Papillomaviruses: a systematic review

Abstract In the last decades, a group of viruses has received great attention due to its relationship with cancer development and its wide distribution throughout the vertebrates: the papillomaviruses. In this article, we aim to review some of the most relevant reports concerning the use of bovines as an experimental model for studies related to papillomaviruses. Moreover, the obtained data contributes to the development of strategies against the clinical consequences of bovine papillomaviruses (BPV) that have led to drastic hazards to the herds. To overcome the problem, the vaccines that we have been developing involve recombinant DNA technology, aiming at prophylactic and therapeutic procedures. It is important to point out that these strategies can be used as models for innovative procedures against HPV, as this virus is the main causal agent of cervical cancer, the second most fatal cancer in women.


A brief history of the papillomavirus (PVs) on carcinogenesis
In the last decades, novel diagnostic methods and therapies have been implemented in an attempt to combat cancer. However, the number of patients that succumb to the disease has increased globally (Varga et al., 2014). This negative result emphasizes the complexity of the oncogenic process, which has a multifactorial cause. Among the etiological factors associated to cancer are the infectious agents, such as bacteria and viruses.
The association between cancer and infectious agents has been discussed for centuries (Graner, 2000). In 1858, George B. Wood stated in his book Practice of Medicine that cancer could be disseminated as an infectious disease (Graner, 2000). However, the association between cancer and infectious agents was only implied in the second half of 19 th century by Rudolf Maier (Graner, 2000;zur Hausen, 2009). The major difficulty in demonstrating this association can be attributed to the time-lapse of 15-40 years be-tween the infection and the development of the first clinical signs that would allow cancer diagnosis (zur Hausen, 2009). Yet, in the last decades, the involvement of infectious agents with cancer has aroused great attention, as one in ten human malignancies is caused by these pathogens (Ribeiro-Müller and Müller, 2014).
there is no epidemiological data about the number of PVassociated incident animal cancers, this association is recognized since 1932 (Shope and Hurst, 1933;Graner, 2000). Moreover, veterinary research demonstrates an increase in both benign and malignant tumors (Misdorp, 1996), particularly in domestic animals (cats and dogs). Furthermore, animal neoplasms are important models for the study of human oncogenic process (Misdorp, 1996), by allowing the identification of molecular mechanisms associated to carcinogenesis (Cotchin, 1962(Cotchin, , 1976 and novel therapeutics (Misdorp, 1996), and emphasizing the importance of comparative oncology.
In this review, we summarize relevant data and advances in papillomaviruses biology, including viral evolution, pathogenic mechanism of viral proteins and oncoproteins, ways of transmission, pathogenesis and oncogenesis. We also discuss the importance of BPV as a study model for HPV-associated oncogenic process.

Evolutionary history of papillomaviruses
Although the virus origin is still uncertain (Bernard, 1994;Holland and Domingo, 1998), studies about PVs suggest that they arose concomitantly with tetrapods in the Carboniferous period of the Paleozoic era (330 million years ago) (Rector and Van Ranst, 2013). This makes the PVs one of the oldest and largest known virus family (Cubie, 2013;Rector and Van Ranst, 2013).
Studies based on molecular phylogeny suggest that these viruses originated in Africa, from where they disseminated to all continents (Bernard, 1994). It was not a pandemic dissemination, occurring over one million years (Bernard, 1994).
PVs genomic diversification occurred together with mammals diversification, being influenced by multiple evolutionary forces (Bravo et al., 2010), such as the addition of sequence boxes, previously present in their host (García-Vallvé et al., 2005). Thus, PVs co-evolved with their respective host (Gottschling et al., 2007). An evidence of this co-evolution is the similarity in guanine and cytosine (G + C) content; HPVs have 41-49% of G + C, and humans 40-42% (Black, 1968). Another evidence of this coevolution is seen with the use of in situ hybridization probe for Shope papilloma virus (CRPV), which presents homology with the rabbit genome sequences (Black, 1968). Moreover, replicative mechanisms of PVs and their host are similar, reinforcing co-evolution (Leatherwood, 1998). These data suggest that PVs could have originated from genomic fragments of amniotes' common ancestor (van Doorlaer and Burk, 2010;Rector and Van Ranst, 2013;van Doorslaer, 2013).
Animal domestication favored the enzootic transmission of PVs (Gottschling et al., 2007). In this process, novel strategies of adaptation were required to guarantee the infection of novel hosts (Gottschling et al., 2011). These adaptations favored PVs evolution, conferring specificity in terms of host to all members of Papillomaviridae family, except for BPV, which is able to promote cross-infection (Bernard, 1994).
Based on this, García-Vallvé et al. (2005) described a hypothetical model to recreate PVs evolution. This model supports the existence of a proto-papillomavirus, com-prised by URR-E1-E2-L2-L1 genomic regions, able to infect primitive amniotes. Along mammals divergence (150 million years ago), this proto-papillomavirus was added with E6 and E7 ORFs. From this moment, PVs' interaction with their host became more specific, resulting in a coevolution. This process also resulted in the addition of E5 ORF in the hot spot between E2 and L2 ORFs (Bravo and Felez-Sanchez, 2015). Phylogenetic analyses point out that PVs only acquired oncogenic potential after they infected humans (García-Vallvé et al., 2005). This suggests that BPV could have originated from HPV transmission to bovines, as a consequence of animal domestication (García-Vallvé et al., 2005)., which would justify the similarities between these viruses, making the BPV a useful model to HPV studies (Campo, 2006;Borzacchiello and Roperto, 2008;Munday, 2014).
The E4 ORF codifies a family of proteins produced by splicing followed by post-translational modifications (Campo, 1997a). The E4 protein is the most expressed protein of PVs (Doorbar, 2013). For this reason, E4 is easily detected in suprabasal and granulosum layers of epidermis (Rampias et al., 2013), being recognized as important hallmark of PVs pathogenic activity (Doorbar, 2013). The E4 protein interacts with cytokeratin filamentous, contributing to viral replication (Campo, 1997a). Moreover, E4 is associated to virus maturation and extracellular matrix (MEC) remodeling (Ferraro et al., 2011).

