RNA Detection pp 143-162 | Cite as

Detection and Automated Analysis of Single Transcripts at Subcellular Resolution in Zebrafish Embryos

  • L. Carine Stapel
  • Coleman Broaddus
  • Nadine L. Vastenhouw
Protocol
Part of the Methods in Molecular Biology book series (MIMB, volume 1649)

Abstract

Single molecule fluorescence in situ hybridization (smFISH) is a method to visualize single mRNA molecules. When combined with cellular and nuclear segmentation, transcripts can be assigned to different cellular compartments resulting in quantitative information on transcript levels at subcellular resolution. The use of smFISH in zebrafish has been limited by the lack of protocols and an automated image analysis pipeline for samples of multicellular organisms. Here we present a protocol for smFISH on zebrafish cryosections. The protocol includes a method to obtain high-quality sections of zebrafish embryos, an smFISH protocol optimized for zebrafish cryosections, and a user-friendly, automated analysis pipeline for cell segmentation and transcript detection. The software is freely available and can be used to analyze sections of any multicellular organism.

Key words

smFISH Zebrafish Cryosections Automated cell segmentation Transcript detection 

1 Introduction

An important feature of multicellular organisms is their large variety of cell types. Each cell type is characterized by a specific gene expression profile which provides information about the function of the cell. Spatial information on gene expression is often obtained by in situ hybridization [1, 2]. However, this technique provides limited cellular resolution and is not quantitative due to nonlinear signal amplification [1, 2]. Quantitative information on gene expression is often obtained by qPCR or RNA-sequencing but these techniques only detect highly abundant transcripts when single cells are analyzed, and precise spatial information is lost [3, 4, 5].

Single molecule fluorescence in situ hybridization (smFISH) is a method to visualize single transcripts at subcellular resolution in their original tissue context [6, 7, 8, 9, 10, 11]. Stellaris smFISH is a popular method because it is straightforward and affordable, and its quantitative nature has been thoroughly tested [7]. The method makes use of up to 48 individually labeled 20 nucleotide-long oligonucleotides that hybridize to their target RNA [7]. The accumulation of a large number of probes on a single transcript produces a diffraction limited signal that can easily be distinguished from background [7]. However, smFISH gives the best results when used on thin samples like single cells or tissue sections, and no protocols were available for zebrafish yet. Furthermore, available analysis pipelines for quantification of transcripts at cellular resolution in multicellular organisms relied on manual cell segmentation which is very labor-intensive and time-consuming [8, 12, 13, 14].

We previously developed a protocol for smFISH on zebrafish cryosections as well as an automated analysis pipeline for transcript detection and cell segmentation (Fig. 1) [15]. As input samples for our smFISH protocol, we use cryosections of zebrafish embryos embedded in OCT. Although sectioning at early zebrafish stages is difficult because cells are large and embryos fragile, our protocol generates high quality sections at a broad range of developmental stages. We then adapted a protocol for Stellaris smFISH on tissue sections [8] for use on these zebrafish sections (Fig. 1a). With the resulting protocol, even very low transcript levels can be detected at high specificity and sensitivity. To increase image analysis speed and throughput, we developed a freely available analysis pipeline for automated transcript detection and semiautomated cell segmentation. This pipeline can be applied to sections of zebrafish and other multicellular organisms (Fig. 1b) [15]. The transcript analysis pipeline enables quantification of transcript levels, as well as quantification of the number of active transcription sites (transcription foci), and the number of transcripts per focus (Fig. 1b6). Nuclear segmentation is integrated in the transcript analysis pipeline to be able to assign transcripts to nuclei or cytoplasm. The membrane segmentation pipeline we developed consists of three parts. First, a random forest pipeline in KNIME is used to predict for each pixel whether it is part of the membrane, a membrane intersection point (vertex), or background (Fig. 1b2). Then, the PathFinder plugin in Fiji uses these predictions to generate a cell mask (Fig. 1b3). Finally, the Fiji Cell annotation tool can be used to correct small errors in the segmentation, and to group cells according to cell type (Fig. 1b4). The resulting cell mask can be used in combination with the transcript analysis pipeline in Fiji to assign transcripts to individual cells and nuclei (Fig. 1b6).
Fig. 1

Protocol and analysis pipeline for smFISH in zebrafish. (a) Overview of the smFISH method on sections of zebrafish embryos. (b) Membrane segmentation and transcript detection pipeline. Scale bar: 10 μm. Membrane staining (b1) is used to calculate the probability that pixels belong to membrane (green), or membrane intersection points (vertices, magenta) (b2). Paths are traced from vertices over the membranes to generate a cell segmentation (b3). Small errors in the segmentation (asterisk and arrowhead in b3) are manually corrected to finalize the cell mask (b4). smFISH signal (b5) is used to identify single transcripts (magenta) and transcription foci (white) and combined with the cell (green) and nuclear (blue) segmentations (b6). Images are maximum projections of 17 z-slices spaced by 0.3 μm

Here, we describe (1) a method for high-quality cryosectioning of zebrafish embryos, (2) an optimized smFISH protocol for use on zebrafish cryosections, and (3) a pipeline for (semi)automated transcript detection and cell segmentation that can be applied to smFISH samples of any multicellular organism.

2 Materials

2.1 Embryo Embedding

  1. 1.

    Zebrafish embryos.

     
  2. 2.

    Embryo medium: 5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4 in deionized water.

     
  3. 3.

    PBS: 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.47 mM KH2PO4 in nuclease-free water. Adjust pH to 7.4 with HCl.

     
  4. 4.

