mRNA fluorescence in situ hybridization to determine overlapping gene expression in whole-mount mouse embryos


Correspondence to: John Cobb, Department of Biological Sciences, 2500 University Drive N.W., University of Calgary, Calgary AB T2N 1N4 Canada. E-mail:


Background: Whole-mount in situ hybridization (ISH) is a prevalent tool to examine the spatial distribution of gene transcripts in intact embryos. Chromogenic-based methods of signal development are commonly used in mouse embryos because of their high sensitivity. Fluorescence techniques, however, offer several advantages over chromogenic methods including the ability to visualize multiple signals in a specimen at once. Results: We describe a procedure for fluorescence in situ hybridization (FISH) for whole mouse embryos up to embryonic day 13.5. We show that this approach successfully produces a bright expression signal for several genes, validating the procedure in multiple tissues. Further, we show that double FISH can be used to visualize the expression of two genes in a single embryo by determining that Hoxd13 and Shh are co-expressed in both the limb bud and the hindgut. Finally, we demonstrate that FISH can be paired with confocal microscopy to take optical sections of interior regions of the embryo. Conclusions: FISH is a valid alternative to chromogenic-based ISH for visualizing gene expression in whole mouse embryos. This work provides a framework to add additional fluorescence signals in the mouse such as visualizing both mRNA and protein by pairing the procedure with immunofluorescence. Developmental Dynamics 242:1094–1100, 2013. © 2013 Wiley Periodicals, Inc.


In situ hybridization (ISH) is a powerful and widely used tool to analyze the expression pattern of a gene of interest during animal development. In contrast to methods that measure RNA levels following its isolation from a tissue (e.g., quantitative PCR), the main advantage of ISH is that the spatial distribution of a gene transcript is visualized. It can be performed with sections to investigate a single plane of tissue, or with an intact embryo, with the added advantage that the expression pattern is visualized in the three-dimensional (3D) context of the embryo (Rosen and Beddington, 1993). Thus it is routinely used to examine gene expression patterns during normal development, and in genetic mutants to explore downstream effects.

Most commonly, chromogenic ISH is performed by enzymatically converting a colorless substrate, usually 5-bromo-4-chloro-3-indolyl-phosphate/4-nitro blue tetrazolium (BCIP/NBT), into a dark precipitate. However, fluorescence in situ hybridization (FISH) is an important alternative to chromogenic methods for several reasons. For one, the development of a fluorescence signal does not increase the opacity of a tissue as a colored precipitate does (Hughes and Krause, 1998; Quintana and Sharpe, 2011). With this, staining in one region does not mask staining in another region. Furthermore, it is possible to detect overlapping expression patterns of two or more genes in a single embryo through multichannel labeling. In this respect, chromogenic methods are limited to genes that have mutually exclusive expression domains, as any overlap is difficult to discern (Bueno et al., 1996). Additionally, as fluorescent labels are non-diffusible, unlike the dark precipitate produced from BCIP/NBT, FISH has higher resolution than chromogenic methods (Hughes and Krause, 1998). Finally, the advantages of confocal microscopy can be utilized with fluorescently-labeled whole embryos, such as taking non-invasive optical sections to examine interior regions of the embryo, and the potential to recombine these optical sections to build an accurate 3D representation of a sample (Quintana and Sharpe, 2011).

The initial procedures for whole-mount FISH were limited to small embryos, such as Drosophila (Hughes and Krause, 1998), as the signal-to-noise ratio was too low for larger embryos. Mouse embryos older than E9.5 had to be sectioned prior to fluorescence labeling (Bueno et al., 1996), thereby losing the advantages of using whole embryos. Although some studies reported FISH in whole-mount vertebrate embryos (e.g., Jowett and Yan, 1996; Palmes-Saloma and Saloma, 2000), the introduction of Tyramide Signal Amplication (TSA) technology has led to vast improvements in several model systems, including Xenopus, zebrafish, and chick (Davidson and Keller, 1999; Denkers et al., 2004; Clay and Ramakrishnan, 2005; Vize et al., 2009; Lauter et al., 2011). TSA technology utilizes Horse Radish Peroxidase (HRP) to convert an amplification reagent, a tyramide that is conjugated to a fluorophore, into a highly reactive intermediate that covalently binds to nearby protein residues (Raap et al., 1995; Zaidi et al., 2000). With this, many fluorophores are deposited near the target mRNA, boosting the fluorescence signal. Thus, although chromogenic methods may still be better able to detect low abundance transcripts, FISH is a viable alternative for examining the expression of many genes.

