Two-photon calcium imaging of visual cortex

When a neuron fires an action potential, voltage-gated calcium channels open and calcium flows into the cell for a fraction of a second. Genetically encoded calcium indicators such as GCaMP, a green fluorescent protein joined to the calcium-binding protein calmodulin, become brighter when they bind this calcium. Imaging the indicator therefore shows which neurons have just fired. The signal reports calcium, not voltage: it peaks about a tenth of a second or more after the spikes and fades over hundreds of milliseconds. Indicators of the GCaMP6 family detect single action potentials in the visual cortex of mice (Chen et al., 2013).

Brain tissue scatters visible light strongly, so ordinary fluorescence microscopes see only a blur a short way below the surface. A two-photon microscope focuses a pulsed infrared laser to a tiny spot and scans it across the tissue. A fluorescent molecule is excited only when two infrared photons arrive at almost the same moment, which happens only at the focus, so each image comes from one thin plane, even several hundred micrometres deep in the living brain (Lichtman and Conchello, 2005). GCaMP is excited in this way by infrared light of about 920 to 940 nm, roughly twice the 488 nm used for one-photon excitation.

Neurons in the primary visual cortex respond best to edges or stripes of one orientation. Showing a mouse drifting gratings at eight orientations while imaging a few dozen neurons gives a tuning curve for each neuron, and the orientation selectivity index (OSI) summarises how sharply tuned it is; Chen et al. (2013) measured tuning in this way in layer 2/3 of mouse V1.

The session includes every decision of a real imaging experiment: the indicator, the time of imaging, the laser wavelength and power, the field of view, the stimulus, motion correction, ROIs, neuropil correction and the ΔF/F baseline. The simulation knows the true spikes and tuning of every neuron, so the effect of each choice on the result can be shown.

Learning objectives

• Explain how GCaMP reports action potentials through calcium, and why the fluorescence lags and smooths the spikes
• Describe the light path of a two-photon microscope and explain why two-photon excitation gives optical sectioning in scattering tissue
• Choose a field of view, laser power and frame rate, and recognise vessel shadows, motion artefacts and photodamage
• Correct motion, draw and check ROIs, correct neuropil contamination and compute ΔF/F with a suitable baseline
• Compute tuning curves and the orientation selectivity index, with a responsiveness criterion set in advance
• Explain why neurons from one mouse are not independent replicates and analyse with the animal as the unit

Ethics and safety

Calcium imaging in living mice involves regulated procedures: viral injection, implanting a cranial window and head-plate under general anaesthesia with pain relief, and repeated head fixation. They need approval by an Institutional Animal Ethics Committee; in India such work is governed by the CCSEA under the Prevention of Cruelty to Animals Act 1960, with equivalent frameworks in the European Union (Directive 2010/63/EU), the United States (Animal Welfare Act and PHS Policy) and the United Kingdom (Animals (Scientific Procedures) Act 1986). Surgery is done by trained, authorised staff following the approved protocol. The 3Rs apply: habituation to head fixation and good post-operative care are Refinements, imaging many neurons in each mouse over several sessions supports Reduction, and simulations such as this one can replace some practice. Studies are reported following the ARRIVE guidelines, developed by the NC3Rs in the United Kingdom and endorsed by journals internationally (Percie du Sert et al., 2020), giving the numbers of mice, fields of view and neurons. Lasers used for two-photon imaging are Class 4: invisible infrared beams can damage the eyes and skin, so only trained users operate the microscope, enclosures and interlocks are kept in place, and laser safety rules are followed.

Key terms

• Calcium indicator: A molecule whose fluorescence changes when it binds calcium. GCaMP is a genetically encoded indicator: the neurons make it themselves from an introduced gene.
• GCaMP: A green fluorescent protein joined to calmodulin and a peptide that calmodulin binds; calcium binding makes it brighter. GCaMP6f is a faster version, GCaMP6s a slower, more sensitive one.
• ΔF/F: The change in fluorescence divided by the baseline fluorescence F0: (F − F0) / F0. A value of 0.2 means 20% brighter than at rest.
• Two-photon excitation: Excitation of a fluorescent molecule by two infrared photons absorbed almost together. It depends on the square of the light intensity, so it happens only at the focus. GCaMP is excited this way at about 920 to 940 nm.
• Optical sectioning: Imaging one thin plane of a thick sample without light from the planes above and below.
• Scattering: The deflection of light by structures in tissue. Scattered light blurs images and loses its focus; infrared light is scattered less than visible light.
• Cranial window: A glass coverslip that replaces a small piece of skull so that the brain can be imaged repeatedly.
• Habituation: Accustoming an animal to a procedure, here head fixation, over several sessions, so that it is calm and moves little.
• Field of view: The area of tissue imaged in each frame, here about 200 µm across in one focal plane.
• Motion correction: Aligning every frame of a movie to a reference image to remove sideways movement of the brain. Movement in depth cannot be corrected this way.
• ROI: Region of interest: the outline of one cell body in the image, whose brightness is measured in every frame.
• Neuropil: The dense mesh of dendrites and axons between cell bodies. Its fluorescence contaminates every ROI and is partly subtracted using a ring around the ROI.
• Drifting grating: A pattern of stripes moving across a screen, used to find the orientation a visual neuron prefers.
• Orientation selectivity index: OSI = (Rpref − Rorth) / (Rpref + Rorth), from 0 for no preference to 1 for no response at the orthogonal orientation. Noise biases it upwards, more so with few trials.