Mechanisms behind the varying severity of Aleutian mink disease virus: comparison of three farms with a different disease status

Aleutian mink disease virus (AMDV) is distributed widely among mink farms and wild mustelids despite ongoing attempts to stop the spread. The severity of Aleutian disease (AD) varies from subclinical to fatal but the reasons for its varying severity are complex and unclear. Recently, breeding of tolerant mink has drawn attention as the possible solution to reduce the effects of AD in farms. The aim of this study was to gather information on the effects of breeding based on overall health, production traits, and antibody titer on AD severity by comparing a positive farm (farm 1) that has been breeding for tolerance in mink to an infected farm without tolerance selection, and an AMDV-free farm. During the 2.5-year follow-up, the mink in farm 1 remained mostly free of clinical AD, had normal pelt quality and litter size, and had low virus copy numbers in tissues and low antibody titers in ELISA. In histopathological studies, most of the farm 1 mink had no/mild lesions in their kidneys. 29-43% of the mink were ELISA negative but PCR positive throughout the follow-up and frequent changes in virus strains and coinfections were observed. Several differences in gene expression between animals from different farms were also detected. These results indicate that the disease burden of AMDV can be reduced, with seemingly normal health and production rates, despite continual circulation of ADMV in cases where eradication attempts are unsuccessful.


Introduction
Aleutian mink disease virus (AMDV), species Carnivore amdoparvovirus 1 and family Parvoviridae, is widespread among farmed and feral mink (ICTVdb, 2021).AMDV has a 4.8 kb ssDNA genome that encodes ve proteins.Left open reading frame (ORF) encodes structural proteins NS1, NS2, and NS3 that are needed for viral replication and right ORF encodes structural proteins VP1 and VP2 that form the capsid (Alexandersen et al., 1988;Bloom et al., 1990;Bloom et al., 1988;Bloom et al., 1980Bloom et al., , 1982;;Christensen et al., 1995;Huang et al., 2014).AMDV is transmitted horizontally directly or indirectly via body uids like blood, feces, urine, and saliva or vertically through the placenta (Gorham et al., 1976;Jensen et al., 2014;Padgett et al., 1967).It is resistant to many standard physical and chemical treatments making it di cult to clear it from infected farms (Cho, 1976;Hahn et al., 1977;Hussain et al., 2014).AMDV has been found in a wide range of surfaces in infected farms, including surfaces that are not in direct contact with infected animals (Prieto et al., 2017).
AMDV causes Aleutian disease (AD), an immune complex disease characterized by a massive number of antibodies and immune complexes that accumulate in tissues (Porter et al., 1969).Clinical signs and lesions in adult mink include anorexia, weight loss, reproductive failure, splenomegaly, lymphadenopathy, plasmacytosis, hypergammaglobulinemia, necrotizing arteritis, plasma cell-dominated mononuclear cell in ltrates in the organs, and glomerulonephritis, and vary in severity from transient to persistent and asymptomatic to fatal (Eklund et al., 1968;Porter, 1986).Antibodies are unable to neutralize the virus and no immunity from future infections is gained (Hadlow et al., 1984;Porter et al., 1969).There is no effective treatment and all attempts to develop a vaccine have been unsuccessful.Vaccines based on inactivated virus or capsid proteins and treatment with passive antibodies have only enhanced the disease, and vaccines based on NS1 led to partial protection, including lower death rates, that was insu cient to prevent the infection (Markarian and Abrahamyan, 2021).
Varying disease severity has been suggested to be connected to both viral and host factors.For example, the highly pathogenic Utah strain usually causes a severe disease in both Aleutian and non-Aleutian mink whereas the Pullman strain causes severe disease mainly in Aleutian mink (Hadlow et al., 1983;Porter, 1986).Amino acid differences in capsid protein sequences between pathogenic strains and nonpathogenic AMDV-G have often been con ned to a relatively small number of residues and a study by Kowalczyk et al.only detected one amino acid difference with the potential to affect the functionality of the protein in strains from two farms with subclinical and clinical infections (Bloom et al., 1988;Kowalczyk et al., 2018;Oie et al., 1996).It has been suggested that stabilization of the pathogen-host relationship plays a role in the clinical picture in farms as viral strains that eliminate the host are likely selected against as they do not provide for long-term transmission to be established in the population (Kowalczyk et al., 2018).When it comes to host factors, mink age and genotype both play a role in pathogenesis.In mink kits, disease manifests as acute interstitial pneumonia instead of the classic adult form of the disease (Alexandersen, 1986).Mink homozygous recessive for Aleutian gene (aa) appear to be more susceptible to the disease than other genotypes and it has been estimated that ¼ of non-Aleutian mink can clear the virus (Eklund et al., 1968).Selecting mink based on phenotypic health and iodine agglutination test (IAT) has been shown to reduce the severity of lesions, and several farms in Canada have applied selection of disease tolerant mink as a form of disease control (Farid and Ferns, 2017;Farid and Hussain, 2020).Selecting mink that have low positive ELISA values has also been recognized (Andersson et al., 2017).
The aim of this project was to study the mechanisms behind varying clinical pictures by conducting a 2.5-year follow-up on asymptomatic mink from a farm that had been controlling the disease for decades through breeding.We studied viral loads, genetic properties, transcriptomes from blood, pathological properties, and antibody response, and compared them to symptomatic mink from a freshly infected farm and healthy mink from an AMDV-free farm.