Although available in multiple model systems, whole-mount FISH has not yet been demonstrated with mouse embryos older than E9.5. Indeed, this bottleneck is part of the basis for the development and use of alternate approaches for imaging gene expression in whole embryos (Sharpe et al., 2002; Quintana and Sharpe, 2011). Here we describe and validate whole-mount FISH in mouse embryos up to E13.5. Critical to the procedure is taking steps to maximize the fluorescence signal-to-noise ratio. This includes extensive washing of the embryo, optimized probe and antibody concentrations, and Tyramide Signal Amplification. We show that this procedure is sufficiently sensitive to clearly visualize the expression of several genes that are expressed in diverse tissues. Furthermore, the technique can be used to reveal overlapping expression patterns of two genes. In addition to demonstrating the immediate benefits of one- and two-color FISH, this work provides a framework for adding more fluorescent labels, which could be additional transcripts, or integrating the procedure with immunofluorescence to visual both mRNA and protein distribution. We anticipate that whole-mount FISH will be a valuable technique to the mouse community for the study of both development and disease.


As a first step to investigate the possible use of FISH with mouse embryos, Hoxd13 was chosen as a suitable test case as it has a dynamic and well-characterized spatio-temporal gene expression pattern (Kmita et al., 2002). RNA probe and anti-digoxigenin (DIG)-HRP concentrations were optimized empirically and TSA employed using a bench-made fluorescein (FITC)-conjugated tyramide on E10.5–E13.5 embryos. The basic hybridization and wash methods were adapted from standard techniques (Chotteau-Lelievre et al., 2006), and the full protocol is detailed in the Experimental Procedures section. At E10.5, this resulted in a bright fluorescence signal in the posterior regions of both the fore and hindlimb bud, and the entire tail bud (Fig. 1A). At E11.5, the Hoxd13 signal expanded anteriorly in the limb bud though it retained a posterior bias, and was also seen in the genital bud (Fig. 1B). Finally, at E12.5 and E13.5, expression became symmetrical in the limb bud, and the outline of the digit condensations could be seen (Fig. 1 C,D). Importantly, these patterns are in good agreement with stage-matched embryos using chromogenic methods (Fig. 1E–H) and previous descriptions of Hoxd13 expression at these stages (Kmita et al., 2002; Tarchini and Duboule, 2006). Therefore, FISH is capable of detecting gene expression up to E13.5. Of note, the success of ISH in general relies on penetration of the reagents into the specimen, so late stages such as E12.5 and E13.5 work best in accessible regions of the embryo, such as the appendages.

Figure 1.

Comparison of expression pattern using fluorescence and chromogenic ISH for several genes. ISH with a Hoxd13 probe at E10.5 (A,E), E11.5 (B,F), E12.5 (C,G) and E13.5 (D,H). The pattern seen through fluorescence ISH (A–D) closely matches that seen by chromogenic ISH (E–H). Insets show close-up view of the forelimb bud at the respective stage of the panel. ISH for Fgf8, Shh, and Shox2 at E10.5, and Epha3 at E12.5, with either a fluorescence signal (I–L) or a chromogenic signal (M–P). aer, apical ectodermal ridge; di, dienchaphalon; drg; dorsal root ganglia; fb, forebrain; fl, forelimb; fp, floor plate; fh, forebrain-hindbrain junction; gt, genital tubercle; hg, hindgut; hl, hindlimb; lp, lateral plate; me, mandible/maxilla epithelium; np, nasal process; pa, pharyngeal arches; so, somites; tb, tail bud; tr, trigeminal ganglion. All embryos of a common stage are shown with the same magnification. Younger embryos are shown at a higher magnification than older embryos. Scale bar = 1 mm.

