Perception and behavior emerge from the coordinated and orchestrated activity of neurons in brain circuits. Individual, functionally coherent neurons form ensembles, which become the building blocks of large-scale circuitry to drive the brain machinery (1). The ability to modulate brain activity in a spatiotemporal pattern with high specificity (millisecond time scale and cellular resolution) has great implications for interfacing with this sophisticated machinery. In fundamental science, it provides a powerful tool to dissect neuronal circuits in very fine detail and study causality among neural activity, circuit dynamics, and behavior (2–6). In translational medicine, it plays an important role in treating brain disorders (7, 8) and holds great promise to become a new tool for precision medicine.
Electrical stimulation is the most mature approach to modulating brain activity. However, penetrating electrodes are highly invasive, and there is a lack of spatial specificity in the targeted brain regions. Therefore, neurons in a large brain volume are indiscriminately stimulated simultaneously regardless of their individual function in the brain circuit and resultant link to behavior. Such low spatial specificity and the associated unspecific off-target effects not only limit the application of these approaches in studying brain circuits, but also pose concerns with regard to overall efficacy and side effects in clinical therapy (9, 10).
Optical methods, particularly when coupled with optogenetics (11, 12), offer a new approach to modulating brain activity with cell-type specificity. Although high spatial specificity can be achieved in two-dimensional (2D) samples, such as thin brain slices, early use of optogenetics in living brains faced the same challenges as electrical stimulation: The dispersed light failed to distinguish individual cells in a 3D volume and instead stimulated all neurons together. Two-photon light resolves the problem of spatial specificity and achieves cellular resolution. Borrowing this technique from laser scanning microscopy, two-photon optogenetics sequentially stimulates neurons one by one (13, 14). Although specificity is high, sequential single-cell stimulation fails to mimic intrinsic activity patterns in the brain, where multiple neurons can fire action potentials simultaneously.
Metaphorically speaking, manipulating neurons in a circuit is akin to pressing the keys of a piano keyboard. Photostimulating neurons one at a time is like playing the piano with a single finger, which would fail to produce a rhythmic and melodious concert piece. To modulate neural activity in a coordinated manner, it is necessary to simultaneously stimulate an ensemble of neurons, distributed in a 3D brain volume, with cellular resolution—as if playing the piano with all 10 fingers. We are among the first to tackle this challenge in vivo (15) and demonstrate the power of optogenetics in studying the link between neural activity and behavior (2) (see the figure).
Two-Photon Holographic Optogenetics
Leveraging the computer-generated hologram, we encoded the 3D spatial information of the targeted neurons into the phase hologram using a spatial light modulator, to develop two-photon 3D holographic techniques for precision optogenetics (15). By projecting a holographic light pattern, which contains beamlets focused on the target neurons in a mouse brain, we can precisely modulate the activity of neuronal ensembles. Two-photon excitation ensures that the light can penetrate deep into the scattering tissue and stimulate the target neurons distributed in a 3D volume with excellent specificity. To maximize the number of neurons that can be stimulated at once without imposing high doses of light (and thus heat) on the brain, we increased the two-photon excitation efficiency by adapting a low–repetition rate femtosecond laser. This allowed us to simultaneously stimulate a large group of neurons (>50) with a minimum amount of light power (a few milliwatts per neuron). By rapidly switching the holograms (millisecond time scale), we can stimulate different groups of neurons with high temporal specificity. Our two-photon holographic optogenetics approach thus enables modulating the neuronal activity in a desired spatiotemporal pattern.
All-Optical Interrogation of the Neuronal Activity
To expertly manipulate brain circuitry, we needed a 3D neuronal map. We built a dual-path microscope with two different lasers, integrating two-photon high-speed volumetric calcium imaging with two-photon holographic optogenetics (15). The imaging path was equipped with an electrically tunable lens for fast 3D imaging (15, 16), and the optogenetics path was equipped with a spatial light modulator to generate the 3D photostimulation pattern. To avoid cross-talk between imaging and optogenetics, we selected indicators and opsins with distinct light excitation spectra: calcium indicator GCaMP6 (17) for imaging and opsin C1V1 (18) for optogenetics. Using this dual-path microscope, we were among the first to demonstrate simultaneous 3D imaging and holographic photostimulation of cortical activity in awake mice (see the figure). Such an all-optical setup allowed us to precisely stimulate an arbitrary group of neurons while monitoring the response of the circuit, and thus enabled closed-loop control of brain activity, all with high temporal specificity and cellular resolution across a large 3D brain volume.
(A) Schematics of simultaneous two-photon volumetric calcium imaging and two-photon 3D holographic patterned photostimulation in a mouse brain. A user-defined group of neurons can be stimulated simultaneously with high spatiotemporal specificity. (B) Closed-loop control of neuronal activity and behavior. The neuronal circuit imaged during animal behavior provides a map to modulate the brain through two-photon holographic optogenetics. We demonstrated mouse performance in a Go/No-Go visual discrimination task can be enhanced by photoactivating only two core ensemble neurons in visual cortex (2).
