Evidence for Increased Activation of Persistent Inward Currents in Individuals With Chronic Hemiparetic Stroke

Abstract

Despite the prevalence of hyperactive stretch reflexes in the paretic limbs of individuals with chronic hemiparetic stroke, the fundamental pathophysiological mechanisms responsible for their expression remain poorly understood. This study tests whether the manifestation of hyperactive stretch reflexes following stroke is related to the development of persistent inward currents (PICs) leading to hyperexcitability of motoneurons innervating the paretic limbs. Because repetitive volleys of 1a afferent feedback can elicit PICs, this investigation assessed motoneuronal excitability by evoking the tonic vibration reflex (TVR) of the biceps muscle in 10 awake individuals with chronic hemiparetic stroke and measuring the joint torque and electromyographic (EMG) responses of the upper limbs. Elbow joint torque and the EMG activity of biceps, brachioradialis, and the long and lateral heads of triceps brachii were recorded during 8 s of 112-Hz biceps vibration (evoking the TVR) and for 5 s after cessation of stimulation. Repeated-measures ANOVA tests revealed significantly (P ≤ 0.05) greater increases in elbow flexion torque and EMG activity in the paretic as compared with the nonparetic limbs, both during and up to 5 s following biceps vibration. The finding of these augmentations exclusively in the paretic limb suggests that contralesional motoneurons may become hyperexcitable and readily invoke PICs following stroke. An enhanced tendency to evoke PICs may be due to an increased subthreshold depolarization of motoneurons, an increased monoaminergic input from the brain stem, or both.

INTRODUCTION

A central abnormality following hemiparetic stroke is the emergence of the hyperactive stretch reflex in the paretic limbs. Although the exact pathophysiological mechanisms underlying the expression of this motor impairment have yet to be determined, hyperactive stretch reflexes may be due to an increased reliance on monoaminergic bulbospinal pathways following stroke-induced losses of corticospinal and corticobulbar projections (Cabaj et al. 2006; Dewald and Beer 2001; Dewald et al. 1995, 2001; Ellis et al. 2007; Ono and Fukuda 1995; Sukal et al. 2007). Metabotropic monoaminergic inputs from the dorsal reticulospinal system have much stronger effects on motoneuron excitability than do ionotropic inputs from the vestibulospinal system or the remaining reticulospinal pathways, which can alter only the recruitment threshold of motoneurons (Heckman et al. 2005). Specifically, monoaminergic input to the cord provides an exceptionally powerful neuromodulatory control of motoneurons, greatly enhancing excitability by hyperpolarizing the spike threshold, depolarizing the resting membrane potential, reducing the spike afterhyperpolarization, and facilitating persistent inward currents (PICs) (Fedirchuk and Dai 2004; Heckman et al. 2003, 2008; Hultborn et al. 2004; Powers and Binder 2001).

The PIC is an especially striking manifestation of monoaminergic actions because it markedly amplifies and prolongs the effects of synaptic input (Hounsgaard et al. 1988; Lee and Heckman 2000). Amplification of motoneuronal responses to excitatory postsynaptic potentials has been studied primarily in animal preparations using a steady synaptic input mediated by tendon vibration, which selectively activates muscle spindle Ia afferents (Matthews 1972). When monoaminergic drive is strong, PICs amplify Ia input as much as fivefold (Hultborn et al. 2003; Lee and Heckman 2000). In low-threshold motoneurons, this amplification is accompanied by self-sustained firing (SSF) that persists after the input stimulus has stopped (Lee and Heckman 1998a,b). This phenomenon was initially referred to as “bistable” behavior in the classic experiments that identified the potent effects of monoamines on motoneuron firing (Crone et al. 1988; Hounsgaard et al. 1988). Tendon vibration has also been applied in human subjects (Eklund and Hagbarth 1966, 1967; Hagbarth and Eklund 1966, 1968b), and although it is not as selective for Ia afferents as in animal preparations, the tonic vibration reflex has been shown to evoke SSF in motor unit firing patterns of human subjects (Gorassini et al. 2002; Kiehn and Eken 1997; Walton et al. 2002).