    PBT: 0.1 v/v % Tween 20 in PBS.

     
  5. 5.

    Fixative: 4% formaldehyde in PBT.

     
  6. 6.

    Cryoprotection solution: 30% sucrose in PBS.

     
  7. 7.
    Tissue-Tek® OCT compound (optimal cutting temperature compound) (Fig. 2a).
    Fig. 2

    Embedding of zebrafish embryos for cryosectioning. (a) Samples are equilibrated in OCT in a 6-well staining dish and embedded in the cap of an Eppendorf tube under a dissecting scope. (b) Embryos are oriented laterally with their animal caps facing in the same direction to ensure optimal section quality. The red arrow on the Eppendorf cap indicates their orientation so that the cryo-block can be mounted on the cryostat with the animal caps facing toward the blade. (c) A beaker with isopentane is cooled to −80 °C in a bucket with liquid nitrogen. A sieve may be used to hold the beaker. (d) The cryo-block is rapidly frozen in the precooled isopentane

     
  8. 8.

    Cryo incubation solution: prepare a solution of 30% sucrose in a 50/50 mixture of PBS with OCT.

     
  9. 9.

    Isopentane.

     
  10. 10.

    Petri dish, 100 mm diameter.

     
  11. 11.

    Caps of Eppendorf tubes (seeNote1).

     
  12. 12.

    Pipet pump for glass pipets to transfer dechorionated embryos (e.g., The Pipet Pump, Bel-Art Scienceware) (Fig. 2a).

     
  13. 13.

    Glass pipet.

     
  14. 14.

    6-well staining plate (Fig. 2a).

     
  15. 15.

    Embryo manipulator (e.g., nickel-plated pin holder with curved pin, Fine Science Tools).

     
  16. 16.

    Large liquid nitrogen container (Fig. 2c).

     
  17. 17.

    Large sieve (Fig. 2c).

     
  18. 18.

    250 mL beaker.

     
  19. 19.

    Sharp forceps.

     
  20. 20.

    Blunt-end forceps.

     
  21. 21.

    Low-temperature thermometer (to −100 °C).

     
  22. 22.

    Incubator set to 28 °C.

     
  23. 23.

    Stereomicroscope.

     

2.2 Cryosectioning

  1. 1.

    Glass cleaning detergent (e.g., Mucasol).

     
  2. 2.

    Milli-Q water.

     
  3. 3.

    100% ethanol.

     
  4. 4.

    0.1 w/v % poly-l-lysine (MW 150,000–300,000 Da) in Milli-Q water.

     
  5. 5.

    22 × 22 mm selected #1.5 coverslips (0.17+/− 0.0005 mm) (seeNote2).

     
  6. 6.
    Coverslip holder (e.g., XL Wash-N-Dry coverslip rack, Diversified Biotech) (Fig. 3b).
    Fig. 3

    Sample preparation for smFISH. (a) The cryo-block is mounted on the cryostat with the animal caps facing toward the blade of the cryostat (red arrow). (b) Coverslips are placed in a coverslip holder and are coated with poly-l-lysine in a slide staining dish to increase adhesion of sections to the coverslips. (c) Multiple sections can be placed on a single coverslip. (d) Cryosections on a coverslip. Note the orientation. While the yolk is damaged, the animal caps are intact. (e) Coverslips with sections are placed section-up in a 6-well plate for smFISH washes. (f) Coverslips are placed section-down on a drop of hybridization mix in a Parafilm-coated petri dish for probe hybridization overnight

     
  7. 7.

    Slide staining dish or beaker to fit coverslip holder (Fig. 3b).

     
  8. 8.

    Anti-roll plate for cryostat.

     
  9. 9.

    Cryostat blade.

     
  10. 10.

    Specimen stage for cryostat.

     
  11. 11.

    Thick and thin brush.

     
  12. 12.

    6-well plate.

     
  13. 13.

    Parafilm.

     
  14. 14.

    Sonication bath.

     
  15. 15.

    Cryostat (e.g., Microm HM560).

     

2.3 smFISH (Incl. Imaging)

  1. 1.

    70% ice-cold ethanol in Milli-Q water (store at −20 °C).

     
  2. 2.

    2× SSC diluted from 20× SSC (commercial, RNase free).

     
  3. 3.

    5 μg/mL proteinase K in 2× SSC.

     
  4. 4.

    smFISH wash buffer: 10% formamide and 2× SSC in nuclease-free water (seeNote3).

     
  5. 5.

    smFISH hybridization buffer: 10 w/v % dextran sulfate, 10 v/v % formamide, 1 mg/mL E. coli tRNA, 2× SSC, 0.02 w/v % BSA, and 2 mM Vanadyl-ribonucleoside complex in nuclease-free water (seeNote3).

     
  6. 6.

    Custom Stellaris smFISH probes, diluted to 25 μM in Milli-Q water (seeNote4).

     
  7. 7.

    1 mg/mL DAPI.

     
  8. 8.

    Phalloidin–Alexa 488 (seeNote5).

     
  9. 9.

    GLOX buffer: 10 mM Tris, pH 7.5, 2× SSC, and 0.4 w/v % glucose (in nuclease-free water).

     
  10. 10.

    GLOX mounting medium: GLOX buffer, 47 μg/mL glucose oxidase, and 0.58 mg/mL catalase suspension.

     
  11. 11.

    Microscope slides.

     
  12. 12.

    Nail polish.

     
  13. 13.

    Epifluorescence microscope with suitable filter sets, a high NA objective, and a sensitive camera (seeNote6).

