Protocol for assessing neuron-astrocyte spatial interactions using the neuron-astrocyte proximity assay

Abstract

Astrocytes are bushy, morphologically complex cells with numerous close contacts with neurons at the level of their soma, branches and branchlets. The smallest astrocyte processes make discrete contacts with synapses at scales that cannot be observed by standard light microscopy. At such contact points, astrocytes are thought to perform both homeostatic and neuromodulatory roles – functions that are proposed to be determined by their close spatial apposition. To study such spatial interactions, we developed a Förster resonance energy transfer (FRET) based approach, which enables observation and tracking of the static and dynamic proximity of astrocyte processes with synapses. The approach is compatible with standard imaging techniques such as confocal microscopy and permits assessment of the most proximate and perhaps likely physiologically relevant contacts between astrocytes and neurons in live tissues. In this protocol, after in vitro testing in HEK293T cells, we used the approach to analyze the contacts between striatal astrocyte processes and corticostriatal neuronal projection terminals onto medium spiny neurons (MSNs). We describe the required procedures in detail, including AAV microinjections, acute brain slice preparation, imaging and post hoc FRET quantification. The protocol provides a detailed description that can be used to characterize and study astrocyte process proximity to synapses in living tissue.

INTRODUCTION

Astrocytes are abundant glial cells that tile the central nervous system (Khakh & Sofroniew 2015). A common feature of astrocyte populations throughout the CNS is their extensive and ramified processes. However, circuit-specific interrogation of astrocyte-synapse dynamics in the central nervous system has been problematic in live tissue, because the relevant distance scales are below the resolution of light microscopy. Alternative strategies to image nanometer scale interactions in biological systems such as electron microscopy require time-consuming fixation, embedding, imaging and reconstruction. Furthermore, this technique is limited to single time point assessments and does not enable tracking of process movement in real time. Based on ideas previously reported (Feinberg et al 2008, Kim et al 2011), the Neuron-Astrocyte Proximity Assay (NAPA) is a FRET-based light microscopy approach that reports astrocyte engagements within intact neural circuits (Octeau et al., 2018). Additionally, NAPA reports dynamic changes in astrocyte contacts with synapses in mouse models of neurodegenerative disease. Herein, we describe the tools, techniques and protocols to image and analyze NAPA FRET and colocalization signals that report proximate and distal astrocyte-synapse interactions.

This protocol provides detailed stepwise instructions on how to perform FRET imaging in cell lines and brain slices. Specifically, we provide instructions on HEK293T culture transfection and imaging for FRET. We used the GFP-mCherry fusion protein as a control for FRET imaging and calculations of donor bleed through and acceptor crosstalk coefficients using an ImageJ-based plugin called PixFRET (Feige et al 2005). Subsequently, we used these experimentally determined coefficients to calculate FRET efficiency using PixFRET. We also describe the experimental setup for detection of FRET in acute brain slices, detailing the process of AAV microinjections for expression of membrane-tethered fluorescent reporter probes. We describe the preparation of live mouse brain slices and how to acquire and process images for FRET efficiency calculations. We suggest that the use of this protocol will make the implementation of NAPA easier and allow researchers to use such FRET-based approaches in multiple applications (Fig. 1).

Figure 1.

Outline of the experimental paradigm for NAPA experiments in situ. On the far left are listed the time considerations for the corresponding steps. In the middle we report the stepwise protocol from project inception culminating in data interpretation. On the right are listed the related figures, and relevant sections of the paper which correspond to steps in the protocol. We estimate the full protocol will take approximately 6 weeks to complete.

NOTE: Protocols such as this one which use live animals must be reviewed and approved by an Institutional Animal Care and Use Committee (IACUC) and/or must conform to local and federal regulations regarding the care and use of laboratory animals.

BASIC PROTOCOL 1

FÖRSTER RESONANCE ENERGY TRANSFER (FRET) IMAGING IN CULTURED CELLS

Basic protocol 1 uses a mCherry-GFP tandem vector expressed in HEK293T cells to measure sensitized emission (SE-FRET), in which the mCherry and the GFP are fused with a linker (24 bp). Cells are transfected with the controls (mCherry and GFP alone) or with the mCherry-GFP tandem and imaged after 2–3 days. Representative imaging parameters and analysis parameters are listed.

Materials

Culturing HEK293T cells

• 1Grow human embryonic kidney cells (HEK293T, ATCC) in a T75 flask with DMEM/F12 + Glutamax (ThermoFisherScientific, cat. no. 10565018) supplemented with fetal bovine serum, penicillin and streptomycin sulfate (10% and 100 mg/ml) in humidified atmosphere of 95% air/5% CO₂ at 37°C in a cell culture incubator.
• 2When confluence reaches 60–90%, split the cells (1 ml of cells in 9 ml of cultured media) to a six well plate. Add 2 ml of cell suspension to each well.

  Mix the cell suspension gently before plating to avoid cells forming focal aggregates in the well.

Transfecting HEK293T cells

• 3Two days after splitting, transfect cells using Effectene transfection reagent following the manufacturer’s instructions. Use 0.5–1 μg plasmid cDNA per 10⁶ cells.

  Total transfected DNA is maintained between experimental groups by co-transfecting empty plasmid vector pcDNA 3.1 if necessary (Life Technologies).

• 4Plate transfected cells on a twelve well plate containing poly-l-lysine coated coverslips. Add 1 ml of cell suspension per well.

  Coat the coverslips 24 h before plating the cells. Leave the poly-l-lysine O/N at 4°C and next day wash the coverslips 3-times with sterile water. Let them dry inside the hood. It is important to make sure that the poly-l-lyse is washed out fully because it is toxic for cells.

Imaging HEK293T cells

• 5Image FRET using FV1000 FluoView confocal laser-scanning microscope (Olympus) or equivalent microscope.

  For HEK293T cells, image FRET 2 – 3 days after transfection depending on the confluence, preferred to be around 60%.

  Settings presented in Table 3 were successfully tested on images acquired with a 40x objective (with a digital zoom of 3x)

• 6Remove carefully the coverslip from the plate and put it in the imaging chamber with the cells facing up towards the objective lens.
• 7Slowly add the HEK imaging media on the coverslip holder until the surface is covered.

  Ensure that the imaging chamber is not leaking. Dripping solutions can cause salts to accumulate on the lens and internal machinery of the microscope. This can lead to equipment failure and expensive repairs.

• 8Place the imaging chamber under the microscope.

  The apparent pixel intensity observed at the acceptor emission wavelength upon donor excitation is a result of three parameters: donor bleed through (leakage of donor emission into the acceptor emission), acceptor cross-talk (excitation of the acceptor at the donor excitation wavelength) and net FRET (emission of the acceptor due to energy transfer from the donor to the acceptor). Thus, first and foremost, assess the donor bleed through and the acceptor crosstalk values in order to calculate the net FRET signal. To this end, GFP (donor control) and the mCherry (acceptor control) conditions are imaged prior to the GFP plus mCherry condition.

