Intraspinal microstimulation and diaphragm activation after cervical spinal cord injury

General neurophysiology protocols.

Rats were anesthetized as above. Rectal temperature was maintained at ∼37.5°C by a heating pad (CWE, Ardmore, PA). A femoral artery was catheterized (PE-50) for blood pressure measurements (Statham P-10EZ pressure transducer, CP122 AC/DC strain gauge amplifier; Grass Instruments, West Warwick, RI) and arterial blood samples. The femoral vein of the same hindlimb was also catheterized (PE-50) for supplemental fluid administration and conversion from isoflurane to urethane anesthesia (1.7 g/kg iv; Sigma, St. Louis, MO). Animals received a tracheotomy and were mechanically ventilated (50–65% O₂, balance N₂; 6–7 ml/kg volume; 70–72 bpm frequency; Harvard Apparatus, Holliston, MA) throughout the experimental procedures. Since spontaneously breathing anesthetized rats can rapidly become hypercapnic, we elected to employ mechanical ventilation for these initial experiments in order to keep blood gases stable and remove Pa_O2 and/or Pa_CO2 fluctuations as confounding variables.

End-tidal CO₂ (Capnogard 1265; Respironics, Wallingford, CT) was continuously monitored, and arterial blood gases (iSTAT1; Abbott, Princeton, NJ) were periodically assessed from 0.1-ml arterial samples. On the basis of these measures, the inspired CO₂ content was adjusted to maintain Pa_CO2 at 40 mmHg. If base excess was greater than −3 meq, this was corrected with intravenous administration of sodium bicarbonate solution [8.4%, Hospira; dose (ml) = 0.3·weight (kg)·standard base excess].

Recordings of EMG activity from respiratory-related muscles (i.e., diaphragm, genioglossus, and intercostal) and an off-target, nonrespiratory muscle [i.e., extensor carpi radialis longus (ECR)] were obtained with pairs of Teflon-coated tungsten hooked wires (A-M Systems, Sequim, WA). Recordings of the genioglossus muscle were obtained at the base of the tongue (Fuller et al. 1999; Fuller and Fregosi 2000) and exhibited a centrally driven inspiratory rhythm that was used to trigger ISMS. For intercostal EMG recordings, a small incision lateral to the sternum was made at the T₂ level and wires were placed 1–1.5 cm lateral to the midline.

The spinal cord was exposed via a cervical middorsal incision followed by laminectomy and durotomy from C₂ to C₅. Raw EMG signals were amplified at 100-10K, band-pass filtered at 100 Hz–10 kHz (A-M Systems, Carlsborg, WA), and, in the case of the genioglossus EMG recording, passed through a moving time averager (50 ms time constant; CWE, Ardmore, PA). The moving time average signal was used to trigger the stimulator for ISMS. All EMG signals were digitized at 25 kHz (CED Power 1401) and recorded (Spike2 v8; CED, Cambridge, UK) to a PC and then analyzed off-line.

ISMS at midcervical spinal cord.

A tungsten microwire electrode (FHC, Bowdoin, ME) was mounted in a stereotaxic micromanipulator (David Kopf Instruments, Tujunga, CA). The microwire had a 100-μm segment at the tip stripped of all insulation. The electrode was placed above the C₄ segment with the dorsal root entry zone as the lateral anatomical landmark. Electrical activity was initially recorded via the stimulating electrode and assessed visually with Spike2 software and audibly with an AM8 Audio Monitor (Grass Technologies, Quincy, MA). The software program simultaneously displayed the inspiratory diaphragm EMG recording; these procedures enabled placement of the stimulating electrode tip in proximity to inspiratory neurons (i.e., targeting phrenic motoneurons). Inspiratory bursting was absent in rats with subacute C2Hx injury. Therefore, a stimulator and constant-current stimulus isolation unit (S88X and SIU-C; Grass Technologies, Warwick, RI) were used to deliver single pulses (0.3 ms duration) during electrode descent with gradually decreasing currents (200, 100, 50 μA). This approach allowed us to determine a location for ISMS that evoked left diaphragm activation.

Stimulation of the spinal cord was initiated (triggered) with the EMG signal recorded from the genioglossus muscle. Specifically, a “threshold crossing” was established such that when the inspiratory integrated EMG burst (50 ms time constant) reached a preset amplitude the spinal cord was stimulated. All stimulations during the inspiratory phase were made at the onset of the genioglossus inspiratory burst, while stimulations made during expiratory periods were accomplished by adding a 400-ms delay to the trigger.

Experimental group 1: ISMS before and after acute C2Hx.

