Rhesus macaque IFITM3 gene polymorphisms and SIV infection.

Interferon-induced transmembrane proteins (IFITMs) have been recognized as important antiviral effectors of the innate immune system, both in cell culture and in infected humans. In particular, polymorphisms of the human IFITM3 gene have been shown to affect disease severity and progression in influenza A virus (FLUAV) and human immunodeficiency virus (HIV) infection, respectively. Rhesus macaques (Macaca mulatta) are commonly used to model human infections and the experimental inoculation of these animals with simian immunodeficiency virus (SIV) is one of the best models for HIV/AIDS in humans. However, information on the role of IFITM3 in SIV infection of rhesus macaques is currently lacking. We show that rhesus macaque (rh) IFITM3 inhibits SIV and FLUAV entry in cell culture, although with moderately reduced efficiency as compared to its human counterpart. We further report the identification of 16 polymorphisms in the rhIFITM3 gene, three of which were exonic and synonymous while the remainder was located in non-coding regions. Employing previously characterized samples from two cohorts of SIV-infected rhesus macaques, we investigated the relationship between these rhIFITM3 polymorphisms and both AIDS-free survival time and virus load. In cohort 1, several intronic polymorphisms were significantly associated with virus load or survival. However, an association with both parameters was not observed and significance was lost in most cases when animals were stratified for the presence of MHC allele Mamu-A1*001. Moreover, no significant genotype-phenotype associations were detected in cohort 2. These results suggest that, although IFITM3 can inhibit SIV infection in cell culture, genetic variation in rhIFITM3 might have only a minor impact on the course of SIV infection in experimentally infected animals.


Introduction
Interferon-induced transmembrane proteins are encoded by a gene family of diverse function in vertebrates [1][2][3]. A subgroup of immunity-related IFITMs (IFITM1-3) has been recognized recently to inhibit multiplication of a broad range of viruses including, but not limited to, a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 influenza A viruses (FLUAV), flaviviruses, coronaviruses, filoviruses and lentiviruses [4][5][6][7][8][9][10][11]. IFITM proteins are type II transmembrane proteins and consist of a variable cytoplasmic Nterminus followed by a conserved intramembrane domain, an intracellular loop, a transmembrane domain and a short variable and luminal C-terminus [12][13][14]. It is well established that IFITMs inhibit viral entry [11]. While the exact mechanism still has to be elucidated, IFITMs seem to act through the arrest of virus-host cell membrane fusion at the hemifusion step by altering membrane fluidity [15,16]. Despite their broad antiviral activity, IFITMs can also promote entry or assembly of selected viruses [17,18], indicating that they belong to a pathway that can positively and negatively regulate viral infection.
All immunity-related IFITMs (IFITM1-3) have been shown to affect early stages of human and simian immunodeficiency virus (HIV, SIV) replication, with IFITM2 and IFITM3 inhibiting virus entry [8,19,20]. In addition, IFITMs may target later stages of the viral replication cycle such as Gag expression and proteolytic processing of Env [8,21,22]. The antiviral activity of IFITM1 can restrict HIV-1 cell-to-cell spread, and mutations in env and vpu have been identified that rescue the virus from inhibition by IFITM1 at the cost of reduced viral fitness [21,23]. Moreover, IFITMs are incorporated into lentiviral particles thereby reducing viral infectivity even though inhibition by IFITMs and virion incorporation of IFITMs were not found to be intimately related [21,24,25]. Finally, unlike other antiviral effector proteins of innate immunity, IFITMs do not seem to be antagonized by viral factors [22].
Different lines of evidence suggest that IFITMs can limit viral spread and reduce disease burden in the infected host. Thus, absence of IFITM3 expression was found to be associated with reduced FLUAV spread and pathogenesis in experimentally inoculated mice [26,27]. Moreover, a polymorphism in the human IFITM3 gene (rs12252 T>C) was found in several studies to be associated with severe influenza [16,[27][28][29], although one study reported an association with mild influenza only [30]. The rs12252 T>C polymorphism is believed to alter IFITM3 mRNA splicing [27] and to result in the expression of an IFITM3 variant lacking the first 21 amino acids, which displays reduced antiviral activity against FLUAV upon directed expression in cell culture [31]. Conversely, the truncated protein exerts increased anti-HIV-1 activity in transfected cells, due to preferential localization at the cell surface [32], the site of HIV-1 entry and exit from target cells. However, an association between rs12252-C and more rapid progression to AIDS, but not risk of HIV-1 infection, has been demonstrated [33], and the reason for the discrepancy between these results and the increased anti-HIV-1 activity of truncated IFITM3 in cell culture is at present unclear. Finally, a recent study demonstrated that HIV-1 variants sexually transmitted between individuals are resistant against blockade by IFITM proteins indicative of a strong selection pressure exerted by IFITMs [34,35]. Moreover, IFITM expression was shown to be a major contributor to the blockade of HIV-1 by treatment of cells with IFN [34], suggesting that IFITMs constitute important defenses against HIV-1 infection.
Non-human primates are an important animal model for infectious diseases in humans. In particular, experimental SIV infection of rhesus macaques has contributed substantially to our current understanding of HIV/AIDS. Polymorphisms in several immune-relevant gene loci such as MHC or KIR are associated with the transmission and course of disease in SIV infected rhesus macaques and HIV-1 infected humans [36,37]. Some primate IFITM gene products were recently shown to have antiviral activity against SIV and HIV-1 [19,20]. However, these analyses did not include rhesus macaque so that it is presently unknown whether rhesus macaque (rh) IFITM3 also inhibits SIV infection in this species. Moreover, no rhIFITM3 gene polymorphisms have been described so far or their modulatory effect upon infection by SIV or other viruses studied. Therefore, we systematically searched for polymorphisms in rhIFITM3 and assessed their association with viral load at set point and disease progression in the context of SIV infection.

