Electrical stimulation therapies for CNS disorders and pain are mediated by competition between different neuronal networks in the brain

Abstract

CNS neuronal networks are known to control normal physiological functions, including locomotion and respiration. Neuronal networks also mediate the pathophysiology of many CNS disorders. Stimulation therapies, including localized brain and vagus nerve stimulation, electroshock, and acupuncture, are proposed to activate "therapeutic" neuronal networks. These therapeutic networks are dormant prior to stimulatory treatments, but when the dormant networks are activated they compete with pathophysiological neuronal networks, disrupting their function. This competition diminishes the disease symptoms, providing effective therapy for otherwise intractable CNS disorders, including epilepsy, Parkinson's disease, chronic pain, and depression. Competition between stimulation-activated therapeutic networks and pathophysiological networks is a major mechanism mediating the therapeutic effects of stimulation. This network interaction is hypothesized to involve competition for "control" of brain regions that contain high proportions of conditional multireceptive (CMR) neurons. CMR regions, including brainstem reticular formation, amygdala, and cerebral cortex, have extensive connections to numerous brain areas, allowing these regions to participate potentially in many networks. The participation of CMR regions in any network is often variable, depending on the conditions affecting the organism, including vigilance states, drug treatment, and learning. This response variability of CMR neurons is due to the high incidence of excitatory postsynaptic potentials that are below threshold for triggering action potentials. These subthreshold responses can be brought to threshold by blocking inhibition or enhancing excitation via the paradigms used in stimulation therapies. Participation of CMR regions in a network is also strongly affected by pharmacological treatments (convulsant or anesthetic drugs) and stimulus parameters (strength and repetition rate). Many studies indicate that treatment of unanesthetized animals with antagonists (bicuculline or strychnine) of inhibitory neurotransmitter (GABA or glycine) receptors can cause CMR neurons to become consistently responsive to external inputs (e.g., peripheral nerve, sensory, or electrical stimuli in the brain) to which these neurons did not previously respond. Conversely, agents that enhance GABA-mediated inhibition (e.g., barbiturates and benzodiazepines) or antagonize glutamate-mediated excitation (e.g., ketamine) can cause CMR neurons to become unresponsive to inputs to which they responded previously. The responses of CMR neurons exhibit extensive short-term and long-term plasticity, which permits them to participate to a variable degree in many networks. Short-term plasticity subserves termination of disease symptoms, while long-term plasticity in CMR regions subserves symptom prevention. This network interaction hypothesis has value for future research in CNS disease mechanisms and also for identifying therapeutic targets in specific brain networks for more selective stimulation and pharmacological therapies.

Fig. 1

Diagram of Hypothetical Competitive Neuronal Network Interaction between a CNS Disorder Network and a Therapeutic Network. In panel A the neuronal network that mediates the brain disorder is shown as being tonically active. At this time the stimulation procedure has not been presented, so the potential therapeutic neuronal network is dormant. Therefore, the input from the brain disorder network to the conditional multireceptive (CMR) regions(s) of the brain, which may include the brainstem reticular formation and amygdala, are activated by the disorder network and contribute importantly to the expression and severity of the symptoms of the disorder. In panel B the stimulation protocol has been instituted, causing the therapeutic network to become active, which does not produce any noticeable effects of its own. However, the therapeutic network also projects to an overlapping population of neurons in the CMR brain regions, and this input competes for dominance of the CMR regions. This network competition reduces the severity of the symptoms of the disorder and may even prevent these symptoms, allowing the stimulation protocol to be an effective therapy for this disorder.

Fig. 2

Diagram of an Example of a Neuronal Network Interaction with Negative Consequences - Audiogenic Seizures (AGS). Elements of the normal network for hearing up to the level of the midbrain are shown in panel A. (Projections to the medial geniculate body and auditory cortex are not shown.) Elements of the normal network for locomotion are shown in panel B. In panel A the inferior colliculus (IC) normally projects to a “minor” degree to the sensorimotor brain regions (elements) of the normal locomotion network, including the deep layers of the superior colliculus (DLSC) and brainstem reticular formation (RF), which project to the substantia nigra reticulata (SNR) and periaqueductal gray (PAG), and PAG, RF and DLSC all project to the spinal cord. A genetically-based or induced defect of γ-aminobutyric acid (GABA) inhibition (γ↓) in the IC allows high intensity acoustic stimuli to induce excessive firing in of IC neurons that project abnormally intense output to the sensorimotor nuclei, and overdrive their output to the spinal cord, resulting in the sound-induced audiogenic seizure (AGS). The behaviors seen in AGS range in severity from wild running (1), to generalized (four limb) clonus (2), and in the most severe form tonic seizures that may result in tonic hind limb extension (3), which leads to behavioral (post ictal) depression. Other abbreviations: CN = cochlear nucleus, SOC = superior olivary complex, DNLL = dorsal nucleus of the lateral lemniscus, MLR = midbrain locomotion region. [Based in part on (11,18)]

