Proteomic profiles and cytokeratin 13 as a potential biomarker of Ovis aries Papillomavirus 4 3-positive and negative cutaneous squamous cell carcinomas

25 Ovis aries papillomavirus 3 (OaPV3) is an epidermotropic PV reported in sheep cutaneous 26 squamous cell carcinoma (SCC). The presence of OaPV3 DNA and its transcriptional activity in 27 cutaneous SCC, as well as its in vitro transforming properties, suggest a viral etiology for this 28 neoplasm. Nevertheless, the reactome associated with viral-host interaction is still unexplored. 29 Here, we investigated and compared the proteomic profiles of OaPV3-positive SCCs, 30 OaPV3-negative SCCs, and non-SCC samples by liquid chromatography tandem-mass 31 spectrometry (LC-MS/MS) analysis, bioinformatics tools, and immunohistochemistry (IHC). 32 OaPV3-positive SCCs (n = 3), OaPV3-negative SCCs (n = 3), and non-SCCs samples (n = 3) were 33 subjected to a shotgun proteomic analysis workflow to assess protein abundance differences among 34 the three sample classes. Proteins involved in epithelial cell differentiation, extracellular matrix 35 organization, and apoptotic signaling showed different abundances in OaPV3-positive SCCs tissues 36 (P ≤ 0.05) when compared to the other tissues. Cytokeratin 13 (CK 13) was among the most 37 increased proteins in OaPV3-positive SCC and was validated by immunohistochemistry on 10 38 samples per class, confirming its potential as a biomarker of OaPV3 infection in SCC. 39 Collectively, results provide a preliminary insight into the reactome associated with viral-40 host interaction and pave the way to the development of specific biomarkers for viral-induced sheep 41 SCC. 42


Introduction
Squamous cell carcinoma (SCC) is a malignant tumor arising from the squamous epithelium of the skin and mucous membranes, widely reported in domestic animals and represents the most common form of skin tumor in sheep (Alberti et al., 2010;Tore et al., 2017;Vitiello et al., 2017;Goldschmidt et al., 2017).
Papillomavirus infection, along with several environmental risk factors, such as prolonged exposure to ultraviolet radiation of poorly pigmented skin, has been claimed to act as a major factor contributing to SCC development (Alberti et al., 2010;Ahmed et al., 2015).PVs are a large group of small, non-enveloped, double-stranded DNA viruses that infect skin and mucosae causing proliferative and neoplastic lesions in domestic and wild vertebrate species (Munday et al., 2010;Lange et al., 2011;Rector et al., 2013;Sardon et al., 2015;de Villiers et al., 2017;Lecis et al., 2020).
Although the putative etiological role of ovine PVs both in cutaneous fibropapilloma and SCC has not yet been fully elucidated, it has been shown that OaPV3 expresses the typical E6/E7 oncogenes, and that OaPV3 E7 binds the retinoblastoma tumor suppressor protein (pRb) much more efficiently than fibropapilloma-related papillomaviruses, similarl to what has been observed in high and lowrisk human PVs (Alberti et al., 2010;Tore et al., 2019).Additionally, the presence of both OaPV3 DNA and transcripts in SCCs, as well as its in vitro transforming properties, point towards a contribution of OaPV3 to tumor development (Alberti et al., 2010;Vitiello et al., 2017;Tore et al., 2019).
Proteomic profiling technologies open the way to the discovery of novel biomarkers with potential as sensitive and specific molecular tools in cancer research (Mabert et al., 2014).The identification of altered proteins occurring in the oncogenesis process, as well as their qualitative and quantitative characterization, can offer valuable information relating to more effective diagnosis, prognosis, and response to therapy.Thus, the application of proteomics to ovine SCC appears particularly interesting in order to map the biological processes associated with viral infection, in which OaPV3 could play a pivotal role.Furthermore, characterizing the proteins involved in virus-host interactions can contribute to the identification of candidate biomarkers indicating viral activity.Based on these considerations, the aim of our study was to discover and validate OaPV3 infection-related proteins in ovine cutaneous SCC by proteomics and immunohistochemistry (IHC), in order to elucidate the pathways involved in viral neoplastic transformation and to identify potential viral biomarkers.

