Phenotypic and Transcriptional Changes of Pulmonary Immune Responses in Dogs Following Canine Distemper Virus Infection

Canine distemper virus (CDV), a morbillivirus within the family Paramyxoviridae, is a highly contagious infectious agent causing a multisystemic, devastating disease in a broad range of host species, characterized by severe immunosuppression, encephalitis and pneumonia. The present study aimed at investigating pulmonary immune responses of CDV-infected dogs in situ using immunohistochemistry and whole transcriptome analyses by bulk RNA sequencing. Spatiotemporal analysis of phenotypic changes revealed pulmonary immune responses primarily driven by MHC-II+, Iba-1+ and CD204+ innate immune cells during acute and subacute infection phases, which paralleled pathologic lesion development and coincided with high viral loads in CDV-infected lungs. CD20+ B cell numbers initially declined, followed by lymphoid repopulation in the advanced disease phase. Transcriptome analysis demonstrated an increased expression of transcripts related to innate immunity, antiviral defense mechanisms, type I interferon responses and regulation of cell death in the lung of CDV-infected dogs. Molecular analyses also revealed disturbed cytokine responses with a pro-inflammatory M1 macrophage polarization and impaired mucociliary defense in CDV-infected lungs. The exploratory study provides detailed data on CDV-related pulmonary immune responses, expanding the list of immunologic parameters potentially leading to viral elimination and virus-induced pulmonary immunopathology in canine distemper.


Virus Loads and Cell Tropism
Viral loads in lung tissue at different stages of CDV infection were evaluated by immunohistochemistry. CDV nucleoprotein was absent in non-infected control dogs (group 1). Lung tissue from affected dogs showed highest numbers of CDV-infected cells in the acute (group 2) and subacute stage (group 3), while CDV was almost eliminated from lungs in the subacute-chronic phase (group 4; Figure 2A). CDV loads corre-

Virus Loads and Cell Tropism
Viral loads in lung tissue at different stages of CDV infection were evaluated by immunohistochemistry. CDV nucleoprotein was absent in non-infected control dogs (group 1). Lung tissue from affected dogs showed highest numbers of CDV-infected cells in the acute (group 2) and subacute stage (group 3), while CDV was almost eliminated from lungs in the subacute-chronic phase (group 4; Figure 2A). CDV loads correlated positively with the severity and extent of interstitial pneumonia (correlation coefficient: 0.658).

Phenotyping of Pulmonary Immune Responses
Spatiotemporal changes of cellular immune responses of CDV-infected lungs were determined by immunohistochemistry in bronchial, bronchiolar and alveolar regions at different disease phases.

Innate Immune Cell Response
A significantly increased expression of MHC-II was found in bronchial, bronchiolar and alveolar regions during the acute (group 2) and subacute phase (group 3) compared to non-infected lungs (group 1), followed by a significant decrease in the subacutechronic phase (group 4). Similarly, the number of Iba-1 + and CD204 + macrophages was significantly elevated in all three anatomical compartments in group 2 and 3 animals, and significantly decreased in lungs of group 4 animals (Figures 3 and 4). Immunohistochemistry revealed a preferential infection of the epithelium and of immune cells in the bronchiolar submucosa and alveolar area ( Figure 2B,C). Immunofluorescence double-labeling confirmed CDV infection of cytokeratin + epithelial cells as well as of submucosal and alveolar Iba-1 + histiocytes ( Figure 2D,E).

Phenotyping of Pulmonary Immune Responses
Spatiotemporal changes of cellular immune responses of CDV-infected lungs were determined by immunohistochemistry in bronchial, bronchiolar and alveolar regions at different disease phases.

Adaptive Immune Cell Response
A mild but significant increase of CD3 + T cells was observed in bronchial and alveolar regions at all disease stages compared to control dogs. CD20 + B cells were decreased in alveolar regions during the acute (group 2) and subacute phase (group 3) compared to control lungs (group 1). A significant increase of CD20 + B cells was found during the subacute-chronic phase (group 4) around bronchi and bronchioles compared to group 2 and group 3 and in alveolar areas compared to group 3 ( Figures 5 and 6).

