ANNALS OF CLINICAL AND LABORATORY SCIENCE, Vol. 24, No. 4 Copyright © 1994, Institute for Clinical Science, Inc. mRNA In-situ Hybridization Using Biotinylated Oligonucleotide Probes: Implications for the Diagnostic Laboratory* JULIANA G. SZAKACS, M.D.t and SANDRA K. LIVINGSTON, HTL (ASCP)tt tDepartment of Pathology, tMolecular Histology Laboratory University of South Florida, College of Medicine Tampa, FL 33612

ABSTRACT It is now possible to detect low copy numbers of messenger ribonucleic acid (mRNA) while retaining good histologic morphology for the determi­ nation of specific gene expression in diseased tissues. This technology will allow the pathologist to provide important prognostic information about tumors (expression of oncogenes and growth factors), to identify the sub­ clones within the tumor which may be most likely to metastasize (expres­ sion of adhesion molecules and proteases) and to identify etiologic genetic aberrations (viral insertions). A technique for in-situ hybridization to mRNA has been developed for use in formalin fixed paraffin embedded tissues which is suitable for a hospital histology laboratory. Optimal con­ ditions for the procedure were determined by using a biotinylated poly (d)T oligonucleotide probe. Results were dependent on the tissue type, fixation time, condition of the tissue prior to fixation, and degree of diges­ tion before hybridization. The temperature and conditions of hybridization were optimized so that the poly d(T) control probe and the longer test probe could be run simultaneously. Streptavidin and avidin alkaline phos­ phatase detection systems were tested using levamisole to minimize back­ ground staining, and a biotin blocking agent to reduce reaction to renal tubular biotin. Increasing the temperature of stringency washes did not significantly improve the specificity but had a markedly detrimental effect on tissue morphology. The mRNA appears to remain stable within routinely fixed surgical mate­ rial over long periods of time allowing for large retrospective studies. A review of c-erbB-2 expression in 16 human breast lesions was carried out comparing mRNA in-situ hybridization to immunoperoxidase and cytosolic methods. By direct localization of both message and antigen, it was possi­ ble to demonstrate focal positivity that cytosolic methods did not detect. Aberrant translation was noted in one case, and c-erbB-2 expression in non-malignant breast was detected in two cases.