Late proteins: L1 and L2
The L1 ORF is the most conserved among PVs (Bernard et al., 2006;Haga et al., 2013), and for this reason it is employed in virus classification (Haga et al., 2013;de Villiers, 2013). The L1 protein has 55 kDa (Buck et al.,2,013). and is able to self-organize in pentameric structures that compose the viral capsid (Ribeiro-Müller and Müller, 2014). It has a central role in viral infection mechanism, allowing the capsid anchorage to heparin sulfate receptors present in cell membrane (Florin et al., 2012). Considering that L1 is a late protein, it is expressed in the most differentiated epithelium layers (Buck et al., 2004). Therefore, L1 immunodetection has been considered the main evidence of productive infection (Nasir and Reid, 1999;Costa and Medeiros, 2014;Melo et al., 2015;Araldi et al., 2015b), which is characterized by the virus assembly (Green, 1972).
The L2 protein has 64-78 kDa (Wang and Roden, 2013). The molecular weight variation is a consequence of post-translational modifications (Wang and Roden, 2013). During the assembly of PV particles, 2 binds to viral DNA, contributing to encapsidation and then to viral release (García-Vallvé et al., 2005;Campo, 2006).
A third structural protein (L3) has been described as present exclusively in BPV-4 (Catroxo et al., 2013). However, its function remains unclear.
PV-1 E5 is a transmembrane protein (Costa and Medeiros, 2014), with 43-44 amino acids, characterized by the presence of a central hydrophobic region, which acts as a transmembrane domain (Burkhardt et al., 1989;Tomita et al., 2007;DiMaio, 2014). E5 also presents two cysteine residues in the C-terminal region that confers stability for the homodimer composed by two E5 monomers (DiMaio, 2014).
The overexpression of p53 represents a threat to viral replication (Shamanna et al., 2013). For this reason, different oncogenic viruses express proteins able to promote p53 deregulation. Among these proteins are antigen T of SV40, adenovirus E1B, hepatitis B virus (HBV) HBx and PVs E6 (Shamanna et al., 2013;Cuninghame et al., 2014). Thus, p53 deregulation is a characteristic shared by oncogenic viruses.
Studies have also demonstrated that E6 oncoprotein can induce cell transformation (Liu et al., 2002) and immortalization (Boon et al., 2015) due to the up-regulation of telomerases (Cuninghame et al., 2014). This occurs because E6 promotes FOXM1 overexpression, resulting in cyclin B1, D1 and cdc25 expression and, therefore, cell proliferation (Chen and Lee, 2015). In addition, E6 is able to bind to LXXL motif of MAML1 transcriptional regulator, inhibiting the Notch signaling (Tan et al., 2012a;White and Howley, 2013). The E6 oncoprotein of both HPV and BPV also promotes energy metabolic deregulation, contributing to cell transformation .
E6 inhibits the SSB repair system, leading to genomic instability. This occurs because the oncoprotein interacts with XRCC1 and O 6 -methylguanosine-DNA-methyltransferase protein, which is recruited during the SSB repair (Wallace and Galloway, 2014). E6 also leads to cyto-genetic damages (Wallace and Galloway, 2014) and stimulates neosis (Araldi et al., 2015a).