     

2.4 Data Analysis

For additional information on how to install the plugins, you may consult the documentation associated with Stapel et al. 2016 [15].
  1. 1.
    Fiji (http://fiji.sc/Fiji)
    1. (a)

      Activate update sites “MS-ECS-2D” and “3D ImageJ Suite”.

       
     
  2. 2.

    KNIME

    Background information and more details on how to install KNIME for the applications described in this chapter can be found at http://tinyurl.com/KNIME-MS-ECS. Briefly:
    1. (a)

      In the KNIME.ini file in your KNIME installation folder: set Xmx to –Xmx4g or –Xmx6g and –XX:MaxPermSize to –XX:MaxPermSize = 512 m.

       
    2. (b)

      Ensure your KNIME update sites include “KNIME Analytics Platform Update Site” and “Stable Community Contributions”.

       
    3. (c)

      Add update site “MPI-CBG” with location ‘https://community.knime.org/download/de.mpicbg.knime.ip.update’

       
    4. (d)
      Install the following tools (seeNote7):
      • KNIME External Tool Support (Labs).

      • KNIME Image Processing.

      • KNIME Image Processing—Supervised Image Segmentation.

      • KNIME Quick Forms.

      • KNIME Quick Forms (legacy).

      • KNIME Virtual Nodes.

       
    5. (e)

      Install the KNIME MS-ECS-2D workflow via “tinyurl.com/KNIME-MS-ECS”. On this page you can also find more background information on how to install KNIME.

       
    6. (f)

      The pipeline is available in versions for Mac or Linux.

       
     

3 Methods

Carry out all washes and rinses in 1 mL unless otherwise specified.

Use RNase-free solutions and wear gloves at all times to prevent RNase contamination.

3.1 Embryo Embedding

  1. 1.

    Collect embryos in the chorion in a petri dish and let them develop in embryo medium to the desired stage (seeNote8).

     
  2. 2.

    Fix embryos in a round bottom 2 mL tube in 4% formaldehyde in PBT overnight at 4 °C or for 4 h at RT.

     
  3. 3.

    Rinse embryos in PBT and manually dechorionate them in PBT under a dissecting scope using sharp forceps. Put embryos back in a 2 mL tube with PBT.

     
  4. 4.

    Equilibrate embryos in 30% sucrose in PBS until they sink (seeNote9).

     
  5. 5.

    Rinse embryos twice in fresh 30% sucrose in PBS (seeNote10).

     
  6. 6.

    Replace with fresh 30% sucrose in cryo incubation solution. Make sure that embryos are mixed well with the medium and leave for 5 days at 4 °C (seeNote11).

     
  7. 7.

    Equilibrate embryos in OCT by moving them through two consecutive baths of OCT (seeNotes12 and 13) (Fig. 2a).

     
  8. 8.

    Embed multiple embryos in the cap of an Eppendorf tube (seeNote14) (Fig. 2b). Fill the cap with OCT so that a dome of OCT extends from the cap.

     
  9. 9.

    Cool ~100 mL isopentane in a 250 mL beaker to −80 °C in a liquid nitrogen bath (seeNote15) (Fig. 2c). Use blunt-end tweezers to carefully immerse the sample in the precooled isopentane until it freezes and the OCT turns white (about 5 s) (Fig. 2d).

     
  10. 10.

    Use blunt-end tweezers to take the frozen block out of the isopentane. Drain off excess isopentane, wrap the block in plastic wrap and aluminum foil (mark stage and date on the foil) and store in a tightly sealed bag at −80 °C (seeNote16).

     

3.2 Coating Coverslips

  1. 1.

    Prepare a sonication bath with clean demi water.

     
  2. 2.

    Load selected #1.5 coverslips (seeNote2) into a coverslip holder and place them in a glass container (Fig. 3b).

     
  3. 3.

    Fill the container with 1:20 Mucasol in Milli-Q water until the coverslips are covered, and sonicate for 10 min.

     
  4. 4.

    Rinse the coverslips and the container with Milli-Q water until all traces of detergent are gone.

     
  5. 5.

    Place the coverslip holder back in the container and fill with 100% ethanol. Sonicate for 10 min.

     
  6. 6.

    Drain excess ethanol from the coverslip holder. Submerge coverslips in 1:10 poly-l-lysine in Milli-Q water for 30 min to coat (seeNote17).

     
  7. 7.

    Drain excess poly-l-lysine from the coverslip holder and let the coverslips air-dry overnight.

     

3.3 Cryosectioning

  1. 1.

    Set cryostat block and blade temperature to −17 °C (seeNote18).

     
  2. 2.

    Take a cryo-block out of the −80 °C freezer and mount it to a specimen stage using OCT (seeNote19). Let the block equilibrate to cryostat temperature for 30 min.

     
  3. 3.

    Align the anti-roll plate parallel to the cryostat blade with a small distance between plate and blade to guide sections between.

     
  4. 4.

    Mount the sample on the cryostat so that the blade will hit the yolk last (seeNote20) (Fig. 3a).

     
  5. 5.

    Make 8–10 μm sections and quickly mount the sections on the coated coverslips that you prepared under Subheading 3.2 (seeNotes2123) (Fig. 3c).

     
  6. 6.

    Let the sections dry at room temperature for a couple of minutes (seeNote23) before storing them at −80 °C (seeNotes24 and 25) (Fig. 3d).

     

3.4 smFISH

  1. 1.

    Take coverslips with sections from −80 °C and place them in a 6-well plate (Fig. 3e). Post-fix the sections in 4% formaldehyde in PBS for 15 min (seeNotes26 and 27).

