Functional Granulocyte–Macrophage Colony-Stimulating Factor (GM-CSF) Delivered by Canine Histiocytic Sarcoma Cells Persistently Infected with Engineered Attenuated Canine Distemper Virus

The immune response plays a key role in the treatment of malignant tumors. One important molecule promoting humoral and cellular immunity is granulocyte–macrophage colony-stimulating factor (GM-CSF). Numerous successful trials have led to the approval of this immune-stimulating molecule for cancer therapy. However, besides immune stimulation, GM-CSF may also accelerate tumor cell proliferation, rendering this molecule a double-edged sword in cancer treatment. Therefore, detailed knowledge about the in vitro function of GM-CSF produced by infected tumor cells is urgently needed prior to investigations in an in vivo model. The aim of the present study was to functionally characterize a persistent infection of canine histiocytic sarcoma cells (DH82 cells) with the canine distemper virus strain Onderstepoort genetically engineered to express canine GM-CSF (CDV-Ondneon-GM-CSF). The investigations aimed (1) to prove the overall functionality of the virally induced production of GM-CSF and (2) to determine the effect of GM-CSF on the proliferation and motility of canine HS cells. Infected cells consistently produced high amounts of active, pH-stable GM-CSF, as demonstrated by increased proliferation of HeLa cells. By contrast, DH82 cells lacked increased proliferation and motility. The significantly increased secretion of GM-CSF by persistently CDV-Ondneon-GM-CSF-infected DH82 cells, the pH stability of this protein, and the lack of detrimental effects on DH82 cells renders this virus strain an interesting candidate for future studies aiming to enhance the oncolytic properties of CDV for the treatment of canine histiocytic sarcomas.


Introduction
Histiocytic sarcoma (HS), which can occur in a disseminated or localized form, is a malignant tumor in humans and dogs that has a comparable poor prognosis in both species. Due to the higher prevalence of canine HS compared with its human counterpart, dogs are an interesting translational model for this neoplastic disease [1][2][3][4][5].

Virus Neutralization
Supernatants from non-infected DH82 cells and DH82 cells persistently infected with canine distemper virus (CDV) strain Onderstepoort (CDV-Ond), along with modified strains of CDV-Ond with the insertion of mNeonGreen (CDV-Ond neon ) and CDV-Ond with the insertion of mNeonGreen and canine GM-CSF (CDV-Ond neon-GM-CSF ), were obtained as described before [10]. Supernatants obtained from DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF and those from non-infected controls were harvested 7 days after seeding and centrifuged (700× g, 10 min, 4 • C). Afterward, the pH of the supernatants was lowered with HCl to a pH of 2 for 30 min to inactivate the CDV. Afterward, NaOH was added to increase the pH of all the supernatants to 7. The measurements of pH were performed using pH indicator strips (Merck, Darmstadt, Germany).

Immunofluorescence
In order to verify a CDV infection or confirm the successful inactivation, Vero.DogSLAM cells were immunolabelled after virus titration using an antibody directed against CDV nucleoprotein as previously described [10]. Briefly, cells were fixed with 4% buffered paraformaldehyde (PFA 4%, pH 7.4) and permeabilized with PBS-Triton X (0.025%). After serum blocking, cells were incubated overnight at 4 • C with the anti-CDV nucleoprotein antibody (Table 1). Afterward, cells were washed with PBS/0.1% Triton and incubated for 2 h with the secondary antibody. For nuclear staining, bisbenzimide (Merck, Darmstadt, Germany) was used. Negative controls included the omission of primary or secondary antibodies. Pictures were taken at 200× magnification using a microscope (Olympus IX-70, Olympus Optical Co. GmbH, Hamburg, Germany) equipped with an Olympus DP-72 camera and Olympus cellSens standard software version 2.3 (Olympus optical Co. GmbH, Hamburg, Germany). CDV-NP-canine distemper virus nucleoprotein; GaM-Cy3-goat anti-mouse cyanine 3-conjugated; GaR-b-goat anti-rabbit biotinylated; IF-immunofluorescence; IHC-immunohistochemistry.

