- Email: [email protected]
Specific localisation of human cytomegalovirus nucleic acids and proteins in human colorectal cancer
MECHANISMS OF DISEASE
Mechanisms of disease
Specific localisation of human cytomegalovirus nucleic acids and proteins in human colorectal cancer Lualhati Harkins, Andrea L Volk, Minu Samanta, Ivan Mikolaenko, William J Britt, Kirby I Bland, Charles S Cobbs
Summary Background Colorectal cancer is the second most frequent cause of death from cancer in the USA, and most tumours arise sporadically with no clear cause or genetic predisposition. Human cytomegalovirus is a -herpesvirus that is endemic in the human population and can cause lifethreatening disease in immunosuppressed adults. In vitro, human cytomegalovirus can transform cells and dysregulate many cellular pathways relevant to colon adenocarcinoma pathogenesis, especially those affecting the cell cycle, mutagenesis, apoptosis, angiogenesis, and cyclo-oxygenase2 (COX-2) expression. We aimed to assess whether gene products of human cytomegalovirus could be detected in colorectal cancers. Methods We obtained formalin-fixed, paraffin-embedded pathological specimens of colorectal polyps, adenocarcinomas, and adjacent normal mucosa from 29 patients. To detect human cytomegalovirus proteins and nucleic acids, we used immunohistochemistry with two different monoclonal antibodies, in-situ hybridisation, and PCR with DNA sequencing. Findings Human cytomegalovirus proteins IE1-72 and pp65 were detected in a tumour cell-specific pattern in 14 (82%) of 17 and seven (78%) of nine colorectal polyps, respectively, and 12 (80%) of 15 and 11 (92%) of 12 adenocarcinomas, respectively, but not in adjacent non-neoplastic colon biopsy samples from the same patients (none of seven and none of two, respectively). Human cytomegalovirus infection of coloncancer cells (Caco-2) in vitro resulted in specific induction of Bcl-2 and cyclo-oxygenase-2 proteins, both of which are thought to contribute to progression of colon cancer. Interpretation Human cytomegalovirus nucleic acids and proteins can be found that specifically localise to neoplastic cells in human colorectal polyps and adenocarcinomas, and virus infection can induce important oncogenic pathways in colon-cancer cells. Lancet 2002; 360: 1557–63
Pathology Service (L Harkins MS), and Surgical Service (M Samanta MD, I Mikolaenko MD, C S Cobbs MD), Birmingham Veterans Affairs Hospital, Birmingham, AL, USA; and Departments of Pathology (A L Volk MD), Paediatrics (Prof W J Britt MD), and Surgery (Prof K I Bland MD, C S Cobbs), University of Alabama School of Medicine, Birmingham, AL Correspondence to: Dr Charles S Cobbs, Departments of Surgery and Cell Biology, UAB Medical Center, MEB 519, 1813 6th Avenue South, Birmingham, AL 35294, USA (e-mail: [email protected])
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
Introduction Colorectal cancer is the third most common cancer and the second most frequent cause of cancer death in many industrialised countries.1 Most colorectal cancers arise sporadically, and about 5% are thought to be due to inherited predisposition syndromes. Despite present understanding of genetic alterations associated with progression of colon cancer, results of epidemiological studies suggest that environmental factors and host immunological characteristics could contribute to initiation and progression of this cancer. Frequency of colon cancer is ten times higher in developed countries than in some developing countries, and 3·6 times higher in immunosuppressed patients than in those who are not immunosuppressed.2,3 Human cytomegalovirus is a -herpesvirus that persistently infects 50–90% of adults in the USA, and in immunocompromised patients with this infection, focal colonic epithelial lesions can arise.4 Data suggest that gene products of human cytomegalovirus can promote mutagenesis, cell-cycle progression, angiogenesis, cell invasion, and immune evasion.5,6 Other data show that the cell-type involved in human cytomegalovirus infection can greatly affect viral replication, gene expression, and protein trafficking. Herpesvirus genomes are transcribed in a temporally defined fashion, and viral genes are categorised with respect to their temporal expression. Human cytomegalovirus genes are classified into IMMEDIATE EARLY, EARLY, and LATE. In human colonic adenocarcinoma cells (Caco-2 cells), human cytomegalovirus infection can only arise when these cells are in a specific state of differentiation, virus does not spread from cell to cell, and productive infection is rare.7 Although researchers have, in the past, attempted to link human cytomegalovirus with colon cancer, data have been inconsistent, and may have been limited by technologies and reagents available at the time.8–11 Our interest in the chronic inflammatory properties of malignant gliomas led us to assess whether human cytomegalovirus might be associated with these tumours, because of its unique biological properties. Unexpectedly, we discovered that the virus is strongly associated with malignant gliomas, and many gene products are expressed in these tumours.12 Because of results of two early reports suggesting that human cytomegalovirus DNA could be detected in colorectal cancers,8,11 we decided to assess the possibility that gene products of this virus are also expressed in these tumours.
Methods Procedures Clinical samples—We obtained formalin-fixed, paraffinembedded, surgical biopsy specimens of normal and neoplastic colon from the pathology archives of the Birmingham Veterans Affairs Hospital, and the University of Alabama at Birmingham, AL, USA. All samples were obtained in accordance with ethics guidelines from the institutional review board of each institution. A pathologist 1557
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
GLOSSARY IMMEDIATE EARLY GENES
The first set of viral genes expressed immediately after infection. These gene products—eg, IE1-72, which codes for a highly expressed regulatory phospho-protein—do not require viral DNA synthesis or proteins to be transcribed. EARLY GENES
These are the second set of viral genes expressed, and these gene products encode for viral DNA polymerases and other viral functions required for viral DNA synthesis, and some viral structural proteins--eg, the pp65 delayed early tegument protein. LATE GENES
Encode mostly for structural proteins used in viral assembly and packaging, and are generally expressed sequentially after early genes.
(ALV) re-examined all cases to confirm diagnosis, and we reviewed patients’ charts to ensure that no one was immunocompromised. Immunohistochemical analyses of paraffin sections—We cut 4-m paraffin sections from biopsy specimens of neoplastic and non-neoplastic colon, removed the paraffin from them, then hydrated the sections. We digested each section with pepsin (37°C, 4 min; BioGenex, San Ramon, CA, USA), retrieved antigen in citrate buffer (pH 6·0, 37°C, 12 h; BioGenex), and blocked non-specific peroxidase activity (3% H2O2, 12 min) and antibody binding (Fc-receptor blocker, 10 min, 20°C; Innovex Sciences, Richmond, CA, USA). We incubated each section with monoclonal antibody (anti-IE1-72 [1:20; BioGenex]; anti-pp65 [1:40; Novocastra, Newcastle-upon-Tyne, UK]; anti-CD34 [1:20; BioGenex]; anti-COX-2 [1:30; Cayman Chemical, Ann Arbor, MI, USA]) or no antibody in tris-buffered saline (pH 7·6) with 0·05% Tween 20 (TBST; 4°C, 12 h). IE1-72 is an immediate early gene product of human cytomegalovirus and pp65 is a delayed early gene product. We used anti-pp65 to assess delayed human cytomegalovirus gene expression. Since cyclo-oxygenase-2 (COX-2) is thought to have an important role in development and malignant progression of human colorectal cancer,13 and because COX-2 mRNA expression is strongly induced in primary human cells after infection with human cytomegalovirus and is needed for viral replication, we used anti-COX-2 to establish if expression of this enzyme was localised to a similar cell population as human cytomegalovirus protein expression.14,15 Negative controls for IE1-72 and pp65 immunohistochemistry were done, which included elimination of antibody from the reaction, substitution of antibody with an equivalent concentration of anti-CD34 isotypeidentical monoclonal antibody, which is specific for endothelial cells of blood vessels, or both. We rinsed slides with TBST and did colorimetric determination with a three-step horseradish peroxidase detection system (BioGenex) with the chromogen diaminobenzidine (Innovex Sciences). A pathologist (ALV)—masked to diagnosis and to whether antibody had been added—scored anti-IE1-72 immunostained specimens and controls as either positive or negative. A positive score showed that specific cell immunoreactivity was present and a negative score that none was present. Immunoreactivity for anti-pp65, antiCD34, and anti-COX-2 was scored similarly, but the pathologist was not masked. In-situ hybridisation of paraffin sections—To confirm that anti-IE1-72 and anti-pp65 were specific for human
1558
cytomegalovirus, and to show the presence of viral nucleic acids, we did in-situ hybridisation on paraffin sections of several of the tumour specimens that were immunoreactive for IE1-72. To detect viral nucleic acids, we obtained a biotinylated 21-base oligonucleotide (5⬘-GTGGTGGCGCTGGGGGTGGCG-3⬘) probe specific for the most abundantly transcribed early gene of human cytomegalovirus (ResGen, Huntsville, AL, USA). As a positive control, we used a biotinylated probe specific for poly-A RNA sequences. As a negative control, we either eliminated the specific human cytomegalovirus probe from the reaction, or we used a biotinylated 21-base oligonucleotide probe specific for herpes simplex virus 1 and 2 mRNA (ResGen), which had a similar GC content to the viral probe. We did enzyme digestion with pepsin (37°C, 15 min; Zymed Labs, San Francisco, CA, USA) and nucleic acid denaturation (90°C, 15 min) of 4-m paraffin sections with a Misha thermocycler (Shandon Lipshaw, Pittsburgh, PA, USA), and then slides were hybridised with probe overnight at 37°C in a humidified chamber. We detected probe with the DNA/mRNA alkaline phosphatase supersensitive detection system with the chromogen nitro-blue tetrazolium (BioGenex, San Ramon, CA). Cells were counterstained with methyl green, or not counterstained, and then mounted with aqueous mount for visualisation. Infection of Caco-2 cells in vitro—We used the Caco-2 human colon epithelial adenocarcinoma cell line, which has been characterised with respect to human cytomegalovirus infection and the effects of COX-2 expression.7,16,17 In culture, Caco-2 cells grow in colonies in which most cells differentiate after a few days. Human cytomegalovirus infection of these cells is dependent on the presence of rare Caco-2 cells that are in a specific state of differentiation, and therefore only a small proportion of cells are typically infected with virus, even at a high multiplicity of infection.7 We obtained human colon epithelial-derived Caco-2 cells from the American type culture collection (ATCC, Rockville, MD, USA) and cultured them at 37°C in 5% CO2 in Dulbecco’s Modified Eagle’s Medium (DMEM) with 10% fetal bovine serum (FBS; BioWhittaker, Walkersville, MD, USA). Caco-2 monolayers were grown for 5 days in culture, at which time they had assumed characteristic colony formation. We washed cells with phosphatebuffered saline (PBS) and infected (or mock-infected) them with human cytomegalovirus Towne strain (obtained from ATCC) by addition of virus to the media at a multiplicity of infection of one in DMEM with no serum. The mock-infected cells were treated identically except that no virus was added. After addition of virus, we allowed infection to proceed for 2 h at 37°C. We then washed cells with PBS, and DMEM with 10% FBS was again added. They were incubated in 5% CO2 at 37°C. After 48 h, cells were fixed with methanol (–20°C, 20 min), air-dried, and processed for double-staining immunohistochemistry or in-situ hybridisation. Immunohistochemistry of infected cells—We rinsed cells with TBST (pH 7·6), blocked endogenous peroxidase activity (3% H2O2, 8 min), and then incubated cells in Fc-receptor blocker (20°C, 10 min; BioGenex). Slides were blotted dry and primary monoclonal antibody against IE1-72 (1:300; BioGenex) was added (4°C, 12 h). We rinsed cells with TBST and detected antigen with a three-step horseradish peroxidase system (BioGenex) with diaminobenzidine black chromogen (Zymed Labs) or diaminobenzidine brown (Innovex
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Sciences). Coverslips were added with an aqueous mounting media. We photographed slides and then soaked them in water to remove coverslips and mounting media. Like COX-2, the anti-apoptotic protein Bcl-2 can be induced in tumour cells infected with human cytomegalovirus and can confer upon these cells a growth advantage and resistance to cytotoxic chemotherapeutic agents.18 Expression of Bcl-2 in colorectal adenocarcinoma is thought to have an important role in early tumour progression and may be activated by COX-2 expression.19,20 For COX-2 double-staining, we added anti-COX-2 (1:200, 4°C, 12 h; BioGenex) to cells immunostained with IE1-72 and diaminobenzidine black and detected antigen with the three-step alkaline phosphatase system with fast red chromogen (BioGenex). For Bcl-2 double-staining, we added anti-Bcl-2 (1:200, 4°C, 12 h; Novocastra) to cells immunostained with IE1-72 and diaminobenzidine brown and detected antigen as described for COX-2. Slides were photographed after mounting with aqueous media and a coverslip.
