GABAERGIC INHIBITION AS A DYNAMIC ORGANIZER OF CORTICAL ACTIVITY EDRIS REZAEI Master of Science, National Institute of Genetic Engineering and Biotechnology, 2018 A thesis submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY in Neuroscience Department of Neuroscience University of Lethbridge LETHBRIDGE, ALBERTA, CANADA © Edris Rezaei, 2025 GABAERGIC INHIBITION AS A DYNAMIC ORGANIZER OF CORTICAL ACTIVITY Edris Rezaei Date of defense: 01/13/2026 Ian Q. Whishaw Professor Ph.D. Supervisor Robert Sutherland Professor Ph.D. Thesis Examination Committee Member Artur Luczak Professor Ph.D. Thesis Examination Committee Member Dustin Hines Associate Professor Ph.D. External Examiner University of Nevada, Las Vegas Robbin Gibb Chair, Thesis Examination Professor Ph.D. Committee III Dedication To my parents, for their unwavering support and sacrifices; to my wife, for her patience and love throughout this journey; and to my daughter, Lena, for bringing hope and happiness into my life. IV Abstract Most studies of cortical function have focused on excitatory neurons and their circuit dynamics, while inhibitory GABAergic dynamics has been examined primarily at the level of local microcircuits. As a result, GABAergic dynamics at the mesoscale and large-scale network level remain poorly understood. To close this gap, mesoscale wide-field imaging employing the genetically encoded GABA sensor iGABASnFR2 was employed to resolve cortical inhibitory dynamics during sensory processing and hippocampal–cortical interactions. During sensory stimulation, GABA release was evoked across multiple modalities, including whisker, forelimb, hindlimb, and visual inputs. Inhibitory responses were spatially specific and localized to the appropriate primary sensory cortices, with stronger activation in the contralateral hemisphere. Inhibitory dynamics appeared to differ with brain state, with quiet wakefulness associated with faster, stronger, and more widespread GABA responses than anesthesia, while the spatial organization of sensory maps was preserved. Increasing extracellular GABA with tiagabine was accompanied by a loss of detectable sensory-evoked GABA responses and changes in large- scale cortical connectivity. Examining cortical GABA dynamics around hippocampal sharp-wave ripples (SWRs) revealed large-scale, coordinated inhibitory activation across the cortex. During NREM sleep, GABA activation emerged earlier and propagated from medial to lateral cortical regions, whereas during wakefulness, activation followed the ripple and propagated from lateral to medial cortex. In both states, a global increase in inhibitory tone was observed, but with distinct spatiotemporal organization. The spatial distribution and timing of these components differed between wakefulness and NREM sleep, indicating that hippocampal ripples recruit distinct inhibitory network depending on behavioral state. These results identified cortical GABA signaling as a large-scale network process involved in sensory processing and hippocampal–cortical communication. V Preface This dissertation is original, and all methods and interventions were approved by the University of Lethbridge’s Animal Care Committee. The work presented in Chapters 2 and 3 was conducted in Dr. Mohajerani’s laboratory at the Canadian Centre for Behavioural Neuroscience. The research in Chapter 2 of this dissertation has been published in Neurophotonics under the title: “Characterization of iGABASnFR2 for in vivo mesoscale imaging of intracortical GABA dynamics” (DOI: 10.1117/1.NPh.12.3.035006) Authors Edris Rezaei¹, Setare Tohidi¹, Mojtaba Nazari¹, Javad Karimi Abadchi² ¹ University of Lethbridge (Canada) ² McGill University (Canada) Author Contributions E. R. conceptualized the study, designed and conducted the main experiments, prepared all figures, and wrote and revised the paper. M. N. contributed to the additional experimental procedures. J. K. A. implemented the strobing imaging system. E. R. and S. T. analyzed the data. VI Chapter 3, titled “Cortical GABAergic inhibition dynamics around hippocampal sharp- wave ripples”, has received comments from reviewers (October 2025), and the revision is ready for submission to Nature Communications Biology. A preprint of this work is available on bioRxiv (DOI: 10.1101/2025.08.17.670701). Authors Edris Rezaei¹, Setare Tohidi¹ ¹ University of Lethbridge (Canada) Author Contributions E. R. conceptualized the study, designed and conducted the main experiments, prepared all figures, and wrote and revised the paper. E. R. and S. T. analyzed the data. VII Chapter 4, titled “GABAergic inhibition as a dynamic organizer of cortical activity”, will be submitted. Authors Edris Rezaei¹ ¹ University of Lethbridge (Canada) Author Contributions E. R. conceptualized the study, prepared all figures, and wrote the manuscript. VIII Ethics Work described in this thesis received research ethics approval from the University of Lethbridge Research Ethics Board under protocol 2209, Project Name “Project Area I: In vivo assessment of subcortical-cortical interactions”. IX AI Generative I would like to acknowledge the use of Grammarly, which provided suggestions for rephrasing, restructuring sentences, and enhancing paragraphs by recommending additional words or alternative expressions. X Acknowledgment I would like to sincerely thank the members of my thesis examination committee for their time, expertise, and valuable feedback: Dr. Robert James Sutherland and Dr. Artur Luczak as Thesis Examination Committee Members, Dr. Dustin Hines as the External Examiner from the University of Nevada, Las Vegas, and Dr. Robbin Gibb as Chair of the Thesis Examination Committee. My deepest thanks go to my supervisor, Dr. Ian Whishaw, for his unwavering guidance, encouragement, and mentorship, and for helping me grow into a true scientist. XI Table of contents 1 General Introduction 1.1 The Concept of Inhibition ...... 1 1.2 The Physiological Discovery of Inhibition ...... 2 1.3 Inhibition in Behavior, Learning, and Brain ...... 3 1.4 Experimental Description of Neural Inhibition ...... 4 1.5 Investigating Inhibition through Hippocampal Lesions ...... 6 1.6 GABA Neurotransmitter ...... 7 1.6.1 GABA Receptors ...... 7 1.6.1.1 GABA_A Receptor ...... 7 1.6.1.2 GABA_B Receptor ...... 8 1.6.1.3 GABA_C Receptors ...... 8 1.7 How Inhibition Shapes Cortical Activity ...... 9 1.7.1 Inhibition Shapes Tuning ...... 9 1.7.2 Inhibition and Gain Control ...... 10 1.7.3 Inhibition and Oscillations ...... 10 1.8 Glycine ...... 11 1.8.1 Glycine Receptors ...... 12 1.9 Other Neurotransmitters and Neuromodulators with Inhibitory Effects ...... 12 1.10 Conceptual Framework and Research Hypotheses ...... 13 2 Characterization of iGABASnFR2 for in vivo Mesoscale Imaging of Intracortical GABA Dynamics 2.1 Introduction ...... 15 2.2 Materials and Methods ...... 16 2.2.1 Animal Subjects ...... 16 2.2.2 Viral Constructs ...... 16 2.2.3 Retro-Orbital Injection ...... 17 2.2.4 Drug Administration ...... 17 2.2.5 Surgical Procedure and Post-Operative Care ...... 17 2.2.6 iGABASnFR2 Imaging Under Anesthesia ...... 18 2.2.7 Sensory Stimulation ...... 21 2.2.8 Habituation ...... 22 2.2.9 iGABASnFR2 Imaging During Wakefulness ….. 22 2.2.10 iGABASnFR2 Imaging During Sleep ...... 22 2.2.11 Preprocessing ...... 23 2.2.12 ROI-based Fluorescence Analysis ...... 23 2.2.13 Motion Detection and Exclusion ...... 24 XII 2.2.14 Seed-Pixel Correlation Analysis ...... 24 2.2.15 Motion Signal Extraction, Alignment, and Sleep Scoring ...... 25 2.2.16 Effect of Isoflurane Anesthesia on Extracellular GABA Dynamics ...... 25 2.2.17 Statistical Analysis ...... 26 2.3 Results ...... 26 2.3.1 Experimental Workflow, Imaging Setup, and Cortical Expression of iGABASnFR2 ...... 26 2.3.2 iGABASnFR2 Reveals Modality- and Hemisphere- Specific Cortical Inhibition ...... 27 2.3.3 Sensory-Evoked and Spontaneous GABA Activity in Quiet Wakefulness Resembles Anesthesia-Induced Patterns ...... 30 2.3.4 Mesoscale Imaging of Cortical GABA Dynamics During Natural Sleep and Wakefulness ...... 33 2.3.5 Intracortical Long-Range GABAergic Correlations Revealed by Seed-Pixel Analysis Across Brain States ...... 34 2.3.6 Tiagabine Elevates Baseline GABA Levels but Dampens Sensory-Evoked Responses and Reorganizes Cortical Inhibitory Connectivity ...... 36 2.4 Discussion ...... 40 2.4.1 Interpreting iGABASnFR2 Signals ...... 41 2.4.2 Cortical GABA Responses to Sensory Input Are Conserved Across Brain States ...... 41 2.4.3 Spontaneous GABA Dynamics Reveal State-Dependent Connectivity ...... 42 2.4.4 Functional Inhibitory Architecture Aligns with Cortical Structural Organization ...... 42 2.4.5 Tiagabine Elevates Extracellular GABA and Disrupts Sensory- Evoked Responses via Altered Cortical Synchrony ...... 43 2.5 Conclusion ...... 44 3 Cortical GABAergic Inhibition Dynamics Around Hippocampal Sharp-Wave Ripples 3.1 Introduction ...... 53 3.2 Materials and Methods ...... 56 3.2.1 Animals and Experimental Conditions ...... 56 3.2.2 Viral Constructs and Retro-Orbital Injection Procedure ...... 56 3.2.3 Surgical Procedure and Electrophysiological Recording Setup ...... 57 3.2.4 Habituation for Head-Restraint Sleep and Wakefulness ...... 57 XIII Experiments 3.2.5 GABA Imaging During Wakefulness ...... 58 3.2.6 GABA Imaging During Natural Sleep ...... 58 3.2.7 GABA Imaging ...... 58 3.2.8 Preprocessing ...... 59 3.2.9 Sleep Scoring ...... 59 3.2.10 SWR Detection ...... 60 3.2.11 Normalization of Peri-SWR Neocortical Activity Using Z- Scoring ...... 60 3.2.12 Statistical Analysis ...... 61 3.2.13 Data Availability ...... 61 3.3 Results ...... 61 3.3.1 Experimental Protocol for Investigating Neocortical GABA Dynamics ...... 61 3.3.2 Cortical GABA Dynamics Across Sleep-Wake State Transitions ...... 63 3.3.2.1 NREM to Awake Transition ...... 63 3.3.2.2 Wake to NREM Transition ...... 64 3.3.2.3 NREM to REM Transition ...... 66 3.3.2.4 REM to Wake Transition ...... 66 3.3.3 State-Dependent Spatiotemporal Patterns of Cortical Inhibition Around SWRs ...... 68 3.3.4 Temporal Mapping of Cortical GABA Peaks Reveals Brain- State-Dependent Propagation During SWRs ...... 72 3.3.5 Spatial and Temporal Modes of Peri-Ripple iGABASnFR2 Activity and Region-Specific Dynamics ...... 75 3.4 Discussion ...... 77 3.4.1 Summary of the Study ...... 77 3.4.2 Cortical GABA Dynamics During State Transitions ...... 78 3.4.3 Ripple-Triggered GABA Responses Are Brain-State Dependent ...... 79 3.4.4 Temporal Gradients Reveal Propagation of Inhibitory Signals ...... 80 3.4.5 SVD Reveals Global vs. Local Components of Ripple- Evoked Inhibition ...... 80 3.5 Conclusion ...... 81 4 GABAergic inhibitions as a dynamic organizer of cortical activity 4.1 Reframing Cortical Inhibition ...... 87 4.2 Fast and slow inhibitory motifs define sensory processing ...... 90 4.2.1 Circuit mechanism ...... 90 4.3 Ramping inhibition coordinates cortical state transitions ...... 92 XIV 4.3.1 Circuit mechanism ...... 92 4.4 Two-phase inhibition during NREM sleep: gating internal replay ...... 95 4.4.1 Circuit mechanism ...... 97 4.4.1.1 The pre-ripple gate ...... 97 4.4.1.2 The ripple-timed inhibition ...... 