Single Molecule Fluorescence In Situ Hybridization Using RNAscope to Study Hematopoietic and Vascular Interactions in the Zebrafish Embryo and Larva

4. Probe(s) hybridization

Notes:

1. The protocol beneath is for incubation with one probe, which is illustrated in Figures 3–7 beneath. Up to three probes can be co-incubated at the hybridization step if they are in the C1, C2, and C3 channels, respectively. Since probes in the C1 channel are delivered diluted in probe diluent (PD), the probe solution can be applied directly to the samples. If the probes are in the C2 or C3 channels, they should be diluted 1/50 in PD before addition to the samples.

2. Control probe(s), ideally in each one of the channels that are used for assays with specific probes in a given channel, should be assayed in parallel to appreciate background (which can also vary with specific organs).

3. For our procedure, we use a heating block with a cap (see the reference in the Equipment list and Figure 1); although never tested, an alternative would be to use a moisture-controlled oven, particularly for the overnight incubation with the probe(s).

a. Incubate embryos/larvae with 100 µL of PD for at least 2 h, on the heating block, at 40 °C.

b. Remove as much PD as possible and replace with 50–100 µL/tube of the probe (or a mixture of several probes).

c. Incubate overnight in the heating block at 40 °C.

Caution: Since the heating block is not agitating (to prevent damaging embryos/larvae), mix gently the content of the tube(s) a couple of times before the overnight incubation by gently tapping the tubes with the finger.

C. Day 2: Amplification and fluorescence reaction

Note: The Multiplex v2 kit allows performing HRP-based tyramide signal amplification (TSA) using fluorescent dyes applicable to up to three probes concomitantly (3-plex assay). Here, we show confocal fluorescence images [Figure 3 (n = 1, 48 hpf embryo, trunk and tail regions), Figures 4 and 5 (n = 1, 5 dpf larva, trunk, tail, and pronephros regions), and Figure 6 (n = 1, 5 dpf larva, trunk and tail regions)] obtained using one probe only (single plex assay), with amplification reactions revealing a 570 nm fluorescent dye (OPAL-570). Assays were also performed with two probes concomitantly but with difficulties regarding the maintenance of the morphology of embryos/larvae due to additional steps of incubation/washes at the HRP reaction step. In this specific case, we revealed the two probes using OPAL-570 and OPAL-690. OPAL-480 (see Reagents) can be used as an alternative to one of the dyes, depending on the strategy followed for the experiment and imaging setup (here, the use of eGFP transgenic fish line is not compatible with this dye).

1. Amplification and dye coupling

Note: Prewarm (for 10 min, on the surface of a heating block, at 30 °C) the concentrate 50× WB (provided with the kit) before shaking the bottle and diluting 1/50. Before each step with specific reagents provided in the kit [AMP(1–3) reagents, FL v2 HRP-C(1–3) reagents, TSA buffer], place aliquots of the required volume (depending on the number of samples) on the heating block for 5 min (prewarming at 40 °C). For replacement of all reagents and WB, sediment embryos/larvae by gravity on the bench and remove media under the binoculars. From time to time, during incubations at 40 °C on the heating block, shake very gently by tapping tubes with the finger.

a. Wash each sample twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

Critical: Overnight incubation with probes in PD and following washes in WB trigger the clearing of embryos/larvae; all steps of reagents and WB removal after incubations need to be performed under binoculars to avoid losing material. This recommendation is important because the proteinase K digestion has already triggered the translucence of material, which is critical particularly for handling 48 hpf embryos. In addition, embryos/larvae will lose density and will take longer to sediment by gravity; do not centrifuge because embryos/larvae will curve, which will make mounting and imaging more difficult.

b. Remove WB and add 50 µL of AMP1. Incubate for 30 min at 40 °C on the heating block. Add 1 mL of WB and sediment embryo/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

c. Remove WB and add 50 µL of AMP2. Incubate for 30 min at 40 °C on the heating block. Add 1 mL of WB and sediment embryo/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

d. Remove WB and add 50 µL of AMP3. Incubate for 15 min at 40 °C on the heating block. Add 1 mL of WB and sediment embryo/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

Note: Even if only one probe is used (for example, in Figures 3–7), incubations with the AMP1, AMP2, and AMP3 solutions are required, particularly for comparative studies using probes in different channels, which will ensure regularity in background levels.

e. Remove WB and, for each sample, add 50 µL of RNAscope Multiplex FL v2 HRP-C(1–3) reagent. Incubate for 15 min at 40 °C on the heating block. Add 1 mL of WB and sediment embryos/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

Note: In Figures 3–7, we used the cmyb-C3 probe. Hence, we have performed the reaction with the RNAscope Multiplex FL v2 HRP-C3 reagent.

f. Remove WB and add, for each sample, 100 µL of TSA buffer. Incubate for 10 min on the heating block at 40 °C.

g. Remove the 100 µL of TSA buffer and add 500 µL of the diluted (1/500) OPAL-570 dye (or any other appropriate OPAL dye depending on the required fluorescence wavelength). Incubate on the heating block for 30 min at 40 °C. Sediment embryos/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

h. Remove WB and, for each sample, add 50 µL of the HRP blocker solution. Incubate on the heating block for 15 min at 40 °C. Add 1 mL of WB and sediment embryos/larvae by gravity. Remove medium and wash twice with 1 mL of WB for 10 min and with gentle rotative agitation on the rocking platform at RT.

