Recurrent loss-of-function mutations reveal costs to OAS1 antiviral activity in primates

SUMMARY

Immune responses counteract infections but also cause collateral damage to hosts. Oligoadenylate Synthetase 1 (OAS1) binds double-stranded RNA from invading viruses, and produces 2′–5′ linked oligoadenylate (2–5A) to activate Ribonuclease L (RNase L), which cleaves RNA to inhibit virus replication. OAS1 can also undergo autoactivation by host RNAs, a potential trade-off to antiviral activity. We investigated functional variation in primate OAS1 as a model for how immune pathways evolve to mitigate costs and observed a surprising frequency of loss-of-function variation. In gorillas, we identified a polymorphism that severely decreases catalytic function, mirroring a common variant in humans that impairs 2–5A synthesis through alternative splicing. OAS1 loss-of-function variation is also common in monkeys, including complete loss of 2–5A synthesis in tamarins. The frequency of loss-of-function alleles suggests that costs associated with OAS1 activation can be so detrimental to host fitness that pathogen-protective effects are repeatedly forfeited.

INTRODUCTION

The protective effect of immune defense provides clear benefits to host fitness but can also exact costs. A growing collection of studies demonstrate that experimentally induced innate immune activation reduces longevity and fecundity (Moret and Schmid-Hempel, 2000; Sadd and Siva-Jothy, 2006; Schwartz and Koella, 2004). Related work shows that increased pathogen resistance is often tied to reduced fitness in the absence of infection (Fellowes et al., 1998; Kraaijeveld and Godfray, 1997). It follows that the cost-benefit balance of immune responses fluctuates between species depending on the intensity and frequency of threats from infectious microbes or changes in host biology. These differing histories of exposure can select for increased defense responses in some host lineages and decreased responses in others (Schmid-Hempel, 2003). How these selective forces might shape the evolution of individual immune pathways remains largely undetermined. Here we investigate diversity in the antiviral Oligoadenylate Synthetase 1 (OAS1)/Latent Ribonuclease (RNase L) pathway in primates as a model system for the evolutionary balance between beneficial and detrimental outcomes of immune functions.

The collateral damage caused by OAS1/RNase L pathway activation provides a useful experimental system for studying the tradeoffs involved in evolution of immune responses. OAS proteins are a critical mediator of innate immunity and function by sensing foreign double-stranded RNA (dsRNA) from invading viruses in the cytosol (Kristiansen et al., 2010a). Upon dsRNA binding, OAS proteins convert ATP into polymer chains joined by 2′–5′ linkage referred as oligoadenylate (2–5A) (Kristiansen et al., 2010a). The only reported role for 2–5A is to activate RNase L, which cleaves viral and host RNAs, leading to a potent block of viral replication and eventually apoptosis in infected cells (Chakrabarti et al., 2010). OAS1 recognizes a general motif of 17 or more base pairs of double stranded RNA with little preference for nucleotide sequence, a pattern frequently occurring in structures within the human transcriptome (Donovan et al., 2013). Indeed, constitutive editing of cellular dsRNA is required to suppress RNase L induced lethality in cultured human cells, highlighting the active measures taken to protect host cells from the deleterious effects of OAS activation in the absence of infection (Li et al., 2017b).

Although OAS1 is the most ancient of the OAS genes, its volatile evolutionary history is consistent with potential costs. Several animal lineages, including insects and teleost fish, have lost OAS1 completely (Kjaer et al., 2009). In contrast, OAS1 has undergone extensive gene amplifications in rodents and even-toed ungulates (Perelygin et al., 2006). In mice, the OAS family has expanded to include eight Oas1 genes (Oas1a-h). Of these OAS1 copies, only two are catalytically active, while OAS1b and OAS1d antagonize 2–5A generation through a dominant negative mechanism, highlighting a need to modulate harmful 2–5A levels (Elbahesh et al., 2011; Kakuta et al., 2002). OAS1 amino acid sequence is also highly variable across species. Phylogenetic and population genetic analysis demonstrate that OAS1 has evolved under recurrent positive selection in primates and other mammals, resulting in remarkable sequence diversity (Hancks et al., 2015; Mozzi et al., 2015). A lack of comparative studies of OAS1 activity, however, leaves the functional consequences of OAS1 diversification unknown.

Here we investigate diversity in OAS1 function among primates and report a recurring pattern of fixed or polymorphic alleles that decrease or eliminate enzymatic activity. In gorillas, we identify a high frequency variant in a single active-site amino acid position leads to subtantially reduced 2–5A production. In contrast, we show that variation in human OAS1 previously associated with reduced 2–5A levels likely results from decreased protein levels rather than enzymatic capacity. Finally, we find that loss of OAS1 activity is common among monkey species, including complete loss of 2–5A production in tamarins. Our findings highlight the cost-benefit conundrum of immune functions on host fitness, where deleterious effects of immune activation may sometimes outweigh the benefits of host defense.