To test whether FISH can be broadly used to determine gene expression patterns and is not limited to a particular transcript or probe, four additional genes were selected to characterize-Fibroblast growth factor 8 (Fgf8), Sonic hedgehog (Shh), Short stature homeobox 2 (Shox2), and Eph receptor a3 (Epha3) (Fig. 1I–L). Each of these genes has a unique expression pattern and is therefore appropriate to assess the performance of FISH in several different tissues. Fgf8 expression is seen in the developing forebrain and midbrain-hindbrain junction, the epithelium between the developing maxilla and mandible, the apical ectodermal ridge of the limb bud, the somites, and the tail bud (Fig. 1I) in good agreement with published findings (Crossley and Martin, 1995; Richman et al., 1997). Analysis of the Shh transcript shows characteristic signal in the floor plate, the diencephalon, the posterior region of the limb bud, and the hindgut (Fig. 1J) (Echelard et al., 1993). It is interesting to note that the diencephalon expression was only weakly visible in our experiments using chromogenic approaches but was strongly labeled using FISH. Shox2 expression is seen in the limb buds, trigeminal and dorsal root ganglia, and nasal process (Fig. 1K) (Blaschke et al., 1998; Semina et al., 1998). Finally, Epha3 is expressed in the developing limbs, brain, lateral plate mesoderm, and pharyngeal arches (Fig. 1L). Epha3 was chosen as it requires a relatively long signal-development period using chromogenic ISH, permitting us to assess the sensitivity of FISH. While the Epha3 fluorescence signal appeared relatively weak when visualizing the entire embryo, most or all expression domains could nonetheless be seen (Fig. 1L). In conclusion, whole-mount FISH successfully detects gene expression in a wide range of tissues, though it may be more difficult to detect a subset of mRNA targets, such as lower abundance transcripts.

We next sought to determine whether we could use double FISH to examine the co-expression of two genes. Dual fluorescence involves simultaneous hybridization of two distinctly labeled RNA probes, and sequential employment and detection of two antibodies using different fluorophore-conjugated tyramides for each. Dinitrophenol (DNP) was selected as a second hapten and Cy3 as the secondary fluorophore, in combination with the DIG-labeled RNA and FITC fluorophore described above, since other studies have shown good sensitivity with these reagents (Vize et al., 2009). In addition to sequential signal development, the activity of the HRP used to amplify the first signal must be destroyed before adding the HRP for the second signal. Without this inactivation step, an individual signal will be readily apparent in both channels used to visualize the specimen, obscuring the result. Shh and Hoxd13 were selected as targets as these genes are both expressed in multiple common tissues of the embryo. In the limb bud, genes of the HoxA and HoxD clusters are collectively required for Shh expression (Kmita et al., 2005); Hoxd13 is sufficient among the Hox genes to promote Shh expression (Tarchini et al., 2006) and HOXD13 protein binds to the Shh limb enhancer (Capellini et al., 2006), collectively suggesting these genes are co-expressed. Expression of each gene could be independently viewed in an individual E10.5 embryo through distinct channels (Fig. 2 A,B). Importantly, inactivation of the HRP through the use of a low-pH glycine treatment was effective, in agreement with Lauter et al. (2011), as the expression of each gene was indistinguishable from single FISH experiments with no additional features of the alternate gene (compare Fig. 2A,B to Fig. 1A,J). Further, the merged image allows their simultaneous visualization, with distinct expression of each gene in the majority of the embryo, and overlapping expression in the limb bud (Fig. 2C). Within the limb bud, Shh expression is confined to the posterior region, and strong Hoxd13 expression partially overlaps with this pattern but also extends distally and anteriorly. To examine their co-expression more closely, single optical planes of the intact limb bud were imaged using confocal microscopy (Fig. 2G–I). The same general pattern was observed in the confocal images as with the whole limb bud images, providing further evidence that these genes are indeed coexpressed and not merely appearing as overlapping layers because of the imaging viewpoint. Hoxd13 and Shh are additionally both expressed in the posterior hindgut endoderm (the cloacal endoderm) (Dolle et al., 1991; Echelard et al., 1993) (Fig. 2J–L). A single optical section of this region shows Shh uniquely expressed in the floor plate of the neural tube, yet it is co-expressed with Hoxd13 in the endoderm (Fig. 2M–O).

Figure 2.

