed in current level (0.5 mA or 1.0 mA) or frequency (5 Hz or 10 Hz) were used. One ms pulse duration was used for each nerve stimulation protocol. The forepaw stimulation protocol included a current of 2.0 mA, a frequency of 10 Hz, and a duration of 3 ms. Each nerve and forepaw stimulation sequence began with an OFF period of 40 sec followed by three repetitions of ON for 20 sec and OFF for 40 sec (total scan time = 3 min 40 sec). The stimulation sequence was computer controlled with LabVIEW software (National Instruments, Austin, TX) and started by a trigger pulse delivered by the scanner. A rapid acquisition with relaxation enhancement (RARE) anatomical image was acquired with a 256 × 256 matrix, TE = 12.5 ms, TR = 2.5 sec, and the same slice geometry as the echo planer imaging (EPI) sequence. Gradient echo scans (single-shot EPI, TE = 18.76 ms, TR = 2 sec, matrix size 96 × 96, FOV = 4 cm, number of repetitions = 110, 10 contiguous interleaved 1-mm slices, acquisition time = 3 min 40 sec) were acquired on a 9.4T Bruker AVANCE MRI scanner (Bruker BioSpin, Billerica, MA) with a 30-cm bore. Two sets of gradient echo images were acquired for each stimulation protocol. Images were acquired using a Bruker receiving surface coil (T9208) and a linear transmit coil (T10325). The total fMRI protocol duration was 2 hr. These studies were performed in compliance with federal regulations and the guidelines of our institution's Animal Care and Use Committee.

Physiologic Monitoring

During the fMRI protocol, the invasive blood pressure, body temperature, respiratory rate (Model 1025, SA Instruments, Stony Brook, NY), pulse oximetry (8600V, Nonin Medical, Plymouth, MN), arterial blood gases (i-Stat, Heska, Loveland, CO), and inspired/expired O2 and CO2 (POET IQ2, Criticare Systems, Waukesha, WI) were monitored (WinDaq Pro, DataQ Instruments, Akron, OH) and maintained within normal physiologic ranges. The rat was euthanized using an overdose of sodium pentobarbital upon completion of the study.

Separate bench experiments using five Sprague-Dawley rats were conducted to record changes in arterial blood pressure, pulse, pulse oximetry, and arterial blood gases during active ulnar nerve or forepaw stimulation. The same surgical and anesthetic conditions were used as described above. After transitioning the anesthetic conditions to pancuronium bromide and medetomidine hydrochloride, there was a 20-min period of equilibration prior to the initiation of the stimulation protocol.

Data Analysis

The EPI scans were registered to an ideal RARE anatomy with a 12 degree of freedom affine transformation using the Linear Image Registration Tool (FLIRT) software from the Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (25). The images for each nerve and stimulation protocol were averaged (3dcalc) using Analysis of Functional NeuroImages (AFNI) software (26). The averaged data for each nerve and stimulation level were then masked (3dAutomask), and activation was determined by an F test (3dDeconvolve) with a P-value threshold of 0.005 using AFNI. The fMRI signal was modeled using a boxcar function as a regressor. Voxels were classified as active if the F statistic was above threshold. The color coding of the voxels was determined by the amplitude of the fit coefficient, with blues signifying negative values and red/orange signifying positive values. Regions of interest (ROIs) were drawn by consulting the activation map and cross-referencing this with the Paxinos and Watson atlas (5). ROIs for the S1FL, S2, sensory motor thalamus (SMT), and Motor 1/Motor 2 (M1/M2) were used to plot time-course data. Only voxels that were considered above statistical threshold were included. Voxels were selected from the center of the ROIs and their time courses are plotted in Fig. 3.

Figure 3.

Representative single voxel time courses in response to direct nerve stimulation. No signal filtering, signal averaging, or baseline correction algorithms were applied. In column a response in S1FL to excitation of four different nerves is shown; in column b the excitation protocol applied to the median nerve is varied and S1FL response is displayed; and in column c the excitation protocol applied to the median nerve is held constant and responses in different brain regions are displayed.