If monoaminergic bulbospinal input to the spinal cord is enhanced following stroke, then its potent effects may play a major role in generating motoneuron hyperexcitability. The present investigation sought to test this hypothesis by quantifying the elbow joint torque and electromyographic (EMG) activity induced by the tonic vibration reflex (TVR) in the paretic and nonparetic upper limbs of 10 participants with chronic hemiparetic stroke. The results of this study show an increased TVR response as well as evidence of sustained EMG and torque generation after vibration cessation, both of which are consistent with increased PICs in the paretic upper limb of chronic stroke survivors.

METHODS

Study participants

A departmental registry was used to recruit individuals with hemiparetic stroke for participation in the study. Ten participants (three females, seven males; mean age, 59 ± 10 yr) fulfilled all criteria for participation in the investigation and subsequently completed the experimental protocol. All participants sustained a cortical or subcortical stroke (non–brain stem) at least 1 yr prior to enrollment in the study (range, 40–124 mo poststroke). Upper limb Fugl-Meyer Motor Assessment (FMA) scores for all subjects ranged from 10 to 37 (of a possible 66), representing severe to moderate impairment (Fugl-Meyer et al. 1975).

For inclusion in the study, passive range of motion for the paretic limb was ≥90° each of shoulder flexion, shoulder abduction, and elbow flexion to interface comfortably with the experimental equipment. Additionally, participants were required to demonstrate some volitional control in elbow flexion and extension for the purposes of obtaining maximum voluntary torques (MVTs) as part of the experimental protocol. Overpressure at the end of the range of motion was used as a medical screening tool to verify the absence of an inflammatory condition at the shoulder, elbow, wrist, and fingers. Subjects were excluded from the study if they had minimal (50–66 on the FMA) or very severe (0–9 on the FMA) impairment, significant impairment of upper extremity tactile sensation and proprioception as determined by tactile localization and detection of movement tasks (O'Sullivan and Schmitz 2001), or difficulty sitting for extended periods of time. Recent changes in the medical management of hypertension or current use of antispastic medications were also grounds for exclusion from this study.

All subjects provided informed consent to participate in the study, which was approved by the Institutional Review Board of Northwestern University in accordance with the ethical standards stipulated by the 1964 Declaration of Helsinki for research involving human subjects.

Experimental setup

Participants were seated in a Biodex experimental chair (Biodex Medical Systems, Shirley, NY) with their trunk immobilized by seat belts. Each subject's forearm, wrist, and hand were placed in a rigid fiberglass cast to prevent unwarranted motion during experimentation and to provide an interface for the direct coupling of the arm to a 6 degrees-of-freedom load cell (Model 45E15A; JR3, Woodland, CA). The load cell was connected to the cast at the level of the subject's wrist and was positioned such that the limb retained the following configuration: 75° of shoulder abduction, 40° of shoulder flexion (horizontal adduction from the frontal plane), and 90° of elbow flexion. The test apparatus supported the full weight of the limb in this posture throughout all experimental protocols.

Experimental protocol

While seated in the experimental apparatus described previously, participants were instructed to generate maximum voluntary elbow flexion and extension torques (MVT) while watching visual torque feedback on a computer monitor. To ensure that an accurate representation of maximum torque was determined for each direction, subjects repeated the MVT trials until they obtained three maximum torque values within 10% of one another without the last repetition being the greatest. Enthusiastic verbal encouragement was consistently given during the performance of MVTs.

Posture was not altered for the TVR response phase of the protocol, although participants were instructed to remain still and relaxed during all trials. The load cell was recalibrated once a vibrator (frequency: 112 Hz, Model 91, Daito-Thrive, Showa-cho, Japan) was placed medially over the distal muscle belly of the biceps brachii. Each TVR trial lasted for 15 s and began when the arm was fully relaxed; the onset of data collection began once the experimenter determined that the participant was fully relaxed. Vibration of the biceps was initiated and terminated by a TTL (transistor–transistor logic) pulse and began 2 s after the onset of data collection and continued for 8 s; at 10 s after the start of a trial the vibratory stimulus was stopped and data collection continued for an additional 5 s. Vibration evoked a net elbow flexion torque in all participants. Following each trial, participants were given a 15 to 20-s rest period to allow the limb time to relax before initiation of the next trial. The experimenter viewed a real-time display of the trial to verify its acceptability for retention and analysis. Trials were not retained if activation of biceps brachii, brachioradialis, or triceps brachii occurred prior to vibration onset, if postural or unconstrained extremity movements occurred during the trial and were associated with mechanical artifacts on the real-time display, or if subjects closed their eyes and appeared to fall asleep during the trial. Subjects were routinely asked to produce antagonist activity (extension) to facilitate the achievement of quiescent flexor activity prior to trial onset. Approximately 25–30 TVR response trials were collected per experimental session. The collection of MVT and TVR response data was repeated within 1 wk of the initial testing of both limbs.

