A proline deletion in IFNAR1 impairs IFN-signaling and underlies increased resistance to tuberculosis in humans

Type I interferons (IFN), best known for their anti-viral functions, have been shown to impair host resistance to intracellular bacteria in mice. However, the precise role of type I IFN signaling in bacterial infection in humans is unclear. Here, we show that genetic variation in the human IFNAR1 gene is associated with decreased susceptibility to tuberculosis and an increased risk of viral hepatitis in Chinese populations. Receptor mutagenesis and cell signaling studies establish that the IFNAR1 mutation corresponding to a proline deletion in the hinge region of the membrane-proximal domain of IFNAR1 decreases the binding affinity of IFNAR1 to IFN-β, impeding type I IFN signaling. Our findings suggest that IFNAR1 signaling underlies an increased risk of tuberculosis in humans and reveals a function for the IFNAR1 inter-domain region in cytokine–cytokine receptor interaction and signal transduction.

T he type I interferon (IFN) family consists of~20 different members including multiple IFN-α subtypes and IFN-β. The signaling complex utilized by type I IFNs comprise IFN alpha and beta receptor subunit 1 and 2 (IFNAR1 and IFNAR2) 1 . Ligand engagement of the extracellular domains of the receptor complex induces the phosphorylation of signal transducer and activator of transcription (STAT) molecules, which subsequently activate a set of Interferon Stimulated Genes (ISG). Structurally, IFNAR1 is unique amongst the class II helical cytokine receptors as its extracellular domain is comprised of a four-domain architecture, referred to as subdomains (SD) 2 . Although the minimal ligand-binding region is localized to the membrane-distal portion (SD1-SD3) 3 , the membrane-proximal domain (SD4) is required to undergo a conformational change necessary for signal transduction across the cell membrane 4 . Similar to many class II helical cytokine receptors, IFNAR1 has paired proline residues in hinge regions between individual extracellular domains. Paired proline residues located in such non-helical linker regions are hypothesized to increase local structural rigidity 5 .
Type I IFNs are established as important in anti-viral immunity 6 ; however, these cytokines also impair immune responses to various pathogens 7,8 . Mice deficient in IFNAR1 (Ifnar1 −/− ) have reduced pathogen loads in response to intracellular bacterial infection 7,[9][10][11][12][13][14] , and administration of type I IFN-inducing poly: (IC) exacerbates listeriosis and tuberculosis (TB) in wild-type (WT), but not Ifnar1 −/− mice 11,15 . In contrast to animal studies, evidence that IFNAR1 signaling contributes to TB disease susceptibility in humans is unclear, although several clinical studies have linked the blood interferon signature to detrimental clinical outcome [16][17][18] . Here we show that a rare IFNAR1 genotype is associated with decreased risk of TB in Chinese populations and the mutation decreases the magnitude of IFN-β-mediated ISG induction in cells from individuals heterozygous for the SNP. We further demonstrate that the SNP decreases type I IFN signaling by reducing the overall binding affinity of IFN-β for IFNAR1. These findings suggest a host-detrimental function for type I IFN signaling in human TB. As we show that the mutation, which occurs at a non-cytokine binding region of IFNAR1, weakens the cytokine receptor interaction, our findings also suggest a function for the membrane-proximal domain in type I IFN binding and signaling.

Results
ISG induction is associated with active TB in Chinese population. As an initial step in understanding the function of type I IFNs in human TB, we examined the expression of ISGs in the leukocytes of healthy controls (HC), latent TB infection (LTBI) subjects and TB patients. LTBI subjects were positive for Interferon-Gamma Release Assay (IGRA) testing, but with no clinical symptoms of active TB and normal chest radiography. In agreement with previous studies performed in African populations 19,20 , we observed that PBMCs of active TB patients expressed significantly higher levels of ISGs than LTBI subjects in Chinese populations (Fig. 1a).