The physiological parameters of mean arterial pressure (MAP), heart rate, and blood oxygen saturation (SpO2) were recorded before, during, and after direct ulnar nerve stimulation (Table 2). Because none of these parameters were significantly different before, during, or after either of the two stimulation currents or two frequencies, they were averaged together. These values are: MAP = 136 ± 17 mmHg; heart rate = 289 ± 6 beats/min; and SpO2 = 98 ± 1%. Similarly, the arterial blood gases did not show any statistically significant changes over the course of the experiment. Averages were: pCO2 = 38 ± 4; pO2 = 156 ± 26; pH = 7.40 ± 0.05. At the end of the experiment the stimulation current was briefly increased to 1.5 mA (two rats) or 2.0 mA (three rats) at 10 Hz. An increase of MAP of ≈11 mmHg was observed.

Table 2. Systemic Circulatory Parameters during Ulnar Stimulation
mAHzMAP (mmHg)SPO2 (%)Heart Rate (beat/min)
  1. Values are mean ± SD.

0.55125 ± 14125 ± 15126 ± 1499 ± 199 ± 1289 ± 27294 ± 34
0.510128 ± 19130 ± 21129 ± 2098 ± 198 ± 1283 ± 21293 ± 28
1.05136 ± 14138 ± 17138 ± 1598 ± 198 ± 2282 ± 23294 ± 29
1.010139 ± 16139 ± 18140 ± 1498 ± 298 ± 1292 ± 37299 ± 46

The left forepaw stimulation was included in the protocol for every rat to validate cortical activation in each experiment. Figure 2b shows a representative blood oxygenation level-dependent (BOLD) time course from a single voxel in response to forepaw stimulation. No signal averaging, filtering, or baseline correction algorithms were applied. Figure 2c shows BOLD activation maps of forepaw response across the 10 slices. Note that the activation is observed contralateral to the stimulated forepaw. The BOLD response is seen in slices 1 through 4, with the most intense activation in slices 2 and 3, as depicted by the ovals. This correlates well with the forepaw region (S1FL) as described in the Paxinos and Watson atlas (5).

Figure 2.

a: Midline sagittal view of the rat brain with the 10 slices acquired with each acquisition. Slice 3 is just anterior to the anterior commissure of the rat (−0.36 mm). b: A representative left forepaw voxel time course in S1FL (2 mA, 10 Hz, 3 ms). c: BOLD maps of forepaw response across the 10 slices.

Figure 3 shows representative BOLD time courses from single voxels in response to various direct nerve stimulation protocols. In column a, the time-course responses in S1FL to stimulation of each of the four nerves investigated in this study are shown for the same stimulation protocol. Columns b and c show the time-course responses to variation of other parameters when using median nerve stimulation. In column b the S1FL time-course response is plotted across the four stimulation protocols, and in column c time courses from four selected brain regions are shown while holding the stimulation protocol fixed. These time courses can be compared with the forepaw time course of Fig. 2b. Figure 3 establishes that response to direct nerve stimulation is robust and generally similar to that seen in S1FL in response to forepaw stimulation.

Figure 4 shows the BOLD activation map for slice 5 when the radial nerve is stimulated (10 Hz, 1.0 mA). Ovals have been drawn around ROIs in the functional activation map to delineate the functional anatomy. Clearly shown are the S1Tr (somatosensory S1 upper arm/trunk region), M1/M2 (motor regions), S2 (sensory region), and the thalamus. A negative BOLD response is observed in the caudate putamen region. Figure 4 serves as a reference for discrimination of structures in the Results and Discussion sections of this article.

Figure 4.

Cortical response in slice 5 to direct radial nerve stimulation (1.0 mA, 10 Hz, 1 ms). Two sensory regions are delineated: the S1 trunk region (S1Tr) and the secondary somatosensory region (S2). Both the primary and secondary motor regions (M1/M2) are active, as well as the thalamus. The negative BOLD response overlies the caudate putamen region of the brain.

Ulnar nerve stimulation indicates BOLD responses in many regions of the brain (Fig. 5). Activation in the thalamic and mediofrontal cortical regions attenuates when the stimulation frequency is increased from 5 Hz to 10 Hz at a constant current (0.5 mA, Fig. 5a,b). Comparing Fig. 5a,b (slices 3 and 4), we see that the sensory region becomes more pronounced with increasing frequency (5 Hz to 10 Hz). The activated region can be localized to the stereotaxic S1FL region and overlaps with the region activated by forepaw stimulation as seen in Fig. 2b, slices 3 and 4. Note that the forepaw signal in Fig. 2b is found contralateral to the sensory region activated by ulnar nerve stimulation because the left forepaw and right ulnar nerve were stimulated. At lower current levels (0.5 mA, Fig. 5a,b), most of the activations are seen in both hemispheres, and these activated areas greatly expand at the higher stimulation current (1.0 mA, Fig. 5c,d). Also at the higher stimulation current, activation is observed in the secondary somatosensory region, S2 (Fig. 5c,d, slices 1–6). A robust negative BOLD response is observed in a region corresponding to the caudate putamen region (Fig. 5a,b, slices 1–5). The negative BOLD response disappears at the higher stimulation current (1.0 mA, Fig. 5c,d), but this is replaced with a smaller somatosensory negative BOLD response bilaterally.

