NSG-Mice Reveal the Importance of a Functional Innate and Adaptive Immune Response to Overcome RVFV Infection

Rift Valley fever (RVF) is a zoonotic disease caused by RVF Phlebovirus (RVFV). The RVFV MP-12 vaccine strain is known to exhibit residual virulence in the case of a deficient interferon type 1 response. The hypothesis of this study is that virus replication and severity of lesions induced by the MP-12 strain in immunocompromised mice depend on the specific function of the disturbed pathway. Therefore, 10 strains of mice with deficient innate immunity (B6-IFNARtmAgt, C.129S7(B6)-Ifngtm1Ts/J, B6-TLR3tm1Flv, B6-TLR7tm1Aki, NOD/ShiLtJ), helper T-cell- (CD4tm1Mak), cytotoxic T-cell- (CD8atm1Mak), B-cell- (Igh-Jtm1DhuN?+N2), combined T- and B-cell- (NU/J) and combined T-, B-, natural killer (NK) cell- and macrophage-mediated immunity (NOD.Cg-PrkdcscidIl2rgtm1WjI/SzJ (NSG) mice) were subcutaneously infected with RVFV MP-12. B6-IFNARtmAgt mice were the only strain to develop fatal disease due to RVFV-induced severe hepatocellular necrosis and apoptosis. Notably, no clinical disease and only mild multifocal hepatocellular necrosis and apoptosis were observed in NSG mice, while immunohistochemistry detected the RVFV antigen in the liver and the brain. No or low virus expression and no lesions were observed in the other mouse strains. Conclusively, the interferon type 1 response is essential for early control of RVFV replication and disease, whereas functional NK cells, macrophages and lymphocytes are essential for virus clearance.


Introduction
Rift Valley fever (RVF) is a zoonotic disease caused by Rift Valley fever virus (RVFV) that is endemic in many regions of Africa and Arabia [1,2]. It poses a threat to many host species including cattle, goats, sheep and humans [2][3][4].
RVFV infection causes massive abortion storms and high mortality rates in neonatal ruminants, which are symptomatic of RVF and result in devastating economic losses of local ruminants [5,6]. Furthermore, the disease affects humans, especially those in close contact with animals [7,8]. While most human cases present as subclinical flu-like symptoms, severe or even fatal complications may occur, and the estimated case-fatality rate varies While some of the aforementioned aspects, e.g., the role of dendritic cells and macrophages as the first location of viral replication, are well-known, other aspects, e.g., the contribution of the different T-cell subtypes within the adaptive immune response, require further investigations; current research efforts focus on the detailed understanding of RVFV pathogenesis and immunogenesis [13]. The hypothesis of this study is that virus replication and severity of lesions induced by the MP-12 strain in immunocompromised mice depend on the specific function of the disturbed pathway. Therefore, the aim is to investigate the clinical disease, viral spread and lesion profile in mouse strains deficient in IFN type 1 signaling (B6-IFNAR tmAgt ), IFN type 2 signaling (C.129S7(B6)-Ifng tm1Ts /J) Toll-like receptor 3 and 7 mediated innate immunity (B6-TLR3 tm1Flv , B6-TLR7 tm1Aki ), defective natural killer (NK) cells and macrophages (NOD/ShiLtJ), helper T lymphocytes (CD4 tm1Mak ), cytotoxic T lymphocytes (CD8a tm1Mak ), B lymphocytes (Igh-J tm1Dhu N?+N2), T-and B lymphocytes (Foxn1 nu-/-) and combined NK-cells, macrophages and T and B lymphocytes (NOD.Cg-Prkdc scid Il2rg tm1WjI /SzJ). These differently impaired mice were chosen to evaluate single aspects of the immune response during RVFV infection.