97 4.5 Internal cortical inhibition during wakefulness: suppressed pre-ripple and time-locked post-ripple dynamic ...... 99 4.5.1 Circuit mechanism ...... 100 4.5.1.1 Pre-ripple phase: loss of anticipatory inhibition ...... 100 4.5.1.2 Ripple-timed phase: localized, propagating inhibition ...... 100 4.6 Concluding remarks and future perspectives ...... 102 XV List of figures Figure 2.3.1 Schematic of experimental workflow, imaging setup, and expression of iGABASnFR2. 20 Figure 2.3.2 Sensory-evoked GABAergic responses in the neocortex measured by iGABASnFR2 imaging. 28 Figure 2.3.3 Sensory-evoked and spontaneous GABA activity in quiet wakefulness resembles patterns observed under anesthesia. 32 Figure 2.3.4 Combined electrophysiological recording and mesoscale iGABASnFR2 imaging of GABA activity during wakefulness and sleep. 35 Figure 2.3.5 Brain state–dependent patterns of extracellular GABA dynamics measured by iGABASnFR2. 37 Figure 2.3.6 Effect of tiagabine administration on extracellular GABA activity in mice expressing iGABASnFR2. 39 Supplementary Figure 1 Imaging acquisition timing and sensory stimulation protocol. 45 Supplementary Figure 2 Sensory-evoked and spontaneous cortical responses under anesthesia using cpSFGFP as a control for iGABASnFR2. 46 Supplementary Figure 3 Auditory-evoked cortical GABA responses under anesthesia measured with iGABASnFR2. 47 Supplementary Figure 4 Intrahemispheric connectivity: Region-based sensory-evoked correlation maps. 48 Supplementary Figure 5 Motion signal, hippocampal LFP activity, and EMG power for sleep scoring in a head-fixed mouse. 50 Figure 3.3.1 Experimental protocol for imaging neocortical GABA dynamics during sleep and hippocampal ripples. 62 Figure 3.3.2.1 Neocortical GABA activity during NREM is used to awaken and awaken NREM transitions. 65 Figure 3.3.2.3 Neocortical GABA Activity During NREM to REM and REM to Awake Transitions. 67 XVI Figure 3.3.3 Peri-SWR neocortical GABA activity: activation and deactivation patterns across sleep and wake states. 71 Figure 3.3.4. Cortical regions exhibit state-dependent peak GABA activation dynamics around SWRs during natural sleep and wakefulness. 74 Figure 3.3.5 Spatial and Temporal Modes of Peri-Ripple iGABASnFR2 Activity and Region-Specific Dynamics. 76 Supplementary Figure 1 Characteristics of SWRs recorded in head- restrained naturally sleeping and wakeful mice. 82 Supplementary Figure 2 State-dependent modulation of neocortical GABA dynamics during SWRs: enhanced regional specificity in sleep compared to wakefulness. 84 Supplementary Figure 3 Sequential GABA Activation Across Neocortical Regions During Hippocampal SWRs: Lateral Dominance in Wakefulness and Medial Dominance in NREM Sleep. 85 Figure 4.1 Historical development of the concept of inhibition in neuroscience. 88 Figure 4.2 External inhibition in sensory cortex: imaging and circuit anatomy. 91 Figure 4.3 Ramping inhibition and circuit reorganization across sleep– wake states. 94 Figure 4.4 State-dependent cortical inhibition during hippocampal sharp- wave ripples. (A) Non-REM sleep. 96 Figure 4.4.1 Circuit mechanisms of cortical inhibition during NREM sleep. (A) Pre-ripple dynamics. 98 Figure 4.5 Mechanisms of cortical inhibition during wakefulness. 101 XVII List of abbreviations GABA Gamma-Aminobutyric Acid NREM Non-Rapid Eye Movement REM Rapid Eye Movement GAT-1 GABA Transporter Type 1 LED Light Emitting Diode AAV Adeno-Associated Virus GCs Genome Copies EMG Electromyography LFP Local Field Potential ROI Region of Interest TTL Transistor–Transistor Logic CCD Charge-Coupled Device ΔF/F Delta F over F RSC Retrosplenial Cortex VISp Primary Visual Cortex VISa Anterior Visual Area BC Barrel Cortex M1 Primary Motor Cortex SST Somatostatin interneuron PV Parvalbumin interneuron cpSFGFP Circularly Permuted Super folder GFP VSD Voltage-Sensitive Dye XVIII BOLD Blood-Oxygen-Level Dependent ANOVA Analysis of Variance 1 Chapter 1 1. General Introduction Inhibition regulates excitatory activity in cortical circuits and determines when, where, and how neurons fire. Inhibitory neurons, which release GABA, consist of special and diverse populations that affect excitatory neurons through feedforward and feedback inhibition, refining the timing, precision, and strength of neuron responses. Cortical GABAergic inhibition is the basis for information processing, such as stimulus selectivity, gain modulation, normalization, and the generation of synchronous oscillations coordinating activity between distributed brain areas. This chapter starts with a brief historical perspective of the inhibition concept. I then explore its molecular mechanisms with respect to inhibitory neurotransmitters like GABA, glycine, and serotonin, and their receptor subtypes. I conclude by discussing how GABAergic inhibition shapes cortical function prior to presenting the conceptual background and experimental hypotheses for the rest of this thesis. 1.1. The concept of inhibition The concept of inhibition comes from Plato and Aristotle, who believed moral Behaviour depended on reason controlling emotion. In the early 1800s, Franz Gall advanced a similar hierarchical model in the context of brain function. He argued that mental abilities (higher faculties)—like reasoning, judgment, and self-control—should guide or regulate more basic impulses or instincts (lower faculties), such as desire, or hunger. Gall didn’t talk about inhibition as we do today. The German philosopher-psychologist Herbart (1776–1841), unlike Gall, did not use the term inhibition in a hierarchy of faculties. Instead, he believed inhibition was what stopped unrelated or conflicting ideas from entering our conscious awareness all at once, allowing only compatible thoughts to stay in focus. He used the term associative inhibition to describe how some memories or pieces of information can block or interfere with others. Today, we call this proactive inhibition (old memories interfere with new ones) and retroactive inhibition (new memories interfere with old ones). Later, psychiatrists used inhibition to explain symptoms of mental illnesses. The German psychiatrist Griesinger believed that thoughts turn into actions unless they are stopped by a person's willpower or self-control. He described how in depression there's too much inhibition (people are blocked or slowed down), and in mania/excitement there's too little inhibition (people 2 do things without control). In this model, inhibition = willpower, and the symptoms of mental illness were viewed as a physiological issue. Consciousness and willpower were linked to inhibition in early theories, but neurophysiological theories subsequently went in another direction, with more emphasis on how the brain itself regulates inhibition, not on concepts such as will or consciousness (Bari & Robbins, 2013). 1.2. The physiological discovery of inhibition Sir Charles Bell was among the first to notice that nerve signals can reduce or stop activity, not just cause it, while studying eye muscles. In his book On the Motions of the Eye, Sir Charles Bell investigated the distinct roles of the recti and oblique muscles in controlling eye movement, proposing that the recti muscles govern voluntary, directional motion while the oblique muscles perform involuntary, protective functions such as rolling the eye upward. When he described the nerves, he found that the fourth cranial nerve, which controls the superior oblique, might not always function by stimulating the muscle to contract; instead, it might sometimes act by allowing the muscle to relax. He wrote, “We have seen that the effect of dividing the superior oblique was to cause the eye to roll more forcibly upwards; and if we suppose that the influence of the fourth nerve is, on certain occasions, to cause a relaxation of the muscle to which it goes, the eyeball must be then rolled upwards.” (Charles, 1824) After Volkmann’s work on frogs—where he suggested that the brain could block or reduce nerve activity, even though he didn’t use the term inhibition—scientists came to agree that the discovery that stimulating the vagus nerve slows the heart was key to forming the first real theories about how nerve signals can inhibit or reduce activity .Later, Weber and Weber in 1845 observed the same thing and were the first to actually call it “inhibition” whereas Lister was the first to talk about an “inhibitory system” in physiology (Bari & Robbins, 2013). Although previous research addressed peripheral inhibition, genuine understanding of central mechanisms started with Sechenov's experiments in 1863. Sechenov used frogs to show that brainstem stimulation had the effect of inhibiting reflex responses (Stuart, 2014). Following Sechenov's research, subsequent researchers started examining inhibition not merely as a physiological phenomenon, but as a fundamental principle in the operation of the nervous system. In 1906, Charles Sherrington, who was working on reflex physiology, defined the mechanisms of inhibition. He introduced the synapse theory and advanced the idea of reciprocal inhibition— 3 showing that activating one group of muscles is accompanied by the inhibition of its antagonistic counterpart (Breathnach, 2004). 1.3. Inhibition in Behaviour, learning and brain The concept of inhibition was also used in theories of Behaviour and learning. Ivan Pavlov argued that all Behaviours can be learned through a process known as conditioning. He introduced two main types of inhibition in classical conditioning: external inhibition and internal inhibition. External inhibition is when an irrelevant stimulus interrupts a conditioned response while internal inhibition is a process where a new conditioned response interferes with an existing one. As the concept of inhibition became more clearly defined through both Behavioural and physiological studies, attention gradually shifted toward understanding where in the nervous system inhibition occurs. Physiologists and neuroscientists early on looked for an inhibition neural "locus"—an area accountable for dampening activity. Inhibitory control in the brain was first envisioned as acting between higher and lower nerve centers, but later perspectives highlighted its ubiquitous nature throughout the brain. Rather than being tied to specific locations, inhibition came to be seen as a general function of higher brain regions influencing lower ones. Early theories proposed cellular mechanisms for inhibition and later linked it to attention, where the mind filters out irrelevant information to focus. Research eventually connected inhibitory control to areas involved in emotion and Behaviour regulation, such as the frontal lobes. Damage to these areas was discovered to disrupt self-control, impulse control, and attention. The same effects were seen with damage of other brain regions, indicating that inhibition is generated by a distributed network and not by a single brain center. In addition to lesion research, researchers employed electrical stimulation to investigate how brain areas such as the cortex, hypothalamus, and reticular formation regulate inhibition. These studies helped map out the brain's inhibitory pathways and showed that multiple regions work together to regulate movement and Behaviour (Bari & Robbins, 2013). 