Imaging of in vitro and in vivo neurons in Drosophila using stochastic optical reconstruction microscopy

Abstract

The Drosophila brain comprises different neuronal cell types that interconnect with precise patterns of synaptic connections. These patterns are essential for the normal function of the brain. To understand the connectivity patterns, we must characterize them at single-cell resolution, for which a fluorescence microscope becomes an indispensable tool. Additionally, because the neurons connect at the nanoscale, the investigation often demands super-resolution microscopy. Here, we adopt one of the super-resolution microscopy techniques, called STORM (Stochastic Optical Reconstruction Microscopy), improving the lateral and axial resolution to ~20 nm. This unit extensively describes our methods and considerations for sample preparation of neurons in vitro and in vivo, conjugation of dyes to antibodies, immunofluorescence labeling, and STORM data acquisition and processing. With these tools and techniques, we open up the potential to investigate cell-cell interactions using STORM in the Drosophila nervous system.

INTRODUCTION

The nervous system of Drosophila is an attractive model for studying the molecular mechanisms of neuronal differentiation and proliferation, neurite outgrowth, synapse formation, and neurodegeneration (Bellen, Tong, & Tsuda, 2010; Clark, Zarin, Carreira-Rosario, & Doe, 2018; Doe, 2017). In particular, the ability to express transgenes has facilitated the use of Drosophila in carrying out these studies (Venken, Simpson, & Bellen, 2011). For instance, in the brain where many neurites are entangled, it is difficult to distinguish a single neurite from others (Scheffer et al., 2020). To resolve each neurite at high contrast, we commonly express membrane markers in small subsets of neurons in the brain. To this end, we can take advantage of a transgenic fly in which the transcriptional activator GAL4 is expressed under the control of a cell-type-specific enhancer (Manning et al., 2012). Complementary to such an in vivo approach, we can also employ primary neuronal cultures from the Drosophila brain (Egger, van Giesen, Moraru, & Sprecher, 2013). The availability of well-defined culture media and advanced culturing techniques allows us to dissociate individual neurons and generate primary neuronal cultures. Primary neuronal cultures have previously been used for high-resolution imaging of neurite arbors, high-throughput screening of genes for developmental brain disorders, and cell-recording of neuronal activity (Campusano, Su, Jiang, Sicaeros, & O’Dowd, 2007; Del Castillo, Muller, & Gelfand, 2020; Perrimon & Mathey-Prevot, 2007). Regardless of which system – in vivo or in vitro – we choose, analysis of complex neuron morphologies requires optical resolution below the diffraction limit of light (< 250 nm), as the nanoscopic processes of neurons (e.g., dendrites < 250 nm in diameter) are of high interest for imaging (Lichtman & Denk, 2011). Super-resolution microscopy techniques provide an excellent solution to this issue, improving the resolution beyond the diffraction-limit (Huang, Bates, & Zhuang, 2009). While many techniques have been established for super-resolution imaging of Drosophila samples (Gao et al., 2019; Jiang et al., 2018; McGorty, Kamiyama, & Huang, 2013; Schnorrenberg et al., 2016), stochastic optical reconstruction microscopy, also called STORM, is already widely available and can be used with an extensive array of fluorophores for our demands.

In this unit, we provide detailed protocols in preparing Drosophila neurons and taking STORM images of various structures in the neurons. BASIC PROTOCOL 1 sets up the establishment of primary neuronal cultures and dissection of embryos for immunostaining. For labeling the neuronal membranes genetically, we take advantage of the UAS-GAL4 system. We start with the appropriate genetic crosses to express membrane markers in particular neurons and train the adults to lay eggs in their new cages. Once we collect the embryos, we can either dissociate the embryos for making a primary neuronal culture or dissect them for in vivo analysis. Finally, we fix the samples before immunostaining, the approach for which can differ depending on the protein of interest (we discuss the difference in BASIC PROTOCOL 2). For immunostaining and imaging the samples, we must choose bright dyes that can overcome reduction of signal due to tissue thickness. In SUPPORTING PROTOCOL, we describe a dye conjugation method to antibodies for the selected far-red dye which is optimal for STORM imaging. BASIC PROTOCOL 2 describes the immunostaining approach for these samples. While in vivo labeling is a time-intensive process, primary neuronal cultures can be labeled and imaged within a couple of hours. Last, we describe our image acquisition approach in BASIC PROTOCOL 3 and our reconstruction approach in BASIC PROTOCOL 4. For STORM imaging, a 647 nm laser line is used to photo-switch the fluorophores. In this protocol, it is critical to optimize acquisition settings to obtain a high-resolution image. Finally, we describe our step-by-step approach to reconstructing the STORM data using the ThunderSTORM plugin in Fiji/ImageJ.

BASIC PROTOCOL 1

PREPARATION OF DROSOPHILA PRIMARY NEURONAL CULTURE AND EMBRYONIC FILLETS

STORM has been previously demonstrated on primary neuronal cultures from the rat cortex (Xu, Zhong, & Zhuang, 2013), which do not deem any complex sample preparation and imaging steps necessary. As opposed to these cultured cells, STORM imaging of the Drosophila nervous system is challenging. Typically, we use a whole embryo for immunofluorescence imaging due to the ease of preparation. However, the surrounding tissues and yolk outside of the nervous system provide background fluorescence. The background fluorescence subsequently degrades the detection and localization of single molecules, resulting in STORM images with fewer localizations and worse resolution. To circumvent this, we dissect the embryos in fillet preparation and remove the intestines and trachea, which contribute to the background fluorescence. Here, we describe our method for culturing and plating primary cells dispersed from Drosophila mid- to advanced-gastrula embryos and dissecting late embryonic samples for STORM imaging.

Materials

Embryo collection

• UAS-myr::GFP flies (Bloomington Drosophila Stock Center, RRID: BDSC_32197)

• elav-GAL4 flies (Bloomington Drosophila Stock Center, RRID: BDSC_8765)

• Embryo collection cages fitting 60-mm petri dishes

• Yeast paste prepared from active dry yeast by mixing with distilled water

• Grape juice agar plates

• Egg basket: Create an egg basket by cutting a conical tube horizontally and cutting out a circle from the cap; then place a 100-μm nylon mesh filter between the cap and the tube.

