Single infrared light pulses induce excitatory and inhibitory neuromodulation

: The excitatory and inhibitory effects of single and brief infrared (IR) light pulses (2 µm) with millisecond durations and various power levels are investigated with a custom-built fiber amplification system. Intracellular recordings from motor axons of the crayfish opener neuromuscular junction are performed ex vivo . Single IR light pulses induce a membrane depolarization during the light pulses, which is followed by a hyperpolarization that can last up to 100 ms. The depolarization amplitude is dependent on the optical pulse duration, total energy deposition and membrane potential, but is insensitive to tetrodotoxin. The hyperpolarization reverses its polarity near the potassium equilibrium potential and is barium-sensitive. The membrane depolarization activates an action potential (AP) when the axon is near firing threshold, while the hyperpolarization reversibly inhibits rhythmically firing APs. In summary, we demonstrate for the first time that single and brief IR light pulses can evoke initial depolarization followed by hyperpolarization on individual motor axons. The corresponding mechanisms and functional outcomes of the dual effects are investigated.


Introduction
Infrared (IR) light pulses in the wavelength range of 1400 nm -2100 nm have been used to stimulate or inhibit neural as well as muscular activities without chemical or genetic manipulation [1][2][3][4][5][6]. Advantages of IR neuromodulation (INM) over conventional electrical stimulation include high spatial-temporal selectivity and contactless delivery, minimizing potential physical intervention. Recent studies have demonstrated that INM can be combined with brain imaging techniques such as magnetic resonance imaging (MRI) [7] and optical coherence tomography (OCT) [8]. The compatibility has positioned INM at a promising position for basic and clinical brain research where simultaneous modulation and imaging are needed. Additional applications of INM range from brain stimulation [7,[9][10][11], cochlear prostheses [12][13][14], and cardiac pacemakers [15] to neural identification and monitoring during surgery [16,17]. It is generally accepted that thermal transients induced by water and tissue absorption of pulsed IR light underlie the IR nerve stimulation (INS) [18,19] and IR nerve inhibition (INI) [20][21][22]. Heat-mediated nerve stimulation [23][24][25] and inhibition [26,27] have also been achieved with visible and near-IR light pulses by utilizing photo-active materials as extrinsic light absorbers and heat converters. With the photo-active materials being in contact with or attached to the target cell membrane, the light-induced heating is likely to be more spatially confined and with different temporal dynamics than those under direct IR light illumination. However, specifics of the biological processes induced by these thermal transients under both the label-free and material-based techniques remain to be further examined.
In accordance with the photothermal effects, various mechanisms have been reported and discussed to explain INS [1]. Brief IR light pulses have been shown to evoke depolarization in multiple biological and artificial membranes. Changes in the membrane capacitive current resulting from alterations in the membrane structure caused by the IR light-induced spatiotemporal temperature gradients have been shown to generate depolarization [19,[28][29][30]. Another proposed mechanism of INS, though less investigated, identified nanoporation in the plasma membrane caused by brief IR light pulses [31]. The nanopores result from the temporary destabilization of the plasma membrane and can act as channels for the flux of ions, which may depolarize cells or activate intracellular pathways [31,32]. The activation of heat-sensitive transient receptor potential vanilloid (TRPV) ion channels can also help explain INS observed in neurons or cells expressing these channels [33,34]. Finally, it has been suggested that IR light pulses can induce intracellular Ca 2+ transients in a wide range of systems [9,[35][36][37][38][39]. Though these Ca 2+ transients exhibit relatively slow dynamics and are unlikely to directly generate action potentials (APs), they can indirectly alter neuronal excitability.
In contrast to INS, INI has been largely attributed to "heat block" effects [20][21][22][40][41][42][43]. Early studies using perfusion bath to raise bath temperature have shown that this manipulation can suppress the AP amplitude and duration or completely block APs [44,45]. These inhibitory effects stem from a combination of altered passive membrane properties and ion channel kinetics involved in generating APs. Accordingly, similar mechanisms have been assumed in INI. Specifically, increases in the cell membrane fluidity, which are correlated with a reduction in membrane resistance, can occur during and following short IR light pulses [46] and can lead to INI of AP initiation [22,43]. Moreover, successful modulation of AP waveforms and dynamics by IR light pulses suggests that temperature-dependent changes in K + and Na + channels kinetics are important contributors to INI [22,43,47]. Studies have also specifically emphasized the importance of tetraethylammonium (TEA)-sensitive and voltage-dependent K + ion channels for INI in Aplysia [41,42]. Overall, it is likely that the relative importance of the mechanisms outlined above may be neuron-specific.
Since the temperature rises induced by IR light pulses nonspecifically impact the passive membrane properties and ion channels, it is expected that both INS and INI effects can potentially be evoked concurrently or sequentially within neural tissues or even in a single neuron. Indeed, a silicon neural microdevice implanted into the rat cortical region both elevated and suppressed cortical multiunit responses to continuous wave (CW) IR light (1550 nm) irradiation [48]. Though the illumination regime with a pulse duration of 2 minutes was quite different from most INM reports, it nevertheless highlights the complex nature of INM, especially of the brain in vivo. So far, most INM studies have mainly focused on either INS or INI separately, rather than evaluating the net outcome of the two coexisting but opposite processes. For example, excitatory TRPV channels have been discussed extensively in INM studies [33,34,[36][37][38][39]49]. However, heat-sensitive two-pore domain TWIK related K + (TREK) channels, which are known to be widely expressed among neurons and whose activation can generate inhibitory effects [50,51], have yet to be evaluated in INM. We believe that contributions of excitatory and inhibitory effects triggered by IR light pulses are likely to occur concurrently and should be considered together to realistically assess potentials of INM applications.
In this report, we use intracellular recordings to investigate both excitation and inhibition evoked by single and brief IR light (2 µm) pulses in single motor axons of the crayfish opener neuromuscular preparation. We examine biophysical processes underlying the dual effects and demonstrate for the first time that single and brief IR light pulses can excite or inhibit axonal APs in different functional contexts. Our data and exploration of the underlying mechanisms provide basic understanding of changes in the functional states of a neuron during INM. This study further establishes a foundation for predicting outputs generated by INM in more complex neural networks [7,11] and for translational applications [17].