RNA Isolation and cDNA Synthesis
Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) with on-column DNA digestion with RNase free DNase (Qiagen, Hilden, Germany) according to the manufacturer's protocols. The RNA concentration was spectroscopically measured at 260 nm using the GeneQuant pro (GE Healthcare, Amersham, Buckinghamshire, UK). RNA to cDNA transcription was performed according to the manufacturer's protocols with OmniScript (Qiagen, Hilden, Germany), RNAseOut (ThermoFischer Scientific, Schwerte, Germany), and Random Hexamers (Promega, Madison, WI, USA). Reverse transcription was performed using a Biometra Thermocycler T-Gradient ThermoBlock (American Laboratory Trading, East Lyme, CT, USA) under the following conditions: 25 • C for 10 min, 37 • C for 1 h, and 93 • C for 5 min.

Primer Design
The primers for qualitative and quantitative RT-PCR for the detection of CDV nucleoprotein mRNA transcripts were taken from the literature [34]. The primers used for qualitative CDV nucleoprotein RT-PCR and quantitative CDV RT-PCR are presented in Table 2.

Reverse Transcription Quantitative PCR (RT-qPCR)
The NucleoSpin ® Gel and PCR Clean-up Kit (Macherey-Nagel, Düren, Germany) was used for the isolation of PCR amplicons from the agarose gel according to the manufacturer's protocol. Standard curves for the estimation of copy numbers were generated using RT-PCR amplicons in a serial dilution from 10 8 to 10 2 copies/µL. RT-qPCR was performed with four samples per condition and negative controls, as previously described [34]. Quantitative PCR was performed using the Brilliant III Ultra-Fast SYBR®Green QPCR Master Mix (Agilent Technologies, Cedar Creek, TX, USA). Primers were diluted to a final concentration of 150 nM each. The detection of CDV mRNA transcripts was performed using the AriaMx Real-time PCR System (Agilent Technologies, Santa Clara, CA, USA) at the following conditions: denaturation at 95 • C for 3 min, 35 cycles at 95 • C for 5 s, and 57 • C for 10 s, followed by a melting curve consisting of one initial denaturation step at 95 • C for 30 s followed by 65 • C, increasing the temperature by 1 • C per cycle.

Immunohistochemistry
Immunohistochemistry for the GM-CSF receptor (CD116) was performed to investigate the expression in DH82 cells. Therefore, formalin-fixed, paraffin-embedded cell pellets obtained from non-infected DH82 cells and DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF were used. HeLa cell pellets with a known expression of the GM-CSF receptor (CD116) served as positive controls. Immunohistochemistry was carried out as previously described [10]. Immunolabeling was performed in triplicate with negative controls, as previously described [10]. Briefly, after the dewaxing, rehydration, and blocking of endogenous peroxidases, sections were blocked with goat serum for 30 min. Subsequently, slides were incubated overnight at 4 • C with the primary GM-CSF receptor (CD116) antibody (rabbit, polyclonal, 1:500; Invitrogen, CA, USA). Afterward, a secondary goat anti-rabbit biotinylated antibody (1:200, Vector Laboratories, Burlingame, CA, USA) and an avidin-biotin complex (ABC) peroxidase kit (Vectastain®Elite®ABC Kit, Vector Laboratories, Burlingame, CA, USA) were applied for 30 min and 20 min, respectively. A 3 3 -diaminobenzidine (DAB) system (Vector Laboratories, Burlingame, CA, USA) was used for the detection of positive reactions (Table 1). Nuclei were counterstained with Mayer's hemalum (Carl Roth GmbH, Karlsruhe, Germany). For the negative controls, the specific primary antibody was replaced by normal rabbit serum. The dilution of the negative controls was chosen according to the protein concentration of the replaced primary antibodies. HeLa cell pellets were used as a positive control. Pictures were taken at 400× magnification using a microscope (Olympus BX51, Olympus Optical Co. GmbH, Hamburg, Germany) equipped with an Olympus D72 camera (Olympus Optical Co. GmbH, Hamburg, Germany). CD116 immunolabeling was qualitatively analyzed.