Fig. 3

Typical example of the major increase in response to an external stimulus of a neuron in a conditional multireceptive (CMR) brain region [brainstem reticular formation (RF)] induced by a convulsant (bicuculine) agent that blocks γ-aminobutyric acid (GABA_A) receptor-mediated inhibition. Panel A shows a bar graph of the neuronal firing changes in panel B. Panel B illustrates the effect on single bulbar RF neuron firing in the unanesthetized cat. The three columns in panel B illustrate peristimulus time histograms (PrSTH) in response to visual stimuli (100 µsec light flash at 0.5 Hz) in control (left column) before drug administration, which has a hint of time-locked firing (peak), suggesting a minor response to the stimulus. After i.v. administration of bicuculline (0.075 mg/kg) (middle column), the same neuron exhibited a considerable increase in firing and a large peak in the PrSTH as well as consistent repetitive firing (inset). The response recovered to the control pattern (right column) 27 min after drug administration ceased. The insets in each PrSTH show oscilloscope trace of the neuron’s extracellular action potentials (3 superimposed trials). PrSTH parameters (25 presentations, bin width = 1 msec, scan length 400 msec). N= number of action potentials per PrSTH, Calibration = 1 mv. [Based on data from (92)]. This effect was seen in 88% of over 700 bulbar and midbrain RF neurons in unanesthetized cat and rat by systemic or direct (iontophoretic) application of drug. [see (30)]

Fig. 4

Typical examples of the increases in single neuron responses to an external stimulus of a conditional multireceptive (CMR) brain region in the midbrain reticular formation induced by a convulsant agent (pentylenetetrazol) that reduces the inhibitory effect of γ-aminobutyric acid (GABA). Panel A shows a bar graph of the neuronal firing changes in panel B. Panels B, C, and D illustrate the effect on the firing of three different single midbrain RF neurons in the unanesthetized cat to three types of electrical stimuli. The panels B, C, and D illustrate poststimulus time histograms (PSTH) in response to single electrical stimuli of the sciatic nerve (B), inferior colliculus, or lateral geniculate nucleus (D) of three separate neurons. The control (left column) shows the firing pattern before drug administration, which have no clear indication of time-locked firing (peak) in response to each stimulus. After i.v. administration of a convulsant drug (middle column) [pentylenetretrazol, 15 (B), 10 (C), or 10 (D) mg/kg], the same neuron in each row exhibited a considerable increase in extracellular action potential firing and a large peak in the PSTH. The responses recovered towards the control pattern (right column) 20 (B), 22 (C), and 29 (D) min after drug administration. Stimulus parameters = 100 µsec single bipolar pulse at 0.5 Hz). PSTH parameters (50 presentations, bin width, 1 msec, scan length 400 msec). N= number of action potentials per PSTH. [Based on data from (92,93)].

Fig. 5

Schematic representation of the different brain regions in which the effects of i.v. administration of convulsant drugs have been examined in the unanesthetized cat. The mean increase of overall firing rate as percentage of control is shown by the scale on the left. The two areas of the brainstem RF exhibited the highest degree of enhancement. AMG = amygdala, HPC = hippocampus, IC = inferior colliculus, LG = lateral geniculate, PC = pericruciate cortex, RF = brainstem reticular formation, SNR = substantia nigra reticulata. [Taken from Faingold and Riaz (55) - Permission to reprint requested from CRC Press.]

Fig. 6

The ability of an uncompetitive NMDA antagonist (ketamine) that produces dissociative anesthesia to reversibly block the responsiveness of neurons in a conditional multireceptive (CMR) brain region in the lateral amygdala (LAMG). Panel A shows a bar graph of the mean effect in 8 different LAMG neurons in 8 rats. Mean number of action potentials/PSTH in control was 182.1 ± 42.0 (SEM) vs. 9.6 ± 2.0 after ketamine. Panel B shows a typical example of the effect of ketamine (30 mg/kg, i.p.) on the LAMG neuronal firing. The LAMG neuron exhibited an onset response before ketamine treatment (panel B top row). The neuronal firing was significantly (p<0.01, paired t test) and almost completely suppressed 15 min after ketamine treatment (panel B middle panel). Four hr after ketamine treatment, the LAMG neuronal response was comparable to that prior to ketamine treatment (panel B bottom row). The insets (in each row of panel B) show examples of the digital oscilloscope tracings for each PSTH. AP amplitude: 300 µV. N=number of action potentials in the PSTH. Treatment was given in previously unanesthetized awake behaving rats (Acoustic stimulus parameters: 12 kHz tone burst, 100 ms duration, 5 ms rise-fall, 100 dB SPL, 0.5 Hz rate, 50 presentations). [Based on data from (63)]

Fig. 7

Mean firing changes and a typical example of the effects of repetitive auditory stimuli [audiogenic seizure (AGS) kindling) on lateral amygdala neurons in an inherited form of AGS (GEPR-9). In panel A the mean firing change is shown with significant firing increases (action potentials/PSTH before kindling 31.4 ± 4.9 (SEM), (36 neurons) vs. 162.1 + 11.8 (30 neurons) after AGS kindling p<0.05, paired t test. In panel B the firing of typical examples before and after repetitive seizures is shown by PSTHs. The insets (in panel B) show examples of the oscilloscope tracings for each PSTH. AP amplitude: 200 µV (left column) and 300 µV (right column). N=number of action potentials in the PSTH in unanesthetized behaving rats. [See Fig. 6 for stimulus and PSTH parameters, based on data from (63)]