Samples
Mink from three farms were sampled.Farm 1 had had AMDV since 1980s and had been aiming to control the disease by selecting weak ELISA-positive but asymptomatic animals with normal litter size and pelt quality for breeding.Before ELISA, selection was done with iodine agglutination test.The farm reported having only an occasional mink with the clinical form of the disease and litter size was comparable to the average litter sizes in Finland.Farm 2 had had AMDV for under ve years and had observed mink with a clinical AD, while farm 3 was an AMDV-free farm.Rectal swabs and blood samples on lter paper were collected from farm 1 in May 2017, October 2017, October 2018, and November 2019, starting with 10 white, brown, and sapphire mink but ending with six brown, ve sapphire, and six white mink since 13 mink for unreported causes died during the follow-up.Five white mink showing signs of anorexia and dehydration were sampled from farm 2 and seven healthy white mink from farm 3.For the 2019 sampling in farm 1 and samplings in farms 2 and 3, mink were sedated with intramuscular dosing of 0.4 ml of 10 µg/ml medetomidine (Domitor, Orion Pharma) combined to 0.4 ml of 50 mg/ml ketamine (Ketalar, P zer animal health) after which serum (BD Vacutainer) and blood samples (Tempus blood RNA tubes, applied biosystems) were taken by cardiac puncture.Mink were euthanized with 2 ml of intracardial pentobarbitalsodium (Euthasol, 400 mg/ml).The health of the mink before samplings were assessed, samples taken, and the necropsies performed by a veterinarian.

Pathological examination
Mink (n=17) from farm 1 and two of the control mink from farm 3 were submitted for necropsy for gross and histopathological examinations.Body weight and weight of the spleen of each mink were measured.Spleen, kidneys, liver, lungs, heart, stomach, small intestine, colon, thyroid glands, adrenals, bladder, muscle, mesenteric lymph nodes, brain, hypophysis, bone marrow, and all other tissues with possible abnormalities based on macroscopic evaluation were sampled for histopathological studies.The samples were xed in 10% neutral buffered formalin, processed routinely, and stained with haematoxylineosin.Spleen and kidneys were also sampled for PCR.Detected cross and histopathological changes in organs were scored from 0-3 (0=no lesions, 1=mild lesions, 2=moderate lesions, and 3=severe lesions).

DNA extraction
DNA was extracted from spleen and kidney tissues with NucleoSpin Tissue kit (Macherey-Nagel) with a standard protocol for tissue samples and from blood samples by incubating approximately 1 cm 2 piece of lter paper in 300 µl of PBS o/n at 4 ˚C and using support protocol for viral DNA from blood samples for DNA extraction.DNA was extracted from stool samples with QIAamp Fast DNA Stool Mini Kit (Qiagen) or with QIAQube HT using DNeasy 96 PowerSoil Pro QIAcube HT Kit.

