Methodological considerations

The development of 2-photon imaging and its application to the living brain opened the possibility to simultaneously study the activity of many identified cells with sub second temporal precision in a relatively non-invasive manner. Calcium changes within a neuron are a suitable approximation for the underlying activity pattern of the cell since each action potential (AP) is associated with a somatic calcium influx. However, one should keep in mind that monitoring calcium fluctuations within a neuron is only an indirect readout of its spiking activity (Lütcke et al. 2013).

To perform in vivo 2-photon calcium imaging in the brain, specific principles and steps have to be considered. 1) The cells of interest need to be labelled using the principle of fluorescence. For the labelling of cells with calcium reporter molecules two prominent approaches have been established: Extracellular application of a synthetic calcium indicator on the brain area of interest (e.g. OGB-1) or the expression of GECIs in specific cells of interest (e.g. GCaMP). The fluorescence intensity change of GECIs is nearly proportional to the cellular calcium concentration (Rose et al., 2014) rendering them a reliable tool to image neuronal calcium dynamics. The great advantage of GECIs over synthetic calcium indicators is the possibility to repeatedly perform calcium imaging of the same neurons over months. 2) In single photon fluorescence, a fluorophore absorbs a single photon of appropriate energy and enters an excited state.

When the fluorescent molecule transits back to the ground state, it emits photons of lower energy and therefore has a longer red-shifted wavelength compared to the wavelength used for excitation. However, the excitation of a fluorophore is also possible with two photons (or more), if their combined energy provides the adequate energy for excitation. 2-photon microscopy makes use of this process: Two low-energy photons (in the red region of the spectrum, ~700 nm) together cause a higher-low-energy transition from a ground state to an excited state in a fluorescent molecule (Denk et al. 1990, Svoboda et al. 2006).

The excitation source for 2-photon microscopy is a focused near infrared laser beam that illuminates a small volume in the brain area of interest at a time. The chance that two photons simultaneously strike and then excite a fluorophore is highest in the focal volume of the laser beam and drops exponentially with decreasing intensity outside the focal volume. Therefore, all emitted photons originate from the focal volume of the infrared laser beam (principle of optical sectioning). The out of focus excitation of fluorophores is drastically reduced leading to much better spatial resolution compared to single photon microscopy. Furthermore, the near infrared laser excitation wavelengths used for 2-photon microscopy

scatter much less in tissue than visible light. This leads to much better penetration of light even in relatively deep brain areas.

1.4.2 Circuit mapping in the brain

Circuit connectivity in acute brain slices can be readily studied using optical approaches. A focused light beam is optimal for the activation of neurons since it can be controlled both spatially and temporally with great precision. Furthermore, the wavelength, shape of light waves and strength of light can be tightly controlled.

In most circuit mapping approaches, whole-cell patch-clamp recordings of either a single postsynaptic cell or multiple postsynaptic cells are performed simultaneously. Since cells cannot be activated by light per se, two main approaches are prominently used for circuit mapping: Photolysis of caged compounds or the expression of channelrhodopsins in presynaptic cells. For LSPS, a focused light beam is rapidly moved across different sites in the tissue activating presynaptic cells. An alternative strategy is to use wide-field illumination leading to the detection of the net synaptic input of the recorded postsynaptic cells.

The most prominently used caged compound for photolysis is the excitatory transmitter glutamate bound to a caging moiety via a photoscissile bond (Katz et al. 1994). Upon stimulation by a focused UV beam, glutamate is locally released and activates endogenous receptors of nearby cells.

Sufficient glutamate release will lead to AP generation in potential presynaptic cells and will only be detected as postsynaptic input if the stimulated pre- and recorded postsynaptic cell(s) are connected.