Data collection

Data collected from the load cell included orthogonal forces and moments generated at the wrist in the x, y, and z planes. Active differential electrodes with 1-cm interelectrode distance (16-channel Bagnoli EMG System; Delsys, Boston, MA) were placed on the skin surface of both arms and recorded the EMG activity of biceps brachii (BIC), brachioradialis (BRD), and the lateral (TRI LAT) and long (TRI LONG) heads of the triceps brachii during all trials.

All data were digitized at 1 kHz. Initial conditioning of raw EMG signals was provided by onboard 1,000× preamplification and first-order high-pass filtering with a cutoff frequency of 20 Hz (16-channel Bagnoli EMG System). An eighth-order low-pass Butterworth filter (cutoff frequency: 500 Hz) was used for subsequent conditioning of EMG data (Model 9016; Frequency Devices, Havelhill, MA).

Data analysis

Custom Matlab software (The MathWorks, Natick, MA) used a Jacobian-based algorithm to convert the collected load cell data into elbow flexion and extension torque curves. All torque curves were filtered with a digital eighth-order low-pass Butterworth filter with cutoff frequency of 50 Hz and then normalized by the appropriate MVT.

Window averages of individual trials (duration: 500 ms) were computed both during vibration (V-ON) and after vibration (V-OFF). Vibration-on torque windows were taken from 9.50 to 10.00 s to ensure that reflex-evoked EMG and torque activity were maximal, whereas V-OFF torque windows spanned from 14.50 to 15.00 s (Fig. 1). The V-OFF range was selected because it represented the last 500 ms window over which data could be analyzed, in accordance with the duration of each trial.

FIG. 1.

Depiction of experimental protocol and window averages. For tonic vibration reflex (TVR) response trials, vibration commenced 2 s after data collection began and lasted for 8 s. At 10 s following the start of a trial, vibration was extinguished and data collection continued for an additional 5 s. Window averages of trials were computed at 9.50–10.00 s (“V-ON”) and 14.50–15.00 s (“V-OFF”) for torque data, and 9.45–9.95 s and 14.45–14.95 s for EMG data. Curves represent ensemble average torque of 22 trials in the paretic (top, light gray) and nonparetic (bottom, dark gray) arms of one subject, bounded by ±2SE.

After full-wave rectification, EMG data were smoothed using a causal one-sided moving average filter (window duration: 250 ms) and normalized by the appropriate maximum EMG signal. EMG data were also baseline corrected by subtracting from all of a trial's data the average EMG activation level between 250 and 500 ms; the first 250 ms of data were not used because of artifacts associated with building the moving filter.

Window averages of EMG data were also computed both during and after vibration. To account for the electromechanical delay inherent to skeletal muscle however, the EMG windows preceded the torque windows by 50 ms; V-ON EMG windows spanned from 9.45 to 9.95 s, whereas the V-OFF EMG windows were taken from 14.45 to 14.95 s after the onset of data collection (see Fig. 1 for conceptual reference).

Statistical analysis

Prior to beginning the statistical analysis of limb behavior, custom Matlab software based on the Shapiro–Wilkes test verified the normality of all data. Statistical analysis was performed with DataDesk software (Data Description, Ithaca, NY). For all statistical measures, a P value of ≤0.05 was considered to be significant. Twenty-two trials were used for all statistical analyses because this value represented the largest number of accepted trials common to all subjects across all experimental sessions.

The effect of limb on V-ON and V-OFF torque and EMG was assessed using a two-factor repeated-measures ANOVA; independent factors for this test included limb (paretic or nonparetic) and trial number (repeated measure) and the dependent variable was elbow flexion torque or EMG activation level. The effect of repeated experimental sessions on paretic limb V-ON and V-OFF torque and EMG activation was also assessed using a two-factor repeated-measures ANOVA; independent factors included trial (repeated measure) and session and the dependent factor was elbow flexion torque or EMG activation. Additionally, a one-factor repeated-measures ANOVA design was used to test for an effect of repeated trials on torque or EMG behavior in the affected limb; the independent factor for this test was trial number (repeated measure) and the dependent variable was elbow flexion torque or EMG activation level.