* Send reprint requests to Juliana G. Szakacs, M.D., Department of Pathology, White River Junction VAM & ROC, White River Junction, VT 05009. 324 0091-7370/94/0700-0324 $02.00 © Institute for Clinical Science, Inc. mRNA IN-SITU HYBRIDIZATION 325 Introduction more sensitive in certain cases than Northern blotting when only a small number of cells are expressing the mes­ Routine molecular histology is now sage in a total quantity too low to detect commercially available in kit form for on a blot. deoxyribonucleic acid (DNA) in-situ Gene activation results in the forma­ hybridization (ISH) to viral genes, and tion of heterologous nuclear RNA recently kits and nonradioactive detec­ (hnRNA) by RNA polymerase II. Within tion systems geared to mRNA in-situ the nucleus a 3'-poly A tail (composed of hybridization have been developed up to 300 adenylyl residues) is added to which greatly simplify this process. Bio- the hnRNA along with a 5'-methyl cap. tinylated oligonucleotide probes have The RNA then undergoes splicing which been found to be both highly sensitive removes the intron or non-coding and specific,1 eliminating the need for sequences resulting in mature mRNA special radiation licenses, waste disposal, which is transported from the nucleus to and the need to label probes fresh with the cytoplasm for translation. The aver­ each run. Biotinylation of probes can be age cell contains approximately 360,000 performed enzymatically in the labora­ molecules of mRNA which are divided tory or oligonucleotide probes can be into 10,000 to 20,000 different species. synthesized using biotinylated nucleo­ This means that many specific messages tides. These remain stable for 6 to 12 will be found in low numbers, less than months. The use of shorter oligonucleo­ 30 molecules per cell.3 Since the majority tide probes as opposed to longer probes of mRNA contains a poly A tail, a poly raised in plasmids has not compromised d(T) probe was used to detect mRNA specificity, and they are much easier to preservation in fixed tissues and to opti­ obtain. They are single stranded DNA mize fixation, digestion, and hybridiza­ which means that no dénaturation step is tion parameters for varying tissue types.3 required prior to hybridization, and there Once this is accomplished, studies of is no decrease in efficiency owing to specific messages may be undertaken. sense strands binding to probe. The These may include studies for the pres­ selection of the probe sequence, how­ ence of oncogene expression for correla­ ever, becomes important in shorter tion with protein expression studies probes to ensure specificity. Computer (c-erbB-2 in breast carcinoma is our programs are now available to screen example) or the development of probes selected sequences against all known for the study of protein products for genetic sequences. These programs will which there is no available antibody or identify the probability of hybridizing to for tissues in which the protein epitopes the selected message and to any other are lost during processing. In-situ genetic material with similar sequences. hybridization to mRNA will allow detec­ It has now been shown by correlation tion of gene activation even if protein is of in-situ hybridization and northern blot not expressed owing to malfunctions of analysis that ISH, using non-radioactive translation; however, correlation with the labels, is sensitive enough to detect as protein product will provide definitive few as 10 molecules of mRNA per cell knowledge of gene activation and protein with some authors claiming the ability to synthesis within a cell type. detect one copy per cell.2 In addition, The identification of oncogene ampli­ ISH will define the specific cell type or fication and expression in tumors is subclone which is the source of the mes­ becoming the standard of care. The sage within the tissue sample and may be immense field of molecular diagnostics is 326 SZAKACS AND LIVINGSTON making its way into the diagnostic labo­ anti-c-erbB-2 immunohistochemistry ratory and soon therapies will be based and the cytosolic measurement of on the presence or absence of onco­ the oncoprotein. genes, drug resistance genes or muta­ tions which have been found to cor­ relate with the progression of disease, Materials and Methods response to therapy and survival. Excel­ lent reviews have been prepared by RNAse free deionized water was pre­ Richardson4 and Yamamoto.5 pared by the addition of 1 ml/L diethyl Within the last few years, oncologists pyrocarbonate (DEPC, Sigma D5758)*, have been looking for prognostic indica­ with overnight incubation followed by tors which will help them determine the autoclaving to remove DEPC. The need for adjuvant therapy in node nega­ DEPC water was used for all steps prior tive breast cancer patients with small pri­ to and including hybridization. All glass­ mary tumors. Although 80 to 90 percent ware was baked at 200°C for a minimum of these patients do well with lumpec­ of four hours to destroy RNAses, and ster­ tomy and radiation alone, a small number ile pipette tips and plasticware were would derive significant benefit from employed where possible. adjuvant chemotherapy. In an attempt to