E7 oncoprotein
The E7 oncoprotein has 127 amino acids (Borzacchiello, 2007), and is able to bind to LXCXE conserved motif of pRb tumor suppressor protein (Moody and Laimins, 2010;White and Howley, 2013). This binding results in pRb phosphorylation, leading to E2F factor translocation to the nucleus . The E2F factor recruits different chromatin modifiers, including histone deacetylases (HDAC) (Moody and Laimins, 2010). Thus, the E7 oncoprotein induces the constitutive expression of E2F-responsive genes, such as cyclin A and E (Moody and Laimins, 2010), leading to S and G2/M cell cycle phase increase (Sacco et al., 2003;Ferraro et al., 2011). The interaction between viral oncoproteins and pRb is not exclusive of PVs, being also observed with adenovirus E1A and antigen T of SV40 (White et al., 2015).

Bovine papillomavirus (BPV): first reports and questions
In 1986 in an intriguing paper by Campo and Jarret (1986) reported the description of six different BPV types (BPV-1 to BPV-6) classified into two subgroups: subgroup A, that promotes fibropapillomas, and subgroup B, which leads to true epithelial papillomas. The report also showed that the BPV-4, a member of subgroup B, was the etiological agent of papillomas of the upper digestive tract, which could become carcinomas in animals feeding on a specific bracken fern pasture (Pteridium aquilinum). Later, Walter- Moura et al. (1988) verified the increase of chromosome aberrations in cells obtained from short term peripheral lymphocytes collected from bovines afflicted with bovine enzootic hematuria (BHE) that were exposed to pastures with bracken fern. These studies were reinforced by following reports, demonstrating that carcinogenesis is associated to the interaction between BPV and carcinogens present in the fern, such as quercetin and ptaquilosides (Pennie and Campo, 1992;Shahin et al., 1999;Potter and Baird, 2000;Beniston et al., 2001;Leal et al., 2003;Lioi et al., 2004). These studies brought new questions to be answered: how can the co-factor act in synergism with the virus? Why can these effects be detected in peripheral blood, considering that the virus is epithelial? Or, could the effect on chromosomes be only related to bracken fern compounds?

BPV and bovine papillomatosis (BP)
BPV is a cosmopolitan virus, worldwide distributed, independently of the level of expertise on livestock exploration (He et al., 2014). It is estimated that 60% of Brazilian cattle herd is infected by BPV (Stocco dos Santos, et al., 1998). However, that rate can be higher, once virus infection can be asymptomatic Silva et al., 2013a). Furthermore, the absence of epidemiological studies about BPV distribution could underestimate the real percentage of infected animals, representing a notorious difficulty in attempting to develop vaccines, once we do not know the prevalent virus types in each country.

Diagnostic methods
Studies about BPV diversity and prevalence are mandatory to develop novel therapeutic methods (Silva et al., 2013b), since immunity is species-specific (Claus et al., 2007). Therefore, diagnosis is crucial.
Although real-time PCR (qRT-PCR) allows to determine the number of viral copies, this method has the lowest reproducibility (Guo et al., 2012). For this reason, PCR followed by DNA sequencing represents the most common method to identify and typify BPVs (Melo et al., 2014;Araldi et al., 2014b). However, there is no "gold-standard" primer employed in PVs identification (Antonsson et al., 2010).
On the one hand, although the specific primers have higher sensitivity than degenerate primers, they cannot identify the 14 BPV types simultaneously (Araldi et al., 2014b). In addition, specific primers do not allow to identify novel virus types and subtypes (Araldi et al., 2014b). Moreover, evidence indicates that a specific primer for BPV-1 can anneal to BPV-2 (Haga et al., 2013), once these virus types are considered serotypes-like (Shafti-Keramat et al., 2009). However, in a comparative study using complete genomes of BPV-1-6 in 2014 we demonstrated the BPV-1 primer specificity (Araldi et al., 2014b).
Among the degenerate primers described in the literature, FAP59/64 (Forslund et al., 1999) is the most employed in both BPV and HPV identification. This primer was designed based on the L1 ORF homology of HPV (Forslund et al., 1999) and later optimized to identify BPV DNA sequences (Ogawa et al., 2004). Furthermore, the use of FAP59/64 primer already allowed to identity novel BPV types, including BPV-9 and 10 (Hatama et al., 2008). However, despite of these advantages of degenerate primers, there are several reports showing their low sensitivity. For instance, MY09/11, another degenerate primer frequently used in HPV diagnosis, was unable to identify BPV and HPV sequences in clinical samples (Martelli-Marzagão et al., 2010;Zhu et al., 2014) or copies of complete cloned genome (Araldi et al., 2014b).
Although PCR is commonly used to identify PV DNA sequences, the method does not allow to identify their localization and physical state (episomal or integrated). Therefore, CISH represents and additional method used to demonstrate the physical state of these viruses (Black, 1968;Munday, 2014;Melo et al., 2015). Another additional technique to identify PVs is L1 immunodetection (Nasir and Reid, 1999;Longworth and Laimins. 2004;Roperto et al., 2011;Munday, 2014;Melo et al., 2015;Araldi et al., 2015b)., which not only allows identification of the viral presence, but also provides an important evidence of productive infection (Nasir and Reid, 1999;Roperto et al., 2011;Melo et al., 2015).
Histopathological analysis of BPV-infected lesions is a differential diagnosis method, which allows to identify intra-epithelial neoplasms with oncogenic potential (Monteiro et al., 2008;Araldi et al., 2015b).