added as described above. We added coverslips, denatured cells (90°C, 15 min), and then cells were hybridised with probe (37°C, 90 min) in a humidified chamber. Cells were soaked in TBST and then probe wash (BioGenex) was added for 10 min and hybridisation was detected as described above with the chromogen nitro-blue tetrazolium. Cells were mounted, coverslips were added, and the cells were photographed. PCR and DNA sequencing—To confirm that our probe was specific for human cytomegalovirus nucleic acids, we extracted DNA from three to six paraffin sections of 10 m size, cut from a subset of the same biopsy specimens described above with QIAamp DNA mini kit (Qiagen, Valencia, CA, USA). To avoid potential PCR contamination, these experiments were done in a laboratory with no previous exposure to human cytomegalovirus, all preparations were processed masked, no positive controls were used in any PCR reactions, and blank paraffin blocks were cut sequentially between each patient’s sample and processed identically. Between each section, we changed the sectioning blade and wiped down the cutting surface with xylene and ethanol. For each specimen, at least two tubes of DNA were eluted and run as separate samples for PCR. From each sample, 250 ng
In-situ hybridisation of infected cells—We fixed cells in methanol (–20°C, 20 min) and then air-dried them. Biotinylated cytomegalovirus probe or control probe was Age (years), Diagnosis sex Patient number 1 77, m 2a 52, m 2b 2c 3 66, m 4 51, m 5 71, m 6 72, m 7 75, m 8a 80, m 8b 9 52, m 10 59, m 11 79, m 12 82, m 13 52, m 14 66, m 15 54, m 16a 61, m 16b 16c 17 71, m 18 66, m 19 52, f 20 74, m 21 52, m 22a 68, m 22b 23a 68, m 23b 24a 66, m 24b 24c 25a 65, m 25b 26a 71, m 26b 27a 67, m 27b 28 71, m 29a 54, m 29b
Non-neoplastic colonic epithelium Non-neoplastic colonic epithelium Polyp Adenocarcinoma Polyp Polyp Adenocarcinoma Polyp Polyp Non-neoplastic colonic epithelium Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Non-neoplastic colonic epithelium Non-neoplastic colonic epithelium Polyp Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Polyp Adenocarcinoma Adenocarcinoma Adenocarcinoma Non-neoplastic colonic epithelium
Immunohistochemistry
In-situ hybridisation
UL55 PCR
Anti-IE1-72 Anti-pp65 Anti-COX-2 Anti-CD34* HCMV Negative control† HSV
Poly-T‡
– – + + + + + + + – + + + – + + + – + + + + – + + + + – + – + .. – – + – – + + – .. ..
.. .. .. .. + + .. .. .. .. .. .. .. .. .. + + .. .. .. .. .. + .. .. .. .. .. + .. + .. .. .. + .. .. .. + .. .. ..
– .. + + .. + + + .. .. + + .. .. + .. + .. .. .. + + .. .. + + + .. + .. + .. .. – + – .. – + – .. ..
.. – + .. + + .. .. .. .. .. + .. .. .. + + .. .. .. .. + + .. + .. .. .. + .. + .. .. .. + .. .. .. .. .. .. ..
– – – .. – – – .. – – .. .. .. – – – – – .. – .. – – – – .. – .. – – – – – – – – .. – – .. .. ..
.. .. .. .. + + .. .. .. .. .. + .. .. .. + + .. + .. + + + + + + .. .. + .. + .. .. .. + .. .. .. + .. .. ..
.. .. .. .. – – .. .. .. .. .. .. .. .. .. – – .. .. .. .. .. – .. .. .. .. .. – .. – .. .. .. .. .. .. .. .. .. .. ..
.. .. .. .. .. – .. .. .. .. .. – .. .. .. – .. .. .. .. .. .. .. .. .. .. .. .. – .. .. .. .. .. .. .. .. .. .. .. .. ..
.. – .. + .. .. .. + .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. + .. .. .. .. .. + – .. .. .. .. .. .. + + + –
HCMV=human cytomegalovirus; IE1-72=HCMV immediate early gene probe; pp65=delayed early gene probe; HSV=herpes simplex virus 1 and 2; UL55=HCMV glycoprotein B gene. ··=test not done. *Immunoreactivity present only in blood vessels. †Elimination of primary antibody or in-situ hybridisation probe from reaction. ‡Positive control probe for poly A RNA.
Immunohistochemical, in-situ hybridisation, and PCR results for paraffin specimens of human colonic polyps, adenocarcinomas, and nonneoplastic tissues
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
1559
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Blast search, and sequences were identical to that of UL55 (Genbank accession number X17403). Statistical analysis We calculated frequencies and proportions to summarise expression of IE1-72 and pp65 immunoreactivity for each sample type (polyps, adenocarcinomas, non-neoplastic epithelium). Pairwise comparisons of proportion of immunoreactivity between non-neoplastic and polyp and between non-neoplastic and adenocarcinoma were done with Fisher’s exact test. For these comparisons, independent patients were included in each group by exclusion of polyp or adenocarcinoma samples from patients with matching non-neoplastic colon. Role of the funding source The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Results
Figure 1: Immunohistochemistry and in-situ hybridisaton of human cytomegalovirus proteins and nucleic acids in human colonic polyps and adenocarcinomas (A) IE1-72 immunoreactivity (brown staining) in basal crypts in a hyperplastic polyp (arrow), and (B) in dysplastic crypts in a tubular adenoma (large arrow), but not in adjacent normal-appearing colon epithelium (small arrow). Low-power (C) and high-power (D) images of IE1-72 immunoreactivity in malignant-appearing cells but not adjacent normal-appearing epithelium in a colon adenocarcinoma (D=boxed area in C). Colon adenocarcinoma shows similar pattern of immunoreactivity for IE1-72 (E and F) and pp65 (G and H), except that inflammatory cells are also immunoreactive for pp65 in areas adjacent to tumour (arrow in H; F=boxed area in E; H=boxed area in G). (I) Blood-vessel immunoreactivity noted after CD-34 immunostaining (negative control); (J) arrows show endothelial-cell staining (J=boxed area in I). (K and L) Diffuse COX-2 immunoreactivity in adenocarcinoma. Low-power (M) and high-power (N–P) images of a different colon adenocarcinoma show IE1-72 immunoreactivity in tumour (N and P) but not adjacent normal crypts (O). pp65 immunoreactivity present in tumour (Q) but not in adjacent normal epithelium (R). (S) COX-2 immunoreactivity colocalised with IE1-72 and pp65 in the same adenocarcinoma shown in M–R. Only inflammatory cells are immunoreactive for COX-2 in normal appearing epithelium from same tumour (arrow in T). Low-power (U) and highpower (V) images of hybridisation of probe specific for human cytomegalovirus early nucleic acids with colon adenocarcinoma epithelium. (W) Probe omitted. (X) Positive control probe hybridisation (specific for poly-A mRNA). Original magnification ⫻40 (E, G, I, K, M, U, W, X), ⫻100 (A, B, C, N, O, Q, R, S, T), ⫻400 (D, F, H, J, L, P, V).
