**Feline Foamy Virus Infection: Characterization of Experimental Infection and Prevalence of Natural Infection in Domestic Cats with and without Chronic Kidney Disease**


Received: 25 June 2019; Accepted: 13 July 2019; Published: 19 July 2019

**Abstract:** Foamy viruses (FVs) are globally prevalent retroviruses that establish apparently apathogenic lifelong infections. Feline FV (FFV) has been isolated from domestic cats with concurrent diseases, including urinary syndromes. We experimentally infected five cats with FFV to study viral kinetics and tropism, peripheral blood mononuclear cell (PBMC) phenotype, urinary parameters, and histopathology. A persistent infection of primarily lymphoid tropism was detected with no evidence of immunological or hematologic perturbations. One cat with a significant negative correlation between lymphocytes and PBMC proviral load displayed an expanded FFV tissue tropism. Significantly increased blood urea nitrogen and ultrastructural kidney changes were noted in all experimentally infected cats, though chemistry parameters were not outside of normal ranges. Histopathological changes were observed in the brain, large intestine, and other tissues. In order to determine if there is an association of FFV with Chronic Kidney Disease, we additionally screened 125 Australian pet cats with and without CKD for FFV infection and found that FFV is highly prevalent in older cats, particularly in males with CKD, though this difference was not statistically significant compared to controls. Acute FFV infection was clinically silent, and while some measures indicated mild changes, there was no overt association of FFV infection with renal disease.

**Keywords:** foamy virus; spumavirus; retrovirus; viral tropism; infection; kidney; cats; chronic kidney disease; chronic renal disease

#### **1. Introduction**

Feline foamy virus (FFV) is a retrovirus belonging to the ancient *Spumaretrovirinae* subfamily that infects domestic cats (*Felis catus*) and was originally discovered following development of cytopathic effects (CPEs) in feline cell lines [1,2]. Foamy viruses (FVs) cause multiple CPEs in vitro including multinucleation, giant cell formation, and vacuolization, leading to cells looking "foamy" (and where the "*spuma*" originates) [1,3–5]. In naturally-occurring and experimental infections of the domestic cat, however, FFV infection does not cause obvious disease, and has not been definitively associated with pathology despite establishing a persistent, life-long infection with a wide tissue tropism [3,6–11]. It is believed the apathogenicity of FVs in general is due to long periods of co-evolution with their hosts that has led to a disease-free or highly-attenuated infection [2,12,13]. FV transmission is thought to primarily occur via salivary shedding and ongoing contact between animals, though alternate routes such as vertical transmission through lactating dams have been reported [7,14]. In cats, biting and amicable prolonged contact, such as grooming, have been suggested as routes of transmission [7,15,16]. Global FFV prevalence in pet and feral domestic cats can be high and varies from 8 to 80% based on geographic location, population sampled, and assay type [16–26]. FFV prevalence studies of cats in the USA have documented infection rates of 10 to 75%, with age and male sex identified as risk factors in some cohorts [7,16,27].

FVs are generally host-specific with the exception of simian foamy virus (SFV) where non-human primates may transmit virus to related species and zoonotically to humans [28–34]. Zoonotic transmission of FFV to humans has not been detected thus far [12,19,35]. Because of the apparent apathogenicity, wide tissue tropism, and large vector cassette packaging capacity, FVs have been used to develop vaccine and gene therapy vectors in multiple species including cats and non-human primates as a model for human therapies [10,13,36–42]. Many aspects about FV biology, including target cells, latency reservoirs, the specific receptor used for cell entry, viral kinetics over time following infection, and peripheral blood mononuclear cell (PBMC) population changes during infection have been poorly documented [6,12,43,44]. Experimental FFV infection studies in disease-free specific-pathogen-free (SPF) domestic cats with age-matched negative controls using modern and specific assays are rare; the majority have been conducted with domestic cats kept as pets or from shelters [3,6–8,11,38,42,45].

While FFV has been detected in apparently disease-free and healthy animals and has historically been considered apathogenic, the virus has been detected in animals suffering from co-infecting pathogens including feline immunodeficiency virus (FIV) [16,17,46,47], feline leukemia virus (FeLV) [45,48,49], feline coronavirus (FCoV) [24,50], and *Bartonella henselae* [18]. German and others reported histopathological changes in kidney and lung following experimental FFV inoculation [6]. A recent study of zoonotic infection with SFV in African hunters found alterations of urinary parameters including blood urea nitrogen (BUN) and serum creatinine, among other hematological changes [30]. These findings in both cats and humans call into question whether chronic infections with FVs are truly apathogenic.

FFV has also been isolated from cats with renal and other urinary tract disease [4,6,24,51–55], polyarthritis [45,47], neoplasia [1,3,24,56], upper respiratory illness [6,24], and myeloproliferative diseases [7]. Chronic kidney disease (CKD) is one of the most commonly diagnosed renal diseases in cats, with prevalence rates reaching up to 85% in geriatric cats [57–59]. CKD is characterized by functional and structural loss of kidney tissue likely resulting from prolonged or repeated renal insults [60–63]. As renal function declines, urine concentrating ability is lost and glomerular filtration rate falls, which eventually manifests as azotemia characterized by elevated BUN and serum creatinine. While the etiologies of CKD are often unknown, a list of comorbidities have been associated with the development of CKD, including retroviral infections [64–67].

Due to the widespread presence of FFV in domestic cat populations and the knowledge gaps that remain about FFV pathogenicity, especially considering its use in vaccine and gene therapy development, biochemical and histopathologic data from an in vivo FFV experimental infection in healthy SPF domestic cats [10] were further analyzed with emphasis on clinical, immunological, and pathological characteristics and changes during early infection. Due to the renal findings in this study and the detection of FFV in cats suffering from urinary disease, we sought to establish if there is an association between FFV and CKD in cats. We compared FFV prevalence rates in pet cats suffering from CKD in Australia (AU) (as measured by increased blood creatinine) and compared findings to age-matched cats without evidence of CKD. We hypothesized that (1) FFV causes subtle perturbations in immunological and hematological parameters in addition to histopathological changes that could potentially lead to disease in domestic cats, and that (2) FFV is associated with chronic kidney disease in cats. We identified mildly altered hematological and biochemical parameters associated with mild histopathological changes in the lung, brain, and other tissues, and ultrastructural changes in the kidney. We additionally found that while FFV is highly prevalent in AU, there was no direct association between FFV infection and CKD pathology in our sampled population.

#### **2. Materials and Methods**

#### *2.1. Cells and Virus Generation*

Plasmid pCF-7 encoding an FFV genome that is replication-competent in vitro and in vivo was used as virus source [10,38]. Crandell feline kidney (CrFK) cells [21,25,68] were used for transfection, viral propagation, and titer determination as described [10]. A CPE end-point dilution assay was used to determine viral titer (50% tissue culture infectious dose, TCID50/mL) for cat inoculations [10,69]. CPEs consistent with FFV infection used in our assay include cytomegaly, vacuolization, and syncytia formation [1,3,70].

#### *2.2. Animals and Study Design for Experimental FFV Inoculation*

Cats were infected with pCF-7-derived FFV as a control group for a previous study testing an experimental molecularly modified chimeric FFV vector [10]. Domestic cats used in our study were acquired from the Colorado State University (CSU, Fort Collins, CO, USA) specific-pathogen-free (SPF) colony, which is free of FFV, and housed in an animal facility at CSU accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. The trial was approved by the Colorado State University Institutional Animal Care and Use Committee (IACUC) on 5 December 2013 and registered under IACUC protocol #: 13-4104A.

Nine cats (male castrated and intact females, aged 6–8 months) were separated into naïve (N) and FFV groups based on inoculum (Figure 1, modified with permission from [10]): cats N1-4 received FFV-negative CrFK culture media and cats FFV1-4 were inoculated with 10<sup>5</sup> FFV particles (derived from a 2.78 <sup>×</sup> 105 TCID50/mL as described previously [10]) in CrFK culture supernatant. Each cat was inoculated with 2 mL, divided into 1 mL intravenously (iv) through the cephalic vein and 1 mL intramuscularly (im) into hindlimb musculature. A fifth cat was inoculated at the start of the study with 10<sup>5</sup> viral particles (5.56 <sup>×</sup> <sup>10</sup><sup>5</sup> TCID50/mL) of the afore-mentioned chimeric FFV but remained PCR negative. This cat was subsequently inoculated with 1.4 <sup>×</sup> 106 TCID50/mL of the pCF-7-derived FFV on day 53 of the study and subsequently became FFV PCR positive [10]. This animal, referred to as FFV5, was included in our analyses to increase the statistical power of this study. The study timeline and sample collection schedule (blood, saliva, urine, and tissues) are shown in Figure 1.

Cats were monitored daily for any clinical signs of disease, and rectal temperature and body weight were recorded weekly. Peripheral blood was used for flow cytometric PBMC phenotype analysis or processed to collect serum and plasma shortly after collection. Whole blood and sera were submitted to the CSU Veterinary Diagnostic Laboratory (VDL) for complete blood count (CBC) and chemistry analyses (normal blood urea nitrogen concentration: 18–35 mg/dl; normal serum creatinine concentration: 0.8–2.4 mg/dl). Saliva was collected by swabbing oral mucosa with a sterile cotton-tip applicator and freezing at −80 ◦C. Urine was collected through cystocentesis and submitted to the CSU VDL for urinalysis, urine sediment, and urine protein:creatinine (UPC) ratio determination. Urine was considered appropriately concentrated if it had a urine specific gravity (USG) over (>) 1.035. UPC ratio

was considered normal if below (<) 0.2 and borderline proteinuric if between 0.2 and 0.4 [71]. On day 176 post-inoculation (pi), cats were euthanized and necropsied to assess gross pathology and harvest tissues for virus detection, histopathology, and renal-specific assays at the International Veterinary Renal Pathology Service (IVRPS) in The Ohio State University (OSU, Columbus, OH, USA).

**Figure 1.** Experimental timeline of feline foamy virus (FFV) (strain pCF-7) inoculation and sample collection in specific-pathogen-free (SPF) domestic cats. Cats were separated into groups based on inoculum type: negative Crandell feline kidney (CrFK) culture media (naïve control cats N1-4) or FFV in CrFK cell culture supernatant (cats FFV1-5). Blood, saliva, urine, and tissues were collected on the dates specified. Sample collection for cat FFV5 was on a different schedule than the rest of the cohort (bottom timeline, adjusted to match rest of FFV cohort). Samples for baseline data were collected on day -21. On day 176 post-inoculation, cats were euthanized to perform necropsy and tissue collection (black **X**). Figure modified with permission from [10].

#### *2.3. Nested and Real-Time Quantitative PCR for Virus Detection and Quantification*

DNA was purified from whole blood, saliva, and plasma using the DNeasy Blood and Tissue Kit (QIAGEN, Hilden, Germany) and amplified using 0.5 μM forward and reverse primers under conditions described [10,72]. Nested PCR (nPCR) products were electrophoresed in agarose gel and stained to identify the desired PCR product. Real-time quantitative PCR (qPCR) was performed in duplicate to triplicate on purified FFV DNA as described [10,72]. Tissue DNA was purified with the DNeasy Blood and Tissue Kit after homogenizing in Buffer ATL and Proteinase K using the FastPrep®-24 Instrument (MP Biomedicals, Santa Ana, CA, USA). Saliva and plasma RNA was purified using the QIAamp Viral RNA Mini Kit (QIAGEN). RNA was reverse transcribed to cDNA using Superscript II and random hexamer primers (Invitrogen, Carlsbad, CA, USA) and resulting cDNA was used for qPCR as described above.