• 9
  To calculate the donor bleed through, two sets of images are generated (see Figure 3):

  1. GFP is excited at 488 nm and the emission is measured at 505 nm. This value corresponds to the donor emission fluorescence.
     In Figure 3a it is referred as “donor emission”.
  2. GFP is excited at 488 nm and the emission is measured at 615 nm. This value corresponds to the amount of fluorescent signal that is measured in the red-shifted channel upon GFP excitation.
     In Figure 3a it is referred as “donor bleed through”.
• 10
  To assess the leak of the acceptor emission under the donor excitation (acceptor cross-talk), image mCherry expressing cells. As with the donor control, two sets of images are generated:

  1. mCherry is excited at 543 nm and the emission is measured at 615 nm. This value corresponds to the acceptor emission fluorescence.
     In Figure 3b appears as “acceptor emission”.
  2. mCherry is excited at 488 nm and the emission is measured at 615 nm. This value corresponds to the amount of mCherry emission fluorescence detected upon 488 nm wavelength excitation.
     In Figure 3b appears as “acceptor cross-talk”.
• 11Take several images from the donor and acceptor control group (~30–40 cells per condition) to accurately calculate the donor bleed through and the acceptor cross-talk. Name them accordingly (i.e. cell_1, cell_2, cell_3, etc).
• 12
  Use the same settings as previously mentioned (sections 9–10) to image the FRET condition. In this case, three sets of image are generated (Figure 3c):

  1. FRET (uncorrected): GFP is excited at 488 nm and mCherry is measured at 615 nm.
  2. Donor emission: GFP is excited at 488 nm and measured at 505 nm.
  3. Acceptor emission: mCherry is excited at 543 nm and measured at 615 nm.
• 13
  When the image acquisition is finalized, there should be the following image sets (Figure 3):

  1. Donor control (GFP alone): donor emission and donor bleed through
  2. Acceptor control (mCherry alone): acceptor emission and acceptor cross-talk
  3. FRET (GFP-mCherry fused): FRET, donor emission and acceptor emission
• 14Evaluate the FRET efficiency with ImageJ (PixFRET plug-in) as indicated in the section below.

Table 3.

Representative image settings used in the HEK293T cell experiments

Condition	Filter set	Excitation wavelength (nm)	Laser power (%)	PMT (V)	Gain (x)	Offset (%)	Pinhole	Dwell time (μs/pixel)
Donor	GFP	488	2	365	1	0	300	2.0
Acceptor	mCherry	543	10	480	1	0	300	2.0
FRET	mCherry	488	2	480	1	0	300	2.0

Figure 3.

Bleed through and cross-talk (SBT) calculations to measure sensitized emission FRET (SE-FRET). A. Diagram of the two channels’ imaging conditions for the GFP alone, i.e. donor bleed through SBT model. B. Diagram of the two channels’ imaging conditions for the mCherry alone, i.e. acceptor cross-talk SBT model. C. Diagram of the three channels’ imaging conditions for the sensitized emission FRET experiment between GFP and mCherry. D. Förster equation used to calculate FRET efficiency, E.

Analysis of SE-FRET with the PixFRET plug-in

The PixFRET plug-in runs under the version 1.33 of ImageJ. Upgraded versions of ImageJ are not compatible with this plug-in. To install PixFRET, download the PixFRET_.jar file from the website listed in the internet resources section of the manuscript. Copy it to the “plugins” folder in the ImageJ program and restart ImageJ.Calculate the spectral bleed through and cross-talk values with PixFRET plug-in.

Donor control bleed through

The donor control group includes two image settings: donor emission (488 nm excitation, 505 nm emission) and donor bleed through (488 nm excitation, 615 nm emission).

• 15Open the first acquired image (cell_1) of a cell or group of cells from the donor control group and split the two channels: Image > Stacks > Stacks to image (Figure 4a).
• 16Close channel 2 (donor bleed through), and keep channel 1 (donor emission) open.
• 17Do the same procedure (sections 15–16) for all the other images acquired.

  Open the images consecutively (cell_1, cell_2, cell_3…) and keep always the same order among conditions. This is important for the subsequent calculations.

• 18Make a stack of the opened images: Image > Stacks > Image to stacks (Figure 4a)
• 19Name it accordingly and select “use titles as labels”.
• 20Once the stack is done, make a montage: Image > Stacks > Make montage. Use scale factor of 1. Leave all the other parameters as defined (Figure 4a).

  Boosting the apparent pixel intensity with the “Brightness and contrast” tool helps to visualize the resulting montage.

• 21Save Stack and Montage files (.tiff) and close them.
• 22Open the donor control images but now use the other channel (bleed through, 488 nm emission and 615 nm excitation) and do the same as described previously. When opening the images, follow the same order as the first stack. At the end, save Stack and Montage files and close them (Figure 4b).
• 23
  Make the Donor stack for the donor control group. Open both Montages in the following order:

  1. Donor bleed through Montage
  2. Donor emission Montage
• 24Stack the montages: Image > Stack > Image to stack. Save it (.tiff). This is the donor control stack.

Figure 4.

Donor bleed through and acceptor cross-talk calculations in HEK293T cells. A. The GFP channel images (ex. 488 nm / em. 505 nm) should be stacked from separate FOVs, before stacking the FRET channel images (ex. 488 nm/ em. 615 nm). B. These stacks can then be used to create FOV donor montages of the GFP and FRET channels, these montages can then be stacked in a single file, the donor montage stack. C. From the montage stack we can graph the raw donor bleed through model and by thresholding the PixFRET viewer, we can determine the final donor model. D. For the acceptor controls, the mCherry channel images (ex. 543 nm/ em. 615 nm) should be stacked from separate FOVs, before stacking the FRET channel images (ex. 488 nm/ em. 615 nm). These stacks can then be used to create multiple FOV acceptor montages of the mCherry and FRET channels. E. The mCherry and FRET channel montages can then be stacked in a single file, the acceptor montage stack. F. From the montage stack we can graph the raw acceptor cross-talk model and by thresholding the PixFRET viewing window, we can determine the final acceptor model.

Acceptor control cross-talk

The acceptor control group contain two sets of images: acceptor emission (543 nm excitation, 615 nm emission) and acceptor cross-talk (488 nm excitation, 615 nm emission).

• 25Open the first acquired image (cell_1) from the acceptor control group and split the two channels: Image > Stacks > Stacks to image (Figure 4d).
• 26Close channel 2 (acceptor cross-talk), and keep channel 1 (acceptor emission) open.
• 27Proceed equally (sections 25–26) for all the images.

  Open the images consecutively (cell_1, cell_2, cell_3…) and keep always the same order among conditions. This is important for the following calculations.