of DNA was amplified by nested PCR with internal and external primers specific for human cytomegalovirus glycoprotein B (UL55) gene.21 We judged samples positive if a band of 140 bp could be seen on agarose gel with ethidium bromide. Amplified DNA products from tumours were visualised on 1·5% agarose gels, bands were cut out, and DNA was extracted (Qiagen) and analysed by automated sequencing (ABI Model 377 DNA Sequencer, Foster City, CA, USA). Confirmation of human cytomegalovirus sequence was done by NCBI
1560
We obtained 45 paraffin-embedded, formalin-fixed biopsy samples from 29 different patients (table). We detected IE1-72 immunoreactivity in 14 (82%) of 17 polyps and 12 (80%) of 15 adenocarcinomas. IE1-72 protein was not typically detected in normal appearing colonic epithelium in areas adjacent to tumour within the same pathological section, nor was IE1-72 detected in tumour-free surgical biopsy specimens of colon from seven of these same patients. Results of pairwise comparisons showed a significant difference in IE1-72 immunoreactivity between non-neoplastic tissue and polyps (p=0·001) and between non-neoplastic tissue and adenocarcinomas (p=0·002). In hyperplastic polyps, IE1-72 immunoreactivity, when present, was detected only in discrete cell populations at the base of colonic crypts (figure 1, A). In tubular adenomas, IE1-72 immunoreactivity was detected in areas of dysplastic epithelium, and generally was not seen in areas of normal-appearing colonic crypts (figure 1, B). Colonic crypts immunoreactive for IE1-72 typically showed morphological characteristics of dysplastic aberrant crypt foci, with increased nuclear/cytoplasmic ratios, absence of uniform architecture, and nuclear migration away from the basement membrane (figure 1, B–D). In moderately and well-differentiated colon adenocarcinomas, IE1-72 immunoreactivity was seen throughout the neoplastic epithelium and was notably absent in adjacent areas of normal-appearing colonic crypts (figure 1, C–F, M–P). Little, if any, IE1-72 immunoreactivity was detected in areas of poorly differentiated adenocarcinoma, even when welldifferentiated areas of adenocarcinoma within the same specimen showed strong IE1-72 immunoreactivity (data available from authors). No tumour-cell immunoreactivity was detected in negative controls for IE1-72 immunohistochemistry (table; figure 1, I and J). We detected pp65 immunoreactivity in neoplastic epithelium and stromal inflammatory cells in neoplastic tumours (table, figure 1, G, H, Q), which accord with the pattern of expression we saw with respect to IE1-72 immunoreactivity. We did not detect pp65 immunoreactivity in normal-appearing colonic epithelium adjacent to tumour (figure 1, G, H, Q, R). In adjacent sections of the same tumour specimen, the pattern of pp65 immunoreactivity closely matched the pattern of IE1-72 immunoreactivity (figure 1, E–H). Results of negative controls for anti-pp65 were closely similar to those of anti-IE1-72. Similar to the pattern of IE1-72 immunoreactivity, pp65 immunoreactivity was noted in
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
M w
b
2b 2c b b
2a 2a
b
b
27b 27b 27b
140 bp M
6
20
w
20 24b 29b 29a
140 bp Figure 2: Nested PCR to detect UL55 gene in specimens of human colon cancers Band at 140 bp is amplified UL55 gene. Numbers above lanes refer to each patient’s biopsy specimen number, shown in the table. Most specimens had two to three DNA elutions made for each, and in some samples amplified products were identified in only one or two samples from a particular tumour. M=DNA ladder lane (underloaded); b=blank paraffin block specimens; w=negative PCR water controls.
seven (78%) of nine polyps and 11 (92%) of 12 adenocarcinomas, whereas non-neoplastic colon from two patients was not immunoreactive. Results of pairwise comparisons showed a significant difference in IE1-72 immunoreactivity between non-neoplastic and adenocarcinoma (p=0·033) but not between non-neoplastic and polyps (p=0·109). The biotinylated oligonucleotide probe specific for human cytomegalovirus early gene mRNA showed hybridisation with viral nucleic acids in neoplastic colon epithelium in several specimens (table, figure 1, U–X). No hybridisation was detected when probe was omitted from the reaction, or when probe specific for HSV-1 and HSV-2 was substituted (negative controls). Poly-A RNA probe hybridised to all cells (positive controls). We could detect amplified UL55 PCR products in six tumours (five adenocarcinomas and one polyp) that were immunoreactive for human cytomegalovirus, but in no non-neoplastic tumour-free margin biopsy specimens from three of these same six patients (figure 2). We detected no human cytomegalovirus amplified products in five blank paraffin-block control samples or four PCR water negative controls that were run in the same nested PCR reaction. Amplified PCR products from the six tumours were cut from agarose gels, and DNA sequencing of these products showed that all six were the human cytomegalovirus UL55 gene. We detected COX-2 expression in regions of tumour epithelium that in general corresponded with areas of IE1-72 and pp65 immunoreactivity, whereas little COX-2 immunoreactivity was detected in normal-appearing colonic epithelium (apart from stromal inflammatory cells; figure 1, K, L, S, T). Since results of these experiments suggested that COX-2 and human cytomegalovirus antigens colocalised in these tumours, we sought to establish if infection of human colon-cancer cells in vitro would lead to specific COX-2 expression in these cells. IE1-72 nuclear immunoreactivity was clearly seen in rare Caco-2 cells 48 h after infection. Double-labelling with anti-COX-2 showed intense COX-2 expression only in cells immunoreactive for IE1-72 (figure 3, C, D). These results were the same when we did pp65 immunostaining, suggesting that both anti-IE1-72 and anti-pp65 were specific for human cytomegalovirus proteins and were not cross-reacting with non-specific colon epithelial cell epitopes (data available from authors). No IE1-72 or pp65 immunoreactivity was noted, and only faint COX-2 immunoreactivity was seen in control (mock-infected) Caco-2 cells (data available from authors). We recorded that Caco-2 cells expressing IE1-72 had increased expression of Bcl-2, which was not seen in uninfected adjacent cells or mock-infected cells (figure 3, E, F).
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
Figure 3: Double-staining immunohistochemistry and in-situ hybridisation for human cytomegalovirus in Caco-2 colon adenocarcinoma cells (A) Rare Caco-2-cell nuclear immunoreactivity (black staining shown by arrows) for IE1-72 (B=boxed cell in A). (C) After double-labelling with anti-COX2, infected cells show increased COX-2 expression (red staining; arrows show same cells in A; D=boxed cell in C). Low-power (E) and high-power (F) views of a colony of Caco-2 cells double-stained with anti-IE1-72 (brown chromogen in nuclei) and anti-Bcl-2 (red chromogen in cytoplasm). In-situ hybridisation of infected cells at low-power (G) and high-power (H) with probe specific for human cytomegalovirus early nucleic acids shows discrete hybridisation pattern of infected cells similar to that seen with IE1-72 immunostaining. Original magnification ⫻40 (A, C, E, G) and ⫻400 (B, D, F, H).
Human cytomegalovirus nucleic acids were seen to hybridise in discrete Caco-2 cells (figure 3, G, H). No evidence of virus nucleic acid hybridisation was noted in mock-infected control cells, nor in virus-infected Caco-2 cells when the human cytomegalovirus probe was omitted from the hybridisation reaction (data available from authors).
Discussion We have shown that human cytomegalovirus proteins and nucleic acids can be found that localise specifically in human colorectal polyps and adenocarcinomas. Present paradigms for development of human colorectal adenocarcinoma show that a multistep progression of genetic mutations takes place over years in colonic epithelium, characterised by phenotypic progression from polyp to invasive adenocarcinoma.1,22 Of these alterations, those that dysregulate the adenomatous polyposis coli and TP53 tumour suppressor pathways and induce COX-2 expression are key. Although our findings are unexpected, they could tie in with these existing paradigms.