     
  2. 2.

    Rinse sections twice with PBS.

     
  3. 3.

    Rinse with 70% ice cold ethanol and replace with fresh 70% ethanol. Keep for 4–8 h at 4 °C to permeabilize the sample (seeNote28).

     
  4. 4.

    Rehydrate the sections in 2× SSC.

     
  5. 5.

    Treat with 5 μg/mL proteinase K in 2× SSC for 10 min while gently shaking (seeNote29).

     
  6. 6.

    Wash 2 × 5 min with 2 mL 2× SSC while gently shaking.

     
  7. 7.

    Rinse sections with wash buffer and replace with fresh wash buffer. Let equilibrate for 5 min.

     
  8. 8.

    Thaw 100 μL hybridization buffer per sample and add 0.7 μL smFISH probe (25 μM) (seeNotes4,30 and 31).

     
  9. 9.

    Coat the bottom of a petri dish with Parafilm and place a 95 μL drop of hybridization buffer with probe on the Parafilm (Fig. 3f). Take a coverslip with sample and carefully remove as much wash buffer as possible with a piece of filter paper without touching the sections. Carefully place the coverslip section-down on the drop of hybridization buffer (Fig. 3f), close but do not seal the petri dish and incubate overnight for 14–16 h at 30 °C.

     
  10. 10.

    The next day, pipet 100 μL wash buffer on the corner of the coverslip and carefully peel the coverslip off the Parafilm, making sure to not dislodge the sample. Place the coverslip in a 6-well plate with the sections facing up.

     
  11. 11.

    Rinse once with wash buffer.

     
  12. 12.

    Wash twice with wash buffer for 30 min at 30 °C without shaking. For membrane staining, add 1:100 phalloidin–Alexa 488 (or another fluorophore) to the second wash step (seeNote5). For DNA staining, add 0.5 μg/mL DAPI to the second wash step.

     
  13. 13.

    Rinse once with GLOX buffer and leave in fresh GLOX buffer at 4 °C until mounting (seeNote32).

     
  14. 14.

    Mount in GLOX mounting medium. Place 25 μL GLOX mounting medium on a microscope slide and slowly lower the smFISH sample down on the drop to prevent bubbles. Remove excess mounting medium by carefully touching a piece of filter paper to the side of the coverslip.

     
  15. 15.

    Tightly seal the sample with nail polish to prevent evaporation of the mounting medium and proceed to imaging.

     
  16. 16.

    Image your samples on an epifluorescence microscope with a high NA objective and a sensitive camera (seeNotes6 and 33) (Fig. 1b5). Use z-spacing of 0.3 μm or less to capture each transcript in multiple z-slices (seeNote34).

     

3.5 Image Analysis

Here, we describe all steps required for image analysis. For more background information on the image analysis tools, you may consult the documentation associated with Stapel et al. 2016 [15].

3.5.1 Membrane Segmentation

These steps can be skipped if one is not interested in assigning transcripts to individual cells.
  1. 1.

    Open your image in Fiji and duplicate the central membrane slice of the z-stack (e.g., slice 10 for a z-stack of 19 slices) (Fig. 1b1) using the function “Duplicate” (seeNote35).

     
  2. 2.

    If you are analyzing multiple images at the same time, resize all images to the exact same size. This is a requirement for the membrane prediction pipeline in KNIME that we will use in the next steps. You can use Fiji function “Canvas size” for this. Save the image in ‘.tiff’ format and collect all membrane images for which you want to generate segmentations in a single folder.

     
  3. 3.
    Start KNIME and open the workflow “MS-ECS-2D_2.0” (Fig. 4). Double click on the node “Image reader” at step 2.1 (Fig. 4) and select the membrane images that you generated in the previous steps (seeNote36).
    Fig. 4

    KNIME cascaded random forest pipeline. The KNIME cascaded random forest pipeline consists of two steps. In step 1 the pipeline is trained to recognize membrane, membrane intersection points (vertex points) and background on a small set of representative membrane images (seeNote36). Details on training can be found at “tinyurl.com/KNIME-MS-ECS”. Once the pipeline has been trained it can be used repeatedly to predict membrane and vertex points in step 2. In step 2.1, membrane images are loaded. In step 2.2 the cascaded random forest pipeline is run. Before running this step, the downsampling factor needs to be adjusted to the pixel size of the images (seeNote38). In step 2.3 the results can be viewed and written to the computer for further processing

     
  4. 4.

    Reset the node “Prediction” at step 2.2 (Fig. 4) by right-clicking on the node and selecting “Reset”.

     
  5. 5.

    Double click on the node “Prediction” and then on “Run cascaded RF” to open these nodes. Now double click on the node “Image Resizer” to configure this node. Set X, Y, and Z to downsample your images and speed up the analysis (seeNote37).

     
  6. 6.

    Wait until the prediction is finished. The node will show a green check mark. Now use the node “Image Writer” at step 2.3 to write the result images to a folder on your computer (Fig. 4).

     
  7. 7.

    Sort the membrane probability and the vertex probability images (Fig. 1b2) into separate folders.

     
  8. 8.

    Open Fiji and start the plugin PathFinder which is part of the MS-ECS-2D update site (seeNote38). Select the folder with your original membrane images (seestep 2), the folder with the membrane probability images (seestep 7), and the folder with vertex probability images (seestep 7) as prompted (seeNote39). While the PathFinder plugin is running, several image windows will pop up and disappear again.

     
  9. 9.