RESULTS

A yeast growth assay for OAS1/RNase L pathway activity

To investigate functional differences among OAS1 and RNase L variants, we developed an assay to efficiently measure changes in cellular viability resulting from RNase L activation by OAS1 signaling (Figure 1A). Budding yeast, which constitutively produce cytosolic dsRNA, are a proven system for studying the function of the dsRNA sensing Protein Kinase R (PKR), where eIF2α phosphorylation by PKR leads to attenuation of protein translation and growth arrest (Dever et al., 1993; Elde et al., 2008). We tested whether heterologous co-expression of human OAS1 and RNase L might similarly arrest yeast growth due to cellular RNA degradation. While galactose-induced expression of either OAS1 or RNase L alone has no effect, co-expression of both proteins robustly inhibits yeast growth (Figure 1B). Furthermore, growth arrest specifically depends on RNase L activation, as yeast expressing OAS1 and a catalytically inactive RNase L^H672A grow normally (Figure 1B). Growth attenuation also corresponds with ribosomal RNA (rRNA) cleavage and reduced protein translation, a hallmark of RNase L activation (Li et al., 2016) (Figure 1C). Thus, OAS1/RNase L pathway function is recapitulated when genetically transplanted into yeast cells, providing a versatile platform to efficiently measure pathway output.

Figure 1: A yeast growth assay for OAS1/RNase L pathway activity.

(A) Schematic of the OAS1/RNase L pathway. OAS1 is activated when bound to dsRNA, producing 2–5A chains of varying length to activate RNase L. Degradation of cellular RNAs by RNase L eventually leads to cell death. (B) Yeast strain w303 was transformed with a galactose inducible expression plasmid encoding the indicated human proteins. Tenfold serial dilutions were plated on either galactose or glucose media and imaged after 24 hours (glucose) or 48 hours (galactose). (C) Immunoblot analysis of OAS1, RNase L, and actin from yeast lysates transformed with the indicated plasmids (top 3 panels). Total RNA integrity was assessed by bioanalyzer from yeast strains induced with galactose for six hours (bottom panel). Black arrow indicates 18S rRNA band, red arrow indicates major RNase L cleavage products. Data are representative of at least three independent experiments.

A high frequency SNP in gorilla OAS1 controls catalytic output

To understand functional diversity of OAS1 in primates, we first compared OAS1 cloned from four hominoid species via coexpression with human RNase L in yeast. In developing the yeast assay, we found expression of the OAS1 inhibitor Us11 encoded by Herpes Simplex Virus 1 increases sensitivity and unmasks differences in OAS1 activity between species. Us11 broadly inhibits dsRNA sensing in infected cells by sequestering dsRNA (Sànchez and Mohr, 2007), and lowers OAS1 activity in yeast below the threshold of complete growth arrest (Figure 2A/B). Two dominant OAS1 isoforms, p42 and p46, are produced by humans and are distinguished by alternative splicing of the final exon. We initially chose to compare the longer p46 isoform to understand how variation in the C-terminus might influence activity. Unexpectedly, OAS1 p46 mRNA cloned from chimpanzee and gorilla fibroblasts harbor early stop codons in exon five or six, respectively, resulting in smaller protein products (Figure 2A/S1A). When expressed in yeast, human, chimpanzee, and orangutan OAS1 similarly activate human RNase L and arrest growth. Yeast expressing gorilla OAS1, however, grow robustly in the presence of Us11 (Figure 2A). These data suggest that gorilla OAS1 is deficient in 2–5A synthesis compared to its hominoid cousins.

Figure 2: A high frequency SNP in gorilla OAS1 controls catalytic output.

(A) Yeast constitutively expressing HSV-1 Us11 from the LEU2 locus (see methods) were transformed with vector pBM272 encoding human RNase L and OAS1 from the indicated species, plated in serial tenfold dilutions on galactose medium and imaged after 48 hours (left). Immunoblot analysis of OAS1, RNase L, and HA-tagged Us11 protein (right). (B) Procedure as in panel A, with yeast expressing gorilla OAS1^C130R and human OAS1^R130C. (C) Space-filling model of crystal structure of porcine OAS1 (blue) bound to RNA (gray) in complex with ATP analog ApCpp donor (tan) and acceptor (yellow) (PDB 4RWN). Residue 130 is highlighted in red. Inset ribbon model highlighting interaction between arginine 130 and acceptor ATP (right). Amino acid sequence alignment of OAS1 residues 127–133 from a panel of mammals, highlighting substrate interacting residue 130 (green triangle) (right). (D) Gel image of ATP polymerization assay. Indicated OAS1-MBP fusion proteins (OAS1 residues 3–346) were expressed and purified from E. coli and incubated with poly I:C and [ɑ−32P] ATP for 2h before resolving by polyacrylamide gel electrophoresis (methods). Lane 1–4 are derived from reactions containing 5, 1, 0.2, or 0.04 μM OAS1, respectively. All data are representative of at least three independent experiments

To determine the basis of reduced activity in gorilla OAS1, we engineered a series of human-gorilla chimeric OAS1 proteins and tested their ability to activate RNase L in yeast (Figure S1B). With this approach, we identified an arginine to cysteine amino acid substitution at codon 130 (R130C) unique to gorilla as a candidate mutation causative of the attenuated phenotype (Figure S1C). We next tested the effects of codon 130 mutation in both human and gorilla OAS1. Reversion from cysteine 130 to the ancestral arginine in gorilla OAS1 restored activity to levels similar to other hominoids, while the converse arginine to cysteine substitution in human OAS1 significantly reduced activity as judged by robust yeast growth (Figure 2B). The R130C substitution in gorilla OAS1 is therefore responsible for its decreased level of 2–5A synthesis in yeast.