• Squirt bottle with distilled water

• 50% household bleach prepared at 1:1 ratio by mixing in distilled water (stored away from light)

Common materials for culturing neurons and embryo dissection

• Dissection microscope (Nikon, SMZ-U)

• Dumont #5 Forceps (Fine Science Tools, #11252-20)

• Capillary tubing with outer diameter of 1.2 mm and an inner diameter of 0.6 without filament for dissection needle and suction needle

Culturing neurons

• Parafilm (Sigma-Aldrich, #P7543)

• Task wipers (KimTech, #34155)

• 60-mm petri dishes (Corning, #430166)

• 70% ethanol

• Coverslip (No. 1.5 thickness, Fisherbrand, #12-541B)

• Imaging spacer (Electron Microscopy Sciences, #70327-20S)

• Bunsen burner

• Concanavalin-A (Con-A; Sigma Aldrich, #C5275-5MG)

• Humid chamber (or lidded Tupperware with wet tissue)

• Supplemented SFX medium (see Reagent and Solutions)

Embryo alignment and dissection

• Modeling ‘Play-Doh’ clay

• 1 mL syringe (BD, #309659)

• Double-sided tape (Scotch, # 665)

• Pre-cleaned glass microscope slides 25 × 75 × 1 mm (Fisher Scientific, #S13943) with vinyl tape (Scotch, #6143)

• Phosphate-buffered saline (PBS)

Special equipment

• Micropipette puller (Narishige PC-100)

Training Flies and Collecting Embryos

• 1In an embryo collection cage, cross the GAL4-driver flies with the UAS-reporter flies at 25°C (Figure 1B, left).

  Having a ratio of 1:5 male-to-females is sufficient for all of the females to lay eggs. A typical cage containing at least 30–50 adult flies should lay a sufficient number of eggs.

• 2Prior to the experiment, acclimate the adult flies to the egg collection cage by allowing them to lay eggs for 1–2 days by changing the yeast-streaked agar plates at least once per day. The plates should be changed until flies can synchronously lay ~500 eggs per collection period in each plate (Figure 1B, right).

  Keep the cages at 25°C with a humidity of about 65% in a 12-hr light and 12-hr dark cycle to facilitate the collection of synchronous embryos.

• 3Allow the adult flies to lay eggs on a yeast-streaked agar plate for 1 hour and collect the plate from the cage and replace with a new yeast-streaked plate.
• 4Incubate the plates with the embryos until the appropriate developmental stage. Mid-gastrula embryos aged 5–7 hours AEL (after egg laying) – or Stage 11 (Campos-Ortega, 1985) – are ideal for neuronal culture preparation. By our hands, we found that embryos aged up to 10 hours AEL (i.e., Stage 12) can be plated as well.
• 5Remove the eggshell from the embryos by adding 50% bleach to the embryo-containing plate for 5 minutes to allow access to the embryo for cell dissociation and dissection.
• 6Filter the eggs through the egg basket (Alternatively, a cell strainer can be used).
• 7Wash the filter thoroughly with distilled water so that the embryos are cleared of chorions and yeast from the plate.

  This step is especially critical for cell culture preparation to ensure there will be no contamination by yeast.

• 8Disassemble the egg basket and blot the mesh filter dry with sterilized task wipers.
• 9Transfer the embryos into a petri dish by flipping the mesh so that the embryos will be facing a clean plate and wash them into the plate set beneath the filter.
• 10Pour the water off the embryos just before collecting.

Figure 1: Equipment for sample preparation.

(A) Tube-connected suction needle. A pipette tip (blue) attached to the tube is used as the mouthpiece (arrowhead). At the other end of the tube, a dissection needle attaches to an adaptor piece (arrowhead). (B) Embryo collection cage with a mesh top and yeast-streaked plate. The yeast-streaked agar plate is used by flies to lay eggs. (C) Coverslip preparation inside a humid chamber. The double-sided adhesive imaging spacer overlays the coverslip. Cells should be plated inside the well chamber. (D) Glass slide preparation with a dissection pool for dissecting the embryos. The embryos (e.g., encircled) are aligned onto a piece of double-sided tape and submerged in saline. The dissection pool is created from vinyl tape on a clean glass slide.

In the last part of this protocol, we report two separate methods for in vitro and in situ sample preparation. In steps 11–20, we describe the preparation of primary cell cultures and through steps 21–30, we describe the preparation of dissected embryos.

Collecting the embryo cell culture

• 11Prepare the plates by laying down a strip of parafilm on the bottom of a 60-mm petri dish and covering a section of the plate wall with a damp task wiper. Ensure there is enough room to later lay down the double-sided adhesive imaging spacer (of the appropriate diameter) (Figure 1C).

  The strip of parafilm beneath should roughly be the size of coverslip/imaging spacer. This allows easier removal of the coverslip/imaging spacer from the plate later.

• 12Using forceps, sterilize the coverslip by submerging it in 70% ethanol and drying it completely over the Bunsen burner. Lay the sterilized coverslip on top of the parafilm strip inside the plate.
• 13Peel the sticker off one side of the spacer and press the adhesive side, facing down, onto the coverslip. Using forceps, firmly press around the circumference of the spacer. This prevents any leakage of ConA or culture media from the coverslip.

  You may skip this step and instead use vacuum grease to mount the sample as described in BASIC PROTOCOL 2 under “Mounting the immunofluorescence labeled samples.”

• 14Spread 10 μL of 0.5 mg/mL Con-A to produce a thin layer on the coverslip. Cover the petri dish with its lid. Place the dish in a humid chamber with humidity ~80%, or a lidded Tupperware with a wet tissue inside. Incubate for 30 minutes at room temperature.
• 15Make the supplemented SFX medium (see REAGENTS AND SOLUTIONS).
• 16Remove the Con-A from the coverslip. Add 100 μL of room temperature supplemented SFX medium on the Con-A-coated coverslip in the plate near the open flame of a Bunsen burner, which provides a sterile environment.
• 17Take a dissection needle and break off the tip until the diameter is ~60 μm or about 1/3 of the embryo’s width, which can be checked in comparison to the embryos. Connect the needle to one end of the suction tube and add a pipette tip to the other end to act as a suction source. (Figure 1A)
• 18Examine the embryos, directly from step 10, underneath the dissection microscope. Use the pipette tip attached to the tube-connected suction needle (Figure 1A) to suck and release the contents of the embryo to break the cells out of the vitelline membranes.
• 19Transfer the cells from the plate onto the center of the coverslip by gently suctioning in and out with the tube-connected suction needle; make sure any large clumps are broken up. Repeat this process until you have dissociated 20 embryos onto the center of the coverslip.

  To prevent contamination, make sure to always place the lid over the plate of your coverslip.

• 20Place the lidded plates into the humid chamber and culture the primary cells for 1–2 days at 28°C.

Aligning and dissecting embryos for in situ preparation

• 21On a glass slide, create a dissection pool using vinyl tape and place a piece of double-sided tape inside the dissection pool horizontally to align the embryos as demonstrated in Figure 1D.
• 22Collect the embryos at 13–15 hours AEL, which has the four-chambered gut morphology, (i.e., Stage 16 described by Campos-Ortega), using forceps and place them onto the tape with the dorsal side facing up.

