Key words

1 Introduction

In cells, it is of vital importance that biomolecules and organelles are transported from one side to the other. For transport of small particles over short distances, thermal energy-driven diffusion can be sufficient, but for larger particles and transport over large distances, active, motor-driven transport is required [1,2,3]. In many cases, single cargoes are transported by teams of motor proteins that use the cytoskeleton as tracks [4]. To unravel the molecular basis of intracellular transport, we perform live-cell imaging of cargoes, motor proteins, and other factors involved. Key aspects of interest are how many motors of what type are involved in transport, how motors of the same or other (often opposite) directionality cooperate, and how transport is regulated. We use a combination of ensemble imaging—where we visualize multi-protein cargo trains carrying numerous fluorescing proteins—and single-molecule imaging, where we visualize individual proteins. Single-molecule imaging can reveal interesting dynamics that are hidden or averaged out in ensemble experiments and can thus provide a more detailed understanding of the transport mechanism.

As a model system for intracellular transport, we use intraflagellar transport (IFT) in the chemosensory cilia of the nematode Caenorhabditis elegans. C. elegans is a widely used model organism, because it is relatively easy to maintain and has a short reproduction cycle, and the knowledge regarding its genome, cell lineage, and connectome is extensive. Furthermore, C. elegans is small (~1 mm long), thin (~100 μm), and transparent, which makes it ideal for fluorescence microscopy. A subset of the neurons in C. elegans, referred to as sensory neurons, are specialized in sensing its surroundings, essential for the animal’s survival. Typically, these neurons have a long dendrite (~50 μm long) with a sensory cilium protruding out at the distal end [5] (Fig. 1). The tips of these cilia are often in contact with the environment outside of the animal and can sense various chemical, mechanical and thermal stimuli, resulting in signal transduction and neuronal activity. Here, we specifically focus on the chemosensory cilia that are membrane-enveloped structures ~8 μm long, having a diameter of ~100–300 nm. A highly ordered microtubule-based axoneme forms the core of the cilium and acts as a template for motor proteins, kinesin-2 and IFT dynein, to drive large IFT trains back and forth, along the length of the cilium. These IFT trains, carrying cargo, including receptors and ciliary components, regulate what enters and exits the cilia and how the proteins inside the cilia are distributed and are therefore crucial for the maintenance and function of cilia.

Fig. 1
A diagram of transport in C. elegans with labels head and tail and an enlarged diagram of the tail. The tail consists of axons connected to cell bodies, dendrites, cilia, and cuticles. In cilia, the I F T particles are transported from the base to the tip and back by motor proteins.

C. elegans has 60 ciliated neurons; positions of cell bodies and dendrites of amphid (head) and phasmid neurons (tail) are schematically shown (top). We focus mainly on four phasmid neurons in the tail of the worm (see Note 1), which are located in pairs, symmetrically on the left and right sides of the worm and have cilia protruding out of the opening in the cuticle (middle). Inside the cilia, IFT, driven by motor proteins with different directionality, transports ciliary components from the base to the tip and back

Full size image

In this chapter, we provide a comprehensive description of the methods we employ to perform in vivo single-molecule fluorescence microscopy in a multicellular organism, with IFT in living C. elegans as an example. First, we explain the key features of our custom-built epi-illuminated widefield fluorescence microscope. Subsequently, we describe sample preparation, including anesthetizing the nematodes and placing them on imaging pads. Finally, we describe the imaging methodology and outline the analysis approach we utilize to quantitatively investigate IFT. In this chapter, we will not address the standard methods to maintain and transform C. elegans, since these have been described elsewhere in great detail [6].

2 Materials

2.1 Anesthetizing and Mounting C. elegans

  1. 1.

    Multipurpose agarose.

  2. 2.

    M9 buffer: 5 g NaCl, 6 g Na2HPO4, 3 g KH2PO4, 1 mL 1 M MgSO4, H2O to 1 L. Sterilize by autoclaving.

  3. 3.

    Microscope slides: 76 × 26 mm.

  4. 4.

    Labeling tape.

  5. 5.

    Anesthetic: 5 mM Levamisole (tetramisole hydrochloride) in M9 (see Note 2).

  6. 6.

    Cover glass: 22 × 22 mm (we use Marienfeld, High Precision No. 1.5H, 0107052).

  7. 7.

    VaLaP: equal parts vaseline, lanolin, and paraffin wax.