PCR, quantitating, and sequencing of virus strains
Blood samples from May 2017, October 2017, and October 2018 collections in farm 1 were tested with pan-AMDV-PCR, which ampli es nt 578-951 (Jensen et al., 2011;Virtanen et al., 2019) and pan-AMDO-PCR, which ampli es nt 1662-2302 (Knuuttila, 2015) (sites according to AMDV-G (M20036.1)throughout the manuscript).All pan-AMDV products were sequenced with Sanger sequencing to be used in phylogenetic analysis.Pan-AMDO results were analyzed based on ampli cation curves, melting curves, or sequencing.Those samples that were positive with at least one PCR were considered positive.Spleen and blood samples from November 2019 collection were tested with pan-AMDV-PCR and all positive results were sequenced.PCR was also performed for kidney samples, and positive kidney samples that had been negative in PCR from spleen were sequenced.All the real-time PCRs were performed with Stratagene Mx3005P (Agilent Technologies).PCR products were puri ed for sequencing by adding 0.5 µl of Exonuclease I and 1 µl of FastAP Thermosensitive Alkaline Phosphatase (Thermo Scienti c) to 5 µl of PCR reaction and incubating them at 37 ˚C for 45 min and 85 ˚C for 15 min.Spleen samples, kidney samples, November 2019 blood samples, and all feces samples from those mink that were alive by the end of the follow-up were also tested with quantitative NS1-probe-PCR, and DNA copy numbers were calculated (Virtanen et al., 2020).Each run contained a dilution series (10 4 , 10 3 , 100, 10, 1, and 0 copies/reaction) of a plasmid containing the PCR product prepared earlier (Virtanen et al., 2020) in three parallel reactions and the samples in two parallel reactions.DNA concentrations were measured with NanoDrop.If both parallel reactions were positive, their average Ct-value with a 0.04 cut off was used in calculations.If one reaction was positive and the other one was negative, the negative reaction was excluded.

Serology
All the blood on lter paper samples and serum samples was tested with AMDV VP2 ELISA (Knuuttila et al., 2009).Blood was extracted from lter paper by incubating a circular piece in 200 µl of dilution buffer (PBS + 0.5% BSA + 0.05% Tween 20) o/n, and undiluted liquid was used for ELISA.Sera were diluted 1:200.Peroxidase-conjugated A niPure Goat Anti-cat IgG (H + L) (Jackson ImmunoResearch) with 1:4500 dilution or Goat anti-ferret IgG (H + L) secondary antibody (Novus) with 1:20 000 dilution was used as a conjugate.ELISA cut-off was determined with a panel of 10 negative samples in seven replicates and adding two standard deviations to the mean absorbance.Each ELISA analysis included at least two blank wells (with dilution buffer only), the mean value of which was subtracted from the sample absorbances before the analysis.

Testing for mink circovirus
Due to indications that circoviruses might increase the severity of secondary infections, the mink were tested for mink circovirus (MiCV).PCR reactions contained 12,5 µl of Fermentas SYBR MasterMix (Thermo Scienti c), 6 µl of H2O, 0.75 µl 10 of µM primers CapF and CapL (Cui et al., 2018) and 5 µl of DNA isolated from spleen.The mixture was incubated at 10 min at 95 ˚C, followed by 40 cycles of 30 s at 95 ˚C, 1 min at 55 ˚C and 1 min at 72 ˚C, and melting curve analysis with 1 min at 95 ˚C, 30 s at 55 ˚C and 30 s at 95 ˚C.PCR products with unclear results were puri ed as earlier and sequenced with Sanger sequencing as described above.