Global Transcriptome Analysis of Canine Distemper Virus-Infected Lung Tissue
In order to gain more detailed insights into antiviral immune mechanisms, pulmonary inflammation and responses to virus-induced injury, bulk RNA-Seq transcriptome analysis of lung tissue from non-infected and acutely CDV-infected dogs was performed. In total, 3252 differentially expressed genes (DEGs, p < 0.05) were detected by direct comparison between non-infected control dogs and CDV-infected dogs (Supplementary Table S1). Hierarchical cluster analysis resulted in five clusters, grouping genes with similar expression patterns. Genes in cluster 1 (n = 353), cluster 2 (n = 515) and cluster 3 (n = 118) were upregulated in CDV-infected dogs, whereas clusters 4 (n = 1373) and 5 (n = 893) contained downregulated genes (Figure 7).   Enrichment analysis using gene ontology (GO) terms in the category "biological function", showed an overrepresentation of terms such as "defense response to virus", "negative regulation of viral process", "response to type I interferon", "mononuclear cell proliferation", "cytokine-mediated signaling pathway" and "regulation of apoptotic process" in cluster 1 and 2. No GO terms were enriched in cluster 3 ( Table 1, for details see  Supplementary Table S2).

Cytokine Expression Analyses
Selected differences detected by RNA-Seq were confirmed by RT-qPCR and immunostaining. In agreement with RNA-Seq data, RT-qPCR analysis revealed a significant increase of TNF-α, IL-6 and IL-12 transcription in CDV-infected lungs compared to tissue from control animals ( Figure 8). TNF-α expression was detected around bronchioles and within alveolar regions by immunohistochemistry. Immunofluorescence revealed Iba-1 + macrophages as a source of TNF-α production within lungs of CDV-infected dogs ( Figure 9). Spearman correlation analyses revealed a positive correlation between CDV RNA loads and IL-6 (correlation coefficient: 0.560) and TNF-α transcription (correlation coefficient: 0.633). Lung IL-2 mRNA transcription was significantly downregulated in CDV-infected dogs, while no significant changes between groups were found for IL1-β, IL-4, IL-8, IL-10, TGF-β and IFN-γ mRNA expression ( Figure 8).

Apoptosis Induction in Canine Distemper Virus-Infected Lungs
Quantification of immunohistochemical staining against cleaved caspase-3 (CC-3) confirmed an increase of apoptotic events as indicated by RNA-Seq data. Scattered CC-3 + apoptotic cells were found within airway and alveolar epithelia as well as within airway lumina of CDV-infected dogs ( Figure 10).

Apoptosis Induction in Canine Distemper Virus-Infected Lungs
Quantification of immunohistochemical staining against cleaved caspase-3 (CC-3) confirmed an increase of apoptotic events as indicated by RNA-Seq data. Scattered CC-3 + apoptotic cells were found within airway and alveolar epithelia as well as within airway lumina of CDV-infected dogs ( Figure 10).

Interferon-Related Genes
Activation of interferon-related responses obtained by RNA-Seq analysis was confirmed by immunohistochemistry for interferon-induced proteins Mx, PKR and ISG15. ISG15 was constitutively expressed in basal cells of the bronchial epithelium and to a lesser extent in bronchial glands of non-infected dogs. Upon CDV infection, airway epithelial cells, pneumocytes and alveolar macrophages showed an increased expression of Mx, PKR and ISG15. Most prevalent changes were detected in alveolar regions of infected dogs. (Figures 11 and 12).

Interferon-Related Genes
Activation of interferon-related responses obtained by RNA-Seq analysis was confirmed by immunohistochemistry for interferon-induced proteins Mx, PKR and ISG15. ISG15 was constitutively expressed in basal cells of the bronchial epithelium and to a lesser extent in bronchial glands of non-infected dogs. Upon CDV infection, airway epithelial cells, pneumocytes and alveolar macrophages showed an increased expression of Mx, PKR and ISG15. Most prevalent changes were detected in alveolar regions of infected dogs. (Figures 11 and 12).

Discussion
CDV infection causes pulmonary infection and pathology in susceptible hosts. The present study showed a transient damage of respiratory epithelia and pulmonary immune responses dominated by innate immune cells in affected dogs. Transcriptome analyses of lung tissue revealed the expression of several genes and pathways involved in antiviral immunity and control of virus infection, respectively.
In agreement with previous reports, pathologic findings in CDV-infected lungs are characterized by interstitial pneumonia with necrotizing bronchiolitis, syncytia formation and viral inclusion bodies [34,61,62]. Transcriptome analysis revealed a downregulation of genes involved in ciliary and epithelial function (cluster 4 and 5), which is in accordance with the observed epithelial pathology in investigated dogs. Disturbed mucociliary transport, either generated by the pathogen itself or excessive immune responses, inhibits self-clearance of airways and protection against invading pathogens, and has been demonstrated in mice and murine air-liquid interface cultures infected with paramyxoviruses (Sendai virus) [63,64]. Not only the pathogen itself, but also host responses, e.g. the production of reactive oxygen species (ROS) by inflammatory cells, have been shown to exhibit a negative impact on ciliary function [65]. Decreased inflammatory changes and pulmonary damage were found in the subacute-chronic phase of canine distemper, indicative of disease remission and viral elimination during the advanced infection phase. Accordingly, the number of CDV-infected cells transiently in-