identify this group of patients, a number of tumor parameters are being studied T is s u e s and have been associated with poor prog­ nosis. These include: aneuploidy, high Fischert 344 rat tissues were used to cycling index, lack of estrogen and pro­ determine optimum fixation and diges­ gesterone receptors, and increased epi­ tion times for the poly d(T) probe and to dermal growth factor receptor (EGFr), establish the procedure. Under general cathepsin-D, p53, and c-erbB-2.6,7 How­ anesthesia, rats were exsanguinated and ever, no single finding is predictive of all tissues were sliced into 5 mm sections outcome on a case by case basis. Most and immediately placed in 10 percent studies of c-erbB-2 (also known as HER2/ buffered formalin for 4 to 72 hours. neu) have been done by protein analysis Screened tissues included: brain, spinal in cytosolic extractions from tumor sam­ cord, eye, tongue, trachea, esophagus, ples8,9,10 or quantification of DNA ampli­ lungs, heart, aorta, liver, muscle, skin, fication by Southern blotting.11,12,13 In stomach, small and large bowel, kidney, both of these techniques, normal and adrenal, bladder, prostate, and testes. malignant cell populations within the Sixteen paraffin blocks of human breast sample are studied, which could result in tissue removed surgically for diagnosis at a small subclone of highly malignant the Moffitt Cancer Center! were tested cells expressing the oncogene being for c-erbB-2 mRNA and oncoprotein. diluted out. Very few correlations of gene These tissues were handled routinely, amplification, expression, and translation and fixation varied from 12 to 72 hours within one tumor have been carried out. prior to processing. Early results suggest that these parame­ ters may not always be concordant.14 Which, if any, of these parameters is most * Sigma Chemical Company, P.O. Box 14508, St. closely associated with tumor progres­ Louis, MO 63178. t Fischer 344 rats, % Charles River Breeders, sion and survival is not known. A study Kingston, NY 12401. has been initiated by us to correlate t Moffitt Cancer and Research Center, 12902 in-situ hybridization to mRNA with Magnolia Dr. Tampa, FL 33612. mRNA IN-S1TU HYBRIDIZATION 327 Q uantitation o f So l u t io n s c-e r b B-2 O n c o p r o t e in 20X SSC ph 7.0 3M NaCl, 0.3 M trisodium citrate Fresh frozen tissue samples were sent 50X Denhardt’s Solution to Dianon Systems § for cytosolic quanti­ Ficoll 400 5 gm fication of c-erbB-2 oncoprotein at the Polyvinyl pyrrolidine 5 gm BSA 5 gm time of surgery. Quantities less than 2000 DEPC water to 500 ml FM/mg are considered a favorable prog­ 2X Hybridization buffer nostic indicator. A section of each of the 20X SSC 2.0 ml 0.2 M NaP04, pH 6.5 7.6 ml tumors was stained by routine immuno- 50X Denhardt’s Solution 0.4 ml histochemical methods with the poly­ 20% Dextran sulfate in clonal antibody anti-c-erbB-2 (Zymed formamide dextran sulfate 10 gm Laboratories, #18-01 ll ).11 absolute formamide 50 ml Prehybridization Solution 2X hybridization buffer 500 ul P r o b e s a n d D e t e c t io n Kit s Vanadyl Ribonucleoside 120 ul 20% dextran S04/formamide 620 ul Probe/Hybridization Solution Probe in prehybridization A c-erbB-2 anti-sense oligonucleotide solution to a final probe was designed to the molecularly concentration of 0.1 to cloned cDNA for c-erbB-2 mRNA, bases 0.5 (Jig/ml 20X Modified Tris buffered 3424 to 3465.15 This segment of the gene saline (TBS) was searched against GenBank® Release 1 M Tris, 3 M NaCl, 40 76.0, 4-15-93 for possible non-specificity mM MgCl2, 2% (w/v) BSA and cross reaction to epidermal growth IX Modified TBS/Triton X factor receptor (EGFr) and fibroblast 20X Modified TBS 50 ml growth factor receptor (FGFr). It was 10% Triton X 10 ml Millipore or sterile 940 ml found to have the smallest Poisson Proba­ autoclaved water bility of 1.3e-10 of mismatch and no cross Phosphate buffered saline reaction to EGFr or FGFr.16 The probe (PBS) pH 7.4 137 mm NaCl, 2.7 mM was synthesized, biotinylated, and puri­ KC1, 8.1 mM Na2HP04, fied by gel-filtration by the DNA Synthe­ 1.5 mM KH2P04 sis Laboratory of the Interdisciplinary IX Modified TBS Center for Biotechnology Research at the 20X Modified TBS 50 ml University of Florida!. Millipore or sterile 950 ml A poly d(T) biotinylated oligonucleo­ autoclaved water tide probe was obtained from British Bio­ technology Limited** (BBS9) as part of the ISH detection kit. Procedure All tissues were formalin fixed (4 to 72 hours), paraffin embedded, and cut in 3 micron sections. Precharged glass slides § Dianon Systems Inc., 200 Watson Boulevard, were used to prevent tissue slippage Stratford, CT 06497. || Zymed Laboratories, 458 Carlton Court, South (table I). Sections were oven dried over San Francisco, CA 94080. night, dewaxed in xylene, soaked in abso­ f DNA Synthesis Laboratory, Box 100266 HSC, lute alcohol, and then air dried for five Rm Rl-244 ARB, University of Florida, Gainesville, FL 32610-0266. minutes. The protocol for in-situ hybrid­ ** R&D Systems (British Biotechnology), 614 ization with the poly d(T) positive control McKinley Place, NE, Minneapolis, MN 55413. probe was followed per BBS kit using 328 SZAKACS AND LIVINGSTON

