A technical review and guide to RNA fluorescence in situ hybridization

Tissue preparation and permeabilization

Tissue preparation is one of the most critical aspects of a FISH protocol. Tissue preparation typically comprises both fixation and tissue permeabilization, and the balance of these is important in determining the degree of probe penetration as well as the morphological integrity of the sample. Prior to fixation, and critical for some species and sample types while less important for others, is the issue of relaxation of the sample of interest; a clear FISH signal can be obscured or rendered uninterpretable if it is concealed by a contracted morphology. Muscle relaxants are extremely species-specific and beyond the scope of this review, however an adequately relaxed tissue preparation (especially for whole-mounts) will make the visualization and interpretation of any signal significantly easier. We encourage the reader to survey the literature for appropriate relaxants for their species of interest. The most common fixatives are 4% formaldehyde or paraformaldehyde (PFA) in phosphate buffered saline (PBS; Nakamura, Nakamura & Hamada, 2013; Neufeld et al., 2013; Kernohan & Bérubé, 2014; Shiura et al., 2014; Oka & Sato, 2015; Thiruketheeswaran, Kiehl & D’Haese, 2016). Formaldehyde is a crosslinking fixative that forms covalent links between macromolecules such as lipids, peptides and DNA; this creates a mesh inside the cells or tissues to hold their components in place and minimize enzymatic degradation over time (Eltoum et al., 2001). PFA solutions produced from a powder will contain pure fixative, however, prepared 4% PFA solutions will produce polymers over time and become less effective as the polymers precipitate from the solution (Thavarajah et al., 2012). Thus, PFA solutions should be made fresh for each experiment. Alternatively, commercial formalin contains 37% monomeric formaldehyde in water and is supplemented with 10% methanol as a stabilizer to prevent polymer formation. Thus, a 1:10 dilution of commercial formalin solution is a common substitute for 4% PFA that does not require fresh preparation for each experiment (Thavarajah et al., 2012).

Fixation protocols are generally consistent among cytological, histological and whole-mount preparations, although whole mounts generally require longer treatments to ensure complete penetration of the fixative. Fixation protocols often consist of a treatment with 4% PFA or formaldehyde in PBS for varied lengths of time and temperatures (Table S1). The following examples, and the link between sample size and density (larger and more dense samples need longer fixation) can provide some scope when estimating a fixation duration for other sample types. Optimal fixation of planarian worms is achieved with 4% formaldehyde for 20 min (Pearson et al., 2009; Rink, Vu & Alvarado, 2011). For bacterial species or eukaryotic cells, 4% PFA is used to fix cells for as little as 10 min or as much as 90 min (Shaffer et al., 2013; Skinner et al., 2013; Chen et al., 2015; Wang et al., 2015; Aistleitner et al., 2018; Cardinale et al., 2018; Rocha, Almeida & Azevedo, 2018). Fruit fly (Drosophila melanogaster) embryos are typically fixed in 4% PFA for 20–30 min (Hauptmann et al., 2016; Jandura et al., 2017; Little & Gregor, 2018; Szabo et al., 2018). Zebrafish (Danio rerio) embryos and the annelid Platynereis dumerilii can be suitably fixed in 4% PFA for 2 h at room temperature (Jékely & Arendt, 2007; Steinmetz et al., 2011) but can alternatively be fixed overnight at 4 °C (Oxtoby & Jowett, 1993; Lauter, Söll & Hauptmann, 2011a, 2011b; Marra et al., 2017). Arms of the brittle star (Amphiura filiformis) are also sufficiently fixed in 4% PFA overnight at 4 °C. Embryos of the brachiopods Terebratalia transversa and Novocrania anomala should be fixed in 4% formaldehyde for 4 h (Schiemann et al., 2017; Gąsiorowski & Hejnol, 2019). The starlet sea anemone (Nematostella vectensis) and an acoelomorph worm (Convolutriloba longifissura) have been successfully fixed for ISH with 3.7% formaldehyde supplemented with 0.3% glutaraldehyde which is another strong cross-linking agent (Finnerty et al., 2003; Martindale, Pang & Finnerty, 2004; Hejnol & Martindale, 2008). Note, however, that glutaraldehyde is known to increase autofluorescence, at least with immunohistochemistry protocols. Whole mouse brains are often fixed in 4% PFA for up to 6 h at room temperature or overnight at 4 °C, although fixation of brain tissue is recommended not to exceed 24 h (Kernohan & Bérubé, 2014; Kasai et al., 2016; Lanfranco et al., 2017; Hua et al., 2018).