#### *2.4. Hematological and Flow Cytometric Analyses for PBMC Phenotyping*

Routine hematology and comprehensive PBMC phenotyping were performed at regular intervals (Figure 1). For flow cytometric PBMC phenotype analysis, EDTA-anticoagulated blood was incubated with fluorescent-labelled antibodies (Table S1) diluted at 4 ◦C flow buffer (PBS with 5% bovine fetal serum and 0.1% sodium azide) and processed by the IMMUNOPREP whole blood lysis method on a Q-Prep EPICS Immunology Workstation (Beckman Coulter, Fort Collins, CO, USA) to lyse red blood cells. Samples were analyzed with a Gallios Flow Cytometer (Beckman Coulter). Output data were analyzed using FlowJo software (FlowJo, Ashland, OR, USA). Data from the CBC were used to determine absolute neutrophil, lymphocyte, and monocyte cell numbers by multiplying the number of nucleated cells by percentages of each cell population. Absolute cell counts were then multiplied by percentages of each cell subpopulation obtained through flow cytometry for each cat per timepoint. PMBC phenotype analyses were divided into two panels: Panel A assayed T lymphocyte populations while Panel B determined number of B cells, natural killer (NK) cells, and monocytes. Markers for activation (CD134+, CD125+, MHCII+) and apoptosis (Fas+) were also assayed. In total, 24 PBMC phenotypes were measured for each cat per timepoint (Table 1). General white blood cell (WBC) activation and apoptosis were determined by multiplying WBC counts by MHCII+ and Fas+ percentages.


**Table 1.** White blood cell (WBC) populations assayed for peripheral blood mononuclear cell (PBMC) phenotype analysis.

<sup>1</sup> T helper, <sup>2</sup> T cytotoxic.

#### *2.5. Gross Necropsy and Histologic Characterization of Tissues*

To evaluate pathologic changes associated with FFV infection, necropsy was performed on cats FFV1-5 and control cat N4 on day 176 pi. The following tissues were collected and stored either frozen at −80 ◦C for viral tropism determination (qPCR) or in 10% buffered formalin for histopathological evaluation by light microscopy: lymph nodes (submandibular, mesenteric, pre-scapular, retropharyngeal, and ileocecocolic), thyroid, tongue, tonsil, oral mucosa, salivary glands, thymus, heart, lung, spleen, liver, kidney, ovary, testis, mammary tissue, brain (cerebrum, cerebellum, brainstem), small intestine (jejunum, ileum), colon, bone marrow, and hindlimb skeletal muscle. For histopathological assessment, formalin-fixed tissue samples were embedded into paraffin and 5 μm sections were collected onto charged slides (Superfrost; CSU CDL, Fort Collins, CO, USA). One slide of each tissue was stained with hematoxylin and eosin (HE) for microscopic examination. Tissues were scored using the following scale: 0 = no apparent pathology/change, 1 = minimal change (minimally increased numbers of small lymphocytes, plasma cells, macrophages, and/or mast cells), 2 = mild change (mild inflammation, edema, and/or parafollicular expansion, secondary follicle formation, and presence of tingible body macrophages within lymph nodes), 3 = moderate change (as previously described, but more extensive), 4 = marked changes (as previously described, but with severe inflammation, edema, and/or lymphoid reactivity).

#### *2.6. Renal Tissue Microscopic and Ultrastructural Examination*

Renal tissues collected from cats FFV1-5 and N4 during necropsy were submitted to the IVRPS for comprehensive analysis with light microscopy (LM), transmission electron microscopy (TEM), and immunofluorescence (IF). Samples were submitted in 10% buffered formalin for LM, 3% glutaraldehyde for TEM, and Michel's transport media for IF, and were processed as previously described [73]. Briefly, formalin-fixed paraffin embedded samples were sectioned at 3 μm thickness and stained with HE, Periodic Acid Schiff (PAS), Masson's Trichrome (MT), and Jones Methenamine silver method (JMS). Samples for TEM were processed routinely and examined with a JEOL JEM-1400 TEM microscope (JEOL USA, Inc., Peabody, MA, USA) and representative electron micrographs were taken with an Olympus SIS Veleta 2K camera (Olympus Soft Imaging Solutions GmbH, Münster, Germany). For IF, samples were washed to remove residual plasma constituents, embedded in Optimal Cutting Temperature (OCT, Sakura Finetek USA INC, Torrance, CA, USA), and frozen until sectioning. The OCT blocks were sectioned at 5 μm thickness and direct IF performed with FlTC-labeled goat anti-feline Immunoglobulin (Ig) G, IgM, and IgA antibodies (Bethyl Laboratories, Montgomery, TX, USA) as well as FITC-labeled rabbit anti-human lambda light chain (LLC), kappa light chain (KLC), and C1q antibodies (Dako-Agilent, Santa Clara, CA). Stained sections were examined using an Olympus BX51 epifluorescence microscope and representative images were taken with a Nikon Digital Sight DS-U2 camera (Nikon, Tokyo, Japan). TEM assessment was not performed on control cat N4.

#### *2.7. Samples from Australian Pet Cats with CKD and Non-Azometic, Age-Matched Controls*

To investigate a possible association between FFV infection and naturally-acquired azotemic CKD, biobanked residual samples from pet cats undergoing routine clinical care at the University Veterinary Teaching Hospital Sydney (UVTHS), Sydney School of Veterinary Science, were obtained. Sample collection was approved by the University of Sydney Animal Ethics Committee, approval numbers N00/7-2013/3/6029 (approved 26 August 2013) and 2016/1002 (approved 17 May 2016). Cases were defined as CKD-positive ("CKD") if serum creatinine was elevated above the reference range, USG (measured concurrently) was <1.035, and clinical signs consistent with CKD were present [71]. Age-matched control samples were from non-azotemic clinically healthy cats with USG > 1.050.

A total of 125 samples (53 whole blood and 34 plasma samples in the CKD group and 38 whole blood samples in the control group) were analyzed for FFV infection by either nPCR (whole blood) or specific FFV ELISA against Gag antigen (plasma) as described previously [10]. Detection of FFV antibodies in sera/plasma is more sensitive for detection than FFV PCR [8,10,16,26,72]. Four different laboratories were used for serum creatinine and BUN determination and values were classified as abnormal based upon the references established for each laboratory: BUN: 7.2–10.7 mmol/L (20.16–29.96 mg/dl), 5.7–12.9 mmol/L (15.96–36.12 mg/dl), 5–15 mmol/L (14–42.01 mg/dl), or 3–10 mmol/L (8.4–28 mg/dl) [74]; serum creatinine: 91–180 μmol/L (1.03–2.04 mg/dl), 71–212 μmol/L (0.8–2.4 mg/dl), 40–190 μmol/L (0.45–2.15 mg/dl), or 80–200 μmol/L (0.9–2.26 mg/dl) [75].

#### *2.8. Statistical Analyses*

For the experimentally FFV-inoculated cats, two-tailed Student's t test were performed on hematology, flow cytometry, BUN, serum creatinine, and USG data sets. *P*-values < 0.05 were considered statistically significant. For cat FFV5, the timeline following re-inoculation on day 53 with FFV was adjusted so that day 53 equaled day 0 pi. To have consistent timelines with the main FFV group, timepoints after day 53 for cat FFV5 were grouped with either the equivalent day or the nearest date post-inoculation to the cats in the FFV cohort. A Pearson's correlation coefficient was calculated to determine presence of a correlation and its significance between lymphocyte population numbers and FFV proviral load over time. To assess distributions of viral load to lymphocyte counts, we ran a generalized linear mixed model (GLMM) with the individual cat as a random factor and lymphocyte count as a fixed factor. Data were only run through the GLMM if viral load was at detectable levels and the data fit a negative binomial distribution.

For the CKD analyses, risk ratios and chi-square tests were performed to assess the independence of three pairs of categorical variables: (1) sex and FFV infection, (2) sex and CKD, and (3) FFV infection and CKD. For each pair of variables, cats were stratified by sex (M or F), FFV infection status, and CKD status. Given that both nPCR and ELISA were used to determine FFV infection, CKD analyses were performed on both assays' results to see if conclusions differed. If a chi-square test produced a *P*-value < 0.05, risk ratios and 95% confidence intervals (95% CI) were calculated as an additional post-hoc test. Risk ratios describe the probability of a health outcome occurring in an exposed group to the probability of the event occurring in a comparison, non-exposed group. A Risk ratio >1 suggests an increased risk of that outcome in the exposed group, and a risk ratio <1 suggests a reduced risk in the exposed group. Cats for which sex data were not known (*n* = 5) were omitted in sex-specific data analyses.

Excel (Microsoft Corporation, Redmond, WA, USA) and Prism (GraphPad, La Jolla, CA, USA) were used to conduct the Student's t tests, calculate the lymphocyte and proviral load Pearson's correlation coefficient, and produce graphs. GLMM and CKD analyses were run using the statistical program R version 3.4.2 [76]. The "fitdistrplus" package [77] was used to determine error distributions of the viral load data and the "glmmTMB" package [78] was used to run the GLMM. Chi-square tests and RRs for CKD were calculated using the 'epitools' package [79].

#### **3. Results**

#### *3.1. FFV-Infected Cats Did Not Show Clinical Signs of Infection Despite a Persistent FFV Proviral Load and Specific Humoral Response*

As previously reported, all FFV group cats became PBMC FFV DNA positive (PCR), starting at 21 d pi (Figure 2) [10]. One cat (FFV3) was not PCR positive until day 42 but maintained a much higher proviral load than the rest of the cohort from that point on (Figure 5 in [10]). Cat FFV5 was FFV PCR positive by 10 days pi with FFV pCF-7 (Figure 2). FFV DNA was consistently detected in PBMC once the animals showed positivity [10]. Out of 80 FFV saliva samples tested, FFV RNA was detected only once (cat FFV4 on day 133 pi, Table 2); this sample was FFV DNA negative. All plasma samples tested were negative for FFV RNA and DNA (Table 2). Cats in the naïve group remained negative at all times. Additional proviral kinetics and anti-FFV antibody responses were reported previously [10].

Despite evidence of productive infection and specific immune response [10], none of the cats developed a fever, had changes in body weight, or displayed signs of clinical illness related to infection (such as anorexia or lethargy). CBC and chemistry values did not change significantly from the pre-inoculation time point (day -21) or indicate disease [80].

**Figure 2.** FFV proviral load in PBMC of cats FFV1-5 with summary of significant findings. FFV-infected cats began showing PBMC provirus 21 days pi. Cat FFV5 was re-inoculated and its timeline adjusted to match the rest of the cohort; this cat showed FFV positivity on day 10 post-reinoculation (†) [10]. Blood urea nitrogen (BUN) was significantly increased in infected cats compared to naïve on days 15, 21, and 28. Cat FFV3 had decreased lymphocytes compared to the rest of its cohort, which was negatively correlated to proviral PBMC load (Results Section 3.3). FFV3 also had borderline proteinuria on days 122 and 142 pi. Histopathological changes found after necropsy on day 176 are shown at the right-hand margin. Graph shows mean of FFV group cats' FFV proviral load with bars denoting standard deviation. Numbers in parenthesis indicate number of cats out of the FFV cohort showing findings, with an asterisk (\*) indicating findings also observed in control cat N4 to a lesser severity.

**Table 2.** Summary of findings for diagnostic assays used in the experimental inoculation experiments. Bold font indicates that at least one cat was positive for the measured value, or differences in values between naïve and FFV-infected animals were significant. Cat FFV5 was on a different inoculation and sample collection schedule following re-inoculation on day 53 pi (Figure 1).


**Table 3.** FFV provirus has a primarily lymphoid tissue tropism. Viral load was determined through DNA qPCR and is presented as viral copies per million cells. Cat FFV3 had altered PBMC FFV DNA kinetics and expanded tissue tropism compared to the other FFV cats. Cat FFV5 was on a different inoculation schedule than the rest of the FFV cats (Materials and Methods Section 2.2 and Figure 1). Bold text indicates difference in either proviral load or viral detection compared to other cats in the group.


#### *3.2. FFV Provirus Tissue Tropism Is Primarily Lymphoid*

FFV DNA was detected in the tissues of four out of the five FFV-inoculated cats primarily in lymphoid tissue including lymph nodes (submandibular, retropharyngeal, and prescapular, which drain lymph from head, neck, and forelimbs), tonsil, and spleen (Table 3). Cat FFV3 showed an expanded tissue tropism to central lymphoid tissues (thymus and bone marrow) in addition to non-lymphoid tissue (oral mucosa). The prescapular lymph node was the tissue with highest viral load (cat FFV1). Cat FFV5 showed an FFV tissue tropism similar to the rest of the FFV group, with the submandibular lymph node having the highest viral load (Table 3). Control cat N4 and cat FFV4 did not have detectable provirus in any of the tissues tested.

