University of California, Berkeley

We here report an optical system for live 3D microscopy that we call the Multifocus 25-camera microscope (M25). M25 can capture 25 simultaneous focal planes at >100 Hz. This design takes advantage of the latest generation of small, fast and sensitive CMOS cameras to simultaneously but separately capture 25 focal planes on individual camera sensors, arranged in a 5×5 array. Aberration-corrected multifocus microscopy (MFM) is a simultaneous 3D imaging technique based on diffractive Fourier optics that allows fast live imaging of biological samples. Classically, MFM has been applied in an optical layout where the entire stack of focal planes is recorded on a single camera. Our new design—that we here demonstrate for the first time— simplifies the multifocus optical layout so that it can be built from affordable off-the-shelf components and a set of custom manufactured diffractive gratings. (We made the M25 diffractive optics in the UCSB NanoFab, an open access facility). We are currently launching M25 in biological research projects to study neural circuit function in model organisms such as drosophila and fish.

University of California, Berkeley

Calcium imaging is a popular technique for tracking neural activity in awake animals. However, the traditional microscopes for calcium imaging are large and bulky, limiting the movement of the test subject. In addition, scanning techniques such as two-photon microscopy force a trade-off between acquisition speed and volume size. Here, we introduce a promising alternative microscope, called DiffuserCam, which is comprised only of a diffuser placed a small distance away from an image sensor. We demonstrate that this simple system can reconstruct 3D information from a single acquisition, enabling capture of large volumes without sacrificing time resolution. In addition, the system is small and lightweight, making it well-suited for head-mounted applications.

University of California, Santa Cruz

We here report an optical system for live 3D microscopy that we call the Multifocus 25-camera microscope (M25). M25 can capture 25 simultaneous focal planes at >100 Hz. This design takes advantage of the latest generation of small, fast and sensitive CMOS cameras to simultaneously but separately capture 25 focal planes on individual camera sensors, arranged in a 5×5 array. Aberration-corrected multifocus microscopy (MFM) is a simultaneous 3D imaging technique based on diffractive Fourier optics that allows fast live imaging of biological samples. Classically, MFM has been applied in an optical layout where the entire stack of focal planes is recorded on a single camera. Our new design—that we here demonstrate for the first time— simplifies the multifocus optical layout so that it can be built from affordable off-the-shelf components and a set of custom manufactured diffractive gratings. (We made the M25 diffractive optics in the UCSB NanoFab, an open access facility). We are currently launching M25 in biological research projects to study neural circuit function in model organisms such as drosophila and fish.

University of Colorado, Boulder

Light scattering limits the ability of current imaging techniques to image individual neurons inside the brain of living animals, to depths of about 1 mm. Multimode fibers (MMF), with their small footprint, high resolution and efficient light collection make good candidates for minimally invasive microendoscopes that can allow us to go deeper inside the brain. Imaging through multimode fibers typically requires a means of controlling the illumination on the object and recording a feedback signal corresponding to a set of such controlled illuminations. Wavefront shaping enables this control by employing an interferometric calibration of the fiber transmission matrix. Here, we present an approach to simplify the calibration process by employing the naturally occurring speckle patterns at the MMF output as structured illuminations for sampling the object. The fluorescence feedback corresponding to a set of different speckle patterns can allow object recovery using a reconstruction algorithm. This approach allows imaging with intensity-only measurements and using fewer number of illuminations than one would need for raster scanning. We further present a method to optimize the compressibility of our measurements by eliminating redundancies in our sampling illuminations. Our algorithm minimizes the correlations in the speckle patterns used for imaging, hence improving sampling efficiency and allowing higher compression.

Delft University of Technology

Archaerhodospin-3 (AR3) is a hepta-helical trans membrane protein, which functions as light-driven proton pump in the archaeon Halorubrum sodomense. It binds a molecule of all-trans retinal as a chromophore, resulting in an absorbance band in the visible region of the solar spectrum (λmax ~560 nm). AR3 and its mutants are widely used in neuroscience as optogenetic neural silencers and as fluorescent indicators of transmembrane potential. Here, we examine the impact of protein and chromophore modification on the spectral properties and activity of AR3. Two AR3 analog pigments from our study generated red-shifted pigments which could be activated by near-infrared light. Furthermore, one analog pigment in particular showed strongly enhanced fluorescence with an emission band in the near-infrared peaking around 815 nm. We anticipate that these near-infrared active AR3 pigments have potential applications in optogenetics, as fluorescent voltage-gated sensors and in bioenergy-based applications.

University of California, Berkeley

Recent advances in opsin development and the use of scanning tools have extended the spatiotemporal reach of optogenetics as a neurostimulation modality. However, scanning techniques for full 3D stimulation across a target volume of brain tissue require the use of dynamic z-axis focusing tools, existing versions of which have become a bottleneck for high stimulation rates. We propose a MEMS micro-mirror device to perform z-axis focusing at much faster refresh rates compared to conventional techniques like electrically tunable lenses or liquid crystal on silicon (LCoS). We have developed a micro-mirror array consisting of 32 individually addressable concentric rings with 30 V drive, 8.75 kHz refresh rate and a focusing full-width-half-maximum (FWHM) to range ratio of 4.8%. The 4.1 mm radius, 48 µm pitch array was designed using a custom optical simulation framework to work for wavelengths of up to 1040 nm and fabricated using the MEMSCAP PolyMUMPs-PLUS® process with post-processing for metallization. Ultimately, with ongoing work towards single pixel addressability, we aim to use this technology for the development of a high speed Spatial Light Modulator (SLM) for 3D optogenetics via scanless holography.

Delft University of Technology

Holographic light-shaping techniques are now successfully applied in all-optical studies of neurophysiology of neuronal circuits, focusing on stimulation and imaging from soma typically 10s of microns in diameter. Optogenetics at the scale of single synapses requires targeting light to spatial scales near, or exceeding, 1 micron with high NA optics, where anisoplanarity and sample-induced phase aberrations become limiting factors. Here we develop a holographic photostimulation system in which all forms of optical aberrations can be corrected, recovering diffraction-limited foci within a 3D volume corresponding to the larval zebrafish brain. We systematically correct for both system- and sample-induced aberrations within the full field of view by employing a sensorless adaptive optics approach. The resulting 3D map of aberrations is subsequently applied to computer-generated holograms of 3D point clouds through a fast compressive sensing weighted Gerchberg –Saxton algorithm. (Pozzi, P.; Maddalena, L.; Ceffa, N.; Soloviev, O.; Vdovin, G.; Carroll, E.; Verhaegen, M. Fast Calculation of Computer Generated Holograms for 3D Photostimulation through Compressive-Sensing Gerchberg–Saxton Algorithm. Methods Protoc. 2019, 2, 2.).

University of Florida

Somatosensory feedback (tactile and proprioceptive) during active sensing is an essential requirement for precise motor control. Most existing strategies to restore lost somatosensation via brain-machine interfaces have focused largely on biomimetic approaches to devising stimulation protocols in primary somatosensory (S1) cortex. While achieving S1 activity that mimics naturally-evoked responses has been studied within the context of perceptual reports, subsequent integration of artificial sensory feedback during active motor control is equally critical for moment to moment action selection during complex movement sequences. In the present study we seek to describe 1) cortico-cortical information transmission between S1 and M1 in response to natural tactile stimulation 2) cortical response to artificial stimulation of cortical and thalamic targets and 3) the differential capacity for biomimetic vs non-biomimetic artificial stimulation protocols during active sensation. In mice, M1 and S1 regions in the caudal forelimb area (CFA) are somatotopically adjacent and therefore simultaneous imaging of both is possible. Herein we developed a platform for interrogating somatosensory feedback circuits in head fixed behaving mice utilizing two-photon volumetric calcium imaging and cell-targeted optogenetic stimulation of cortical ensembles. Mice chronically implanted with cranial windows and virally-induced to co-express the calcium indicator GCaMP6f and redshifted opsin C1V1 in a custom-built behavior apparatus are trained to perform a simple sensory-guided motor task. The apparatus consists of computer screens to facilitate virtual navigation, lickports for delivering liquid reward, and a floating ball capable of wirelessly delivering vibrotactile feedback necessary for goal-directed navigation. M1 and S1 CFA responses to passive and active forepaw touch are first analyzed to functionally tag GCaMP6f+ neuronal ensembles. These ensembles are then photostimulated to achieve firing dynamics—in the absence of natural stimulation—that are either biomimetic or arbitrarily defined non-biomimetic states, and the resulting impact on task performance is assessed. In addition, subcortical microstimulation of VPL/VL/VM thalamus is being explored as an alternative upstream target for injecting artificial stimuli with concurrent calcium imaging of the same cortical targets. These data will describe the comparative capacity for multiple stimulation paradigms to induce artificial percepts and, critically, drive behavior.