  Here we chose to look at Stage 16 as synaptogenesis at the neuromuscular junction occurs during this stage, however, it is possible to dissect and image at older or younger stages as needed.

• 23Add PBS to prevent the embryos from desiccating.
• 24Using a micropipette puller at 12V, pull the capillary tubing to create a sharp needle with a taper of ~0.4 cm in length to make a dissection needle. Mount the dissection needle onto a 1 mL syringe filled with some modeling clay.
• 25Using the dissection needle prepared, cut through the midline of the embryo under a dissection microscope inserting the needle from the surface of posterior end pushing to the anterior end.
• 26Gently lift the embryo with the needle which should be slightly underneath the brain lobes of the embryo and pull it up-and-away from the tape.
• 27Reposition the dissected embryo onto the glass slide. Here, the embryos should adhere to the glass slide by themselves as long as the buffer is not saturated with proteins and the embryos do not have cuticles.

  Please refer to the Commentary and Troubleshooting for further details.

• 28Using the yolk and the intestines that come out as cushion for the dissection needle, place the epithelial tissues to either side on the glass. With this, there should be epithelial tissues on both sides of the ventral nerve cord and the brain.
• 29Create another tube-connected suction needle by repeating step 17. To suction the internal organs after dissection, the diameter should be around 300 nm which is about the size of the intestines. Any larger or smaller tip sizes will change the speed of suction.
• 30Clean up the sample by using a tube-connected suction needle (Figure 1A) and blowing air or aspirating to detach and remove the dorsal tracheal trunks and any remaining intestines prior to moving onto immunostaining process.

  If the saline has a lot of tissues or cells floating around, you can replace it by using a micropipette to wash the sample with new saline to ensure the sample is clear of debris.

BASIC PROTOCOL 2

IMMUNOFLUORESCENCE LABELING OF SAMPLES

For immunofluorescence labeling of samples, there are a couple of considerations necessary to obtain the best quality of STORM images. One important consideration is the fixation approach applied to the samples. As a result of insufficient fixation, membrane structures and macromolecule architectures can be broken. In order to avoid this type of artifact, different cellular targets have been optimized to be fixed with certain fixatives. For instance, glutaraldehyde (GA)-based fixation for tubulin is optimal in preserving the microtubule structures, while fixation with paraformaldehyde (PFA) alone is sufficient for preserving membrane structures. Because of its high resolution, STORM can readily uncover such disruptions. Therefore, the choice of proper fixatives is crucial. Next, the density of fluorescent labels should also be considered. When immunostaining membrane structures with a membrane marker, the low density of the marker makes a STORM image discontinuous. With this inadequate labeling density, it is not easy to evaluate whether the structures are related. Here, we describe our method for fixing, immunostaining, and mounting the neuronal culture and brain tissues.

Materials

• Drosophila primary culture on coverslip or dissected embryos on glass slide

• Phosphate-buffered saline (PBS)

• 4% Paraformaldehyde (PFA; Electron Microscopy Sciences, #15710) in PBS.

• 0.1% Glutaraldehyde (GA; Sigma Aldrich, #111-30-8) in PBS.

• PBS with 0.1% Triton-X100 (TBS)

• 0.1% TBS + 0.06% bovine serum albumin (BSA) (TBSB)

• Primary antibodies

  • 1:200 anti-GFP rabbit monoclonal antibody diluted in TBSB (ThermoFisher, #G10362, RRID: AB_2536526)
  • 1:200 DM1A anti-tubulin antibody: (Cell Signaling Technology #3873, RRID: AB_1904178)
  • 1:5 anti-HRP antibody: (JacksonImmuno, #123-005-021, RRID: AB_2338952)
• Secondary antibodies

  • 1:200 anti-rabbit IgG secondary antibody conjugated with AF647 dye (see SUPPORTING PROTOCOL for dye to antibody conjugation) diluted in TBSB

• Aspirator (Bio-Rad, Model #1651754P)

• Orbital Shaker (CB, KJ-201BD)

• Imaging medium (see Reagents and Solutions)

• Clean coverslip No. 1.5 thickness (Fisherbrand, #12-541B)

• Pre-cleaned glass microscope slides 25 × 75 × 1 mm (Fisher Scientific, #S13943)

• Vacuum grease or some other spacer to cushion between the coverslip and the tissues (e.g., Secure Seal Imaging Spacer from Electron Microscopy Sciences, #70327-20S)

• Task wipers (KimTech #34155)

• Nail polish for sealing the sample

Permeabilization and Staining

• 1Using an aspirator, remove the culture medium or saline buffer and replace with 4% PFA in PBS for membrane labeling and with 3% PFA+ 0.1% GA in PBS for microtubule labeling.
• 2Incubate on an orbital shaker to fix for 5 minutes at room temperature.
• 3Then, wash for 5 minutes each with TBS, 3 times.
• 4Block the sample for 10 minutes for cell culture (or 1 hour for dissected embryos) at room temperature using TBSB to reduce/eliminate non-specific binding of primary and secondary antibodies.
• 5Exchange the TBSB buffer with the primary antibody in TBSB. Please see the Critical Parameters section under Commentary for our discussion of advantages and disadvantages of using labeled primary antibodies and indirect labeling with secondary antibodies.

  For primary neuronal culture samples, incubate the primary antibody for at least 1 hour at room temperature.

  For dissected embryo samples, incubate the primary antibody overnight at 4°C.

• 6Before staining samples with secondary antibody, remove the primary antibody and wash with TBSB for 5 minutes each, 3 times.
• 7Dilute the secondary antibody IgG conjugated with AF647 dye to a working concentration in TBSB.
• 8Exchange the blocking buffer with the secondary antibody using an aspirator. Ensure that the samples are not dried out during this exchange process.
• 9Stain the sample by incubating at room temperature on the orbital shaker.

  For primary neuronal culture samples, incubate the secondary antibody for 1–2 hours.

  For dissected embryo samples, incubate the secondary antibody for 2 hours.

• 10Wash the sample with TBS for 5 minutes each, 3 times.
• 11Post-fix the sample using the 4% PFA in PBS for 5 minutes at room temperature to crosslink the antibodies strongly and to eliminate floating dyes, which can contribute to creating background signal.
• 12Finally, wash with TBS 3 times to remove the fixative from the sample.

Mounting the immunofluorescence labeled samples

• 13Using task wipers or an aspirator, remove excess buffer from the sample but ensure the samples do not dry out.

  In case of the in-vivo preparation, remove the tapes that are on the glass slide using forceps first.

• 14Exchange the buffer with 100 μL imaging medium.

  Make sure to use as much imaging medium as possible to prevent any large air bubbles from forming in the next step.

• 15
  Mount the coverslip onto the glass slide.