As an alternative to formaldehyde, some protocols employ alcohol-based fixation using either ethanol (Schurter, LeBrun & Harrison, 2002) or methanol (Legendre et al., 2013). Ethanol and methanol are coagulant fixatives that replace free water in the tissue to dehydrate cells and destabilize hydrophobic and hydrogen bonds (Eltoum et al., 2001). Alcohol-based fixation is common for cultured cells and ice-cold (−20 °C) ethanol and methanol have been used to fix multiple cultured cell lines in as little as 10 min (Shaffer et al., 2013). To fix tissue sections or whole-mounts, alcohol is commonly combined with other fixatives such as formaldehyde (Finnerty et al., 2003; Martindale, Pang & Finnerty, 2004; Hejnol & Martindale, 2008; Pearson et al., 2009). Although methanol has been used successfully with immunofluorescence (Levitt & King, 1987), methanol has a propensity to disrupt native protein structure and is generally not recommended for use in multiplex FISH and immunohistochemistry (Fowler et al., 2011). Methanol will strip membrane lipids to improve permeability (Hoetelmans et al., 2001) and ethanol can strip the external wax and lipids from plant tissues (Bleckmann & Dresselhaus, 2016). Thus, if cross-linking is also desired, formalin may improve tissue permeability over PFA due to the added methanol. For example, fixative solutions that contain alcohol and formaldehyde improve permeability in gram-positive bacterial preparations and may retain higher DNA quality in cytological preparations (Manz et al., 1994; Shaffer et al., 2013).

Beyond the choice of fixative, fixation temperature can also have a substantial impact on the final tissue quality (Fox et al., 1985; Thavarajah et al., 2012). With the use of formaldehyde, heat can accelerate the fixation process; although heat also increases the release of formaldehyde fumes which are hazardous to human health (Fox et al., 1985; Titford, 2001). Additionally, heat can denature proteins and cause a loss of antigenicity which would negatively affect multiplex FISH and immunohistochemistry (Fowler et al., 2011). For nucleic acid visualization, reduced temperatures of 4 °C have been shown to preserve RNA throughout the fixation process (Bussolati et al., 2011). Additionally, ice-cold solutions of alcohol fixatives are recommended as the reduced temperatures will reduce the risk of over-permeabilization and subsequent leakage of target molecules.

Following fixation, samples are generally permeabilized to allow for proper penetration of hybridization reagents. Detergent treatment of fixed tissue is commonly employed at a concentration of 0.1% as it substantially improves permeability of the tissues via disruption of cellular membranes. The use of Tween-20 is common but other detergents including sodium dodecyl sulfate and Triton X-100 can also be used. The detergent 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) is commonly used as an additive detergent to multiplex FISH with immunofluorescence as it effectively protects the native structure of proteins (Meyer, Garzia & Tuschl, 2017; Sepsi et al., 2018). Whole-mount preparations generally require stronger detergent treatments compared to cytological preparations or sectioned tissue, thus, a more aggressive detergent treatment such as 4% Triton X-100 can be effective in whole-mounts (Croll et al., 1999).

