Analysis of messenger RNA expression by in situ hybridization using RNA probes synthesized via in vitro transcription

Abstract

The analysis of the spatial patterning of mRNA expression is critically important for assigning functional and physiological significance to a given gene product. Given the tens of thousands of mRNAs in the mammalian genome, a full assessment of individual gene functions would ideally be overlaid upon knowledge of the specific cell types expressing each mRNA. In situ hybridization approaches represent a molecular biological/histological method that can reveal cellular patterns of mRNA expression. Here, we present detailed procedures for the detection of specific mRNAs using radioactive RNA probes in tissue sections followed by autoradiographic detection. These methods allow for the specific and sensitive detection of spatial patterns of mRNA expression, thereby linking mRNA expression with cell type and function. Radioactive detection methods also facilitate semi-quantitative analyses of changes in mRNA gene expression.

1. Introduction

The biological processes involved in transmitting genetic information are of great interest to a diverse range of biomedical scientists. One aspect of particular interest is gene expression, the process of producing messenger RNA that can then be translated into proteins. Messenger RNAs (mRNAs) are transcribed from genomic DNA in the nucleus. Double-stranded DNA serves as the sequence source; the strand that is used during RNA transcription is known as the template strand. Single-stranded RNA is subsequently synthesized by an RNA polymerase, and the resulting RNA is complementary to the DNA template strand. This full-length RNA form (termed heteronuclear RNA or hnRNA) ranges from hundreds to thousands of nucleotides in length. The final mRNA form is created from this longer version by spliceosome machinery, which removes intron sequences and links exon sequences together. Following splicing, mRNAs are then exported into the cytoplasm where ribosomes can bind with the mRNAs. Ribosomes transfer the information in the mRNA nucleotide sequence into a protein product via translation. This process of converting information from DNA to RNA to protein is known as “the central dogma of biology” and enables functionality on the cellular scale [1].

1.1 Gene expression analysis and need for correlated anatomical data

Gene expression studies are an important element of biomedical research because (1) gene expression patterns dictate cell identity and (2) the complement of genes expressed by any cell can change and is subject to extensive regulation. Global gene expression heavily influences cell specification; for instance, the genes a brain cell expresses make it a brain cell and not a fat cell or liver cell. In the brain, gene expression is exceptionally heterogeneous; certain mRNAs are enriched or only expressed in specific brain regions and often within specific neurons or glial cells. These localizations implicate specific RNAs and their corresponding proteins in the functions associated with a given brain region or specific neuronal or glial cell type within this brain region. For example, within the hypothalamus (a main site of hormone regulation), several mRNAs are specifically expressed in different nuclei implicated in the regulation of circadian rhythms [2], appetite and feeding behavior [3], and stress axis regulation [4]. These mRNA can also serve as reliable markers of specific cell types associated with these nuclei. Similarly, mRNA expression patterns also vary between neurons, astrocytes, and oligodendrocytes [5]. In addition, transcription and the production of mRNA is a highly regulated process. Hence, the complement of mRNAs expressed by cells is dynamic, altered by time and both intrinsic as well as extrinsic factors. Some pertinent biological examples include developmental processes [6], aging [7], disease [8], and medical treatments [9]. While gene expression can be a highly regulated process, the total number of genes that are regulated can vary considerably and likely reflects a range of environmental factors, including the nature of the cellular stimuli, the duration of the stimuli, and the state of the cell receiving the stimuli.

The idea of “where” (i.e. in vivo cellular location) an mRNA molecule is expressed having major implications for its function is especially important for structures such as the brain, an anatomically complex organ with numerous functional roles associated with location and connectivity (e.g. neural circuits). Since alterations in mRNA levels are a basis of functional changes in a cellular environment, techniques that visualize the anatomical location of mRNA expression can define where a given mRNA is located and thus enable additional functional studies to understand cellular mechanisms of that location (e.g. within a given cell type or organ). With numerous researchers examining global gene expression changes using microarrays and now next-generation sequencing, there is a growing need to complement this genome-wide expression data with experimental techniques that address anatomical complexity.

1.2 In situ hybridization; overview of a useful tool for anatomical analysis

A useful tool developed to visualize the spatial expression patterns of mRNAs is in situ hybridization (ISH), a technique used to localize nucleic acid sequences in tissue sections or cell culture samples. ISH involves using a nucleotide sequence that is tagged or labeled with a detection molecule. ISH can be used to localize chromatin, DNA, or RNA; for this report, we are focusing on mRNA detection. In an ISH protocol, a tagged nucleotide sequence (i.e. “the probe”) complementary to the sequence of the mRNA to be detected is applied to the tissue or cell culture sample. During this application/incubation, the probe associates with the mRNA of interest via nucleic acid base-pair interactions, a process called hybridization. Once bound to this mRNA, the presence of the tag enables detection and subsequent visualization of the location of the mRNA of interest. Standard controls for these methods include using RNase pre-treatment (demonstrates that detection is dependent upon the presence of RNA) and opposite-strand nucleic acid sequences as probes (demonstrates sequence-specific nucleic acid strand dependency).

There are a number of factors to consider when designing ISH studies. Nucleic acid probes for ISH can be made of either DNA or RNA molecules (detailed in [10–15]). Both of these approaches require careful selection of the specific probe sequence for the targeted mRNA because some segments of mRNAs derived from single copy genes can contain sequence domains that are highly homologous across gene families, leading to potential cross-hybridization and non-specific hybridization results (elaborated in section 2.1). In the early days of ISH methodology, DNA probes were the predominant form because of the ease with which DNA oligonucleotides could be synthesized. With the continued development of methodologies involving RNA polymerases, the in vitro synthesis of RNA probes (often referred to as complementary RNA or cRNA) are now the predominant form for a couple reasons. First, cRNA probes can be easily synthesized from plasmid vectors containing genes of interest and subsequent in vitro transcription (IVT); these techniques are now commonplace molecular biological methods (see [13]). Second, cRNA-mRNA hybridization complexes are more stable than comparable cDNA-mRNA complexes, permitting the experimenter to use stringent hybridization conditions specifically conducive to the formation of targeted probe-mRNA complexes. Finally, excess/non-hybridized cRNA probes can be easily digested from experimental preparations via RNase, thereby lowering non-specific signal. Overall, both DNA and cRNA probes can be used with different detection methods chosen based on the hypothesis or question of interest.