  8. 8.

    C. elegans: transgenic young adults with one row of eggs (see Note 3), expressing fluorescently labeled proteins of interest (see Notes 4 and 5), maintained at 20 °C.

2.2 Microscope Setup

We use an epi-illuminated widefield fluorescence microscope, with key components as schematically illustrated in Fig. 2.

  1. 1.

    The system is built on the basis of a commercial, inverted microscope body (Nikon, Eclipse Ti), equipped with an eyepiece and bright-field imaging capabilities for searching nematodes.

  2. 2.

    Excitation light is provided by continuous-wave lasers with wavelengths close to the maximum of the absorption spectra of the fluorescent probes used. In our experiments, we mostly use 491 nm and 561 nm 50 mW lasers. Their beams are combined by a set of dichroic mirrors to form a single excitation path (Fig. 2, dichroic mirror 1).

  3. 3.

    An acousto-optic tunable filter (AOTF, AA Opto-Electronics) enables the user to select laser lines, individual or together, and tune their intensity.

  4. 4.

    Excitation light is circularly polarized with an achromatic quarter-wave plate, and homogeneous and speckle-free illumination is obtained using a rotating ground-glass diffuser.

  5. 5.

    To ensure uniform excitation intensity across the field of view, the beam diameter is widened by a set of Galilean beam expanders before it reaches the epi lens (see Note 6).

  6. 6.

    As objective lens, a Nikon, CFI Apo TIRF 100×, N.A.: 1.49 oil-immersion objective is used. Fluorescence signal is collected by the same objective.

  7. 7.

    Excitation and emission light are separated using a dichroic mirror (Semrock, Brightline dual-edge, Di01-R488/561–25×36) (Fig. 2, dichroic mirror 2).

  8. 8.

    Emission light from certain fluorophores is then separated and filtered by a dichroic long-pass filter (Fig. 2, dichroic mirror 3) and band-pass filters (Fig. 2, emission filters) covering a large part of the fluorophore emission spectra inside a two-way image splitter (Cairn Research, Optosplit II). For single-color imaging, one of the light paths is blocked.

  9. 9.

    Fluorescence images are detected using an EMCCD camera (Andor, iXon 897, DU-897E-COO-#BV).

  10. 10.

    The microscope is operated by Micro-Manager software (version 1.4, https://www.micro-manager.org).

Fig. 2
A schematic representation of the microscope. It consists of dichroic mirror 1, beam block, lambda by 4, beam expander, diffuser, epi lens, dichroic mirror w, objective lens, sample, tube lens, dichroic mirror 3, emission filter, and E M C C D.

Schematic layout of the microscope

Full size image

3 Methods

3.1 Preparing Imaging Pads

  1. 1.

    Prepare microscope slides that each have two pieces of tape, of about 5 cm in length, on top of each other, as molds in order to obtain a reproducible agarose-pad thickness of ~0.27 mm. These molds can be reused.

  2. 2.

    Place a not-taped slide between two slides with tape. Repeat for as many slides as required. Adjust a pipet to 550 μL and place it next to the slides.

  3. 3.

    Add 20 mL M9 buffer to 0.4 g of multipurpose agarose and microwave it until fully dissolved (see Note 7).

  4. 4.

    Pipet 550 μL of agarose in M9 on the middle of the first microscope slide that is positioned between two slides with tape (Fig. 3a).

  5. 5.

    Gently place a new (not-taped) microscope slide on top of the agarose, such that it spans from one slide with tape to the other (see Note 8) (Fig. 3b). Repeat this step for all the prepared slides or till you notice that the agarose starts solidifying, then move to the next step.

  6. 6.

    Carefully remove the solidified agarose that spilled out between the two microscope slides with a scalpel (see Note 9) (Fig. 3c). Repeat for all slides.

  7. 7.

    Pick up the two slides with agarose between them. Gently slide the top one from the bottom slide. The agarose should now be on the bottom slide (Fig. 3d).

  8. 8.

    Using the now detached top slide, remove the agarose that is hanging over the edges of the slide with the agarose pad (Fig. 3e). The flat part of the detached top slide can be pressed against the side with the agarose sticking out. This should result in a square agarose pad in the middle of the bottom slide (Fig. 3f).

  9. 9.

    Obtained imaging pads can be stored in a vertical slide holder in an airtight container with a moisturized Kimwipe at the bottom, for at least 2 weeks.