    Once the run is finished, all pop-up windows will have disappeared and you will find a folder “results” inside the folder with your original membrane images. Open the image that ends on “_scaled” in Fiji and duplicate the first channel using the “Duplicate” function. This channel contains a mask of the segmented cells (“Cell Masks”, Fig. 1b4). In addition, open the original membrane image that you used to run the segmentation (seestep 2).

     
  10. 10.
    Start the Fiji plugin “Cell annotation” that is part of the MS-ECS-2D update site (Plugins > Cell transcript > Cell annotation) (seeNote40) (Fig. 5). Select the Original image and its matching “Cell Masks” image. You can ignore the Nuclei channel and Membrane channel fields.
    Fig. 5

    The Cell annotation plugin. The Cell annotation pipeline in Fiji can be used to correct over- and under-segmentations (Correction mode) by drawing missing lines and breaking excessive lines. In the Annotation mode, cells can be assigned to a specific group, for example based on cell type. In the Label inspection mode (depicted here), the final cell segmentation results can be checked

     
  11. 11.

    In the Cell annotation tool, set the mode to “Correction”. Correct any under-segmentations by drawing missing lines pressing the left mouse button. Correct any over-segmentations by breaking excessive lines by pressing the left mouse button and the Alt key simultaneously. Go to “Label inspection” mode to check whether you have made all necessary corrections (Fig. 5).

     
  12. 12.

    Use “get Masks” or simply close the Annotation control window to finalize your masks image and save it (seeNote41) (Fig. 5).

     

3.5.2 Transcript Detection

  1. 1.

    Open the original z-stack image that you acquired on the microscope. In addition, open the Cell mask image that you generated in the previous steps in Fiji if you would like to assign transcripts to cells (seeNote42).

     
  2. 2.
    Start the “Cell transcript analysis” plugin which is part of the MS-ECS-2D update site (plugins > Cell transcript > Cell transcript analysis). Select the smFISH image and cell mask and adjust the settings to your sample (Fig. 6a).
    Fig. 6

    The Transcript analysis plugin. (a) User interface for the Transcript analysis plugin. Before running the Transcript analysis plugin in Fiji, the user is prompted to select smFISH input and mask images, and to set parameters for nuclear segmentation, transcript detection, foci detection and data display in a user interface. The values depicted in this screenshot are good starting values for the analysis of pregastrulation stage embryos imaged on systems similar to ours. Below, we will explain the function of each parameter. Nuclear segmentation parameters: (1). Nuclei channel: channel that contains the image of the nuclei. (2). Nuclei size: radius of the largest expected nucleus in the sample in pixels. This can be measured in the image with the line tool in Fiji. (3). Nuclei threshold adjustment: this value only needs to be modified if nuclear detection is poor. This value should be decreased if weak nuclei are not detected properly. Transcript segmentation parameters: (1) Transcript channel(s): channel(s) containing smFISH results. Separate channels should be separated by a comma. (2) Transcript typical radius: can be measured in the image and is typically two or three pixels. (3) Spot (= transcript) minimum intensity: determine the value of this parameter based on the spot intensity distribution (see panel B), after running the pipeline a first time. Before the transcript detection threshold has been determined, fill in arbitrary values for each smFISH channel, separated by commas. Foci segmentation parameters: (1). Foci intensity and (2). Volume threshold adjustments can be made if foci detection is poor. Values should be increased in case of over-detection and decreased in case of under-detection. A value should be entered for each smFISH channel, separated by commas. (3) Nuclei enlargement: this value is used to capture transcription foci that are located at the edge of the nucleus and usually does not need to be changed. (4) Foci maximum radius: this parameter sets a maximum to the foci size to prevent that large, nonspecific accumulations of probe are detected as foci. The maximum radius can be measured in the image. The standard setting of 10 pixels works well for our samples. Miscellaneous parameters: provide options for data display and storage and are self-explanatory. (b) Setting a transcript detection threshold. After running the Transcript analysis pipeline a first time with an arbitrary transcript detection threshold, a “Maxima distribution” image will be generated for each analyzed smFISH channel. This plot can be used to determine the optimal intensity threshold for transcript detection. (b1) The axes of the original plot are not well suited to determine the transcript detection threshold due to the large number of background spots that is detected. The user will need to zoom into the area of interest of the plot, which is located right next to the background signal by dragging a box around this area (red dashed line). (b2) After zooming in to the area of interest, a unimodal peak for the transcript intensities can be observed. The transcript detection threshold (black line) should be set between the background signal (left) and the unimodal peak for the transcript signal (right). The same threshold should be used for all images from the same sample (coverslip) that were acquired with the same microscope settings

     
  3. 3.

    First, run the Transcript analysis plugin once to generate a plot of maximum distributions (‘_MaxDistrib_*.png’) (see Fig. 6b). This plot will help you to determine the appropriate transcript detection threshold. The threshold should be set between background and transcript signal peaks (see Fig. 6b). You can zoom into the plot by dragging a rectangle around the area of interest (see Fig. 6b2). Identify the transcript detection threshold that is appropriate for your smFISH sample.

     
  4. 4.

    Rerun the Cell transcript analysis pipeline with the identified transcript detection threshold (seeNote43).

     
  5. 5.

    Check the file that ends in “_ResultImg.tif” to determine if transcripts, transcription foci and nuclei were detected correctly. Adjust the parameters for transcription foci and nuclei detection if necessary (seeNote44).

     
  6. 6.

    The file that ends on “_cell.txt” (seeNote45) contains the quantitative data of the transcript detection pipeline including cell size, nuclear size, number of transcripts per cell, number of transcription foci per cell, number of transcripts in each transcription focus, cell type (if you generated a cell type mask) and cell position. You can import this data into your favorite program (e.g., Excel, Prism, R) to analyze it.