BPV infection pathway and histopathological alterations
BPV transmission can occur by direct (animal-animal) or indirect contact with contaminated surfaces (Muro et al., 2008;Cubie, 2013). Studies also show that the virus can be transmitted by flies (Finlay et al., 2009) and ticks (Muro et al., 2008).
The L1 binding to heparin sulfate leads to conformational changes in capsid icosaedric structure . This exposes the L2 N-terminal to be cleaved by furin protein, present in the cell membrane . This cleavage induces a second capsid conformational change, allowing L2 to bind to different receptors, such as integrin a2b4 . Next, the virions are internalized by an clathrin-dependent endocytose mechanism, resulting in cytoplasmic vesicles that associate to lysosomes (Day et al., 2003). The lysosomal acid content release promotes pH alterations in capsid proteins, resulting in viral DNA release (Day et al., 2003). The BPV genome is found in epissomal form (Campo, 2002;Munday, 2014;Cota et al., 2015), while HPV can integrate in fragile sites of the host genome (Monte and Peixoto, 2010;Moody and Laimins, 2010;Munday, 2014). A current study based on qRT-PCR, showed that cutaneous papillomas have about 2.210 4 viral copies (Cota et al., 2015).
Virus assembly is observed in the most differentiated epithelium layers (Munday, 2014), where the virion release occurs by cell degeneration (Brobst and Hinsman, 1966;Buck et al., 2013). This process results in koilocyte formation. The term koilocyte comes from the Greek word koillos, which means "cavity" (Ferraro et al., 2011). Koilocyte formation results from the cytopathic effect of E5 and E6 oncoproteins, although the molecular mechanism that results in cell vacuolization remains unclear (Krawczyk et al., 2008). However, the cytoplasmic vacuolization contributes to cell fragility and virion release (Krawczyk et al., 2008;Wang and Kieff, 2013). In this sense, koilocytes are cells destined to apoptosis, which emerges as a consequence of DNA replication and macromolecule synthesis inhibition.
After viral assembly, virions are released in the corneum layer, allowing infiltration in the keratin matrix (Brobst and Hinsman, 1966;Buck et al., 2013). This mechanism confers an immune evasion, since the icosaedric morphology of PVs is immunoreactive . In addition, keratin confers physical protection for virions, as they are non-enveloped.

