DNA Isolated From Parafin-Embedded Tissues and Dot Hybridization

Transcription

1 American Journal of Pathology, Vol. 127, No. 3, June 1987 Copyright American Association of Pathologists RAPID COMMUNICATION Detection of Cytomegalovirus Infection by Means of DNA Isolated From Parafin-Embedded Tissues and Dot Hybridization EDWARD SETO, BS, and T. S. BENEDICT YEN, MD, PhD From the Department ofpathology, School ofmedicine, University of California, San Francisco A specific and rapid method for detecting the presence of cytomegalovirus (CMV) DNA in formalin- or mercuric chloride-fixed, paraffin-embedded tissue has been developed. With dot hybridization with a biotinylated CMV DNA probe, the presence of CMV infection can be readily detected even if only a minority of the cells in the tissue block are infected. The sensitivity THE DIAGNOSIS of infectious diseases in biopsy specimens and postmortem examinations is an important, yet difficult aspect of histopathology. Many infectious agents are difficult or impossible to visualize by light microscopy even with special stains, while others produce nonspecific histopathologic findings. Furthermore, speciation is rarely possible. Therefore, definitive diagnosis of infection usually depends on culturing. However, cultures must be performed on fresh tissues and are not available for certain microorganisms. Recently it was reported that DNA could be recovered from paraffin-embedded tissue blocks and analyzed by Southern blot hybridization."2 Together with the development ofcloned DNA probes specific for infectious agents, this points to the possibility of diagnosing infections in tissue blocks by the techniques of molecular biology. We here report the sensitive and specific detection of cytomegalovirus (CMV) infection in paraffin blocks of tissues by extracting DNA and performing dot blots with CMVspecific probes. Materials and Methods Preparation of Tissue DNA DNA from paraffin blocks was isolated according to the method of Goelz et al' with slight modifica- is further increased by about two orders of magnitude if high specific-activity 32P-labeled probe is used. This method of detection is more sensitive than histologic diagnosis, appears to be roughly comparable to that by virus isolation in culture, and should be applicable to the detection of other infectious agents. (Am J Pathol 1987, 127: ) tions. Briefly, thick sections were cut from paraffinembedded tissue blocks and as much ofthe paraffin as possible was removed from the tissues with a razor blade. The tissues were finely minced and suspended in 5-10 ml of 500 mm Tris-HCl (ph 9), 20 mm EDTA, 10 mm NaCl containing 1% sodium dodecyl sulfate (SDS), and 500,ug/ml proteinase K. After vortexing at high speed for 3 minutes, the mixture was incubated at 48 C for 24 hours. The sample was vortexed again for 3 minutes, and additional proteinase K and SDS were added for a final concentration of 1 mg/ml and 2%, respectively. The mixture was incubated for an additional 40 hours and vortexed again. DNA was then purified by three to four extractions with phenol and chloroform, precipitated with sodium acetate and ethanol, and collected by centrifugation.3 After being washed with 70% ethanol, the DNA pellets were dried under a vacuum desiccator before being resuspended in 3 mm Tris-HCl (ph 7.2) and 0.2 mm EDTA(TE). The concentration of DNA was determined by absorbance at wavelength 260 nm. Supported by FPC USA. Accepted for publication March 18, Address reprint requests to T. S. Benedict Yen, Department of Pathology, HSW-501, Box 0506, School of Medicine, University of California, San Francisco, CA

