Abstract
In situ hybridization (ISH) methods for detection of nucleic acid sequences have proved especially powerful for revealing genetic markers and gene expression in a morphological context. Although target and signal amplification technologies have enabled researchers to detect relatively low-abundance molecules in cell extracts, the sensitive detection of nucleic acid sequences in tissue specimens has proved more challenging. We recently reported the development of a branched DNA (bDNA) ISH method for detection of DNA and mRNA in whole cells. Based on bDNA signal amplification technology, bDNA ISH is highly sensitive and can detect one or two copies of DNA per cell. In this study we evaluated bDNA ISH for detection of nucleic acid sequences in tissue specimens. Using normal and human papillomavirus (HPV)-infected cervical biopsy specimens, we explored the cell type-specific distribution of HPV DNA and mRNA by bDNA ISH. We found that bDNA ISH allowed rapid, sensitive detection of nucleic acids with high specificity while preserving tissue morphology. As an adjunct to conventional histopathology, bDNA ISH may improve diagnostic accuracy and prognosis for viral and neoplastic diseases.
Keywords
C
Among the methods used for detection of molecular markers, in situ hybridization (ISH) offers the unique advantage of visualizing and even quantifying clinically relevant molecules in a morphological context (Unger 2000). The application of nucleic acid target and signal amplification technologies to ISH has enabled researchers to detect as few as one or two copies of specific DNA molecules in cell preparations (Femino et al. 1998; Player et al. 2001). However, the sensitive detection of nucleic acid sequences in tissue biopsy specimens has proven more challenging. In addition to the general problems of ISH, such as diffusion of signal, fluorescence quenching, and preservation of cell morphology, effective detection of nucleic acid sequences in biopsy specimens must overcome challenges unique to tissue sections, including more limited accessibility to target sequences, increased background from nonspecific hybridization, endogenous reporter activity, and preservation of complex tissue architecture.
We recently reported the development of a branched DNA (bDNA) ISH method for detection of DNA and mRNA in whole cells (Player et al. 2001). The bDNA ISH method is a signal amplification system in which target nucleic acid sequences are hybridized to a series of synthetic oligonucleotide probes and visualized through generation of chromogenic or fluorescent signals in an alkaline phosphatase (AP)-catalyzed reaction. The bDNA ISH method is highly sensitive and can detect one or two copies of DNA per cell. Moreover, the bDNA ISH method is specific and provides precise localization, yielding positive signals that are retained within the subcellular compartment in which the target nucleic acid is localized. Our earlier work showed that bDNA signal amplification overcomes many of the challenges of ISH when applied to cell preparations. Damage to cell morphology caused by heat exposure is minimized because bDNA ISH does not require repeated cycling through elevated temperatures. Background caused by endogenous biotin is avoided because bDNA ISH does not utilize an avidin-biotin reporter system. Diffusion of the signal is not a problem because reporter probes for bDNA ISH are physically linked to the spatially fixed nucleic acid targets. Whereas our earlier studies were performed on cervical cancer cell lines, we wished to test whether bDNA ISH would provide the same sensitivity, specificity, and precise localization of nucleic acid sequences in tissue sections.
In this study we evaluated the bDNA ISH method for detection of nucleic acid sequences in clinical tissue specimens. Using normal and HPV-infected cervical biopsy specimens as a model system, we explored the cell type-specific distribution of HPV DNA and mRNA by bDNA ISH with oligonucleotide probes designed for HPV-16 and HPV-18 E6/E7 gene sequences. We assessed the distribution of HPV DNA and mRNA among the various cell types present in cervical intraepithelial neoplasias (CIN), explored the cell type-specificity of HPV localization by comparing epithelial regions with and without dysplasia, and demonstrated the specificity of bDNA ISH by comparing HPV detection in sections from HPV-16- and HPV-18-infected CIN II and normal cervical biopsies.
