Epithelial to Mesenchymal Transition (EMT) in a Laryngeal Squamous Cell Carcinoma of a Horse: Future Perspectives

Simple Summary Squamous cell carcinoma (SCC) is one of the most common cancers in horses, and it can arise at any site on the skin and mucosae. Recent studies associated equine papillomavirus type 2 (EcPV2) infections with this type of cancers of the oral tract and genitals. Larynx and pharynx are frequently recognized as sites of SCC. In humans, squamous cell carcinoma of the larynx (SCCL) is a common cancer associated with papilloma virus (PV) infection and epithelial to mesenchymal transition (EMT). EMT can occur under different biological conditions, upon the same programmed changes: embryogenesis and organ development fibrosis, wound healing, and cancer metastases. This work reports for the first time in a SCCL of a horse a wide immunohistochemical EMT characterization, by analyzing main epithelial markers (E-cadherin, β-catenin, and pan-cytokeratin AE3/AE1), main mesenchymal markers (N-cadherin and vimentin), and the main EMT-related transcription factors (TWIST-1, ZEB-1, and HIF-1α). This work illustrates an example of tumor cell adaptation during the metastatic process in the equine SCCL, taking also into consideration the potential influence provided by EcPV2 oncoproteins on the EMT process. Abstract Squamous cell carcinoma (SCC) is one of the most frequent tumors of skin and muco-cutaneous junctions in the horse. Equine papillomavirus type 2 (EcPV2) has been detected in equine SCC of the oral tract and genitals, and recently also in the larynx. As human squamous cell carcinoma of the larynx (SCCL), it is strongly etiologically associated with high-risk papillomavirus (h-HPV) infection. This study focuses on tumor cells behavior in a naturally occurring tumor that can undergo the so-called epithelial to mesenchymal transition (EMT). A SCCL in a horse was investigated by immunohistochemistry using antibodies against E-cadherin, pan-cytokeratin AE3/AE1, β-catenin, N-cadherin, vimentin, ZEB-1, TWIST, and HIF-1α. EcPV2 DNA detection and expression of oncogenes in SCC were investigated. A cadherin switch and an intermediate filaments rearrangement within primary site tumor cells together with the expression of the EMT-related transcription factors TWIST-1, ZEB-1, and HIF-1α were observed. DNA obtained from the tumor showed EcPV2 positivity, with E2 gene disruption and E6 gene dysregulation. The results suggest that equine SCCL might be a valuable model for studying EMT and the potential interactions between EcPV2 oncoproteins and the EMT process in SCCL.

. Details of the antibodies used for immunostaining, including primary antibody, host species, clonality, epitope retrieval method, blocking serum, dilution of primary antibody, secondary antibody, and positive control.

Detection of EcPV2 and Evaluation of Oncogene Expression
DNA and RNA extraction was performed to check EcPV2 presence and expression of oncogenes in SCCL. One sample of kidney obtained from a healthy horse was used as negative control. To prevent cross-contamination, a different blade was used for each sample. Four sections (5 µm) were obtained from FFPE samples for total acid nucleic extraction, which was performed using an AllPrep DNA FFPE Kit (Qiagen, Hilden, Germany) and an AllPrep RNA FFPE Kit (Qiagen, Hilden) in accordance with the manufacturer's instructions.
During extraction, samples were treated with DNase (RNase-Free DNase set, Qiagen). To evaluate DNA amplifiability, β-actin gene amplification was performed [7]; then, EcPV2-L1 DNA presence was tested using a Real-Time protocol previously described [7]. Briefly, TaqMan ® probe-based real-time PCR (CustomProbe 2× qPCR Master Mix, Canvax, Cordoba, Spain, cat. E0339) was performed using a CFX96™ Real-Time System (Bio-Rad, Rome, Italy). The reaction mix had a final concentration of 1× TaqMan ® master mix, 200 nM probe, 100 nM each primer combination, and 200 ng of DNA. Reverse transcription (RT) was performed using a SuperScript™ IV VILO™ Master Mix with ezDNase™ Enzyme (Invitrogen, ThermoFisher Scientific (Waltham, MA, USA), cat. 11766050) according to the manufacturer's instructions; cDNA was used to check gene expression by TaqMan ® probe-based real-time PCR (CustomProbe 2× qPCR Master Mix, Canvax, Spain, cat. E0339) performed using a CFX96™ Real-Time System (Bio-Rad, Paris, France). The PCR mix (25 µL) had a final concentration of 1× TaqMan ® master mix, 200 nM probe, 100 nM each primer combination, and 5 µL of template (cDNA or DNA and RNA used as negative control to exclude possible contamination by genomic DNA). The thermal profile used for amplification was the following: 95 • C for 10 min, followed by 40 cycles at 95 • C for 15 s, and 60 • C for 60 s for annealing/extension and detection of the fluorescence signal. The fluorescence threshold limit was set automatically. Moreover, the expression of E2, E6, and L1 genes was evaluated by RT-real-time PCR using primers and probes (Table 2) and a protocol described Animals 2020, 10, 2318 5 of 13 by Porcellato and co-workers [7]. Briefly, RT was performed using a SuperScript™ IV VILO™ Master Mix (Invitrogen, ThermoFisher Scientific) according to the manufacturer's instructions. The cDNA was used to check gene expression: 5 µL of template were added to 20 µL of PCR mix at a final concentration of 1× master mix (iTaq Universal Probes Supermix, Bio-Rad, Italy), 200 nM probe, and 100 nM each primer combination. RNA was used as control to exclude possible contamination by EcPV2 genomic DNA. Nuclease-free water was used as negative control. Each sample was tested in triplicate.