     

4 Notes

  1. 1.

    We use Eppendorf caps to embed our samples. Alternatively, commercial molds can be used but we prefer the small size of Eppendorf caps. For early stages, coloring the bottom of the cap/embedding mold with a black marker can improve visibility of the embryos (Fig. 2b).

     
  2. 2.

    Mounting sections on coverslips instead of slides improves light transmission and image quality because it places the sample directly at the microscope objective without a barrier of mounting medium. We prefer using 22 × 22 mm coverslips. Although protocols from the company that sells Stellaris smFISH probes suggest to use #1 coverslips, we obtain optimal results with #1.5 coverslips. Selected #1.5 coverslips are optimized for the light path of most microscopes.

     
  3. 3.

    The concentration of formamide in the smFISH and hybridization buffers can be increased from 10% to up to 25% to reduce background signal. However, we find that this often results in a loss of signal and find that optimizing probe concentration is a more efficient way to increase the signal to noise ratio (seeNote28).

     
  4. 4.

    Probes can be designed using freely available software from Stellaris (https://www.biosearchtech.com/support/tools/design-software/stellaris-probe-designer). We recommend blasting the probe sequences that are suggested by this webtool against the zebrafish genome to make sure they exclusively hybridize to your transcript of interest. When choosing a fluorophore for probe labeling, select fluorophores in the red or far-red spectrum. Fluorophores with shorter wavelengths produce poor results due to auto fluorescence in zebrafish samples. Which exact fluorophore is optimal for your needs will depend on the filter sets on your microscope.

     
  5. 5.

    We use phalloidin to mark cell outlines and enable automated cell segmentation. Alternatively, membranes can be visualized using a transgenic line (e.g., a lyn::fluorescent protein line), or by injecting mRNA that codes for a membrane localized fluorescent protein at the 1-cell stage. For the choice of fluorescent protein, we have obtained good results with tdTomato, which was still visible after the smFISH protocol. GFP, on the other hand, loses its activity in the smFISH protocol and will need to be detected using an antibody (seeNote29).

     
  6. 6.

    We use a Delta Vision system equipped with an Olympus UPlan SApochromat 100× 1.4 oil objective, an EDGE/sCMOS camera and the following filter sets: 435/48 (DAPI), 525/36 (Alexa 488), 594/45 (CalFluor 610-labeled smFISH probes), 676/34 (Quasar 670-labeled smFISH probes. We have also obtained good results with a 60× 1.3NA silicon objective. It is also possible to image your samples on a spinning disk microscope with comparable objectives and camera, but results might be more difficult to interpret for probe sets with low signal to noise ratios.

     
  7. 7.

    Uncheck “Group items by category” to be able to search through an alphabetically ordered list of all tools. This makes it easy to find the required tools.

     
  8. 8.

    While we optimized the protocol for embryonic stages up to gastrulation (shield stage), it also works well at later stages. We will indicate small changes in the embedding procedure that can be taken into account when working with post-gastrulation stages.

     
  9. 9.

    Embryos at pregastrulation stages will sink within 30 min; later stages might take longer to sink.

     
  10. 10.

    It is important to rinse the embryos several times in sucrose/PBS to remove any Tween remaining from the fixation step, as this decreases embedding quality.

     
  11. 11.

    While a 5-day incubation time seems excessive, in our hands it has led to greatly improved section quality at (pre)gastrulation stages without loss of RNA signal. Make sure to keep embryos in the dark if you are working with a fluorescent transgenic line (seeNote5). At post-gastrulation stages, 5-day incubation is not necessary and one can proceed with the next steps of the protocol as soon as embryos sink to the bottom of the tube.

     
  12. 12.

    We prefer using 6-well staining plates and an embryo manipulator (Fig. 2) to move embryos through OCT (see Subheading 2.1, items 14 and 15; Fig. 2a).

     
  13. 13.

    From this step onward keep embryos and liquids at 4 °C as much as possible to enhance freezing speed and improve embryo integrity.

     
  14. 14.

    Make sure to orient all embryos in the same direction and note the orientation on the embedding cap/mold with a marker. This will improve section quality later (seeNote20) (Fig. 2b).

     
  15. 15.

    Check the temperature with a low-temperature thermometer. After use, isopentane can be stored in a glass bottle and reused.

     
  16. 16.

    Section quality will be optimal if blocks are sectioned within 1 week. However, it is possible to obtain good sections when blocks have been stored for several months.

     
  17. 17.

    Poly-l-lysine for coverslip coating can be stored in a plastic container and reused up to three times.

     
  18. 18.

    Temperatures might need to be optimized depending on your cryostat. However, we recommend to use relatively high temperatures for the best results. As a reference, we use block and blade temperatures of around −20 °C for many other sample types on the same cryostat.

     
  19. 19.

    After mounting the cryo-block to the specimen stage, remove the mold and check whether any cracks are present. If so, repair cracks with OCT (use the quick-freeze option on your cryostat to rapidly freeze the newly applied OCT). Cracks can occur due to rapidfreezing of the OCT in a confined space. Although cracks can be prevented by freezing at lower temperatures, this also decreases embryo integrity and is therefore not recommended.

     
  20. 20.

    The yolk can fracture during sectioning. By sectioning through the yolk last, you ensure that this does not affect embryo integrity.

     
  21. 21.

    Sectioning results are best when the cryo-block is relatively warm. If the block is too cold and your sections contain cracks or roll up tightly it can help to briefly warm up the block before cutting a section by pressing your thumb to the block for a couple of seconds. Make sure to wear clean gloves to not contaminate the sample with RNases!