4 1.4 Experimental description of neural inhibition The mid-20th century provided physiological evidence of inhibitory mechanisms at the neural level, with Renshaw’s work being a key example. He investigated the effects of antidromic motor volleys—electrical impulses traveling backward along motor neuron axons—on spinal cord activity. The investigation was prompted by the earlier findings of Müller and collaborators that stimulation of the central end of a transected ventral root failed to produce muscular contractions or to activate other motor neurons. Renshaw sought to investigate the inhibitory mechanism underlying these observations. He performed his experiments on decerebrate or lightly anesthetized rabbits and cats, taking great pains to isolate motor pathways by cutting the dorsal roots to eliminate all sensory feedback. Electrical stimuli were then applied to the ventral root to induce antidromic volleys, and fine microelectrodes were inserted into the ventral horn of the spinal cord to record the electrical responses of nearby interneurons. He discovered interneurons in the ventral horn responded to a single antidromic volley with brief, high-frequency bursts of action potentials. These discharges were consistent in shape and timing, localized to individual cells, and reliably evoked only in specific regions of the spinal cord. He proposed that the interneurons were likely activated by recurrent collaterals of motor neurons (Renshaw, 1946). Following up on Renshaw's discovery, Eccles and his co-workers examined the mechanism by which antidromic volleys in motor axons produce inhibitory actions on spinal motoneurons. They used cats as their experimental animals and applied electrophysiological methods—both intracellular and extracellular recordings—on the lumbar spinal cord. They aimed to find out if impulses in the motor axon collaterals activate the ventral horn interneurons, already described by Renshaw, and if these interneurons, in turn, have an inhibitory action on motoneurons. They established that antidromic stimulation caused prolonged discharges in the interneurons and simultaneous inhibitory post-synaptic potentials (IPSPs) in the motoneurons. Pharmacological experiments revealed that cholinergic antagonists such as dihydro-β-erythroidine inhibited the interneuron discharges and anticholinesterases such as eserine prolonged them. They also showed that acetylcholine could evoke similar interneuron activity. From these findings, they concluded 5 that motor axon collaterals excite these interneurons via cholinergic synapses, and the interneurons inhibit motor neurons through a direct synaptic pathway (Eccles et al., 1954). Building on spinal models of recurrent inhibition, researchers soon turned their attention to cortical structures like the hippocampus to determine whether similar inhibitory mechanisms existed in higher brain areas. Andersen and colleagues investigated recurrent inhibition in the hippocampus, aiming to identify the inhibitory cell responsible and describe its synapses. They focused on the hippocampal pyramidal cells and their responses to stimulation from three input pathways: commissural, septal, and local. The study was conducted using anesthetized cats, in which the neocortex was removed to expose the hippocampus. Using intracellular microelectrodes filled with potassium citrate, potassium chloride, or sodium acetate, they recorded from CA3 pyramidal cells while stimulating the input pathways. They found that virtually all successfully penetrated cells produced large inhibitory postsynaptic potentials (IPSPs) of long duration in response to each of the three inputs. The authors observed that the inhibitory effect appeared at the soma of the pyramidal cells, based on the distribution of extracellular potentials and latency measurements. They proposed that the inhibition was mediated by a specific type of interneuron. Based on physiological and anatomical features, they suggested that the basket cell matched the expected properties: its axon ramifies extensively, contacts the soma of many pyramidal cells, and can be activated by all three inputs. The study concludes that basket cells are probably the inhibitory neurons that are accountable for recurrent inhibition in the hippocampus (Andersen et al., 1963). Whereas Andersen and colleagues were concerned with the cellular origin of recurrent inhibition in the hippocampus, Kandel et al. took this line of research further by making a comparison between the electrophysiological characteristics of hippocampal pyramidal cells and spinal motoneurons. Employing adult cats anesthetized with Evipal, they exposed the hippocampus through suction decortication and made intracellular recordings using 2M potassium citrate-filled microelectrodes. Stimulation was applied to the fornix, fimbria, alveus, and subiculum. In some experiments, the fornix and commissure were cut. They found that pyramidal cells could be identified by antidromic activation and showed sequential spike invasion. Subiculum 6 stimulation produced excitatory postsynaptic potentials, while fimbria stimulation evoked inhibitory potentials (Kandel et al., 1961). 1.5 Investigating inhibition through Hippocampal lesions Building on physiological insights into hippocampal inhibition, researchers also turned to lesion studies to understand the behavioural role of this brain region. An example is the work of Robert J. Douglas, who examined how hippocampal damage affects the ability of animals to inhibit dominant responses. From evidence on behavioural tasks—maze learning, avoidance learning, discrimination, and sequential tasks—Douglas observed that animals with hippocampal lesions do worse when inhibition is required but, where it is not, they perform better. He also compared the findings to lesions in other parts of the brain and established that the pattern is specific to the hippocampus. He elaborated on the Douglas-Pribram model, which views the hippocampus as a regulator of attention by inhibitory gating of sensory input. This model is substantiated in behavioural data and fits into Pavlov's description of internal inhibition. He proposed that the hippocampus acts to screen out irrelevant information to protect memory and regulate behavioural accordingly (Douglas, 1967). Continuing to elucidate the role of the hippocampus in the regulation of behavioural ,Daniel P. Kimble investigated whether the structure plays a role in internal inhibition, as initially formulated in Pavlovian learning theory. Using rats as subjects, Kimble compared three groups: normal rats, rats with bilateral hippocampal lesions, and rats with lesions to the neocortex above the hippocampus. All rats were trained on a brightness discrimination task in a Y-maze, followed by discrimination reversal, overtraining (50 repeated trials), and extinction (removal of the reward). The results showed that all groups learned the initial task equally well, but the hippocampal- lesioned rats were impaired in reversal learning. Under overtraining, normal rats and rats with cortical lesions ceased to respond after a period of approximately 25 trials, but rats with hippocampal lesions persisted in responding throughout all 50 trials. In extinction, the hippocampal group also exhibited a more gradual decrement in responding. Kimble interpreted that hippocampal damage disrupts internal inhibition and results in more rigid and perseverative behavioural (Kimble, 1968). 7 1.6. GABA neurotransmitter Although GABA was detected in biological tissues as early as 1910, it wasn’t until the 1950s that it was identified in mammalian brain tissue. In 1950, Eugene Roberts (Roberts & Frankel, 1950) and Awapara (Awapara, 1950) independently reported the discovery of γ-aminobutyric acid (GABA) as a naturally occurring compound in brain tissue. Bazemore and colleagues discovered an inhibitory substance found in brain extracts, known as Factor I. This substance had shown the ability to block nerve signals in the crayfish stretch receptor neuron, and the team wanted to isolate and identify it. They extracted and purified the active compound from beef brain using chemical processes and bioassays on crayfish neurons. The final crystals were confirmed to be γ- aminobutyric acid (GABA) through various chemical tests. They also showed that synthetic GABA had identical effects, and concluded that Factor I is GABA, at least in the crayfish nervous system (Bazemore et al., 1957). Then, Florey and McLennan (1959) investigated if GABA could explain all the effects of Factor I, not just in neurons but in smooth muscle tissues. They tested both Factor I and synthetic GABA on guineapig and rabbit ileum, as well as on the oesophagus of sea-urchins, using different stimulant drugs like acetylcholine and nicotine to trigger contractions. They observed that GABA and Factor I showed similar effects, but GABA couldn't fully explain Factor I's actions, suggesting that Factor I may include other active components. Therefore, they suggested that GABA might be a natural inhibitory neurotransmitter (Florey & McLennan, 1959). In the 1950s and 60s, GABA's role as a neurotransmitter was questioned due to unclear inactivation mechanisms and conflicting data. However, invertebrate studies supported its inhibitory function, and key findings in 1967 by Krnjevic and Schwartz confirmed GABA’s action on cerebral cortical neurons (Bari & Robbins, 2013). 1.6.1 GABA receptors 1.6.1.1 GABA_A receptor GABA_A receptor is an ionotropic ligand-gated chloride channel that facilitates fast inhibitory neurotransmission. When GABA acts on these receptors, the channel opens and Cl⁻ ions enter the 8 neuron, resulting in membrane hyperpolarization. In certain situations, however, GABA can be excitatory. When the intracellular concentration of Cl⁻ is higher than the outside, Cl⁻ flows out and depolarizes the neuron. This naturally happens in neonatal neurons, where the Cl⁻ gradient is reversed. Thus, the action of GABA relies on the Cl⁻ electrochemical gradient and the developmental stage (Ben-Ari et al., 2007). Structurally, the GABA_A receptors are pentameric, formed by different combinations of subunits (e.g., α, β, γ, δ, ε, and π), which dictate their functional and pharmacological characteristics. The diversity of subunits is the reason the receptor can respond differently to modulators such as benzodiazepines, barbiturates, neurosteroids, and zinc. Their distinct subunit distribution and sophisticated regulation highlight the fundamental role of GABA_A receptors in the maintenance of inhibitory control in the central nervous system (Bormann, 2000). 1.6.1.2 GABA_B receptor The GABA_B receptors are metabotropic (G-protein-coupled) receptors that allow for slow and sustained inhibition. Presynaptically, activation of the GABA_B receptor suppresses the release of neurotransmitters—GABA itself, glutamate, dopamine, and serotonin—by inhibition of voltage-gated calcium channel activity and thus calcium influx. Postsynaptically, the receptors open potassium channels (GIRKs), causing membrane hyperpolarization. Structurally, the functional receptor is a heterodimer of two subunits: GABA_B R1 (with splice variants like R1a and R1b) and GABA_B R2, which have differential patterns of expression in neuronal populations (Bettler et al., 2004). 