ABSTRACT
Human papillomavirus type 16 (HPV-16) E6 (16E6) binds the E3 ubiquitin ligase E6AP and p53, thereby targeting degradation of p53 (M. Scheffner, B. A. Werness, J. M. Huibregtse, A. J. Levine, and P. M. Howley, Cell 63:1129–1136, 1990). Here we show that minimal 16E6-binding LXXLL peptides reshape 16E6 to confer p53 interaction and stabilize 16E6 in vivo but that degradation of p53 by 16E6 requires E6AP expression. These experiments establish a general mechanism for how papillomavirus E6 binding to LXXLL peptides reshapes E6 to then act as an adapter molecule.
TEXT
Papillomavirus E6 oncoproteins interact with target cellular proteins through docking on short peptides often containing the sequence LXXLL (4, 6, 26). The cancer-associated human papillomavirus type 16 (HPV-16) protein E6 (16E6) binds to an LXXLL motif (LQELL) on the cellular E3 ubiquitin ligase E6AP (10); 16E6-E6AP recruits and ubiquitinates p53. Neither 16E6 nor E6AP interacts alone with p53 (8–10, 17, 18) (20). A form of E6AP with a C833A mutation still binds E6 and p53 but fails to ubiquitinate or degrade p53 (19).
The necessity of E6AP for p53 degradation by 16E6 is controversial. While E6AP is required for in vitro degradation systems (9) and immortalized mouse fibroblasts (5), E6AP-null mice that express 16E6 in the skin do not accumulate p53 when irradiated, and mouse embryo kidney cells from those E6AP-null mice are reported to lose p53 upon overexpression of 16E6 (12, 22). Further, ubiquitin-independent degradation of p53 by E6 has been described (2). These disparate observations raise the questions of whether of E6AP is necessary for degradation and how the specificity of 16E6 for p53 is determined.
16E6 binding to the LQELL peptide alone recruits p53 to 16E6 in the absence of full-length E6AP.Our previous studies in yeast (Saccharomyces cerevisiae) had shown that p53 was very weakly recruited by LexA fusions to 16E6, but coexpression of LexA_16E6 with the E6AP C833A mutant (here termed E6AP-Ub−, for ubiquitin ligase negative) strongly recruited p53 to 16E6 (5). We reasoned that the binding of 16E6 solely to the minimal LQELL peptide of E6AP or similar LXXLL peptides might be sufficient to induce a p53-binding conformation of 16E6. To test this, we fused the 10-residue E6AP LQELL peptide between LexA and 16E6 to provide 16E6 with an LQELL docking site in cis; we similarly fused a mutated LXXLL peptide (LQEAS) between LexA and 16E6 (Fig. 1A). We tested interactions of these LexA fusions by yeast two-hybrid assays with a B42 transactivator-tagged LQELL peptide, B42_E6AP-Ub−, or with B42_E6AP-Ub− molecules where LQELL was progressively mutated from LQELL to LQELS or LQEAS. Fusion of the LQELL peptide in cis to 16E6 blocked interaction with B42_LQELL peptide in trans as expected, and mutation of LexA_LQELL_16E6 to LexA_LQEAS_16E6 restored interactions in trans (Fig. 1B, spots 4C and 4D). This is consistent with the LQELL peptide binding in cis to the 16E6 portion of the fusion protein and blocking trans interaction with the B42-fused peptide. Interaction of LQELL_16E6 with B42_E6AP-Ub− in trans was not blocked, indicating that the interaction of 16E6 with the full E6AP protein is more robust than the interaction with the free LQELL peptide but is still dependent upon an intact LQELL peptide on E6AP, as progressive mutation of the LQELL sequence in E6AP-Ub− to LQELS and finally LQEAS progressively diminished interactions (Fig. 1B). This correlates with the prior measurements of lower LQELL peptide interaction affinities with 16E6 in vitro (low micromolar [3]) compared with the affinities of 16E6 for the full-length E6AP in vitro (low nanomolar [21]).