Infrared neuromodulation laser system
To systematically explore the effects of brief IR light pulses on the excitability of the motor axons, a custom-built high-power thulium-doped fiber amplifier (TDFA) was designed and constructed ( Fig. 1(a)) that boosted the power of a 2-µm CW diode laser (FPL2000S, Thorlabs) up to 1 W. The thulium-doped double clad fiber (SM-TDF-10P/130-HE, Nufern) was used as a gain fiber for highly efficient operation between wavelengths of 1900-2100 nm. A 793-nm diode laser system (BWT Beijing Ltd., China) was used to pump the amplifier in a backward-pumping configuration. An acousto-optic modulator (AOM) (T-M250-0.3C16Z-3-F2P, Gooch & Housego) was incorporated into the setup after the amplification stage to shape clean pulses and to avoid unwanted background illumination resulting from amplified spontaneous emission. The AOM, which has a transmission rate of ∼50% at 2 µm wavelength, was used to modulate the amplified output at the desired pulse duration and peak power. The amplified 2 µm light was delivered to the target motor axon via an optical fiber with a core diameter of 105 µm, as shown in Fig. 1(b). For each preparation, the delivery fiber was positioned in line with the axon main branch, slightly above the axon surface at an angle of 28°t o the horizontal plane. A red laser diode was coupled into the delivery fiber to facilitate the alignment. The delivery fiber was cleaved before each experiment and the power was measured at the free end of the delivery fiber. Single IR light pulses with various durations (0.2-10 ms) and average power levels (50-400 mW) were applied. The pulse energy ranged between 0.08-0.5 mJ/pulse. The resting membrane potential of the target axons was monitored during the recording. The stability of the resting membrane potential was used as the main criterion of a healthy axon [52]. Irreversible depolarization in the resting membrane potential was only observed when pulse energies significantly higher than 0.5 mJ/pulse were deposited.