Reduced 2–5A synthesis by gorilla OAS1 might reflect reduced dsRNA affinity or reduced catalytic capability. To distinguish between these possibilities, we mapped residue 130 to a previously published crystal structure of OAS1 in complex with ATP analog ApCpp (Lohöfener et al., 2015). Consistent with a role in catalysis, Arginine 130 is located in the OAS1 active-site cleft and hydrogen bonds with the α-phosphate of the acceptor ATP molecule (Figure 2C). To directly test the impact of the R130C mutation on catalytic function, we measured OAS1 activity in vitro with purified components. This technique allows for the direct visualization of 2–5A products generated from purified proteins with excess available dsRNA. Consistent with our observations in yeast, gorilla OAS1 and human OAS1^R130C produce far less 2–5A than wild-type human OAS1 or gorilla OAS1^C130R (Figure 2D), with both proteins displaying approximately 5-fold lower activity relative to wild-type human OAS1 (Table S2). Importantly, we confirmed that gorilla RNase L functions equivalently to human RNase L in yeast and is not more sensitive to the enzymatic products of the low activity OAS1 variant (Figure S1E), ruling out compensatory co-evolution. Together, these data show that the gorilla OAS1 variant recovered is catalytically impaired by a mutation in a critical active site residue.

Variation in OAS1 has previously been shown to modulate 2–5A production within human populations (Bonnevie-Nielsen et al., 2005). Extensive polymorphism also exists in nonhuman primate OAS1 (Fish and Boissinot, 2015), leading us to investigate whether the OAS1 allele we recovered is fixed or variable in gorilla populations. We found evidence for polymorphism by examining a whole-genome sequencing dataset generated from 31 individual gorillas (Prado-Martinez et al., 2013). Among this cohort, the C130 allele occurs at a frequency of 36%, while the remainder encode the full activity ancestral arginine (Table S1). In western gorillas (Gorilla gorilla gorilla), 2/24 individuals are homozygous for the low activity allele (TT), 12/24 are heterozygous (CT), and 10/24 are homozygous for the ancestral full activity allele. The low activity allele is also present in eastern gorilla individuals (Gorilla gorilla graueri), suggesting an ancient origin (Table S1). Western gorillas are estimated to have diverged from eastern gorillas approximately 261 Kya, with some gene flow continuing between the populations until between approximately 77–150 Kya (McManus et al., 2015; Thalmann et al., 2007). Therefore, the R130C variant has been maintained in gorilla populations for least 77 thousand years.

Variation in human OAS1 activity does not reflect altered catalytic capacity

Similar to our findings in gorillas, high frequency variation in human OAS1 has previously been associated with decreased 2–5A synthesis. This variation in OAS activity results from a single G/A SNP in the OAS1 exon 6 splice-acceptor (rs10774671), which influences the repertoire of alternatively spliced OAS1 isoforms. Five human OAS1 isoforms have been described at the RNA level, each encoding alternative C-terminal exons that result in altered molecular weight: p42, p44, p46, p48, and p52. Individuals with the ancestral G allele predominantly produce the p46 enzyme, while those with the derived A allele transcribe the p42, p44, p48 and p52 isoforms (Figure 3A). OAS activity is also significantly reduced in individuals with one or two copies of the A allele (Bonnevie-Nielsen et al., 2005), leading to speculation that human isoforms differ in catalytic capacity. We investigated the basis of decreased activity in individuals with the A genotype. When expressed in yeast, both p42 and p46 robustly activate RNase L and arrest cell growth. The p44, p48, and p52 OAS1 isoforms, however, only weakly inhibit growth (Figure 3B). The abundance of these isoforms is also decreased, correlated with their ability to arrest growth, suggesting a defect at the level of protein accumulation (Figure 3C).

Figure 3: Variation in human OAS1 activity does not reflect altered catalytic capacity.

(A) Summary of mRNA splice variants made by individuals with the A or G variant at exon 6 splice acceptor site (rs10774671). Individuals with the A allele have lowered OAS activity (see text). (B) Tenfold serial dilutions of yeast expressing human RNase L and the indicated human OAS1 isoform imaged after 48 hours of growth on galactose medium. (C) Immunoblot analysis of OAS1, RNase L and Us11 expression from yeast (D) OAS1 Immunoblot of human cell lines A549 and a primary Human Fibroblast (HF) with or without IFN-α stimulation. Isoform lane is a mixture of 293T cell lysates transfected with pcDNA6 expression vector encoding human OAS1 p42 or p46 (top) Sanger sequencing traces at the OAS1 exon 6 splice junction for each cell lines are shown (bottom) (E) Gel image of ATP polymerization assay. Indicated human OAS1 isoforms were expressed and purified from E. coli as MBP fusion proteins and incubated with poly I:C and [ɑ−32P] ATP for 2h before resolving by polyacrylamide gel electrophoresis (methods). Lane 1–4 are derived from reactions containing 5, 1, 0.2, or 0.04 μM OAS1, respectively. All data are representative of at least two independent experiments.

