Methods to Enhance Signal Using Isotopic In Situ Hybridization

Abstract

Isotopic in situ hybridization (ISH) has been established as a uniquely powerful tool for the study of gene expression in specific cell types. This technique allows the visualization and quantification of gene expression and gene expression changes in cells. In our study of biological and molecular phenomena, we have increasingly encountered the need to detect small changes in gene expression as well as genes of low abundance, such as the oxytocin receptor (OTR) and the tuberoinfundibular peptide of 39 residues (Tip39). To increase the sensitivity of isotopic ISH for detection of rare mRNAs, we performed ISH on cryostat sections of rat hypothalamus and thalamus with ³⁵S-labeled riboprobes and amplified the signal by hybridizing over 2 nights as well as labeling the probe with both [³⁵S]-UTP and [³⁵S]-ATP. These two methods of enhancement independently and in combination demonstrated a dramatic increase in signal, allowing the visualization of low levels of gene expression previously undetectable by conventional methods.

In situ hybridization (ISH) is a powerful and versatile tool that allows the direct visualization and localization of specific nucleic acid sequences in tissues or specific cell types. The identification of gene expression patterns in tissues can provide critical spatial and temporal information and is the first step in understanding gene function. An extensive amount of literature already exists describing different methodologies for ISH. These techniques, however, vary considerably in a number of important steps as well as in general efficacy and reproducibility, especially when applied to situations in which gene expression levels are low or cells that express a particular gene are few.

The basic underlying principle of ISH can often be complicated by excessively complex methodology. A nucleic acid probe, tagged with either a radiolabeled nucleotide or other molecules allowing colorimetric or fluorescent detection, is hybridized directly to cell RNA made accessible by sectioning and processing a tissue of interest. Traditionally, [³⁵S]-uridine triphosphate (UTP) is used to label the antisense RNA probe. [³⁵S]-UTP has proved to be the most sensitive and reliable label for detecting mRNA because it provides the greatest compromise between time and resolution of signal. The recent development of non-isotopic techniques to amplify fluorescent or colorimetric detection has been well documented. These methods are of comparable sensitivity in the detection of relatively abundant genes and can be useful for visualization of multiple genes in a single cell (Breininger and Baskin 2000). However, fluorescence does not allow quantification and is often not sensitive enough to detect small changes in gene expression or rare message. Such data can be obtained only with radioactively labeled probes. To date, however, there has been little literature describing the enhancement of radioactive signal.

Given the basic premise of ISH, it should follow in a logical manner that labeling the RNA probe with two ³⁵S nucleotides vs one should amplify signal of expression roughly twofold. Likewise, hybridizing the sections with probe over 2 nights vs over 1 night should also significantly enhance signal intensity because the probe has a greater chance to hybridize to its complementary mRNA. We now describe an extensive methodological analysis and introduce a new approach to further optimize expression profiles of genes of low abundance. Moreover, we describe in detail a simple method for ISH that has eliminated many unnecessary steps.

Materials and Methods

Tissue Preparation

Adult male Sprague–Dawley rats (Taconic; Germantown, NY) were housed at the Merck Research Laboratories animal care facility in accordance with policies and regulations of the IACUC. Rats were sacrificed with CO₂ and brains were carefully removed, frozen on dry ice, and stored at −80C until sectioning. Twenty-μm coronal cryostat sections were collected from the ventromedial nucleus of the hypothalamus (Bregma −2.8) (Paxinos and Watson 1986) through the posterior commissure (Bregma −4.8) and were mounted on cold RNase-free gelatin-coated microscope slides. The section-mounted slides were briefly dried on a slide warmer maintained at 42C and then stored in desiccated slide boxes at −80C. Before the tissues were processed, the desiccated slide boxes were slowly warmed to room temperature [-20C for 1 hr; 4C for 1 hr; room temperature (RT) for 1 hr] to eliminate formation of condensation on slides and thus minimize tissue and RNA degradation. The dry slides were loaded into metal racks and processed as follows under RNase-free conditions. Tissues were postfixed in 4% paraformaldehyde (J.T. Baker; Phillipsburg, NJ), pH 9.0, for 5 min at 4C. After fixation, sections were rinsed in 1 × PBS (Biowhittaker; Rockland, ME) pH 7.4 for 2 min at 4C, equilibrated in 100 mM triethanolamine (J.T. Baker), pH 8.0, for 1 min at RT, and transferred to 100 mM triethanolamine with acetic anhydride (TEA), made fresh at RT. Slides were maintained in the stirring TEA solution for 10 min at RT to remove positive charges that may cause excess background. Tissue was then washed in 2 × SSC (0.3 M NaCl, 0.03 sodium citrate, pH 7.0) (Biowhittaker) for 2 min and dehydrated through a graded series of ethanols (70%, 95%, 100%) each for 2 min at RT. Slides were then placed in chloroform (J.T. Baker) for 5 min at RT to delipidate tissue. Finally, tissue was transferred into 100% ethanol for 2 min at RT and air-dried for at least 1–2 hr before applying probe.