Post-Mortem and Microscopical Analysis
A 17-year-old female, 550 Kg, Maremmano Horse was referred with severe clinical signs due to a dorso-cranial dislocation of the epiglottis which caused the reduction of 70% of the laryngeal lumen, diagnosed by an endoscopic exam, performed under sedation with butorphanol combined with detomidine [23]. The animal was humanely euthanized due to the poor condition. Necropsy revealed a locally expansive, multilobular, white-yellowish, firm mass localized at the base of the larynx. The lesion was partially ulcerated, with irregular margins and central necrotic areas (Figure 1). The neoplasia expanded to the nearby tissues and regional lymph nodes were markedly enlarged. Microscopically, the laryngeal tumor was completely effacing the submucosa, non-encapsulated, densely cellular, and poorly demarcated. Neoplastic cells were variably arranged in anastomosing bands and chords occasionally forming lobules embedded in a moderate amount of fibrous collagen stroma. Neoplastic cells were large, variably from polygonal to spindle-shaped with indistinct cell borders and an intermediate to high nuclear/cytoplasmic ratio ( Figure 2). The cytoplasm was moderate and eosinophilic nuclei were large, round to oval, with vesicular chromatin and 1 or 2 round, basophilic nucleoli. Anisocytosis and anisokaryosis were high, and mitoses ranged from 0-2 per HPF (400×). Multifocally, there were wide areas of coagulative necrosis within the tumor. A moderate amount of small mature lymphocytes and plasma cells were found within and surrounding the tumor area. Remarkably, scattered tumor cells displayed a more prominent spindled shape. Interestingly, the regional lymph node was 80% effaced by necrotic and metastatic events ( Figure 2). Numerous neoplastic cells with a morphology and histological pattern similar to those in the laryngeal tumor were found among resident lymphoid cells. Given this aggressive behavior of the neoplasia and this particular phenotypical change in morphology of the neoplastic cells which were variably from polygonal to elongated, an immunohistochemical panel for the EMT phenomenon was performed.

