Nineteen fibropapilloma samples were obtained from 19 bovines (T = 19), aged between 7 months to 3 years and were clinically identified according to their gross pathology, consisting of a cauliflower appearance. The lesions were located mainly on the head, lateral neck and shoulders, with a diameter ranging from 0.5 to 3 cm. All samples included in this study were obtained from a public slaughterhouse. In the present study, we used the same tumor samples that were employed in the previous work of Bocaneti Daraban et al. (26) and all tumor samples were known to be BPV-2 positive (26). Additionally, four normal skin samples (NS = 4) were obtained from healthy bovines, harvested from auricular conchae inner skin and further tested separately to detect BPV-1/-2 DNA, following the same protocol as mentioned above; all 4 skin samples resulted to be BPV-1/-2 negative. Tissue samples were fixed in formalin for routine histologic diagnosis and immunohistochemical analysis. Moreover, in order to prevent cross-contamination, for each paraffin-embedded tissue sample, 20 μm thick sections were cut with separate microtome blades and stored in sterile tube for molecular biology analysis. Moreover, samples T1 - T9 and NS1-NS4 were appropriately processed and frozen at −80 °C for western blotting and zymography analysis.

Fibropapilloma samples were 10 % formalin fixed paraffin-embedded for routine histological processing and stained with haematoxylin and eosin for light microscopy evaluation, according to the guideline proposed by Goldschmidt and Goldschmidt (27).

Paraffin sections of 19 fibropapillomas and 4 normal skins were dewaxed in xylene, dehydrated in graded alcohols and washed in 0.01 M phosphate-buffered saline (PBS), pH 7.2–7.4. Endogenous peroxidase was blocked with hydrogen peroxide 0.3 % in absolute methanol for 20 min at room temperature (RT). The immunohistochemical (IHC) protocol (streptavidin biotin- peroxidase method - Novolink Polymer Detection System; Leica Biosystems, NewCastle, United Kindom) was described by the authors in a previous study (23). Primary antibodies used in this study are listed in Table 1. The antibodies anti-MM-2, anti-MMP-7, anti-MMP-9 and anti-MMP-14 were applied for 1 h at RT, while the antibodies anti-TIMP-1 and anti-TIMP-2 were applied overnight at 4 °C. The treatment with diaminobenzidine was used to visualize the specific immunoreactivity. The immunolabelling procedure included negative control sections, where primary antibodies were omitted and incubated with PBS instead. Canine mammary carcinoma samples were employed as positive control, since were demonstrated to express these MMPs and TIMPs (28).

MMPs and TIMPs immunoreactivity were scored as previously described by Martano et al. (15): –, negative; +, weak, individual positive cells; ++, foci of moderate positivity; +++, > 50% of cells moderately to strongly positive. The immunoreactivity was scored by two observers (FDB and GB) under blinded conditions.

Biochemical analysis was performed on four normal skin samples (NS1–NS4) and nine fibropapilloma samples (T1-T9); tissue biopsies were each cut with separate, disposable scalpel blades and brand-new steel beads were employed for mechanical homogenation. Protein extraction, electrophoresis and blotting were performed as previously described by Altamura et al. (29). The membranes were blocked with 5% bovine serum albumin (BSA) in TBS-0.1% Tween buffer (10 mM Tris–HCl, pH 7.4, 165 mM NaCl, 0.1% Tween) at RT, washed with TBS-0.1% Tween, and incubated with the following primary antibodies: anti-MMP-2 (1:500), anti-MMP-7 (1:1,000), anti-MMP-9 (1:1,000), anti-MMP-14 (1:1,000), anti-TIMP-1 (1:500) and anti-TIMP-2 (1:500). After appropriate washing steps in TBS-0.1% Tween buffer, goat anti-mouse (GE Healthcare #LNA931V/AH) and donkey anti-rabbit (GE Healthcare #LNA934V/AH) secondary antibodies conjugated with horseradish peroxidase were applied for 1 h at RT, and protein bands were visualized by enhanced chemiluminescence (ECL, Bio-Rad) using ChemiDoc gel scanner (Bio-Rad). The blots were stripped and reprobed against mouse anti-β-actin antibody (C-2: sc-8432, Santa Cruz) at 1:500 dilution to ensure comparable amounts of proteins for each sample. Densitometric analysis for protein quantization was achieved by using Image Lab software (Bio-Rad). The protein concentrations were normalized to the β-actin levels and expressed as the densitometric ratio.