Infection and Study Design
All animal experiments were conducted in accordance with German animal welfare laws and authorized by the responsible authority (Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern, permission LALLF 7221.3-1-038/17).
Group size was determined using a Cox's proportional hazards model superiority by a margin analysis employing literature-based a priori estimates of the median survival of the different strains [61,62]. Infection groups of six, nine or twelve specific pathogen-free mice as well as placebo groups of six mice of the same strain were randomly divided into three mice per cage (two to four cages per infection group, Table 1). They were kept in ventilated isocages (Tecniplast S.p.A., Buguggiate, Italy) and provided food (Ssniff Spezialdiäten GmbH, Soest, Germany) and water ad libitum. High caloric food from the same supplier was given to the NUDE mice, as they require increased energy intake due to their lack of fur.
After 14 days of acclimatization, the mice were infected subcutaneously in the neck with RVFV MP-12 (TCID 50 : 1.43 × 10 3 , 100 µL DMEM), while placebo groups were mock infected following the same inoculation route with the same amount of virus-free DMEM. All mice were observed for 14 days and daily body weight and clinical signs ( Table 2) were noted.
If the animals developed severe clinical signs, defined as a score of 3 in one category or a score of 2 in all categories, or after reaching 14 days post infection (dpi), they were euthanized by isoflurane anesthesia and subsequent cardiac puncture and blood collection. Necropsy was performed and organ samples were collected, including the brain, spleen, thymus (if applicable), liver, heart, kidney and lungs. Mice were euthanized after reaching a score of 3 in one category or a score of 2 in all categories.

RNA Isolation and Reverse Transcription Quantitative Polymerase Chain Reaction
Mouse tissue samples of 3-78 µg (mean 30 µg) liver, spleen and brain were lysed in cell culture medium using the QIAGEN TissueLyser II ® (QIAGEN GmbH, Hilden, Germany) and after centrifugation, RNA was isolated using the NucleoMag ® VET Kit (Machery & Nagel GmbH & Co. KG, Düren, Germany) and the automated KingFisher™ Flex Purification System (Thermo Scientific, Inc., Waltham, MA, USA). As internal extraction control, an MS2 bacteriophage was added to each sample [63]. The presence of RVFVand MS2-derived RNA was verified using a primer-probe based quantitative real-time RT-PCR (qRT-PCR), with a detection limit of five copies per reaction [64]. A synthetic RNA calibrator was utilized for quantification [65].

Serum Neutralization Test
Blood samples from mice were centrifuged and serum was tested for its neutralization ability in decreasing dilutions (serum neutralization test, SNT) as described previously [66]. Briefly, 100 TCID 50 of MP-12 were added to duplicates of serial two-fold diluted (from 1:10 to 1:120) and heat inactivated sera. Following an incubation of 30 min at 37 • C and 5% CO 2 , Vero-76 cells were added to each well. Plates were incubated at 37 • C, 5% CO 2 for six days. Neutralizing doses of 50%(ND 50 ) were expressed as the reciprocal of the serum dilution that still inhibited > 50% of cytopathogenic effect and calculated as described by Spearman and Kaerber [43,44]. The cytopathogenic effect was macroscopically evaluated after fixation with 4% paraformaldehyde and staining with 1% crystal violet.

Histology and Immunohistochemistry
Mouse tissue samples of brain, spleen, liver, heart, kidney and lungs were fixed in 4% neutral buffered paraformaldehyde for 21 days, were cut according to Registry of Industrial Toxicology Animal-data (RITA) trimming guides and were subsequently embedded in paraffin wax [67]. Four µm thick sections were stained with hematoxylin and eosin (HE) stain. In addition, immunohistochemistry (IHC) with a primary antibody targeting RVFV nucleoprotein (Np) was performed on 2 µm sections. Briefly, slides were dewaxed and rehydrated, microwaved in citrate buffer (700 Watt, 20 min, pH: 6.01) and blocked by 1 h incubation with rabbit serum. Thereafter, heat-inactivated serum from a RVFV-infected sheep was applied overnight at 4 • C as a primary antibody, followed by biotinylated rabbit anti-sheep secondary antibodies. ABC Kit Vectastain (PK 6100 Biozol Diagnostica Vertrieb GmbH, Eching, Germany) and AEC Substrat Chromogen Ready to use (K3464 Dako Denmark A/S, Glostrup, Denmark) were applied according to the manufacturer's instructions, followed by hematoxylin counterstain. RVFV-infected and non-infected cell pellets were used as positive and negative controls, respectively. Details of the staining protocols have been published previously [47,68,69].
Sections were evaluated by qualitative description of HE stained lesions and manual cell count of immunohistochemically labeled cells per 30 high power fields (1 HPF = 0.159 mm 2 ; using a Zeiss standard light microscope (Carl Zeiss AG, Oberkochen, Germany) with a Zeiss Kpl-W10x/18 ocular and a Zeiss 40/0.65 objective). Semi-quantification of immunohistochemistry was counted as follows: −: no findings; +: up to five positive cells per HPF; ++: five to 20 positive cells per HPF; +++: over 20 positive cells per HPF/diffuse expression of RVFV Np. Pictures were taken with an Olympus BX51 microscope and a DP72 Camera using the manufacturer's operating software cellSens, version 1.18 (Olympus Solutions Inc., Tokyo, Japan).