Origin of the samples
This study included a total of 30 archival tissue samples, of which 10 were OaPV3-positive SCCs, 10 were OaPV3-negative SCCs, and 10 were non-SCC samples collected from the udder (n = 17) and the head (n = 13) of 30 Sarda breed sheep.All tissues belonged to a previous sampling study (Vitiello et al., 2017) in which specimens were divided in two aliquots, one of which was formalinfixed, paraffin-embedded (FFPE), and used for histological evaluation and immunohistochemistry, while the other was frozen at -80°C for later proteomic analyses.In that study (Vitiello et al., 2017) the presence of OaPV3 in sheep SCC was assessed by conventional PCR, and its cellular localization and transcriptional activity were evaluated by ISH and RT-PCR.All SCCs were classified as moderately differentiated (except 1 classified as poorly differentiated) according to the modified Anneroth's multifactorial histological grading system by two experienced (EA, SP) and one board-certified pathologist (GPB) (Vitiello et al., 2017).
Experiment permission was not required from the University's Animal Care Ethics Committee because all the samples were retrieved from the abattoir.

Protein extraction, in-gel digestion, shotgun analysis, and protein identification
For proteomic analysis, frozen udder tissue samples from 3 moderately differentiated OaPV3-positive SCCs, 3 moderately differentiated OaPV3-negative SCCs, and 3 non-SCCs were selected among the 30 archival samples listed in the previous paragraph and characterized in our previous study (Vitiello et al., 2017).For protein extraction, the tissue replicates stored at -80°C were included in Optimal Cutting Temperature medium (Tissue-Tek, Sakura Finetek, Torrance, CA, USA), cut into 20 serial cryosections (Leica CM 1950, Heidelberg, Germany) at 10-µm, and collected in a 1.5-ml sterile tube.In parallel, serial cryostat sections (3-µm thick) from the same tissue were histologically evaluated in order to confirm their classification and to ensure that the lesions were present in the portion of the tissue subjected to proteomic analysis, as well as to reassess non-SCC tissues.Total proteins were extracted from each tissue by incubating the respective 10-µm cryosections in 200 µl of lysis buffer containing 2% sodium dodecyl sulphate (SDS), 0.4 % Tween-20, 130 mM dithiothreitol (DTT), 500 mM Tris HCl (pH 8.8) plus SIGMAFAST™ Protease Inhibitors (Sigma, St. Louis, MO, USA) at the concentration recommended by the manufacturers, at 300 rpm for 15 min at 95 °C using a Thermomixer comfort (Eppendorf, Hamburg, Germany).After centrifugation at 10.000 x g for 10 min at 4° C, supernatants were quantified with the PierceTM 660 nm Protein Assay (Thermo Scientific).Protein extracts were processed to obtain peptide mixtures by means of the Filter Aided Sample Preparation (FASP) as previously described, starting from 100 μg of protein extract (Wisniewski et al., 2009;Tanca et al., 2013).Then, liquid chromatographytandem mass spectrometry (LC-MS/MS) analysis of tryptic digests was performed on a Q-TOF hybrid mass spectrometer with a nano lock Z spray source, coupled on-line with a NanoAcquity chromatography system (Waters).Each peptide mixture was analyzed in duplicate and each sample was first concentrated, washed with an enrichment column, fractionated over a 250 min gradient on a C18 reverse-phase column and then analyzed by a data-dependent MS/MS mode as described previously (Ghisaura et al., 2019).Raw files were processed by ProteinLynx software (Version 2.2.5) to produce the peak lists as pkl files.All pkl files were first converted into MGF files, and subsequently, Proteome Discoverer software (version 1.4; Thermo Scientific) was used for protein identification.Two technical replicates were analyzed as merge to generate a unique list of proteins for each biological sample using a workflow assembling by different nodes: Sequest-HT as a search (peptide confidence: q-value < 0.01) (Kall et al., 2007;The et al., 2016).Peptide and protein grouping according to the Proteome Discoverer's algorithm were allowed, applying the strict maximum parsimony principle.

