Oncogene (2008) 27, 2532–2541 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE involved in cell adhesion, cell motility and mitogenic signaling are altered due to HPV 16 E5 expression

N Kivi1, D Greco2, P Auvinen2 and E Auvinen1

1Department of Virology, Haartman Institute, University of Helsinki and Helsinki University Central Hospital, Helsinki, Finland and 2Institute of Biotechnology, University of Helsinki, Helsinki, Finland

We investigated the effects of the human papillomavirus leads to indefinite proliferation and immortalization of type 16 E5 oncogene on cellular expression in human keratinocytes (Munger and Howley, 2002) and induces epithelial cells using cDNA microarray. In a genome-wide premalignant neoplasms together with v-Ha-Ras in vivo microarray assay, the expression of 179 genes was found (Schreiber et al., 2004). Both E6 and E7 oncoproteins to be significantly altered due to E5 expression. The can also independently immortalize human cells, but expression of lamin A/C was downregulated at protein with reduced efficiency (Band et al., 1990; Halbert et al., level. The expression of protein kinase C-d and phosphoi- 1991). The role of the E5 protein in cellular transforma- nositide-3-kinase was found to be upregulated. tion is much less understood. HPV 16 E5 is an 83-amino We also observed increased motility of E5-expressing acid membrane-associated protein found in the Golgi cells. We conclude that the E5 protein affects several apparatus, endoplasmic reticulum and nuclear mem- cellular pathways involved in cell adhesion, cell motility brane (Conrad et al., 1993). It is able to transform and mitogenic signaling. These alterations may together murine fibroblasts and keratinocytes (Valle and Banks, lead to inhibition of apoptosis and facilitate the establish- 1995; Chen et al., 1996). E5 also enhances the ment of persistent infection in the epithelium. immortalization potential of E6 and E7, and stimulates Oncogene (2008) 27, 2532–2541; doi:10.1038/sj.onc.1210916; the proliferation of human and mouse primary cells in published online 5 November 2007 cooperation with E7 (Bouvard et al., 1994; Sto¨ ppler et al., 1996). The E5 protein is known to inhibit gap- Keywords: HPV; papillomavirus; E5; microarray; cell junctional intercellular communication (Oelze et al., adhesion; cell motility 1995), which may lead to disturbance of normal epithelial maintenance and differentiation. It has also been shown that E5 alters the growth and differentiation of stratified epithelia and induces epithelial tumors at a high frequency in mice (Genther Williams et al., 2005). Introduction Moreover, important functions in late stages of viral replication have been suggested (Fehrmann et al., 2003; Cervical cancer is one of the most common causes of Genther et al., 2003). cancer-related deaths in women worldwide (Pisani et al., Alterations in cellular signaling pathways commonly 2002). A prerequisite for cervical cancer is infection by a contribute to cancer. A number of viruses have developed high-risk human papillomavirus (HPV; Walboomers mechanisms to modulate cellular signaling pathways for et al., 1999), out of which HPV 16 is the predominant reprogramming host cells and supporting their life cycles, type in all continents (Mun˜ oz et al., 2004). Cancer- or for controlling host defense responses (DiMaio et al., associated HPV types encode three oncogenes, E5, E6 1998; Burgert et al., 2002; Duerst and Morrison, 2003). and E7, and the E6 and E7 proteins have a significant We examined genome-wide by cDNA role in malignant transformation (Hawley-Nelson et al., microarrays in a stable epithelial cell line to investigate 1989; Munger et al., 1989; Leechanachai et al., 1992; holistic effects of the E5 oncogene. The expression of Pim et al., 1992; Straight et al., 1993). E6 and E7 several genes of the phosphatidylinositol phosphate stimulate cell proliferation by interfering with the kinase pathway appeared altered in microarray and were function of regulatory proteins in cells, including the validated by quantitative PCR. Protein kinase C-d tumor suppressors p53 and pRB (Dyson et al., 1989; (PKC-d) and phosphoinositide-3-kinase p55 regulatory Werness et al., 1990). Coexpression of both E6 and E7 subunit (PI3KR3) were found to be upregulated, and lamin A/C was downregulated at both mRNA and protein level. E5 protein seems to affect several cellular Correspondence: N Kivi, Department of Virology, Haartman pathways involved in cell adhesion, cell motility and Institute, University of Helsinki, PO 21 (Haartmaninkatu 3), Helsinki mitogenic signaling. We were also able to verify altered 00014, Finland. E-mail: niina.kivi@helsinki.fi cell motility of E5-expressing cells by live-cell imaging. Received 1 March 2007; revised 31 August 2007; accepted 10 October This is the first report on a large-scale transcriptome 2007; published online 5 November 2007 analysis of HPV E5 effects in epithelial cells. HPV 16 E5 alters cellular gene expression N Kivi et al 2533 Results Validation of gene expression profiling of selected genes by RT–PCR Gene expression profiling using cDNA microarrays On the basis of the microarray results, altogether 27 genes We performed a cDNA microarray experiment to screen were selected according to their biological relevance for for the expression of 16 000 transcripts in human qRT-PCR validation (Table 1). The 24-h time point was epithelial HaCaT cells expressing the E5 protein of used for validation. The microarray result was considered HPV 16 (HaCaT-E5). E5 expression was induced with valid, if the P-value of the qRT-PCR triplicates was o0.05. dexamethasone and confirmed by northern blotting and For 15 out of 27 genes (56%), the microarray result could quantitative real-time RT-PCR (qRT-PCR). No anti- be confirmed. Similar validation rates of cDNA microarray body to the HPV 16 E5 protein is available. E5 results have been reported by other authors (Rajeevan transcription was observed already after short induc- et al., 2001; Chuaqui et al., 2002). Upregulation of PI3K tion, and high expression level of E5 mRNA was and inositol polyphosphate-4-phosphatase was confirmed. reached at 24 h (Table 2). For this reason, 24 h induction Also upregulation of fibronectin (FN1), 67-kDa laminin was selected for the microarray experiment. Moreover, receptor and PKC-d transcripts was observed with both we have previously shown that prolonged expression of methods. Of matrix metalloproteinases (MMPs), down- the E5 protein for several days, albeit at a high level in regulation of MMP-7 was validated. Twelve genes gave transient transfections, is lethal to cells (Auvinen et al., discordant fold change as compared to microarray, 2004). Cellular gene expression in E5 cells was compared including MMP-16 and lamin A/C. to that in control cells transfected with the empty vector To obtain more profound information about the (HaCaT-pMSG), cultured and treated in a similar dynamics of E5 effect on host cell pathways, we selected manner. a set of further genes from the same We initially assumed that even a minor change in gene families, including some downstream target genes. expression may be of importance, and thus no cutoff Another 34 genes were chosen, and the expression of was set as to the fold change of gene expression. Since these altogether 61 genes was studied by qRT-PCR after the availability of reliable statistical methods, P-values E5 induction for 0, 2, 4, 24, 48 and 72 h to follow the have been launched for validation. P-value o0.01 was effect on cellular gene expression along with time. We employed to infer statistical significance of the observa- found that already at earlier time points, E5 expression tions. Using this P-value cutoff, we identified 117 affects transcription of a majority of cellular genes upregulated and 62 downregulated genes in E5-expres- studied, including MMP-7, -12, -16 and PKC-d (Table 2). sing cells from the microarray assay (Supplementary The transcript levels of several genes oscillated along Table 1). These genes were functionally grouped with time. This may partly represent true alterations in according to the Gene Ontology Biological Process gene expression, although further analysis is needed to (BP) database (Ashburner et al., 2000; Figure 1). The better understand this finding. statistically significant over-represented gene categories (Fisher’s exact test P-value o0.05) were metabolism, Changes of the selected genes and their downstream protein modification and various responses to external effectors at protein level stimulus, for example stress, biotic stimulus and immune Of the genes whose altered expression was confirmed by responses. qRT-PCR, 15 were also investigated by western blotting