TABLE I

Outline of Protocol

Gloves Must be worn throughout procedure to protect slides from RNAses on skin. Tissue Formalin fixed, paraffin embedded, cut in 3 micron sections on precharged slides (Fisher #12-550-15), dried overnight and deparaffinized. Digestion Proteinase K 20 mg/mi (AMRESCO E-195) 0, 10, 20, 40, 80, or 100 |4.g/ml, 37° C, 10 min. RNAse 61 mg/ml (Sigma R-9134) control slides only, 0.1 mg/ml, 37° C, 1 hour. Biotin block Avidin-biotin blocking kit (Vector #SP2001) added to buffer of renal sections prehybridization only. Postfixation 0.4% paraformaldehyde, 4° C, 10 min. Prehybridization In humidified chamber, cover tissue sections with prehybridization solution, incubate at 37° to 42° C, 1 hour. Hybridization Drain slides (except “no probe control”), apply hybridization solution to cover tissue, incubate in humid chamber 37° to 42° C, overnight. Washes Two changes, 5 minutes each: 4X SSC 37°, 42°, 56°, or 60° C. 2X SSC 37°, 42°, 56°, or 60° C. 0.2X SSC, 37° C. or, 2 changes, 5 min each: 4X SSC plus 30% formamide, 37° C. 2X SSC plus 30% formamide, 37° C. 0.2X SSC, 37° C. 1X modified TBS/Triton (Sigma T6878), room temperature, 15 min. Detection In situ hybridization and detection kit from British Biotechnology Ltd (BBS9). Unconjugated streptavidin alkaline phosphatase, BCIP/NBT detection kit (DAKO K600). Conjugated streptavidin alkaline phosphatase, BCIP/NBT detection kit (DAKO K601). Unconjugated avidin alkaline phosphatase, naphthol fast red (Vectastain ABC-AP, AK 5000, Vector Red SK 5100). Mount Aqueous mounting solution Control slides 1-poly d(T) Detects preservation of mRNA. 1 - no probe Reveals nonspecific reactions to detection system.1 1 - RNAse with poly d(T) Confirms RNA is target. 1 - RNAse with probe Confirms RNA is target.

