Mesenchymal to epithelial transition driven by canine distemper virus infection of canine histiocytic sarcoma cells contributes to a reduced cell motility in vitro

Abstract Sarcomas especially of histiocytic origin often possess a poor prognosis and response to conventional therapies. Interestingly, tumours undergoing mesenchymal to epithelial transition (MET) are often associated with a favourable clinical outcome. This process is characterized by an increased expression of epithelial markers leading to a decreased invasion and metastatic rate. Based on the failure of conventional therapies, viral oncolysis might represent a promising alternative with canine distemper virus (CDV) as a possible candidate. This study hypothesizes that a CDV infection of canine histiocytic sarcoma cells (DH82 cells) triggers the MET process leading to a decreased cellular motility. Immunofluorescence and immunoblotting were used to investigate the expression of epithelial and mesenchymal markers followed by scratch assay and an invasion assay as functional confirmation. Furthermore, microarray data were analysed for genes associated with the MET process, invasion and angiogenesis. CDV‐infected cells exhibited an increased expression of epithelial markers such as E‐cadherin and cytokeratin 8 compared to controls, indicating a MET process. This was accompanied by a reduced cell motility and invasiveness. Summarized, these results suggest that CDV infection of DH82 cells triggers the MET process by an increased expression of epithelial markers resulting in a decreased cell motility in vitro.


| INTRODUC TI ON
Malignant neoplasms represent nowadays one of the most common causes of death in humans and companion animals due to their often rapid and lethal progression. 1,2 Histiocytic sarcomas (HS) are malignant tumours that may occur in a systemic or localized form in both humans and dogs, with a comparable poor prognosis. [3][4][5][6] Therefore, dogs may represent an interesting translational model for this neoplastic disease due to the higher prevalence of HS in the canine species than in humans. 3 Patients with HS often have short survival times due to the high metastatic rate of this neoplasm which is furthermore characterized by a limited and ineffective response to conventional therapies including surgery, chemo-and radiotherapy. 4,5 Therefore, novel and more effective approaches against this neoplasia are highly demanded. Since the beginning of the twentieth century the idea to use an oncolytic virus against neoplastic cells took place 6 considered the ability of several viruses to preferentially infect and destroy cancer cells with direct and indirect mechanisms. 7,8 In human medicine, several oncolytic viruses (OVs) are currently used in clinical trials, including adenovirus, herpes simplex virus, vaccinia virus, reovirus, Seneca Valley virus and measles virus (MV). 9 The latter is a morbillivirus that belongs to the Paramyxoviridae family. 9 Another morbillivirus, closely related to MV, is canine distemper virus (CDV), which shares many common features with the first, including the ability to infect and induce apoptosis in lymphoid cells. 10,11 Therefore, CDV represents a promising candidate for future applications as an oncolytic virus for canine hematopoietic tumours.
CDV demonstrated the ability to persistently infect canine histiocytic sarcoma cells (DH82 cells), influencing the expression of reversion-inducing cysteine-rich protein with Kazal motifs (RECK), matrix metalloproteinases (MMP) −2 and −9 and tissue inhibitors of matrix metalloproteinases (TIMP) −1 and −2, 12 altering cortactin distribution within the cytoskeleton, 13 and reducing the expression of genes known to interfere with angiogenesis. 14 Taken together, all these findings provide a robust basis to confirm CDV as a promising oncolytic virus for HS in dogs and use it as a model for the corresponding human disease.
During the last decade, the knowledge about factors influencing the biological behaviour of malignant neoplasms constantly increased. Specifically, the transition of cells from an epithelial to a mesenchymal state (EMT process) has been extensively studied and validated as one of the major features correlated to invasiveness and metastatic rate of carcinomas. 15,16 In contrast, the reverse transition known as mesenchymal to epithelial transition (MET process) came into the research focus only recently. 17 The latter process is characterized by the expression of markers typical of epithelial cells in sarcomas, which is often linked with a favourable clinical outcome and a better prognosis. 17 For example, in human synovial sarcoma, the epithelial cell markers E-cadherin and β-catenin are considered as potential positive prognostic factors. 18 Additionally, longer survival time has been associated with E-cadherin expression both at protein and mRNA level in a subset of human leiomyosarcomas. 19 E-cadherin has also been implicated as a tumour suppressor due to its protective role against epithelial to mesenchymal transition (EMT) at the primary site in carcinomas. 20 The MET process in sarcomas is characterized by an increased expression of classical epithelial markers, whereas the classical mesenchymal markers still predominate in the tumour cells therefore determining the so-called 'metastable phenotype'. 17,20,21 Typical epithelial-like markers include proteins such as cytokeratin, CD44, CD34, β-catenin and E-cadherin. 17 N-cadherin, vimentin, desmin and alpha-smooth muscle actin (α-SMA) are considered among the typical mesenchymal markers. 17 The hypothesis underlying the aim of this study is that a persistent infection of histiocytic sarcoma cells (DH82 cells) with CDV, strain Onderstepoort (CDV-Ond), triggers the MET process by increasing the expression of epithelial markers, resulting in a less invasive phenotype with decreased motility of the neoplastic cells. 13

| Morphological analysis using phase contrast microscopy
The morphology of non-infected DH82 and DH82Ond pi cells was analysed using a phase contrast microscope (Olympus IX-70, Olympus Optical Co. GmbH) equipped with an Olympus DP72 camera and Olympus cell sense standard software version 2.3. Cells were observed at 6 hours, 12 hours, 24 hours, 36 hours, 48 hours, 60 hours, 3 days, 4 days, 5 days, 6 days and 7 days after seeding.
Afterwards, cells were counted according to their morphology and grouped in 4 different categories: round-shaped, triangle-shaped, cigar-shaped and slender (supplementary material and Figure S1).

