Human papillomavirus 16 E6 polymorphisms in cervical lesions from different European populations and their correlation with human leukocyte antigen class II haplotypes
Version of Record online: 30 AUG 2001
Copyright © 2001 Wiley-Liss, Inc.
International Journal of Cancer
Volume 94, Issue 5, pages 711–716, 1 December 2001
How to Cite
Zehbe, I., Tachezy, R., Mytilineos, J., Voglino, G., Mikyškova, I., Delius, H., Marongiu, A., Gissmann, L., Wilander, E. and Tommasino, M. (2001), Human papillomavirus 16 E6 polymorphisms in cervical lesions from different European populations and their correlation with human leukocyte antigen class II haplotypes. Int. J. Cancer, 94: 711–716. doi: 10.1002/ijc.1520
- Issue online: 30 OCT 2001
- Version of Record online: 30 AUG 2001
- Manuscript Accepted: 29 JUN 2001
- Manuscript Revised: 27 JUN 2001
- Manuscript Received: 7 MAR 2001
- Wilhelm Sander-Stiftung, Munich, Germany. Grant Number: 98.065.1
- Granting Agency, Ministry of Health, Prague, Czech Republic. Grant Number: 5959-3
- NATO Collaborative Linkage. Grant Number: LST.CLG 975794
- E6 and E7 genotypes;
- cervical lesion;
- human leukocyte antigen
Infection with high-risk human papillomavirus (HPV) is necessary for the development of a cervical lesion, but only a fraction of precursor lesions progress to cancer. Additional factors, other than HPV type per se, are likely to increase the probability for progression. Intratype genome variations have been reported to be associated with viral persistence and the development of a major cervical disease. We have recently shown that the prevalence of specific HPV16-E6 variants in invasive cervical cancer (ICC) varies between Italian and Swedish women. To extend our initial study we have analyzed E6 variants in cervical lesions from Czech women, ranging from low-grade cervical intraepithelial neoplasia (LCIN) to ICC and scaled up the sample size of our initial study of Swedish and Italian women. In addition, we have correlated the cases of cancers with human leukocyte antigen (HLA) class II haplotypes. In line with our earlier observation, the distribution of specific HPV16-E6 genotypes in CIN and ICC varied in the 3 cohorts. For instance, the HPV16-E6 L83V variant, which has been found to be positively associated with ICC in Swedish women (p = 0.002), was more prevalent in LCIN than in ICC in Italian and Czech women (p = 0.01 and = 0.03, respectively). These data indicate that host genetic factors, such as HLA polymorphism, may determine the potential oncogenicity of the HPV16-E6 L83V variant. Indeed, the DR04-DQ03 haplotype, which is approximately 3-fold more abundant in the normal Swedish population than in those in Italy and the Czech Republic, was found to be positively associated with HPV16-E6 L83V in the 3 cohorts investigated (p = 0.01). This observation may explain why L83V is a risk factor more in Sweden than in the other 2 countries. © 2001 Wiley-Liss, Inc.
Certain human papillomavirus (HPV) genotypes are etiologic agents for cervical cancer development.1 HPV16 is the most frequently detected genotype in cervical carcinoma and its precursors.2–5 Biochemical data have demonstrated that the products of 2 early genes, E6 and E7, are the major oncoproteins of HPV16.6 HPV16 E6 and E7 associate with and inactivate the biologic function of several cellular proteins, including the regulators of cell cycle checkpoints, p53 and pRb.7
Infection with high-risk HPV is the principal risk factor for the development of an invasive lesion. However, most HPV infections regress spontaneously and, for the cases that do progress to cancer, a long period of latency is normally required. It is clear that additional risk factors play a role in disease progression. Recently, independent studies have provided evidence that specific intratype HPV genome variations may influence the persistence of the infection and the progression of a precursor lesion to cancer.8–10 Several North American and South American studies have shown that nonprototype-like HPV16 variants, which harbor several nucleotide changes within the long control region and the E6 gene, confer a higher risk for the development of anogenital high-grade lesions than the prototype-like variants, which contain no or only few nucleotide changes within the same regions.11–15 In a follow-up study, an HPV16 E6 variant, E-G350, harboring an amino acid change at position 83 substituting a leucine for a valine (L83V), was found to be associated with progression from cervical intraepithelial neoplasia-I (CIN-I) to CIN-III.8 We have recently shown that the same E6 variation (L83V) is more prevalent in cases of invasive cervical carcinoma (ICC) from Swedish women than the prototype.10, 16 Our initial findings in an Italian cohort have indicated that in this group the HPV16-E6 L83V variant is mainly associated with precursor lesions.9 Thus, the potential oncogenicity of L83V appears to be dependent on genetic differences between populations. However, a full confirmation of this hypothesis requires an analysis of HPV16-E6 polymorphisms in other cohorts.