Fig. 3
A schematic representation. a to c. A slide with 2 tapes and 2% agarose in M 9, place perpendicular slide over, then remove extra agarose with a scalpel. Let it solidify for 1 minute. d to f. Slide the top one from the bottom slide and remove the extra agarose and the imaging pad is ready.

Schematic of imaging pads preparation (a–f)

Full size image

3.2 Mounting C. elegans

  1. 1.

    Place a coverslip on a clean microscope slide and pipet 5 μL of 5 mM (room temperature) levamisole on the center of the coverslip (Fig. 4a).

  2. 2.

    Using a dissecting stereo microscope, pick six to eight healthy young adult C. elegans and place them in the drop of levamisole solution (see Note 10) (Fig. 4b).

  3. 3.

    Once the worms have been in the levamisole solution for about 10 min and have stopped moving (see Note 11), orient them in the droplet such that they do not overlap and are not too far from the center of the coverslip.

  4. 4.

    Then gently lower a prepared microscope slide with agarose pad, with the agarose pad down, on the coverslip (see Note 12) (Fig. 4c).

  5. 5.

    Lift the top microscope slide, the coverslip will stick to the agarose pad. Now the worms are “sandwiched” between an agarose pad on a microscope slide and a coverslip (Fig. 4e). Seal the edges of the coverslip with VaLaP, connecting them to the microscope slide (Fig. 4d). This prevents agarose and worms from drying out (see Note 13).

  6. 6.

    Label your microscope slide and wait for at least 20 min for the worms to stop moving before you start imaging (see Notes 14 and 15).

Fig. 4
An illustration of imaging sample preparation. a. A coverslip with levamisole in M 9. b. Slide has 7 to 8 worms. Anasthetize for 10 minutes. c. Place a slide on the coverslip. d. Remove the top slide and seal with VaLaP. e. The worm is in between the coverslip and the agarose pad.

Imaging sample preparation steps (a–d) and schematic cross-sectional view of the sample (e)

Full size image

3.3 Imaging

  1. 1.

    Ensure that the temperature in the imaging room is steady and close to 20 °C.

  2. 2.

    Once the worms are mounted on the fluorescence microscope, check, using the ocular and bright-field imaging, whether the worms are not moving. Focus and position the region of interest of the worms into the approximate field of view of the camera (see Note 16).

  3. 3.

    Switch from bright-field to fluorescence imaging and bring the region of interest carefully into focus (see Notes 17 and 18).

  4. 4.

    Choose the correct imaging settings (see Note 19) and start recording.

  5. 5.

    Photobleach the sample up to a point when single fluorescence spots can be clearly distinguished (see Note 20).

  6. 6.

    Continue imaging until all fluorescent proteins are bleached or until you have collected enough data (see Note 21).

  7. 7.

    If you aim to image two different fluorescent proteins at the same time, reaching the single-molecule regime for both of them might be tricky due to different photostability and overlap in the excitation spectra. The solution that works well in our hands is using alternating excitation, described in detail in [7]. However, photobleaching is a stochastic process, so the probability that the same protein complex will have two different labels emitting at the same time is very low. Alternatively, one can image one fluorescent protein (the more photostable) in bulk regime and the other in single molecule (see Note 22)—in this way colocalization events occur way more frequently. Additionally, for a precise dual-color overlay, it is advisable to perform calibration using multicolor fluorescent beads (see Note 23).

3.4 Data Analysis

Here, we will describe how we analyze the single-molecule data acquired for IFT components in the phasmid cilia of C. elegans. In general, analysis methods can be divided into two: using kymographs (Fig. 5, left panel) or using single-particle tracking (Fig. 5, right panel). Kymographs are space-time plots that display intensity along a chosen spline across time. They allow visualizing particle dynamics both in bulk and single-molecule regimes, and it is possible to extract parameters such as velocity, frequency, and intensity of tracks. However, in kymograph analysis, the spatial information is reduced to 1D, while the initial images are 2D (see Note 24). Single-particle tracking allows to overcome this limitation and enables more detailed analysis of the protein dynamics but is generally more time-consuming and requires data with a higher signal-to-noise ratio.

Fig. 5
2 illustrations of Kymograph analysis and Single-particle tracking. A. The generation and analysis are done by processing 3 steps in kymograph clear and detection and analysis trajectories in kymograph direct. B. It is detected, then selected a spline and converted to trajectories.