Fisher Scientific, 7445 Exchange Drive, Orlando, FL 32809. AMRESCO, Inc., 30175 Solon Industrial Parkway, Solon, OH 44139. Vector Laboratories, Inc., 30 Ingold Road, Burlingame, CA 94010. SigmaChemicalCo.,P.O. Box 14508,St. Louis, MO 63178. British Biotechnology, Ltd., 614 McKinley Place, NE, Minneapolis, MN 55413. Dako Corporation, 6392Via Real, Carpinteria, CA 93013. non-formamide containing buffers and (xg/ml to determine optimum digestion hybridization solutions provided. This for each tissue type and length of fixation. protocol was modified by adding a pro­ Digestion was followed immediately by teinase K digestion step following depar- immersion in cooled paraformaldehyde affinization. Proteinase K was heated to to protect the newly exposed message 37°C for 40 minutes to remove RNAses from RNAses. prior to digestion of tissue sections. The Slides were rinsed in two changes of concentration was varied from 0 to 100 phosphate buffered saline, pH 7.4 (PBS) mRNA IN-SITU HYBRIDIZATION 329 then DEPC water and placed in IX modi­ ABC-APt systems. Slides were counter­ fied tris buffer. Control slides for RNAse stained with either nuclear fast red, digestion were removed at this point and malachite green or methyl green and incubated at 37°C in RNAse solution for coverslipped using an aqueous mount­ one hour followed by thorough rinsing ing medium. with DEPC water and phosphate buff­ ered saline (PBS). All slides were drained (but not allowed to dry) and Results placed in a humid chamber. Adequate prehybridization solution to cover the tis­ sue sections (150 to 250 |xl) was applied ISH T e c h n iq u e and slides were incubated at 37 to 42°C for one hour. All slides except the Multiple variables in the development no-probe controls were drained prior to of the in-situ hybridization procedure application of the probe mixture and were studied. Tissues required rapid incubated at 37 to 42°C in the humid immersion in fixative. Time to fixation chamber 1, 4 or 18 hours. Probe mixtures greater than 15 minutes resulted in weak­ were prepared to yield a final concentra­ ening of mRNA signal. Partially necrotic tion of probe between 0.1 and 0.5 (jug/ml. tissue, such as that seen in rapidly grow­ Renal sections were treated with an avi- ing tumors, also tended to lose mRNA din/biotin block during prehybridization. more rapidly. Fixation times were varied The avidin solution was mixed 1:2 v/v from 4 to 72 hours with mRNA detection with prehybridization buffer and incu­ possible in all tissues. A “rim” effect was bated for one hour at 42°C in the humid noted particularly in liver sections with chamber. This was followed by a brief decreased mRNA signal from the more rinse with warm PBS. The biotin solution fixed rim around the section. Increasing was applied and incubated at 42°C for 30 the proteinase K concentration revealed minutes, then rinsed with warm PBS and mRNA in this rim but resulted in destruc­ DEPC water. Hybridization was then tion of the central portion of tissues. carried out as per protocol. Digestion with proteinase K was nec­ Following hybridization, all slides essary for all tissue types. As the concen­ were rinsed using a wash bottle with 4X tration increased initially, the mRNA sig­ SSC at 37°C and placed in Coplin jars for nal became stronger to a point where stringency washes. Two changes for five cell-to-cell discrimination/variation was minutes each of the graded salt solutions lost and all cells were darkly stained. were carried out at either 37°, 42°, 56°, or Above this concentration, signal was lost 60°C, or 30 percent formamide was added and eventually tissue morphology was to the graded salt solutions and used at destroyed (figure 1). The concentration 37°C. All slides were then washed with needed to produce this range of changes IX modified TBS/Triton X for 15 minutes varied by tissue type and length of fixa­ at room temperature. Detection of the tion. For tissue fixed less than 24 hours biotinylated probe was carried out using optimum concentrations of proteinase K either the British Biotechnology BBS9,* were: (a) liver, lung and highly vascular Dako K600t, Dako K601t, or Vector tissues: 20 |xg/ml, (b) heart, muscle, GI: 40 |Ag/ml, and (c) brain: 100 |xg/ml. * R&D Systems (British Biotechnology), 614 McKinley Place, NE, Minneapolis, MN 55413. t Dako Corporation, 6392 Via Real, Carpinteria, t Vector Laboratories, Inc., 30 Ingold Road, Bur­ CA 93013. lingame, CA 94010. 330 SZAKACS AND LIVINGSTON

TABLE II

Calculation of the Melting Point in Formamide Containing Buffers

DNA-DNA Tm = 81.5 + 16.6 X Log [Na+] - 0.65 X (% formamide) + 41 X (G + C content) RNA-RNA Tm = 79.8 + 18.5 X Log [Na+j - 0.35 X (% formamide) + 58.4 X (G + C content) + 11.8X(G + C content)

Tm = melting point.