histone deacetylation response to biotic phosphorylation response to pest\, stimulus pathogen or parasite defense response macromolecule metabolism immune response

intracellular transport amino acid biosynthesis response to stimulus intracellular protein transport response to external biotic response to stress antigen processing antigen presentation protein modification phosphorus metabolism protein metabolism

phosphate metabolism Figure 1 Gene Ontology chart. Clustering analysis was performed on 114 genes that were altered due to HPV 16 E5 expression in the microarray analysis. Distribution of major functional categories for biological process terms is shown in the pie chart.

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2534 Table 1 Validation of microarray results by qRT-PCR at 0, 2, 4, 24, 48 and 72 h time points (Figure 2a). The Gene name PCR FC MA FC selection of these genes was largely determined by the (P-value) (P-value) availability of antibodies. E5 cells were compared to pMSG control cells induced in a similar manner. The PCR-validated, concordant expression of each protein was normalized against the Major histocompatibility 1.5 (0.0025) 1.8 (0.0053) complex, class II, DQ expression of b-actin, which was used as a loading beta 1 control in each blot. Interferon, alpha- 1.5 (0.0016) 2.4 (0.0014) PI3K p55 regulatory subunit (PI3KR3) and class 2 inducible protein (clone beta polypeptide, phosphatidylinositol transfer protein IFI-15K) as well as inositol polyphosphate-4-phosphatase were Proteasome (prosome, 1.1 (0.0094) 4.7 (0.0077) macropain) subunit, found to be upregulated in the microarray experiment, alpha type, 4 along with downregulation of phosphatidylinositol-4- Inositol polyphosphate-4- 1.4 (0.0014) 1.2 (0.0045) phosphate-5-kinase. Slight increment of PI3KR3 was phosphatase, type I, also seen at protein level (Figure 2a). The effect of E5 107 kDa Protein kinase H11 5.1 (1.12EÀ06) 1.8 (0.0027) expression on phosphorylation and expression of Akt, Protein kinase C, delta 1.4 (0.0094) 1.6 (0.0089) the main effector of PI3K, was also investigated. Laminin receptor 1 1.3 (0.0086) 1.7 (0.0008) Interestingly, modest increase in Akt expression was (ribosomal protein SA, seen in E5 cells (Figure 2b) and little if any increase was 67 kDa) seen in Akt phosphorylation (Figure 2b). The E5 protein Fibronectin 1 2.5 (0.0004) 1.4 (0.0096) v-abl Abelson murine 1.6 (0.0064) 1.2 (0.0057) may thus have an effect on promoting cell survival by leukemia viral oncogene enhancing the expression of Akt. homolog 1 The protein levels of PKC-d appeared to be upregu- Homocysteine-inducible, 2.3 (0.0004) 1.2 (0.0054) lated, while lamin A/C was downregulated in E5- endoplasmic reticulum stress-inducible, ubiqui- expressing cells, similar to the measured transcript tin-like domain member 1 levels (Figure 2). No differences between E5-expressing Hepatitis B virus x 2.3 (0.0037) 1.3 (0.0094) cells and control cells were observed in the amount interacting protein of intracellular proteins MMP-7, -12, -16, keratin-8, Phosphoinositide-3- 2.3 (0.0018) 1.4 (0.0089) a-tubulin, g-tubulin, PAK6, symplekin, MCM4, kinase, class 2, beta polypeptide EphA1, FN1 and osteoprotegerin as studied by western Macrophage stimulating À2.6(0.0009) À2.1 (0.0027) blotting (data not shown), even though alterations in 1 receptor (c-met-related mRNA level were observed. tyrosine kinase) Among the PCR-validated genes, upregulation of key Golgi associated, gamma À1.2 (0.0071) À1.5 (0.0040) adaptin ear containing, cell-adhesion components such as 67-kDa laminin ARF binding protein 3 receptor and laminin-a4 important in integrin signaling Matrix metalloproteinase À1.7 (0.0007) À2.7 (0.0004) was observed. We wanted to investigate whether this 7 (matrilysin, uterine) would imply increased activity in cell adhesion. How- ever, no differences in the expression or phosphorylation PCR-validated, discordant Guanine nucleotide- À1.3 (0.0192) 1.2 (0.0029) of the downstream effector focal adhesion kinase (FAK) binding protein were observed (data not shown). Nevertheless, upregu- (G protein), beta-5 lated expression as well as increased phosphorylation of Lamin A/C À2.5 (2.09EÀ04) 1.5 (0.0050) paxillin downstream of FAK in HaCaT-E5 cells as Polymerase (DNA 1.3 (3.18EÀ03) À1.4 (0.0042) directed) kappa compared to control cells was observed (Figure 2c), Polymerase (DNA 1.4 (3.53EÀ03) À1.7 (0.0030) suggesting enhancement of cell adhesion or cell motility. directed) iota Cell division cycle 25B À2.6 (3.32EÀ03) 1.3 (6.56EÀ04) Matrix metalloproteinase À8.5 (4.61EÀ04) 1.4 (0.0048) HPV E5 increases the rate of wound healing in HaCaT 16 (membrane-inserted) Homo sapiens tubulin, À1.8 (4.96EÀ04) 1.8 (0.0085) cell cultures alpha, ubiquitous To elucidate whether HPV E5 has an effect on cell Symplekin; Huntingtin À565.6 (2.97EÀ08) 1.3 (0.0059) motility or cell proliferation, a wound-healing assay was interacting protein I performed using confluent monolayers of HaCaT-E5 Zinc finger protein 259 À11.5 (4.73EÀ03) 1.3 (0.0061) Interleukin 2 receptor, À1.3 (0.0168) 1.9 (0.0043) and HaCaT-pMSG control cells. Wound closure was gamma monitored over time using a live-cell imaging system Laminin, alpha 4 5.3 (0.0001) À1.5 (0.0018) (Figure 3; Supplementary Movies). HaCaT-E5 cells Hepatitis delta antigen- À1.4 (0.0115) 1.4 (0.0038) migrated into the wound at a higher rate than control interacting protein A cells. Total cell number was counted from each picture to differentiate whether the wound closure was due to Abbreviations: FC, fold change; MA, microarray. The genes analysed by qRT-PCR are classified into two groups (PCR-validated, con- cell division or cell migration. Statistical analysis cordant and PCR-validated, discordant) depending on the correlation revealed no differences in cell number along with time with the microarray results. PCR fold changes and in the second (data not shown) between HaCaT-E5 and -pMSG cells, column the microarray fold changes are presented. indicating that wound closure occurred mainly due to cell movement.