#### *3.3. Significant PBMC Phenotypic Changes Were Rare though a Negative Correlation Was Found between Lymphocytes and FFV Proviral Load in Cat FFV3*

Out of the 24 cell populations and activation or apoptosis markers assayed for each cat per timepoint (Table 1), there were only 9 instances where significant differences (*P* < 0.05) were found between infected and control animals (Table 2). Significantly increased populations were found between FFV (1–5) and N (1–4) groups in the following instances: (1) absolute monocyte numbers on days 15 (*P* = 0.036) and 42 (*P* = 0.025) pi and (2) CD21+MHCII+ cells on d 86 pi (*P* = 0.0076). FFV-group cats had decreased populations compared to controls in the following instances: (1) CD8+CD25+ cells on d 112 pi (*P* = 0.044), (2) CD8+CD134+ cells on d 10 pi (*P* = 0.031), (3) CD8+FAS+ on d 10 pi (*P* = 0.015), (4) CD56+ cells on d 112 pi (*P* = 0.00038), and (5) CD56+MHCII+ cells on days 15 (*P* = 0.049) and 112 pi (*P* = 0.00070).

We further evaluated WBC populations in cat FFV3 due to the altered PBMC FFV provirus pattern observed [10]. This cat appeared to have lower lymphocytes and a trend for decreasing lymphocyte count over time compared to the rest of the infected and naïve cats (Figure 3A, blue line) as PBMC proviral load increased over time (Figure 3B, black line). A Pearson's correlation coefficient test for this cat showed a significant negative correlation between lymphocyte cell number and viral load over time (*r* = −0.653, *P* = 0.006). There was no correlation found in the rest of the infected cats (data not shown) and there was no significant relationship between viral load and lymphocyte count when we analyzed all cats as a group (GLMM Estimate −0.530, *P* = 0.596).

**Figure 3.** High viral load correlates with decline in circulating lymphocytes in cat FFV3. (**A**) Absolute lymphocyte population numbers determined through complete blood count for cat FFV3 (blue line) appeared to decrease over time compared to all other cats in the study. Naïve cats are grouped on the black line and the rest of the FFV-group cats are displayed on the red line; (**B**) A significant negative correlation (*r* = −0.653, *P* = 0.006) was found between lymphocytes and FFV proviral load [10] in cat FFV3 as lymphocyte population numbers (blue line) decreased and proviral load (determined by qPCR, black line) increased over time. Bars denote standard deviation.

#### *3.4. Subtle Di*ff*erences in Urine and Hematological Parameters Were Detected between Experimentally Infected and Control Cats*

UPC ratios were 0.1 (normal) for all cats throughout the study, with the exception of two timepoints in cat FFV3 (d 122 and 142 pi) where its UPC ratio increased from 0.1 to 0.2 (borderline proteinuric), before decreasing back to normal (0.1) on day 176 pi (Figure 2) [81]. This mild transient increase in UPC coincided with the timepoint when this cat's PBMC FFV proviral load was highest (d 142 pi) (Figure 3B) [10]. BUN concentration remained within normal ranges (18–35 mg/dl) for all cats throughout the study, however values were higher in infected cats compared to naïve controls on three consecutive timepoints: days 15 (*P* = 0.012), 21 (*P* = 0.039), and 28 (*P* = 0.025) (Figures 3 and 4). All cats had USG > 1.035 and urinalyses and urine sediment were unremarkable throughout the study. Serum creatinine concentrations were at or below 1.8 mg/dl at all timepoints, and there were no significant differences in serum creatinine between infected and control cats [82].

**Figure 4.** Blood urea nitrogen (BUN) levels are within normal range, but higher in FFV-infected versus naïve control cats. While BUN, one of the biomarkers used to assess renal health, remained within normal range (18–35 mg/dl) for all cats, concentrations tended to be higher in infected cats (red line) compared to naïve cats (black line) on days 15, 21, and 28 pi (red asterisks, *P* < 0.05). Lines represent mean of BUN measurements for the cats in each group. Vertical lines denote the standard deviation for each grouped measurement.

#### *3.5. Mild Lymphoplasmacytic Infiltrates and Lymphoid Hyperplasia of Multiple Tissues Were Associated with FFV Exposure*

No significant clinical or pathological findings were grossly observed in control (N4) or FFV-infected cats (FFV1–5) during necropsy. Microscopic evaluation of tissues from FFV-infected cats revealed mild (*n* = 3) to moderate (*n* = 2) lymphoid hyperplasia in retropharyngeal, submandibular, mesenteric, and prescapular lymph nodes, characterized by numerous secondary follicles that contain abundant tingible-body macrophages. The tonsils of infected animals exhibited minimal (*n* = 1), mild (*n* = 3), and moderate (*n* = 1) lymphoid hyperplasia with multifocal infiltration of small numbers of lymphocytes beyond the capsule in one mildly affected animal. Two infected cats exhibited mild thyroiditis characterized by multifocal infiltrates of small lymphocytes, plasma cells, and macrophages within the interstitium and surrounding colloid-filled follicles of varying size. Within the ileum, Peyer's patches were minimally (*n* = 1) to mildly (*n* = 4) hyperplastic, and one animal exhibited multifocal lymphoplasmacytic infiltrates extending deep into the submucosa. Additionally, minimal (*n* = 3) to mild (*n* = 1) lymphoplasmacytic colitis was observed in FFV-infected cats, with lymphoplasmacytic infiltration into the submucosa that caused disruption of the submucosal architecture (*n* = 3), as well as small numbers of degenerate neutrophils scattered within the submucosa (*n* = 1). In the cerebrum of FFV-infected cats, there were minimally (*n* = 2) to mildly (*n* = 3) increased numbers of glial cells (gliosis) and paired astrocytes (astrocytosis) surrounding scattered neurons within the gray matter (satellitosis), a feature that was most prominently noted within the frontal lobe and thalamus (Figure 5B). Cat FFV5 also had small numbers of small lymphocytes within the meninges (lymphocytic meningitis) (Figure 5C). Scattered neurons within these regions were multifocally swollen, rounded, and demonstrated mild central dispersion of Nissl substance (chromatolysis), as well as rare, scattered neurons that exhibited hypereosinophilic and/or fragmented cytoplasm (potentially indicative of neuronal necrosis) (*n* = 2) (Figure 5D). One cat had mild multifocal lymphohistiocytic mastitis. Histologic changes in cat FFV5 were slightly more pronounced when compared to the other infected animals and included moderate lymphoid hyperplasia in the tonsil with moderate numbers of lymphocytes and macrophages within the tonsil medullary sinus, and enlarged germinal centers in the Peyer's patches. Within the lung of this cat, the parabronchial interstitium and alveolar septa were multifocally expanded by small numbers of small lymphocytes, intact neutrophils, and macrophages (interpreted as mild interstitial pneumonia). Alveoli were occasionally filled with small numbers of alveolar macrophages and frequently lined by plump, cuboidal epithelial cells, indicating type 2 pneumocyte hyperplasia, with occasional clubbing of alveolar walls due to mild smooth muscle hypertrophy.

**Figure 5.** FFV-infected cats exhibit early neurodegenerative changes in the central nervous system. (**A**) Neurons in the CNS of control cat N4 contain uniform, round nuclei, abundant basophilic Nissl substance, and are flanked by few glial cells (black arrow). Frontal lobe, Hematoxylin-eosin (HE) 400×. Scale bar = 100 μm. (**B**) Neurons in the CNS of FFV-infected cat FFV3 exhibit moderate satellitosis, characterized by increased numbers of glial cells (black arrows). Thalamus, HE 400×. Scale bar = 100 μm. (**C**) The meninges of FFV-infected cat FFV5 are expanded by minimal numbers of mature small lymphocytes (red arrows) and plasma cells (red arrowheads). Cerebellum, HE 400×. Scale bar = 100 μm. Neurons in the frontal lobe of this animal (inset) are shrunken, with hypereosinophilic cytoplasm, and exhibit moderate satellitosis (black arrows). Frontal lobe, HE 400×. (**D**) Neurons in the CNS of an FFV-infected cat are swollen and rounded, with an indistinct nucleus and a dispersed Nissl substance (chromatolysis). Thalamus, HE 400×. Scale bar = 100 μm.

Non-specific histologic findings in control cat N4 included mild lymphoid hyperplasia in the mesenteric lymph node, tonsil, and thymus, minimal to mild inflammatory infiltrate in the tongue, and mammary tissue, and minimal chromatolysis in the cerebrum. Findings in this control cat ranged from very subtle to mild, and were less severe than in infected animals.

#### *3.6. Ultrastructural Changes Were Noted in the Kidneys of FFV-Infected Cats*

Histopathology of the kidneys from cats FFV1-5 (Table 4) demonstrated a few small foci of tubular degeneration encompassing fewer than 15 tubular cross-sections per focus in cats FFV1 and FFV2; cat FFV1 also had associated atrophy of the tubules. Glomeruli from the remaining cats in the FFV cohort and control cat N4 were within normal limits.

Ultra-structural TEM evaluation of glomeruli from cats FFV1-5 demonstrated minimal to moderate segmental effacement of podocyte foot processes in all infected cats (Figure 6A and Table 4). There were a few small segments of wrinkled glomerular capillary walls in cat FFV5 (Table 4). Electron-dense whorls resembling myelin figures appeared free in the cytoplasm or within cytoplasmic vacuoles in tubular epithelial cells of three of the infected cats (Table 4). Cytoplasmic vacuolization of parietal or tubular epithelial cells was present in two cats (Table 4).



+ = minimal, +/+ = minimal to mild, ++ = mild, +++ = moderate, NT = not tested.

**Figure 6.** Transmission electron microscopy (TEM) documents podocyte foot process effacement (**A**). Examples of organized linear structures in tubular epithelial cell cytoplasm are depicted in panels **B**–**D**. These structures ranged from polygonal (**B** and **D**) to ovoid (**C**). Some structures were composed of a single electron dense line (**B**), whereas others were composed of numerous parallel electron dense lines (**C**,**D**) separated by regularly spaced electron lucent lines (**C**,**D**).

Within the cytoplasm of the proximal tubular epithelial cells of four of these cats, there were small electron-dense spirals and linear structures of 10–15 nm in length arranged in pairs, stacks, polygonal shapes, or spirals, and of variable length (Figure 6B–D). Sometimes the linearly shaped ones had a beaded appearance or formed structures resembling a zipper. Mitochondria occasionally wrapped around the structures. In cat FFV5, the structures were similar to the ones found in the other FFV cats but appeared significantly more organized. TEM evaluation was not conducted on cat N4.

Immunofluorescence did not demonstrate definitively positive (granular) labeling for any of the antibodies (IgG, IgM, IgA, LLC, KLC, and C1q). Cat FFV3 had weak blush to linear staining with IgM of the glomerular mesangium and some capillary walls but based on the pattern of staining, it was considered non-specific. Immunofluorescence was negative in tissues from control cat N4.

#### *3.7. FFV Is Highly Prevalent in Australian Domestic Pet Cats*

We determined FFV prevalence by either nPCR or ELISA in groups of cats with and without CKD to evaluate associations between FFV infection and the incidence of CKD. Overall FFV prevalence was 67% (Table S2) with no significant difference between CKD and control groups (Figure 7). Overall FFV prevalence was similar between males (68%) and females (69%). Males had a higher FFV prevalence in CKD (73%) versus males without CKD (58%), although this difference was not significant (Figure 7).

**Figure 7.** FFV prevalence is high in Australian domestic pet cats. Males with CKD had higher FFV prevalence than male control cats, although not significantly. M = male; F = female.

Chi-square test statistics and P-values are reported in Table S2. Our results suggest there is no significant association between CKD and FFV infection. Chi-square and relative risk ratio results did not differ between nPCR and ELISA indicating that neither assay detected an association between CKD and FFV infection in all groups evaluated.