There are two main categories of ISH detection, non-radioactive detection methods and radioactive detection methods. In the following sections, we present an overview of the various options along with their individual strengths and weaknesses.

1.3 In situ hybridization: non-radioactive detection methods

Non-radioactive detection methods are well-suited for answering many research questions related to anatomical localization, co-localization and cellular compartmentalization or subcellular localization. Most non-radioactive methods make used of labeled nucleotide analogs that enable detection either directly (e.g. fluorescence) or contain haptens (e.g. biotin, digoxigenin) that can be subsequently detected using antibodies conjugated to a detection entity (e.g. fluorescent dye, HRP/AP enzyme). Hybridization signals using HRP enzymes can also be enhanced using tyramide amplification systems (TSA, review [16]). Non-radioactive ISH approaches are particularly advantageous for whole mount preps of translucent tissue (e.g. zebrafish, embryos). Also, fluorescence-based detection methods are amenable to confocal microscopy, which allows refined 3-D resolution of hybridization signals compared to the surface-based detection of other techniques (e.g. autoradiography). Along with these beneficial features, there are specific challenges associated with using these methods as well. Non-radioactive ISH methods routinely employ high nucleic acid probe concentrations that can increase non-specific nucleic acid interactions, and hybridization signals derived from these methods are typically generated using non-linear detection methods (e.g. enzymatic detection methods). Non-radioactive ISH methods can and have been used for semi-quantitative analysis (see examples [17–19]). However, the dynamic range of non-radioactive methods is often thought to be less than radioactive methods due in part to the non-linear enzymatic amplification methods necessary for visualizing low expressing transcripts. Non-radioactive ISH quantitative measurements can be carried out using computer software (e.g. Adobe Photoshop) and/or image analysis platforms (e.g. Ventana VIAS system). When quantifying non-radioactive ISH signals, one also needs to ensure that measurements collected are within a linear range of detection (e.g. use of standard curve [19]). These complexities can significantly limit semi-quantitative assessments of RNA amounts (e.g. differential gene expression). However, non-radioactive detection methods do not require laboratories to deal with radioactivity (e.g. potential laboratory safety and waste disposal challenges) or radioactive detection and analysis methods (e.g. autoradiography including extensive exposure times and densitometry that can require additional software analytical tools).

1.4 In situ hybridization: radioactive detection methods

Radioactive detection methods are well-suited for answering questions relating to differential gene expression because the techniques can yield semi-quantitative measurements of mRNA amounts. Since there can be a linear relationship between hybridization signal detected (via autoradiography and subsequent densimetric analysis) and mRNA levels, one can derive relative comparisons of mRNA levels across samples. In our hands, we also find ISH radioactive cRNA probes to be more sensitive than some non-radioactive cRNA probes even for the same gene/mRNA sequence (i.e. some mRNAs cannot be readily detected when using non-radioactive probes while radioactive probes yield positive and specific hybridization patterns) (Thompson, unpublished). There are likely many reasons for such differences. A short list of such explanations include the absolute copy number of the mRNA per cell to be detected, the absolute robustness of the utilized detection method, and the ability to distinguish authentic hybridization signal from non-specific hybridization signal (i.e. background signal).

Additional components in designing radioactive ISH cRNA probes include choosing the type of radioactive isotope to be used. Factors to be considered and balanced include radionucleotide costs, isotope half-life, and quality of autoradiographic detection (both resolution and sensitivity) (see Table 1).

TABLE 1.

Half-life and resolution range of common radioactive isotopes [20]

Isotope	Half-life	Resolution*
3H	12.4 years	1–5 µm
32P	14.3 days	20–30 µm
33P	25.4 days	15–20 µm
35S	87.4 days	10–15 µm

^*Other authors have reported higher levels of resolution for ³H [21]

Common isotope choices include ³H, ³²P, ³³P, or ³⁵S molecules conjugated to ribonucleotides. Although radioactivity has inherent safety dangers, these risks are relatively low and vary only modestly across the mentioned isotope types. There are advantages and disadvantages in using each of these isotopes. Initial radioactive ISH methods used tritium-labeled nucleic acid probes (³H) based largely upon commercially availability [22]. ³H has largely been replaced by other isotopes for most ISH applications because it is by comparison an energetically weak isotope. ³H-based methods can be useful in that the isotope has an extensive half-life and can yield high autoradiographic resolution results because of the confined energetic emission, although the detection of ³H can require very long exposure times (e.g. months). ³²P-labeled nucleotides are substantially more energetic than other commonly used radioactive isotopes, which often leads to improved sensitivity (see Table 2). Unfortunately, nucleic acids labeled with high energy isotopes such as ³²P have short half-lives and often yield more diffuse autoradiographic signals due to the broader area of silver grains (e.g. on x-ray film, photographic emulsion) or pixels (e.g. on phosphorimaging plates) developed in the detection process [23]. ³³P-labeled nucleotides are intermediate choices in terms of cRNA probe energetics; ³³P-labeled probes have extended half-lives and yield improved anatomical resolution as compared to hybridization results using ³²P-labeled cRNA probes. However, ³³P-labeled nucleotides are often the most expensive to purchase. Both ³²P and ³³P can be attached to ATP as γ-labeled nucleotides; this arrangement makes them amenable to end-labeling of synthetic nucleic acid probes via kinase reactions, a convenient methodology for some ISH applications (sample [24]). ³⁵S-labeled nucleotides have isotope energetics that are similar to ³³P-labeled nucleotides, have a dramatically longer half-life, have still better anatomical resolution properties, and are often more affordable than ³³P. Longer half-life cRNA probes permit extensive autoradiographic exposure times (e.g. weeks to months) to detect weaker hybridization signals with fewer concerns about radioactive decay. Unfortunately, ³⁵S-labeled nucleotides are only commercially available as α-labeled nucleotides and are thus not compatible with nucleic acid end-labeling strategies. In ³⁵S ISH applications, ³⁵S-αUTP, ³⁵S-αCTP, or a combination of the two (sometimes referred to as double-labeled cRNA) is commonly used for probe synthesis using in vitro transcription reactions. For mRNA ISH applications, we routinely favor ³⁵S-based labeling for radioactive detection applications because of the favorable characteristics mentioned (i.e. long half life, reasonable energetics, affordability).