Note: If the reaction is done with one channel only, the protocol can be stopped here (as performed for results shown in Figures 3–7). If required, a DAPI staining of nuclei can be performed. In any case, at the end, keep the embryos/larvae in PBST until mounting. If more than one probe was hybridized, continue the protocol by returning to step C1e; incubate with HRP coupled to another channel [e.g., HRP-C2 if the first reaction was performed with HRP-C1 and if the second probe was synthesized in the C2 channel (or HRP-C3 if synthetized in the C3 channel)]. In the latter case(s), the DAPI reaction can be performed at the end of the protocol (and/or continuing with immunofluorescence if required).

D. Day 3: Mounting and imaging

Note: Mounting can also be performed at the end of Day 2. Ideally, imaging should be performed on Day 3 but imaging on day 4 is also possible without significant degradation of the biological material (even without 0.02% NaN₃ or any other preservative, with mounted samples kept in a cold room at 4 °C).

1. Mounting

Critical: The mounting step is critical because embryos/larvae are transparent, sticky, and very fragile. Perform all steps under the binoculars. To prevent losing embryos/larvae, mounting is processed in the same Ibidi dish as the dissection of yolk and protruding eyes (removing eyes with 25G needles, particularly the one facing the glass bottom of the Ibidi dish, will ensure flatness of the mounting). More specifically, the eye facing the glass needs to be removed if imaging anterior niches (the thymus, the pronephros, or any other organ in the anterior region). Any dissection is performed on a restricted area of the glass (e.g., the upper part; see Figure 8).

a. Remove WB (in case the mounting is performed directly after amplification and dye coupling) and wash twice for 10 min in PBST with gentle rotative agitation on the rocking platform at RT.

b. Transfer up to 6–7 embryos/larvae from the Eppendorf tube to the Ibidi dish (maximum volume of 1 mL; if the volume of animals and PBST is below 1 mL, complement up to 1 mL with PBST; this volume will cover the surface of the glass bottom).

c. Carefully remove the eyes and protruding tissues that may hamper the flatness of the mounting (with 25G needles).

d. Remove medium and debris and add 1 mL of PBST.

e. Move the embryos/larvae with a slightly trimmed Eppendorf microloader to the middle/bottom of the dish; the fact that they stick to the glass will help to maintain them longitudinally on the lateral side (Figure 8).

f. Slowly add 1 mL of 2% agarose (melted by warming up at 85 °C on the heating block; see the note below) while mixing very gently to limit perturbations to the embryos/larvae alignment. If necessary, reorient them longitudinally. Wait until the agarose is solidified before moving the dishes.

Note: Prepare the 2% low melting agarose in 1× PBS. Typically, 30 mL of 1× PBS is added to a 50 mL Falcon tube. Warm up the solution in a microwave until nearly reaching the boiling point with the tube uncapped (wear protective glasses and gloves). Add the agarose powder slowly and stepwise, close the cap of the Falcon tube, and shake vigorously. Repeat warming up, adding agarose stepwise and mixing until the entire solution is nearly entirely dissolved. Complete the volume to 50 mL, warm up, close the tube, and shake. Aliquot in 2 mL Eppendorf tubes. When needed for mounting, warm up the Eppendorf tubes containing 2% agarose to 85 °C until complete melting. Transfer 1 mL of the agarose in a new 2 mL Eppendorf tube and immediately perform 3–4 ups and downs to homogenize and cool the temperature down before mounting by slowly mixing the 2% agarose with 1 mL of PBST containing embryos/larvae properly dissected in the Ibidi dish. The appropriate temperature when initiating mounting can be maintained by slowly releasing the 2% agarose from the 1 mL tip at the external plastic margin of the dissection area (see Figure 8); no vapor should be seen on the margin (which, before the addition of the 2% agarose, should remain dry while the 1 mL PBST is contained in the area delimited by the glass bottom of the Ibidi dish).

Pause point: Imaging can be performed on the day following the mounting. In this case, keep the dishes protected from light in a cold room.