Riboprobes

The two riboprobes used for these experiments are both known to be weakly expressed in the rat brain. One is complementary to the rat tuberoinfundibular peptide of 39 residues (Tip39) mRNA sequence and the other to part of the rat oxytocin receptor (OTR) mRNA sequence. Antisense and sense (control) RNA probes were synthesized from a linearized PCRII plasmid containing a 180-bp fragment of Tip39 cDNA (obtained from Dr. Hao Wang, Merck), or a linearized PGEM7 plasmid containing a 283-bp fragment of OTR cDNA (obtained from Drs. Tracy Bale and Dan Dorsa) (Bale and Dorsa 1995a, b). The probes were labeled with either [³⁵S]-UTP (NEN, NEG039H; Boston, MA), [³⁵S]-ATP (NEN, NEG033H), or double labeled with a cocktail of both [³⁵S]-UTP and [³⁵S]-ATP in a transcription reaction using the T7 RNA polymerase (Roche; Mannheim, Germany). Transcription reaction components included 1 × transcription buffer, 10 mM dithiothreitol (DTT), 10 U RNase inhibitor, and 500 μM cocktail of unlabeled nucleotides, (A)/(U)/C/GTP. After a 2.5-hr incubation at 37C, the probes were purified over a Nick column (Amersham; Arlington Heights, IL) according to the manufacturer's instructions to remove unincorporated nucleotides, and the probes' quality was assessed on an acrylamide gel.

In Situ Hybridization

The probes were diluted with TED (10 mM Tris, 5 mM EDTA/100 mM DTT) and heated at 65C for 3 min and quickly cooled on ice to stabilize the single-stranded probes. The probes were then added to the hybridization buffer composed of 50% deionized formamide, 5% dextran sulfate, 0.3 M NaCl, 10 mM Tris, pH 8.0, 1 mM EDTA, 1 × Denhardt's, and 0.2 M DTT. Processed section-mounted slides (1 × 3 inch) were hybridized with 200 μl of the anti-sense probe mixture (6 × 10⁶ disintegration per minute (DPM)/slide or 12 × 10⁶ DPM/slide for double-labeled probes). Slides were incubated without coverslips in an open-air hybridization chamber (Figure 1). To prevent drying, the Plexiglass chamber (Trident Plastics; Ivyland, PA) was humidified with strips of blotting paper (Figure 1B) supersaturated (190 ml) with a mixture of 50% formamide/600 mM NaCl, pre-warmed to 65C, and sealed tightly with black electrical tape. The slides were divided into two groups. One group was hybridized at 55C over 1 night (∼16 hr) and the second group was hybridized over 2 nights (∼40 hr) at 55C. After each corresponding incubation time, the group of slides was placed in metal racks immersed in 2 × SSC/10 mM DTT to remove the excess hybridization mixture. The racks were then transferred to a large container (Rubbermaid; Wooster, OH) filled with 2 × SSC/10 mM DTT. When all of the slides were in the larger container, the racks were then transferred again to a new container and washed in 2 × SSC/10 mM DTT for 15 min at RT with gentle agitation. The slides were then transferred into RNase buffer (10mM Tris, 500 nM NaCl, 1 mM EDTA) and treated with 20 μg/ml of RNase A for 30 min at 37C. Slides were then washed for 15 min in 1 × SSC at RT with a final wash in 0.1 × SSC for 40 min at 65C. Slides were allowed to cool briefly to RT before dehydration through a graded series of ethanols (70%, 95%, 100%). Air-dried slides were apposed to X-ray film (Kodak; Rochester, NY) for 5 days and then dipped in NTB2 nuclear emulsion (Kodak) diluted 1:1 with deionized H₂O. The slides were exposed for 6 weeks in black desiccated boxes at 4C, protected from light. After appropriate exposure time, slides were developed under safelight, stained with cresyl violet, and coverslipped with permanent mounting medium. These basic methods concerning the preparation of tissue, synthesis of probe, and in situ hybridization were adapted from methods previously reported (Miller et al. 1989; Shughrue et al. 1996). Modifications to these methods are noted herein.