Amphipathic peptides that interact with 16E6 confer 16E6 interaction with p53. (A) Yeast two-hybrid plasmids express the LexA DNA binding domain fused to 16E6 or 16E6 with the intervening LXXLL peptide of E6AP (406-ELTLQELLGEE-416) or the indicated mutant peptide and then a 6-amino-acid linker (GGSGGS). The 16E6 carboxy-terminal PDZ binding motif (PBM) interacts with the PDZ domain of tyrosine phosphatase PTPN3 to demonstrate expression of the LexA-E6 fusions. In the lower part of the panel are B42 transactivator fusions to E6AP containing the C843A mutation, which ablates ubiquitin ligase activity (E6AP-Ub−). The LXXLL motif, which binds 16E6, and mutant versions of the LXXLL motif are indicated. LQELS refers to the 11-amino-acid E6AP peptide sequence that has been mutated to ELTLQELSGEE, and LQEAS refers to the sequence ELTLQEASGEE. The LexA_FDELL_16E6 fusion expresses LexA fused to a 16E6-binding amphipathic peptide (MEGVFDELLGE) (23), then a 6-amino-acid linker (GGSGGS), and then 16E6. (B) LXXLL peptide interaction with 16E6 in cis blocks LXXLL peptide interaction in trans and reduces interaction with E6AP in trans. LexA reporter strains with the indicated LexA fusions were mated to strains expressing the indicated B42-transactivator fusions as previously described (5). Mated products were selected and then patched onto X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) plates to assay for protein interaction, indicated by dark color development. (C) LXXLL peptide binding to 16E6 confers interaction with p53. (D) E6 interaction with non-E6AP peptide sequences can confer interaction with p53. The LexA_FDELL_16E6 fusion expresses LexA fused to a 16E6-binding amphipathic peptide (MEGVFDELLGE) (23), then a 6-amino-acid linker (GGSGGS), and then 16E6. (E) Unfused 16E6 can recruit p53 to the minimal LQELL peptide. Strains with the indicated LexA fusions to E6AP-Ub−, the indicated fragments of E6AP-Ub−, or the minimal 11-amino-acid E6 binding motif of E6AP were mated to strains expressing 16E6, p53, or the combination of 16E6 and p53. p53 does not interact with E6AP unless 16E6 is present, and the minimal 11-amino-acid LQELL peptide of E6AP is sufficient to bind E6 and recruit p53. (F) Quantitative assay of beta-galactosidase activity of yeast strains from panel E. Diploid mated yeasts from the indicated spots in part E were grown in liquid galactose medium (to induce p53 expression) for 4 h and then assayed for beta-galactosidase activity using o-nitrophenyl-β-d-galactopyranoside (ONPG) as a substrate as described previously (5); the asterisk indicates a P value of <0.005 by Student's t test. Results are averages from three experiments, normalized to data for spot 4B.
LQELL_16E6 associated with p53 in yeast (Saccharomyces cerevisiae) in the absence of E6AP, indicating that cis interaction of the LQELL peptide with 16E6 reshapes 16E6 to interact with p53 in vivo (Fig. 1C, spot 2C). Mutation of LQELL to LQEAS largely reduced the interaction with p53 (Fig. 1C, spot 2D). To test if the E6AP LQELL peptide was uniquely required to reshape 16E6 to associate with p53, we utilized a different but similar peptide that had been selected for association with 16E6 in yeast (MEGVFDELLGE) (23); it too reshaped 16E6 to interact with p53 in vivo without E6AP (Fig. 1D, spot 3D).
To confirm these results without using E6 fusion proteins, we expressed LexA fused to either E6AP-Ub−, deletion mutants of E6AP-Ub−, E6AP-Ub− mutants with point mutations in the LQELL motif, or just the 11-amino-acid LQELL motif of E6AP (406-ELTLQELLGEE-416) together with native 16E6 and p53, so that native 16E6 acts as an adapter to bring native p53 to the LexA_E6AP-Ub− or LexA_LQELL peptide. Recruitment of p53 required only the minimum LXXLL motif and expression of native 16E6 (Fig. 1E, spot 4H). However, the LXXLL motif alone was substantially less effective than LXXLL in the context of E6AB-Ub−, correlating to the enhanced affinity of E6 for E6AP-Ub− compared to that for the isolated LXXLL motif (Fig. 1F).
Both LQELL_16E6 and 16E6 bound to E6AP-Ub− interact with the p53 core DNA binding domain.Fragments of p53 were expressed as B42 transactivator-fused proteins and assayed for association with LexA_16E6 compared to LexA_LQELL_16E6 (Fig. 2A and B). LQELL_16E6 could associate with full-length p53-(1–393) and p53-(92–393) but not the other p53 deletion mutants. In contrast, the LexA_16E6 + E6AP-Ub− complex required the full-length p53-(1–393) and did not associate with the 92–393 fragment (Fig. 2C). Both native p53 and B42-tagged p53 activated transcription from p53 binding sites in yeast (data not shown). Thus, LQELL_16E6 associates with the p53 core domain but is less stringent for p53 association than the 16E6–E6AP-Ub− complex, which required the transactivation domain as well (Fig. 2C). p53 with missense mutations in the DNA binding domain failed to associate with LexA_LQELL_16E6, consistent with LQELL_16E6 interaction with the p53 core DNA-binding domain in vivo (Fig. 2D).