BPV as a model for HPV
HPV is species-specific, infecting exclusively humans (Koller and Olson, 1972;Campo, 2002). This specificity was demonstrated in the 1970s, when calves, hamsters, ponies and Rhesus monkeys were inoculated with BPV and HPV virions isolated by ultracentrifugation (Koller and Olson, 1972). However, only BPV was able to infect the species (Koller and Olson, 1972). Due to its capability to infect different species and the pathogenic and morphological characteristics shared with HPV (Munday, 2014), BPV has been used as a prototype to study PV biology and oncology (Koller and Olson, 1972;Campo, 2002;Liu et al., 2005;Campo, 2006;Costa and Medeiros, 2014). Therefore, research involving BPV has contributed with the understanding of viral oncogenesis (Costa and Medeiros, 2014;Munday, 2014). Based on these data, the next sections will focus on BPV and its host interaction.
Equine sarcoids are lesions with intense fibroblastic proliferation, in which fibroblasts are disposed in fusiform bundles or spirally organized, presenting a morphology of fibropapilloma-like (Martens et al., 2000). Another characteristic of this neoplasia is the presence of anaplastic and pleomorphic fibroblasts, perpendiculary orientated in relation to basal membrane, and being observed in the dermo-epidermal junction (Bogaert et al., 2010). The epidermal component is only present in verrucous and mixed sarcoids (Bogaert et al., 2010). The sarcoid invasiveness capability can be attributed to the high levels of expression of metalloproteinase. MMP1 promotes the laminin and collagen IV degradation, resulting in extracellular matrix remodeling (Mosseri et al., 2014) Equine sarcoid has a multifactorial cause (Bergvall, 2013). However, the Deltapapillomavirus is recognized as an etiological factor (Nasir and Reid, 1999;Chambers et al., 2003b;Bogaert et al., 2008b;Bergvall, 2013). The association between BPV and equine sarcoid was first described by Cook in 1951 (Brandt et al., 2008). BPV-1 and 2 DNA sequences are identified in 100% of equine sarcoids (Martens et al., 2001a;Gaynor et al., 2015b). Furthermore, DNA sequences of these virus types are identified in 2/3 of asymptomatic horses (Bravo et al., 2010). These data demonstrate that the virus can be asymptomatic, remaining latent in epidermis and dermis (Bogaert et al., 2008b;Brandt et al., 2008;Bergvall, 2013).
As verified in PB, sarcoids are frequently seen in sites most susceptible to traumatism (Angelos et al., 1991;Otten et al., 1993;Martens et al., 2000), confirming the need of tissue micro-injury for virus infection. In addition, as in bovines, it is believed that insects can contribute to BPV transmission, because BPV-1 DNA sequences were already identified in Musca automnalis, Fannia carnicularis and Stomoxys calcinatrans flies (Bergvall, 2013).
Studies also show that the Arabic breed is the most susceptible to sarcoid development (Knottenbelt, 2005;Bogaert et al., 2008b). The reason for this is the presence of W3 and B1 MHC-II haplotypes that facilitate BPV infection persistence (Chambers et al., 2003b;Bogaert et al., 2008b).
Currently, treatments for sarcoid, as well as for PB, are almost inefficient (Bergvall, 2013). Treatment methods are: (1) surgical excision of the neoplasia, with recurrence in 50-64% of the cases within six months (Lancaster et al., 1977;Martens et al., 2001b;Bergvall, 2013;Mosseri et al., 2014), (2) laser therapy, where recurrence is 38% (Martens et al., 2001b); cryotherapy, which is inefficient for lesions larger than 2 cm 2 (Carr, 2009), and chemotherapy using cisplatin or 5'-fluouracyl (5-FU), that can cause nephro and hepatotoxicity (Stewart et al., 2006). The recurrence of disease is argued to be a consequence of BPV presence in the surgical margin (Martens et al., 2001a). However, a current study showed a lack of correlation between BPV DNA in surgical margins and recurrence of equine sarcoids (Taylor et al., 2014). Meanwhile, Brandt et al. (2008) identified BPV DNA sequences in peripheral blood of sarcoidaffected equines. These data suggest that, as verified in bovines (Stocco dos Santos et al., 1998;Roperto et al., 2011;Araldi et al., 2013;Melo et al., 2015), the peripheral blood can be argued as being a vehicle of viral dissemination.

Breaking paradigms
After the first reports (Campo and Jarret, 1986;Walter-Moura et al., 1988), our purpose of investigation was to examine the role of peripheral blood as a potential BPVtransmitting agent. The first data came from Stocco dos Santos et al. (1998), which showed high levels of chromosomal aberrations in lymphocytes infected by BPV-2. That study also described the presence of BPV-2 DNA sequences in peripheral blood of donor and recipient animals and in the progeny of recipient animals. These results were the first evidence of vertical transmission . In this study, we verified the presence of BPV DNA sequences and DNA damages in peripheral blood samples of animals from different areas of Brazil (Diniz et al., 2009;Melo et al., 2011;Araldi et al., 2013). In addition, we also demonstrated the presence of different BPV types in peripheral blood and cutaneous papilloma (Araldi et al., 2014a). All these data suggest a viral activity in blood cells (Stocco dos Santos et al., 1998;Araldi et al., 2013). The presence of BPV DNA sequences in peripheral blood mononuclear cells (PBMCs) was also described in the literature in both bovines Silva et al., 2013a) and equines (Brandt et al., 2008, reinforcing our results. Following studies also showed the presence of BPV transcripts and the L1 capsid protein in PBMCs, demonstrating the productive infection in blood cells Melo et al., 2015). We also described the presence of BPV DNA sequences in different non-epithelial tissues such as spermatozoa, urine and milk (Lindsey et al., 2009). These data demonstrate the need to review the natural history of papillomavirus, as currently proposed by Munday (2014).
We verified that primary culture cell lines from BPV-infected cutaneous and esophageal papilloma have chromosomal aberrations similar to those verified in peripheral blood (Campos et al., 2013). In addition, using BPV-4 E7 oncoprotein transformed PALF cell lines, we demonstrated the mutagenic potential of quercetin, a flavonoid found in bracken fern P. aquilinum, which is recognized as a co-factor to BPV-associated upper gastric cancer (Leal et al., 2003). In a current study, we also showed that the BPV induced metabolic alteration in host cells, increasing reactive oxygen species and DNA damages (Araldi, 2016;Araldi et al., 2016). A summary of our main results is shown in Figure 3. These data suggest that BPV-infected cell lines are a useful model to study the pathogenic mechanisms that lead to cancer. Our contribution has allowed to know the BPV prevalence and distribution in Brazil (Diniz et al., 2009;Carvalho et al., 2013;Lunardi et al., 2013;Melo et al., 2014;Araldi et al., 2014b;Cota et al., 2015;Gaynor et al., 2015a;Grindatto et al., 2015;Dong et al., 2016;Martano et al., 2016). These studies showed that the co-infection of at least two BPV types is frequent (Araldi et al., , 2014aCarvalho et al., 2013), demonstrating the importance to develop multivalent vaccines. 10 Araldi et al.  (Melo et al., 2015); C) BPV virions identified in cytoplasmic vesicles of PBMCs (Melo et al., 2015); D) Immunodetection of BPV L1 and E2 in lymphocytes (Melo et al., 2015); E) cytogenetic aberrations (breaks) observed in BPV-infected lymphocyte (Stocco dos Santos et al., 1998)