Figure 1

Hybridization chamber (A) constructed for ISH histochemistry using an open-air hybridization method. The section-mounted slides are placed on raised plastic strips of the chamber, elevating the slides from the floor. When the lid of the chamber is tightly secured with tape, the chamber provides a constant humidified environment and prevents dehydration of the tissues. (B) The section-mounted slides are covered with probe/hybridization mixture and placed on the raised surface without coverslips. The floor of the hybridization chamber is covered with strips of filter paper supersaturated with hybridization chamber buffer, providing a humidified atmosphere and resulting in an overall decrease in background.

Figure 2

OTR mRNA hybridization signal in the ventromedial hypothalamic nucleus of the hypothalamus, VMH, and central medial amygdaloid nucleus, CeM, (indicated by arrows). Effects of hybridization time, 1-night hybridization (left) and 2-night hybridization (right) on signal intensity of representative ISH images (a-c, g-i, film; d-f, j-l, emulsion). Twenty-μm coronal brain sections were hybridized over 1 night with the OTR [³⁵S]-ATP-labeled antisense riboprobe (a,d); hybridized over 2 nights with the OTR [³⁵S]-ATP-labeled antisense riboprobe (g,j); hybridized over 1 night with the OTR [³⁵S]-UTP-labeled antisense riboprobe (b,e); hybridized over 2 nights with the OTR [³⁵S]-UTP-labeled antisense riboprobe (h,k); hybridized over 1 night with the [³⁵S]-ATP- and [³⁵S]-UTP-labeled antisense riboprobe (c,f); and hybridized over 2 nights with the [³⁵S]-ATP- and [³⁵S]-UTP-labeled antisense riboprobe (i,l). The ability to detect very low-abundance genes not seen in conventional in situ methods (b) is demonstrated by the appearance of signal in the central medial amygdaloid nucleus (c,h,i).

Evaluation

Film autoradiographs were used to assess Tip39 and OTR mRNA expression in the rat brain. The autoradiographs were digitized with a computer image analysis system (MCID M5; Imaging Research, St. Catharines, ON, Canada), processed for brightness/contrast enhancement, and imported into Photoshop (Adobe; Tucson, AZ), where the images were excised from background and arranged into plates. Emulsion-coated slides, which were used to generate autoradiographic images, were also used to evaluate the density of grains at the cellular level using bright and darkfield microscopy (Zeiss; Oberkochen, Germany).

Figure 3

Tip39 mRNA hybridization signal in the paraventricular nucleus of the thalamus, PVP (indicated by arrow). Effects of 1-night hybridization (left) and 2-night hybridization (right) times on signal intensity of representative in situ hybridization images (a-c, g-i, film; d-f, j-l, emulsion). Twenty-μm coronal brain sections were hybridized over 1 night with the Tip39 [³⁵S]-ATP-labeled antisense riboprobe (a,d); hybridized over 2 nights with the Tip39 [³⁵S]-ATP-labeled antisense riboprobe (g,j); hybridized over 1 night with the Tip39 [³⁵S]-UTP-labeled antisense riboprobe (b,e); hybridized over 2 nights with the Tip39 [³⁵S]-UTP-labeled antisense riboprobe (h,k); hybridized over 1 night with the [³⁵S]-ATP- and [³⁵S]-UTP-labeled antisense riboprobe (c,f); and hybridized over 2 nights with the [³⁵S]-ATP- and [³⁵S]-UTP-labeled anti-sense riboprobe (i,l).

Quantitative Analysis

OTR and Tip39 film autoradiographs digitized into the MCID M5 computer image analysis system were used to produce relative optical density (ROD) values to compare intensity of signal between ISH conditions. The average of six measurements of a specified area of the VMH (for OTR) and PVP (for Tip39) were taken and data was evaluated through ANOVA and Bonferroni post-hoc tests. The personal computer program StatView (SAS Institute; Cary, NC) was used for statistical analysis of the data. All statements of non-difference imply that p>0.05.