LXXLL_16E6 associates with the p53 core DNA binding domain. (A) Diagram of p53 domains, adapted from reference 14. (B) Association of p53 domains with LXXLL-bound 16E6. LXXLL-peptide-bound 16E6 does not require the p53 transactivation domain and associates with p53-(92–393). The vertical white line indicates matched X-Gal indicator plates used in this experiment. (C) LXXLL_16E6 interacts with p53-(92-393), while the 16E6–E6AP-Ub− complex requires the full-length p53 protein. The vertical black line indicates the position of a removed irrelevant sample in the same plate, and the vertical white lines indicate additional separate but matched X-Gal indicator plates used in this experiment. (D) LXXLL peptide-bound 16E6 interaction with p53 is sensitive to missense mutations within the p53 core DNA-binding domain. The H179Q and V143A point mutants of p53 are cancer-associated mutants. p53 mutants with mutations in the transactivation domain were a gift from Jiayuh Lin (Nationwide Hospital, Columbus, OH) (11).
LQELL interaction stabilizes 16E6 expression but fails to target the degradation of p53.E6AP-null 8B9 cells are murine embryonic kidney cells from E6AP-null mice transformed by adenovirus E1a and mutation-activated Ras; in contrast to results obtained previously with these cells (12), we observed no targeted degradation of p53 by wild-type 16E6 in the absence of exogenously re-expressed wild-type E6AP (Fig. 3A). Expression of LQELL_16E6 without E6AP also failed to target p53 degradation unless WT E6AP was coexpressed (Fig. 3B). 16E6 expression was enhanced by fusion of the LQELL or the FDELL sequence shown in Fig. 1C but less so by fusion of LQEAS to the amino terminus of E6, indicating that LXXLL-type peptide interactions stabilize the expression of 16E6 without E6AP (Fig. 3C). However, only in the presence of E6AP-Ub− could a stable complex of E6 and p53 be immunoprecipitated (Fig. 3C), indicating that the E6–E6AP-Ub−–p53 complex is more stable than the LQELL_16E6–p53 complex. Interestingly, transient overexpression of some epitope-tagged 16E6 molecules can result in the appearance of p53 loss in the absence of E6AP, but these results appear to be secondary to toxicity of overexpression (compare loss of p53 and cotransfected hemagglutinin-green fluorescent protein [HA-GFP] in Fig. 3D and E and the failure to detect HA-16E6 compared to WT 16E6). Toxicity due to overexpressed epitope-tagged 16E6 was repeatedly observed with HA-tagged 16E6 and less so with FLAG_16E6-tagged protein but was not observed with HA_LQELL_16E6- or EE_16E6-tagged proteins (Fig. 3C and data not shown); this indicates that caution is needed in interpreting transient-overexpression experiments using tagged 16E6 proteins.
LXXLL peptide interaction stabilizes 16E6 expression but does not target p53 for degradation in the absence of E6AP. 8B9 cells are E6AP-null mouse kidney epithelial cells transformed by adenovirus E1a and activated ras oncogenes (gift from Lawrence Banks, ICGEB, Trieste, Italy) which were cultured in Dulbecco's modified Eagle medium and 10% fetal bovine serum and transfected using polyethyleneimine at an early passage. (A) Re-expression of E6AP is required for 16E6 to efficiently target the degradation of p53. 8B9 cells were cotransfected with human p53 (0.5 μg), HA-tagged GFP as an internal transfection control (0.25 μg), 1.0 μg of untagged E6AP or E6AP_Ub−, and the indicated amount of E6 (in μg) and lysed 24 h posttransfection. Transfected human p53 was detected with human specific monoclonal antibody Ab-8 (Oncogene Science). (B) Re-expression of E6AP is required for LQELL_16E6 to efficiently target the degradation of p53. Experiments were carried out as described for panel A, but 8B9 cells were transfected with HA_LQELL_16E6 instead of WT 16E6. (C) LXXLL peptides stabilize 16E6 expression but do not target p53 degradation. 8B9 cells were transfected with the indicated plasmids as described for panel A. All the LXXLL fusions to 16E6 were epitope tagged with HA. Cotransfected LacZ was used as an internal transfection control, and HA-NS refers to a nonspecific HA-reactive cellular band used as a loading control. The white line indicates removal of an irrelevant sample. The sample was divided, with total cell lysate being used for Western blotting and the soluble fraction being immunoprecipitated with rabbit anti-HA followed by Western blotting with HA monoclonal antibody. (D and E) Epitope-tagged 16E6 can artifactually appear to target p53 degradation in an E6AP-independent manner. 8B9 cells were transfected with cDNA expression plasmids (all based on pcDNA3) using polyethyleneimine, 150 ng HA_GFP, 250 ng human p53, 650 ng E6AP or E6AP-Ub−, and the indicated amounts of wild-type 16E6 or HA-tagged 16E6 and balancing amounts of empty pcDNA3 vector to a total of 4 μg and harvested 24 h later with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer. While wild-type 16E6 and LQELL_16E6 targeted only p53 degradation in the presence of wild-type E6AP, p53 was lost in lane 5 of cells transfected with HA-16E6. However, the loss of cotransfected HA_GFP in this lane suggests that the loss of p53 expression is due to cellular toxicity of overexpressed HA-16E6 in 8B9 cells. Results in panels D and E are representative of three replicate experiments.