Double FISH using Shh and Hoxd13 probes at E10.5. Epifluorescence images of the entire embryo (A–C), the forelimb bud (D–F), and the hindgut endoderm (J–L). Confocal images of a single plane of the forelimb bud (G–I), and the hindgut (M–O). Arrowheads in D–I point to the posterior region of the forelimb bud where Shh and Hoxd13 are coexpressed. Arrowheads in J–O point to the hindgut endoderm where Shh and Hoxd13 are coexpressed. Arrows in M and O point to the floor plate of the neural tube. Scale bar = 50 μm.

In conclusion, we have described and validated mRNA FISH in whole-mount mouse embryos at several embryonic stages. Using multiple genes, we show that this technique performs well as compared to established chromogenic methods, and allows the expression patterns of two genes to be visualized in a single mouse embryo. Furthermore, the fluorescence signal(s) can be paired with confocal microscopy to visualize deeper regions of the embryo without the need for physical sectioning.



All solutions should be made up in high-quality deionized water that is free of RNAse.

  • FITC and Cy3-conguated tyramides as previously described (Vize et al., 2009)
  • PBS (Phosphate-Buffered Saline): 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, in H2O, final pH 7.4.
  • PBST (PBS-Tween): PBS with 0.1% (vol/vol) Tween 20
  • TBS (Tris-Buffered Saline): 137 mM NaCl, 2.7 mM KCl, 25 mM Tris-HCl pH 7.4, in H2O.
  • TBST (TBS-Tween): TBS with 1% (vol/vol) Tween 20
  • TBTI (TBST-Imidazole): TBS with 0.05% Tween 20 (vol/vol) and 10 mM imidazole.
  • Tween® 20 (Polyethylene glycol sorbitan monolaurate) MP Biomedicals (Solon, OH)
  • 4% PFA (Paraformaldehyde): PBS with 4% (wt/vol) paraformaldehyde (Sigma-Aldrich, St. Louis, MO). Filter sterilized and stored at −20°C.
  • Methanol EMD Millipore (Billerica, MA)
  • Vials (4 mL) Fisher Scientific (Pittsburgh, PA)
  • H2O2 30% vol/vol solution (VWR)
  • ProK (Proteinase K) Roche (Indianapolis, IN)
  • Torula RNA: Ribonucleic acid from Torula yeast, type VI (Sigma-Aldrich)
  • Heparin Sigma-Aldrich
  • BSA (Bovine Serum Albumin): H2O with 100 mg/ml Albumin (Sigma-Aldrich) as stock solution. Filter sterilized and stored at −20°C.
  • Glycine Sigma-Aldrich
  • Fetal Calf Serum Gibco (Carlsbad, CA)
  • Anti-DIG-AP Roche
  • Anti-DIG-HRP Roche
  • Anti-DNP-HRP Perkin-Elmer (Waltham, MA)
  • Formamide (Deionized) Sigma-Aldrich
  • SDS (Sodium Dodecyl Sulfate) CalBioChem (San Diego, CA)
  • SSC (Standard Saline Citrate) (20X): 3 M NaCl, 300 mM sodium citrate. pH of solution adjusted to 4.5. Autoclaved and stored at Room Temperature (RT).
  • SDS (Sodium dodecyl sulfate): 20% wt/vol in H2O (stock solution) pH=7.2. Stored at RT.
  • Agarose Invitrogen (Carlsbad, CA)
  • Low-Melt Agarose Sigma-Aldrich
  • BM Purple Roche
  • Buffer H1: 50% (vol/vol) deionized formamide, 5× SSC (pH=4.5), 1% SDS, 0.1% Tween 20. Made in H2O. Stored at −20°C.
  • Buffer H2: Buffer H1 with 5 mg/mL Torula RNA and 50 μg/ml Heparin. Stored at −20°C.
  • Buffer L1 = Buffer H1
  • Buffer L2: 50% (vol/vol) deionized formamide, 2× SSC (pH=4.5), 0.1% Tween 20. Made in H2O. Stored at −20°C.
  • Buffer L3: 2×SSC, 0.1% Tween 20. Made in H2O. Stored at −20°C.
  • Blocking Solution: TBST with 20 μl/ml of 100 mg/mL BSA and 20 μl/ml heat-inactivated fetal calf serum. Made fresh day of use.
  • NTMT: 100 mM NaCl, 100 mM Tris (pH=9.5), 1% Tween 20. Made in H2O. Stored at RT.