Scratch Wound Assay
The scratch assay was performed as previously described [29]. Briefly, non-infected DH82 cells were seeded at a density of 0.3 × 10 5 cells/well into 96-well microtiter plates (ThermoFischer Scientific, Schwerte, Germany) with 200 mL of Minimal Essential Medium (MEM) with Earle's salts (Merck, Darmstadt, Germany) supplemented with 10% fetal bovine serum (Capricorn Scientific, Ebsdorfergrund, Germany), 1% penicillin/streptomycin (Sigma-Aldrich, Taufkirchen, Germany) and 1% non-essential amino acids (Sigma-Aldrich, Taufkirchen, Germany). Cells were cultured for 2 days under standard conditions (37 • C, 5% CO 2 , water-saturated atmosphere). When cells reached 99% confluence, the cell monolayer was scratched in a straight line with a p100 pipette tip. The medium was discarded, and the cell monolayer was washed with a washing medium (Minimal Essential Medium (MEM) with Earle's salts (Merck, Darmstadt, Germany)). Acidified supernatants obtained Pathogens 2023, 12, 877 6 of 18 from non-infected DH82 cells or DH82 cells infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF , respectively, were added to the scratched wells. Culture supernatant supplemented with 2 ng/mL of canine GM-CSF (R&D Systems, Systems, Minneapolis, MN, USA) was used for additional samples. Pictures were taken at the same position directly after performing the scratch (T 0 ) and after 24 h (T 24 ) using a phase contrast microscope (Olympus IX-70, Olympus Optical Co. GmbH, Hamburg, Germany) equipped with an Olympus DP-72 camera and Olympus cellSens standard software version 2.3 (Olympus Optical Co. GmbH, Hamburg, Germany). The percentage of cell-free area was calculated with ImageJ 1.52p according to the following formula: 100 − percentage of cell-covered area, as previously described [39]. The change in the cell-covered area was calculated according to the following formula: (|Area T 0 − Area T 24 |)/Area T 0 * 100. Commercially available human and canine GM-CSF was supplemented at 5 µg/mL. A culture medium supplemented with caGM-CSF or rhGM-CSF was used in an amount of 500 µL/well. The acidified supernatants obtained from DH82 cells infected with CDV-Ond, CDV-Ond neon , or CDV-Ond neon-GM-CSF were diluted to 2:3 with MEM. Cell numbers were quantified after 6 and 12 h. Therefore, cells were detached from the bottom of the well with 0.05% trypsin/0.02% EDTA (Sigma-Aldrich, Taufkirchen, Germany). Trypsinized cell suspensions were centrifuged for 10 min at 250× g at 4 • C. The obtained cell pellets were re-suspended in 30 µL MEM medium, and 20 µL of the cell suspension was added to 40 µL of trypan blue to obtain a final 1:2 dilution. Counting was performed using a 0.100 mm Neubauer counting chamber (Assistent, Sondheim vor der Rhön, Germany) and a light microscope (Axiovert 10, Zeiss, Oberkochen, Germany) [40]. The cell number was obtained according to the following formula: Commercially available human and canine GM-CSF was supplemented at 5 µg/mL. A culture medium supplemented with caGMCSF or rhGM-CSF was used in an amount of 500 µL/well. The acidified supernatants obtained from DH82 cells infected with CDV-Ond, CDV-Ond neon , or CDV-Ond neon-GM-CSF were diluted to 2:3 with DMEM. The cell number was quantified after 6 and 12 h. Thereafter, cells were detached from the bottom of the well with 0.05% trypsin/0.02% EDTA (Sigma-Aldrich, Taufkirchen, Germany). Trypsinized cell suspensions were centrifuged for 5 min at 300× g at 4 • C. Obtained cell pellets were re-suspended in 30 µL of DMEM, and 20 µL of the cell suspension was added to 40 µL of trypan blue to obtain a final 1:2 dilution. Counting was performed as described above.

Statistical Analysis
For descriptive statistics, the median and range were calculated. For the analysis of data obtained from the virus titration, immunofluorescence, RT-qPCR, and immunoblotting and proliferation assays, the non-parametric Mann-Whitney U and Wilcoxon signedrank tests were used. Statistical analysis was performed with SAS software version 7.1.5.0 (SAS Institute, Cary, NC, USA, www.sas.com). The level of significance was set at p ≤ 0.05. Graph creation was carried out using GraphPadPrism version 8.0.1 for Windows (GraphPad Software, La Jolla, CA, USA, www.graphpad.com).