In addition, blood samples were also collected from a group of active pulmonary TB patients before and after the initiation of the standard 2 months of isoniazid (H), rifampicin (R), pyrazinamide (Z), and ethambutol (E) during intensive phase and 4 months of HR in continuation phase (2HRZE/4HR) treatment regimen recommended by the WHO guideline 21 . We observed that the elevated ISG expression declined rapidly after initiation of anti-TB drug treatment (Fig. 1b). In addition, pleural fluid mononuclear cells (PFMCs) consistently expressed higher levels of ISGs compared to paired PBMC samples from the same pleural TB patients (Fig. 1c). Altogether, these findings confirm that active mycobacterial infection/disease is associated with elevated IFN signaling.
SNP rs72552343 in IFNAR1 reduces risk of developing active TB. To understand the function of type I IFN signaling in TB susceptibility in humans, we determined the association of IFNAR1 genetic polymorphisms and the risk of developing active TB disease in Chinese populations. Human IFNAR1 is located on chromosome 21 adjacent to IL10RB and IFNGR2 ( Supplementary  Fig. 1a)  Each symbol/paired symbols/line represents an individual. With the exception of the study described in b, all blood samples were collected before the initiation of anti-TB chemotherapy. Statistical significance was determined using one-way analysis of variance (ANOVA) for multi-group comparison (in a and b), and Student's t-test for two-group comparison (in c). *p < 0.05 and **p < 0.01. The data shown are fold change in ISG expression relative to the house-keeping gene GAPDH individuals carrying the SNP (rs72552343) with the deletion of nucleotides TCC in IFNAR1 (TCC/Del), showed significantly decreased risk of developing active TB (multiplicative model: p = 0.0002; dominant model: p = 0.0003) ( Table 1). Further sequencing analysis confirmed that SNP rs72552343 is associated with the in-frame deletion of nucleotides TCC ( Supplementary  Fig. 1b).
To validate this genetic association, the distribution of the SNP rs72552343 genotype was further determined in a second TB cohort, which was established independently in Chengdu city in Sichuan province located in south-western China 23 . This validation cohort consists of 832 active TB cases (367 pulmonary TB and 465 extra-pulmonary TB) and 1084 controls. The significant association of IFNAR1 SNP rs72552343 (TCC/Del) and reduced risk of TB was replicated in both this validation cohort (p = 0.004, multiplicative and dominant models) and the combined discovery and validation cohorts (2365 cases and 2529 controls, p < 0.0001, multiplicative and dominant models, Table 1). SNP rs72552343 in IFNAR1 increases hepatitis B susceptibility. Since type I IFNs are known to mediate anti-viral immunity, we next examined whether SNP rs72552343 influenced disease pathogenesis in viral infection in humans. To this end, the rs72552343 genotype distribution was determined in hepatitis B virus surface antigen (HBsAg)-positive hepatitis B (HepB) patients and HC recruited for a previously published study 24 . In contrast to mycobacterial infection, IFNAR1 rs72552343 TCC/ Del genotype was observed to be associated with significantly increased susceptibility to viral hepatitis (multiplicative model: p = 0.01; dominant model: p = 0.02, Table 2). Therefore, IFNAR1signaling can differentially regulate resistance to intracellular bacterial and viral infection in humans.
TCC deletion in IFNAR1 is associated with reduced TB pathology. We next analyzed the association of rs72552343 with the clinical manifestations of TB diseases, sputum smear and mycobacterial culture positivity and lung cavity formation in the combined TB cohort. We found that while the rs72552343 TCC deletion was associated significantly with decreased risk of both pulmonary (PTB vs. HC, OR: 0.47; p < 0.0001; dominant model) and extra-pulmonary TB disease (ETB vs. HC, OR: 0.40; p = 0.005; dominant model), it did not preferentially influence the development of the either type of the disease (Supplementary  Table 3). Similarly, the host protective effect of TCC deletion on pulmonary TB was present in TB patients who were mycobacterial culture and sputum smear positive or negative (Supplementary Tables 4 and 5). In addition, analysis of highresolution computed tomography (HRCT) findings revealed that pulmonary TB patients carrying rs72552343 TCC/Del exhibited reduced risk of developing pulmonary cavities compared to their TCC/TCC counterparts (cavity− vs. cavity+, p = 0.02; multiplicative and dominant models, Supplementary Table 6).