2 410 SETO A YEN Unfixed DNA was prepared from K562 leukemia cells (ATCC CCL243) by standard techniques.3 Preparation of DNA Probes Biotinylated CMV probe comprising a mixture of two fragments ofcmv (Towne strain) genomic DNA sequences with sizes of 17.2 kb and 25.2 kb was purchased from Enzo Biochem Inc. (New York, NY). Biotinylated human Alu repetitive sequence4 was prepared by nick-translation with biotin- 1 I-dUTP with the use of a kit from Bethesda Research Laboratories (Gaithersburg, Md). For preparation of high specific-activity 32P-CMV probe, a plasmid clone (prl- 1) containing a 25.2-kb CMV genomic DNA fragment of the UL region was kindly provided by R. LaFemina and G. Hayward.5 Bacterial growth, plasmid amplification, and DNA isolation were done according to Maniatis et al.3 The insert fragment was purified by electrophoresis in agarose gel and then labeled with a-32p-dctp to a specific activity ofapproximately 108 cpm/,ug ofdna by means of the Nick Translation Kit from Bethesda Research Laboratories (Gaithersburg, Md). Unincorporated 32P-dCTP was removed from the preparation by spin-column procedure.3 The specificity of this probe was confirmed by Southern blotting of CMV, herpes simplex Types I and II, and Epstein-Barr viral DNA, which revealed hybridization only to the CMV DNA (data not shown). Preparation of Dot Blots and Hybridization Conditions DNA was sheared by passing through a 25-gauge needle, diluted to the appropriate concentrations in a final volume of 30,ul of TE buffer, and heat denatured.3 After the addition of 100,ul ice-cold 20 X SSC solution (3 M sodium chloride and 0.3 M sodium citrate, ph 7.0), the entire sample mixture was aspirated onto a Schleicher & Schuell BA-85 nitrocellulose filter (prewetted in 2 X SSC) using an S & S manifold. The filter was then baked at 80 C for 2 hours in a vacuum oven. For biotinylated CMV probe, all hybridization procedures were performed as described by Bethesda Research Laboratories' manual for the BluGene Nonradioactive Nucleic Acid Detection System at a probe concentration of 300 ng/ml. The sensitivity of the method was estimated by blotting several dilutions of known quantities of the CMV DNA on the same filter. For 32P-labeled CMV probe, all procedures followed closely to Maniatis et al,3 with a few modifications. Nitrocellulose filters were prehybridized with 6 X SSC, 0.5% SDS, 5 X Denhardt's solution, and 100,ug/ml denatured herring sperm DNA for 6 hr at 68 C. The prehybridization solution was replaced with a mixture containing 6 X SSC, 0.01 M EDTA, 5 X Denhardt's solution, 0.5% SDS, 100,ug/ml denatured herring sperm DNA, and 10 ng/ml 32P-labeled denatured CMV-probe DNA. Hybridization was carried out overnight at 68 C, and the blots were washed twice for 1 hour each in 0.1 X SSC, 0. 1% SDS at 68 C (high stringency). The extent of hybridization was determined by autoradiography. The sensitivity was estimated as described for the nonisotopic hybridization above. Results DNA isolated from the paraffin blocks (yield of ug/mg tissue) had OD260/OD280 ratios ranging from to 1.896, whereas DNA recovered from the fresh K562 cells had a ratio of Thus, our DNA preparations were essentially free of protein contamination. One of the DNA samples isolated from the paraffin blocks was chosen at random for use in comparing the hybridizability of DNA from such preparations to the DNA isolated from fresh cells using a cloned Alu repetitive DNA fragment as probe. The results (Figure 1) indicate that DNA isolated from paraffin-embedded tissue is suitable for hybridization experiments, and the process of fixation and paraffin embedding does not significantly alter the ability of DNA to hybridize to probes. For testing the ability of our CMV DNA probe to detect viral infection in tissue blocks, paraffin blocks of lung from 5 autopsy cases with clear histologic evidence of CMV infection were obtained, DNA extracted, and dot blots prepared with the biotinylated probe. As seen in Figure 2A and Table 1 (Blocks Fresh Fixed Ng AJP * June 1987 Figure 1-Dot blot analysis of DNA isolated from cultured K562 cells (fresh) and DNA recovered from paraffin blocks (fixed). The indicated amounts of DNA were sheared, transferred onto nitrocellulose filters, and hybridized to biotinylated human Alu DNA probe. There is no indication for preferential hybridization to DNA isolated from cell cultures.