Discussion
CDV infection causes pulmonary infection and pathology in susceptible hosts. The present study showed a transient damage of respiratory epithelia and pulmonary immune responses dominated by innate immune cells in affected dogs. Transcriptome analyses of lung tissue revealed the expression of several genes and pathways involved in antiviral immunity and control of virus infection, respectively.
In agreement with previous reports, pathologic findings in CDV-infected lungs are characterized by interstitial pneumonia with necrotizing bronchiolitis, syncytia formation and viral inclusion bodies [34,61,62]. Transcriptome analysis revealed a downregulation of genes involved in ciliary and epithelial function (cluster 4 and 5), which is in accordance with the observed epithelial pathology in investigated dogs. Disturbed mucociliary transport, either generated by the pathogen itself or excessive immune responses, inhibits self-clearance of airways and protection against invading pathogens, and has been demonstrated in mice and murine air-liquid interface cultures infected with paramyxoviruses (Sendai virus) [63,64]. Not only the pathogen itself, but also host responses, e.g. the production of reactive oxygen species (ROS) by inflammatory cells, have been shown to exhibit a negative impact on ciliary function [65]. Decreased inflammatory changes and pulmonary damage were found in the subacute-chronic phase of canine distemper, indicative of disease remission and viral elimination during the advanced infection phase. Accordingly, the number of CDV-infected cells transiently increased during the acute and subacute phase and only residual infection was present in the lung of dogs with subacute-chronic infection. Similarly, reduced viral loads have been shown in the spleen and brain of dogs during subacute-chronic CDV infection in previous studies [56,66,67]. Noteworthy, the reduction of inflammatory and pathologic changes in the lung contrasts with findings in the CNS of CDV-infected dogs, where inflammation progresses in the subacute-chronic phase despite of decreased viral burdens, indicating an immune-mediated neurologic disorder in the advanced infection phase [51,67,68]. Hence, the correlation between viral loads and severity of interstitial pneumonia found in the present study clearly indicates a primarily virus-mediated lung pathology in canine distemper and effective pulmonary antiviral immunity. The innate immune system is the first line of defense against pathogens and plays a major role for antigen presentation and induction of virus-specific adaptive immunity [49,69]. Results of immune cell phenotyping indicate a local activation of innate immune responses in bronchial, bronchiolar and alveolar regions of CDV-infected dogs during the acute and subacute infection phase, with MHC-II expression by immune cells and pulmonary epithelium as well as a dominance of Iba-1 + /CD204 + histiocytes. This process is assumed to be a consequence of pro-inflammatory cytokine expression, detected by transcriptome analysis and RT-qPCR, which enhances pulmonary recruitment of mononuclear cells and antigen presenting capacity [42]. In the lung, type II pneumocytes and airway epithelial cells, as part of the primary barrier, were reported to express MHC-II, albeit not as efficiently as professional antigen-presenting cells [70][71][72][73][74]. MHC-II upregulation by resident and infiltrating cells is a frequent finding in morbillivirus infection. For instance, it has been observed in CDV-and measles virus-infected brains as well as in lungs of CeMV-infected dolphins [41,51,53,60,75,76]. In chronic CNS lesions of CDV-infected dogs, MHC-II expression remained upregulated despite strongly reduced CDV loads, suggesting a trigger function of non-viral antigens for persistent neuroinflammatory responses [51].
Robust adaptive immune responses are crucial for viral elimination and disease recovery as well as protection from reinfection by induction of an immunological memory. Morbilliviruses target lymphoid cells, which causes depletion of lymphoid organs and long-lasting immunosuppression [25,77,78]. In the present study, a minor but significant infiltration of CD3 + T cells was observed in alveolar interstitial regions upon CDV infection, likely in response to pulmonary innate immune responses. By contrast, B cell numbers initially decreased within alveolar regions upon infection, which can be explained by the marked lymphotropism of CDV and induction of lymphoid apoptosis [79,80]. B cell depletion has been described also in the bronchus-associated lymphoid tissue (BALT) of measles patients and in measles virus-infected macaques, leading to B cell exhaustion and impaired humoral responses [59,81]. In the subacute-chronic phase of CDV infection, B cell repopulation was observed in lung samples, as previously described in CDV-infected lymphoid tissues [26,82]. Similarly, measles virus-infected macaques show transient leukopenia and subsequent restoration of peripheral lymphocyte populations [81]. Repopulation of lymphocytes indicates an intact proliferative capacity of lymphoid cells during the convalescent phase. However, these immune cells are supposed to be virus-specific and bystander lymphocytes, masking a depletion of pre-existing memory lymphocytes, which accounts for prolonged immune suppression and enhanced susceptibility for secondary infections after virus clearance [81]. Moreover, ex vivo experiments revealed that peripheral blood lymphocytes of measles patients fail to respond to expansion stimuli and show altered cytokine profiles during and after acute virus infection, indicating virus-induced T cell silencing despite normal lymphocyte counts (aka measles paradox) [83,84].
In order to obtain a more detailed view on pulmonary immune mechanisms in canine distemper, transcriptome analysis of lung tissue was performed. Data revealed a preferential regulation of genes related to antiviral defense mechanisms, including altered cytokine expression, cell death processes and interferon type I-related pathways. Cytokines orchestrate innate and adaptive immune responses in infectious disorders, while imbalanced cytokine expression leads to organ dysfunction and immunopathology [42,85,86]. Molec-ular analysis demonstrated an enhanced pro-inflammatory cytokine response, including TNF-α, IL-6 and IL-12 transcription, in lungs infected with CDV. TNF-α is produced mainly by macrophages and activated T cells in response to various stimuli, which is pivotal for protective antiviral immunity, but on the other hand potentially fosters immune mediated tissue damage [85,87,88]. Its beneficial effects in viral infection include the induction of antiviral immune responses by recruiting inflammatory cells and inhibiting viral replication, either directly or by inducing cell death of infected cells [46,[89][90][91]. In the lung, TNF-α regulates epithelial sodium channels in type II pneumocytes and alveolar edema development [92][93][94][95][96]. Another destructive effect of TNF-α is disruption of the alveolar epithelial barrier by death signaling [97]. TNF-α also contributes to leakage of the endothelial barrier, via induction of reactive oxygen species (ROS) production [98][99][100][101] or rearrangement of microtubules [102][103][104]. The pro-oxidative effect of TNF-α is supported by its direct inhibition of the antioxidant glutathione [105]. In the present study, double labeling showed TNF-α expression by Iba-1 + cells in CDV-infected lungs, indicating a pro-inflammatory M1 phenotype of pulmonary macrophages in canine distemper. Increased TNF-α levels have been reported also within early brain lesions of CDV-infected dogs and are thought to enhance immune mediated damage demyelination in the CNS [41,53,106]. In the spleen, TNF-α expression might induce lymphocyte death during acute CDV infection [56,107]. Similar to CDV-infected tissues, TNF-α levels are increased also in measles virus-infected human glial cells and have been demonstrated in the spleen, lung and brain of measles virus-infected children [108,109].