GRAPHIC: H. BISHOP/SCIENCE BASED ON W. YANG
” data-icon-position=”” data-hide-link-title=”0″>
(A) Schematics of simultaneous two-photon volumetric calcium imaging and two-photon 3D holographic patterned photostimulation in a mouse brain. A user-defined group of neurons can be stimulated simultaneously with high spatiotemporal specificity. (B) Closed-loop control of neuronal activity and behavior. The neuronal circuit imaged during animal behavior provides a map to modulate the brain through two-photon holographic optogenetics. We demonstrated mouse performance in a Go/No-Go visual discrimination task can be enhanced by photoactivating only two core ensemble neurons in visual cortex (2).
GRAPHIC: H. BISHOP/SCIENCE BASED ON W. YANG
Modulating Behavior By Triggering Neuronal Ensemble Activity
Understanding the role of neuronal ensembles could lead to new insight into how behaviors emerge as well as innovative therapies for brain diseases (8). Using our alloptical method, we studied the causal link between ensemble activity and behavior and demonstrated an efficient approach to modulate behavior (2) (see the figure).
We designed a Go/No-Go visual discrimination task, in which two orientations of drifting gratings were randomly displayed and the mouse discriminated between them by licking a waterspout. We hypothesized that modulation of ensemble activity could affect behavior. We holographically photoactivated a random group of unspecific neurons in the mouse visual cortex during the task. Unsurprisingly, the resultant “noise” in the visual cortex decreased task performance. We then asked if directed neuronal modulation could improve the task outcome. Using a machine learning algorithm, we extracted the neuronal ensembles and the core ensemble neurons related to the “Go-cue” of the visual stimuli in the visual cortex. Surprisingly, holographic photoactivation of only two core ensemble neurons during the Go-cue could enhance task performance (2).
Through imaging, we observed that the activation of core ensemble neurons drove widespread recruitment of other neurons within the ensemble. Such a pattern completion mechanism, potentially involving recurrent neural networks, eventually amplified the activation effect of core ensemble neurons and ultimately modulated the behavioral outcome. This effect was so pronounced that the holographic activation could elicit mouse licking associated with the Go-cue even when the Go-cue was not physically presented (2).
Compared to previous approaches whereby large brain regions are stimulated at once, either electrically or through single-photon optogenetics, our holographic approach provides much greater specificity and efficiency. Not only does our study prove the functional and behavioral relevance of neuronal ensembles and provide a direct illustration of pattern completion, but the ability to precisely write information into the brain to trigger behavior opens a new avenue in precision medicine to correct the pathophysiology of mental disorders (8).
An Outlook of Precision Optogenetics
The invention of optogenetics has given neuroscientists a new tool to modulate cell-type–specific neuronal activity. Our in vivo two-photon holography technique has brought optogenetics into a new, precision era. Today and in the near future, 4D spatiotemporal modulation patterns, which parallel the intrinsic physiology of the neural system, could be applied to elicit recurrent activity and recruit downstream activity and behavior.
We have demonstrated the triggering of visually guided behavior through two-photon holographic optogenetics in the mouse visual cortex (2), and others have applied this technique to the mouse hippocampus (4), to drive spatial behavior. In other animal models such as the larval zebrafish (6), the method has been used to elicit motor behavior. In each case, activation of only a small number of neurons was able to modulate animal behavior.
In addition to studying circuit causality, two-photon holographic optogenetics is an ideal tool to induce network plasticity through Hebbian plasticity (19). By repeatedly photostimulating a group of neurons, we demonstrated that functional connectivity increased in a subset of these neurons (20). When pairing the holographic photoactivation of a behaviorally unspecific ensemble with a behavioral reward, it was shown that the animal could learn to associate the ensemble activation with the award (3, 21, 22). Such findings suggest that two-photon holographic optogenetics could be used to reprogram the brain and create an artificial link between neuronal activity and cognitive states. This result has tremendous translational importance and could potentially be used to reestablish brain functions of a damaged region in a new region.
The past 3 years have witnessed a new wave of findings enabled by two-photon holographic optogenetics in awake mice. Much could be done to further exploit its potential, particularly in translational medicine. In our noninvasive demonstrations, target neurons were confined to the cortical layers. The ability to target deep brain regions with a noninvasive or minimally invasive approach will greatly broaden its application. The closed-loop, real-time control of imaging, optogenetics, and monitoring of behavior could potentially create a new type of brain machine interface. As the first type of precise brain modulation modality, we envision that two-photon holographic optogenetics (15, 23–27) will continue to play a pivotal role in both fundamental neuroscience and translational medicine.
PHOTO: COURTESY OF WEIJIAN YANG
FINALIST
Weijian Yang
Weijian Yang received his undergraduate degree from Peking University and a PhD from the University of California, Berkeley. After completing his postdoctoral fellowship at Columbia University, he started his laboratory in the Department of Electrical and Computer Engineering at the University of California, Davis in 2017. His research aims to develop advanced optical methods and neurotechnologies to interrogate and modulate brain activity, with a goal to understand how neural circuits organize and function and how behaviors emerge from neuronal activity. www.sciencemag.org/content/373/6555/635