TABLE 2.

Comparisons of radioisotope energetics [20]

Isotope Comparison	Energetic Fold-Difference
33P/35S	1.50
32P/33P	6.84
35S/3H	9.28
32P/35S	10.24

Comparisons between the emission energetics of commonly utilized radioisotopes. The first column (Isotope Comparison) notes specific comparisons made and the calculated ratios are shown in the second column (Energetic Fold-Difference). While radioisotope choice is only one important ISH experimental design consideration, such energetic comparisons highlight why ³²P-labeled probes are often thought to yield the most robust detection (i.e. most sensitive) when compared to either ³⁵S-labeled ISH probes (10.24x more energetic) or ³³P-labeled ISH probes (6.84x more energetic).

As mentioned previously, ISH autoradiographic data varies by radioactive isotope, which influences both the optimal autoradiographic exposure time and the resolution of the autoradiographic image. Higher-energy isotopes generally yield desired signal levels in shorter time frames than lower-energy isotopes, allowing the experimenter to determine if an ISH hybridization study yielded useful results in less time. However, higher-energy isotopes can achieve lower anatomical/cellular resolution based on the spread of energy emission onto the photographic material. Since the silver grains in film or emulsion are in contact with the sample at a fixed distance from the radiolabeled probe within the tissue, point sources of higher-energy isotopes tend to expose more silver grains at greater distances compared to lower-energy isotopes. Nucleic acid probes labeled with high-energy isotopes can sometimes yield less well-defined anatomical patterns of gene expression when compared to lower-energy isotopes (e.g. fuzzier edges to cellular mRNA expression patterns). In contrast, nucleic acid probes labeled with lower-energy isotopes expose silver grains that are more proximal to the actual hybridization event. Isotopes should be chosen based on the greatest priorities for a given experiment (e.g. weight of time vs. resolution). Further details on the differences between the signals of various isotopes have been described elsewhere [25].

1.5 Summary

There are different types of detection molecules that can be used; each type contains attractive characteristics as well as challenges or shortcomings, and researchers should choose the appropriate detection methodology based on their specific application (Table 3).

TABLE 3.

Different detection methodologies for in situ hybridization

Category	Detected substance	Detection method	Citation Example
Non-radioactive	Biotin	Colorimetry/fluorescence/chemiluminescence	[26]
	Digoxigenin	Colorimetry/fluorescence/chemiluminescence	[27]
	Fluorescein	Fluorescence/colorimetry	[28]
	Fluorescent dyes	Fluorescence	[12]
	Metallography	Light microscopy/electron microscopy (e.g. Ag/Au)	[29]
Radioactive	3H-labeled nucleotides	Autoradiography	[22]
	32P-labeled nucleotides		[30]
	33P-labeled nucleotides		[24]
	35S-labeled nucleotides		[31]

ISH techniques can reveal the cellular and subcellular location of specific classes of nucleic acids. Given the anatomical complexity of the brain, ISH techniques offer neuroscientists the opportunity to overlay mRNA expression patterns upon neurocircuitry knowledge to further define molecular mechanisms of neural processes. However, the anatomical use of in situ hybridization to identify cells expressing specific nucleic acids has broad applications, such as developmental biology and organogenesis (e.g. cardiovascular system [32], respiratory system [33], bone growth [34], etc.). In our lab, we are interested in mRNA expression patterns and gene expression changes using tissues and experimental models relevant to psychiatric disease and drug abuse. We also routinely study mRNAs that are expressed at low levels within discrete populations of cells. Given these parameters, our choice of ISH methodology must be both highly sensitive and compatible with semi-quantitative measurements of mRNA levels. In our experience, radioactive detection methods are the most suitable option for our applications. In the following sections, we describe detailed methods for ISH detection of mRNAs using in vitro transcribed cRNA probes synthesized with ³⁵S-labeled nucleotides. As an example, we present ISH data revealing the specific location in rat brain tissue sections of tyrosine hydroxylase (TH) mRNA, the key rate limiting enzyme responsible for the synthesis of dopamine, a major neurotransmitter in the central nervous system.

2. RNA probe synthesis via in vitro transcription using ³⁵S-αUTP

2.1 Background: probe design

Targeting an mRNA for ISH experiments begins with obtaining a cRNA probe that is complementary to the target (i.e. antisense). Ideally this probe can be obtained from an available plasmid; if not, the desired probe must be obtained using reverse transcription and PCR methods to obtain the cDNA fragment of interest. Once a cDNA fragment is amplified, it can be cloned into an appropriate plasmid vector for subsequent ISH experiments (for detailed molecular cloning methods, see [13]). As mentioned in the introduction, sequence specificity is an important component of successful probe design. Use of nucleic acid sequences that are known to be similar across gene families can lead to non-specific in situ hybridization results. To address this issue, we recommend using probes targeting the 3’UTR domain of a target mRNA because 3’UTR domains of most eukaryotic mRNAs are typically less well conserved across gene families than protein coding domains and are thus more likely to yield RNA-type specific ISH results. The selection of a given sequence within any mRNA domain can be evaluated using sequence-database comparison software such as BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The goal with BLAST is to define a nucleic acid sequence that (1) perfectly matched to the mRNA of interest and (2) displays low levels of homology to any other known mRNA.

Yet another component to consider is nucleic acid fragment size or length; the longer the sequence, the more labeled nucleotides can be incorporated into the ISH nucleic acid probe, maximizing detection sensitivity. However, very large nucleic acid probes may not penetrate tissues/cells with the same efficiency as smaller nucleic acid probes. One way to address this caveat is to fragment a large probe into smaller probes. For instance, large cRNA probes (~>1000 nt) can be fragmented by limited base hydrolysis to yield nucleic probe sizes (200–400 nt) to increase cellular penetration [23]. We find that cRNA probes ranging from 300–500 nt in length often worked well and represent a blend of high probe sensitivity and sufficient tissue penetration without the need to fragment probes or alter tissue morphology (e.g. Proteinase K digestion, see section 4.6.1).