• 28Make a stack of the images opened: Image > Stacks > Image to stacks (Figure 4d).
• 29Name it accordingly and select “use titles as labels”.
• 30Once the stack is done, make a montage: Image > Stacks > Make montage. Use scale factor of 1. Leave all the other parameters as defined (Figure 4e).

  Boosting the apparent pixel intensity with the “Brightness and contrast” tool helps to visualize the resulting montage.

• 31Save Stack and Montage files (.tiff) and close them.
• 32Open the acceptor control images but now use the other channel (cross-talk, 488 nm emission and 615 nm excitation). Follow the same steps as indicated previously. At the end, save Stack and Montage files and close them (Figure 4d, e).
• 33
  Make the Acceptor stack for the acceptor control group. Open both the Montages in the following order:

  1. Acceptor cross-talk Montage
  2. Acceptor emission Montage
• 34Stack the montages: Image > Stack > Image to stack. Save it (.tiff). This is the acceptor control stack.

Calculation of donor control bleed through and acceptor control cross-talk (SBT) values with PixFRET

Donor bleed through

• 35Open PixFRET: Plugins> PixFRET.
• 36Use default settings (Gaussian blur 3.0 and Threshold 1.0).
• 37Select for output: FRET Efficiency (%).
• 38Do not select “show blurred images”.
• 39Open the Donor control stack created previously.
• 40Use the polygon selection tool to select the background from the Montage.

  We recommend to boost the signal intensity (“Brightness and Contrast”) to distinguish the cells from the background. Before establishing the background, check both channels of the stack to make sure there is no real signal selected as a background.

• 41Select “Get” in the PixFRET window’s Background Donor section. Following that, select “Accept”.
• 42Use the rectangle selection tool to select the entaire montage.
• 43Select “get” under the Model Donor in the PixFRET window.
• 44A distribution model is generated representing the donor control background (Figure 4c).
• 45Some of the data points reported in the Donor SBT Model are noise and therefore will negatively interfere with the spectral bleed through calculation. Thus, use the yellow drag tool and the magnifying icon (top right corner) to adjust and restrict the window to the data points that accurately fit the equation (Figure 4c).

  There is no standard method to restrict the window to fit the data. However, the selected equation should include most of the data points represented in the graph, while eliminating low intensity noise on the far left of the graphs.

• 46Select the equation (constant, linear or exponential) that represents more precisely the data points. For the donor control model, select “constant”. Following that click “Accept”. Representative values for the donor SBT are located in Table 7 below.

Table 7.

Donor control model adjusted values

Condition	HEK293T cells			Brain slices
Model	a	b	e	a	b	e
Constant	0.92	-	-	0.22	-	-
Linear	0.94	−0.00004	-	-	-	-
Exponential	0.92	0.28	−0.003	0.22	0.0023	0.0009

Acceptor cross-talk

The calculation of the acceptor cross-talk is very similar than the donor bleed through, but using the Acceptor Control Stack.

• 47Open the “Acceptor Control Stack.”.
• 48Select the background using the polygon tool.
• 49Select “Get” under the Background Acceptor.
• 50Use the rectangle tool to select the entire Montage. Select “Get” under the Model Acceptor window.
• 51Restrict the data points accordingly with the magnifying tool (Figure 4f).
• 52Select the equation (constant, linear or exponential) that represents more precisely the data points. In this protocol the acceptor model is represented as an exponential equation. Following that, select “Accept”.
• 53Select “Save Parameters” on the FRET tab (PixFRET window) in order to save the background parameters for the FRET efficiency calculation. Representative values for the acceptor SBT are located in Table 8 below.

Table 8.

Acceptor control model adjusted values

Condition	HEK293T cells			Brain slices
Model	a	b	e	a	b	e
Constant	0.05	-	-	0.16	-	-
Linear	0.06	0.00009	-	-	-	-
Exponential	0.05	0.004	0.001	0.16	0.003	0.001

Calculation of FRET efficiency

• 54
  Open the following images from the FRET image set in the order mentioned below (Figure 5a):

  1. FRET channel (excitation 488 nm, emission 615 nm)
  2. Donor emission channel (excitation 488 nm, emission 505 nm)
  3. Acceptor emission channel (excitation 543 nm, emission 615 nm)
• 55Make a stack of the three images: Image > Stack > Image to stack (Figure 5a).
• 56Use the polygon tool to draw the global background making sure that the selected area does not contain real signal in any of the three images that compose the stack (Figure 5a). Increase the brightness to enhance the contrast between the background and the real signal. In the FRET tab, Select “Get” under the background window.
• 57Select as an output “FRET efficiency”.
• 58Select “Compute FRET”. Two new images will appear (% FRET efficiency and FRET stack).
• 59Select the window “% FRET efficiency” and determine the Threshold. Set the minimum value (% of FRET efficiency) to be considered a positive signal.
• 60Quantify the % of FRET efficiency: Analyze > Analyze particles > Measure
• 61A table is generated with the mean FRET efficiency per ROI. The values should range from 0.1 to 100 (Figure 5c).
• 62Choose a color pattern to visualize the FRET signal based on % of efficiency: Image> Lookup tables > (i.e. fire, cold…) (Figure 5b).

  In the current protocol the % of FRET efficiency is depicted with the color LUT “fire”.

  To confirm that the measured FRET values are a result of the energy transferred between the two fluorophores, we recommend to go back to the Stack image and evaluate whether the identified ROIs correspond to a detectable fluorescent signal in the FRET channel.

Figure 5.

Calculation of sensitized emission FRET efficiency in HEK293T cells. A. Example of the image order that should be used in creating the three channel FRET stacks before quantification in PixFRET. B. Representative image of FRET efficiency from the GFP-mCherry fusion experiment. C. Masked image of FRET from panel B, showing the calculated FRET efficiency from individual cells.

BASIC PROTOCOL 2

FÖRSTER RESONANCE ENERGY TRANSFER IMAGING WITH NAPA IN ACUTE BRAIN SLICES

Basic protocol 2 describes a genetically targeted neuron-astrocyte proximity assay whereby changes in astrocyte processes-to-synapse proximity are visualized using confocal microscopy. AAVs encoding Napa-a and Napa-n are locally microinjected in the motor cortex and the dorsal striatum, respectively. In the current protocol, we aimed to characterize the astrocyte process–to-synapse proximity of corticostriatal inputs by measuring sensitized emission FRET (SE-FRET). An overview of the entire protocol is shown in Figure 1.