Neuromuscular preparation and electrophysiological configuration
Crayfish (Procambarus clarkii) of both sexes were purchased from Niles Biological Supplies (Sacramento, CA). The opener neuromuscular preparation from the first pair of walking legs were dissected to expose both the inhibitory and the excitatory motor axons (20-30 µm in diameter) for this study. The control physiological saline contained (mM): 195 NaCl, 5.4 KCl, 13.5 CaCl 2 , 2.6 MgCl 2 , 10 HEPES (pH 7.4). The saline was circulated by a peristaltic pump (Cole-Parmer, IL, USA) at a rate of around 1 ml/min during the recording. Two-electrode current clamp (TECC) recordings from the axon were performed with a MULTICLAMP 700B (Molecular Devices, CA, USA) and AXOCLAMP-2A (Axon Instruments Inc., CA, USA). Microelectrodes were filled with 0.5 mM KCl (40-60 M ). The recording electrode (voltage (V) electrode, Fig. 1(b)) was placed distal to the axon branch point, 350-400 µm away from the tip of the delivery fiber, to minimize the confounding effects of direct IR light illumination on electrodes. The stimulation electrode (current (I) electrode, Fig. 1(b)) was positioned on the axon main branch more proximal to the recording electrode. Both electrodes were arranged with an angle of ∼28°to the horizontal plane. To monitor the IR light-induced temperature transients experienced by the target axons (see section S1 and Fig. S1 in Supplement 1), an open patch pipette filled with saline (∼10 M ) was positioned slightly above the axon membrane and close to the illumination center. The electrophysiological recording was performed under an Olympus BX51 microscope with a 60X water immersion lens. All experiments were conducted at room temperature around 20-21°C. All chemicals were purchased from Sigma-Aldrich unless specified otherwise.

Data analysis
Data acquisition and analysis were performed with Igor Pro (WaveMetrics). Voltage signals were filtered at 5 kHz and sampled at 50 kHz (NI USB-6363). Each preparation (N) represented a set of data recorded from an axon dissected from an animal. Statistical results were presented as average ± the standard error of mean (SEM) and the Student's t-test was used to determine the statistical significance with the significance level α set as 0.05.

Brief infrared light pulses evoke axonal membrane potential depolarization followed by hyperpolarization
With the customized TDFA system depicted in Fig. 1(a), we first scanned for optimal durationpower combination that maximized depolarizing amplitude without tissue damage. Figures 2(a) -2(d) illustrate membrane depolarizations evoked by IR light pulses of varying durations (0.2 ms -10 ms) at four power levels. The axonal membrane potential (V m ) depolarized during the IR light pulses and decayed to resting level after the pulses. The decay descended into hyperpolarization in trials with higher energy levels. The amplitudes of the depolarization increased with increased power levels and durations (Figs. 2(a) -2(d)). Results of the parameter scan is summarized by plotting the amplitudes of depolarization against the pulse duration for each power level ( Fig. 2(e)). In general, the depolarization amplitude increased linearly with durations up to two milliseconds and plateaued as the pulse duration increased further. The bar plots in Fig. 2(f) summarize the depolarization amplitudes evoked by IR light pulses with the same energy level (0.4 mJ/pulse) for four pulse durations (N = 8). The depolarization amplitude was larger when the pulse energy was concentrated within shorter durations (see also the comparison in Fig. S2).
Within the parameter space we scanned, the largest depolarization was achieved by a pulse of 1-ms duration with 400-mW power. For the rest of the study, IR light pulses of 400-mW average power and 1-ms duration were used for further investigation. We also measured temperature transients, at the surface of motor axons, induced by the IR light pulse combinations used in Fig. 2. The local temperature increased rapidly during the IR light pulses and decayed to baseline within 100 ms ( Fig. S1(a)). For the same pulse energy, higher peak temperature rises were achieved with shorter pulse durations ( Fig. S1(a)) and the temperature rise increased as the total energy deposition went up ( Fig. S1(b)). A maximum temperature rise of ∼23°C was recorded for the 400-mW and 1-ms IR light pulses, which is comparable to measurements reported in previous studies [19,33]. The overall shapes of the temperature transients are similar to the V m changes shown in Fig. 2, with two notable differences. First, the decay of the temperature transient is slower than that of the V m . Whereas the temperature decayed to a low steady state ∼50 ms after the IR light pulse, V m typically had returned to baseline or undergone hyperpolarization by that time. Second, for the lower power level of 50 mW, the depolarization plateaued for durations longer than 4 ms (Fig. 2(e)) while the maximal temperature continued to rise for longer durations (Fig. S1(b)).