Acidic Inactivation of the Supernatant Results in CDV Neutralization
Virus replication was comparatively investigated using virus titration, which revealed no significant differences in the titers produced by DH82 cells infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF (Mann-Whitney U test; p > 0.05; Figure 1). After inactivation by acidification (pH 2), a significant reduction in the 50% log10 tissue culture infectious dose per milliliter (TCID 50 /mL) was observed compared with non-acidified supernatants independent of the virus strain (Wilcoxon signed-rank test; p ≤ 0.05, Figure 1). In acidified supernatants from CDV-infected cultures, neither a cytopathic effect nor cells immunopositive for CDV nucleoprotein were observed, independent of the virus strain (Table 3, Figure 1). Table 3. CDV titers of different virus strains before and after neutralization by acidification (n = 3 per condition and group). At the molecular level, the number of CDV nucleoprotein mRNA transcripts was similar in the supernatants obtained from DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF , respectively (Mann-Whitney U test; p > 0.05). After acidification, the number of CDV nucleoprotein mRNA transcripts was significantly lower in supernatants from all treated cultures compared with corresponding native controls (CDV-Ond: p = 0.0152, CDV-Ond neon : p = 0.0259 and CDV-Ond neon-GM-CSF , p = 0.0152; Wilcoxon signed-rank test; Figure 2 and Table 4). In supernatants of non-infected DH82 cells, CDV nucleoprotein mRNA transcripts were not present. CDV-canine distemper virus; GM-CSF-granulocyte and macrophage colony-stimulating factor; TCID50/mL-50% log10 tissue culture infectious dose per milliliter; w/o-without.

Acidic Inactivation of the Supernatant Does Not Affect the GM-CSF Prot
Immunoblotting with an anti-CDV nucleoprotein antibody displa at 58 kDa in supernatants from DH82 cells infected with CDV-Ond, C CDV-Ond neon-GM-CSF , while no bands were present in non-infect Supernatants obtained from DH82 cells infected with CDV-Ond conta lower amounts of CDV nucleoprotein than supernatants obtained fr infected cultures (p = 0.0404) and CDV-Ond neon-GM-CSF -infected cultures(p Whitney U test; Table 3, Figure 3). Similar results were obtained for acid

Acidic Inactivation of the Supernatant Does Not Affect the GM-CSF Protein Content
Immunoblotting with an anti-CDV nucleoprotein antibody displayed protein bands at 58 kDa in supernatants from DH82 cells infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF , while no bands were present in non-infected supernatants. Supernatants obtained from DH82 cells infected with CDV-Ond contained significantly lower amounts of CDV nucleoprotein than supernatants obtained from CDV-Ond neon -infected cultures (p = 0.0404) and CDV-Ond neon-GM-CSF -infected cultures (p = 0.0404) (Mann-Whitney U test; Table 3, Figure 3). Similar results were obtained for acidified supernatants with significantly lower amounts of CDV nucleoprotein in supernatants obtained from CDV-Ond-infected DH82 cells compared with supernatants obtained from cultures infected with both other virus strains (CDV-Ond neon : p = 0.0404; CDV-Ond neon-GM-CSF : p = 0.0404; Mann-Whitney U test; Table 5, Figure 3). The comparison of native compared with corresponding acidified supernatants showed no significant differences in the amount of CDV nucleoprotein (Wilcoxon signed-rank test; p > 0.05). CDV-Ond-canine distemper virus strain Onderstepoort; pi-persistently infected; GMgranulocyte and macrophage colony-stimulating factor; IOD-integrated optical density   CDV-Ond-canine distemper virus strain Onderstepoort; pi-persistently infected; GM-CSF-granulocyte and macrophage colony-stimulating factor; IOD-integrated optical density Immunoblotting with an anti-GM-CSF antibody revealed protein bands at 14 kDa in native and acidified supernatants obtained from DH82 cell cultures infected with CDV-Ond neon-GM-CSF . The amount of GM-CSF did not differ between native and acidified supernatants obtained from DH82 cells infected with CDV-Ond neon-GM-CSF (Wilcoxon signedrank test; p > 0.05; Table 6, Figure 4). In supernatants obtained from DH82 cells infected with CDV-Ond, CDV-Ond neon , and non-infected controls, GM-CSF was not detectable in native or in acidified samples (Table 4, Figure 4). In native and acidified supernatants of CDV-Ond neon-GM-CSF -infected DH82 cells, the amount of GM-CSF was significantly higher than in supernatants from non-infected controls and DH82 cells infected with CDV-Ond and CDV-Ond neon (Mann-Whitney U test; p ≤ 0.05). Table 6. Amount of GM-CSF within DH82 cell culture supernatants (n = 3 per condition and group).

