Simultaneous high-resolution detection of multiple transcripts combined with localization of proteins in whole-mount embryos

1Whole-mount in situ hybridization (WISH) is used to study the RNA expression pattern of genes in the context of tissues in which the RNAs function. Chromogenic in situ hybridization (ISH) techniques are commonly used in different model organisms for this purpose and are based on an enzymatic reaction that converts a colorless substrate into a dark visible precipitate. Employing chromogenic ISH, up to three different mRNAs can be detected simultaneously in Drosophila embryos. However, the detection of not more than two different transcripts has thus far been achieved in zebrafish embryos using this method.

2Compared to chromogenic ISH, fluorescent in situ hybridization (FISH) offers a higher resolution and an improved detection of overlapping gene expression patterns in a single sample. Combining FISH with confocal microscopy can thus provide spatial information concerning the expression of the investigated RNA within three-dimensional complex samples such as embryos. The signal generated in FISH relies on the binding of a horse radish peroxidase (HRP)-conjugated antibody to modified ribonucleotides of the probe, followed by tyramide signal amplification (TSA), where HRP converts a tyramide conjugated to a fluorophore into a reactive fluorescent intermediate that covalently binds to nearby amino-acid residues, typically tyrosine. Whereas multicolor FISH can potentially be used for detecting multiple transcripts by employing different fluorescent tyramide substrates, the method is less efficient, particularly for RNAs expressed at low levels. To overcome this drawback of the method, modifications to the original FISH protocol have been put forward that significantly enhance the sensitivity of the assay. Nevertheless, as other WISH methods, this protocol also relies on the generation of diffusible products that potentially decrease resolution of transcript localization.

3A newly developed alternative method has been reported that allows the simultaneous fluorescent labeling of up to five transcripts by employing an orthogonal amplification with hybridization chain reactions (HCRs). In this method, a probe set of one to nine probe species is used to target each mRNA molecule. Following probe hybridization, fluorescent RNA hairpins self-assemble into fluorescent amplification polymers. More recently, a novel technology called RNAscope has been developed for RNA detection in situ. This method is based on fluorescent signal amplification upon target probe hybridization (Figure 1), and unlike other RNA ISH methods including HCR, it is a rapid protocol that can be completed within less than 2 days. Importantly, as a result of innovative probe design and a detection strategy that relies on non-diffusible fluorogenic products, this method allows the detection of rare transcripts at high resolution, while generating a low background signal. In this method, special probes are designed such that they hybridize to the target mRNA, with each probe containing a different tail sequence that provides the base for the assembly of a signal-amplifying scaffold. As this scaffold will only form on a pair of flanking probes, signal specificity is dramatically increased (Figure 1). This method allows the simultaneous detection of up to four mRNA species in formalin-fixed paraffin-embedded cancer cell lines and tumor tissue sections using either fluorescent or chromogenic detection. Thus far, RNAscope has been used in medical research and for detection of histopathological biomarkers in clinical diagnosis. As a sensitive and specific assay with multiplex capabilities, RNAscope allows quantitative measurements of RNA levels in the context of the tissue as well as reliable detection of low abundance RNA molecules by flow cytometry. The introduction of relatively small probes prior to the generation of immobile amplification products can contribute to good penetration into the embryo and efficient labeling of rare transcripts.

4Whereas RNAscope was successfully employed for sectioned material, determination of RNA expression patterns in larger biological samples, such as intact embryos, would be extremely beneficial in biomedical research and developmental biology and when analyzing large clinical samples. Adapting this method for intact embryos would allow simultaneous high-resolution analysis of developmentally important genes and provide detailed quantitative information concerning expression patterns and expression levels within the developing embryo and tissue.

5Based on the RNAscope ISH principle, we provide a fine-tuned protocol that we developed for large tissue samples, as exemplified here by using whole-mount zebrafish embryos. The devised protocol includes steps promoting effective preservation of embryo integrity, while allowing efficient penetration of the probes and an excellent signal-to-noise ratio. It is a rapid and robust protocol that facilitates the simultaneous quantitative detection of multiple transcripts within an embryo. Importantly, in addition to revealing the spatial distribution of RNA molecules, the protocol allows the simultaneous determination of protein expression patterns and protein subcellular localization.