RESULTS

The elbow flexion torque and EMG responses of four elbow muscles were measured in both arms of 10 stroke survivors during and after the application of a vibratory stimulus to the distal muscle belly of biceps brachii. Across subjects the maximum voluntary elbow flexion torque for the paretic limb averaged 36.44 Nm, whereas the nonparetic elbow flexion MVT averaged 62.09 Nm, indicating that the paretic arm retained slightly greater than half of the flexion strength of the nonparetic limb (58.7%). As a consequence of this strength disparity, vibration-evoked elbow flexion torque was analyzed as both a raw and a normalized quantity. Representative elbow flexion torque, flexor EMG, and extensor EMG traces of single trials and ensemble averages can be seen in Fig. 2. Importantly, statistical analysis revealed no significant effects of repeated sessions or trials (P ≤ 0.05) in the results presented herein.

FIG. 2.

Representative elbow flexion torque and electromyographic (EMG) data. The left column represents elbow flexion torque, biceps brachii (BIC) EMG, and lateral head of triceps brachii (TRI LAT) EMG of individual TVR-response trials for a single subject. Right column represents ensemble average data for one subject with 22 trials constituting each curve. Positive torque deflections represent elbow flexion. Vibration commences at 2 s and terminates at 10 s; gray lines represent the paretic limb; black traces represent the nonparetic limb. Data have been filtered as described in methods for clarity of presentation. x-axis: time (ms); y-axis: normalized torque or EMG activation (×100 for % MVT).

During vibration

Significant differences were noted in the elbow flexion torque, brachioradialis (BRD) EMG, triceps lateral head (TRI LAT) EMG, and triceps long head (TRI LONG) EMG elicited from the paretic compared with the nonparetic limb (P ≤ 0.05, Figs. 3 and 4). These differences were observed in both raw and normalized data. The average raw elbow flexion torque generated across subjects during vibration was 1.4 Nm in the paretic limb, whereas the average raw elbow flexion torque generated during the same interval in the nonparetic limb was 0.4 Nm (Fig. 4). When normalized to MVT, these values translated into elbow flexion torques of 4.5% MVT in the paretic limb and 0.7% MVT on the nonparetic side.

FIG. 3.

Normalized ensemble average activation values for the V-ON and V-OFF windows. V-ON: 9.5.0–10.00 s (torque), 9.45–9.95 s (EMG); V-OFF: 14.50–15.00 s (torque), 14.45–14.95 s (EMG). BIC: biceps brachii; BRD: brachioradialis; TRI LAT: lateral head of triceps brachii; TRI LONG: long head of triceps brachii. Gray: paretic limb; black: nonparetic limb.

FIG. 4.

Unnormalized (raw) ensemble average elbow flexion torque values for V-ON and V-OFF windows. Gray: paretic limb; black: nonparetic limb. V-ON window: 9.5–10.0 s; V-OFF window: 14.5–15.0 s. Top: V-ON window. Bottom: V-OFF window. Each bar represents average raw elbow flexion torque across 10 participants, with 22 trials each. Statistically significant increases in raw torque production can be seen both during and following vibration in the paretic limb relative to the nonparetic limb.

The paretic biceps (BIC) was active at 17.0% MVT during vibration and the paretic BRD was active at 13.0% MVT; nonparetic BIC and BRD were active at 7.0 and 5.0% of MVT during vibration, respectively. Elbow extensors were also active during vibration, with TRI LAT and TRI LONG achieving 13.0 and 9.0% of MVT, respectively, in the paretic limb and both muscles active at 3.0% of MVT on the nonparetic side.

