In vitro neuronal depolarization and increased synaptic activity induced by infrared neural stimulation

: Neuronal responses to infrared neural stimulation (INS) are explored at the single cell level using patch-clamp electrophysiology. We examined membrane and synaptic responses of solitary tract neurons recorded in acute slices prepared from the Sprague-Dawley rat. Neurons were stimulated using a compact 1890 nm waveguide laser with light delivered to a small target area, comparable to the size of a single cell, via a single-mode ﬁber. We show that infrared radiation increased spontaneous synaptic event frequency, and evoked steady-state currents and neuronal depolarization. The magnitude of the responses was proportional to laser output.


Introduction
Infrared neural stimulation (INS) is an experimental nerve stimulation modality, first reported by Wells et al. [1], that uses infrared laser pulses to directly activate excitable biological tissues. Although currently at an early stage of development, the enhanced spatial resolution offered by laser stimulation makes INS a potentially attractive alternative to electrical stimulation for the development of next generation neuro-prosthetic devices. This is because INS only activates cells directly within the laser spot, whereas electrical currents spread more diffusely from an electrode, resulting in recruitment of cell bodies and fibers around the stimulation site [2]. In contrast to optogenetic techniques, in which light sensitivity of a neuron is conferred by artificial expression of a light-sensitive cation-channel [3], INS results in direct stimulation of unmodified neuronal tissue, making it potentially easier to translate to human applications. Further, INS does not require the addition of exogenous drugs as a transduction substrate, as is the case with the use of caged neurotransmitters [4,5]. However, debate regarding the mechanisms that underlie INS, and variability in the results obtained in the limited number of INS studies described in the literature, have confounded the rapid investigation of this technology. It seems likely that differences in the biological preparation used, the method used to assay neuronal activation, and the parameters used for laser stimulation have contributed to this debate. Nonetheless, interest remains, as indicated by recent studies examining the applicability of INS for human neuroprostheses [6,7].
Studies of compound action potentials from nerve bundles and single neurons recorded in vitro have allowed investigation into the mechanisms that drive INS responses. These studies have indicated a photothermal effect as the likely driver behind INS, after eliminating the possibility of significant contributions from pressure, electric field, and photochemical effects [8]. In vitro whole cell recordings of membrane responses to high-intensity (230 mW) INS at 1875 nm suggest that heat sensitive TRPV4 channels contribute to the depolarizing effect of INS, at least in some cell types [9,10]. A more general mechanism involving a change in the target cell's electrical capacitance was suggested by Shapiro et al. as the likely physical pathway involved in INS for pulse lengths in the thermal confinement regime [11]. Shapiro et al. additionally demonstrated the role of water absorption as the primary mechanism by which energy from INS is absorbed by substituting heavy water, with its lower absorption coefficient, and eliciting a marked reduction in the amplitude of response. Further, nanosecond pulses have been shown to evoke potentials in mechanosensitive cells in the cochlea via an optoacoustic mechanism [12]. However, there is currently only limited research focusing on interrogation of INS in cells of the central nervous system. One such study relied on calcium imaging to probe the visual cortex's response to INS and demonstrates a fast and slow component of the calcium signal [13], with another study showing that INS increased the spontaneous discharge rate of visual cortex neurons recorded in vivo in the Macaque through extracellular recordings of unit activity [14]. A similar study instead described the use of infrared light to cause neural inhibition [15]. The laser parameters, cell types and recording methods of these studies of the mechanism by which INS operates are detailed in Table 1. These studies do not provide detail as to the shape, cellular cause, and threshold onset time and change in holding potential, with the remainder of measurements relying on signal optical imaging. Thus while some level of stimulation due to laser illumination can be inferred, the specific underlying cause remains to be determined. In order to deconvolve each of the underlying mechanisms contributing to INS, studies are required which investigate the response of specific neural processes such as synaptic activity. Sciatic nerve [8] CAPs 750-2120 5 µs-10 ms 0.3-1 Ganglion neurons [9] WCPC 1875 1-15 ms 20-60 Ganglion neurons [10] WCPC 1470-1875 2-30 ms 15-100 Oocytes, HEK cells [11] WCPC 1889 0.1-10 ms 0.28-7.3×10 −3 Inner ear hair cells [12] CAPs 420-2120 3-5 ns 69-92×10 −3 Visual cortex [13] Ca 2+ imaging 1875 250 µs 0.1-0.88 Visual cortex [14] SOI, WCPC 1875 75-300 µs 0.5-1.3 Somatosensory cortex [15] SOI, WCPC 1875 250 µs 0.01-0.55 To the best of our knowledge, only one previous study has investigated the synaptic activity of cells undergoing INS and reported on an invertebrate animal model [16]. Here we present preliminary findings that indicate that infrared illumination at low power densities evokes intensity-dependent reproducible inward currents in individual neurons recorded in mammalian acute brainstem slices, and that such effects are also accompanied by evidence of increased synaptic activity, even at illumination intensities that are insufficient to directly drive neuronal depolarization.