Production of MLV particles
For the production of MLV particles encoding IFITM proteins, we followed a published protocol [6,7]. In brief, 293T cells were cotransfected with vector pQCXIP harboring rhIFITM3, huIFITM3 or CAT, an MLV gag-pol-encoding plasmid and a plasmid encoding vesicular stomatitis virus glycoprotein (VSV-G). For the production of reporter particles required for the analysis of IFITM3-mediated inhibition of viral entry, 293T cells were cotransfected with vector pQCXIP encoding firefly luciferase, an MLV gag-pol encoding plasmid and a plasmid encoding the viral glycoprotein to be tested. Medium was changed at 8 h and supernatants were harvested at 48 h post transfection, passed through a 0.45 μm filter, aliquoted and stored at -80˚C.

Analysis of the antiviral activity of rhesus macaque IFITM3
The antiviral activity of rhIFITM3 was analyzed as described previously [6]. In brief, 293T cells seeded at 10 4 cells per well in a 96-well plate were spin-oculated [38] at 4,000×g for 30 minutes with IFITM3 or CAT encoding vectors. After incubation for 48h at 37˚C, the medium was replaced by 50μl of fresh culture medium followed by the addition of 50μl of MLV vectors harboring the viral glycoproteins to be analyzed and normalized for equal luciferase activity upon transduction of control cells. After incubation for 8h, medium was replaced by fresh culture medium and luciferase activities in cell lysates were analyzed at 72h post infection, employing a microplate reader, Plate CHAMELEON V (Hidex, Turku, Finland), and a commercial luciferase assay system (Promega, Mannheim, Germany) or Beetle-Juice (PJK, Kleinblittersdorf, Germany) kits.

Animals
Genetic and phenotypic data from two cohorts of SIV-infected rhesus macaques (Macaca mulatta) of Indian origin, comprising a total of 94 individuals (19 female, 75 male), were screened retrospectively for rhIFITM3 polymorphisms ( Table 1). Samples of MHC-typed animals were selected that were infected intravenously, intrarectally or via tonsils through a single dose of SIVmac239, SIVmac251 or SIVmac239/32H and which displayed the whole spectrum of disease progression, from rapid progression to long-term non progression, for each combination of virus and infection route. All macaques were either bred at the German Primate Centre and had known ancestry, or were from breeders in France, United Kingdom or the United States. The definition of early AIDS-defining illness was based on clinical, necropsy and histopathological findings as described [39]. The 39 animals included in the first cohort were previously used as a 'screening cohort' to search for polymorphisms influencing AIDS-free survival time and viral load in SIV infection [39]. It should be noted that no viral load data were available for some of the animals included in the survival analysis. The second cohort of 55 animals consisted of macaques more recently infected, and viral load data were available for all animals.