Image analysis pipeline. Left panel schematically illustrates steps involved in the generation and analysis of a kymograph using KymographClear and KymographDirect. Right panel schematically illustrates single-particle tracking procedure, converting XY trajectory coordinates to parallel and perpendicular displacement along the cilia, and a few example of the results that can be derived using such analysis: (i) determining the distribution of turnaround events along the ciliary length for different IFT components [4], (ii) distinguishing directed transport from diffusive and subdiffusive motion [8], (iii) obtaining super-resolution images from single-particle localizations and revealing ciliary ultrastructure [9]

Full size image

3.4.1 Kymograph Analysis

We use the ImageJ plugin KymographClear to generate kymographs and a stand-alone LabView-based software KymographDirect for subsequent analysis of obtained kymographs. Both of these software packages are open-source and can be found on https://www.nat.vu.nl/~erwinp/downloads.html.

  1. 1.

    First, inspect the image stack visually and find parts of the movie where the worm is lying still and cilia are in focus (see Note 25).

  2. 2.

    Since generating a kymograph with KymographClear takes quite long, it is advisable to create separate folders with 150–400 frames of the high-quality parts of the movie (see Note 26). Note that in its current version, KymographClear only works with folders with separate images and not with images stacks.

  3. 3.

    Make an average projection of the stack to see the shape of the cilia and use segmented line tool to draw a spline along which you want to track the motion.

  4. 4.

    Choose the spline thickness and make a kymograph. For each frame, the intensity along the spline will be averaged for the selected thickness. A thicker spline ensures that all the particles are covered but at the same time decreases the signal-to-background ratio, since the intensity of the signal-containing pixels will be averaged with the “empty” ones. KymographClear will create five separate kymographs and automatically save them in the folder containing your movie: a raw kymograph (Fig. 6c), kymographs with Fourier-filtered forward, backward motion, static signal, and a color-coded overlay of Fourier-filtered kymographs.

  5. 5.

    For the subsequent analysis of kymographs generated by KymographClear, we use the stand-alone program KymographDirect. This software detects tracks of chosen direction (Fig. 6d), calculates velocity and intensity over time and position and allows exporting data to Microsoft Excel or other software. A more detailed description of KymographClear and KymographDirect can be found in reference [10].

Fig. 6
An illustration of the tail region of C. elegans. a. A microscopic image of Cilia. b. A microscopic image of the bases and tips of the tail region. c to e illustrate the trajectories with the forward and backward track of the single molecule.

(a) Tail region of C. elegans expressing eGFP-labeled tubulin isotype TBB-4. This protein allows visualizing, from left to right: axons, cell bodies, dendrites, and cilia of the phasmid neurons. (b–e) Single-molecule data analysis in KymographClear, KymographDirect and FIESTA: (b) Single-molecule trajectories of eGFP-labeled IFT dynein in the cilia, overlayed with a maximum projection of the image stack. Trajectories from two different cilia are colored cyan and magenta. (c) A kymograph along the bottom cilium shown on (b) made with KymographClear plugin, using line width 5 pixels. (d) Forward (red) and backward (green)-directed tracks were detected with KymographDirect. (e) Single-molecule trajectories shown on (b) overlayed with a kymograph along the bottom cilium. Note that in the distal region two cilia become close to each other, and trajectories occurring in the other cilium (cyan) become visible

Full size image

3.4.2 Single-Particle Tracking

  1. 1.

    For tracking single molecules and connecting their trajectories, we use a MATLAB-based application, FIESTA [11]. It is a user-friendly open-source software that determines locations of single fluorescent particles by approximating their intensity profiles by a 2D-Gaussian function followed by linking fitted positions from individual frames into tracks. The tracks can be displayed in an overlay with the original movie, time projections, or a kymograph and can be manually edited, merged, or deleted (Fig. 6b, e). The output data is stored as MATLAB tables containing position and time information for each individual particle. The algorithm allows tuning various parameters to achieve the best tracking efficiency, such as intensity threshold, velocity, length of the trajectories, etc. The program also works with two-channel movies. Localization precision that can be achieved by this algorithm is 2 nm [11]; however, it largely depends on the acquisition time, signal-to-noise ratio, and motility of the particles [12].

  2. 2.

