Address correspondence to Steven M. Rothman, MD, Department of Pediatrics (Clinical Neuroscience), University of Minnesota Medical School – MMC 486, 420 Delaware Street S.E., Minneapolis, MN 55455-0374,U.S.A. E-mail: email@example.com
Purpose: To determine if a small ultraviolet emitting diode (UV LED) could release sufficient γ-aminobutyric acid (GABA) from a caged precursor to suppress paroxysmal activity in rat brain slices.
Methods: Electrophysiologic recordings were obtained from rat brain slices bathed with caged GABA: 4-[[(2H-benzopyran-2-one-7-amino-4-methoxy)carbonyl]amino]butanoic acid (BC204), at concentrations between 3 and 30 μm. Seizure-like activity was induced by perfusing slices with extracellular medium lacking magnesium and containing 4-aminopyridine (4-AP; 100 μm). A small, high-power UV LED was used to uncage BC204 and determine whether an increase in ambient GABA could alter normal or paroxysmal activity in the slice.
Results: UV LED illumination, in the absence of BC204, had no effect on CA1 population spikes or seizure-like activity. The light did induce a small temperature elevation (<0.15°C) over the current intensities and exposure durations used in these experiments. In the presence of BC204, UV light decreased the CA1 population spike and seizure-like activity. The BC204 effect can be best accounted for by release of GABA: The reduction of population spikes and seizure-like activity was blocked by the GABA antagonist picrotoxin, and BC204 illumination produced a membrane polarization that reversed at the expected potential for GABAA receptors.
Discussion: These experiments establish that illumination of a low concentration of caged GABA with a tiny UV LED can release sufficient GABA to attenuate seizure-like activity in brain slices. Because our seizure model is very severe, it is probable that this technique would have a robust effect in human focal epilepsy.
The therapy of epilepsy remains problematic for approximately one-fourth to one-third of patients who fail to respond to conventional medical therapies. In addition to the burden of uncontrolled epilepsy, many of these patients have to tolerate a variety of toxicities attributed to their antiepileptic drugs (AEDs). Although surgery has clearly benefited many patients with focal epilepsy, a substantial number of patients do not get complete remission after operation, and even those patients who do remit are at risk for neurologic or neuropsychological deficits arising from the resection (Helmstaedter & Elger, 1996; Jokeit et al., 1997; Spencer & Huh, 2008). The deficiencies of current therapy have stimulated many investigators to explore newer technologies to terminate or even prevent seizures. Up to the present, much of this work has been focused on electrical stimulation, either of the vagus nerve, the epileptogenic region of cortex, or deeper brain structures, such as the anterior nucleus of the thalamus (Nakken et al., 2003; Morrell, 2006). Results from this approach have been mixed, with success rates in the range of 50–60%.
Potentially more appealing than resective surgery or electrical stimulation is the possibility of directly reducing the excitability of neurons within epileptogenic cortex. This has already been achieved by focal cooling, which can rapidly terminate paroxysmal activity in rodent and human brain (Karkar et al., 2002; Rothman et al., 2005; Imoto et al., 2006).
Recent advances in optics, genetics, and organic chemistry have suggested ways to exploit the temporal and spatial resolution of optical techniques to diminish excitability within epileptogenic cortex. Several investigators have transfected light-sensitive channels directly into neuronal membranes (Banghart et al., 2004; Volgraf et al., 2006; Han & Boyden, 2007). This technique allows flashes of light to directly inhibit the transformed neurons, although it permanently alters the genome of the light-sensitive neurons.
