Genetic Labeling of Synapses

Abstract

A major challenge in neuroscience is to unravel how the synaptic contacts between neurons give rise to brain circuits. A number of techniques have been developed to visualize the synaptic organization of neurons. In this chapter, we focus on genetic methods to mark specific types of synapses so that synaptic sites can be visualized throughout the entire dendritic or axonal arbor of single neurons. Genetic synaptic labeling can be achieved by cell-type-specific viral or transgenic delivery of synaptic proteins tagged by fluorescent proteins. Sparse genetic labeling of neurons permits semiautomated quantification of the distribution and densities of selected types of synapses in segregated domains of the axonal and dendritic trees. These approaches can reduce the complexity and ambiguity of attributing synaptic sites to unravel principles of the synaptic organization of identified neuronal types in the circuit.

Appendices

Appendix: Gene Delivery of Genetic Synaptic Markers with Retroviruses and Quantification of Synapses

This appendix describes some specific procedures to deliver retroviruses carrying genetically encoded synaptic markers into the brain of rodents. General details about production and titration of lentiviral and retroviral vectors can be found elsewhere.

Injection of Viruses into the Brain

Viral prep: It is critical that the viral suspension is very clean. During the preparation of the viruses, there will be some cellular debris that can be strongly autofluorescent. To eliminate this debris it is useful to centrifuge the viral prep with a 20 % sucrose cushion.

Stereotaxic injection: It is critical to minimize the damage to the brain during injection. In particular, bleeding associated with the injection will cause very high levels of autofluorescence that will make quantification of synapses very difficult. To minimize damage it is useful to use thin borosilicate pipettes pulled to an outer diameter of approx. 15–20 μm. It is not advisable to inject through metallic needles as this will cause severe tissue damage on the injection site. Similarly, it is advisable to inject the viral prep slowly, at a rate of approx. 1 μl over 5 min. Rapid injection can severely damage and distort the tissue. Regarding the timing of imaging after injection, for lentiviral vectors, the expression of the transgene peaks as early as 3 days, but there will likely be some acute damage in the injection area at this early time. Thus, it is advisable to wait at least a week so that the autofluorescence due to damage is resolved before perfusion of the animal.

Acquisition and Analysis of Genetically Labeled Synapses

This section describes the procedure that we optimized to visualize the synaptic organization of single genetically labeled neurons, which can be easily modified for individual experimental needs. The procedure is divided into three main steps, and technical issues are highlighted that are critical in our experience: preparation of the tissue (“Preparation of Tissue”), image acquisition by confocal microscopy (“Image Acquisition”), and semiautomated image analysis (“Quantification of Synaptic Clusters”).

3.1 Preparation of Tissue

• 1. The protocol is described for small rodents, but can be easily adapted to other species. Animals are transcardially perfused initially with phosphate buffered saline (PBS, 1×) for 10–15 s, followed by 4 % paraformaldehyde (PFA) for 3–5 min. The animal should become rigid within the first 30 s to 1 min of perfusion with PFA. It is optimal to use an overdose of an injectable anesthetic drug (such as Ketamin/Xylazin) and to start perfusion when the heart is still beating. PBS should be infused at a pressure such that the liver becomes pale within 10–15 s and clear PBS flows out of the right atrium after this period. It is equally important to perfuse with relatively low pressure, because at high perfusion pressure, the small capillaries in the brain will break and perfusion will not be homogeneous throughout the brain. PBS should be set to pH 7.0–7.4 and pre-warmed to 32–37 °C to prevent contraction of smaller blood vessels in the brain. Following PBS, perfusion should be immediately switched to room-temperature PFA. Incorrect perfusion leads to delayed fixation with PFA, which results in beaded structures of dendrites and loss of genetically labeled synaptic clusters. After perfusion is complete, the brains are extracted from the skull and post-fixed in 4 % PFA overnight at 4 °C.

• 2. After preparation of floating sections with a vibratome (e.g., 50 μm sections), tissue can be incubated (overnight at 4 °C) with primary antibody raised against the fluorescent protein tagged to the synaptic marker. The following day sections are rinsed in PBS and stained with a secondary fluorescent antibody for two hours at room temperature. Sections are then washed with PBS and mounted with an aqueous mounting medium that preserved fluorescent molecules. This procedure allows for the visualization of the neuronal tree and to attribute synaptic clusters to a neuron and specific dendritic domains. Blocking solutions (PBS containing 1 % bovine serum albumin or related serum proteins) for antibody incubation usually contain a detergent, i.e., Triton X-100, to permeabilize the tissue. We keep the procedure and times as constant as possible to avoid introducing additional variability, i.e., by differentially affecting the brightness of the fluorescence of the synaptic clusters.