1561
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Human herpes viruses are presently implicated in the pathogenesis of at least four human malignancies. Epstein-Barr Virus (EBV) is associated with Burkitt’s lymphoma, nasopharyngeal carcinoma, and B-cell lymphoproliferative syndromes, and human herpes virus 8 (HHV-8) is associated with Kaposi’s sarcoma.23 EBVrelated malignancies arise after years of viral dormancy and are accompanied or triggered by viral reactivation. Present ideas for the pathogenesis of HHV-8 in Kaposi’s sarcoma suggest that tumour initiation and progression is associated with cytokine-mediated reactivation of HHV-8, and may arise only in individuals with specific immunogenetic phenotypes.24 In early-developing Kaposi’s sarcoma lesions, HHV-8 viral load may be low or undetectable, and latently expressed viral genes are thought to induce cell transformation.24 Our data accord with preferential replication of human cytomegalovirus in dysplastic colon epithelial cells. These cells are known to harbour replicative virus in immunocompromised patients, in whom focal colonic ulcerative lesions often arise.4 Jarvis and colleagues7 reported that human cytomegalovirus infection of Caco-2 colon epithelial cells could only happen when these cells are in a specific state of differentiation, which accords with our in-vitro data. Should this occurrence be the case in vivo, dysplastic colonic epithelial cells with impaired cell-cycle control mechanisms may provide a unique reservoir for human cytomegalovirus persistence with no effect on the overall tumour phenotype. Alternatively, long-term, persistent human cytomegalovirus infection and expression in dysplastic colonic epithelial cells could be important in promotion of oncogenic events directly implicated in malignant progression. Human cytomegalovirus infection activates cellular proto-oncogenes, cyclins, and kinases involved in cell mitogenic and cell survival pathways, including c-myc, c-fos, c-jun, cyclin-B, cyclin-E, mitogen-activated protein kinase, ERK 1/2, and PI3-kinase.25–29 This virus also induces the NF-B transcription factor, which might have an important role in tumour-cell survival pathways.30 We, and others,18 have shown that human cytomegalovirus infection of tumour cells can induce Bcl-2 expression, which confers resistance to apoptotic signalling during tumour progression. The human cytomegalovirus UL111A gene product has been reported to transform cells in vitro and dysregulate cell-cycle control pathways by binding TP53 and disrupting its cellular transcription.31 The virus gene products IE1-72 and IE2-86 can also dysregulate cell-cycle checkpoint controls by interaction with the TP53 and retinoblastoma tumour-suppressor proteins and downstream pathways.32 Even in the presence of functional TP53, IE2-86 can block downstream TP53 cell-cycle arrest pathways by degradation of CDKN1A (p21cip1), a principal mediator of the TP53induced G1 cell-cycle arrest.33 When combined with the adenovirus E1A protein, which impairs retinoblastoma tumour-suppressor protein function, IE1-72 and IE2-86 can induce oncogenic transformation of cells.34 IE1-72 and IE2-86 can also block apoptosis mediated by tumour necrosis factor ␣ and are highly mutagenic; thus expression of these gene products could strikingly increase genomic instability in a population of chronically infected cells.34,35 In addition to evasion of apoptotic and cell-cycle control pathways, tumour cells infected with human cytomegalovirus might also escape normal host antitumour cytotoxic T cell and natural killer cell immune
1562
responses, since the virus encodes for many gene products that block MHC class 1 and class 2 antigen synthesis and function of natural-killer cells.36 Our data, and those of others, show that human cytomegalovirus induces expression of COX-2.14 Since COX-2 can promote angiogenesis and invasion of colon adenocarcinoma cells, virus-infected colon-cancer cells might acquire an additional growth advantage through these mechanisms.16,17 In view of the many cellular modulatory properties of this virus, our data justify further studies to establish whether human cytomegalovirus participates in pathogenesis of colorectal cancer. Contributors L Harkins did immunohistochemistry, in-situ hybridisation, and PCR experiments. A L Volk interpreted immunostaining data. M Samanta did cytomegalovirus infections in vitro and PCR sequencing. I Mikolaenko did DNA extractions and PCR optimisation. W J Britt and K I Bland contributed to study design and writing of the report. C S Cobbs contributed to study design, did immunohistochemistry, in-situ hybridisation, PCR experiments, interpreted data, and wrote the report.
Conflict of interest statement None declared.
Acknowledgments We thank Heidi Weiss for assistance with statistical analysis. CSC is supported by a Merit Review Grant for cancer research from the Veterans Administration.
References 1
2
3
4
5
6
7
8
9
10 11 12
13 14
15
16
Fearon ER. Molecular biology of gastrointestinal cancers. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: principles and practice of oncology, 6th edn. Philadelphia: Lippincott, Williams, and Wilkins, 2001: 1037–49. Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide incidence of eighteen major cancers in 1985. Int J Cancer 1993; 54: 594–606. Birkeland SA, Storm HH, Lamm LU, et al. Cancer risk after renal transplantation in the Nordic countries, 1964–1986. Int J Cancer 1995; 60: 183–89. Britt WJ, Alford CA. Cytomegalovirus. In: Fields BN, Knipe DM, Howley PM, eds. Fields virology, vol 2, 3rd edn. Philadelphia: Lippincott, Williams, and Wilkins, 1996: 2493–523. Doniger J, Sumitra M, Rosenthal LJ. Human cytomegalovirus and human herpesvirus 6 genes that transform and transactivate. Clin Microbiol Rev 1999; 12: 367–82. Cinatl J Jr, Cinatl J, Vogel JU, et al. Modulatory effects of human cytomegalovirus infection on malignant properties of cancer cells. Intervirology 1996; 39: 259–69. Jarvis MA, Wang CE, Meyers HL, et al. Human cytomegalovirus infection of caco-2 cells occurs at the basolateral membrane and is differentiation state dependent. J Virol 1999; 73: 4552–60. Huang ES, Roche JK. Cytomegalovirus DNA and adenocarcinoma of the colon: evidence for latent viral infection. Lancet 1978; 1: 957–60. Hashiro GM, Horikami S, Loh PC. Cytomegalovirus isolations from cell cultures of human adenocarcinomas of the colon. Intervirology 1979; 12: 84–88. Hart H, Neill WA, Norval M. Lack of association of cytomegalovirus with adenocarcinoma of the colon. Gut 1982; 23: 21–30. Ruger R, Fleckenstein B. Cytomegalovirus DNA in colorectal carcinoma tissues. Klin Wochenschr 1985; 63: 405–08. Cobbs CS, Harkins L, Samanta M, et al. Human cytomegalovirus infection and expression in human malignant glioma. Cancer Res 2002; 62: 3347–50. Williams C, Shattuck-Brandt RL, DuBois RN. The role of COX-2 in intestinal cancer. Ann N Y Acad Sci 1999; 889: 72–83. Zhu H, Cong JP, Mamtora G, et al. Cellular gene expression altered by human cytomegalovirus: global monitoring with oligonucleotide arrays. Proc Natl Acad Sci USA 1998; 95: 14470–75. Zhu H, Cong JP, Yu D, Bresnahan WA, Shenk TE. Inhibition of cyclooxygenase 2 blocks human cytomegalovirus replication. Proc Natl Acad Sci USA 2002; 99: 3932–37. Tsujii M, Kawano S, Tsuji S, et al. Cyclooxygenase regulates angiogenesis induced by colon cancer cells. Cell 1998; 93: 705–16.
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
17 Tsujii M, Kawano S, DuBois RN. Cyclooxygenase-2 expression in human colon cancer cells increases metastatic potential. Proc Natl Acad Sci USA 1997; 94: 3336–40. 18 Cinatl J Jr, Cinatl J, Vogel JU, et al. Persistent human cytomegalovirus infection induces drug resistance and alteration of programmed cell death in human neuroblastoma cells. Cancer Res 1998; 58: 367–72. 19 Sheng H, Shao J, Morrow JD, et al. Modulation of apoptosis and Bcl-2 expression by prostaglandin E2 in human colon cancer cells. Cancer Res 1998; 58: 362–66. 20 Merritt AJ, Potten CS, Watson AJ, et al. Differential expression of bcl-2 in intestinal epithelia: correlation with attenuation of apoptosis in colonic crypts and the incidence of colonic neoplasia. J Cell Sci 1995; 108: 2261–71. 21 Kuhn JE, Wendland T, Eggers HJ, et al. Quantitation of human cytomegalovirus genomes in the brain of AIDS patients. J Med Virol 1995; 47: 70–82. 22 Kinzler KW, Vogelstein B. Lessons from hereditary colorectal cancer. Cell 1996; 87: 159–70. 23 Howley PM, Ganem D, Kieff E. DNA viruses. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: principles and practice of oncology, 6th edn. Philadelphia: Lippincott, Williams, and Wilkins, 2001: 168–73. 24 Sturzl M, Zietz C, Monini P, Ensoli B. Human herpesvirus-8 and Kaposi’s sarcoma: relationship with the multistep concept of tumorigenesis. Adv Cancer Res 2001; 81: 125–59. 25 Hagemeier C, Walker SM, Sissons PJ, Sinclair JH. The 72K IE1 and 80K IE2 proteins of human cytomegalovirus independently transactivate the c-fos, c-myc and hsp70 promoters via basal promoter elements. J Gen Virol 1992; 73: 2385–93. 26 Johnson RA, Wang X, Ma XL, et al. Human cytomegalovirus up-regulates the phosphatidylinositol 3-kinase (PI3-K) pathway: inhibition of PI3-K activity inhibits viral replication and virus-induced signaling. J Virol 2001; 75: 6022–32.
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
27 Johnson RA, Huong SM, Huang ES. Activation of the mitogenactivated protein kinase p38 by human cytomegalovirus infection through two distinct pathways: a novel mechanism for activation of p38. J Virol 2000; 74: 1158–67. 28 Jault FM, Jault JM, Ruchti F, et al. Cytomegalovirus infection induces high levels of cyclins, phosphorylated Rb, and p53, leading to cell cycle arrest. J Virol 1995; 69: 6697–704. 29 Bresnahan WA, Albrecht T, Thompson EA. The cyclin E promoter is activated by human cytomegalovirus 86-kDa immediate early proteins. J Biol Chem 1998; 273: 22075–82. 30 Yurochko AD, Kowalik TF, Huong SM, Huang ES. Human cytomegalovirus upregulates NF-kappa B activity by transactivating the NF-kappa B p105/p50 and p65 promoters. J Virol 1995; 69: 5391–400. 31 Muralidhar S, Doniger J, Mendelson E, et al. Human cytomegalovirus mtrII oncoprotein binds to p53 and down-regulates p53-activated transcription. J Virol 1996; 70: 8691–700. 32 Castillo JP, Yurochko AD, Kowalik TF. Role of human cytomegalovirus immediate-early proteins in cell growth control. J Virol 2000; 74: 8028–37. 33 Chen Z, Knutson E, Kurosky A, Albrecht T. Degradation of p21cip1 in cells productively infected with human cytomegalovirus. J Virol 2001; 75: 3613–25. 34 Shen Y, Zhu H, Shenk T. Human cytomegalovirus IE1 and IE2 proteins are mutagenic and mediate “hit and run” oncogenic transformation in cooperation with the adenovirus E1A proteins. Proc Natl Acad Sci USA 1997; 94: 3341–45. 35 Zhu H, Shen Y, Shenk T. Human cytomegalovirus IE1 and IE2 proteins block apoptosis. J Virol 1995; 69: 7960–70. 36 Loenen WA, Bruggeman CA, Wiertz, EJ. Immune evasion by human cytomegalovirus: lessons in immunology and cell biology. Semin Immunol 2001; 13: 41–49.
