A cyclin-binding motif in human papillomavirus type 18 (HPV18) E1^E4 is necessary for association with CDK-cyclin complexes and G2/M cell cycle arrest of keratinocytes, but is not required for differentiation-dependent viral genome amplification or L1 capsid protein expression

Abstract

The G2/M arrest function of human papillomavirus (HPV) E4 proteins is hypothesized to be necessary for viral genome amplification. Full-length HPV18 E1^E4 protein is essential for efficient viral genome amplification. Here we identify key determinants within a CDK-bipartite consensus recognition motif in HPV18 E1^E4 that are critical for association with active CDK-cyclin complexes and in vitro phosphorylation at the predicted CDK phosphorylation site (threonine 23). The optimal cyclin-binding sequence ((43)RRLL(46)) within this E4 motif is required for G2/M arrest of primary keratinocytes and correlates with cytoplasmic retention of cyclin B1, but not cyclin A. Disruption of this motif in the E4 ORF of HPV18 genomes, and the subsequent generation of stable cell lines in primary keratinocytes revealed that this motif was not essential for viral genome amplification or L1 capsid protein induction. We conclude that the HPV18 E4 G2/M arrest function does not play a role in early vegetative events.

Figure 1. HPV18 E1^E4 colocalizes with CDK-cyclins in epithelial cells

Confocal microscopy of COS-1 cells transfected with the plasmid expressing the codon-optimized HPV18 E1^E4 protein showed colocalization between HPV18 E1^E4 and keratin 18 (a), cyclin A (c), cyclin B1 (d), and CDK1 (e), but not cyclin E (b). Green, E4: Red, keratin 18, cyclin proteins and CDK1: Blue, DAPI-stained nuclei, and yellow indicates colocalization. Bar, 10 μm.

Figure 2. E4 sequence contains a conserved bipartite CDK consensus recognition motif

(A) Homology between a bipartite CDK consensus recognition motif and E4 coding sequences of HPV types of the alpha genus. Six examples are given, but other types within the genus also contain the motif within the E4 ORF. They are: HPV61, 72, 81, 83 (species 3); HPV18, 39, 45, 59, 68 (species 7); HPV43 (species 8); HPV16, 31 (species 9); HPV6, 11, 13 (species 10); HPV54 (species 13). The potential CDK phosphorylation sites (S/TPX[K/R]) and cyclin binding motifs (RXL) in E4 are identified by a grey-shaded box. A threonine dipeptide in HPV16 E1^E4 that is a key determinant in cytoplasmic retention of CDK-cyclin complexes is underlined (Davy et al., 2005). (B) Details of a deletion and substitutions made within the bipartite CDK consensus recognition sequence of HPV18 E1^E4; the substitutions are underlined.

Figure 3. Disruption of RXL tripeptides in HPV18 E1^E4 abrogates cytoplasmic retention of cyclins A and B1

(A) Immunofluorescence microscopy of COS-1 cells transfected with empty vector (control) or plasmids expressing HPV18 E1^E4 wildtype or the various mutants. Cells were co-stained with antibodies to E4 (green) and cyclin A (red), and nuclei counterstained with DAPI (blue). Cells showing sequestration of cyclin A to E4 inclusions is only evident in cells expressing the wild-type or deletion mutant (E4Δ21-23) E1^E4 proteins with examples identified with arrowheads. In a small sub-population of cells expressing the RXL2 mutation cyclin A was found to be associated with the collapsed E4-kerat in structures (arrowhead). Bar, 30 μm. (B) Staining of transfected COS cells for E4 (green) and cyclin B1 (red), and nuclei counter stained with DAPI (blue). In cells transfected with the plasmid expressing RXL2, rare cells are observed that show a degree of E4 and cyclin B1 co-staining (indicated by an arrow). Bar, 20 μm. The percentages of dual-positive cells (co-expressing E4 and cyclins A or B1) showing colocalization are given as means ± standard deviation from data derived from counting a minimum of 100 cells from three independent experiments.