To assess the role of viral genome variations in cervical disease, we analyzed the sequences of the E6 gene in HPV16-positive cervical biopsies from 169 Swedish, 91 Italian and 107 Czech women. The samples range from low-grade lesion to cancer. In addition, we analyzed human leukocyte antigen (HLA) class II haplotypes in the cases of cancers. We have conducted a cross-sectional investigation because prospective studies of the progression from high-grade lesion to cancer cannot be conducted due to ethical reasons. Here, we have shown that HPV16-E6 genotypes have a different distribution in various grades of cervical lesions in the same cohort and that the prevalence of specific HPV16-E6 polymorphisms varies among European countries. In addition, a positive association between the HLA class II DR04-DQ03 haplotype and the L83V variant was found in the cohorts investigated.
MATERIAL AND METHODS
We have undertaken 3 cross-sectional studies and included exclusively HPV16-positive biopsies, which were available at the Department of Genetics and Pathology, University Hospital Uppsala, Sweden (n = 169); at the Servizio di Anatomia Patologica, Ospedale Sant'Anna in Turin, Italy (n = 91); and at the Department of Experimental Virology, Institute of Hematology and Blood Transfusion in Prague, Czech Republic (n = 107). Only specimens from women of the same geographical area were accepted for each individual study. According to these criteria, we selected in Sweden 34 cases of low-grade CIN (LCIN; mean age 29 years, range 18–34 years), 61 cases of high-grade CIN (HCIN; mean age 33 years, range 23–48 years) and 74 cases of ICC of squamous cell origin (mean age 53 years, range 29–83 years); in Italy 21 cases of LCIN (mean age 33 years, range 23–43 years), 23 cases of HCIN (mean age 36 years, range 27–47 years) and 47 cases of ICC (mean age 53 years, range 29–47 years); and in the Czech Republic 17 cases of LCIN (mean age 40 years, range 23–58 years), 47 cases of HCIN (mean age 33 years, range 20–53 years) and 43 cases of ICC (mean age 49 years, range 27–87 years).
The E6 sequence was previously shown to be highly polymorphic, whereas the E7 sequence hardly varied.9 Therefore we analyzed all cases with regard to E6 and determined E7 where enough material was left, that is, in 105 Swedish samples (7 LCIN, 54 HCIN, 44 ICC), 30 Italian samples (15 LCIN, 7 HCIN, 8 ICC) and 37 Czech samples (2 LCIN, 17 HCIN, 18 ICC).
HLA-DR and -DQ alleles could be typed in 68 cases of ICC of the Swedish population, in 39 cases of ICC of the Italian population and in 36 cases of ICC of the Czech population by sequence-specific primer PCR (PCR-SSP). For cases in which PCR-SSP was not successful, or where not enough DNA was available, HLA-DR and -DQ typing was performed with the LIPA assay (Abbot, Wiesbaden, Germany).
HPV16-E6- and E7-specific PCR was performed with a hot start modification using the TaqStart antibody PT1576-1 (Clontech, Palo Alto, CA) with primers flanking the coding region of HPV16-E6 (nucleotides 52–575): 5′-CGAAACCGGTTAGTATAA-3′ and 5′-GTATCTCCATGCATGATT-3′ and (PCR) E7 (nucleotides 480–985): 5′-ATAATATAAGGGGTCGGTGG-3′ and 5′-CATTTTCGTTCTCGTCATCTG-3′. Forty amplification cycles were run in the GeneAmp PCR System 2400 (Applied Biosystems, Weiterstadt, Germany) with a 94°C denaturation step (1 min), a 55°C annealing step (1 min) and a 72°C extension step (2 min), including an initial denaturation step of 3 min and a final extension step of 7 min resulting in 524- and 506-bp products for E6 and E7, respectively. PCR products were checked by ethidium bromide agarose gel electrophoresis.