Laser parameters
The experiments performed utilized a monolithic thulium doped ZrF 4 − BaF 2 − LaF 3 − AlF 3 − NaF (ZBLAN) glass waveguide laser that was custom built using in-house facilities [17]. The 1.89 µm output light was coupled to a 14 µm mode field diameter, single mode fiber (SMF) (N A = 0.11, Thorlabs, SM2000), with a nominal power output of 7.8 mW measured at the fiber tip. The minimum stable output power was 0.4 mW. SMF was used to maximize the power density as well as demonstrate the use of a small core fiber, which has been shown to cause less trauma during insertion into the human cochlea [18]. The fiber was positioned in a micromanipulator at 45 • to the horizontal and positioned approximately 100 µm from the target cells, giving a pulse energy calculated at the surface of the cell of 261 J/cm 2 over a full one second pulse for the fiber parameters listed previously in a medium of water with n = 1.33. The calculated spot size after divergence was 42.8 µm × 26.7 µm, which is comparable to the cell width of ∼ 20 µm, and enables improved specificity when targeting individual cells. The power output and pulse length of the laser was controlled using an external digital trigger.

Slice preparation and whole cell recording
The effects of laser stimulation on neuronal membrane properties and synaptic activity were assessed in whole-cell recordings obtained from juvenile rat brain slices as previously described [19]. Data presented were obtained from a single female Sprague-Dawley pup (post-natal day 36). The rat was anesthetized with isoflurane and immediately decapitated. The brainstem was rapidly removed and placed in ice-cold artificial cerebrospinal fluid (aCSF). Coronal sections (300 µm) were cut using a vibrating microtome (Lecia VT 1200S). Slices of the medial nucleus of the solitary tract (Bregma: -13.92 to -13.96 [20]) were collected and placed in warm (34 • C) aCSF and allowed to equilibriate for up to 60 min before recording. Recordings were performed at room temperature in the recording chamber of an Olympus microscope superfused at 1.5-2 mL/min with carbogen-bubbled aCSF.
The Whole cell recordings were made using a Multiclamp 700B patch-clamp amplifier (Molecular Devices, USA) in voltage-or current-clamp modes using borosilicate pipettes with 1.5-2 µm tip diameters (3-6MΩ). Recordings were obtained after formation of a gigaseal; series resistance compensation of 70-80% was used in voltage-clamp recordings. Recorded parameters were digitized using a Power 1401 Mark II (Cambridge Electronic Design, UK) running Spike2 software.

Experimental protocols
The following protocols were used to establish the effects of laser illumination on neuronal activity, once stable recordings had been established. All pulses were delivered for a duration of 1 s for the data shown. Pulse lengths ranging from 20 ms to 1 s were evaluated, with significant responses only noted at 1 s. As a result, 1 s was used for the remainder of the experiment.
The first protocol examined the effect of laser power on membrane responses recorded in voltage clamp or current clamp mode. The power was pulsed for 1 s at increments of approximately 0.9 mW in the first cell tested, with larger power intervals of ∼ 2 mW for all other cells, to a maximum output of 7.8 mW with 10 s between the onset of laser pulses. The larger intervals allowed for a greater number of trials to be carried out more quickly on the remainder of cells as cell viability degrades over time. Further, clear trends were still able to be confirmed with the larger interval in laser power.
The second protocol examined the reversal potential for laser-evoked current flow; reversal potential was defined as the potential at which laser-evoked current flow became zero. Each cell was stepped alternatively from a holding potential of -60 mV to between -110 mV and +10 mV, at 10 mV increments. An interval of 5 seconds was used between steps. Midway through each step a 1 s laser pulse (maximum output) was delivered. In one cell the reproducibility of INS-evoked currents was assessed by repeatedly stimulating the neuron (7.8 mW, 1 s) every 30 seconds for 30 minutes.

Data analysis
Data recorded using Spike2 software were exported to MATLAB (Mathworks, Natick, MA) for analysis. A custom script was utilized to graph raw data from cellular responses, holding current, and holding potential in Fig. 1(a), 2(a), and 3(a). The script also calculated the laser power using the measured diode driver current. The calculation was performed using the measured maximum output power combined with previously measured laser output power levels for a specified diode laser driver current. These previous measurements also provided the threshold diode driver current for lasing. The statistical analysis and plots of Fig. 1(b), 2(b), 3(b), 5(b), and 5(c) were performed using GraphPad Prism.
Synaptic events were detected from voltage clamp recordings using an event-detection function in Spike 2. The membrane current channel was first DC-filtered to eliminate slow changes in holding current, differentiated to facilitate detection of overlapping synaptic events, and then window-discriminated to detect downward deflections corresponding to post-synaptic currents [21]. Peristimulus time histograms of laser-evoked synaptic events were constructed from sequential laser stimuli using 100 ms bin width in Spike 2.