Consistent with our results in yeast, only OAS1 p42 and/or p46 were observed following IFN-α treatment of human cells heterozygous or homozygous for the A allele, while p44, p48, and p52 were undetectable (Figure 3D) These results agree with previous studies (Li et al., 2017a), which have demonstrated a lack of the p44, p48, and p52 isoforms in human cells despite transcriptional induction, providing a basis for lowered OAS activity. To rule out differences in catalytic activity between isoforms, we directly tested enzymatic activity in vitro using equimolar amounts of full-length purified protein. Similar to yeast and human cells, p44, p48 and p52 are poorly expressed in E. coli, resulting in much lower protein purity (Figure S2). Despite these differences in purity, each of the five isoforms is capable of robust 2–5A synthesis, albeit with lower activity from the p44 and p52 preparations at low concentration (Figure 3E/Table S2). These data indicate that reduced OAS activity in humans with the A allele is best explained by decreased OAS1 protein abundance, rather than lower catalytic activity of any individual isoform.

OAS1 activity is variable among monkeys and lost in tamarins

To determine if loss-of-function variation in human and gorilla reflects a general pattern in OAS1 evolution, we next tested OAS1 function from a more diverse sampling of primate species. To capture a broad taxonomic and genetic distribution, we cloned OAS1 p46 from five Old World and three New World monkeys and tested their ability to activate human RNase L in yeast. Similar to gorilla, dusky titi OAS1 p46 contains an early stop codon in exon 6, resulting in a smaller protein product (Figure 4A/S1A). Yeast growth is strongly inhibited by expression of African green monkey, rhesus macaque and baboon OAS1 with human RNase L. Both talapoin and white faced saki OAS1, however, display an intermediate phenotype with lowered relative levels of RNase L activation (Figure 4A). Consistent with reduced yeast growth inhibition, in vitro 2–5A synthesis assays confirmed that talapoin OAS1 produces less 2–5A compared to baboon or African green monkey OAS1 (Figure 4B/Table S2). In contrast, yeast expressing saddleback tamarin OAS1 grew robustly, even in the absence of Us11, indicating a complete loss of OAS1 enzymatic function (Figure 4A/4D). OAS1 activity is therefore highly variable among the monkey species sampled in our study, with alleles ranging from high, intermediate, or zero capacity to activate RNase L in yeast.

Figure 4: OAS1 activity is variable among monkeys and lost in tamarins.

(A) Tenfold serial dilutions of yeast expressing human RNase L and OAS1 from the indicated species (AGM – African green monkey) were imaged after 48 hours of growth on galactose medium (Left). Immunoblot analysis of OAS1, RNase L and HA-Us11 from yeast cell lysates after 6h of galactose induction (Right) (B) Gel image of ATP polymerization assay. Indicated OAS1-MBP fusion proteins (residues 3–346) were expressed and purified from E. coli and incubated with poly I:C and [ɑ−32P] ATP for 2h before resolving by polyacrylamide gel electrophoresis (methods). Lane 1–3 are derived from reactions containing 5, 1, or 0.2 μM OAS1, respectively. (C) Space-filling model of crystal structure of porcine OAS1 (blue) bound to RNA (gray) in complex with ATP analog ApCpp donor (tan) and acceptor (yellow), and Mg²⁺ (purple) (PDB 4RWN). Residues 64, 74, 77, and 130 mutated in tamarin OAS1 are highlighted in red (top). Amino acid sequence alignment of OAS1 residues 61–78 and 128–132 from the indicated mammalian species (mouse OAS1a). Green triangles indicate mutations in conserved residues implicated in substrate interaction, red triangles indicate mutations in catalytic residues (bottom). (D) Serial dilutions of yeast strain w303 transformed with empty integrating pRS405 vector at the LEU2 locus and pBM272 encoding human RNase L and mustached or saddleback tamarin OAS1 were grown on galactose medium and imaged after 48 hours of growth (left). Immunoblot analysis of tamarin OAS1 and human RNase L expressed in yeast (right). (E) Gel image of ATP polymerization assay. Indicated OAS1 MBP fusion proteins (residues 3–346) were expressed and purified from E. coli and incubated with poly I:C and [ɑ−32P] ATP for 2h before resolving by polyacrylamide gel electrophoresis (methods). Lane 1–4 are derived from reactions containing 5, 1, 0.2, or 0.04 μM OAS1, respectively. Data are representative of at least three independent experiments.

