RESEARCH ARTICLE www.advancedscience.com Control of Brain State Transitions with a Photoswitchable Muscarinic Agonist Almudena Barbero-Castillo, Fabio Riefolo, Carlo Matera, Sara Caldas-Martínez, Pedro Mateos-Aparicio, Julia F. Weinert, Aida Garrido-Charles, Enrique Claro, Maria V. Sanchez-Vives,* and Pau Gorostiza* 1. Introduction The ability to control neural activity is essential for research not only in basic neuroscience, as spatiotemporal control of activity is a fundamental All perceptions, memories, and behaviors are based on the communication between experimental tool, but also in clinical neurology for therapeutic brain the billions of neurons that constitute the interventions. Transcranial-magnetic, ultrasound, and alternating/direct brain.[1–4] Individual neurons transmit in- current (AC/DC) stimulation are some available means of spatiotemporal formation using chemical and electrical sig- controlled neuromodulation. There is also light-mediated control, such as nals, and are organized in groups or cir- optogenetics, which has revolutionized neuroscience research, yet its clinical cuits involved in different functions.[3,5,6] translation is hampered by the need for gene manipulation. As a drug-based The electrochemical interactions of neu- ronal ensembles result in electrical activ- light-mediated control, the effect of a photoswitchable muscarinic agonist ity emerging from the brain, which can (Phthalimide-Azo-Iper (PAI)) on a brain network is evaluated in this study. show synchrony across populations in the First, the conditions to manipulate M2 muscarinic receptors with light in the form of brain rhythms and waves that experimental setup are determined. Next, physiological synchronous propagate[7–9] or asynchronous discharges, emergent cortical activity consisting of slow oscillations—as in slow wave depending on the brain state.[10] Differ- ent brain states (slow wave sleep, REM, sleep—is transformed into a higher frequency pattern in the cerebral cortex, wakefulness, anesthesia, etc.) are associated both in vitro and in vivo, as a consequence of PAI activation with light. These with specific network activity parameters results open the way to study cholinergic neuromodulation and to control such as brain rhythms of certain frequency spatiotemporal patterns of activity in different brain states, their transitions, and synchronization across areas or partic- and their links to cognition and behavior. The approach can be applied to ular functional connectivity patterns, while different organisms and does not require genetic manipulation, which would brain states are also associated with spe- cific behaviors.[1,11] Transitions across brain make it translational to humans. states are therefore linked to substantial changes in brain activity patterns.[12] For ex- ample, deep sleep is dominated by large and Dr. A. Barbero-Castillo, S. Caldas-Martínez, Dr. P. Mateos-Aparicio, Dr. F. Riefolo, Dr. C. Matera, Dr. A. Garrido-Charles, Prof. P. Gorostiza Dr. J. F. Weinert, Prof. M. V. Sanchez-Vives Network Biomedical Research Center in Bioengineering Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS) Biomaterials, and Nanomedicine (CIBER-BBN) Barcelona 08036, Spain Madrid 28029, Spain E-mail: msanche3@clinic.cat Dr. C. Matera Dr. F. Riefolo, Dr. C. Matera, Dr. A. Garrido-Charles, Prof. P. Gorostiza Department of Pharmaceutical Sciences Institute for Bioengineering of Catalonia (IBEC) University of Milan The Barcelona Institute for Science and Technology Milan 20133, Italy Barcelona 08028, Spain Prof. E. Claro E-mail: pau@icrea.cat Institut de Neurociències and Departament de Bioquímica i Biologia Molecular Unitat de Bioquímica de Medicina Universitat Autònoma de Barcelona (UAB) The ORCID identification number(s) for the author(s) of this article Barcelona 08193, Spain can be found under https://doi.org/10.1002/advs.202005027 Prof. M. V. Sanchez-Vives, Prof. P. Gorostiza © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH. Catalan Institution for Research and Advanced Studies (ICREA) This is an open access article under the terms of the Creative Commons Barcelona 08010, Spain Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. DOI: 10.1002/advs.202005027 Adv. Sci. 2021, 8, 2005027 2005027 (1 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com synchronous brain waves, which transition towards a more of spontaneous emerging slow oscillations (SO) in neuronal desynchronized and higher-frequency content with waking circuits. PAI cis-to-trans photoisomerization decreases the Down up.[8,13,14] These changes in brain and behavioral states and the state durations and increases oscillatory frequency in cortical concomitant changes in electrophysiological activity are physi- slices. In addition, PAI allows the reversible manipulation of the ologically driven largely by neuromodulators, of which acetyl- cortical oscillatory frequency in anesthetized mice using light. choline (ACh) is one.[12,15] Indeed, it is known that the activation Two different species, ferret (in vitro) and mouse (in vivo) were of cholinergic nuclei in the upper brainstem contributes to the used in this study, since ferret cortical slices are more suitable transition from sleep to awake states[16–18] and associated brain than those of rodents to spontaneously generating network patterns.[19] However, it is not fully understood how the different activity under physiological conditions.[38] Thus, we demon- cells expressing ACh receptors contribute to the alteration of the strate that photopharmacology allows us to selectively control global cortical state, or what the contribution of different mus- SO both in vitro and in vivo, opening the way for the analysis carinic receptor subtypes to this transition is. of their spatiotemporal dynamics and their effects on brain Cholinergic receptors include ionotropic nicotinic ion chan- and behavioral state transitions. Photoactivation of mAChR nels and muscarinic metabotropic G protein-coupled receptors. driven by PAI offers the potential to modulate neural circuits They modulate brain activity, and in particular the activity of in numerous regions distributed throughout the neocortex. In the cerebral cortex,[20] being involved in crucial neocortical func- this first work we studied the impact of global activation on the tions such as attention,[21–23] learning,[24–26] memory,[27] as well temporal dynamics of cortical patterns. Patterned illumination as sensory and motor functions.[28–30] In the neocortex, ACh is will allow us to carry out more spatially targeted studies in the released mostly at cholinergic terminals from neurons with so- future. mas in the basal forebrain nuclei. Electrical stimulation of the In our design, we considered that changes in cortical rhythms nucleus basalis can evoke the release of ACh in the neocortex, al- accompany behavioral state transitions, and endogenous ACh though in an unselective manner, as ascending projections from actions play a central role in such variations.[11,39–46] It is basal forebrain nuclei not only comprise cholinergic axons, but known[16–18] that ACh contributes to the shift of the neocor- also GABAergic and glutamatergic axons.[31] tical network state from synchronous to asynchronous activ- Selective stimulation of cholinergic projections in the neocor- ity (associated with awake states), thus we hypothesized that tex from basal forebrain nuclei has been demonstrated with op- a preparation expressing synchronized activity—physiological togenetics and has enabled the disruption of neocortical syn- slow oscillations—would be an appropriate testbed to investigate chronous activity during certain sleep states.