6Whereas the RNAscope signal amplification method can be used on tissue sections, a different protocol had to be devised for whole-mount embryonic samples. This conclusion was initially based on our finding that embryos disintegrated when subjected to the protocol designed for staining thin sections (Additional file 1A). While the RNA expression pattern of myoD, a universal marker for myogenic cells, appeared to be correct (Additional file 1D, arrowhead) in a small proportion of the embryos that withstood the procedure (Additional file 1B), a significant background signal in the notochord was present in both channels, myoD and dapB, a bacterial mRNA that serves as a negative control (Additional file 1C,D, arrows). Furthermore, following the original protocol, 4’,6-diamidino-2-phenylindole (DAPI) staining for visualizing the cell nuclei proved inefficient, presumably due to limited tissue penetration (Additional file 1E). Last, a high background signal was observed in the yolk in all channels (Additional file 1C,D, stars).

7Given the shortcomings described above, we sought to optimize the procedure and incubation conditions such that embryo integrity is preserved, while enhancing tissue penetration of reagents and maintaining a low background signal (see Table 1 for a short protocol, Table 2 for reagents, Additional file 2 for a detailed version of the procedure, Additional file 3 for troubleshooting, Table 3 for the probes and Table 4 for label probe combinations and filter set specifications).

8As a first step in establishing the protocol, we modified the composition of the solutions such that the method preserves embryo integrity, while allowing penetration of the relevant reagents. A likely source for embryo damage during washing steps when employing the original RNAscope protocol (Additional file 1) is the inclusion of lithium dodecyl sulfate in the hybridization, amplification and wash buffers. Indeed, using 0.2× SSCT buffer (saline-sodium citrate buffer + 0.01% Tween-20) or 1× PBT (phosphate buffer + 0.01% Tween-20) instead of the original wash buffer resulted in good preservation of the embryos. An additional key factor contributing to embryo preservation was the fixation step. For example, fixing 20-hours post fertilization (hpf) embryos using 4% paraformaldehyde (PFA) in PBS for 1 hour at room temperature (RT) yielded the best results with respect to high signal level and low background, while maintaining tissue integrity. Shorter fixation duration (e.g., 30 min) for 20-hpf embryos resulted in dissociation of the embryos, while such short fixation times were compatible with 24-hpf or older embryos. Considering the importance of the fixation conditions for embryo preservation, we included an additional fixation step (post-fixation, Table 1, point 6) following hybridization of the probes (Table 1, point 4).

9The initial embryo-drying step (Table 1, point 1) also appeared to be a crucial factor in preserving the integrity of embryos (Additional file 3). We found that air-drying the embryos for 30 min after methanol (MeOH) removal, followed by digestion with Pretreat solution for 20 min yielded the best results. A combination of both fixation steps and the use of 0.2× SSCT as a wash buffer resulted in optimal preservation of embryos, whose general morphology following RNAscope appears to be similar to that of fixed embryos prior to the procedure (e.g., Additional file 4).

10A critical issue in methods designed to detect RNA expression patterns in tissues is the achievement of a high signal-to-noise ratio. Three sources could potentially increase background noise in the course of RNAscope: (i) non-specific signal amplification attributed to the amplifiers and subsequent labels not associated with the specific probes, (ii) signal detection due to non-specific hybridization of the probes and (iii) autofluorescence of the tissue.