Within each of these tissues, certain the same time by using formamide in the cell types were also more susceptible to hybridization mix. For the poly d(T) digestion, with the vascular endothelium probe which has a melting point of being most sensitive. Increasing the ~42°C, no formamide was added so that length of time of slide immersion in the hybridization could be achieved at proteinase K solution as opposed to 37-42°C. For the c-erbB-2 probe which increasing concentration did not improve has a melting point of ~65°C, 50 percent digestion quality, in fact, prolonged formamide in the hybridization mix times (> 1 0 minutes) resulted in loss of reduced the relative temperature of dis- mRNA signal. When no digestion was association to ~42°C. To calculate the performed even overnight hybridization temperature of hybridization for a DNA- failed to give a signal. RNA hybrid, the mean of the DNA-DNA The poly d(T) probe obtained from and RNA-RNA may be used17 (table II). BBS is a short ~30 base pair (bp) oligo­ Stringency washes to remove nonspe- nucleotide which gave a strong signal cifically bound probe and improve predominantly in the nuclei of all cell background were tested. Washes at 56° types. In all cases of appropriately fixed and 60°C resulted in loss of tissue mor­ and digested viable tissue, mRNA was phology. Washes at 37° or 42°C gave ade­ identified. Tumor tissues with necrotic quate specificity and background control; regions were the least likely to have pre­ however, to improve specificity, the served mRNA. Blocks of archived mate­ addition of 30 percent formamide to the rial up to two years old were found to first two graded salt washes was under­ contain mRNA. taken with good results and maintenance Hybridization time requirements were of tissue morphology. Poly d(T) hybrid­ studied. A one hour hybridization ized slides were washed with graded salt resulted in no signal at any concentration solutions without formamide at 37°C with of proteinase K digestion. A four hour good results. hybridization time resulted in a weak sig­ Detection systems from British Bio­ nal at higher levels of proteinase K diges­ technology Systems* (BBS), Dakot and tion. Overnight incubation (12 to 20 Vectort were tested, and all gave ade­ hours) gave the strongest signal for all quate staining when following the manu- proteinase K concentrations. Small probes such as the poly d(T) required * R&D Systems (British Biotechnology), 614 slightly less time for hybridization than McKinley Place, NE, Minneapolis, MN 55413. the longer c-erbB-2; however, this was t Dako Corporation, 6392 Via Real, Carpintería, not significant, and both probes could be CA 93013. X Vector Laboratories, Inc., 30 Ingold Road, Bur­ run side by side in the procedure. lingame, CA 94010. Hybridization temperatures could be X Vector Laboratories, Inc., 30 Ingold Road, Bur­ adjusted to run short and longer probes at lingame, CA 94010. mRNA IN-SITU HYBRIDIZATION 331

FIGURE 1. Messenger ribonucleic acid in-situ hybridization (mRNA ISH) with the poly d(T) probe to liver varying the concentration of proteinase K digestion. (A) No digestion performed, no mRNA detected; (B) 40 |xg/ml proteinase K digestion, detection of mRNA in the nuclei; (C) 60 (xg/ml proteinase K digestion, detection of mRNA in nuclei and cytoplasm; (D) 100 jJLg/ml proteinase K digestion, loss of morphology. Original magnification 400x. 332 SZAKACS AND LIVINGSTON facturers’ directions. The BBS kit gave the most specific staining pattern with the least background to our way of think­ ing. The nitroblue tetrazolium / 5-bromo- 4-chloro-3-indolyl phosphate (NBT/ BCIP) substrate gave a clearer, more readily localized signal than the naphthol red. Its granular precipitate at the site of hybridization gave good contrast to most counterstains and did not fade. The NBT/ BCIP substrate is sensitive to organic sol­ vents, and any prolonged exposure resulted in the formation of large blue crystals and loss of all specificity. Background staining owing to endoge­ nous alkaline phosphatases could be revealed by omitting the streptavidin or avidin alkaline phosphatase step from the procedure. The addition of levamisole to the substrate eliminated almost all back­ ground owing to endogenous alkaline F ig u r e 2. Focal positivity for c-erbB-2 messen­ phosphatase. Levamisole, however, ger ribonucleic acid (mRNA) by in-situ hybridiza­ could not prevent staining owing to alka­ tion (ISH) in an infiltrating duct carcinoma, not line phosphatase in the small bowel detected by cytosolic studies for oncoprotein. Origi­ nal magnification 630x.