3 Vol. 127 * No. 3 A M 14 U C" Da 7 B mo" yg DNA ig DM Figure 2-Illustrative dot blots of DNA isolated from tissue blocks and probed for CMV DNA. DNA isolated from paraffin blocks was diluted into different concentrations are spotted onto nitrocellulose filters and hybridized to either (A) biotin or (B) 32 P-labeled CMV probes. See Table 1 for description of the different cases. CMV DNA FROM PARAFFIN BLOCKS ), all showed strong positivity, indicative ofthe expected hybridization. Moreover, 3 cases ofcultureproven CMV pneumonia gave positive results, despite the lack of typical CMV inclusions by histology (Table 1, Cases 20-22). However, DNA from another case that was CMV-positive by culture but negative by histology did not produce a detectable signal with this method (Figure 2A and Table 1, Blocks 23 and 24). We therefore repeated the dot blots, but using a 32P-labeled probe instead of the biotinylated probes. As can be seen from Figure 2B and Table 1 (Blocks 14-18, 23, and 24), all ofthe DNA samples now gave a positive signal, including DNA extracted from the histologically unremarkable kidney (Block 26) of the patient from whom came Blocks 23 and 24. Thus, with radiolabeled probes, our blotting technique is more sensitive than light-microscopic histopathology in detecting CMV infection. The technique works equally well with Zenker's-fixed tissues (Figure 2B and Table 1, Blocks 27 and 28). Conversely, dot blots are quite specific, in that no false-positive results were obtained from three blocks of CMV culture-negative pneumonia (Figure 2B and Table 1, Blocks 1-2 and 19), as well as 11 blocks of histologically normal lung (Blocks 3-13) chosen randomly from autopsies of subjects over age 35 with no history of immunodeficiency. These results also provide convincing evidence that lungs from persons with previous exposure to CMV (as evidenced by positive serology), but with no active infection, do not give positive results by dot blots. That is, because the incidence of positive CMV serologic tests is at least 60% in the general American population over age 35,6 the probability is >99.99% (1-0.4") that at least Table 1 -Detection of CMV Infection With DNA Isolated From Paraffin-Embedded Tissues and Dot Blot as Compared With Other Available Clinical Diagnoses Block t 25t 26t Tissue Liver Kidney Fixative Zenker Zenker Histology l CMV infection detected by Culture Biotin Dot blots *Histologic study of Block 19 shows basophilic smudging of the lining alveolar cells, as well as one cell with nuclearchanges suggestive, but not diagnostic, of CMV infection. tblocks were samples obtained from the same patient (see text). Each of the other blocks originated from different patients., not done.

4 412 SETO A YEN one ofthe 11 randomly chosen people would be serologically positive; yet none gave a positive result by dot blot. The absolute sensitivity of our detection methods was estimated by dotting known quantities of cloned prl- CMV DNA on filters and probing with either biotinylated or radiolabeled probes. As seen in Figure 3, the former method could easily detect 100 pg of CMV DNA, whereas 1 pg was detectable with the latter method. Because on our blots 1 jug of normal human DNA gave a blank spot with no nonspecific signal (Figure 2B, Block 1), the lower limit of detection is 10-4 g ofcmv DNA/g tissue DNA for biotinylated probes. As each human cell contains approximately 3.5 X 109 pairs ofdna versus 2.5 X I04 base pairs for the cloned CMV fragment, this is equivalent to the ability to detect the presence of 14 CMV genomes per cell. Because an infected cell can contain up to 106 CMV genomes,7 it is likely that this method can detect CMV in tissues even early in infection, when only a minority ofcells are infected. In addition, use of radiolabeled probes can increase the sensitivity by almost two orders of magnitude. However, further studies with more cases would be needed to determine whether these techniques are as sensitive as those using cell culture. Discussion Using the methodology developed by Goelz et al,' we have been able to extract DNA from paraffin-embedded blocks of biopsy and autopsy specimens. Although the DNA was too fragmented for Southern blotting (data not shown), in all cases tested it was Bioti 32 p 10@ pg DNA Figure 3-Determination of sensitivity of CMV DNA detection using dot hybridization. Known amounts of CMV DNA purified from prl-1 plasmid were immobilized onto nitrocellulose filters and hybridized to biotinylated or 32P-labeled CMV cdna. While the lower limit of detection appears to be 100 pg of DNA using biotinylated probe, as little as 1 pg of DNA can be detected with 32P-labeled probe. AJP * June 1987 suitable for dot blots. The presence ofcmv infection in these tissues could be detected with high sensitivity and specificity by probing dot blots with a cloned CMV DNA fragment. The entire procedure can be finished in 4-6 days, compared with up to 6 weeks for observing viral cytopathic effects in tissue culture.8 Thus, this can be a possible adjunct to CMV cultures, when a diagnosis is needed