Wild-type mice on a mixed C57BL/6-129/Sv background were used. The Life and Environmental Sciences Animal Care Committee at the University of Calgary approved all animal experiments.

RNA probes

Standard methods were used to generate RNA probes. Briefly, between 700 and 1,000 ng of linearized plasmid was used as a template for transcription with one of T7, T3, or SP6 polymerases (Promega, Madison, WI) using either DIG RNA labeling mix (Roche), or DNP RNA labeling mix (Perkin Elmer). Nucleic acids were purified using a G-50 microspin column (GE Healthcare, Piscataway, NJ). To assess relative quantity of labeled RNA to DNA template, 2 μl of the probe was analyzed on an agarose gel. If the RNA was not approximately 10 times more abundant than the DNA (often the case for DNP-labeled RNA), the DNA was degraded with DNAse (Invitrogen). All probes were previously described: Hoxd13 (Herault et al., 1996), Epha3 (Cobb and Duboule, 2005), Shox2 (Cobb and Duboule, 2005), Fgf8 (Crossley and Martin, 1995), and Shh (Echelard et al., 1993).

Synthesis of tyramides

FITC and Cy3 tyramides were synthesized as previously described (Vize et al., 2009). We note that other groups also successfully generate bench-made tyramides (e.g., Davidson and Keller, 1999; Lauter et al., 2011). However, commercial tyramides have also been recently reported to function in whole-mount Xenopus embryos (Lea et al., 2012).

Fluorescence in situ hybridization procedure

Collection and dehydration of embryos

Dissect embryos in PBS and immediately fix in 4% PFA overnight at 4°C. Dehydrate embryos on ice in a methanol series (vol/vol with PBST): 5 min in 30% methanol, 5 min in 50% methanol, 5 min in 70% methanol, and 2 times for 5 min in 100% methanol. Embryos can be kept at −20°C for several months.

Day 1: Rehydration, preparation, and RNA hybridization

For the entire procedure, embryos are kept in 4-mL vials (to promote thorough washing later in the procedure); 1-mL solutions per vial are used at all steps unless otherwise noted. Wash steps are performed on a rocker.

  • 1. Rehydrate embryos on ice (vol/vol in PBST): 5 min in 70% methanol, 5 min in 50% methanol, 5 min in 30% methanol, and 2 times for 5 min each in PBST.
  • 2. Bleach embryos in 6% H2O2 in PBST on ice (block from light).
  • 3. Wash 2 times for 5 min each with PBST on ice, and then 1 time for 5 min at RT.
  • 4. Permeabilize embryos with PBST containing 10 μg/mL of ProK at RT for the following times: E9.5 embryos for 9 min, E10.5 embryos for 10 min, E11.5 embryos for 13 min, E12.5 embryos for 14 min, and E13.5 embryos for 17 min.

Although ProK treatment helps reagents penetrate the specimen, ProK (partially) digests the ectoderm, so may be inappropriate for examining ectodermal gene expression.

  • 5. Wash 3 times for 5 min with PBST on ice.
  • 6. Postfix for 20 min at RT with 4% PFA in PBS.
  • 7. Wash 4 times for 5 min each with PBST on ice, then 1 time for 5 min with PBST at RT.
  • 8. Incubate embryos for 5 min in preheated Buffer H1 at 68°C.
  • 9. Incubate embryos for 2 hr in preheated Buffer H2 at 68°C.
  • 10. Meanwhile, denature RNA probes for 10 min at 80°C in Buffer H2 at approximately 1 ng/μl (normally dilute probe stock 1/100). Add probe solution to embryos, with enough to completely cover them. For double FISH, both probes are added together (i.e., 1 ng/μl of probe #1 and 1 ng/μl of probe #2). Incubate overnight at 68°C.
Day 2: Post-hybridization washes and antibody incubation
  • 11. Wash 3 times for 30 min each with preheated Buffer L1 at 68°C.
  • 12. Wash 3 times for 30 min each with preheated Buffer L2 at 68°C.
  • 13. Wash 15 min in preheated Buffer L3 at 68°C. Place embryos at RT for 15 min.
  • 14. Wash embryos 3 times for 5 min each in TBST at RT.
  • 15. Incubate embryos in Blocking Solution for 2 hr at RT.
  • 16. Incubate embryos overnight at 4°C with 1/2000 anti-DIG-HRP antibody or 1/800 anti-DNP-HRP antibody in Blocking Solution.