GM-CSF Produced by CDV-Ond neon-GM-CSF -Infected DH82 Cells Is Functional and Doe Affect DH82 Cell Behavior
After the verification of the presence of GM-CSF in supernatants obtained from cells persistently infected with CDV-Ond neon-GM-CSF and the confirmation that neutralization by acidification did not affect the amount of GM-CSF, the next step w analyze the effect on DH82 cells and to assess its functionality.
To assess the potential growth-stimulating effect of GM-CSF on histiocytic sa cells (DH82 cells), the presence of the GM-CSF receptor (CD116) was investigated i infected DH82 cells and DH82 cells persistently infected with CDV-Ond, CDV-O and CDV-Ond neon-GM-CSF . DH82 cells expressed CD116 independently of the inf status, and no significant differences in the CD116 protein amount were observed. ( 5; Mann-Whitney U test; p > 0.05).
To investigate the functional impact of GM-CSF on DH82 cell motility, a s wound assay was performed using non-infected DH82 cells and acidified supern obtained from non-infected DH82 cells and DH82 cells persistently infected with Ond, CDV-Ond neon , or CDV-Ond neon-GM-CSF . Commercially available canine GM (caGM-CSF) was added to the non-conditioned medium as a control. There we

GM-CSF Produced by CDV-Ond neon-GM-CSF -Infected DH82 Cells Is Functional and Does Not Affect DH82 Cell Behavior
After the verification of the presence of GM-CSF in supernatants obtained from DH82 cells persistently infected with CDV-Ond neon-GM-CSF and the confirmation that virus neutralization by acidification did not affect the amount of GM-CSF, the next step was to analyze the effect on DH82 cells and to assess its functionality.
To assess the potential growth-stimulating effect of GM-CSF on histiocytic sarcoma cells (DH82 cells), the presence of the GM-CSF receptor (CD116) was investigated in noninfected DH82 cells and DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF . DH82 cells expressed CD116 independently of the infection status, and no significant differences in the CD116 protein amount were observed. (Figure 5; Mann-Whitney U test; p > 0.05).

0.0303).
Overall, the present study demonstrated that GM-CSF in supernatan DH82 cells persistently infected with CDV-Ond neon-GM-CSF did not influ sarcoma cell proliferation and migration, but at the same time, it exerte stimulating HeLa cell proliferation in vitro.  Black arrows indicate positive cells. Bars = 400 µm. Western blot analysis of CD116 using an anti-GM-CSF receptor antibody (C). Intracellular amount of CD116 in % of GAPDH. Immunoblotting of CD116 showed no significant differences in the sizes and intensities of bands at 66 kDa in HeLa cells, non-infected DH82 cells, and DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , and CDV-Ond neon-GM-CSF (C). Similarly, the relative protein expression of CD116 did not differ between the different groups. Box plots represent minimum, median, and maximum (Mann-Whitney U test; p > 0.05) (D). GAPDH was used as a housekeeping protein. n = 3 per condition and group.
To investigate the functional impact of GM-CSF on DH82 cell motility, a scratch wound assay was performed using non-infected DH82 cells and acidified supernatants obtained from non-infected DH82 cells and DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , or CDV-Ond neon-GM-CSF . Commercially available canine GM-CSF (caGM-CSF) was added to the non-conditioned medium as a control. There were no significant differences in the percentage of scratch closure after 24 h between groups (Mann-Whitney U test; p > 0.05; Figure 6). In addition, the effect of GM-CSF on DH82 cell proliferation was examined using a cell duplication assay. The analysis yielded no significant differences among all groups (Mann-Whitney U test; p > 0.05, Figure 7). thogens 2023, 12, x FOR PEER REVIEW Figure 6. Representative images of the scratch wound assay at time point 0 ( (B1-B5). The medium of non-infected DH82 cells was supplemented with obtained from non-infected DH82 cells (A1) or DH82 cells persistently infecte CDV-Ond neon (A3), CDV-Ond neon-GM-CSF (A4), or commercially available canine A5). The scratch closure was similar in all groups after 24 h of treatment with obtained from non-infected DH82 cells (B1), DH82 cells persistently infected CDV-Ond neon (B3), CDV-Ond neon-GM-CSF (B4), or commercially available canin B5). Graphical presentation of the percentage of scratch closure (C). Box plot first quartile, median, third quartile, and maximum. Mann-Whitney U test r differences between the groups: p > 0.05. n = 4 per condition and group.