A more practical approach to harnessing the potential of optical techniques takes advantage of caged compounds (Lester & Nerbonne, 1982). Caged γ-aminobutyric acid (GABA) analogs, which are based on a carbon or ruthenium core, release the natural inhibitory neurotransmitter GABA, when illuminated with ultraviolet (UV) or visible light, respectively (Curten et al., 2005; Rial Verde et al., 2008). These caged GABA analogs could be delivered into the subarachnoid space over the cortical convexities and allowed to penetrate into the underlying cortex. A tiny, light-emitting diode (LED) placed over the cortex could release GABA locally, reducing paroxysmal activity associated with a focal seizure. The uncaged GABA would likely target the extrasynaptic, high-affinity GABAA receptors, which are nondesensitizing and sensitive to low micromolar levels of extracellular GABA (Semyanov et al., 2004; Farrant & Nusser, 2005; Scimemi et al., 2005; Krook-Magnuson & Huntsman, 2007). Caged GABA could be superior to local administration of GABA, because it is released almost instantaneously in the brain area where the seizure focus is located, whereas GABA itself would have to be delivered relatively slowly by a pump at high concentration to overcome the avid GABA reuptake system (John et al., 2007).
The work described in subsequent text demonstrates that a small UV LED can pass sufficient light into a mammalian brain slice to uncage a newer, bath-applied caged GABA: 4-[[(2H-benzopyran-2-one-7-amino-4-methoxy) carbonyl]amino]butanoic acid (abbreviated as BC204), and attenuate paroxysmal activity almost instantly (Curten et al., 2005). Because human epilepsy may not be nearly as severe as interictal bursting induced by the extreme pharmacologic provocations in these slice studies, GABA uncaging might exert robust effects on clinical seizures.
Materials and Methods
Male Sprague-Dawley rats were anesthetized with halothane and rapidly decapitated using a protocol approved by the University of Minnesota Institutional Animal Care and Use Committee. We used animals at three different ages because we needed slices appropriate for three different types of experiments:
(1) For hippocampal population spike experiments, 4–5 week-old animals were used, because this age provides slices that are most suitable for extracellular recording; (2) for hippocampal intracellular recording experiments, >6 week-old animals were used because it is easier to obtain intracellular impalements from the brains of larger animals; and (3) for in vitro interictal burst experiments, 2–3 week-old animals were used, because it is easier to elicit paroxysmal activity from the entorhinal neocortex of very young animals (Wong & Yamada, 2001). The brains were removed and briefly immersed in ice-cold artificial cerebrospinal fluid (ACSF) containing (mm): 124 NaCl, 5 KCl, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 22 NaHCO3, and 10 glucose, continuously bubbled with a 95% O2/5% CO2 gas mixture. They were then placed on their dorsal surface on ACSF-dampened filter paper. The cerebellum and brainstem were removed with a scalpel, and the portion of the brain anterior to the optic chiasm was removed with one coronal cut.
For hippocampal slices, the flat frontal surface was then rotated down, and the ventral surface was placed against an agarose block in a Vibratome pan (Pelco, St Louis, MO, U.S.A.). The pan was filled with oxygenated ice-cold ACSF, and the Vibratome well was filled with ice water. We cut 500-μm transverse slices using the highest setting of blade vibration amplitude and an excursion speed that allowed the blade to pass through the brain in 20–30 s. The hemispheres of each slice were separated, and incubated in a submerged oxygenated holding chamber at 25°C for at least 1 h, before transfer to a submerged recording chamber for electrophysiology. For entorhinal neocortical slices, we used similar methods except that the slice was positioned with the dorsal side down in the Vibratome pan and the slice thickness was 600 μm.
Electrophysiologic recordings were made in a submerged chamber (Warner Instruments, Hamden, CT, U.S.A.) perfused with oxygenated ACSF (95% O2/5% CO2) flowing at 2 ml min−1 and maintained at 33°C by heating the inflow and the chamber itself. For extracellular recording, microelectrodes were made from 1.2 outer diameter (o.d.) 0.68 inner diameter (i.d.) (mm) borosilicate glass (WPI, Sarasota, FL, U.S.A.) and had resistances of 4–6 mΩ when filled with ACSF. We elicited population spikes in CA1 by delivering constant current pulses (50 μs; WPI 305) through a bipolar tungsten microelectrode (David Kopf Instruments, Tujunga, CA, U.S.A.) placed in the stratum radiatum of CA3. The stimulation current was adjusted to produce a half-maximal response. The CA1 population spikes were fed into a conventional DC amplifier (Axoclamp 2 A; Axon Instruments, Union City, CA, U.S.A.), digitized at 10 kHz, and stored on a personal computer using a commercially available A/D converter and software (Digidata 1322 and pClamp 9; Axon Instruments). The spikes were analyzed off-line to determine peak magnitude.