The genioglossus EMG signal was used to trigger ISMS during two consecutive inspiratory cycles, followed by ISMS delivered during two consecutive expiratory cycles. In both cases, repeated 250-ms stimulus trains (200 μA) were delivered with 0.3 ms pulse duration and 100 Hz stimulus frequency. Once these initial stimulations were complete, ISMS was delivered during the inspiratory cycle for 1 min. All of the aforementioned ISMS was done with the spinal cord intact. After C2Hx, the ISMS protocol was repeated, as described above. Subsequent to protocols performed with spinal cord intact or after C2Hx, an additional bout of ISMS was administered after neuromuscular blockade via intravenous pancuronium bromide (2.5 mg/kg; Hospira) to confirm unequivocally that the evoked activity was not contaminated by a stimulus artifact.

Selection of stimulus parameters.

Pilot experiments were done with a manually triggered, open-loop approach in spinal-intact animals. Repeated 250-ms trains of ISMS were targeted to the left phrenic motor nucleus, and EMG activity was recorded in both hemidiaphragms, the genioglossus, and ipsilateral intercostal and forelimb (ECR) muscles (see, e.g., Fig. 1). A range of stimulation frequencies (50, 100, 200, 300 Hz) and currents (50–200 μA) were tested to optimize parameters for eliciting compound motor unit action potentials (MUAPs) in the left diaphragm. Activation of the ipsilateral hemidiaphragm was more pronounced when 100-Hz stimulation was used, with stimulation frequencies above 200 Hz producing large contractions of nonrespiratory muscles. Preliminary experiments also indicated that progressive attenuation of ISMS-induced phrenic MUAP amplitude occurred during sustained stimulation (e.g., 1 min) at 200 Hz but not at lower stimulus frequencies (50 Hz, 100 Hz). In the first series of experiments, a stimulus current of 200 μA was used to activate the ipsilateral diaphragm.

Open in a separate window

Fig. 1.

Cervical ISMS activates the ipsilateral diaphragm A: schematic diagram illustrates the EMG recording sites relative to the placement of the stimulating electrode (S). The stimulating electrode was placed in the immediate vicinity of the phrenic motor nucleus, and ISMS was initiated via a trigger signal based on inspiratory tongue (genioglossus) EMG activity. LI, left intercostal. B: pre-C2Hx representative EMG recordings from the genioglossus, extensor carpi radialis (ECR), and both sides of the diaphragm showing baseline (prestimulation) activity and a period of genioglossus-triggered cervical ISMS (100 Hz, 200 μA; 0.3-ms pulse duration, 250-ms train duration) represented by a gray box in the ∫tongue trace. C: expanded trace of the EMG recording from the left diaphragm during the period of ISMS shows each stimulus artifact, indicated by asterisks, and each subsequent MUAP. D: overlay of each elicited MUAP, aligned by the stimulus artifact, demonstrates constant latency and amplitude.

Experimental group 2: ISMS after subacute C2Hx.

The intraspinal electrode placement was determined as described above. Triggered ISMS was then delivered during two consecutive expiratory periods, followed by one continuous minute of stimulation during the inspiratory phase. In both cases, a continuous 250-ms train was delivered with 0.3-ms pulses at 100 Hz and 100 μA. Stimulations also were repeated after neuromuscular blockade as described above.

Consistent with other reports, we used cessation of the ipsilateral diaphragm inspiratory EMG burst and histology as functional and anatomical verification, respectively, of the subacute C2Hx lesions (Goshgarian 1981). After the electrophysiology protocols, the animals were perfused with saline followed by 4% paraformaldehyde (Sigma). The cervical spinal cords from the subacute C2Hx animals were subsequently harvested, and a tissue block including the C₂ region was paraffin embedded, sectioned (8 μm), and counterstained with cresyl violet.

Biomechanical impact of ISMS.

Changes in tracheal pressure were assessed as a crude indicator of the biomechanical impact of ISMS. It should be noted, however, that this was not considered to be a primary outcome variable but rather an accessory measurement intended to provide some insight regarding the functional impact of spinal stimulation. The pressures associated with lung inflation and deflation were measured at the tracheal cannula. Measurements were taken continuously and used to compare baseline conditions to stimulation periods. For each period of interest, tracheal pressure deflections for 20 consecutive breaths were averaged and presented in a box and whisker plot (SPSS; IBM, Armonk, NY). Passive tracheal pressure values during exhalation were obtained in each preparation after neuromuscular blockade and thus provided a baseline measurement that was devoid of any respiratory contribution from the animal (for additional details, see results).

Data acquisition and analyses.

For the first group of experiments, results from the spinal-intact condition were compared to data collected 10 min after acute C2Hx. In addition to the evoked responses (i.e., response during ISMS), we also evaluated the impact of the ISMS on spontaneous EMG activity (i.e., following the period of stimulation). To assess ISMS-entrained activation of motor units, the EMG signals were averaged with respect to stimulus pulses within each trial. The resulting stimulus-triggered average (McPherson et al. 2015; Moritz et al. 2007) served to minimize activity unrelated to ISMS. The resultant waveform averages represented ISMS-entrained MUAPs. All data were collected with a CED Spike2 data acquisition system and subsequently analyzed with Spike2 v8 software on a standard PC. Values are reported as means ± SD.

