Part I:

A. INTRODUCTION
B. BASIC IN SITU HYBRIDIZATION PROTOCOL

1. Tissue Preparation
2. Sectioning of Frozen Tissue
3. Slide Preparation/Tissue Adherence
4. 35S-labeled Riboprobe Preparation

Part II:

B. BASIC IN SITU HYBRIDIZATION PROTOCOL (continued)

5. Hybridization
6. Modifications for Paraffin Sections or Cultured Cells
7. Hybridizations Using Synthetic Oligomers
8. Controls
9. Non-radioactive In Situ Hybridization
10. Size Reduction of Riboprobes
11. In Situ PCR
12. In Situ PCR or PCR and In Situ Hybridization?

Part III:

C. SUMMARY
D. Selected Bibliography-In Situ Hybridization Protocol
E. References
F. Protocols and Reagents

1. In Situ of Frozen Sections Using 35S Riboprobes
2. 35S-Riboprobe Synthesis
3. Labeling of Synthetic Oligomers with 35S-CTP
4. Tissue Preparation Protocol
5. Buffers and Solutions of In Situ

Part IV:

F. Protocols and Reagents (continued)

6. Non-radioactive In Situ Using Biotin-labeled Riboprobes
7. Synthesis of Biotinylated Riboprobes
8. Sample Calculations for Non-radioactive In Situ
9. Solutions for Non-radioactive In Situ (biotinylated)
10. Non-radioactive In Situ Using Digoxigenin-labeled Riboprobes
11. Transcription of Digoxigenin-labeled Riboprobes

G. Sources of Material for In Situ
H. Acknowledgments

INTRODUCTION

In situ hybridization is an invaluable tool for the examination of gene expression. The technique was first described as a method for the localization of DNA:RNA hybrids in cytological preparations by Gall and Pardue in 1969 (Gall, Pardue, 1969). Subsequently, papers appeared in which the technique was used to localize mRNAs for globin (Harrison et al 1973; Conkie et al 1974), visna virus (Brahic, Haase, 1978), growth hormone (Hudson et al 1981), beta-endorphin/ACTH (Hudson et al 1981; Gee et al 1983; Gee, Roberts, 1983) and prolactin (Pochet et al 1981). A major advance in the method was achieved with the description in 1984 of in situ hybridization using single stranded RNA probes (Cox et al 1984a) otherwise known as "riboprobes". In situ hybridization is very similar to Northern blots and depends on the hybridization of a labeled nucleic acid probe (RNA or DNA) to a complementary sequence of mRNA. The two techniques differ in that the starting material for a Northern blot is a tissue digest while the primary material for in situ hybridization is a histological tissue section. All of the cellular relationships are lost with Northern blots and mRNA levels are averaged from all of the cells contained in the original sample. However, in situ hybridization is exquisitely sensitive and can detect the amount of mRNA contained in a single cell. Furthermore, since in situ hybridization is a histological technique cellular relationships are maintained and it is possible to precisely identify cell types expressing the gene of interest. Thus, potentially important interactions between cells that express different proteins may be uncovered.

A major advantage of in situ hybridization is that it allows the maximum use of tissues that may be in short supply (i.e. clinical biopsies, embryos, cultured cells). A tissue digest from a surgical biopsy might yield sufficient RNA for one or two Northern blots. These Northern blots would only yield information regarding the presence or absence of mRNAs without any other information as to the cellular source of that RNA. While Northern blots can be probed multiple times for different mRNAs, in fact this may be limited to four or five different probes at most, as the transfer membranes break down and there can be some RNA loss with repeated stripping of the blots. However, literally hundreds of different hybridizations can be performed on the same piece of tissue using situ hybridization. A single surgical biopsy of a human atherosclerotic plaque coming from carotid endarterectomy surgery, for example, might yield up to 1000 tissue sections each of which can be used for a separate hybridization. Kept at -70°C with desiccant, these tissue sections can be stored for more than six years without significant loss of the hybridization signal (Wilcox, unpublished observations). It is possible therefore to make libraries of tissues, stored as sections in the freezer that can be probed in the future as new probes become available or new questions arise. The repeated hybridization of these tissues with multiple probes over a long period of time also allows for the eventual serial reconstruction of a tissue with all of the information concerning mRNA content, cell identity, and regional localization available for study.

