Two conserved amino acids differentiate the biology of high‐risk and low‐risk HPV E5 proteins

Abstract The high‐risk alpha human papillomaviruses (HPVs) are responsible for 99% of cervical cancers. While the biological functions of the HPV E6 and E7 oncoproteins are well‐characterized, the function of E5 has remained elusive. Here, we examined gene expression changes induced by E5 proteins from high‐risk HPV‐16 and low‐risk HPV‐6b in multiple pools of primary human keratinocytes. Surprisingly, microarray analysis revealed that over 700 genes were significantly regulated by HPV‐6b E5, while only 25 genes were consistently and significantly regulated by HPV‐16 E5 in three biological replicates. However, we observed that more than thousand genes were altered in individual sample compared with vector. The gene expression profile induced by 16E5 in primary genital keratinocytes was very different from what has been previously published using immortalized HaCaT cells. Genes altered by HPV‐16 E5 were unaffected by HPV‐6b E5. Our data demonstrate that E5 proteins from the high‐ and low‐risk HPVs have different functions in the HPV‐host cell. Interestingly, conversion of two amino acids in HPV‐16 E5 to the low‐risk HPV‐6b sequence eliminated the induction of high‐risk related cellular genes.

The correlation of phylogeny with cancer risk suggested that HPV-16 E5 might also contribute to tumorigenesis. 9,11,21,22 The main oncoproteins of HPV-16 are E6 and E7, which are both necessary and sufficient for cell immortalization. E5 is neither necessary nor sufficient for immortalization. Besides E2, E5 is one of the other proteins that is assumed to be disrupted during viral integration. 23,24 Estimations for the percentage of HPV-induced cervical cancers that have integrated DNAand therefore potentially no E5-varies greatly, from 15% to 86%. [25][26][27] One study estimated that only 60% of HPV-16 induced cervical cancers might express E5. 4,28 For several reasons, E5 is considered the third oncoprotein of HPV. First of all, the lack of E5 at later stages of malignant transformation does not mean that early E5 expression is not essential in establishing a successful and persistent infection (the precursor to dysplasia and cancer). It has been suggested that E5 helps to expand the initial population of HPV-infected basal cells, perhaps by enhancing EGFR activation. 4,10,12,17,29,30 Second, while E5 is present in all high-risk viruses, many low-risk types either lack an E5 ORF altogether or lack a translation start codon. 9,21 Finally, E5 is able to enhance the transformation of cells by E6 and E7 in-vivo. For example, it was shown that estrogen-treated transgenic mice expressing HPV-16 E5 in addition to E6 and E7 developed a larger number of tumors than mice expressing E6 and E7 alone. 11,31 How E5 actually causes these observed phenotypes is still under debate, although there are several possibilities, including EGFR activation, activation of c-jun and c-fos, binding of v-ATPase, disruption of gap junctions, immune evasion, formation of koilocytes, and binding of nuclear transport proteins. 4,[32][33][34] We previously reported that 16E5, as well as HPV-6b E5 (6bE5), induce koilocytosis in collaboration with E6. 33 The mechanism behind these 16E5-induced phenotypes is unknown. However, the ability of 16E5 to bind several cellular proteins, including the 16-kDa subunit of the vacuolar H + -ATPase, [35][36][37] BAP31, 38 HLA, [39][40][41][42][43] ErbB4, 44 calnexin, 43 and karyopherin β3 32 might account for some of its biological activities. Little is known about the biologic functions and cellular partners of E5 proteins of low-risk HPVs. Here, we examined gene expression changes induced by E5 proteins from high-risk HPV-16 and low-risk HPV-6b in multiple pools of primary human keratinocytes. Our microarray analysis revealed that over 700 genes were significantly regulated by HPV-6b E5, while only 25 genes were consistently and significantly regulated by HPV-16 E5 in three biological replicates. Genes altered by HPV-16 E5 were unaffected by HPV-6b E5. Conversion of two amino acids in HPV-16 E5 to the low-risk HPV-6b sequence eliminated the induction of high-risk related cellular genes. Our data demonstrate that E5 proteins from the high-and low-risk HPVs have different functions in the HPV-host cell. To transduce cells, 1.5 ml retroviral stock supplemented with 1.5 μl polybrene was added to cells in T75 flasks at 40%-60% confluency. Cells were incubated with the retrovirus on a gentle rocker at 37°C. After 2 h, the retrovirus was removed and replaced with media appropriate to the cell type. Cells were allowed to grow to approximately 80%, which occurred within 1-3 days. For cell selection, geneticin (G418) (Invitrogen) at a concentration of 75-100 μg/ml was used and selection was maintained until all the cells in the control (uninfected) flask died. Then, nuclei were stained with 0.5 mg/ml Hoeschst stain for 3 min at room temperature. Coverslips were then removed and inverted over slides with 30 μl mounting media (Invitrogen) and allowed to rest at room temperature for several hours until the mounting media hardened. Slides were stored at 4°C overnight and viewed the next day using a Zeiss Axioskop microscope (Carl Zeiss, Inc.). Cells were imaged using a 63X objective, Hammamutsu CCD camera, and Openlab 3.0.7 software.