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2535 Table 2 Genes analysed by quantitative RT-PCR in different E5 induction time points

0h 2h 4h 24h 48h 72h HPV-16 E5 7366.0 (0.0008) 750070.5 (1.4E-05) 12007.7 (0.0112) 161368.6 (0.0076) 270.0 (0.0011) 5765.9 (0.0021) Laminin, gamma1 -1.8 (0.0044) -2.6 (0.0002) -2.3 (0.0003) -2.0 (0.0002) -1.8 (0.0002) -3.3 (0.0039) p21-activated kinase 6 -2.6 (0.0196) 2.2 (0.0043) -1.3 (0.0127) -1.4 (0.0043) -2.0 (0.0040) -2.7 (0.0011) osteoprotegerin 1.4 (0.004) -1.4 (0.0318) 1.3 (0.0437) 3.3 (0.0318) 7.0 (3.7E-05) 6.6 (1.8E-06) RNA polymerase polypeptide D -1.3 (0.0010) -1.4 (0.0044) -1.3 (0.0027) 1.1 (0.0044) 1.0 (0.4208) -1.6 (0.0350) elongation factor, RNApol II, 2 -1.2 (0.0037) 1.1 (0.0446) 1.2 (0.0051) 1.5 (0.0446) 1.2 (0.1156) -1.1 (0.0666) CDC28 protein kinase regulatory subunit 1B -1.7 (0.0027) -1.7 (0.5975) -1.8 (0.0031) -1.4 (0.5975) -1.5 (0.0404) -1.5 (0.0049)

PI3K, regulatory subunit, polypeptide 3 -1.3 (0.0148) 1.4 (0.4620) 1.1 (0.0242) 1.4 (0.4620) 1.6 (0.0024) 1.6 (0.0259) CDK inhibitor 2B 1.4 (0.0124) 1.4 (0.0012) 1.2 (0.0016) 2.0 (0.0012) 1.3 (0.0026) -1.6 (0.0975) MMP-12 -3.4 (0.0006) -3.7 (0.0008) -13.8 (0.0001) -1.3 (0.0008) -8.5 (0.0007) -14.0 (0.0029) fibronectin leucine rich transmembrane protein 0.9 (0.4188) -1.0 (0.8688) 1.0 (0.9305) 1.1 (0.8688) -1.2 (0.0055) -2.0 (0.0012) glycoprotein A33 -3.0 (0.0003) -3.7 (0.0007) -1.2 (0.2353) 4.0 (0.0007) 1.5 (0.1334) -2.4 (0.0149) ATPase, H+ transporting, lysosomal, beta polypeptide -4.3 (0.0016) 2.2 (0.0005) 1.1 (0.2545) 2.5 (0.0005) 1.7 (0.0050) 1.6 (0.0025) INPP1 -1.5 (0.0820) -1.4 (0.0033) -1.4 (0.0010) 1.7 (0.0033) -1.8 (0.0262) -2.3 (0.0022)

beta-catenin-interacting protein ICAT -1.9 (0.0292) -1.6 (0.0321) -1.9 (0.0005) -1.3 (0.0321) -2.1 (0.0015) -2.0 (0.0061) MCM4 -2.0 (0.0265) -1.2 (0.0126) -1.2 (0.1682) -1.1 (0.0126) -1.1 (0.1848) -1.4 (0.0296) interferon-induced transmembrane protein 2 1.1 (0.0468) 1.2 (0.0852) 1.1 (0.3432) 1.0 (0.0852) -1.0 (0.7002) -1.4 (0.0002)

B-cell receptor-associated protein 31 0.9 (0.2128) -1.3 (0.0420) -1.4 (0.0030) -1.2 (0.0420) -1.7 (0.0037) -1.8 (0.0005) 5'-AMP-activated protein kinase, gamma-1 subunit 1.1 (0.4694) -1.7 (0.0112) 1.1 (0.3812) 1.1 (0.0112) -1.2 (0.1444) -1.8 (0.0228)

collagen-16 alpha-1 chain precursor -3.0 (0.0029) -2.6 (0.0010) 1.3 (0.6119) 2.6 (0.0010) 1.3 (0.0230) -2.1 (0.0630)

lumphocyte antigen 6 complex, locus E 1.2 (0.1598) 1.8 (0.0017) 1.1 (0.7684) 2.1 (0.0017) 2.0 (0.0001) 1.7 (0.0001) thrombomodulin -3.3 (0.0006) 1.6 (8.0E-06) -2.7 (1.4E-05) 1.1 (8.0E-06) -1.6 (0.0039) -1.9 (0.0021) synaptotaqmin-like protein 2 2.3 (0.0010) 1.4 (0.0002) 2.3 (2.2E-05) 1.9 (0.0002) 2.5 (0.0001) -1.2 (0.0197) ubiquitin spesific protease 39 1.0 (1.0000) -1.04 (0.1017) -1.3 (0.0008) 1.0 (0.1017) -1.2 (0.4124) -1.3 (0.0428) HLA-C -1.4 (0.2560) -3.2 (0.0871) 1.6 (0.0017) -1.1 (0.6743) 1.3 (0.0348) 1.3 (0.1154) HLA-DQB -1.1 (0.4793) -8.9 (0.0027) 1.5 (0.0172) 1.5 (0.0025) 2.3 (0.0008) 3.8 (0.0033) interleukin 15 precursor -1.6 (0.1153) 1.2 (0.1427) 1.0 (0.8720) 1.0 (0.9136) -1.7 (0.0182) -2.1 (0.0033) interferon alpha-inducible protein -1.2 (0.5110) -1.4 (0.0122) 2.2 (0.0129) 1.5 (0.0016) 1.1 (0.5450) 1.5 (0.0002) interleukin 2 receptor, gamma 2.1 (0.0156) -3.9 (0.0149) 2.7 (0.2326) 1.7 (0.1665) 7.4 (0.0232) 12.6 (0.1101) proteasome subunit, alpha type, 4 -1.3 (0.0581) -337.8 (0.0128) 1.3 (0.1624) 1.1 (0.0094) -1.6 (0.0035) -1.8 (0.0015)