There are many different ways to do in situ hybridization including probing with cDNAs, cRNAs, or synthetic oligonucleotides. There are also multiple ways of labeling these probes using radioactive (3H, 32P, 33P, 35S) or non-radioactive (biotin, alkaline phosphatase, digoxigenin, fluorescent) nucleotides. Over the past 10 years we have compared most of these methods in my laboratory and have come to the conclusion that 35S riboprobes represent the most sensitive method for the detection of mRNA in tissue sections (Table I).

Table I. Comparison of different labeling strategies for in situ hybridization ranked according to their relative sensitivity.

Riboprobes > cDNAs > Synthetic oligomers 35S > 32P > 3H >> biotin/digoxigenin Frozen > Paraffin

BASIC IN SITU HYBRIDIZATION PROTOCOL

Based On 35S-Labelled Riboprobes And Frozen Sections

The goal of this presentation will be to present an overview of in situ hybridization first introducing the various uses of the procedure in the analysis of local gene expression in tissues and then focusing on some of the more technical aspects of the procedure dealing with the choice of probe and the question of appropriate controls. I will begin by describing the primary in situ hybridization protocol used in my laboratory which is based on 35S-labeled riboprobes and frozen sections. We have used this particular protocol with great success and reliability on a variety of different tissues ranging from surgical biopsies to animal tissues including well over 1000 different tissues and over 100 different probes without modification. The success rate of this method with a new probe and tissue is over 90% on the first series of hybridizations thus providing rapid results without the necessity of a series of experiments establishing optimal conditions individualized for a specific tissue or probe. It is hoped that this protocol may provide a starting point for individuals wishing to begin in situ hybridization in their own laboratory.

The method we employ for in situ hybridization is basically a modification of the original protocol for radiolabeled riboprobes as presented by Cox and colleagues (Cox et al 1984a) and has been used by our lab for many different projects (Majesky et al 1990; Nelken et al 1991; Rosenthal et al 1987; Wilcox, 1992; Wilcox, Derynck, 1988; Wilcox et al 1989a; Wilcox et al 1989b; Wilcox et al 1988). This method can be adapted for synthetic oligomers (40-50mers), cultured cells or paraffin embedded tissues.

Tissue Preparation:

The tissues are fixed by immersion in 4% paraformaldehyde 0.1M NaPO4. The paraformaldehyde should be freshly prepared at least weekly to prevent the formation of degraded aldehyde by-products which may cause an increase in the autoradiography background. The optimum period for fixation is from 1-3 hrs for tissue from 1 - 10mm3. This process does not necessarily fix the entire tissue block. Large tissue blocks will have a core of relatively unfixed tissue while small blocks will be fixed to a much greater extent. However, the purpose of the fixation step is not to completely fix the tissue but only to preserve and harden the tissue sufficiently to improve the tissue morphology from that of unfixed snap frozen material. Complete fixation is not critical at this point in the protocol since a second fixation step on the tissue sections is performed just prior to hybridization which will fix all sections equally.

While tissues may be fixed for periods longer than 3 hrs, extended fixation times create additional problems. Over fixation often results in a greater loss of tissue from the slides during the hybridization process. Excessive cross-linking of the tissue by paraformaldehyde may reduce potential cross-linking to charged molecules on the coating the slide which are responsible for tissue adhesion. Over fixation also reduces the in situ hybridization signal presumably by limiting probe access to cellular mRNAs. This can sometimes be recovered by increasing the concentration or time in proteinase K (Armstrong et al 1992).

Fresh frozen tissue without fixation can be used for in situ hybridization. Tissue should be rinsed in saline or PBS and frozen in liquid nitrogen in O.C.T. blocks as outlined above. While not optimal, tissue can also be snap frozen without an embedding matrix and used for hybridization as well. The main determinant of the fixation method is based on the tissue morphology after hybridization. Some tissues such as lung are not suitable for frozen sections and need to be prepared as paraffin blocks for the best cellular morphology. Paraffin embedded tissues can be used for in situ hybridization with an approximate 25% mRNA signal loss. This loss of signal intensity is acceptable especially in those tissues in which frozen sections are not practical. Additional details on using paraffin embedded sections will be covered later in this review.