IL-6 and IL-12 are secreted by macrophages and dendritic cells in response to virus infection. In concert with TNF-α, they initiate Th1 responses of CD4 + T cells, which are supposed to trigger CNS immunopathology during CDV infection [41,106,110,111]. Interestingly, IL-2 was shown to be downregulated in CDV-infected lungs. IL-2 is vital for memory T cell development, and its absence contributes to the loss of immune memory and a prolonged immunosuppressive state [112,113]. Moreover, IL-2 plays a major role in Foxp3 + regulatory T cell function, maintaining self-tolerance and preventing immunopathology [56,114,115]. Reduced IL-2 transcription and disturbed T cell function is observed also in spleens during acute CDV infection of dogs [56]. Similarly, deficient IL-2 production in peripheral blood mononuclear cells was found during measles virus infection [84]. Of note, expression of programmed death ligand-1 (PDL-1) was found in CDV-infected lungs by transcriptome analysis (cluster 1) in the present study, which acts as negative regulator of T cell effector responses and IL-2 transcription [116,117]. Interestingly, blockage of the programmed cell death-1/PDL-1 checkpoint pathway by measles virus enhances effector memory T cell response in vitro and in mouse models, representing a potential target for immunotherapy [118][119][120]. Noteworthy, immunomodulatory cytokines such as IL-4, IL-10 and TGF-β showed no significant changes in CDV-infected lung tissue in the present study. These cytokines suppress M1 macrophage functions and are involved in disease remission and tissue repair [121][122][123]. Similarly, previous reports have shown that IL-4, IL-10 and TGF-β expression is limited during early CDV infection in the CNS [41,53]. In addition, no increased expression of these cytokines was detected in morbillivirus-infected lungs of cetaceans [60]. These observations support the hypothesis of an imbalanced cytokine response towards a pro-inflammatory environment in canine lungs upon CDV infection.
During viral infections, apoptosis poses an antiviral defense response that results in elimination of infected cells, thereby inhibiting viral replication and enhancing adaptive immunity [124,125]. However, lymphoid cell death also contributes to leukopenia and immunosuppression during morbillivirus infections [79,80,126]. The present study revealed increased apoptotic events, most prominent within airway epithelia. Enhanced caspase 3 activation and apoptotic cell death were detected also in the CNS and lung of dolphins following CeMV infection [60]. Apoptosis induction has been reported in vitro directly by morbillivirus infection and indirectly in bystander cells by the release of toxic factors [126][127][128][129][130].
Transcriptome analysis also revealed an activation of interferon I pathways with an upregulation of interferon-related genes (IRGs). Distributions of Mx, PKR and ISG15 expressing cells in lung tissue were determined by immunohistochemistry. Type I interferons are key contributors to an effective innate antiviral response, which are also involved in modulation of cell differentiation and growth, the promotion of apoptosis, positive regulation of adaptive immune responses [131,132]. By binding to their receptors on host cells, they induce a signaling cascade, resulting in the production of IFN-stimulated proteins, such as Mx proteins, ISG15, protein kinase R and RNAse L [133]. Mx proteins appear to induce antiviral activity by interference with viral replication after sensing of viral nucleocapsid-like structures, as has been demonstrated in several RNA virus infections, including bunyavirus, influenza virus and measles virus infection [134][135][136][137]. In respiratory epithelial cells, it triggers early inflammatory responses following influenza virus infection [138]. Elevated Mx protein expression can also be found in the CNS of CDVinfected dogs [55,139]. Interestingly, toothed whales are devoid of Mx proteins, suggesting a constrained antiviral response to morbillivirus infection [140]. ISG15 is a ubiquitin-like protein, either inhibiting or activating diverse signaling cascades, such as NF-κB and retinoic acid-inducible gene I (RIG-I) pathways by protein-binding ("ISGylation") [141,142]. Its effect varies among different species [143]. Antiviral activity by interference with viral replication has been suggested for pulmonary canine influenza virus infection and CDV brain infection [55,144]. Constitutive expression of ISG15 has been found in basal cells of the bronchial epithelium in the present study. Similarly, ISG15 is expressed in neurons and endothelial cells of non-infected canine cerebella, suggesting a function in normal protein turnover [55]. Protein kinase R inhibits viral and cellular mRNA translation via phosphorylation of eukaryotic translation initiation factor 2 alpha (eIF2α) [145,146]. By in vitro studies, it has been proposed to promote apoptosis [147]. The ISG RNAse L, which was upregulated in CDV-infected lungs, is an enzyme involved in degradation of cellular and viral RNA [148,149]. Additionally, RNAse L prevents viral entry into the host cell via interference with the cytoskeleton and propagates type I IFN response via a positive feedback mechanism by stimulation of the type I interferon inducing RIG-I pathway by its cleaved products [150,151]. Several viruses have been reported to antagonize RNAse L as an immune evasion strategy, including respiratory syncytial virus, influenza A virus and Theiler's murine encephalomyelitis virus [149,152]. Taken together, these findings are comparable with analyses of IFN-related genes during natural CDV infection in the canine CNS [55,139]. In CDV-infected canine air-liquid interface cultures, an induced disturbance of interferon signaling by blockage of the interferon-induced JAK/STAT pathway enhanced cytopathic effects and facilitated viral spread, underlining the important role of type I interferons in the defense against respiratory CDV infection [153]. These results highlight the potential of interferon-based antiviral therapy in CDV infection, as shown by in vitro studies [154]. However, morbilliviruses also developed strategies to evade the IFN response. For instance, viral V protein of CDV and MV interfere with mda-5 and STATs, important regulator molecules of the IFN response [155][156][157][158]. The variable ability of morbilliviruses to inhibit the IFN response might be explained by the influence of different factors, including viral strain, infected species, infected tissue type, interferences with other signaling pathways and differences between in vivo and in vitro systems [132,[159][160][161].
In conclusion, the present study delineates spatiotemporal phenotypic changes and lesion development in the canine lung, as well as molecular aspects of the antiviral response to CDV infection. The findings indicate the development of a pro-inflammatory environment driven by innate immune responses and impaired lymphocyte functions. The study represents the first report of CDV-related immune responses in lungs of its natural host, and lines up with previous studies in other target organs, which have shown a pro-inflammatory and potentially harmful immune response during CDV infection. The exploratory study contributes to an augmented understanding of morbillivirus pathogenesis and expands the list of immunologic parameters potentially contributing to viral elimina-tion and virus-induced pulmonary immunopathology in canine distemper, providing a broad basis for further mechanistic research.