  • To mount the cells: Remove the remaining sticker from the spacer and place the glass slide on top of the spacer and make sure it is completely sealed.
    Alternatively, apply vacuum grease to each corner of the coverslip then place the glass slide on the coverslip.
  • To mount the embryos: Place the coverslip on the glass slide and push it down to minimize the space between the coverslip and the sample until it touches the CNS.
    This will allow for maximum brightness of fluorophores while imaging for the tissue sample.
• 16Remove the excess buffer that has leaked out and seal the edges of the coverslip with nail polish and image or incubate at 4°C until imaging.

  It is ideal if the sample is imaged within 3 hours after mounted in buffer under minimal oxidation conditions for optimal results.

BASIC PROTOCOL 3: SINGLE MOLECULE FLUORESCENCE IMAGING

In this section, we describe the equipment and steps required for successfully acquiring raw single-molecule data. Although we use a custom-built STORM imaging system for the acquisition (Figure 2), one can adopt any of the commercial systems currently available, e.g., N-STORM (Nikon).

Figure 2: Microscopy setup.

(A) Schematic of the single molecule localization microscope. Three lasers at Wavelengths 647nm (Coherent, OBIS LX 647-120mW), 561nm (Coherent, OBIS LS 561-50mW), and 405nm (Coherent, OBIS LX 405-100mW), are combined with Dichroics D₂ (Thorlabs DMLP605T) and D₃ (Thorlabs, DMLP425T). Laser Line filters are used to filter out spontaneous emission for the 647 laser (Semrock, LL02-657-12.5) and the 561 laser (Semrock, LL02-561-12.5). Lenses L₂ (Newport, PAC052; efl=100mm) and L₃ (Opto-Sigma, 026-1132; efl=120.1mm) expand the laser beams to a diameter (1/e²) of 0.96mm. The excitation beams are then demagnified by 60x by the tube lens, TL (Olympus, 180mm) and the objective lens, O₁ (Olympus 60x) to a beam diameter of 16 μm. The emitted light is separated from the excitation light by the dichroic D₁ (Omega Optical, XF2054, 485-555-650 TBDR). The emission from the sample is imaged on to the EMCCD Camera (Andor, Ixon-897 Life) first to 60x magnification by the objective lens O₁ and the tube lens TL and then by an additional 2.5x by lenses L₃ (120 mm) and L₄ (300 mm) for a total magnification of 150. The camera pixel size is 16 μm so the effective pixel size at the sample is 107 nm. Two emission filters (Semrock, FF01-446/523/600/677-25 and FF01-680/42) and a notch filter (Semrock, NF01-488/647) are placed before the camera to block excitation and stray light. (B) Photograph of the system. The excitation paths and emission path are traced in red, yellow, purple, and green respectively.

Materials

• Immersion oil (n=1.52) (Cargille, #16242)

• Lens paper (Thorlabs, MC-5)

Microscopy Setup

• Custom-built inverted microscope (see also Figure 2B)

• Light source

  • 405-nm excitation laser (Coherent, OBIS 405 nm LX 100 mW)
  • 561-nm excitation laser (Coherent, OBIS 561 nm LS 50 mW)
  • 647-nm excitation laser (Coherent, OBIS 647 nm LX 120 mW)
• Laser control unit

  • Arduino Uno
  • Micro-manager Arduino software (https://micro-manager.org/wiki/Arduino)
• Spectral filters

  • Excitation laser clean-up filters: 516 nm (Semrock, LL02-561-12.5)
    647 nm (Semrock, LL02-647-12.5)
  • Dichroic mirror (Omega, XF2054,485-555-650TBDR)
  • Emission filters: Multi-bandpass (Semrock, FF01-446/523/600/677-25)
    Red detection (Semrock, FF01-605/15)
    Far-red detection (Semrock, FF01-680/42)
  • Notch Filters: Red detection (Chroma, ZET561NF)
    Far-red detection (Semrock, NF01-488/64)

• 60X Oil Objective lens (Plan Apo N, Olympus, N.A. 1.42)

• EMCCD camera (Andor, iXon-897 Life)

• Image acquisition software (μManager 2.0, RRID:SCR_016865)

• Computer workstation (Intel Xeon CPU ES-1603 v4, 2.80 GHz, 16GB RAM) with a 500GB solid-state hard drive

Optimizing image acquisition for cell culture and whole mount immunofluorescence

1. At least 30 minutes prior to imaging, turn on the lasers and camera to allow components to reach a stable operating state.

2. Clean the objective with lens paper.

3. Launch image acquisition software (μManager 2.0) and let the camera temperature stabilize (−70°C for Andor iXon897).

4. Place a drop of immersion oil on the objective lens and place the glass-slide containing the sample in the sample holder with the coverslip facing downwards.

5. Set camera exposure and use a white light source to bring the sample to focus.

   Typical setting for the exposure of an EMCCD camera is 10 ms.

6. Record a single image using the 647 nm laser at an area on the sample where there are no cells. These images can be used to subtract the background later from the raw single molecule data of the respective channels.

7. Using a 647 nm laser record a wide-field image of the cell with low laser power of 1 mW and short exposure time of 10–20 ms.

8. Using the 647 nm laser, start the STORM imaging acquisition using empirically derived camera and laser settings. We derived these settings by inversely adjusting the exposure time and the laser power until the fluorophore emission events are far enough from each other (sparsely distributed) to be individually identified (Figure 3A, top panel). Our typical settings for laser power and exposure time at the beginning of acquisition were 15 ms exposure time and laser power of 30 mW. The frame transfer mode is set to off in order to reduce noise, and the EM gain is set to 500. These settings can be found in Micromanager under Devices → Device Property Browser corresponding to each connected device (i.e., camera and lasers).

   We can typically collect at least 40,000 images using the AF647 dye with these settings. The number of images to collect (i.e., count under ‘Time Points’) is set in the Micromanager acquisition menu with an interval of 0 seconds.

9. Turn on the 405 nm laser and let it illuminate the sample in continuous wave (CW) mode to facilitate regeneration of the fluorescent dye molecules.

   At the beginning of the acquisition set the 405 nm laser power to the lowest possible setting (1mW). Increase the laser power of the 405 nm laser about 5 mW every 5,000 frames as the acquisition time increases to record the maximum number of images until all dyes cease to photo-switch. The laser power might increase to 20 mW or more. The laser power is set in the Micromanager Device Property Browser and can be adjusted manually during acquisition. The two lasers are controlled by the laser control unit.

10. After the acquisition ends, the images are available as an image stack in a file in .tiff format. This file can then be opened in ImageJ or other image processing software for analysis.

Figure 3: Workflow to detect single molecules and localize the detected molecules.