     
  22. 22.

    The anti-roll plate will keep the section flat. Lift the plate to mount your section to a coverslip. You can use a thin brush to prevent the section from rolling up (Fig. 3c).

     
  23. 23.

    We use 6–12 coverslips per embryo and put sections obtained from multiple positions in the embryo on each single coverslip (see Fig. 2e). Make sure to not leave the sections out at RT for more than 20 min before storing them at −80 °C as this can affect smFISH quality.

     
  24. 24.

    We store sections in 6-well plates, sealed with Parafilm. Sections can be stored at −80 °C for a long time. Up to 6 months, we have not observed a decrease in sample quality.

     
  25. 25.

    Sections should be kept at −80 °C for at least 1 h before continuing with the smFISH protocol. This will improve smFISH results.

     
  26. 26.

    Add the fixative to each section immediately after taking it out of the freezer, without letting it thaw. This will prevent RNA degradation.

     
  27. 27.

    Add solutions to the side of the well and not directly on the coverslip to avoid dislodging the sections.

     
  28. 28.

    In some protocols, sections are kept in ethanol for up to a month. However, in our hands this reduces smFISH quality.

     
  29. 29.

    This treatment is optional but improves signal/noise for most of our probe sets. Mild proteinase K treatment can digest proteins that cover the RNA of interest and reveal smFISH probe binding sites. Make sure to optimize treatment conditions to your aliquot of proteinase K since excessive protein digestion will affect sample integrity.

     
  30. 30.

    If signal/noise is poor in your samples, the best way to optimize this is by changing probe concentrations in the hybridization buffer. We usually test ranges from 0.2–1.5 μL probe (25 μM stock solution) per 100 μL buffer.

     
  31. 31.

    smFISH can easily be combined with antibody staining to, for example, boost the signal of a transgenic line, or to detect the protein encoded by your mRNA of interest. For antibody staining, add the primary antibody to the overnight hybridization step (Subheading 3.4, step 9) and the secondary antibody to the first wash step (Subheading 3.4, step 12). Antibody staining can also be performed after smFISH staining. Be aware that this will lead to a reduction of smFISH signal.

     
  32. 32.

    GLOX mounting medium is a simple and quick mounting medium which can be used when imaging within 24 h of sample preparation (coverslips can be stored in GLOX buffer at 4 °C until mounting and imaging). For longer-term storage, other mounting mediums like prolong GOLD or Vectashield are recommended. When using mounting mediums like prolong GOLD or Vectashield, make sure that no liquid is left on the sample before mounting. A small amount of liquid can form a barrier between the sample and the mounting medium, preventing the mounting medium from penetrating the sample. This will lead to rapid bleaching of your sample.

     
  33. 33.

    Acquire images starting with the longest wavelength and work your way down to shorter wavelengths. This will minimize bleaching across fluorophores because shorter wavelengths carry higher energy levels and induce more bleaching.

     
  34. 34.

    Using z-spacing of maximally 0.3 μm ensures that individual transcripts are detected in more than one z-slice and that no transcripts are missed in the acquisition. This aids computational transcript detection.

     
  35. 35.

    Pressing the letter “l” in Fiji will open a command searching window. Although all commands are accessible through the menu, this is a quick way to find commands without having to browse through the menu structure.

     
  36. 36.

    The KNIME workflow consists of a training phase and a prediction phase (Fig. 4). After training the workflow once, it can be used to produce predictions for different samples. We have trained the pipeline on zebrafish membrane images, and have produced high-quality predictions for samples from zebrafish as well as other species using this pipeline. Therefore, we suggest that you use our pretrained pipeline on your samples. If the available pipeline produces poor results for your samples, you can consider retraining the pipeline with your own samples. Details on how to do this can be found on the following webpage: “tinyurl.com/KNIME-MS-ECS”.

     
  37. 37.

    The downsampling factor that you use in the prediction phase should correspond to the downsampling factor that was used to train the pipeline. To train the pipeline we used a factor of 0.33 for data with pixel size 0.1072 μm. If the pixel size in your samples is different, you should scale the downsampling factor for the prediction phase accordingly. For example, if your pixel size is 1.5× larger than ours, use a downsampling factor of 0.5.

     
  38. 38.

    We have found that the standard settings of the PathFinder plugin will give good results in many cases. If you observe strong over- or under-segmentation you may change the Segmentation parameters. Increasing the threshold for vertex, membrane path, or membrane pixel detection will reduce over-segmentation. Decreasing the small cell removal parameter will also reduce over-segmentation.

     
  39. 39.

    Image names do not need to match between folders, but make sure that the order of images is identical between folders as the PathFinder plugin will work its way down the file list.

     
  40. 40.

    Cell segmentation will not be fully accurate. You can use the Cell annotation tool (Fig. 5) to correct small segmentation errors that might have been made in the PathFinder tool. In addition to cell segmentation correction, the “Cell annotation” tool can be used to define different cell types. Options are preset for early embryonic cell types, namely EVL (enveloping layer, #1), YSL (yolk syncytial layer, #2), DEL (deep layer, #3), and Outside (#4) for regions outside of the embryo.

     
  41. 41.

    Before opening the Cell annotation tool again to start with your next image, you will need to Reset startup tools in Fiji (press ≫ in the Fiji menu bar and then Restore Startup Tools).

     
  42. 42.

    If you are not interested in assigning transcripts to single cells, you do not need to use the membrane segmentation pipeline to generate a cell mask. Instead, you can use the “Cell annotation” plugin to outline your areas of interest and use the resulting file as a mask or use no mask at all.

