Background
Human papillomavirus (HPV) family groups a heterogeneous number of viruses able to infect squamous stratified epithelia and cause benign papillomas, warts and anogenital lesions, depending on the viral genotype, time of persistence and possible integration into the eukaryotic host genome. The virus also correlates with oropharyngeal malignancies, strongly rising because of sexual behaviors changes [
1].
HPV vaccine, the first one developed in 1991, till now represents, with its 2-, 4- and 9-valent formulations, the best way to prevent and control both infection and genital cancer progression that are mainly mediated by high risk HPV16 and 18 oncoproteins. Ten years of phase III clinical trials revealed their optimal effectiveness with success rates ranging from 90 to 100%, although to be confirmed by a 20 years overall period of clinical evidences [
2,
3].
Despite this and the strict screening measures, anogenital cancer, mainly promoted by the over-expression of the high risk α-HPV16 E6 and E7 oncoproteins which destabilize the host genome over time after primary infection [
4‐
6], still represents the fourth most common tumor and the second largest cause of death among women in the world, with about more than 5 hundred thousand new cases/year, as evidenced by the World Health Organization (WHO).
Cell replication fidelity is the result of the host DDR. Unlike many DNA viruses that promote their life cycle through the DDR inactivation, HR-HPVs encourage cells proliferation despite the DDR turned on. Why and how it occurs has only partially elucidated.
Although a lot of research has been published in the area [
7‐
9], further insights are needed to deepen the replication and host genome destabilizing properties of the virus, from the early phases of the tumorigenesis process, in order to discover new target therapies for unvaccinated HPV infected patients and for those in whom vaccination is not the right approach.
Well established reports demonstrate that HR-HPVs can efficiently amplify their genome thanks to the independent, but cooperative ability of E6 and E7 oncoproteins to manipulate the DDR and repair machinery of the targeted host’s epithelial cells [
10‐
14], activated just after the oncogene-induced replicative stress, leading to genomic instability (GIN) and cancer progression [
15‐
18].
It is well known that HR-HPV16 E6 oncoprotein is highly expressed during the early phases of infection and responsible for p53 degradation, via the proteasome machinery, through the binding with E6AP, a HECT domain ubiquitin ligase [
19]. Both E6 and E6AP alone are unable to bind p53, suggesting that their interaction is crucial for p53 degradation [
20‐
22]; indeed, structure analysis reveals that such binding induces conformational changes in E6AP that allow the link with p53 [
23].
The E6AP binding relies on a short leucine rich aminoacidic sequence, called LXXLL binding motif; when deleted and/or mutated, the formation of the E6-E6AP complex is prevented, thus failing to target p53 for degradation [
21,
23‐
25].
Similarly, E7 increases DDR through several mechanisms: by competing with E2F1-pRb interaction, thus leading to the inactivation of pRb, and promotion, E2F1 mediated, of DDR genes translation, by binding to the pRb-like proteins CBP/p300 and p107, that also harbour LXXLL sequence, and via the interaction and activation of several DDR proteins.
Because of E6 and E7 interaction with several LXXLL binding motif containing proteins [
26‐
30], we hypothesized their binding also to one of the LXXLL motifs contained into DNA damage sensors.
On these bases, we show experimental evidences to support our theory by using an in vitro reconstructed 3D infected epithelium [
31‐
34], made of HPV16E6E7 transduced keratinocytes and already consolidated for HPVs study, as a surrogate of an in vivo lesion. H/E and immunofluorescence stainings for the main viral markers (HPV16-DNA, E6 and E7), DSBs sensors γH2A.X and 53BP1 and cyclins (E2 and B1) were firstly made in order to reproduce, inside the 3D model, what in vivo happens.
Afterwards, despite the existence of many DNA damage sensor targets such as 53BP1, well-recognized DDR mediator/adaptor, MDC1 (mediator of DNA Damage Checkpoint 1), BRCA1 (Breast Cancer 1, early onset), TOPBP1 (Topoisomerase II-binding protein 1) and Claspin, we supposed that E6 and E7 could directly interfere just on 53BP1 activity since they all respectively link to p53 and pRb for the control of anti-tumorigenic cell-fate decisions and for their rapid proteasome-mediated degradation and inactivation respectively.
After highlighting E6 and E7 co-expression with 53BP1 by IF analysis, [
35‐
41] and found the presence of a LXXLL binding motif within the 53BP1 BRCT2 domain (LKVLL), we finally evidenced the complex and direct interaction between E6 and E7 HPV16 oncoproteins with 53BP1 via the highly specific and sensible in situ PLA system [
42‐
46], well known and validated assay able to identify and characterize interactions in native proteins in their correct tissue/cell context under near natural/physiological conditions.