Statistical Analysis
The assumption of normal distribution of RT-qPCR results was rejected using Shapiro-Wilks' test and visual assessment of the qq plots of the model residuals. For descriptive statistics, measures of location and statistical dispersion were depicted as median and range with min and max values. The survival rate of all RVFV-infected mouse strains and placebo groups was calculated and depicted using the Kaplan-Meier analysis. SNT data were tested for differences between the infected groups and the placebo controls using a Kruskal-Wallis test for independent samples followed by bilateral multiple pairwise comparisons (Dwass, Steel, Critchlow-Fligner method). Figures and statistical workup of the gathered data were performed using the commercially available software Graphpad Prism (GraphPad Software, Inc., San Diego, CA, USA, version 9.0.0) and SAS 9.4 (SAS Institute, Inc. Cary, NC, USA).

Survival and Clinical Signs
The RVFV-infected B6-IFNAR tmAgt mice died or had to be euthanized 3 dpi (Figure 1) due to severe clinical signs of disease, including lack of spontaneous or provoked movement, apathy, labored breathing, ruffled fur, hunched posture and final weight loss of over 15% of the original body weight. Furthermore, one RVFV-infected NSG mouse had to be euthanized at 11 dpi due to similar severe clinical signs. All other animals in this group and in the other mouse strains survived until the end of the experiment at 14 dpi and did not show any clinical signs besides mild weight loss (<15%).

Reverse Transcription Quantitative Polymerase Chain Reaction
Detailed RT-qPCR results are presented in Table 3. Table 3. RT-qPCR detection of RVFV RNA shown as positive animals/mouse strain and copies/µL RNA in liver, spleen and brain.