To investigate further the association between SNP rs72552343 and pulmonary pathology in TB patients, HRCT images of pulmonary TB patients enrolled in Shenzhen study were further assessed and scored based on the patterns, profusion and distribution of pulmonary abnormalities including cavity formation 25 . This analysis revealed that pulmonary TB patients carrying TCC/Del genotype showed significantly lower HRCT scores (less severe TB disease) than those with the common allele (Fig. 2a). When radiographic data collected before and 2 years after treatment were compared in a subset of the TB patients, we found that IFNAR1 TCC deletion was also associated with a more favorable disease outcome following a standard 2HRZE/4HR treatment regimen (Fig. 2b). Altogether, these data establish an association between type I IFN receptor signaling and increased tissue pathology in active pulmonary TB.  TCC deletion in IFNAR1 impairs type I IFN signaling in PBMCs. To determine whether IFNAR1 SNP rs72552343 plays a role in the cellular response to IFN stimulation, PBMCs from individuals carrying WT (TCC/TCC) or mutant (TCC/Del) IFNAR1 genotypes were stimulated with recombinant human IFN-β, IFN-γ, or both cytokines, and ISG expression analyzed using qRT-PCR. We observed that MX1, IFIT1, OAS1, STAT2, and STAT1, which were highly sensitive to IFN-β induction, were up-regulated to a lesser extent in cells carrying the TCC deletion than in their WT counterparts following IFN-β stimulation (Fig. 3). In contrast, the upregulation of CXCL10 and IRF1 driven predominately by IFN-γ at this time-point was not affected by the IFNAR1 mutation. In addition, there was no significant difference in the level of IFNAR1 or IFNAR2 between genotypes. These findings collectively suggested TCC deletion is associated with defective type I, but not type II, IFN signaling. As IFNAR1 signaling can regulate type I IFN production 26,27 , we examined whether the IFNAR1 mutation has an impact on the production of type I IFNs. We first compared the levels of IFN-α and IFN-β in the plasma of HC and TB patients carrying TCC/ TCC or TCC/Del genotypes. Although IFN-α was not measureable in plasma, a low level of circulating IFN-β was detected in all TB cases irrespective of their genotype ( Supplementary Fig. 2a). Similarly, a comparable induction of IFN-α and IFN-β was observed in the culture supernatants of M. tuberculosis-infected macrophages carrying TCC/TCC or TCC/Del IFNAR1 (Supplementary Fig. 2b), suggesting that the TCC/Del in IFNAR1 does not appear to impact type I IFN production in TB patients and M. tuberculosis-infected macrophages.
TCC deletion in IFNAR1 decreases type I IFN signal transduction. To establish definitively that the identified IFNAR1 deletion affects type I IFN signaling, mammalian expression vectors encoding human WT IFNAR1 and IFNAR1Del were generated for transfection studies. Together with an IFNAR2 encoding vector and an ISRE-driven luciferase reporter plasmid, the WT or mutant IFNAR1 vectors were transfected into Ifnar1 −/− mouse embryonic fibroblasts (MEFs). Ifnar1 −/− MEFs were used to ensure that the activation of the IFN signaling pathway by human IFNs depends exclusively on the transfected IFNAR1. Because both IFN-β and IFN-α can signal via IFNAR1 1 , we also included IFN-α in this series of experiments to determine whether the signaling defect resulting from the TCC deletion is restricted to IFN-β stimulation. Upon stimulation with human IFN-α or IFN-β, WT IFNAR1-transfected cells exhibited significantly stronger luciferase activity compared to their IFNAR1Del-transfected counterparts, whereas IFN-γ induced minimal luciferase activity, similar to levels seen in the untransfected cells (Fig. 4a). Similarly, WT IFNAR1-transfected MEFs expressed significantly higher levels of IFN-inducible Isg15, Ifit1, and Oas1 than the IFNAR1Del-transfected cells when measured at 3 and 6 hr after IFN stimulation (Fig. 4b). The reduced IFNinducible response in IFNAR1Del-transfected MEFs correlated with similarly impaired STAT1 phosphorylation following IFN-β stimulation ( Fig. 4c and d).