2. Confocal imaging and interpretation of RNAscope information

Confocal imaging is specific to each setup, depending on available equipment; for an example of imaging conditions of 48 hpf embryos treated for RNAscope, see [8]. Hence, this part of our protocol is not aimed at describing image acquisition but at guiding the reader to critically analyze the outcome of the RNAscope procedure, here applied to zebrafish 48 hpf embryos/5 dpf larvae but that can be extended to other stages or organisms. This section is then organized as a series of points (Figures 3–6, Videos 1 and 2) to guide throughout specific and relevant points of RNAscope analysis after image acquisition, including the characteristics of the RNAscope signals with the multiplex assay (notes 1 and 2), the specific case of in-depth signals (note 3), and the choice of appropriate biological material and configuration to quantify RNAscope signals per cell type (note 4, linking to the Data analysis section).

a. To obtain information throughout the entire organ/region of the embryo and at the cellular resolution, a series of z-sections needs to be acquired. In Figure 3, a 48 hpf embryo expressing the eGFP controlled by the vascular kdrl promoter [Tg(kdrl:eGFP)] was imaged through a series of z-sections in the trunk and in the tail regions (observed with a 40× objective). Images reveal the typical RNAscope spotty signals obtained with the multiplex assay.

To quantify the number of spots and avoid saturation, all fluorescent confocal images were obtained in the linear range of the camera and analyzed without modification of the brightness and contrast. Importantly, quantitative analyses using RNAscope can only be applied to spot numbers and not intensity (in theory, 1 spot = 1 mRNA; for more details the Data analysis section). Larger and more intense spots are assumed to correspond to signals produced by several mRNAs accumulating in an area below the resolution of the optic microscope. Regarding spot fluorescence intensity, their respective minimal and maximal intensities give an estimation of the dynamic range of signals in comparison to the background. Typically, for the results shown in Figure 3, RNAscope signals are, for the highest intensity spots, 53 times higher than the background; for the lowest intensity spots, signals are 3.3 times higher than the background.

b. For 5 dpf larvae, in the trunk and in the tail regions, the RNAscope spots are detected with high density, mostly because hematopoietic cells are organized into clusters (see Figure 4). The conditions of image acquisition were the same as for Figures 3–7; raw images, without modification of brightness and contrast, reveal individual spots (relatively well individualized in the z-sections shown in Figure 4, in which, however, image brightness and contrast was modified from the raw images for visibility). For a comparison with the signals analyzed in Figure 3, the fluorescence intensity is 98 times higher than the background for the highest intensity spots and 7.6 times higher for the lowest signals. To evaluate, in the 5 dpf larva, the association of RNAscope spots with individual hematopoietic cells (using the cmyb-C3 probe), see Figure 6.

c. In the pronephros region of 5 dpf larvae, in addition to the comments made for Figure 4 in point 2 above, imaging the region of interest in depth is a major issue. The peri-glomerular area is localized beneath the notochord, deep in the anterior region (see Figure 5 and the z-projection image on the left), hence decreasing image resolution and signal intensity as well as increasing diffusion of the fluorescence signal and overall background. In these conditions, using deep red dyes is not recommended (such as for example the OPAL-690 dye). One advantage is the capture, by the notochord, of the fluorescent dye that is useful to localize the peri-glomerular region as well as the more distant pronephric tubules and vascular structures around which HSPCs are niching (see Figure 5, Video 1, produced by assembling the individual slices of the z-stack, and Video 2, obtained by reconstituting the z-stack in 3D using the Imaris software).

d. In 5 dpf larvae, the RNAscope signal can be superposed to hematopoietic cells when using a transgenic fish line that expresses a fluorescent reporter driven by a hematopoietic promoter. For example, we used here a Tg(runx1+23:eGFP) line that allows visualizing HSPCs (mainly organized in clusters), expressing cytosolic eGFP as well as cmyb mRNAs (Figure 6). The RNAscope spots can then be colocalized with HSPCs expressing eGFP by 3D reconstitution of the z-stack using the Imaris software (see Figure 7).

Data analysis

A. Image analysis overview

Images acquired with the confocal microscope are multichannel z-stacks that can be analyzed with open-source image analysis software [including but not limited to ImageJ (FIJI), Icy, and Napari [17–19]] or with proprietary software such as Imaris. Here, the foundation of the quantitative analysis lies in the segmentation of cellular volumes (labeled, in our case, by eGFP transgene fluorescence) as well as the segmentation of RNAscope spots and the possibility of detecting which cells contain spots. On more general grounds and in different biological contexts, analyses can lead to a diversity of quantitative results, such as (i) the total number of positive cells, (ii) the repartition of positive cells within an organ or—the case in our current studies—in embryos/larvae, (iii) the variation of percentage of positive cells in interfering conditions, and (iv) co-expression analyses in multiplex RNAscope assay.

The quantification of the number and size of spots per cell is possible but should be done carefully as confounding factors [the variation of the TSA amplification step duration between multiple tubes or across conditions, the number of ZZ-pairs used to detect each gene, and the localization of mRNA molecules within the cell (homogeneously dispersed or clustered together)] might bias the interpretation of the analysis by modifying the intensity and size of RNAscope spots.

Importantly and to avoid misinterpretation, image analysis should be performed with software and workflows designed for 3D analysis. For instance, working with 2D projections (such as maximum intensity projections; see, for example, the z-projections in Figures 3–6) of z-stacks will lead to the observation of falsely positive cells, when RNAscope signal overlaps with transgene fluorescence in the x- and y-axis but not the z-axis.