The enzymatic activity of MMP-2 and MMP-9 was analyzed by gelatin zymography, with 7.5 % polyacrylamide gel containing 0.1% gelatine (Sigma Aldrich) as substrate. Twenty μg of each protein extract sample were diluted 1:1 in non-reducing loading buffer (Bio-Rad) and electrophoresed in non-reducing electrophoresis conditions at 120 volts for 1 h and 30 min. After electrophoresis, two washing steps of 30 min each were performed in zymogram developing buffer (2.5% Triton X-100, 50 mM Tris-HCl, pH 7.5, 5 mM CaCl₂, 1 μM ZnCl₂). Further, the gel was incubated with zymogram incubation buffer containing co-factors necessary for the gelatinase reaction (1% Triton X-100, 50 mM Tris-HCl pH 7.5, 5 mM CaCl₂, 1 μM ZnCl₂) overnight at 37 °C and then stained with staining buffer (40% methanol, 10 % acetic acid, 50 ml deionized water, 0.5 g Commasie Blue) for 1 h on a rotary shaker at RT. The gel was destained with destaining solution (40 % methanol, 10 % acetic acid, 500 ml deionized water), until the bands were clearly visible and digital images were acquired using ChemiDoc gel scanner (Bio-Rad Laboratories). Densitometric analysis was performed as described above.

For statistical analysis, Student's t-test was performed using SPSS 17.0 software (SPSS Inc.), and differences were considered to be statistically significant for ^*P < 0.05 or ^**P < 0.01.

The main lesional characteristics consisted in mild hyperkeratosis, epithelial and dermal hyperplasia, accompanied by rete pegs (Figure 1A). Numerous cells found in the granular epithelial layer displayed keratohyalin granules and cytoplasmic vacuolation was evident in some cells. Moreover, koilocytes were also observed, being suggestive of papillomavirus infection (Figure 1B). Therefore, histological features were consistent with a diagnosis of cutaneous fibropapillomas.

The expression patterns of MMP-2, MMP-7, MMP-9, MMP-14, TIMP-1 and TIMP-2 in 19 bovine cutaneous fibropapillomas and 4 normal skin samples are summarized in Table 2.

In 4/4 normal skin samples, MMP-2 was faintly expressed in basal, spinosum and granular epithelial layers, consisting in a fine light brown granular cytoplasmic and perinuclear pattern (Figure 2A). Four out of 19 (21 %) tumor samples did not show any specific immunoreactivity, while in 15/19 (79 %) fibropapillomas, MMP-2 was expressed in the cytoplasm of basal, spinosum and granular epithelial layers, although a weaker immunoreactivity was noted in the basal cell layer (Figure 2B). Fibropapillomas showed a variable immunoreactivity scoring: 6/19 (32 %) samples displayed a weak positivity, 6/19 (32 %) showed a moderate positivity, while 3/19 (16 %) samples displayed a strong immunosignal (Figure 2C). Moreover, in 11/19 (58 %) samples, a moderate labeling was recorded in the dermal layer, in cytoplasm of fibroblast cells (Figure 2B). Fibropapillomas sections with the first antibody anti-MMP-2 omitted and incubated with PBS (negative control) did not exhibit positive staining (Figure 2D).

Furthermore, MMP-7 was expressed in all normal skin samples, exhibiting predominantly a moderate cytoplasmic, perinuclear and nuclear immunoreactivity in basal and parabasal layers (Figure 3A). Nineteen out of 19 (100 %) fibropapilloma samples showed specific cytoplasmic and perinuclear immunoreactivity, with a variable degree of expression confined to basal, spinosum and granular epithelial layers (Figure 3B); a strong positivity was recorded in 6/19 (32 %) samples (Figure 3C), a moderate intensity was noted in 7/19 samples (36 %), while 6/19 (32 %) of samples showed a weak immunolabeling. Four out of nineteen 4/19 (21 %) tumor samples showed a weak cytoplasm expression in fibroblasts from dermal layer (Figure 3B). However, the nuclear pattern seen in the normal skin samples was not observed in the fibropapillomas. The tumor sections with the first antibody anti-MMP-7 omitted and incubated with PBS (negative control) did not exhibit positive staining (Figure 3D).

MMP-9 was expressed predominantly in the cytoplasm of basal cell layer in 4/4 skin samples, where a weak immunoreactivity was recorded (Figure 4A). In tumor samples, a cytoplasmic and perinuclear immunoreactivity was evident in epithelial cells from basal, spinosum and granular epithelial layers and fibroblast cells from dermal layer (Figure 4B), scoring from a weak expression in 2/19 (11 %) tumors, to moderate in 7/19 (36 %) samples and strong in 10/19 (53 %) samples. Indeed, an intense immunoreactivity was observed in the basal cell layer, concomitantly with scattered cells from spinosum stratum that displayed the same pattern. Moreover, a moderate positivity was noted in the dermal blood vessels (Figure 4C). The negative controls consisting in fibropapillomas sections with the first antibody anti-MMP-9 omitted and incubated with PBS did not exhibit positive staining (Figure 4D).

With respect to MMP-14, in 4/4 skin samples its expression was detected as moderate cytoplasmic, perinuclear and nuclear immunosignal in basal, spinosum and granular epithelial layers (Figure 5A). Fibropapillomas featured a weak positivity in 11/19 (58 %) samples, moderate in 4/19 (21 %) and strong in 1/19 (5 %) sample (T19), characterized by a finely granular pattern in the cytoplasm (Figure 5B) and interestingly, in the nucleus (Figure 5C) of some keratinocytes cells from spinosum and granular layers. In 5/19 (26 %) fibropapillomas, a weak immunoreactivity was detected in the cytoplasm (Figure 5B) of fibroblast from dermal layer. The tumor sections with the first antibody anti-MMP-14 omitted and incubated with PBS (negative control) did not exhibit positive staining (Figure 5D).