After post-PCR clean up with the JetQuick Spin PCR Purification kit (Genomed, Bad Oeynhausen, Germany), the HPV16-E6 products were sequenced by the fluorescent dye dideoxy termination method using an ABI Prism 377 DNA sequencer (Applied Biosystems). For the sequencing reaction, the same primers were used as for the PCR reaction. The sequences were determined for both strands. In approximately half of the samples, the PCR amplification and sequencing were repeated to exclude PCR artifacts.
An isolate was classified as variant if it had at least 1 amino acid change (polymorphism or variation) when compared with the reference isolate.17 Cases with nucleotide changes not leading to amino acid changes were assigned to the prototype. Genotype denotes either variant or prototype of HPV16-E6 or -E7. Proportions of viral genotypes in cases were compared statistically using the Pearson chi-square test. The Cochran-Armitage test for a linear trend was used to assess a trend of increasing or decreasing HPV16-E6 genotypes in CIN and ICC.
Population control HLA class II frequencies are quoted from published information of women with normal cytology from the Uppsala Region (Sweden)18 and from the Proceedings of the 11th International Histocompatibility Workshop and Conference (Italy and Czech Republic).19 For comparing E6 genotypes with HLA class II haplotypes, we performed the Pearson chi-square test or in the case of small sample sizes the Fisher exact test. Statistical analysis was done with StatXact (Version 4.01 Cytel Software Corporation).
Distribution of HPV16-E6 and -E7 genotypes in Swedish and Italian women with cervical lesions
To expand our initial data on the distribution of HPV16 polymorphisms in cervical lesions from Swedish and Italian women,9, 10 we scaled up the sample size from 42 to 169 of the Swedish group and from 36 to 91 of the Italian group. All E6 and E7 genotypes analyzed are listed in Figure 1.
In the Swedish cohort, the E6 sequence corresponding to the prototype was found in 71 of 169 lesions (42%), whereas in the Italian cohort 14 of 91 lesions (15%) harbored this E6 genotype. All other cases contained nucleotide exchanges in the E6 gene leading to amino acid changes. The most frequent E6 polymorphism in both groups was found to be L83V, having been detected alone or in combination with another variant in approximately 80% of the variant cases in each group. The remaining E6 variants contain multiple polymorphisms (from 3 to 5 amino acid substitutions) of the African, Asian-American or North American lineages20, 21 or single amino acid changes found at other sites than residue 83.
We subsequently determined whether specific HPV16-E6 polymorphisms were preferentially associated with a particular type of cervical lesion. Based on the nucleotide exchanges described above, the HPV16-E6 genotypes were subdivided into 4 different groups: L83V with or without another variation (L83V+), non-European multivariants, prototype and other variants20, 21 (Table I). In the Swedish cohort, L83V+ was significantly enriched in ICC (p = 0.002) and increased linearly with the severity of the lesion (p = 0.001). The opposite result was obtained in the Italian group, as L83V+ was enriched in LCIN (p = 0.01) but decreased linearly with the severity of the lesion (p = 0.01). In contrast to L83V+, the E6 multivariants are almost exclusively associated with ICC in the Italian but not in the Swedish group, being 10-fold higher in ICC than in CIN (p = 0.003). The E6 prototype distribution did not differ significantly in preinvasive and invasive lesions in the Italian group, whereas this genotype was enriched in LCIN (p = 0.001) but decreased in a linear fashion in HCIN and cancer of the Swedish group (p = 0.0005). Other E6 variants were not preferentially associated with a certain type of lesion in the Swedish group. In the Italian group, 4 such cases were found among patients with cancer.