11To examine the effect of the hybridization temperature on the procedure (point (ii) above), we determined the consequences of altering this parameter on the signal and the background levels using vasa mRNA, a primordial germ cell (PGC) marker. A complete lack of signal was observed when employing the standard 65°C hybridization temperature used in FISH experiments with zebrafish. Conducting RNAscope using a 55°C or 60°C hybridization temperature resulted in high background or low specific signal (Additional file 5A,B, respectively). Thus, hybridization of the embryos at 50°C and 40°C (the standard temperature of the original RNAscope protocol) was associated with high specific signal and low background (Additional file 5C,D). To reduce the accumulation of excess probes or unbound amplification scaffold in the tissue (points (i) and (ii) above), we extended the washing incubations to three cycles of 15-min washes, which efficiently eliminated the background signal in the notochord. Employing these optimization steps reduced the background, as seen by the lack of a significant signal when using probes directed against the bacterial dapB RNA (Additional file 5E). As judged by omitting the fluorescent label from the protocol, the tissue autofluorescence generated in the course of the procedure is low (Additional file 5F).

12Interestingly, the fixation step was found to be important not only for preservation of embryo integrity, but for reducing the background signal as well. A possible explanation is that the amplification scaffolds could accumulate in samples that were subjected to non-optimal fixation conditions. In this analysis we found that reducing the fixation time to 30 min at RT for 24-hpf embryos significantly enhanced the specific signal, while reducing the non-specific one (Additional file 5H). Fixation of the same-stage embryos for 24 hours at 4°C resulted in increased background signal (Additional file 5G, star and arrow), while including 0.5 to 2.5% glutaraldehyde in the fixation solution led to a complete elimination of the signal.

13To enhance the signal in the thick embryo samples, we prolonged the duration of the hybridization from 2 hours in the original protocol (Additional file 5K) to overnight (O/N), which indeed yielded an enhanced specific signal (Additional file 5L). Employing probes already used once, a specific but weaker signal was detected (Additional file 5J) compared to the results obtained using fresh probes (Additional file 5I). While the original 30-s DAPI treatment led to nuclear staining of the outer cell layers of the embryo (Additional file 1E), O/N incubation at 4°C allowed DAPI to penetrate also into deeper cell layers to visualize the tissue context (Table 1, point 16). Counterstaining the DNA using Hoechst stain provided results similar to those obtained using the DAPI stain. In summary, we found that fixation, hybridization and washing conditions are critical for achieving an optimal signal-to-noise ratio (Additional file 6) and defined the conditions needed to detect the localization of RNA molecules reliably using the RNAscope technology in whole-mount samples.

14Both chromogenic WISH and FISH are commonly used to study the expression pattern of transcripts in cells and tissues. To compare the RNAscope-based procedure described here with existing ISH methods, we performed chromogenic WISH, whole-mount FISH and RNAscope WISH on 24-hpf zebrafish embryos using the myoD probe (Figure 2A–H). Specific expression of myoD can be detected in morphologically intact embryos using the chromogenic WISH method (Figure 2A). However, high-resolution imaging of the myoD expression revealed its limitations (Figure 2B), as the colored precipitate increased the opacity of the tissue and therefore interfered with imaging of deeper parts of the tissue. On the other hand, expression of myoD could be efficiently visualized at higher resolution when whole-mount FISH was employed (Figure 2C,D). Unlike the precipitates seen for chromogenic WISH, employing FISH for zebrafish embryos gave less diffusion of the label, thereby increasing the resolution of the assay and hence, enabling the detection of overlapping expression domains. In spite of the potential for simultaneous detection of multiple transcripts using FISH, the standard procedure appears to be limited to highly expressed RNA molecules and generated non-specific staining in our assay (arrow in Figure 2C).

15In addition to the protocol employed above, we examined an optimized whole-mount FISH protocol for zebrafish embryos using a bench-made TSA substrate, permitting higher signal intensities and visualization of less abundant transcripts. In comparison to the results obtained using standard FISH, this protocol resulted in a brighter signal for myoD mRNA (Figure 2E,F), coupled with reduced background signal from the yolk (Figure 2E, arrow).