As noted previously, cRNA probes can be derived from plasmids containing nucleic acid fragments corresponding to the mRNA of interest (hereby referred to as the cDNA fragment). Once a plasmid is available, the major requirements here are that the experimenter know (1) the specific plasmid vector, (2) how the cDNA fragment is insert or cloned into this plasmid vector (including orientation), and (3) the precise sequence of the cloned cDNA fragment. It is assumed in the following sections that the experimenter knows all of these details. The preparation of the plasmid DNA can be accomplished using commonplace and commercially available molecular biology plasmid DNA kits. Once plasmid DNA is in hand, the experimenter can determine the plasmid DNA concentration using spectrophotometric measures. In sections below, we will refer to these DNA template samples as plasmid DNA.

2.2 Plasmid DNA restriction enzyme digestion

In vitro transcription requires that (1) the plasmid DNA be linearized with a restriction endonuclease such that subsequent transcription will terminate shortly after the desired endpoint of the cRNA (often the end of the cloned cDNA fragment) and (2) the appropriate RNA polymerase be selected to synthesize a cRNA fragment that is complementary to the target mRNA. Commercial plasmid vectors often have a series of unique restriction enzyme sites and two different RNA polymerase promoters that flank the inserted cDNA fragment. This arrangement allows both strands of the cDNA fragment to be synthesized into cRNA fragments. Knowledge of the orientation of the cDNA fragment within the plasmid DNA to be used provides the researcher sufficient information to generate the desired cRNA probe (e.g. sense-strand RNA probe versus antisense RNA probe). Care should be executed to ensure that the chosen restriction enzymes yield expected plasmid digestion patterns (e.g. no unexpected cuts within the insert sequence; digestion patterns can be analyzed using agarose gel electrophoresis to confirm expected fragment sizes). The orientation of the cDNA fragment within a plasmid DNA can be validated by DNA sequencing.

For a single transcription reaction, digest 2 µg of plasmid DNA per enzyme manufacturer’s protocol. Confirm that template digestion has been completed by analyzing an aliquot of the digest by 2% agarose gel electrophoresis (100 ng is usually sufficient for visualization). The visualized DNA fragment should confirm that the probe size is correct in comparison to a known standard (e.g. DNA ladder). Precipitate the plasmid DNA by standard ethanol precipitation and reconstitute in sterile water at a concentration of 1 µg/µL. Phenol-chloroform extraction can performed prior to ethanol precipitation to minimize contamination of the plasmid DNA, but we have found it unnecessary for most small-volume digests. Extensive description and discussion of restriction enzyme digestions and agarose gel characterization of DNA methods have been previously described [13].

2.3 In vitro transcription reaction

NOTE: In the specific reactions noted below, we have included specific reagents and suppliers commonly utilized in our laboratory, although there are likely other sources for many of these reagents that can be substituted with near or equal success.

Transcription of a ³⁵S-αUTP-labeled cRNA probe from linearized plasmid DNA can be prepared as follows:

To a 1.5 mL RNase-free microcentrifuge tube, add:

5x transcription buffer (Promega #P1181)	3 µL
Sterile water	2.2 µL
10mM ATP/CTP/GTP (Roche #111409)	1 µL each
0.1 M DTT (Promega #P1171)	1.5 µL
1 µg/µL template DNA	1.5 µL
12.5 µCi/µL 35S-αUTP (Perkin Elmer #NEG039H)	7 µL
RNase Inhibitor (Invitrogen #10777-019)	0.6 µL
RNA polymerase (typically T3, T7, or SP6) (Promega #P208, #P207, #P108)	1.2 µL

**Total volume = 20 µL

Mix gently by pipetting and briefly centrifuge to pool contents. Incubate at 37°C for 1.5–2 hrs. Add 1 µL RNase-free DNase (Roche 04716728001) to remove plasmid DNA. Mix gently and briefly centrifuge. Incubate at 37°C for 15 min. Isolate the radiolabeled cRNA probe from unincorporated ³⁵S-αUTP using Micro Bio-Spin 30 Chromatography Columns (Bio-Rad 732–6202) according to manufacturer’s instructions. Add DTT to a final concentration of 10mM (prevents oxidation of thiol-ester linkage in newly synthesized cRNA). The quantity of radioactivity incorporated in the cRNA probe is determined using a scintillation counter. Radiolabeled cRNA probe can be stored at −20°C (one to two days) or −80°C (up to one week) prior to use.

3. Preparation of solutions and reagents

NOTE: Unless otherwise stated, commonly available reagents can be used to prepare the following solutions needed to prepare and subsequently treat tissue sections. Commonly available reagents are referenced to Fisher Scientific products or their equivalent.

3.1

0.5 M NaPhos buffer, pH 7.4 (1 L)

NaH2PO4-H2O (monobasic; Fisher 5465-500)	13.8 g
Na2HPO4 (dibasic; Baker 3828-05)	56.8 g

Dissolve in one liter of filtered water.

Available from Fisher (#50843071, pH 7.0)

3.2

PFA solution (4% paraformaldehyde in 0.1M NaPhos buffer, 500mL)

CAUTION: perform all steps involving paraformaldehyde in a fume hood. Add 20 g PFA (Fisher #AC41678-5000) to 350 mL of filtered water. Heat to 60°C and mix with stirring. Add 10 N NaOH drop-wise and continue mixing at 60°C until the PFA dissolves completely. Add 100 mL 0.5 M NaPhos buffer to the PFA solution. Stop heating and stir until the solution cools. Adjust the pH to between 7.0 and 7.4 with concentrated HCl. Add filtered water to 500 mL. Filter the solution using Whatman Grade 2 Filter Paper. This solution can be stored at 4°C for up to 1 week.

NOTE: Our experience is exclusively with the use of freshly prepared PFA solution. 4% PFA solution can be commercially purchased (e.g. Fisher NC9245948); if purchased, long-term usefulness should be evaluated empirically.

3.3

20x SSC (1 L)

NaCl (Fisher 271-10)	175.3 g
NaCitrate (dihydrate; Fisher 279-3)	88.2 g

Dissolve in 800 mL of filtered water. Adjust the pH to 7.0. Adjust the volume to 1 liter with filtered water. Filter sterilize using a 0.2 µm filter.