| Cell lysis and protein concentration
For direct western, whole-cell lysates were made by plating cells on 100 mm dishes (BD Falcon) and allowing them to grow to 80% confluence. Plates were washed with cold PBS, and cells scraped in 300 μl of two times Laemlli buffer. Lysates were kept on ice, then boiled for 10' at 110°C, allowed to cool for 2 min, and frozen on dry ice. Before protein assay, lysates were thawed in a 37°C water bath.
Before loading, up to 45 μl of sample (40-60 μg protein) was mixed with a volume of β-mercaptoethanol (Sigma-Aldrich) equal to 10% of the final loading volume. For immunoprecipitation, cells were scraped instead with 1.2 ml radioimmunoprecipitation assay (RIPA) buffer with 12 μl protease inhibitor cocktail set 1 (Calbiochem, 100X stock) and frozen on dry ice. Before protein assay, lysates were thawed in a 37°C water bath, DNA was sheared with a 23 G needle, and lysates were spun down at 2 K rpm. Protein concentration for both lysates was determined using the BioRad D c Protein Assay (Bio-Rad Laboratories) per the manufacturer's protocol.

| Immunoprecipitation
Equal amounts of protein (up to 600 μg) per sample were added to 40 μl Protein A Plus beads (Pierce). After washing with 1 ml PBS, beads were rotated for 90' end-to-end with antibody. After being spun down for 1' at 2k rpm, beads were washed with 1 ml cold RIPA buffer with protease inhibitors, followed by an additional 5' rotation and 1' centrifugation. This was repeated two times more, followed by three consecutive washes with PBS (no rotation). Beads were pelleted and then resuspended in 47 μl two times Laemmli with 10% βme. No βme was added if reducing conditions were not to be used (as for E5 dimerization studies). After 20 min in a 37°C water bath, beads were boiled for 6 min at 110°C before being frozen on dry ice.
Before gel loading, samples were thawed in a 37°C water bath.

| Western blot
Samples were electrophoretically separated on Tris-Glycine gels (Invitrogen) and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). Membrane was blocked for 30' in either PBS with 5% nonfat dry milk or in wash buffer with 2% bovine serum albumin, depending on the antibody. Primary antibody anti-AU1 was left overnight at 4°C on a rocker. ß-actin (Sigma-Aldrich) at a final dilution of 1:10 000, served as the loading control. Membrans was washed two times for 15' with either PBS + .05% Tween or wash buffer (Fisher Scientific). Membranes were then probed with a secondary antibody, anti-mouse IgG. (A) Less than 25 genes were found to be consistently changed (>1.5 fold in each array, p < 0.01) in 16E5-expressing cells as compared to LXSN-expressing cells.

| RESULTS AND DISCUSSION
To define the biological activities of E5 proteins, we analyzed 16E5 expression in these cells induces high levels of apoptosis, requiring the use of an inducible promoter. Consequently, analysis is temporally limited following the induction of 16E5 and is potentially confounded by the apoptotic and genetic changes in these cells.
Rather than using immortalized cells as a target, we chose to use primary genital keratinocytes to more closely mimic the effect of high-risk HPV-16 E5 and low-risk HPV6B E5 on cellular gene expression in vivo. To ensure that these changes were reproducible and physiologically relevant, we performed the microarray assays in triplicate. We first verified that all E5s expressed at similar levels using RT-PCR ( Figure 1A) and IP/WB ( Figure 1B). Surprisingly, we found that 16E5 consistently regulated fewer than 25 genes across all arrays conducted (>1.5 fold change in each array, p-value < 0.01) (Table 1A), even though individual array had more than thousand genes altered by 16E5 compared with LXSN (Supporting Information Table). Interestingly, we also found that all of these consistently regulated genes were downregulated. These genes were functionally grouped using the Gene Ontology Biological Process (BP) database 45 as shown in  were also submitted for functional grouping according to the Gene ontology BP database (Table 2C). We noted interesting numbers of genes overlapped in arrays, overlapped gens from 16E5 and 6bE5 arrays were 47, this was similar between 16E5 and 16HA (49 genes), while we noticed 237 genes were overlapped in 6bE5 and 16HA.
These data further demonstrated that two amino acids in 16E5 contributed to different biological functions of HR and LR HPV E5 proteins.
We next attempted to define the protein domain that might account for the biological differences between the low-and highrisk E5 proteins. An alignment of the E5 amino acid sequences from several low and high-risk HPV types was performed Immunofluorescence was used to confirm the expression and localization pattern of the mutant construct, which merged with the ER-marker calnexin in stably-expressing primary human cells and transfected COS-1 cells ( Figure 2B). 46 Our previous study found that this 16HA was unable to repress COX-1 mRNA and XBP-1 splicing in primary keratinocytes. 46  Real-time RT-PCR was used to confirm the downregulation of four genes affected by 16E5 ( Figure 3). All four genes were downregulated; however, only two were statistically significant. Interestingly, three of these genes were not altered by either the 6bE5 or the 16HA mutant, suggesting that the ability of the high-risk E5 protein to downregulate these genes may be dependent upon two highly-conserved C-terminal amino acids. In addition, five genes affected by 6bE5 in the microarray were selected and confirmed by real-time PCR (Figure 4). However, similar to the wild-type 16E5, the 16HA mutant failed to induce gene expression changes in a similar manner to 6bE5 (Figure 4). This suggests that the two C-terminal amino acids conserved in low-risk HPVs are not sufficient to confer properties of the low-risk 6bE5 when introduced in isolation into a 16E5 sequence.
In brief, our results with primary keratinocytes differ very significantly from those obtained in the previously published HaCat cell study, 19