(A) Illustration of background subtraction (Min) and peak detection (Max). The background intensity was estimated from an area with no emitting fluorophores (yellow box) using the Measure function in ImageJ, and the minimum background value was then subtracted from all frames. (B) Demonstration of fluorophore localization. The image shows the wavelet filtered image which is used to find the fluorophores. (C) The image identifies the emitting fluorophores. The local maximum approach with ‘8-connected neighborhoods’ is used to set parameters.

BASIC PROTOCOL 4: LOCALIZATION AND VISUALIZATION OF SINGLE MOLECULE DATA

In STORM, an algorithm is used to precisely measure the positions of single molecules from raw camera frames. The algorithm has a fundamental effect on the resolution and fidelity of the final rendered STORM image. The entire process involved in the rendering of a STORM image can be divided into five steps: detecting single molecules from raw camera frames, fitting and localizing the detected molecules, processing the positions of these molecules (e.g., drift correction, density filtering, and intensity thresholding), rendering the final image of the molecular localizations, and finally statistically analyzing these localizations. Many software packages have been developed and published to help non-expert users perform these steps efficiently. Our protocol below gives a brief description of how to choose the critical parameters for image reconstruction using ThunderSTORM (Ovesny, Krizek, Borkovec, Svindrych, & Hagen, 2014), an open-source plug-in for ImageJ. We note that the selected parameters and options described in this protocol are not the only choices. Users need to refer to the manual for a detailed guide to select the proper parameters based on their demands.

Materials

• Fiji/ImageJ software (RRID:SCR_002285)

• Software for localizing and visualizing single molecules data (e.g., ThunderSTORM, RRID:SCR_016897; use version phasor-intensity-1) (Martens, Bader, Baas, Rieger, & Hohlbein, 2018)

Localization of single molecule data

1. Drag and drop the data file collected using 647 nm laser from the file explorer to ImageJ, or select File → Open in ImageJ, browse to the data file and press Open. This will load the entire image stack into memory.

2. Open the dark frame recorded using the 647 nm laser as mentioned in step 1 in Basic Protocol 3. Using Process → Image Calculator, subtract the dark frame from the dataset to remove the background (Figure 3A).

3. Set the camera pixel size in the sample plane, conversion factor between photons and digital units, base level offset of the camera digitizer, and EM gain of the camera using Plugins → ThunderSTORM → Camera setup. For our setup, the pixel size is 107nm, the gain is 500, and the conversion factor is 4.83 photons / ADU. The conversion factor and offset are specific to the camera and camera settings and can be found in the documentation that came with your camera.

   Analog-to-digital units (ADU) are the integer values the camera uses to record intensity. The analog-to-digital converter (ADC) converts the electrons recorded in each pixel into a digital signal measured in ADUs, which is used to quantify the incident number of photons per pixel.

4. After opening the dataset, select Plugins → ThunderSTORM → Run analysis. Select the Wavelet filter (order 3, scale 2) to perform band-pass filtering on the dataset (Figure 3B).

5. Under the Approximate localization of molecules tab, select Local Maximum for Method and choose a peak intensity threshold between 0.5 to 2 times the standard deviation of the 1st wavelet level, e.g., 1*std (Wave.F1). If the threshold is set too low, some additional peaks are identified due to noise. If the threshold is set too high, then some of the true fluorophores are missed.

6. Select Phasor-based localization 2D as the method for sub-pixel localization of the single molecules with the fit radius being an integer number close to 3*sigma (Figure 3C).

   The initial size of sigma can be found by running ThunderSTORM on few images of the data sequence. A histogram of the fitted sizes of sigma (in pixels) can help to find the initial value.

7. The localized data can be visualized using averaged shifted histograms, with a magnification of 20 and 4 lateral shifts. Once the parameters for localization and visualization have been set, press OK (Figure 4A).

8. The order of post-processing steps is user-specified. However, we recommend the following order: filtering, density filtering (to remove outliers) and finally drift correction. Typically, we will use the filtering step to remove localizations from the first frames of the acquisition when too many fluorophores are emitting. Density filtering is used to remove localizations from unattached, isolated fluorophores that do not label the structure of interest (Figure 4B). Drift correction accounts for movement of the sample during the acquisition process.

9. Drift correction by cross-correlation can be performed by clicking the drift correction tab (Figure 4C). Typical settings for the parameters are 5 bins, 5x magnification, and 1.0 for trajectory smoothing. Cross-correlation images with detected peaks can be viewed by checking the “Show cross-correlations” checkbox to fine tune the parameters for different datasets. As shown in Figure 4C, successful drift correction results in an image with higher resolution and less blurring. Compare the image before drift correction (the middle panel in Figure 4C) to the image after drift correction (the right panel in Figure 4C).

10. The output of this process is the table of filtered and corrected localizations and the final super-resolution figure.

Figure 4: Visualization of single molecule data.

(A) Clicking on the visualization button on the Thunderstorm results window opens the Visualization window. Set the parameters and click OK to create the image. (B) The density filter removes unattached fluorophores by removing localizations that are not close to other localizations and are therefore not part of a structure. Here the cutoff is 8 localizations within an 8 nm radius. As these numbers increase, more localizations are removed which can start to affect continuous structures as well as unattached fluorophores. (left) Thunderstorm Density filter window and settings. (right-top) Image of microtubules before application of the density filter. (right-bottom) After application of the density filter. Notice that there are fewer isolated spots between the microtubules, and the microtubule on the right is less continuous as well. (C) Drift correction corrects for movement of the sample during data acquisition which leads to a blurring of the image. We select the cross-correlation method with the following settings: number of bins, 5; magnification, 5; trajectory smoothing factor, 1.0. (left) Thunderstorm drift correction window and plot of sample trajectory vs. frame. (middle) Image of microtubules before drift correction. (right) Image of microtubules after drift correction. Scale bar: 0.5 μm.

SUPPORTING PROTOCOL

CONJUGATION OF ANTIBODIES WITH STORM COMPATIBLE DYES

STORM images are constructed based on single-molecule imaging of photo-switchable fluorescent probes. Compared to labeling for epi-fluorescence imaging, a single fluorophore emits a weak signal, which makes its detection challenging. Additionally, tissue-induced spherical aberration and light scattering cause a further loss of fluorescence signal and thus degrade localization precision (Kamiyama & Huang, 2012). Various bright probes, including organic dyes and quantum dots, have previously been characterized to circumvent this problem. Among those probes, Alexa Fluor 647 (AF647; a far-red photo-switchable dye) is exceptionally bright, producing high-quality images of central nervous system (CNS) neurons (Dempsey, Vaughan, Chen, Bates, & Zhuang, 2011). Here, we demonstrate the conjugation of AF647 to antibodies.