Ethical Statement
The present study was conducted in accordance with the German Animal Welfare Act. The authors confirm that no animals were experimentally infected or sacrificed for the purpose of this retrospective pathological study. All dogs used in the present study were dead at the time of submission to routine necropsy service. Some of the control tissues were collected from dogs deriving from an animal experiment, which was approved and authorized by the local authorities (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (LAVES), Oldenburg, Germany, permission number 08A580). All dog owners provided written consent for the collection of the dogs' tissue and its usage for research purposes.

Animals, Tissue Samples and Processing
Lung tissues from 37 dogs with natural CDV-infection (groups 2-4) and 7 non-infected control dogs (group 1) were examined by histology, immunohistochemistry and immunofluorescence. Snap frozen lung tissues from nine CDV-infected dogs and six non-infected control dogs were used for molecular analyses (bulk RNA sequencing and RT-qPCR). Clinical signs of CDV-infected dogs included seizures, ataxia, diarrhea, vomitus, dyspnea, coughing, fever and occasionally nasodigital hyperkeratosis. They either died spontaneously or were euthanized due to poor prognosis. CDV infection was confirmed post mortem via immunohistochemistry. Age, sex, breed and mode of death of dogs are listed in Supplementary Tables S3 and S4. During necropsy, lung tissue was collected and either immersion fixed in 10% neutrally buffered formalin or mounted in O.C.T.™ embedding compound (Tissue Tek ® ; Sakura Finetek Europe, Alphen aan den Rijn, The Netherlands) and stored at −80 • C until use for molecular analyses. To generate formalin-fixed and paraffin embedded slides, formalin-fixed tissue was dehydrated by ascending series of alcohols and subsequently embedded in paraffin (Thermo Fisher Scientific, Langenselbold, Germany). Then, 2-4 µm thick serial sections of lung tissue and cerebellum were cut with a rotary microtome (Leica Biosystems, Wetzlar, Germany), mounted on Superfrost ® Plus slides (Gerhard Menzel, Braunschweig, Germany) and either stained with hematoxylin and eosin (HE) for histological evaluation or subjected to immunohistochemistry or immunofluorescence.