Discussion
Cell replication fidelity is the result of the DDR, a complex signaling proteins network that finds, reports, and repairs lesions that occur to injured DNA. Many DNA viruses actively promote their life cycles through the inactivation of the host DDR, that ordinarily acts by blocking the cell cycle progression [
61]. On the contrary, despite the DDR turned on that HR-HPVs have learned to drive, these viruses highly promote the proliferation of the infected squamous epithelial cells. Why and how HR-HPVs activate the DDR through E6 and E7 oncoproteins, thus favoring cancer progression, is still a matter of debate; what it is known is that these events are not so useful for maintaining, but rather for amplifying viral DNA replication.
In particular, during HPV16 infection early phases, E6 oncoprotein manipulates the DDR response through the link to p53 and degradation, mediated by LXXLL E6AP binding motif; unfortunately, a direct role for E6 in the DDR response has not clearly identified yet.
Similarly, E7 increases DDR proteins levels through several main mechanisms: i) by competing with E2F1-pRb interaction, thus leading to the inactivation of pRb, and to the promotion, E2F1 mediated, of DDR genes translation, ii) by binding to the pRb-like proteins CBP/p300 and p107, that also harbour LXXLL sequence, and iii) via the interaction and activation of several DDR proteins, most of them yet unknown.
Therefore, it can be supposed that the DDR is bypassed by the over-expression of the viral oncogenes that promote cell cycle progression. Furthermore, several recent studies sustain the hypothesis that HPV chromatin its-self is modified by the DDR. In particular Gillespie et al. demonstrated that γH2A.X, a marker for cellular response to DNA damage, localizes to HPV replication compartments, inside nuclear foci, whose size directly increases together with virus productive replication [
12]. Importantly, γH2A.X was found to link to viral DNA, suggesting the enrollment of cells repair factors into specific viral replication sites. In support of this, DDR components that rely on γH2A.X for recruitment to DNA breaks, including 53BP1, Nbs1, BRCA1, and Rad51, also localize to HPV replication compartments. In particular E7 promotes DNA breaks accumulation inside cells harboring γH2A.X nuclear foci, while, even if E6 can also increase DNA breaks, it seems not to promote γH2AX foci number [
14]. To support these evidences, Park et al. [
62‐
64] demonstrated, in genetically engineered murine models expressing E6 and E7 HPV16 oncoproteins, that HPV16 E7, alone or together with E6, was able to promote an accumulation of γH2A.X nuclear foci inside epithelial cells, while E6 alone did not.
In order to add evidences regarding E6 and E7 role in the DDR, we firstly produced an in vitro 3D epithelium, made of HPV16 E6E7 transduced keratinocytes, already consolidated study model for HPVs, and assessed a H&E, BrdU, HPV16 DNA and E6E7 proteins staining.
It is well known that γH2AX is the most suitable DNA damage and repair sensor marker and that 53BP1 protein is one of the DDR components recruited to DNA breaks [
65‐
67]. To ascertain the DDR in the 3D in vitro HPV16 epithelial model, we evaluated both γH2A.X and 53BP1 positive nuclear punctuate signals; they were both present, but not in the normal counterpart, as expected.
Then we evaluated if also cyclins expression was in line with what happens in vivo during HR-HPV infection. As described by the literature, cyclin E is physiologically produced in late G1 until cells enter in S phase and then decreases. Moreover, DNA damaged S cells normally inhibit activation and nuclear import of cyclin B1-Cdk, master regulator of the entry in M phase, therefore stopping in G2 to avoid oncogenic transformation. In DNA damaged HPV16 infected cells these cyclins are no more modulated; indeed they are overexpressed, indicating a clear propensity towards tumoral transformation [
68‐
70]. All these in vivo events, we observed in vitro
, are displayed in Fig.
4.
Several proteins, such as 53BP1, MDC1, BRCA1, TOPBP1 and Claspin sense the DNA damage. Since 53BP1, like E6 and E7, is a binding partner of the central DNA binding domain of p53, pRb, CBP/p300 and p107 pRb-like proteins, with whom it cooperates in tumor suppression [
35‐
37,
71‐
73], we hypothesized that E6 and E7 could bind to this DNA damage detector. In support to our hypothesis came the fact that defects in DNA damage control checkpoints are able to promote chromosome translocations and tumorigenesis, particularly in the context in which p53 and pRb dependent apoptosis is abrogated, like in case of E6 and E7 HPV16 overexpression [
39].