Infrared neural stimulation evoked inward membrane currents
INS consistently evoked intensity-dependent inward currents in neurons in which the fiber optic was aligned with the recorded neuron. INS-evoked inward currents ranged from 3-22 pA (Mean = 10.0 pA ± 7.9 pA) at the maximum laser output of 7.8 mW, with an increase in response of 1.15 pA/mW (p = 0.99) (Fig. 1). Laser-evoked responses were dependent on illumination of the target cell, as they were abolished by repositioning the fiber optic or by interrupting the laser path, indicating that the measured responses represented a biological effect and not an electromagnetic artifact. Further, the same protocol produced null results in the absence of a cell, indicating that heating of the recording electrode was not responsible for the observed effects. While previous reports have shown heating of the recording electrode with less total energy deposited than in our experiments, the rate at which energy is deposited in these instances is higher. For example, a 0.7 mJ pulse in 0.2 ms [11] equates to 7000 J/s, thus representing a greater rate of change in temperature over time (dT/dt) than our 261 J/s. Although we observed considerable variability in the amplitude of INS-evoked responses between neurons (see Fig. 1(b)), the reproducibility of INS-evoked responses was highly reproductible within individual neurons.
Comparable results were observed when recordings were performed in current-clamp mode; INS consistently depolarized neurons, with a mean response of +4.15 mV ± 2.55 mV observed at maximum laser output (Fig. 2). On no occasion did INS evoke action potentials.
The amplitude and polarity of INS-evoked current was altered by manipulating the holding current. The relationship between holding potential and the amplitude of INS-evoked current was linear between -100 and -30 mV, with an estimated reversal potential of −47.8 mV±3.2 mV(R 2 = 0.99, p < 0.0001) measured in 4 neurons (Fig. 3).

INS-evoked responses are stable and reproducible
The final cell interrogated was subjected to an extended exposure time (t = 30 min) with laser power set to the maximum 7.8 mW (261 J/cm 2 over a one second pulse). This extended exposure ensured that there was strong correlation between laser stimulus and cellular response, as well as showing that the effect was constant over a long exposure time (Fig. 4). This suggests that INS at the pulse energies and wavelength used in this experiment is non-damaging and reversible.
During this test the beam was briefly interrupted between the laser head and fiber coupler. This blocking was performed in order to confirm that the observed stimulus responses were caused by the laser stimulation and not the result of electrical noise in the system or other extraneous variables. The shaded region in the time averaged trace of Fig. 4 indicates the time during which the fiber coupling was inhibited. The inhibited region has no clear stimulation coinciding with

Synaptic response
Ongoing synaptic activity is a feature of central nervous system neurons recorded in acute slices. We observed an increase in the synaptic activity of cells when exposed to INS. Synaptic activity is shown in the raw trace of a single cell in Fig. 5(a) as rapid, transient downward deflections that deviate significantly from the baseline current. Increased synaptic activity was observed during laser irradiation under ascending stimulus intensity protocols and current stepping protocols in five cells (Fig. 5(b)) but was most clearly illustrated in a stimulus-triggered histogram of synaptic event frequency generated from 30 minutes of intermittent stimulation at 261 J/cm 2 over a one second pulse (Fig. 5(c)).

Discussion
This study presents the first known instance of INS causing an increase in synaptic activity concurrently with evoked inward membrane currents, which reversed at -47.8 mV. However, under the stimulus parameters and recording conditions employed in the current study, stimuli were insufficient to drive action potential generation under current-clamp recording conditions, at least in the limited number of cells surveyed. This is important to consider when interpreting the increased synaptic activity that was consistently observed in response to irradiation, and suggests that infrared activation may influence the probability of neurotransmitter release occurring independent of the action potential propagation. The absence of action potential generation with increased synaptic activity correlates with previous findings [16].
Excitation of cells at a level below that required to reliably elicit APs was shown as an inward current flow in the cells, with a mean change in holding current of 10.0 pA at maximum laser power. Clear correlation between an increase in the laser power delivered and the magnitude of the cellular response was noted, in keeping with previously reported cellular response curves [22]. Importantly, we have demonstrated INS has the capacity to elicit a specific synaptic response, even when a cell is not stimulated to a level required to trigger a full AP. To the best of our knowledge, this is the first report of INS enhancing synaptic activity in a vertebrate animal model, with a previous report focusing on C. elegans [16]. However, the previous report noted that their effect was due to a large change in temperature over a short time interval e.g. a large dT/dt. In contrast, our experiments reveal that similar effects can be observed under long (1 s) pulse lengths with low average power, where dT/dt is comparatively small. Additionally our experiments utilized a small cross-sectional area of stimulation from a SMF with a correspondingly higher total energy deposited than reports in Table 1. Thus further investigation into the mechanism responsible for the synaptic effect of INS is warranted, as it will contribute to the overall understanding of INS.

Conclusion
This study demonstrates that direct laser stimulation of naïve mammalian neurons in acute brain slices causes reproducible membrane currents and, for the first time, that such events are accompanied by an increase in synaptic activity. These preliminary results highlight the need for further investigation into cellular mechanisms that underlie INS; the ability to 'tune' synaptic activity, independent of directly driving action potentials, may have useful applications that go beyond direct nerve stimulation.