Available from Fisher (#BP1325-1).

3.4

1 M triethanolamine, pH 8.0 (10x stock)

Add 66.5 mL triethanolamine (Fisher # AC42163-1000) to 400 mL of water. Adjust pH to 8.0 with concentrated HCl. Adjust volume to 500 mL with filtered water.

3.5

Acetic Anhydride

Available from Sigma (#A6404).

3.6

100x Denhardt’s solution (250 mL)

Bovine serum albumin (Sigma #A7206)	10 g
Polyvinylpyrrolidone (Sigma #P5288)	10 g
Ficoll 400 (Sigma #F4375)	10 g

Make a slurry in DEPC-treated water and then dilute to a final volume.

Filter sterilize and store in aliquots at −20°C.

Available from Fisher (50X, #NC9385183)

3.7

Yeast tRNA (50 mg/mL)

Available from Invitrogen (#15401-029). Dissolve 50 mg of yeast tRNA in 1 mL RNase-free water. Store at −20°C.

3.8

Hybridization Solution

In a 50 mL conical tube, combine:

Formamide (Sigma #F7503)	25 mL
20x SSC	7.5 mL
0.5 M NaPhos buffer, pH 7.4	5 mL
100x Denhardt’s solution	0.5 mL
50 mg/mL yeast tRNA	0.1 mL
Dextran Sulfate (Sigma #D8906)	5 g

Heat to 37°C and rotate in a water bath or shaking incubator to dissolve dextran sulfate. Add filtered water to 50 mL. Alternatively, we have found a commercial hybridization solution (Amresco #0973-50ML) to be an acceptable substitute.

3.9

RNase A Solution (200 µg/mL, 500 mL)

1 M Tris buffer, pH 8.0 (made from solid; Fisher T1503-1KG)	5 mL
5M NaCl	50 mL
10 mg/mL RNase A stock (Fisher #BP2539250)	10 mL
1 M EDTA (made from solid; Fisher BP120-500)	1 µL

Add filtered water to 500 mL.

3.10

Permount

Available from Fisher (#SP15-100).

3.11

0.5 % Cresyl violet solution, (500 mL)

In a 1 L flask, combine:

Cresyl violet acetate (Sigma C5042)	2.5 g
Filtered water	400 mL
Glacial acetic acid (Fisher A38)	1.5 mL

Add filtered water to 500 mL. Heat to 37°C and rotate in a water bath or shaking incubator overnight to dissolve cresyl violet. Filter using Whatman Grade 1 Filter Paper.

3.12

Ethanol

Available from Fisher (#A405P)

3.13

Xylenes

Available from Fisher (#X5)

4. mRNA ISH procedure

4.0 Theoretical considerations for mRNA ISH procedures

In our procedure, tissues should be cut with a cryostat onto standard microscope slides (e.g. 10–12 µm sections). A number of elements should be considered to promote a successful ISH result (reviewed in [25, 35–36]). Sections are ideally stored at −80°C to limit RNA degredation by RNAses. Alternatively, if no −80°C freezer is available, sections can be fixed immediately after cryostat sectioning and then stored at −20°C until use, although it is not known how long such tissue preparations are useful. If possible, this protocol should be done in an RNAse-free environment, and reagents should be carefully monitored to prevent RNAse contamination for subsequent experiments. Paraformaldehyde acts as a protein crosslinker, setting molecular components in place while retaining cell structure [37]. Acetic anhydride treatment acetylates proteins and affects the charge interactions of the nucleic acid probe and cellular proteins [38–39]. The negatively-charged acetyl groups become attached to positively-charged protein residues and thus limit the affinity of the negatively-charged nucleic acid probes to these residues, thereby decreasing nonspecific interactions. Given the small volume of hybridization solution applied to the tissue, drying the tissue with alcohols prior to hybridization prevents excess water from diluting the probe in an unpredictable manner (i.e. slide to slide variability). Following probe hybridization, RNAse treatment lowers nonspecific signal (i.e. background) by digesting single-stranded RNA, including unbound RNA probe as well as non-specifically hybridized RNA probe [23]. The high-stringency washes also attempt to lower background by removing residual RNA/RNA probe interactions that are not energetically strong enough (i.e. low sequence homology) to be maintained through a higher temperature wash. The final alcohol dehydration prevents water from interfering with hybridization signal detection since moisture exposes autoradiographic films. Detection methods rely on the interaction of the emission energy from the radioactive isotopes with the silver molecules in the photographic material, revealing the location of the target mRNA.

4.1 Day 1 protocol

4.1.0

Special equipment requirement: ISH methods require special glassware containers for placing slides in the various solutions. Some glassware containers hold the slides vertical with the sections oriented toward the bottom, such as Coplin jars (e.g. Fisher # 08-816); these containers require less solution to cover the sections. Containers holding slides in a vertical orientation can sometimes be a source of variable signal or background due to temperature gradients in the container (see section 4.6.3). Alternatively, other glassware containers hold the slides horizontal and immerse the entire slide in each solution, such as Wheaton cell staining containers (e.g. Fisher #08-812). We routinely use the Wheaton containers with removable racks since this arrangement enables us to prepare solutions in separate glass containers and directly transfer the rack of slides from solution to solution.

4.1.1

Sections are removed from the −80°C freezer and immediately placed into PFA solution for 1 hour at room temperature in a fume hood. While incubating slides in PFA solution, make 2x SSC (for step 4.1.2).

4.1.2

Wash slides three times in 2x SSC for 2 min each.

4.1.3

Wash slides in filtered water for 1 min.

4.1.4

Acetic anhydride treatment: make 0.1 M TEA by diluting 20 mL 1 M Triethanolamine (TEA) (pH 8.0) in 180 mL filtered water. In fume hood, add 0.5 mL acetic anhydride to 0.1 M TEA and make certain that acetic anhydride is well dispersed. Incubate slides in this solution for 10 min.

4.1.5

Wash slides two to three times in water for 1 min each.