Figure 4. HPV18 E1^E4 association with active CDK-cyclin complexes is dependent on RXL motifs

(A) The presence of cyclin A, B1 and E proteins in co-precipitates formed in a HeLa cell lysate using GST-fusion proteins of wild-type (WT) and mutant HPV18 E1^E4 proteins (immunoblot). Two forms of cyclin B1, indicated with arrows, are detected in the cell lysates, the slower migrating form being a phosphorylated form of the cyclin. Two forms of cyclin E (identified by arrowheads) were recognised by the antibody, the upper represents the mature form of cyclin E, ~ 47 kDa. Ponceau stain identifies loading of all of the GST-E4 proteins, with GST being in excess. (B) Phosphorylation of histone H1 protein by GST and GST-HPV18 E1^E4 co-precipitates prepared from HeLa cell lysates in the presence and absence of the CDK-inhibitor roscovitine (autoradiograph). The level of GST and histone H1 proteins is visualised by a Coomassie stain. (C) Phosphorylation of histone H1 by GST, wild-type (WT) and mutant GST-18E4 coprecipitates prepared from HeLa cell lysates. The Coomassie-stained gel shows loadings of the fusion proteins.

Figure 5. CDKS phosphorylate HPV18 E1^E4 on threonine 23 in vitro.

(A) In vitro p42 MAP kinase phosphorylation assay of GST, GST-HPV16 and GSTHPV18 E1^E4. (B) In vitro phosphorylation of wild-type (WT) and mutant GST-18 E4 proteins by recombinant CDK2-cyclin A. (C) GST-HPV18 E4 containing substitutions at serine 22 (S22N) and threonine 23 (T23I) were used in an in vitro phosphorylation assay containing CDK2-cyclin A (a) and CDK1-cyclin A (b). The level of GST and GST fusion proteins in the reactions is shown in the Coomassie-stained gels.

Figure 6. Integrity of RXL tripeptides is required for HPV18 E1^E4 induced G2/M arrest in primary foreskin keratinocytes

(A) Cell cycle profiles of primary human foreskin keratinocytes (HFK) infected with recombinant adenoviruses expressing β-galactosidase (βGal), HPV18 E1^E4 (WT) or the mutants T23I or RXLD for 36 h. The percentages of cells in G1 and in G2/M were used to calculate the G2+M:G1 ratios. Results are show for keratinocytes derived from two different donors and are representative of at least four independent experiments. (B) Western blot analysis of cell lysates showing the level of expression of the different proteins using β-actin levels as a loading control. (C) Cellular distribution of cyclin B1 (red) in HFK mock-infected (i), infected with the virus expressing (βGal) (ii), or with viruses expressing E4WT, E4T23I or E4RXLD (βGal and E4 proteins shown in green). Cyclin B1 is sequestered to both fully collapsed E4 filamentous networks (iii) as well as those that have undergone a partial collapse (iv). Mutation of threonine 23 (v) did not abrogate the association. Cyclin B1 was not sequestered to the collapsed filamentous networks of RXLD (vi and vii), although occasional cells showed a partial co-distribution (viii). Nuclear stain (DAPI) is shown in blue. (D) Cyclin A (red) distribtion in HFK, either mock-infected or infected with viruses expressing βGal, E4WT or E4RXLD proteins (all shown in green). Nuclear stain (DAPI) is shown in blue. Bar, 10 μm. The percentages of dual-positive cells is given as the means ± standard deviation, in the top-right of the panel s. Data was derived from three counts, each of a minimum of 100 cells.

Figure 6. Integrity of RXL tripeptides is required for HPV18 E1^E4 induced G2/M arrest in primary foreskin keratinocytes

(A) Cell cycle profiles of primary human foreskin keratinocytes (HFK) infected with recombinant adenoviruses expressing β-galactosidase (βGal), HPV18 E1^E4 (WT) or the mutants T23I or RXLD for 36 h. The percentages of cells in G1 and in G2/M were used to calculate the G2+M:G1 ratios. Results are show for keratinocytes derived from two different donors and are representative of at least four independent experiments. (B) Western blot analysis of cell lysates showing the level of expression of the different proteins using β-actin levels as a loading control. (C) Cellular distribution of cyclin B1 (red) in HFK mock-infected (i), infected with the virus expressing (βGal) (ii), or with viruses expressing E4WT, E4T23I or E4RXLD (βGal and E4 proteins shown in green). Cyclin B1 is sequestered to both fully collapsed E4 filamentous networks (iii) as well as those that have undergone a partial collapse (iv). Mutation of threonine 23 (v) did not abrogate the association. Cyclin B1 was not sequestered to the collapsed filamentous networks of RXLD (vi and vii), although occasional cells showed a partial co-distribution (viii). Nuclear stain (DAPI) is shown in blue. (D) Cyclin A (red) distribtion in HFK, either mock-infected or infected with viruses expressing βGal, E4WT or E4RXLD proteins (all shown in green). Nuclear stain (DAPI) is shown in blue. Bar, 10 μm. The percentages of dual-positive cells is given as the means ± standard deviation, in the top-right of the panel s. Data was derived from three counts, each of a minimum of 100 cells.