University of Southern California

Norepinephrine is a neuromodulator that plays important roles in the regulation of arousal, memory formation and focused attention. The spatiotemporal dynamics of norepinephrine release and propagation in cortex during behavior is unknown. We use a newly-developed genetically encoded GPCR activation based norepinephrine indicator to monitor norepinephrine dynamics with ~100ms temporal resolution and single cell spatial resolution in awake behaving mice. We image these dynamics in superficial layers of barrel cortex during a head-fixed, whisker-guided object location task from the first day of training till task mastery. We observe highly dynamic fluctuations of norepinephrine at the sub-second level. These are coupled to sensorimotor features and reward. We find a sustained increase in NE during Hit and False alarm trials, relative to Miss and Correct Rejections. Auditory cues signaling the start and end of object availability drive strong NE release. Rewarded licks also drive strong NE release on the first day of training, but this release is attenuated across training sessions. We are currently analyzing the relationship of these dynamics to touch and whisker motion, as well as the spatial distribution of NE release and propagation across cortical columns.

Stanford University

Studying neural activities in brain requires rapid volumetric imaging in deep tissue. Conventional two-photon microscopy has a decent penetration depth in the scattered brain tissue, but its sequential point-scanning method limits the imaging volume and speed. We developed a multi-modality two-photon Bessel light sheet microscope that enables large field of view light-sheet imaging and high-speed 3D projection imaging for neural activities.The microscope uses a Bessel beam to illumination a 250-mm-wdie field of view. It has two distinct imaging mode: the light-sheet scanning mode and the projection mode. The light-sheet scanning mode enables sub-micron 3D imaging over the entire field of view. Structured illumination was added to reject the scattered emission and enhance image clarity on fine neural structure in deep brain tissue layers. The projection mode uses the two-photon Bessel beam to scan across tissue at up to 100 volumes per second. It captures dual-view projection images with two objectives oriented at 90-degree angle. Neural activities are imaged in two orthogonal projections simultaneously and analyzed in 3D at the millisecond timescale. Using this approach, we could acquire sub-micron resolution volumetric imaging at a maximal speed of 100 volumes per second.

Vision Institute

Temporally precise control of action potential generation of individual cells from a neuronal ensemble is desirable for dissecting circuit mechanisms underlying perception and behavior. Here we demonstrate that such degree of precision is achievable by using two-photon (2P) temporally focused computer-generated holography (TF-CGH) to control neuronal excitability at the supragranular layers of anesthetized and awake visual cortex in both male and female mice. Using 2P-guided whole-cell or cell-attached recordings in positive neurons expressing either of the three opsins ReaChR, CoChR or ChrimsonR we investigate the dependence of spiking activity on the opsin’s channel kinetics and show that in all cases the use of brief illumination (≤ 10 ms) induces spikes of millisecond temporal resolution and sub-millisecond precision, which are preserved upon repetitive illuminations up to tens of Hz. We also demonstrate that using large illumination spot covering the entire cell body and amplified laser at high peak power enable to reach such degree of temporal precision by using low excitation intensity (in average ≤ 0.2 mW/µm2), thus minimizing the risk for nonlinear photodamage effects. Finally, by combining 2P holographic excitation with electrophysiological recordings and calcium imaging using GCaMP6s, we investigate the factors, including illumination shape and intensity, opsin distribution in the target cell, and cell morphology, which affect the spatial selectivity of single- and multi-cell holographic activation. Parallel optical control of neuronal activity with cellular resolution and millisecond temporal precision should be particularly advantageous for probing neuronal connections and further yielding causal links between microcircuit dynamics and brain functions.

Harvard University

The stability of neural dynamics arises through a tight coupling of excitatory (E) and inhibitory (I) signals.  Genetically encoded voltage indicators (GEVIs) can report both spikes and subthreshold dynamics in vivo, but voltage alone only reveals the combined effects of E and I synaptic inputs, not their separate contributions individually.  Here we combine optical recording of membrane voltage with simultaneous optogenetic manipulation to probe E and I individually in barrel cortex Layer 1 (L1) neurons in awake mice.  Our studies show that L1 neurons integrate thalamocortical excitation and lateral inhibition to produce precisely timed responses to whisker stimuli.  Top-down neuromodulatory inputs drive additional excitation in L1.  Together, these results suggest a model for computation in L1 consistent with its hypothesized role in attentional gating of the underlying cortex.


University of California, Berkeley

The cortical microcircuit contains a variety of cell types with distinct inputs, biophysical properties, and outputs. While these have been investigated in slice or even under anesthesia, the interactions of local networks in awake, behaving animals are poorly understood. Studies of single cell stimulation have revealed surprisingly complex effects, but cells do not usually act in isolation, and synchronous activity has particular implications for activity propagation. To understand the complex cortical network, we have worked to refine our methods of ensemble stimulation. By characterizing the unique optogenetic input-output function of each cell, we can balance power delivery within ensembles to improve the fidelity and precision of stimulation while minimizing off-target effects. With these improvements, we are able to quickly stimulate spikes synchronously in groups of neurons. Using this, we’ve begun to investigate the effects of different spatiotemporal patterns of cell stimulation on recruitment of inhibition and excitation in an awake, behaving animal. We reveal how modulating the timing, synchrony, and cell distribution of spikes affects population activity, with a focus on the recruitment of inhibition. This provides insight into fundamental properties of the cortical circuit.

University of Sussex

Neural network computation occurs over a wide range of spatial scales including brain-wide circuits. Studying brain-wide computation therefore requires the ability to simultaneously monitor activity of arbitrary groups of neurons largely independent of their individual 3D positions. Using 2-photon imaging, this remains a challenging task: rapidly guiding the excitation laser between any two positions in 3D space is limited by the speed of movements of optical elements such as scanning mirrors or z-focusing mechanisms. Here, we use a simple optical trick to dramatically ameliorate this problem, thus allowing for rapid random-access 3D mesoscale imaging. With an investment below £1,000, we simplified a standard Sutter-MOM 2-photon setup into a non-collimated design, thereby extending the standard 0.5 mm x-y field of view by 7-fold to 3.5 x 3.5 μm using a standard x20 objective. In hand, an electrically tunable lens (ETL) placed in the still expanding pre-galvo laser permits rapid z-travel by up to 0.6 mm. Together, these simple modifications allow running arbitrary 3D scan-paths across a volume of 3.5×3.5×0.6 mm with a maximal travel time of 2 ms between any two points. Our design opens up a wide range of eminently useful scan-options, including mesoscale random-access scans as well as vertical and 3D curved scans that acknowledge the 3D structure of biological samples. We demonstrate the capability of this setup using a wide range of examples from mice, zebrafish and Drosophila.


Propagation of coherent light through a scattering medium generates a speckle pattern, which is detrimental for imaging applications. Since 2007, various wavefront shaping techniques have emerged to control this very complex figure of interference. In particular, they can be used to focus light behind scattering media. Still, all of these approaches require some feedback signal from the targeted focal point. Usually, the feedback is either measured with a detector placed behind the scattering medium, or recovered from implanted guide stars. Both approaches are invasive and thus not directly applicable when it comes to imaging at depth. To go beyond this major limitation, wavefront shaping has recently been combined with techniques such as acoustics or nonlinear optics. However, focusing non-invasively on extended fluorescent objects with linear excitation remain to date unresolved. Here we report on a new method allowing focusing inside a scattering medium using an incoherent linear optical signal as feedback, for instance fluorescence, in an epi detection geometry. Contrary to [Galya, S., Katz, Ori. “Noninvasive focusing through scattering layers using speckle correlations.” Optics Letters 44.1 (2019): 143-146.], our technique does not rely on speckle correlations and is also efficient in the multiple scattering regime. We use an optimization procedure to find the incident phase correction that maximizes the spatial variance of the linear fluorescence speckle, retro-reflected by the medium. The speckle pattern variance is the product of the contrast and the intensity of the speckle, meaning it is maximal when all the excitation light is focused on a single fluorescent target, since fluorescent signals emitted by multiple targets are summed incoherently. Experimentally, we demonstrate diffraction-limited focusing of light scattered after propagation through multiple layers of parafilm using the variance of the fluorescence speckle pattern as a feedback signal. This approach should be adaptable to several microscopy techniques and linear optical signals.