1.6.1.3 GABA_C receptors GABA_C receptors are also chloride ion channels, as are GABA_A receptors, and consist of ρ (rho) subunits. In spite of their structural similarity to GABA_A receptors, they possess different functional characteristics, such as greater sensitivity to GABA, lower current responses, and an absence of desensitization. These ionotropic receptors are found primarily in the retina, where they mediate long-term inhibitory responses that are necessary for the processing of visual information (Bormann, 2000). 9 1.7 How inhibition shape cortical activity In the cortex, excitatory and inhibitory signals work together. When a neuron is stimulated by sensory stimulation or spontaneous activity, a wave of inhibition follows a latency. This dynamic balance is flexible and adaptive, adjusting to the specific needs of each neuron, stimulus, and moment, ensuring the brain remains in balance. As excitation becomes too intense, inhibition comes into play to modulate it. When too little excitation threatens to result in silence, inhibition reduces. The result is a system that is both stable and responsive, capable of precision as well as adaptability. Inhibition, then, is not merely a suppressive force; instead, it is a chief architect of neural computation (Isaacson & Scanziani, 2011). The architects of cortical inhibition are a diverse class of GABA-releasing interneurons, making up about 20% of cortical neurons (Markram et al., 2004). There is a strong classification system in place that incorporates molecular markers to divide cortical interneurons into three main and distinct classes: those expressing parvalbumin (PV), those expressing somatostatin (SST), and those expressing vasoactive intestinal peptide (VIP) (Rudy et al., 2011). These classes correlate with specific morphological, electrophysiological, and functional properties and hence make separate contributions to network computations and behavioural control. Interneurons form local networks through feedback and feed-forward inhibition. In feedback inhibition, they receive input from excitatory neurons and send inhibition back. In feed-forward inhibition, long-range excitatory signals activate inhibitory cells first, allowing early and strong inhibition to follow. Moreover, GABAergic interneurons inhibit one another, fine-tuning the timing, strength, and spatial distribution of inhibition across cortical circuits. 1.7.1 Inhibition shapes tuning Cortical neurons are tuned to features of stimuli—a particular tone, orientation, or whisker deflection. A key player in the generation and refinement of this tuning is inhibition. For example, pharmacological blockade of GABA_A receptors causes neurons to fire more, losing their selective response. Early models proposed that tuning was sharpened by lateral inhibition, whereby neurons inhibit the activity of surrounding cells tuned to a different feature, similar to mechanisms 10 in the retina. Yet in most rodent cortical regions where spatial maps of stimulus features do not exist, such a model cannot apply. Instead, inhibition tends to be co-tuned with excitation, i.e., inhibitory and excitatory inputs are maximal in response to the same preferred stimuli (Isaacson & Scanziani, 2011). Inhibition regulates neural activity through multiple mechanisms. The “iceberg effect” occurs when inhibition reduces the resting membrane potential, silencing subthreshold excitatory inputs and allowing only the strongest inputs to generate spikes. Inhibiting responses to less preferred stimuli, since inhibition tends to be more widely tuned than excitation. Reducing the time window for firing by tracking excitation only a few milliseconds later. Collectively, these mechanisms demonstrate that inhibition does not simply counteract excitation, determining when and how neurons fire, ultimately increasing the fidelity and specificity of cortical sensory processing (Isaacson & Scanziani, 2011). 1.7.2 Inhibition and gain control In addition, inhibition is essential in controlling the gain of a neuron, or its sensitivity to incremental inputs. Like how altering the volume of a speaker amplifies sound without altering the song, gain control scales the amplitude of neural responses without altering their selectivity. Such a process is essential for processes such as attention, where the brain amplifies relevant signals. In the brain's dynamic environment, inhibition that is coincident with excitation is important in controlling the size of neuronal responses without affecting their targets, an effect known as multiplicative gain modulation. Inhibition also maintains the fidelity of sensory representations even as their amplitude changes. Inhibition also extends the dynamic range of the brain. If not for inhibition, minor increments in excitatory input may cause unconstrained recruitment of huge neural populations, which would decrease the brain's power to separate subtle differences. Nevertheless, as inhibition strengthens together with excitation, neural activation turns out to be more gradual and constrained. Such coordination enables the brain to depict a wider, more subtle extent of inputs with high accuracy (Isaacson & Scanziani, 2011). 1.7.3 Inhibition and oscillations Inhibition also creates rhythmic cortical activity, particularly in the beta (20–30 Hz) and gamma (30–80 Hz) frequency ranges. Such oscillations allow for temporal coordination, information 11 binding, and communication between cortical areas. Parvalbumin-positive interneurons play a key role in this process, synchronizing with high precision due to their mutual connection. Two main models explain rhythm generation: PING (excitatory-inhibitory loops) and ING (interneuron- driven inhibition). In both scenarios, inhibition does more than regulate—it organizes. By setting precise timing and synchrony, it allows distributed neurons to function as a unified, pulsing network (Isaacson & Scanziani, 2011). 1.8. Glycine Glycine is the simplest in structure and its journey to recognition as a major inhibitory neurotransmitter in the spinal cord and brainstem has been detailed. The first suggestion that glycine acts as a neurotransmitter came nearly 60 years ago, when Aprison and Werman (1965) observed that glycine concentrations in the spinal cord were significantly higher than in other parts of the brain (Aprison & Werman, 1965). Using techniques developed by Curtis and Watkins, subsequent studies showed that application of glycine to spinal neurons consistently decreased their firing rates, furnishing physiological proof of its inhibitory role (Curtis & Watkins, 1960). The glycine receptor (GlyR) usually acts as an inhibitory receptor because it allows chloride ions (Cl⁻) to flow in and bring the neuron's membrane potential closer to the Cl⁻ equilibrium, which is typically negative. This either keeps the neuron stable or makes it less likely to fire (hyperpolarization). Even small changes that don’t reach firing threshold can still suppress neuron activity through a process called shunting inhibition. However, in developing (embryonic) neurons, the situation is different. These cells have high internal chloride levels, so when GlyR opens, Cl⁻ flows outward, causing a strong depolarization that can actually excite the neuron and trigger important developmental events, like calcium influx and formation of glycinergic synapses. As the nervous system develops, the neuron starts to express KCC2, a transporter that reduces internal Cl⁻ concentrations. This change renders GlyR's action inhibitory once again, representing a fundamental switch from developmental excitation to mature inhibition in the nervous system (Dutertre et al., 2012). 12 1.8.1 Glycine receptors Researchers, including Betz and colleagues, first purified glycine receptors (GlyRs) from the rat spinal cord using affinity chromatography with aminostrychnine. Glycine receptors (GlyRs) are found in both homomeric and heteromeric forms, composed of α (α1–α4) and β subunits. During development, homomeric α2 GlyRs dominate in embryonic neurons but are replaced by heteromeric α1β GlyRs after birth, which are the main inhibitory receptors in the adult spinal cord and brainstem. Homomeric α1 and α3 GlyRs are rare in adults and likely extrasynaptic, while α4 is a pseudogene in humans and functionally irrelevant. Heteromeric α2β GlyRs persist in some adult retinal neurons, and α3β GlyRs are found in pain pathways, where they are regulated by inflammation. Overall, eight GlyR types are identified: four homomeric (α1–α4) and four heteromeric (α1β–α4β), with α1β being the most functionally important in adults (Dutertre et al., 2012). 1.9 Other Neurotransmitters and Neuromodulators with Inhibitory Effects Although GABA is the principal inhibitory neurotransmitter, inhibitory control of neuronal excitability is implemented through a broad network of neuromodulators that act alongside GABAergic transmission. Several endogenous molecules exert powerful inhibitory regulation by suppressing synaptic release, hyperpolarizing neurons, and stabilizing network activity. Taurine (Wu & Prentice, 2010) and β-alanine (Wu et al., 1993) act as inhibitory neuromodulators through activation of GABA_A and glycine receptors, contributing to neuroprotection and control of excitability, particularly in brainstem and spinal circuits. Opioid peptides such as enkephalins and dynorphins produce strong inhibition by activating Gi/o-coupled opioid receptors that suppress presynaptic calcium entry and enhance potassium conductance, thereby reducing neurotransmitter release and neuronal firing (Rysztak & Jutkiewicz, 2022). The neuropeptide galanin exerts widespread inhibitory effects in hippocampal, hypothalamic, and cortical networks by dampening excitatory transmission and regulating arousal, memory, and metabolic homeostasis (Lang et al., 2015). Adenosine serves as a global homeostatic inhibitor of brain activity, accumulating during prolonged wakefulness and suppressing synaptic transmission via A1 receptors to promote sleep pressure and network stabilization (Huang et al., 2024). Also, classical monoamines including 13 dopamine (Beaulieu & Gainetdinov, 2011), serotonin (Albert & Ansari, 2019), and norepinephrine (Jones & Cohen, 2005) exert potent inhibitory modulation through specific receptor subtypes which suppress neuronal firing and transmitter release across basal ganglia, cortical, and brainstem circuits . 