Classification of Disease Phases
Disease phases were determined based on the type of white matter lesions in the cerebellum. As shown before by experimental infections, CDV-induced leukoencephalitis in dogs develops in a sequential order. Acute lesions, characterized by white matter vacuolization and glial infection in the cerebellum, can be observed 16-24 days post infection. Subacute lesions with demyelination but without perivascular lymphohistiocytic cuffs occur 24-32 days after infection. Subacute to chronic lesions with demyelination together with perivascular lymphohistiocytic cuffs and reduced numbers of CDV + cells can be found after a minimum of 29-63 days post infection in the brain of infected dogs [37,[162][163][164][165][166][167]. Accordingly, dogs were classified into four groups: group 1 consisted of non-infected dogs without CNS disease (control); group 2 included dogs with cerebellar lesions with focal vacuolization and gliosis (acute phase); group 3 comprised dogs with demyelinating encephalitis without perivascular mononuclear infiltrates (subacute phase); group 4 included dogs with demyelinating encephalitis and perivascular mononuclear cuffing (subacute-chronic phase) [56,66].

Histological Scoring of Lung Lesions
HE-stained lung sections were evaluated independently by four scientists (E.C., P.P., J.K. and A.B.) for the intensity of inflammatory changes (mononuclear cell infiltration, type II pneumocyte hypertrophy and hyperplasia, bronchial/bronchiolar necrosis, fibrin extravasation and edema) and the presence of syncytia and viral inclusion bodies by light microscopy (Carl Zeiss, Jena, Germany). A semi-quantitative score was generated from the severity and extent of interstitial pneumonia as follows: 0 = no changes, 1 = severity: mild changes/extent: >25% of section affected, 2: severity: moderate changes/extent: 25-50% of section affected and 3: severity: marked changes/extent: >50% of section affected. Scores of both parameters were added.