BPV-associated malignant neoplasms Urinary bladder carcinoma
Urinary bladder carcinoma represents 0.01% of all bovine cancers (Roperto et al., 2015). It is estimated that urinary bladder cancer has caused economic losses of 4 million between 2000-2006 in the Azores (Costa and Medeiros, 2014).
The etiopathogenic role of BPV in urinary bladder carcinoma was first described in 1955 in Brazil and South Africa (Plowrigh, 1955). Currently, the BPV promoter action in urothelial carcinoma is well established. This is because sequences of BPV-1, 2, 13 and 14, as well as the expression of E5 oncoprotein are detected in this neoplasms (Wosiacki et al., 2006;Balcos et al., 2008;Roperto et al., 2016;Russo et al., 2016).
P. aquilinum presents high levels of immunosuppressor and carcinogenic compounds, including quercetin, ptaquilosides and shikimic acid (Shahin et al., 1999;Beniston et al., 2001;Bonadies et al., 2004). The immunosuppressor activity of these metabolites contributes to BPV infection persistence and represents an additional source of DNA mutation (Leal et al., 2003). According to Tokarnia et al. (2000), the diary intake of 10 kg of Pteridum ssp. in a year can lead to BEH. Moreover, the consumption of this bracken fern can result in urinary bladder and esophageal carcinoma (Masuda et al., 2011).

