|Home | About | Journals | Submit | Contact Us | Français|
Interferon α (IFN-α) levels are elevated in many patients with systemic lupus erythematosus (SLE); however it is not known whether high serum IFN-α activity is a cause or a result of the disease. We studied 266 SLE patients and 405 of their healthy relatives, and frequently found high serum IFN-α activity in both patients and healthy relatives as compared to healthy unrelated individuals. High IFN-α activity was clustered in specific families in both SLE patients and their healthy first-degree relatives, suggesting a heritable trait. Heritability was also supported by quantitative familial correlation of IFN-α activity, concordance in affected sib pairs and frequent transmission of the high IFN-α activity trait from parents to offspring. Autoantibodies to RNA-binding proteins and double-stranded DNA were associated with high IFN-α activity in SLE patients; however these autoantibodies were very uncommon in healthy family members and did not explain the observed familial correlations. The frequency of high IFN-α activity was similar across all studied ethnic backgrounds. These data suggest that high serum IFN-α activity is a complex heritable trait, which plays a primary role in SLE pathogenesis.
Systemic lupus erythematosus (SLE) is a prototypic systemic autoimmune disease characterized by rash, arthritis, nephritis, serosal inflammation, cytopenias and autoantibodies directed at nuclear antigens. The pathogenesis of SLE is multifactorial and likely driven by a complex combination of genetic risk factors and environmental influences, which lead to an irreversible break in immunologic self-tolerance. Interferon α (IFN-α) is a pleiotropic type I IFN with the potential to break tolerance to self by activating antigen-presenting cells after uptake of self material.1 IFN-α serum levels are elevated in SLE patients, and correlate with disease activity.2,3 Additionally, a number of patients treated with recombinant human IFN-α for malignancy and chronic viral hepatitis have developed de novo SLE, which typically resolves after the IFN-α is discontinued.4,5 Our group and others have shown that SLE patients exhibit coordinate overexpression of IFN-α-induced genes in their peripheral blood mononuclear cells (PBMCs) as compared with healthy individuals and patients with other inflammatory diseases such as rheumatoid arthritis.6–8 Increased IFN-α-induced gene expression correlates with greater disease activity and particular autoantibody profiles and disease phenotypes in SLE patients.3
It is not known whether increased IFN-α activity is a cause or a result of SLE. SLE family members are at higher risk of developing not only SLE, but also other autoimmune diseases such as antibody-mediated cytopenias, autoimmune thyroid disease and others.9,10 A similar spectrum of non-SLE autoimmune disease has been induced by exogenous IFN-α administration, including autoantibodies, autoimmune thyroid disease, autoimmune cytopenias and others.11 A heritable predisposition to increased activation of the IFN-α pathway in SLE families could explain some of the burden of both SLE and non-SLE autoimmunity in this population. Possible genetic variability in endogenous IFN-α signaling has been suggested by the association of single nucleotide polymorphisms (SNPs) in the IFN-α pathway genes IRF5 and TYK2 with SLE,12,13 although the impact of these polymorphisms on IFN-α activity in vivo is not known.
We have characterized serum IFN-α activity using a sensitive and reproducible reporter cell assay in both affected and unaffected individuals in two large SLE family registries. We have discovered a frequent tendency to high serum IFN-α activity in SLE patients as well as their healthy first-degree relatives as compared with unrelated controls. These data implicate high serum IFN-α activity as a heritable SLE risk factor. Autoantibodies against double-stranded DNA (dsDNA) and RNA-binding proteins (RBP) are strongly associated with high IFN-α activity in SLE patients, however these autoantibodies do not account for the observed familial IFN-α activity patterns. The tendency toward high IFN-α activity is frequently transmitted from parents to offspring independent of SLE disease, while autoantibodies against dsDNA and RBP are found predominantly in the SLE-affected individuals. IFN-α activity is similar in SLE families of all ethnic backgrounds and appears to play a primary role in SLE pathogenesis.
To measure serum IFN-α activity, we used a sensitive and reproducible functional bioassay which has been developed and validated in our lab.14 In this assay, reporter cells are used to measure the ability of patient sera to cause IFN-induced gene expression. We use this functional assay because enzyme-linked immunosorbent assay (ELISA) methods for detection of IFN-α in human serum have been complicated by low reproducibility and poor correlation with functional assays in our hands and in other studies,15 possibly due to detection of a similar epitope on a non-IFN-α protein, or a stable but biologically inactive IFN-α breakdown product.15 In the reporter assay, cells are exposed to patient or family member sera, and then IFN-induced gene expression is measured in the reporter cells for three representative genes known to be specifically induced by type I IFN. IFN-induced gene expression from the patient and family member samples is then compared to that induced by healthy unrelated donor sera in the same assay. Serum type I IFN activity is almost completely blocked by preincubation with anti-IFN-α antibody, so while the assay is capable of measuring global type I IFN activity, it appears that IFN-α is the major type I IFN that is overexpressed in the SLE patient and healthy family member samples. Serum IFN-α activity data are presented quantitatively as the number of standard deviations of IFN-α-induced gene expression above the mean of healthy donors. Data are also analyzed categorically using a method which has been developed in our laboratory for use with real-time PCR gene expression data and validated in a cohort of SLE patients.7 Patients and family members are considered to have high serum IFN-α activity if either of the two following criteria are met: (1) two of the three tested IFN-α-induced genes are expressed >1 s.d. above healthy donors, and at least one is greater than or equal to 2 s.d. above healthy donors or (2) one IFN-α-induced gene is expressed greater than 4 s.d. above healthy donor sera. Please see the ‘Patients and methods’ section for more complete details regarding the assay.