Immunohistochemistry
CDV antigen detection and phenotyping of cellular immune responses was performed using the avidin-biotin complex method as described previously with slight modifications [51]. For ISG15, the EnVision visualization system was used. Respective primary antibodies are detailed in Table 2. In brief, formalin-fixed and paraffin-embedded (FFPE) slides were deparaffinized by ROTICLEAR ® (Carl Roth, Karlsruhe, Germany) and rehydrated through a series of graded alcohols for 2-3 min each. To suppress endogenous peroxidase activity, they were incubated with H 2 O 2 (0.5%) in 85% ethanol for 30 min, followed by antigen retrieval in citrate buffer (pH 6.0) for 20 minutes in a microwave (800 W). Unspecific reactions with the secondary antibody were blocked by incubation with goat normal serum for 20 min at room temperature (except ISG15). The primary antibodies (CDV-NP, MHC-II, Iba-1, CD204, CD3, CD20, TNF-α, cleaved caspase-3 (CC-3), Mx, PKR and ISG15), diluted in phosphate-buffered saline (PBS) with 1% bovine serum albumin (BSA, Carl Roth GmbH), were incubated overnight (18 h) at 4 • C. Negative controls were treated with ascites fluid from non-immunized BALB/c mice (CDV-NP, MHC-II, CD204, TNF-α and Mx) or serum from non-immunized rabbits (Iba-1, CD3, CD20, CC-3, PKR and ISG15) instead of the primary antibody. As positive controls, immunohistochemically confirmed CDV-positive canine lung and cerebellar tissue (CDV-NP, Mx, PKR and ISG15) and canine lymph node tissue (MHC-II, Iba-1, CD204, CD3, CD20, TNF-α and CC-3) were used. Secondary labeling was performed using polyclonal biotinylated antibodies diluted 1:200 in PBS (goat anti-mouse, Vector Laboratories, Burlingame, CA, USA, BA-9200; goat anti-rabbit, Vector Laboratories, BA-1000) for 45 min at room temperature, followed by a treatment with the avidin-biotin-peroxidase complex (Vectastain Elite ABC Kit, Vector Laboratories) for 20 min at room temperature. For ISG15, the EnVision+ anti-rabbit HRPlabelled polymer antibody (Dako, Glostrup, Denmark; K4003) was incubated for 30 min at room temperature instead of using the biotinylated antibodies and the avidin-biotinperoxidase complex. Subsequently, the reaction was visualized by incubation in 0.05% 3.3 -diaminobenzidine tetrahydrochloride (DAB, Carl Roth) in PBS and H 2 O 2 (0.03%) for 5 min at room temperature. Nuclei were counterstained with Mayer's hemalum solution (Carl Roth) for 30 s with subsequent dehydration in 70% ethanol, 96% ethanol, isopropanol and n-Butyl acetate.  Immunoreactivity in lung sections was evaluated quantitatively by counting absolute numbers of positive cells using a morphometric grid (number of positive cells/0.0625 mm 2 ) in 10 randomly chosen high power fields within affected and control lungs. Subsequently, the mean value of immunopositive cells per 0.0625 mm 2 was calculated.

Immunofluorescence Double Labeling
To determine the cell tropism of CDV in infected lungs, immunofluorescence labeling of CDV nucleoprotein, either combined with Iba-1 (histiocytic cell antigen) of pancytokeratin (CK, epithelial cell antigen) was performed on lung tissue of 12 CDV-infected dogs. Additionally, immunofluorescence was applied to demonstrate TNF-α production in Iba-1 + macrophages in lung sections (TNF-α/Iba-1 double labeling). Used primary antibodies are listed in Table 2. Following deparaffinization by ROTICLEAR ® (Carl Roth), rehydration through graded alcohols and rinsing in PBS with a stirring tool (3 × 5 min), the slides were either pretreated with citrate buffer (pH 6.0, CDV/Iba-1 and CDV/CK) for 20 min or Tris-EDTA buffer (pH 9.0, TNF-α/Iba-1) for 30 min in a microwave (800 W). Unspecific bindings of the secondary antibody were blocked by 20% goat normal serum in PBS with 1% BSA and 0.1% Triton-X100 (Sigma-Aldrich, St. Louis, MO, USA) for 30 min. Both primary antibodies were diluted simultaneously in PBS with 1% BSA and 0.1% Triton X-100 and were incubated overnight (18 h) at 4 • C. Negative controls were treated with ascites fluid from non-immunized BALB/c mice and rabbit normal serum instead of the primary antibodies. After washing thrice in PBS, the secondary polyclonal antibodies (Alexa Fluor ® 488-conjugated goat anti-mouse (Jackson ImmunoResearch Europe, Ely, UK, 115-545-003) and Cy3-conjugated goat anti-rabbit (Jackson ImmunoResearch Europe, 111-165-144)) were diluted at 1:200 in PBS with 1% BSA and 0.1% Triton X-100 and subsequently incubated for 45 minutes at room temperature in the dark, followed by rinsing in PBS for 3 × 5 min. The slides were washed twice with distilled water preceding an autofluorescence-reduction treatment (Vector TrueVIEW Autofluorescence Quenching Kit, Vector Laboratories, Burlingame, CA, USA) for 5 min, followed by rinsing thrice with PBS and twice with distilled water. Nuclei were counterstained with bisbenzimide Hoechst 33,258 (1:100 in sterile bidistilled water; Sigma-Aldrich Chemie, Taufkirchen, Germany) for 8 min, and slides were mounted with fluorescence mounting medium (Dako, Glostrup, Denmark), followed by curing and storage in the dark at 4 • C until inspection.