[13] Optogenetic the effect of muscarinic ligands, including the novel photosensi- stimulation of basal forebrain cholinergic neurons also revealed tive agonist PAI. their influence in awake cortical dynamics: cholinergic neuro- modulation was found to be relevant for visual discrimination 2. Results tasks, modulating the encoding properties of V1 neurons and ac- tivating cortical transitions faster than previously presumed.[13] 2.1. Non-Specific Activation of mAChRs Evokes Neuronal However, optogenetics cannot target endogenous signaling as it Hyperexcitability in Cortical Slices is limited by the overexpression of microbial proteins using ge- netic manipulation, which can distort synaptic physiology.[32,33] We started by studying the effect of Iperoxo, a muscarinic non- It also raises safety and regulatory concerns regarding therapeu- selective agonist,[47,48] on cortical SO spontaneously generated in tic applications.[32] Illumination can be applied locally with im- cortical slices. The goal was to evaluate the potential of mAChRs planted devices like optical fibers and LEDs, or transcranially to modulate the dynamics of the neocortical network, while avoid- when using long wavelengths, which are less scattering and more ing the simultaneous activation of nicotinic cholinergic receptors penetrating in tissue.[34,35] (nAChRs).[20] In addition, these experiments would allow the val- The control of neuronal signaling with photopharmacology is idation of whether Iperoxo-based photoswitches like PAI could based on synthetic ligands that target endogenous proteins, and be useful photopharmacological tools to control neuronal activ- thus its physiological relevance spans from circuit to sub-cellular ity, and characterizing their performance. levels. Since neuronal receptors are highly conserved, photo- Ferret cortical slices spontaneously generate cortical SO,[38] switchable ligands can generally be used in multiple species, and a hallmark of activity during deep sleep or anesthesia.[38] We their safety and regulation can be established in the same man- recorded this spontaneous oscillatory activity (control) and then ner as other drugs. Given that the cholinergic system is key to the activity under different concentrations (1, 10, 100 × 10–9 the modulation of a variety of Central Nervous System (CNS) m; n = 7) of Iperoxo. Slow oscillations consist of alternating functions,[20,36] the use of selective and photoswitchable cholin- periods of activity or high neuronal firing (Up states) and ergic drugs to achieve a spatiotemporal modulation of cortical periods of near silence (Down states). In order to quantify activity might have relevant scientific and clinical implications. Up and Down state duration, firing rates and frequencies, we We recently developed a photoswitchable agonist of M2 mus- estimated the MUA (multiunit activity reflecting the activity carinic acetylcholine receptors (mAChRs), Phthalimide-Azo-Iper of the local neuronal population) and ran an algorithmic sep- (PAI),[37] but its action on neuronal circuits and networks has aration of Up and Down states by establishing a threshold never been explored. from a bimodal distribution of MUA that separates firing Herein, we report that the cholinergic-dependent brain periods (Up states) from silent periods (Down states).[7] A state transitions in the neocortex can be controlled and thus more detailed account of this quantification can be found in investigated in detail with light using photosensitive drugs. the Experimental Section. The activation of mAChRs by Iper- In particular, we have found that PAI enables the modulation oxo resulted in a global change in the network’s dynamics Adv. Sci. 2021, 8, 2005027 2005027 (2 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com Figure 1. Activation of mAChRs with non subtype-specific agonist Iperoxo (IPX) evokes neuronal hyperexcitability in cortical slices. A) On the left, the experimental setup: 16-channel multielectrode array (MEA); WM, white matter; L1-L6, layer 1–6. On the right, raw local field potential (LFP) traces illustrating network activity showing the increase in oscillatory frequency corresponding to the spectrogram of panel B with 100 × 10–9 m IPX. B) Spectrogram from the same time recording of LFP traces on panel A: control, 100 × 10–9 m and periods of seizure-like discharges. C) Oscillatory frequency (Hz) and FR (a.u.) during the Up-states in control (ctrl) conditions and after IPX (1, 10, 100 × 10–9 m; n = 7 ferret brain slices). Muscarinic activation of the brain slice with 100 × 10–9 m IPX produces a significant increase in oscillatory frequency compared to control conditions, and no significant changes in the firing rate (FR) during Up-states. Data are reported as mean ± standard error of mean (SEM). Analyses were performed with one-way ANOVA test (Brown–Forsythe and Welch test, unpaired t with Welch’s correction). *p-value <5 × 10−2 . D) Averaged power spectral density (PSD) of oscillatory activity showing low (<1 Hz), delta (1–4 Hz) and alpha (7–12 Hz) frequency component. in a dose-dependent way, displaying enhanced excitability (12–30 Hz) and gamma frequencies (30–100 Hz) (Figure 1D). (Figure 1A,B). At 100 × 10–9 m Iperoxo, the oscillatory However, our interest was to avoid epileptiform discharges and frequency—or frequency of the Up/Down state cycle— to activate muscarinic receptors within the range of physiological significantly increased (from 0.85 ± 0.1 Hz in the control to activity, therefore with the activation achieved below 100 × 10–9 1.42 ± 0.18 Hz with 100 × 10–9 m Iperoxo, p-value = 3 × 10−2 ), m Iperoxo. while the firing rate (FR; see Experimental Section)[49] during Up-states did not change significantly (from 0.97 ± 0.15 a.u. to 0.74 ± 0.09 a.u. with 100 × 10–9 m Iperoxo) (Figure 1C). At 2.2. Effect of PAI Isomers on Slow and Fast Oscillations In Vitro concentrations equal or higher than 100 × 10–9 m Iperoxo, the oscillatory activity evolved to periods of full-blown epileptiform The hyperexcitable network state induced with Iperoxo reflects discharges (Figure 1A,B), as previously characterized in this the impact of mAChR activation on cortical networks and brain preparation[49] and also described in vivo following muscarinic states.[14] In order to remotely control these states, we aimed at activation,[50,51] as in the pilocarpine model of epilepsy.[52] Such the muscarinic neuromodulation with light using PAI, a photo- epileptiform or seizure-like discharges are a clear sign of hy- switchable Iperoxo-derivative that allows the reversible activation perexcitability, displaying intense firing for several seconds and of M2 mAChRs in vivo.[37] The light-dependent behavior of PAI long periods of silence, and characterized by a rather different is achieved with a molecular switch in its structure that is based spectrogram showing larger amplitudes in different frequency on azobenzene. PAI exists in two forms, trans and cis, which are bands: slow (<1 Hz), delta (1–4 Hz), and alpha (7–12 Hz) in dynamic equilibrium with each other (Figure 2A). A distribu- frequency components, and including enhancement of beta tion of 87% in favor of the trans form (13% of the cis) is found in Adv. Sci. 2021, 8, 2005027 2005027 (3 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com the dark or after illumination with visible light (white light (WL) for 2 min). After illumination with ultraviolet (UV) light (365 nm for 1 min) the ratio between the two configurations rapidly changes to about 77% cis (23% trans). Both PAI isomeric mixtures (respectively termed “trans” and “cis” for simplicity) are thermally stable for hours and their biological effects have been character- ized in vitro and in vivo.