Ethics statement
The animals were housed and treated at the German Primate Centre under standard conditions which are in accordance with the German Animal Protection Act and the European Union guidelines on the use of non-human primates for biomedical research as described [40]. Briefly, all experiments were approved by an external ethics committee authorized by the Lower Saxony State Office for Consumer Protection and Food Safety (project licenses: 509.42502/08-02.95, 509.42502/08-13.98, 504.42502/08-03.90 (V+Ä ), 33.9-42502-04-12/0820, 509-42502/08-04.03, 33.9.42502/04/017/07, 33.9.42502/04/72/08). According to §11 of the German Animal Welfare Act, the DPZ is permitted to breed and house non-human primates under license number 392001/7 issued by the local veterinary authorities. All macaques were under daily surveillance by veterinarians and animal caretakers. During quarantine, the animals were usually housed in groups of 2-3 animals per cage, with a minimum enclosure height of 180 cm and a cage volume of 3 to 4.5 m 3 . Animals in experiments had to be housed singly (cage dimensions [in cm]: 190h × 90w × 90d), but with constant visual, olfactory and acoustic contact to their roommates. Procedures for animal welfare and to minimize discomfort and suffering were undertaken in accordance with the recommendations of the Weatherall report "The use of nonhuman primates in research". Monkeys were fed with dry monkey biscuits supplemented with fresh fruit or vegetables twice daily and fresh water access ad libitum. For environmental enrichment, animals were offered feeding puzzles, varying toys and wood sticks for gnawing. In addition, music was played. For DNA preparation, blood samples were drawn under anesthesia with 10 mg ketamine i.m. per kg body weight. Animals were humanely euthanized by an overdose of Pentobarbital-Natrium (Narcoren 1 , Merial, Hallbergmoos, Germany) under anesthesia either for experimental reasons without exhibiting clinical symptoms or in cases of suffering predefined by a scoring system of termination criteria that was approved by the external ethics committee and corresponds to the IACUC endpoint guidelines as described [40]. For the purpose of this study, no monkey was sacrificed since archival samples were used.

Statistical analysis
AIDS-free survival analysis, including a log-rank test for rhIFITM3 genotype-related differences, was performed using the LIFETEST procedure of the SAS software version 9.4 (SAS Institute Inc., Cary, NC). The association between rhIFITM3 polymorphisms and virus load was assessed for statistical significance by means of either a Kruskal-Wallis test (three genotypes) or a Mann-Whitney test (two genotypes), as appropriate, using the GraphPad Prism software. All p-values were two-sided. The level of statistical significance was depicted as follows, Ã , p<0.05; ÃÃ , p<0.01; ÃÃÃ , p<0.001.

Rhesus macaque IFITM3 inhibits viral entry
One IFITM1 gene and two IFITM3 genes have been identified in rhesus macaques. For the purpose of the present study, we will distinguish between the two rhIFITM3 genes as rhIFITM3 (LOC697829) and rhIFITM3(2) (LOC697564) because rhIFITM3(2) shows a slightly higher sequence homology to the human IFITM2 gene than rhIFITM3. Here, we investigated whether rhIFITM3 inhibits viral entry in a way similar to human (hu) IFITM3, employing a pseudotyping approach. Expression of huIFITM3 in 293T cells did not appreciably inhibit entry driven by the MLV envelope protein (Env) or the Machupo virus glycoprotein (MACV-GPC) whereas entry driven by the FLUAV-hemagglutinin (HA) was robustly inhibited (Fig 1), in keeping with published findings [4][5][6]. Entry driven by SIV-Env was also reduced upon huI-FITM3 expression (Fig 1), again in agreement with published data [19,20]. A similar tendency was observed for rhIFITM3 even though inhibition of SIV-Env-and FLUAV-HA-driven entry was less robust than in the case of huIFITM3 (Fig 1), despite comparable expression levels [6]. Thus, rhIFITM3 exerts antiviral activity but does so less efficiently than huIFITM3.