Molecular Investigations
In order to investigate immune responses and CDV RNA loads on a molecular level, RNA sequencing analysis and quantitative reverse transcription polymerase chain reaction (RT-qPCR) were performed.

RNA Isolation
OCT-embedded frozen lung tissue from CDV-infected dogs and non-infected control dogs as well as frozen lymph node tissue from healthy dogs (used for controls) was cut with a cryostat microtome at 50 µm (Leica CM1950, Leica Biosystems Nussloch GmbH, Nussloch, Germany). Isolation and purification of total RNA was achieved using the RNeasy ® Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions, including an on-column DNAse treatment (Qiagen). The obtained RNA amount was calculated by measuring the optical density at 260 nm with a spectrophotometer (Multiskan TM GO microplate spectrophotometer, µDrop TM plate, SkanIt TM software version 5.0.0.42, Thermo Fisher Scientific, Braunschweig, Germany).

RNA Sequencing Analysis
For analysis of transcriptional changes during CDV-infection in the lung, the isolated RNA of four control animals and five acutely infected dogs was selected. Quality and integrity of total RNA was controlled on Agilent Technologies 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). The RNA sequencing library was generated from 50ng rRNA depleted total RNA (QIAseq FastSelect rRNA HMR, Qiagen) using NEB Ultra II Directional-RNA Seq Library-Kit (New England Biolabs, Ipswich, MA, USA) according to the manufacture's protocol. The libraries were sequenced on Illumina NovaSeq 6000 using NovaSeq 6000 S1 Reagent Kit (100 cycles, paired end run) with an average of 5 × 10 7 reads per RNA sample. Each FASTQ file obtained a quality report generated by FASTQC tool. Before alignment to the reference genome, each sequence in the raw FASTQ files was trimmed on base call quality and sequencing adapter contamination using Trim Galore! wrapper tool. Reads shorter than 20 bp were removed from the FASTQ file and trimmed reads were aligned to the reference genome (Canis_lupus_familiaris.ROS_Cfam_1.0,  [169,170]. Expression data was log2 transformed and TMM normalized followed by calculation of differential gene expression by the R package edgeR (version 3. 38

Reverse Transcription Quantitative PCR (RT-qPCR) Analysis
Quantities of CDV cDNA loads and mRNA expression of cytokines in CDV-infected lung samples and uninfected controls were assessed together with the standard dilution series and no template controls via RT-qPCR using the AriaMx Real-Time PCR System (Agilent Technologies; Agilent Aria software version 1.71). The Brilliant III Ultra-Fast SYBR ® Green QPCR Master Mix (Agilent Technologies) was used following the manufacturer's instructions with primers at a concentration of 200 nmol/L and carboxy-X-rhodamine (ROX) as a reference dye. Annealing temperatures were adjusted to 57 • C (CDV), 60 • C (IL-1-β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, TNF-α, TGF-β and IFN-γ) or 64 • C (GAPDH). Copy numbers were calculated by comparing with the standard curves and were normalized against GAPDH as a housekeeping gene. Reaction specificity was assessed by melting curve analysis.

Statistical Analysis
Statistical analysis of non-normally distributed data obtained by immunohistochemistry and quantitative PCR was performed by employing the non-parametric Kruskal-Wallis H test for two independent samples using the IBM "Statistic Package for Social Sciences" SPSS program for Windows (version 26, SPSS ® , IBM, Ehningen, Germany). To detect a possible correlation between CDV loads and interstitial pneumonia and the cytokine transcription, respectively, the Spearman rank correlation coefficient was calculated for all investigated variables. p-values ≤ 0.05 were designated as statistically significant when comparing differences between groups. Graphs were generated using GraphPad Prism ® for Windows (version 9.9.0, GraphPad Software, San Diego, CA, USA).