Materials

Adeno Associated virus (AAV) microinjections

• 1Use both male and female mice between 6–8 weeks of age (for example, C57BL/6N).
• 2Anesthetize the mice using isoflurane (induction at 5%, maintenance at 1–2.5% v/v); place the mouse inside the induction chamber for 2–3 minutes and move it to the breathing circuit, securing the nose cone to the animal’s face.
• 3Prior to surgery, administrate 0.05 ml of buprenophrine (Buprenex diluted to 0.1 mg/ml) subcutaneously.
• 4After induction of anesthesia, place the nose of the mouse into a stereotaxic frame and secure the head using blunt ear bars.
• 5Clean surgical area with 10% povidone iodine and 70% alcohol swabs.
• 6Perform unilateral viral injections using a stereotaxic apparatus to deliver the virus into the brain region of interest (see Athos and Storm 2001 for a more detailed stereotaxic surgery protocol). The coordinates (axes relative to Bregma) are as follows: dorsolateral striatum: Anterior-Posterior (AP) +0.8/ Medial-Lateral (ML) +2.0/ Dorsal-Ventral (DV) −2.4, motor cortex: AP −0.1/ ML +0.7/ DV −0.3.

  Use comparable titers between both viruses. In this study, we used AAV2/5 GfaABC₁D Napa-a (1.85 × 10¹³ gc/mL) and AAV2/1 hSynapsin1 Napa-n (6.7 × 10¹² gc/mL). Napa-a was diluted 1:2.

• 7Using a glass needle at the corresponding coordinates, start infusing with an injection rate of 0.2 μl/min. The injected volume of Napa-a was 0.5 μl and 1.2 μl for Napa-n.

  The virus volume will vary depending on the area of interest. Smaller areas should be targeted with lower volumes to ensure that there is not spread to neighboring brain regions.

• 8Leave the needle in place for 10 minutes after injection to allow diffusion of the virus in the brain.
• 9Pull up the needle slowly over the course of 2 minutes.
• 10Close the wound with external silk suture.
• 11After surgery, place mice in cages halfway on a low-voltage heating pad.
• 12Provide Trimethoprim and sulfadiazine diet to mice for ad lib feeding.
• 13Weight the mice one time per day up to three days after surgery to ensure a proper recovery.

  A video of the AAV microinjection procedure has been produced by our colleagues (Jiang et al 2014, video available at: https://dx.doi.org/10.3791%2F51972)

Brain slice preparation

• 14Prepare 500 ml of slicing solution and 1–2 liters of recording solution (aCSF). See Reagents and Solutions section for detailed recipe.
• 15Fill the slice holder with aCSF and keep it at 32–34 °C. Make sure the solutions are constantly saturated with 95% O₂ and 5% CO₂.
• 16Fill the vibratome chamber with ice-cold slicing solution previously saturated with 95% O₂ and 5% CO₂ for ~ 30 minutes (on ice).
• 17Decapitate the mouse swiftly with a pair of sharp shears.
• 18Place the head of the mouse in an ice-cold recipient with slicing solution. Remove the scalp using small tweezers.
• 19Carefully extract the mouse brain from the skull and discard the cerebellum and olfactory bulbs using a clean sharp blade.
• 20Pour ice-cold slicing solution on the brain in order to maintain it cold and oxygenated.
• 21Mount the brain onto the vibratome tray using super glue. Wait 20 seconds until the glue is dry.
• 22Fill the vibratome tray with ice cold slicing solution.
• 23Cut the tissue sections on the coronal plane with 300 μm thickness. Collect the slices containing the region of interest.
• 24Transfer the slices to the slice holder. Allow for the slices to recover at 32–34 °C for 20 minutes.
• 25Place the slice holder at room temperature for a minimum of 30 minutes before proceeding to imaging.

Imaging of neuron-astrocyte proximity in brain slices

As advised in “critical parameters” section, we recommend generating a small data set (donor control, acceptor control and FRET conditions) in which ~ 3 different laser powers are tested to excite 488 nm and 543 nm. Furthermore, it is especially important to follow this advice for the quantification of SE-FRET in brain slices due to the higher amount of background that considerably affects the signal-to-noise ratio.

• 26For brain slices, examine FRET 18–22 days following AAV microinjections.
• 27Connect the flow (aCSF in constant bubbling) to the imaging chamber. Place one end of the tube in the aCSF bottle and the other end in the imaging chamber. Position the suction on the opposing site of the running solution so there is always a constant flow coming in the imaging chamber. Let it run for 5 minutes to make sure there is no leakage.
• 28Put the slice in the imaging chamber and immobilize it with a harp.

  A harp is a slice anchor that is used to secure the slice in the imaging chamber. It is made of platinum and 3 or 4 thin threads of nylon. It can be made in the laboratory ).

• 29Place it strategically so the region of interest is not damaged or occluded by the harp.
• 30Find the region of interest with the 10x magnification objective and then switch to 40x to begin FRET imaging.

  In order to visualize one entire astrocyte per field of view, it is advised to use a zoom factor of 3. See table 6 for detailed image acquisition parameters.

  Take several images (~30 per mouse) and name them accordingly (i.e. cell_1, cell_2, cell_3, etc).

  To assess the bleed through and the cross-talk values, the “donor control” and “acceptor control” groups are imaged in the first place.

• 31
  For the donor control, two images for each acquisition are generated (Figure 3a):

  1. Napa-a (GFP) is excited at 488 nm and the emission is measured at 505 nm. This value corresponds to the donor emission signal.
     In Figure 3a appears as “donor emission”.
  2. Napa-a (GFP) is excited at 488 nm and the emission is measured at 615 nm. This value corresponds to the amount of signal that it’s measured in the mCherry channel upon GFP activation.
     In Figure 3a appears as “donor bleed through”.
• 32
  To assess the leakage of the acceptor emission under the donor excitation (cross-talk), mCherry expressing mice are imaged. Two images are generated:

  1. Napa-n is excited at 543 nm and the emission is measured at 615 nm. This value corresponds to the acceptor emission signal.
     In Figure 3b appears as “acceptor emission”.
  2. Napa-n is excited at 488 nm and the emission is measured at 615 nm. This value corresponds to the cross-talk, the amount of mCherry emission upon 488 nm wavelength excitation without the presence of the donor.
     In Figure 3b appears as “acceptor cross-talk”.
• 33
  Following the imaging of the donor and the acceptor control, proceed with the mice that express both Napa-a (GFP) and Napa-n (mCherry). Keep the same image settings as used previously for the donor and acceptor controls. For the NAPA group, three images are generated (Figure 3c):

  1. FRET (uncorrected): GFP is excited at 488 nm and the emission is measured at 615 nm.
  2. Donor emission: GFP is excited at 488 nm and the emission is measured at 505 nm.
  3. Acceptor emission: mCherry is excited at 543 nm and the emission is measured at 615 nm.
• 34
  At the end there should be the following image sets (Figure 3):

  1. Donor control (GFP alone): donor emission and donor bleed through.
  2. Acceptor control (mCherry alone): acceptor emission and acceptor cross-talk.
  3. FRET (GFP & mCherry): FRET, donor emission and acceptor emission.
• 35Evaluate the FRET efficiency with ImageJ (PixFRET plug-in) as indicated in the section below.

Table 6.