Identification of polymorphisms in rhesus macaque IFITM3
We next aimed to characterize polymorphisms in the rhIFITM3 gene based upon a previously characterized cohort of 39 SIV infected animals (cohort 1) (Table 1), where relationships between animals and confounding factors had been minimized [39]. To obtain results from a larger collection of animals, we randomly selected additional 55 animals for inclusion into a second cohort (cohort 2) ( Table 1). The rhIFITM3 gene has two exons leading to a mature mRNA of 636 bp (XM_001085567.2) whereas the genomic region encoding the primary transcript comprises 1202 bp. We devised a PCR strategy to selectively amplify the genomic region of the rhIFITM3 gene so as to yield a 1340 bp amplification product (Fig 2). The DNA sequence was determined directly from purified PCR products, and sequence variations were identified at 16 positions (Fig 3 and Table 2). Three polymorphisms were located in the coding region but did not change the amino acid sequence. Twelve polymorphisms (including one single nucleotide deletion) were intronic and one polymorphism was located upstream of the coding sequence. Seven polymorphisms had a low level of heterozygosity ( 10%) but, with few exceptions, the two cohorts were found to be characterized by similar major allele   Table 2). Polymorphism g.488 C>A was only detected in the second cohort. Polymorphisms g.279 G>A and g.678 G>T showed higher heterozygosity in cohort 2, while the opposite was true for g.693 delA, g.708 A>G, g.709 G>A and g.747 T>G. Finally, a polymorphism in the 5´region of the coding sequence that corresponds to rs12252 T>C in the human IFITM3 gene was not identified in our sample.

Rhesus macaque IFITM3 polymorphisms in SIV infection
To assess whether the detected sequence variations had an influence on the course of SIV infection, we first analyzed their effect on AIDS-free survival time (Fig 4, Table 3). Polymorphisms g.270 C>T (genotype GT) and g.678 G>T (genotype GG) showed a nominally significant association with faster disease progression when all animals from both cohorts were considered. However, these associations were not observed when each cohort was analyzed separately and were not significant when animals with MHC allele Mamu-A1 Ã 001 were excluded from the analysis (not shown). Polymorphism g.279 G>A showed a highly significant association of the AA genotype with faster disease progression (log-rank test < 0.0001) and this association was also detected when carriers of Mamu-A1 Ã 001 (log-rank test < 0.0001) were excluded (Fig 4, Table 3). However, this variant was only little polymorphic (MAF: 0.95), and the significance of the association hinged heavily on a single AA animal in cohort 1. For a subset of 81 animals, virus load data were available that allowed the potential association between rhIFITM3 polymorphism and SIV amplification to be analysed. However, no statistically significant association with plasma viral RNA copies at 20 weeks after infection or at time of death was noted for any of the polymorphisms (Fig 5, Table 4). Only in cohort 1, we found a nominally significant association of polymorphisms g.693 delA (genotype -/-), g.708 A>G (genotype GG) and g.709 G>A (genotype AA) with increased viral load and of polymorphism g.678 G>T (genotype GT) with lower virus load (Table 4). However, these findings were not replicated in cohort 2 and were not detected when the full set of animals was analyzed.