Northwestern University

Light-sheet microscopy techniques have helped make great strides in the field of developmental biology and neuroscience. Many variants of light-sheet systems have been developed which are targeted towards better resolution, lesser bleaching, deeper and faster imaging. Among these, a class of light-sheet systems prioritize steric access around sample. They are based on single high numerical aperture microscope objective facing the sample, which is responsible for both light-sheet creation and fluorescence detection. An off-axis illumination is used to provide oblique plane illumination from the main objective. These systems consist of three sequentially arranged microscope objective & tube-lens pairs and a remote scanning arrangement for volumetric imaging. However, existing two classes of remote scanning arrangements suffer from additional scan position based spherical aberrations or scan position dependent tilt in the light-sheet orientation. Therefore, we introduce a new planar mirror scanner based approach to enable tilt-invariant scanning of the light-sheet. Our scanned oblique plane illumination (SOPi) microscopy approach allows for rapid, true perspective 3D imaging of samples. Moreover, because of the tilt-invariance, we can also stitch multiple volume-scans together to perform large-scale imaging of a given sample. We use our microscope to show rapid volumetric functional imaging of live and behaving Zebrafish larvae at 10 volumes per second. With one -photon light-sheet, we also show imaging of larger than 1 mm3 volume in a thick mouse brain slice highlighting the superior depth penetration and cellular resolution imaging at larger than 0.3 mm depth.

Delft University of Technology

Recording membrane potential from multiple neurons simultaneously, in behaving animals, will have a transformative impact on neuroscience research, especially when sensitivity and time resolution of the recording are sufficient for high-resolution mapping of sub-cellular voltage dynamics. Genetically encoded voltage indicators (GEVIs) are a promising tool for this purpose, but robust voltage imaging in tissue remains a technical challenge. Building on recent results showing recordings of voltage dynamics from multiple neurons simultaneously in mouse hippocampus, in vivo, I will present our latest voltage imaging advances encompassing the evolution of photoactivatable near infrared GEVIs, and the creation of high speed microscopes and optical stimulation schemes which, together, enabled detailed imaging of dendritic voltage at unprecedented depths in acute mouse cortex tissue. These tools open the possibility for detailed explorations of subcellular signaling and network dynamics in the context of behavior.

University of California, San Diego

Hebb’s postulate states that correlated firing between two synaptically coupled neurons strengthens the synapse, forming the basis of learning, but a gap in knowledge remains regarding how changes at the synaptic level result in circuit-level learning. We are studying a network motif in the hippocampus where direct and indirect streams of information converge to facilitate comparison. The entorhinal cortex (EC) receives sensory information and sends it to CA3 and CA1 both directly and indirectly by routing it through the dentate gyrus (DG). Thus, CA3 and CA1 form convergence points for direct and indirect pathways and a hypothesized substrate for learning. To measure real-time network changes, we are building a novel two-photon microscope capable of recording calcium activity of multiple identified cell types within the DG and CA1/CA3 simultaneously in awake animals. We employ remote focusing and adaptive optics to image two optical sections in the mouse brain separated in depth by over 0.5mm for the first time. By manipulating inputs from the EC via electrical stimulation and during behavior, and measuring simultaneous network activity in multiple hippocampal subregions, we ask how information in converging pathways modifies network activity during learning.

Institut de l’Audition – Institut Pasteur, Paris, France

Over the past decade, there have been a number of seminal theories on the conditions for propagating information faithfully while maintaining stability in neuronal networks. Testing these theories is difficult as it requires accurate information about network architecture and precise control of the input. To overcome these limitations and test theoretical predictions, we developed a model based on cultures of cortical neurons and used patterned optogenetic stimulation techniques to stimulate the network with high spatiotemporal resolution. We varied the rate, correlation and number of stimulated neurons to show that the neuronal dynamics that was predicted by theory hold even under conditions far from ideal limits. These results – obtained in generic cultures, predicted by theory and observed in the intact brain – suggest that the neuronal dynamics is an emergent property of networks in general. Neural coding schemes include rate or temporal coding. To study what type of information is propagated in a network composed of several connected layers of neurons, we combined microfabrication techniques with neuronal culture to build multilayer networks in vitro. We optogenetically activated neurons from the first layer and vary systematically the temporal dispersion of a packet of light pulses applied to these neurons. We found that a brief stimulus with different temporal precisions resulted in the modulation of the firing rate. Whereas temporal information was better preserved in networks of low density, dense networks could propagate firing rate information. These results indicate that both temporal and rate information can be propagated in multilayer networks, depending not only on the input but also on network architecture and cell density.

Institute of Mediterranean Neurobiology INMED, INSERM

The developmental journey of cortical interneurons encounters several activity-dependent milestones. Between the end of the first and the second postnatal week, GABAergic neurons are transient preferential recipients of thalamic inputs and undergo activity-dependent migration arrest, wiring and programmed cell-death. Despite their importance for the emergence of sensory experience, and the role of activity in their integration into cortical networks, the collective dynamics of GABAergic neurons during that neonatal period remain unknown. Here, we study coordinated activity in GABAergic cells of the mouse barrel cortex using in vivo calcium imaging. We uncover a transient structure in GABAergic population dynamics that disappears in a sensory-dependent process. Its building blocks are anatomically-clustered GABAergic assemblies mostly composed by prospective parvalbumin-expressing cells. These progressively widen their territories until forming a uniform perisomatic GABAergic network. Such transient patterning of GABAergic activity constitutes a hidden functional scaffold that links the cortex to the external world prior to active exploration.

Aix-Marseille University, Inmed UMR1249

Today we have a detailed knowledge of the basic elements that make up the brains microcircuits such as computational capacity, plasticity, electrophysiological and neurochemical properties of single neurons. In contrast, the principles of how single cell properties relate to the emergent network dynamics in the developing mammalian cortex are still poorly understood. Neural systems as well as other biological networks show common organizational properties. Here we focus on the characteristically heavytailed distribution of connections between neurons, a hallmark of “hub neurons”. We characterize the emergence of this cell type in the barrel cortex of unanesthetized pups during development in vivo, with a specific focus on different topologies of GABAergic and glutamatergic populations. Using 2photon calcium imaging and holographic optogenetic stimulation we then entangle the necessary and sufficient conditions of how these neurons contribute to and synchronize network activity.

University of Strathclyde

Sleep is a physiological phenomenon observed throughout the animal kingdom. Humans, mammals and birds typically cycle between wakefulness, non-rapid eye-movement sleep (NREM) and rapid eye-movement (REM). REM sleep, associated with dream state in humans, is the shortest behavioural state and is thought to play a critical role in memory consolidation, but the mechanisms underlying this are not fully understood. Many brainstem and hypothalamic structures have been implicated in REM sleep regulation, however, the non-laminar nature of these structures make it difficult to physically isolate individual nuclei. One example of this are brainstem neurons within the pedunculopontine and laterodorsal tegmental nucleus (PPT/LDT). Cholinergic PPT/LDT neurons are active during wakefulness and REM sleep, but the nucleus also contains glutamatergic and GABAergic cell-types. As a result, we still lack a clear understanding of cell-type specific neural dynamics during sleep/wake states. To address this, we have developed an inexpensive, easy-to-build fibre photometry system to monitor cell-type specific population activity. We expressed GCaMP6s in pontine cholinergic neurons and monitored fluorescence signals, cortical EEG and EMG in freely moving mice during the sleep/wake cycle. We confirm state-dependent cholinergic activity during REM sleep and wakefulness. This system is a versatile tool that can be used to better understand the sleep regulatory circuits in a cell-type-specific manner and can also be easily modified to suit individual needs.