16Employing the optimized RNAscope-based procedure, specific, high-resolution detection of myoD mRNA was achieved (Figure 2G,H). The staining obtained using the RNAscope method appeared to be more distinct and to reflect more accurately the subcellular localization of the transcripts (Figure 2H), compared to the staining in the standard and the optimized whole-mount FISH, which appears more aggregated (Figure 2D) and more diffuse (Figure 2F). The RNAscope method thus allows clear subcellular detection of RNA spots within the cytoplasm and the nuclei of zebrafish embryos (see Figure 2I and Additional file 7 for myoD RNA). To further demonstrate the capacity of the RNAscope technique to probe subcellularly localized RNAs, we detected vasa mRNA in four-cell stage embryos, a stage when the transcript is confined to the cleavage furrows (Figure 2J).

17We have shown that RNAscope is an equivalent alternative to the optimized FISH technique, while presenting several important advantages, in particular with respect to speed and ease of use coupled with high resolution and a low background signal.

18We next sought to determine whether the RNAscope technology would allow the detection of multiple transcripts simultaneously in the three-dimensional context of zebrafish embryos. In principle, up to four distinct mRNA molecules can be probed using this technique and thus reveal the spatio-temporal distribution of several mRNA molecules in a single sample. As shown in Figure 3, employing the fine-tuned protocol, cxcl12a, an RNA encoding for a chemoattractant for PGCs, is specifically detected at a high resolution in 12-hpf embryos (Figure 3A). vasa-expressing PGCs (Figure 3B) are positioned within the cxcl12a expression domains (Figure 3E), with the developing somites labeled with the myoD probe (Figure 3C). In this sample, we have thus successfully visualized transcripts of cxcl12a, vasa, myoD and DAPI-stained nuclei (Figure 3A–D, respectively). Importantly, the RNAscope technology allows the quantification of the RNA level, by measuring the fluorescence signal emitted by the tissue. This enables the RNA expression level to be represented using pseudocolors, correlating cell behavior with signal distribution (e.g. PGCs arriving at subdomains where cxcl12a is highly expressed in the left half of the embryo, as shown in Figure 3F), or quantifying the signal distribution within the cxcl12a expression domain (the graph in Figure 3F shows the signal values along the white line). The simultaneous and high-resolution quantitative detection of multiple mRNA molecules in whole embryos constitutes an important tool that is particularly useful in developmental and cell biology.

19As the RNAscope technology benefits from special probe design and detection strategy, we examined the detection of rarely expressed transcripts using this technique. cxcr4b encodes for the chemokine receptor expressed by PGCs, cells of the lateral line primordium and within the nose and hindbrain neurons of a wild-type 24-hpf embryo. Employing chromogenic WISH to detect the RNA of cxcr4b, no staining could be observed in PGCs after a short staining reaction (Figure 4A,B). However, extended staining allowed the detection of cxcr4b mRNA also in the PGCs (Figure 4C). As an example of a very weakly expressed transcript in PGCs, we thus examined cxcr4b expression in these cells in 24-hpf embryos using RNAscope. Indeed, cxcr4b mRNA was readily detected in the PGCs in the gonad region (Figure 4D, arrow), as confirmed by the expression of a known PGC marker, nanos (Figure 4E,H, arrows). In addition, very weak cxcl12a expression can also be detected at 24 hpf in the gonad using RNAscope (Figure 4F, arrow). Thus, the RNAscope technique allows the detection of weakly expressed transcripts in whole-mount zebrafish embryos.

20The ability to combine localization of RNA and fluorescent proteins would enable investigation of RNA expression patterns relative to specific embryonic or cellular structures that express fluorescent proteins. We presumed that the relatively low temperature applied in the RNAscope-based protocol (40°C) was likely to preserve the function of fluorescent proteins. We tested this supposition by using 20-hpf zebrafish transgenic embryos, whose PGCs express either EGFP on their plasma membrane or DsRed in the perinuclear structures that characterize PGCs (Figure 5A–F). Indeed, PGCs that expressed vasa transcripts in their cytoplasm exhibited a strong EGFP signal on their membrane (Figure 5A,B,C). Similarly, DsRed expressed in PGC granules could be observed along with vasa RNA (Figure 5D,E,F), clearly presenting vasa localization relative to the granules within PGCs. Increasing the temperature to 50°C, which is compatible with RNA detection, negatively affected the survival of protein fluorescence. We able to show that the RNAscope technology is compatible with immunohistochemistry experiments, where protein localization is determined by use of specific antibodies. As shown in Figure 5G,H,I, we could visualize myoD RNA together with E-cadherin labeled by antibodies directed against the protein. The RNAscope method thus allows the simultaneous detection of RNA expression patterns with protein localization using antibodies or fluorescent-protein fusions.