Figure 5.

a: Cortical response to direct ulnar nerve stimulation (0.5 mA, 5 Hz, 1 ms). b: Similar to a except with a higher frequency of stimulation (0.5 mA, 10 Hz, 1 ms). c: Similar to a except with a higher current (1.0 mA, 5 Hz, 1 ms). d: Similar to a except at a higher frequency and current (1.0 mA, 10 Hz, 1 ms).

Median nerve stimulation (Fig. 6) produces a similar pattern to ulnar nerve stimulation. In addition, median nerve stimulation results in activation in the motor cortex (Fig. 6a, slice 4) at the lowest levels of stimulation. Comparing Fig. 6a,b, we see an attenuation of the activation in the motor and thalamic regions at the higher frequency of 10 Hz. The activation seen in the sensory region at the lowest level of median nerve stimulation (Fig. 6a, slices 2–4) corresponds to S1FL and overlaps with the forepaw sensory region (Fig. 2b). The region of sensory activation is increased in size at the higher stimulation frequency (compare Fig. 6a,b, slices 2–4). Activation in most regions is enhanced at the higher stimulation current (0.5 mA vs. 1.0 mA, Fig. 6c,d). Similar to the ulnar nerve, there is some activation in the secondary somatosensory region (S2) that is most pronounced at the higher stimulation current (Fig. 6a–d, slices 5 and 6). Note that this sensory activation is below threshold in Fig. 6b, slice 6. A negative BOLD response in the caudate putamen region is observed with median nerve stimulation, but the area of activation is reduced compared to that of ulnar nerve stimulation.

Figure 6.

a: Cortical response to direct median nerve stimulation (0.5 mA, 5 Hz, 1 ms). b: Similar to a except with a higher frequency of stimulation (0.5 mA, 10 Hz, 1 ms). c: Similar to a except with a higher current (1.0 mA, 5 Hz, 1 ms). d: Similar to a except at a higher frequency and current (1.0 mA, 10 Hz, 1 ms).

Stimulation of the radial nerve produces the cortical activation pattern described in Fig. 7. As with ulnar and median nerve stimulation, radial nerve stimulation produces activation in mediofrontal cortical regions at low frequency and current (0.5 mA, 5 Hz, Fig. 7a). This activation is attenuated with an increased frequency (0.5 mA, 10 Hz, Fig. 7b) but increases considerably at the higher stimulation current (1.0 mA, Fig. 7c,d). Activation is seen in the somatosensory region at low stimulation frequency and current (Fig. 7a, slices 3–6). This activation increases in intensity and is observed in additional slices at the higher stimulation frequency and current (Fig. 7b–d, slices 3–9). The negative BOLD response is more pronounced in the caudate putamen region with radial nerve stimulation (Fig. 7, slices 1–5) compared to negative BOLD response observed with ulnar and median nerve stimulation. Some additional lateral somatosensory negative BOLD responses are present bilaterally.

Figure 7.

a: Cortical response to direct radial nerve stimulation (0.5 mA, 5 Hz, 1 ms). b: Similar to a except with a higher frequency of stimulation (0.5 mA, 10 Hz, 1 ms). c: Similar to a except with a higher current (1.0 mA, 5 Hz, 1 ms). d: Similar to a except at a higher frequency and current (1.0 mA, 10 Hz, 1 ms).

Figure 8.

a: Cortical response to direct musculocutaneous nerve stimulation (0.5 mA, 5 Hz, 1 ms). b: Similar to a except with a higher frequency of stimulation (0.5 mA, 10 Hz, 1 ms). c: Similar to a except with a higher current (1.0 mA, 5 Hz, 1 ms). d: Similar to a except at a higher frequency and current (1.0 mA, 10 Hz, 1 ms).