Treatment with a nonspecific protease such as proteinase K will permeabilize the tissues after fixation and can also release target nucleic acid molecules from bound proteins (such as RNA binding proteins), making them more accessible for hybridization. There is generally an inverse relationship between fixation time and the strength of the proteinase treatment as more highly fixed tissues will require a stronger protease digestion to become permeable to the probe. A protease treatment is not always necessary for bacterial or eukaryotic cells as a detergent is usually sufficient, however, a brief treatment with a dilute solution of proteinase K should be considered if probe penetration is the suspected cause of an issue (Carr et al., 2005). Zebrafish embryos are treated with 10 µg/mL proteinase K in PBST for 2–20 min depending on the age (Oxtoby & Jowett, 1993; Marra et al., 2017). The same treatment is also recommended for snail embryos as well as whole-mount planarian worms and is sometimes applied to fruit fly embryos, although several other permeabilization strategies including acetone are also frequently used for Drosophila (Paré et al., 2009; Pearson et al., 2009; Jackson, Herlitze & Hohagen, 2016; Hauptmann et al., 2016; Trcek et al., 2017). Some protocols call for brain sections to be treated with proteinase K, however, many protocols omit this step as permeability is less of an issue with sectioned material (Kasai et al., 2016; Hua et al., 2018). The proteinase K treatment will require careful optimization as too little digestion will prevent probe penetration whereas too much digestion will destroy the morphology of the tissue and lead to increased background (Tessmar-Raible et al., 2005; Bleckmann & Dresselhaus, 2016). As the degree of permeabilization with proteinase K can be a critical factor in the success of a FISH experiment, we recommend the use of accurately and consistently assayed batches of proteinase K enzyme such as supplied by New England Biolabs (Catalog: P8107S). As an alternative to proteinase K, pepsin has also been used to achieve more mild digestion of the tissue. Pepsin is preferred for cultured cells (Buxbaum, Wu & Singer, 2014) and tissue sections (Moorman et al., 2001; Teng et al., 2017) but potentially could be adapted to whole embryos. A treatment of 1 mg/mL pepsin in 0.01 N HCl is a common treatment, although the treatment length varies from 30 s to 10 min depending on the sample type (Moorman et al., 2001; Buxbaum, Wu & Singer, 2014; Teng et al., 2017).

Further permeabilization treatments are available as an alternative or an addition to protease treatments. A treatment of 1 M HCl at 37 °C for 30–50 min is effective to improve permeability of mycolic-acid-containing bacterial cells whereas other bacteria (including Escherichia coli) can be permeabilized in only 10 min (Macnaughton, O’Donnell & Embley, 1994). The addition of Triton X-100 or other detergent directly to the fixative in the initial fixation protocol has also been used to improve the permeability of bacterial cells through its interaction with cell envelope lipid molecules (Jackson, Herlitze & Hohagen, 2016; Rocha, Almeida & Azevedo, 2018). Protease-free detergent-based methods have also been successful for permeabilization of Drosophila embryos (Boettiger & Levine, 2013). Zebrafish embryos that are stored in methanol can be treated with 2% H₂O₂ for 20 min at room temperature to improve permeability (Lauter, Söll & Hauptmann, 2011b). This H₂O₂ treatment can also quench endogenous peroxidase activity and bleach tissues to reduce background in horseradish peroxidase-based assays (Marra et al., 2017). Organic solvents such as acetone have been used as an alternative to protease digestion of fragile embryos, and this method can also retain antigenicity for immunohistochemistry (Nagaso et al., 2001). In the preparation of whole-mounts with particularly tough integument, a digestion with 0.25% collagenase can be incorporated to improve permeability of dermal layers (Wyeth & Croll, 2011). Ultimately, careful optimization of the balance between fixation (strength, length and temperature thereof) and a proteinase based permeabilization is necessary to achieve a consistently high signal to noise ratio.

Hybridization

For efficient and complete hybridization of probe to target, the optimal environment must be provided. The hybridization reaction can contain an array of different components (Table S1). In addition to the tissue, most documented hybridization solutions comprise a saline-sodium citrate buffer (SSC) with formamide, vanadyl-ribonucleoside complex (VRC), dextran sulfate, bovine serum albumin (BSA), competitor tRNA or DNA, and the probe (Pinkel et al., 1988; Singer, 1998; Shaffer et al., 2013; Kernohan & Bérubé, 2014; Oka & Sato, 2015). Alternative components include Denhardt’s solution, ethylenediaminetetraacetic acid (EDTA), and Tween-20 (Langenbacher et al., 2015; Parker et al., 2019). In addition to the recipe of the hybridization solution, there are several reaction conditions that must be considered, including salt concentration, pH, and the temperature and duration of the hybridization reaction.