| Cumulative population doubling assay
The population doubling assay was performed as previously described 13 by evaluating non-infected DH82 cells and DH82Ond pi cells over 14 weeks. Briefly, cells were seeded into 75-cm 2 flasks (Nunc GmbH & Co. KG, Thermo Scientific, Langenselbold, Germany) and counted at every weekly passage over 14 weeks.
Population doubling was determined using the following population doubling (PD) formula: PD = log10 (cells harvested-initial cell number)/log2. 22 Then, the cumulative population doubling was determined by adding the PD of every weekly passage to the previous one.
Statistical analysis as well as graphical visualization was performed using GraphPad Prism version 8.0.1 for Windows (GraphPad Software, La Jolla California USA, www.graph pad.com). The values were analysed with non-parametric Wilcoxon-Mann-Whitney two-sample test, setting the significance level at P ≤ .05.

| Microarray data analysis using a manually generated list of gene symbols related to MET and invasiveness
Data of a previously published microarray dataset of non-infected DH82 and DH82Ond pi cells (ArrayExpress; http://www.ebi.ac.uk/ array express; accession number E-MTAB-3942) 13,14,23 were used to evaluate the potential influence of a persistent CDV infection on genes associated with EMT/MET and cellular motility. In a hypothesis-driven approach, the present study focused on a manually generated list of selected genes associated with EMT/MET, invasion and angiogenesis. Selected gene symbols were based on previously published lists, 13,14 which were further modified and extended (Table S1). According to the biological function of the corresponding protein(s), each selected gene symbol was assigned to 'EMT/MET' and/or 'invasion and angiogenesis' functional group. Genes were considered as differentially expressed between non-infected DH82 and DH82Ond pi cells combining a fold change (FC) filter (FC ≥ 1.5 or ≤−1.5) 24 with a statistical significance filter (Mann-Whitney U test; P ≤ .05).
Non-infected DH82 and DH82Ond pi cells were seeded at a density of 0.03 * 10 6 cells/0. 33  ranged from 65 to 151 (the specific number of frames for each z-stack is reported in the caption of the corresponding figure).
For each z-stack set, the background was set to black by standard software settings. Subsequently, top view and section view of the 3D reconstructions were created for each staining. Section view allowed to analyse protein localization within the cells.

| Immunoblotting
Immunoblotting of non-infected and persistently CDV-infected DH82 cells was comparatively carried out in three independent samples for each cell type, as formerly described. 27 Following cell lysis, the correct amount of each sample required for the analysis was calculated based on the protein concentration as determined applying the Bradford method. Immunoblotting was performed using as primary antibod-

| Scratch assay and invasion assay
For the scratch assay, non-infected DH82 and DH82Ond pi cells were

| Persistent CDV-Ond infection of DH82 cells leads to morphological changes while growth features remain unaltered
The   Figure 1N).

| DH82Ond pi cells display an increased expression of epithelial markers on a protein level
However, DH82Ond pi cells displayed a significantly (P < .0001) higher 'membranous to cytoplasmic' expression of this protein compared to non-infected controls ( Figure 1O), while the diffuse cytoplasmic localization did not reach statistical significance (P = 0,8340).
Interestingly, cytokeratin 8 displayed a focal cytoplasmic expression that was significantly (P < .0001) more often observed in non-infected controls compared to persistently CDV-infected DH82 cells ( Figure 1O). Similarly to E-cadherin, the focal cytoplasmic localization of this marker in non-infected DH82 cells was initially confirmed by laser scanning confocal microscopy of single-labelling immunofluorescence stains ( Figure 1K, insert). Further analyses employing 3D reconstructions of double-labelling immunofluorescence combining the Golgi apparatus. Occasionally, also a membranous to cytoplasmic expression of the protein was detected in non-infected controls.
DH82Ond pi cells were analysed accordingly confirming a membranous to cytoplasmic expression of cytokeratin 8, which was frequently arranged in variably sized aggregates ( Figure S6). Additional single-labelling immunofluorescence pictures displaying the intracellular distribution of β-catenin, E-cadherin and cytokeratin 8 in non-infected and persistently CDV-infected DH82 cells at different confluences are available as Figure S7. In order to confirm the immunofluorescence results, an immunoblotting for all the investigated epithelial markers was performed ( Figure 3A). Beta-actin was used as a house-keeping protein, lacking differential expression at both the gene (fold change: 1.06; p value < 0.001) and the protein level

| DH82Ond pi cells retain mesenchymal marker expression
Mesenchymal marker immunolabelling was analysed as shown in

| Molecular expression of mesenchymal and epithelial markers in DH82Ond pi is suggestive of MET
Selection of gene symbols and proteins associated with EMT/MET, invasion and angiogenesis resulted in a manually generated list of 84 canine gene symbols (Table S1)

| MET in DH82Ond pi cells is associated with a decreased cell motility and invasiveness
Analysis of the aforementioned microarray dataset revealed that among 19 selected genes classified within the functional group 'invasion and angiogenesis', all 13 differentially expressed gene symbols between DH82Ond pi and non-infected controls were Note: Microarray data were obtained from a previously published dataset 13,14 and were filtered according to a combination of the fold change (FC ≥ 1.5 or ≤−1.5) and the level of significance (P ≤ .05). Down-regulated genes are highlighted in green, while up-regulated genes are labelled in red.

| D ISCUSS I ON
The aim of the current study was to investigate the impact of CDV- Dr Reiner Müller-Peddinghaus Foundation.

CO N FLI C T O F I NTE R E S T
The authors declare no potential conflicts of interest.