1.10 Conceptual framework and research hypotheses These findings provide the foundation for the present GABAergic inhibition theory, which proposes that GABA is the primary regulator of neuronal excitability and network stability across brain systems. Therefore, I build my study around the proposed GABAergic inhibition theory. The GABAergic inhibition theory This theory proposes that GABA exerts an inhibitory effect on neuronal activity throughout the nervous system. Hypothesis 1 Therefore, GABAergic neurons play a role in regulating cortical responses to both external and internal stimuli. Hypothesis 2 Furthermore, cortical responses to other brain signals, such as hippocampal sharp-wave ripples (SWRs), are expected to involve inhibitory GABAergic mechanisms. 14 Chapter 2 Characterization of iGABASnFR2 for in vivo mesoscale imaging of intracortical GABA dynamics Abstract Although genetically encoded sensors have advanced the study of cortical excitation, tools for large-scale imaging of inhibition remain limited. Visualizing extracellular gamma-aminobutyric acid (GABA) dynamics in vivo is essential for understanding how inhibitory networks shape brain activity across sensory, behavioural, and pharmacological states. My goal is to validate and apply the genetically encoded sensor iGABASnFR2 for wide-field imaging of extracellular GABA and to characterize how cortical inhibition reorganizes across brain states, sensory modalities, and after GABA transporter blockade. I performed mesoscale imaging in head-fixed C57BL/6 mice systemically expressing iGABASnFR2. Recordings were conducted under isoflurane anesthesia, during quiet wakefulness, natural sleep [non-rapid eye movement (NREM) and rapid eye movement], and after administration of the GAT-1 inhibitor tiagabine. I analyzed both sensory- evoked and spontaneous GABA signals using timeseries, spectral, and seed-pixel correlation analyses. iGABASnFR2 demonstrated strong and modality-specific GABAergic responses to sensory stimulation, with faster and stronger activation in the contralateral cortex. Although the general spatial patterns of sensory-evoked GABA responses were consistent across anesthesia and quiet wakefulness, the amplitude, timing, and spread of these responses were significantly greater during wakefulness. During spontaneous activity, cortical GABA levels and connectivity modulated by brain state: GABA amplitude and interhemispheric synchrony, were highest during quiet wakefulness but reduced during NREM sleep. Tiagabine elevated baseline GABA levels, abolished stimulus-evoked responses, and enhanced local and long-range inhibitory synchrony. iGABASnFR2 enables reliable, high-resolution imaging of cortical GABA dynamics in vivo. These results demonstrate that inhibitory signaling is dynamically structured across brain states and can be pharmacologically modulated. This tool offers opportunities to explore the role of inhibition in health and disease at the mesoscale level. 15 2.1 Introduction The nervous system is composed of two basic cell types: neurons and glial cells. Neurons can be categorized into excitatory and inhibitory types. Excitatory neurons primarily release glutamate which is the major neurotransmitter in the nervous system to facilitate communication among neurons across different brain regions (Zhou & Danbolt, 2014),whereas inhibitory neurons mainly release gamma-aminobutyric acid (GABA) to stabilize neural networks by balancing excitatory activity and preventing excessive neuronal firing (Froemke, 2015). Maintaining this precise balance between excitation and inhibition is essential for sensory processing, memory formation, and motor control (Tatti et al., 2017; Yuste, 2005). Through temporally and spatially precise modulation of neuronal activity, inhibitory signaling contributes to a wide range of brain functions, including sensory processing, circuit refinement, and the regulation of oscillatory dynamics (Ferguson & Gao, 2018; Kepecs & Fishell, 2014; Markram et al., 2004; Tremblay et al., 2016; Urban-Ciecko & Barth, 2016). Disruptions in GABAergic signaling underlie a variety of neurological and psychiatric disorders, including epilepsy, schizophrenia, and autism spectrum disorders, highlighting the clinical importance of understanding GABAergic modulation in both health and disease (Curley & Lewis, 2012; de Lanerolle et al., 1989; Gonzalez-Burgos & Lewis, 2008; Lewis & Moghaddam, 2006; Marín, 2012; Rubenstein & Merzenich, 2003). Despite the importance of GABA in regulating cortical activity, direct, real-time visualization of GABA dynamics in vivo remains unknown. The development of genetically encoded neurotransmitter sensors has significantly expanded the ability to monitor neural activity across spatial and temporal scales (Marvin et al., 2013). For instance, the glutamate sensor iGluSnFR enabled high-resolution, real-time imaging of excitatory neurons, providing insights into cortical connectivity and sensory-evoked activity in awake and anesthetized animals (Xie et al., 2016).Building on advances in sensor engineering, the genetically encoded fluorescent sensor iGABASnFR2 was developed to detect extracellular GABA dynamics (Marvin et al., 2019) This sensor permits real-time, in vivo monitoring of extracellular GABA, providing a powerful tool for visualizing GABA’s spatial and temporal dynamics across large-scale cortical regions. Wide-field imaging with iGABASnFR2 enables comprehensive mapping of inhibitory circuits. However, a detailed characterization of 16 iGABASnFR2 functionality across different brain states, sensory modalities, and pharmacological conditions is still needed. In this study, I characterized spontaneous and sensory-evoked GABAergic activity across the dorsal cortex using wide-field imaging of iGABASnFR2 sensor in natural sleep, awake mice, and under 1% isoflurane anesthesia. In addition, I assess the effects of tiagabine, a GABA reuptake inhibitor, to further characterize sensor performance and demonstrate its sensitivity to pharmacological manipulation of cortical GABA levels. These results not only validate the robustness and sensitivity of iGABASnFR2 as a practical tool for monitoring GABAergic neurotransmission in vivo but also lay a solid foundation for future studies into inhibitory dynamics in health and disease, including conditions such as epilepsy, schizophrenia, and autism spectrum disorders. 2.2 Material and Methods 2.2.1 Animal Subjects The University of Lethbridge animal care committee approved all procedures, which adhered to the guidelines set forth by the Canadian Council on Animal Care and Use. I used 20 young adult (6 weeks old) C57BL/6 mice (12 males and 8 females). Mice were accommodated in transparent plastic cages within a 12-h light–dark cycle, with lights turning on at 7:30 AM, and provided unrestricted food and water access. Room temperature was maintained at 24°C, and relative humidity was kept between 40% and 50%. 2.2.2 Viral Constructs The viral constructs originated from the Viral Vector Core of the Canadian Neurophotonics Platform (RRID: SCR_016477). Plasmids encoding iGABASnFR2 and cpSFGFP were obtained from Addgene, and AAV2/PHP.N-CAG-iGABASnFR2 and AAV2/PHP.N-CAG-cpSFGFP viral vectors were subsequently designed, packaged, and purified by the CNP Viral Vector Core at a final concentration of 1.5 X1013 genome copies (GCs)/mL. 17 2.2.3 Retro-Orbital Injection For retro-orbital viral injections (Yardeni et al., 2011), 4- to 6-week-old C57BL/6 mice were anesthetized with 3% isoflurane and maintained at 2% to 2.5% during the procedure. Metacam (5 mg/kg, subcutaneous) was administered for analgesia, and body temperature was maintained using a heating blanket. Topical anesthesia (0.5% proparacaine hydrochloride) was applied to the eye, followed by gentle pressure to induce mild eye protrusion. A 30-G needle was then inserted through the medial canthus at a 30-deg angle into the retro-orbital sinus. Mice received 1.4 X 1011 GCs of either AAV2/ PHP.N-CAG-iGABASnFR2 or AAV2/PHP.N-CAG-cpSFGFP, using the PHP.N capsid for widespread CNS expression, as originally described by Ref. (Chan et al., 2017). iGABASnFR2 has a secretion signal (IgG kappa) and a transmembrane domain (from PDGFR). Combined those two sits target it to the plasma membrane, facing out. Without those, it will be expressed in the cytosol, and in theory could be used to detect cytosolic GABA. In addition to neuronal synaptic release, astrocytes are known to contribute to extracellular GABA via non- vesicular release mechanisms that generate tonic inhibition. Therefore, the iGABASnFR2 signal likely reflects the combined contribution of neuronal and glial sources of extracellular GABA. 2.2.4 Drug Administration Tiagabine hydrochloride (Cat #SML0035, Sigma-Aldrich Canada, Oakville, Canada) was purchased from Sigma-Aldrich. Tiagabine was dissolved in sterile saline to achieve a concentration of 5 mg/ml. For intraperitoneal injection, mice received tiagabine at a dosage of 10 mg/kg body weight. Control animals were administered an equivalent volume of sterile saline. 2.2.5 Surgical Procedure and Post-Operative Care C57BL/6 mice received buprenorphine (0.05 to 0.1 mg/kg, subcutaneously) ∼30 min before surgery, followed by anesthesia with isoflurane (1% to 2% in oxygen) delivered via a nose cone. After shaving and sterilizing the scalp, lidocaine (0.5%, 5 mg/mL, subcutaneously) was administered at the incision site for local anesthesia (0.04 to 0.08 mL for mice weighing 25 to 55 g). A midline incision was made to expose the skull, and the overlying skin was carefully removed to avoid damaging the underlying bone. A custom-designed head plate was affixed to the skull 18 using C&B Metabond Quick Base (Parkell, Brentwood, New York, United States) mixed with C&B Metabond Clear L-Powder (3 g, Product Code S399, Parkell, Tokyo, Japan). A sterile 12- mm circular glass coverslip (Carolina Biological Supply, Cat. No. 633005, Burlington, North Carolina, United States) was placed on the skull surface and sealed in place with the same adhesive. For hippocampal recordings, a bipolar electrode made of two twisted 50-μm Tefloncoated stainless-steel wires (A-M Systems, Sequim, Washington, United States) was slowly inserted through a craniotomy at a 57-deg angle relative to vertical. Electrode placement was guided by continuous monitoring of signal quality using both visual and auditory feedback. The electrode was secured to the skull using Krazy Glue, followed by dental cement. An electromyography (EMG) electrode was also implanted into the neck muscles to monitor muscle activity. Following surgery, animals were housed individually in temperature-controlled recovery cages and received subcutaneous injections of Baytril (enrofloxacin, 10 mg/kg), meloxicam (5 mg/kg), and 1 mL of warm sterile saline. These injections were administered once every 24 h for 3 days postoperatively, in accordance with institutional guidelines. After this recovery period, animals were monitored twice daily for the remainder of the experiment. 