Interestingly, despite having the greatest mean activation of any muscle sampled, the paretic BIC activation level was not significantly greater than that of the nonparetic BIC during vibration (P = 0.1172). Further analysis revealed however that the lack of significance was attributable to a single subject who exhibited an unusually high degree of V-ON BIC activation, particularly in the nonparetic limb (paretic BIC = 45.7% MVT; nonparetic BIC = 42.4% MVT). Although removing this subject from the data set resulted in a mean paretic BIC activation across all remaining subjects of 14.3% MVT, the absence of this subject reduced the mean nonparetic BIC activation level to 4.0% MVT. Removal of this outlier resulted in a statistically significant difference in BIC activation across the remaining subjects (P ≤ 0.05). As expected, paretic BIC continued to display the greatest activation among sampled muscles, even in the absence of the outlier.

Postvibration

Following the cessation of vibration, the paretic limb in all cases displayed a significantly enhanced activation over the nonparetic limb (P ≤ 0.05, Figs. 3 and 4). Raw elbow flexion torque in the paretic limb following vibration averaged 0.23 Nm, whereas raw elbow flexion torque in the nonparetic limb averaged 0.05 Nm (Fig. 4). Normalized elbow flexion torque in the paretic limb was 0.7% MVT and the average normalized nonparetic elbow flexion torque was 0.08% MVT. The paretic BIC exhibited the greatest residual activity of all sampled muscles, with a magnitude of 1.0% MVT. By comparison, the nonparetic BIC was active only at 0.08% MVT. The paretic BRD was active at 0.2% MVT, with nonparetic BRD active at 0.03% MVT. The lateral and long heads of the paretic triceps were both active at 0.2% MVT, with nonparetic TRI LAT and TRI LONG maintaining activation levels of 0.01 and 0.03% MVT, respectively.

DISCUSSION

This study used the tonic vibration reflex (TVR) response of the paretic and nonparetic arms of individuals with chronic hemiparetic stroke as a probe to investigate the presence of motoneuronal hyperexcitability and TVR-induced sustained torque and EMG activation consistent with PIC-mediated SSF of motoneurons. The presence of significantly augmented elbow flexion torque and EMG activity in the paretic compared with the nonparetic upper limb during vibratory input to the distal muscle belly of the biceps brachii suggests that contralesional spinal alpha motoneurons may become hyperexcitable and invoke PICs following stroke. Additionally, the presence of stable EMG and torque activity persisting for at least 5 s following cessation of the vibratory stimulus is consistent with PICs and SSF in motoneurons innervating the paretic upper limb.

Animal preparations

Although existing animal preparations for investigating intrinsic motoneuron properties are not intended to simulate human stroke, the loss of descending corticofugal projections and the increased bulbospinal monoaminergic drive exhibited by the decerebrate cat parallel many of the physiological changes that arise following unilateral brain lesion in humans (Dewald and Beer 2001; Dewald et al. 1995, 2001; Ellis et al. 2007; Leblond et al. 2000; Sukal et al. 2007). In the reduced state, the monoamines norepinephrine (NE) and serotonin (5-HT) profoundly increase spinal motoneuronal excitability by acting at postsynaptic metabotropic NE alpha1 and 5-HT₂ receptors to modulate indirectly the state of voltage-gated Ca_v1.3 channels located on the dendrites of alpha motoneurons (Harvey et al. 2006; Heckman et al. 2003; Hounsgaard et al. 1988; Li et al. 2007; Lipscombe et al. 2004; Simon et al. 2003). Once activated, these channels are capable of generating persistent inward currents (PICs) characterized by a noninactivating depolarizing inward ion flux (Schwindt and Crill 1980). Increasing concentrations of monoamines lead to robust increases in PIC magnitude, and if monoaminergic drive is sufficiently high, dendritic PICs may result in plateau potentials that facilitate motoneuronal hyperexcitability with self-sustained firing, which can serve as a quantifiable marker for the presence of increased spinal concentrations of monoamines (Hounsgaard et al. 1988; Kiehn et al. 1996).

Although PICs and SSF can exist in the nonpathological state, it should be reiterated that they are virtually absent when the motoneuron is deprived acutely of monoamines and conversely are overexpressed in the presence of increased monoaminergic concentrations (Hounsgaard et al. 1988). Of great significance for the pathogenesis of hyperactive stretch reflexes following stroke, these properties serve to heighten the sensitivity of motoneurons to ionotropic inputs, thereby augmenting cellular responses to motor commands and afferent input.