Discussion
Immune-related IFITM proteins have been established as important antiviral effectors of the interferon response, and a polymorphism in the human IFITM3 gene has been found to be associated with disease severity and progression in FLUAV and HIV-1 infection [27,33]. Rhesus macaques are important animal models for human infections but it was hitherto unknown whether the rhIFITM3 gene is polymorphic and whether possible polymorphisms would impact upon the course of SIV infection. We demonstrated that rhIFITM3 displays antiviral activity and identified 16 polymorphisms in rhIFITM3. Some of the polymorphisms were associated with either disease progression or viral load. However, none of the polymorphisms impacted on both parameters, and significant associations were not observed in cohort 2. Moreover, most associations failed to attain statistical significance when Mamu-A1 Ã 001 animals were excluded from the analysis, arguing against a major impact of rhIFITM3 polymorphisms on SIV infection. Recently, a comprehensive characterization of the antiviral activity of IFITM proteins of non-human primates revealed that these proteins, like their human counterparts, can block viral infection [20]. However, IFITM proteins from rhesus macaques were not included in these studies. Our results demonstrate that rhIFITM3 and huIFITM3 target the same viral glycoproteins, but with somewhat different efficiency. Thus, entry driven by FLUAV-HA and SIV-Env appears to be more efficiently inhibited by huIFITM3 than by rhIFITM3. These results are in agreement with our previous analysis demonstrating that huIFITM3 inhibits entry driven by filovirus glycoproteins with greater efficiency than rhIFITM3, despite comparable expression in cells [6]. In summary, rhIFITM3 inhibits entry of SIV in cell culture which raised the question whether rhIFITM3 is polymorphic and whether possible polymorphisms would impact the course of SIV infection.
Our sequence analysis identified variations at 16 positions present in at least two out of 94 animals. Six of these polymorphisms were highly polymorphic (MAF <75%) and homozygous animals could be detected for both alleles. Three of the polymorphisms identified in our study were located within the coding sequences, but led to silent changes. This is noteworthy since IFITM3 polymorphisms in SIV-infection the N-terminal residues of IFITM3 control intracellular localization and spectrum of antiviral activity [32,43,44] and were suggested to be shaped by selective pressure during primate evolution [32]. One polymorphism was located upstream of the coding sequence; all other polymorphisms were identified within the intron. We found no evidence for a polymorphism that would lead to the production of a truncated protein similar to rs12252-C in human IFITM3, a variant that is associated with the clinical severity of influenza disease [27]. Our results were in keeping with a previous report suggesting that IFITM1 and IFITM3 in non-human primates probably underwent purifying selection [20]. Thus, we did not find any non-synonymous polymorphisms, indicating that maintenance of the primary amino acid sequence is essential for the function of the protein in rhesus macaques. Our analysis was based upon DNA sequences derived from assembly Mmul_051212 [41]. In the meantime, a new genomic framework (Mmul_8.0.1) has been published [45] where the sequences at the 3' end of the coding regions of rhIFITM3 (LOC697829) and rhIFITM3(2) (LOC697564) have been exchanged in comparison to the previous assembly (Mmul_051212). However, our previously reported cloning and sequencing of cDNAs from both loci [6] and the analysis presented here show better agreement with the rhIFITM3 locus as annotated in the older genomic assembly. Therefore, we continued to use Mmul_051212 throughout this study. A reconstruction of haplotypes using PHASE (v. 2.1) [46] based upon our results revealed 16 haplotypes for the rhIFITM3 gene, all of which clustered together and were clearly distinct from rhIFITM3(2) (data not shown). This reinsures us that all detected polymorphisms are true rhIFITM3 polymorphisms.
Silent mutations may affect translational efficiency and protein folding [47] whereas polymorphisms in untranslated regions or introns may alter mRNA stability or splicing [48]. In consequence, we analyzed the association between the polymorphisms and survival time and viral load in SIV-infected rhesus macaques. For this, we used a sample that previously allowed detection of an association between TLR7 polymorphisms and survival time (cohort 1, [39]) and a sample of randomly selected animals (cohort 2). When the full set of animals was considered, we identified three polymorphisms (g.270 C>T, g.279 G>A, g.678 G>T) with a nominally significant association between a genotype and AIDS-free survival, but not with virus IFITM3 polymorphisms in SIV-infection load at set point (20 weeks post infection). However, for two of these polymorphisms the association was not significant if the two cohorts were analyzed separately or when animals with MHC allele Mamu-A1 Ã 001 were excluded from the analysis. The polymorphism g.279 G>A genotype AA showed a nominally significant association with AIDS progression even after adjustment for MHC alleles. However, this position is only modestly polymorphic and the association observed was dependent on a single animal with AA genotype. Collectively, these results suggest that rhIFITM3 polymorphism might not play a major role in SIV infection. In summary, our study not only demonstrated that rhIFITM3 displays antiviral activity, but also led to the identification of polymorphisms in the coding and non-coding region of the IFITM3 gene of rhesus macaques. Notably, all polymorphism in the coding region were silent and strong evidence for an association of rhIFITM3 polymorphisms with disease progression and viral load in SIV infected animals was not obtained. We cannot exclude that the analysis of a substantially larger number of animals might reveal such associations. However, the frequency of potentially affected animals and/or the effect of the polymorphisms in terms of viral inhibition would be very low and are therefore not expected to significantly influence results obtained in the macaque model of AIDS.