University of Florida

The posterior parietal cortex (PPC) has been implicated in a number of higher order cognitive functions, including perceptual decision making (PDM) and working memory (WM). However, the extent to which PPC neurons link mnemonic information of previous trial cues and outcomes with the accumulated evidence in the current trial remains highly elusive. Here we trained head-fixed mice on a two alternative, instructed delay forced choice task that requires the animals to discriminate a sequentially presented ‘cloud of tones’ sampled randomly from high and low frequency bands to earn water reward through corresponding left or right lick ports. Two-photon Calcium imaging of the genetically encoded calcium indicator GCaMP6f revealed differential engagement and sequential activation of many GCaMP6f+ neurons during evidence accumulation and delay intervals. A cell-targeted event-triggered closed loop photostimulation strategy is used to artificially synchronize GCaMPf+ neurons co-expressing C1V1+ that were not sequentially activated in previous trials. Changes in discharge rate of sequence dependent and non-sequence dependent neurons is characterized for both single cell and network level using our previously established functional connectivity techniques. This all optical neural interface permits studying the extent to which event-triggered stimulation can drive plasticity at the single cell and network levels during PDM and WM.

Delft University of Technology

Lattice light-sheet microscopy (Chen, et al., Science, 2014) is a recent innovation in biological imaging that reduces phototoxicity and improves spatial resolution over traditional cylindrical-focus light-sheet microscopy. While lattice light-sheet microscopy stands to become a workhorse for biological research, current instrumentation is expensive and complex for non-experts. Here we present a simple, custom lattice light-sheet module that improves on the state-of-the-art in two ways:

  1. we greatly simplify the theoretical framework for optimal lattice generation by selecting a limited number of operational modalities;
  2. we introduce a binary amplitude pupil mask that can be inexpensively generated by 3D printing.

We demonstrate the performance of our lattice light-sheet in live zebrafish embryos.


University of Modena and Reggio Emilia

Adaptive optics can be employed to correct for phase aberrations in optical microscopy and improve image resolution, contrast and signal intensity. However, conventional adaptive optics can only correct a single phase aberration for the whole field of view (isoplanatic correction) while, due to the highly heterogeneous nature of biological tissues, the sample induced aberrations in microscopy often vary throughout the field of view (anisoplanatic aberration), limiting significantly the effectiveness of adaptive optics. We present the possibility of using computer generated holography to create arrays of independently corrected foci within a sample, which can be used as excitation sources for highly parallelized, anisoplanatically corrected laser scanning microscopy applications. Results will be presented for two systems: a linear excitation system tested in Drosophila brains as well as Zebrafish embryos, and a multiphoton system for murine brain slices and in-vivo microscopy currently under development.

Imperial College London

Light field (LF) imaging [Levoy et al., 1996], also known as plenoptic imaging, is a 3D optical imaging technique that allows to simultaneously capture both the spatial and angular information of the incident light in a 4D plenoptic representation. Therefore, it enables the computational reconstruction of 3D volume or reconstruction of images at different planes and perspectives from a single camera frame. By placing a microlens array (MLA) at the intermediate imaging plane, light field microscopy (LFM) [Levoy et al., 2006] is able to acquire such 4D data. Combining with calcium imaging technique, LFM achieves rapid scan-less volumetric functional imaging of biological specimens and facilitates the recording of neuronal population activity at high frame rates.

In this work, we propose an efficient approach to demix calcium transients from functional LFM data. The procedure includes: (1) Localizing overlapping neurons using our EPI based location detection technique; (2) Using a LF wave optics model [Broxton et al., 2013] to generate initial neuron footprints; (3) Solving a non-negative matrix factorization (NNMF) problem in an alternating manner to factorize spatial components from temporal components, that is, separating the neuron footprints from their calcium transients. There have been a variety of algorithms to solve the NNMF problem. We adopt alternating least-squares (ALS) algorithm [Berry et al., 2007] considering its fast convergence and good consistency property. Owing to the EPI based location detection, one of the merits of the proposed approach is that we do not need to perform 3D deconvolution, which alleviates the time/memory consumption issue.

New York University Langone Health

A central goal of systems neuroscience is to decipher the spatiotemporal code of neuronal activity. Holographic two-photon (2P) optogenetics combined with calcium imaging can be used for all-optical imaging and stimulation. Here, we developed and used such a system to probe the detection of evoked neuronal activity at cellular and single action potential resolution, with millisecond precision matched to the demanding characteristics of the olfactory system. We performed 2P stimulation and imaging in the olfactory bulb of awake, head-fixed mice. To boost excitation efficiency and temporal precision, our system used a low-repetition rate, amplified laser for stimulation. We performed two-photon guided cell-attached electrophysiological recordings and were able to generate action potentials in the cell during photostimulation and up to a few ms after stimulation. We next trained mice to detect 10 ms, ~single spike 2P stimulation of a specific subset of 30 neurons. We further explored this by testing detection performance while varying the number of targeted neurons. Detection performance varied monotonically as a function of the number of neurons targeted. The data suggest that mice are exquisitely sensitive to changes in the synchronous activation of <20 olfactory bulb neurons, on the order of several spikes. To rule out behavioral detection of confounds, we used a sham stimulation that didn’t evoke spiking. We increased the laser pulse duration thereby reducing the 2P signal. When presenting sham photostimulations, detection performance dropped to chance level, demonstrating that photostimulation detection performance is determined by the spiking of targeted neurons.

Columbia University

Perception of sensory information by the brain is highly dependent on the information context, where responses to the most informative, or “deviant”, stimuli are selectively amplified. Deficits of deviance detection appear in patients with schizophrenia, suggesting its high importance for normal information processing and cognitive function. Here we use two-photon calcium imaging in awake mice undergoing the “oddball” paradigm and learn the neural activity structure encoding informational context in visual and auditory cortices. Analysis of trial to trial and averaged stimulus evoked responses reveals a low dimensional organization, consisting of independent, contextually selective subpopulations of neurons (“ensembles”). One major ensemble includes cells selectively responding to stimuli that are deviant, i.e. deviance detectors, number of which increases from primary to secondary sensory cortex. Imaging in schizophrenia mouse models revealed that the low dimensional context specific structure is disorganized in disease. Further we use two-photon holographic stimulation along with go/no-go deviance detection behavioral readout to assess the functional importance and stability of deviance detection ensembles. We optically manipulate the low dimensional activity patterns in the primary sensory cortex in conjunction with the oddball paradigm to imprint an altered context specific activity and behavioral response. Furthermore, with optical imprinting in in schizophrenia mice we attempt to enhance the deviance detection behavior aiming at improving the functional phenotype of the disease.

Imperial College London

Light Field Microscopy (LFM) is a high-speed imaging technique that allows to capture a three-dimensional (3D) image of an entire volume in a single camera snapshot. Furthermore, a Light Field microscope is a simple modification of the standard microscope: The image sensor is pushed back, and a micro lenses array is interposed between the tube lens and the image sensor [Levoy et al., 1996]. Therefore, LFM is a tool that will potentially allow to monitor the activity of the intricate network of neurons that form the brain tissue. Specifically, this process can be studied through non-invasive techniques that use calcium-sensitive proteins or voltage-sensitive dyes since dyes work as great optical reporters of the neuron activity [Pegard et al.,2016]. In this work, we present a technique that uses aside information coming from a wide-field stack to guide the 3D Light Field deconvolution. This method is implemented to deconvolve static samples coming from Ca2+ LFM data. Our method uses the LF model proposed in [Broxton et al., 2013], but an approximation is used to accelerate computation, essentially, the super sample factor defined in [Broxton et al., 2013] is reduced to 1. Then, a 3D mask is computed from the widefield stack. The mask selects the regions where we will enforce smoothness, the regions where the neurons are located. Finally, ADMM is used to solve the constrained inverse problem. One contribution of the method is that the computation is alleviated since an approximation is used to model the system.

Georgia State University

Cortical processing of sensory events is influenced by context. For instance, neural responses to a repetitive or “redundant” stimulus are attenuated, but, if the same stimulus is novel or “deviant”, responses are augmented. This contextual modulation of cortical processing is likely a fundamental function of neural circuits, yet an understanding of how it is computed is still missing. Using volumetric two-photon calcium imaging in awake mice presented with a visual “oddball” paradigm, we identified three distinct, spatially intermixed subsets of neurons in V1 which respond to the same stimulus under separate contexts, including a subnetwork of neurons which responds selectively to novel events (“deviance detectors”). Similar subnetworks, including robust deviance detectors, were also present in primary auditory cortex (A1) during an auditory “oddball” paradigm. Deviance detector neurons were distributed across layers 2-5, though they were more common in superficial layers. Further, we show that contextual preferences likely arise locally since they are not present in bottom up inputs from the thalamus and do not depend on top-down inputs from prefrontal cortex. The functional parcellation of cortical circuits into specialized ensembles that encode stimulus novelty provides a circuit mechanism for the brain’s processing of context, a function which is key for survival and fundamentally deficient in many psychiatric disorders.