Sectioning of Frozen Tissue:

The tissues are sectioned on a cryostat to 5-10µm thickness, thaw-mounted onto coated slides (0.2N NaOH-PLL, Vectabond, or Fisher Superfrost/Plus), and immediately re frozen by placing the slides in plastic slide boxes kept in cytostat or on dry ice. After completion of the sectioning the slides are stored at -70°C with Humi-Cap desiccant capsules (United Desiccants, Pennsauken, NJ) in the slide boxes. The mRNA on the slides is stable when stored in this fashion and we have used tissue sections for up to 8 years after sectioning with excellent hybridization results. This can be very valuable when working with human tissues or small tissue biopsies and enables the researcher to conduct multiple studies on related adjacent sections over the years as new probes become available. In addition it is possible to store hundreds of sections typical of a particular animal model and continue to examine the expression of additional mRNAs without having to prepare new tissues. Given that the major effort of doing in situ hybridization lies in the preparation of the tissue sections, it is relatively easy for a researcher to hybridize tissue sections from a wide range of animal models, human biopsies, and normal tissues when probing with new cDNAs. This type of random screening does not take very much additional time but may yield a wealth of information and suggest new directions of research.

Slide preparation/tissue adherence:

One of the biggest problems with in situ hybridization is keeping the tissue sections on the slide during the hybridization and washing procedures. A variety of slide coatings have been tested (Table II and III) and while some seem to be better than others none are absolutely perfect. We have found that the best coatings for tissue retention include the poly-L-lysine and Vectabond (Vector Laboratories, Burlingame, CA). Vectabond slide coating is a proprietary modification of the amino-acyl silane procedure and has given excellent in situ hybridization results with good tissue morphology after the hybridization procedure. We have recently begun to use Superfrost/Plus microscope slides (Fisher Scientific) for all of our frozen tissue sectioning. All three methods of slide coating appear to have similar properties as regards tissue morphology/retention on the slide after the hybridization procedure (Table III). The advantage of using Superfrost/Plus slides is they require no preparation time in the laboratory and are competitive in terms of cost when you consider technician time and reagent expenses.

Sections placed on poly-L-lysine, Vectabond, or Superfrost Plus slides do not require coverslips during the hybridization procedure. With gelatin coated slides the hybridization buffers tend to roll off to one side of the tissue because the gelatin layer wets unevenly. Poly-L-lysine and silane coatings on the other hand do not wet and the high surface tension in the hybridization bubble keeps the buffer in place. As long as the hybridizations are conducted in a humid chamber containing a buffer (4xSSC, 50% formamide) with a similar vapor pressure as the hybridization buffer the sections will not dry out during the overnight incubation. Water alone should not be used to maintain the humid atmosphere as the hybridization solutions are hygroscopic and will take up water during the incubation diluting out the probe.

To compare different slide coatings we have prepared tissue sections from different tissues that in particular tended not to adhere well during the in situ hybridization procedure. These tissue sections were then carried through a mock hybridization procedure (no labeled probe added) and the ultimate tissue morphology evaluated after the final washing and drying steps. The following slide coatings were evaluated.

Maple's: Slides were acid washed by boiling at 130°C for 10min in H2SO4:HNO3 (9:1), then allowed to cool to room temperature and washed thoroughly in dH2O. The washed slides were then soaked in 2% aqueous solution of triethoxy-3-aminopropyl silane pH3.0 overnight at 65°C, washed in dH2O, immersed in 10% glutaraldehyde pH6.5 for 60min, and washed in dH2O. Prior to use the slides were activated by treatment with 0.15M sodium metaperiodate for 30min followed by washing in dH2O and drying. Some slides were coated as above except substituting dH2O and ethanol washes similar to that used in the PLL method below for the acid wash step (ETOH/Maple's).