Esophageal carcinoma (EC)
While the association between HPV and EC remains under discussion despite all evidences (Antonsson et al., 2016), the etiological role of BPV in EC is well established (Borzacchiello et al., 2003). EC in cattle is a self-limiting disease, being directly associated to BPV-4 infection (Borzacchiello et al., 2003;Masuda et al., 2011).
Although there is no report about BPV transmission in humans, a study performed in Germany showed a high incidence of warts in veterinarians that had contact with bovines (Bosse and Christophers, 1964). Another study in Central Asia discussed the association between milk consumption and EC (Nasrollahzadeh et al., 2015). Considering that BPV DNA sequences were already detected in milk (Lindsey et al., 2009) and the thermal resistance of viral capsid (Módolo et al., data not published), it seems that the virus can survive to pasteurization process (zur Hausen, 2012). Therefore, more efforts are necessary to verify this possible cross-infection.
EC is the eighth most prevalent human malignancy, being considered the sixth causa mortis for cancer globally (Antonsson et al., 2010;Herbster et al., 2012). Considered the third most common gastrointestinal cancer (Felin et al., 2008), EC has a high incidence in men (Bjørge et al., 1997). In 2002, 462,000 novel cases of EC in the world were reported (Antonsson et al., 2010). Brazil had 10,780 novel cases of EC in 2014 and 7,636 deaths due to the disease in 2011 (INCA, 2016).; its incidence has increased in the last years (INCA, 2016). This data leads to concern, since EC has a mortality rate that is 25% higher than cervical cancer (Han et al., 1996;Kahrilas and Hirano, 2013).
Among the clinical signs of EC in humans are progressive dysphagia, weight loss, odynophagia, anorexia, fever and retrosternal pain (Felin et al., 2008;Haster and Owyang, 2013), which are similar to those verified in bovines (Borzacchiello et al., 2003). Scarce epidemiological data about EC in cattle are available. In humans, EC has a variable incidence according to geographic area (Bjørge et al., 1997;Syrjänen, 2002;Guo et al., 2012;Nasrollahzadeh et al., 2015;Antonsson et al., 2016). Among countries with high EC incidence are China, Singapore, Iran, South Africa and Brazil (Bey et al., 1976;Han et al., 1996;Syrjänen, 2002;Guo et al., 2012). Due to the high number of EC cases in Central Asia (100/100,000) in relation to North America and Western Europe (5-10/100,000), the region is known as the Asian Esophageal Cancer Belt (Nasrollahzadeh et al., 2015). Although the reason for EC incidence variation is unknown (Syrjänen, 2002;Nasrollahzadeh et al., 2015), studies indicate the contribution of environmental factors, in addition to infectious agents, in the oncogenic process (Chang et al., 1992;Antonsson et al., 2010).
Smoking and alcohol consumption are pointed out as etiological factors for human EC (Han et al., 1996;Lagergren et al., 1999;Syrjänen, 2002;Haster and Owyang, 2013). However, due to sociocultural reasons, smoking does not justify the high incidence of EC in Central Asia (Nasrollahzadeh et al., 2015), reinforcing the contribution of infectious agent in esophageal carcinogenesis, and evi-dence shows the etiopathogenic contribution of different infectious agents in EC, such as cytomegalovirus (CMV), Epstein-Barr virus (EBV), herpes simplex and HPV (Syrjänen, 2002).
Although the association between HPV and premalignant cervical lesions is known since the 1970s (Syrjänen, 2002), the participation of this virus in EC had its history marked by controversial reports in the literature (Lavergne and de Villiers, 1999). The association between HPV and EC was first proposed by Syrjänen (1982). On the one hand, HPV protein expression was observed in samples by IHC (Winkler et al., 1985;Hille et al., 1986;Kulski et al., 1986) and viral DNA sequences were identified by CISH in EC (Chang et al., 1993;Togawa et al., 1994;Cooper et al., 1995),. On the other hand, serological studies contested these results (Dillner et al., 1995;Han et al., 1996;Lagergren et al., 1999). However, despite these controversial results, the association between HPV and EC is currently accepted, being discussed in the 18 th edition of Harrison's Principles of Internal Medicine (Kahrilas and Hirano ,2013). Thus, the absence of HPV detection by serological methods reflects a problem still observed nowadays: the use of antibodies against late (L1) proteins (Dillner et al., 1995;Lagergren et al., 1999). However, considering that EC does not present an HPV productive infection, the late protein expression is not expected. This occurs because HPV presents a "hit and run" mechanism (Han et al., 1996;Bjørge et al., 1997).
Not only serological studies are controversial. Studies based on PCR also show a low correlation between HPV and EC, mainly in clinical samples from Australia (Antonsson et al., 2010;Antonsson et al., 2016). These results can be attributed to the origin of clinical samples, since Australia is not considered a high-risk area for HPVassociated EC. In addition, problems involving primer sensitivity are extensively discussed in the literature of both BPV and HPV (Lavergne and de Villiers, 1999;Silva et al., 2013b;de Villiers, 2013;Araldi et al., 2014b). However, evidence of PVs etiopathogenic action in EC has accumulated along the last 20 years. Among these are: (1) the presence of koilocytes in EC biopsies (Bjørge et al., 1997;Syrjänen, 2002;Vieira et al., 2013), (2) loss or mutation in p53 as a consequence of the PV E6 oncoprotein (Chang et al., 1994(Chang et al., , 1995Herbster et al., 2012), (3) increase in telomerase activity (Syrjänen, 2002), and detection of DNA sequences of HPV-6, 11, 18, 31 and 33 in EC (Vieira et al., 2013).