#### **4. Discussion**

This study further characterized early FFV infection and host immune response in healthy SPF domestic cats following experimental FFV inoculation using a well-characterized replicative FFV molecular clone that achieves comparable viral titers, similar gene expression and comparable humoral immune response to wild-type virus in experimentally inoculated cats [42]. Though it is difficult to exactly recapitulate natural exposure route and dose, the FFV experimental exposure protocol used in this study is comparable to inoculation doses others have used in experimental FFV inoculation studies, and resulted in viral loads and antibody responses comparable to those recorded in naturally infected animals [6,8,11,42]. In addition to clinical monitoring, assessing viral kinetics and tropism, determining specific antibody response, and a histopathological assessment of different tissues, we conducted assays to expand infection characterization. This included flow cytometric assessment of specific white blood cell subsets suggested to be involved in FFV infection and renal-specific assays to determine the extent of FFV involvement in renal health or disease. Samples analyzed were obtained from control experimental cats reported in a previous study in which an FFV-based vaccine candidate was tested [10]. Based on microscopic findings from the FFV-infected cohort showing evidence of injury in renal tissues during acute FFV infection (Figure 6A and Table 4), we investigated the potential association of FFV with CKD in Australian client-owned cats following natural FFV infection.

FFV established a detectable, persistent, and clinically apathogenic infection during the relatively acute 6-month time period of our study [6,8,10,38]. Provirus was primarily isolated from lymphoid tissues including PBMC, the retropharyngeal lymph node, and spleen as previously reported (Table 3) [6,11,27]. The expanded tissue tropism found in cat FFV3, with 10-fold higher PBMC proviral load than the other cats, also included the oral mucosa, suggesting that animals with higher viral loads could have dissemination to non-lymphoid sites. We were unable to detect virus in the tissues of one cat (FFV4) which also tended to have lower PBMC proviral load [10]. Viral RNA was detected once in the saliva of cat FFV4 (Table 2), indicating that limited amounts of virus was being shed, and that salivary excretion may not correlate with widespread tissue distribution. A recently published report by Cavalcante and others on FFV-infected cats in Brazil found that pet and feral cats were more likely to be FFV positive in PBMC than in buccal swabs, as tested by PCR [83]. This has also been reported in primates due to a delay in salivary excretion of virus compared to blood [84]. A wide variability of FV positivity has been shown in different nonhuman primate species and even within the same species [85–87]. It is possible that due to the acute time period in our study, viral titer had not yet reached high enough levels in the oral cavity to be detected in the saliva.

A significant negative correlation between lymphocyte numbers and proviral load was found in cat FFV3 (Figure 3B), despite the fact that PBMC phenotype analysis did not indicate increased cell death or lymphocyte population contraction in FFV-infected cats (Table 2). Thus, PBMC phenotyping appears not to be a useful indicator of FFV infection. These findings, however, indicate that a subset of FFV-infected cats may experience higher viral loads with the potential for concurrent lymphocyte decline. Further work analyzing correlates of FFV-infected cats with hematologic indices is warranted to investigate this limited observation.

Necropsy and histologic analysis of experimentally infected cats yielded minimal to moderate changes in the lymphoid compartment, CNS, large intestine, lung, and thyroid. FFV-associated lung lesions have previously been noted in another experimental FFV infection study which reported mixed cellular infiltrates and eosinophilic fluid within alveolar walls [6], similar to findings reported here. Alterations in CNS histopathology of FFV-infected cats suggest viral replication or associated inflammation as is seen in other retroviral infections [88–90]. FFV has previously been isolated from the CNS of cats, and may therefore indicate that FFV is capable of productive CNS infection and subtle neurologic alterations [91]. These findings suggest a potential role for FFV in the development of mild acute inflammation in a variety of parenchymal organs and brain. FFV proviral load did not appear to correlate to the presence or severity of histopathology and histopathological changes were observed in tissues that were not FFV PCR positive and vice-versa. The changes observed were mild and consistent with inflammation due to nonspecific immune stimulation, and cannot be unequivocally associated with FFV infection, which would be ideally confirmed by viral propagation from isolated tissues or immunohistochemistry [11]. Nevertheless, the findings of consistent mild inflammation in multiple tissues of SPF cats exposed to FFV suggest that microscopic alterations occur during acute FFV infection.

FFV has previously been associated with urinary syndrome pathology [4,6,24,51–53]. BUN concentrations remained normal during FFV-infection, but were statistically elevated in FFV-infected compared to naïve groups on days 15, 21, and 28 pi, which coincided with the days when animals were first FFV PCR and ELISA positive [10]. These findings are intriguing given recent reports of chronic zoonotically SFV-infected people with increased BUN compared to un-infected controls [30]. Borderline proteinuria (increased UPC ratio to 0.2 [71]) was also recorded in cat FFV3 on days 122 and 142 pi, which coincided with the timing of the highest viral load measured in the study. While recorded changes in BUN and UPC ratio are interesting, these findings may not specifically indicate renal proteinuria and could be considered a transient systemic response to infection.

Ultrastructural kidney changes (glomerular podocyte foot process effacement, myelin figures, vacuolization, and wrinkled glomerular capillary walls, Figure 6A and Table 4) are non-specific and reversible changes. If enough podocytes are irreversibly injured, then the patient can develop segmental to global glomerulosclerosis, a disease process in humans and small animals that can cause proteinuria, usually with UPC >2. Although a cut-off for the number of irreversibly injured podocytes has not been established in cats, a model of glomerulosclerosis in rats estimated that >40% of the podocytes have to die and detach in order for glomerulosclerosis to develop [92]. Notably, glomerulosclerosis was not identified in any of the cats in the present study. Tubular atrophy noted on histopathology is an irreversible lesion and often seen in cats with clinical evidence of CKD [62,63]. Overall, our findings indicate subtle reversible renal alterations, indicating that FFV does not induce dramatic changes to renal function within 6 months of infection, however it is unknown how the lesions could have progressed in a chronic infection. We did not detect correlations between proviral PBMC load and presence or severity of TEM changes.

Electron-dense structures identified in proximal tubular epithelial cells in the kidney (Figure 6B–D) could represent viral structures at immature stages of assembly before forming the spherical shapes of FFV virions reported in the literature [1,93,94]. Similar structures have been found in the central nervous system in both cats and humans. Cook and others described similar tubular structures as "paramyxovirus nucleocapsid-like" in the cytoplasm of oligodendrocytes taken from demyelinating lesions in the optic nerve of three clinically healthy adult cats [95]. These inclusion-body like structures were of 16–17 nm in diameter and fused into penta- to septa-laminar shapes and 900 nm in length. The authors suggested a possible viral etiology. Wilcox and others (1984) reported similar structures in optic nerves and brains of 24 clinically healthy cats from which FFV was isolated [91]. While also finding 10–18 nm wide and 500 nm long structures, they also reported structures in much smaller shapes, appearing as "short, disorganized fragments," located next to where budding virions were observed, in addition to intranuclearly. These structures were, however, found in the cytoplasm of cells that did not display CPEs and thus these lesions were not attributed to FFV but perhaps a morbillivirus [91].

FFV was highly prevalent in Australian domestic cats as reported previously [16,19], and could be due to the lifestyle of Australian cats which are commonly allowed outside [96,97], allowing more opportunities to interact with other potentially infected cats. The samples analyzed also consisted of sera/plasma assayed by ELISA and whole blood analyzed by PCR. PCR is not as sensitive as ELISA for detection and can yield false negatives [8,10,16,26,72], therefore it is possible that FFV prevalence is even higher in Australia. Similar prevalence rates in males (68%) and females (69%) indicates that there are no sex-related differences in prevalence rates in Australian pet cats [16]. A recent study of feral cats in the USA found an association between FFV infection and male sex [27]. However, our epidemiological results were obtained from client-owned desexed animals. Another epidemiological study of FFV in Australia did not find an association between sex and FFV infection in desexed domestic cats, but did see higher incidences of FFV in female feral cats [16]. Our findings suggest that while overall there appears to be no association between FFV and CKD, CKD in male cats may be associated with FFV infection, though any effect is mild, and evaluation of a much larger cohort is required to assess actual associations. Male sex has not been found to be an overall risk factor for CKD, however males can be overrepresented in certain age groups [98].

#### **5. Conclusions**

Collectively, our findings reinforce and expand on the current established notion that FFV is widely prevalent and apathogenic over an acute time period, supporting decades long assumptions that FFV is well adapted to the domestic cat host. Evaluation of a multitude of hematologic and immunological parameters did not reveal significant host responses to infection. However, our detailed analysis of serum chemical and histopathological changes indicates sub-clinical alterations that could contribute to metabolic or degenerative diseases over time, supporting our hypothesis and work conducted by earlier researchers, and recent reports in humans [1,3,4,6,7,24,30,45,47,51–56]. The negative correlation between lymphocyte count and viral load in one cat with higher viral load suggests that a differential susceptibility and potential pathogenicity may exist in some individuals. Studies with larger cohorts of animals with well-characterized disease phenotypes may reveal subtle relationships between FFV and feline health, since the virus is clearly widely distributed in free-ranging cat populations.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/11/7/662/s1, Table S1: Antibody marker combinations used for PBMC phenotype analysis through flow cytometry. Table S2: FFV prevalence and chi-square analysis data for CKD studies.

**Author Contributions:** Conceptualization, C.L.-F., R.M.T. and S.V.; Data curation, C.L.-F.; Formal analysis, C.L.-F., N.D., R.G. and S.V.; Funding acquisition, C.L.-F., M.L. and S.V.; Investigation, C.L.-F., R.M.T., X.Z., C.M., R.C. and M.B.; Methodology, C.L.-F., R.M.T., M.L. and S.V.; Project administration, C.L.-F. and S.V.; Resources, C.L.-F., M.B., J.B., M.L. and S.V.; Supervision, S.V.; Validation, C.L.-F. and X.Z.; Visualization, C.L.-F., C.M., R.C. and J.Q.; Writing—original draft, C.L.-F., C.M., R.C., N.D., R.G. and S.V.; Writing—review & editing, C.L.-F., R.M.T., C.M., R.C., N.D., R.G., J.B., J.Q., M.L. and S.V.

**Funding:** C.L.F. was funded by Morris Animal Foundation (grant number: D16FE-402) and The National Institutes of Health (NIH T32 grant number: 2T32OD010437-16). Funding bodies played no role in the design of the study, collection, analysis, and interpretation of data, or in writing the manuscript.

**Acknowledgments:** We would like to acknowledge Martha McMillan and Esther Musselman for in vitro FFV training, animal care, and help with regulatory matters (IACUC).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Insights into Innate Sensing of Prototype Foamy Viruses in Myeloid Cells**

**Maïwenn Bergez 1,**†**, Jakob Weber 2,3,**†**, Maximilian Riess 1, Alexander Erdbeer 2,3, Janna Seifried 1, Nicole Stanke 2,3, Clara Munz 2,3, Veit Hornung 4, Renate König 1,5,6,\* and Dirk Lindemann 2,3,\***


Received: 30 September 2019; Accepted: 22 November 2019; Published: 26 November 2019

**Abstract:** Foamy viruses (FVs) belong to the *Spumaretrovirinae* subfamily of retroviruses and are characterized by unique features in their replication strategy. This includes a reverse transcription (RTr) step of the packaged RNA genome late in replication, resulting in the release of particles with a fraction of them already containing an infectious viral DNA (vDNA) genome. Little is known about the immune responses against FVs in their hosts, which control infection and may be responsible for their apparent apathogenic nature. We studied the interaction of FVs with the innate immune system in myeloid cells, and characterized the viral pathogen-associated molecular patterns (PAMPs) and the cellular pattern recognition receptors and sensing pathways involved. Upon cytoplasmic access, full-length but not minimal vector genome containing FVs with active reverse transcriptase, induced an efficient innate immune response in various myeloid cells. It was dependent on cellular cGAS and STING and largely unaffected by RTr inhibition during viral entry. This suggests that RTr products, which are generated during FV morphogenesis in infected cells, and are therefore already present in FV particles taken up by immune cells, are the main PAMPs of FVs with full-length genomes sensed in a cGAS and STING-dependent manner by the innate immune system in host cells of the myeloid lineage.