To perform qualitative analysis and visualize images, Napari and Imaris Viewer are free options that can allow exploring data with fast 3D reconstitution (3D rendering is the default visualization mode), while ImageJ (Image → Stack → 3D project…) and Icy (3D VTK) propose slower and less manipulable renderings, with a noticeable reduction of the displayed image quality. All allow to hide or show specific channels, zoom on the data, and change the color and intensity display, but 360° rotation of the 3D stack is not supported in ImageJ.

For the creation of a quantitative analysis workflow, we use the Imaris software (version 10.1.0) with the complete workflow detailed below (see also https://imaris.oxinst.com/tutorials for various Imaris tutorials). Because Imaris requires a license, we will briefly mention two free alternative software and possible workflows (ImageJ and Napari).

1. ImageJ

With ImageJ, the 3D ImageJ Suite and MorpholibJ suite [20,21] propose solutions to perform segmentation of cells and spots (Analyse → 3D object counter, Plugins → 3D → 3D spot segmentation). To facilitate segmentation, pre-processing (filtering: Plugins → 3D → 3D fast filters) can be used. Once segmented objects have been obtained, segmentation data can be modified (merge, split, remove) and saved using the 3D object manager (Plugins → 3D → 3D manager). With the same manager, .csv files containing information regarding segmented objects can be exported, including fluorescence intensity, size, position, or geometrical information. The segmentation workflow can be made into a macro (Plugins → macro → record), allowing you to run your different steps, recording it as a script, and saving it (see https://imagej.net/scripting/macro for detailed use of the macro tools), and reusing it automatically on multiple files (Plugins → macro → run). The main limitation of the ImageJ suite is the absence of object-object interaction analysis: spots and cells can be segmented, but one cannot automatically get information on the position of spots relative to cells. One can deduce if a segmented cell is positive, based on the intensity of fluorescence in the RNAscope channel, but the number or size of RNAscope spots within a cell cannot be extracted. A script (macro) is available as an example at https://doi.org/10.5281/zenodo.14936123.

2. Napari

With Napari, the Zelda plugin [22] provides a single package that allows for pre-processing, segmentation, relation analysis between multiple object types, and generation of data tables and plots from extracted information. Napari is a relatively recent Python interface for image visualization and analysis; as such, it requires minimal knowledge of coding to use the majority of its plugins or to create a custom pipeline. One of the advantages of Napari is its versatility and the constant creation and upgrading of the available packages by the user community. The Zelda plugin provides a graphical interface to facilitate its utilization. One of the shortfalls of the Zelda package compared to the Imaris software is the reduced ability to interact with the objects after segmentation (manual curating: split, merge, suppression, classification, sorting). Moreover, Zelda allows for the concomitant analysis and segmentation of only two object types, and therefore simultaneous analysis of multiple RNAscope targets in multiplex experiments is not supported.

B. 3D cells and RNAscope signal segmentation with Imaris, Cell Biologist package

1. Convert the acquired individual confocal image files into proprietary .ims files using the Imaris file converter software (Add files). Specify, if needed, the voxel size (Set Voxel Size) and convert one or multiple files simultaneously (Start All) (Figure 7A). New .ims files are automatically generated in the same folder as input files.

2. Open the .ims file and select the 3D View in the Surpass mode to generate a 3D rendering of the z-stack (Figure 7B). The channel color display can be changed as well as the displayed intensity (Edit → Show Display Adjustment).

3. Depending on your needs, Imaris offers multiple segmentation tools (Spots, Surface, Cell, Filaments). Here, we used the Cell tool, a combination of the Spots and Surface tools, that allows for the simultaneous segmentation of cells and objects (nuclei, vesicles) within the cells. Once selected, one can choose to use either the creation wizard with modifiable pre-defined parameters or manually parameter all steps. Of notice, the manual configuration can be used to import objects already segmented (such as the already defined Spots and Surface objects to generate integrated Cell objects). Here, we used the creation wizard and will detail each step (Figure 7C).

a. First step. Here, one can choose to load saved parameters from previous experiments, by clicking the tool button next to the Favorite Creation Parameters, and loading from the computer a .icpx file. Similarly, at the end of segmentation, one can export parameters for future experiments (Store Parameters for batch). Here, we went through all eight steps to generate Cell objects. One can move to the next step by clicking the blue arrow at the bottom of the window or fast forward to the end of the wizard by clicking on the double green arrow (after loading pre-defined parameters, for example).

Caution: Parameters can be reused for similar experiments (same transgenic background with identical pattern of fluorescence, same RNAscope probe), but must be re-evaluated for each new experimental setup; thresholding for background intensity and cell or spot size, for example, will differ in different transgenic lines and with different expression of the genes investigated.