inositol polyphosphate-4-phosphatase -4.4 (0.0001) -1.1 (0.1070) -2.3 (0.0593) 1.4 (0.0014) -1.8 (0.0176) 1.1 (0.7404) PI3K C2 beta-polypeptide 2.1 (0.0027) -177.7 (0.2918) 1.3 (0.7326) 1.4 (0.0680) -1.2 (0.0339) 1.7 (0.0035) Protein kinase H11 3.0 (0.0012) 438.1 (0.4699) 1.7 (0.2313) 5.1 (1.1E-06) 1.7 (0.3565) 3.2 (0.0035) Protein kinase C -6.3 (0.0200) -1.2 (0.0269) 2.4 (0.0553) 1.2 (0.0484) 3.4 (0.0059) 3.4 (0.7469) MAPKK 3 2.3 (2.4E-06) -6.7 (0.0015) 1.5 (0.7160) 1.0 (0.8058) 1.7 (0.0248) -1.4 (0.0447) MSP receptor -11.5 (0.0637) -23.2 (0.0002) 1.1 (0.5359) -1.3 (0.1046) 1.1 (0.6656) -4.6 (0.0011) guanine nucleotide-binding protein beta subunit 5 -1.8 (0.1285) -1.5 (0.0139) -1.3 (0.0081) -1.3 (0.0192) -2.0 (0.0001) -3.3 (0.0160) Rho guanine nucleotide exchange factor (GEF) 17 1.3 (0.4602) 2.8 (0.0149) -1.6 (0.2787) 1.2 (0.4953) -2.1 (0.0172) -1.1 (0.8440) laminin alpha 4 1.8 (0.4701) 3.0 (0.0052) 5.4 (0.0024) 5.3 (0.0001) 2.5 (0.0006) -2.0 (0.0473) 67 kDa laminin receptor 1.4 (0.5470) -3.8 (0.0002) 1.2 (0.0006) 1.3 (0.0086) -1.6 (0.0021) -3.0 (0.0001) lamin A/C -14.6 (0.0010) -2.8 (0.0202) -1.4 (0.1025) -2.5 (0.0002) -1.6 (0.0438) 1.0 (0.6546) fibronectin 1 -1.3 (0.0201) 4.8 (0.0788) -1.0 (0.9786) 2.5 (0.0004) 3.5 (0.0006) 3.8 (0.0284) Kallmann syndrome sequence -2.1 (0.0032) 3.4 (0.0076) 2.3 (0.0001) 1.0 (0.8109) -1.0 (0.9968) 2.1 (0.0160) polymerase (DNA directed) kappa -1.8 (0.0093) -31.2 (0.0002) -1.5 (0.2713) 1.3 (0.0032) -1.4 (0.2834) -4.0 (0.0011) polymerase (DNA directed) iota 1.2 (0.4723) -2.2 (0.0004) 2.2 (0.0107) 1.4 (0.0035) 1.1 (0.1978) -3.7 (1.7E-05) cdc 25B -13.2 (0.0003) -24.9 (3.7E-06) 1.3 (0.0890) -2.6 (0.0033) -1.0 (0.9271) -2.9 (0.0083) RAB2 -1.4 (0.0001) -4.4 (3.7E-05) 1.3 (0.2996) 1.0 (0.9517) -1.3 (0.3421) 2.9 (0.5750) V-ATPase 116 kDa isoform a2 -1.4 (0.0006) -1.3 (0.0341) 1.5 (0.1390) -1.1 (0.3747) 1.4 (0.0227) -2.3 (0.0003) reticulocalbin 1 -1.1 (0.4768) -1.8 (9.6E-06) 1.1 (0.6513) -1.0 (0.6171) 1.5 (0.0586) -1.4 (0.0459) homocysteine-responsive ER-resident ubiquitin-like domain member 1 protein -1.4 (0.0144) 1.2 (0.0181) 1.6 (0.0407) 1.3 (0.2080) 1.1 (0.3850) -3.2 (0.0001) keratin 8 -2.5 (0.0002) 1.0 (0.9316) 1.3 (0.1831) -1.1 (0.4911) -1.6 (0.0395) -1.8 (0.0664) EphA1 -2.8 (0.0007) 52.5 (0.0036) -1.9 (0.0608) -1.4 (0.2851) -2.5 (0.001) -4.5 (0.0059) v-abl abelson murine leukemia viral oncogene homolog 1 -1.4 (0.3665) -11.3 (0.0019) 2.8 (0.0009) 1.6 (0.0064) 3.2 (0.0189) 1.2 (0.1427) prohibitin -2.0 (0.0005) 2.6 (0.1423) 1.9 (0.1544) 1.0 (0.8269) 1.1 (0.4544) -1.4 (0.0122) MMP 7 -2.2 (0.1290) -6.1 (0.0003) 2.1 (0.0051) -1.7 (0.0007) -1.5 (0.1030) -3.9 (0.0149) MMP 16 -672.5 (0.0003) -11.6 (0.2480) -23.9 (0.2084) -8.5 (0.0005) -17.5 (0.1723) -337.8 (0.0128) tubulin alpha-1 -2.0 (0.0001) 3.8 (0.0001) 1.2 (0.1013) -1.8 (0.0005) 1.0 (0.9126) -1.1 (0.0918) symplekin -8.2 (0.0001) -81.2 (0.0177) -1568.3 (0.1245) -565.6 (3.0E-08) -2.2 (0.0060) -4.6 (0.2918) zinc finger protein 259 -1.1 (0.2524) -1.4 (0.0215) 1.9 (0.0003) -11.5 (0.0047) 1.4 (0.0256) -1.5 (0.0002) nuclear receptor subfamily 2, group c, member 1 -2.0 (1.27E-05) -2.3 (0.0004) 1.2 (0.0256) 1.3 (0.1210) -1.2 (0.1818) -1.7 (0.4699) hepatitis delta antigen interacting protein A -2.1 (0.0320) -14.7 (0.0002) 1.6 (0.0311) -1.1 (0.6843) 1.8 (0.0093) -1.2 (0.0269)

hepatitis B virus x interacting protein 1.2 (0.1626) -5.9 (0.0002) 1.4 (0.0326) 2.3 (0.0037) -5.4 (0.0069) -1.6 (0.0703)

Results are presented as fold changes (P-values) in E5 cells as compared to control cells. Green, upregulated genes; red, downregulated genes; white, nonsignificant.

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2536 0 2 4 24 48 72 (hours) E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG

18.6 1.0 2.8 1.0 1.1 1.0 1.5 1.0 1.1 1.0 1.0 1.0

← PI3KR3 (55 kDa)

← β-actin (43 kDa)

-1.8 1.0 1.4 1.0 2.5 1.0 3.6 1.0 2.7 1.0 1.6 1.0

← PKC-delta (79 kDa)

← β-actin (43 kDa)

-1.7 1.0 -1.1 1.0 -2.6 1.0 -1.6 1.0 -1.9 1.0 -1.2 1.0

← Lamin A (74 kDa) ← Lamin C (65 kDa)

← β-actin (43 kDa)

02 4244872(hours) E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG

2.2 1.0 1.5 1.0 1.6 1.0 1.5 1.0 1.3 1.0 1.3 1.0

← Akt (60 kDa)

1.1 1.0 1.2 1.0 1.4 1.0 1.2 1.0 1.3 1.0 1.1 1.0

← P-Akt (60 kDa)

← β-actin (43 kDa)

02 4244872(hours) E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG E5 pMSG

1.3 1.0 2.1 1.0 2.1 1.0 1.3 1.0 1.0 1.0 2.6 1.0

← Paxillin (68 kDa)

1.8 1.0 3.8 1.0 1.7 1.0 3.6 1.0 3.2 1.0 7.2 1.0

← P-Paxillin (68 kDa)

← β-actin (43 kDa)

Figure 2 Quantitative western blotting using the Odyssey Infrared Imaging System. (a) Expression of PI3KR3, PKC-d and lamin A/C in HaCaT-E5 and -pMSG cells at 0, 2, 4, 24, 48 and 72 h induction with dexamethasone. (b) Expression of Akt and P-Akt after 0, 2, 4, 24, 48 and 72 h induction. (c) Expression of paxillin and P-paxillin after 0, 2, 4, 24, 48 and 72 h induction. Blotting for b-actin was used as a loading control and is shown separately for each blot in the figure. Numeric values above each lane are quantitation results as expressed in fold changes in E5 cells as compared to pMSG control cells at each time point, and normalized against b-actin expression. PI3KR3, phosphoinositide-3-kinase p55 regulatory subunit; PKC-d, protein kinase C-d.