Data analyses
Statistical analysis was performed with IBM SPSS statistics 27.Normality of the data was assessed with Shapiro-Wilk test and equality of the variances with Levene's test (signi cance level of 0.05).Differences of means between independent groups of normally distributed data were tested with Independent-Samples T-test (two groups) or one-way ANOVA (more than two groups).Pairwise differences of one-way ANOVA were determined with Bonferroni analysis.If the data was not normally distributed, corresponding analysis was performed with Mann-Whitney U-test (two groups) or Kruskal-Wallis test (more than two groups).Correlations between variables were assessed with Spearman's rho values.
Poor quality Sanger sequences and sequences that seemed to have two or more overlapping sequences were excluded from the sequence data and the rest of the AMDV sequences were submitted to GenBank under accession numbers OM142153-OM142203.Sequences were aligned with MEGA6 (Tamura et al., 2013) using ClustalW (Thompson et al., 1994) together with Finnish strains published in GenBank and a representative of strains from other countries (picked based on a tree built with all published AMDVsequences of nt 578-951 (Virtanen et al., 2019)).Phylogenetic tree without molecular clock was built with BEAST 1.8.2 (Drummond et al., 2012), Tracer v1.6 (Rambaut et al., 2014), and FigTree v1.4.2 (Rambaut, 2014) with a 20,000,000 chain length, Hasegawa-Kishino-Yano model (HKY+G), and Bayesian skyline.To analyze the molecular clock, an additional tree, containing all Finnish sequences with known sampling years and months, was built using a lognormal relaxed clock.Effective sample size values were checked to be over 100.

IgG levels were consistently below the detection limit in several asymptomatic PCR positive mink
The proportion of antibody positive mink in farm 1 varied between 44-71% and PCR positive mink between 52-100% during the follow-up period (Table 1, detailed information about individual samples in Table S1).30% (8/27) of mink from farm 1 that were ELISA negative were PCR positive in March 2017, 29% (8/28) in October 2017, 43% (9/21) in October 2018, and 41% (7/17) in November 2019.All mink from farm 2 were positive and farm 3 were negative both in ELISA from serum and PCR from spleen and kidney.When IgG levels of different color types of farm 1 are compared, there was a statistically signi cant difference between white and brown mink in October 2018 (p=0.027,detailed information of statistical test in Table S2) but not between other color types or other time points or between farms 1 and 2 (Fig. 1).Only up to 30% of tested feces samples were PCR positive.13/17 mink from farm 1, all mink from farm 2, and one out of two tested mink from farm 3 were MiCV positive.Due to almost all the mink being positive, correlations between MiCV and AD severity were not further analyzed.3.2.Mild histopathological lesions were detected in asymptomatic, AMDV-positive mink.
In the necropsy, 14 mink from farm 1 were classi ed as fat, two as in normal body condition, and one (F1/D) as thin.Both control animals from farm 3 classi ed as fat.Spleen weight as a percentage of body weight was higher in sapphire (mean=0.81%,SD=0.21, n=5) mink as compared to brown (mean=0.29%,SD=0.11, n=6, and p=0.026) and white mink (mean=0.24%,SD=0.086, n=6, and p=0.033) (Fig. S2).In farm 3, values were similar to those of white mink from farm 1 (mean=0.21%,SD=0.038, n=2).In gross pathological examination, grey mink F1/D had ndings typical for AD i.e., cachexia, dehydration, pale, enlarged and mottled kidneys, and splenomegaly (Fig. S3).No signi cant macroscopic lesions typical of AD were detected in other mink from farm 1 or AMDV negative control mink from farm 3, but mild unspeci c changes were observed in some of the mink (fatty liver, hyperemia in gastrointestinal track, occasional enlarged lymph nodes etc.).
In histopathological examination, both AMDV negative control mink had mild or moderate lesions in lungs (peribronchial and perivascular in ltration of mononuclear cells) and spleen (extramedullar hematopoiesis) (means=1.50).One of them had mild chronic eosinophilic enteritis in the intestines and one mild lipidosis in the liver (mean=0.5).Neither of the mink had lesions in kidneys, brain and meninges, or in other organs.In farm 1, lesions were most severe in the spleen, followed by liver and lungs.Lesions were also detected in kidney and brain (Table 2, Fig. S4 and S5, and Table S3).The clinically sick mink (F1/D) had typical lesions for Aleutian disease, i.e., severe arteritis in various organs, severe glomerulonephritis, and moderately increased number of plasma cells in the spleen.Accumulations of mononuclear cells, predominantly plasma cells on heart, liver, kidneys, and gastrointestinal tract, as well as perivascular cu ng in the meninges, were also detected.Lesions in clinically healthy mink were generally milder or absent (Fig. 2).Mild to moderate non-speci c changes were observed in the kidneys (calci cation, chronic interstitial nephritis), spleen (extramedullar hematopoiesis, congestion), liver (chronic (neutrophilic) cholangiohepatitis, lipidosis), lungs (congestion, alveolar odema), and intestines (chronic eosinophilic enteritis).No changes were observed in skeletal muscle, thyroid, or adrenal glands in any of the mink and they were not included in out analysis.No notable difference between color types was observed.Results of individual mink can be viewed in Table S1 and results sorted by color type in Table S3 3.3.AMDV genome copy number in tissues was higher in a farm with clinical disease In total, all blood samples, 12/17 spleen samples, and 15/17 kidney samples of mink from farm 1 were PCR positive in the last sampling.AMDV genome copy numbers (as copies/ng of DNA) were successfully quantitated from all blood samples, 13 kidney samples, and 8 spleen samples.In the other samples, the amount of DNA was too small to be quanti ed.All ve spleen and kidney samples from farm 2 were PCR positive and successfully quanti ed.Comparison of AMDV genome copy numbers between farms, tissues, and color types are presented in Table 3 and Fig. 3. Copy number was signi cantly higher in farm 2 as compared to farm 1 in both tissues (spleen: p = 0.006, kidney: p = 0.011).There was no statistically signi cant difference between spleen and kidney (farm 1: p = 0.352, farm 2: p = 0.686) or between color types of farm 1 (p = 0.609).Virus strains changed in at least nine of the mink during the follow-up based on the sequencing from blood.From the 2019 sampling, usable sequences from both blood and tissue (spleen or kidney) were acquired from seven mink.In three of those (F1/20, F1/B, and F1/D), the sequences in spleen/kidney and blood were similar.In mink F1/1, there was a 13 nt difference between sequences from spleen and blood, in F1/28 the difference was 2 nt, in F1/29 it was 8 nt and in F1/E it was 7 nt.
Based on molecular clock analysis (Fig. S9