To confirm the above findings in a human system, we knocked down endogenous IFNAR1 in HEK293 cells using CRISPR/Cas9 gene-editing technology and then transfected the cells with human WT IFNAR1 or IFNAR1Del plasmids together with an ISRE-driven luciferase reporter plasmid. Similar to the studies  performed using MEFs, following type I IFN stimulation, HEK293 cells transfected with WT IFNAR1 exhibited significantly stronger luciferase activity compared to IFNAR1Deltransfected counterparts (Fig. 5a). WT IFNAR1-transfected cells also expressed significantly higher levels of IFN-inducible ISG15, IFIT1, and OAS1 compared to the IFNAR1Del-transfected counterparts when measured at 6 h after IFN stimulation (Fig. 5b).
Altogether, these biochemical studies performed using both murine and human cells provide direct evidence demonstrating that the TCC deletion and the resulting deletion of proline 335 (Pro335) impairs IFNAR1-dependent type I IFN signaling.
IFNAR1 TCC deletion decreases binding affinity of IFN-β to IFNAR1. The TCC deletion in IFNAR1 results in the subsequent deletion of one of the paired proline residues (Pro335) located in the inter-domain hinge between SD3 and SD4 of human IFNAR1 (Fig. 6a). Evidence from the literature suggests that deletion of Pro335 from IFNAR1 would increase flexibility of the hinge domain between SD3 and SD4 of IFNAR1, thereby altering ligand-binding kinetics 5,28 . To test this hypothesis, we generated recombinant forms of the hIFNAR1-extracellular domain (ECD) and also a form with the Pro335 deletion (IFNAR1-ECD-Del) in an insect cell expression system, and used these to assess the effect of Pro335 deletion on hIFN-β affinity for IFNAR1 using surface plasmon resonance (SPR). We observed in three independent experiments that deletion of Pro335 from IFNAR1 significantly reduced the overall affinity of the ligand-receptor interaction (WT IFNAR1 affinity was determined to be 15.8 ± 3.4 nM (mean ± standard deviation, n = 3 independent experiments) compared to IFNAR1Del affinity at 28.1 ± 3.1 nM), whereas the rate of association and dissociation of hIFN-β to and from the receptor were not significantly different for both WT IFNAR1 and IFNAR1Del (Fig. 6b, c).

Discussion
Animal studies have suggested that type I IFNs can impair resistance to a diverse range of pathogens. In particular, immunity against intracellular bacteria is often compromised by type I IFNs 7,13 . However, whether type I IFN-dependent regulatory mechanisms dictate the outcomes of bacterial infection in humans are unknown. In this study, we have combined genetic and biochemical approaches to elucidate type I IFN function in human TB. We identify that a genetic variant in the human IFNAR1 gene (the deletion of nucleotides TCC) decreases type I IFN signaling and reduces the risk of active TB. Moreover, pulmonary TB patients carrying the TCC/Del genotype exhibit less severe tissue pathology in their lungs than their TCC/TCC counterparts. Finally, the same IFNAR1 mutation is associated with increased susceptibility to viral hepatitis, revealing a In both TB and leprosy, an IFN-inducible transcriptional signature presented in periphery leukocytes is associated strongly with detrimental clinical outcomes [16][17][18] . Although the observation is indicative of an involvement of IFN signaling in mycobacterial diseases, it remains unclear whether the IFN signature is driven by type I, type II, or both types of IFNs, and whether the activation of an IFNAR1 signaling pathway exacerbates TB in humans. Addressing these questions in human studies has been challenging. In contrast to animal studies, where the function of any given gene is commonly established by analyzing genetically modified animals, contribution of individual genes to host resistance in a clinical setting is commonly defined by linking genetic variations to disease phenotype. Focusing on the role of IFNAR1 in human TB, we first identified IFNAR1 SNP rs72552343 is associated with increased resistance to TB. Further analysis using mutagenesis, gene-editing, and transfection experiments allowed us to conclude that the rs72552343 associated TCC deletion decreased the cellular response to type I IFN stimulation thereby linking the disease-associating genotype to cytokine receptor signaling. These findings collectively provide evidence for a hostdetrimental function for type I IFNs in human mycobacterial disease.