First, for our image analyzed here (Figure 7C), we selected amongst the four options the one corresponding to our data (A Surface object containing multiple Spots objects). Since our image is relatively small, we performed the segmentation of the whole image straightaway; for bigger images, the Segment only a region of interest option allows running the wizard in a small ROI and only after expanding the segmentation to the whole image, to speed up the calculations during the configuration steps.

b. Configure cell type: Chose whether the fluorescence signal is cytoplasmic (our case) or restricted to the membrane (such as a ZO1-eGFP signal), select the channel used for the cell segmentation, and apply pre-processing filters (Smooth: gaussian filter, Background subtraction).

c. Set the Cell Threshold (Background Subtraction): The threshold proposed by the wizard, corresponding to the limit between signal and background noise, can be manually modified. Here, if cells are tightly packed and touching each other, manual splitting can be performed, using the Split Touching Cells function. In our case, because we lacked a nuclear signal, we used the Split by Seed Points option and set the average diameter of a single cell (in our case to 5µm). This step will allow creating seeds for each single cell within the segmented surface that will be used as a reference point to separate touching cells using a watershed algorithm downstream.

d. Set a threshold for the Filter Cell Seed Points function: We established a threshold based on the quality measurement (fluorescence intensity at the center of the object). Other parameters could be taken into account, such as maximum/mean/median/minimum intensity. At this stage, it is helpful to temporarily deselect the rendering of the cell surface (in the Settings tab) so that the seed for all individual cells can be evaluated and the threshold adjusted accordingly (the surface rendering is not transparent and hides the cell seed points).

e. Set a threshold for the Filter Cell function: here, we filtered out cells based on size only (Number of voxels), but other parameters such as distance to neighbors, sphericity, distance to border, volume, or sphericity could also be used.

f. Segmenting the Spots (RNAscope signals) within segmented Cell Surfaces. Select the channel used for spot segmentation, configure the average spot size (Estimated Diameter), and select options for Background Subtraction and variable size spots [Different Vesicle Size (Region Growing)]. At this stage, several types of spots can be segmented simultaneously (for multiplexed RNAscope experiments with signals in different channels, for example).

g. Set a threshold for the Filter Vesicles function: We established a threshold based on the quality measurement (fluorescence intensity at the center of the object), but other parameters could be used alone or in combination. This generates a unique seed for all Spots, and all seeds have the same size at this step.

h. Configure the Vesicle Region Growing function. This function takes the seeds generated in the previous steps and expands or reduces their volume according to the size of the RNAscope fluorescence signal.

4. Once the Cell objects are generated, segmentation can be modified manually: merge or split objects, remove artefactual objects (in our example, artefactual cells were detected in the notochord because of its autofluorescence), duplicate, subset, or classify them. The classification can be manually curated (one selects cells and creates several categories; these categories are then usable for further statistical analysis) or automatically generated based on quantifiable parameters: number of spots contained in the cell, intensity of the spots, localization, size… The display of the object (color, transparency) can also be modified: cells can be colored uniformly, randomly, or with a statistic-based code (with the parameter of your choice, such as the ones aforementioned). Each object has an ID number and can be identified when in post-processing analysis.

Imaris allows to save screenshots of images, create movies from 3D rendering (see Videos 2 and 3), and hide and show any layer (e.g., 3D grid, fluorescence channels, objects).

C. Quantitative analysis

1. Imaris proposes a quantitative data visualization module, the Vantage mode (Figure 9A), that allows for all objects created in the opened image to generate plots to analyze data.

a. Plots can be 1D view (boxplots with any variable of interest on the y-axis), scatter plots (2D View, with each point characterized by two variables plotted on the x- and y-axis, such as the number of spots as a function of cell size, as shown in Figure 9A), or grid/gallery views (with all segmented objects aligned on a grid and classified depending on the variable that was chosen).

b. The Spatial View proposes an analysis of the position of objects [for example, the cumulative number of RNAscope Spots as a function of the distance to Cells surface in the Cumulative Distribution Function (CDF) plot], compares it to a random distribution in space, and interprets it as an attraction or repulsion between objects. In our case (Figure 9B), when looking at spots, the attraction distance is negative (as expected), the data line (solid pink line) is higher than the confidence interval’s random distribution (dashed pink line and light pink interval), and almost 100% of points are located within a distance of 0 µm from the cell surface.

2. Imaris only plots data for one image and, to compare data from multiple embryonic regions or multiple replicates, all data regarding spatial, fluorescence, geometrical, and interaction features can be saved in .csv or .xls files (Statistical tab, Figure 9C left) for downstream analysis using alternative data analysis software, such as R, Prism (which requires a license), or Matplotlib. We used RStudio and specific packages [23,24] to read the data (multiple files in a row), perform statistical analysis, and generate plots for publication (Figure 9C, right). Below are code snippets for data reading and plotting. More information on the use of the ggstatsplot package can be found on https://indrajeetpatil.github.io/ggstatsplot/.

a. R library setup

library(stringr)

library(dplyr)

library(ggplot2)

library(ggstatsplot)

library(readxl)

b. Read multiple .xls files generated by Imaris

# set you working directory

setwd("your path")