Materials

• Illustra Nap^™-5 columns, (20 Columns, 0.5 mL) (Cytiva, #17-0853-01)

• 1 M NaHCO₃ prepared from powder (JT Baker, #3506-01)

• Anti-rabbit IgG (JacksonImmuno, #711-001-003, RRID: AB_2340584)

• Anti-mouse IgG (JacksonImmuno, #115-001-003, RRID: AB_2338443)

• Anti-HRP (JacksonImmuno, #123-005-021, RRID: AB_2338952)

• 1.5 mL Eppendorf tubes (for conjugation mixture and for collecting product fractions)

• Phosphate-buffered saline (PBS)

• Alexa Fluor^™ 647 NHS Ester (ThermoFisher, #A20006; or any other STORM compatible dye)

• Dimethyl sulfoxide (DMSO; Sigma Aldrich, #276855)

• Aluminum foil

• Nutator (BD Clay Adams^™ Nutator Mixer, BD Diagnostics)

• Micropipettes (Gilson, SKU #F167380)

Conjugating the primary and secondary antibody with STORM compatible dye

1. For anti-HRP conjugation in an Eppendorf tube, add:

   • 60 μL of anti-HRP
   • 20 μL of 1x PBS
   • 10 μL of 1 M NaHCO₃
   • 1 μL of AF647 dye dissolved in 10 μL DMSO
2. For secondary antibody conjugation in an Eppendorf tube, add:

   • 20 μL of secondary antibody
   • 60 μL of 1x PBS, 10 μL of 1M NaHCO₃
   • 6 μL (for anti-rabbit IgG) or 3 μL (for anti-mouse IgG) of AF647 dye dissolved in 10 μL DMSO.
     The amount of antibody or dye may not always provide the same degree of labeling conditions. These amounts may change depending on the conditions of the ingredients.

3. Cover the tube with aluminum foil to protect from light.

4. Incubate at room temperature for 15 minutes on a nutator to allow for reaction to take place.

5. During the reaction, wash the Sephadex column at 4°C with PBS 3 times to equilibrate the resin pores with PBS and drain the PBS from the column.

6. After the reaction is finished, add the reaction mixture to the column slowly with a micropipette.

7. Once the mixture has set into the resin, elute with 400 μL of PBS. In AF647 conjugation, we can see two bands forming; the lower band is the conjugated dye, and the upper band is the unconjugated dyes.

8. Collect the product in 100 μL increments by adding PBS and eluting into different tubes.

   By looking at the color, we can tell the best conjugated fraction. Typically, the most dye concentrated fractions in the middle– i.e., fractions 3 or 4– are the best conjugated fractions.

   To find the degree of labeling we can use UV/Vis absorption measurements.

   Using the molar extinction coefficients and the absorptions of dyes at 280 nm, calculate the degree of labeling. The number of dye molecules per antibody (DOL) should be between 1–2 for STORM, and in case of quantitative studies, DOL closer to 1 is ideal.

9. Store the antibodies at 4 °C for up to 1 month or aliquot the antibodies into small tubes and store them at −20°C for up to 1 year.

REAGENTS AND SOLUTIONS

Supplemented SFX Medium

Prepare the medium in the cell culture hood. When handling the medium afterwards (such as preparing plates), work near the Bunsen burner to avoid bacterial contamination. Use deionized, distilled water in all steps and recipes unless otherwise specified.

• To 9.4 mL HyClone SFX-Insect cell culture media (Cytiva, #SH30278.02), add:

  • 400 μL fetal bovine serum (R&D Systems, #S11150)
  • 100 μL Penicillin-Streptomycin (100X) (R&D Systems, #B21110)
  • 100 μL Gibco Insulin-Transferrin-Selenium (Thermo-Fisher, #41400045)
  • Aliquot into 200 μL fractions.
  • Store up to 12 months at 4°C.

Imaging medium

β-mercaptoethanol is toxic if inhaled; therefore, it is critical to keep it in a well-ventilated area and use a fume hood when making the imaging medium.

• 80 μL PBS

• 20 μL 50% Glucose (w/v)

• 1μL β-mercaptoethanol (Sigma Aldrich, #M3148)

• 1 μL Glucose oxidase with catalase: Prepared by dissolving 1 mg of glucose oxidase in 10 μL PBS (100 mg/mL, Sigma Aldrich, #G2133) and adding 2 μL of catalase (17 mg/mL, Sigma Aldrich, #C40) solution.

Typically, when preparing this buffer, we add the catalase to the glucose oxidase solution. This way, we can store the catalase aliquot of 17 mg/ml in −20°C for multiple uses in the future. This buffer should be switched out after 3 freeze-thaw cycles. Glucose oxidase with catalase has a much shorter lifespan in 4°C, which should be prepared freshly after 3–4 days.

COMMENTARY

a. Background Information

Cultured Drosophila neurons have been used for investigating the cytoskeletal dynamics in neurons, conducting comparative analysis studies with in vivo experiments, carrying genetic RNAi screening (Mohr, Bakal, & Perrimon, 2010), and identifying cell-autonomous neuronal mechanisms such as membrane compartmentalization or presynaptic differentiation (Katsuki, Ailani, Hiramoto, & Hiromi, 2009). Although such in vitro approaches play an important role in revealing cellular phenotypes, developmental and functional findings are crucial for understanding neural circuitry development. To investigate the underlying mechanisms, we use embryonic motoneurons in vivo (Furrer, Vasenkova, Kamiyama, Rosado, & Chiba, 2007; Kamiyama & Chiba, 2009; Kamiyama et al., 2015; Sharifai et al., 2014). This in vivo system is amenable to targeted genetic manipulation in a cell-type-specific manner, allowing us to study neural specification, axon and dendritic guidance, partner selection, and synaptogenesis. In furthering our understanding of synaptogenesis, the neuromuscular junction serves as an excellent model system due to the stereotypical motoneuron targeting (Keshishian, Broadie, Chiba, & Bate, 1996). For partner recognition to result in a mature synapse, the neuronal growth cone must interact with the muscle through actin-based hairlike structures found in both cells (Ritzenthaler, Suzuki, & Chiba, 2000). These fine structures, called filopodia, explore the local environments to match partner cells together (Figure 5). While some developmental time points have been identified, the nature of the filopodial interactions remains elusive because of their fine structures. However, the protocol described here can be further applied to labeling the neuromuscular junction and applying dual-color STORM, so the details of the filopodial interactions can be further investigated.

Figure 5: in vivo central nervous system.

The central nervous system is labeled with anti-HRP antibody (magenta) and a single muscle (M12) is labeled using 2702-GAL4 driver with GFP (green). The boxed region highlights the neuromuscular junction of the embryo. The right panel is a representative growth cone and muscle filopodia interaction during synaptogenesis. Once the initial contact is made, the filopodia undergo morphological changes by clustering at the synaptic site to form the synapse. Images were acquired with 10x objective (left) and 100x oil immersion objective (right). Scale bars: (left) 50 μm and (right) 5 μm. Genotype: UAS-CD4-tdGFP, 2702-GAL4 / TM3.