2.2.6 iGABASnFR2 Imaging Under Anesthesia Following, the animals were anesthetized with isoflurane (2.5% for induction, followed by 1% for maintenance). The depth of anesthesia was confirmed by assessing reflexes. Once adequately anesthetized, each mouse was positioned in a head-stage, and the head was securely head-fixed. A homeothermic blanket was utilized to maintain their body temperature, and isoflurane was administered via a nosepiece. The isoflurane concentration was adjusted to 1% to initiate the imaging procedure. Images were acquired using a microscope consisting of a front-to-front pair of video lenses with a field of view measuring 8.6 X 8.6 mm. The camera’s focal plane was positioned 0.5 to 1 mm (∼0.04) below the cortical surface. A 12-bit charge-coupled device (CCD) camera (1M60 Pantera Dalsa, Waterloo, Ontario, Canada) and an EPIX E8 frame grabber with XCAP 3.8 imaging software (EPIX, Inc., Buffalo Grove, Illinois, United States) were used to capture images at a frame rate of 80 Hz. These imaging parameters have been employed in previous studies (Mohajerani et al., 2013; Silasi et al., 2016; Vanni & Murphy, 2014). Carefully designed data collection protocols support the robustness of our findings. Sequential illumination was achieved 19 using alternating blue and green light-emitting diodes (LEDs) (Xiao et al., 2021). The timing of LED alternation is illustrated in Fig. S1 in the Supplementary Material. Blue light (473 nm, filtered through a 467- to 499-nm bandpass) was used to excite the iGABASnFR2 indicator, and green light (530 nm, filtered through a 527/42-nm bandpass) was 20 Fig. 1 Schematic of experimental workflow, imaging setup, and expression of iGABASnFR2. (A) Experimental timeline. AAV2/PHP.N-CAG-iGABASnFR2 or the control virus AAV2/PHP.N-CAGcpSFGFP was systemically administered via retro-orbital injection into 6- week-old C57BL/6 mice. Four weeks after viral injection, the animals underwent cranial window 21 surgery. Following a 7-day recovery period, animals gradually habituated to the head-fixation setup over the course of another week. Sensory-evoked imaging was then conducted under 1% isoflurane anesthesia. After this session, longitudinal recordings during quiet wakefulness and natural sleep were carried out for up to 6 weeks. Mice were subsequently euthanized, and the brains were perfused for histological analysis. (B) Imaging setup. (i) Illustration of the cranial window implantation over the skull following scalp removal. (ii) Representative image showing the cortical imaging area, indicated by the dashed white line. (iii) Schematic of the wide-field imaging setup: a blue LED (470 nm) was used for excitation, and signals were captured using a CCD camera at 530 nm emission. (iv) Map of the bilateral craniotomy showing the targeted cortical regions based on the Allen Mouse Brain Atlas. (C) Expression of iGABASnFR2. (i) Schematic of coronal sectioning locations (a–f) along the anterior–posterior axis. (ii) Coronal sections (a–f) show strong iGABASnFR2 expression in the cortex and hippocampus. (iii) The region marked by a white dashed rectangle in panel C(iid) was used to extract fluorescence profiles across animals (n . 4), highlighting the unique expression and inter-animal expression consistency. (iv) The average profile with SEM as a shaded region. used for intrinsic signal imaging of blood volume. A bandpass emission filter (shown in Fig. 1) was positioned in front of the CCD camera to enable selective detection of either fluorescence or reflectance signals. Blue and green LEDs were synchronized and alternated on a frame-by-frame basis using transistor-transistor logic (TTL) triggering, resulting in interleaved acquisition of fluorescence and reflectance images at 40 Hz per channel. In addition, images of reflectance, crucial for blood artifact corrections, were evaluated within the current pipeline (Kramer & Pearlstein, 1979; Ma et al., 2016; Scott et al., 2018). Anesthetized iGABASnFR2 imaging of spontaneous activity was conducted without sensory stimulation for 15-min sessions. 2.2.7 Sensory Stimulation I captured the iGABASnFR2 signal in response to varied peripheral simulations while utilizing urethane anesthesia, following the methodology outlined in previous studies. Sensory stimuli were employed to map the functional areas of the hindlimb somatosensory, forelimb somatosensory, auditory, visual, and barrel cortices (Mohajerani et al., 2013). Sensory stimuli were applied to map the cortical regions corresponding to forelimb, hindlimb, whisker, visual, and auditory modalities. For forelimb and hindlimb stimulation, a piezoelectric bending actuator delivered a single 300-ms tap via a square pulse directly to the skin of one forelimb or hindlimb. Whisker stimulation targeted the whisker, which was attached to a piezoelectric actuator (Q220-A4-203YB, Piezo Systems, Inc., Woburn, Massachusetts, United States) and deflected using a single 300-ms square pulse. Visual stimuli consisted of a single 20-ms pulse of 435-nm light (LED), delivered at a fixed distance and 22 height relative to the right eye. For each sensory modality, 40 stimulus presentations were delivered with a 10-s interstimulus interval to calculate the average cortical responses. The timing of stimulus delivery is illustrated in Fig. S1 in the Supplementary Material. 2.2.8 Habituation After the 7-day recovery period from surgery, mice gradually habituated to the head restraint in the recording environment. Initially, each mouse was placed individually on the recording platform along with Cheerios cereal, allowing them to explore freely and become comfortable. Mice were progressively acclimated to eating the cereal while head-fixed, beginning with 5-min sessions and increasing by 5 min each day until reaching 60 min. 2.2.9 iGABASnFR2 Imaging During Wakefulness Following habituation, wakefulness recordings began. Each animal underwent recording sessions every 3 days, completing three to four sessions per mouse. Recordings were consistently performed at the same time of the day to reduce variability and stress. Among sessions, mice were returned to their home cages for rest and recovery before the next recording. After finishing all wakefulness sessions, mice proceeded to the sleep recording phase. 2.2.10 iGABASnFR2 Imaging During Sleep To optimize conditions for natural sleep under head restraint, mice were transferred from their colony housing to a separate room at noon the day before recording. Sleep was restricted for 6 h by gentle stimulation (using a cotton-tip stick) whenever signs of drowsiness were observed. Following 6 h of sleep deprivation, mice were placed overnight in large, enriched cages containing a running wheel, Cheerios, and a water container to promote exploration and natural sleep. The next morning (∼9∶00 AM), animals were transferred to the imaging platform for sleep recordings. Afterward, they were returned to their home cages for at least 3 days of recovery before any further recordings. This sleep deprivation protocol is commonly used to induce moderate but physiologically meaningful sleep pressure (Vyazovskiy et al., 2008), which is known to trigger a homeostatic increase in slow-wave activity during subsequent non-rapid eye movement (NREM) sleep. 23 2.2.11 Preprocessing Image stacks were first de-interleaved to separate the GABA-sensitive fluorescence signal (blue channel) and the hemodynamic reflectance (green channel) signal. The correct channel assignment was verified by computing pixel-wise correlations between the first frame of each stack and reference images corresponding to each illumination wavelength. To quantify extracellular GABA dynamics, the relative fluorescence change (ΔF/F) was calculated on a per- pixel basis using a 2-second pre-stimulus window to define baseline fluorescence (F₀). The ΔF/F signal was defined as: ΔF/F(t) = (F(t) - F₀) / F₀ where F(t) is the fluorescence intensity at time t, and F₀ is the baseline fluorescence. To correct for hemodynamic artifacts (Kramer & Pearlstein, 1979; Ma et al., 2016; Scott et al., 2018) , a pixel-wise linear regression was applied, in which a scaled version of the reflectance signal R(t) was subtracted from the fluorescence trace: Fcorrected(t) = F(t) - α · R(t) The corrected fluorescence signal was then normalized to the baseline (ΔF/F) and bandpass filtered to remove low-frequency drift and high-frequency noise. Trial-averaged ΔF/F responses were generated to assess sensory-evoked activity. All preprocessed data—including corrected ΔF/F, raw fluorescence, and reflectance signals—were saved in float32 format. 2.2.12 ROI-based Fluorescence Analysis Following preprocessing, imaging data from specific regions of interest (ROIs)—defined as 3×3-pixel areas (~40,401 µm²) centered around anatomical coordinates corresponding to stimulation sites)— were extracted. Baseline correction was conducted by subtracting the mean fluorescence signals calculated from a 1-second pre-stimulus period from the post-stimulus fluorescence signals. The signals were further filtered to eliminate slow baseline drifts using a 24 high-pass filter (>0.1 Hz) and to reduce high-frequency noise using a low-pass filter (<5 Hz). From the filtered signals, several key parameters were derived, including peak amplitude (maximum ΔF/F₀ within a 1-second post-stimulus interval), time-to-peak, and decay time (duration for the fluorescence to fall to 50% of peak amplitude). For visualization, the corrected ΔF/F₀ signals were plotted with indicators marking peak and decay times, facilitating clear interpretation of response dynamics. Mean responses across trials were computed separately for each sensory modality, with variability assessed by plotting the standard error of the mean (SEM) as shaded regions surrounding the mean trace. 2.2.13 Motion Detection and Exclusion Motion artifacts were identified and excluded from analyses using electromyography (EMG) signals recorded simultaneously with imaging data. EMG recordings were smoothed using a median filter and squared to enhance the detection of muscle activity periods. An activity threshold was established at the 95th percentile of the processed EMG signal to identify movement onset and offset events. These EMG-detected motion periods were temporally aligned with imaging frames via synchronized camera clock signals. Frames coinciding with detected movements were subsequently removed from analysis, ensuring the seed pixel correlation analysis reflected only stationary periods free of motion-related artifacts. 2.2.14 Seed-pixel Correlation Analysis Seed pixel correlation analysis was performed to evaluate functional connectivity based on spontaneous GABA activity from sleep and anesthetized mice. preprocessed data were spatially registered to the Allen Brain Atlas, enabling anatomical alignment and inter-subject comparisons. Regions of interest (ROIs)—specifically the barrel cortex (BC), visual cortex (VC), hindlimb (HL), and forelimb (FL)—were defined using anatomical coordinates derived from the atlas. Within each ROI, seed pixels were selected to serve as reference points for correlation-based connectivity analysis. 