Amino-acyl silane (AAS): The slides were washed in dH2O, dried in an oven, and dipped for 10-15 sec in a solution of 0.5, 1.0, or 2.0% triethoxy-3-aminopropyl silane (Sigma) in acetone. The slides were then dried, rinsed in at least 3 changes of dH2O followed by drying at 50°C and storage.

Poly-L-lysine (PLL): Slides were washed for 2min each in dH2O and 100% ethanol twice followed by drying in an oven at 42°C. The slides were dipped in 100µg/ml poly-l-lysine (poly-L-lysine MW 47,000, Sigma) in sterile dH2O for 20sec. The slides were allowed to dry for 30min at room temperature, dipped again in the poly-L-lysine solution, and then dried overnight at 42°C. The slides can be stored at room temperature for up to 6 months.

NaOH/PLL: This method included initial immersion of the slides in 0.2N NaOH for 30min, followed by 2 dH2O washes, 2 ethanol washes and poly-L-lysine coating as above. Some slides were etched in 1.0N NaOH or by boiling in H2SO4:HNO3 as outlined in the Maples procedure (Acid/PLL) prior to PLL coating but this did not appear to improve tissue retention on the slides.

Gelatin Chrome Alum Subbing: This was a standard coating technique used in many histology laboratories consisting of 2.5g/l gelatin and 0.25g/l chrome alum dissolved in dH2O and used to coat ethanol washed slides.

Vectabond: This was used as directed by the manufacturer. Slides were washed for 5 min in acetone, immersed in a Vectabond/acetone solution (1bottle Vectabond with 350mls of acetone) for 5 min, rinsed for 1 min in dH2O with some agitation, drained and dried overnight at 42°C.

Neoprene: Neoprene (1.5g) was dissolved in 300ml iso-pentyl acetate in a fume hood over a 6 hr period (do not apply heat as the iso-pentyl acetate is highly flammable). The slides were then treated for 5 min in iso-pentyl acetate, 5 min in 0.5% neoprene/iso-pentyl acetate, followed by a wash for 30 sec in dH2O, and allowed to dry overnight at 37°C before using.

Table II. Tissue adherence to glass microscope slides prepared with different coatings. At least 6 slides from 2 tissues in each experiment were evaluated for tissue retention and morphology after a mock in situ hybridization. The results were graded as complete tissue loss (0), few fragments of tissue left (1), fair morphology with moderate folding or tissue loss (2), good tissue morphology without tissue loss (3).
PLL	1.8
Acid/PLL	1.8
NaOH/PLL	2.7
0.5% AAS	1.8
1.0% AAS	1.6
2.0% AAS	1.3
Maple's	2.0
Gelatin Chrome Alum	0
Vectabond	3.0

Table III. Comparison of Vectabond and Superfrost/Plus slides for tissue adherence after in situ hybridization. Mock hybridizations were conducted as indicated in Table II and scored from 0-3 for the amount of tissue left after hybridization.
Superfrost/Plus	2.9
Vectabond coated Superfrost/Plus	2.2
Sigma Silane-Prep	0
NaOH/PLL	2.9
Neoprene	1.9
Vectabond	1.9

35S-labeled Riboprobe Preparation:

Probes are labeled by transcription (Melton et al 1984) using 35S-labeled UTP (specific activity >1000 Ci/mmol; Amersham). An aliquot of the 35S-labeled nucleotide is pipetted into a micro centrifuge tube and dried in under vacuum with centrifugation in a Speed-Vac (Savant). The transcription reaction is begun by adding in sequential order: 2.0µl 5x Transcription buffer (containing 200mM Tris-HCl, pH 7.5, 30mM MgCl2, 10mM spermidine, 50mM NaCl; Promega), 1.0µl 100mM DTT, 1.0µl RNasin (20-40 units; Promega), 1.0µl linearized plasmid DNA as a template (1.0µg/µl), 2.0µl of a GCA solution containing 2.5mM each of GTP, CTP, and ATP in H2O, and 2.0µl sterile dH2O. The components are thoroughly mixed to ensure that the labeled nucleotide is in solution and then briefly centrifuged to bring the mixture to the bottom of the tube. The reaction is then started by the addition of 1.0µl of the appropriate RNA polymerase (SP6, T7, or T3) depending on the orientation of the cDNA insert or the vector used. The polymerase is added and mixed gently by pipetting without vortexing or centrifugation and the reaction incubated for 60min at 37°C. To increase the amount of probe produced additional enzyme can be added after 60min and the reaction allowed to continue for another 60min. The size of the probes produced should be checked on a 5.2% acrylamide, 7M urea gel and autoradiography. While there will be many small bands on the gel the major band of transcribed material should be equal to the size of the cDNA insert.