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2537 HaCaT-E5 HaCaT-pMSG

0 h

24 h

48 h

Figure 3 Motility of HaCaT-E5 and -pMSG cells. A wound was made to a confluent monolayer of HaCaT-E5 and -pMSG cells, and the area was marked for identification and imaged at 1-h intervals over a 48-h period. Representative time points (0, 24 and 48 h) after wounding are presented.

Discussion The PI3K pathway is essential in maintaining the undifferentiated phenotype of basal keratinocytes, as Here we present a global analysis of gene expression the pathway is switched off when basal cells detach from modulations due to the HPV 16 E5 oncogene. Genes the basement membrane and enter the differentiation involved in antigen presentation, mitogenic signaling, program of epithelial cells (Sayama et al., 2002). PI3K cell adhesion, cell proliferation, differentiation and activity generates phosphoinositides, which control cell motility were found to be altered in E5-expressing cells survival, proliferation, migration and cellular trafficking as compared to control cells. Leykauf et al. (2004) (Cantley, 2002; Czech, 2003). Inositol polyphosphate-4- earlier reported a careful analysis of E5 protein effects phosphatase is a regulator of PI3K, and it has been on membrane-protein composition using nanoelectros- implicated in PI3K signaling at distinct subcellular pray ionization mass spectrometry. These authors junctions (Ivetac et al., 2005). Constitutive upregulation noticed differential expression of proteins involved in of cell-survival signaling mediated by PI3K/Akt has maturation and transport of MHC class I complexes to already been implicated in oncogenic transformation the plasma membrane. Further, Bravo et al. (2005) and tumor development, as it prevents apoptotic cell studied intracellular lipid dynamics in human keratino- death during uncontrolled proliferation (Chang et al., cytes, and showed that E5 modifies the availability of 2003). It has been shown that the HPV E5 protein free cholesterol in the plasma membrane and alters the protects cells from ultraviolet-B-induced apoptosis and glycerophospholipid composition of the membranes. As promotes survival of human foreskin keratinocytes by compared to our work, different sets of data were activating PI3K/Akt and MAPK signaling downstream presented in those studies, which were specifically of the epidermal growth factor receptor (Zhang et al., designed to investigate E5 effects on membranes. We 2002). It has also been suggested that the induction of present here the largest report so far on the effect of the PI3K/Akt-dependent protection from apoptosis by E5 E5 protein on cellular gene expression. contributes to the HPV E7-mediated oncogenesis Prevention of apoptosis is a mechanism developed by (Zhang et al., 2002). Akt activation due to E5 has a number of viruses either to prevent premature death of recently been shown to increase vascular endothelial the host cell to maximize virus progeny, or to facilitate growth factor expression downstream of epidermal the establishment of persistent infection (O’Brien, 1998). growth factor receptor, which may enhance angiogenesis

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2538 in cervical cancer (Kim et al., 2006). The modest gap-junctional intercellular communication (Lampe upregulation of Akt observed in this work suggests that et al., 2000). The HPV 16 E5 protein has been reported the E5 protein may contribute to tumorigenesis by to inhibit gap-junctional intercellular communication preventing apoptosis and enhancing cell survival. (Oelze et al., 1995), and this could at least partially be Adhesion to fibronectin as mediated by integrins explained by the increase in PKC-d expression. triggers downstream signaling to FAK and paxillin. Upon normal epithelial development, basal epithelial Increased activity of these proteins has been implicated cells have to detach from integrin and laminins of the in cervical carcinogenesis, but it has previously been basement membrane, cease to proliferate and enter the addressed to the E6 and E7 proteins in HPV-immorta- differentiation program. Instead, in epithelial dysplasia lized cells (McCormack et al., 1997). Increased FAK cancer, cells do not differentiate but continue to activity has also been found in cells immortalized by proliferate. The HPV 16 E5 protein seems to play an other factors than HPV (McCormack et al., 1997). We important role in these events, as it protects cells from observed upregulation of total and phosphorylated apoptosis, affects cell–matrix adhesion and interferes paxillin downstream of FAK. Also, increased levels of with epithelial differentiation. Our findings contribute to FN1 transcript in E5-expressing cells were shown, the understanding of the mechanisms of cellular whereas increase in intracellular FN1 protein was not transformation and carcinogenesis. detected. In addition, increased amount of laminin-a4 and laminin 67-kDa receptor transcripts were measured. High expression of 67-kDa laminin receptor has Materials and methods previously been reported in HPV-associated cervical neoplasia (Demeter et al., 1992), and it has been shown Cell culture to play a key role in tumor invasion and metastasis HaCaT human keratinocytes stably transfected with HPV 16 (Castronovo, 1993). Because several of these genes have E5 (HaCaT-E5) under the control of a dexamethasone- been implicated in cell motility, we further wanted to inducible promoter or with the empty vector pMSG (Ha- investigate whether the E5 effect would be seen in a live- CaT-pMSG) as a control (Oelze et al., 1995) were used. Cells cell wound-healing assay. Cell motility was indeed were grown in Dulbecco’s modified Eagle’s growth medium shown to be increased by E5 expression. Thomsen supplemented with 10% fetal bovine serum, glutamine and et al. (1999) earlier reported a reduction in the motility penicillin–streptomycin to 70–80% confluency. Cells were of E5-expressing cells as compared to control cells. Their serum starved for 24 h and induced with 1 mM dexamethasone (Sigma-Aldrich Inc., Saint Louis, MO, USA) for 0 and 24 h for experimental setting was specifically designed to study the microarray analysis. To study the different kinetic patterns the motility of individual cells, whereas we were of altered gene expression by quantitative RT-PCR, the cells interested in directional movement of a population of were induced for 0, 2, 4, 24, 48 or 72 h. The MMTV promoter epithelial cells with the help of intracellular junctions. of the pMSG vector is somewhat leaky, and low E5 expression Also, mouse embryonic fibroblasts were used by takes place even without induction. Comparisons were Thomsen et al. (1999), while our model system consisted performed between E5 and control cells treated in a similar of human epithelial cells. manner. Matrix metalloproteinases are proteins that degrade the extracellular matrix and are thus essential in a RNA extraction and hybridization to cDNA microarrays number of physiological processes and pathological Total RNA was isolated from confluent cells that had been conditions including embryonic development, differen- serum starved and induced with dexamethasone for 0 or 24 h. tiation, tissue remodeling and cancer (Hojilla et al., RNA concentration was quantitated by ultraviolet spectro- 2003). Expression of MMP-7 has been found to photometer and the quality was confirmed by Agilent 2100 correlate with invasive property and advanced tumor Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). stage (Werner et al., 2002). It has also been speculated Total RNA (5 mg per sample) was transcribed into cDNA that MMP-7 expressed on the surface of carcinoma cells using an oligodT primer harboring the T7 promoter sequence. may cleave E-cadherin, an inherent protein of adherent cDNA was copied into double-stranded DNA from which RNA was transcribed in vitro by T7 RNA polymerase (Amino junctions of epithelial cells, and facilitate the detach- Allyl MessageAmp aRNA kit, Ambion, Foster City, CA, ment of carcinoma cells from the site of primary tumor USA). Twenty micrograms of purified antisense RNA was (Shiomi and Okada, 2003). Somewhat surprisingly we labeled with monoreactive Cy3 label or Cy5 label (Amersham observed downregulation of MMP-7 and -16 transcripts Biosciences, Buckinghamshire, UK) according to the manu- in qRT-PCR but no change at protein level. We also facturer’s instructions (Ambion). Labeled antisense RNA was found an increase in the expression of fibronectin in purified and dried in a vacuum along with blocking agents qRT-PCR, which has been shown to be a substrate for (human CoT DNA (20 mg; Invitrogen, Carlsbad, CA, USA), MMP-7, -12 and -16 (Chakraborti et al., 2003). poly-(dA) (20 mg; Amersham) and tRNA (8 mg; Sigma, Interestingly, we found upregulation of PKC-d ex- St Louis, MO, USA). This probe complex was then dissolved pression due to E5. Earlier reports as well as our own in 80 ml of hybridization buffer (50% formamide, 6 Â SSC, 0.5% sodium dodecyl sulfate, 5 Â Denhardt’s solution). The observations suggest a role for the E5 protein in several mixture was dissolved and incubated at 95 C for 5 min. It was cellular pathways involving PKC (Ueda et al., 1996; then cooled down to 4 C and transferred to ice. Brodie and Blumberg, 2003), and PKC may thus mediate The cDNA microarrays (Finnish DNA-Microarray Centre, several biological activities of E5. Intriguingly, PKC is Turku, Finland) were prehybridized at 65 C in prewarmed 6 Â known to phosphorylate connexins and downregulate SSC, 0.5% sodium dodecyl sulfate and 1% bovine serum