Results

In these studies, the distributions of OTR and Tip39 gene expression were initially examined by the traditional methodology of ISH, labeling one nucleotide of the probe with [³⁵S]-UTP and hybridizing over 1 night. As predicted, the results of this single-nucleotide, single-night hybridization in this study showed weak expression localized in the ventromedial hypothalamic nuclei (VMH) for OTR (Figures 2b and 2e) and the paraventricular thalamic nucleus (PVP) for Tip39 (Figures 3b and 3e). The quality of the ISH was consistent with previous unpublished studies using an open-air hybridization chamber (Figure 1), resulting in a lower level of nonspecific binding and background levels that are commonly seen with other in situ methods. Typically, the use of a slide coverslip or Parafilm often results in increased variability in background levels among sections, especially if the humidity of the chamber is low during incubation, in which case the hybridization mixture will dehydrate and lead to drying of the tissue and high background levels (Mercer et al. 1996). In our experiments, Plexiglas chambers were used for hybridization, with blotting paper placed on the floor of the chamber and chamber buffer used to humidify the chamber. When the lid is placed on top of the chamber and sealed with tape, a consistent humidity can be maintained, which ultimately results in the demonstrated increase in signal-to-noise ratio (Figure 1).

Figure 4

Quantitative analysis of OTR (A) and Tip39 (B) mRNA expression. Comparison of level of signal intensity produced using one vs two ³⁵S nucleotide labelings and 1- vs 2-night hybridizations using relative optical densities (ROD) of specified areas of expression. Mean ± SEM of UTP 1 night, ATP 1 night, ATP 2 night, UTP 2 night, ATP/UTP 1 night, ATP/UTP 2 night. ^∗ p<0.05; ^∗∗ p<0.01; ^∗∗∗ p<0.001. Bonferroni corrected t-tests after significance found with ANOVA.

The isotopic ISH protocol with the double-label and 2-night hybridization procedure produced silver grains concentrated in the VMH for OTR (Figure 2) and in the PVP for Tip39 (Figure 3), while producing no signal in the sense controls (not shown). In the 1-night hybridization with [³⁵S]-ATP-labeled OTR, faint signal was seen in the VMH (Figures 2a and 2d). In the 1-night hybridization with [³⁵S]-UTP-labeled OTR, the expected signal was again seen in the VMH (Figures 2b and 2e), of slightly greater intensity than the sections hybridized with [³⁵S]-ATP. Although the increased intensity of signal was not statistically significant, an increase in signal was consistently seen throughout the sections and was therefore believed not due to artifact. In fact, there was indeed a significant difference in sections hybridized with [³⁵S]-ATP-vs [³⁵S]-UTP-labeled Tip39 (Figure 4A; ^∗ p<0.05). This could perhaps be a reflection of the number of ATPs within the OTR and Tip39 mRNA sequences vs the number of UTPs. In the 2-night hybridizations, the density of silver grains was significantly elevated (^∗ p<0.05) compared with the 1-night [³⁵S]-UTP-labeled hybridization (Figure 5). Within this 2-night hybridization group, OTR signal in the central medial amygdala could now be identified, whereas in the 1-night hybridization it was not detected above background levels. The 1-night hybridization with [³⁵S]-ATP and [³⁵S]-UTP demonstrated an even greater increase in signal (^∗∗∗ p<0.001), clearly delineating specific expression in the VMH and in the central medial amygdala. In the 2-night hybridization with [³⁵S]-ATP and [³⁵S]-UTP the most intense signal was seen (^∗∗∗ p<0.001), clearly increasing the number of silver grains deposited for this low-abundance gene (Figure 5B). However, although the intensity of signal was increased the background levels also increased. In this case, this increase in background appears to be a good compromise for the increased intensity of specific signal. Consistent with the OTR results, Tip39 (Figures 4B and 5D) demonstrated the same general pattern of increased specific signal with a corresponding increase in background.