21As investigating the RNA expression pattern in internal organs of embryos of more advanced developmental stages is often desired, we evaluated the RNAscope method in this context. To this end, we used 1-day post fertilization (dpf) Tg(elavl3:egfp) and 3-dpf Tg(fli1a:egfp) transgenic embryos that express egfp RNA and EGFP protein in neuronal cells or in their vasculature, respectively. egfp RNA could be detected in internal tissues of 1- and 3-day-old larvae (Figure 6, Additional files 4E and 8). Importantly, consistent with the idea that the 20-bp short probes used in the procedure can readily penetrate the tissue, the signal provided by RNAscope was comparable to that generated by the endogenous EGFP fluorescence (which withstands RNAscope and as such is not limited by reagent penetration) at different tissue depths (Figure 6B). To demonstrate the penetration of RNAscope probes and reagents into deeper tissues of 3-day-old embryos, irrespective of imaging-related constrains, we sectioned the Tg(fli1a:egfp) transgenic embryos following whole-mount RNAscope. egfp RNA could be detected in internal vessels at 170 and 250 μm tissue depths (frontal view, Additional file 9).

22The RNAscope method allows the detection of RNA molecules in a wide range of developmental stages and tissue depths in whole-mount zebrafish embryos.

23ISH is a fundamental method for studying gene expression in cells, tissues and whole organisms. Using zebrafish as an example, we present the development of RNAscope-based detection of multiple transcripts in whole-mount embryos, offering a procedure that is compatible with protein localization. The improved protocol preserves embryo integrity and is remarkably rapid. The uniquely designed probes for RNAscope and the amplification of target signals allow the detection of rarely expressed transcripts. In addition, the exclusion of antibodies and the lack of diffusible reaction staining products contributed to increased resolution and excellent signal-to-noise ratios. While not tested in this study, we assume that minor modifications to the final protocol would allow its application to embryos of other species and large tissue samples as well. The implementation of such modifications in whole-mount samples would contribute to biomarker analysis of clinical samples, diagnostics and basic research.

24Zebrafish (Danio rerio) with the AB genetic background and transgenic fish carrying Tol-kop-egfp-f’-nanos 3′ UTR, Tol-kop-gra-dsRed-nanos 3’ UTR, fli1a-egfp or the elavl3-egfp transgene were used. The zebrafish were handled according to the laws of the state of North Rhine-Westphalia, supervised by the veterinarian office of the city of Muenster.

25Zebrafish embryos were fixed according to their developmental stage (Additional file 2), in 4% PFA in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7) at RT prior to (for embryos younger than 24 hpf), or after (for 24-hpf embryos and older) hand dechorionation. For each experimental point, 20 embryos were processed in one 1.5 ml Eppendorf tube. A series of increasing MeOH concentrations (25%, 50%, 75%, 2× 100%) in 0.1% PBT (0.1% Tween-20 in PBS, pH 7.4) was used to dehydrate the embryos stepwise in 5-min washes. After the last MeOH wash, embryos were stored at -20°C for at least one night. The embryos were then air-dried for 30 min at RT and subjected to the RNAscope-based signal amplification (Advanced Cell Diagnostics). Protease digestion of embryos using Pretreat 3 was performed for 20 min at RT followed by rinsing the embryos three times in 0.01% PBT (0.01% Tween-20 in PBS, pH 7.4). Target probe hybridization (hybridization buffer 1: 6× SSC (1× SSC is 0.15 mol/L NaCl, 0.015 mol/L Na-citrate), 25% formamide, 0.2% lithium dodecyl sulfate, blocking reagents) was performed at 40°C O/N to allow the preservation of protein fluorescence. A hybridization temperature of 50°C results in comparably high signal intensities in the sample and can thus be employed for background problems when the activity of fluorescent proteins is not of importance. Following recovery of the probes, the embryos were washed three times for 15 min in 0.2× SSCT (0.01% Tween-20, 3 mM NaCl, 0.3 mM TriNaCitratdihydrate, pH 7) at RT. An additional fixation step was performed using 4% PFA for 10 min at RT.