Brief infrared light pulses-induced depolarization is tetrodotoxin-insensitive and voltage-dependent
In order to further characterize the IR light-induced depolarization, we first examined whether it was boosted by Na + influx. Figure 3(a) shows that the depolarizations remained unchanged before (dashed blue) and after (red) the Na + channel blocker tetrodotoxin (TTX) (200 nM) was applied. Thus, Na + channels did not bias the amplitude or duration of the depolarization (p = 0.08684, N = 4). We next explored whether the IR light-induced depolarization exhibits any voltage dependence. The V m of the target axon was shifted in both directions by a series of current steps delivered with a second intracellular stimulation electrode (Fig. 3(b)). V m was recorded for nine different levels, with (red) and without (dashed blue) IR light pulses. The recordings were carried out in the presence of TTX (200 nM) and ZD 7288 (10 µM, Tocris), which were both added 20 min before the recordings. These two drugs linearize and expand the membrane polarization range for this study. TTX can inhibit the AP firing at large depolarization levels. ZD 7288 is a blocker of hyperpolarization-activated cation channels (I h ) which contribute to the resting membrane potential, input resistance as well as the "sag" during hyperpolarization (Fig. S3). As the current step went from -20 nA to 20 nA, the peak depolarization amplitude decreased while the hyperpolarization component became more noticeable. Subtracting the IR traces from the Control traces in Fig. 3(b) and aligning the baselines of the IR light-evoked potentials show the depolarization (and hyperpolarization) evolution clearly in Fig. 3(c). When the depolarizing amplitude was plotted against their corresponding V m , a linear relationship with a slope of -0.054 was observed (Fig. 3(d), R 2 = 0.8872, N = 4). These results demonstrated that the membrane depolarization induced by short IR light pulses is voltage-sensitive and its amplitude decreases as the V m depolarizes.

Brief infrared light pulses-induced hyperpolarization reverses its polarity near potassium equilibrium potential
A second and slower potential component of the IR light-induced membrane potential change could be observed following the initial and fast IR light-induced depolarization. The amplitude and polarity of this second component depended on the V m . As shown in Fig. 3(b) (red), the second component was hyperpolarizing when the axon was depolarized and depolarizing when the axon was hyperpolarized. Given its physiological function illustrated below, we named the second component hyperpolarization based on its polarity at resting state and depolarized potentials. Figure 4(a) illustrates aligned traces recorded from different membrane potentials (from -100 mV to -40 mV). At resting membrane potential, the hyperpolarization in general lasted 50-100 ms. When the hyperpolarization amplitudes, measured at the time point of 7 ms after the IR light pulse, as indicated by the dashed gray line in Fig. 4(a), are plotted against the V m biased by current injection, the hyperpolarization reversed around the resting membrane potential of -68 mV ( Fig. 4(b), red cross). The reversal potential of the hyperpolarization shifted as the extracellular K + concentration ([K + ] o ) was varied, in the depolarizing direction for high [K + ] o and in the hyperpolarizing direction for low [K + ] o (Fig. 4(c)). While the actual reversal potential of the IR light induced-hyperpolarization varied considerably among different preparations, the dependence on [K + ] o examined in four preparations exhibited a consistent trend (Fig. 4(d)), which supports the hypothesis that the K + conductance likely contributed to the second component. For these measurements, 200 nM TTX and 10 µM ZD 7288 were used.