To determine the basis of the loss of 2–5A synthesis by saddleback tamarin OAS1, we first performed pairwise amino acid alignments to identify possible mutations in conserved active-site residues. Remarkably, saddleback tamarin OAS1 has three mutations mapping to the p-loop, a short helical turn motif that coordinates substrate nucleotides during catalysis (Donovan et al., 2013; Lohöfener et al., 2015), in residues otherwise strictly conserved in mammals (S63, S74 and D77) (Figure 4C). Notably, residue 77, one of three catalytic aspartates that coordinate with Mg²⁺ to facilitate catalysis, is replaced with asparagine in saddleback tamarin. In addition, serine 74, which hydrogen bonds with the β-phosphate of the donor ATP, is replaced with a hydrophobic phenylalanine in tamarin. Finally, like the hypomorphic gorilla OAS1 allele, tamarin OAS1 independently acquired a mutation at residue 130 (arginine to histidine), additionally altering its interaction with the acceptor ATP. Thus, saddleback tamarin OAS1 contains multiple debilitating mutations in conserved substrate interacting and catalytic residues critical for 2–5A synthesis.

To chart the evolutionary history of OAS1 loss-of-function in tamarins, we PCR amplified and sequenced OAS1 exons 1–6 from genomic DNA of the related mustached tamarin. Mustached tamarin shares mutations in noncatalytic residues (S63F, S74I, and R130H) but has an intact aspartate triad, similar to a partial sequence available from the Midas tamarin (George et al., 2011) (Figure 4C). We next tested whether mustached tamarin OAS1 retains residual 2–5A synthesis activity compared to saddleback tamarin. Yeast expressing saddleback or mustached tamarin OAS1 and human RNase L, however, display no growth arrest (Figure 4D), indicating a lack of activity by both species. Furthermore, purification (Figure S3A) and visualization of enzymatic activity in vitro revealed no 2–5A production above background from either species (Figure 4E). Finally, we investigated whether tamarin OAS1 might have evolved to preferentially activate tamarin RNase L through another mechanism. When co-expressed with human OAS1 in yeast, tamarin RNase L displays slightly lower activity than human RNase L but is not activated by tamarin OAS1 (Figure S3B). Thus, evolution of OAS1 in the tamarin lineage involved several inactivating mutations in conserved substrate coordinating residues 63, 74, and 130 in a common ancestor, and the subsequent loss of catalytic aspartate 77 in saddleback tamarins.

DISCUSSION

Previous studies on the impact of altering the OAS/RNase L pathway in healthy and infected animals revealed contrasting outcomes consistent with evolutionary compromises. Mice lacking RNase L exhibit greatly increased susceptibility to viral infection (Zhou et al., 1997), supporting a fundamental role for the OAS/RNase L pathway in immunity. In the absence of infection, however, RNase L deficient mice display significantly increased longevity, potentially due to the lack of chronic low level activation of OAS proteins by cellular dsRNAs (Andersen et al., 2007). Indeed, RNA extracted from various cell types can activate OAS1 in vitro (Dan et al., 2012), including specific mRNAs (Molinaro et al., 2006), consistent with detectible 2–5A in uninfected tissues (Hearl and Johnston, 1987). In addition, recent work demonstrates that cellular RNAs activate OAS proteins in the absence of RNA editing by ADAR1, leading to RNase L mediated cell death (Li et al., 2017b), highlighting cellular mechanisms to prevent immune activation by endogenous dsRNA. Lineage-specific increases in the repertoire of cellular RNAs that activate OAS1 could therefore account for the benefit of forfeiting 2–5A synthesis in some species. OAS2 and OAS3, which have distinct dsRNA recognition patterns (Donovan et al., 2015), provide additional means of RNase L activation, potentially alleviating the loss of OAS1 activity during infection. These observations lead us to favor a model where recurrent loss of function mutations in primate OAS1 reflect ongoing balancing of deleterious and antiviral effects of 2–5A synthesis.

An alternative explanation for recurrent loss-of-function mutations in OAS1 is neutral evolution, where loss of OAS1 activity is merely tolerated as it provides no meaningful benefit to host fitness. This scenario seems less likely given several lines of evidence. Previous studies of OAS1 evolution revealed evidence of strong and recurrent positive selection in primates, including humans (Hancks et al., 2015; Mozzi et al., 2015). Such evolutionary patterns in immunity genes often coincide with evasion of pathogen encoded inhibitors, suggesting that OAS1 antagonism is a key step in infection by multiple viruses. Consistent with this idea, OAS1 activity has been implicated for specific immune function against several human viruses. Notably, the low activity human OAS1 allele has been shown to increase susceptibility to West Nile Virus infection (Lim et al., 2009). In addition, overexpression of OAS1 inhibits Dengue virus infection in human cells in a RNase L dependent manner (Lin et al., 2009). Finally, although we recovered several OAS1 loss-of-function alleles in this study, we did not identify any OAS1 pseudogenes interrupted by premature stop codons precluding translation of the core enzyme. Our repeated finding of mutations in critical catalytic residues in otherwise intact proteins supports previously proposed RNase L-independent antiviral functions for OAS1 (Kristiansen et al., 2010b; Li et al., 2016).

Lineage specific changes in regulation of OAS1, either pre- or post-translational, could provide a mechanism to buffer the loss of OAS1 activity we observed in several species. The subsequent evolution of such compensatory mechanisms after loss of OAS1 activity seems unlikely given that selection against the initial debilitating mutation provides the simplest path to restoring OAS function. Compensatory increases RNase L activity provide another mechanism to buffer decreased OAS1 function. We ruled out this scenario as well, showing that gorilla and tamarin RNase L do not have increased activity or are more sensitive to the properties of their species OAS1 in yeast.