|Patients||L83V+ (%)||Multivariants (%)||Prototype (%)||Other E6 variants (%)|
|LCIN (n = 34)||10 (29)||1 (3)||21 (62)||2 (6)|
|HCIN (n = 61)||28 (46)||2 (3)||27 (44)||4 (7)|
|ICC (n = 74)||45 (61)||2 (3)||21 (28)||6 (8)|
|LCIN (n = 21)||18 (86)||1 (5)||2 (10)||0|
|HCIN (n = 23)||17 (74)||0||6 (26)||0|
|ICC (n = 47)||26 (55)||11 (23)||6 (13)||4 (9)|
|LCIN (n = 17)||13 (76)||0||3 (18)||1 (6)|
|HCIN (n = 47)||19 (40)||2 (4)||25 (53)||1 (2)|
|ICC (n = 43)||20 (46)||2 (5)||18 (42)||3 (7)|
Next, we determined the E7 sequence in 105 Swedish specimens and in 30 Italian specimens. In agreement with previous studies,10, 22, 23 we observed that amino acid changes in the HPV16-E7 gene are rare events (Fig. 1). In the Swedish group, 4 cases (4%) contain amino acid changes and in the Italian group 1 case (3%). Silent mutations were identified in 13 cases (12%) of the Swedish lesions and in 4 cases (13%) of the Italian lesions.
Distribution of HPV16-E6 and -E7 genotypes in Czech women with cervical lesions
To accumulate further evidence that the prevalence of specific HPV16-E6 genotypes in ICC varies in different populations, we performed an additional cross-sectional study in another European country. We selected 107 HPV16-positive lesions from Czech women (Fig. 1). The percentage of the HPV16-E6 prototype was identical to the Swedish cohort (42%). Also in the Czech cohort L83V+ was the most prevalent HPV16-E6 polymorphism, being present in 85% of the variant cases. The remaining E6 polymorphisms comprised 4 non-European multivariants20, 21 and 5 different genotypes with 1 amino acid substitution. We then determined their distribution in preinvasive lesions and cancers (Table I). As observed in the Italian cohort, the L83V polymorphism was highly represented in LCIN, whereas its prevalence decreased significantly in HCIN (p = 0.007) and in ICC (p = 0.03). In contrast, the E6 prototype was detected with a low frequency in LCIN and was found to be increased in HCIN (p = 0.01) and in ICC (p = 0.08). However, no significant linear trend with lesion grade was observed for either genotype, being p = 0.1 (L83V) and p = 0.2 (prototype). The distribution of the multivariants, which was clearly enriched in ICC from Italian women, did not differ significantly in preinvasive and invasive lesions of the Czech cohort, being detected only twice in HCIN. Other variants were not associated with a particular lesion grade. E7 sequences were established in 37 cases (Fig. 1). None showed an amino acid substitution. One silent mutation in base 822 (AG change) was detected in 5 cases.
Together, these results are in line with our initial investigation indicating that variants in the HPV16-E6 protein are common, whereas variants are uncommon for the E7 protein. Moreover, the distribution of certain HPV16-E6 genotypes can vary in cervical lesions in different geographical regions.9
Correlation of HLA class II polymorphisms with HPV16-E6 genotypes
Amino acid substitutions in the E6 protein may result in an alteration of immunogenicity with a consequent evasion of host immune surveillance. To evaluate this hypothesis, we analyzed the correlation between specific E6 genotypes and HLA class II haplotypes. To obtain more conclusive results, only cases of cancer representing the final stage of cervical disease were included in the HLA analysis (n = 143). Indeed, precursor lesions have still the potential to regress. Therefore, they were not considered in this context. We compared L83V or the remaining genotypes with HLA. The results are summarized in Table II. In all groups, the DR04-DQ03 showed a trend for a positive association with the cases of cancer containing L83V+. Although the increase between L83V+ vs. the other E6 genotypes was high, a statistically significant result was obtained only after pooling the samples (p = 0.01). This finding might have resulted from the small sample size in each group. No other haplotype was affected in all 3 groups. Several haplotypes were found positively or negatively associated with L83V+ in 1 or 2 of the cohorts. For instance, DR07-DQ02 was positively associated with L83V+ in the Italian group [not significant (NS)], whereas the same haplotype was negatively associated with L83V+ in the Czech group (p = 0.05). However, a positive association between the DR07 allele and L83V was observed in a Dutch study (p = 0.08).24 DR13-DQ06 was found to be negatively associated with L83V+ in the Swedish (NS) and the Czech group (NS).
|DR-DQ haplotypes||Controls %||L83V+ %||Cases other E6 genotypes %||All cases of ICC %|
|Swedish women||n = 3501||n = 82||n = 54||n = 136|
|Italian women||n = 10142||n = 44||n = 34||n = 78|
|Czech women||n = 1502||n = 38||n = 34||n = 72|
Our data indicate a link between certain HLA class II haplotypes and specific HPV16-E6 genotypes. Indeed, DR04-DQ03 was found to be associated with L83V+ in all cohorts analyzed in this study. For the other HLA-E6 genotype associations larger studies are required to finally assess a possible correlation.