Note that the protocol presented here does not call for base hydrolysis of the riboprobe prior to use in hybridization studies. We use full length (which is really a mixture of sizes) transcripts from probes up to 1.3 kb in length. If the probe is longer than that it would be preferable to find internal restriction sites for linearization so the resulting template will be less than 1.3 kb. Alternatively short non-overlapping 5' and 3' fragments of the probe could be subcloned into riboprobe vectors. These can be linearized and used individually or as a cocktail for transcription and hybridization. Base hydrolysis produces fragments with a wide range of sizes, the smaller fragments produced may lack specificity and increase the spurious binding of the probe in the tissue thus increasing background (see later section on "Size Reduction of Riboprobes" for more discussion of this issue).

The concentration of 35S-labeled UTP in the transcription reaction should exceed 10µM without adding any unlabeled material. It is generally more difficult to get full length transcripts using nucleotide concentrations less than 10µm as the labeled nucleotide becomes rate limiting and the reaction stutters. It is important that the 35S-labeled UTP is used immediately after thawing and that any remaining label should be discarded rather than re frozen for use at a later time. We have found that the single greatest contribution to high backgrounds using 35S -labeled probes comes from repeated freezing and thawing of the isotope prior to use. According to Amersham there is about a 5% degradation of the nucleotide with freezing and thawing that might contribute to background problems. It is therefore convenient to use only 250µCi vials of nucleotide, dividing each vial into 2 transcription reactions with 125µCi each. If the specific activity of the isotope is 1200Ci/mmol and the final volume of the transcription reaction 10µl, the final concentration will exceed 10µM.

After the transcription reaction 1.0µl of the reaction mixture is removed, diluted 1:1000 and an aliquot pipetted onto a small piece of DE81 paper (Whatman) to monitor incorporation of the 35S label. The unincorporated nucleotide is washed off the paper with 3 x 5min washes in 0.5M sodium phosphate buffer pH7.4. The filter paper is rinsed briefly in H2O and ethanol, dried, and counted in the scintillation counter.

After removing an aliquot to monitor incorporation of nucleotide 1.0µl of RQ1 DNase (1 unit; Promega) is added and the reaction allowed to proceed for an additional 15min at 37°C to degrade the template. The probe is then cleaned up by a single phenol chloroform extraction after the addition of 20µl 1xTE (to increase the volume) and 1.0µl of tRNA (50mg/ml in dH20) to act as a carrier, RNase sink, and to provide a visible pellet in future ethanol precipitations. After the phenol chloroform extraction the aqueous extract is brought up to 100µl with the addition of 1xTE and ethanol/sodium acetate added to precipitate the RNA. After centrifugation, washing in 70% ethanol, and drying, the pellet is resuspended in 1xTE to a final concentration of 300,000 CPM/µl using the total incorporation determined above as a guide. Aliquots of the probe (100µl) can then be reprecipitated with ethanol/sodium acetate and the probe stored this way for up to 1 week at -70°C. It is convenient to make multiple aliquots of the probe for a series of hybridizations over the following week rather than continuously thaw and refreeze a single aliquot or to prepare probe daily. We have examined probe stability over time when stored under ethanol/sodium acetate and have found that there is detectable degradation of the probe size after a week which could affect the background or hybridization result. Given the relative ease of probe preparation relative to the amount of effort placed in the hybridization procedure and the time one invests in long autoradiography exposures it is preferable to use a freshly transcribed probe for these studies.