For intracellular recording we used an identical recording configuration, except that “sharp” microelectrodes were fabricated from 1.0 mm o.d./0.50 i.d. (mm) borosilicate glass (Friedrich and Dimmock, Millville, NJ) and had 80–120 mΩ impedances when filled with 4 m potassium acetate. Signals were recorded with the same amplifier in current clamp mode and digitized at 2 kHz.
Seizure-like activity was provoked by addition of the convulsant 4-aminopyridine (4-AP; 100 μm) and removing magnesium from the extracellular perfusate. The recording electrode was placed in the entorhinal cortex. The standard deviation of the voltage recording and number of spikes in a defined epoch during seizure-like activity were measured using programs within pClamp.
Caged GABA, BC204, was perfused into the preparation for at least 30 min prior to the first illumination. To maximize slice exposure to the UV light and rapidly uncage BC204, a high-intensity UV LED package (NCSU033A; Nichia, Japan) was positioned on a copper pedestal 1 mm below the slice coverslip (Fig. 1). The UV LED has an emission maxima of 365 nm, close to BC204’s absorption peak at 350 nm and generates 54 mW of optical power at 100 mA current flow. We did not determine the precise geometry of slice illumination, but centered the slice over the LED.
All of the slice experiments involving BC204 were carried out in a darkened room with only occasional illumination with a red lamp to visualize the slice. In preliminary experiments, we directly monitored the slice temperature during UV exposure with a miniature thermocouple inserted into a 33 gauge needle (HYP-1;Omega, Stamford, CT, U.S.A.). The thermocouple was inserted into the top of the slice, but it is impossible to determine the extent of the slice that the probe actually interrogates.
All specialty chemicals were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.) except for the BC204, which was a custom synthesis (Curten et al., 2005).
Experimental values are expressed as ± standard deviation (SD). The error bars on graphs are also standard deviations, not standard errors, which explains their size. Statistical comparisons were made on the same slices, before and after light exposure, using paired t-test or Friedman repeated-measures analysis of variance (ANOVA) on ranks followed by Student-Newman-Keuls test (for multiple comparisons) (Sigma Stat 3.1; Systat, Point Richmond, CA, U.S.A.).
The possibility that UV LED activation could heat slices was assessed in our initial set of experiments. We first activated the light for 5 s, the exposure time that we planned to use in later attempts to terminate seizure-like activity. When the UV LED illuminated the slice with currents ranging from 50–400 mA, the peak temperature increase was less than 0.35°C at highest current setting (Fig. 2). At 100 mA, the current setting used in many of the electrophysiology experiments, the peak temperature increase was less than 0.15°C after 5 s. With 30 s light exposures, 100 mA of LED current still increased slice temperature by only 0.2°C (Fig. 2).
When population spikes were elicited in CA1 by CA3 stimulation, we found that illumination of control slices for 5 s before and during electrical stimulation using 50 mA (n = 4 slices), 100 mA (n = 8 slices), or 200 mA (n = 9 slices) LED current had no effect on peak amplitude (Figs. 3A and 4A). In each case, the magnitude of the population spike without illumination was compared to the population spike in the same slice during illumination 1 min later. However, after the addition of BC204 (30 μm), illumination using UV LED currents of 50 mA (n = 19 slices), 100 mA (n = 17 slices), or 200 mA (n = 15 slices) for 3 s prior to and during stimulation, produced a statistically significant reduction of the peak of the population spike (Figs. 3B and 4B). The population spikes recovered within a minute after all three levels of slice illumination. We believe that this decrement of the population is most likely explained by tonic activation of high-affinity GABAA receptors on dendrites and soma of CA1 pyramidal neurons. This should partly shunt excitatory postsynaptic currents and reduce the subsequent population spike arising from synaptic activation. If this is the case, the BC204 response should be reduced or eliminated by co-application of the GABAA antagonist picrotoxin. As anticipated, in the presence of BC204 (10 μm) plus picrotoxin (30 μm), illumination of the slice had no effect upon the population spike (Figs. 3C and 4C).