FIGURE 3. Intraductal carcinoma showing strong positivity for c-erbB-2 messenger ribonucleic acid (mRNA) by in-situ hybridization (ISH) (A), but negative for oncoprotein by immunoperoxidase (IPX) in a serial section (B). Original magnification 250x and 400x. mRNA IN-SITU HYBRIDIZATION 333

FIGURE 4. Weak staining for c-erbB-2 messenger ribonucleic acid (mRNA) in an intraductal portion of comedo carcinoma (A), but strong staining for the oncoprotein by immunoperoxidase (IPX) in a serial section (B). Original magnification 250 x. Stronger staining for mRNA by in-situ hybridization (ISH) is seen in an invasive component of the same tumor (C), which is positive for the oncoprotein on a serial section (D). Original magnification 630x. 334 SZAKACS AND LIVINGSTON

FIGURE 5. Apocrine metaplasia in a fibrocystic breast is strongly positive for c-erbB-2 messenger ribo­ nucleic acid (mRNA) by in-situ hybridization (ISH) (A) and oncoprotein by immunoperoxidase (IPX) (B). Original magnification 100 x. lumen. The addition of levamisole prior fast red tended to blend with the purple to the substrate or conjugate steps did not color of the NBT and a slightly better make a significant difference in back­ contrast was achieved with a green coun­ ground. Endogenous biotin, found in terstain. Because of the sensitivity of large quantities in the renal tubular epi­ NBT to organic solvents, attention was thelium, interfered with the detection paid to the type of counterstain cho­ system. In our studies it was found that as sen, and all slides were mounted in aque­ the hybridization times increased there ous media. Attempts at permanent was an increase in nonspecific staining in mounting resulted in eventual loss of the renal tubules. This could be avoided stain or crystallization. with the use of a biotin block during pre­ hybridization. Non-mRNA binding of probe could be detected by using an Comparison o f c-erb B -2 mRNA ISH RNAse treated control slide. As the and Anti-c-erbB-2 probes aged (> 6 months), nonspecific Immunoperoxidase t o binding became more evident. Cytosolic Quantitation Counterstaining to improve identifica­ tion of cell morphology was carried out Nine cases of infiltrating duct carci­ using nuclear fast red, methyl green, or noma, one intraductal carcinoma, two malachite green. In slides with very invasive lobular carcinomas, one adeno­ weak NBT/BCIP staining, the nuclear carcinoma, one comedo carcinoma, and mRNA IN-SITU HYBRIDIZATION 335 TABLE III ing pattern for both mRNA by ISH and protein by IPX. Within this case, hyper­ Detection of c-erbB-2 by Method in 16 Cases plastic, nonmalignant glands were nega­ of Human Breast Lesions tive for both mRNA and protein. The case Number Positive / Total Number of Cases of comedo carcinoma was strongly posi­ Cytosolic mRNA tive for protein by the cytosolic method Histology Protein ISH IPX (11308 FM/mg) which corresponded with the IPX method. However, mRNA detec­ Infiltrating duct 2/9 4/9 4/9 tion by ISH was weak in the ducts which Intraductal 0/1 focal focal contained a marked amount of protein Lobular 0/2 0/2 0/2 (figures 4A and 4B). Typical central Adenocarcinoma 0/1 0/1 0/1 necrosis was apparent in these ducts and Comedo 1/1 1/1 1/1 destruction of the mRNA may be postu­ Fibrocystic disease 0/2 2/2 2/2 lated. In areas of micro invasion, malig­ nant cells demonstrated positivity for mRNA = messenger ribonucleic acid. both mRNA by ISH and protein by IPX ISH = in-situ hybridization. (figures 4C and 4D). IPX = immunoperoxidase. Two cases of fibrocystic disease were negative for cytosolic oncoprotein; how­ two cases of fibrocystic disease were ever, by ISH and IPX both message and studied (table III). By cytosolic evalua­ protein were strongly identified in areas tion, two of the infiltrating duct carcino­ of apocrine metaplasia (figure 5). Hyper­ mas were positive for c-erbB-2 oncopro­ plastic glands and normal benign tissues tein (2707 and 7567 FM/mg). Both of were negative. these cases were positive for mRNA by ISH and protein by immunoperoxidase (IPX). In addition, two cases which were Discussion not identified as containing c-erbB-2 pro­ tein by the cytosolic method were found The identification of mRNA by ISH to be positive for mRNA by ISH and for requires the timely handling of tissues. protein by IPX. Both of these revealed Tissues which remained on the bench for focal staining patterns of a minority of prolonged periods (i.e., while frozen sec­ malignant cells within the tumor. In one tions were being performed) lost mRNA case the cells which stained positively owing to cellular RNAses. Routine forma­ were of higher grade and in smaller clus­ lin fixation and processing provided tis­ ters within the extracellular matrix (fig­ sue samples that were adequate for ure 2). Five cases of infiltrating duct car­ mRNA detection. The addition of a diges­ cinoma, two cases of lobular carcinoma tion step with proteinase K improved and one case of adenocarcinoma were detection of message in all tissues, and negative for both mRNA by ISH and pro­ determination of optimum digestion was tein by IPX. obtained for each tissue type by varying One case of intraductal carcinoma, the concentration of proteinase K in solu­ negative for protein by the cytosolic tion. In general 20 to 40 (xg/ml proteinase method, was studied. There was very K at 37°C for 10 minutes was a good start­ strong focal staining for mRNA by ISH in ing point. Tissues fixed for more than 12 glands that were negative for protein by hours required a higher concentration of IPX on serial sections (figure 3). In other proteinase K. Highly vascular (liver) or glands, there was a heterogenous stain­ porous tissues (lung) required less, while 336 SZAKACS AND LIVINGSTON brain was found to require the highest Detection systems were tested per pro­ concentration of proteinase K (100 fxg/ tocol with varying incubation times from ml). To maintain optimum morphology 1 to 16 hours. The BBS and DAKO K601 and to be able to provide a semiquantita- kits gave excellent results, the unconju­ tive evaluation of message from cell to gated DAKO K600 was slightly less sen­ cell, it was found that the best digestion sitive, and the BBS kit gave the least was at a concentration of 20 (xg/ml less background. The NBT/BCIP with its than that which provided the strong­ granular precipitate tended to be more est signal. specific than the naphthol red substrate, Prehybridization and hybridization which tended to spread throughout the solutions containing vanadyl ribonucleo- cell. It was important not to expose BCIP/ sides reduced non-specific binding and NBT stained slides to excessive organic protected message from RNAses.18 Dex- solvents after counterstaining, this tran sulfate and polyvinyl pyrrolidine resulted in crystallization and eventual increased the viscosity of the solution to loss of stain. The routine use of levami- slow random molecular motion. Hybrid­ sole in the substrate minimized non­ ization temperature was optimized at specific staining owing to endogenous 20° to 25°C below the melting point (Tm) alkaline phosphatase and studies of the of the probe duplex using 50 percent kidney required a biotin block. formamide to increase the stringency of In general, detection of mRNA within hybridization without increasing