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2539 albumin, for 60 min. After prehybridization, the slides were expression was confirmed in all cDNAs from HaCaT-E5 cells washed with fresh water at room temperature, with gentle by qRT-PCR. The primer set used to amplify E5 mRNA was agitation. The slides were then dried by brief centrifugation at 50-ATTACTGGCGTGCTTTTTGCT-30 (sense) and 50-CAAAT 2000 r.p.m. A lifter slip (Erie Scientific Company, Portsmouth, ATAATATATACAATAAAACACCTAAACG-30 (antisense). NH, USA) was placed onto the spots and the sample was added onto the slides. Western blotting The slides were then placed in a humid hybridization chamber (GeneMachines; Genomic Solutions, Ann Arbor, Dexamethasone-treated E5 and control cells were lysed in lysis M M N N0 MI, USA), submerged in water and hybridized for 16 h at 42 C. buffer (150 m NaCl; 50 m -2-hydroxyethylpiperazine- - 2-ethanesulfonic acid, pH 7.4; 0.1% Igepal; 5 mM EDTA) After incubation, the slides were washed with 0.1 Â SSC, 0.1% sodium dodecyl sulfate and water and dried by centrifugation containing proteinase inhibitors (complete proteinase inhibi- (2000 r.p.m.). The slides were scanned with GenePix 4200AL tors; Roche Applied Science). Total protein was quantitated scanner (Molecular Devices Corporation, Sunnyvale, CA, using the Bio-Rad protein assay (Bio-Rad Laboratories, USA) using appropriate setting for Cy3- and Cy5-labeled Hercules, CA, USA). Forty micrograms of total protein was samples. All microarray experiments were performed in separated in 7.5 or 10% sodium dodecyl sulfate-polyacryla- triplicates. mide gel electrophoresis and transferred to Protran nitrocellu- lose membranes (Schleicher & Schuell, Dassel, Germany). One membrane was prepared for each protein studied. The cDNA microarray analysis membranes were blocked with 5% nonfat milk in phosphate- Segmentation, computation of the absolute foreground in- buffered saline, followed by incubation with primary anti- tensities and local background estimation from the scanned bodies in 5% nonfat milk in PBST overnight at 4 1C. The images were performed with the software Axon GenePix Pro v. membranes for PI3KR3 blot were blocked with 2% gelatine, 6.0 (http://www.moleculardevices.com/pages/software/gn_ge- 0.1% normal goat serum (Vector Laboratories Inc., Burlin- nepix_pro.html). game, CA, USA) and 0.1% normal rabbit serum in phosphate- The raw data were then imported into the R package (http:// buffered saline to reduce background signal. The antibodies www.r-project.org) BioConductor (Gentleman et al., 2004; were to PI3KR3 (Abgent, San Diego, CA, USA), Akt (Cell http://www.bioconductor.org). Background correction and Signaling Technology, Beverly, MA, USA), P-Akt (Ser473; quality assessment of the data set were carried out using the Cell Signaling Technology), paxillin (Cell Signaling Technol- BioConductor packages limma (Smyth, 2004) and arrayMagic ogy), P-paxillin (Tyr118; Cell Signaling Technology), PKC-d (Buness et al., 2005). The background-corrected log ratios (BD Biosciences) and lamin A/C (BD Biosciences). Subse- were then normalized using LOWESS regression method. quently, the membrane was incubated with fluorescent Paired permutation-corrected t-test (BioConductor package secondary antibodies: IRDye 800CW goat anti-mouse (LI- siggenes) was carried out to find differentially expressed genes COR Biosciences, Lincoln, NE, USA) and IRDye 680 goat in the HaCaT-E5 versus -pMSG comparisons at 0 and 24 h anti-rabbit (LI-COR Biosciences). Images were acquired with separately. Median false discovery rate less than 1 was chosen the Odyssey infrared imaging system (LI-COR Biosciences) as significance cutoff (Tusher et al., 2001). and analysed by the software program provided by the Hierarchical clustering of samples on the normalized data manufacturer. The program measures the amount of light was performed to evaluate the normalization accuracy. produced. For lamin A/C (Figure 2a), both bands were Functional clustering of the genes presenting significant quantitated separately and the results were then combined. variations to the statistical tests was performed according to Two-color western detection was performed by blotting each the Gene Ontology annotation, BP database (Ashburner et al., membrane simultaneously for the protein of interest and for b- 2000). Fisher’s exact test was performed to evaluate the actin (Sigma-Aldrich) to control for loading. Quantitation overrepresentation of functional ontologies, using P-value results are expressed as fold changes in E5 cells as compared to cutoff o0.05. This analysis was carried out using the pMSG cells (Figure 2). First, a coefficient was calculated for BioConductor package GOstats (Gentleman et al., 2004). each time point in each blot separately by comparing the b- actin signal from E5 cells to the b-actin signal from pMSG cells. This gave us a measure of loaded protein amounts in Quantitative real-time RT-PCR each lane. This coefficient was then used to normalize the Total cellular RNA was extracted from confluent cell cultures signal for the proteins of interest against the protein amount using the TriPure reagent (Roche Applied Science, Indianapolis, loaded. Next, a value of 1 was set for the signal of each protein IN, USA). QRT-PCR was performed by Taqman technology of interest in pMSG cells, and a fold change was then and a sequence detector in ABI PRISM 7700 (Applied calculated by comparing the expression of the proteins of Biosystems, Foster City, CA, USA) using SYBR Green PCR interest in E5 cells to pMSG cells at each time point. Master Mix and RT-PCR kit (Applied Biosystems). The cDNA was synthesized from 2 mg of total RNA with reverse transcriptase using the oligo d(T)16 primer. From the reaction Cell motility assay mixture, 15 ng cDNA was then subjected to PCR amplification Cell motility was assessed using wound-healing assay and live- with specific forward and reverse primers (Supplementary Table cell imaging. HaCaT-E5 and -pMSG cells were grown as a 2). The specificity and purity of the amplification reaction were confluent monolayer in Nunclon 24-well plates (Thermo also determined by performing a melting curve analysis. The Fisher Scientific, Waltham, MA, USA), serum starved for expression level of b-actin was used to normalize the abundance 24 h and induced with dexamethasone for another 24 h. of the tested transcripts. The comparative threshold cycle (CT) Monolayers were then wounded using a disposable pipette method was used to determine fold changes of transcript present tip. Wound closure was followed for 48 h using the Cell-IQ in E5-expressing cells as compared to control. All qRT-PCRs real-time image capture system (Chip-Man Technologies Ltd, were performed in three independent experiments. Moreover, Tampere, Finland). To study whether the wound closure was Student’s t-test was employed to statistically evaluate the due to cell division or cell migration, cells were counted and results, using P-value cutoff o0.05. The primers were designed analysis of variance test was performed to find overall changes. using the PrimerExpress program (Applied Biosystems). E5 Tukey honest test was performed to find differences while