Although the preceding results were consistent with GABA release triggered by BC204 uncaging, we wanted to obtain more direct evidence for tonic activation of GABAA receptors by BC204 uncaging. We used intracellular recording to determine the effect of UV illumination and BC204 on resting membrane potential. These experiments were done with sharp electrodes, because placement of the UV LED below the coverslip precluded direct visualization of individual neurons with infrared optics and use of patch pipettes and single electrode voltage clamp. In the presence of BC204 (30 μm), illumination (100 mA UV LED current) produced a sustained effect on membrane potential that varied with voltage (Fig. 5A). When normalized to the response at −80 mV, the average voltage dependency of the BC204 effect reversed at approximately −60 mV, close to the accepted reversal potential for mammalian GABAA receptors in physiologic extracellular chloride solution (n = 4 neurons; Fig. 5B). Assuming a linear current–voltage relationship, this corresponded to 17% increase in membrane conductance.
We then examined the effect of control UV illumination and BC204 uncaging on interictal bursts induced in neocortical slices by addition of 4-AP and removal of extracellular magnesium. We applied light for 5 s, which we knew, from preliminary experiments and tissue culture results, was of sufficient duration to attenuate paroxysmal activity (Rothman et al., 2007). Three parameters that reflected the intensity of seizures were analyzed:
1 The standard deviation of the baseline voltage recording over the 4.5 s immediately preceding, and coinciding with, light application. The light was programmed to activate every 10 min and we always analyzed the 4.5 s before and after illumination. This analysis algorithm eliminated the potential for investigator bias, because the segments picked for analysis were selected prior to any investigator examination and a control segment immediately preceding illumination was always compared to the illuminated segment that followed. The 10 min interval also ensured that “fresh” BC204 was available for uncaging, although we have not systematically examined the rate of BC204 replenishment in slice. We ignored the first 0.5 s after illumination because there was sometimes a stimulation artifact.
2 Number of spikes in the 4.5 s intervals immediately preceding, and coinciding, with light application, which occurred every 10 min; spikes were defined as voltage excursions from baseline greater than 3 times the baseline standard deviation.
3 Number of bursts in the 4.5 s intervals immediately preceding and coinciding with light application, which occurred every 10 min; bursts were defined as clusters of spikes with no sustained return to baseline voltage.
For each experimental group, we analyzed data from six separate slices, in which each slice was illuminated between 2 and 8 times (usually 4–6 times), depending upon the level of paroxysmal activity. We saw no effect of UV LED illumination alone on the baseline voltage standard deviation, spike number, or burst count (Figs. 6A and 7). There were, however, significant reductions of these same three parameters after UV LED activation in the presence of 10 μm and 30 μm BC204 (Figs. 6B and 7). These differences disappeared when the BC204 was reduced to 3 μm or UV LED current was reduced from 100 mA to 50 mA. Adding picrotoxin (100 μm) to the BC204 (30 μm) significantly reduced the BC204 effect (Figs. 6C and 7). The return of seizure-like activity after illumination is not surprising, because the slices remain in ACSF containing 4-AP and low extracellular magnesium.
The experiments described in this article document the ability of a small UV LED to uncage GABA from BC204 and attenuate seizure-like activity in rodent brain slices. These strongly positive results, in an epilepsy model far more severe than the naturally occurring disease, suggest that this technique could translate to human epilepsy. In order to accomplish this we envision a programmable pump delivering a caged compound into the subarachnoid space overlying epileptic cortex (Albright et al., 2006). One or more tiny UV LEDs could be placed on a grid immediately over this region. The UV LEDs could be activated either on a predetermined schedule, much like the present vagal nerve and thalamic stimulation units, or by a responsive seizure detection unit. There are already precedents for both in regular clinical use or trials (Kerrigan et al., 2004; Kossoff et al., 2004; D’Alessandro et al., 2005).