Transcriptomes
421 differentially expressed genes were detected (out of 30220 genes) between farms 2 and 3 (Table S4).237 of those were upregulated and 184 were downregulated in farm 2. When farms 1 and 2 are compared, 531 genes were differentially expressed with 273 being downregulated and 294 being upregulated in farm 2. Out of 199 genes that were differentially expressed between farms 1 and 3, 63 were upregulated and 136 downregulated in farm 1.Only seven genes (DDIT4, DTHD1, GZMA, LOC122900174, LOC122915485, STYK1, and WSCD2) were upregulated in both infected farms (as compared to uninfected mink of farm 3) and none were downregulated in both infected farms.HPGD and NNMT were upregulated in farm 1 and downregulated in farm 2. Out of these genes, DDIT4 and GZMA are involved in immune response and the rest in other cellular processes.
In addition to the highly differentially expressed genes listed above, several other genes related to immune response were also upregulated in mink from farm 2. These include IL-1β, a macrophageproduced proin ammatory cytokine and IL-27 that has several pro-and anti-in ammatory properties like

Discussion
In this study, we report a comparison of mink from three farms, one of which had been living with AMDV for decades and had mostly asymptomatic mink, one that had been infected for a couple of years and had mink with clinical AD, and one that was AMDV-free.The results show correlations and differences between antibody response, viremia, virus copy numbers, pathology, and transcriptomes between the tree farms.We also conducted a 2.5-year follow-up in farm conditions and observed a high proportion of mink that remained ELISA negative but PCR positive throughout the follow-up.
Due to the lack of vaccine or treatment, AMDV control relies greatly on diagnostics with serological tests like CIEP and ELISA.Interestingly, up to 43% of ELISA negative mink were PCR positive from blood in our follow up.A high proportion of antibody-negative and PCR positive mink has also been noted earlier by Farid et al. who found that 16.5% and 40.0% of CIEP negative mink were PCR positive from spleen (Farid, 2020;Farid and Ferns, 2017).In China, 12.5% of CIEP negative mink from 5 farms were PCR positive from spleen (Wang et al., 2014) and in Sweden, 4.5% of free-ranging mink were ELISA negative but PCR positive from spleen (Persson et al., 2015).As the sensitivity of ELISA has been shown to be comparable to that of CIEP (Andersson and Wallgren, 2013;Knuuttila et al., 2009), differences in the protocol are unlikely to fully explain the high proportion in our study.One possible explanation for the ELISA negative results of AMDV-infected individuals would be fresh infections, as detectable antibody response may form later than detectable viremia (Farid et al., 2015).However, it is most likely not the only explanation, as many of the mink remained ELISA negative throughout the follow-up.It is also likely that decades of breeding have led to the selection of low antibody producers, as was also suggested by Farid et al. (Farid and Ferns, 2017).It should be noted that ELISA negative results does not necessarily mean no AMDV antibodies at all, but it means that they were below the detection limit that was determined using a commonly accepted guideline.It has also been shown that in some cases (e.g.low inoculum doses), antibody titer may decrease over time (Farid and Hussain, 2020), but as these mink are constantly reexposed by diverse strains circulating in the farms, that is not the most likely explanation for the low antibody titers in this study.Mink from farm 1 showed some uctuation in antibody titers and also occasional high titers despite being asymptomatic, which is not unheard of, as asymptomatic non-Aleutian mink may also have at least transiently high antibody titers (Bloom et al., 1975).
Due to decades of breeding, it is di cult to say how common consistently seronegative but PCR positive mink are in other farms, but false negative ELISA/CIEP results would explain the unsuccessful attempts to fully eradicate AMDV from infected farms.Further studies should be conducted on ELISA/CIEP negative mink in other AMDV infected farms with different disease status (freshly infected naïve animal population vs long-term infected farm with established pathogen-host relationship) to determine the extent of this phenomenon, as false negative results in diagnostics greatly hinder the eradication attempts.These results and the results from previous studies (Farid and Ferns, 2017) also bring up the option to co-exist with the virus by breeding tolerant mink as eradication has proven to be di cult and no fully effective vaccine exists.However, virus may cause problems when entering the naïve non-tolerant mink population, which is supported by the fact that farm 1 reported usually temporarily having more symptomatic mink when new mink were introduced into the farm.