# list all .xls files save from Imaris

ls<-list.files("your path/", pattern = ".xls")

# create an empty dataframe were data will be store

df<-data.frame()

# function to read all listed files automatically

for (i in 1:length(ls)) {

# print the name of the file currently being read

  print(ls[i])

# read the sheet of interest in the .xls file you are reading (here for example the sheet corresponding to the number of vesicle per cell) and create a dataframe containing all information from this sheet

  temp<-read_excel(ls[i], sheet = "Cell Number Of Vesicles Ves-2",

                   col_names = TRUE, skip = 1)

# add a “file” column to your dataframe containing the name of the file you read. This information is important when reading multiple files, as it allows to retrieve information on the origin of your data after having merged everything together

  temp$file<-as.character(substr(ls[i], start = x, stop = y))

# same as above, allows you to read multiple sheets from the same file depending on the information you want (here two example, the sheet with information on cell size [number of voxel], and the sheet with information on fluorescence signal in the RNAscope channel

  temp2<-read_excel(ls[i], sheet = "Cell Number of Voxels",

                    col_names = TRUE, skip = 1)

  temp3<-read_excel(ls[i], sheet = "Cell Intensity Sum Ch=2 Img=1",

                    col_names = TRUE, skip = 1)

# merge the columns of all temporary dataframes generated, the column used for merging is “ID”, that corresponds to the unique ID that identifies all Cell objects.

  temp<-merge(temp, temp2, by="ID", all.x = TRUE)

  temp<-merge(temp, temp3, by="ID", all.x = TRUE)

# at each loop of the function, append the new opened data to the dataframe containing previously opened data

  df<-rbind(df, temp)

}

c. Generate plots with ggstatsplot package

# generate boxplot with automated statistical comparisons between several conditions (here non-parametrical testing).

ggbetweenstats(data=df, x=size, y=`Cell Number Of Vesicles`, type="nonparametric", pairwise_comparisons=TRUE, pairwise.display = "none",

               point.args = list(position = ggplot2::position_jitterdodge(dodge.width = 0.5), alpha = 2, size=4 ),

               centrality.point.args = list(color="black", size=3),

               centrality.label.args = list(nudge_x = 0.5, segment.linetype = 4,min.segment.length = 0),

               results.subtitle = TRUE,

               ggplot.component = list(labs(title="", x="Cell size", y="Cell Number of RNAscope Spots")))

# generate scatterplots to identify correlations between two variables, with automated correlation coefficient calculation.

ggscatterstats(

  data = df,

  x = `Cell Number Of Vesicles`,

  y = `Cell Intensity Sum`,

  xlab = "Cell Number of RNAscope Spots",

  ylab = "Cell cytoplasmic sum of fluorescence intensity at 561nm",

  title = ""

)

Validation of protocol

This protocol was validated for embryos in [8]. For its translation to the 5 dpf larvae, see [15].

In [8], the protocol applied to 48 hpf embryos was used in Figure 3/figure supplement 1A (with the cmyb probe) and Figure 4 (with the pard3ba probe); Figure 3/figure supplement 2 (with the pard3aa, pard3ba, and pard3ab probes in panels A, B, and C, respectively) shows large fields of the entire body region at the level of the trunk and encompassing the peripheral nervous system (spinal cord), the notochord, the dorsal artery, the vein, and the gut region; Figure 3/figure supplement 2D shows magnifications of aortic regions to visualize RNAscope dots for the three pard3 probes at the cellular level; Figure 3/figure supplement 3B (with the pard3ba probe) shows aortic regions; and Figure 3/figure supplement 4 (with the pard3aa and pard3ab probes in panels A and B, respectively) shows magnifications of aortic regions.

In [15], the protocol was applied to 5 dpf larvae using gata2b (ACD Biotechne, catalog number: 551191-C2), cd34/podocalyxin (ACD Biotechne, catalog number: 1223761-C1), cmyb (ACD Biotechne, catalog numbers: 558291-C1, 558291-C2, 558291-C3), and timp4.2 (ACD Biotechne catalog number: 1255031-C3) probes.

General notes and troubleshooting

General notes

1. The enzymatic digestion of the biological material is critical, particularly when the mRNAs of interest are expressed in deep tissues in animals (e.g., the periglomerular region in the larva here; see Figure 5). The efficiency of enzymatic batches may also be subject to change. Of note, the Multiplex Fluorescent Reagent kit v2 contains a set of proteases (information not provided by the supplier) that can be tested, in addition to the proteinase K used in this protocol. Testing these proteases may be useful for other stages, organs, and animals.

2. Bio-techne is now providing a reagent kit with new TSA (tyramide signal amplification) Vivid^TM dyes that provide brighter signals [TSA Vivid dyes 520 (green), 570 (orange), and 650 (red)]. These dyes are suitable for multiplexing (three probes concomitantly), and their optical properties are compatible with the use of DAPI for staining of nuclei. The combination of detection of three mRNAs and one protein target using antibodies is also possible.