This could become a very specific and powerful method for modulating focal epilepsy without subjecting intractable patients to toxic doses of medication or irreversible brain damage from epilepsy resections. We believe that this technique is far superior to local drug application and specifically local GABA application (John et al., 2007). Optical stimulation would be far more rapid than any mechanical device for direct intracerebral administration of drug. There would also be no direct brain trauma from a device that had to penetrate the cortex itself. It should be possible to configure a grid or matrix of small UV LEDs so that uncaging would occur in a very defined region of epileptic cortex below the LEDs. Furthermore, using a caged compound to make GABA instantaneously available at the receptors, obviates the need to apply enormous concentrations of GABA to overcome the robust GABA uptake systems (John et al., 2007).
There is of course concern that UV light will penetrate poorly through neocortex. Other slice experiments have indicated that about 50% of 365 nm light is absorbed over each 200 μm of brain, so there will be substantial light attenuation within the first millimeter of cortex (unpublished). However, there is still good reason to believe that positioning the LED at the cortical surface could transmit sufficient light to abort clinical seizures. First, the very striking suppression of paroxysmal activity in the slices verifies that there is sufficient light penetration from a UV LED to uncage even within a section of brain tissue 500 μm thick. Second, we utilized an especially severe seizure model in which potassium channel function was blocked by 4-AP, and N-methyl-d-aspartate receptors and calcium channels were unblocked by removal of magnesium. This model leads to far greater excitability than any naturally occurring human epilepsy.
Since completing our experimental work, we have become aware of another series of caged GABA analogs that are activated within the visible light range (Rial Verde et al., 2008). These could be even more applicable to clinical use, due to the greater tissue penetration of visible over UV light.
In addition, several important pharmacologic questions will need to be answered about BC204 or related caged compounds. There is, as yet, little information about the brain permeability of any caged compound. They will not have to cross blood–brain barrier, but they would still need to move from subarachnoid space into cortex. Based upon its structure, there is no reason to believe that BC204 will not move across the pia, but we still require confirmatory evidence. There is also no information about BC204’s stability in biologic tissue at 37°C. However, the original article that characterized its optical properties showed that it was stable for at least several days when diluted, so there is evidence that it will be stable at physiologic pH (Curten et al., 2005).
No pathologic studies have explored possible toxic effects of BC204, but preliminary studies have failed to show any anatomic neuronal alterations in cultured neurons after 24–48 h of BC204 exposure. Likewise, we will need to examine whether the caging moiety by itself is toxic. We do not anticipate cellular injury from the UV LED, because we are using a wavelength that is beyond the generally accepted range for UV toxicity and anticipate illumination for only brief time-periods. Even if there were some subtle injury caused by the UV LED or BC204 itself, this would have to be balanced by the consequences of continued seizures or permanent brain resection.
The optimal physical arrangement for an in vivo UV LED illumination system also has to be determined. The tiny diode could be removed from the much larger package, attached to a silastic grid, and placed over epileptogenic cortex in contact with the dura (Fig. 1). A heat pipe could be used to dissipate the temperature gradient between LED and dura (Hilderbrand et al., 2007). We saw only a tiny rise in slice temperature caused by UV illumination per se, so we would anticipate that robust blood flow would prevent any temperature rise in brain (Sukstanskii & Yablonskiy, 2004). As previously indicated, the duty cycle for illumination would be very brief, minimizing battery power consumption. There are already usable algorithms for human seizure detection, which could be adapted to trigger brief UV LED illumination (Litt et al., 2001).
We recognize that the present study is a very preliminary step toward the possibility of using caged drugs as treatments for epilepsy. There are still numerous experiments that need to be performed in vivo before clinical applications can be seriously considered. However, the light intensity of LEDs has dramatically increased over the last decade, making us optimistic that an LED-based implantable device is feasible. The technology for pump fabrication continues to improve, so drug administration into the subarachnoid space is achievable. The selection of the optimal caged compound for this sort of therapy remains uncertain. Although BC204 remains very attractive, ruthenium-based compounds that uncage with visible illumination may be superior (Rial Verde et al., 2008). These questions should be experimentally resolvable over a relatively short period.
This work was supported by NIH (R01 NS42936 to SMR).
We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Disclosure: None of the authors has any conflict of interest to disclose.