Figure 7. The CDK-bipartite recognition motif in E4 is not essential for establishment of the viral genome or differentiation-dependent viral genome amplification

(A) Southern blot analysis of HFK stably transfected with HPV18 wild-type (WT), RXLD and E4M17 mutant genomes. Equal amounts of total DNA was digested with DpnI to remove residual input bacterial DNA and either BglII (B), which does not have a site in the HPV18 genome or EcoR1 (E), which has a single recognition site in the HPV18 genome and produces linear forms of the viral genome. HPV18 episomes migrate as supercoils and open circles. Whilst in one donor (#1), the RXLD genomes established at a higher copy number than the wild type genomes this was not a consistent observation across all donors (e.g. 2). (B) Southern analysis of equal amounts of total genomic DNA isolated from keratinocyte cell lines containing wild-type (WT) or mutant (T23I, RXLD, E4M17) HPV18 genomes grown in monolayer cell culture (0 h) or suspended in methylcellulose (MC) for 24 and 48 h. DNA was digested with DpnI and BglII and HPV18 episomes migrate as supercoils and open circles. (C) Bargraph of phosphoimager analysis of Southern blots to show levels of HPV18 DNA genome amplification. Data were derived from at least fifteen (WT and RXLD) or six (E4M17 and T23I) experiments of each cell line in a total of three different donor keratinocytes. (D) Western blot analysis of the total protein extracts isolated from the different HPV18 genome-containing cell lines to determine expression levels of HPV18 E1^E4 protein. The induction of differentiation was monitored by determination of the expression levels of the epidermal differentiation-marker involucrin. Levels of β-actin were used as a loading control.

Figure 8. Disruption of the RXL tripeptides within E4 does not block expression of major capsid protein L1 in organotypic rafts

(A) Sections of organotypic rafts were stained with haematoxyl in and eosin (H and E) or immunohistochemistry performed for detection of E4 (green) and nuclei counterstained with DAPI (blue). (B) Sections stained for E4 (green), L1 (red) and counter stained with nuclear stain DAPI (blue). Areas of the rafts formed from cells containing HPV18 wild type genomes or the mutant RXLD showed expression of the capsid protein in E4-positive cells of the superficial cell layers. Rafts generated from donor keratinocytes (HFK) were negative for E4 and L1 expression. (C) Following co-staining of sections for E4 and cyclin B1, E4-positive cells containing cytoplasmic cyclin B1 are present in rafts generated from cells containing the wild-type or the RXLD genomes (arrowheads). Nuclear cyclin B1 is present in a few E4-positive cells in both wild-type and RXLD rafts (inset). Images of organotypic rafts are typical of staining observed between at least three organotypic rafts of each cell line.

Figure 8. Disruption of the RXL tripeptides within E4 does not block expression of major capsid protein L1 in organotypic rafts

(A) Sections of organotypic rafts were stained with haematoxyl in and eosin (H and E) or immunohistochemistry performed for detection of E4 (green) and nuclei counterstained with DAPI (blue). (B) Sections stained for E4 (green), L1 (red) and counter stained with nuclear stain DAPI (blue). Areas of the rafts formed from cells containing HPV18 wild type genomes or the mutant RXLD showed expression of the capsid protein in E4-positive cells of the superficial cell layers. Rafts generated from donor keratinocytes (HFK) were negative for E4 and L1 expression. (C) Following co-staining of sections for E4 and cyclin B1, E4-positive cells containing cytoplasmic cyclin B1 are present in rafts generated from cells containing the wild-type or the RXLD genomes (arrowheads). Nuclear cyclin B1 is present in a few E4-positive cells in both wild-type and RXLD rafts (inset). Images of organotypic rafts are typical of staining observed between at least three organotypic rafts of each cell line.

Figure 9. Homology between amino acid sequence of HPV E1 and E1^E4 proteins containing RXL motifs

Comparison of the amino acids sequences of E1 and E1^E4 proteins that contain an RXL tripeptides (Cy motif) and the cellular factor p21, are shown with sequence identity, or close conservation of amino acid type, highlighted by grey-shading. Numbers indicate amino acid positions.