Troubleshooting

1. A major problem for zebrafish embryos and larvae is that the RNAscope wash buffer (WB) is harsh; consequently, the animals become translucent, smooth, and easily damaged. The advantage of the harsh buffer is that it limits background; this is an essential parameter. If necessary, washing steps can be optimized by diluting the WB or using an alternative buffer [e.g., SSC buffer at different concentrations (UltraPure 20×, Invitrogen, catalog number: 15557-044)]. Because animals become smooth, centrifugation should be avoided, and animals should be sedimented between washes only by gravitation. The removal of any medium (WB and media with reagents) should be, at all steps, monitored under the binoculars.

Note also that regarding the morphology of animals (which is more critical for embryos), an alternative that is faster and probably less damaging is to use the HiPlex assay (https://acdbio.com/rnascope-hiplex-assays), with which three fluorophores can be added together directly after HiPlex AMPs (1–3). The limitation of using HiPlex is its cost, which is approximately twice the price of the Multiplex assay and which uses different probes.

2. The reaction gives large patches and not individual spots (e.g., Figure 5), which does not occur in the majority of cases; this precludes quantification of dots. Probes should not be diluted (recommendation from the supplier). One step that may be worked on is the incubation time and temperature of the dye coupling in TSA buffer (this parameter was not tested). If the information is critical, the user may contact the supplier to reduce the number of ZZ pairs. Another alternative may be to expand the tissue to decrease the density of the signal, for example by using expansion microscopy; for one such example in the zebrafish larva, see [25,26]. Still, the translation of RNAscope to expansion conditions needs to be investigated; see [27].

Acknowledgments

We wish to warmly thank Philippe Herbomel for providing resources for image acquisition [spinning disk station from Andor, financed by a grant from the Fondation pour la Recherche Médicale (#DEQ20160334881)]. We warmly thank Laure Bally-Cuif for her support. We thank Catherine Vivier for her excellent technical assistance and our fish facility members Sebastien Bedu, Nathan Guibert, and Karim Sebastien for their daily help and commitment. This work was supported by Institut Pasteur, CNRS, and grants from La Ligue Contre le Cancer, Comité de Paris (RS21/75-5 and RS22/75-9) to AS. LT was the recipient of a PhD fellowship from the Collège Doctoral, Sorbonne Université, and from the Labex Revive (ANR-10-LABX-73).

Competing interests

The authors declare no conflicts of interest

Ethical considerations

The fish maintenance at the Pasteur Institute follows the regulations of the 2010/63 UE European directives and is supervised by the veterinarian office of Myriam Mattei.