Cadherin Switch and Intermediate Filaments Rearrangements Suggest an EMT Phenomenon in Equine Laryngeal Squamous Cell Carcinomas
Immunohistochemical analysis of primary site tumor cells revealed an increased number of cells with cytoplasmic E-cadherin expression rather than membranous, together with a gradual overall loss of cells expressing this adhesion molecule moving towards the invasive front of the tumor ( Figure 3).
Interestingly, the number of cells expressing N-cadherin resulted to be increased within primary site tumor cells, even though it was expressed at nuclear level instead of being membranous ( Figure 3). Interestingly, moving from the tumor center towards the invasive front, a decreased number of cells expressing cytokeratin was observed, while scattered neoplastic cells with a prominent mesenchymal morpholgy acquired cytoplasmic vimentin expression ( Figure 3). Noteworthy, numerous neoplastic cells were scattered throughout the tumor area and expressed the EMT-related transcription factors TWIST-1, ZEB-1, and HIF-1α ( Figure 4). In addition, β-catenin was found to be frequently expressed at a nuclear level, or (less frequently) cytoplasmically, rather than being membraneous (Figure 3). On the other hand, immunohistochemical analysis of the regional lymph node (medial retropharingeal lymph node) revealed a multifocal strong cytokeratin expression together with a more frequent membranous E-cadherin immunolabeling ( Figure 5). Neoplastic cells did not express vimentin within the lymph nodes, but still expressed N-cadherin at nuclear level ( Figure 5). Moreover, we detected a slight decrease of the number of neoplastic cells expressing the EMT-related transcription factors (TWIST-1, ZEB-1, and HIF-1α) and nuclear β-catenin ( Figures 5 and 6). These findings demonstrate that the morphological changes and the aggressive behavior are most likely due to the EMT process activated within these tumor cells. Considering that the current literature suggests an emerging role for EcPV2 in several equine squamous cell carcinomas [11] and given that a correlation between SSCL and h-HPV infection [6] was demonstrated also in humans, we investigated whether, also in this case, the papillomavirus infection played a role, making the equine species a promising model for this type of tumor. Animals 2020, 10, x 6 of 14

EcPV2 Detection Suggests a Potential Role in Equine Laryngeal Squamous Cell Carcinomas
DNA obtained from the tumor sample was amplifiable ( Table 2) and it showed positivity for EcPV2 L1 (Table 2), with a mean Cq of 22.5 ± 0.17, obtained as mean of three replicates. The negative sample was amplifiable, but negative for EcPV2. Regarding gene expression, only the E6 gene was expressed (Table 3), with a mean Cq of 30.4 ± 0.24. Data are expressed as mean Cq (quantitation cycle) ± SD (standard deviation) of 3 replicates; B2M= beta-2-microglobulin.

Discussion
Based on the current findings, a laryngeal squamous cell carcinoma with a regional lymph node metastasis was diagnosed. The wide immunohistochemical panel used in the current study allowed us to detect a gradual decreased number of cells expressing epithelial markers within the primary tumor site invasive front, together with a gradual increased number of cells expressing mesenchymal markers and key transcription factors for the EMT process. On the other hand, lymph node metastasis revealed a moderate switch in markers expression compared to the primary carcinoma site that might suggest a partial/incomplete mesenchymal to epithelial transition (MET) process. Specifically, lowered numbers of cells expressing E-cadherin and cytokeratin were detected among primary site cancer cells together with aberrant N-cadherin expression and a moderate number of cells immunolabeled for vimentin. These findings were further supported by numerous cells displaying nuclear staining for TWIST-1, ZEB-1, β-catenin, and HIF-1α. In contrast, the metastatic lymph node

EcPV2 Detection Suggests a Potential Role in Equine Laryngeal Squamous Cell Carcinomas
DNA obtained from the tumor sample was amplifiable ( Table 2) and it showed positivity for EcPV2 L1 (Table 2), with a mean Cq of 22.5 ± 0.17, obtained as mean of three replicates. The negative sample was amplifiable, but negative for EcPV2. Regarding gene expression, only the E6 gene was expressed (Table 3), with a mean Cq of 30.4 ± 0.24. Data are expressed as mean Cq (quantitation cycle) ± SD (standard deviation) of 3 replicates; B2M = beta-2-microglobulin.