1563
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Mechanisms of disease
Specific localisation of human cytomegalovirus nucleic acids and proteins in human colorectal cancer Lualhati Harkins, Andrea L Volk, Minu Samanta, Ivan Mikolaenko, William J Britt, Kirby I Bland, Charles S Cobbs
Summary Background Colorectal cancer is the second most frequent cause of death from cancer in the USA, and most tumours arise sporadically with no clear cause or genetic predisposition. Human cytomegalovirus is a -herpesvirus that is endemic in the human population and can cause lifethreatening disease in immunosuppressed adults. In vitro, human cytomegalovirus can transform cells and dysregulate many cellular pathways relevant to colon adenocarcinoma pathogenesis, especially those affecting the cell cycle, mutagenesis, apoptosis, angiogenesis, and cyclo-oxygenase2 (COX-2) expression. We aimed to assess whether gene products of human cytomegalovirus could be detected in colorectal cancers. Methods We obtained formalin-fixed, paraffin-embedded pathological specimens of colorectal polyps, adenocarcinomas, and adjacent normal mucosa from 29 patients. To detect human cytomegalovirus proteins and nucleic acids, we used immunohistochemistry with two different monoclonal antibodies, in-situ hybridisation, and PCR with DNA sequencing. Findings Human cytomegalovirus proteins IE1-72 and pp65 were detected in a tumour cell-specific pattern in 14 (82%) of 17 and seven (78%) of nine colorectal polyps, respectively, and 12 (80%) of 15 and 11 (92%) of 12 adenocarcinomas, respectively, but not in adjacent non-neoplastic colon biopsy samples from the same patients (none of seven and none of two, respectively). Human cytomegalovirus infection of coloncancer cells (Caco-2) in vitro resulted in specific induction of Bcl-2 and cyclo-oxygenase-2 proteins, both of which are thought to contribute to progression of colon cancer. Interpretation Human cytomegalovirus nucleic acids and proteins can be found that specifically localise to neoplastic cells in human colorectal polyps and adenocarcinomas, and virus infection can induce important oncogenic pathways in colon-cancer cells. Lancet 2002; 360: 1557–63
Pathology Service (L Harkins MS), and Surgical Service (M Samanta MD, I Mikolaenko MD, C S Cobbs MD), Birmingham Veterans Affairs Hospital, Birmingham, AL, USA; and Departments of Pathology (A L Volk MD), Paediatrics (Prof W J Britt MD), and Surgery (Prof K I Bland MD, C S Cobbs), University of Alabama School of Medicine, Birmingham, AL Correspondence to: Dr Charles S Cobbs, Departments of Surgery and Cell Biology, UAB Medical Center, MEB 519, 1813 6th Avenue South, Birmingham, AL 35294, USA (e-mail: [email protected])
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
Introduction Colorectal cancer is the third most common cancer and the second most frequent cause of cancer death in many industrialised countries.1 Most colorectal cancers arise sporadically, and about 5% are thought to be due to inherited predisposition syndromes. Despite present understanding of genetic alterations associated with progression of colon cancer, results of epidemiological studies suggest that environmental factors and host immunological characteristics could contribute to initiation and progression of this cancer. Frequency of colon cancer is ten times higher in developed countries than in some developing countries, and 3·6 times higher in immunosuppressed patients than in those who are not immunosuppressed.2,3 Human cytomegalovirus is a -herpesvirus that persistently infects 50–90% of adults in the USA, and in immunocompromised patients with this infection, focal colonic epithelial lesions can arise.4 Data suggest that gene products of human cytomegalovirus can promote mutagenesis, cell-cycle progression, angiogenesis, cell invasion, and immune evasion.5,6 Other data show that the cell-type involved in human cytomegalovirus infection can greatly affect viral replication, gene expression, and protein trafficking. Herpesvirus genomes are transcribed in a temporally defined fashion, and viral genes are categorised with respect to their temporal expression. Human cytomegalovirus genes are classified into IMMEDIATE EARLY, EARLY, and LATE. In human colonic adenocarcinoma cells (Caco-2 cells), human cytomegalovirus infection can only arise when these cells are in a specific state of differentiation, virus does not spread from cell to cell, and productive infection is rare.7 Although researchers have, in the past, attempted to link human cytomegalovirus with colon cancer, data have been inconsistent, and may have been limited by technologies and reagents available at the time.8–11 Our interest in the chronic inflammatory properties of malignant gliomas led us to assess whether human cytomegalovirus might be associated with these tumours, because of its unique biological properties. Unexpectedly, we discovered that the virus is strongly associated with malignant gliomas, and many gene products are expressed in these tumours.12 Because of results of two early reports suggesting that human cytomegalovirus DNA could be detected in colorectal cancers,8,11 we decided to assess the possibility that gene products of this virus are also expressed in these tumours.
Methods Procedures Clinical samples—We obtained formalin-fixed, paraffinembedded, surgical biopsy specimens of normal and neoplastic colon from the pathology archives of the Birmingham Veterans Affairs Hospital, and the University of Alabama at Birmingham, AL, USA. All samples were obtained in accordance with ethics guidelines from the institutional review board of each institution. A pathologist 1557
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
GLOSSARY IMMEDIATE EARLY GENES
The first set of viral genes expressed immediately after infection. These gene products—eg, IE1-72, which codes for a highly expressed regulatory phospho-protein—do not require viral DNA synthesis or proteins to be transcribed. EARLY GENES
These are the second set of viral genes expressed, and these gene products encode for viral DNA polymerases and other viral functions required for viral DNA synthesis, and some viral structural proteins--eg, the pp65 delayed early tegument protein. LATE GENES
Encode mostly for structural proteins used in viral assembly and packaging, and are generally expressed sequentially after early genes.
(ALV) re-examined all cases to confirm diagnosis, and we reviewed patients’ charts to ensure that no one was immunocompromised. Immunohistochemical analyses of paraffin sections—We cut 4-m paraffin sections from biopsy specimens of neoplastic and non-neoplastic colon, removed the paraffin from them, then hydrated the sections. We digested each section with pepsin (37°C, 4 min; BioGenex, San Ramon, CA, USA), retrieved antigen in citrate buffer (pH 6·0, 37°C, 12 h; BioGenex), and blocked non-specific peroxidase activity (3% H2O2, 12 min) and antibody binding (Fc-receptor blocker, 10 min, 20°C; Innovex Sciences, Richmond, CA, USA). We incubated each section with monoclonal antibody (anti-IE1-72 [1:20; BioGenex]; anti-pp65 [1:40; Novocastra, Newcastle-upon-Tyne, UK]; anti-CD34 [1:20; BioGenex]; anti-COX-2 [1:30; Cayman Chemical, Ann Arbor, MI, USA]) or no antibody in tris-buffered saline (pH 7·6) with 0·05% Tween 20 (TBST; 4°C, 12 h). IE1-72 is an immediate early gene product of human cytomegalovirus and pp65 is a delayed early gene product. We used anti-pp65 to assess delayed human cytomegalovirus gene expression. Since cyclo-oxygenase-2 (COX-2) is thought to have an important role in development and malignant progression of human colorectal cancer,13 and because COX-2 mRNA expression is strongly induced in primary human cells after infection with human cytomegalovirus and is needed for viral replication, we used anti-COX-2 to establish if expression of this enzyme was localised to a similar cell population as human cytomegalovirus protein expression.14,15 Negative controls for IE1-72 and pp65 immunohistochemistry were done, which included elimination of antibody from the reaction, substitution of antibody with an equivalent concentration of anti-CD34 isotypeidentical monoclonal antibody, which is specific for endothelial cells of blood vessels, or both. We rinsed slides with TBST and did colorimetric determination with a three-step horseradish peroxidase detection system (BioGenex) with the chromogen diaminobenzidine (Innovex Sciences). A pathologist (ALV)—masked to diagnosis and to whether antibody had been added—scored anti-IE1-72 immunostained specimens and controls as either positive or negative. A positive score showed that specific cell immunoreactivity was present and a negative score that none was present. Immunoreactivity for anti-pp65, antiCD34, and anti-COX-2 was scored similarly, but the pathologist was not masked. In-situ hybridisation of paraffin sections—To confirm that anti-IE1-72 and anti-pp65 were specific for human
1558
cytomegalovirus, and to show the presence of viral nucleic acids, we did in-situ hybridisation on paraffin sections of several of the tumour specimens that were immunoreactive for IE1-72. To detect viral nucleic acids, we obtained a biotinylated 21-base oligonucleotide (5⬘-GTGGTGGCGCTGGGGGTGGCG-3⬘) probe specific for the most abundantly transcribed early gene of human cytomegalovirus (ResGen, Huntsville, AL, USA). As a positive control, we used a biotinylated probe specific for poly-A RNA sequences. As a negative control, we either eliminated the specific human cytomegalovirus probe from the reaction, or we used a biotinylated 21-base oligonucleotide probe specific for herpes simplex virus 1 and 2 mRNA (ResGen), which had a similar GC content to the viral probe. We did enzyme digestion with pepsin (37°C, 15 min; Zymed Labs, San Francisco, CA, USA) and nucleic acid denaturation (90°C, 15 min) of 4-m paraffin sections with a Misha thermocycler (Shandon Lipshaw, Pittsburgh, PA, USA), and then slides were hybridised with probe overnight at 37°C in a humidified chamber. We detected probe with the DNA/mRNA alkaline phosphatase supersensitive detection system with the chromogen nitro-blue tetrazolium (BioGenex, San Ramon, CA). Cells were counterstained with methyl green, or not counterstained, and then mounted with aqueous mount for visualisation. Infection of Caco-2 cells in vitro—We used the Caco-2 human colon epithelial adenocarcinoma cell line, which has been characterised with respect to human cytomegalovirus infection and the effects of COX-2 expression.7,16,17 In culture, Caco-2 cells grow in colonies in which most cells differentiate after a few days. Human cytomegalovirus infection of these cells is dependent on the presence of rare Caco-2 cells that are in a specific state of differentiation, and therefore only a small proportion of cells are typically infected with virus, even at a high multiplicity of infection.7 We obtained human colon epithelial-derived Caco-2 cells from the American type culture collection (ATCC, Rockville, MD, USA) and cultured them at 37°C in 5% CO2 in Dulbecco’s Modified Eagle’s Medium (DMEM) with 10% fetal bovine serum (FBS; BioWhittaker, Walkersville, MD, USA). Caco-2 monolayers were grown for 5 days in culture, at which time they had assumed characteristic colony formation. We washed cells with phosphatebuffered saline (PBS) and infected (or mock-infected) them with human cytomegalovirus Towne strain (obtained from ATCC) by addition of virus to the media at a multiplicity of infection of one in DMEM with no serum. The mock-infected cells were treated identically except that no virus was added. After addition of virus, we allowed infection to proceed for 2 h at 37°C. We then washed cells with PBS, and DMEM with 10% FBS was again added. They were incubated in 5% CO2 at 37°C. After 48 h, cells were fixed with methanol (–20°C, 20 min), air-dried, and processed for double-staining immunohistochemistry or in-situ hybridisation. Immunohistochemistry of infected cells—We rinsed cells with TBST (pH 7·6), blocked endogenous peroxidase activity (3% H2O2, 8 min), and then incubated cells in Fc-receptor blocker (20°C, 10 min; BioGenex). Slides were blotted dry and primary monoclonal antibody against IE1-72 (1:300; BioGenex) was added (4°C, 12 h). We rinsed cells with TBST and detected antigen with a three-step horseradish peroxidase system (BioGenex) with diaminobenzidine black chromogen (Zymed Labs) or diaminobenzidine brown (Innovex
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Sciences). Coverslips were added with an aqueous mounting media. We photographed slides and then soaked them in water to remove coverslips and mounting media. Like COX-2, the anti-apoptotic protein Bcl-2 can be induced in tumour cells infected with human cytomegalovirus and can confer upon these cells a growth advantage and resistance to cytotoxic chemotherapeutic agents.18 Expression of Bcl-2 in colorectal adenocarcinoma is thought to have an important role in early tumour progression and may be activated by COX-2 expression.19,20 For COX-2 double-staining, we added anti-COX-2 (1:200, 4°C, 12 h; BioGenex) to cells immunostained with IE1-72 and diaminobenzidine black and detected antigen with the three-step alkaline phosphatase system with fast red chromogen (BioGenex). For Bcl-2 double-staining, we added anti-Bcl-2 (1:200, 4°C, 12 h; Novocastra) to cells immunostained with IE1-72 and diaminobenzidine brown and detected antigen as described for COX-2. Slides were photographed after mounting with aqueous media and a coverslip.