Discussion

The Tip39 and OTR plasmids were selected for this study on the basis of a number of considerations. These genes provide interesting models for several reasons, including the fact that they are expressed in low abundance in well-defined areas of the brain. The expression of the OTR has been studied extensively in the rat brain by radioligand binding and ISH analysis (Bale and Dorsa 1995a, b). In these studies, expression was most consistent and intense in the VMH of the male rat brain. Relative to such abundant genes as vasopressin or its own ligand, oxytocin, the expression of OTR is quite scarce. Likewise with the novel Tip39 gene (Usdin et al. 1999; Dobolyi et al. 2002), previous studies have shown it to be specifically expressed in the PVP at very low levels. Using the OTR probe as a positive control, we could further explore the expression of Tip39 in the PVP and other regions of the brain using enhanced methods of isotopic ISH. Because this work aimed only to validate the technique, the physiological implications of the findings are not addressed.

Figure 5

Comparison of [³⁵S]-ATP-labeled antisense riboprobe hybridized over 1 night (A,C) and [³⁵S]-ATP- and [³⁵S]-UTP-labeled antisense riboprobe hybridized over 2 nights (B,D). High magnification view brightfield photomicrographs of hybridization signals over individual cells for OTR mRNA in the VMH (A,B) and for Tip39 mRNA in the PVP (C,D).

Having established a consistent and distinct pattern of gene expression for OTR and Tip39 using traditional in situ methodologies, we explored different radioactive methods in an attempt to enhance the weak signal of these low-abundance genes. Conventional methods of ISH call for the labeling of one nucleotide in the RNA probe, typically the UTP with ³⁵S. This, in most circumstances, is sensitive enough to detect the expression and localization of many genes. For scarce genes, however, enhanced methods of isotopic ISH can be explored to optimize signal. Because traditional techniques involve the labeling of one nucleotide of the probe, it should theoretically follow that labeling two nucleotides would doubly enhance the signal. Likewise, hybridizing over 2 nights vs 1 night should allow for twice the amount of time for probe to hybridize to mRNA in the tissue. This method is frequently seen in immunohistochemistry protocols because primary antibodies are often incubated over multiple nights to allow full penetration of the antibody into the tissue. Although these methods have the potential to be very sensitive, their use requires careful consideration to procedural details to avoid artifacts and/or excessive background. The fact that labeled nucleotides do not incorporate into the RNA probe as efficiently as unlabeled nucleotides must be taken into account. Therefore, it is important to properly assess the probe by gel analysis and counts on the scintillation counter. In addition, during the 2-night hybridization the humidified conditions of the open-air hybridization chamber are now of critical importance, because the longer length of time allows further opportunity for the slides to dry during incubation. Extra care should be taken to ensure a level chamber and oven shelf to avoid uneven incubation of hybridization buffer.

Under conventional ISH methods, these genes were hardly detectable above background levels. The two labeled nucleotides and 2-night hybridization procedures significantly increased signal and identified mRNA expression in novel nuclei of the rat brain. Although the increase in signal was also accompanied by an increase in background, this seems to be an adequate concession for the increased elucidation of specific signal. It is important, however, to take this into account in future studies.

A number of ISH protocols have been developed using either radioactive or non-radioactive labeled probes. The emphasis has largely been on the development of methods to enhance the visualization of mRNA that may be in low abundance in tissues or cells. Several strategies have been proposed using radioactive as well as non-radioactive probes. Non-radioactive probes are often favored due to considerations of time and cell resolution. In addition, many have used amplification techniques with non-radioactive probes, such as tyramide signal amplification (TSA) or enzyme-labeled fluorescent (ELF) (Breininger and Baskin 2000). Although these methods may prove to be useful for the visualization of multiple genes in a single cell, only radioactive probes allow precise quantitative measurements and, as we demonstrate here, can be sensitive enough to detect rare message. Here we report a sensitive and reliable method of isotopic ISH and its amplification to enhance rare and low-abundance signal. In our hands, a humidified open-air hybridization chamber consistently provided a lower level of background, and thus a better signal-to-noise ratio, compared with previous studies with cover-slipped samples (Mercer et al. 1996). To amplify the low-expressed signal, 2-night hybridizations and double nucleotide labeling of the probe were used. Based on our observations, these methods significantly increased the signal intensity of rare message in a consistent and reliable manner. These techniques not only boost the intensity of signal but may even elucidate very low expression of mRNA in nuclei not seen by conventional methods. In situ hybridization has made an extraordinary contribution to our understanding of gene expression and cellular events within tissues. Amplified radioactive detection may now even further improve the potency of this approach, providing an even greater sensitivity in understanding molecular and cellular events.