Histology and Immunohistochemistry
Histology and immunohistochemistry results of B6-IFNAR tmAgt and NSG mice are presented in Table 4.
RVFV-infected B6-IFNAR tmAgt mice exhibited a severe, diffuse, hepatocellular necrosis characterized by hypereosinophilic cellular debris and nuclear pyknosis as well as karyorhexis. Apoptosis was also present as shown by swollen, eosinophilic hepatocytes (resembling Councilman bodies). Furthermore, multifocal hemorrhages were detected. The same hepatic changes were found to a lesser (mild) degree in two RVFV-infected NSG mice ( Figure 4, Table 4). In addition to the liver findings, B6-IFNAR tmAgt mice showed moderate lymphocytolysis in the red pulp of the spleen and mild lymphocytolysis in splenic follicles (Figure 4, Table 4). All other mice, including placebo controls, did not show any lesions. Moreover, the CNS, lung, heart and kidney lacked significant microscopic lesions in all groups.   Immunohistochemistry revealed RVFV Np in a multifocal to diffuse pattern throughout the liver tissue of RVFV-infected B6-IFNAR tmAgt mice ( Figure 5, Table 4). Furthermore, there were abundant amounts of RVFV antigen present within the cytoplasm of sinusoidal macrophages in the spleen, and there were single neurons labeled positive in 3/6 (50%) mice ( Figure 5, Table 4). In contrast, RVFV-infected NSG mice exhibited RVFV antigen within the cytoplasm of neurons and small groups of hepatocytes ( Figure 5, Table 4). Two and three animals from the RVFV-infected B6-IFNAR tmAgt and NSG mice, respectively, exhibited immunohistochemical labeling of RVFV Np in epithelial cells of kidney tubules and lung bronchioles ( Figure 6). However, no histological lesion was associated with these signals. The remaining groups and tissues, including the placebo controls, lacked expression of RVFV Np.
In all host species, innate immunity is a key component of early RVFV detection and activation of further effector cells, e.g., macrophages, via release of cellular mediators including IFNs [24,27,34,45,70]. The B6-IFNAR tmAgt mice lack the IFN type I receptor (IFNAR) and showed severe RVF characterized by severe clinical signs, necrotizing hepatitis and virus spread to various organs as described previously [37,50]. However, other knockout mice strains with a deficient innate immunity, including lack of TLR 3, TLR 7 and IFN γ activity, respectively, were characterized by absence of clinical disease, virus antigen expression and lesion development in the investigated organs [52,70,71]. These results show the pathogenicity of the RVFV MP-12 strain in an IFN type 1 deficient immune system [37]. Clinical signs, pathological lesions, immunohistochemistry results and RT-qPCR detection of high amounts of viral RNA in the B6-IFNAR tmAgt mice resemble previously published descriptions of severe RVF in susceptible mice, infected with different RVFV strains including MP-12, and due to their resemblance to human RVF courses, mice have been used as a model species [18][19][20][21]72]. The effect of the other knockouts of the innate immune system, however, appears to be compensated by other mechanisms. IFN γ is the main activator of macrophages, which together with other antigen-presenting cells, initiate the adaptive immune response [73]. However, activation of macrophages can be obtained by other stimuli, e.g., tumor necrosis factor (TNF), which is a substantial component of the interleukin pathway produced by Th17 helper cells [73]. TLR 3 and 7 are pivotal components for RNA detection, and their contributions to the early immune response within viral infection have been shown for multiple viral diseases, e.g., dengue fever [29,74]. TLR activation is another initial step in the immune response triggering cascade, as both receptors recognize double-strained or single-stranded RNA, respectively, and therefore initiate antiviral immune responses [29]. As neither B6-TLR3 tm1Flv nor B6-TLR7 tm1Aki mice developed clinical RVF, initial recognition of the virus seems to be sufficiently achieved by one of the TLR or another intracellular receptor of viral RNA. Alongside TLR 3 and 7, other receptors of viral RNA, e.g., RIG-I, have been shown to be associated with severe RVF in humans [33]. Therefore, other RNA receptors might represent promising targets of future investigations [13]. RT-qPCR results indicate that mice lacking TLR 7 may require more time for successful virus elimination, as shown by 5/12 mice that exhibited small loads of viral genetic material in the spleen at 14 dpi indicating a less efficient activation of virus clearance mechanisms, e.g., macrophage activation, when lacking TLR 7. This difference in receptor efficacy might be explained by the respective triggered pathways. While TLR 3 triggers an IFN response via a TRIF/TRAM-dependent pathway, TLR 7 achieves this by activating the MyD88/TIRAP pathway [29]. Furthermore, different virus infections highlight the different roles of TLR 3 and 7. Although TLR 3 signaling appears crucial to the immune response against some infections, e.g., herpes simplex virus, the importance of TLR 7 was shown for others, e.g., avian influenza [30,75,76].
In most viral diseases including RVF, the key aspect of the adaptive immunity is the lymphocyte response that includes the production of antibodies and the development of a long-lasting immunity after virus clearance [45]. This protection by RVFV-targeted antibodies has been shown in several studies evaluating RVF vaccination attempts in ruminants [77][78][79]. CD4 + , CD8 + , T lymphocyte or B lymphocyte impaired mice were able to control the disease and developed neither clinical disease nor significant viral loads as determined by RT-qPCR and immunohistochemistry. This shows that a full elimination can be achieved by an intact IFN response followed by effector cells, e.g., macrophage activity, the role of which during RVF development was already studied with regard to initial virus spread [80,81]. Furthermore, as the adaptive immune response plays a crucial role in the later course of RVF, a successful elimination of the virus may even be achieved predominantly by the innate immune response as RVFV strain MP-12 is viewed as widely attenuated in IFN competent mice [20,37]. Over and above the deficiency in T-and B-cell-mediated adaptive immunity shared with NUDE mice, NSG mice are also deficient in innate immune pathways; namely, they are deficient in NK cells and their macrophages exhibit functional defects, leading to defective antigen presentation and altered immunoregulation [57][58][59]. They