Although identified on the basis of their anti-viral activity, type I IFNs are now known to play a broader role in host response to infection than originally anticipated. Our genetic association studies indicate that type I IFNs can have a different role in viral and bacterial infections in humans. Similar to mouse studies, the question remains as to how the same cytokine receptor signaling pathway leads to distinct disease outcomes in humans. As IFNs are pleiotropic cytokines, the downstream outcome of type I IFN signaling will depend on the type and location of the cells activated by the cytokines. It is also possible that the major function of type I IFNs in infection in vivo is directed at the host rather than the pathogen. Interestingly, our data suggest that type I IFN signaling is associated with more severe immunopathology in pulmonary TB. Discovery of additional genetic associations of IFNAR1 signaling components and susceptibility of TB and other infectious diseases in future studies will assist in validating the observations reported here and in dissecting the complex function of IFNs in humans. It is also worth noting that not all identified IFNAR1 SNPs associated with viral susceptibility were associated with increased resistance to mycobacterial infection. IFNAR1 SNP rs2843710, shown previously to exhibit increased susceptibility to enterovirus infection 29 , did not display a significant association with active TB in the current study (Supplementary Table 1).
The extracellular domains of class II helical cytokine receptors, of which IFNAR1 is a member, are composed of discrete subdomains separated by distinctive poly-proline, inter-domain hinges 30 . The presence of such poly-proline hinges between functional receptor subdomains leads to structural rigidity of the protein thereby dictating the conformational state of the receptor 5 . Factors that perturb protein rigidity, including mutation or substitution of proline residues in inter-domain hinge regions can increase protein flexibility thereby decreasing ligand-binding affinity and altering protein conformational states and protein function [31][32][33] . Although the minimal ligand-binding domain of IFNAR1 is limited to its three membrane-distal subdomains (SD) 1-3, the membrane-proximal subdomain (SD4) of IFNAR1 is required to undergo a conformational change for optimal IFNinduced signal transduction across the cell membrane 4 . Given this evidence from the literature, we hypothesized that deletion of Pro335 from IFNAR1-ECD would reduce the rigidity of the interdomain hinge between IFNAR1 SD3 and SD4, thereby increasing inter-domain flexibility and altering the IFN-β binding kinetics. The fact that IFNAR1-ECD-Del, compared to its WT counterpart, shows reduced affinity for IFN-β strongly suggests that deletion of Pro335 from IFNAR1-ECD has indeed increased flexibility of the receptor. Taken together, our results suggest that SNP rs72552343 has reduced IFNAR1 functionality by altering IFN binding kinetics and thereby modifying receptor response upon ligand engagement, reducing the magnitude of STAT1 activation and subsequent ISG induction measurable from the mutated receptor.
It remains to be determined how the proline deletion attenuates IFNAR signaling in the heterozygous individuals. There is, however, a precedent in the literature for the heterozygous expression of a SNP that results in a missense mutation in human IFNAR1 to alter the function of the receptor. SNP rs2257167, which causes substitution of Leucine 168 to Valine in IFNAR1, has been the focus of a number of studies [34][35][36][37][38] . For example, in a severe malaria susceptibility study, the presence of SNP rs2257167 in the heterozygous state was associated with increased susceptibility to this disease 34 . There are other examples where the heterozygous state of SNPs in cell surface receptors, such as in IFNGR1, a helical cytokine receptor related to IFNAR1, is also associated with disease outcome 39 .