Here, we demonstrate STORM, which utilizes photo-switchable organic fluorescent dyes to precisely localize individual molecules. Unlike conventional light microscopy, STORM can achieve resolutions of ~20 nm laterally and ~50 nm axially (Huang et al., 2009). There are numerous examples in which STORM has been applied to biological research (Kamiyama & Huang, 2012). In particular, STORM is a valuable asset to conduct anatomical studies of filopodia. Because the filopodia are so tiny (~200 nm in width and ~5 μm in length), their architecture can be characterized through STORM. Furthermore, we believe that multicolor STORM allows us to visualize filopodia between the motoneuron and the muscle (as demonstrated by confocal microscopy; Figure 5). For this purpose, we have evaluated many organic dyes at different emissions (M.A.I., and D.K., unpublished data). Although the dyes are photo-switchable under specific buffer conditions, Alexa Fluor 647 exhibits the best photo-switching ability. As an alternative to the STORM technique, multiple groups have recently developed a new approach, called DNA-PAINT (points accumulation for imaging nanoscale topography) (Jungmann et al., 2014; Jungmann et al., 2010). DNA-PAINT is compatible with the imaging platform of standard STORM. This approach takes advantage of DNA specificity for target recognition using a docking strand of DNA conjugated to an antibody and an imager strand of DNA conjugated to a fluorescent dye. Contrary to the traditional STORM technique, the blinking effect is achieved by the transient binding of these DNA strands. We can apply different orthogonal docking sequences with unique fluorophores to the same sample, which allows the observation of multiple proteins. This approach also has the potential for multicolor imaging of Drosophila tissues.

b. Critical Parameters

Primary neuronal culture:

To set up for a healthy neuronal culture, adult flies in the mating cage should not be older than one week. This ensures healthy embryos to begin plating neurons. Plating an appropriate density of cells (~1.2×10⁵ cells/cm²; see also Figure 6) is key to axonal growth, secured adhesion to the coverslip, and reduced clustering and layering. However, the researcher should keep in mind that a cell density too low may risk the survival of the neurons.

Figure 6: Neuronal cell culture expressing GFP on the plasma membrane.

The boxed region details the fine structural components of the neurite. The components such as filopodia and lamellipodia are important in forming mature synapses with their partners. Scale bar: 15 μm. Genotype: UAS-myr::GFP; elav-GAL4.

Tissue preparation:

For embryonic dissection, selection of the developmental age is important (Inal, Banzai, & Kamiyama, 2020). Beyond the age of ~15 hours after embryo laying, the embryos will start depositing cuticles which will prevent the embryos from sticking to the glass slide. While older embryos can also be dissected, the method described here may not be compatible for those experiments. In addition, the number of embryos that can be dissected and placed onto the glass slide becomes limited as the amount of protein in the buffer will be saturated and coat the glass preventing any more embryos from sticking to the slide. When washing the embryos after dissection in the subsequent steps of the protocol, care should be taken to prevent embryos from lifting off the slide or being destroyed due to surface tension.

Fixation:

We usually fix cellular proteins with 4% paraformaldehyde. In contrast, we use the mixed concentration of paraformaldehyde and glutaraldehyde to preserve microtubule structures (Kamiyama & Huang, 2012). To maintain cells in their most natural state, we optimize a fixation approach for the structure of interest. The optimal fixative for the ideal image should be best for preserving the target structure while avoiding structural artifacts.

Immunostaining:

Since the quality of a STORM image depends on labeling density, the immunostaining needs to provide the best labeling density for taking the best image. To achieve this, antibody concentrations higher than the recommended should be used. Additionally, longer incubation times allow for more effective labeling. Therefore, if possible, the primary antibody should be incubated overnight at 4°C and secondary antibody for ~1–2 hours at room temperature. Selecting antibodies that are highly specific for the antigens will minimize nonspecific labeling and background signals. Furthermore, it may be helpful to use primary antibodies that are directly conjugated with the fluorophore to minimize linkage error which results in increased apparent size of structures (Huang et al., 2009).

Antibody conjugation:

The degree of antibody labeling should be between 1–2 for STORM imaging, and this can be measured using a spectrophotometer with the given calculation approach in the methods. The conjugation from one batch to another may not give the same DOL and even within the same batch, the labeling efficiency may vary. In this case, it is advisable to collect the conjugated mixture in smaller (~100 μL) fractions.

Mounting/Imaging buffer:

During long imaging processes the pH value of the buffer decreases as oxygen reacts with the oxygen scavenging component of the imaging buffer. This event can hinder the photo-switching abilities of the fluorophore. Therefore, the freshest buffer should be used and the oxygen concentration in the buffer should be minimized by sealing the sample which can improve the lifespan of imaging buffer. Additionally, the fluorophores may lose photon counts if the sample is far away from the cover-glass. Therefore, the mounting approach, especially for tissues, should minimize the distance between the sample and the cover-glass (Dani, Huang, Bergan, Dulac, & Zhuang, 2010). This will evidently increase the photon counts and resolution.

Acquisition:

The resolution of an image acquired using STORM is inversely proportional to the number of photons emitted by a single molecule (Huang et al., 2009). Apart from the freshness of the imaging buffer and the quality of the dye, the laser power with which the dye is excited plays a vital role in determining the number of photons emitted by the dye. Every dye has a maximum number of photons that it can emit (Dempsey et al., 2011), and the number of photons emitted by a dye is linearly related to the laser power until it reaches the maximum number of photons emitted at which point the dye is saturated and the number of photons emitted remains constant at the maximum. Therefore, to achieve the best acquisition, it is important to excite the sample with the appropriate laser power sufficient to facilitate blinking but not too high that one bleaches the dye too fast.