    The next step is to convert the x-y coordinates to cilium-based coordinates, to enable pooling and comparing the data from multiple cilia and worms. For this purpose, we employ a custom-written MATLAB script, which allows to manually draw a spline on the stack average projection or the detected trajectories and select the reference point on this spline. Then the trajectories are converted to the new coordinates, and motion parallel to the cilium is distinguished from perpendicular.

    Further, one can extract various parameters such as velocity, starting positions of directed tracks, positions and durations of turnaround events, localizations of molecules along and perpendicular to the cilium, etc. To determine whether a molecule is moving in a directed manner, diffusing or bound to a structure, we use an MSD-based script developed in our lab [7, 8].

4 Notes

  1. 1.

    For single-molecule imaging, phasmid chemosensory cilia are the most suitable to image, for several reasons: (i) the ciliary architecture is simple, with only two phasmid cilia grouped together (in comparison, amphid cilia is a compact bundle of multiple cilia), (ii) the cilia often tend to lay in a single imaging plane, and (iii) there is significantly lower background autofluorescence in the nematode tail.

  2. 2.

    Choosing the right anesthetic is vital for the success of the experiment. One needs to consider the duration of the experiment and the process being investigated. The key is to obtain good-quality data while minimally disturbing the biological system. For example, sodium azide, a commonly used anesthetic in C. elegans studies, inhibits adenosine triphosphate (ATP) synthase and cytochrome c oxidase, both essential for many cellular processes [13,14,15]. Sodium azide cannot be used in our case, since active transport critically depends on ATP production. Another drug, levamisole, is an anesthetic that immobilizes worms by opening a subgroup of AChR channels that results in muscle contraction leading to paralysis [16], but has no impact on intracellular transport in the nervous system. However, since the muscle activity is inhibited, worms cannot feed during imaging, which limits the total acquisition window to 2–2.5 h. Additionally to the abovementioned chemical anesthetics, worms can be physically immobilized, for example, by being embedded in a hydrogel [17, 18], restricted with polystyrene beads [19], or placed in microfluidic chips designed in various ways [20,21,22]. In our hands, levamisole works efficiently enough, unless the worms are additionally stimulated while imaging.

  3. 3.

    To minimize the physiological differences between individual worms, one should try to be as consistent as possible to pick worms of similar age (we use young adults with no more than about eight eggs) that are well-fed. Also, care should be taken to avoid contamination appearing on the plates.

  4. 4.

    Most of the endogenously labeled strains we use in our lab were generated by MosSCI insertions [23], however, the recently developed CRISPR/Cas9 system [24] provides a quicker and more straightforward way to label proteins endogenously. Labeling proteins that are expressed in a subgroup of cells, for instance, cilia-specific proteins for imaging IFT, allows us to minimize out-of-focus background autofluorescence. It is also possible to express the labeled protein of interest under a cell-specific promoter, allowing us to image, for instance, IFT in a specific cilium in the amphid channel (in a bundle of nine cilia). However, using a non-native promoter could lead to a different expression level, potentially resulting in undesired effects.

  5. 5.

    The specifications vary widely among the rich color palette of fluorescent proteins. For the detection of single molecules, fluorescent proteins with a high quantum yield, brightness, and photostability are advisable. In our studies, we typically use eGFP to observe single-molecule trajectories, but other fluorophores with better fluorescent characteristics, like mNeonGreen, StayGold, etc., will likely improve the quality of imaging. Even though mCherry appears to have less suitable photo properties in comparison to eGFP, it can be suitable for single-molecule imaging in some cases: (i) The autofluorescence from the body of the worm is much lower in the red wavelength. (ii) While mCherry bleaches faster than eGFP in general, there is a sub-population of mCherry-labeled proteins that remain active for significantly long periods (photobleaching is not described by a single exponential decay), resulting in several long single-molecule trajectories. If the objective of the study is to obtain longer single-molecule trajectories, albeit, acquired at a lower framerate (since quantum yield and brightness is lower, frame time has to be longer to acquire more photons per frame), mCherry might be more suitable than eGFP.

  6. 6.