[37] Despite the partial photoconversion (which is characteristic of azobenzene-based switches) the trans- and the cis-enriched forms of PAI display different pharmaco- logical activity: the trans configuration is a more potent activa- tor of M2 receptors than the cis, allowing the reversible manip- ulation of muscarinic functions with light over many cycles.[37] Prior to testing PAI in the brain, which shows high expression of both M1 and M2 receptors, we tested its activity profile in these two mAChR subtypes using in vitro GTPgammaS assays (Figure S1.2, Supporting Information) and calcium imaging as- says (Figure 2B) in cultured cell lines separately overexpressing M1 and M2 receptors. Our results show significant M2 selectiv- ity (Figure 2B) in agreement with the dualsteric design and in vivo properties of PAI.[37] M2 receptors are highly expressed in cardiac tissue and in the brain, where they play a relevant role in several CNS disorders.[36] Controlled spatiotemporal regulation of M2 mAChR activity and subsequent effects on cortical neu- ronal networks may provide new therapeutic opportunities for diseases involving the cholinergic system. Note that subtype se- lectivity can be advantageous to interpret (photo)pharmacological experiments but muscarinic activation with light is sufficient for the aim of this work. We applied PAI to spontaneously active neocortical brain slices and recorded their oscillatory activity be- fore and after photoactivating the drug. We first obtained dose- response curves of the two drug forms separately, trans- (dark- adapted state) and cis-PAI (pre-illuminated with UV light) with the purpose of identifying the concentration range displaying dif- ferences in brain activity with light. The baseline activity (charac- terized by SO) was recorded as a control, prior to bath-application of artificial cerebrospinal fluid (ACSF) with increasing PAI con- centrations (10 × 10–9 m, 100 × 10–9 m, 300 × 10–9 m, and 1 × 10–6 m, n = 6 ferret brain slices for each PAI form, trans and cis) (Figure 2C). Both trans-PAI and cis-PAI significantly modulated the Up- and Down-state sequence in a dose-dependent manner, with cis-PAI displaying weaker effects in agreement with the re- ported PAI properties in cells expressing M2 receptors[37] (Fig- ure 2C,D,E; the quantification and significance versus control for Figure 2. Effect on SO of mAChRs activation by trans-PAI (dark-relaxed) and cis-PAI (pre-illuminated with UV). A) Chemical structures of trans- and *p < 5 × 10−3 . C) Raw local field potential (LFP) example recordings show- cis-PAI. B) Subtype selectivity of trans-PAI (3, 10 × 10–12 m) was studied ing the different ability of trans- and cis-PAI to increasing the oscillatory by comparing the amplitude of fluorescence calcium imaging responses of frequency. Note that trans-PAI is a stronger agonist of M2 mAChR than cis- cells expressing M1 (3 × 10–12 m, n = 232 cells; 10 × 10–12 m, n = 232 cells; PAI.[37] D) Raster plots showing the FR during the Up-states (color coded) IPX, n = 258) or M2-GqTOP (3 × 10–12 m, n = 70 cells; 10 × 10–12 m, n = under control conditions and different trans- and cis-PAI concentrations. 200 cells; IPX, n = 300) mAChRs, using the procedure described in Riefolo E) oscillatory frequency (Hz) and FR during the Up-states (a.u.) of the et al. 2019[37] and by pre-incubating cells with OGB-1AM (10 × 10–6 m two different PAI isomers, trans- (blue, n = 6 ferret brain slices) and cis- for 30 min) as calcium indicator. M2 mAChR transfected cells gave a sig- PAI (pink, n = 6 ferret brain slices) at different concentrations. Data are nificantly higher response than M1 mAChR expressing cells (33% for M2 reported as mean ± SEM. Analyses were performed with multiple t-test compared to 8% for M1). Data were normalized over the maximum re- (Mann-Whitney). Significant differences between cis- and trans-PAI (**p- sponse obtained with the nonselective orthosteric agonist IPX in satura- value < 10−2 ) are observed in the oscillatory frequency at 100 × 10–9 m tion conditions at 30 × 10–12 m. These results in cell lines are in agreement and in the FR at 1 × 10–6 m. These experiments are aimed at estimat- with previous reports (Riefolo et al., 2019),[37] in which higher concen- ing the concentration of effective photoswitching of cortical oscillations, trations were required for cardiac assays in vivo. Bars are mean ± SEM which was set at 200 × 10–9 m in subsequent experiments (see Figures 3 and compared using a t-test of two samples assuming equal variances. and 4). Adv. Sci. 2021, 8, 2005027 2005027 (4 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com Figure 3. Photocontrol of brain waves in vitro using PAI and direct illumination with white light. A) Representative local field potential (LFP) traces (top) and raster plots of firing rate (FR) during the Up-states under control conditions, 200 × 10–9 m of cis-PAI and 200 × 10–9 m trans-PAI after photoconversion with white light (WL) (n = 17 ferret slices) (bottom). B) Representative spectrogram under control condition, 200 × 10–9 m of cis-PAI and 200 × 10–9 mtrans-PAI (WL). C) Oscillatory frequency (Hz) and FR during the Up-states (a.u.) at 200 × 10–9 m PAI after pre-illumination with 365 nm (cis-PAI), and photoswitching with WL (trans-PAI). Data are reported as mean ± SEM. Analyses were performed with one-way ANOVA test (Brown–Forsythe and Welch test, unpaired t with Welch’s correction). **p-value < 10–2 ; ***p-value < 10–3 , ****p-value < 10–4 . D) Averaged power spectral density (PSD) of oscillatory activity under control conditions, 200 × 10–9 m of cis-PAI and 200 × 10–9 mtrans-PAI after WL activation (color code). each isomer are detailed in Figure S2, Supporting Information). the control to 1.04 ± 0.14 Hz with cis-PAI, p-value = 2.8 × 10−3 ), The differences between cis-PAI and trans-PAI were significant and no significant effects in the FR of the Up-states (from 0.98 ± for the oscillatory frequency at 100 × 10–9 m and for the FR dur- 0.09 a.u. in the control to 0.86 ± 0.10 a.u. with cis-PAI, p-value = ing Up-states at 1 × 10–6 m (Figure 2E). Note that epileptiform 2.9 × 10−1 ) in comparison to the control situation (Figure 3A,C). activity was never observed with PAI up to concentrations of 1 × Subsequent direct illumination of the slices with WL produced 10–6 m, in contrast to the effect of Iperoxo. a robust increase in oscillatory frequency (from 0.53 ± 0.04 Hz These results obtained with a limited number of animals for in the control to 1.68 ± 0.13 upon illumination, p-value < 10−4 ; each isomer (n = 6) already suggested that intermediate concen- from 1.04 ± 0.14 Hz with cis-PAI to 1.68 ± 0.13 upon illumina- trations between 100 and 300 × 10–9 m might be suitable to pro- tion, p-value = 1.1 × 10−3 ), a significant decrease in FR of the duce changes in the oscillatory frequency and in the firing rate Up-states (from control values of 0.98 ± 0.09 a.u. to 0.52 ± 0.06 of the Up-states upon cis-trans photoisomerization (Figure 2E). a.u. upon illumination, p-value = 2 × 10−4 ; from 0.86 ± 0.10 a.u. Thus, in subsequent experiments with light we expanded the with cis-PAI to 0.52 ± 0.06 a.u. upon illumination, p-value = 2.2 sample (n = 17) and focused on the concentration of 200 × 10–9 × 10−3 ), and a noticeable change in the activity regime of the net- m in order to photomodulate cortical SO using PAI. work (Figure 3A,C). The changes in the power spectrum in the population were incremental, as illustrated in Figure 3B,D. These differences are in agreement with PAI photoconversion to the ac- 2.3. Photoswitchable PAI Effectively Modulates Cortical Slow tive form (trans). At this stage, the modulation of cortical activity Oscillations In Vitro was not reversible upon illumination with 365 nm light (to iso- merize PAI to the cis form in situ) in slices, which could be due Once the concentration range of drug to obtain different oscilla- either to pharmacological irreversibility (which was ruled out in tory activity evoked by cis- and trans-PAI was determined in vitro experiments in vivo, see next Section 2.4) or to the reduced pen- (Figure 2), we moved on to control the rhythmic activity with etration of UV light in brain tissue. light in cortical slices (Figure 3). We took advantage of the ther- mal stability of both PAI forms to apply initially the less potent 2.4. Photoswitchable PAI Can Reversibly Modulate Brain Wave one (cis-PAI) at 200 × 10–9 m in cortical slices (n = 17), in the Activity In Vivo absence of WL to avoid photoconversion to trans-PAI during the recordings.