neither develop follicles in spleen and lymph nodes, nor Natural Killer cells, B-cells nor detectable antibody titers in serum samples [54,57]. These mice can be used for cancer or xenograft research studies as they are resistant to lymphoma development and do not show inflammatory responses to xenograft implants [54,57,82]. It seems reasonable, that the combination of defective innate pathways and lacking adaptive immunity in NSG mice as compared to NOD and NUDE mice is responsible for the lack of host control of ongoing virus replication in the NSG mice. Concerning the molecular pathway responsible for reduced or lacking function of NK-cells, the Il2rg tm1Wjl null mutation in NSG mice, which results in the loss of most of the extracellular domain and all of the transmembrane and cytoplasmic domains of the protein, is different from their background strain NOD, which exhibits an Il2 m1 hypoactive polymorphism [83]. This difference may be the reason for a more profound deficiency in functional NK cells in NSG mice compared to the NOD strain. An increased activation of and a suggested alleviating effect on the early stage of viral infections by NK cells has been shown for various viral infections such as dengue virus, Zika virus, hantavirus or tick-borne encephalitis virus [84]. While their role in systemic RVFV infection remains unclear in detail, a rise of NK cells in liver and lymph nodes of RVFV-infected mice has been observed previously [44]. Therefore, a lack of NK cells could be a contributing factor for the lack of virus clearance in NSG mice and therefore warrants further investigations.
Clinical disease was observed in only one of six NSG mice, although all six animals exhibited viral loads in the RT-qPCR analysis and in four animals, viral loads were detected in the liver, spleen and brain, indicating a viremic phase and infection of the CNS as described during the course of RVF [3,4]. However, RVFV antigen was found in hepatocytes and neurons, while no CNS lesions and only small foci of hepatocellular damage were observed. These results suggest that NSG mice are unable to control viral spread, but at the same time do not regularly succumb to or even develop clinical disease. The IL15-receptor is a heterodimer of the IL-2/IL-15 receptor beta chain (CD122) and the common gamma chain (gamma-C, CD132), the latter being the target of the null mutant in NSG mice [83]. There are no studies on the importance of NK cells in human or ruminant RVF pathogenesis. However, it has been shown that clinically ill Puumala hantavirus-infected humans exhibit a marked increase and activation of NK cells due to an IL15-mediated mechanism [85,86]. This NK overactivation has been suggested to be a factor involved in the pathogenesis of Puumala hanta virus-induced disease, leading to collateral damage on surrounding host cells regardless of their infection status [85,86]. Therefore, it seems reasonable that the lack of NK cells in NSG mice is mediated by a lack of IL2rg-function and may be the reason for both, the lack of MP-12 RVFV clearance as well as an explanation for the lack of lesions and clinical disease in the current study. Further studies are needed to unravel the role of NK-cell mediated mechanisms in the pathogenesis of RVFV-induced hepatitis and encephalitis in other host species such as humans and cattle.
Regarding the applicability of the present results and its conclusions to other host species, two limiting factors must be mentioned. Although mice are an established animal model of RVF and widely used, the present mouse strains are genetically modified and lack certain aspects of their immune responses [18,19,87]. Although the variety of investigated knockout and background strains within the present study allows scientifically sound conclusions with respect to the murine immune system, the importance of different cell types of the immune response, including the NK cell response, may differs in other species, especially humans and ruminants. These aspects still require detailed investigations [13]. Furthermore, the RVFV strain used in the present study is not a wild-type RVFV strain. RVFV MP-12 is attenuated by its heat sensitivity and a combination of several mutations in all three segments of its genome [20,88]. These multiple mutations were derived by repeated mutagenic passages in 5-fluorouracil [89]. However, the main pathogenicity factor, NSs, is only partially attenuated in RVFV MP-12 and retains some of its virulence [90]. Moreover, RVFV MP-12 still has been shown to be lethal to certain, susceptible mouse strains, e.g., IFNAR -/or STAT1 -/mice [20,37]. Furthermore, its use is allowed under biosafety level (BSL) 2 conditions, and it is even conditionally licensed as a vaccine in the United States [20,37,88]. A non-attenuated wild-type isolate of RVFV may be less likely to show similar results as RVFV MP-12 did in the present study, due to its increased virulence.
During its evaluation as a RVF vaccine, RVFV MP12 was shown to induce seroconversion and protection against re-infection within ruminants and primates [91,92]. The seroconversion was measured in the present study, and although no statistical significance was yielded, a trend toward successful seroconversion in a subset of infected mouse strains was observed. B-cell deficient mice and mice lacking CD4 + T-helper cells are naturally unable to produce neutralizing antibodies, and the B6-IFNAR tmAgt succumbed to disease too quickly to develop a detectable adaptive immune response [93]. The remaining mouse strains exhibited inconsistent results ranging from no to high antibody titers with exception of the NOD mice that did not show seroconversion at all, most likely due to their deficiency in antigen presentation and immune regulation [57][58][59].
In summary, the present study shows that a deficiency of IFN type 1 signaling results in fatal RFV disease in mice, whereas a lack of functional NK-cells and combined lack of Tand B-cells as well as defective macrophages is associated with continuous viral replication without development of clinical disease and lesions in the liver and CNS in NSG mice. However, as mice are an established model species but do not represent a natural host to RVF and as an attenuated RVFV strain was used, the results must be interpreted with caution and warrant further investigation regarding their relevance and applicability to humans and ruminants.

Data Availability Statement:
The data presented in this study are not publicly available but are available upon reasonable request.