Human investigations

Shortly after its initial characterization in healthy human adults (Eklund and Hagbarth 1966; Lang and Vallbo 1967), the TVR and electrical stimulation of 1a afferents began to be used as tools to investigate the role of afferent feedback in neurological movement disorders (Eklund and Hagbarth 1967; Hagbarth and Eklund 1966, 1968a,b; Lance et al. 1966). Results of these early studies were highly variable however, most likely due to their inclusion of participants with a wide range of impairment etiologies and severities. The physiological underpinnings of the TVR were initially ascribed to posttetanic potentiation, gamma motoneurons, and reverberating spinal circuits (Eklund and Hagbarth 1966, 1967; Lang and Vallbo 1967) and it was not until the work of Crone (Crone et al. 1988) and subsequently Baldissera (Baldissera et al. 1991, 1994) that intrinsic motoneuron properties were used to explain the response of muscle to vibration or electrical stimulation of 1a afferents. Later, following the development of paired motor unit recording techniques by Gorassini, Kiehn, and colleagues (Gorassini et al. 1998; Kiehn and Eken 1997), it became increasingly accepted that PICs underlie the self-sustained firing, muscle contractions, and joint torques that persist following mechanical and electrical stimulation of spindle afferents and motor nerves in humans (Baldwin et al. 2006; Collins et al. 2001, 2002; Gorassini et al. 1998; Kiehn and Eken 1997; Walton et al. 2002). Soon thereafter, the notion that PICs could be readily evoked in control populations became commonplace.

Although investigations of myokymia and spinal cord injury (SCI) in humans clearly suggested that PICs were present during abnormal muscle activation in those populations (Baldissera et al. 1991, 1994; Gorassini et al. 2004; Nickolls et al. 2004; Norton et al. 2008), it remained unknown whether and how the expression of PICs was altered following stroke. The present study found evidence consistent with increased expression of PICs in the paretic upper limb of individuals with chronic hemiparetic stroke, manifested as increased EMG and elbow flexion torque in the paretic limb during vibration and for at least 5 s following its cessation. Importantly, these findings were absent in the nonparetic limb, which exhibited minimal EMG and torque activation during vibration and returned quickly to quiescence following termination of the stimulus.

One explanation for the differential activation pattern of the two arms is that when the nonparetic limb is reflexively activated from rest, its descending bulbospinal monoaminergic drive is not sufficient to induce long-lasting PICs with SSF. Conversely, the presence of significantly augmented TVR responses in the paretic limb—notably with strong postvibration effects—suggests that following loss of corticofugal and corticospinal motor tracts the paretic side may rely more extensively on and/or receive an enhanced bulbospinal monoaminergic drive (Cabaj et al. 2006; Dewald and Beer 2001; Dewald et al. 1995, 2001; Ellis et al. 2007; Sukal et al. 2007). Indeed, individuals with chronic hemiparetic stroke do exhibit abnormal muscle coactivation patterns that are consistent with enhanced bulbospinal involvement (Brunnstrom 1970; Dewald and Beer 2001; Dewald et al. 1995, 2001; Ellis et al. 2007; Sukal et al. 2007). Such an influence of residual monoaminergic reticulospinal projections could increase motoneuronal excitability directly by increasing the spinal concentration of monoamines. In fact, Walton and colleagues (2002) demonstrated similar bulbospinal facilitation in healthy humans by showing that provocation of NE and 5-HT release from the locus coeruleus and raphe nucleus by caffeine administration led to an increased incidence of SSF following vibratory muscle stimulation (Walton et al. 2002).

Given that our study involved activating muscles reflexively from rest rather than activating muscles with a combination of volitional and reflexive means, we do not anticipate large changes in descending monoaminergic drive during the TVR response trials. It seems likely then that the large paretic limb TVR and minimal nonparetic limb TVR may reflect different resting potentials of the respective motoneuron pools, as animal preparations have shown that PIC-mediated SSF occurs only when repeated synaptic inputs are applied to a cell resting near the threshold for sodium spiking (Lee and Heckman 1996). Therefore had the potentials been equal prior to stimulation, the responses of the limbs may have been more similar. Consequently, care should be taken not to interpret these findings as suggesting that PICs cannot be evoked in the nonparetic limb; rather, the results indicate that PICs may be more readily evoked from rest in the paretic than in the nonparetic limb, ostensibly due to a depolarized resting potential that reflects an increased contralesional bulbospinal monoaminergic drive.