Nuclear families with diverse structures were included in the study. A total of 147 families were studied, all with at least one SLE-affected offspring and variable numbers of unaffected offspring. Ninety-six of the nuclear families had both parents available for analysis, and 51 nuclear families had a single parent available. Fifty-nine affected sib pairs were available. One hundred and twenty-six of 266 (47%) SLE patients in the registries demonstrated high serum IFN-α activity, which is similar to previous reports.7,8 Many healthy family members of the SLE patients also had high IFN-α activity, including 27 of 135 (20%) healthy mothers of SLE patients, 21 of 110 (19%) healthy siblings, 14 of 108 (13%) healthy fathers and 3 of 6 (50%) healthy daughters. Sixty-five of the 359 (18%) first-degree relatives had high IFN-α, compared to 5 of 105 (4.8%) healthy unrelated donors (Fisher's exact P = 2.36−10−4). Two of 37 (5.4%) second-degree relatives had high IFN-α, which was not significantly more frequent than healthy unrelated donors, and none of nine third-degree relatives had high IFN-α. Analyzing the IFN-α data quantitatively, healthy first-degree relatives of SLE patients as a group had higher IFN-α activity than healthy unrelated donors (P = 0.0029). Unrelated donor samples were matched by sex and self-reported ethnicity to the healthy first-degree SLE relatives. Comparison of demographic information between healthy first-degree SLE relatives and controls is presented in Table 1. When healthy first-degree relatives were stratified by their relationship to an SLE patient, all first-degree relative groups showed higher IFN-α activity than healthy unrelated donors. Quantitative comparison of IFN-α data are shown in Figure 1. There was a trend toward higher IFN-α activity in healthy mothers as compared to healthy fathers (P = 0.096), however there was no difference between healthy sisters and brothers, and they are presented together as ‘siblings’ on the graph.
High IFN-α activity was clustered in certain families: 54% (43 of 79) of high IFN-α SLE patients had a first-degree healthy relative with high IFN-α, compared to 15% (14 of 87) of low IFN-α SLE patients (Fisher's exact P = 2.43 × 10−7, OR = 6.80, 95% CI = 3.25–14.21). All SLE patients with two or more healthy family members with IFN-α data were included in the above calculation. The relative recurrence risk ratio (λ) is 3.77 for all healthy first-degree relatives, and 1.13 for all healthy second-degree relatives.16 A restricted recurrence risk ratio (λ*)17 can be calculated for the high IFN-α activity trait in which only healthy relatives of a high IFN-α SLE proband are examined. Restricted recurrence risk ratios are 5.41 for all healthy first-degree relatives and 1.44 for healthy second-degree relatives. The increase in the restricted relative recurrence risk as compared to the standard relative recurrence risk (χ2 P = 0.039) is also supportive of the trait clustering in specific families. There was no significant difference in relative recurrence ratios between parents and siblings (3.52 vs 3.98 and 5.29 vs 5.37 parents vs sibs for λ and λ*, respectively) excluding a significant component of dominance variance.16 The large decline in recurrence risk ratios between first- and second-degree relatives is suggestive of a polygenic-multiplicative or epistatic model of inheritance.16 There was no evidence for Mendelian or X-linked inheritance, and no parent of origin effect was detected.