added as described above. We added coverslips, denatured cells (90°C, 15 min), and then cells were hybridised with probe (37°C, 90 min) in a humidified chamber. Cells were soaked in TBST and then probe wash (BioGenex) was added for 10 min and hybridisation was detected as described above with the chromogen nitro-blue tetrazolium. Cells were mounted, coverslips were added, and the cells were photographed. PCR and DNA sequencing—To confirm that our probe was specific for human cytomegalovirus nucleic acids, we extracted DNA from three to six paraffin sections of 10 m size, cut from a subset of the same biopsy specimens described above with QIAamp DNA mini kit (Qiagen, Valencia, CA, USA). To avoid potential PCR contamination, these experiments were done in a laboratory with no previous exposure to human cytomegalovirus, all preparations were processed masked, no positive controls were used in any PCR reactions, and blank paraffin blocks were cut sequentially between each patient’s sample and processed identically. Between each section, we changed the sectioning blade and wiped down the cutting surface with xylene and ethanol. For each specimen, at least two tubes of DNA were eluted and run as separate samples for PCR. From each sample, 250 ng
In-situ hybridisation of infected cells—We fixed cells in methanol (–20°C, 20 min) and then air-dried them. Biotinylated cytomegalovirus probe or control probe was Age (years), Diagnosis sex Patient number 1 77, m 2a 52, m 2b 2c 3 66, m 4 51, m 5 71, m 6 72, m 7 75, m 8a 80, m 8b 9 52, m 10 59, m 11 79, m 12 82, m 13 52, m 14 66, m 15 54, m 16a 61, m 16b 16c 17 71, m 18 66, m 19 52, f 20 74, m 21 52, m 22a 68, m 22b 23a 68, m 23b 24a 66, m 24b 24c 25a 65, m 25b 26a 71, m 26b 27a 67, m 27b 28 71, m 29a 54, m 29b
Non-neoplastic colonic epithelium Non-neoplastic colonic epithelium Polyp Adenocarcinoma Polyp Polyp Adenocarcinoma Polyp Polyp Non-neoplastic colonic epithelium Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Polyp Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Non-neoplastic colonic epithelium Non-neoplastic colonic epithelium Polyp Adenocarcinoma Non-neoplastic colonic epithelium Adenocarcinoma Polyp Adenocarcinoma Adenocarcinoma Adenocarcinoma Non-neoplastic colonic epithelium
Immunohistochemistry
In-situ hybridisation
UL55 PCR
Anti-IE1-72 Anti-pp65 Anti-COX-2 Anti-CD34* HCMV Negative control† HSV
Poly-T‡
– – + + + + + + + – + + + – + + + – + + + + – + + + + – + – + .. – – + – – + + – .. ..
.. .. .. .. + + .. .. .. .. .. .. .. .. .. + + .. .. .. .. .. + .. .. .. .. .. + .. + .. .. .. + .. .. .. + .. .. ..
– .. + + .. + + + .. .. + + .. .. + .. + .. .. .. + + .. .. + + + .. + .. + .. .. – + – .. – + – .. ..
.. – + .. + + .. .. .. .. .. + .. .. .. + + .. .. .. .. + + .. + .. .. .. + .. + .. .. .. + .. .. .. .. .. .. ..
– – – .. – – – .. – – .. .. .. – – – – – .. – .. – – – – .. – .. – – – – – – – – .. – – .. .. ..
.. .. .. .. + + .. .. .. .. .. + .. .. .. + + .. + .. + + + + + + .. .. + .. + .. .. .. + .. .. .. + .. .. ..
.. .. .. .. – – .. .. .. .. .. .. .. .. .. – – .. .. .. .. .. – .. .. .. .. .. – .. – .. .. .. .. .. .. .. .. .. .. ..
.. .. .. .. .. – .. .. .. .. .. – .. .. .. – .. .. .. .. .. .. .. .. .. .. .. .. – .. .. .. .. .. .. .. .. .. .. .. .. ..
.. – .. + .. .. .. + .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. + .. .. .. .. .. + – .. .. .. .. .. .. + + + –
HCMV=human cytomegalovirus; IE1-72=HCMV immediate early gene probe; pp65=delayed early gene probe; HSV=herpes simplex virus 1 and 2; UL55=HCMV glycoprotein B gene. ··=test not done. *Immunoreactivity present only in blood vessels. †Elimination of primary antibody or in-situ hybridisation probe from reaction. ‡Positive control probe for poly A RNA.
Immunohistochemical, in-situ hybridisation, and PCR results for paraffin specimens of human colonic polyps, adenocarcinomas, and nonneoplastic tissues
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
1559
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Blast search, and sequences were identical to that of UL55 (Genbank accession number X17403). Statistical analysis We calculated frequencies and proportions to summarise expression of IE1-72 and pp65 immunoreactivity for each sample type (polyps, adenocarcinomas, non-neoplastic epithelium). Pairwise comparisons of proportion of immunoreactivity between non-neoplastic and polyp and between non-neoplastic and adenocarcinoma were done with Fisher’s exact test. For these comparisons, independent patients were included in each group by exclusion of polyp or adenocarcinoma samples from patients with matching non-neoplastic colon. Role of the funding source The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Results
Figure 1: Immunohistochemistry and in-situ hybridisaton of human cytomegalovirus proteins and nucleic acids in human colonic polyps and adenocarcinomas (A) IE1-72 immunoreactivity (brown staining) in basal crypts in a hyperplastic polyp (arrow), and (B) in dysplastic crypts in a tubular adenoma (large arrow), but not in adjacent normal-appearing colon epithelium (small arrow). Low-power (C) and high-power (D) images of IE1-72 immunoreactivity in malignant-appearing cells but not adjacent normal-appearing epithelium in a colon adenocarcinoma (D=boxed area in C). Colon adenocarcinoma shows similar pattern of immunoreactivity for IE1-72 (E and F) and pp65 (G and H), except that inflammatory cells are also immunoreactive for pp65 in areas adjacent to tumour (arrow in H; F=boxed area in E; H=boxed area in G). (I) Blood-vessel immunoreactivity noted after CD-34 immunostaining (negative control); (J) arrows show endothelial-cell staining (J=boxed area in I). (K and L) Diffuse COX-2 immunoreactivity in adenocarcinoma. Low-power (M) and high-power (N–P) images of a different colon adenocarcinoma show IE1-72 immunoreactivity in tumour (N and P) but not adjacent normal crypts (O). pp65 immunoreactivity present in tumour (Q) but not in adjacent normal epithelium (R). (S) COX-2 immunoreactivity colocalised with IE1-72 and pp65 in the same adenocarcinoma shown in M–R. Only inflammatory cells are immunoreactive for COX-2 in normal appearing epithelium from same tumour (arrow in T). Low-power (U) and highpower (V) images of hybridisation of probe specific for human cytomegalovirus early nucleic acids with colon adenocarcinoma epithelium. (W) Probe omitted. (X) Positive control probe hybridisation (specific for poly-A mRNA). Original magnification ⫻40 (E, G, I, K, M, U, W, X), ⫻100 (A, B, C, N, O, Q, R, S, T), ⫻400 (D, F, H, J, L, P, V).