4.1.6

Dehydrate slides in a graded series of 50–100% ethanol. Incubate slides in separate ethanol containers sequentially for 4 min each as follows: 50%, 75%, 95% (2x)100% (3x)

NOTE: repeated washes should be carried out using fresh solutions in separate containers

4.1.7

Hybridization: Per slide, dilute 1.0–2.0x10⁶ cpm of ³⁵S-labeled RNA probe into 40–70 µL of hybridization buffer (40 µL is typically sufficient for a 22 mm² coverslip). An IVT reaction as described above typically yields enough probe for 20–40+ slides using 22 mm² coverslips. Add 1 M DTT to a final concentration of 10 mM. Pipette solution onto coverslip. Invert slide with sections toward the coverslip and place slide onto coverslip to pick up the coverslip (cover/incubate the sections). Flip slide upright (i.e. cover slip on top) and place covered slide in a sealed, humidified chamber (e.g. place filter paper on the bottom of a Nunc sterile bioassay dish; soak the filter paper with 50% formamide and then place slides on plastic supports above filter paper; seal by wrapping in Saran wrap).

4.1.8

Hybridize overnight at 55°C in a laboratory oven.

4.2 Day 2 protocol

4.2.1

Using water baths, warm the RNase A solution to 37°C (for step 2.4) and 0.1x SSC solutions to 67°C (for step 2.6). Make 2x SSC (for steps 2.2, 2.3., and 2.5). Make 1x SSC, 0.5x SSC, and 0.1xSSC (for steps 2.5 and 2.7).

4.2.2

Remove coverslips by dipping slides in a Coplin jar containing 2x SSC. Place dipped slides into a separate container of 2x SSC until all coverslips have been removed. Note: use a gentle vertical motion to avoid damaging sections.

4.2.3

Wash slides two times in 2x SSC for 1 min each.

4.2.4

RNase A Treatment: Incubate slides in the pre-warmed 200 µg/mL RNase A solution for one hour at 37°C. If possible, RNAse treatment should be done in a separate lab location than other steps to limit RNAse contamination of other reagents. If RNAse treatment must be done in an adjacent area, keep the RNAse away from other reagents as much as possible (e.g. separate water bath, regular cleaning of area, use sterilizing products such as RNAse Away (Molecular BioProducts #7003)).

4.2.5

Wash slides in 2x SSC for 5 min, 1x SSC for 5 min, 0.5x SSC for 5 min, and twice in 0.1x SSC for 5 min each.

4.2.6

High-stringency wash: Wash slides two times in 0.1x SSC for 30 min at 67°C.

4.2.7

Wash slides in room-temperature 0.1x SSC for 5 min.

4.2.8

Dehydrate slides in a graded series of 50–100% ethanol. Incubate slides in separate ethanol containers sequentially for 4 min each as follows: 50%, 75%, 95% (2x)100% (3x)

NOTE: repeated washes should be carried out using fresh solutions in separate containers

4.2.9

Allow slides to air dry.

4.2.10

Detection of radioactively labeled probes: Expose slides to X-ray film (Kodak BioMax MR film, Fisher # 05-728-24) in an appropriate x-ray film cassette (e.g. Fisher # FB-XC-810). Exposure times must be determined empirically and can vary from several hours to several weeks depending on the relative abundance of the mRNA target within tissue regions. Optimal exposure time can also vary depending on type of experiment and need to be determined empirically (e.g. visualizing presence/absence of hybridization signal vs. semi-quantitative measurements of mRNA levels). For visualizing mRNA presence, an initial 24-hour exposure often serves as a useful reference in estimating necessary exposure length for many mRNAs (i.e. reasonable time to get feedback on whether experiment successfully produces data, can determine additional exposure time for better signal). Initial exposures also provide a linear scaling factor for calculating additional exposure time (e.g. if one wants an exposure that is twice as dark as the 24 hour exposure, additional film is developed for 48 hours). High resolution autoradiographic images can be collected using photographic emulsions (e.g. Kodak NTB2/3 or Ilford) with potentially longer exposure times (determined empirically) followed by photographic development in standard photographic development solutions. For semi-quantitative measurements, some assessment of autoradiographic image intensity is necessary to ensure that measurements are captured within the linear response range of the autoradiographic matrix (i.e. x-ray film/photographic emulsion not saturated). This need can be addressed by use of a standard curve (e.g. using dilutions of radioactive samples [40], commercially available standards [10]).

4.3 Cresyl violet cell staining

In order to reveal cellular morphology that is useful for subsequent semiquantitative analysis, tissue sections can be stained with a generic counterstain following hybridization. The following steps outline a protocol for cresyl violet cell staining used in our lab.

4.3.1

Rehydrate slides in a series of graded ethanol washes. Incubate slides in separate ethanol containers sequentially for 1 min each as follows: 100% (2x), 95%, 90%, 75% (2x), 50%

4.3.2

Wash slides in filtered water for 30 seconds.

4.3.3

Incubate slides in 0.5% Cresyl violet for 2–3 min.

4.3.4

Wash slides in distilled tap water for 30 seconds. Rinse slides 2–3 times in distilled tap water.

4.3.5

Dehydrate slides in a graded series of 50–100% ethanol. Incubate slides in separate ethanol containers sequentially for 1 min each as follows: 50%, 75% (2x), 90%, 95%, 100% (2x) NOTE: The same ethanol washes can be used for dehydration and rehydration of cresyl violet stained slides.

4.3.6

Wash slides two times in xylenes for 5 min each.

NOTE: all work with xylenes should be performed in a fume hood while wearing the appropriate personal protective equipment.

4.3.7

Remove each slide from xylenes. Using a disposable Pasteur pipette, apply a drop of Permount along the edge of the slide adjacent to the tissue. Place the edge of a cover slip along the edge of slide to which Permount was applied and carefully lower the cover slip over the tissue.

4.3.8

Allow slides to dry in the fume hood overnight.

NOTE: In our experience, ISH-processed slides often yield less robust cell stain intensity than directly-processed fresh-frozen or perfusion-fixed tissues. An extended incubation in the cresyl violet may result in modestly increased staining intensity (e.g. 5 min vs. 2–3 min). If staining is insufficient, cover slips can be removed by an extended wash in xylenes (minutes to hours, depending of the time elapsed since cover slip mounting), followed by repeating steps 4.3.1 to 4.3.7 with a longer incubation in 0.5% Cresyl violet.