Barium-sensitive potassium channels contribute to the infrared light-induced hyperpolarization
We next aimed to explore the origin of the IR light-induced hyperpolarization. Based on the findings of the [K + ] o -dependence of the hyperpolarization and its voltage-dependence (Fig. 4), we sought K + channels that are also heat-activated. The temperature-sensitive two-pore domain TWIK-related K + (TREK) channels satisfy both requirements [53]. More importantly, the IR-light induced temperature rises (Fig. S1) are within the window of the activation temperature (30-45°C) of TREK channels [50]. Thus, we tested this hypothesis by first examining the effects of a TREK channel blocker, Ba 2+ (5 mM), on IR light-induced hyperpolarization (in the presence of TTX and ZD 7288). Consistent with the role of TREK channels in setting the resting membrane potential, the application of 5 mM Ba 2+ depolarized the resting membrane potential (7.2 ± 0.47 mV (N = 4, p = 0.00061)) and increased the axonal membrane input resistance ( Fig. 5(a), red). Ba 2+ -mediated block of the IR light-induced hyperpolarization was apparent at depolarized V m (red) (Fig. 5(b)). However, the blocking effects of Ba 2+ appeared to be absent when the V m was significantly hyperpolarized (Figs. 5(b) and 5(c)). This is consistent with previous studies showing that TREK channels exhibited a strong outward rectification, namely these channels were nearly impermeable to K + when the V m was below the K + equilibrium potential [54][55][56]. On average, Ba 2+ reduced the IR light-induced hyperpolarization amplitude by 61% ± 11% (N = 4, p = 0.01044) when +10 nA current steps were applied. Since the application of Ba 2+ also changed the resting membrane potential and input resistance of the axon, which could potentially lead to underestimation of the blockade, we further compared the K + conductance (g K ) underlying the IR light-induced hyperpolarization before and after Ba 2+ application using the following equation: Here, V m is the membrane potential depolarized by +10 nA current steps, ∆V h is the amplitude of the IR light-induced hyperpolarization measured using the V m as the baseline, E rv is the reversal potential of the IR light-induced hyperpolarization measured in control saline without Ba 2+ , and R in is the input resistance. V m , ∆V h , and R in were measured separately before and after Ba 2+ application. Under this condition, Ba 2+ blocked 78% ± 6% (N = 4, p = 0.00079) of the g K underlying the IR light-induced hyperpolarization (Fig. 5(c)). As the Ba 2+ -sensitive TREK channels are outward rectifiers, the reversal potential of the hyperpolarization shifted to the right with Ba 2+ (Fig. 5(d)). Ba 2+ consistently right-shifted the reversal potential of the IR light-induced hyperpolarization by an average of 14.9 ± 1.68 mV (N = 4, p = 0.00303). The addition of 5 mM tetraethylammonium (TEA), a non-specific blocker of voltage-gated K + channels and Ca 2+ -activated big K + (BK) channels, did not result in a further block (Fig. 5(d)). Reversing the order of the application of Ba 2+ and TEA showed that TEA slightly reduced the amplitude of the hyperpolarization (Fig. S4), which may suggest a potential contribution from voltage-gated K + channels [41,42]. However, applying TEA first did not change the effectiveness of Ba 2+ to significantly suppress the IR light-induced hyperpolarization (Fig. S4).

Single infrared light pulses facilitate AP generation and disrupt rhythmic AP firing
We next investigated the functional implications of the IR light-induced depolarization as well as hyperpolarization. Five preparations (N = 5) were used without any TTX or ZD 7288 treatment for this purpose. For the crayfish motor axon and the optical parameters adapted here, the maximum depolarization amplitude obtained at resting state was 3.0 ± 0.16 mV (N = 8), which was relatively small. The crayfish motor axon generally requires relatively large depolarization (10-15 mV) in order to fire APs. Indeed, we did not observe axonal AP firing induced solely by IR light pulses under the constraints of sub-damaging IR light intensity. However, in one of the five preparations, single IR light pulses successfully facilitated axonal AP generation (red) when the axonal V m was depolarized by current injection just below the firing threshold, see Figs. 6(a) and 6(b). A delay in the onset of the AP firing was observed in both cases (Figs. 6(a) and 6(b)). The delay is likely a result from the interactions of the voltage-gated Na + and K + currents at near threshold membrane potentials, as well as the IR light-activated K + conductance. For the other four preparations, the IR light-induced depolarization, though not large enough to trigger AP firing, occurred during subthreshold current stimulation. On the other hand, when the axon was depolarized to a stable and rhythmic firing state, the application of a single IR light pulse reversibly disrupted the depolarizing trajectory between APs and suppressed AP firing during the time window of IR light induced hyperpolarization (Figs. 6(c) and 6(d)). The IR light-induced initial depolarization, though small, is visible as highlighted by the arrows. The reproducibility of the IR light-induced inhibition is illustrated in Fig. S5. The inhibition of AP firing was consistently observed with the five preparations tested. Collectively, the results demonstrate the functional relevance of the IR light induced depolarization and hyperpolarization in terms of AP firing.