Label-free quantitation and data analysis
Spectral counts (SpC) were used to estimate protein abundances and to compare the abundance of the same proteins between different sample groups (Addis et al., 2011;Pisanu et al., 2011;Tanca et al., 2013).The Normalized Spectral Abundance Factor (NSAF) of proteins was used to express their relative abundance and the SpC log Ratio (RSC) to express the log fold change of proteins between different experimental groups (Old et al., 2005;Zybailov et al., 2006).Proteins identified with less than one SpC in at least one replicate or with fewer than two SpCs in more than one replicate were excluded from the differential analysis, in order to increase the accuracy of the analysis (Addis et al., 2011;Addis et al., 2013).Among proteins identified in the databases of Bos taurus, Ovis aries, and Capra hircus proteins, only those with the highest number of peptides and PSMs were considered.

Functional analysis of differential proteins
All proteins showing statistically significant differences in OaPV3-positive SCCs, OaPV3-negative SCCs, and non-SCC tissues were subjected to pathway analysis based on the Gene Ontology (GO) database (biological processes (BP), molecular functions (MF), and cellular components (CC) ) and STRING (Szklarczyk et al., 2015;Pisanu et al., 2018).To enable pathway analysis, the UniProt codes for Bos taurus, Ovis aries, and Capra hircus were replaced with the UniProt codes for the closest human protein equivalent by sequence alignment of identified peptides with human sequences using Basic Local Alignment Search Tool (BLAST).

Evaluation of immunohistochemical data
The extent of CK 13 immunopositivity was evaluated by considering the cytoplasmic signals of the malignant squamous cells or non-neoplastic skin cells.Immunoreactivity was semiquantitatively scored considering the number of positive cells in 10 HPF (grade 0: no positive cells; 1: < 10%; 2: 11-30%; 3: 31-60%; 4: > 60%) and the intensity of staining (weak: 1; moderate: 2; strong: 3).Then, a combined immunoreactivity score (IRS) ranging from 1 to 12 was calculated for each specimen.Tissues were imaged using Nikon Eclipse 80i and digital computer images were recorded with a Nikon Ds-fi1 camera.

Statistical analysis
All the statistical analyses, including descriptive statistics, were performed using Stata 11.2 software (StataCorp LP), with statistical significance set as P ≤ 0.05.To evaluate differentially abundant proteins (Rsc) among the experimental groups, we applied the beta-binomial test (Pham et al., 2010).Only proteins with Rsc ≥ 1.5 or ≤ −1.5 and with a P-value ≤ 0.05 obtained by the betabinomial test were considered significant in the comparison between OaPV3-positive SCCs, OaPV3-negative SCCs, and non-SCC samples.To evaluate the differences between the principal biological processes obtained by STRING, we used NSAF values and Student t-test after checking the normality with the Shapiro-Wilk test.Abundances of the different biological processes were calculated by the sum of the NSAF values for each protein associated to a biological process (Addis et al., 2011).

Proteomic analysis of SCC and non-SCC sheep tissue samples
A total of 476 proteins were successfully identified (Sheet 1 "1.All identified proteins", Supplementary File).Of these, 242 were eligible for the differential analysis between OaPV3positive SCCs, OaPV3-negative SCCs, and non-SCC tissues, by label-free quantitative proteomics.
Differential proteomics results are reported in Table 1.