**Keywords:** retrovirus; foamy virus; spumavirus; innate sensing; cGAS; STING

#### **1. Introduction**

Spuma or foamy viruses (FVs), which constitute several genera in the retrovirus subfamily *Spumaretrovirinae* [1], display a replication strategy with features common to both other retroviruses (*Orthoretrovirinae*) and hepadnaviruses (reviewed in [2,3]). FVs are unique amongst retroviruses, as the initiation of reverse transcription (RTr) of the packaged viral genomic RNA (vgRNA) occurs in a significant fraction of virions (5–10%) during viral assembly [4–7]. Thereby, unlike to orthoretroviruses, both vgRNA and/or viral genomic DNA (vgDNA) containing virions are found in the supernatant

of FV infected cells. It is generally accepted that the vgDNA containing virions contribute to the majority of new productive infection events during spreading of FVs in cultures, at least in vitro [5,6]. However, a low level of reverse transcription, probably derived from vgRNA containing virions, has been observed during uptake of FVs at very low multiplicities of infection (MOI) [4,7].

FVs are naturally endemic to most non-human primates (NHPs), including New and Old World monkeys and apes, cats, cows, horses, tree shrews, sea lions, and bats (reviewed in [2,8,9]). In addition, endogenized copies of FV genomes were identified in sloths, the aye-aye, the Cap golden mole [10,11], cod [12], platyfish [12], zebra fish, and the coelacanth. Nowadays, humans are not considered as a natural host, but frequent zoonotic transmissions of NHP simian FVs (SFVs), but not feline FV (FFV) or bovine FV (BFV), have been observed in workers occupationally exposed to NHPs—bush meat hunters in central Africa, and in various contexts of human–NHP interspecies contact in South and Southeast Asia. Cases of spread from human to human have not been reported. The best-studied and characterized isolate to date is the so-called prototype FV (PFV; formerly known as human FV, HFV), which was originally isolated from an African patient who presumably was infected zoonotically by a chimpanzee FV [13–15].

Another characteristic of FVs is their extremely broad tropism. In vitro, only very few species and cell types are known to be non-permissive to FVs or FV Env-mediated entry [16]. FV infection in vitro is highly cytopathic to most cell types, except cell lines or primary cells of myeloid or lymphoid origin, which can become chronically infected [17,18]. The cellular targets of FVs in vivo remain poorly characterized. In infected monkeys, the viral genome is detectable in many tissues but appears to be largely in a latent state, as viral replication is reported to be mainly restricted to the superficial epithelial layer of the oral mucosa [19]. This explains the major transmission mode between monkeys and zoonosis to humans through bites. In the blood, proviral DNA is detected in CD8+- and CD4<sup>+</sup> T-cells (memory and naïve) as well as B-cells, and to a lower extent also in monocytes and NK-cells [20–22]. Other types of immune cells that encounter FVs during the course of an infection in vivo have not been characterized.

FVs have co-evolved with their hosts for at least 60 million years and are considered to be non-pathogenic in natural hosts and zoonotically infected humans. The immune system seems to control FV infection very efficiently in natural hosts and zoonotically infected humans, as replication and viral load of FV stays low [20–22]. However, the mechanisms involved are poorly understood (reviewed in [8]). In particular, the interaction of FV with the innate immune system remains to be fully clarified. FVs are known to respond in vitro in a cell type-dependent manner towards IFNs [23]. Furthermore, treatment of cells of different origin with type-I or II IFNs, impairs spreading of FVs or inhibits early steps in FV replication [24–28]. In line with this observation, restriction of FVs by several IFN-induced cellular proteins has been reported, although their relevance for FV replication in vivo has not been tested so far [29–32]. Although FV replication in vitro is impaired by addition of exogenous IFNs, infection of a variety of tissues does not seem to mount an innate immune response [28]. Only recently, stimulation of IFN secretion by human hematopoietic cells, in particular, plasmacytoid dendritic cells (pDCs), upon incubation with SFV virions or SFV infected cells, was demonstrated [23]. pDCs, as the main producer of type-I IFN, detect FV RNA by the TLR7-mediated pathway. However, the contribution of myeloid cells, besides pDCs, to innate sensing and IFN induction has as of yet not been investigated. Above all, conventional dendritic cells (DCs) play a critical role in detecting retroviruses, as shown for the lentivirus HIV-1 [33], and promote, thereby, the activation of adaptive immunity [34]. For HIV-1 and other lentiviruses, it has been demonstrated that the reverse transcribed DNA products generated upon viral entry in DCs mediate the activation of transcription factors, such as IRF3, resulting in ISG expression and IFN synthesis [33,35]. Thereby, the viral reverse transcribed components are sensed by the DNA sensor cyclic GAMP synthase (cGAS) [36] together with polyglutamine binding protein 1 (PQBP1) [37]. Interestingly, and in contrast to lentiviruses, FV infected cells release both vgRNA and vgDNA containing virions, and harbor RTr products late during the assembly steps. Therefore, exposure of DCs and other myeloid cells to

FVs may result in interactions with the innate immune system different to those of other retroviruses. For instance, they could include pathways encountered by hepatitis B virus (HBV), as HBVs, like FVs, reverse transcribe their genome during particle morphogenesis (reviewed in [38]).

The aim of this study was to investigate whether the innate immune system of cells of the human myeloid lineage is capable of sensing FVs. If so, the viral pathogen-associated molecular patterns (PAMPs), the pattern recognition receptors (PRRs), and sensing pathways involved were to be characterized. We found that the innate immune system responds to sensing of RTr products of full-length PFV genomes but not of minimal vector genomes in the cytoplasm of human myeloid cells with an efficient interferon-stimulated gene (ISG) induction within hours of virus exposure. Sensing of PFV RTr products was dependent on cellular cGAS and STING expression and largely insensitive to reverse transcriptase inhibition during viral entry, suggesting that the already viral DNA (vDNA) containing PFV particles are the main stimulator.

#### **2. Materials and Methods**

#### *2.1. Cells and Culture Conditions*

The human embryonic kidney cell line 293T (ATCC CRL-1573) [39] and a proteoglycan-deficient variant 293T-25A (described elsewhere), the human fibrosarcoma cell line HT1080 (ATCC CCL-121) [40], and the clonal variant HT1080 PLNE containing a PFV LTR driven *EGFP* reporter gene expression cassette [41], were cultivated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (*v*/*v*) heat-inactivated fetal calf serum and antibiotics. The human monocyte cell line THP-1 (ATCC TIB-202) and knockout (KO) variants ΔIFNAR1, ΔSAMHD1 [42], ΔcGAS [43], ΔMAVS [43], ΔMyD88, and ΔSTING [43] were cultivated in Roswell Park Memorial Institute 1640 Medium (RPMI 1640) supplemented with 10% (*v*/*v*) heat-inactivated fetal calf serum, antibiotics, 2.5 g/L glucose, 10 mM Hepes, and 10 mM sodium pyruvate. THP-1 cells were differentiated into macrophage-like cells by the addition of phorbol-12-myristyl-13-acetate (PMA) at 30 to 50 ng/mL final concentrations for 48 h prior to exposure of viral supernatants.

Human buffy coats, of anonymous blood donors, were obtained from German Red Cross Blood Donor Service Baden–Württemberg Hessen. Primary human monocytes were isolated from peripheral blood mononuclear cells (PBMCs) using Ficoll density gradient and subsequent isolation of CD14<sup>+</sup> monocytes [44]. Briefly, PBMCs were separated by density gradient centrifugation (30 min, 980× *g*, RT), using Ficoll (Histopaque 1077 Sigma-Aldrich Biochemie GmbH, Hamburg, Germany). The mononuclear layer of the interphase was recovered and washed twice with PBS. PBMCs were subsequently incubated with 0.86% (*w*/*v*) ammonium chloride for erythrocyte lysis (37 ◦C, 10 min). PBMCs were washed twice and filtered (0.7 μm). CD14<sup>+</sup> monocytes were positively selected using CD14-MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's protocol, and separated from unlabeled cells using an AutoMACS device (Miltenyi Biotec). Subsequently, cells were differentiated into monocyte derived dendritic cells (MDDC) or monocyte derived macrophages (MDMs) by cultivating for 5 days in RPMI-1640 medium supplemented with 2 mM l-Glutamine; 10% (*v*/*v*) FCS; 1% (*v*/*v*) HEPES, 1 mM sodium pyruvate; 280 U/mL granulocyte-macrophage colony-stimulating factor (GM-CSF) (Leukine® Sargramostim, Genzyme, Boston, MA, USA); 800 U/mL IL-4 (PeproTech GmbH, Hamburg, Germany) for MDDCs and 560 IU/mL GM-CSF only for type 1 proinflammatory MDMs. The same amounts of medium and cytokines were added on day 3. On day 5 of differentiation, MDMs were detached by a short incubation with PBS-EDTA and MDDCs by gentle resuspension in PBS; they were counted and plated for cell stimulation experiments and simultaneously checked for MDDC and MDM surface markers and their differentiation status.

#### *2.2. Recombinant Plasmid DNAs*

A four-component PFV vector system, consisting of the expression-optimized packaging constructs pcoPG4 (PFV Gag), pcoPE (PFV Env), and pcoPP (Pol), and an enhanced green fluorescent protein (EGFP)-expressing PFV transfer vector pMD9 or puc2MD9, has been described previously [16,45,46].

The CMV-driven proviral expression vector pczHSRV2 (wt) and its variants pczHSRV2 M69 (iRT), expressing a Pol protein with enzymatically inactive RT domain (YVDD312–315GAAA mutation), pczHSRV2 M73 (iIN), with enzymatically inactive IN domain (D936A mutation), and pczHSRV2 EM271 (ΔEnv), with inactivated Env translation start (M1T, ATG to ACG, M5T, ATG to ACG; M16T, ATG to ACG mutation) were described previously [5,47,48]. For this study the variants pczHSRV2 EM284 (ΔGag), with inactivated Gag translation start (M1L, ATG to CTG; S3Stop, TCA to TAA mutation); pczHSRV2 EM270 (ΔPol), with inactivated Pol translation start (M1L, ATG to CTG mutation); pczHSRV2 EM273 (ΔGPE), with simultaneously inactivated Gag, Pol, and Env translation starts; pczHSRV2 EM020 (iFuse), with inactivated Env SU/TM furin cleavage site (R571T mutation); and pczHSRV2 EM010, with inactivated Tas translation start (M1L, ATG to TTG mutation), were generated. All constructs were verified by sequencing analysis. Primer sequences and additional details are available upon request.

For VLP-Vpx production the lentiviral vector pSIV3+, derived from SIVmac251 was used, as previously described [49]. For single-round HIV-1 reporter virus production the plasmid pBR-NL43-Env−-IRES-eGFP-nef<sup>+</sup> [50] was used. In both cases the envelope vector pCMV-VSVg was used for pseudotyping.

#### *2.3. Transfection, Virus Production, and Titration*

Cell culture supernatants containing recombinant PFV particles and respective mock controls were generated by transfection of the corresponding virus encoding or mock control plasmids (mock A: pUC19; mock B: pczHSRV2 EM273 (ΔGPE)) into 293T cells using polyethyleneimine (PEI) as described previously [41,46], or 293T-25A cells using calcium phosphate (described elsewhere). Cell-free supernatants were generated by passing through a 0.45 μm filter and were either stored in aliquots at −80 ◦C when used for stimulation experiments or processed further for additional analysis. Viral titers of recombinant, EGFP-expressing PFV vector particles (PFV-SRVs) by fluorescence marker-gene transfer assay on HT1080 cells were determined as described previously [51]. Virus particles generated by use of proviral expression plasmids (PFV-RCPs) were titrated on HT1080 PLNE cells harboring a Tas-inducible nuclear *EGFP* ORF in their genome as described previously [41].

VLP-Vpx and HIV-1 GFP reporter viruses were produced as previously described [52]. Briefly, 2 <sup>×</sup> 10<sup>7</sup> HEK293T/17 cells per T175 flask were seeded. The next day, 15.2 <sup>μ</sup>g pSIV3+ and 2.3 <sup>μ</sup>g pCMV-VSVg for VLP-Vpx production and 11.6 μg of pBR-NL43-Env−-IRES-eGFP-nef<sup>+</sup> and 5.9 μg pCMV-VSVg, for HIV-1 reporter virus production, per flask, were transfected using 18 mM PEI (Sigma-Aldrich). Medium was changed approximately 16 h later and viral supernatants were harvested 48 and 72 h post-transfection. Supernatants were centrifuged (10 min at 4 ◦C; 1500 rpm), filtered (0.45 μm), and DNaseI digested (1 U/mL) for one hour. Viral supernatants were purified by ultracentrifugation through 20% (*w*/*v*) sucrose (2 h, at 4 ◦C; 25,000 rpm); virus pellets from day one and two were resuspended in PBS, pooled, aliquoted and stored at −80 ◦C. HIV-1 reporter viruses were titrated via serial dilutions on TZM-bl reporter cells using the beta-galactosidase colorimetric assay. VLP-Vpx were titrated according to their ability to target the restriction factor SAMHD1 for degradation, using Western blot.

