Abstract
Unraveling how neural networks process and represent sensory information and how these cellular signals instruct behavioral output is a main goal in neuroscience. Two-photon activation of optogenetic actuators and calcium (Ca(2+)) imaging with genetically encoded indicators allow, respectively, the all-optical stimulation and readout of activity from genetically identified cell populations. However, these techniques locally expose the brain to high near-infrared light doses, raising the concern of light-induced adverse effects on the biology under study. Combining 2P imaging of Ca(2+) transients in GCaMP6f-expressing cortical astrocytes and unbiased machine-based event detection, we demonstrate the subtle build-up of aberrant microdomain Ca(2+) transients in the fine astroglial processes that depended on the average rather than peak laser power. Illumination conditions routinely being used in biological 2P microscopy (920-nm excitation, ∼100-fs, and ∼10 mW average power) increased the frequency of microdomain Ca(2+) events but left their amplitude, area, and duration largely unchanged. Ca(2+) transients in the otherwise silent soma were secondary to this peripheral hyperactivity that occurred without overt morphological damage. Continuous-wave (nonpulsed) 920-nm illumination at the same average power was as damaging as femtosecond pulses, unraveling the dominance of a heating-mediated damage mechanism. In an astrocyte-specific inositol 3-phosphate receptor type-2 knockout mouse, near-infrared light-induced Ca(2+) microdomains persisted in the small processes, underpinning their resemblance to physiological inositol 3-phosphate receptor type-2-independent Ca(2+) signals, whereas somatic hyperactivity was abolished. We conclude that, contrary to what has generally been believed in the field, shorter pulses and lower average power can help to alleviate damage and allow for longer recording windows at 920 nm.