25 2.2.15 Motion Signal Extraction, Alignment, and Sleep Scoring Behavioural videos were used to monitor animal movement during imaging. Motion signals were extracted using FaceMap (Syeda et al., 2024), which computes frame-to-frame pixel intensity changes in user-defined regions of interest (ROIs). Five ROIs—nose, whisker pad, ear, shoulder, and trunk—were selected to capture both facial and body movements. To synchronize video with neural data, camera frame pulses were recorded on analog channels during acquisition. These were used to align motion traces with electrophysiological recordings (LFP and EMG, sampled at 2 kHz). Traces were interpolated or trimmed as needed, z-scored, and averaged across ROIs to generate composite facial and body motion signals. This enabled detection of both gross movement and small twitches, such as those occurring during REM sleep, which may not be captured by EMG alone. With motion aligned, vigilance states were classified as wakefulness, NREM, or REM sleep using combined behavioural and physiological features. Wakefulness was marked by visible movement and high EMG activity. NREM sleep was defined by low EMG, a low hippocampal theta-to-delta power ratio, and the presence of Large Irregular Activity (LIA) in the LFP. REM sleep was characterized by minimal EMG activity, a high theta-to-delta ratio, and continuous hippocampal theta. In head-fixed recordings, pupil constriction served as an additional marker of sleep onset (Karimi Abadchi et al., 2020; Yüzgeç et al., 2018). This multimodal approach enabled robust, accurate classification of sleep stages across different experimental conditions. 2.2.16 Effect of Isoflurane Anesthesia on Extracellular GABA Dynamics At concentrations around 1%, isoflurane induces a slow-wave cortical state characterized by synchronized network activity and enhanced inhibitory tone. Isoflurane potentiates synaptic GABA_A receptor–mediated currents (Topf et al., 2003), enhances tonic inhibition via extrasynaptic GABA_A receptors that are sensitive to ambient extracellular GABA (Jia et al., 2008), and can directly activate GABA_A receptor channels at higher concentrations (Neumahr et al., 2000). In addition, isoflurane preferentially suppresses pyramidal neuron excitability relative to parvalbumin interneurons, biasing cortical circuits toward inhibition (Qiu et al., 2023). Under these slow-wave conditions, the iGABASnFR2 sensor reports physiologically meaningful 26 extracellular GABA dynamics within an inhibition-dominated network regime (Neumahr et al., 2000; Topf et al., 2003; Jia et al., 2008; Qiu et al., 2023) 2.2.17 Statistical Analysis All data processing and analyses were performed using custom scripts written in MATLAB R2024a. 2.3 Results 2.3.1 Experimental Workflow, Imaging Setup, and Cortical Expression of iGABASnFR2 To characterize extracellular GABA dynamics, I first injected AAV2/PHP.N- CAGiGABASnFR2 and AAV2/PHP.N-CAG-csGFP systemically via retro-orbital injection [Fig. 1(a)]. Imaging was performed under isoflurane anesthesia, during quiet wakefulness, NREM, and REM sleep. The imaging setup [Fig. 1(b)] utilized a bilateral cranial window and a CCD-based wide-field microscope that captured fluorescence across an 8.6 Å~ 8.6 mm field of view. Sequential blue (470 nm) and green (530 nm) LED illumination enabled alternating frame acquisition and hemodynamic correction (Kramer and Levitan, 1979; Ma et al., 2016; Scott et al., 2018). To validate sensor expression, I conducted a histological analysis [Fig. 1(c)]. A robust expression of iGABASnFR2 was revealed in cortical and hippocampal regions. The coronal brain sections [Fig. 1(c)ii] demonstrated consistent expression across animals. To assess inter-animal variability and regional expression strength, fluorescence intensity profiles were extracted using ImageJ from defined cortical areas [Fig. 1(c)iii], and to further quantify and visualize overall trends, these profiles were averaged across animals, with the mean±SEM shown [Fig. 1(c)iv] confirming uniform sensor expression suitable for quantitative cortical imaging. 27 2.3.2 iGABASnFR2 Reveals Modality- and Hemisphere-Specific Cortical Inhibition Under Anesthesia Previous studies on sensory processing have primarily focused on excitatory neuronal responses, often using calcium or glutamate indicators to map stimulus-evoked activity across the cortex (Xie et al., 2016; Mohajerani et al., 2013; Chen et al., 2012). However, much less is known about how sensory stimuli engage inhibitory networks, particularly at the mesoscale. Inhibitory interneurons play a crucial role in shaping sensory responses, modulating cortical excitability, and controlling the timing and precision of neural coding (Atallah et al., 2012; Wilson et al., 2012).To examine the spatiotemporal dynamics of GABA activity across the cortex, I delivered contralateral whisker, hindlimb, forelimb, and visual stimulation under 1% isoflurane anesthesia and recorded GABA activity using the iGABASnFR2 sensor. As shown in Fig. 2(a), each sensory modality shows a localized increase in GABAergic fluorescence within the corresponding primary sensory cortex, with response onsets occurring ∼75 to 150 ms after stimulus. All sensory modalities elicited more robust and spatially localized GABAergic activity. Temporal response profiles [Fig. 2(b)] demonstrated stronger GABAergic activation in the contralateral hemisphere relative to the ipsilateral side across all sensory modalities. 28 Fig. 2 Sensory-evoked GABAergic responses in the neocortex measured by iGABASnFR2 imaging. (a) Montages of the wide bilateral craniotomy, with bregma marked by a white circle. Cortical GABAergic activation patterns are shown in a mouse anesthetized with isoflurane (1%) following (i) whisker stimulation (300 ms), (ii) hindlimb stimulation (300 ms), (iii) forelimb stimulation (300 ms), and (iv) visual stimulation (20 ms) of the eye using an LED. Sensory-evoked extracellular GABA signals were detected using the iGABASnFR2 sensor. Activation is observed within 50- to 375-ms post-primary sensory cortex activation. Responses represent an average of 40 trials. The second image in the second row indicates anterior (A), posterior (P), medial (M), and lateral (L) directions. (b) Time series of sensory-evoked GABA responses. The time series of GABA responses for each sensory stimulation was measured from the respective primary sensory regions. Contralateral responses are shown in red, and ipsilateral responses are shown in black. 29 Data are presented as mean ± SEM, with responses extracted from 3 X 3-pixel ROI (∼40; 401 μm2), n = number of animals. (c) Summary of sensory-evoked GABA response features. Decay time (ms), peak amplitude (ΔF⁄F), and time to peak (ms) for contralateral and ipsilateral responses. Data are shown as mean ± SEM, with contralateral responses in red and ipsilateral responses in black. A significant region X laterality interaction was found for time to peak (p = 0.0055), indicating lateralization effects vary across sensory regions (Video 1, avi, 12.7 MB [URL: https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s1];Video 2, avi, 12.9 MB [URL: https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s2];Video 3, avi, 13.3 MB [URL: https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s3];Video 4, avi,13.6 MB [URL: https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s4]). Specifically, the contralateral visual cortex exhibited the highest peak amplitude, followed by hindlimb, whisker, and forelimb cortices, whereas ipsilateral responses were generally weaker and slower. To quantify these differences, I extracted response features including decay time, time to peak, and peak amplitude across all sensory regions [Fig. 2(c)]. A two-way analysis of variance (ANOVA) revealed significant main effects of sensory region on all three measures of inhibitory response dynamics. Time to peak (p= 0.0092), decay time (p= 0.0006), and peak amplitude (p= 0.0011) all varied significantly across sensory regions, indicating region-specific characteristics of GABAergic inhibition. Laterality (ipsilateral versus contralateral) had a significant main effect only on peak amplitude (p < 0.0001), with contralateral responses consistently showing higher amplitudes. No significant effects of laterality were observed for time to peak (p= 0.35) or decay time (p= 0.77). Furthermore, there were no significant interactions between region and laterality for any of the three metrics (time to peak: p= 0.22, decay: p=0.83, and peak amplitude: p=0.78), suggesting that hemispheric differences in GABAergic inhibition were consistent across modalities and not dependent on specific sensory regions. To confirm that these delayed signals were specific to GABAergic dynamics and not artifacts of hemodynamics or sensor excitation, I used cpSFGFP-expressing mice under identical imaging conditions as a negative control. As shown in Fig. S2 in the Supplementary Material, cpSFGFP mice showed no significant sensory-evoked or spontaneous fluorescence changes, confirming that the iGABASnFR2 signals reflect GABAergic activity. Auditory stimulation under anesthesia evoked clear iGABASnFR2 responses in the auditory cortex (Fig. S3 in the Supplementary Material), further validating sensor specificity. To assess cortical coordination during sensory processing, I performed seed-pixel correlation analysis across 10 anatomically defined cortical regions. https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s1%5D;Video https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s2%5D;Video https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s3%5D;Video https://doi.org/10.1117/1.NPH.XX.XX.XXXXXX.s4 30 This revealed structured and modality-specific inhibitory networks, with the strongest intrahemispheric connectivity observed contralateral to the stimulus (Fig. S4 in the Supplementary Material). Across sensory modalities—including whisker (Video 1), hindlimb (Video 2), forelimb (Video 3), and visual (Video 4) stimulation—sensory input activates thalamocortical projections targeting layer 4 of the primary sensory cortices, leading to early excitatory responses (Petersen, 2007; Ridder and Nusinowitz, 2006).These are primarily mediated by excitatory neurons and refined by fast feedforward inhibition from parvalbumin (PV)-expressing interneurons (Gabernet et al., 2005 ; Isaacson and Scanziani, 2011).Subsequently, a delayed GABAergic response emerges, largely driven by somatostatin (SST)-expressing interneurons providing feedback inhibition to modulate dendritic activity and maintain cortical stability (Urban-Ciecko and Barth, 2016; Muñoz et al., 2017; Gentet et al., 2012). This response is consistently stronger and earlier in the contralateral hemisphere, which receives direct thalamic input, whereas the ipsilateral hemisphere exhibits a weaker and more delayed GABAergic response, likely due to slower callosal transmission and reduced excitatory drive (Ferezou et al., 2007; Minamisawa et al., 2018).Together, these results demonstrate that sensory-evoked GABA dynamics follow a consistent temporal structure across modalities and hemispheres and confirm that iGABASnFR2 reliably detects extracellular GABA responses with spatiotemporal precision across cortical regions, establishing its utility for mesoscale mapping of inhibitory dynamics. 