Biological and functional pathways involving differential proteins
The proteins showing statistically significant changes belonged to numerous biological pathways related to the disease.The most relevant ones were response to stress (RS, 28 proteins), regulation of apoptotic signalling pathways (RASP, 7 proteins), negative regulation of apoptotic process and cell death (NRAP, 10 proteins), tissue development (TD, 18 proteins), epithelial cell differentiation (ECD, 8 proteins), extracellular matrix disassembly (EMD, 5 proteins) and organization (EMO, 9 proteins), and glycosaminoglycan catabolic process (GCP, 4 proteins).
Differential proteins belonging to these pathways and significant in at least one sample group (bold type) are indicated in Table 1 with asterisks in each respective pathway column.Detailed information is reported in Supplementary File, Sheet 5 "5.Biological Process_STRING" and sheet 6 "6.Table with Gene Ontology".
The biological pathways most represented in each SCC type and in non-SCC tissues were then assessed by considering the total abundance of all the proteins belonging to each pathway within the different sample groups (Fig. 1).As a result, OaPV3-positive SCCs had a significantly higher abundance of proteins belonging to the functional classes RASP, NRAP, and NRCD when compared to both OaPV3-negative SCC and non-SCC tissues.OaPV3-positive SCCs also contained higher amounts of proteins belonging to the functional class RS when compared to non-SCC tissues.On the other hand, OaPV3-positive SCCs had lower amounts of proteins belonging to TD and ECD when compared to both OaPV3-negative SCC and non-SCC tissues.OaPV3-positive SCCs did also have lower amounts of proteins belonging to EMD and EMO when compared to non-SCC tissues.The proteins participating in each pathway are indicated in Table 1 with the respective abbreviations, and are detailed in the Supplementary file, Sheet 5 "Biological Process_STRING".Protein abundance values are detailed in Sheet 7 "7.NSAF".
Reactome analysis was also carried out on all differential proteins to highlight common pathways involved in the development of SCC.The resulting protein network is reported in Fig. 2.
The most significantly represented pathway was "neutrophil degranulation", followed by "innate immune system".Of interest in the context of the disease was the significant involvement of "formation of the cornified envelope", "degradation of the extracellular matrix", as well as "Chk1/Chk2(Cds1) mediated inactivation of "Cyclin B:Cdk1 complex".Results are detailed in Sheet 8 "8.Reactome_STRING", Supplementary file.

Immunohistochemistry
Among differentially expressed proteins, CK 13 was higher in OaPV3-positive SCCs when compared to either OaPV3-negative SCCs or non-SCC tissues, in both cases with high RSC values.