Genes encoding immune functions are rich with genetic variation in humans (Lazarus et al., 2002). In some cases, common genetic variants in key mediators of cellular immunity lead to partial or complete loss of protein function and can also be associated with disease. For example, a common human variant of STING, a central component of several innate immunity pathways, results in decreased IFN-b production in response to viral DNAs (Patel et al., 2017) (but also see (Sivick et al., 2017)). The human OAS1 variant associated with lower in vivo 2–5A levels is also associated with increased risk of viral infection (Lim et al., 2009) and development of Sjögrens syndrome (Li et al., 2017a), an autoimmune disease associated with Epstein-Barr and cytomegalovirus infection. Despite these risks, the low activity allele rose to near fixation in ancient African populations, proceeded by reintroduction of the ancestral high activity allele via introgression from Neandertals to non-African populations (Mendez et al., 2013). Taken in isolation, the prevalence of such alleles compromising defense functions is difficult to explain. Our identification of multiple independently evolved loss-of-function OAS1 alleles is consistent with a common underlying evolutionary pressure frequently favoring lowered OAS activity in humans and nonhuman primates alike. Together, these findings illustrate how selection for diminished defense function can be a pervasive force in the evolution of immune defenses.

STAR METHODS

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Nels Elde (nelde@genetics.utah.edu)

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Yeast culturing techniques

Yeast strain w303 was used exclusively in this study. Yeast were cultured at 30C in selective standard minimal medium (BD Difco Yeast Nitrogen Base Cat# DF0919-07-3) with either 2% glucose, galactose, or raffinose.

Mammalian cell culture

All cells were grown at 37C with 5% CO₂. Human fibroblasts (male) isolated from unidentified and discarded foreskin with patient consent and IRB approval (Swedish Medical Center, Seattle, WA) were obtained as a gift from Adam Geballe/Denise Galloway (Fred Hutchinson Cancer Research Center) and 293T cells (female) were grown in High Glucose Dulbecco’s Modified Eagles Medium (GE Healthcare Life Sciences cat# SH30022.01) with10% FBS and 2mM L-glutamine. A549 cells (male) were grown in Ham’s F12K medium (ThermoFisher cat# 21127022) + 10% FBS. Primate fibroblasts (sexes detailed in Key Resource Table) were cultured in a 1:1 mixture of Minimum Essential Medium Alpha Modification (GE Healthcare Life Sciences cat# SH30265.01) High Glucose Dulbecco’s Modified Eagles Medium (GE Healthcare Life Sciences cat# SH30022.01) with 10% FBS and 2mM L-glutamine.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-OAS1	Sigma-Aldrich	Cat#HPA00365 7;RRID:AB 18 54747
Mouse monoclonal anti-HA	Covance	Cat#MMS-101P;RRID:AB 2314672
Mouse monoclonal anti-RNase L	Santa Cruz	Cat#sc-23955; RRID:AB 6282 22
Mouse monoclonal anti-actin	BD biosciences	Cat#612657; RRID:AB 3999 01
Bacterial strains
Escherichia coli	(Kranzusch et al., 2014)	BL21(DE3) RU
Biological Samples
Human fibroblasts, male (Homo sapiens)	Laboratory of Adam Geballe/Denise Galloway	N/A
Chimpanzee fibroblast, male (Pan troglodytes)	Coriell Institute	Cat#PR00226; RRID:CVCL 2 V65
Gorilla fibroblast, male (Gorilla gorilla gorilla)	Coriell Institute	Cat#AG21765; RRID:CVCL L K50
Orangutan fibroblast, male (Pongo pygmaeus pygmaeus)	Coriell Institute	Cat#AG05252; RRID:CVCL 1 G38
Rhesus macaque fibroblast, female (Macacca mulatta)	Coriell Institute	Cat#AG06252; RRID:CVCL 21 33
Olive baboon fibroblast, male (Papio anubis)	Coriell Institute	Cat#PR00036; RRID:CVCL 2 V14
African Green Monkey fibroblast, male (Cercopithecus aethiops)	Coriell Institute	Cat#PR01193; RRID:CVCL 2 X98
Talapoin fibroblast, male (Miopithecus talapoin talapoin)	Coriell Institute	Cat#PR00716; RRID:CVCL 2 X12
Colobus fibroblast, male (Colobus guereza)	Coriell Institute	Cat#PR00240; RRID:CVCL 2 V75
Dusky titi fibroblast, female (Plecturocebus moloch)	Coriell Institute	Cat#PR00742; RRID:CVCL 2 X22
White faced saki fibroblast, male (Pithecia pithecia)	Coriell Institute	Cat#PR00239; RRID:CVCL 2 V74
Saddleback tamarin fibroblast, female (Saguinus fuscicollis)	Coriell Institute	Cat#AG05313; RRID:CVCL 21 28
Mustached tamarin genomic DNA (Saguinus labiatus)	Coriell Institute	Cat#NG05308
Chemicals, Peptides, and Recombinant Proteins
Universal Type 1 Interferon	PBL assay scientific	Cat#11200
Experimental models
Saccharomyces cerevisiae	Laboratory of Tom Dever	W303
A549 human cells	Laboratory of Susan Weiss	N/A
Human fibroblasts	Laboratory of Adam Geballe/Denise Galloway	N/A
Critical Commercial Assays
Superscript III First-Strand Synthesis kit	ThermoFisher Scientific	Cat#18080051
quick-RNA miniprep	Zymo Research	Cat#R1050
YeaStar RNA Kit	Zymo Research	Cat#R1002
Deposited Data
0AS1 coding sequences	This paper	GenBank: MH188449-MH188461]
Recombinant DNA
pBM272 dual GAL yeast expression plasmid	Laboratory of Mark Hochstrasser	N/A
pRS405 integrating yeast expression plasmid	Laboratory of David Stillman	N/A
Oligonucleotides
See data in table S3	Integrated DNA Technologies	N/A
Software and Algorithms
ImageQuant (v5.2)	GE life sciences	https://www.gelife-sciences.com/en/ch/shop/protein-analysis/molecular-imaging-for-proteins/imaging-software