As only virulent mycobacteria induce type I IFN production in infected mice and human macrophages [40][41][42] , it is hypothesized that type I IFN signaling is associated with mycobacterial virulence and increased host susceptibility. Although the exact mechanisms by which type I IFNs exacerbate TB are not fully understood, IFNAR1 deficiency or therapeutic inhibition of type I IFN production have been shown to be host beneficial in resistance to M. tuberculosis infection in mice 43 . Our findings here demonstrate that reduced type I IFN signaling resulting from a naturally occurring mutation in IFNAR1 decreases the risk of developing active TB in humans. Together, these experimental and clinical studies highlight a pivotal role for type I IFN signaling in determining the outcome of mycobacterial infection. Identification of the molecular mechanisms mediating the antiviral and pro-bacterial functions of type I IFNs may lead to a better understanding of basic IFN biology, pathogenesis of human infectious diseases, and assist in developing novel therapeutics.

Methods
Study populations. We analyzed samples from three case-control cohorts 23,24,44 . All subjects were genetically unrelated Southern Han Chinese adults. Active TB was defined based on the WHO guideline for the diagnosis of non-HIV related TB with the following criteria: clinical signs and symptoms, chest radiography, acid-fast bacilli (AFB) identification (sputum smear or M. tuberculosis culture positive), and response to anti-TB chemotherapy 21 . Patients with allergic diseases, diabetes, cancer, immune-compromised conditions, and HIV infection were excluded. The discovery TB cohort includes 1533 active TB patients and 1445 HC recruited at Shenzhen Third People's Hospital, Shenzhen, Guangdong Province 44 . Among the 1533 active TB patients, 1432 were diagnosed with pulmonary TB, 101 with extra-pulmonary TB including tuberculous lymphadenitis (n = 62), tuberculous meningitis (n = 13), and osteoarticular TB (n = 26). The healthy controls were asymptomatic individuals with negative T cell reactivity to M. tuberculosis-specific antigens tested in an in-house Interferon-Gamma Release Assay (IGRA) 45 , and normal chest radiography. For the validation TB study cohort, 832 active TB cases (367 pulmonary TB and 465 extra-pulmonary TB) and 1084 controls were recruited at West China Hospital, Sichuan University in Chengdu, Sichuan province 23 . Individuals with Tibetan background have been excluded from the study.
Hepatitis B (HepB) study includes 848 patients and 894 controls 24  SNP selection and genotyping. Genomic DNA was prepared from PBMC using QIAamp DNA Blood Minikit (Qiagen, Hilden, Germany). To select SNP candidates in IFNAR1 gene, we used a Position Weight Matrices (PWM)_SCAN algorithm to scan IFNAR1 sequence in Jaspar, UniPROBE, TRANSFAC and PITA databases. For the SNP within the putative binding sites, Sr value was calculated to reflect the change of binding scores between the putative binding site and putative binding protein, and then converted into a p-value based on the FastPval program, the SNP with Sr p-value < 0.01 was selected for genotyping 47 , in addition, nonsynonymous substitutions in the exon region were also selected. In TB discovery study, three SNPs in the promoter (rs2834191 T>G, rs2843710 C>G, rs17875752 G>T), 1 SNP in exon (rs72552343 TCC>DEL) and 2 SNPs in the intron (rs1012334 A>T, rs1041868 G>A) of IFNAR1 (Supplementary Table 1) were genotyped in health controls and TB patients using the MassARRAY system (Sequenom, San Diego, CA). The relative height (intensity) of the peaks and the signal-to-noise (SNR) ratio were analyzed using Caller software to call genotypes in real-time. After cluster analysis using Typer software, manual curation of spectra was performed to further validate the outcome. In TB validation and HBV cohort studies, the SNP rs72552343 was genotyped in control and case samples using TaqMan assays (ABI, Carlsbad, CA) by CT Bioscience (Suzhou, China).