Acquisition settings tested to detect FRET in brain slices

Condition	Filter set	Excitation wavelength (nm)	Laser power (%)	PMT (V)	Gain (x)	Offset (%)	Pinhole (μm)	Dwell time (μs/pixel)
Donor	GFP	488	14–24	585	1	0	300	8.0
Acceptor	mCherry	543	25–35	610	1	0	300	8.0
FRET	mCherry	488	14–24	610	1	0	300	8.0

Analysis of sensitized-emission FRET with PixFRET plug-in

Calculate the spectral bleed through and cross-talk values with PixFRET plug-in

Donor control bleed through

The donor control group includes two image settings: donor emission (488 nm excitation, 505 nm emission) and donor bleed through (488 nm excitation, 615 nm emission).

• 1Open the first acquired image (cell_1) of an astrocyte from the donor control group and split the two channels: Image > Stacks > Stacks to image (Figure 7a).
• 2Close channel 2 (bleed through), and keep channel 1 (donor emission) open.
• 3Do the same procedure (sections 1–2) for all the other images acquired.

  Open the images consecutively (cell_1, cell_2, cell_3…) and keep always the same order among conditions. This is important for the following calculations.

• 4Make a stack of the opened images: Image > Stacks > Image to stacks (Figure 7a)
• 5Name it accordingly and select “use titles as labels”.
• 6Once the stack is done, make a montage: Image > Stacks > Make montage. Use scale factor of 1. Leave all the other parameters as defined.

  Boosting the apparent pixel intensity with the “Brightness and contrast” tool helps to visualize the resulting montage.

• 7Save Stack and Montage files (.tiff) and close them.
• 8Open the donor control images but now use the other channel (bleed through, 488 nm emission and 615 nm excitation) and do the same as described previously. When opening the images, follow the same order as the first stack. At the end, save Stack and Montage files and close them (Figure 7b).
• 9
  Make the Donor stack for the donor control group. Open both Montages in the following order:

  1. Donor bleed through Montage
  2. Donor emission Montage
• 10Stack the montages: Image > Stack > Image to stack. Save it (.tiff). This is the donor control stack (Figure 7b).

Figure 7.

Calculation of donor bleed through and acceptor cross talk in brain slices. A. The GFP channel images (ex. 488 nm/ em. 505 nm) should be stacked from separate FOVs, before stacking the FRET channel images (ex. 488 nm / em. 615 nm). B. These stacks can then be used to create FOV donor montages of the GFP and FRET channels. The channels can then be stacked in a single file, the donor montage stack. C. From the montage stack we can graph the raw donor bleed through model and by thresholding the PixFRET viewer, we can determine the final donor model. D. For the acceptor controls, the mCherry channel images (ex. 543 nm/ em. 615 nm) should be stacked from separate FOVs, before stacking the FRET channel images (ex. 488 nm/ em. 615 nm). These stacks can then be used to create multiple FOV acceptor montages of the mCherry and FRET channels. E. The mCherry and FRET channel montages can then be stacked in a single file, the acceptor montage stack. F. From the montage stack we can graph the raw acceptor cross-talk model and by thresholding the PixFRET viewing window, we can determine the final acceptor model.

Acceptor control cross-talk

The acceptor control group contain two sets of images: acceptor emission (543 nm excitation, 615 nm emission) and acceptor cross-talk (488 nm excitation, 615 nm emission).

• 11Open the first acquired image (cell_1) from the acceptor control group and split the two channels: Image > Stacks > Stacks to image (Figure 7d).
• 12Close channel 2 (cross-talk), and keep channel 1 (donor emission) open.
• 13Proceed equally (sections 11–12) for all the images left.

  Open the images consecutively (cell_1, cell_2, cell_3…) and keep always the same order among conditions. This is important for the following calculations.

• 14Make a stack of the images opened: Image > Stacks > Image to stacks (Figure 7e).
• 15Name it accordingly and select “use titles as labels”.
• 16Once the stack is done, make a montage: Image > Stacks > Make montage. Use scale factor of 1. Leave all the other parameters as defined (Figure 7e).

  Boosting the apparent pixel intensity with the “Brightness and contrast” tool helps to visualize the resulting montage.

• 17Save Stack and Montage files (.tiff) and close them.
• 18Open the acceptor control images but now use the other channel (cross-talk, 488 nm emission and 615 nm excitation). Follow the same steps as indicated previously. At the end, save Stack and Montage files and close them (Figure 7d, e).
• 19
  Make the Acceptor stack for the acceptor control group. Open both the Montages in the following order:

  1. Acceptor cross-talk Montage
  2. Acceptor emission Montage
• 20Stack the montages: Image > Stack > Image to stack. Save it (.tiff). This is the acceptor control stack.

Calculation of donor control bleed through and acceptor control cross-talk values with PixFRET

Donor bleed through

• 21Open PixFRET: Plugins> PixFRET.
• 22Use default settings (Gaussian blur 3.0 and Threshold 1.0).
• 23Select for output: FRET Efficiency (%).
• 24Do not select “show blurred images”.
• 25Open the Donor control stack created previously.
• 26Use the polygon selection tool to select the background from the Montage.

  We recommend to boost the signal intensity (using the ImageJ function “Brightness and Contrast” to adjust the LUTs) to distinguish the cells from the background. Before establishing the background, check both channels of the stack to make sure there is no real signal selected as a background.

• 27Select “Get” in the PixFRET window’s Background Donor section. Following that, select “Accept”.
• 28Use the rectangle selection tool to select the entaire montage.
• 29Select “get” under the Model Donor in the PixFRET window.
• 30A distribution model is generated representing the donor control background (Figure 7c).
• 31Some of the data points reported in the Donor SBT Model are noise and therefore will negatively interfere with the spectral bleed through calculation. Thus, use the yellow drag tool and the magnifying icon (top right corner) to adjust and restrict the window to the data points that accurately fit the equation (Figure 7c).

  There is no standard method to restrict the window to fit the data. However, the selected equation should include most of the data points represented in the graph, while eliminating low intensity noise on the far left of the graphs.

• 32Select the equation (constant, linear or exponential) that represents more precisely the data points. For the donor control model, select “constant”. Following that click “Accept”. Representative values for the donor SBT are located in Table 7.

Acceptor cross-talk

The calculation of the acceptor cross-talk is very similar than the donor bleed through, but using the Acceptor Control Stack.

• 33Open the “Acceptor Control Stack”.
• 34Select the background using the polygon tool.
• 35Select “Get” under the Background Acceptor
• 36Use the rectangle tool to select the entire Montage. Select “Get” under the Model Acceptor window.
• 37Restrict the data points accordingly with the magnifying tool (Figure 7f).
• 38Select the equation (constant, linear or exponential) that represents more precisely the data points. In this protocol the acceptor model is represented as an exponential equation. Following that, select “Accept”.
• 39Select “Save Parameters” on the FRET tab (PixFRET window) in order to save the background parameters for the FRET efficiency calculation. Representative values for the acceptor SBT are located in Table 8.