Insitut de la Vision

Over the past years, progress in opsin engineering and light delivering approaches have enabled neuroscientists to drive and read neural circuits with single action potential precision and cellular resolution. Many variants in microbial opsins have been recently engineered, to fasten their kinetics, improve their conductance, confine their expression and shift their absorption peak. On the other hand, advanced wave-front shaping approaches have been developed to optimize temporal resolution and precisely guide light through tissues using either scanning or parallel two-photon(2P) excitation. If joint progresses in these two fields have significantly widened the possible experimental configurations for 2P optogenetics, they have also made it more difficult to choose the combination of opsin and illumination configuration that better matches the experimental requirements. In this work, we investigated, experimentally and theoretically, the temporal dynamics of photoevoked currents using opsins with different kinetics under different 2P-illumination conditions. By using a fast (f-Chrimson), an intermediate (CoChR) and a slow opsin (ReachR) in cell culture and brain slices, we studied the role of opsin kinetics on temporal resolution and precision under 2P-scanning and -parallel illumination. We then used 3- and 4-states models to reproduce the experimentally photoevoked currents under both illumination conditions. This characterization and the prediction of the model can provide crucial information to guide the opsin choice and development and to optimize the design of two-photon optogenetics experiments.

University College London

To understand brain function, it is essential to identify how information is represented in neuronal population activity and how it is transformed by individual neurons as it flows through microcircuits. Two-photon microscopy is a core tool for this because it enables neuronal activity to be monitored at high spatial resolution deep within brain tissue in behaving animals. However, the temporal resolution of conventional galvanometer-based two-photon microscopy severely limits measurements of fast signaling in 3D neuronal circuits. Acousto-optic lens (AOL) microscopy, which enables fast focussing and selective random-access imaging of regions of interest distributed within the imaging volume, has substantially improved the temporal resolution of 3D two-photon microscopy. We have recently extended the functionality of this technology by utilizing nonlinear acoustic drives to enable ultra-fast line scanning (up to 40 kHz) in any arbitrary direction in X, Y and Z. This allows structures extending in the Z-dimension, such as dendritic trees to be imaged much more efficiently. Our prototype nonlinear AOL two-photon microscope, which also includes real time correction of brain movement, can selectively image entire 3D dendritic arbours at higher temporal resolution than previously possible. We demonstrate the utility of this new technology for high-speed multiscale 3D imaging of neural circuits in awake behaving animals by recording from the entire dendritic tree of L2/3 cortical pyramidal cells, while simultaneously monitoring the neuronal population activity in the surrounding network.

Janelia Research Campus

Short-term memory (STM) is associated with persistent neural activity without sustained input, arising from the interactions between populations of memory-less neurons. A variety of neural circuit motifs can account for the measured neural activity. A mechanistic understanding of the emergence of persistence requires probing network connectivity between functionally characterized neurons. We used targeted photostimulations of small (< 10) groups of neurons, while imaging the response of hundreds of other neurons, to probe network interactions in anterior-lateral motor cortex (ALM) of mice performing a delayed response STM task. Mice were instructed with brief auditory stimuli to make directional movements, but only after a three second delay epoch. ALM contains neurons with delay epoch activity selective for both left and right choices. Targeted photostimulation of groups of neurons during the delay epoch enables observation of functional organization of recurrent interactions underlying short-term memory. Photostimulation produced changes in activity that persisted for several seconds following the offset of photostimulation. Photostimulation revealed strong coupling between neurons sharing similar patterns of selectivity and produced behavioral biases that were predictable based on the selectivity of the perturbed neuronal population. These results are consistent with networks in which persistence is produced by local excitatory modules of recurrently coupled neurons.

University College London

To identify causal links between sensory stimuli, neural activity, and behaviour it is important to combine psychophysics with approaches for measuring and manipulating neural activity with cellular resolution. Recent developments in simultaneous ‘all-optical’ two-photon calcium imaging and optogenetic photostimulation are ideally suited for this type of causal investigation (Packer et al, 2015). Importantly, the coupling of a programmable spatial light modulator (SLM) into the photostimulation light path enables neurons to be selectively manipulated based on their functional identity – which is not possible using conventional optogenetic strategies. We harnessed this novel technology in combination with a two-alternative forced choice (2AFC) decision-making task to investigate sensory stimulus encoding in head-fixed mice. Co-expression of a calcium indicator and an optogenetic actuator in layer 2/3 barrel cortex permitted all-optical interrogation of neural circuitry during behaviour. During task performance, a small proportion of sampled L2/3 neurons showed responses that were modulated by contralateral stimulus amplitude and behavioural choice. We are currently performing targeted two-photon photostimulation experiments in 3D to manipulate stimulus and choice-informative neurons specifically during behaviour to explore the mechanisms by which sensory information is encoded in cortical circuits, and how this information may influence behaviour.

University College London

Understanding how the structure of connectivity underlies the processing carried out by cortical circuits is a fundamental problem in neuroscience. Layer 2/3 of mouse visual cortex consists of functionally distinct subnetworks of recurrently connected neurons. Neurons sharing similar stimulus response properties (i.e. cotuned to the same stimuli) preferentially share monosynaptic connections. This specific synaptic connectivity rule may facilitate and maintain robust representations of visual stimuli even under situations when those stimuli are weak or degraded. Here we have trained mice on a visual detection task and used simultaneous two-photon calcium imaging and two-photon optogenetics to ask: 1. How does this pattern of paired connectivity extend to, and influence, activity at the population level in vivo? and 2. How does the functional signature of subnetworks impact upon the neural representation, and ultimately the behavioural salience, of weak or ambiguous stimuli? To address these questions, we performed targeted photostimulation of ensembles of either cotuned, non-cotuned or non-stimulus responsive L2/3 visual cortex neurons and observed the response of the local network as well as the animal’s performance in reporting the current visual stimulus. Our preliminary results show that when a sufficient number of cells are photostimulated during reduced contrast visual stimulus presentation, the behavioural response rate to that stimulus is enhanced. We are currently dissecting how enhancing the behavioural response to a sensory stimulus depends on the functional identity of the photostimulated ensembles and their engagement of the local network. These results provide a bridge between connectomics, sensory stimulus coding and behaviour.

University of Tokyo

Genetically encoded calcium indicators (GECIs) are widely used for detecting calcium transients in somata and processes that are triggered by neuronal activities. In vivo calcium imaging with two-photon microscopy is usually achieved by viral gene delivery using adeno associated virus (AAV). However, viral injection sometimes produces heterogeneous and nonstationary expression in neighboring neurons. In addition, current GECI expressing transgenic mice are regulated by Cre and tTA. Here, we generated a novel transgenic line expressing high SNR green GECI, named G-CaMP9a, in Flp recombinase dependent manner in various neuronal subpopulations from the Rosa26 locus, driven by a strong CAG promoter. Combining this reporter mouse with appropriate Flp delivery methods (Flp-expressing transgenic mice, AAV, or in utero electroporation) produced robust and stable G-CaMP9a expression in defined neuronal populations in the cortex. In vivo imaging with two-photon microscopy revealed that this transgenic mouse can detect spontaneous and whisker-evoked Ca2+ transients in excitatory neurons with cellular resolution in the barrel cortex. Also, in pyramidal neurons of the visual cortex, visually evoked G-CaMP9a signals showed orientation and direction selectivity. Furthermore, this mouse could reliably detect spontaneous activities from cortical interneurons. Our results show that this new reporter line allows investigation of neuronal activity in defined populations in vivo and will notably facilitate dissecting functional relationships of neural networks.


Laboratoire Kastler Brossel, CNRS

Fluorescence represents nowadays an irreplaceable tool to non-invasively probe neuronal activity in the mammalian brain. However, when neurons are very deep, fluorescent light propagation through the upper layers of tissues scatter and scramble seemingly beyond recovery the original information. This problem can be circumvented in many ways, however a non-invasive method to record fluorescent functional activity from several sources simultaneously, in the multiple scattering regime, is still lacking. Our work is based on recent signal processing insight, indicating that multiple scattering of light does not destroy its information content. We generate temporally varying fluorescent sources, emulating signals from neuronal calcium activity reporters, and we let the light go through a highly scattering mouse skull. We demonstrate we can retrieve the temporal traces of individual neurons in such regime. Importantly, we do not rely here on ballistic light, nor on the presence of speckle correlation across the field of view.