METHOD DETAILS

Cloning of OAS1 and RNase L cDNAs

Total RNA was isolated from primate cell lines (quick-RNA miniprep, Zymo Research cat# R1050) followed by preparation of cDNA (Superscript III First-Strand Synthesis kit, ThermoFIsher scientific cat# 18080051). Unless otherwise noted, cell lines were obtained from Coriell Cell Repositories. OAS1 and RNase L cDNAs were PCR amplified from cDNA using primers indicated in Table S3. Mustached tamarin OAS1 p46 exons 1–6 were PCR amplified and sequenced directly from genomic DNA. The assembled mustached tamarin OAS1 coding sequence was synthesized by Life Technologies. OAS1 and RNase L variants were cloned into the yeast CEN Gal10/Gal1 dual expression plasmid pBM272. OAS1 was cloned via EcoRI/NotI restriction sites, and RNase L through BamHI sites introduced by PCR primers. Human OAS1 p42 and p46 were cloned into vector pcDNA6 via EcoRI/NotI sites for mammalian cell expression. C-terminally hemagglutinin tagged Us11 (Us11-HA) was codon optimized for budding yeast and synthesized by Life Technologies (based on Genbank YP_009137147.1) followed by cloning into integrating yeast expression vector pRS405 via SalI/SpeI cloning sites.

Yeast growth assays

Strains were generated by transformation of EcoRV linearized vector pRS405 containing either an empty site (-Us11) or HA-Us11 (+Us11), resulting in integration at the leu2 locus. Yeast strains were transformed with pBM272 containing OAS1/RNase L variants. Transformants were grown in S -leu -ura liquid media containing 2% glucose overnight and washed before plating. Tenfold serial dilutions were prepared (OD₆₀₀ 3.0, 0.3, 0.03, 0.003) and a multichannel pipette was used to plate 2.3 μl of each dilution on 2% glucose or 2% galactose medium. Unless otherwise noted, glucose plates were imaged after 24 h of growth, galactose plates after 48 h of growth.

RNA integrity analysis

Transformed yeast were grown to log phase in 2% raffinose S -leu -ura growth medium, followed by induction with 2% galactose for 6 h. Total RNA was extracted using a YeaStar RNA kit (Zymo research) and quantified. RNA integrity was assessed using an RNA ScreenTape assay with the Agilent TapeStation instrument.

Western Blotting

Transformed yeast were grown to log phase in 2% raffinose S -leu -ura growth medium, followed by induction with 2% galactose for 6 h. Cell lysates containing equal amounts of yeast were prepared by treatment with 0.1 M NaOH for 5 min, followed by lysis in 2× SDS loading buffer. Mammalian cell lysates were prepared by direct lysis in 2× SDS buffer containing 8 M Urea and 3 M thiourea. Total protein was resolved by Mini-PROTEAN GTX polyacrylamide gel electrophoresis (Bio-rad). Proteins were detected using anti-OAS1 (Sigma cat# HPA003657, 1:1000), anti-RNase L (Santa Cruz cat# sc-23955, 1:200), anti-HA (Covance cat# MMS-101P, 1:1000), and anti-Actin (BD cat# 612657, 1:1000) antibodies. Blots were visualized using film or c-digit chemiluminescent imager (LI-COR).

Mammalian OAS1 alignment

The following OAS1 coding sequences were retrieved from genbank for multialignment: human (Homo sapiens, NM_016816.3), mouse OAS1a (Mus musculus, NM_145211.2), rabbit (Oryctolagus cuniculus, XM_017349609.1), megabat (Pteropus Alecto, NM_001290162.1), cow (Bos Taurus, NM_001040606.1), dog (Canis lupus familiaris, NM_001048131.1), pig (Sus scrofa, NM_214303.2), dolphin (Lipotes vexillifer, XM_007468577.1), elephant (Loxodonta africana, XM_010598964.2), star-nosed mole (Condylura cristata, XM_012730965.1), opossum (Monodelphis domestica, XM_001378771.3). Translated sequences were aligned with tamarin or gorilla OAS1 sequences using clustalw.