HRCT and radiological scoring. HRCT were performed at 10 mm section interval (120 kV, 50-450 mAs, 1 mm slice thickness, 1.5 s scanning time) with a window level between 2550 and 40 Hounsfield Units (HU) and window width between 300 and 1600 HU using the Toshiba Aquilion 64 CT Scanner (Toshiba, Tokyo, Japan). HRCT scans were examined by two independent chest radiologists and final conclusions on the findings were reached by consensus. Radio-pathological changes were quantified using a scoring system developed by Ors et al. with the following parameters: (1) micronodule; (2) nodule; (3) consolidation; (4) ground glass opacity; (5) cavity; (6) bronchial lesion; (7) parenchymal bands 25 . The total score for two lungs is 168.
qRT-PCR. RNA was extracted with Trisure according to the manufacturer's instructions (Bioline, Australia). Total RNA (2 μg) was reverse transcribed using the Tetro cDNA synthesis Kit with random primers according to the manufacturer's instructions (Bioline). All qRT-PCR were performed using SYBR NoROX master mix (Bioline) on a Roche LightCycler480. Forward and reverse qRT-PCR primers are listed in Supplementary Table 7. Data were calculated by the 2 −ΔΔCT method using 18S or GAPDH as the house-keeping genes.
Recombinant protein expression and purification. For expression in insect cells, the clones of hIFNAR1-ECD and hIFNAR1-ECD-Del in modified pFastBac-1 (above) were used to generate recombinant bacmids and baculoviral stocks using transfection of Spodoptera frugiperda (Sf)-9 incest cells following the protocols in the Bac-to-Bac Expression System manual (LifeTechnologies). Recombinant forms of hIFNAR1-ECD and hIFNAR1-ECD-Del were expressed as soluble proteins and secreted into the culture supernatant of High Five™ (BTI-TN-5B1-4 from Trichoplusia ni purchased from ThermoFisher Scientific, Cat. No. B85502). Recombinant proteins were purified by affinity and size exclusion chromatography following protocols established in our laboratory for protein purification from insect cell culture 52 . All purified proteins were assessed for purity by reducing SDS-PAGE and protein concentration (mg/ml) measured by absorbance at 280 nM/ extinction co-efficient (per M/cm) (using a molecular weight of 49,000 for hIF-NAR1-ECD).
Surface plasmon resonance. SPR was performed to measure and compare the affinity of ReBif (hIFN-β) to recombinant forms of hIFNAR1-ECD and hIFNAR1-ECD-Del. All experiments were carried out in the absence of IFNAR2. All SPR experiments were performed on a ProteOn XPR36 (Bio-rad Labs) using a HTG chip for His-tagged proteins and TBS as the running buffer. Both ligands (hIF-NAR1-ECD and hIFNAR1-ECD-Del) were immobilized to the nickel activated chip via their C-terminal His-tags after dilution to 25 µg/ml in TBS. ReBif, (hIFNβ) previously dialyzed into TBS, was also diluted in TBS to various concentrations. All data were referenced according to the manufacturer's instructions (Bio-rad) and analyzed using the Langmuir binding model. Data were considered for inclusion in the analysis only if the Chi 2 value (the measure of error between measured and fitted values) was less than 10% of the R max as per the manufacturer's instructions (Bio-rad). K a (1/Ms), K d (1/s), and K D (nM) were calculated by the ProteOn Manager software (Bio-rad) and are represented as mean ± S.D. from at least three independent experiments.
Statistical analysis. For genetic association studies, the Hardy-Weinberg Equilibrium for IFNAR1 SNP distribution was analyzed in patients and healthy controls. The Pearson χ 2 test was used to compare the genotypic and allelic frequencies of SNPs between cases and controls. Unconditional logistic regression adjusted by gender and age was performed to calculate the odd ratios (ORs), 95% confidence intervals (CIs), and corresponding p values using four alternative models (multiplicative, additive, dominant and recessive). For statistical analysis for gene expression and biochemical analysis, one-way analysis of variance (ANOVA) with Newman-Keuls multiple comparison test was used to compare the differences among multiple groups. Student's t-test was used to compare the difference between two experimental groups. All statistical tests were performed using Prism 6.0 (GraphPad). p < 0.05 is considered statistically significant.