of DNA was amplified by nested PCR with internal and external primers specific for human cytomegalovirus glycoprotein B (UL55) gene.21 We judged samples positive if a band of 140 bp could be seen on agarose gel with ethidium bromide. Amplified DNA products from tumours were visualised on 1·5% agarose gels, bands were cut out, and DNA was extracted (Qiagen) and analysed by automated sequencing (ABI Model 377 DNA Sequencer, Foster City, CA, USA). Confirmation of human cytomegalovirus sequence was done by NCBI
1560
We obtained 45 paraffin-embedded, formalin-fixed biopsy samples from 29 different patients (table). We detected IE1-72 immunoreactivity in 14 (82%) of 17 polyps and 12 (80%) of 15 adenocarcinomas. IE1-72 protein was not typically detected in normal appearing colonic epithelium in areas adjacent to tumour within the same pathological section, nor was IE1-72 detected in tumour-free surgical biopsy specimens of colon from seven of these same patients. Results of pairwise comparisons showed a significant difference in IE1-72 immunoreactivity between non-neoplastic tissue and polyps (p=0·001) and between non-neoplastic tissue and adenocarcinomas (p=0·002). In hyperplastic polyps, IE1-72 immunoreactivity, when present, was detected only in discrete cell populations at the base of colonic crypts (figure 1, A). In tubular adenomas, IE1-72 immunoreactivity was detected in areas of dysplastic epithelium, and generally was not seen in areas of normal-appearing colonic crypts (figure 1, B). Colonic crypts immunoreactive for IE1-72 typically showed morphological characteristics of dysplastic aberrant crypt foci, with increased nuclear/cytoplasmic ratios, absence of uniform architecture, and nuclear migration away from the basement membrane (figure 1, B–D). In moderately and well-differentiated colon adenocarcinomas, IE1-72 immunoreactivity was seen throughout the neoplastic epithelium and was notably absent in adjacent areas of normal-appearing colonic crypts (figure 1, C–F, M–P). Little, if any, IE1-72 immunoreactivity was detected in areas of poorly differentiated adenocarcinoma, even when welldifferentiated areas of adenocarcinoma within the same specimen showed strong IE1-72 immunoreactivity (data available from authors). No tumour-cell immunoreactivity was detected in negative controls for IE1-72 immunohistochemistry (table; figure 1, I and J). We detected pp65 immunoreactivity in neoplastic epithelium and stromal inflammatory cells in neoplastic tumours (table, figure 1, G, H, Q), which accord with the pattern of expression we saw with respect to IE1-72 immunoreactivity. We did not detect pp65 immunoreactivity in normal-appearing colonic epithelium adjacent to tumour (figure 1, G, H, Q, R). In adjacent sections of the same tumour specimen, the pattern of pp65 immunoreactivity closely matched the pattern of IE1-72 immunoreactivity (figure 1, E–H). Results of negative controls for anti-pp65 were closely similar to those of anti-IE1-72. Similar to the pattern of IE1-72 immunoreactivity, pp65 immunoreactivity was noted in
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
M w
b
2b 2c b b
2a 2a
b
b
27b 27b 27b
140 bp M
6
20
w
20 24b 29b 29a
140 bp Figure 2: Nested PCR to detect UL55 gene in specimens of human colon cancers Band at 140 bp is amplified UL55 gene. Numbers above lanes refer to each patient’s biopsy specimen number, shown in the table. Most specimens had two to three DNA elutions made for each, and in some samples amplified products were identified in only one or two samples from a particular tumour. M=DNA ladder lane (underloaded); b=blank paraffin block specimens; w=negative PCR water controls.
seven (78%) of nine polyps and 11 (92%) of 12 adenocarcinomas, whereas non-neoplastic colon from two patients was not immunoreactive. Results of pairwise comparisons showed a significant difference in IE1-72 immunoreactivity between non-neoplastic and adenocarcinoma (p=0·033) but not between non-neoplastic and polyps (p=0·109). The biotinylated oligonucleotide probe specific for human cytomegalovirus early gene mRNA showed hybridisation with viral nucleic acids in neoplastic colon epithelium in several specimens (table, figure 1, U–X). No hybridisation was detected when probe was omitted from the reaction, or when probe specific for HSV-1 and HSV-2 was substituted (negative controls). Poly-A RNA probe hybridised to all cells (positive controls). We could detect amplified UL55 PCR products in six tumours (five adenocarcinomas and one polyp) that were immunoreactive for human cytomegalovirus, but in no non-neoplastic tumour-free margin biopsy specimens from three of these same six patients (figure 2). We detected no human cytomegalovirus amplified products in five blank paraffin-block control samples or four PCR water negative controls that were run in the same nested PCR reaction. Amplified PCR products from the six tumours were cut from agarose gels, and DNA sequencing of these products showed that all six were the human cytomegalovirus UL55 gene. We detected COX-2 expression in regions of tumour epithelium that in general corresponded with areas of IE1-72 and pp65 immunoreactivity, whereas little COX-2 immunoreactivity was detected in normal-appearing colonic epithelium (apart from stromal inflammatory cells; figure 1, K, L, S, T). Since results of these experiments suggested that COX-2 and human cytomegalovirus antigens colocalised in these tumours, we sought to establish if infection of human colon-cancer cells in vitro would lead to specific COX-2 expression in these cells. IE1-72 nuclear immunoreactivity was clearly seen in rare Caco-2 cells 48 h after infection. Double-labelling with anti-COX-2 showed intense COX-2 expression only in cells immunoreactive for IE1-72 (figure 3, C, D). These results were the same when we did pp65 immunostaining, suggesting that both anti-IE1-72 and anti-pp65 were specific for human cytomegalovirus proteins and were not cross-reacting with non-specific colon epithelial cell epitopes (data available from authors). No IE1-72 or pp65 immunoreactivity was noted, and only faint COX-2 immunoreactivity was seen in control (mock-infected) Caco-2 cells (data available from authors). We recorded that Caco-2 cells expressing IE1-72 had increased expression of Bcl-2, which was not seen in uninfected adjacent cells or mock-infected cells (figure 3, E, F).
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
Figure 3: Double-staining immunohistochemistry and in-situ hybridisation for human cytomegalovirus in Caco-2 colon adenocarcinoma cells (A) Rare Caco-2-cell nuclear immunoreactivity (black staining shown by arrows) for IE1-72 (B=boxed cell in A). (C) After double-labelling with anti-COX2, infected cells show increased COX-2 expression (red staining; arrows show same cells in A; D=boxed cell in C). Low-power (E) and high-power (F) views of a colony of Caco-2 cells double-stained with anti-IE1-72 (brown chromogen in nuclei) and anti-Bcl-2 (red chromogen in cytoplasm). In-situ hybridisation of infected cells at low-power (G) and high-power (H) with probe specific for human cytomegalovirus early nucleic acids shows discrete hybridisation pattern of infected cells similar to that seen with IE1-72 immunostaining. Original magnification ⫻40 (A, C, E, G) and ⫻400 (B, D, F, H).
Human cytomegalovirus nucleic acids were seen to hybridise in discrete Caco-2 cells (figure 3, G, H). No evidence of virus nucleic acid hybridisation was noted in mock-infected control cells, nor in virus-infected Caco-2 cells when the human cytomegalovirus probe was omitted from the hybridisation reaction (data available from authors).
Discussion We have shown that human cytomegalovirus proteins and nucleic acids can be found that localise specifically in human colorectal polyps and adenocarcinomas. Present paradigms for development of human colorectal adenocarcinoma show that a multistep progression of genetic mutations takes place over years in colonic epithelium, characterised by phenotypic progression from polyp to invasive adenocarcinoma.1,22 Of these alterations, those that dysregulate the adenomatous polyposis coli and TP53 tumour suppressor pathways and induce COX-2 expression are key. Although our findings are unexpected, they could tie in with these existing paradigms.