Previous studies have shown that AMDV can persist in tissues even when viremia in blood, feces, and mouth is transient (Farid and Hussain, 2020;Jensen et al., 2014).After the rst sampling, we consistently detected virus in blood samples, which is somewhat contradictory to that nding, but may be explained by the constant re-exposure to the virus in this study as compared to experimental infections.AMDV was, on the other hand, only transiently detected in feces which indicated that mink did not frequently shed the virus in feces.Phylogenetic analysis from blood showed the virus strains frequently changed between samplings.Different sequences in different samplings can result either from virus evolution within the host (Canuti et al., 2016;Virtanen et al., 2019) or from clearance of one virus strain and infection with another.In this case, the reason is probably a combination of both.In many cases, virus strains locate in totally different branches, with tMRCA being up to 21 years in different samplings, which speaks more for virus clearance and reinfection by another strain.Sanger sequence raw data also frequently showed overlapping AMDV sequences, most likely resulting either from coinfection or within-host evolution.
Another interesting observation was different virus strains in tissues and blood, but this result is not uncalled for as the same phenomenon has been detected with other viruses like HIV (Haggerty and Stevenson, 1991) and can result from either coinfection or within-host evolution.As the farm has been infected for an exceptionally long time, great variation in virus strains is not a surprise, considering AMDV has been shown to have an exceptionally high substitution rate, most likely due to the intense farming practices (Virtanen et al., 2019).Multiple introductions into the farm are another possibility.
After years of selecting low antibody producers for breeding, farm 1 appeared to have been able to breed an AMDV-tolerant herd and their mink had litter sizes and pelt quality comparable to the average in Finland.Mink often produced low number of antibodies and had lower copies of virus in their tissues as compared to the mink from farm 2. In histopathological evaluation, mink from farm 1 had mild lesions similar to the lesions seen in AMDV infections in kidneys, spleen and liver, indicating that virus may have caused some tissue damage despite the lack of visible symptoms.Also, one of the mink that remained healthy most of the follow-up developed severe AD with typical histopathological lesions by the last sampling despite the fact that the farmers reported rarely having mink with clinical AD.The virus strain of that mink was the same as in some of the asymptomatic mink so the change of virus strain to a more pathogenic one is not the most likely explanation for the sudden onset of symptoms.More likely, this might be connected to the fact that mink were kept alive longer than they normally would have been.
Possibly the tolerant mink were not completely unaffected by the virus but had a very slowly progressing form of the disease and some of the other mink in the follow up may have also developed AD if the follow-up had been continued.Kidneys have a good reserve capacity, and clinical signs of kidney failure may not be detected until kidney function declines to 25% or less (Cianciolo and Mohr, 2015).
With regard to other differences between color types, some of the previous studies have detected more antibodies in Aleutian type mink (Porter et al., 1984).We did not detect higher ELISA absorbances in sapphire mink and, on the contrary, white mink had the highest mean ELISA values in ¾ samplings even though the difference was statistically signi cant only in one sampling.Breeding might play a role in low numbers of antibodies, even in sapphire mink.Also, we did not observe differences in AMDV genome copy numbers in asymptomatic, AMDV-positive mink of different color types.However, our results are in uenced by the fact that we focused on mainly asymptomatic mink (excluding one sick mink in the last sampling) and the results might be different in a naïve mink population that has not been previously exposed to the virus.The only difference detected between color types was that sapphire mink, which are considered more