    The laser beam has a Gaussian intensity profile, so the fluorescence excitation is the highest at the center of the beam and decreases toward the edges. In the widefield microscope, a parallel excitation beam is focused by an epi lens on the back focal plane of the objective and becomes parallel again upon passing through the objective lens (Fig. 2). The change in the beam diameter is determined by the ratio of focal distances of the objective and epi lenses: \( \frac{D\ \left(\mathrm{after}\ \mathrm{obj}\right)}{D\ \left(\mathrm{before}\ \mathrm{epi}\right)}=\frac{F\ \left(\mathrm{obj}\right)}{F\ \left(\mathrm{epi}\right)} \). So, knowing the camera pixel size, the magnification to be used, and the objective focal length, one can estimate how large the beam needs to be before the epi lens, and which lens to use. To achieve a more uniform excitation across the camera field of view (FOV), we aim to have the beam diameter such that only the central part of the beam having one standard deviation (1σ width) of the intensity distribution excites the area visible on the camera chip.

  7. 7.

    We usually dissolve agarose in a 50 mL upright standing centrifuge tube. Place the tube in the microwave, place the lid on top, but do not fasten tightly: a fully closed lid will result in too much pressure built up in the tube, and an open lid will allow water to evaporate. At our lowest (350 W) microwave setting, the M9 quickly boils over once it is warm. At this point, the agarose is not fully dissolved yet. Frequent shaking is advised in order to prevent burning the agarose in the bottom of the tube. Dissolving the agarose works best by heating in the microwave oven while constantly watching the tube, quickly turning it off once it starts to boil over, shaking the tube, and repeating till agarose in completely dissolved.

  8. 8.

    Once the agarose is melted and taken out of the microwave, it will solidify quickly. Steps 4 and 5 are therefore best performed quickly. When pipetting agarose, avoid creating air bubbles. Place the top microscope slide on one of the taped microscope slides first, without it touching the agarose. Gently lower it over the agarose, wait till one can see condensation on the top microscope slide, and then press it down on both microscope slides with tape. This should not be done with too much force, as it will make the agarose pads too thin.

  9. 9.

    Try to prevent moving the bottom and top slide too much by gently pressing them down while one removes the excessive agarose.

  10. 10.

    The bacteria in the agarose plate, on which the worms feed on, are highly autofluorescent and, when passed on to the imaging pads, can reduce the quality of the single-molecule imaging in the worms. It is hence best to pick worms from outside the bacterial lawn, or transfer them there and let them crawl a bit (to get rid of the bacteria sticking onto their body) before drop** them into the anesthetic solution. It is recommended to put not more than seven to eight worms in one droplet to avoid them being too close to each other.

  11. 11.

    Ensure that worms lie still before proceeding to the next step. Sealing the sample too early often causes worms to end up with sharply bent tails because they keep moving while being partly restricted. In order to prevent the levamisole drop from evaporating, it is advisable to put a cell culture plate lid on top of the droplet. Further, while the worms are in the anesthetic solution, the imaging pad (stored in a moist airtight container) should be allowed to dry a bit (~10 min). It is crucial for the agarose bed in the imaging pads to have an optimal level of moisture. If they are too dry, they tend to absorb too much moisture from the body of the worm. On the other hand, too moist pads will cause worms move during imaging.

  12. 12.

    In our hands, this works best while holding the microscope slide with the agarose pad at a 90° angle with respect to the one with the coverslip with worms on it (Fig. 3c). Make sure that agarose pad covers the coverslip completely.

  13. 13.

    Melt the VaLaP by heating to approximately 75 °C before one sedates the worms. Use a cotton swab to apply the VaLaP on the corners of the coverslip to prevent it from moving and then seal the agarose edges between the coverslip and microscope slide. Attempt to apply as little VaLaP as possible on the coverslip, as VaLaP can dissolve in immersion oil, causing a lower image quality.

  14. 14.

    During first 20–30 min after immobilizing, worms seem to keep on twitching, which makes long-duration single-molecule imaging very challenging.

  15. 15.

    It is often useful to make a schematic drawing of how the worms are oriented in the imaging pad. Once one worm is localized under the microscope, it becomes easier to find the rest of the worms with reference to the first worm, by looking into the scheme.

  16. 16.

    Let the sample equilibrate to the objective temperature for at least 10 min before acquiring. In the meanwhile, one can search for the regions of interest, save their positions using Micro-Manager, adjust the imaging settings, and name the files appropriately.

  17. 17.

    Since the diameter of an adult C. elegans is around 100 μm, the location of proteins of interest with respect to the coverslip is a factor to keep in mind. The worms are randomly oriented in the imaging sample, which implies that in some cases, the cilia may not lie in a single focal plane, and in some cases, the light has to pass through the entire thickness of the worm body, which results in excess background autofluorescence and aberrations. Such worms should be ignored and not imaged.