4.4

As an example using the listed radioactive in situ hybridization procedure with subsequent cresyl violet cell staining, see Figure 1 (ISH detection of tyrosine hydroxylase mRNA in the rat midbrain as detected using X-ray film (Figure 1, AS/S/RNase) and emulsion autoradiography (Figure 1, BF-AS/DF-AS)).

FIGURE 1. In situ hybridization detection of tyrosine hydroxylase (TH) mRNA in rat brain sections using ³⁵S-labeled cRNA probes.

At left; X-ray film images from Antisense cRNA probe (AS), Sense strand cRNA probe (S) and RNase-pre-treated antisense cRNA probe (RNase) ISH results (X-ray film exposure 3 days (BioMax film; Kodak)). At right; high resolution autoradiographic images captured on a Ziess Axiophot microscope (10x objective (insert with 20x objective)) using bright field (BF-AS) and dark field illumination (DF-AS) from a photographic emulsion-coated, ISH-processed microscope slide (Kodak NTB2 emulsion; photographic development 2 min D-19 (Kodak), 3 min Rapid fix (Kodak); exposure time 8 days at 4°C in light-tight container). Clear TH ISH signal can be detected with antisense cRNA probes (dark signal in AS) in the substantial nigra and ventral tegmental area of the rat midbrain as compared to controls (S and RNase images yield only background hybridization patterns). The use of emulsion autoradiography reveals cellular patterns of TH mRNA expression (BF-AS/DF-AS). Clear cellular resolution of TH ISH signal can be seen in the larger boxed area in BF-AS image (arrow head marking TH ISH labeled cell while many non-labeled cells (cresyl violet stained nuclei) lack silver grains (full arrow notes one example of a cell that lacks detectable TH mRNA)). The small box in the BF-AS image notes specific location of higher magnification insert. Note the many non-labeled cells are visible in the bright field image as a result of the cresyl violet counterstain but are not visible in the dark field image (e.g. compare upper right portion of BF-AS vs. DF-AS)

4.5 Overview of semiquantitative analysis using radioactive-based ISH

Semi-quantification of radioactive-based ISH can be done using various versions of computer-assisted optical densitometry (e.g. [10, 41–44], overview in [45]). In all ISH semi-quantitative methods, processing all tissue samples simultaneously under the same experimental conditions can yield ISH data compatible with mRNA expression comparisons. Images can be quantified using publicly-available NIH ImageJ software (Scion Image). In principle, ISH mRNA levels can be quantified by the following steps:

1. Identifying and outlining the area of interest in a number of tissue sections per sample (e.g. using digitized images from a CCD camera)

2. Defining the actual signal above background and filtering out background signal (actual signal often defined by the number of standard deviations (S.D.) above background/area with no apparent hybridization signal; we routinely use 3.5 S.D. above background)

3. Determining an integrated density value of the actual signal for the area of interest (e.g. background-subtracted radioactive intensity per pixel or per unit area)

4. Averaging the density value across sections to give a mean, single value per sample

4.6 Additional protocol information: sample preparation, troubleshooting

Details on additional considerations and options for ISH experimental procedures has been previously described from our lab [24], including further information on collecting tissue sections and preparing cell culture samples for ISH, RNase-free working conditions, gel purification of labeled RNA probes.

4.6.1 Tissue preparation and fixation issues

Tissue fixation also influences nucleic acid probe penetration. While our laboratory typically utilizes fresh frozen tissue only lightly treated with tissue fixatives, perfusion-fixed tissues work well for ISH if tissue preparations are lightly digested with proteases (e.g. Proteinase K) prior to hybridization [23]. The one-hour fixation steps recommended here work for a variety of tissue sample types but are not universally ideal. Thin, delicate tissues may decrease in structural integrity if fixed for too long in the PFA solution and lose some portion of cellular contents, thus reducing ISH signal (e.g. retina sections). If low signal is observed, one possible solution is shorter fixation times; optimal conditions would need to be determined empirically.

4.6.2 Unstable temperature control

Limiting temperature variation in the protocol is important for maximizing specific hybridization signal while limiting non-specific background. In general, hybridization temperatures above theoretically ideal temperatures yield low non-specific hybridization background but often also yield low specific hybridization signals. In contrast, the use of hybridization temperatures below theoretically ideal temperatures yield more abundant specific hybridization signal but can also yield increased non-specific hybridization signal. If the probes have been used for Northern blot experiments, theoretical or ideal hybridization temperatures are often extrapolated based upon the hybridization conditions used in these experiments as long as such conditions do not harm tissue section quality. All temperature-specified solutions should be preheated using a water bath. If probe-based signal is not optimal, changing hybridization temperatures of steps can be done to attempt to improve signal; such decisions should be made with the mentioned caveats in mind.

4.6.3 Variable signal and/or background intensity between sections on the same slide

Signal variation between multiple sections on a single slide can be indicative of unequal processing conditions based on heterogeneous solutions (e.g. unmixed), improper coverslip placement or removal (Figure 2B, right portion of tissue section), or intraslide temperature variation (Figure 2C). For example, temperature-specified solutions used in a water bath need to completely cover the sample container (e.g. RNAse treatment covers all of sections). Also, the solution container needs to be surrounded by water to maintain uniform temperature (e.g. water bath water level is above solution level in container). Coverslips sometimes slide off some or all of the tissue section, thus allowing the hybridization solution to dry upon the tissue as well as altering the concentration of volatile reagents (e.g. DTT) (Figure 2A). We also find that pipetting hybridization solution directly onto tissue versus onto coverslips can lead to unexplained non-specific hybridization signals. We recommend visually inspection of all slides with coverslips before overnight hybridization and care taken to keep the hybridization chamber flat when moving to the incubator. On rare occasion, we have seen differential hybridization signal when using ISH dishes that hold the slide vertical (e.g. Copplin jars) compared to horizontal (e.g. Wheaton staining dish); the vertical-holding containers may show differential signal due to inefficient reagent mixing and/or uneven solution temperatures. Although we often use vertical-holding containers, we recommend using a horizontal-holding container as a potential solution to variable intensity.

FIGURE 2. Examples of In Situ Hybridization Artifacts.

Shown are three undesired autoradiographic in situ hybridization results that can occur. Panel A demonstrates an X-ray film autoradiographic image that reveals very high non-specific background throughout the entire brain tissue section. This type of result occurred in this case due to the absence of dithiothreitol (DTT) in the hybridization buffer. Panel B shows a distinct linear background variation that is due to inappropriate placement or movement of the hybridization coverslip (top to bottom in the image, see arrow). This same image also shows tissue damage through the hippocampal regions likely due to a general lack of care during coverslip removal (note the absence of signal in scrape across section midline). The speckles in the upper right hand corner are due to silver grain exposure likely caused by moisture from that portion of the slide Panel C displays uneven ISH hybridization signal at the dorsal edge of the top brain image that is clearly absent in the lower brain image, yet clear specific hybridization signal is seen in both brain images (see arrow heads).

4.6.4 High nonspecific background signal

High background can be a common challenge for ISH studies and often results from experimental procedure errors. Typical issues include accidentally omitting reagents or steps (e.g. omitting DTT (Figure 2A), not purifying labeled probe with spin column), low wash temperatures, unexpectedly high concentration of labeled probe, inappropriate hybridization conditions, and ill-timed autoradiographic exposure (e.g. too long). All reagents and steps should be carefully monitored and recorded for reference so errors can be determined if there is high non-specific hybridization background signal. All water baths should be regularly checked for correct temperature readings (e.g. thermometer confirmation of reported temperature). Probe calculations need to be recorded and repeatable; we recommend using computer software (e.g. Microsoft Excel) for such calculations so they can be reviewed as need be.

4.6.5 Lack of or limited probe signal

Absence of signal can be caused by degradation of the RNA probe (e.g. RNase contamination during probe purification), problems with the RNase A treatment, or inefficient nucleic acid probe labeling. Although rare, RNase A powder may contain contaminating ribonucleases that can degrade dsRNA. This possibility can be addressed by boiling the RNase A or using a different lot of RNase A. Signal may also be reduced by excess RNase A treatment; if need be, different concentrations of RNase A solution should be tested for optimal use. Another possibility for radioactive probes is inefficient incorporation of the radionucleotide during probe synthesis. This possibility can be clarified by characterizing an aliquot of the radioactively-labeled RNA probe using denaturing gel electrophoresis and exposing that gel to film, preferably with an additional successful reaction as a positive control; an inefficient reaction will have a lower autoradiographic signal of labeled RNA probe compared to free, unincorporated radiolabeled nucleotide (Note: the two entities are distinguished by size).

4.6.6 Non-specific blotches/speckling on x-ray film

Some x-ray film autoradiographic images from in situ hybridization processing may display dark blotches or speckles. This situation can occur if the sample is not completely dry as moisture reacts with the x-ray film (Figure 2B, upper right hand corner of image). If this phenomenon is observed, slides should be left out to dry for a longer period of time before being placed into the x-ray film cassette. Slides can also be dehydrated through the graded ethanol series for longer periods (e.g. 8 min vs. 4 min) to ensure water removal.

5. Discussion

Detection of spatial expression patterns of an mRNA of interest is an important component of gene expression analysis in biomedical research. We have outlined detailed methods for ISH detection of mRNAs using in vitro transcribed RNA probes synthesized with ³⁵S-labeled nucleotides. Beyond the more obvious anatomical location information as documented in this work, this technique can be extended to address additional biological questions. First, as stated earlier, ISH using radiolabel detection methods can be used for semi-quantitative analysis of differential gene expression. mRNA levels documented to change in vitro (e.g. cell culture) can be examined in an anatomical context using ISH approaches, enhancing the physiological significance of the in vitro data. Comparative measurements can also be made between brain regions or across biological samples derived from different treatments (e.g. anatomical location, experimental vs. control). Second, given that the ISH methodology uniquely identifies cells expressing specific mRNAs, there exists the possibility that the mRNA is subject to post-transcriptional or translational regulation, prompting the desire to combine ISH methods with immunocytochemistry methods (ICC) that detect the encoded (or other) proteins. The combination of ISH with ICC has been performed successfully (e.g. [46–47]). This coupling allows correlative anatomical analysis between these two types of molecules (e.g. mRNA with cell type marker or known interacting protein; are they in the same cellular localization?). However, this approach can present additional technical challenges involving the order in which the two methods are combined (i.e. ISH followed by ICC, ICC followed by ISH). Also, ISH+ICC techniques can in some cases limit both the quality of the anatomical identification (typically reduced tissue/section quality) and the overall sensitivity of one or both detection methods. Often these combined approaches are influenced by the robustness of each detection method alone and the quality and purity of the antibody preparations, which both require some empirical testing.

Variations on the ISH methods outlined in this paper can be used to detect different forms of RNA molecules. For example, many mRNAs are subject to alternative splicing and thus have multiple mRNA transcripts from a single gene that may have differential effects on subsequent protein function. ISH RNA probes can thus be designed to distinguish between mRNA forms based on the specific nucleic acid sequences inserted into the transcription-based plasmid constructs (e.g. form-specific nucleic acid probes). This approach can enable isoform-specific hybridization and analysis (e.g. substance P vs. substance K, [48]). ISH methods can also be used to detect RNA forms other than mRNAs, such as microRNAs (miRNAs). ISH techniques have become an important component in the emerging field of miRNA functional analysis. Our lab has collaboratively developed ISH methods to detect miRNAs using RNA oligonucleotide probes that are end-labeled by kinases instead of in vitro transcribed cRNA probes because of the short length of miRNAs (~20–22nt) [24, 49]. Such applications highlight the continued and evolving need for ISH techniques.

6. Concluding Remarks

In situ hybridization methods are useful for visualizing gene expression patterns in tissue and cell culture samples. ISH complements methods that analyze global gene expression changes (e.g. microarray, next-generation sequencing); with the frequent usage of these methods, ISH is an important follow-up technique to verify findings through revealing the anatomical distribution of mRNA expression patterns. Given the critical linkage that emerges by documenting anatomical location to function, ISH methods continue to be important components in determining the function of different genes and changes in gene expression due to experimental treatments or disease conditions. The method outlined in this paper can support biomedical researchers in their continuing efforts to characterize their cellular mechanisms of interest.