susceptible to the disease, had larger spleens in relation to their body weight.This is logical considering that AMDV infection is known to cause enlarged spleen both in farmed and freeranging mink (Eklund et al., 1968;Zalewski et al., 2021), but a better comparison would require the inclusion of uninfected controls to take natural differences between color types into account.Another logical observation was that spleen size correlated positively with ELISA absorbances.ELISA absorbance also showed positive correlation with AMDV genome copy number in spleen (as copies/ng of DNA) and possible but not statistically signi cant correlation with copy number in kidneys.This is similar to the earlier ndings that genome copy number in blood was greater in the farm with a clinical course of infection as compared to the farm with subclinical infection (Kowalczyk et al., 2018).One mink with the clinical form of AD from farm 1 had clearly higher copy numbers in both tissues as compared to clinically healthy mink.Mean copy number in kidney was greater than mean copy number in spleen in the farm with clinically sick mink, but the small sample size prevents any strong conclusions.
Transcriptome analysis revealed several up-or downregulated genes in infected mink as compared to non-infected mink and symptomatic mink as compared to non-symptomatic mink.These genes were involved not only directly in immune response but also other cellular processes.Many of the genes that were highly upregulated in symptomatic mink as compared to asymptomatic mink were related to innate immunity, which may partly be explained by the fact that mink from farm 1 had been infected for a long time but mink from farm 2 may have had an acute infection.The most signi cantly upregulated gene in mink from farm 2 (compared to farm 1) was ONECUT2 (log 2 FC=7.21), which activates the transcription of several liver genes.Interestingly, Karimi et al. detected several genes involved in liver development to be strongly selected between groups of different disease severity (Karimi et al., 2021).Bloom et al (Bloom et al., 1994), on the other hand, suggested a predominance of Th1 response (macrophage activation) over Th2 response (B cell activation and antibody production) in mink lacking the progressive disease.In addition to some proin ammatory genes, several genes involved in suppression of in ammatory response and activation of Th1 response were upregulated in farm 2. Also, considering the fact that farm 2 reported that the number of symptomatic mink was slowly decreasing, it appears that mink from farm 2 were also starting to tolerate the virus by suppressing in ammation and antibody production.It should be noted that some differences may have been caused by other factors than AMDV, including age, other infections, and possible differences in environmental conditions, but to minimize their effect, mink of same sex and color type were chosen, and the samples were collected at the same time of the year.These results give information about gene-level differences in mink with different AMDV status and help understand the mechanisms behind varying symptoms and immune response.Further studies, e.g., genome-wide association analysis, are needed to better understand the roles of environmental and genetic factors in AD severity.
One limitation to this study is the small sample size and the small number of farms.However, all the mink represented the same gender, excluding one male mink from farm 2. Limited sample size should also be taken into account when interpreting statistical tests.Especially with the non-signi cant correlations, results might have been different with a bigger sample size and non-signi cant differences should also be considered when planning further studies.Causes of death for the mink that died during the follow-up were also not reported, so it is not known if some of them died of AD.However, it is expected that some breeding females are lost during a long follow-up like this one for example due to nursing sickness.Even though decades of breeding and living with the virus is the most likely explanation for the different numbers of symptomatic mink in the farms, the effect of virus strain cannot be excluded either, as the two infected farms had different virus strains.More thorough analysis of virus sequences or complete genome sequencing was not conducted because frequent co-infections in farm 1 would affect the reliability of the sequencing results.However, the very diverse virus strains in farm 1 indicate that the dominance of the host effect on different disease severity in farms.