In 4/4 normal skin samples, a moderate cytoplasmic TIMP-1 immunoreactivity was noted in epidermis, more precisely in basal cells and only few cells from spinosum and granular layers showed moderate positivity (Figure 6A). In 5/19 (26 %) tumor samples, no specific TIMP-1 immunoreactivity could be detected. However, its expression was variable in the rest of tumor samples: weak in 10/19 (53 %) and moderate in 4/19 (21 %); a finely granular staining pattern was recorded in the cytoplasm of keratinocytes from basal, spinosum and granular layers (Figure 6B), as well as in 20–30 % of fibroblasts from the dermal layer (Figure 6C). The tumor sections with the first antibody anti-TIMP-1 omitted and incubated with PBS (negative control) did not exhibit positive staining (Figure 6D).

With respect to TIMP-2 expression, a cytoplasmic moderate immunoreactivity restricted to the basal epithelial cells was detected in all normal skin samples (Figure 7A). Fibropapillomas exhibited predominantly cytoplasmic TIMP-2 immunoreactivity confined to basal, spinosum and granular cell layers (Figure 7B), with intensity ranging from weak in 10/19 (53 %) samples, to moderate in 8/19 fibropapillomas (42 %); in 1/19 (5 %) (sample no T18) no specific immunoreactivity was detected. Most cells from spinosum and granular layers exhibited perinuclear immunosignal (Figure 7C). A weak expression was noted also in the fibroblast cells of almost all samples. In parallel, negative controls represented by fibropapillomas sections with the first antibody anti-TIMP-2 omitted and incubated with PBS did not exhibit positive staining (Figure 7D).

Nine bovine fibropapillomas (T1-T9) samples and four normal skin samples (NS1-NS4) were subjected to Western blotting analysis. The anti-MMP-2, anti-MMP-7, anti-MMP-9, anti-MMP-14, anti-TIMP-1 and anti-TIMP-2 antibodies yielded a band of expected molecular weight, confirming the specificity of IHC staining. MMP-2 appeared over-expressed in 8 out of 9 tumor samples (89 %), with particularly high levels in sample T3, when compared to normal skin samples, where its expression was very low (Figure 8A). Densitometric analysis of individual samples (Figure 8B) as well as mean tumor vs. normal skin values (Figure 8C) confirmed the results, although no statistically significant difference between samples was recorded, mostly due to high variability in the fold of upregulation among fibropapillomas.

Further, an over-expression of MMP-7 was detected in all tumor samples (Figure 8D) as confirmed also by individual (Figure 8E) and mean (Figure 8F) densitometric values of tumor vs. normal skin samples, and the difference between the two groups was statistically significant (t-test; P < 0.05).

Similarly, all fibropapillomas showed an increased expression of MMP-9 pro-form of 92 kDa when compared to normal skin samples (Figure 8G). Data were confirmed by densitometric analysis of individual samples (Figure 8H) as well as mean tumor vs. normal skin values (Figure 8I), although the difference was not statistically significant, probably for the reason stated above for MMP-2. Next, an additional band of ~82 kDa, likely corresponding to MMP-9 active form was observable in tumor samples, while it was very faint to undetectable in normal skin specimens.

When MMP-14 was analyzed, its expression was readily detectable in all normal skin samples, whilst in 9/9 tumor samples (100%), a lower expression was evident and confirmed by densitometric analysis (Figures 8J,K). Mean densitometric values revealed a statistically significant difference (t-test; ^**P < 0.01) (Figure 8L).

TIMP-1 appeared down-regulated in most of fibropapillomas, when compared to normal skin samples (Figure 9A). Furthermore, densitometric analysis of individual samples (Figure 9B) as well as mean tumor vs. normal skin values (Figure 9C) confirmed the results, although the difference between samples was not statistically significant.

WB for TIMP-2 (Figure 9D) followed by densitometric analysis revealed very similar results (Figures 9E,F).

In addition to protein expression levels, enzymatic activity of MMP-2 and MMP-9 was analyzed in normal skin and fibropapilloma samples using gelatin zymography. Pro- and active forms of MMP-2 were recorded with gelatin zymogram electrophoresis (Figure 10A). However, fibropapillomas showed an upregulation of the 62 kDa active MMP-2 in comparison with normal skin samples as confirmed by individual densitometric analysis, while the mean densitometric values revealed a statistically significant difference (Figures 10B,C) (t-test; ^*P < 0.05). Impressively, when analyzing MMP-9 activity, its activated form of 82 kDa was readily detectable in all fibropapilloma, whilst it was present at low to undetectable levels in the normal skin samples, in agreement with western blotting results, denoting a strong activity during tumor development. Moreover, the individual and mean densitometric analysis revealed similar results as mentioned above, although no statistically significant difference was recorded (Figures 10D,E) (t-test; P < 0.05).