Discussion
Based on the current findings, a laryngeal squamous cell carcinoma with a regional lymph node metastasis was diagnosed. The wide immunohistochemical panel used in the current study allowed us to detect a gradual decreased number of cells expressing epithelial markers within the primary tumor site invasive front, together with a gradual increased number of cells expressing mesenchymal markers and key transcription factors for the EMT process. On the other hand, lymph node metastasis revealed a moderate switch in markers expression compared to the primary carcinoma site that might suggest a partial/incomplete mesenchymal to epithelial transition (MET) process. Specifically, lowered numbers of cells expressing E-cadherin and cytokeratin were detected among primary site cancer cells together with aberrant N-cadherin expression and a moderate number of cells immunolabeled for vimentin. These findings were further supported by numerous cells displaying nuclear staining for TWIST-1, ZEB-1, β-catenin, and HIF-1α. In contrast, the metastatic lymph node displayed a partial reversion of the aforementioned epithelial and mesenchymal markers, thus suggesting an MET process. It is well reported that TWIST-1 is considered as the main regulator of EMT [24,25] and is up-regulated in a large number of malignant tumors determining the onset of the metastatic process, via promoting invasiveness in both spontaneous and experimental models [25,26]. Noteworthy, the tumor in the present study had both a high number of cells expressing TWIST-1 and a very low number of cells expressing membranous E-cadherin or, interestingly, the cells exhibited cytoplasmic internalization of the protein. This aberrant cytoplasmic expression has been recently related in equine penile carcinoma to a more aggressive behavior due to AKT/MAPK pathway activation [27,28]. Interestingly, the decreased number of cells expressing E-cadherin (E-cadherin loss) was replaced by the increased number of cells expressing nuclear N-cadherin. The aforementioned results might be in line with a study on human nasopharyngeal carcinoma by Luo and colleagues, reporting a correlation between nuclear N-cadherin and a poorer prognosis [29]. Another aberrant localization found in the current case is represented by β-catenin in both primary and metastatic sites. Normally, the membranous E-cadherin/β-catenin complex maintains the intercellular tight junction and minor free β-catenin cytoplasmic molecules are controlled by multiprotein complexes, while on the other hand, nuclear localization of β-catenin is essential for the progression of various human cancers, such as nasopharyngeal carcinoma, via transcriptional upregulation of downstream genes [30]. These findings are in line with the low number of cells expressing β-catenin at a cytoplasmic level, compared to the higher number of cells expressing β-catenin at a nuclear level found in this study. It is well documented that solid tumors generally have a hypoxic microenvironment [31]. Previous studies suggested that moderate hypoxic conditions might trigger an EMT process via HIF-1α, leading different human cancer cells to significantly increase their invasiveness [32]. Noteworthy, Yang and colleagues demonstrated that HIF-1α directly binds to the hypoxia response element (HRE) in the TWIST promoter, regulating the expression of this transcription factor [33]. These results seem to be in agreement with the HIF-1α expression found in this study. However, all the results regarding the EMT process obtained at a protein level in the present study by immunohistochemistry will need to be further confirmed also at a genetic level in future studies aimed at increasing the knowledge on these aspects by focusing also on the possible role of some microRNA (miRNA) families. The molecular analysis which demonstrated the presence of EcPV2 and its oncogene expression is in agreement with previous studies [11,34,35]. In particular, our data showed E6 but not E2 expression. The lack of E2 expression suggests virus genome integration and loss or disruption of the E2 gene. This can cause deregulation of E6 expression and, in turn, the increase of this event triggers cancer progression [36]. Moreover, a previous study in humans demonstrated a role for E6 in innate immune gene repression [36]. It is also important to consider the potential role of papillomavirus oncoproteins in triggering the EMT process. Interestingly, Liu and colleagues reported that E6 and E7 oncoproteins enhance the expression of HIF-1α, as well as of ZEB-1, SNAIL-1, SLUG, and TWIST-1 in non-small cell lung cancer (NSCLC) cells, thus promoting the EMT process [37]. Recently, a study using human lung samples confirmed the E7 oncoprotein role in promoting EMT in human lung cancers, reporting correlations with E7 and E-cadherin, N-cadherin, and TGF-β expression [38]. According to the authors, the findings of the present work about EcPV oncoproteins and the EMT-related transcription factors and structural/adhesion proteins are in line with the literature and might represent a promising starting point to be further investigated.

Conclusions
In conclusion, this interesting case of equine metastatic SCCL provides an example of tumor cell adaptation during the metastatic process in the equine species, taking also into account the possible influence of EcPV2 oncoproteins on the EMT process. This is an opportunity to propose the equine species in future studies for evaluation of the potential interactions between EcPV2 oncoproteins and the EMT process both in human and animal cancers, thus opening new study perspectives in this field.