References

1. Eilken, H. M., Nishikawa, S. I. and Schroeder, T. (2009). Continuous single-cell imaging of blood generation from haemogenic endothelium. Nature. 457(7231): 896–900. https://doi.org/10.1038/nature07760
2. Boisset, J. C., van Cappellen, W., Andrieu-Soler, C., Galjart, N., Dzierzak, E. and Robin, C. (2010). In vivo imaging of haematopoietic cells emerging from the mouse aortic endothelium. Nature. 464(7285): 116–120. https://doi.org/10.1038/nature08764
3. Bertrand, J. Y., Chi, N. C., Santoso, B., Teng, S., Stainier, D. Y. R. and Traver, D. (2010). Haematopoietic stem cells derive directly from aortic endothelium during development. Nature. 464(7285): 108–111. https://doi.org/10.1038/nature08738
4. Kissa, K. and Herbomel, P. (2010). Blood stem cells emerge from aortic endothelium by a novel type of cell transition. Nature. 464(7285): 112–115. https://doi.org/10.1038/nature08761
5. Jaffredo, T., Gautier, R., Eichmann, A. and Dieterlen-Lièvre, F. (1998). Intraaortic hemopoietic cells are derived from endothelial cells during ontogeny. Development 125(22): 4575–4583. https://doi.org/10.1242/dev.125.22.4575
6. de Bruijn, M. F., Ma, X., Robin, C., Ottersbach, K., Sanchez, M. J. and Dzierzak, E. (2002). Hematopoietic Stem Cells Localize to the Endothelial Cell Layer in the Midgestation Mouse Aorta. Immunity. 16(5): 673–683. https://doi.org/10.1016/s1074-7613(02)00313-8
7. Lancino, M., Majello, S., Herbert, S., De Chaumont, F., Tinevez, J. Y., Olivo-Marin, J. C., Herbomel, P. and Schmidt, A. (2018). Anisotropic organization of circumferential actomyosin characterizes hematopoietic stem cells emergence in the zebrafish. eLife. 7: e37355. https://doi.org/10.7554/elife.37355
8. Torcq, L., Majello, S., Vivier, C. and Schmidt, A. A. (2024). Tuning apicobasal polarity and junctional recycling in the hemogenic endothelium orchestrates the morphodynamic complexity of emerging pre-hematopoietic stem cells. eLife. 12: e91429. https://doi.org/10.7554/elife.91429
9. Tian, Y., Xu, J., Feng, S., He, S., Zhao, S., Zhu, L., Jin, W., Dai, Y., Luo, L., Qu, J. Y., et al. (2017). The first wave of T lymphopoiesis in zebrafish arises from aorta endothelium independent of hematopoietic stem cells. J Exp Med. 214(11): 3347–3360. https://doi.org/10.1084/jem.20170488
10. Murayama, E., Kissa, K., Zapata, A., Mordelet, E., Briolat, V., Lin, H. F., Handin, R. I. and Herbomel, P. (2006). Tracing Hematopoietic Precursor Migration to Successive Hematopoietic Organs during Zebrafish Development. Immunity. 25(6): 963–975. https://doi.org/10.1016/j.immuni.2006.10.015
11. Tamplin, O. J., Durand, E. M., Carr, L. A., Childs, S. J., Hagedorn, E. J., Li, P., Yzaguirre, A. D., Speck, N. A. and Zon, L. I. (2015). Hematopoietic Stem Cell Arrival Triggers Dynamic Remodeling of the Perivascular Niche. Cell. 160: 241–252. https://doi.org/10.1016/j.cell.2014.12.032
12. Gross-Thebing, T., Paksa, A. and Raz, E. (2014). Simultaneous high-resolution detection of multiple transcripts combined with localization of proteins in whole-mount embryos. BMC Biol. 12(1): 1–14. https://doi.org/10.1186/s12915-014-0055-7
13. Gross-Thebing T. (2020). RNAscopeTM Multiplex Detection in Zebrafish. In: Nielsen BS, Jones J, editors In Situ Hybridization Protocols. Springer US, New York, NY; pp 195–202.
14. Jin, S. W., Beis, D., Mitchell, T., Chen, J. N. and Stainier, D. Y. R. (2005). Cellular and molecular analyses of vascular tube and lumen formation in zebrafish. Development 132(23): 5199–5209. https://doi.org/10.1242/dev.02087
15. Torcq, L., Vivier, C., Schmutz, S., Loe-Mie, Y. and Schmidt, A. A. (2024). Single-cell and in situ spatial analyses reveal the diversity of newly born hematopoietic stem cells and of their niches. bioRxiv: e618250. https://doi.org/10.1101/2024.10.14.618250
16. Schmidt, A. (2022). Shining a light on hematopoietic stem cells. eLife. 11: e81963. https://doi.org/10.7554/elife.81963
17. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nat Methods. 9(7): 676–682. https://doi.org/10.1038/nmeth.2019
18. de Chaumont, F., Dallongeville, S., Chenouard, N., Hervé, N., Pop, S., Provoost, T., Meas-Yedid, V., Pankajakshan, P., Lecomte, T., Le Montagner, Y., et al. (2012). Icy: an open bioimage informatics platform for extended reproducible research. Nat Methods. 9(7): 690–696. https://doi.org/10.1038/nmeth.2075
19. Sofroniew N et al. (2024). napari: a multi-dimensional image viewer for Python.
20. Ollion, J., Cochennec, J., Loll, F., Escudé, C. and Boudier, T. (2013). TANGO: a generic tool for high-throughput 3D image analysis for studying nuclear organization. Bioinformatics. 29(14): 1840–1841. https://doi.org/10.1093/bioinformatics/btt276
21. Legland, D., Arganda-Carreras, I. and Andrey, P. (2016). MorphoLibJ: integrated library and plugins for mathematical morphology with ImageJ. Bioinformatics. 32(22): 3532–3534. https://doi.org/10.1093/bioinformatics/btw413
22. D’Antuono, R. and Pisignano, G. (2022). ZELDA: A 3D Image Segmentation and Parent-Child Relation Plugin for Microscopy Image Analysis in napari. Front Comput Sci. 3: e796117. https://doi.org/10.3389/fcomp.2021.796117
23. Patil, I. (2021). Visualizations with statistical details: The 'ggstatsplot' approach. J open source softw. 6(61): 3167. https://doi.org/10.21105/joss.03167
24. Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., et al. (2019). Welcome to the Tidyverse. J open source softw. 4(43): 1686. https://doi.org/10.21105/joss.01686
25. Cárdenas-García, S. P., Ijaz, S. and Pereda, A. E. (2024). The components of an electrical synapse as revealed by expansion microscopy of a single synaptic contact. eLife. 13: e91931. https://doi.org/10.7554/elife.91931
26. Ijaz, S., Cárdenas-García, S. and Pereda, A. (2024). Expansion Microscopy of Synaptic Contacts on the Mauthner Cells of Larval Zebrafish. Bio Protoc. 14(1353): e5067. https://doi.org/10.21769/bioprotoc.5067
27. Cho, I. and Chang, J. B. (2022). Simultaneous expansion microscopy imaging of proteins and mRNAs via dual-ExM. Sci Rep. 12(1): 3360. https://doi.org/10.1038/s41598-022-06903-3