#### *2.4. Myeloid Cell Stimulation and qPCR Analysis of ISG Induction*

For stimulation experiments, THP-1 cells were plated at 2 <sup>×</sup> 106 cells/well (3 <sup>×</sup> 104 cells/well) in a total volume of 2 mL (100 μL) in 6-well (96-well) plates and PMA was added to 30 ng/mL (50 ng/mL) final concentration. Forty-eight hours later the medium was replaced by the respective virus supernatant, mock control supernatants (mock A: pUC19, mock B: PFV-RCP ΔGPE mutant) or medium

(medium) as indicated. At different time points post virus exposure, as indicated, viral supernatants were aspirated and cells were snap frozen at −80 ◦C and stored until subsequent nucleic acid extraction. MDDCs and MDMs were plated at 3 <sup>×</sup> 104 cells/well in the differentiation medium without cytokines, 24 h prior to infection. Virus supernatants, either PFV or a VSV-G pseudotyped full-length HIV-1 GFP reporter virus, with or without the addition of VLP-Vpx were added. AZT (Sigma) was added during infection in the indicated experiment at a final concentration of 100 μM. MDDC and MDM infections were conducted by spinoculation at 1200 rpm, 32 ◦C for 1.5 h. Viral supernatants were replaced by fresh medium and cells were cultivated at 37 ◦C, 5% CO2. At different time points post infection, cells were either lysed for RNA extraction and subsequent RT-qPCR or stained for FACS analysis.

#### *2.5. Quantitative PCR Analysis*

Cellular nucleic acids from cultures in 6-well plates were extracted using the RNeasy Mini kit (QIAGEN, Hilden, Germany) according to the manufacturer's protocol. qPCR analysis of cellular mRNA expression using Maxima Probe qPCR Master Mix including ROX dye (ThermoFisher Scientific, Dreieich, Germany), a StepOnePlus (Applied Biosystems, Foster City, CA, USA) quantitative PCR machine, and plasmid standard curves was performed as previously described [48]. Primers, Taqman probes, and cycling conditions are summarized in Table A1. Cellular nucleic acids from cultures in 96-well plates were extracted using NucleoSpin® RNA Plus Kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's protocol. Relative expression levels of *ISG54* and *RPL13A* were determined using QuantiTect SYBR Green RT-qPCR Kit (QIAGEN) with the respective specific primers on a LightCycler® 480 Instrument (Roche, Basel, Switzerland). Relative mRNA expression levels were normalized to the housekeeping gene *RPL13A* and analyzed using the 2ˆ(−ΔΔCT) method, finally depicted as fold inductions over mock A, mock B, or medium, as indicated. Primers and cycling conditions are summarized in Table A2. Primer efficiencies have been tested before in 10-fold serial dilutions and were calculated to have >90% efficiency.

#### *2.6. Flow Cytometry Analysis*

Purity of MDMs and MDDCs was assessed via flow cytometry analysis. Triple stainings, of 1 <sup>×</sup> 10<sup>5</sup> cells with CD14-Pacific blue (BioLegend, San Diego, CA, USA), CD163-PE (BD), CD206-APC (BD) and CD1a-PE (BioLegend), and CD11c-Vio Blue (BioLegend) and CD16-APC (BioLegend) were performed with the matching IgG controls, listed in Table A3. In order to determine CD86 activation, marker expression upon infection with different PFV mutants, 24 h post infection, 6 <sup>×</sup> 104 cells were stained with CD86-PE (Biolegend) or the corresponding isotype control. Briefly, after 5 days of differentiation, MDMs were detached by a short incubation with PBS-EDTA, MDDCs by gentle resuspension in PBS. Cells were washed twice with FACS staining buffer (PBS containing 10% (*v*/*v*) FCS), FC-Block (1:10, BD) was added to prevent unspecific binding via Fc-receptors (10 min, RT), cells were resuspended in the specific antibody dilutions or corresponding isotype-controls, and they were stained for 20 min on ice. Subsequently, cells were washed twice with FACS staining buffer, fixed with ice-cold 2% (*w*/*v*) Paraformaldehyde, washed twice, and analyzed via flow cytometry using MACSQuant Analyzer 10 (Miltenyi) and FCS Express software (De Novo Software, Glendale, CA, USA). For MDM and MDDC purity, percentages of positive cells of the specific markers are depicted in Figure S1. For CD86 expression of infected MDDCs, mean fluorescent intensities (MFIs) normalized to the IgG controls of each condition are depicted as relative MFIs normalized to the wt treatment. Infection levels by HIV-1 GFP with and without AZT were assessed by determining the percentage of GFP-positive cells (Figure S2). The cut-off was set to 0.1% with the IgG controls or the non-infected controls.

#### *2.7. Analysis of PFV Particle Protein and Nucleic Acid Composition*

PFV particles were concentrated from cell-free supernatants by centrifugation at 4 ◦C and 25,000 rpm for 3 h in a SW32Ti rotor (Beckman Coulter GmbH, Krefeld, Germany) through a 20% (*w*/*v*) sucrose cushion. The particulate material was resuspended in phosphate-buffered saline (PBS) and used immediately for further analysis or stored at −80 ◦C. Western blot analysis was performed after protein sample buffer addition as described before using PFV Gag and PFV Env leader peptide (LP) specific antisera [48]. DNase digestion of viral particles, extraction of particle-associated nucleic acids, and analysis of nucleic acid composition by qPCR was done using the primer–probe sets listed in Table A1, as described before [46,48,53].

#### *2.8. Analysis of Cellular Protein Expression*

For immunoblot analysis of IRF3 phosphorylation and SAMHD1 degradation, 5 <sup>×</sup> <sup>10</sup><sup>5</sup> MDDCs/<sup>12</sup> wells were seeded and exposed to either PFV-RCP or VLP-Vpx. Six or twenty-four hours post-exposure, cells were harvested by resuspension in ice cold PBS, centrifuged (300× *g*, 6 min, 4 ◦C), and lysed in 25 μL RIPA-lysis buffer (100 mM NaCl; 10 mM EDTA (pH 7.5), 20 mM Tris (pH 7.5); 1% (*v*/*v*) Triton X-100; 1% (*w*/*v*) sodium deoxycholate) containing protease and phosphatase inhibitor cocktails (Complete Protease Inhibitor Cocktail; PhosSTOP Phosphatase Inhibitor Cocktail, Roche) for 45 min on ice. Lysates were centrifuged (17,000× *g* for 15 min at 4 ◦C), and protein concentration was determined based on the Bradford assay using the Bio-Rad Protein Assay Dye Reagent Concentrate. Samples containing 20 μg protein were prepared with NuPAGE LDS sample buffer (4×) and NuPAGE Sample Reducing Agent (10×), to a final 1× concentration and denatured at 70 ◦C for 10 min. Proteins were separated on precasted NuPAGE™ 4–12% Bis-Tris gradient gels (Invitrogen). The gel was run in 1× MOPS buffer (1 M MOPS,1MTris, 69.3 mM SDS, 20.5 mM EDTA Titriplex II) supplemented with 200 μL NuPage Antioxidant 10× (inner chamber) at 200 V for 1 h 10 min. Proteins were transferred to a Hybond P 0.45 PVDF membrane (GE Healthcare, Chicago, IL, USA) using the XCell IITM blotting system with 1× NuPAGE transfer buffer (Invitrogen) at 35 V for 1 h 40 min. Membranes were blocked in 5% (*w*/*v*) BSA (Carl Roth) in 0.01% (*v*/*v*) Tris-buffered saline with Tween 20 (TBST) for 2 h at 4 ◦C with subsequent incubation in primary antibody dilutions at 4 ◦C overnight. Horseradish peroxidase (HRP)-linked goat anti-rabbit or horse anti-mouse IgG (heavy and light chain) secondary antibodies (Cell signaling, Danvers, MA, USA) were applied for 2 h at 4 ◦C. For detection Pierce® ECL Western Blotting Substrate (ThermoFisher Scientific) or ECL Prime (GE Healthcare) were used and the emitted chemiluminescence was detected at different exposure times on autoradiography films (Amersham Hyperfilm ECL, GE Healthcare). The following primary antibodies were used and applied at 4 ◦C, overnight: Anti-Phospho-IRF-3 (Ser396) (Cell Signaling, number 4947); Anti-IRF3 (Epitomics, number 2241-1); Anti-GAPDH (Cell Signaling, number 2118); and Anti-SAMHD1 (Proteintech; number 12586-1-AP). In order to remove phospho-IRF3 antibody, probed membrane was incubated in stripping buffer (2% (*w*/*v*) SDS, 62.5 mM Tris-HCl (pH 6.8), 100 mM β-mercaptoethanol), rotating for 45 min at 65 ◦C.

#### *2.9. Statistics*

All the statistical analyses were performed using GraphPad Prism 8. The numbers of experimental replicates and information on the statistical methods used for determination of two-tailed *p*-values are described in the individual figure legends. Symbols represent: \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001; ns: not significant (*p* ≥ 0.05).

#### **3. Results**

#### *3.1. ISG Induction in Myeloid Cells upon Exposure to Replication-Competent PFV*

PMA-differentiated THP-1 monocytic cells represent an in vitro model system recapitulating the functional properties of macrophages and dendritic cells exposed to retroviruses [54]. In order to analyze whether FVs are sensed by cells of the myeloid lineage, replication-competent PFV supernatants derived from full-length, wild type proviral expression constructs (PFV-RCP) (Figure A1) were first used to infect PMA-differentiated THP-1 cells (Figure 1a,b). Relative transcription levels of *ISG54* or *ISG56* were determined as readouts for an IRF-3 dependent stimulation, since the selected ISGs are directly downstream transcriptional targets of IRF3 [55,56].

**Figure 1.** PFV-mediated ISG induction in myeloid cells. (**a**,**b**) Kinetics of *ISG56*/*ISG54* induction in PMA-differentiated THP-1 wild type cells incubated with different amounts of wild type PFV-RCP (a, MOI 0.2) as well as pUC19 (mock A) mock supernatants. ISG mRNA levels normalized for *RPL13A* mRNA levels were determined by qPCR at the indicated time points post exposure. Means ± SDs of *ISG56* (*n* = 4; a) or *ISG54* (*n* = 4; b) induction relative to mock A treatment are shown. (**c**,**d**) Primary human MDDC (**c**) or MDM (**d**) were incubated with wild type PFV-RCP (PFV; MOI 0.25), ΔGPE (mock B) mock supernatant, or medium (medium) for 6, 12, or 24 h, as indicated. Means ± SEMs, plus individual data points, of *ISG54* (*n* = 5–8) induction normalized to *RPL13A* relative to medium treatment are shown. Mixed-effects analysis with Holm–Sidak's multiple-comparisons test was used to assess significance. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001; ns: not significant (*p* ≥ 0.05).

Exposure of PMA-differentiated THP-1 cells to PFV-RCPs led to a strong IRF3-dependent ISG induction (Figure 1a,b). Furthermore, a significant, dose-dependent ISG induction was detectable, which peaked at 8 to 12 h and declined slowly thereafter (Figure 1a,b). To corroborate this finding, primary human MDMs and MDDCs were analyzed. As it cannot be ruled out that DCs and macrophages may possess slightly different sets of proteins aiding to sense DNA, we exposed both cell types to wild type PFV-RCP (PFV), mock (mock B) supernatants obtained after 293T transfection of a proviral expression construct with inactivated viral structural protein expression (ΔGPE), or medium (medium). Interestingly, in both MDDCs and MDMs we detected a robust and high ISG induction (Figure 1c,d) after PFV-RCP but not mock supernatant exposure, suggesting that replication-competent PFV derived from full-length proviral expression constructs are efficiently sensed by the innate immune system in different myeloid cell types. In line with this, a strong phosphorylation of IRF3 was detectable in PFV-RCP treated but not in medium treated MDDCs at 6 h post exposure (Figure S3).