Discussion
The treatment of canine histiocytic sarcomas is often of limited success and leads only to partial or transient regressions or a delay in disease progression [41][42][43]. Therefore, therapeutic advances are greatly needed. Former studies in a murine xenotransplantation model of canine histiocytic sarcomas demonstrated a complete spontaneous regression following the transplantation of persistently CDV-Ond-infected tumor cells. Interestingly, this regression was accompanied by a decreased microvessel density [33], highlighting the importance of agents being able to modulate the tumor microenvironment. One important cytokine known to be involved in angiogenesis and anti-tumoral immune response is GM-CSF [13,16,20]. Therefore, a CDV-Ond strain was genetically engineered to express GM-CSF to enhance the oncolytic properties of this virus [10]. The present study intended to characterize the functionality of GM-CSF produced by CDV-Ond neon-GM-CSF -infected canine histiocytic sarcoma cells (DH82 cells) and explore the effect on non-infected canine histiocytic sarcoma cells in vitro.
To allow an investigation of the effect of GM-CSF produced by CDV-Ond neon-GM-CSFinfected DH82 cells, the first step was to inactivate the infectious virus, which was also present in the supernatant of these cultures, to avoid additional CDV infections, which would hamper the interpretation of observed effects. Morbilliviruses are stable at pH 4.5-9 [44], whereas GM-CSF maintains its structural stability and thus its functionality in For the verification of the overall functionality of GM-CSF produced by CDV-Ond neon-GM-CSF -infected DH82 cells, a cell duplication assay was performed with HeLa cells. HeLa cell culture media were supplemented with acidified supernatants obtained from non-infected DH82 cells and DH82 cells persistently infected with CDV-Ond, CDV-Ond neon , or CDV-Ond neon-GM-CSF . The culture media of HeLa cell control groups were supplemented with commercially available canine and recombinant human GM-CSF (caGM-CSF; rhGM-CSF). After 6 h, the cell duplication assay did not reveal significant differences between the groups (Mann-Whitney U test; p > 0.05, Figure 7). Interestingly, after 12 h, the number of cells was significantly higher in HeLa cell samples supplemented with acidified supernatants obtained from DH82 cells persistently infected with CDV-Ond neon-GM-CSF compared with samples supplemented with acidified supernatants obtained from noninfected DH82 cells (p = 0.0152) or from DH82 cells infected with CDV-Ond (p = 0.0152) or CDV-Ond neon (p = 0.0147) or supernatants supplemented with commercial caGM-CSF (p = 0.0303).
Overall, the present study demonstrated that GM-CSF in supernatants obtained from DH82 cells persistently infected with CDV-Ond neon-GM-CSF did not influence histiocytic sarcoma cell proliferation and migration, but at the same time, it exerted its function by stimulating HeLa cell proliferation in vitro.