     
  43. 43.

    To ensure reproducibility, we recommend to use the same transcript detection thresholds for all images of the same sample (coverslip) that were acquired in the same imaging session.

     
  44. 44.

    To ensure reproducibility, we recommend to use the same parameters for nuclei segmentation and foci segmentation for all images of similar embryonic stages that were acquired at the same pixel size.

     
  45. 45.

    Additional data files that are generated are: a file with the analysis parameters (‘_parameters.txt’), a file with transcript detection information per channel (‘_spot_*.txt’), a log file (‘.log’), and a scatter plot that indicates which spots have been identified as transcripts and which have been identified as foci, based on their intensity and size (‘_ScatterPlot_*.png’). These files provide more background information on the results and the analysis.

     

Notes

Acknowledgments

This work was supported by MPI-CBG core funding, a Human Frontier Science Program Career Development Award [CDA-00060/2012-C] to NLV; and a Boehringer Ingelheim Fonds PhD fellowship to LCS. We thank Pavel Vopalensky for critically reading the manuscript, Julia Eichhorn for taking photos of the sectioning procedure, and Jan Philipp Junker and Alexander van Oudenaarden for initial advice on smFISH.

References

  1. 1.
    Thisse C, Thisse B (2008) High-resolution in situ hybridization to whole-mount zebrafish embryos. Nat Protoc 3:59–69. doi:10.1038/nprot.2007.514 CrossRefPubMedGoogle Scholar
  2. 2.
    Tomancak P, Berman BP, Beaton A et al (2007) Global analysis of patterns of gene expression during drosophila embryogenesis. Genome Biol 8:R145. doi:10.1186/gb-2007-8-7-r145 CrossRefPubMedPubMedCentralGoogle Scholar
  3. 3.
    Grün D, Kester L, van Oudenaarden A (2014) Validation of noise models for single-cell transcriptomics. Nat Methods 11:637–640. doi:10.1038/nmeth.2930 CrossRefPubMedGoogle Scholar
  4. 4.
    Junker JP, Noël ES, Guryev V et al (2014) Genome-wide RNA tomography in the zebrafish embryo. Cell 159:662–675. doi:10.1016/j.cell.2014.09.038 CrossRefPubMedGoogle Scholar
  5. 5.
    Satija R, Farrell JA, Gennert D et al (2015) Spatial reconstruction of single-cell gene expression data. Nat Biotechnol 33:495–502. doi:10.1038/nbt.3192 CrossRefPubMedPubMedCentralGoogle Scholar
  6. 6.
    Raj A, Peskin CS, Tranchina D et al (2006) Stochastic mRNA synthesis in mammalian cells. PLoS Biol 4:e309. doi:10.1371/journal.pbio.0040309 CrossRefPubMedPubMedCentralGoogle Scholar
  7. 7.
    Raj A, van den Bogaard P, Rifkin SA et al (2008) Imaging individual mRNA molecules using multiple singly labeled probes. Nat Methods 5:877–879. doi:10.1038/nmeth.1253 CrossRefPubMedPubMedCentralGoogle Scholar
  8. 8.
    Lyubimova A, Itzkovitz S, Junker JP et al (2013) Single-molecule mRNA detection and counting in mammalian tissue. Nat Protoc 8:1743–1758. doi:10.1038/nprot.2013.109 CrossRefPubMedGoogle Scholar
  9. 9.
    Wang F, Flanagan J, Su N et al (2012) Technical advance. J Mol Diagn 14:22–29. doi:10.1016/j.jmoldx.2011.08.002 CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Little SC, Tikhonov M, Gregor T (2013) Precise developmental gene expression arises from globally stochastic transcriptional activity. Cell 154:789–800. doi:10.1016/j.cell.2013.07.025 CrossRefPubMedPubMedCentralGoogle Scholar
  11. 11.
    Boettiger AN, Levine M (2013) Rapid transcription fosters coordinate snail expression in the drosophila embryo. Cell Rep 3:8–15. doi:10.1016/j.celrep.2012.12.015 CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Bahar Halpern K, Tanami S, Landen S et al (2015) Bursty gene expression in the intact mammalian liver. Mol Cell 58:147–156. doi:10.1016/j.molcel.2015.01.027 CrossRefPubMedGoogle Scholar
  13. 13.
    Itzkovitz S, Lyubimova A, Blat IC et al (2011) Single-molecule transcript counting of stem-cell markers in the mouse intestine. Nat Cell Biol 14:106–114. doi:10.1038/ncb2384 CrossRefPubMedPubMedCentralGoogle Scholar
  14. 14.
    Oka Y, Sato TN (2015) Whole-mount single molecule FISH method for zebrafish embryo. Sci Rep 5:8571. doi:10.1038/srep08571 CrossRefPubMedPubMedCentralGoogle Scholar
  15. 15.
    Stapel LC, Lombardot B, Broaddus C et al (2016) Automated detection and quantification of single RNAs at cellular resolution in zebrafish embryos. Development 143:540–546. doi:10.1242/dev.128918 CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media LLC 2018

Authors and Affiliations

  • L. Carine Stapel
    • 1
  • Coleman Broaddus
    • 2
  • Nadine L. Vastenhouw
    • 1
  1. 1.Max Planck Institute of Molecular Cell Biology and GeneticsDresdenGermany
  2. 2.Center for Systems Biology DresdenMax Planck Institute of Molecular Cell Biology and GeneticsDresdenGermany

Personalised recommendations