3.2 Image Acquisition

Neurons expressing synaptic marker proteins can be conveniently imagined using confocal laser scanning microscopy. In most experimental conditions, it is advantageous to image sections that are sparsely labeled, where individual neurons are clearly separated from each other. In this case it is easy to analyze the full dendritic arbor of a single neuron without having to deal with neurites that could belong to neighboring cells. Confocal stacks are acquired at high magnification (with a 60–63× oil immersion objective) to efficiently capture emitted light from the clustered fluorescent proteins. The pixel size should be sufficiently small to obtain high intensity of all the pixels that are grouped in individual synaptic clusters, and to easily distinguish them from the occasionally observed noise that may result in random isolated pixels with high intensity. As most synaptic clusters have a size around 1 μm, we found a cluster size between 0.2 × 0.2 and 0.3 × 0.3 μm² most reliable for subsequent analysis. Laser excitation intensities should be set to levels that result in little or no obvious bleaching of the clusters. This can be easily tested by imaging the same neuron twice in the same day and comparing the clusters among the two images. A reference section containing neurons with good synaptic cluster intensity should be used to guarantee comparable acquisition conditions over time. The sensitivity of the photomultipliers (PMT) should be set to a level that clusters just saturate but low enough that individual clusters do not become confluent due to overexposure. Similarly, the pinhole size should be kept in the recommended range [51] for the chosen magnification. Once the settings are initially defined with a test sample, the conditions should be kept constant throughout the different imaging session. Upon acquisition of confocal stacks, maximum density projections are prepared for further image analysis. Two-dimensional projections are generally used for analysis, as current version of most image processing software cannot handle 3D data for quantification.

3.3 Quantification of Synaptic Clusters

Analysis of densities and distribution of genetically labeled synapses can be semiautomated. Fully automated analysis is currently limited by the still challenging task for computers to reconstruct neuronal trees due to overlap of labeled neurons and discontinuities in the processes deriving from histological processing and incomplete filling with fluorescent proteins. Thus, reconstruction of processes has to be performed individually or at least be supervised.

Contrary to reconstruction of dendritic trees, analysis of clusters can be fully automatized provided the original image quality has a good signal-to-noise ratio. Signal-to-noise ratio for these experiments means high intensity of fluorescence in the synaptic cluster and low levels of autofluorescent background outside of synaptic sites. Similarly, it is important that there is a low level of fluorescence originating from diffusely distributed XFP in the cells’ processes outside the synapses. Another potential source of “contamination” with artifactual autofluorescent clusters can be due to lipofuscin granules observed in some brain structures and species. The appearance of these autofluorescent granules is difficult to predict. For example, we have found them in mouse olfactory bulb and dentate gyrus, but not in the rat olfactory bulb or mouse neocortex. These autofluorescent artifacts can be easily diagnosed as they are excited by all wavelengths. In contrast, real genetic synaptic markers containing XFPs can only be detected at a specific wavelength (e.g., 550 nm for GFP). In addition, these autofluorescent granules can usually be excluded from analysis as they are mostly present in cell bodies.

Given these considerations the analysis is relatively straightforward using different analysis software packages. We will describe the different steps of analysis and particularly refer to the ImageJ-based MacBiophotonics software (www.macbiophotonics.ca/). Similar procedures can be performed in Metamorph software from Molecular Probes.

Steps:

• 1. Open maximum density projection (File>Open).

• 2. Define pixel size for subsequent distance measurements (Analyze>Set scale).

• 3. Split color channels (Image>Color>Split channels).

• 4. Choose the channel that displays the fluorescent synaptic clusters.

• 5. Set inclusive threshold so that only clusters are included (Image>Adjust>Threshold). The threshold value should be kept constant throughout the analysis. Therefore, it proves useful to use a reference as described in the acquisition part to set the threshold.

• 6. Draw a contour using the freehand selection tool to define a region of interest to measure a specific dendritic domain and exclude neighboring neurons.

• 7. Perform region measurement (Analyze>Analyze particles). Desired parameter data can be set in the results window (Analyze>Set Measurements) and copied to a data sheet of a given statistics program.

Measure the length of the neuronal processes in the region of interest using the freehand line tool and Analyze>Measure.