Generation of Chimeric and mutated OAS1 proteins

Human-gorilla OAS1 chimeras were generated using PCR stitching. Overlapping PCR products were generated using primers listed in table S3 and gel purified. 50ng of each PCR product was then added to a new PCR reaction and allowed to anneal and extend for 12 cycles before addition of primers to generate full length chimeric OAS1 coding sequences. Plasmids encoding human OAS1^R130C, gorilla OAS1^C130R, and RNase L^H672A were generated by site directed mutagenesis.

Human cell culture and transfection

5 × 10⁵ 293T cells in 6-well plates were transfected with 3 μg of expression vector pcDNA6 (Invitrogen) encoding human OAS1 p42 or p46 isoforms using FuGENE HD (Promega cat# E2311) transfection reagent according to the manufacturer’s specifications. Cells were lysed 48 h post transfection in 2× SDS loading buffer and mixed in a 1:1 ratio as a positive control for western blot analysis of primary human cell lines. A549 and Human Fibroblasts were grown to confluence in 6 well plates before collection for western blot analysis. Interferon stimulated cells were treated with 100U/mL universal type I interferon (PBL assay scientific) for 16 hours before collection.

Primate OAS1 Isoform Purification

Human and other primate OAS1 isoforms were PCR amplified and cloned into a custom pET vector optimized for expression of an N-terminally tagged 6×His-MBP fusion protein, and proteins were expressed and purified from BL21(DE3) RIL E. coli as previously described (Kranzusch et al., 2014). Briefly, competent cells already containing the pRARE2 tRNA plasmid (Agilent) were transformed with OAS expression plasmid and grown in 2 × YT media at 37°C to an OD₆₀₀ of ~0.6 before cooling on ice for 15 min, induced with 0.5 mM IPTG and incubated with shaking at 16°C for ~18 h. Cell pellets were re-suspended in lysis buffer (20 mM HEPES-KOH pH 7.5, 400 mM NaCl, 10% glycerol, 30 mM Imidazole, 1 mM DTT), lysed by sonication, and protein was isolated from clarified lysate using Ni-NTA resin (Qiagen) and gravity-flow chromatography. Resin was washed with lysis buffer supplemented to 1 M NaCl to remove bound nucleic acid, and MBP–OAS protein was eluted with lysis buffer supplemented to 300 mM Imidazole. MBP–OAS proteins were diluted to 50 mM Imidazole and 5% glycerol and concentrated to ~30 mg ml⁻¹ prior to overnight digestion with Tobacco Etch Virus protease at 4°C. Digested protein was separated from MBP and protease using a 5 ml Ni-NTA column (Qiagen) and subsequently purified by Superdex 75 16/60 size-exclusion chromatography in gel-filtration buffer (20 mM HEPES-KOH pH 7.5, 250 mM KCl, 10% glycerol, 1 mM TCEP-KOH). Purified OAS was concentrated to ~10 mg ml⁻¹, flash-frozen in liquid nitrogen and stored at −80°C for biochemical experiments. MBP-tagged human OAS1 splice isoforms and MBP-tagged mustached tamarin and saddleback tamarin OAS homologs were purified using identical conditions, except proteins were dialyzed into storage buffer (20 mM HEPES-KOH pH 7.5, 250 mM KCl, 10% glycerol, 1 mM TCEP-KOH) overnight at 4°C immediately following Ni-NTA elution and directly stored without tag removal or gel-filtration purification.

OAS Enzymatic Assay

Purified recombinant OAS enzyme was incubated in reaction buffer (10 mM Tris-HCl pH 7.5, 25 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 0.1 mg ml⁻¹ BSA) supplemented with 200 μM ATP and [α−³²P] ATP (~2 μCi). 10 μl reactions included final OAS enzyme concentrations as indicated, or enzyme titration with final concentrations of 0.04, 0.2, 1, and 5 μM, and included ~5 μM poly I:C (Invivogen) for dsRNA stimulation. Reactions were incubated at 37°C for 2 h and terminated by adding an equal volume of stop solution (95% deionized formamide, 20 mM EDTA) and incubation at 95°C for 2 min. Reactions were analyzed by denaturing gel electrophoresis on a 32-cm tall 20% polyacrylamide 7 M urea gel with 0.5 × TBE running buffer, and 2′–5′ oligoadenylate products were visualized using a Typhoon phosphoimager (GE Healthcare).

QUANTIFICATION AND STATISTICAL ANALYSIS

OAS1 enzymatic activity was quantified after phosphorimaging using ImageQuant (v5.2). Briefly, total enzyme activity was measured by summing the total pixel intensity in each corresponding lane, and normalized to the value obtained for a WT hOAS1 (5 μM) control included in each experiment. For human OAS1 isoform experiments the hOAS1 isoform p46 was treated as WT for normalization. Data are representative of independent replicates. Experimental replication ‘n’-values are indicated in each Figure legend.

DATA AND SOFTWARE AVAILABILITY

OAS1 and RNase L coding cDNA sequences used in the study are available on genbank with accession numbers MH188449-MH188461.