Discussion
In this report, we used intracellular recordings from single axons to analyze the coexisting excitatory and inhibitory effects induced by single and brief IR light pulses. For excitation, we observed a TTX-independent depolarization whose amplitude was voltage-dependent (Fig. 3).
This depolarization was more pronounced for shorter IR light pulses of higher pulse power (Fig. 2). The depolarization could trigger APs when axons were near their firing threshold (Figs. 6(a) and 6(b)). For inhibition, we showed a K + channel-mediated hyperpolarization that followed the initial depolarization and persisted up to 100 ms (Figs. 4 and 5). This hyperpolarization interrupted rhythmic AP firing by blocking APs (Figs. 6(c) and 6(d)). Pharmacological analysis suggested that TREK channels partially contributed to the IR light-mediated hyperpolarization.
These findings are complementary to our earlier reports on the modulatory effects of IR light pulses on AP initiation and propagation and downstream post synaptic activities [22,43]. In those studies, only inhibition in axonal excitability was observed when IR light pulses with significant longer pulse duration (up to 500 ms) and lower pulse power

Infrared light-induced membrane depolarization and its voltage-dependency
We observed a maximum depolarization from a resting state of ∼3 mV induced by a single IR light pulse delivered by an optical fiber that illuminated a small part (∼200 µm) of the crayfish motor axons. For the same total energy deposition, the larger depolarization amplitude was achieved with shorter pulse duration and higher pulse power (Figs. 2(e) and 2(f)). The depolarization component was TTX-insensitive ( Fig. 3(a)), which agrees with the mechanism of IR light-induced membrane capacitance changes reported previously [19,[28][29][30]. The results suggest that short IR light pulses of high pulse power are more efficient in terms of stimulating neurons with IR light-induced capacitive currents and should be preferred when achieving INS.
We also presented that the depolarization exhibited voltage-dependence: the amplitude decreased when the membrane potential was depolarized from -120 mV to -40 mV (Fig. 3(d)). This general trend of voltage-dependence of the IR light-induced depolarization is consistent with results using other preparations such as oocytes and HEK293T cells [19], mouse hair cells [28], neuromuscular junction of Caenorhabditis elegans [29], and rat dorsal root ganglion neurons [57]. However, values of the reversal potential varied considerably in these previous studies. For instance, the reversal potentials of the depolarization in oocytes and HEK293T cells were above +100 mV, while the reversal potential of the depolarization in cultured ganglion neurons, which are more comparable to our motor axons, was around -40 mV. This variation in reversal potentials could potentially be due to differences in the intracellular and extracellular ionic compositions and in the composition of the lipid molecules and other factors that can affect the membrane charge distributions [19,29]. The crayfish motor axons preclude adequate space clamp and thus a quantitative estimate of the reversal potential of the depolarization current is not possible. However, our ex vivo data was the first example obtained from peripheral axons and further demonstrated that IR light pulses can induce voltage-sensitive depolarizing current, which is independent of major membrane channel currents.
Functionally, we showed that IR light-induced depolarizations successfully triggered APs only when the target axons were near their firing threshold (Figs. 6(a) and 6(b)). It highlights a "selectivity" of INS, namely that only axons and neurons hovering around firing threshold can be synchronized by IR light pulses. The efficacy of the IR light-induced AP initiation reported here is similar to a previous study using oocytes co-expressing voltage-gated Na + and K + channels. In that study, the membrane potential of the transfected oocytes was depolarized to be within 0.5-1 mV of their firing threshold in order for a single IR laser pulse to elicit an AP [19]. The large current step size (5 nA) we used in this study might have missed the membrane potential right below the firing threshold of target axons, thus led to a condition that was less favorable for observing IR light-evoked AP firing. While the small amplitudes of the depolarization seem limiting in its excitation capability, it is worth noting that only a small fraction of the target ex vivo crayfish motor axons and the transfected oocytes were illuminated. Expanding the area of illumination may increase the amplitude of the depolarization mediated by a capacitive current, though at the expense of decreased spatial selectivity. Moreover, functional consequences of the depolarization should be considered in the context that most mammalian central neurons are active spontaneously. Thus, a few millivolts depolarization can selectively stimulate cortical neurons that are already close to their firing threshold, while leaving other neurons unstimulated.