Importantly, caged compounds can only be used to study local connectivity since the somata of the presynaptic cells as well as their axons need to be present within the same slice as the recorded postsynaptic cell. For single-photon LSPS using caged glutamate, recordings are performed on a single cell or multiple postsynaptic cells while the intralaminar as well as translaminar presynaptic partners are stimulated. This approach enables single cell resolution on the postsynaptic but not on the presynaptic side. Ionotropic glutamate receptors are mostly found in the soma membrane as well as along the dendrite of a neuron and therefore glutamate uncaging does not activate en passant axons, making this technique a suitable approach to map the local translaminar circuitry. The resolution of this technique depends mostly on the point spread function of the UV light source as well as the scattering of light by neuronal tissue. Importantly, the resolution should be calibrated by adjusting the glutamate concentration and the laser intensity to restrict action potential firing of all different cell types across

layers to their peri-somatic region. Once correctly calibrated, this technique provides sublaminar resolution for the presynaptic input (~50 µm, Shepherd et al. (2003), Anastasiades et al. (2012), Xu et al.

(2016)). LSPS by glutamate uncaging has been used in many studies in different brain areas to understand the translaminar connectivity of specific cell types within a cortical region and the development as well as plasticity-related alterations of these (Shepherd et al. 2003, Bureau et al. 2004, Shepherd et al. 2005, Bureau et al. 2006, Brill et al. 2009, Hooks et al. 2011, Apicella et al. 2012, Kuhlman et al. 2013, Kratz et al. 2015, Xu et al. 2016, Deng et al. 2017, Meng et al. 2017).

The second prominent approach for studying cortical connectivity is the photostimulation of channelrhodopsin-expressing presynaptic cells of interest. In contrast to caged compounds, optogenetic approaches can be applied for both local and long-range circuit mapping since light-sensitive opsins are expressed throughout dendrites and axons. Therefore, the presynaptic soma does not need to be preserved in the same slice as the recorded postsynaptic cell(s). Severed axons can still be activated in acute slices because synaptic terminals remain intact and presynaptic release can be triggered using brief light pulses. By combining optogenetic stimulation with LSPS (so called ChR2-assisted circuit mapping, CRACM, Petreanu et al. (2007)) it is possible to map the input of ChR2-expressing neurons across brain areas onto postsynaptic cells in different layers. Further refinement of this approach, enables mapping the subcellular location of ChR2-postive axon terminals onto target cells by blocking fast transient sodium channels using TTX and potassium channels by 4-AP (subcellular CRACM, sCRACM, Petreanu et al. (2009)).

By blocking sodium channel-mediated action potential conductance along the axons and at the same time blocking the repolarization of the axon mediated by potassium channels, it is possible to map monosynaptic inputs across the postsynaptic dendrite.

A further advancement of circuit mapping combined with optogenetic is the possibility to use dual-channel photostimulation in order to map multiple types of presynaptic input onto the same postsynaptic cell (Hooks et al. 2015). For this, two channelrhodopsin variants that are excited by different wavelengths are expressed in two neuronal populations and then the convergence of these neuronal populations is mapped onto single presynaptic cells (Klapoetke et al. 2014).

1.4.3 In vivo / in vitro approaches

High-resolution analysis of circuit connectivity and cellular and synaptic properties can only partially be performed in vivo. Therefore, in vitro methods are necessary to characterize neuronal circuits at high resolution. In order to directly correlate in vivo measured response properties with the underlying cellular

and synaptic properties as well as the neuronal connectivity the challenge is to re-identify the very same neurons between in vivo as well as in vitro.

Two different experimental approaches for re-identifying neurons between the in vivo and in vitro preparation have been described so far. In the first approach, neurons acutely labelled with a synthetic calcium indicator are matched using precise alignment and transformation of image stacks recorded in vivo and in corresponding in vitro brain slices (Ko et al. 2011). The synthetic calcium indicator OGB-1 labels basically all cells and therefore the goal is to record as many cells as possible in the brain slice. The actual matching of cells between the in vivo and in vitro condition is done post-hoc. In the second approach, specific neurons of interest are labelled in vivo by optically activating photo-activatable-GFP (pa-GFP) allowing these cells to be targeted for further analysis in vitro (Lien et al. 2011, Peter et al. 2013). The neuronal activity is here recorded with a synthetic calcium indicator expressed in all cells.

Taken together, these in vivo / in vitro approaches have not yet been applied to GECIs which would enable performing long-term 2-photon calcium in vivo experiments and then re-identifying neurons in brain slices to characterize neuronal circuits.