Prophylactic and therapeutic vaccines against BPV and HPV
Currently, there are few forms of treatment against BP available (Muro et al., 2008). Among the possibilities is the papilloma surgical excision (Muro et al., 2008). Although frequently employed, this method is inefficient in cattle with high incidence of BPV, because it is impracticable to perform the excision of papillomas of all cattle. Another strategy frequently employed is self-hemotherapy (Leto et al., 2011). This method consists in the removal and intramuscular reinjection of a volume of 10 mL of venous blood, inducing a nonspecific immune stimulus that can promote the "shedding of the warts" (Leto et al., 2011). However, this technique does not avoid BP reoccurrence, thus being a palliative method. Another possibility to reduce the incidence of BP is to control ectoparasite populations, since it was demonstrated that the biological control of ticks reduces the incidence of BPV (William et al., 1992).
Few medical interventions proposed in the last century can match the effects that immunization exerts on longevity (Schuchat and Jackson, 2013). For this reason, vaccination is considered the best form of prevention, control and eradication of viral etiology diseases (Ribeiro-Müller and Müller, 2014). Moreover, immunization reduces both transmission and dissemination of the infectious agent (Schuchat and Jackson, 2013).
Two prophylactic vaccines against HPV are available in the market since 2006: (1) Cervarix, produced by Glaxo-Smith Klein (GSK) and (2) Gardasil, produced by Merck (Ribeiro-Müller and Müller, 2014). These vaccines are based on virus-like particles (VLPs) of the L1 structural protein (Marigliani et al., 2012). Cervarix is a bivalent vaccine able to confer protection against HPV-16 and 18, employing L1 VLPs produced in Baculovirus in Trichoplusia ni insect cells, using aluminum hydrophosphate as adjuvant (Ribeiro-Müller and Müller, 2014). Gardasil is able to confer protection against two high-risk HPV types (HPV-16 and 18) and two low-risk ones (HPV-6 and 11), associated to genital warts. This vaccine is composed by of L1 VLPs produced in Saccharomyces cerevisae, employing lipid A 3-O-diacilete-44-monophosphoryl (ASO4) as adjuvant. . Both vaccines are considered safe and well tolerated (Ribeiro-Müller and Müller, 2014). For these reasons, more than 30 countries, including Brazil, adopted immunization programs against HPV based on these vaccines.
Australia, the first country to adopt the vaccination against HPV, observed a reduction of 70% in HPV-6, 11, 16 and 18 infection incidence. Similar results were also verified in Denmark, Finland and Sweden (Ribeiro-Müller and Müller, 2014). However, available vaccines are not able to protect against all HPV types, since there are more than 200 described. Moreover, these vaccines have a high cost of production. Thus, it is necessary to invest in novel multivalent vaccines, with lower production cost. Vaccines based on recombinant protein expressed in Escherichia coli have demonstrated to be a useful alternative, because they have a lower cost, and not requiring the L1 VLPs, they are more stable (Ribeiro-Müller and Müller, 2014).
Although there are two vaccines against HPV, there is no vaccine yet against BPV available to date. The idea of developing a vaccine to combat BPV began with the infection of Shope papillomas extract in the 1940 decade (Shope, 1937). Since then, different vaccine models were proposed (Jarret et al., 1991;Gaukroger et al., 1996;Góes et al., 2008;Love et al., 2012;Mazzuchelli-de-Souza et al., 2013). However, none of them became a commercial product. This denotes the notorious difficulty to obtain a safe and efficient vaccine, and reflects the reduced number of research groups dedicated to develop a vaccine against BPV.
Studies have demonstrated that BPV early proteins (E6 and E7) show a therapeutic action, while later proteins (L1 and L2) have a prophylactic action (Campo, 1997b). Vaccines based on E6 and E7 have also been discussed against HPV (Borysiewicz et al., 1996;He et al., 2000;Yao et al., 2013), reinforcing the usefulness of the BPV model not only for comprehending the pathogenic mechanisms of HPV, but also for vaccine biotechnology. However, in a current study based on BPV-1 E6 recombinant oncoprotein, we demonstrated that this oncoprotein is able to induce clastogenesis and neosis per se (Araldi et al., 2015a). This data emphasizes the necessity of an in silico analysis of E6 and E7 oncoproteins, aiming to obtain novel variants that are more antigenic and less mutagenic. In order to guarantee the immunization and safety of products, our group is now focusing on the development of a prophylactic vaccine based on L1 recombinant protein (Módolo et al., data not published).

Conclusions
Papillomaviridae comprises the most extensive known family of viruses, able to infect all vertebrates including humans, in which it is responsible for 27-30% of all infectious agent-associated incident cancer cases. Moreover, PVs represent an important problem in the veterinary field, inducing papillomas in dogs, felines and cattle. Also, BPV infects equines, resulting in sarcoids. Although novel discoveries contributed to the understanding of the PVs oncogenic role, some carcinogenic mechanisms remain unknown, especially those following cancer initiation . In addition, current studies have collected evidence of BPV productive infection in sites earlier considered as not permissive, such as PBMCs Melo et al., 2015), placenta (Roperto et al,. 2012) and primary cell cultures (Campos et al., 2008;Campos et al., 2013). Similar results have also been described for HPV (Chiou et al., 2003;Foresta et al., 2013;Pessoa, 2014). However, despite of these advances, the natural history of PVs remains dependent on cell differentiation, emphasizing the need to review the replication cycle of these viruses (White and Howley, 2013;Munday, 2014). In addition, the available vaccines against HPV do not confer protection against all virus types, providing only a limited protection. The veterinary field lives a most dramatic scene, since there is no vaccine available against BPV. Over the last 30 years, our group has dedicated efforts to elucidate the pathogenic mechanisms of PVs, focusing on BPV, once the virus is considered a prototype for HPV studies. Although our contributions brought important advances, more studies are necessary to propose efficacious and safe prophylactic and therapeutic measures.

Dedication
This paper is dedicated to Dr. Maria Luiza Beçak on occasion of her 80 th birthday. She and her husband were the founders of the Laboratory of Genetics of the Instituto Butantan, in 1961. They were Brazil's pioneers in introducing human and vertebrate cytogenetics and the first to use cytogenetics as an important tool in human genetic counseling. Dr. Beçak's contribution in classic and recent papers in the genetic literature is very important and remarkable.