Note: Concentration of antibodies should be determined empirically to determine optimum concentrations.

Day 3: Washing

To ensure effective washing, perform these washes in 4-ml vials that are approximately 80% full with vials lying on their sides on a rocker giving moderate embryo movement.

  • 17. Wash 3 times for 5 min each in TBST at RT.
  • 18. Wash 6 more times for 1 hr each in TBST at RT.
  • 19. Wash overnight in TBST at 4°C.
Day 4: Fluorescence signal development
  • 20. Incubate embryos for 1 hr at RT with TBTI solution containing 2 mg/ml BSA (0.2% BSA).
  • 21. Replace solution with TBTI solution containing 10 mg/ml BSA (1% BSA). Add 1 μl of tyramide (either FITC or Cy3) per 100 μl of solution. Incubate for 30 min at RT. Do not remove solution. Proceed to next step.

To conserve tyramide, only use enough solution to cover embryos.

  • 22. To solution already covering the embryos, add 1 μl of 1% H2O2 per 100 μl of solution and momentarily shake vial gently to mix. Incubate for 30 min at RT. Crucial step. Do not exceed 30 min, as background may increase.

Fluorophores are sensitive to light. Block embryos from light from this point forward.

  • 23. Wash embryos 2 times for 5 min each in TBST.
  • 24. Wash 3 times for 1 hr each in TBST.
  • 25. Wash overnight in TBST at 4°C on rocking table.
Day 5: Visualization for single FISH, deactivation of HRP, and antibody incubation for double FISH
  • 26. When single FISH is performed, embryos are now photographed.
  • 27. When performing double FISH, deactivate coupled HRP by incubating embryos with 100 mM glycine-HCl (pH=2.0) for 2.5 hr on ice.

Note: This step can be lengthened if the HRP is not completely deactivated (i.e., inappropriate cross-talk between channels when visualized through a fluorescence microscope). Alternatively, develop the gene with the weaker signal first, as the activity of the HRP will be easier to dampen.

  • 28. Wash embryos 2 times for 5 min each in TBST at RT.
  • 29. Wash embryos 3 times for 30 min each in TBST at RT.

Repeat steps 15 through 25. However, add alternate anti-hapten antibody and alternate tyramide.

Chromogenic in situ hybridization

Embryos are treated the same as for FISH for the first 3 days of the procedure, however use anti-DIG-AP for the overnight antibody incubation. Development of the chromogenic signal on day four is performed as follows. Embryos are washed 3 times for 10 min each in NTMT. BM purple is then added to the embryos, which are then blocked from light. Signal is developed for several hours until the staining pattern is sufficiently dark, or background is increasing. Embryos are then washed several times in TBST.


For bright field or epifluorescent images, embryos were photographed on solidified 2% agar in PBS. Chromogenic ISH images were taken with a Leica MZ12.5 stereomicroscope using associated Leica software, with no image processing. Fluorescence in situ hybridization images were taken with a Leica MZFIII stereomicroscope with HQF 41001 (FITC) and HQ-R NBX (Cy3) filters. Images were taken in grey-scale using Qcapture imaging software. Brightness and/or contrast of the entire image was altered using Adobe Photoshop if deemed appropriate, followed by false-colorization. For double FISH, images were merged using ImageJ software.

For confocal microscopy, specimens were embedded in 1% low-melt agarose in PBS.

Optical sections were taken at 40× using a Zeiss LSM-510 META confocal microscope with associated Zeiss software, and no post-acquisition modifications were made to the images.


We thank T.K. Lapointe and C. Altier for assistance with confocal microscopy. This work was sponsored by Natural Sciences and Engineering Research Council of Canada (NSERC) grants RGPIN/355731-2008 to J.C. and 288142-2012 to P.D.V.