To gain evidence on our hypothesis, we firstly performed double immunofluorescent and biochemical assays. Comforted by the evidence of the similar punctuate localization pattern of 53BP1 and E6 and E7 proteins in the 3D model (Fig.
4A, A’, A”, D, E, E’, E” and H), we reinforced the association between HPV infection, DDR and genome injury amplification also in CaSki and E6E7HPV16 cells. CaSki, whose HPV16 genome is known to be of about 600 copies of integrated DNA/cell, highly express the E6E7 ORFs [
59]. These cells, often used to study chromosomal rearrangements and genomic instability induced by HR-HPV infection and integration, were here utilized as control cells [
10,
74].
Martinez-Zapien et al. [
23] show that LXXLL motif containing proteins could target E6 and E7; we therefore analyzed 53BP1 sequence and we found a LXXLL (LKVLL) motif within its BRTC2 domain. By comparing this site with those of p53 (LWKLL), we observed a similar total charge. On the contrary, we detected the same total charge among the E6AP- (LQELL), CBP/p300- (LQDLL) and p107 (LDQLL) domains. This analysis let us suppose that hydrophobicity and a slightly positivity of the leucin rich motif could guarantee the optimal conformation to allow the interaction with E6 and E7. Considering that E6 and E7 LXXLL motifs both contain hydrophilic, acid and negative charged aminoacids, we supposed that 53BP1 pocket could enable not only E6- but also E7-53BP1 interactions.
Finally, we performed an in situ PLA technique, which allowed us to detect the direct protein-protein E6-53BP1 and E7-53BP1 interactions within CaSki, E6E7HPV16 NHEKs cells and, more consistently, inside the in vitro reconstructed infected tissue. The PLA signal is different between CaSki (2D) and 3D epithelial cultures; precisely, in CaSki is cytoplasmic/perinuclear, while in 3D is mainly nuclear.
Our data are in agreement with those of Dreier et al.; the subcellular localization of endogenous HPV16 E7 oncoprotein varies during the cell cycle: many interphase (G1, S or G2 phases) CaSki cells showed a predominantly diffuse subcellular cytoplasmic HPV16 E7 expression with a ring structure surrounding the nucleus (that occurs shortly in the early G1 phase), while only few mitotic cells displayed a faint nuclear pattern. It is likely that the nuclear E7 structures reflect well established nuclear E7 functions, such as deregulation of the p16Ink4A/pRb pathway.
In the 3D model, E7 is mainly intranuclear where it co-localize with 53BP1 that is present in mitotic cells during telophase and cytokinesis [
75].
The E6-53BP1 signal was mainly in the perinuclear compartment in big complexes just where E6 is necessary to degrade p53 [
76], while the smallest E7-53BP1 perinuclear punctuate signals are also present in the extracellular milieu of the 3D model, sign of E7 release into the extracellular compartment where it is needed to exert its immunosuppressive role in the in vivo context [
77].
These findings were reached thanks to the excellent PLA technology, with its high sensibility and specificity that allows to detect, as distinct spots, and with sub-cellular resolution and molecular precision (0–40 nm), single-molecule protein-protein interaction events. The exponential rolling circle amplification that occurs produces very strong and visible signals, also in case of limited number of interacting molecules. This assay can in situ identify, validate and characterize interactions in native proteins in their correct tissue/cell context under near natural/physiological conditions, without requiring protein content extraction from tissue or the need to over-express the target proteins. Conversely, as also observed by other authors, since co-IP needs high concentrations of both target proteins to generate effective signals and is often dependent on antibodies affinity [
42,
43,
52‐
58], it can’t be the right way to see protein interactions in our 3D models where cellular density and consequently target proteins are not so highly expressed.
In situ PLA has 5-times higher specificity and sensibility than IF [
78], that conversely allows the detection of single signals if only at a distance of approximately 0.2 μm.
In conclusion this study highlight, for the first time in our knowledge, the interaction between HPV16 E6 and E7 with the 53BP1 protein that is recruited inside nuclear foci after DNA damage. Further investigations on mutants of the LVKLL motif are necessary to definitively assess if the bindings occur just through this site.
The discovery of such interactions is important not only to better understand proteins function and behavior, specifically of E6 and E7, but also to predict the biological processes and pathways in which those proteins are involved in. In our opinion these interactions could explain why during HPV induced carcinogenesis 53BP1 doesn’t correctly process the DDR signal and doesn’t define DSB repair pathway choice in the G1 and S/G2 phases.