c. Troubleshooting

Protocol	Problem	Possible Cause	Solution
Sample preparation	Poor neuronal growth/differentiation	Suboptimal temperature incubation	Incubate cell culture at a higher temperature (e.g., 28 °C)
	Embryo does not stick to glass slide	Unclean glass slide Embryo cuticles	Use clean glass slide Use younger embryos for dissection
	Embryo is floating or destroyed	Sample is exposed to meniscus of the buffer	When exchanging buffers use large volumes for washes to ensure sample remains entirely covered to prevent drying
	Cells are dispersing or floating	Pressure from the objective lens forcing the cells to float if the buffer leaks out	Make sure the coverslip perimeter is well-coated and dried in nail polish Adjust the objective lens to make sure it does not compress the coverslip Minimize the distance between coverslip and the slide which will increase the working distance for the objective lens
	Filopodia are moving	Poor fixation	Increase the length of fixation or the concentration of the fixative
	Poor signal and/or high background	Non-specific labeling Large distance from sample to coverslip	Use highly specific antibodies Increase antibody concentration Use blocking buffer/increase blocking time Minimize distance from the coverslip by pressing the coverslip down. If it is not being imaged, the ventral nerve cord can also be removed via suctioning after fixation
	Inconsistent labeling efficiency	Variable antibody degree of labeling	Collect conjugates in smaller fractions and test degree of labeling
Acquisition	Low photon count from fluorophores	Excitation laser is not aligned perfectly through the objective lens. Reflectivity of mirrors in excitation path is low causing loss of laser power before reaching sample	Re-align the system Measure reflectivity of mirrors in the excitation pathway and change mirrors accordingly. We recommend the use of dielectric mirrors for the highest reflectivities.
	Fluorophores are not turning off/blinking	Low laser power Buffer is not fresh	Increase the laser power until blinking can be observed Make sure the enzymes (glucose oxidase and catalase) in the imaging buffer are freshly prepared Imaging buffer has an expiration after 2–3 hours
	Fluorophores are not turning on	Fluorophore is photobleached Buffer is not fresh	Alternatively use an activation laser such as UV (405 nm) and increase its laser power See above about imaging buffer.
Post-processing	Axons appear blurry once reconstructed	Lateral or optical drift	Ensure the stage is stabilized prior to acquisition Perform drift correction on reconstructed image
	Reconstructed image has too many localizations	Inaccurate intensity thresholding	Filter out unwanted molecules by changing the intensity threshold

d. Understanding Results

Reconstruction of super-resolution images acquired via STORM

We must consider for reconstruction, the parameters such as background intensity levels and the Min/Max values in selecting individual blinking events to achieve the highest resolution without compromising the quality of the reconstructed image. In BASIC PROTOCOL 4 (see also Figures 3 and 4), we show our workflow to identify the optimal parameters for our samples. These parameters can vary between setups, sample types (culture neuron vs. in vivo neuron) and even individual acquisition samples. Although there is not a one-size-fits-all approach, the variability among acquisition samples is not significantly different to require testing many different parameters each time. Therefore, the results of reconstruction should be easily reproducible among multiple samples.

STORM imaging of in vitro and in vivo neurons

We show a conventional widefield image of a neuron in culture (Figure 6) and the reconstruction of the same neuron using STORM imaging (Figure 7). With the protocol we describe here, we can image not only the membranes of the neuronal cells in culture but also the cytoskeletal architecture of them. In Figure 7A, we show images of the microtubules acquired from primary neuronal cultures. As the inset indicates, we can clearly see each microtubule bundle, compared to the inset of the widefield image. In Figure 7B and C, we show different approaches to label the neuronal membrane using anti-HRP (pan-neuronal membrane marker) or anti-GFP (expressed on the cell membrane through UAS-GAL4 system) antibodies. And in Figure 7D we demonstrate that we can also use photoconvertible fluorescent proteins such as tdEos to image the membrane structures of the neurons. Using similar labeling approaches, we also apply these tools to in vivo neuronal imaging. We demonstrate in Figure 8A, the neurons labeled with the anti-HRP antibody and also in Figure 8B, labeled with the UAS-GAL4 system and immunostained against GFP with the AF647 dye.

Figure 7: Reconstructed super-resolution images of various proteins labeled in primary neuron culture.

(A) Conventional wide-field (left) and super-resolution STORM (right) images of microtubules from primary neuronal cells, immunostained with Alexa Fluor 647. The super-resolution image is reconstructed from 60,000 acquired frames. The insets show the magnified regions of the white boxes. (B-D) Neurons immunostained with Alexa Fluor 647 (B and C) or labeled with tdEos (D). Scale bars: 2 μm. Genotypes: (A) UAS-myr::GFP; elav-GAL4, (B and C) UAS-myr::GFP; elav-GAL4, and (D) UAS-myr::tdEos; elav-GAL4.

Figure 8: Reconstructed super-resolution images of neuronal membranes in dissected embryos.

(A) Conventional wide-field (left) and super-resolution STORM (right) images of motor neuron RP5 in anti-HRP labeled embryos, immunostained with AF647. The insets show the magnified regions of the white boxes. (B) An example of 3D STORM image. 3D reconstructed super-resolution image of aCC/pCC motoneurons labeled with GFP and immunostained with AF647. The image is reconstructed from six z-slices that cover a total of 5 μm. The inset shows a hollow cross-section of the axon from the boxed region. Scale bars: (A) 2 μm, (B) 3 μm. Genotypes: (A) UAS-CD4-tdGFP, 2702-GAL4 / TM3 and (B) UAS-CD8-GFP, eve-GAL4^RN2.

While we demonstrate a limited set of examples for using these tools, other proteins can also be imaged with this method using STORM to individually localize and elucidate their cellular roles at resolutions of ~20 nm. Additionally, other photo-switchable dyes or fluorescent proteins can be used to look at cellular or molecular interactions in multi-color STORM imaging.

e. Time Considerations

Flies need to be trained for 2 days prior to primary culture and embryonic sample preparations. The culturing step for primary neurons takes an additional ~2 days for neuronal morphology to differentiate before immunostaining. On the other hand, the embryos can be collected immediately following the 2-day training period and dissected. Dissection of embryos can take as little as 5 minutes for 10 embryos for an experienced individual. A beginner may take 5 minutes or longer for dissecting a single embryo at first. Depending on the efficiency of the individual the number of embryos to be dissected can be modified especially for developmentally time sensitive experiments. After the fixation step which takes 5 minutes, permeabilization and blocking take around 1 hour and 15 minutes. This is a good point for pausing, if necessary. This is also a good time to prepare dye-conjugated antibodies which can be stored for subsequent experiments. For indirect labeling, primary antibody incubation takes place overnight and secondary antibody incubation takes 2 hours. In case of direct labeling incubation takes maximum 2 hours at room temperature. The subsequent washing steps and post-fixation should take ~25 minutes.

The image acquisition is controlled by μManager running on a Dell workstation (Intel Xeon CPU ES-1603 v4, 2.80 GHz, 16GB RAM) with a 500GB solid-state hard drive. Acquiring 40,000 raw frames with a 15ms exposure time takes about 15 minutes. On the same computer, the analysis can take up to 20 minutes depending on the number of localizations that are found and the analysis steps that are used. Opening the acquired data takes 5 minutes. The image stack is then analyzed as described in BASIC PROTOCOL 4. Running the analysis in Thunderstorm takes 3 to 4 minutes. Running the density filter take 3–4 minutes, and if drift correction is performed that will take an additional 5 minutes. The post-processing steps must typically be run a few times to optimize the parameters in order to generate an optimal final image.