Calculation of FRET efficiency

• 40
  Open the following images from the FRET image set in the order mentioned below (Figure 8b):

  1. FRET channel (excitation 488 nm, emission 615 nm)
  2. Donor emission channel (excitation 488 nm, emission 505 nm)
  3. Acceptor emission channel (excitation 543 nm, emission 615 nm)
• 41Make a stack of the three images: Image > Stack > Image to stack (Figure 8b)
• 42Use the polygon tool to draw the global background making sure that the selected area does not contain real signal in any of the three images that compose the stack (Figure 7a). Increase the brightness to enhance the contrast between the background and the real signal. In the FRET tab, Select “Get” under the background window (Figure 8b).
• 43Select as an output “FRET efficiency”.
• 44Select “Compute FRET”. Two new images will appear (% FRET efficiency and FRET stack).
• 45Select the window “% FRET efficiency” and determine the Threshold. Set the minimum value (% of FRET efficiency) to be considered a positive signal.
• 46Quantify the % of FRET efficiency: Analyze > Analyze particles > Measure
• 47A table is generated with the mean FRET efficiency per ROI. The values should range from 0.1 to 100.
• 48Choose a color pattern to visualize the FRET signal based on % of efficiency: Image> Lookup tables (LUT) > Select one from the menu (e.g. fire, cold, etc.) (Figure 8c)

  In the current protocol the % of FRET efficiency is depicted with the LUT “fire”.

  To confirm that the measured FRET values are a result of the energy transferred between the two fluorophores, we recommend to return to the stack image and evaluate whether the identified ROIs correspond to a detectable fluorescent signal in the FRET channel.

Figure 8.

Calculation of sensitized emission FRET efficiency in brain slices. A. Diagram of astrocyte-synapse proximity assay on corticostriatal projections. B. Example of the image order that should be used in creating the three channel FRET stacks before quantification in PixFRET. C. Representative image of all imaging channels and FRET efficiency from the NAPA experiment. The white outline represents estimated edge of astrocyte territory based on NAPA-a expression. D. Quantification of FRET efficiency from 11 astrocytes (Mean ± SEM).

REAGENTS AND SOLUTIONS

HEK293T imaging buffer

For 1 L recording solution:

150 mM NaCl	(8.8 g; MilliporeSigma, cat. no. S5886)
1 mM CaCl2	(1 ml of 1 M stock solution; MilliporeSigma, cat. no. 21108)
1 mM MgCl2	(1 ml of 1 M stock solution; MilliporeSigma, cat. no. M2670)
10 mM Glucose	(1.8 g; MilliporeSigma, cat. no. G7528)
10 mM HEPES	(2.4 g; MilliporeSigma, cat. no. H3375)

The final solution should be slowly buffered to pH 7.4 dropwise with NaOH. Do not add excess NaOH to the solution or the osmolarity will be negatively affected.

Acute brain slice preparation solutions

For 500 ml of slicing solution (modified-aCSF):

194 mM sucrose	(33.2 g; MilliporeSigma, cat. no. S7903)
30 mM NaCl	(0.88 g)
4.5 mM KCl	(0.17 g; MilliporeSigma, cat. no. P9333)
10 mM D-glucose	(0.9 g)
1.2 mM NaH2PO4	(0.07 g; MilliporeSigma, cat. no. S0751)
26 mM NaHCO3	(1.09 g; VWR, cat. no. 0865)
1 mM MgCl2	(0.5 ml of 1 M stock solution)

To minimize the risk of precipitation MgCl₂ should be added after all other solutes are completely dissolved. Final composition should be ~330 mOsm. After dissolving all reagents, the solution should be saturated with 95% O₂ and 5% CO₂ by bubbling for 15 minutes. Prior to cutting, solution should be cooled to 4°C on ice. This solution can be stored at 4°C for 4–5 days.

For 2 L of recording solution (aCSF):

124 mM NaCl	(14.49 g)
4.5 mM KCl	(0.67 g)
26 mM NaHCO3	(4.36 g)
1.2 mM NaH2PO4	(0.28 g)
10 mM D-glucose	(3.6 g)
2 mM CaCl2	(4 ml of 1 M stock solution)
1 mM MgCl2	(2 ml of 1 M stock solution)

To minimize the risk of precipitation, CaCl₂ and MgCl₂ should be added after all other solutes are completely dissolved. Final composition should be ~330 mOsm. After dissolving all reagents, the solution should be saturated with 95% O₂ and 5% CO₂ by bubbling for 15 minutes, and the pH should be 7.3 – 7.4. Prior to cutting slices, a beaker containing this solution should be heated to between 32–34°C. All materials that contact aCSF for acute slices should be thoroughly cleaned beforehand.

COMMENTARY

BACKGROUND INFORMATION

Astrocyte morphology is remarkably complex and the potential role(s) for astrocyte process dynamics at synapses are largely unexplored, although important progress has been made. Single mature rodent astrocytes encompass 20–80 10³ cubic micrometers per territory and have been estimated to contact >100,000 hippocampal synapses (Bushong et al 2002). Significant remodeling of astrocyte processes has been documented in the context of lactation and dehydration (Theodosis et al 2008). Pioneering early work in the hippocampus using electron microscopy approaches demonstrated that astrocytes make varied contacts with synaptic structures over narrow distances on the order of nanometers (Ventura & Harris 1999). More recent studies in organotypic tissue culture using virally expressed fluorescent proteins have suggested that astrocyte processes dynamically move in relation to adjacent synapses (Bernardinelli et al 2014, Haber et al 2006, Perez-Alvarez et al 2014). However, these approaches are limited to the observation of movements on the order of micrometers.

To document proximity and smaller distance scales, we developed NAPA as a method of interrogating the relationship and dynamics of astrocyte processes at synapses with sub micrometer resolution. NAPA uses FRET to quantify spatial relationships between appropriately labelled proteins. FRET is typically quantified by three common methods: donor dequenching, ratiometric imaging and sensitized emission (SE-FRET). Additional methods such as fluorescence lifetime imaging (FLIM-FRET) require highly specialized imaging equipment and those methods are not discussed here. That said, if the equipment is available time-resolved FLIM probably represents the best available method for accurate FRET quantification and should be the technique of choice.

The simplest method of FRET assessment is donor dequenching. This method relies on the fact that donor photonic emission is diminished during FRET due to loss of energy from the excited state to the acceptor via non radiative dipole-dipole coupling. When the acceptor is optically destroyed it is gradually photobleached, which reduces dipole-dipole coupling leading to dequenching of the donor and consequent enhanced photonic emission. This method has two limitations: it takes time to photobleach the acceptor, which decreases the temporal resolution of the experiment and of course it is destructive, which limits the assessment to single FRET measurements.

Time series ratiometric imaging enables the user to track dynamic changes in the amount of apparent relative FRET on a per-pixel basis. In this method, the fluorophore emission intensities in the two fluorophore channels are quantified and the apparent FRET measurement is expressed as a ratio between the two images. The limitation of this method is that the initial FRET efficiency is not determined, because the readout is expressed as a change in fluorophore intensity ratio. However, the method is ratiometric and reciprocal changes in donor and acceptor intensities are a convenient way to assess changes in FRET relative to baseline.

Sensitized emission is a straightforward method to quantify FRET in the initial state and for dynamic changes over time (this was the main reason to use it for NAPA) as described in this protocol. It is optically non-destructive, can be used for live imaging and can be easily implemented using standard light microscopes on a pixel-by-pixel basis. To assess FRET efficiency through sensitized emission images, we subtract the donor bleed through and the acceptor crosstalk from the apparent FRET signal in order to get the net FRET signal. The methods to do this have been described in the preceding sections.

However, SE-FRET does have some limitations. First, sensitized emission is influenced by the local fluorophore concentration. In the past we have performed a series of controls to confirm that the vast majority of ROIs containing FRET fall within the quantifiable range of fluorescence parameters. Second, most sensitized emission-based FRET indices respond nonlinearly to changes in the degree of molecular interaction and depend on the optical parameters of the imaging system (Zal & Gascoigne 2004). Thus, selecting the correct image settings and keeping them consistent among the different imaging conditions is essential for successful NAPA experiments.

In summary, FRET is a powerful method that permits assessment of proximity between appropriately labelled molecules, such as proteins tagged with fluorescent proteins (Lakowicz 2006). Using such approaches, NAPA enabled us to measure both stable and dynamic interactions between presynaptic terminals and astrocyte processes using confocal microscopy. Hence, NAPA can be used to evaluate the relative proximity between astrocyte processes and various neuronal inputs; collateral inputs form closest proximal interactions with striatal astrocytes compared to thalamic, cortical and SNc inputs. Moreover, NAPA can also be used to estimate the closest synaptic contacts (i.e. within 10 nm distance). For example, in the striatum we found there was no differences between the astrocyte contacts for either D1 or D2 collateral projections (Octeau et al 2018). Under the correct conditions and for the appropriate questions, NAPA is a powerful technique that enables interrogation of the dynamic spatial relationships between astrocyte processes and synapses in intact neural circuits. It is likely the method could be further improved to extend the reach of FRET to longer distances and to target cognate fluorophores to additional cellular compartments such as axons, dendrites, somata and microglial processes. Furthermore, as the fluorescent proteins used for FRET are improved iteratively for expression, stability and intensity, these could be incorporated into the NAPA constructs.

CRITICAL PARAMETERS

Donor and acceptor stoichiometry

An important factor that can limit FRET detection is the relationship of donor-acceptor stoichiometry. For FRET measurements of protein–protein interactions, the donor/acceptor ratio should be between 0.1 and 10 (Berney & Danuser 2003). Determine the intensities of FRET ROIs in the donor and acceptor channels as a first approximation of donor-acceptor stoichiometry. Additionally, confirm that these intensities are significantly elevated above background, by at least 2-fold. Accordingly, the FRET ROIs that display donor/acceptor ratios >10 or <0.1 should not be included in the analysis, because these can contribute to over-and under-estimation of FRET efficiency. FRET pairs expressed in a tandem construct yields increased FRET efficiency because the donor and the acceptor stoichiometry are equal and all fluorophores exist at a pre-determined distance from each other. We recommend measuring Napa-n and Napa-a intensities in this way to verify that the measured SE-FRET efficiency does not artificially depend on high or low donor or acceptor expression and therefore likely reflects the real distance between the astrocyte and neuronal membranes. Tables 7 and 8 report representative donor and acceptor SBT models used to measure FRET in HEK293T cells and in brain slices.

Signal-to-noise ratio

FRET efficiency reflects the energy transfer between two fluorophores (e.g. GFP and mCherry) depending on their proximity. This relationship is described by the Fӧrster equation (Figure 3d). Hence, an important step of this protocol is to determine suitable image settings to enhance detection of energy transferred between the two FRET pairs, without saturating pixels in either channel. Thus, we would advise the user to first generate a small data set (donor control, acceptor control and FRET conditions) to test approximately 3 different laser powers to excite at 488 nm and 543 nm respectively. This intermediate step would allow the user to find desired image settings to measure FRET (in vitro and in situ) prior to the real experiment. Table 3 and 4 show representative settings used to measure FRET in HEK293T cells and in brain slices, respectively.

Table 4.

Materials and reagents used for measuring FRET in brain slices

Reagent or resource	Source	Identifier
Laboratory mice (C57BL/6N), both male and female mice (between 8–12 weeks of age)	Jackson Labs	Jack stock # 005304; RRID: IMSR_JAX:005304
Fine borosilicate glass micropipettes	WPI	1B100-4
Artificial tears ointment	Bausch & Lomb	NDC 24208-785-55
Isoflurane (gas)	Henry Schein Animal Health	NDC 11695-6776-2
Low-voltage heating pad	Sunbeam	Cat no. 732500000U
Buprenorphine (0.3 mg/ml stock)	MWI	Cat. no. 060969
10% povidone iodine	Purdue Products L.P.	NDC 67618-150-09
70% alcohol swabs	BD	HRI 326895
Cotton swab and dental cotton balls	Richmond	Cat. no. 100110
Nylon suture	Surgical Specialties	REF752B
Trimethoprim & sulfadiazine diet	Envigo	TD.06596
AAV2/5 GfaABC1D NAPA-a SV40	This manuscript, UPenn Vector Core	Available upon request from UPenn
AAV2/1 hSynapsin 1 Napa-n SV40	This manuscript, UPenn Vector Core	Available upon request from UPenn

TROUBLESHOOTING

STATISTICAL ANALYSIS

For all statistical comparisons we use the following protocol to select appropriate statistical measures. First, all groups are tested for normality. If all groups are found to contain data that is drawn from a normally distributed population than parametric tests are used for direct comparisons. If any group is found to be drawn from a non-normally distributed population than non-parametric tests are used. For non-parametric tests we used Mann-Whitney U tests, for parametric tests we used Student’s T tests. For statistical comparisons between more than two conditions we first perform an ANOVA to check for any significance amongst all conditions and then perform appropriate post-hoc tests for parametric or non-parametric datasets. For statistical analysis, we recommend to image multiple astrocytes per mice (25–35 cells), with a minimum of 4 mice per condition.