How E6 proteins interact with cellular targets has been the subject of intensive study. Our observations suggest that E6 oncoproteins are unstable until bound to an appropriate amphipathic LXXLL-type peptide sequence and that binding to a suitable LXXLL-type peptide then induces conformational changes that enable greater protein stability in vivo. This correlates with recent findings that knockdown of E6AP destabilizes E6 expression in cancer cell lines (25). It may be that solvent exposure of the hydrophobic region of 16E6 that interacts with the hydrophobic leucines of LXXLL renders 16E6 unstable and that interaction of 16E6 with that LXXLL peptide of E6AP covers that hydrophobic surface and replaces the hydrophic surface with an acidic surface supplied by the acidic residues of the LXXLL peptide, thereby improving the stability of the 16E6-LXXLL complex. The observation that knockdown of E6AP in cancer cell lines destabilizes 16E6 suggests that the LQELL peptide of E6AP is the principal stabilizing peptide in cancer cell lines. Our results also show that LXXLL binding by 16E6 promotes interaction with secondary substrates such as p53. It is unclear if this is due to an unmasking of an intact p53-interacting site on 16E6 upon binding LXXLL or if the interaction of 16E6 with LXXLL peptides causes a de novo p53 interaction site to be formed by allosteric modification of 16E6. A recent nuclear magnetic resonance (NMR) solution structure of 16E6 shows strong conservation of the N-terminal (E6N) and C-terminal (E6C) zinc structured domains and poor resolution of the interdomain connecting region in the absence of bound LXXLL peptide (27). Since this solution structure poorly resolved the linker region of E6, it remains unclear how the structure of E6 is altered upon peptide binding or how peptide binding could alter the fold of either the E6N or E6C domain. The same study demonstrated that 16E6 dimerizes through self-association of the E6-N region and that this is required for p53 degradation, but prior work from this group indicated that this self-association of E6 is not involved in p53 association (15). It is likely that additional complexity is involved both in the association of E6 with p53 in solution and indirectly in the yeast-based assays employed in this study; since p53 is a tetramer and LexA itself is a dimer in solution as well (13), there is the possibility of complex multimeric structures being formed that would allow low affinity interactions to gain avidity by multimerization. E6 proteins from disparate groups of papillomaviruses are most sequence divergent at the very amino-terminal sequences, suggesting that other secondary non-p53 targets remain to be discovered for these other E6 types. Our results raise the possibility that 16E6 could recruit p53 to other non-E6AP protein complexes to which E6 is bound by LXXLL-type interactions. Cutaneous papillomavirus E6 proteins have recently been shown to interact with MAML proteins that are coactivators of Notch-induced signaling by docking on LXXLL peptides of MAML and thereby repressing MAML transcriptional activation (1, 16, 24); it is possible that such LXXLL peptide interactions could reshape these E6 proteins to associate with secondary targets, as was observed with 16E6 in this study. Some of the beta-group E6 proteins also associate with p300 (7); additional work will be required to determine if the p300 association with beta-group E6 is direct through p300 LXXLL interactions or if it could be analogous to the interaction of p53 with 16E6, with p300 association being triggered through prior interaction with MAML LXXLL peptides. Our work also suggests that while drugs designed to interact with the LXXLL-binding pocket of 16E6 could prevent the degradation of p53, they could also promote the association of 16E6 with p53 and perhaps stabilize E6 expression, with difficult-to-predict consequences. Our failure to detect the degradation of p53 by 16E6 in E6AP-null cells that had previously been reported (12) may be due to differences in experimental techniques or the exact E6 molecules employed.
ACKNOWLEDGMENTS
We declare no conflicts of interest.
This work was supported by NCI grant CA134737 and CA120352 to S.V.
We thank Gilles Trave, Katia Zanier, and Sebastian Charbonnier at CNRS (Strasbourg, France) for useful conversations and comments.
FOOTNOTES
- Received 17 May 2012.
- Accepted 6 August 2012.
- Accepted manuscript posted online 15 August 2012.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.