1561
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
Human herpes viruses are presently implicated in the pathogenesis of at least four human malignancies. Epstein-Barr Virus (EBV) is associated with Burkitt’s lymphoma, nasopharyngeal carcinoma, and B-cell lymphoproliferative syndromes, and human herpes virus 8 (HHV-8) is associated with Kaposi’s sarcoma.23 EBVrelated malignancies arise after years of viral dormancy and are accompanied or triggered by viral reactivation. Present ideas for the pathogenesis of HHV-8 in Kaposi’s sarcoma suggest that tumour initiation and progression is associated with cytokine-mediated reactivation of HHV-8, and may arise only in individuals with specific immunogenetic phenotypes.24 In early-developing Kaposi’s sarcoma lesions, HHV-8 viral load may be low or undetectable, and latently expressed viral genes are thought to induce cell transformation.24 Our data accord with preferential replication of human cytomegalovirus in dysplastic colon epithelial cells. These cells are known to harbour replicative virus in immunocompromised patients, in whom focal colonic ulcerative lesions often arise.4 Jarvis and colleagues7 reported that human cytomegalovirus infection of Caco-2 colon epithelial cells could only happen when these cells are in a specific state of differentiation, which accords with our in-vitro data. Should this occurrence be the case in vivo, dysplastic colonic epithelial cells with impaired cell-cycle control mechanisms may provide a unique reservoir for human cytomegalovirus persistence with no effect on the overall tumour phenotype. Alternatively, long-term, persistent human cytomegalovirus infection and expression in dysplastic colonic epithelial cells could be important in promotion of oncogenic events directly implicated in malignant progression. Human cytomegalovirus infection activates cellular proto-oncogenes, cyclins, and kinases involved in cell mitogenic and cell survival pathways, including c-myc, c-fos, c-jun, cyclin-B, cyclin-E, mitogen-activated protein kinase, ERK 1/2, and PI3-kinase.25–29 This virus also induces the NF-B transcription factor, which might have an important role in tumour-cell survival pathways.30 We, and others,18 have shown that human cytomegalovirus infection of tumour cells can induce Bcl-2 expression, which confers resistance to apoptotic signalling during tumour progression. The human cytomegalovirus UL111A gene product has been reported to transform cells in vitro and dysregulate cell-cycle control pathways by binding TP53 and disrupting its cellular transcription.31 The virus gene products IE1-72 and IE2-86 can also dysregulate cell-cycle checkpoint controls by interaction with the TP53 and retinoblastoma tumour-suppressor proteins and downstream pathways.32 Even in the presence of functional TP53, IE2-86 can block downstream TP53 cell-cycle arrest pathways by degradation of CDKN1A (p21cip1), a principal mediator of the TP53induced G1 cell-cycle arrest.33 When combined with the adenovirus E1A protein, which impairs retinoblastoma tumour-suppressor protein function, IE1-72 and IE2-86 can induce oncogenic transformation of cells.34 IE1-72 and IE2-86 can also block apoptosis mediated by tumour necrosis factor ␣ and are highly mutagenic; thus expression of these gene products could strikingly increase genomic instability in a population of chronically infected cells.34,35 In addition to evasion of apoptotic and cell-cycle control pathways, tumour cells infected with human cytomegalovirus might also escape normal host antitumour cytotoxic T cell and natural killer cell immune
1562
responses, since the virus encodes for many gene products that block MHC class 1 and class 2 antigen synthesis and function of natural-killer cells.36 Our data, and those of others, show that human cytomegalovirus induces expression of COX-2.14 Since COX-2 can promote angiogenesis and invasion of colon adenocarcinoma cells, virus-infected colon-cancer cells might acquire an additional growth advantage through these mechanisms.16,17 In view of the many cellular modulatory properties of this virus, our data justify further studies to establish whether human cytomegalovirus participates in pathogenesis of colorectal cancer. Contributors L Harkins did immunohistochemistry, in-situ hybridisation, and PCR experiments. A L Volk interpreted immunostaining data. M Samanta did cytomegalovirus infections in vitro and PCR sequencing. I Mikolaenko did DNA extractions and PCR optimisation. W J Britt and K I Bland contributed to study design and writing of the report. C S Cobbs contributed to study design, did immunohistochemistry, in-situ hybridisation, PCR experiments, interpreted data, and wrote the report.
Conflict of interest statement None declared.
Acknowledgments We thank Heidi Weiss for assistance with statistical analysis. CSC is supported by a Merit Review Grant for cancer research from the Veterans Administration.
References 1
2
3
4
5
6
7
8
9
10 11 12
13 14
15
16
Fearon ER. Molecular biology of gastrointestinal cancers. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: principles and practice of oncology, 6th edn. Philadelphia: Lippincott, Williams, and Wilkins, 2001: 1037–49. Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide incidence of eighteen major cancers in 1985. Int J Cancer 1993; 54: 594–606. Birkeland SA, Storm HH, Lamm LU, et al. Cancer risk after renal transplantation in the Nordic countries, 1964–1986. Int J Cancer 1995; 60: 183–89. Britt WJ, Alford CA. Cytomegalovirus. In: Fields BN, Knipe DM, Howley PM, eds. Fields virology, vol 2, 3rd edn. Philadelphia: Lippincott, Williams, and Wilkins, 1996: 2493–523. Doniger J, Sumitra M, Rosenthal LJ. Human cytomegalovirus and human herpesvirus 6 genes that transform and transactivate. Clin Microbiol Rev 1999; 12: 367–82. Cinatl J Jr, Cinatl J, Vogel JU, et al. Modulatory effects of human cytomegalovirus infection on malignant properties of cancer cells. Intervirology 1996; 39: 259–69. Jarvis MA, Wang CE, Meyers HL, et al. Human cytomegalovirus infection of caco-2 cells occurs at the basolateral membrane and is differentiation state dependent. J Virol 1999; 73: 4552–60. Huang ES, Roche JK. Cytomegalovirus DNA and adenocarcinoma of the colon: evidence for latent viral infection. Lancet 1978; 1: 957–60. Hashiro GM, Horikami S, Loh PC. Cytomegalovirus isolations from cell cultures of human adenocarcinomas of the colon. Intervirology 1979; 12: 84–88. Hart H, Neill WA, Norval M. Lack of association of cytomegalovirus with adenocarcinoma of the colon. Gut 1982; 23: 21–30. Ruger R, Fleckenstein B. Cytomegalovirus DNA in colorectal carcinoma tissues. Klin Wochenschr 1985; 63: 405–08. Cobbs CS, Harkins L, Samanta M, et al. Human cytomegalovirus infection and expression in human malignant glioma. Cancer Res 2002; 62: 3347–50. Williams C, Shattuck-Brandt RL, DuBois RN. The role of COX-2 in intestinal cancer. Ann N Y Acad Sci 1999; 889: 72–83. Zhu H, Cong JP, Mamtora G, et al. Cellular gene expression altered by human cytomegalovirus: global monitoring with oligonucleotide arrays. Proc Natl Acad Sci USA 1998; 95: 14470–75. Zhu H, Cong JP, Yu D, Bresnahan WA, Shenk TE. Inhibition of cyclooxygenase 2 blocks human cytomegalovirus replication. Proc Natl Acad Sci USA 2002; 99: 3932–37. Tsujii M, Kawano S, Tsuji S, et al. Cyclooxygenase regulates angiogenesis induced by colon cancer cells. Cell 1998; 93: 705–16.
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
MECHANISMS OF DISEASE
17 Tsujii M, Kawano S, DuBois RN. Cyclooxygenase-2 expression in human colon cancer cells increases metastatic potential. Proc Natl Acad Sci USA 1997; 94: 3336–40. 18 Cinatl J Jr, Cinatl J, Vogel JU, et al. Persistent human cytomegalovirus infection induces drug resistance and alteration of programmed cell death in human neuroblastoma cells. Cancer Res 1998; 58: 367–72. 19 Sheng H, Shao J, Morrow JD, et al. Modulation of apoptosis and Bcl-2 expression by prostaglandin E2 in human colon cancer cells. Cancer Res 1998; 58: 362–66. 20 Merritt AJ, Potten CS, Watson AJ, et al. Differential expression of bcl-2 in intestinal epithelia: correlation with attenuation of apoptosis in colonic crypts and the incidence of colonic neoplasia. J Cell Sci 1995; 108: 2261–71. 21 Kuhn JE, Wendland T, Eggers HJ, et al. Quantitation of human cytomegalovirus genomes in the brain of AIDS patients. J Med Virol 1995; 47: 70–82. 22 Kinzler KW, Vogelstein B. Lessons from hereditary colorectal cancer. Cell 1996; 87: 159–70. 23 Howley PM, Ganem D, Kieff E. DNA viruses. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: principles and practice of oncology, 6th edn. Philadelphia: Lippincott, Williams, and Wilkins, 2001: 168–73. 24 Sturzl M, Zietz C, Monini P, Ensoli B. Human herpesvirus-8 and Kaposi’s sarcoma: relationship with the multistep concept of tumorigenesis. Adv Cancer Res 2001; 81: 125–59. 25 Hagemeier C, Walker SM, Sissons PJ, Sinclair JH. The 72K IE1 and 80K IE2 proteins of human cytomegalovirus independently transactivate the c-fos, c-myc and hsp70 promoters via basal promoter elements. J Gen Virol 1992; 73: 2385–93. 26 Johnson RA, Wang X, Ma XL, et al. Human cytomegalovirus up-regulates the phosphatidylinositol 3-kinase (PI3-K) pathway: inhibition of PI3-K activity inhibits viral replication and virus-induced signaling. J Virol 2001; 75: 6022–32.
THE LANCET • Vol 360 • November 16, 2002 • www.thelancet.com
27 Johnson RA, Huong SM, Huang ES. Activation of the mitogenactivated protein kinase p38 by human cytomegalovirus infection through two distinct pathways: a novel mechanism for activation of p38. J Virol 2000; 74: 1158–67. 28 Jault FM, Jault JM, Ruchti F, et al. Cytomegalovirus infection induces high levels of cyclins, phosphorylated Rb, and p53, leading to cell cycle arrest. J Virol 1995; 69: 6697–704. 29 Bresnahan WA, Albrecht T, Thompson EA. The cyclin E promoter is activated by human cytomegalovirus 86-kDa immediate early proteins. J Biol Chem 1998; 273: 22075–82. 30 Yurochko AD, Kowalik TF, Huong SM, Huang ES. Human cytomegalovirus upregulates NF-kappa B activity by transactivating the NF-kappa B p105/p50 and p65 promoters. J Virol 1995; 69: 5391–400. 31 Muralidhar S, Doniger J, Mendelson E, et al. Human cytomegalovirus mtrII oncoprotein binds to p53 and down-regulates p53-activated transcription. J Virol 1996; 70: 8691–700. 32 Castillo JP, Yurochko AD, Kowalik TF. Role of human cytomegalovirus immediate-early proteins in cell growth control. J Virol 2000; 74: 8028–37. 33 Chen Z, Knutson E, Kurosky A, Albrecht T. Degradation of p21cip1 in cells productively infected with human cytomegalovirus. J Virol 2001; 75: 3613–25. 34 Shen Y, Zhu H, Shenk T. Human cytomegalovirus IE1 and IE2 proteins are mutagenic and mediate “hit and run” oncogenic transformation in cooperation with the adenovirus E1A proteins. Proc Natl Acad Sci USA 1997; 94: 3341–45. 35 Zhu H, Shen Y, Shenk T. Human cytomegalovirus IE1 and IE2 proteins block apoptosis. J Virol 1995; 69: 7960–70. 36 Loenen WA, Bruggeman CA, Wiertz, EJ. Immune evasion by human cytomegalovirus: lessons in immunology and cell biology. Semin Immunol 2001; 13: 41–49.
1563
For personal use. Only reproduce with permission from The Lancet Publishing Group.