#### *3.2. PFV is Sensed by the Cellular cGAS-STING Pathway*

To identify the particular innate pathways that are triggered by PFV, we exposed a panel of THP-1 KO cell lines deficient in key molecules of various sensing pathways to wild type PFV-RCP for 8 or

24 h, respectively. Interestingly, PMA-THP-1 cells deficient in cGAS or STING expression, which are key molecules of the DNA-sensing pathway, failed to mount any measurable ISG-response upon PFV exposure (Figure 2). In contrast, cells deficient in MAVS, a key node of the RIG-I/MAVS RNA-sensing pathway, or MyD88 that lies downstream of the endosomal TLR7/9 pathway showed a reduced but clearly detectable ISG induction. These results suggest that FVs are mainly sensed by the cytosolic DNA-sensing pathway in myeloid cells. Furthermore, IFNAR1 deficient PMA-THP-1 cells displayed a similar ISG response as wild type cells indicating that events downstream of IFN production do not influence sensing of FV. Intriguingly, KO of SAMHD1, previously demonstrated to strongly enhance HIV-1 sensing in various myeloid cell types [33,36,37], did not further stimulate the ISG response upon PFV exposure. On the contrary, PFV-mediated ISG induction in SAMHD1 KO cells was moderately reduced, at levels comparable to those observed for MAVS and MyD88 KO cells.

**Figure 2.** cGAS and STING-mediated sensing of PFV-associated reverse transcription products. PMA-differentiated THP-1 wild type cells and KO variants with deficiencies in components of various innate sensing pathways, as indicated, were incubated with wild type PFV-RCP (MOI 0.2) or pUC19 (mock A) mock supernatants. ISG mRNA levels normalized for *RPL13A* mRNA levels were determined by qPCR at the indicated time points post exposure. Means ± SEM, plus individual data points, of *ISG56* (*n* = 3) induction normalized to *RPL13A* relative to mock A treatment are shown. Two-way ANOVA with Tukey's multiple-comparisons test was used to assess significance. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001; ns: not significant (*p* ≥ 0.05).

Thus, innate sensing of PFV in host cells of the myeloid lineage occurs mainly in a cGAS and STING dependent manner.

#### *3.3. Innate Sensing of PFV Requires Cytoplasmic Access and Enzymatically Active Reverse Transcriptase*

Since we determined the cytoplasmic DNA-sensing pathway as the main innate pathway, we aimed at identifying the PAMPs responsible of PFV sensing and at determining the cellular sublocation. For this purpose, myeloid target cells were exposed to viral supernatants derived from various mutant proviral expression constructs, which varied in their nucleic acid compositions and had different blocks in early steps of viral replication (Figure A1, Figure 3a). A significant ISG induction was detectable in PMA-THP-1 and MDDCs not only by wild type PFV-RCP, but also by variants that either failed to express the PFV transactivator Tas and accessory protein Bet (ΔTas-Bet) after proviral integration due to inactivation of the translation start site of the *tas* ORF, or encoded an enzymatically inactive

integrase (iIN) and were, therefore, largely integration-deficient (Figure 3b,c). This indicates that FV sensing does not require proviral integration or de novo viral transcription capacity.

**Figure 3.** Differential ISG induction profiles of PFV mutants (described in detail in Figure A1 and Material and Methods) varying in their protein and nucleic acid composition. (**a**) Particle-associated nucleic acids extracted from viral particles pelleted by ultracentrifugation of virus supernatants used in (**b**–**d**) were analyzed by qPCR to quantify the particle-associated viral (vDNA: viral DNA; vRNA: viral RNA) and cellular (vGAPDH mRNA) nucleic acid composition. Mean copy numbers ± SDs per mL supernatant determined from duplicates are shown. (**b**,**c**) ISG induction profile of PMA-differentiated THP-1 wild type cells (**b**) or MDDCs (**c**) incubated with identical amounts of wild type PFV-RCP (MOI 0.1 THP-1; MOI 0.25 MDDCs) supernatants, variants thereof, and pUC19 (mock A) and ΔGPE (mock B) mock supernatants, or medium, as indicated, for 8 h and 24 h. Means ± SEMs, plus individual data points, of *ISG56* (*n* = 6) or *ISG54* (*n* = 5–7) mRNA induction normalized to *RPL13A* mRNA relative to mock A or medium treatment are shown. One-way ANOVA with Tukey's multiple-comparisons test (**b**) or mixed-effects analysis with Tukey's multiple-comparisons test (**c**) was used to assess significance. (**d**) CD86 cell surface expression profile of MDDCs 24 h post exposure to wild type PFV-RCP (MOI 0.25) supernatants and variants thereof as indicated. Means ± SEMs (*n* = 5), plus individual data points, relative to medium treatment are shown. Mixed-effects analysis with Tukey's multiple-comparisons test was used to assess significance. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001; ns: not significant (*p* ≥ 0.05).

In contrast, ISG induction was strongly reduced or undetectable in cells exposed to supernatants of PFV-RCPs either encoding an enzymatically inactive reverse transcriptase (iRT) or a fusion-deficient envelope glycoprotein (iFuse); and a control supernatant derived from a proviral expression construct with simultaneous inactivation of *gag*, *pol*, and *env* ORF translation start sites (ΔGPE, mock B), which failed to assemble virions (Figure 3b,c). Strikingly, for all individual virus supernatants a perfect correlation between their potentials to induce ISG expression and the upregulation of CD86 cell surface expression in MDDCs was observed (Figure 3d).

Taken together, these results point to the dependence on cytoplasmic access and enzymatically active reverse transcriptase for PFV-mediated ISG induction and activation of MDDCs.

#### *3.4. PFV ISG Induction Does Not Require Reverse Transcription upon Target Cell Entry and Is Not Suppressed by SAMHD1*

The above analysis revealed that PFV-RCP mediated induction requires an enzymatically active RT. FVs RTr has been reported to occur both during virus's assembly and upon target cell uptake and entry [4–6]. We therefore aimed to determine whether RTr products already present in PFV-RCP particles or those newly generated during target cell entry, as reported for other retroviruses like HIV-1 [36,37], are the main ISG inductors in myeloid cells. For that purpose, MDDCs were exposed to wild type PFV-RCP and HIV-1 GFP reporter viruses in the presence or absence of AZT, preventing de novo RTr upon target cell entry (Figure S2). Quantification of ISG induction at 24 h p.i. revealed a strong, 5- to 10-fold reduction in *ISG54* induction for HIV-1 GFP exposed samples by AZT treatment (Figure 4a,b). In contrast, AZT treatment diminished the ISG induction potential of PFV-RCP only marginally, a maximum of 2-fold (Figure 4a,b).

Furthermore, we examined whether SIV-Vpx-mediated degradation of endogenous SAMHD1 influences ISG induction mediated by PFV RTr products in MDDCs. MDDCs were exposed to PFV-RCP or HIV-1 GFP supernatants in the presence or absence of SIV-VLPs containing Vpx (Figure S4). Analysis of *ISG54* induction at 6 and 24 h p.i. confirmed the previously reported enhancement of HIV-1 sensing, up to 10-fold at 24 h p.i. (Figure 4c). In contrast, *ISG54* induction levels mediated by PFV-RCPs in MDDCs were not significantly altered by simultaneous Vpx-mediated SAMHD1 inactivation. This is in line with the *ISG56* induction capacity of PFV-RCP in THP-1 SAMHD1 KO cells shown before (Figure 2).

Thus, vDNA or RTr products generated from the encapsidated vRNA genome during virus morphogenesis are the major PFV PAMPs, which are sufficient for efficient sensing of PFV in myeloid cells. Furthermore, unlike HIV, PFV sensing in myeloid cells cannot be enhanced by SIV Vpx pretreatment, arguing against a role of endogenous SAMHD1 in the regulation of PFV sensing.

**Figure 4.** Influence of RTr inhibition and VLP-Vpx treatment on PFV-mediated ISG induction during target-cell entry. (**a**,**b**) MDDCs were incubated with wild type PFV-RCP (MOI 0.25), HIV-1 GFP (MOI 2), and VLP-Vpx or ΔGPE (mock B) mock supernatants in the absence or presence of AZT (100 μM) as indicated. *ISG54* mRNA levels normalized for *RPL13A* mRNA levels were determined by qPCR at 24 h post exposure. (**a**) Mean values ± SEMs, plus individual data points of *ISG54* (*n* = 3–4) induction relative to medium incubated samples are shown. (**b**) Mean values ± SEMs, plus individual data points of *ISG54* (*n* = 4) induction relative to the respective sample incubated with the same virus type without AZT addition are shown. One-way ANOVA with Sidak's multiple-comparisons test was used to assess significance. (**c**,**d**) MDDCs were incubated with wild type PFV-RCP (MOI 0.25), HIV-1 GFP (MOI 2) supernatants, or ΔGPE (mock B) mock supernatants in the absence or presence of VLP-Vpx as indicated. *ISG54* mRNA levels normalized for *RPL13A* mRNA levels were determined by qPCR at 6 and 24 h post exposure. (**c**) Mean values ± SEMs, plus individual data points, of *ISG54* (*n* = 7) induction relative to medium incubated samples are shown. (**d**) Mean values ± SEMs, plus individual data points, of *ISG54* (*n* = 7) induction relative to the respective sample incubated with the same virus type without VLP-Vpx addition are shown. One-way ANOVA with Sidak's multiple-comparisons test was used to assess significance. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001; ns: not significant (*p* ≥ 0.05).

#### *3.5. PFV ISG Induction Requires Reverse Transcription of Full-Length Viral Genomes*

Reports on level of innate sensing of HIV-1 in various target tissues appear to be strongly influenced by the specific type of HIV-1 viruses (full-length genome replication-competent versus single-round versus minimal vector genome) [57,58]. The characterization of the ISG induction profile of the various PFV-RCP mutants described above and the wild type like ISG profile of IFNAR1 THP-1 KO cells suggested that viral spreading in the culture is not required. This and the time course analysis of ISG induction suggest that the ISG response is induced shortly after cytoplasmic entry. Therefore, we also examined whether single-round PFV vector particles containing minimal viral genomic sequences (PFV-SRV) (Figure A1) induce an ISG response in myeloid cells (Figure 5). Surprisingly, whereas exposure of PMA-THP-1 cells to wild type PFV-RCPs lead to a strong *ISG56* induction, *ISG56* induction levels in cells to PFV-SRVs were strongly reduced (Figure 5a).

**Figure 5.** Differential ISG induction profile of single round PFV vector particles harboring full-length or minimal viral genomes. (**a**) PMA-differentiated THP-1 wild type cells were incubated with the indicated relative amounts of wild type PFV-RCP (MOI 0.3) supernatants and variants thereof, or the different PFV-SRV supernatants with variable Pol content (μg Pol packaging plasmid used for supernatant production is indicated; MOI 3 at 2.5 μg Pol) or supernatant from 293T cells transfected with pUC19 (mock A). *ISG56* mRNA levels normalized for *RPL13A* mRNA levels were determined by qPCR at 8 h post exposure. Mean values ± SEMs, plus individual data points of *ISG56* (*n* = 1–3) induction relative to mock A treatment are shown. (**b**,**c**) PFV supernatants characteristics. (**b**) Particle protein composition. Western blot analysis of protein composition of viral particles pelleted by ultracentrifugation of virus supernatants used in panel A employing PFV Gag (α-PFV Gag) and PFV Env LP (α-PFV Env LP) specific polyclonal antisera. The identity of individual protein bands is indicated to the left, the molecular weight to the right. (**c**) Particle nucleic acid composition. Particle-associated nucleic acids extracted from viral particles pelleted by ultracentrifugation of virus supernatants used in panel A were analyzed by qPCR to quantify the particle-associated viral (vDNA: viral DNA; vRNA: viral RNA) and cellular (vGAPDH mRNA) nucleic acid composition. Mean copy numbers ± SDs per mL supernatant determined from duplicates are shown. Viral titers were determined for PFV-SRV supernatants by *EGFP* reporter gene transfer assay on HT1080 cells and for PFV-RCP supernatants by Tas-dependent *EGFP* reporter gene induction assays on HT1080 PLNE cells. Mean titers ± SDs per mL supernatant determined from technical duplicates are shown.