In this investigation, we have characterized HPV16-E6 polymorphisms in preinvasive and invasive cervical lesions from Swedish, Italian and Czech women. Consistent with our initial study,9 these data show that the distribution of a specific HPV16-E6 variant in LCIN, HCIN and ICC can vary in different cohorts. The HPV16-E6 L83V polymorphism was the most frequent genotype in the 3 groups examined and increased with the severity of the lesion in Swedish women, whereas the opposite result was obtained for the other 2 cohorts. A similar pattern seen in Swedish women was observed in Norwegian women (Andrew Jenkins, personal communication), whereas a similar pattern seen in Italian and Czech women was observed in a Dutch population.25 In a German study, L83V was reported to be evenly distributed in precursor lesions and cancer.22 The L83V trend observed in Italian women in the current study was less pronounced than in our previous report, because this E6 genotype decreased respectively from approximately 80% in LCIN to 50% in ICC and 70% in CIN to 15% in ICC.9 Interestingly, E6 non-European multivariants were increased 10-fold in cancer of Italian women as compared with LCIN, whereas in Swedish and Czech women multivariants were not preferentially associated with a particular type of cervical lesion. Also the HPV16-E6 prototype had a different distribution in the 3 cohorts. In Swedish and Czech women, the prototype was inversely represented in ICC, whereas in Italian women, the prototype was rarely detected in all types of lesions analyzed and had a similar prevalence in LCIN and ICC.
The fact that the distribution of HPV16-E6 polymorphisms in cervical lesions varied in different European cohorts, provides evidence for a possible link between E6 variants and HLA haplotypes, which are population related. The involvement of HLA class II in HPV-induced cervical disease has been suggested by several independent studies.18, 26–29 However, in these investigations, the characterization of E6 genotypes was not considered. Here, we provide evidence for a positive association between the HLA class II haplotype DR04-DQ03 and HPV16-E6 L83V in the 3 European groups investigated, whereas DR04-DQ03 was negatively associated with the other HPV16 E6 genotypes. Interestingly, the prevalence of the DR04-DQ03 haplotype in the healthy control populations differed in Sweden, Italy and the Czech Republic, accounting for 23%, 6% and 10% in the respective populations. Thus, the enrichment of L83V in ICC as compared with LCIN in Swedish women can be explained by the fact that DR04-DQ03 is the most prevalent HLA class II haplotype in the Swedish population. Other associations between HLA haplotypes and L83V could be observed in 1 or 2 of the cohorts examined. However, they may become more evident when a larger number of cases are analyzed. Based on our findings, an E6 rather than an E7 cytotoxic T-cell (CTL) response appears to be important for counteracting an HPV16 infection. Indeed, Nakagawa and colleagues30 have shown that in women with an E6 CTL response the viral infection cleared, whereas in women with a persistent infection an E6 CTL response did not occur. In addition, an E7 CTL response did not show such correlation, because it was detected in women with regressive or progressive lesions. These findings provide an explanation why amino acid substitutions in the E7 protein are rarely detected, whereas amino acid substitutions are frequent in the E6 protein.
In summary, the data in the current study favor the notion that cervical carcinogenesis is, at least in part, influenced by genetic polymorphisms of both HPV and HLA. However, to finally prove the role of HLA in HPV-induced cervical disease, large-scale epidemiologic studies are required.
We thank Professor Harald zur Hausen for kind support.
- 19Allele and haplotype frequencies for HLA and complement loci in various ethnic groups. In: AizawaM, SasazukiT, eds. HLA 1991. Proceedings of the 11th International Histocompatibility Workshop and Conference, vol. 1. Oxford University Press, Oxford, UK: 1992., , , et al.
- 21The biological significance of human papillomavirus type 16 variants for the development of cervical neoplasia. Papillomavirus Rep 1999;10: 105–16., .