Infrared light-induced membrane hyperpolarization and the TREK channels
The photothermal heating generated by IR light pulses is expected to affect both the excitatory and the inhibitory components of the neurons and can thus induce opposite effects. In addition to the initial depolarizing current, a second hyperpolarization component was observed, which lasted up to 100 ms after the end of the IR light pulse. The amplitude and polarity of the hyperpolarization were found to be dependent on the membrane voltages (Figs. 4(a) and 4(b)). Our data suggests that g K underlies the hyperpolarization, as the changes in [K + ] o shifted the reversal potential of the hyperpolarization in the direction predicted by the Nernst equilibration (Figs. 4(c) and 4(d)). This hyperpolarization, when applied to the target axons with suprathreshold stimulation, reversibly suppressed AP firing (Figs. 6(c) and 6(d)).
Two-pore domain potassium channels like TREK channels are known to exist in single-celled eukaryotes and above [58]. However, the more specific TREK channel types present in crayfish are unknown due to lack of genomic data. Nevertheless, pharmacological data with barium is consistent with the presence of TREK-type channels in the crayfish motor axons. The finding that the application of Ba 2+ reduced 78% of the potassium conductance underlying the IR light-induced hyperpolarization strongly suggests that TREK-type K + channels contributed to the main component of the hyperpolarization. This partial reduction indicates that there were other players being activated by the IR light pulses. Given the diversity of ion channels expressed in different cell types, mechanisms underlying the hyperpolarization are likely to be cell specific [19,28,59]. In hair cells [28], for example, IR light-induced outward current was attributed to the voltage-activated delayed rectifier and BK-type channels activated by the IR light-evoked calcium influx and neurotransmitter release from efferent neurons. A previous study [60] using gold nanorods as photothermal transducers and near-infrared (785 nm) CW light as illumination source showed that TREK-1 activation contributed to thermally-mediated inhibition in cultured hippocampal neurons.
While effects of 5 mM Ba 2+ described in this study strongly suggest that TREK channels may mediate the IR-light induced hyperpolarization, Ba 2+ is known to also block other K + channels. We therefore also tested Spadin (Tocris), a TREK-1 specific blocker [56,61]. Spadin at 100 nM reduced the amplitude of the IR light-induced hyperpolarization at depolarized membrane potentials but to a smaller degree than 5 mM Ba 2+ (Fig. S6). We believe that there are at least two reasons that could explain the smaller effects observed with Spadin (see section S2 in Supplement 1 for further explanations). Nevertheless, the observed Spadin effects are consistent with the importance of TREK channels in INM.
It is worth noting that TREK channels are widely expressed in mammalian brain [62,63]. Our data demonstrates and highlights the IR light-induced inhibitory role of TREK channels, which should be taken into account when applying and assessing INM. On the one hand, when brief IR light pulses are applied to neurons expressing TREK channels, TREK channel-mediated inhibition could occur simultaneously with other inhibitory effects that are due to heat-dependence of Na + and K + channel kinetics or passive membrane properties. These effects can work together to inhibit AP initiation and propagation. The duration of the TREK-mediated inhibition, which can be up to 100 ms following the initial depolarization, also provides a useful guidance in the selection of the frequency of IR light pulses if repetitive excitation is the goal. It suggests that the depolarization resulting from subsequent light pulses may occur during the hyperpolarization period activated by the previous light pulses if the pulse frequency is significantly higher than 10 Hz. On the other hand, in neurons not expressing TREK channels, depolarization generated by individual light pulses can be more effective in exciting neurons.

Conclusion
In summary, this study reported and systematically examined for the first time the coexisting excitatory and inhibitory effects as well as their functional significance that can be induced by a single and brief IR light pulse on an individual motor axon. The cellular events observed with intracellular recordings and the fundamental mechanisms explored here are likely to be common to most neurons exposed to IR light irradiation. Our findings offer critical perspectives into the understanding and application of INM. They suggest that the INM outcomes are largely the combined results of INS and INI, which further depend on the physiological states of the target neurons and the IR light pulse illumination regimes. They highlight the importance of the collective evaluation of INS and INI and the knowledge of heat-sensitive ion channels expressed in target neurons in order to better interpret and control INM.

Disclosures. The authors declare no conflicts of interests.
Data availability. Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request. Supplemental document. See Supplement 1 for supporting content.