Alternatively, it is possible that increased firing of monoaminergic brain stem nuclei or spinal interneurons contributed to the paretic limb's TVR response by enhancing and prolonging motoneuronal firing. Indeed, reticulospinal neurons are capable of directly facilitating and inhibiting the state of Ib and group II fibers via interneurons and thus changes in descending synaptic drive following stroke may affect the relative contributions of these fibers to the motoneuronal state (Cabaj et al. 2006; Jankowska 1992). Significant contributions from either Ib or group II fibers seem unlikely though, as group II fibers are less sensitive to vibratory inputs than group Ia, and group Ib fibers are largely insensitive to vibration unless it is accompanied by baseline contraction (Fallon and Macefield 2007). Moreover, although direct contributions from hyperexcitable interneurons cannot be excluded, monoamines appear to be inhibitory to spinal interneurons that process high-threshold afferent input (Jankowska 1992). In fact, it has been suggested that the delayed onset and depressed magnitude of reflex activity in short-latency flexion reflex afferents in the paretic arm of individuals with chronic stroke may be due in part to a monoamine-mediated decrease in interneuronal excitability (Dewald et al. 1999). Low-threshold proprioceptive input, especially that from group I afferents, however, is facilitated by monoaminergic input (Hammar and Jankowska 2003) and thus a contribution to reflex hyperexcitability from interneurons is possible. Posttetanic potentiation (PTP) of spinal motoneurons also may have contributed to the results observed herein, as others have posited a role for PTP in the development of large contractile forces during and following vibration or electrical stimulation of 1a afferents in humans (Baldwin et al. 2006; Eklund and Hagbarth 1966).

Insofar as glutamate can mediate the induction of persistent inward currents via metabotropic group I receptors on spinal motoneurons (Rekling et al. 2000; Svirskis and Hounsgaard 1998) or dorsal horn neurons (Derjean et al. 2003), the colocalization of norepinephrine and glutamate in coeruleospinal projections (Fung et al. 1994) suggests that glutamatergic neuromodulation may also play a role in motoneuronal hyperexcitability poststroke. Additionally, a reduction in GABAergic inhibition following stroke could have contributed to the apparent finding of enhanced PICs in the paretic limb, as pharmacological studies using both baclofen and nifedipine have shown a calcium-dependent component to the PICs instantiated following administration of GABA_B antagonists (Russo et al. 1998). Indeed, a reduction of GABAergic inhibition has been suggested to contribute to the development of spasticity following SCI (Bennett et al. 2001b), although the influence of descending inhibitory pathways following stroke remains unclear.

One important distinction should be drawn between the mechanistic implications of this study and those derived from investigations of SCI, however. In the case of SCI, it has been estimated that only 2–15% of 5-HT (Newton and Hamill 1988) and 2–3% of NE (Roudet et al. 1994) remain caudal to the lesion and this chronic paucity of monoamines leads to an adaptation that results in an increased expression of PICs (Bennett et al. 2001a,b; Li and Bennett 2003). In contrast, an abundance of monoamines is hypothesized to accompany the putatively increased reliance on reticulospinal fibers following stroke and it is not yet known whether a chronic increase in the concentration of spinal monoamines could also yield changes in monoaminergic receptor sensitivity. Consequently, such adaptive physiological effects cannot be eliminated from consideration.

Clinical implications

Results from this study suggest that PICs with SSF contribute to the expression of altered reflexes in the paretic limb following stroke. Although no studies to date have systematically investigated the potential relationship between intrinsic motoneuron properties and the clinical presentation of spasticity following stroke, we expect the monoaminergic facilitation of contralesional motoneurons to be of direct consequence for the development of poststroke spasticity.

Future work that seeks to identify conclusively the impact of increased bulbospinal monoaminergic drive and intrinsic motoneuron properties following stroke will be essential for our understanding of the pathogenesis of motor deficits measured clinically as well as for the development of more targeted rehabilitation interventions. The well-characterized receptor pharmacology of descending brain stem projections and spinal motoneurons lends itself well to precise modulation by targeted therapeutics and holds promise for the discovery of pharmacological agents capable of treating the underlying causes of specific motor impairments following stroke.

GRANTS

This work was supported by National Institute of Neurological Disorders and Stroke Grant R01-NS-054269.