[37] As shown in Figure 3, 200 × 10–9 m cis-PAI evoked Having established the unique ability of PAI to alter cortical os- an increment of the oscillatory frequency (from 0.53 ± 0.04 Hz in cillatory activity with light in slices, we then aimed at obtaining Adv. Sci. 2021, 8, 2005027 2005027 (5 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com Figure 4. In vivo photomodulation of brain waves. A) Representative raw traces of local field potential (LFP) (top, in mV) and multiunit activity (bottom, in arbitrary units), showing the differences in oscillatory frequency and firing rate (FR) during the Up-states between the control, 1 x 10–6 m cis-PAI (pre-illuminated with 365 nm), and trans-PAI after photoswitching with white light (WL). B) Individual (left) and mean (right) quantification of oscil- latory frequency (Hz) (n = 8 mice), showing a significant increase upon illuminating cis-PAI with WL. Data are reported as mean ± SEM. Analyses were performed with one-way ANOVA test (repeated measures [RM], Geisser–Greenhouse correction—no sphericity—and uncorrected Fisher’s LSD). **p-value < 10–2 ; ***p-value < 10–3 , ****p-value < 10–4 . *p-value < 5 × 10−2 . C) The mean quantification of FR during the Up-states (a.u.) is not sig- nificantly affected by illumination of cis-PAI (n = 8). D) Averaged power spectral density (PSD) of oscillatory activity at different concentrations displays significant enhancement at delta, theta, alpha, and gamma frequency bands after WL activation (left). Mean ± SEM quantification of PSD from D-left panel (n = 8) (right). Analyses were performed with Friedman test and the Wilcoxon post-hoc tests corrected for multiple comparisons. *p-value < 5 × 10−2 . a proof of concept of photocontrolling the cortical state in vivo. delta (p = 1.6 × 10−2 ) and gamma (p = 2.3 × 10−2 ) frequencies Cortical activity was recorded from C57BL6/JR mice (n = 8) with band compared with control conditions (Figure 4D), while the an electrode inserted through a craniotomy across which we used FR during Up states was decreased to the control values (0.83 ± to carry out the drug application and brain illumination (see Ex- 0.09 a.u., Figure 4C). This light-dependent increase in oscillatory perimental Section). Initially, we induced deep anesthesia in the frequency activity of PAI in vivo is in partial agreement with the animals, a state that is known to reproduce the slow wave sleep in vitro results observed in ferret slices, which also displayed a state,[7,41,53] and which is characterized by the generation of corti- reduction in FR (Figure 3). The difference may be due to the in- cal SO similar to the slow frequency waves observed in our exper- tegrity of the neural network and the cholinergic activity in vivo. iments in slices under control conditions (Figures 1–3).[7] Such In brain cortical sections the effect of trans-PAI is not re- SO activity in anesthetized mice was recorded for 500 s under WL versible using 365 nm light and we aimed to understand whether illumination of the brain, and the characteristic parameters ob- this was due to muscarinic activation mechanisms or just to the tained (oscillatory frequency 0.60 ± 0.05 Hz, FR during the Up- limited tissue penetration of UV wavelengths, which could hin- states 0.80 ± 0.11 a.u.) were taken as the baseline, control con- der isomerization in deeper regions. The illumination of the mice dition in vivo. A 100 µL drop of 1 × 10–6 m cis-PAI solution was brain using a more powerful 365 nm source (100 W bulb, see initially applied to the brain surface in the absence of WL, to avoid Experimental Section) did not significantly reduce the oscilla- cis to trans photoisomerization of PAI. The oscillatory frequency tory frequency from the trans-PAI-excited situation (n = 4, Fig- was not significantly altered by cis-PAI (0.58 ± 0.06 Hz) (Figure ure S4A, Supporting Information). Since tissue scattering can- 4B), while it caused only a minor increase in the FR during the not be avoided at this wavelength we tried to revert the effect of Up-states (0.86 ± 0.14 a.u.) (Figure 4C). Subsequent illumination trans-PAI by outcompeting it by direct application of its cis iso- with WL significantly increased the oscillatory frequency (from mer, which binds to the same receptor site with similar strength 0.60 ± 0.05 Hz in the control to 0.73 ± 0.06 Hz upon illumina- that trans (see competition binding experiments, Figure S1.1, tion, p = 2.8 × 10−2 ) (Figure 4B), and induced an increment of the Supporting Information) but causes weaker or no activation. For Adv. Sci. 2021, 8, 2005027 2005027 (6 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com this purpose, we significantly increased the oscillatory frequency Figure 2, PAI was pre-illuminated in order to administer the in- with 200 × 10–9 m trans-PAI, and then applied 1 × 10–6 m cis- active cis form, which is then stable over hours at 37 °C. This PAI. Indeed, we observed the return to the control-like oscillatory allowed us to compare the activities of both forms at different activity (n = 4, Figure S4BC, Supporting Information), demon- concentrations and identifying 200 × 10–9 m PAI as a suitable strating that the process is pharmacologically reversible and that concentration for direct photoswitching of cortical activity in vitro the lack of reversibility is probably due to the limited penetra- (Figure 3) and in vivo (Figure 4). Overall, this proof-of-concept tion of UV light. Thus, the optimization of both the light delivery demonstration of photocontrol of brain waves has allowed us to system and/or drug photophysical properties (e.g., photoswitch- identify two important limitations of PAI that will be addressed ing at less scattering wavelengths) have the potential to reversibly in future drug designs. Namely, that PAI is active in the dark and photocontrol neural muscarinic actions in vivo. that it is not readily photo-reversible in brain tissue due to the limited penetration of UV light. We have promising evidence that 3. Discussion both limitations can be overcome, respectively using a “bridged” azobenzene to obtain a dark-inactive analog drug with equiva- Different brain states are associated with distinct brain emer- lent pharmacology,[64] and using tissue-penetrating infrared light gent patterns and behaviors. In order to establish strong causal from a pulsed laser.[37] Other aspects of PAI that can be opti- links between them, neural activity must be manipulated and mized include its activation wavelength under non-pulsed illu- recorded to observe the effects on specific behaviors. Pharma- mination, photosensitivity, reversibility, photoswitchable concen- cological tools have proven very useful but they affect brain tration range (“therapeutic window”), safety profile, and perme- waves in a systemic way and have relatively slow dynamics.[54] ability to the blood-brain barrier. Still, as a small molecule, PAI Thus, understanding the mechanisms of brain and behavioral is less likely to trigger adverse immune responses than the over- state transitions requires new techniques to manipulate neuronal expression of microbial opsins.