NTNU / Kavli institute

Recent developments in miniaturized microscopes have furthered the quest to visualize brain activity and structural dynamics in freely moving animals engaged in self-determined behaviors. Recently, we have published a fast, high-resolution, miniaturized two-photon microscope (FHIRM-TPM) capable of imaging commonly used biosensors at high spatiotemporal resolution. Here we present an upgrade of this technique, named FHIRM-TPM 2.0. A customized miniature objective and scan lens now enable larger field of view scanning and at the same time offer larger working distances which facilitates imaging through chronic glass windows. With FHIRM-TPM 2.0 we were able to simultaneously record the activity of hundreds of cortical neurons in freely moving mice . We anticipate that this new technique will provide a new generation of miniaturized, high-performance imaging tools for neuroscience to image structural and functional dynamics in freely-behaving animals over multiple spatial and temporal scales.

Institut de la Vision

Visual stimuli are initially processed in the retina where information is transferred from the intermediate bipolar cell layer – which integrates the photoreceptor responses – to the ganglion cell layer – the output of the retina. This processing results in different type of ganglion cells, each encoding for a specific visual feature. Direction selective ganglion cells (DSGC) respond selectively to a motion direction and maintain such a tuning across a broad range of light levels, despite different circuits shape their responses from bright to dim light. In particular, it remains unclear whether the rod bipolar cell (RBC) primary pathway, a key player in night vision, is involved in the modulation of directional selectivity. In order to understand its contribution to DSGC response, we designed an all-optical system to activate single cells in the BC layer by two-photon (2P) temporally focused holographic-based illumination, while recording the activity in the ganglion cell layer by 2P Ca-imaging. We observed that RBCs provide an asymmetric input to DSGCs, suggesting they contribute to their direction selectivity. This indicates that every circuit providing an input to direction selective cells in the retina can generate direction selectivity by itself. This method paves the way towards complete functional connectomics of the retina and is extendable to other brain regions, opening the way for a precise interrogation in vitro and in vivo of multi-layered circuits.

University of California, Berkeley

We present an algorithm to guide neural modulation experiments with the objective of estimating the neural network connection under investigation in the optogenetics setting. Specifically, we introduce an algorithm based on optical stimulation experiments and calcium trace imaging (GCaMP) as a proxy for neuron membrane potential to optimally estimate the neural network at the scale of hundreds of microns. 3D-SHOT, an optical neural modulation technique that allows stimulation of custom ensembles of neurons with cellular resolution and millisecond time precision, is applied to conduct the network estimation experiment. With fluorescent calcium trace indicator, GCaMP, that serves as an indicator of an action potential event, we can observe the activity of both the stimulated and neighboring neurons also with cellular resolution in a three-dimensional voxel. With these technologies at hand, we introduce a deterministic, linear autoregressive model of an arbitrary order with thresholding to describe the population-level calcium trace dynamics of the neurons under one field of view. We leverage the ideas in compressed sensing and parallel computing to efficiently estimate the parameters of the aforementioned model, and subsequently establish the connectivity matrix that characterizes the effective interactions among the observed neurons as well as the autoregressive coefficients that describe the temporal dynamics of GCaMP corresponding to the sampling rate of imaging. With the connectivity matrix describing the partial dynamics of neuronal interactions, we hope to provide both spatial and temporal signatures of a local region of the brain.

University of Chicago

Place cells in the hippocampus are thought to form a cognitive map of space and a memory of places. How this map forms when animals are exposed to novel environments has been the subject of a great deal of research. Numerous technical advances over the past decade greatly increased our understanding of the precise mechanisms underlying place field formation. In particular, it is now possible to connect cellular and circuit mechanisms of integration, firing, and plasticity discovered in brain slices, to processes taking place in vivo as animals learn and encode novel environments. Using high resolution imaging methods capable of resolving activity from fine structures such as dendrites and axons during behavior, we reveal the dendritic mechanisms most likely responsible for the formation of place fields in novel environments.

Institut de la Vision

In recent decades, optogenetics has been transforming neuroscience research, enabling neuroscientists to drive and read neural circuits. The recent development in illumination approaches combined with two- photon (2P) excitation, either sequential or parallel, has opened the route for brain circuit manipulation with single-cell resolution and millisecond temporal precision. Yet, the high excitation power required for multi-target photostimulation, especially under 2P illumination, raises questions about the induced local heating inside samples. Here, we present and experimentally validate a theoretical model that makes it possible to simulate 3D light propagation and heat diffusion in optically scattering samples at high spatial and temporal resolution under the illumination configurations most commonly used to perform 2P optogenetics: single- and multi-spot holographic illumination and spiral laser scanning. By investigating the effects of photostimulation repetition rate, spot spacing, and illumination dependence of heat diffusion, we found criteria that could allow designing a multi-target 2P optogenetics experiment with minimal sample heating.

Champalimaud Foundation

Goal-directed actions depend on brain circuits that integrate movement and value information to maximize reward rate. Cortical and striatal circuits are key structures in this process and regulate motor behavior via the striatonigral (direct) and striatopallidal (indirect) pathways. Importantly, both pathways increase activity during movement. Some models of striatal function propose that the direct pathway promotes specific actions whereas the indirect pathway simultaneously suppresses conflicting actions. However, we and others have shown that both pathways are rather action-specific during natural behaviors. How can these two apparently conflicting views be reconciled? Here we investigated this question by studying the organization of motor cortical and striatal direct- and indirect-pathway neurons in a probabilistic three-alternative choice task. In this task, mice self-initiate trials to obtain water rewards under changing reward contingencies. Importantly, this behavioral paradigm allowed to distinguish: (i) movement- versus value-related, and (ii) action-promoting versus action-suppressing neural signals. Using in vivo microendoscopic calcium imaging, we measured the activity of large populations of cortical and striatal projection neurons. Simultaneously, behavior was recorded using video, lick sensors, and a head-mounted motion sensor for the precise quantification of the motor output. These experiments test current and novel action selection models of cortico-striatal function.

Northwestern University

Resolving the electron momentum in photoemission spectroscopy allows experimental determination of the energy dispersion relations in solid state materials. The new generation of momentum-resolved photoemission spectroscopy promises simultaneous 3D detection of electrons in the (kx, ky, E) momentum space, giving rise to the electron momentum microscopy. The high sensitivity of low-energy photoelectrons to external field perturbations results in symmetry distortions in the image-based datasets and has become a major issue affecting downstream data analysis. To preserve intensity features but lacking in examples for training algorithms, the physical knowledge of the symmetry should be invoked to assist the algorithm design. We propose an optimization-based feature-preserving symmetrization scheme that operates at the image level and is translatable to the single-event data for distortion correction. The algorithm has been integrated into our distributed data processing pipeline that handles single event streams and may find broader use in the electron and X-ray diffraction community as well as time-of-flight imaging spectroscopies.

University of Minnesota

Cortical activity underlying cognitive function is classically thought to be exclusively mediated by neurons. In contrast, astrocytes are considered to play homeostatic roles, without being directly involved in brain function. Yet, astrocytes are emerging as important cells in brain physiology because they interact with neurons at tripartite synapses, responding to neurotransmitters with rises in internal calcium levels and releasing of gliotransmitters that regulate synaptic function. While astrocyte calcium and consequent synaptic regulation has been largely documented at the cellular level, astrocyte network activity and its impact on neuronal network function has been minimally explored. We have investigated this issue by combining simultaneously acquired two-photon microscopy in vivo to monitor astrocyte activity and electrocorticogram (ECoG) recordings to monitor neuronal network activity in the somatosensory cortex (S1) during hind paw stimulation. We found: (1) sensory-evoked astrocyte calcium responses were reliable to repeated stimulations; (2) astrocytes responded to different intensities, durations and frequencies of the stimuli in a stimulus-dependent manner; (3) somatosensory-evoked astrocyte responses were associated with rises in cortical gamma power (30-80 Hz), but not low frequency delta activity (0-4 Hz); (4) sensory-evoked cortical gamma activity was increased in transgenic mice that had abated calcium activity; (5) specific activation of S1 astrocytes selectively expressing Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) reduced the sensory-evoked gamma power. We conclude that cortical astrocytes respond to sensory inputs with calcium rises that in turn regulate sensory-evoked neuronal network activity.