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2540 comparing cells. Each experiment was performed in altogether assistance and Anu Planken for assistance in PCR technology. eight replicates. Light Microscopy Unit (Institute of Biotechnology, University of Helsinki, Finland) is acknowledged for their support in cell Acknowledgements motility assays. N Kivi was supported by grants from the University of Helsinki Research Foundation, the Oskar We thank Professor Angel Alonso (DKFZ, Heidelberg, O¨ flund Foundation, the Biomedicum Helsinki Foundation, Germany) for providing the HaCaT-E5 and -pMSG cell lines. Fund of Laboratory Medicine Research and the K Albin We acknowledge Rita Fingerroos for excellent technical Johansson Foundation.

References

Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM Dyson N, Howley PM, Mu¨ nger K, Harlow E. (1989). The human et al. (2000). Gene Ontology: tool for the unification of biology. The papillomavirus-16 E7 oncoprotein is able to bind to the retino- Gene Ontology Consortium. Nat Genet 25: 25–29. blastoma gene product. Science 243: 934–937. Auvinen E, Alonso A, Auvinen P. (2004). Human papillomavirus type Fehrmann F, Klumpp DJ, Laimins LA. (2003). Human papilloma- 16 E5 protein colocalizes with the antiapoptotic Bcl-2 protein. Arch virus type 31 E5 protein supports cell cycle progression and Virol 149: 1745–1759. activates late viral functions upon epithelial differentiation. J Virol Band V, Zaychowski D, Kulesa V, Sager R. (1990). Human 77: 2819–2831. papillomavirus DNAs immortalize normal human mammary Genther SM, Sterling S, Duensing S, Mu¨ nger K, Sattler C, Lambert PF. epithelial cells and reduce their growth factor requirements. Proc (2003). Quantitative role of the human papillomavirus type 16 E5 Natl Acad Sci USA 87: 463–467. gene during the productive stage of the viral life cycle. JVirol77: Bouvard V, Matlashewski G, Gu ZM, Storey A, Banks L. (1994). The 2832–2842. human papillomavirus type 16 E5 gene cooperates with the E7 gene Genther Williams SM, Disbrow GL, Schlegel R, Lee D, Threadgill DW, to stimulate proliferation of primary cells and increases viral gene Lambert PF. (2005). Requirement of epidermal growth factor receptor expression. Virology 203: 73–80. for hyperplasia induced by E5, a high-risk human papillomavirus Bravo IG, Crusius K, Alonso A. (2005). The E5 protein of the human oncogene. Cancer Res 65: 6534–6542. papillomavirus type 16 modulates composition and dynamics of Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S membrane lipids in keratinocytes. Arch Virol 150: 231–246. et al. (2004). Bioconductor: open software development for Brodie C, Blumberg PM. (2003). Regulation of cell apoptosis by computational biology and bioinformatics. Genome Biol 5: e-pub protein kinase C d. Apoptosis 8: 19–27. ahead of print 15 September 2004. Buness A, Huber W, Steiner K, Su¨ ltmann H, Poustka A. (2005). Halbert CL, Demers GW, Galloway DA. (1991). The E7 gene of arrayMagic: two-colour cDNA microarray quality control and human papillomavirus type 16 is sufficient for immortalization of preprocessing. Bioinformatics 21: 554–556. human epithelial cells. J Virol 65: 473–478. Burgert HG, Ruzsics Z, Obermeier S, Hilgendorf A, Windheim M, Hawley-Nelson P, Vousden KH, Hubbert NL, Lowy DR, Schiller JT. Elsing A. (2002). Subversion of host defense mechanisms by (1989). HPV16 E6 and E7 proteins cooperate to immortalize human adenoviruses. Curr Top Microbiol Immunol 269: 273–318. foreskin keratinocytes. EMBO J 8: 3905–3910. Cantley LC. (2002). The phosphoinositide 3-kinase pathway. Science Hojilla CV, Mohammed FF, Khokha R. (2003). Matrix metallopro- 296: 1655–1657. teinases and their tissue inhibitors direct cell fate during cancer Castronovo V. (1993). Laminin receptors and laminin binding development. Br J Cancer 89: 1817–1821. proteins during tumor invasion and metastasis. Invasion Metastasis Ivetac I, Munday AD, Kisseleva MV, Zhang XM, Luff S, 13: 1–30. Tiganis T et al. (2005). The type I alpha inositol polyphosphate Chakraborti S, Mandal M, Das S, Mandal A, Chakraborti T. (2003). 4-phosphatase generates and terminates phosphoinositide 3-kinase Regulation of matrix metalloproteinases: an overview. Mol Cell signals on endosomes and the plasma membrane. Mol Biol Cell 16: Biochem 253: 269–285. 2218–2233. Chang F, Lee JT, Navalonic PM, Steelman LS, Shelton JG, Blalock Kim SH, Juhnn YS, Kang S, Park SW, Sung MW, Bang YJ et al. WL et al. (2003). Involvement of PI3K/Akt pathway in cell cycle (2006). Human papillomavirus 16 E5 up-regulates the expression of progression, apoptosis, and neoplastic transformation: a target for vascular endothelial growth factor through the activation of cancer chemotherapy. Leukemia 17: 590–603. epidermal growth factor receptor, MEK/ERK1,2 and PI3K/Akt. Chen SL, Huang CH, Tsai TC, Lu KY, Tsao YP. (1996). The Cell Mol Life Sci 63: 930–938. regulation mechanism of c-jun and junB by human papillomavirus Lampe PD, TenBroek EM, Burt JM, Kurata WE, Johnson RG, type 16 E5 protein. Arch Virol 141: 791–800. Lau AF. (2000). Phosphorylation of connexin43 on serine368 by Chuaqui RF, Bonner RF, Best CJ, Gillespie JW, Flaig MJ, Hewitt SM protein kinase C regulates gap junctional communication. J Cell et al. (2002). Post-analysis follow-up and validation of microarray Biol 7: 1503–1512. experiments. Nat Genet 32(Suppl): 509–514. Leechanachai P, Banks L, Moreau F, Matlashewski G. (1992). The E5 Conrad M, Bubb VJ, Schlegel R. (1993). The human papillomavirus gene from human papillomavirus type 16 is an oncogene which type 6 and 16 E5 proteins are membrane-associated proteins which enhances growth factor-mediated signal transduction to the nucleus. associate with the 16-kilodalton pore-forming protein. J Virol 67: Oncogene 7: 19–25. 6170–6178. Leykauf K, Salek M, Sclu¨ ter H, Lehmann W, Alonso A. (2004). Czech MP. (2003). Dynamics of phosphoinositides in membrane Identification of membrane proteins differentially expressed in retrieval and insertion. Annu Rev Physiol 65: 791–815. human papillomavirus type 16 E5-transfected human keratinocytes Demeter LM, Stoler MH, Sobel ME, Broker TR, Chow LT. (1992). by nanoelectrospray ionization mass spectrometry. J Gen Virol 85: Expression of high-affinity laminin receptor mRNA correlates with 1427–1431. cell proliferation rather than invasion in human papillomavirus- McCormack SJ, Brazinski SE, Moore Jr JL, Werness BA, Oldstein DJ. associated cervical neoplasms. Cancer Res 52: 1561–1567. (1997). Activation of the focal adhesion kinase signal transduction DiMaio D, Lai CC, Klein O. (1998). Virocrine transformation: the pathway in cervical carcinoma cell lines and human genital epithelial intersection between viral transforming proteins and cellular signal cells immortalized with human papillomavirus type 18. Oncogene 15: transduction pathways. Annu Rev Microbiol 52: 397–421. 265–274. Duerst RJ, Morrison LA. (2003). Innate immunity to herpes simplex Munger K, Howley PM. (2002). Human papillomavirus immortalization virus type 2. Viral Immunol 16: 475–490. and transformation functions. Virus Res 89: 213–228.