[32,65] M2 mAChRs are involved activity[1] that enable the neural modulation of specific brain re- in several CNS diseases like major depressive[36,66] and bipolar gions and neuronal circuits in a fast and reversible way. They can disorders,[36,67] Parkinson’s[36,68] and Alzheimer’s[36,68] diseases, be defined by a multimodal stimulation approach (e.g., electrical but also in alcohol, smoking, and drug dependence.[36,69] These stimulation with implanted microelectrodes), photostimulation disorders are thus susceptible to drug-based photomodulation with cell-specific optogenetics[55–58] and with neurotransmitter- in vivo without requiring genetic manipulation. We envisage specific photopharmacology.[59] that once a suitable drug concentration has been determined Electromagnetic stimulation pioneered the noninvasive mod- for optimal photoswitching, spatiotemporal patterns of stimula- ulation of brain activity and is used therapeutically to treat CNS tion can be applied to ask specific questions. For example, how diseases.[60] For example, transcranial alternating current stimu- does a focal, acute activation propagate in space and time? What lation has been applied to modulate alpha and beta waves in the is the behavioral outcome? How does it interfere with the on- motor cortex[61] and transcranial magnetic stimulation has been going (endogenous) wave activity? How does the size, duration, used to modulate gamma oscillations in the prefrontal cortex.[62] and location of the activated region relate to the behavioral out- However, further improvements are required to enhance their come and its reversibility? Beyond the fundamental understand- spatiotemporal and spectral performance, both for fundamen- ing of the mechanisms of brain waves, if these experiments could tal research and therapeutic purposes.[60] Optogenetics[55–58] has be performed with safe photoswitchable drugs and using non- emerged as an alternative to electromagnetic stimulation, al- invasive transcranial illumination, these questions would gain lowing the activation or inhibition of specific cell populations. great clinical relevance and feasibility. Beyond photoswitchable For example, photocontrolling the release of ACh, which mod- muscarinic agonists, these concepts could be also exploited with ulates the transitions between different brain states,[12,15] can be drug-like antagonists (Riefolo et al, in review) and adrenergic achieved by overexpressing photosensitive proteins in choliner- ligands[70] among many others, eventually allowing us to build gic neurons of mice neocortex.[13,63] However, genetic manipu- 3D functional pharmacological charts that could be correlated lation is required in this approach, limiting so far its usability with anatomical maps. in humans.[32] The photopharmacological approach presented In summary, the manipulation of brain state transitions, by here is, to date, the only way to directly photomodulate brain means of photocontrolling the frequency of cortical oscillations, state transitions in intact tissue. We first studied the effect of has been achieved with a photoswitchable dualsteric agonist of the muscarinic agonist Iperoxo[47] on isolated cortical slices (Fig- M2 mAChRs. This result opens the way to: a) dissecting the ure 1) in order to demonstrate that a slow oscillatory state can be spatiotemporal distribution and pharmacology of brain states, controlled by selectively manipulating muscarinic receptors (in- namely mapping how they depend on agonists, antagonists, and cluding all M1–M5 subtypes) at their physiological location and modulators of the different muscarinic subtypes expressed in the context. The oscillatory frequency of the network was increased CNS; and b) investigating the neuronal dynamics and causality with 100 × 10–9 m Iperoxo, eventually leading to seizure-like dis- that regulate brain state transitions in the cerebral cortex and be- charges, in agreement with the outcome of muscarinic stimu- yond. In particular, two-photon stimulation of PAI using pulsed lation using knockout mice and the pilocarpine-induced model infrared light[37] should enable deep penetration and subcellular of epilepsy.[50,51] Photocontrol of muscarinic signaling was sub- resolution in 3D,[71] as recently demonstrated with endogenous sequently achieved in vitro and in vivo with the photochromic mGlu5. [72] Compared to the local and often inhomogeneous ex- Iperoxo derivative PAI[37] (Figures 2–4). We showed that PAI is pression patterns achieved with viral injections of optogenetic selective for M2 versus M1 subtype receptors and thus it only ac- constructs, diffusible small molecules like PAI can in principle be counts for part of the effects of Iperoxo. In the experiments of applied to larger brain regions to control neuronal oscillations.[73] Adv. Sci. 2021, 8, 2005027 2005027 (7 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com Thus, remote control of brain waves based on the photophar- cis form allows the administration of the less active drug (inactive in the macological manipulation of endogenous muscarinic receptors cerebral cortex at concentrations lower than 1 × 10–6 m) and subsequent may reveal the complex 3D molecular signaling underlying brain activation of M2 receptors in the target region with WL.[37] The authors first investigated the efficacy of PAI in cortical neuronal circuits in vitro by states and their transitions, in order to link them with cognition obtaining the dose-response curves of trans- and cis-PAI solutions applied and behavior in a diversity of wild-type organisms. separately. The more active PAI isomer (trans) was tested by applying its dark-adapted form (87% of trans-PAI), and cis-PAI was obtained by illumi- nating 1 mm stock solutions with 365 nm light (77% of cis-PAI, which is 4. Conclusion the maximum that can be achieved at the photo stationary state). Because of the high concentration of the stock solution (1 mm) and for the sake of A method for directly manipulating neural activity and brain having the maximal possible photoconversion of PAI into the cis form, UV rhythms with light-controlled drugs is demonstrated. It is based light irradiation (Vilber Lourmat Dual Wave Length UV Lamps, 365 nm 6 on a photoswitchable muscarinic small molecule and does not W) was performed over 10 min, even if it was demonstrated that the max- require gene therapy. The photocontrol of endogenous receptors imal percentage of the cis can be reached by a shorter time exposure to UV light (2 min).[37] Increasing concentrations of both trans- and cis-PAI and their functions in the central nervous system, such as the (10 × 10–9 m, 100 × 10–9 m, 300 × 10–9 m, and 1 × 10–6 m) were bath transition between different brain states, is an achievement for applied in order to build up the dose-response curves. neuromodulation technologies that is useful as a tool in basic LFP Recording and Data Analysis from In Vitro Recordings: The authors’ neuroscience research and in future brain therapies and stimu- objective in this study was to identify the modulation of network dynam- lation. Thus, photopharmacological neuromodulation combined ics exerted by photoswitchable muscarinic agonists. To this end, they ob- with implantable optoelectronic devices offers the ability to ex- tained multiple LFP recordings and their correspondent multiunit activity (MUA) in the way described below. No single units were identified since ploit the untapped potential of neuropharmacology, by control- they aimed at capturing the dynamics of the population and not of indi- ling drug action in precise spatiotemporal patterns. vidual neurons. The recordings started after allowing at least 2 h of re- covery of the slices. Extracellular recordings were obtained with flexible arrays of 16-electrodes arranged in columns as in Figure 1A. The multi- electrode array (MEA) covered a large part of the area occupied by a cor- 5. Experimental Section tical slice.[78] It consisted of six groups of electrodes positioned to record Slice Preparation: For the in vitro experiments the authors used iso- electrophysiological activity from superficial and from deep cortical lay- lated cortical slices from ferret, because they robustly reproduce the corti- ers (692 µm apart) and from what should correspond to three different cal SO compared with other animal models.[74] Twenty-seven ferrets (4- to cortical columns (1500 µm apart). The unfiltered field potential (raw sig- 6-months-old) were anesthetized with sodium pentobarbital (40 mg kg−1 ) nal) was acquired at 10 kHz with a Multichannel System Amplifier (MCS, and decapitated. The entire forebrain was rapidly removed and placed in Reutlingen, Germany) and digitized with a 1401 CED acquisition board oxygenated cold (4–10 °C) bathing medium.[75] Ferrets were treated in ac- and Spike2 software (Cambridge Electronic Design, Cambridge, UK). The cordance with protocols approved by the Animal Ethics Committee of the MUA was estimated from the power of the frequencies between 200 and University of Barcelona, which comply with the European Union guide- 1500 Hz in 5 ms windows.[7,49] The spectrum in this frequency band is lines on the protection of vertebrates used for experimentation (Directive a good estimate of the firing of the neuronal population, since it is pro- 2010/63/EU of the European Parliament and the Council of 22 September portional to the density of the Fourier components at high frequencies.[79] 2010). Coronal slices (400 µm thick) from primary visual cortex (V1) were The MUA signal values were logarithmically scaled to compensate for the used.[76] To increase tissue viability the authors used a modification of the high fluctuations in the firing of neurons that are very close to the elec- sucrose-substitution technique,[77] such that during slice preparation, the trode, thus obtaining the logMUA signal (which the authors refer to as tissue was placed in a solution in which NaCl was replaced with sucrose firing rate; FR). A bimodal distribution of the MUA the two peaks of the while maintaining the same osmolarity. After preparation, the slices were distribution corresponded to the samples of the activity network belonging placed in an interface-style recording chamber (Fine Sciences Tools, Foster to the Up and Down state, respectively. Thus, a threshold value separating City, CA, USA). During the first 30 min the cortical slices were superfused the two modes of the distribution was set between the two peaks, such that with an equal mixture in volume of the normal bathing medium, artificial samples belonged to the Up or to the Down states depending on their po- cerebral spinal fluid (ACSF), and the sucrose-substituted solution. Follow- sition with respect to this threshold. After Up and Down state detection, ing this, normal bathing medium was added up to the recording cham- mean Up and Down state durations were obtained.[7] The frequency of ber and the slices were superfused for 1–2 h; the normal bathing medium the SO was the inverse of the duration of the entire Up–Down cycle. The contained (in mm): NaCl, 126; KCl, 2.5; MgSO4 , 2; Na2 HPO4 , 1; CaCl2 , 2; Up state detection necessary was performed by setting a threshold in the NaHCO3 , 26; dextrose, 10; and was aerated with 95% O2 , 5% CO2 to a log(MUA) time series as previously described to quantify frequency of the final pH of 7.4. Then, a modified slice solution was used throughout the SO.[7,49] Firing rate (FR) of the Up states was quantified from the trans- rest of the experiment; it had the same ionic composition except for dif- formed log(MUA) signal as mean of absolute value of log(MUA). To study ferent levels of the following (in mm): KCl, 4; MgSO4 , 1; and CaCl2 , 1.[75] the variability of power spectral densities (PSD) of the local field potential Bath temperature was maintained at 34–36 °C. (LFP), the authors used Welch’s method with 50% overlapped Hamming Drug Application and Photostimulation in Brain Slices: Iperoxo and PAI, window with a resolution of 1 Hz. All off-line estimates and analyses were both prepared as previously reported from commercially available starting implemented in MATLAB (The MathWorks Inc., Natick, MA, USA). All vari- materials[37] were bath-applied at the concentrations range of 1 to 100 × ables in the experimental conditions were compared with the control (no 10–9 m for Iperoxo and 10 × 10–9 m to 1 × 10–6 m for PAI, as discussed chemical added) condition. in the Results section. The authors typically waited more than 1000 s after The In Vivo Preparation: Cortical electrophysiology experiments were the application of the drug in order to let it act, to obtain a stable pat- carried out in 2–3-month-old C57BL6/JR mice (n = 8) in accordance with tern of electrical activity, and to ensure a stable concentration in the bath, the European Union Directive 2010/63/EU and approved by the local this being an interface chamber. PAI effectively photomodulates the ac- ethics committee. Mice were kept under standard conditions (room tem- tivity of M2 receptors in vitro and in vivo: its dark-adapted state (trans perature, 12:12-h light-dark cycle, lights on at 08:00 a.m.). Anesthesia was form) behaves as a strong M2 agonist, then upon illumination with UV induced by intraperitoneal injection of ketamine (30 mg kg−1 ) and medeto- light (365 nm), PAI switches to its less active state (cis form). PAI can be midine (100 mg kg−1 ). After this procedure, the mouse was placed in a switched back to its “full on-state” with WL, or using two-photon excita- stereotaxic frame, and air was enriched with oxygen. Body temperature tion with pulsed near-infrared light.[37] The high thermal stability of PAI was maintained at 37 °C throughout the experiment.[7] A craniotomy was Adv. Sci. 2021, 8, 2005027 2005027 (8 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com performed in each mouse: AP −2.5 mm, L 1.5 mm (primary visual cor- rection) of GraphPad Prism 9. The analyses in brain slices between trans- tex, V1).[80] Cortical recordings were obtained from infragranular layers and cis-PAI oscillatory frequency and FR during the Up-state values were with 1–2 MΩ single tungsten electrode insulated with a plastic coating ex- performed with multiple t-test (Mann-Whitney) of GraphPad Prism 9. All cept for the tip (FHC, Bowdoin, ME, USA). Spontaneous LFP recordings the in vivo statistical analyses of oscillatory frequency and firing rate (FR) from the V1 area provided information about the local neuronal popula- during the Up-state were performed with one-way ANOVA test (repeated tion activity—within 250 µm.[81] MUA estimation, Up-state detection and measures [RM], Geisser–Greenhouse correction—no sphericity—and un- quantification of relative FR was performed as previously described. All corrected Fisher’s LSD) of GraphPad Prism 9. PSD statistical analysis in these parameters were used to compare spontaneous activity during anes- vivo was performed with Friedman test and the Wilcoxon post-hoc tests thesia (control), after application of the pre-illuminated less active drug corrected for multiple comparisons of MATLAB R2020b. Calcium imaging form (cis-PAI) and drug activation with WL (trans-PAI). Cis-PAI was locally analyses of 3 and 5 independent cell preparations expressing M1 and M2- delivered to the cerebral cortex surface and activity was recorded while ap- GqTOP mAChRs, respectively, were first normalized over the basal fluo- plying a commercial red filter on the WL source to avoid the activation of rescence of the cells. The fluorescence amplitudes of cellular PAI-induced the drug. The uncovered brain was illuminated from a 3 cm distance with calcium oscillations were normalized over IPX responses at the saturating a WL source (Photonic Optics Optics Cold Light Source LED F1) in order concentration of 30 × 10–12 m, reported as mean ± SEM, and compared to activate the drug in situ (trans-PAI). The trans-to-cis photoconversion using a t-test of two samples (assuming equal variances) using Origin Pro was performed by illuminating the brain from a 5 cm distance with a very 8.5. powerful 365 nm lamp (Spectroline, Spectronics FC-100-F, 230 V, 365 nm, 100 W Spot Bulb) in order to improve the penetration of light into the tis- sues. The electrophysiological signal was amplified with a multichannel Supporting Information system (Multi Channel Systems), digitized at 20 kHz with a CED acqui- Supporting Information is available from the Wiley Online Library or from sition board, and acquired with Spike 2 software (Cambridge Electronic the author. Design) unfiltered.[82] In Vitro Specific M1 and M2 Muscarinic Agonist Functional Assays: Sub- type selectivity of trans-PAI was studied by comparing the amplitude of Acknowledgements fluorescence calcium imaging responses of cells expressing M2-GqTOP and M1 mAChRs. Human M2-GqTOP mAChR was transfected as de- A.B.-C. and F.R. contributed equally to this work. The authors thank Miquel scribed in Riefolo et al. 2019.[37] Human M1 mAChR (Addgene) was trans- Bosch for comments on the manuscript, José Manuel Fernández and fected with X-tremeGENE 9 DNA Transfection Reagent (Roche Applied Álvaro Villarroel for useful discussions, Diana Casas and Jose Manuel Science) following the manufacturer’s instructions. The day after trans- Sánchez for collaborating during some in vivo and ex vivo experiments, fection, cells were harvested with accutase (Sigma-Aldrich) and seeded respectively. This research received funding from European Union Re- onto 16-mm glass coverslips (Fisher Scientific) pretreated with collagen search and Innovation Programme Horizon 2020 (Human Brain Project (Sigma-Aldrich) to allow cell adhesion. Preconfluent cultures were used SGA2 Grant Agreement 785907 and SGA3 No. 945539), DEEPER (ICT-36- for experiments between 48 and 72 h after transfection. In vitro single- 2020-101016787), European Research ERA-Net SynBio programme (Mod- cell calcium imaging was performed as described in Riefolo et al. 2019,[37] ulightor project), and financial support from Agency for Management of pre-incubating cells with OGB-1AM (10 × 10–6 m for 30 min) as calcium University and Research Grants/Generalitat de Catalunya (CERCA Pro- indicator. The subtype selectivity of 3 and 10 × 10–12 m trans-PAI emerged gramme; 2017-SGR-1442 project), Ministry of Economy and Competitive- by comparing the amplitude of fluorescence calcium imaging responses ness (MINECO)/FEDER (Grant CTQ2016- 80066-R to PG and BFU2017- of cells expressing M1 (3 × 10–12 m, n = 232 cells; 10 × 10–12 m, n = 232 85048-R to MVSV), the Fundaluce foundation, the Commission for Uni- cells; IPX, n = 258) or M2-GqTOP (3 × 10–12 m, n = 70 cells; 10 × 10–12 m, versities and Research of the Department of Innovation, Universities, and n = 200 cells; IPX, n = 300) mAChRs. Cells transfected with M2 mAChR Enterprise of the Generalitat de Catalunya -AGAUR- (IU16-011508) and co- gave significantly higher responses than M1 mAChR expressing cells (33% financed by the European Union Regional Development Fund within the for M2 compared to 8% for M1). Data were normalized over the maximum framework of the ERDF/FEDER Operational Program of Catalonia 2014– response obtained with the nonselective agonist Iperoxo (IPX) at saturat- 2020 with a grant of 50% of total eligible cost to PG and MVSV. ing concentration of 30 × 10–12 m (t-test of two samples assuming equal variances. p = 1.58 × 10−3 ). Conflict of Interest A preliminary comparison between mAChR subtypes was obtained us- ing an in vitro muscarinic agonist functional assay (Eurofins GTPgammaS The authors declare no conflict of interest. assay) which allows us to estimate possible activation of M1 and M2 re- ceptors (as expressed in the Eurofins specifications). In particular, differ- ent concentrations of the active isomer trans-PAI were tested on human Data Availability Statement recombinant M1 and M2 receptors expressed in CHO-K1 (Chinese ham- The data that support the findings of this study are available from the cor- ster ovary) cells. Compound agonism is defined by quantitation measured responding author upon reasonable request. of bound [35 S]GTP𝛾S. PAI was preincubated with 0.045 mg ml−1 mem- branes and 3 × 10–6 m GDP for 30 min at 30 °C in modified HEPES buffer pH 7.4 with SPA beads are added for another 60 min at 30 °C. The reac- Keywords tion is then initiated by the addition of 0.3 × 10–9 m [35 S]GTP𝛾S for 30 brain states, light-mediated control, muscarinic acetylcholine receptors, min. Trans-PAI-induced increase of [35 S]GTP𝛾S binding by 50% or more neuromodulation, photopharmacology (≥50%) relative to 1 × 10–9 m (M1) or 1 × 10–6 m (M2) oxotremorine M responses indicates possible M1 or M2 agonist activity, as indicated in the Eurofins assay specifications. Experiments were performed in duplicates Received: December 30, 2020 (n = 2) and were accepted in accordance with Eurofins Quality Control Revised: March 19, 2021 Unit’s validation standard operating procedure. Published online: May 21, 2021 Statistical Analysis: In vitro and in vivo oscillatory frequency, firing rate (FR) during the Up-state, PSD of oscillatory activity values are reported as mean ± SEM. All the oscillatory frequency and FR during the Up-state anal- yses in brain slices of Iperoxo (IPX), PAI dose-response curves, and 200 × [1] A. S. Andalman, V. M. Burns, M. Lovett-Barron, M. Broxton, B. Poole, 10–9 m PAI photoconversion experiments were performed with one-way S. J. Yang, L. Grosenick, T. N. Lerner, R. Chen, T. Benster, P. Mourrain, ANOVA test (Brown-Forsythe and Welch test, unpaired t with Welch’s cor- M. Levoy, K. Rajan, K. Deisseroth, Cell 2019, 177, 970.e20. Adv. Sci. 2021, 8, 2005027 2005027 (9 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com [2] S. Wang, O. Tudusciuc, A. N. Mamelak, I. B. Ross, R. Adolphs, U. [36] E. Scarr, CNS Neurosci. Ther. 2012, 18, 369. Rutishauser, Proc. Natl. Acad. Sci. USA 2014, 111, E3110. [37] F. Riefolo, C. Matera, A. Garrido-Charles, A. M. J. Gomila, R. Sortino, [3] D. Krehbiel, B. Bartel, M. Dirks, W. Wiens, Behav. Neural Biol. 1986, L. Agnetta, E. Claro, R. Masgrau, U. Holzgrabe, M. Batlle, M. Decker, 46, 30. E. Guasch, P. Gorostiza, J. Am. Chem. Soc. 2019, 141, 7628. [4] J. Obermayer, M. B. Verhoog, A. Luchicchi, H. D. Mansvelder, Front. [38] M. V. Sanchez-Vives, D. A. McCormick, Nat. Neurosci. 2000, 3, 1027. Neural Circuits 2017, 11, 100. [39] D. A. McCormick, J. Neurosci. 1992, 12, 278. [5] I. Hanin, Environ. Health Perspect. 1978, 26, 135. [40] M. Steriade, F. Amzica, A. Nunez, J. Neurophysiol. 1993, 70, 1385. [6] A. Peyrache, N. Dehghani, E. N. Eskandar, J. R. Madsen, W. S. An- [41] M. Steriade, Prog. Brain Res. 1993, 98, 345. derson, J. A. Donoghue, L. R. Hochberg, E. Halgren, S. S. Cash, A. [42] S. E. Jacobs, S. L. Juliano, J. Neurosci. 1995, 15, 1099. Destexhe, Proc. Natl. Acad. Sci. USA 2012, 109, 1731. [43] Z. Xiang, J. R. Huguenard, D. A. Prince, Science 1998, 281, 985. [7] M. Ruiz-Mejias, L. Ciria-Suarez, M. Mattia, M. V. Sanchez-Vives, J. [44] M. R. Picciotto, M. J. Higley, Y. S. Mineur, Neuron 2012, 76, 116. Neurophysiol. 2011, 106, 2910. [45] E. Zagha, D. A. McCormick, Curr. Opin. Neurobiol. 2014, 29, 178. [8] M. V. Sanchez-Vives, M. Mattia, Arch. Ital. Biol. 2014, 152, 147. [46] E. H. Buhl, G. Tamás, A. Fisahn, J. Physiol. 1998, 513, 117. [9] L. Muller, A. Reynaud, F. Chavane, A. Destexhe, Nat. Commun. 2014, [47] E. Barocelli, V. Ballabeni, S. Bertoni, C. Dallanoce, M. De Amici, C. De 5, 3675. Micheli, M. Impicciatore, Life Sci. 2000, 67, 717. [10] D. A. McCormick, D. B. Nestvogel, B. J. He, Annu. Rev. Neurosci. 2020, [48] R. Schrage, W. K. Seemann, J. Klöckner, C. Dallanoce, K. Racké, E. 43, 391. Kostenis, M. De Amici, U. Holzgrabe, K. Mohr, Br. J. Pharmacol. 2013, [11] D. Gervasoni, S. C. Lin, S. Ribeiro, E. S. Soares, J. Pantoja, M. A. L. 169, 357. Nicolelis, J. Neurosci. 2004, 24, 11137. [49] M. V. Sanchez-Vives, M. Mattia, A. Compte, M. Perez-Zabalza, [12] S.-H. Lee, Y. Dan, Neuron 2012, 76, 209. M. Winograd, V. F. Descalzo, R. Reig, J. Neurophysiol. 2010, 104, [13] L. Pinto, M. J. Goard, D. Estandian, M. Xu, A. C. Kwan, S.-H. Lee, T. 1314. C. Harrison, G. Feng, Y. Dan, Nat. Neurosci. 2013, 16, 1857. [50] F. Yi, E. DeCan, K. Stoll, E. Marceau, K. Deisseroth, J. J. Lawrence, [14] S. Ramaswamy, C. Colangelo, H. Markram, Front. Neural Circuits Epilepsia 2015, 56, 297. 2018, 12, 77. [51] S. E. Hamilton, M. D. Loose, M. Qi, A. I. Levey, B. Hille, G. S. McK- [15] M. E. Hasselmo, M. Sarter, Neuropsychopharmacology 2011, 36, 52. night, R. L. Idzerda, N. M. Nathanson, Proc. Natl. Acad. Sci. U. S. A. [16] M. Steriade, R. Curro Dossi, D. Pare, G. Oakson, Proc. Natl. Acad. Sci. 1997, 94, 13311. USA 1991, 88, 4396. [52] G. Curia, D. Longo, G. Biagini, R. S. G. Jones, M. Avoli, J. Neurosci. [17] M. Steriade, S. Datta, D. Paré, G. Oakson, R. C. Dossi, J. Neurosci. Methods 2008, 172, 143. 1990, 10, 2541. [53] M. Steriade, D. Contreras, R. Curró Dossi, A. Nuñez, J. Neurosci. 1993, [18] M. H. J. Munk, P. R. Roelfsema, P. König, A. K. Engel, W. Singer, Sci- 13, 3284. ence 1996, 272, 271. [54] M. Koch, C. Schmiedt-Fehr, B. Mathes, Int. J. Psychophysiol. 2016, 103, [19] M. Xu, S. Chung, S. Zhang, P. Zhong, C. Ma, W. C. Chang, B. Weiss- 62. bourd, N. Sakai, L. Luo, S. Nishino, Y. Dan, Nat. Neurosci. 2015, 18, [55] J. A. Cardin, M. Carlén, K. Meletis, U. Knoblich, F. Zhang, K. Deis- 1641. seroth, L.-H. Tsai, C. I. Moore, Nature 2009, 459, 663. [20] Y. A. Hay, P. Jarzebowski, Y. Zhang, R. Digby, V. Brendel, O. Paulsen, [56] V. S. Sohal, F. Zhang, O. Yizhar, K. Deisseroth, Nature 2009, 459, V. Magloire, Eur. J. Neurosci. 2020, 53, 1378. 698. [21] G. Buzsàki, F. H. Gage, in Central Cholinergic Synaptic Transmission. [57] H. Adesnik, A. Naka, Neuron 2018, 100, 1028. Experientia Supplementum, Vol. 57 (Eds: M. Frotscher, U. Misgeld), [58] A. Naka, J. Veit, B. Shababo, R. K. Chance, D. Risso, D. Stafford, B. Birkhäuser, Basel 1989. https://doi.org/10.1007/978-3-0348-9138- Snyder, A. Egladyous, D. Chu, S. Sridharan, D. P. Mossing, L. Panin- 7_16 ski, J. Ngai, H. Adesnik, eLife 2019, 8, e43696. [22] J. L. Herrero, M. J. Roberts, L. S. Delicato, M. A. Gieselmann, P. [59] K. Hüll, J. Morstein, D. Trauner, Chem. Rev. 2018, 118, 10710. Dayan, A. Thiele, Nature 2008, 454, 1110. [60] W. Paulus, e-Neuroforum 2014, 5, 29. [23] A. Thiele, Annu. Rev. Neurosci. 2013, 36, 271. [61] B. Pollok, A.-C. Boysen, V. Krause, Behav. Brain Res. 2015, 293, 234. [24] J. M. Conner, A. Culberson, C. Packowski, A. A. Chiba, M. H. Tuszyn- [62] J.-P. Lefaucheur, N. André-Obadia, A. Antal, S. S. Ayache, C. Baeken, ski, Neuron 2003, 38, 819. D. H. Benninger, R. M. Cantello, M. Cincotta, M. de Carvalho, D. De [25] M. Kilgard, Neuron 2003, 38, 678. Ridder, H. Devanne, V. Di Lazzaro, S. R. Filipović, F. C. Hummel, S. [26] D. Ramanathan, M. H. Tuszynski, J. M. Conner, J. Neurosci. 2009, 29, K. Jääskeläinen, V. K. Kimiskidis, G. Koch, B. Langguth, T. Nyffeler, A. 5992. Oliviero, F. Padberg, E. Poulet, S. Rossi, P. M. Rossini, J. C. Rothwell, [27] J. Winkler, S. T. Suhr, F. H. Gage, L. J. Thal, L. J. Fisher, Nature 1995, C. Schönfeldt-Lecuona, H. R. Siebner, C. W. Slotema, C. J. Stagg, J. 375, 484. Valls-Sole, U. Ziemann, W. Paulus, L. Garcia-Larrea, Clin. Neurophys- [28] R. W. Berg, B. Friedman, L. F. Schroeder, D. Kleinfeld, J. Neurophysiol. iol. 2014, 125, 2150. 2005, 94, 699. [63] S. Zucca, V. Pasquale, P. L. de Leon Roig, S. Panzeri, T. Fellin, Curr. [29] M. P. Kilgard, M. M. Merzenich, Science 1998, 279, 1714. Biol. 2019, 29, 1481.e6. [30] A. A. Disney, C. Aoki, M. J. Hawken, Neuron 2007, 56, 701. [64] G. Cabré, A. Garrido-Charles, À. González-Lafont, W. Moormann, D. [31] P. Henny, B. E. Jones, Eur. J. Neurosci. 2008, 27, 654. Langbehn, D. Egea, J. M. Lluch, R. Herges, R. Alibés, F. Busqué, P. [32] B. D. Allen, A. C. Singer, E. S. Boyden, Learn. Mem. 2015, 22, 232. Gorostiza, J. Hernando, Org. Lett. 2019, 21, 3780. [33] G. Miesenböck, Annu. Rev. Cell Dev. Biol. 2011, 27, 731. [65] H. Tomita, E. Sugano, H. Isago, M. Tamai, J. Genet. 2009, 88, 409. [34] A. S. Chuong, M. L. Miri, V. Busskamp, G. A. C. Matthews, L. C. Acker, [66] A. S. Gibbons, E. Scarr, C. McLean, S. Sundram, B. Dean, J. Affective A. T. Sørensen, A. Young, N. C. Klapoetke, M. A. Henninger, S. B. Ko- Disord. 2009, 116, 184. dandaramaiah, M. Ogawa, S. B. Ramanlal, R. C. Bandler, B. D. Allen, [67] D. M. Cannon, R. E. Carson, A. C. Nugent, W. C. Eckelman, D. O. C. R. Forest, B. Y. Chow, X. Han, Y. Lin, K. M. Tye, B. Roska, J. A. Kiesewetter, J. Williams, D. Rollis, M. Drevets, S. Gandhi, G. Solorio, Cardin, E. S. Boyden, Nat. Neurosci. 2014, 17, 1123. W. C. Drevets, Arch. Gen. Psychiatry 2006, 63, 741. [35] R. Chen, F. Gore, Q. A. Nguyen, C. Ramakrishnan, S. Patel, S. H. Kim, [68] M. A. Piggott, J. Owens, J. O’Brien, S. Colloby, J. Fenwick, D. Wyper, E. M. Raffiee, Y. S. Kim, B. Hsueh, E. Krook-Magnusson, I. Soltesz, K. Jaros, M. Johnson, R. H. Perry, E. K. Perry, J. Chem. Neuroanat. 2003, Deisseroth, Nat. Biotechnol. 2020, 39, 161. 25, 161. Adv. Sci. 2021, 8, 2005027 2005027 (10 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH www.advancedsciencenews.com www.advancedscience.com [69] D. M. Dick, A. Agrawal, J. C. Wang, A. Hinrichs, S. Bertelsen, K. K. Bu- [74] M. V. Sanchez-Vives, R. Reig, M. Winograd, V. F. Descalzo, Res. Sign- cholz, M. Schuckit, J. Kramer, J. Nurnberger, J. Tischfield, H. J. Eden- post 2007, 37, 27. berg, A. Goate, L. J. Bierut, Addiction 2007, 102, 1131. [75] M. Mattia, M. V. Sanchez-Vives, Cognit. Neurodyn. 2012, 6, 239. [70] D. Prischich, A. M. J. Gomila, S. Milla-Navarro, G. Sangüesa, R. Diez- [76] L. S. Krimer, P. S. Goldman-Rakic, J. Neurosci. 2001, 21, 3788. Alarcia, B. Preda, C. Matera, M. Batlle, L. Ramírez, E. Giralt, J. Her- [77] G. K. Aghajanian, K. Rasmussen, Synapse 1989, 3, 331. nando, E. Guasch, J. J. Meana, P. Villa, P. Gorostiza, Angew. Chem., [78] M. D’Andola, B. Rebollo, A. G. Casali, J. F. Weinert, A. Pigorini, R. Villa, Int. Ed. 2021, 60, 3625. M. Massimini, M. V. Sanchez-Vives, Cereb. Cortex 2018, 28, 2233. [71] M. Wenzel, J. P. Hamm, D. S. Peterka, R. Yuste, J. Neurosci. 2019, 39, [79] M. Mattia, P. Del Giudice, Phys. Rev. E 2002, 66, 051917. 8562. [80] M. G. Golmohammadi, D. G. Blackmore, B. Large, H. Azari, E. Esfan- [72] S. Pittolo, H. Lee, A. Lladó, S. Tosi, M. Bosch, L. Bardia, X. Gómez- diary, G. Paxinos, K. B. J. Franklin, B. A. Reynolds, R. L. Rietze, Stem Santacana, A. Llebaria, E. Soriano, J. Colombelli, K. E. Poskanzer, G. Cells 2008, 26, 979. Perea, P. Gorostiza, Proc. Natl. Acad. Sci. U. S. A. 2019, 116, 13680. [81] S. Katzner, I. Nauhaus, A. Benucci, V. Bonin, D. L. Ringach, M. Caran- [73] S. Zucca, G. D’Urso, V. Pasquale, D. Vecchia, G. Pica, S. Bovetti, C. dini, Neuron 2009, 61, 35. Moretti, S. Varani, M. Molano-Mazón, M. Chiappalone, S. Panzeri, T. [82] P. Castano-Prat, M. Perez-Zabalza, L. Perez-Mendez, R. M. Escori- Fellin, eLife 2017, 6, e26177. huela, M. V. Sanchez-Vives, Front. Aging Neurosci. 2017, 9, 141. Adv. Sci. 2021, 8, 2005027 2005027 (11 of 11) © 2021 The Authors. Advanced Science published by Wiley-VCH GmbH