Methodological caveats including off-target effects of cervical ISMS.

The method described here is not truly a “closed-loop” system because the ISMS intensity did not vary with the relative strength of endogenous “respiratory drive.” Varying the stimulus intensity or duration in proportion to the relative magnitude of the EMG burst used to trigger ISMS (vs. a simple “onset trigger”) could potentially address that issue, but this could also prove problematic since ISMS needs to occur at the onset (vs. peak) of the inspiratory effort. Another caveat is that rats were mechanically ventilated because in our experience spontaneously breathing and anesthetized rats rapidly develop arterial hypercapnia and respiratory acidosis. However, mechanical ventilation will impact endogenous respiratory drive and thereby alter the interactions between spontaneous and ISMS input to the phrenic region. It should also be noted that tongue EMG activity is unlikely to provide an effective ISMS trigger signal in the vagally intact and spontaneously breathing rodent or human. It is possible to record inspiratory-related discharge from the tongue muscles in the awake human and rodent, but available data indicate that tongue EMG may not provide a consistent inspiratory trigger signal (Bailey 2011; Sood et al. 2005). In the present experiments on anesthetized rats, however, the tongue EMG output enabled a rigorous test of the fundamental hypothesis, and we suggest that future work should focus on additional potential sources of respiratory output to serve as endogenous trigger signals or alternative detection algorithms (Dow et al. 2006).

Neural substrate considerations.

One of the more challenging issues related to electrical stimulation of the spinal cord is identification of the neural substrate being affected directly or indirectly. Here our immediate goal was to target ISMS of gray matter at the level of the phrenic motor nucleus. Whether phrenic motoneurons are being directly activated by ISMS cannot be determined from the present findings, although latencies in the range of 1.5–2.5 ms suggest indirect stimulation via polysynaptic pathways. In the subacute C2Hx group, stimulation of ipsilateral bulbospinal inputs to phrenic motoneurons can be effectively ruled out because these pathways will be undergoing degeneration. It is possible that ISMS activated commissural bulbospinal projections from the contralateral cord [i.e., pathways associated with the “crossed phrenic phenomenon” (Goshgarian 2003; Goshgarian et al. 1991)]. However, past studies in which intraspinal stimulation of the cervical cord was used to activate crossed phrenic pathways reported latencies of <1 ms (Fuller et al. 2003), and these are considerably shorter than the values reported here. Collectively, the evidence is most consistent with indirect activation of phrenic motoneurons via ISMS, possibly via spinal prephrenic interneurons (Lane 2011) or even ascending afferent projections originating below the injury (Decima and von Euler 1969).

ISMS is capable of discrete activation of circuitry with minimal undesired physiological effects (Pikov et al. 2007). Here we observed clear responses in the ipsilateral diaphragm, as intended, but also off-target motor unit recruitment. Very small and inconsistent activation of the contralateral diaphragm, intercostal muscles, and genioglossus were present during ISMS. This indicates current spread beyond the target region or activation of other circuits via interneurons (Perlmutter et al. 1998). For example, in the case of ISMS-evoked responses in the contralateral diaphragm and intercostal muscles, our previous transneuronal tracing studies provide an interneuronal basis for those off-target responses (Lane et al. 2008). Intercostal and/or contralateral diaphragm recruitment could in fact be advantageous when cervical SCIs involve much larger-scale compromise of diaphragm and inspiratory intercostal function compared with C2Hx. Activation of the genioglossus was barely above the threshold for detection and probably reflected activation of ascending projections to medullary respiratory control neurons. For example, we previously reported that some cervical interneurons associated with the phrenic motor circuit have ascending projections to the medulla (Lane et al. 2008).

The relatively stronger coactivation of forelimb muscles induced by ISMS presents a more challenging technical issue. As with the ipsilateral diaphragm, constant-latency entrainment occurred in the forelimb, and this may reflect overlap of phrenic and forelimb circuits along the rostro-caudal axis (Gonzalez-Rothi et al. 2015). ECR motoneurons, however, appear to occupy a more dorso-lateral position than the phrenic motoneuron pool (Tosolini and Morris 2012). This raises the possibility that even if ISMS was precisely restricted to phrenic motoneurons, concomitant excitation of ECR axons as they course toward their ventral roots could still occur. Further investigations are needed to sort out the neuroanatomical basis for direct vs. indirect effects of ISMS. The latter may be especially beneficial by activating desired motoneuron pools in a more natural order than achieved with direct stimulation.