clinically before the culture is completed. Additionally, in cases where fresh tissue is not saved for viral cultures, this technique would make a strong contribution to definitive diagnosis. Naturally, it would be imperative to perform positive and negative controls for each clinical specimen, such as probing with Alu repetitive DNA and plasmid DNA, respectively. In situ hybridization is a somewhat similar technique that over the past few years has been applied to the diagnosis of viral infections in tissue blocks.9-" It is difficult to evaluate the relative sensitivities of that method versus the present one without doing both procedures simultaneously on many tissue blocks. However, it appears that with proper controls, both techniques are over 90% sensitive (compared with culture methods) and essentially 100% specific.'2 In situ hybridization has the advantage of being able to pinpoint the exact cell types being infected, and can be speeded-up to be performed in 8 hours, albeit with some loss of sensitivity (62.5%). 13 On the other hand, the method described here appears to require less training for performing and interpreting the tests, and has the potential for being easily and fully automated. Therefore, we believe the two methods are complementary, with the present method possibly useful for more routine cases and large-scale screening, and in situ hybridization suitable for research, urgent cases, and extremely small blocks of tissue. It should be feasible to apply the method described here to the diagnosis of other infectious diseases. At the present limit of sensitivity, only microorganisms with DNA present in several copies per host cell can be detected. This group includes most DNA viruses as well as eukaryotes such as fungi and certain protozoans that have reiterated DNA sequences. With the use of recently developed DNA amplification techniques, 14 the sensitivity can, in theory, be increased by several orders of magnitude, making the detection of bacterial organisms possible.this technique would be most profitably employed in cases ofinfection caused by organisms that are not readily cultured, such as Leishmania. 15 Other applications of this technique can be envisaged. For example, several tumors are known to contain viral DNA sequences, and this method would enable large-scale epidemiologic and retrospective

5 Vol. 127 * No. 3 CMV DNA FROM PARAFFIN BLOCKS 413 studies on these tissues. Preliminary data from our laboratory indicate that human papillomavirus sequences can indeed be so detected in cervical and vaginal carcinoma specimens. With the use of oligonucleotide probes,'4 even single base-pair mutations in genes can be detected, opening up the possibility of diagnosing and tracing genetic diseases from tissue blocks of deceased people. References 1. Goelz SE, Hamilton SR, Vogelstein B: Purification of DNA from formaldehyde fixed and paraffin embedded human tissue. Biochem Biophys Res Commun 1985, 130: Dubeau L, Chandler LA, Gralow JR, Nichols PW, Jones PA: Southern blot analysis of DNA extracted from formalin-fixed pathology specimens. Cancer Res 1986, 46: Maniatis T, Fritsch EF, Sambrook J: Molecular Cloning: A Laboratory Manual. New York, Cold Spring Harbor Laboratory, Deininger PL, Jolly PJ, Rubin LM, Friedmann T, Schmitt CW: Base sequence studies of 300 nucleotide renatured repeated human DNA. J Mol Biol 1981, 151: LaFemina RL, Hayward GS: Structural organization of the DNA molecules from human cytomegalovirus, Animal Virus Genetics. Edited by BN Fields, R Jaenisch. New York, Academic Press, 1980, pp Ho M: Cytomegalovirus: Biology and Infection. New York, Plenum Publishing, Luria SE, Darnell JE, Baltimore D, Campbell A: General Virology. New York, John Wiley & Sons, Spector SA and Spector DH: Use of DNA probes in studies of human cytomegalovirus. Clin Chem 1985, 31: Brigati DJ, Myerson D, Leary JJ, Spalholz B, Travis SZ, Fong CKY, Hsiung GD, Ward DC: Detection of viral genomes in cultured cells and paraffin-embedded tissue sections using biotin-labeled hybridization probes. Virology 1983, 126: Haase A, Brahic M, Stowring L, Blum H: Detection of viral nucleic acids by in situ hybridization. Methods Virol 1984, VII: Myerson D, Hackman RC, Nelson JA, Ward DC, McDougall JK: Widespread presence of histologically occult cytomegalovirus. Hum Pathol 1984, 15: Myerson D, Hackman RC, Meyers JD: Diagnosis of cytomegaloviral pneumonia by in situ hybridization. J Infect Dis 1984, 250: Unger ER, Budgeon LR, Myerson D, Brigati DJ: Viral diagnosis by in situ hybridization: Description of a rapid simplified colorimetric method. Am J Surg Pathol 1986, 10: Saiki RH, Scharf S, Mullis KB, Hem GT, Erlich HA, Arnheim N: Enzymatic amplification of a-globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 1985, 230: Wirth DF, Pratt DM: Rapid identification of Leishmania species by specific hybridization of kinetoplast DNA in cutaneous lesions. Proc Natl Acad Sci USA 1982, 79: Acknowledgments We thank G. Hayward for providing the prl plasmid, J. Ou for providing the Alu plasmid, and M. Warnock for helpful discussions.