Fondazione Istituto Italiano di Tecnologia

Imaging deep brain areas over large spatial scales is a major challenge which often requires the removal of substantial portions of brain tissue overlying the target region. Here, we developed a new generation of aberration-corrected microendoscopes that combine an extended field of view (up to 400 microns x 400 microns) with minimal invasiveness (probe diameter ≤ 500 microns). Corrected microendoscopes had up to nine folds larger field-of-view compared to uncorrected probes. We used these corrected probes to monitor hundreds of GCaMP6s-expressing neurons in the ventral posteromedial thalamic nucleus of awake head-restrained mice while monitoring whisking, locomotion, and pupil diameter. We found that resting periods – characterized by no locomotor activity and no whisker movement – displayed calcium signals that were sparsely distributed both across neurons and time. In contrast, active periods – characterized by locomotion, enlarged pupil diameter, and whisking in air – showed an increase in both the frequency and amplitude of calcium events across thalamic neurons compared to resting periods. Preliminary analysis also evidenced that active periods were characterized by the emergence of a number of neural pairs displaying significantly correlated activity over hundreds of microns, suggesting the presence of distributed functional subnetworks. These findings demonstrate that corrected microendoscopic probes are a powerful method to measure the activity of distributed neural networks in deep brain regions and to reveal how these activities are dynamically modulated by locomotion and arousal with minimal invasiveness.

University College London

Hippocampal place cells fire when an animal occupies a specific location in an environment and are thought to play an important role in memory formation and spatial navigation. However, the causal role of place cell firing in driving decision making during spatial navigation has remained elusive. Addressing this question requires targeted manipulation of specific place cells in navigating animals. Here we have used an “all-optical” approach for targeted manipulation of place cell activity in head-fixed mice navigating a virtual reality environment. Using simultaneous two-photon calcium imaging and two-photon optogenetic holographic stimulation, we functionally define and manipulate the activity of populations of place cells with a specific firing field location and examine the impact on behaviour. Mice were trained to stop and lick for reward at a location on a linear track. Place cells were identified and grouped into neurons which fired either at the rewarded zone or an equivalent area near the start of the track. We then selectively activated either the reward zone or non-reward zone place cells at a separate equidistant location on the virtual track to test whether the behavioural output associated with the originally encoded location can be retrieved. Our results demonstrate that the activation of place cells with a specific spatial tuning can drive behaviour towards that which is associated with the represented location. This approach enables us to probe the causal relationship between hippocampal neural activity patterns, spatial navigation and memory retrieval to guide behaviour.

University of Oxford

Perception is shaped not only by sensory input but also by our past experiences, indicating sensory processing involves integration of both external inputs and internal information. While we have a good understanding of the flow of sensory information through the ascending sensory pathways, comparatively little is known about how top-down internal states influence sensory processing. To address this question, we focus on one of the major descending circuits in the mouse auditory system. Specifically, the shell of the inferior colliculus (IC). One of the defining features of this auditory midbrain structure is the dense feedback projection which links it with the auditory cortex (AC), making it a prime model system in which to study the role of descending cortical projections. By utilising an adeno-associated virus-mediated anterograde transsynaptic approach, we selectively targeted neurons in the shell of the IC that receive input from the AC and used two-photon calcium imaging to monitor the activity of these neurons while the animals learned an auditory detection task. In some of the animals, we removed the AC. This left the animals’ ability to learn and perform the detection task largely intact and, thus, allowed us to assess how sensory and task information is represented in the shell of the IC, and how it is affected by the absence of cortical feedback.

Institut Pasteur

Transforming similar sensory information from the outside world into distinct memory representations is a major challenge for the brain. The hippocampus has been suggested to play a key role in producing segregated memories of similar objects and events in a process commonly referred to as “pattern separation”. Where and how this function is accomplished in the hippocampal circuit is currently debated. To address this question, we perform 2-photon imaging from the hippocampus of head-fixed mice navigating in a linear virtual-reality environment while performing a visual discrimination task. We find that the population activity in the input region of the hippocampus, the dentate gyrus, is sensitive to small changes in visual cues, producing decorrelated activity patterns in response to small changes in wall textures. By contrast, larger changes in the environment, including replacement of all spatial landmarks, are required to produce decorrelation in CA1. Relating behavioural performance with neuronal population activity patterns suggests that decorrelation in the dentate gyrus depends on task engagement of the animal. Thus, pattern separation in the dentate gyrus seems to be determined by behavioural variables and the degree of differences in the spatial environment. Our findings provide an understanding of how and where in the hippocampal circuit distinct memory representations of similar experiences are produced.

University of North Carolina at Chapel Hill

Optical imaging methods using calcium indicators are critical for monitoring the activity of large neuronal populations in vivo. Imaging experiments typically generate a large amount of data that needs to be processed to extract the activity of the imaged neuronal sources. While deriving such processing algorithms is an active area of research, most existing methods require the processing of large amounts of data at a time, rendering them vulnerable to the volume of the recorded data, and preventing real-time experimental interrogation. In this talk I will describe CaImAn Online, a framework for the analysis of streaming calcium imaging data, including i) motion artifact correction, ii) neuronal source extraction, and iii) activity denoising and deconvolution. Our approach combines and extends previous work on online dictionary learning and calcium imaging data analysis, to deliver an automated pipeline that can discover and track the activity of hundreds of cells in real time, thereby enabling new types of closed-loop experiments. We apply our algorithm on several experimental datasets, benchmark its performance on manually annotated data, and show that it reaches near-human performance.

Sorbonne Université

Managing photo-induced heating effects is essential during in vivo 2-photon holographic optogenetic stimulations. To measure temperature, we developed a minimally-invasive fiber-based microscale probe with thermal properties closely matching those of tissues. The same dual-core glass fiber is used to excite (980 nm) and collect fluorescence from a ~10 µm Er/Yb co-doped crystal. The temperature is deduced from the monitoring of two fluorescence bands. Temperature changes inside the mouse brain are measured during in vivo holographic stimulation with ms and sub-Kelvin range temporal and temperature resolutions. In order to control temperature and understand its effects, we developed a method using a spatial light modulator to tailor the illumination sent on an array of absorbing plasmonic nanoparticles. Calculating the optimal heat source map allows to pre-compensate heat diffusion effects and to control the shape and value of temperature increase profiles at the micrometer scale. This fast (sub-ms), accurate and spatially reconfigurable temperature control method can be used in vitro to thermally target a given population of cells or organelles of interest. This microscale combination of measurement and shaping of the temperature opens new strategies to optimize optogenetic stimulation and locally study the response of neurons to thermal activation.

Northwestern University

Measuring the quality of the neural code is a critical part of many experiments. For example, hippocampal place cells fire in response to an animal’s location in an environment. For extracellularly electrical recordings, a cell’s “spatial information” has been measured by estimating the mutual information between the action potential train and the position of the animal. This provides a measure of the “information rate”, in either bits-per-second or, more commonly, bits per action potential. Information rate can also be applied to measuring information about other variables, such as heading direction, speed, and time. However, it depends on assuming a short-timescale counting process where events are independent, such as in spike trains. These assumptions are violated for fluorescence traces collected via calcium imaging, which are nonstationary and can take on non-integer values. Statistical analysis when assumptions are violated can be done, but result in biases that will change the way we interpret our results. To investigate the effects of indicator dynamics on the information rate, we first acquired behavioral datasets (locomotion speed and track position vs. time) from mice navigating in virtual linear tracks. To these we added simulated spike trains containing varying amounts of spatial and speed information. The spike trains were convolved with different calcium transient shaped kernels, and noise was added. This dataset provided the opportunity to examine the correspondence between information rates estimated from calcium traces versus spike trains, from which we provide recommendations on when its use is valid. Our results may also prove useful for estimating information contained in fluorescence traces from other functional indicators.

Weizmann Institute of Science

The medial prefrontal cortex (mPFC) mediates a variety of cognitive functions, from attention and long-term memory to social behavior, via its vast and diverse connections with cortical and subcortical structures. Understating the patterns of synaptic connectivity that comprise the mPFC local network is crucial for deciphering how this circuit processes information and relays it to downstream structures. Here we present a robust, high-throughput technique for mapping functional synaptic connectivity among identified mPFC neuronal populations. We co-express a channelrhodopsin and a calcium indicator in a cell population of interest, allowing us to simultaneously perform two-photon optogenetic stimulation and calcium imaging at single-cell resolution using a single laser source. To assay functional synaptic connectivity, we first apply automated detection of cell bodies in three dimensions in an acute brain slice; we then perform serial stimulation and imaging over all individual cells, while recording synaptic responses from a single postsynaptic neuron using patch-clamp whole-cell electrophysiology. Calcium-related fluorescence collected during optogenetic stimulation serves to verify spiking in the targeted cells. Using this method, we can probe synaptic inputs onto each recorded neuron from up to several hundreds of cells in the local circuit, thereby reconstructing detailed information regarding the probability, strength and spatial distribution of synaptic connections. This allows unbiased detection of sparse synaptic connections as well as mapping of connectivity parameters onto behavioral performance within individual animals. This methodology can be applicable to any network of cells and can be easily implemented in most two-photon systems.