Discussion
Proteomic analysis appears particularly interesting in cancer research, as both the study of tumor cell biology and the identification of altered cellular processes and of the specific proteins involved in these pathways represent key tools for investigating tumors (Srivastava et al., 2018).
Investigating the reactome of diseased animals does also provide new insights in veterinary medicine, helping to clarify molecular mechanisms dictating initiation and progression of different conditions, and allowing the identification of specific biomarkers that may be useful for establishing effective intervention and treatment control actions (Ceciliani et al., 2014;Lippolis et al., 2016).
Accordingly, when looking at the functional pathways involving the differential proteins identified in this study and at their reactome, a significantly higher abundance of proteins belonging to regulation of apoptotic signalling pathways, negative regulation of apoptotic process, and negative regulation of cell death, was found in OaPV3-positive SCC.These pathways are involved in SCC development and were related to the presence or absence of OaPV3 infection.On the other hand, other pathways associated with normal tissue and extracellular matrix organization were more abundant in non-SCC tissues, followed by OaPV3-negative SCC.
When investigating the reactome, the most significantly altered pathways in the investigated SCC tissues were related to neutrophil degranulation and the innate immune system as well as platelet degranulation and activation and, more in general, to the hemostasis system.It seems plausible that this might be related to the inflammatory reaction due to the high frequency of SCC ulceration, as observed in our cases, and frequently described also in goats (Gibbons et al., 2015).
Furthermore, the role of tumor-associated neutrophils (TANs) has gained attention in cancer and has been linked to the overall survival of human patients with both oesophageal and head and neck squamous cell carcinoma (Shaul et al., 2019).
Concerning differential proteins, several were involved in SCC development and were related to the presence or absence of OaPV3 infection.Among proteins significantly higher in OaPV3-positive SCCs when compared to non-neoplastic samples, transgelin-2, annexin, pyruvate kinase, alpha-1-acid glycoprotein have been reported as key factors for the progression of human SCC (Croce et al., 2001;Calmon et al., 2013;Meng et al., 2017;Kurihara-Shimomura et al., 2018).Also, previous studies reported Annexin A1 overexpression in penile carcinomas positive for highrisk HPVs (Calmon et al., 2013).
Additionally, 14-3-3 proteins theta and zeta/delta (Table 1) were more abundant in SCCs harboring OaPv3 compared to non-neoplastic samples.The 14-3-3 proteins comprise a large family of highly conserved phosphoserine/threonine-binding proteins related to intracellular signaling, apoptosis signal transduction, and cell cycle regulatory pathways, and being also negative regulators of cell death and cellular senescence (van Hemert et al., 2001).Furthermore, their aberrations are involved in cellular transformation and tumorigenesis, due to their role as oncoproteins and tumor suppressor proteins regulators (Morrison, 2009;Pennington et al., 2018).Interestingly, similarly to what observed in OaPV3-positive SCC, 14-3-3 zeta protein increases in high-risk human papillomavirus (HPV)-related cervical cancer, providing further evidence of the relationship between PVs and this family of phosphoserine/threonine-binding proteins (Boon et al., 2013).
Considering the specific interaction of the 14-3-3 zeta protein with the phosphorylated PDZ binding motif (PBM) of the E6 viral protein reported in cervical cancer, it seems plausible that the higher abundance of 14-3-3 proteins in OaPV3-positive SCCs might be exclusively related to the viral presence and that it may contribute to maintaining high levels of OaPV3 E6 protein (Boon et al., 2013).Future studies involving the dissection of the role of E6 and of its possible relation and effects on 14-3-3 activity, as well as the 14-3-3 downstream pathway including, for example, the AKT and P53 pathway, should provide fascinating insights into the function of the E6 OaPV3 oncoprotein, considering also the unsolved p53 deregulation in ovine SCC cell proliferation (Tore et al., 2019).
Among the proteins decreased in OaPV3-positive SCCs when compared with the nonneoplastic tissue we found filaggrin 2, a protein involved in maintaining cell-cell adhesion and in the protection against UVB light, and transgelin-2, a protein that regulates actin cytoskeleton through actin binding and involved in cytoskeletal remodelling.Both proteins have been linked to human HPV-related cancer (Skaaby et al., 2014;Meng et al., 2017;Yang et al., 2019).Of interest is the observed overall decrease of galectins, including galectin-3, a β galactoside-binding protein, involved in tumor growth, progression, and metastasis, and considered a potential target to prevent cancer metastasis (Ahmed et al., 2015).Similar to what observed in OaPV3-positive SCC, a decrease in galectin-3 was reported in human squamous and basal cell carcinomas (Kapucuoglu et al., 20019).Among the 17 proteins significantly higher in OaPV3-positive SCCs when compared to OaPV3-negative SCCs, prosaposin, a lysosomal compartmental protein involved in catabolism of sphingolipids with small sugar chains, has been validated as oesophageal squamous cell carcinoma biomarker (Pawar et al., 2011).
Another interesting observation in the comparison of OaPV3-positive SCCs vs OaPV3negative SCCs was the decreased abundance of proteins involved in epithelial cell differentiation and tissue development, such as type I and II cytokeratins (CKs), including CK 3, CK 6 and 15.This finding could be related to viral infection.Indeed, in our study, the markers for epidermal differentiation CK 3 and CK 6 were less abundant in OaPV3-positive SCC and more abundant in OaPV3-negative SCC, similarly to what reported in human cutaneous squamous cell carcinomas (Moll et al., 2008;Mommers et al., 2000).Nevertheless, in vitro studies have shown that the E1^E4 PV proteins are able to specifically bind CKs by a DEAD-box protein-mediated interaction, inducing the collapse of the cell cytoskeletal network (Raj et al., 2004).Likewise, our results support the deregulation of CK 6 induced by viral infection, and suggest an active role of OaPV3 in tissue development and epithelial cell differentiation.
Conversely, the increase in abundance of CK 13, a marker of epithelial differentiation for non-keratinizing epithelium such as the esophagus, appeared ambiguous considering also the observed decreased in the other proteins involved in epithelial cell differentiation and tissue development, as galectin-7 and transgelin-2 ( Lam et al., 1995).This finding appears of interest since no altered levels of CK 13 were observed in OaPV3-negative SCC and in non-neoplastic tissues compared to OaPV3-positive SCC samples, suggesting that the increased abundance of this protein might be tightly related to viral infection.Interestingly, Hudson and co-workers reported overexpression of this protein in human cutaneous SCC (Hudson et al., 2010).Our hypothesis is further reinforced by the IHC results, with a strong (60% of cases) and diffuse expression of CK 13 in OaPV3-positive malignant squamous cells, while no signal was detected in 100% of nonneoplastic tissue and in most (80%) of OaPV3-negative SCC.
Nevertheless, our data conflicted with previous reports showing the ability of HPV 16 E1^E4 to bind CKs, inducing the collapse of the cell cytoskeletal network (Raj et al., 2004).
Overall, it seems conceivable that the decrease in CK 6 and CK 3 may activate the well-known mechanism of cytokeratins compensation in which group I CK 13 is expressed in order to compensate the absence of group II CKs (CK 6 and 3) (Kanaji et al., 2007).In particular, type I and type II CKs have been shown to form obligate 1:1 heteropolymers, suggesting that dynamic changes must occur in their expression levels, particularly when one CK is suppressed.
A limitation of this study could be related to sampling size and storage conditions.In particular, a larger number of samples need to be systematically analyzed in order to make and further validate general assumptions.Moreover, storage conditions could have nuanced the proteomic features of virus-induced SCC.However, the storage condition used in this study (freezing at -80°C immediately after sampling) is commonly used in proteomics studies (Tanca et al., 2012).In addition, since both negative and positive samples underwent the same storage process, we hypothesize that the differential proteomic results are reliable and are significantly descriptive of the differences among OaPV3 positive, OaPV3 negative SCC, and non-SCC tissues.