the tissue sections could be evaluated on a temperature. This maximized the spe­ semi-quantitative basis, grading degree cific binding and minimized the interac- of stain intensity on a 1 to 4 scale similar tion with partially homologous to immunohistochemistry. However, sequences and permitted us to hybridize because of the variation in fixation time, the shorter poly d(T) probe and longer degree of loss of mRNA prior to fixation, c-erbB-2 probe at the same temperature. digestion variability from cell to cell and Hybridization times from one hour to tissue to tissue, this quantitation could overnight were studied, and it was found not be used to compare one tissue sample that times less than four hours were to another. inadequate. Overnight incubations of 12 The initial findings of ISH and IPX to 20 hours provided good results and compared to the cytosolic method for could be more easily worked into the detection of the oncoprotein c-erbB-2 laboratory schedule. suggest that multiple methods of detec­ Stringency washes were tested using tion may be warranted to obtain an accu­ ionic solutions, heat, and formamide to rate understanding of the presence or determine optimum specificity. While all absence of the oncogene. Sampling error of these methods work well with hybrid­ may be greatest with the cytosolic ization procedures on nitrocellulose, it method, in which small numbers of posi­ was found that high temperature washes tive malignant cells may be diluted out (>56°C) compromised tissue morphology. by the normal stromal component. Detec­ The combination of 2X SSC and 30 per­ tion by ISH of mRNA may be more accu­ cent formamide at 37°C provided excel­ rate in identifying small numbers of cells lent results with our probes whose melt­ expressing the oncogene; however, ing temperatures were between 62° and necrosis of adjacent tumor tissue may 65°C. For the poly d(T) probe, no form­ release RNAses and other enzymes amide should be used since its Tm is which result in the rapid destruction of quite low (~42°C). the message, or prolonged time between mRNA IN-SITU HYBRIDIZATION 337 surgery and fixation of the tissue may Acknowledgements result in message loss owing to intrinsic Appreciation is extended to Aaron Binstock, medi­ RNAses. Immunoperoxidase staining by cal student, for assistance in the laboratory, and to Jeno polyclonal antibody is fairly specific and E. Szakacs, M.D. for the superb photomicroscopy. identifies small numbers of cells, but in rare cases the protein may not be present References despite oncogene expression owing to 1. Leary JJ, Brigatii DJ, Ward DC. Rapid and sen­ aberrant translation. At this time it is not sitive colorimetric method for visualizing known whether or not these cases fall biotin-labelled DNA probes hybridized to DNA or RNA immobilized on nitrocellulose: into the category of poorer prognosis, and Bio-blots. Nucleic Acids Res 1985;13:1777-99. further studies are warranted. As with all 2. Greil R, Fasching B, Huber H. In-situ hybrid­ antibodies, specificity is important. Occa­ ization for the detection of low copy numbers of c-abl oncogene mRNA in lymphoma cells: tech­ sional staining of fibroblasts within the nical approach and comparison with results tumor was noted suggesting cross reac­ with anti-oncoprotein antibodies. Lab Invest tivity of this antibody to FGFr. The find­ 1989;60:574-82. 3. Pringle JH, Primrose L, Kind CN, Talbot IC, ing of both gene expression and transla­ Lauder I. In situ hybridization demonstration tion in apocrine metaplastic cells indi­ of poly-adenylated RNA sequences in formalin- cates that c-erbB-2 is not in itself an fixed paraffin sections using a biotinylated oli­ gonucleotide poly d(T) probe. J Pathol 1989; indicator or cause of malignant transfor­ 158:279-86. mation, and cytosolic studies of benign 4. Richardson GE, Johnson BE. The biology of lesions should be reported cautiously. lung cancer. Semin Oncol 1993;20:105-27. 5. Yamamoto T. Molecular basis of cancer: onco­ genes and tumor suppressor genes. Microbiol Immunol 1993; 37:11-22. Conclusions 6. Berns EMJJ, Klijn JGM, van Stavern IL, Por- tengen H, Noordegraaf E, Foekens JA. Preva­ lence of amplification of the oncogene c-myc, A two to three day mRNA in-situ HER2/neu, and int-2 in one thousand human hybridization protocol requiring a total of breast tumors: correlation with steroid recep­ tors. Eur J Cancer 1992;28:697-700. four to five hours of hands on time was 7. Machotka SV, Garrett CT, Schwartz AM, Calla­ successfully incorporated into our his­ han R. Amplification of the proto-oncogenes tology laboratory. The use of the BBS int-2, c-erbB-2 and c-myc in human breast can­ ISH kit for the poly d(T) positive control cer. Clin Chim Acta 1989;184:207-18. 8. Bacus SS, Bacus JW, Slamon DJ, Press MF. probe allowed initial establishment of HER-2/neu oncogene expression and DNA fixation, digestion and hybridization ploidy analysis in breast cancer. Arch Pathol Lab Med 1990;114:164-9. parameters and provided a sensitive and 9. Dati C, Muraca R, Tazartes O, Antoniotti S, Per- specific detection system. Probes hybrid­ roteau I, Giai M, Cortese P, Sismondi P, Saglio ized under conditions which preserved G, De Bortoli M. C-erbB-2 and ras expression levels in breast cancer are correlated and show tissue morphology were found to provide a co-operative association with unfavorable out­ excellent results. Initial correlation stud­ come. Internat J Cancer 1991;47:833-8. ies of c-erbB-2 oncogene and protein 10. McCann A, Johnston PA, Dervan PA, Gullick WJ, Carney DN. C-erbB-2 oncoprotein expres­ expression proved to be more sensitive in sion in malignant and nonmalignant breast tis­ identifying small subclones of positive sue. LJMS 1989;June: 137-^0. cells within a tumor sample than the 11. Garcia I, Dietrich PY, Aapro M, Vauthier G, Vadas L, Engel E. Genetic alterations of c-myc, cytosolic method and were also able to c-erbB-2, and c-Ha-ras protooncogenes and detect cases in which the oncogene mes­ clinical associations in human breast carcino­ sage was expressed although the protein mas. Cancer Res 1989;49:6675-9. 12. Heintz NH, Leslie KO, Rogers LA, Howard PL. was not translated. Further investigation Amplification of the c-erbB-2 oncogene and into the prognostic importance of these prognosis of breast adenocarcinoma. Arch cases is warranted. Pathol Lab Med. 1990;114:160-3. 338 SZAKACS AND LIVINGSTON 13. Lonn U, Lonn S, Nylen U, Winblad G, Sten- c-erb-B-2 gene to epidermal growth factor kvist B. Amplification of oncogenes in mam­ receptor. Nature 1986;319:230-4. mary carcinoma shown by fine needle biopsy. 16. Altschul SF, Gish W, Miller W, Myers E, Lip- Cancer 1991;67:1396-1400. man DJ. Basic local alignment tool. J Molec 14. Bandyopadhyay, D. Correlations between Biol 1990;215:403-10. c-erbB-2 oncogene amplification and the 17. Kricka, Larry J. Nonisotopic DNA Probe Tech­ expression of its mRNA and protein in human niques. New York, Academic Press, Inc, Har- breast carcinomas. Oncology 1992;49:9-14. court Brace Jovanovich, Publishers, 1992:66-9. 15. Yamamoto T, Ikawa S, Akiyama T, Semba K, 18. Ogilvie AD, Wood NC, Dickens E, Wojtacha D, Nomura N, Miyajima N, Saito T, Toyoshima K. Duff GW. In situ hybridization. Ann Rheum Similarity of protein encoded by the human Dis 1990;49:434-9.