  18. 18.

    We advise to use the Nikon Perfect Focus System (PFS) to ensure that the region of interest stays in focus all the time because once in the single molecule regime it is hard to maintain the focus manually. PFS keeps the distance between the coverslip and the objective constant, compensating for the temperature-caused expansion effects. However, it is not useful if the worm itself is moving—in that case, adjust the focus manually, or skip this worm and move to the next one, if the movement persists.

  19. 19.

    Some key imaging parameters are magnification, exposure time, and excitation laser intensity. The optimal magnification for diffraction-limited widefield microscopy would be such that 1 pixel of the camera chip equals 80–100 nm. With smaller pixel size, the signal will be spread over more pixels, thereby decreasing signal-to-noise ratio without adding any additional information (oversampling), and with larger pixels, the particles close to each other will not be resolved (undersampling). Exposure time depends on the process you aim to capture. For example, directed transport in the cilia moves with an average velocity of 1 μm/s, so imaging at 100–250 ms per frame enables capturing such movement. On contrast, freely diffusing proteins require faster imaging, around 15–50 ms per frame, depending on the size of the protein. EMCCD camera’s speed is limited by the readout time; thus to enable faster imaging, one needs to crop the field of view before acquiring. Faster imaging also requires higher excitation intensity, to ensure similar signal brightness, which in turn leads to faster photobleaching. One needs to find the right balance between the excitation intensity, exposure time, and duration of imaging, which will be different for different proteins, depending on their fluorescent label, level of expression, subcellular localization, and dynamics. Besides that, keep in mind that high intensity light might have both physical (local heating) as well as biological effects. It has been shown that short-wavelength light, especially in the UV spectrum, activates certain neurons in C. elegans and triggers avoidance behavior [25,26,27].

  20. 20.

    Spatially distinguishable individual FPs and invariant intensity between them is a good indication for single molecules. Comparing the intensity of single, purified FPs on glass with the intensity of the same type in vivo, and looking at the bleaching steps of the FPs in vivo can also help determining whether one is observing single molecules.

  21. 21.

    An elegant trick we use to prolong the imaging duration is exposing only the region of interest (~10 μm in case of phasmid cilia) with the excitation light. Fluorescent proteins outside the region of interest remain unbleached and can be readily visualized as soon as they enter the region, thereby allowing us to perform single-molecule imaging in the same worm for 1–2 h if needed. To achieve this, we insert an iris diaphragm in the excitation path. By closing and opening it, we can manually control the size of the excitation beam [28].

  22. 22.

    In case of two-color imaging using eGFP and mCherry, mCherry will usually reach the single-molecule regime first since it fluoresces (and gets photobleached) not only when exciting with 561 nm light but also with the 491 nm light used to excite eGFP. In the meantime, eGFP can stay in the bulk regime (when excited with low intensity 491 nm light) or reach single-molecule regime as well (when excited with high intensity 491 nm light).

  23. 23.

    During dual-color imaging, the emission light is separated using a spectral beam splitter (we use Cairn Research, Optosplit II) installed between the microscope body and the camera. The signals from two different fluorescent proteins are recorded simultaneously on the two halves of the camera chip. Though the positions of the two images can be manually adjusted, it is not possible to achieve a precise overlay. This problem can be solved by tracking multicolor fluorescent beads that emit in both channels and creating an offset model aligning the channels that can be then applied to the actual imaging data. For such calibration procedure, we use Fiji plugin Descriptor-based registration (2d/3d) or FIESTA, an open-source MATLAB-based tracking software [11].

  24. 24.

    It is important to keep in mind that the actual protein dynamics take place in 3D space, with the images acquired using single-molecule localization microscopy, essentially providing us a 2D projection of the 3D motion of molecules, which implies that some information is lost. It should be possible to extend the described imaging setup for 3D imaging, by implementing one of the many methods developed in recent years [29].

  25. 25.

    In some cases, drift in the X-Y plane can be corrected, for example, using the Image Stabilizer plugin in ImageJ [30] or the Mesmerize application [31]. However, these algorithms require detection of clearly recognizable bright structures in each frame, and so far, we have not managed to successfully use it for single-molecule data, in our system.

  26. 26.

    For the initial inspection of the movie and finding the best parts, we use the Fiji plugin Multi Kymograph. It is quick and robust, however, kymographs obtained are noisier than those made using KymographClear, which employs subpixel interpolation.