Columbia University

One of the biggest challenges in neuroscience is to decipher the neural code, i. e., the relation between activity of neurons and behavior or mental states by understanding the function of the neural circuits and the spatiotemporal patterns of activity across neuronal populations. We hypothesize that it may be possible to use synthetic biology approaches to generate in vitro living 3D biological networks with increased functional and structural complexities, while still maintaining systems-level experimental access for observation and high spatio-temporal activation/de-activation with sculpted light. To explore this in vitro, we are developing 3D assemblies of highly intra-connected groups of neurons (called neurospheres), and inter-connecting them in a desired configuration to yield patterned cultures of groups of neurons, named neurosphere networks (NNet). An AAV viral vector was used for expressing GCaMP6f (calcium activity reporter) in all neurons to allow rapid imaging of the entire network activity at sub-cellular resolution. Experiments using NNets from mouse embryonic hippocampal cells demonstrate the feasibility and the efficacy of this approach in modeling some of the brain complexities, i. e. during NNet development we can capture the activity synchronization from single neurons, to neurosphere level to the entire network. However, this is impaired in NNets from mouse models of schizophrenia, where the full network synchrony is never reached.

This approach provides several advantages: (1) no practical limit on the size and patterns of NNets, and therefore their complexity, (2) precise experimental access to individual neurons, connections, modules and even the entire network, (3) multi-scale intra-/inter-/supra-modular activity and connectivity patterns, (4) possible optical activation/inhibition of NNet activity with high spatio-temporal precision, while simultaneously observing the entire network activity state. Possible applications include: (1) modeling complex brain disorders to understand the pathophysiology of neural circuits at systems level, (2) automated and high-throughput drug screening and (3) personalized medicine for psychiatric disorders.

University of Southern California

Perception of surface angle is a key component of tactile shape recognition. How surface angle is represented in primary somatosensory cortex during active tactile exploration is unknown.

We trained mice on a novel head-fixed single whisker task: two-choice discrimination of forward (45˚) or backward (135˚) angled poles. Mice learned to differentiate pole angle in 1-3 weeks of training. We quantified the motor program and sensory input by videography of whisker motion and deformation in 3D. We simultaneously recorded neural activity daily by volumetric two-photon calcium imaging of excitatory cells of layers 2-4 of barrel cortex.

Before and after learning, we presented the pole at 15˚ steps on the interval from 45˚ to 135˚ while varying pole location to identify activity patterns associated with angle, position, and choice. We mapped touch-activated neurons that showed angle selective tuning across all three cortical layers. Tuning patterns were variable in shape, including sharp, broad, multi-peak, inhibited, and bimodal tuning. Neurons tuned to extreme angles (45, and 135) were overrepresented in the population, even in naïve mice. These representations reorganized across learning. We are currently working on identifying the sensorimotor features that underlie angle tuning and layer-specific differences in representation reorganization.

Our results support a circuit model of neuronal representations of tactile shape recognition, their role in shape perception, and reorganization of these representations during sensory-motor learning.

Institut de la Vision

Optical wavefront shaping, combined with optogenetics, is a powerful technique to manipulate neuronal activity with light at the single cell and single action potential level. Recently, we demonstrated multiplexed temporally focused light shaping (MTF-LS), a new optical scheme, based on the spatio-temporal shaping of a pulsed laser beam, to project several tens of spatially confined two photon excitation patterns in a large volume, suitable for the precise photo-stimulation of several neurons. Here, we extend MTF-LS to a micro-endoscope based on the use of a gradient index (GRIN) lens, which constitutes an important step for the precise optical study and manipulation of deep brain circuits. We first show the complete optical characterization of the system, which, thanks to the use of temporal focusing, achieves sufficient optical performances to allow experiments with single cell resolution. Thanks to the flexibility of MTF-LS, different light shaping methods are compatible with our micro-endoscope, thus enhancing the potential spectrum of applications. Finally, we demonstrate proof of principle optogenetic activation of single and multiple neurons in vivo in the mouse visual cortex, thus demonstrating the potential of the novel system.

University of Minnesota

The nucleus accumbens (NAc) controls multiple facets of impulsivity, but is a heterogeneous brain region with complex microcircuitry. Here, we studied the regulation of impulsive behavior by parvalbumin-positive fast-spiking interneurons (FSIs) using combinatory optogenetic and calcium imaging techniques. We found that FSIs, although comprising a small proportion of cells within the NAc, provide strong GABA-mediated inhibition onto local medium spiny neurons (MSNs). To study FSI involvement in impulsive action, we used the 5-choice serial reaction time (5-CSRT) task, for which a mouse must withhold an operant response during a variable inter- trial interval. In mice performing the 5-CSRT task, fiber photometry calcium imaging revealed that FSI activity was sustained on trials ending with correct responses, yet declined on trials ending with premature responses. Optogenetic silencing of NAc FSIs during the inter-trial period significantly increased the number of premature responses, but did not confound attentional control or locomotor activity. This effect was replicated with chemogenetic inhibition during the entire testing session. These experiments provide strong evidence that FSIs within the NAc constrain impulsive actions via GABA-mediated synaptic inhibition of MSN output. Our findings may provide insight into the pathophysiology of impulse control disorders, and inform the development of circuit-based therapeutic interventions.

The Vision Institute

The retina transforms the visual scene in spikes sent to the brain. Photoreceptors transduce light into electrical currents, bipolar cells process this signal and transmit it to ganglion cells, the retinal output. A key component of retinal processing is the information transfer from the intermediate bipolar cell layer to the ganglion cell layer. Our understanding of this transfer is limited: while multi-electrode arrays allow recording large populations of ganglion cells, bipolar cells cannot be easily recorded or stimulated in the intact retinal circuit. Here we present a novel method where we combined several techniques to record ganglion cells with multi-electrode arrays while perturbing individual bipolar cells using optical and optogenetic tools. We used an AAV and a specific promoter to express light sensitive proteins selectively in rod bipolar cells. We then used 2 photon digital holography, a technique to pattern light to stimulate individual neurons, while simultaneously recording ganglion cells with a multi-electrode array. Thanks to this combination of optical and electrophysiological tools, we could stimulate selectively rod bipolar cells and record the impact of this stimulation on the spiking activity of ganglion cells. Our method also allowed us stimulating several bipolar cells simultaneously to measure the impact of complex stimulation patterns on the ganglion cell layer. We are currently using this technique to understand the role of rod bipolar cells in shaping the surround of ganglion cells. This method allows a precise probing of the retinal circuit and paves the way towards complete functional connectomics of the retina.

University College London

Following Channelrhodopsin-2, opsins with different ion selectivities, spectral tuning, photocurrents and kinetics have been developed. Here, we generated nine zebrafish transgenic lines for targeted opsin expression via the Gal4/UAS system [ChR2(H134R), CheRiff, ChrimsonR, Chronos, CoChR, eArch3.0, eNpHR3.0, GtACR1 and GtACR2]. We then compared these new lines by assessing the ability of opsin activation to elicit/block neural activity in vivo using electrophysiological and behavioural readouts. In assays testing excitation, CoChR and ChrimsonR outperformed ChR2(H134R) in both photocurrent amplitude and ability to induce high-frequency (40-50 Hz) spiking. Additionally, CoChR and ChrimsonR were the most effective opsin lines in inducing tail bouts when stimulating sensory trigeminal or spinal motor neurons using blue or red light, respectively. In assays testing inhibition, GtACR1 and GtACR2 induced the largest inhibitory photocurrents and were the most effective lines in suppressing spontaneous behaviour. In conclusion, we provide an expanded optogenetic toolkit for effective manipulation of neural activity in zebrafish. Our comparative analyses will guide opsin line selection for interrogation of neural circuit function and precise control of behaviour.