Musculocutaneous nerve stimulation (Fig. 8) shows a different pattern of activation compared to ulnar, median, and radial nerve stimulation. Activation in the mediofrontal cortical regions is present at the lowest stimulation frequency and current, and increases at the higher stimulation current (Fig. 8a–d). However, there is some attenuation of the activation in this region with the higher frequency (10 Hz vs. 5 Hz) at the high stimulation current (1.0 mA, Fig. 8c,d). Activation in the somatosensory region is not observed at the low current stimulation (0.5 mA, Fig. 8a,b). At the higher current (1.0 mA), activation in the somatosensory region is observed (Fig. 8c, slices 2–4). This activated region overlaps with S1FL. Activation in the thalamic region is present at the low stimulation frequency and current, and is more pronounced at the higher stimulation level and frequency (Fig. 8a–d). There is a similar attenuation of thalamic activation with higher frequency (10 Hz vs. 5 Hz) at the high stimulation current (1.0 mA, Fig. 8c,d). A robust negative BOLD response in the caudate putamen region is observed at the lower stimulation current (0.5 mA, Fig. 8a,b, slices 1–5). This negative BOLD response is attenuated at the higher stimulation current (1.0 mA, Fig. 8c,d).


Many of the studies relating to definition of the sensory or motor regions of the rat brain have used electrophysiological mapping methods (27, 28). To our knowledge, the use of direct nerve stimulation in the upper extremity of the rat has not previously been reported. Chang and Shyu (29) described the use of direct nerve stimulation of the sciatic nerve to study cortical response to the delivery of noxious and nonnoxious stimuli in alphachloralose-anesthetized rats. They found that nociceptive A-δ and C fibers were recruited when the sciatic nerve was stimulated at 10 and 20 (1.6–3.2 mA) times muscle twitch threshold (0.16 mA). This also confirms some of our own findings where current levels of 1.5 mA to 2.0 mA elicited increases in physiologic parameters such as blood pressure and pulse. In our hospital, current intensities of 1 mA to 10 mA and variable frequencies of 2 Hz to 20 Hz are used in the intraoperative setting to evaluate direct nerve or somatosensory evoked potentials. Collectively, these served as guidelines to select our current and frequency levels. Our forepaw stimulation parameters (2.0 mA, 10 Hz, 3 ms) provided robust reproducible somatosensory activation and served mainly as a validation that our experimental conditions were suitable for BOLD fMRI activation. We reduced the current (0.5 mA to 1.0 mA) and duration to 1 ms for direct nerve stimulation since we were still able to elicit BOLD activation. Additional experiments examining the BOLD response to lower durations direct nerve stimulation would be beneficial to examine how much duration contributes to widespread intensity, particularly at the higher current and frequency.

The response to direct nerve stimulation shows a distinct pattern of cerebral activation for each nerve and stimulation parameter. Although the patterns are distinct, there are some overlapping regions of activation. For example, the sensory component for each nerve overlaps to varying degrees with the forepaw region (S1FL) of the somatosensory cortex. This concurs with the anatomical distribution of the nerves because each nerve provides innervation to the different parts of the forepaw. There is also considerable overlap of the motor regions of the brain, which concurs with the anatomical distribution of the nerves to different muscles that have some overlap in function. Several muscles, for instance, may contribute to wrist flexion or extension. These overlapping muscle representations may be a consequence of two different nerves that contribute to innervation. One interesting observation is that at lower levels of stimulation there are more distinct areas of cortical activation.

In response to higher current levels, in particular, there are much higher levels of activation seen in varying regions of the brain. The higher levels of stimulation may indicate additional responses that include multiple sensory, motor, and proprioception thalamic inputs. When the threshold P-value is set at a higher level using AFNI, cortical activation in areas such as the thalamus are more easily defined (data not reported). Direct nerve stimulation to mixed motor and sensory nerves may provide the brain with combined motor-sensory inputs that would not be observed with a pure forepaw stimulation model. In addition, there is bilateral activation in thalamic and motor regions that becomes more pronounced at the higher current and frequency, while the somatosensory regions S1FL, S1Tr, and S2 tend to favor the contralateral activation. When a pure sensory nerve to the rat upper arm (i.e., lateral cutaneous nerve to the brachium) was directly stimulated, a more focal cortical response in the sensory region was observed in only the contralateral somatosensory region with respect to the stimulated side (data not reported). Thus, it appears that stimulating a motor and sensory mixed nerve enables bilateral thalamic activation that extends to corresponding motor regions associated with each nerve. The functional neuroconnectivity within and across hemispheres in the motor and sensory networks appears extensive.