Conclusions
To the best of our knowledge, this is the first study applying a comprehensive proteomic approach for investigating the deregulation of proteins related to viral infection in one of the most common ovine tumors, SCC.The altered biological processes as well as the SCC associatedreactome might be related to viral pathogenesis pathways, suggesting that OaPV3 can represent a driving force in neoplastic transformation, as proposed for several papillomavirus-related tumors.
The identification of altered molecular pathways involved in cell cycle and apoptosis, frequently reported in the literature as related to viral activity, envisages a specific virus-host interaction in which OaPV3 may favor malignant transformation.Considered together, our findings support a role of OaPV3 in the progression of cutaneous squamous cell carcinomas and recognized CK 13 as a promising putative biomarker of OaPV3 infection in ovine cutaneous SCCs, especially useful when the virus is undetectable in the tumor, and considering that OaPV3 specific antibodies suitable for IHC have not yet been developed.
Differences in abundance are reported as fold changes in the respective sample comparisons.Statistically significant differential values are in bold: Rsc ≤ −1.5 or ≥ 1.5, beta-binomial test P ≤ 0.05.Protein identities obtained upon BLAST search.Response to stress (RS), regulation of apoptotic signaling pathway (RASP), negative regulation of apoptotic process (NRAP), negative regulation of cell death (NRCD), tissue development (TD), epithelial cell differentiation (ECD), extracellular matrix disassembly (EMD), extracellular matrix organization (EMO), glycosaminoglycan catabolic process (GCP) engine (Protein Database: database homemade composed by concatenation of different databases obtained by UniProtKB; Taxonomy: Bos taurus, Ovis aries and Capra hircus sequences from SwissProt and Papillomaviridae sequences from TrEMBL; Enzyme: Trypsin; Maximum missed cleavage sites: 2; Precursor mass tolerance: 50 ppm; Fragment mass tolerance: 0.4 Da; Static modification: cysteine carbamidomethylation; Dynamic modification: N-terminal Glutamine conversion to Pyro-glutamic acid and methionine oxidation), and Percolator for peptide validation

Fig. 1 .
Fig. 1.Comparative abundance of biological processes related to the differential proteins observed