UNDERSTANDING RESULTS

Herein we have provided a protocol for assessment of FRET by sensitized emission using the PixFRET plugin for ImageJ (Figure 1). In the case of FRET measured with mCherry-GFP fusion proteins in HEK293T cells, both fluorophores are expressed in equal amounts in the same cell and separated by a known distance (Figure. 2a). On the other hand, FRET based evaluations of astrocyte process-to- synapse proximity are based on the independent expression of the two fluorophores; GFP is targeted to striatal astrocytes and mCherry is located at presynaptic corticostriatal terminals (Figure. 2b). The efficiency of FRET under the conditions of the protocol described is determined by 1) spectral overlap, 2) quantum yield of the donor, 3) the relative orientation of the donor and acceptor dipoles, and 4) the distance between the donor and the acceptor with a steep drop in the energy transfer summarized in the Förster equation (Figure. 3d).

Figure 2.

Neuron-astrocyte proximity assay (NAPA) experimental setup in vitro and in situ. A. Experimental workflow to measure FRET in cultured HEK293T cells. B. Experimental workflow to measure FRET ex vivo in acute brain slices following viral transduction in vivo.

Experiments for pixel-wise quantification of FRET can be performed in cultured cells using appropriate donor (Figure. 4a–c) and acceptor (Figure. 4d–f) spectral bleed through (SBT) model parameters. Using the experimentally determined SBT values and three-channel image stacks we can calculate FRET efficiency values in a subset of image pixels (Figure. 5a–c) and perform analysis of multiple images to generate average FRET efficiency for a particular dataset (Figure. 6). In this way we can use FRET to experimentally determine the distance between fluorophores in live cells.

Figure 6.

FRET efficiency based on tandem GFP-mCherry in HEK293T cells. A. Representative images of the HEK293T cells expressing, GFP alone (top panels), mCherry alone (middle panels) and tandem GFP-mCherry (bottom panels). B. Quantification of FRET efficiency from 20 cells in these three conditions (Mean ± SEM).

This method also requires independent experimental determination of donor (NAPA-a, Figure. 7a–c) and acceptor (NAPA-n, Figure. 7d–f) SBT parameters. We can then utilize these values to calculate FRET efficiency in the NAPA assay using three channel image stacks (Fig. 8a–d). However, the detection of FRET between the astrocyte process and the presynaptic terminal is dependent on the stability of the astrocyte process-synapse interactions as well as the relative orientation of the fluorophores. To be able to quantify FRET, there should be no change in this distance during the time required to perform the measurement.

TIME CONSIDERATIONS

Basic Protocol 1 in HEK293T cells can be completed over 5–6 days. For Basic Protocol 2 in acute brain slices the experiments and preliminary analysis can be completed in approximately 5–6 weeks. The AAV microinjections take approximately 1–2 days depending upon the number of mice and conditions desired. In-vivo expression requires a delay of 3–4 weeks for maximum expression of AAVs in neurons and astrocytes. Imaging takes about 1 day per 2 mice imaged.

Table 1.

Materials and reagents used for measuring FRET in HEK293T cells

Reagent or resource	Source	Identifier
HEK293T cells	ATCC	Cat# CRL-3216
DMEM/F12 GlutaMAX	GIBCO	Cat#10-565-018
Penicillin and Streptomycin solution	Sigma Aldrich	Cat# P0781
Fetal bovine serum	Thermo Fisher	Cat# A31605
Poly-l-lysine	Sigma Aldrich	Cat# A-003-M
Effectene transfection reagent	QIAGEN	Cat# 301425
CMV GFP SV40	Octeau et al., 2018	Addgene; plasmid #92277
CMV mCherry BGH	Octeau et al., 2018	Addgene; plasmid #92278
CMV mCherry GFP BGH	Octeau et al., 2018	Addgene; plasmid #92280
ImageJ v1.30	ImageJ, NIH	N/A

Table 2.

Instruments used for measuring FRET in HEK293T cells

Instrument	Source	Identifier
HeNe 543 laser	CVI Melles Griot	05-LPL-915-065
Argon 488 laser	Showa Uptronics Co.	GLS3135
mCherry filter set	Olympus	BA560-620
GFP filter set	Olympus	BA505-525
FV1000 FluoView confocal laser-scanning microscope	Olympus	RRID:SCR_016840
HEPA filter Laminar biosafety cabinet (cell culture)	N/A	N/A

Table 5.

Instruments used to measure FRET in brain slices

Instrument	Source	Identifier
FV1000 Fluoview confocal laser-scanning microscope	Olympus	RRID:SCR_016840
Stereotaxic apparatus	David Kopf Instruments	Model 940
Injector pump	David Kopf Instruments	Model UMP3
Induction chamber	Stoelting co.	Cat. no. 53915

Table 9.

Troubleshooting common problems in NAPA experiments

Problem	Common cause(s)	Solution(s)
NAPA-n expression too low to visualize	Injection location	Repeat injection with correct coordinates
	Time for viral expression of AAV	Allow 4 weeks for sufficient viral transduction to occur
	Imaging settings	Increase 543 nm laser power setting or open confocal aperture
Overlapping NAPA-a signal from adjacent astrocytes	High AAV NAPA-a titer	Dilute NAPA-a to 1:5 or 1:10 with sterile saline solution and repeat the injections
No or more SE-FRET ROIs observed than expected	NAPA-n signal	Increase 543 nm laser power or PMT voltage for NAPA-n and FRET channels
		Choose FOV with greater NAPA-n expression
	NAPA-a signal	Choose FOV with lower astrocyte expression of NAPA-a
	Background settings	Select an ROI with lower background for SE-FRET calculation
	SBT values	Repeat calculation of SBT values
	Raw FRET image	Confirm raw FRET images contain visible ROIs
SE-FRET images have abnormally high or low background pixel intensities	Image stack order	Confirm images are in the appropriate order before calculating SE-FRET
Raw 3rd channel FRET image shows no/low signals	Excitation of donor	Increase 488 nm laser power
	Expression of NAPA-n	Select FOV with NAPA-n ROIs
		Wait an additional week post AAV injection before imaging
Abnormally high intensity ROIs of SE-FRET within astrocyte somas	Expression of donor	Select FOV with modest level of NAPA-a expression
		Remove somatic signals after SE-FRET calculation
SE-FRET ROIs have efficiency values <0 or >100	SBT values	Repeat calculation of SBT values

SIGNIFICANCE STATEMENT.

Astrocytes, the most numerous glial cells in the central nervous system, are highly branched and morphologically complex. A large fraction of an astrocyte’s surface area comprises thousands of processes. These fine fingerlike structures make discrete contacts with synapses at distance scales below the resolution of standard light microscopy. Furthermore, astrocytes perform regulatory and neuromodulatory roles at synapses via their fine processes, which are likely to be highly dependent on the extent of their spatial interactions and proximity. Previously we reported a Förster resonance energy transfer (FRET) based approach called the Neuron-Astrocyte Proximity Assay (NAPA), which enables observation and tracking of nanometer scale contacts between astrocytes processes and synapses using standard confocal light microscopy. This protocol describes NAPA and provides step-by-step instructions on how to implement it as a routine workhorse.