Direct nerve stimulation strongly projects the areas involved in the neuronal network. Previous forepaw stimulation studies by Keilholz et al. (3, 4) have shown small regions of activation in the thalamus. In the current set of results, the thalamus is delineated by BOLD activation in slices 4 through 9 in multiple stimulation levels of the different nerves. Cutaneous electrodes stimulate the somatosensory nerve fibers orthodromically and the motor nerve fibers antidromically, elucidating the underlying network.

The spinothalamic cells in the spinal cord have variable physiologic response properties. Low-threshold units react to mechanical stimuli; wide dynamic range units react to both light and high-intensity stimuli; and high-threshold units respond to nociceptive stimuli. Stimulation of the peripheral nerve at increasing intensities activates these various cells in the above order. This activation is transmitted through the spinothalamic tract to the contralateral thalamus, but also through the spinoreticular cells and tracts to both thalami and the somatosensory areas in both hemispheres. Most of the spinoreticular fibers terminate in the intralaminar thalamic nuclei, where they project to wide areas in both hemispheres. This is likely the explanation for the more widespread pattern of signal observed in our experiments. The activation of the primary somatosensory cortex is also associated with simultaneous activation of the motor cortex, and rapid spread to the secondary (associative) sensory areas, also resulting in generalized activation. Another possible mechanism is that antidromic stimulation of motor nerve fibers activates spinothalamic and spinoreticular cells in the dorsomedial ventral gray matter of the spinal cord (layers V through VII) without being necessarily nociceptive. This may explain the bilateral BOLD signal even at low-intensity stimulation.

An interesting result from this work is the negative BOLD response located especially in the caudate putamen region. A search of the literature turned up no electric stimulation studies in rats with negative or positive BOLD signals in this region. However, a negative BOLD response in the caudate putamen region was evident in several pharmacological fMRI studies in rats (30, 31). The caudate putamen is involved in both the somatosensory and motor neural network of the rat (5). These regions have direct connections to the motor cortex, the intralaminar thalamic nuclei, or both, and can be activated or inhibited after they receive the nociceptive impulses. Perhaps antidromic stimulation of motor nerve fibers is causing this response. An analogous cerebral blood volume (CBV) experiment using monocrystalline iron oxide nanoparticle (MION) (32) and direct stimulation of the musculocutaneous nerve reversed the sign of the fMRI response in both the caudate putamen (positive) and somatosensory (negative) regions (unreported data). This may indicate that a decrease in CBV occurs in the caudate putamen region with a concomitant increase in CBV in the somatosensory system during direct nerve stimulation. Further studies of the caudate putamen region are ongoing to elucidate the mechanism of this finding.

There are some limitations to the results. In almost every experiment, there was cortical activity seen in the first (slice 0) and last slice (slice 9) of the acquired data. This would suggest that there are other regions of the brain—for example, the cerebellum—that were not scanned and may be involved in the motor and sensory pathways (33). Superior midline activation in slices 0 through 2 may have some draining vein contribution from the superior sagittal sinus. However, the anticipated unilateral activation in M1/M2 on the contralateral side to stimulation argues that a significant portion is coming from the hyperemic response in the motor regions.

The brachial plexus is one of the most complex neuronal regions of the body. Detailed knowledge of the anatomy and function of each nerve is important to understand the clinical implications when these nerves are injured. This study required a coordinated effort between clinicians who actively treat patients with brachial plexus and peripheral nerve injuries and basic scientists who have extensive experience in rodent fMRI. The main goal of this multidisciplinary effort is to develop a rodent fMRI model that can be leveraged in the future to aid in the evaluation of new techniques in the diagnosis and treatment of peripheral nerve injuries.


This study describes a novel approach for mapping the motor and sensory pathways in the rat brain by directly stimulating the nerves of the rat upper extremity. Additional studies will be aimed at directly stimulating smaller branches of the major nerves to refine the motor and sensory pathways and provide a greater level of detail of the motor and sensory ratunculi of the upper extremity. This goal of developing a rodent fMRI model for brachial plexus and peripheral nerve injuries appears within reach.


The authors thank B. Douglas Ward for helpful comments and analysis of the data.