Conclusions
This study provides a long-term follow-up on AMDV positive mink in farm conditions.It shows that mink can have normal production rates and be clinically healthy despite being infected with different strains of AMDV.After decades of breeding and living with AMDV, mink from farm 1 frequently had low antibody titers, low number of virus in tissues and only mild histopathological changes from AD.A large proportion of ELISA negative but PCR positive mink raises the need to conduct further studies on antibody negative mink from farms with a different disease status to evaluate the reliability of antibody tests in diagnostics and disease control.Further studies are also needed to better understand the host factors behind disease tolerance to help select mink for breeding and avoid causing additional health issues by overbreeding.

Funding statement
This work was supported by the Finnish Fur Breeders' Association, Finnish Veterinary Association, and Finnish Veterinary Foundation funds.

inducing
Th1 response and suppressing Th2 and Th17 responses.Receptors for IL-12 (stimulating NK cells and inducing differentiation of Th1 cells), IL-2 and IL-10 (negative regulation on in ammatory response and positive regulation of T-cell development), and IL-15 (differentiation of natural killer cells) were also upregulated as well as IRF7 (activation of interferons α and β), CXCL8 (a chemoattractant of neutrophils, basophils, and T-cells), CXCR6 (chemotaxis), CCL5 (a chemoattractant of monocytes, T-cells, and eosinophiles), CD8A (T cell activation), and TNFAIP3 (a negative regulator of innate immune response).Genes that were downregulated in farm 2 included IL1RL1 (negative regulation of Th1 response and positive regulation of in ammatory response), TNFRSF13C (positive regulation of B and T cell proliferation), and IL5RA (B cell proliferation).

Figure 1 Comparison
Figure 1

Figure 3 Comparison
Figure 3

Table 1
Percentage of positive samples in PCR and ELISA in farm 1 during follow up.Number of positive individuals/number of tested individuals is reported in parentheses.

Table 2
Summary of histopathological studies of mink from farm 1 a Number of mink with lesions/number of mink with a readable sample b Mean severity of lesions (0=no lesions, 1=mild lesions, 2=moderate lesions, and 3=severe lesion)

Table 3
Average copy numbers per ng of DNA in samples of PCR-positive mink from farms 1 and 2. Samples with copy numbers of 0 have not been included.

Table 4
Spearmans correlation coe cients between ELISA, spleen weight, and quantitation results.P-values are shown in parenthesis and signi cant values are bolded.