Materials and Methods
Specimens
Formalin-fixed, paraffin-embedded uterine cervix Leep cone biopsy specimens were obtained from Clinomics Laboratories (Frederick, MD) with Internal Review Board approval. Clinical diagnosis was provided in surgical pathology reports that specified moderate cervical dysplasia (CIN II). The catalog numbers of three HPV-positive and three normal specimens were 00B01-624, 00B01-627, 00B01-625, and H-1188-88, 00B01-631, 00B01-632, respectively. Koilocytes and dysplastic cells were identified using the Bethesda System and as described in diagnostic cytology texts (Koss 1992; Kurman and Solomon 1994).
Oligonucleotide Probes
Target-specific oligonucleotide probes for HPV-16 and HPV-18 E6/E7 and for GAPDH were designed using Probe-Designer Software (Bayer Diagnostics; Berkeley, CA) as described previously (Bushnell et al. 1999). Other DNA oligonucleotide probes comprising the signal amplification system, including preamplifier, amplifier, and alkaline phosphatase (AP)-conjugated label probe, also have been described (Collins et al. 1997). The non-natural nucleotides 5-methyl-2′-deoxyisocytidine (isoC) and 2′-deoxyisoguanosine (isoG) were included in the binding sites of the target, preamplifier, amplifier, and AP-conjugated label probes to reduce nonspecific hybridization (Collins et al. 1997).
Pretreatment of Tissue Sections
Formalin-fixed paraffin sections of cervical tissue were dewaxed and rehydrated with standard histological methods using xylenes and graded ethanol series. For DNA detection, tissue was digested first with 100 μg/ml RNase in RNase buffer (0.5 M NaCl, 10 mM Tris, pH 9.0, 1 mM EDTA) at 37C for 1 hr and then with 12 μg/ml proteinase K in PBS (0.01 M phosphate buffer, pH 7.5) at 37C for 10 min. Immersion in Antigen Unmasking Solution (DAKO; Carpinteria, CA) for 5-40 min at 92C was substituted for proteinase K digestions in some experiments for HPV detection in basal cells and for HPV detection in tissue sections that exhibited higher background staining. After digestion, proteinase K was inactivated by postfixation in 4% paraformaldehyde in PBS at 4C for 10 min. Tissue sections were washed several times in PBS and then acetylated in 1 M triethanolamine and 0.1 M acetic anhydride. Tissue sections were then dehydrated in a graded ethanol series, denatured at 92C for 5 min in 80% formamide in 2 × SSC (1 × SSC is 0.15 M NaCl, 0.015 M Na-citrate) in a humidity chamber (Hybaid Omnislide instrument; Phoenix Research Products, Hayward, CA), followed by immersion in cold 70% ethanol and dehydration. For detection of mRNA, the procedure was the same except that there was no RNase digestion or high-temperature denaturation.
bDNA ISH
After pretreatment (described above), tissue sections were incubated with hyb 1 buffer (50% formamide, 0.2% casein, 3 × SSC, 10% dextran sulfate, 100 μg/ml salmon sperm DNA) at 40C for 30 min. HPV-specific target probes were then added in 80 μl of fresh hyb 1 buffer at a concentration of 10 fmol/μl to each slide and coverslips were placed over slides. Hybridization was carried out in a humidified slide chamber on a slide warmer at 40C for 1-3 hr. After hybridization with target probes, tissue sections were washed in a graded series of SSC buffers from 2 × SSC to 0.1 × SSC for a total of 15 min of wash time. Coverslip-covered tissue sections were incubated with preamplifier probes at a concentration of 1 fmol/μl in hyb 2 buffer (5 × SSC, 10% dextran sulfate, 0.1% sodium dodecyl sulfate, 1 mM ZnCl2, 10 mM MgCl2) at 55C for 25 min, and washed in three changes of 0.1 × SSC for 5 min. Tissue sections were then incubated with amplifier probes at a concentration of 1 fmol/μl in hyb 2 buffer at 55C for 25 min and washed in three changes of 0.1 × SSC for 5 min. Finally, tissue sections were incubated with AP-conjugated label probe at a concentration of 1 fmol/μl in 80 μl hyb 2 buffer at 55C for 15 min, washed in three changes of 0.1 × SSC for 5 min, and washed once in 100 mM Tris-(hydroxymethyl)aminomethane, pH 8.0, 0.1% BRIJ-35, 1 mM ZnCl2, 10 mM MgCl2. Fast Red staining buffer (K0597; DAKO) containing 5 μm levamisole was prepared immediately before application to tissue sections and color was developed for 4-10 min. Nuclei were counterstained for 40 sec with Gills 1 hematoxylin (American Histology Reagent; Modesto, CA), and coverslips were placed on slides for microscopy. Digital images were captured using a Nicon Eclipse microscope, Polaroid digital camera and software, Dell computer with Microsoft Windows 2000 operating system, and combined into figures using Adobe Photoshop 5.5.
Results
Specific Detection of HPV E6/E7 mRNA in CIN II Lesions
A total of 50 serial sections of CIN II tissue from cervical Leep cone biopsy number 627 were analyzed in 12 consecutive experiments using bDNA ISH. A montage of several photographs depicting the distribution of HPV-16 E6/E7 mRNA in CIN II tissue sections is shown in Figure 1. As shown in Figure 1A, hybridization with HPV-16 E6/E7 target probes in tissue sections prepared for mRNA detection consistently yielded signal in the top one third to one half of the squamous epithelia in areas of cells with koilocytic changes, indicating that HPV-16 E6/E7 mRNA was present in these cells. HPV-16 E6/E7 expression was associated with regions of cells that exhibited several of the cytopathic changes that characterize high-grade squamous intraepithelial lesions (HSIL) and/or HPV infection, including high nuclear/cytoplasmic ratios, atypical nuclei with a heterogeneous size and shape, binucleation, and perinuclear clear zones (Koss 1992; Kurman and Solomon 1994). Given the unique perinuclear halo and nuclear atypia, the HPV-16 E6/E7-expressing cells appeared to be koilocytes (Koss 1992). Located between the patches of HPV-16 E6/E7-expressing koilocytes were regions of dysplasia in which HPV-16 E6/E7 mRNA was not detected. In some sections, HPV-16 E6/E7 mRNA was detected at the transformation zone (not shown). In this and other sections of this biopsy, HPV-16 E6/E7 mRNA typically was not detected in regions of epithelial dysplasia, basal cells, proliferating squamous cell precursors, columnar epithelium of the endocervix, or underlying stromal cells.
Localization of HPV-16 E6/E7 mRNA and DNA in sections of CIN II tissue (specimen 627) using bDNA ISH. The presence of viral sequences is visualized by deposition of the Fast Red alkaline phosphatase reaction product (red). For visualization of cell nuclei, sections were counterstained with hematoxylin (blue). (
Spurious staining of the mucus-rich columnar epithelia associated with follicles and other tissue regions was observed in some specimens more than others, regardless of HPV-16 E6/E7 mRNA detection. In most specimens this nonspecific staining was reduced significantly by the addition of acetylation and/or antigen retrieval pretreatment. Another kind of spurious staining was observed in which random deposition of the Fast Red reporter substrate occurred over a few stromal cells (right side of Figure 1A) and over a few cells at lower left of Figure 1D. This staining pattern was not reproduced on adjacent sections and was therefore clearly different from cell type- and probe-specific staining observed in koilocytes that was reproducible on adjacent sections throughout the biopsy.
The highly localized substrate deposition and low background staining of bDNA ISH allowed identification of distinct HPV-16 E6/E7-expressing cells within a tightly packed heterogeneous epithelium containing many cells without detectable HPV-16 E6/E7 mRNA. Figure 1B is a higher-magnification view of the boxed area in Figure 1A. In this region there was a mixture of stained and unstained cells, and several of the cells that stained positive for HPV-16 E6/E7 mRNA also featured cytopathic changes such as multiple nuclei or koilocytic atypia. The subcellular distribution of HPV-16 E6/E7 mRNA was predominantly localized to plasma membrane-associated cytosol and in the nuclei of a few cells but was excluded from the clear perinuclear portion of the cytoplasm. Because the AP-conjugated reporter probe was linked to the target nucleic acid, the AP staining reaction resulted in the precipitation of reporter substrate Fast Red in specific cells without diffusion into adjacent cells. Therefore, individual koilocytes that expressed HPV-16 E6/E7 mRNA were clearly distinguished from adjacent cells that were negative for HPV-16 E6/E7 mRNA.
As a control, an adjacent section prepared for mRNA detection was hybridized with GAPDH-specific target probes. As shown in Figure 1E, this yielded signal that was distributed in cells throughout the squamous epithelium uniformly between koilocytes and dysplastic regions, and also in endocervical columnar epithelium and stromal cells. Detection of GAPDH transcripts throughout the squamous epithelia is consistent with the well-known ubiquitous expression pattern of this housekeeping gene and confirms that this tissue contained RNA available for hybridization to probes.
Comparative Distribution of HPV-16 E6/E7 mRNA and DNA
Adjacent sections of CIN II tissue were analyzed to determine the relative distribution of HPV-16 E6/E7 mRNA and DNA. As shown in Figure 1C, hybridization with HPV-16 E6/E7 target probes in adjacent tissue sections prepared for DNA detection reproducibly yielded signal that was localized primarily to cell nuclei, indicating that HPV-16 E6/E7 DNA was present in regions of cells that expressed HPV-16 E6/E7 mRNA. An adjacent section hybridized with HPV-18 E6/E7 target probes yielded no signal above background (Figure 1D), indicating that the signal detected with HPV-16 E6/E7 target probes was specific. The distribution of HPV-16 E6/E7 DNA in squamous epithelia was generally more extensive than that of HPV-16 E6/E7 mRNA throughout the multiple tissue biopsies examined. For example, HPV-16 E6/E7 DNA was detected in individual basal cells and parabasal cells in addition to areas of cells with koilocytic changes. HPV-16 E6/E7 DNA also was detected in a subset of cells within dysplastic regions. Although a number of cells in dysplastic regions exhibited very strong staining for HPV-16 E6/E7 DNA, staining was not detected in a number of neighboring cells in the dysplastic region. In contrast, throughout the CIN II biopsies examined HPV-16 E6/E7 mRNA was not typically detected in basal cells, proliferating squamous cell precursors, columnar epithelium of the endocervix, or underlying stromal cells, and was not necessarily detected in regions of epithelial dysplasia.
Detection of HPV-16 E6/E7 DNA in Epithelia with Abnormal Squamous Cell Maturation
To determine if the distribution of HPV-16 E6/E7 DNA was associated with regions exhibiting histopathic changes associated with viral infection or neoplasia, 50 sections of each of the HPV-positive biopsies were extensively examined using bDNA ISH. Figures 2A and 2B show two adjacent epithelial regions from the same tissue section of specimen 624. Figure 2A shows a region of normal squamous cell maturation in which HPV-16 E6/E7 DNA was not detected. By contrast, Figure 2B shows a nearby region of the same epithelium that exhibited highly abnormal squamous cell maturation, with moderate dysplasia marked by overgrowth of basal cells and mixing of undifferentiated cells, differentiated squamous cells, koilocytes, and dysplastic cells. In this region, HPV-16 E6/E7 DNA was detected in nuclei of cells exhibiting koilocytic changes and in some cells close to the basement membrane. However, similar to specimen 627, HPV-16 E6/E7 DNA was predominantly detected in regions of cells with koilocytic changes and was not necessarily detected in areas of dysplasia.
Detection of HPV-16 E6/E7 DNA in epithelial regions with abnormal squamous cell maturation. The presence of viral sequences is visualized by deposition of the Fast Red-AP reaction product (red). For visualization of cell nuclei, sections were counterstained with hematoxylin (blue). (
Detection of HPV-16 E6/E7 DNA in Basal Cells of CIN II Lesions
With the bDNA ISH method it was also possible to identify individual basal cells that contained HPV-16 E6/E7 DNA. As shown in Figure 2C, the relatively weak signal for HPV-16 E6/E7 DNA detection in basal cells and parabasal cells near the basement membrane was more clearly distinguished from background by incorporating an antigen-unmasking pretreatment step in the bDNA ISH protocol. Although HPV-16 E6/E7 DNA was detected in some basal cells throughout specimen 624, it was not detected in basal cells of other CIN II specimens examined. In addition to HPV-16 E6/E7 DNA detection in basal cells, specimen 624 also exhibited a greater number of HPV-infected cells and a greater preponderance of signal in these cells.
Genotype-specific detection of HPV E6/E7 DNA in CIN II lesions. The presence of viral sequences is visualized by deposition of the Fast Red alkaline phosphatase reaction product (red). For visualization of cell nuclei, sections were counterstained with hematoxylin (blue). (
Genotype-specific Detection of HPV-16 and HPV-18 E6/E7 DNA in CIN II Lesions
Tissue sections from multiple CIN II biopsies were analyzed to compare the distribution of HPV-16 and HPV-18 E6/E7 DNA. Figure 3 shows sections from two CIN II cervical tissues that were analyzed with bDNA ISH. HPV-16 E6/E7 DNA was detected in CIN II specimen 624 (Figure 3A) but not in CIN II specimen 625 (Figure 3C). By comparison, HPV-18 E6/E7 DNA was detected in CIN II specimen 625 (Figure 3D) but not in CIN II specimen 624 (Figure 3B). These results demonstrate that the detection of HPV DNA in CIN II lesions using bDNA ISH was genotype-specific.
Discussion
In our previous study we demonstrated single-copy gene detection sensitivity for bDNA ISH using cell lines (Player et al. 2001). The present study demonstrates the efficacy of bDNA ISH for detecting nucleic acid sequences in clinical tissue specimens. Our results show that bDNA ISH yielded rapid, sensitive detection of HPV-16 or HPV-18 E6/E7 DNA and mRNA with high specificity while preserving cervical tissue morphology. The specificity of probe hybridization in CIN II biopsies was sufficient to identify biopsy specimens infected with HPV and to distinguish among different HPV genotypes. Furthermore, cell morphology was preserved well enough to distinguish certain cell types that were infected by HPV from neighboring cells in which HPV was not detected. Even in tightly packed heterogeneous epithelia, individual koilocytes that expressed HPV-16 E6/E7 mRNA were clearly distinguished from neighboring cells that were negative for HPV-16 E6/E7 mRNA. Although in cell lines bDNA ISH has been shown to detect as few as one or two copies of DNA per cell (Player et al. 2001), our study did not directly assess the sensitivity of bDNA ISH in tissue sections and it is difficult to make direct comparisons between cell lines and tissues. However, our observation that HPV-16 E6/E7 DNA was detected in individual basal cells, which are thought to have low copy numbers of HPV (Stoler and Broker 1986), indicates that the sensitivity of bDNA ISH in tissues is sufficient to detect low-abundance molecules.
In overcoming the challenges for localization of nucleic acid sequences in clinical tissue sections, bDNA ISH offers many advantages over target amplification methods such as PCR ISH. For example, the histopathological features of formalin-fixed, paraffin-embedded tissue sections are well preserved with bDNA ISH, in part because bDNA signal amplification does not require repeated heat cycling (Nuovo et al. 1991). Because bDNA ISH does not need polymerase activity, it is not affected by the unpredictable presence of polymerase inhibitors in tissue specimens that may affect reproducibility of PCR ISH methods (Daniel et al. 2000). Moreover, because the AP-conjugated reporter bDNA probe is physically linked to the target nucleic acid, the reaction product is retained at the site, thus allowing precise subcellular localization of the target nucleic acid and avoiding the problems of diffusion and false-positives associated with PCR ISH methods (Nuovo et al. 1991; Wiedorn et al. 1999).
Differences in technology distinguish bDNA from other signal amplification systems such as tyramide signal amplification (TSA) applied to ISH. TSA and bDNA ISH have similar sensitivities in cell lines and are effective in detection of target sequences in tissues (Kerstens et al. 1995; Plummer et al. 1998; Player et al. 2001). Although both bDNA and TSA improve the sensitivity of ISH by increasing the numbers of binding sites for reporter molecules, they use different mechanisms to achieve amplification of the signal. TSA is based on the application of biotinylated tyramide peroxidase substrate to the avidin-biotin system and therefore relies on the high affinity constant and specificity of the avidin-biotin complex (Bayer and Wilchek 1978; Wilchek and Bayer 1988). In contrast, bDNA is based on hybridization of nucleic acids and therefore relies on stringent hybridization conditions involving complementary sequences of oligonucleotides, elevated temperatures, the addition of non-natural nucleotides that reduce nonspecific binding, reduced probe concentration, and stringent washing conditions involving buffers with decreasing salt concentrations. Because bDNA signal amplification is based on the hybridization of deoxyoligonucleotides rather than on the protein-protein interactions of TSA, it affords a number of operational opportunities to control the signal amplification specificity not available to the TSA system. The two methods also differ in the type of reporter system used. TSA ISH uses peroxidase-dependent tyramide technology that is either fluorescent or chromogenic (van de Corput et al. 1998), whereas bDNA ISH uses an AP-catalyzed reporter system that provides the opportunity to use AP substrates such as the highly sensitive Fast Red chromogen substrate (DAKO) that is both chromogenic and fluorescent (Player et al. 2001).
Although this study focused primarily on the efficacy of bDNA ISH for detection of HPV in cervical tissue specimens, bDNA ISH may be applicable to genotype- and cell type-specific detection of high-risk HPV genotypes in other tissues with HPV-associated cancers, such as head and neck (Gillison et al. 2000; Mork et al. 2001), lung (Cheng et al. 2001), and melanoma (Dreau et al. 2000). In addition, because oligonucleotide probes for bDNA signal amplification can be readily designed (Bushnell et al. 1999), bDNA ISH can be developed for detection of other nucleic acid sequences that are related to cancer and infectious disease. For example, we have reproducibly detected HIV-1 RNA in lymph node tissue from HIV-1-infected asymptomatic individuals (data not shown). These results suggest that bDNA ISH may provide a specific and sensitive means for detecting genetic markers and gene expression in a variety of specimen types.
Studies using sensitive techniques such as bDNA ISH to further investigate the topographical distribution of HPV and neoplasia in cervical biopsies might offer unique insight into the disease process. Although the potential clinical role of HPV testing is controversial (Kaufman and Adam 1999; Manos et al. 1999; Bishop et al. 2000; Bovicelli et al. 2000), infection with high-risk HPVs has been associated with the progression of intraepithelial precursor lesions to cervical carcinoma (zur Hausen 1996), and HPV-16 and HPV-18 have consistently been found in over 90% of cervical carcinomas and in 50-80% of CIN lesions (Arends et al. 1998). In this study we found that HPV-16 E6/E7 mRNA and DNA were present in areas of cells with koilocytic changes but not necessarily in regions of dysplasia. Our observations suggest that HPV E6/E7 gene expression may not be directly associated with the overgrowth of epithelial progenitors and dysplastic cells but rather may influence neoplastic processes indirectly. However, examination of a greater number of CIN biopsies as part of a comparative clinical study would be needed to determine the association between the distribution of E6/E7 mRNA and DNA from high-risk HPV genotypes and the development of high-grade squamous intraepithelial lesions.
In summary, bDNA ISH is a reproducible, sensitive, specific, and precise method for localizing nucleic acid sequences in clinical tissue sections with exquisite preservation of tissue morphology. As an adjunct to conventional histopathology, bDNA ISH may contribute to the understanding, diagnosis, and prognosis of disease.
Footnotes
Acknowledgment
We thank Linda Wuestehube for critical reading of the manuscript and editorial expertise.