26For RNA detection, incubation with the different amplifier solutions was performed in a water bath at 40°C. The pre-amplifier (2 nmol/L) was in hybridization buffer 2 (20% formamide, 5× SSC, 0.3% lithium dodecyl sulfate, 10% dextran sulfate, blocking reagents). The amplifier (2 nmol/L) was in hybridization buffer 2. The label probe (2 nmol/L) was in hybridization buffer 3 (5× SSC, 0.3% lithium dodecyl sulfate, blocking reagents).

27After each hybridization step, the embryos were washed three times with 0.2× SSCT for 15 min. The embryos were then incubated with DAPI ready-to-use solution (Advanced Cell Diagnostics) O/N at 4°C with slow agitation. Prior to imaging, embryos were rinsed in 0.01% PBT, mounted in 1% low melting point agarose (LMP) and imaged in 1× PBS solution.

28One-color WISH was performed as previously described, with modifications according to. Whole-mount FISH was performed based on and the optimized FISH according to. Digoxigenin-labeled myoD [GenBank:NM_131262] and cxcr4b [GenBank:NM_131834] probes were synthetized using T7 or Sp6 polymerase, respectively, according to the manufacturer’s protocol (Roche, Basel, Switzerland).

29Following the RNAscope protocol, 28-hpf embryos were embedded in 4% LMP and sectioned into 100-μm slices using a vibratome (Leica, Wetzlar, Germany VT1000E). Sections were permeabilized (0.1% Tween-20 and 0.3% Triton-X-100 in PBS) for 1 hour at RT and incubated in blocking buffer (0.3% Triton X-100 and 4% BSA in PBS) O/N at 4°C. Incubation with a mouse monoclonal primary antibody targeting E-cadherin (BD, San Jose, CA, USA Biosciences, 610181) in blocking buffer (1:100) was performed O/N at 4°C followed by three washes for 5 min in 0.3% Triton X-100 in PBS. The secondary antibody (rabbit Alexa 568 anti-mouse IgG, Invitrogen) was applied at 1:1000 dilution in blocking buffer and incubated O/N at 4°C in the dark. The sections were then washed three times, mounted on slides using fluorescence mounting medium (Dako, Hamburg, Germany) and imaged.

30Fluorescent images were acquired on an LSM 710 microscope (Zeiss, Oberkochen, Germany). Chromogenic ISH images were obtained on an Axioplan2 epifluorescent microscope (Zeiss, Oberkochen, Germany). Whole-clutch bright-field images were captured using a SteREO Discovery.V12 (Zeiss, Oberkochen, Germany) and AxioCam MRc5 camera (Zeiss, Oberkochen, Germany). Images were processed using ImageJ software (National Institutes of Health) and Imaris (Bitplane AG, Zurich, Switzerland).

31Signal intensity measurements to assess RNAscope probe penetration were performed on confocal images after subtracting background levels using the rolling ball algorithm. The measurements were performed at four different selected expression sites in the vessels. Measurements were repeated for three independent embryos. Signal-to-noise ratio was determined employing the ROI manager of ImageJ, dividing the average pixel intensity of an area expressing egfp under the control of the fli1a promoter with that in a region where egfp is not expressed. This procedure was repeated six times in different confocal planes.