Discussion
The treatment of canine histiocytic sarcomas is often of limited success and leads only to partial or transient regressions or a delay in disease progression [41][42][43]. Therefore, therapeutic advances are greatly needed. Former studies in a murine xenotransplantation model of canine histiocytic sarcomas demonstrated a complete spontaneous regression following the transplantation of persistently CDV-Ond-infected tumor cells. Interestingly, this regression was accompanied by a decreased microvessel density [33], highlighting the importance of agents being able to modulate the tumor microenvironment. One important cytokine known to be involved in angiogenesis and anti-tumoral immune response is GM-CSF [13,16,20]. Therefore, a CDV-Ond strain was genetically engineered to express GM-CSF to enhance the oncolytic properties of this virus [10]. The present study intended to characterize the functionality of GM-CSF produced by CDV-Ond neon-GM-CSF -infected canine histiocytic sarcoma cells (DH82 cells) and explore the effect on non-infected canine histiocytic sarcoma cells in vitro.
To allow an investigation of the effect of GM-CSF produced by CDV-Ond neon-GM-CSFinfected DH82 cells, the first step was to inactivate the infectious virus, which was also present in the supernatant of these cultures, to avoid additional CDV infections, which would hamper the interpretation of observed effects. Morbilliviruses are stable at pH 4.5-9 [44], whereas GM-CSF maintains its structural stability and thus its functionality in acidic environments [45,46]. Therefore, cell culture supernatants obtained from DH82 cells persistently infected with CDV-Ond neon-GM-CSF were acidified and inactivated at pH 2. Supernatants of all control groups (DH82 cells persistently infected with CDV-Ond and CDV-Ond neon , respectively) were treated analogously. Supernatants of non-infected DH82 cell cultures served as an additional control. Successful treatment was confirmed by the absence of cytopathic effects in Vero.DogSLAM cells upon virus titration. Furthermore, a significantly lower number of CDV mRNA transcripts was present in the acidified supernatants of infected cultures, while the protein amount of CDV nucleoprotein remained unchanged. Despite the acidification, the protein amount of GM-CSF did not change significantly compared with untreated supernatants, proving the stability of GM-CSF at low pH values.
The function of GM-CSF is transmitted by the GM-CSF receptor (CD116), which is composed of subunits α and β, and leads to activation of the Janus Kinase 2 (JAK2)-dependent signaling pathway [16]. The effect of GM-CSF depends on the concentration of available GM-CSF [16,47]. CD116 subunit α binds GM-CSF with a low affinity, whereas subunit β does not bind GM-CSF itself but forms a high-affinity receptor with subunit α [47]. At low concentrations of GM-CSF, phosphatidylinositol 3 kinase (PI3K) and mitogen-activated protein kinases (MAPK) are activated [16]. By contrast, a high concentration of GM-CSF leads to the activation of the signal transducer and activator of the transcription 5 (STAT-5)-dependent pathway [16]. The latter can lead to tumor cell proliferation [16,[47][48][49]. Therefore, the next step was to analyze CD116 expression of DH82 cells. All DH82 cells, independently of the infection status, expressed CD116, allowing them to possibly respond to growth stimulation by GM-CSF. This might bear the risk of the unintended stimulation of neoplastic cells as an adverse effect. Interestingly, despite the expression of CD116, the substitution of cell culture media of non-infected DH82 cells with acidified supernatants obtained from CDV-Ond neon-GM-CSF -infected DH82 cells or commercially available canine GM-CSF did not result in increased cell proliferation or cell motility. This might be caused by defects in the intracellular signaling pathway since GM-CSF is known to play an important role in myelopoiesis and the function of myeloid cells [16]. However, the lack of a detrimental effect on histiocytic sarcoma cells due to GM-CSF produced by genetically engineered viruses is highly desirable. This might enhance the efficacy of these viruses since GM-CSF can stimulate the non-neoplastic, infiltrating inflammatory cells of the host and evolve an amplified anti-tumor immune response.
To access the overall functionality of GM-CSF, HeLa cells were used. These cells are known to show increased proliferation after stimulation with GM-CSF [17]. HeLa cells cultured in a medium supplemented with acidified supernatants of DH82 cells persistently infected with CDV-Ond neon-GM-CSF displayed increased proliferation after 12 h of stimulation, confirming the functionality of GM-CSF produced by CDV-Ond neon-GM-CSF -infected DH82 cells.

Conclusions
To summarize, CDV-Ond neon-GM-CSF -infected DH82 cells secreted increased amounts of functionally active GM-CSF, while the proliferation and migration of histiocytic sarcoma cells was not influenced. Consequently, this genetic modification might result in an increased were capacity of this CDV-Ond strain compared with that of the non-modified parenteral strain. However, the present results only depict in vitro findings and can only partially mimic the complex interactions in the tumor microenvironment. Therefore, the functionality and efficacy of CDV-Ond neon-GM-CSF need to be further detailed in more complex situations, such as three-dimensional co-cultures of neoplastic cells with other components of the tumor microenvironment, and especially in controlled in vivo situations, such as a murine xenotransplantation model.