Formamide reduces the free energy of binding of nucleic acid strands to allow hybridization to take place at lower temperatures without a loss in specificity, thus improving structural preservation of the tissue (McConaughy, Laird & McCarthy, 1969; Bauman et al., 1980; Blake & Delcourt, 1996; Fontenete et al., 2016). As formamide stabilizes free bases and single-stranded DNA in solution, the melting temperature of DNA is decreased in a linear fashion by 2.4–2.9 °C per mole of formamide in the hybridization buffer (Blake & Delcourt, 1996). Formamide generally composes between 10% and 50% of the final volume of the hybridization buffer, but this range may be exceeded under specific circumstances (Table S1). Formamide is a toxic substance and, therefore, proper safety precautions must be made to avoid inhalation and direct contact with formamide (Warheit et al., 1989). Protocols that use safer alternatives to formamide, such as urea (Sinigaglia et al., 2018) have been developed but have yet to gain popularity (Volpi, 2017).

Vanadyl-ribonucleoside complex is an RNase inhibitor that is used to protect RNA-based probes or targets from enzymatic degradation (Berger & Birkenmeier, 1979; Frazier & Champney, 2012). VRC is typically added to the hybridization buffer at a final concentration of 10 mM as a precautionary measure. VRC is not compatible with solutions that contain EDTA as an equimolar concentration of a chelating agent will sequester the cations required for proper VRC function (Puskas et al., 1982). An RNase inhibitor is not absolutely necessary for successful ISH, but one should be considered if RNase contamination is a suspected problem.

Dextran sulfate is an anhydroglucose polymer that absorbs water molecules to reduce the free water in the reaction. This forces the probe and the target closer together, an effect referred to as molecular crowding, which enhances the rate of hybridization of the probe to the target (Lederman, Kawasaki & Szabo, 1981). Dextran sulfate can also improve fluorescent signals (Van Gijlswijk et al., 1996; Franks et al., 1998). Dextran sulfate is a synthetic analog of heparin which can also be used in the hybridization buffer and has also been reported to reduce background signal (Singh & Jones, 1984). Dextran sulfate is most often employed at a concentration of 50–100 mg/mL (Table S1; Singer & Ward, 1982; Oka & Sato, 2015; Parker et al., 2019).

Bovine serum albumin is used as a blocking agent to reduce background signal and thus improve the contrast of the probe (Choo, 2008). BSA blocks nonspecific binding of probe molecules to nucleic acid binding sites on proteins within the tissue as it can saturate the binding sites prior to the introduction of the probe. The use of BSA as a blocking agent may be especially important when using antibody-based detection methods. BSA is generally used at a concentration of 1 mg/mL (Thiruketheeswaran, Kiehl & D’Haese, 2016) up to 10 mg/mL (Singer & Ward, 1982).

Finally, sheared salmon sperm DNA or tRNA from E. coli or yeast is usually included in the hybridization buffer. The purpose of competitive nucleic acids is also to saturate nonspecific binding sites for probes to reduce background. Additionally, the competitor tRNA may protect target mRNA molecules via nonspecific blocking of RNase molecules that may have contaminated the solution. The optimal concentration of tRNA within the hybridization buffer should be empirically determined as it may vary widely depending on the tissue sample and the probe (Table S1; Langenbacher et al., 2015; Liu et al., 2019).

There are several alternative hybridization buffer components that can be used to facilitate an optimal hybridization environment. Denhardt’s solution is a broad blocking reagent composed of BSA, Ficoll type 400 and polyvinylpyrrolidone that can be used in place of BSA alone. EDTA is a chelating agent that can be added to a final concentration of 10 mM to remove free divalent ions such as magnesium. As EDTA can inactivate the VRC, these components are mutually exclusive.