Oncogene HPV 16 E5 alters cellular gene expression N Kivi et al 2541 Munger K, Phelps WC, Bubb V, Howley PM, Schlegel R. (1989). The Sto¨ ppler MC, Straight SW, Tsao G, Schlegel R, McCance DJ. (1996). E6 and E7 genes of the human papillomavirus type 16 together are The E5 gene of HPV-16 enhances keratinocyte immortalization by necessary and sufficient for transformation of primary human full-length DNA. Virology 223: 251–254. keratinocytes. J Virol 63: 4417–4421. Straight SW, Hinkle PM, Jewers RJ, McCance DJ. (1993). Mun˜ oz N, Bosch FX, Castellsague X, Diaz M, de Sanjose S, The E5 oncoprotein of human papillomavirus type 16 trans- Hammouda D et al. (2004). Against which human papillomavirus forms fibroblasts and effects the down-regulation of the types shall we vaccinate and screen? The international perspective. epidermal growth factor receptor in keratinocytes. J Virol 67: Int J Cancer 111: 278–285. 4521–4532. O’Brien V. (1998). Viruses and apoptosis. J Gen Virol 79: 1833–1845. Thomsen P, Rudenko O, Berezin V, Norrild B. (1999). The HPV-16 E5 Oelze I, Kartenbeck J, Crusius K, Alonso A. (1995). Human oncogene and bafilomycin A1 influence cell motility. Biochim papillomavirus type 16 E5 protein affects cell–cell communication Biophys Acta 1452: 285–295. in an epithelial cell line. J Virol 69: 4489–4494. Tusher VG, Tibshirani R, Chu G. (2001). Significance analysis of Pim D, Collins M, Banks L. (1992). Human papillomavirus type 16 E5 microarrays applied to the ionizing radiation response. [Erratum in: gene stimulates the transforming activity of the epidermal growth Proc Natl Acad Sci USA 98: 10515] Proc Natl Acad Sci USA 98: factor receptor. Oncogene 7: 27–32. 5116–5121 . Pisani P, Bray F, Parkin DM. (2002). Estimates of the world-wide Ueda Y, Hirai S, Osada S, Suzuki A, Mizuno K, Ohno S. (1996). prevalence of cancer for 25 sites in the adult population. Int J Cancer Protein kinase C activates the MEK–ERK pathway in a manner 97: 72–81. independent of Ras and dependent on Raf. J Biol Chem 271: Rajeevan MS, Vernon SD, Taysavang N, Unger ER. (2001). 23512–23519. Validation of array-based gene expression profiles by real-time Valle GF, Banks L. (1995). The human papillomavirus (HPV)-6 and (kinetic) RT-PCR. J Mol Diagnosis 3: 26–31. HPV-16 E5 proteins co-operate with HPV-16 E7 in the transforma- Sayama K, Yamasaki K, Hanakawa Y, Shirakata Y, Tokumaru S, tion of primary rodent cells. J Gen Virol 76: 1239–1245. Ijuin T et al. (2002). Phosphatidylinositol 3-kinase is a key regulator Walboomers JM, Jacobs MV, Manos MM, Bosch FX, Kummer JA, of early phase differentiation in keratinocytes. J Biol Chem 277: Shah KV et al. (1999). Human papillomavirus is a necessary cause of 40390–40396. invasive cervical cancer worldwide. J Pathol 189: 12–19. Schreiber K, Cannon RE, Karrison T, Beck-Engeser G, Huo D, Werner JA, Rathcke IO, Mandic R. (2002). The role of matrix Tennant RW et al. (2004). Strong synergy between mutant ras and metalloproteinases in squamous cell carcinomas of the head and HPV16 E6/E7 in the development of primary tumors. Oncogene 23: neck. Clin Exp Metastasis 4: 275–282. 3972–3979. Werness BA, Levine AJ, Howley PM. (1990). Association of human Shiomi T, Okada Y. (2003). MT1-MMP and MMP-7 in invasion and papillomavirus types 16 and 18 E6 proteins with p53. Science 248: metastasis of human cancers. Cancer Metastasis Rev 22: 145–152. 76–79. Smyth GK. (2004). Linear models and empirical Bayes for assessing Zhang B, Spandau DF, Roman A. (2002). E5 protein of human differential expression in microarray experiments. Stat Appl Genet papillomavirus type 16 protects human foreskin keratinocytes from Mol Biol 3: Article 3. UVB-irradiation-induced apoptosis. J Virol 76: 220–231.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene