In situ hybridization, also referred to a hybridization histochemistry, was introduced in 1969 (Buongiorno-Nardelli and Amaldi, 1969; John et al., 1969). The basic technique utilizes the fact that DNA and RNA will undergo hydrogen bonding to complimentary sequences of DNA or RNA. By labeling sequences of DNA or RNA of sufficient length (approximately 50-300 base pairs), selective probes can be made to detect particular sequences of DNA or RNA. The application of these probes to tissue sections allows DNA or RNA to be localized within tissue regions and cell types. In situ hybridization is a powerful technique and unique in the way that it allows one to study the macroscopic distribution and cellular localization of DNA and RNA sequences in a heterogeneous cell population.

In the past 30 years many improvements have been made on the basic in situ hybridization technique and its application has been carried out in all different tissues. The technique is particularly useful in neuroscience where the tight regulation of gene transcription is vital to the operation of the brain. Using in situ hybridization, it is possible to localize gene expression to specific cell types in specific regions and observe how changes in this distribution occur throughout development and correlate with behavioral manipulations. Particularly interesting for many neuroscientists is the fact that regulation of plasticity in the brain requires gene transcription. In this way, using in situ hybridization may allow a snap-shot of the whole brain to help investigators understand how synaptic plasticity in various cells types from different regions in the brain may be working simultaneously to mediate processes underlying learning and memory.

This paper will describe the basic principles of in situ hybridization and advantages and disadvantages of different methodologies that can be used.

Probe Selection

sparing hybridized RNA. One problem with RNA probes are that they can have higher levels of non-specific binding to tissue components, which can lead to higher background and lower penetration of the probe into the tissue. Finally, it is possible to directly manufacture short oligonucleotide sequences that are generated from labeled nucleotides. The advantages of this are that no molecular biology expertise is required, the generated molecules are stable, there is good tissue penetration due to the small size of the probes, a precise amino acid sequence can be specified from published data, and because single stranded probe is generated, there is no competition between probes. Some disadvantages of oligonucleotide probes are that due to the process of oligonucleotide generation the labeling methods are limited, limitations on label quantity per probes leads to reduced sensitivity, and the hybrids are less stable because of the short length of the probes. Additional disadvantages come from the fact that access to a DNA synthesizer is necessary and that one is dependent on published sequences to generate probes.

Probe Generation

There are a number of methods that can be used in order to generate a labeled probe for in situ hybridization (described in Brady & Finlan, 1990; Chan & McGee, 1990). Many of the methods available are specifically useful for the generation of DNA probes. One of the oldest and most popular techniques is nick translation. This method uses two enzymes, DNase I and Pol I. In the presence of Mg²⁺, DNase I randomly hydrolyses the phosphodiester bonds of the individual strands of dsDNA, causing ‘nicks’ along the length of each strand of the dsDNA. The holoenzyme, Pol I, has both exonuclease activity and polymerase activity, which removes nucleotides 3’ of the nicks created by DNase I and replaces them with nucleotides in the solution based on the template provided by the other strand of the DNA. If labeled DNA is in high concentration in the solution, each strand of the DNA molecules in solution incorporate labeled nucleotides. The degree of label incorporation can be controlled by the concentration of labeled nucleotide, the concentration of Pol I, and length of the reaction time. If the reaction is optimally controlled, nick-translated probes can give the highest sensitivity compared to probes labeled with other methods. Another method for label incorporation is random-primed labeling. This method utilizes the Klenow fragment of Pol I, which lacks the exonuclease activity. DNA is first denatured by heating and hexadeoxyribonucleotides with no sequence specificity are added to the mixture. The hexadeoxyribonucleotides bind to the denatured DNA and act as a primer region for Pol I to begin DNA synthesis. By varying the ratio of primer to template, the length of the labeled DNA strands can be controlled, where high concentration of primer gives short labeled strands and vice versa. The major drawback of this method is that both labeled and unlabeled probes compete in the hybridization reaction, but the Klenow fragment is more resistant to inhibition by contaminants in the reaction than is the Pol I holoenzyme. Another method for generating labeled dsDNA is called end-tailing. This simple method uses terminal transferase to catalyze the addition of labeled nucleotides to the end of single-stranded or double-stranded DNA in a template-independent fashion. The disadvantages of this technique are that only a small number of nucleotides will be labeled relative to probes generated with other methods and that the generated nucleotide tail may cross-hybridize with stretches of complementary sequences, leading to non-specific labeling. A final method for the generation of DNA probes is called T4 DNA polymerase replacement, which utilized the fact that T4 DNA polymerase has both exonuclease and polymerase activity. While the exonuclease activity is very high in the absence of free nucleotides, in the presence of all four nucleotides the exonuclease activity is inhibited. To create labeled probes, dsDNA is mixed with T4 DNA polymerase. In the absence of free nucleotides, the T4 DNA polymerase begins chewing away at the 3’ ends of each strand of the DNA. After a short time, labeled nucleotides are added to the mixture, at which time the T4 DNA polymerase begins replacing the removed nucleotides with labeled nucleotides. This creates dsDNA with individual strands that are labeled at one end. This method can be advantageous because these DNA strands can be purified to a ssDNA probe by restriction digestion followed by purification in an agarose gel.

The generation of RNA probes utilizes in vitro transcription. In this method, the sequence is cloned into a vector containing phage transcription promoters that can initiate transcription in the presence of the corresponding RNA polymerase. If two different phage transcription promoters are placed on both sides of the polylinker cloning sites of the vector in opposite orientations, it is possible to selectively transcribe sense and anti-sense RNA probes. Transcribing only one strand at a time prevents reannealing of complimentary strands during hybridization. By inclusion of either labeled or unlabeled nucleotides, it’s possible to make labeled or unlabeled probes that can be used for different control procedures used in in situ hybridization.

In addition to these enzyme-mediated methods, it is possible to label probes with non-radioactive labels by direct chemical labeling. Direct chemical labeling is generally simpler and easier, but enzymatic methods are typically more satisfactory (Chan & McGee, 1990). Among the chemical methods, photobiotin can be directly attached to biomolecules like DNA and RNA, but this method tends to have high background labeling. Biotin hydrazide can also be used to label unpaired cytosine, and biotin ester can label unpaired cytosine that has been transaminated with sodium bisulphate and ethylenediamine. Likewise, guanine residues in nucleic acids can be modified by treatment with N-acetoxy-N-2-acetylaminofluorene and subsequently recognized by antibodies.

Probe Labels

Both radioactive and non-radioactive probe labels are commonly used for in situ hybridization. A comparison of their basic attributes can be seen in table 2. Radioactive labeling has several advantages including that radioisotopes are readily incorporated into probes using many common enzymes, probes

There are several systems available for the non-radioactive labeling of in situ hybridization probes (Lewis & Baldino, 1990). Biotin can be chemically linked to probes as discussed earlier or can be attached

to nucleotides and enzymatically incorporated into nucleic acids. Biotin labeled probes can be detected by avidin or preferably streptavidin molecules that are covalently linked to reporter molecules, such as fluorophores, enzyme reactants (e.g. peroxidase, alkaline phosphatase), or heavy metals (e.g. ferritin, colloidal gold). It is also possible to detect biotinilated nucleotides by antibiotin IgGs, but the affinity is considerably less than avidin-biotin. One caution that must be exercised in using biotin is that some tissues have endogenous biotin (vitamin H), which can lead to false positive results. Another labeling method is to directly couple alkaline phosphatase or horseradish peroxidase to nucleotides for incorporation into nucleic acid probes. The sensitivity of this method can reportedly exceed that of biotinilated probes (Lewis & Baldino, 1990). Hapten-conjugated probes provide another method for in situ hybridization probe labeling. This can either be achieved by chemical methods or by incorporation of digoxigenin-dUTP into nucleic acid probes. Detection can be performed by IgG conjugated to a reporter molecule. This method offers increased sensitivity because multiple labels can be incorporated into the probe and detection with IgG allows antibody bridging, which results in considerable signal amplification. Finally fluorophores can be chemically or enzymatically incorporated into probes. The disadvantage of fluorescent probes is decreased sensitivity, but they have the substantial advantage that multiple probes can be simultaneously detected with spectrally discriminable probes. Fluorescent probes have also been reported to provide superior resolution.

Fixation of Tissue

There are several variables that need to be considered when fixing tissue for in situ hybridization. When detecting mRNA levels it is important to recognize that mRNA is normally synthesized and degraded at a high rate. Likewise, the levels of many mRNAs can be dramatically changed in vivo by stress to the animal. These two facts require that for mRNA detection animals must be sacrificed with as little stress as possible and the tissue fixed quickly. Likewise it is very important that all procedures are carried out to avoid the contamination of the tissue with RNases on skin and laboratory instruments (Simmons et al., 1989).

The fixation methods must be chosen to balance 1) the accessibility of the probe to the target sequence, 2) retention of maximal levels of cellular target DNA or RNA, and 3) maintenance of morphological details in the tissue (Tecott et al., 1987; Hofler, 1990). Using precipitating fixatives like acetic acid-alcohol mixtures provide the best probe penetration, but may permit the loss of RNA from tissue. Glutaraldehyde, on the other hand, provides the best RNA retention and tissue morphology, but because of extensive protein cross-linking the probe penetrance is low. Likewise, paraffin and formalin fixation leads to decreased sensitivity possibly resulting from increased cross-linking or loss of mRNA during embedding. To increase penetrance with these methods it is possible to treat the tissue with detergents or proteases, but these may lead to the loss of DNA and RNA. The most widely successful fixative has been 4% paraformaldehyde solution. This provides a satisfactory compromise between the variables and good sensitivity. It is also common to snap freeze the tissue immediately after excision in liquid nitrogen. The tissue is subsequently fixed in paraformaldehyde after sectioning. The disadvantage of this method is that tissue morphology is compromised, but it allows tissue to be used both for in situ hybridization and for DNA or RNA extractions.

Hybridization and Washing

Hybridization is performed by placing a small amount of solution containing the hybridization probe on a cover slip, which is then placed on the slide containing tissue sections to incubate overnight. It is important to work out all bubbles from under the cover slip or some regions will not receive labeling. The next day, washes are serially applied to the slides to remove the probe that is not bound to target DNA/RNA. During hybridization and washing there are several parameters including, probe construction, temperature, pH, and formamide and salt concentration in the hybridization buffer that all play a crucial role in the specificity of target labeling. By varying these it is possible to change the affinity of the probe for its target sequence. In general, higher temperature, higher pH, lower ion concentration and higher formamide concentration all lead to a lower affinity of the probe for the target sequence. For the hybridization reaction, the solution conditions are set to favor binding of the probe to the target sequence. The subsequent washes, aimed at removing non-selectively and partially bound probe, use increasing temperatures and decreasing salt concentrations. The hybridization and washing conditions must be varied to accommodate differing lengths of probes and different labeling methods, which can affect hybridization affinity. If washing conditions are too stringent a loss of sensitivity can occur and if washing conditions aren’t stringent enough, a high background labeling will occur.

Control Procedures

In using in situ hybridization it is very important to use several control procedures to be sure that the observed labeling is specific to the target sequence (Lewis & Baldino, 1990). One such control procedure is to use alternate sequence or sense (rather than anti-sense) probes to determine whether the observed labeling pattern is specific to the target sequence or would be shared by any nucleic acid probe. Another control procedure is to use multiple probes for the same gene sequence. This controls for fact that the probe sequence could be detecting similar DNA/RNA sequences other than the desired target sequence. Another control is to pretreat sections with non-labeled probe before treating with labeled probe. This should eliminate the specific labeling of target sequences, leaving only background labeling. In order to reduce background labeling it is possible to preabsorb the tissue with specific non-labeled material like ficoll, bovine serum albumin, and polyvinyl pyrrolidone. When detecting mRNA, it is possible to apply RNase to the tissue after hybridization to reduce background. It is also possible to apply RNase to the tissue before hybridization as a control procedure that should eliminate labeling specific to the target sequence. Another control is to use Northern analysis of tissue to show that the labeled probe binds to an mRNA target of the correct molecular size. An especially convincing, but more demanding control procedure is to perform immunohistochemistry for the protein product of the gene target to show that the protein and gene precursor co-label. A negative result with this control, however, isn’t necessarily evidence of failed in situ hybridization, but could be due to either non-translated mRNA or mRNA that is translated to protein that is rapidly transported away from the translation site. In the use of radioactive probes it is possible to control for autoradiographic artifacts by omitting the probe.

Discussion

In situ hybridization is a unique and powerful technique that allows the localization of DNA and RNA to specific regions and cells within these regions. The technique specifically has several advantages over immunocytochemistry in anatomical localization. The nature of the Immunohistochemistry procedure makes it non-quantifiable and is frequently fettered with both false negative and false positive results (Nilaver, 1986). While radioimmunoassays allow quantitative information to be gathered from extracts of particular brain regions, they only allow average measurements of changes within a given area, and cannot specify whether a subpopulation of neurons is responsible for the results. In order to obtain quantitative results from immunocytochemistry, it is necessary to set conditions such that variations in staining intensity are linear with respect to amount of antigen and also take in account the deleterious effects of fixatives in altering tissue antigenicity (Nilaver, 1986). While in situ hybridization cannot provide information on translation and post-translational processing, the levels of mRNA provide information that cannot always be determined from measuring peptide levels alone. This is evident from the studies showing larger treatment-induced changes in mRNA levels than in the coded for protein (as cited in Nilaver, 1986).

Despite the advantages that in situ hybridization provides, it is important to be cautious when interpreting the results from in situ hybridization studies. It is extremely important that the proper controls are performed to show that the labeling is due to hybridization of the target rather than non-specific labeling. In addition, one must be cautious in extrapolating increased mRNA to mean that increased protein is generated. It is possible that in some cases there are mechanisms to maintain protein levels even in the presence of increased RNA. Using immunocytochemistry in concert with in situ hybridization can provide converging evidence to support functional interpretations of the data. Finally, one must be cautious because in situ hybridization is inherently a correlative measure. Like any correlational measure, causative interpretations are speculative. In order to substantiate causative interpretations, careful manipulations must be performed. One such manipulation that has been used recently is the local injection of antisense oligodeoxynucleotides into a brain region to bind mRNA and inhibit protein expression (Guzowski et al., 2000). This allows causal and functional hypotheses generated from in situ hybridization data to be tested directly.