When the reagent recipe has been established to create a supportive hybridization solution, the hybridization conditions must also be determined to facilitate optimal hybridization. We believe attention should be first given to the following parameters regarding hybridization: salt concentration, pH, hybridization temperature and duration of hybridization. Optimal hybridization will occur under conditions that allow the hybridization of the probe to the target but prevent the formation of nonspecific hybrids. Conditions that promote the sole formation of highly stable hybrids are known as highly stringent conditions whereas more permissive conditions that may allow the formation of nonspecific hybrids are considered less stringent. The stringency of the hybridization is affected by the concentration of salt in the hybridization solution (lower concentrations are more stringent) as well as the hybridization temperature (higher temperatures are more stringent). It is most common to keep the salt concentration constant (750 mM NaCl, 87.5 mM sodium citrate), with pH roughly between 7.0 and 8.5, and simply adjust the hybridization temperature to achieve the ideal stringency (Pearson et al., 2009; Zhang et al., 2012; Jackson, Herlitze & Hohagen, 2016). An initial denaturation step of 75 °C for 10 min can be used to denature all target and probe RNA to facilitate hybridization, the sample is then immediately adjusted to the designated hybridization temperature (Jékely & Arendt, 2007; Jackson, Herlitze & Hohagen, 2016). The optimal hybridization temperature is dependent on the length and composition of the probe. Although the hybridization temperature should be empirically optimized for every probe individually, short oligonucleotide probes (20–50 nucleotides) typically require lower hybridization temperatures of 37 °C whereas longer riboprobes of 1,000+ nucleotides may hybridize at temperatures >55 °C (Pearson et al., 2009; Jackson, Herlitze & Hohagen, 2016; Fontenete et al., 2016). Generally, the hybridization step cannot be over-incubated. Thus, an extended hybridization should be performed to allow probes to completely occupy available targets. Most often, 12–24 h is sufficient, regardless of the probe type (Carleton et al., 2014; Jackson, Herlitze & Hohagen, 2016; Meyer, Garzia & Tuschl, 2017; Jandura et al., 2017). Rapid hybridization has been achieved in cultured cells in as little as five minutes with the Turbo FISH method (Shaffer et al., 2013), but this is not a prudent point of entry for new protocols, especially for whole-mount material. Ultimately, salt concentration, hybridization temperature, and hybridization duration can be adjusted to create the optimal hybridization conditions with enough stringency to exclude non-specific labeling.

Post-hybridization treatments

The purpose of the post-hybridization washes is to separate nonspecific hybrids and remove unbound probe molecules from the tissue to minimize background signal. Samples are typically subjected to increasingly stringent washes in SSC buffer containing formamide and a detergent (Table S1; Jackson, Herlitze & Hohagen, 2016; Thiruketheeswaran, Kiehl & D’Haese, 2016). Increased stringency can be achieved through sequential washes with incrementally reduced salt concentrations while the wash temperature is matched to the hybridization temperature (Martindale, Pang & Finnerty, 2004; Hejnol & Martindale, 2008; Jackson, Herlitze & Hohagen, 2016; Schiemann et al., 2017; Gąsiorowski & Hejnol, 2019). At the end of washing, the goal is to allow only the specific and stable hybrids to remain. A wash progression that finishes with a higher concentration of salt (or at a lower temperature, that is, lower stringency) will be less likely to denature and remove nonspecific hybrids, but also may preserve greater intensity of specific labeling.

In addition to nonspecific hybrids, autofluorescence and excessive background are issues that can diminish the visibility of true signal and influence the interpretation of the results. Treatment with 0.1% Sudan Black B in 70% ethanol is effective to minimize autofluorescence in sectioned brain tissue as well as cultured cells (Oliveira et al., 2010; Qi et al., 2017). If background signal is an issue, tissues can also be acetylated with 0.3% acetic anhydride in triethanolamine for 5–10 min (Jackson, Herlitze & Hohagen, 2016). This acetylation blocks positively charged proteins and amine groups (exposed during enzymatic permeabilization) in the tissue that could otherwise engage in electrostatic interactions with negatively charged probes.

The final process prior to visualization of results is tissue clearing to prevent lateral light scattering within the tissue (Richardson & Lichtman, 2015). Common methods of tissue clearing may involve either dehydration or hyperhydration of the tissue sample. An organic solvent-based method of clearing via a two-to-one mixture of benzyl benzoate and benzyl alcohol has been successfully used to visualize whole snail embryos (Jackson, Herlitze & Hohagen, 2016), however, the tissue must first be dehydrated with a graded series of ethanol. One potential issue with solvent-based clearing is that the dehydration process can cause substantial shrinkage of tissues (Richardson & Lichtman, 2015). Other methods of clearing that involve hyperhydration include the formamide-based ClearT (Kuwajima et al., 2013) as well as the urea-based CUBIC (Susaki et al., 2015; Tainaka et al., 2014). Methods of hyperhydration often involve large quantities of detergent and are most suitable when it is desirable to remove the majority of lipids from the tissue sample. A more advanced method of tissue clearing involves the use of anchor probes to fix the hybrids within a polymer matrix with subsequent digestion of non-RNA material (Moffitt et al., 2016a), however, this technique is most suitable for highly multiplexed FISH experiments.