In addition to their replication capacities there are two major differences between PFV-SRV and PFV-RCP. First, PFV-SRVs, unlike PFV-RCP, package Pol also in a vRNA genome independent manner, which leads to higher particle associated RT levels [41]. Second, the packaged and reverse transcribed minimal vRNA genome of PFV-SRVs lacks large genomic regions and does not encode any viral proteins [45].

To examine whether any of these differences are responsible for the differential ISG induction profile, different PFV-SRV and PFV-RCP supernatants were produced. For PFV-SRV, various supernatants were produced, keeping the amounts of vector genome Gag and Env expression vectors constant but using different amounts of Pol packaging plasmid. This resulted in virus supernatants containing similar physical amounts of PFV particles (Figure 5b), which contained similar amounts of viral and cellular RNA, but differed in their vDNA content (Figure 5c) and specific viral infectivity over a 50 to 100-fold range (Figure 5c). When PMA-THP-1 cells were exposed to identical amounts of these different PFV-SRVs, only a very weak ISG response was detectable, which was not influenced by the vDNA content and was at least 10-fold lower than that of PFV-RCP controls (Figure 5a).

Next, PFV-RCPs variants were used in which the translation of individual (ΔGag, ΔPol, ΔEnv) or all (ΔGPE) structural or enzymatic viral ORFs were abolished by point mutagenesis to determine whether the presence of functional ORFs for PFV structural genes in the viral genome is required for efficient innate sensing (Figure A1). Infectious virus supernatants were generated by complementing the individual defective structural functions by using the respective packaging constructs (+G, +P, +E, +G/P/E) also employed for production of PFV-SRVs. RCP virus supernatants contained similar physical amounts of virus particles with similar amounts of vRNA as PFV-SRVs (Figure 5b,c). Mutant PFV-RCP particles contained 2- to 5-fold lower amounts of vDNA and 2- to 8-fold higher amounts of cellular RNA compared to wild type PFV-RCP. Notably, the vDNA of mutant PFV-RCP was similar to that of PFV-SRVs generated with the highest amounts of PFV Pol packaging plasmid (Figure 5c). Strikingly, all mutant PFV-RCP supernatants, including the ΔGPE +G/P/E mutant, did induce an ISG response at wild type level (Figure 5a). When PMA-THP-1 cells were exposed to reducing amounts of PFV-RCP wt and ΔGPE +G/P/E, a dose-dependent decline in the ISG response was observed, with its level correlating well with the amount of vDNA present in the respective supernatants (Figure 5c).

Taken together, these results suggest that the type of PFV RNA genome encapsidated and reverse transcribed is crucial for innate sensing of PFV rather than the amounts of particle-associated Pol and vDNA.

#### **4. Discussion**

FV infections of natural hosts and zoonotic transmission to humans appear to be efficiently controlled by the immune system. We have only limited knowledge on the immunological mechanisms involved in this process.

Here, we examined the interaction of PFV with the innate immune system in immune cells of the myeloid lineage. Using an in vitro model cell line, THP-1, and primary human MDDC and MDM cultures, we observed an efficient stimulation of the innate immune system, determined as an IRF3-dependent stimulation of ISG expression and IRF3 phosphorylation, by replication-competent PFV generated from proviral expression constructs (PFV-RCP).

Furthermore, PFV innate sensing was neither dependent on viral transactivator Tas-mediated de novo viral transcription nor on viral integrase enzymatic activity. This indicates that productive infections, late steps of the viral replication cycle and virus spreading are not a prerequisite for PFV innate sensing in myeloid cells.

By use of various PFV-RCP mutants we demonstrated that PFV sensing occurs predominantly in the cytoplasm of myeloid cells as fusion-defective PFV-RCPs failed to stimulate ISG expression.

In contrast, efficient ISG induction required an enzymatically active reverse-transcriptase, indicating that vDNA or RTr products generated during reverse transcription are the major PFV PAMPs sensed by the innate immune system. Since FV RTr is observed both late in the replication

cycle after capsid assembly and early during host cell entry [4–6], we investigated whether vDNA and RTr products already being present in PFV particles or those newly generated during uptake are the major PAMPs. Inhibition of RTr during entry by RT inhibitor led only to a minor reduction in PFV-mediated ISG induction, whereas that of HIV-1 was strongly reduced. This indicates that the vDNA and RTr products already present in PFV particles are sufficient for efficient induction of an innate immune response. However, since we are unable to prevent RTr during PFV assembly and at the same time allow subsequent RTr to take place during virus entry, as AZT incorporation leads to dead-end products, we cannot formally exclude that the latter may contribute to a certain extent to the innate immune response.

In line with the ISG induction potentials of the various PFV-RCP mutants, we observed that inactivation of essential key molecules of cellular DNA-sensing pathways, cGAS and STING, in THP-1 cells, abolished PFV-RCP-mediated innate immune stimulation. In accordance with vDNA and RTr products already present in PFV particles before host cell entry, representing the main PFV PAMPs, inactivation of cellular SAMHD1, either by gene KO in THP-1 cells or VLP-Vpx co-delivery in MDDCs, had only minor negative effects on PFV-RCP mediated ISG induction. This is also in agreement with previous reports of SAMHD1, unlike for lentiviruses, not being a restriction factor for PFV [59].

pDCs were shown by Rua and colleagues to mount an innate immune response as a consequence of TLR7-mediated sensing of PFV RNA [23]. Our results suggest that in myeloid cells, the contribution of vRNA sensing to PFV-mediated innate immune stimulation appears to be negligible. This is underlined by clearly detectable, although slightly reduced ISG induction in THP-1 KO cells having key molecules of cellular RNA-sensing pathway, MAVS or MyD88, inactivated. Furthermore, PFV-mediated ISG induction was almost completely abolished when myeloid cells were incubated with PFV-RCP with enzymatically inactive RT, which did not contain vDNA but harbored similar levels of vRNA as wild type virus.

The most striking finding of this study was the requirement of vDNA and/or RTr products to be derived from full-length vRNA genomes (PFV-RCP) instead of RNA genomes with minimal cis-acting viral sequences of PFV single-round vectors (PFV-SRV) for efficient ISG induction in myeloid cells. Interestingly, in contrast to the minimal genome of single-round vectors (PFV-SRV wt), which had a strongly impaired ISG induction capacity, the RTr of single-round vectors encapsidating a full-length genome containing point mutations (PFV-RCP ΔGPE + G/P/E) led to an ISG induction profile similar to replication-competent, wild type PFV particles (PFV-RCP wt). Our results obtained with various kinds of PFV-SRVs and PFV-RCP mutants can rule out differences in the replication capacity, the encoding of structural proteins, and the particle-associated RT levels or vDNA copy numbers as causative for this difference in innate stimulatory capacity. This underscores that most likely the encapsidated and reverse transcribed full-length genome represents the immunostimulatory component. Currently we can envision several potential underlying mechanisms.

A very attractive but perhaps also the most unlikely explanation might be the presence of immunostimulatory determinants in full-length PFV genomes that are absent in minimal PFV vector genomes. A specific **v**iral **s**timulatory **s**equence **e**lement (vSSE), and/or secondary structure thereof, present only in RTr products of full-length PFV genomes, could potentially be sensed.

Alternatively, the size difference between full-length PFV proviral (11,024 nt) and minimal SRV genomes (4348 nt), and not, or not only, a specific vSSE absent in the latter, may be responsible for or contribute to their differential ISG induction potential. This would fit to reports of cGAS activation based on DNA length and based on long DNA pre-structured by host proteins to strongly stimulate DNA sensing [60–62]. HIV-1 minus strand strong stop DNA ((-)sssDNA) was reported to contain short stem-loop structures with flanking unpaired guanosines highly stimulatory for cGAS activity. Notably, full-length PFV-RCP and minimal PFV-SRV genomes are identical up to the translation start of the *gag* ORF, thereby leading to identical PFV (-)sssDNA's (Figure A1). Therefore, even if PFV (-)sssDNA harbors cGAS stimulatory structures analogous to HIV-1 (-)sssDNA, it cannot be the cause for the

differential ISG stimulatory capacity of RTr products of full-length PFV-RCP compared to minimal PFV-SRV genomes.

Finally, we cannot rule out that currently unknown differences in stability, structural integrity, or uncoating of the infecting viral cores containing full-length wild type or point mutation-containing genomes in comparison to minimal vector genomes are causing the difference in the innate response.

Further studies, including a detailed bioinformatic analysis of secondary structure prediction of PFV RTr products and their experimental verification, are required to provide experimental evidence for any of the proposed mechanisms, and combinations thereof, or they may reveal a currently unknown way of innate sensing of FV vDNA or RTr products and may identify additional cellular factors involved in this process.

#### **5. Conclusions**

To the best of our knowledge, this study is the first to demonstrate that replication-competent FVs can efficiently stimulate an innate immune response in human immune cells of the myeloid lineage. With FV particles known to contain significant amounts of reverse transcribed viral genome, it was not surprising that vDNA and/or RTr intermediates represent the PFV PAMP and cGAS, the main cellular PRR that are responsible for mounting an IRF3-dependent ISG response in this cell type. Elucidation of the underlying mechanism responsible for the differential innate stimulating capacities of full-length and minimal vector genomes in further studies may result in a more detailed characterization of the PFV genomic structures or sequence elements recognized by the host cell's innate immune system.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4915/11/12/1095/s1. Figure S1: Characterization of MDDC and MDM differentiation status; Figure S2: AZT-mediated inhibition of MDDC transduction by HIV-1 GFP; Figure S3: PFV-mediated IRF3 phosphorylation in MDDCs; Figure S4: Degradation of endogenous SAMHD1 upon VLP-Vpx transduction.

**Author Contributions:** Conceptualization, M.B., J.W., R.K., and D.L.; validation, M.B., J.W., R.K., and D.L.; formal analysis, M.B., J.W., R.K., and D.L.; investigation, M.B., J.W., M.R., A.E., J.S., N.S., and C.M.; resources, V.H.; data curation, M.B., J.W., A.E., N.S., R.K., and D.L.; writing—original draft preparation, M.B., R.K., and D.L.; writing—review and editing, M.B., R.K., and D.L.; visualization, M.B. and D.L.; supervision, R.K. and D.L.; project administration, R.K. and D.L.; funding acquisition, R.K. and D.L.

**Funding:** This research was funded by grants from the Deutsche Forschungsgemeinschaft (DFG: LI 621/10-1; SPP1923 project LI 621/11-1 to D.L. and DFG: SPP1923 project KO4573/1-1 to R.K.). J.W. was supported by a fellowship of the Else Kröner-Promotionskolleg of the Faculty of Medicine of the Technische Universität Dresden.

**Acknowledgments:** We would like to thank Frank Kirchhoff for providing pBR-NL43-Env−-IRES-eGFP-nef+. We would like to thank Christiane Tondera, Heike Schmitz, Michaela Neuenkirch, and Lavinia Schmitt for technical support. We acknowledge support by the Open Access Publication Funds of the SLUB/TU Dresden.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Appendix A**

**Figure A1.** PFV genome organization and viral expression systems used. Schematic outline of the wild type (wt) PFV proviral genome structure with *gag*, *pol*, and *env* ORFs. The locations of the point mutations for the individual proviral mutants resulting in variants with enzymatically inactive integrase (iIN) or reverse transcriptase (iRT); fusion-deficient Env (iFuse); or deficiencies in the translation of, Tas (ΔTas-Bet); Gag (ΔGag); Pol (ΔPol); Env (ΔEnv); or Gag, Pol, and Env (ΔGPE), due to inactivation of the respective translation initiation sites. PFV-RCP supernatants contain full-length viral genomic RNA (vgRNA) whereas PFV-SRV supernatants harbor vgRNA comprising the minimal essential cis-acting viral sequences (CAS). LTR: long terminal repeat; U3: unique 3 LTR region; U5: unique 5 LTR region; R: repeat LTR region; PR: protease; RT/RN: reverse transcriptase—RNase H; IN: integrase; LP leader peptide; SU: surface subunit; TM: transmembrane subunit; IP: internal promoter; ©: cap; An: poly-A tail; PBS: primer binding site; 3 PPT: 3 poly purine tract; cPPT: central PPT.




**Table A2.** Primer–probe set for SYBR Green qPCR analysis.


#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