2.3.3 Sensory-Evoked and Spontaneous GABA Activity in Quiet Wakefulness Resembles Anesthesia-Induced Patterns Cortical brain states vary across behavioural conditions, shaping spontaneous activity and sensory processing. During active behaviour, such as locomotion or whisking, cortical activity becomes desynchronized, and inhibition is modulated to refine sensory gain (Pakan et al., 2016; Gentet et al., 2010; Keller and Mrsic-Flogel, 2018).In contrast, during quiet wakefulness and under light anesthesia, neuronal activity is dominated by slow, synchronized fluctuations that reflect reduced arousal and a shift toward global inhibitory tone (Muñoz et al., 2017; McGinley et al., 2015) .Although previous studies have shown that GABAergic interneurons contribute significantly to these state dependent dynamics, (Chen et al., 2012; Pfeffer et al., 2013) it remains unclear whether the spatiotemporal profile of extracellular GABA during quiet wakefulness 31 resembles that observed under anesthesia. To examine whether GABAergic responses to sensory stimulation and spontaneous activity in quiet awake mice exhibit spatiotemporal dynamics like those observed under anesthesia, I performed wide-field imaging of the cortex using the iGABASnFR2 sensor. Head-fixed mice were imaged in both quiet wakefulness and anesthetized states, the latter induced by 1% isoflurane. EMG recordings from a neck muscle electrode, along with video monitoring of body and whisker movement, were used to classify behavioural state [Fig. 3(a)]. Power spectral analysis of EMG signals confirmed a reduction in muscle tone under anesthesia compared with the quiet awake state. I recorded cortical GABA signals evoked by contralateral visual or whisker stimulation, averaging responses across 40 trials for each condition [Fig. 3(b)]. Both anesthetized and quiet awake states showed robust stimulus-evoked increases in extracellular GABA in primary sensory areas, including the primary visual cortex (VISp), anterior visual area (VISa), BC, and primary motor cortex (M1). In Fig. 3(b)i, which shows cortical GABA responses to visual stimulation, the quiet awake state is characterized by an earlier onset and more widespread GABA release across the visual cortex. In contrast, under anesthesia, the GABA signal appears later and is more spatially restricted. Similarly, in Fig. 3(b)ii, following whisker stimulation, the quiet awake state shows strong and broad activation of the contralateral BC and associated motor areas (M1). This response is both faster and more spatially extensive than in the anesthetized state. To assess how GABAergic responses vary across brain states, I analyzed their temporal dynamics during visual and whisker stimulation [Fig. 3(c)]. Responses were stronger and more spatially distinct during quiet wakefulness than under anesthesia—VISp > VISa for visual input and BC > M1 for whisker input. Quantitative analysis [Fig. 3(d)] confirmed significantly higher peak amplitudes and longer decay times in the awake state. Time to peak was generally shorter in the awake state, suggesting faster inhibitory onset, though most differences were not statistically significant. However, BC in wakefulness responded significantly faster than M1 under anesthesia. Decay times were longer under anesthesia, especially in BC, suggesting more prolonged inhibition when cortical activity is suppressed. These results highlight the influence of both brain state and region on the strength and timing of GABAergic responses. 32 Fig. 3 Sensory-evoked and spontaneous GABA activity in quiet wakefulness resembles patterns observed under anesthesia. (a) Individual example of neck muscle EMG (>1 Hz) from a head restrained mouse under isoflurane (1%) anesthesia (top trace) and quite awake (bottom trace). The power spectra of the EMG signals differ between states of anesthesia (blue) and wakefulness (red). (b) Representative cortical GABA signals are taken from the iGABASnFR2 sensor in response to contralateral visual or whisker stimulation during anesthesia or in quiet awake states. The images represent an average of 40 trials of stimulation. (c) Quantification of GABA signals in response to sensory stimulation under anesthesia and quiet wakefulness. Plots show averaged responses from 3 X 3-pixel ROIs (∼0.04 mm2) within VISp (blue) and VISa (red) for visual stimulation and BC (blue) and M1 (red) for whisker stimulation (Ciii and Civ). Shaded areas represent SEM. (d) Statistical comparison of peak amplitude, time to peak, and decay time of GABA responses across states. P < 0.05, P < 0.01, one-way ANOVA. Error bars indicate SEM. 33 2.3.4 Mesoscale Imaging of Cortical GABA Dynamics During Natural Sleep and Wakefulness Although many studies have explored cortical dynamics across sleep and wake states using electrophysiological methods and excitatory activity sensors (Dash et al., 2009; Nazari et al., 2023; Karimi Abadchi et al., 2023), the ability to track inhibitory signaling at mesoscale resolution across natural brain states remains limited. To further assess iGABASnFR2 performance across different brain states, I examined cortical GABA dynamics during quiet wakefulness, NREM sleep, and REM sleep. Mesoscale iGABASnFR2 imaging was combined with simultaneous hippocampal LFP and EMG recordings in head-fixed mice. Animals were allowed to transition naturally among vigilance states while cortical GABA levels were monitored [Figs. 4(a)–4(c)]. GABA signals were highest during wakefulness and reduced during NREM sleep [Fig. 4(d)]. Spectral analysis showed a reduction in low-frequency GABA fluctuations during REM compared with both wakefulness and NREM [Fig. 4(e)], suggesting diminished slow GABA oscillations during REM sleep. To further assess the spatial coordination of cortical GABA activity, I computed pairwise correlation maps across the cortex during each brain state over multiple days. As shown in [Fig. 4(f)], cortical GABA signals exhibited strong bilateral synchrony during quiet wakefulness, which was markedly reduced during NREM sleep and only moderately diminished during REM. This pattern is evident in interhemispheric correlation heatmaps [Fig. 4(f)], where NREM shows the most substantial decrease in bilateral synchrony, whereas REM correlations remain relatively higher. Quantitative analysis confirmed that the mean interhemispheric correlation values were lower during NREM sleep compared with both wakefulness and REM [Fig. 4(g)]. During transitions from REM to wakefulness, I observed a sharp increase in cortical GABA levels [Fig. 4(h)]. This transition was accompanied by a broad re-engagement of cortical GABAergic activity across multiple regions [Fig. 4(i)], suggesting rapid reinstatement of the GABA response upon arousal from REM sleep. A detailed overview of the motion-based and electrophysiological features used for behavioural state classification is provided in Fig. S5 in the Supplementary Material, which illustrates the temporal alignment of whisker pad and nose motion, EMG power, body motion, theta-to-delta ratio, and hippocampal LFP signals across NREM and REM sleep transitions. Together, these findings demonstrate that iGABASnFR2 reliably captures spontaneous, brain state-dependent fluctuations in extracellular GABA with high temporal and 34 spatial resolutions. It supports its utility for long-term mesoscale imaging of cortical inhibition under natural physiological conditions. 2.3.5 Intracortical Long-Range GABAergic Correlations Revealed by Seed-Pixel Analysis Across Brain States To assess the organization of spontaneous extracellular GABA dynamics in the cortex, I used seed-pixel correlation mapping of iGABASnFR2 fluorescence to investigate intracortical long- range connectivity across different brain states. By placing seed pixels in primary sensory regions, I generated correlation maps of spontaneous GABA dynamics across awake, NREM, and REM sleep states [Fig. 5(a)]. In the awake state, these maps revealed strong bilateral synchrony and widespread long-range connectivity among distant cortical regions. These spatial patterns of functional connectivity closely resemble those previously observed with excitatory signals using iGluSnFR and Ca2. imaging, suggesting that spontaneous GABA activity also reflects underlying anatomical connectivity and shared network drive. During NREM sleep, seed-pixel correlations were reduced, indicating a decoupling of large-scale inhibitory networks consistent with cortical slow-wave activity. In contrast, REM sleep preserved many of the bilateral and local connections seen in wakefulness, although with moderate reductions in correlation strength. I further quantified interhemispheric connectivity using pairwise correlation matrices of bilateral cortical regions [Fig. 5(b)]. These patterns, consistently captured using iGABASnFR2, highlight the sensor’s sensitivity to state-dependent fluctuations in extracellular GABA and its utility for mapping mesoscale inhibitory networks in vivo. 35 Fig. 4 Combined electrophysiological recording and mesoscale iGABASnFR2 imaging of GABA activity during wakefulness and sleep. Spatiotemporal dynamics of GABA activity over a 1-s period during wakefulness, NREM, and REM sleep. Scale bar, 2 mm. (b) Representative traces of GABA activity [retrosplenial cortex (RSC)], hippocampal LFP, and EMG power during wakefulness, NREM, and REM sleep in a head-fixed mouse. Baseline fluorescence (Fo) was 36 calculated as the mean signal over the recording session. (c) Expanded view of GABA activity, LFP, and EMG power corresponding to the time windows in panel (b). (d) Group mean normalized GABA signal across wakefulness, NREM, and REM sleep (n . 5 mice, Kruskal–Wallis test with Nemenyi post hoc correction; P . 0.003 overall, all pairwise comparisons significant at P < 0.05) (e) Spectral power of GABA signal in RSC across wakefulness, NREM, and REM sleep. (f) Cortical GABA activity correlation maps across quiet wakefulness, NREM, and REM sleep over two recording days. (g) Mean interhemispheric correlation of cortical GABA activity across wakefulness, NREM, and REM sleep (n . 3 mice). The mean correlation among units significantly increased from quiet wakefulness to NREM and REM sleep across all mice (P < 0.05, repeated measures ANOVA with Bonferroni-corrected paired t-tests). (h) Simultaneous recordings of cortical GABA signals, hippocampal LFP spectrogram, and EMG power during a REM-to- wakefulness transition. (i) Time-lapse montage showing cortical GABA dynamics during the REM-to-wake transition shown in (H). Scale bar: 2 mm. 2.3.6 Tiagabine Elevates Baseline GABA Levels but Dampens Sensory-Evoked Responses and Reorganizes Cortical Inhibitory Connectivity To examine how pharmacological inhibition of GABA reuptake influences cortical GABA dynamics, I administered tiagabine—a selective GAT-1 inhibitor (Dalby, 2000) —under 1% isoflurane anesthesia and monitored extracellular GABA levels using iGABASnFR2. Mice were head-fixed throughout the experiment. I first recorded visually evoked GABA responses (∼6 min), followed by a ∼15- min baseline period of spontaneous activity. Although the mouse remained head-fixed, tiagabine was injected intraperitoneally. Spontaneous activity was then recorded for another ∼15 min before delivering a second round of visual stimulation to assess post-tiagabine responses. B