In SLE families with one high IFN-α parent, 48% of all offspring have high IFN-α, as compared with 27% of all offspring of parents who are both low IFN-α (P = 0.0086, OR = 2.45, 95% CI = 1.34–4.49) (Table 2). In the SLE families with two high IFN-α parents, 83% of all offspring have high IFN-α (P = 0.018, OR = 13.07, 95% CI = 1.48–115.0). When parent–offspring proportions are compared for only SLE-affected offspring, the proportion of offspring with high IFN-α in each category is slightly higher, although the trait is frequently transmitted to healthy offspring as well. The degree of high IFN-α activity was also familial, as the quantitative IFN-α activity values for SLE patients were correlated with the average of the IFN-α activity values of their available first-degree relatives (Spearman's r = 0.24, P = 0.0021). This correlation is represented visually using a heat map-style plot in which families are arrayed ordinally by the IFN-α activity value of the SLE-affected member to show variance in familial IFN-α activity as it relates to SLE affecteds (Figure 2). In 59 SLE-affected sib pairs, high vs low IFN-α status was concordant 64% of the time (Fisher's exact P = 0.038, OR = 3.40, 95% CI = 1.31–5.49). There was also a trend toward correlation of quantitative IFN-α values in SLE concordant sib pairs (Spearman's r = 0.17, P = 0.19). An estimate of heritability in the broad sense (H2) for the phenotypic trait of high serum IFN-α activity can be calculated from the correlation coefficient between SLE patients and their first-degree relatives. The H2 value for this comparison is 0.48, meaning that 48% of the observed variance in IFN-α activity of IFN-α in SLE patients is due to familial influence.
SLE patients often had much higher IFN-α activity than their healthy relatives with high IFN-α, so it is likely that disease-specific factors also play a role in IFN-α generation in SLE patients. Data from our lab and others3,18 suggest that anti-RBP autoantibodies such as anti-Ro, anti-La, anti-Sm and anti-RNP, as well as anti-dsDNA are associated with high IFN-α levels in SLE patients. In vitro studies have shown that serum containing anti-RBP antibodies can trigger IFN-α production in dendritic cells when combined with dead cellular material.18 In our family cohorts, SLE patients with anti-RBP antibodies had higher IFN-α activity than those who did not have anti-RBP antibodies (P<0.0001). Similarly, SLE patients with anti-dsDNA antibodies had higher IFN-α activity than those without dsDNA antibodies (P<0.0001). When the SLE patients were stratified by both anti-RBP and anti-dsDNA antibodies, both types of autoantibodies were independently associated with high IFN-α activity, and the two classes of autoantibodies show an additive influence on IFN-α activity, as shown in Figure 3.
Anti-RBP antibody data were known for 293 healthy family members, and of these 5 were positive. Of the five healthy family members with anti-RBP antibodies, only one had high IFN-α, which is similar to the rate for all healthy family members. None of the 293 healthy family members had anti-DNA antibodies. So although healthy family members of SLE patients can rarely have SLE-associated autoantibodies, these autoantibodies do not account for the frequent high IFN-α activity in healthy SLE family members.
SLE-affected sib pairs were concordant for anti-RBP antibodies 66% of the time, and were concordant for anti-dsDNA antibodies 59% of the time. However, the concordance in IFN-α activity in SLE-affected sib pairs was not dependent upon concordance in either anti-RBP or anti-dsDNA autoantibodies individually (P = 0.55 and 0.28 for dependence respectively). When both anti-RBP and anti-dsDNA autoantibodies are taken into account simultaneously and considered to be equal influences, there is a stronger but still nonsignificant trend toward concordant IFN-α activity in SLE-affected sib pairs being dependent on concordance in autoantibodies (P = 0.15, OR = 1.53, 95% CI = 0.73–3.20).
Antinuclear antibody (ANA) testing data were available for 293 healthy family members, and of these 49% had a positive test. There was no relation between ANA test result and IFN-α activity in the healthy family members. However 50% of the healthy relatives with high IFN-α were ANA positive, and 49% of the low IFN-α relatives were ANA positive (P = 0.85 for dependence). So although healthy family members frequently had a positive ANA test, there was no relationship with IFN-α activity as the two variables are assorted in a completely independent fashion.
In the subset of families in which none of the affected or unaffected members have anti-RBP or anti-dsDNA autoantibodies, the percentage of high IFN-α offspring from two low IFN-α parents is 12.6%. This is significantly lower than the percentage for all SLE families, which is 26.8% high IFN-α offspring from two low IFN-α parents (Fisher's exact P = 0.0068). The percentage of high IFN-α offspring from parents with the one-parent high/one-parent low pattern does not change significantly in the anti-RBP/anti-dsDNA autoantibody negative group (44.4 vs 48.4% for all families), as shown in Table 3. This suggests that anti-RBP and anti-dsDNA autoantibodies are interacting in a recessive or disease-specific way with an inherent familial tendency to high IFN-α activity. The relative rarity of anti-RBP or anti-dsDNA autoantibodies in healthy relatives of SLE patients would suggest that these autoantibody traits are more recessive or conditionally expressed, and the concept of anti-RBP and anti-dsDNA autoantibodies resulting from a recessive susceptibility is supported by previous studies.19
Patients of diverse self-reported ethnic backgrounds were included in this study, and there was no difference in IFN-α activity in SLE patients or their healthy family members either quantitatively or qualitatively between self-reported ethnic backgrounds (Figure 4). SLE affects people of diverse ethnic backgrounds, and high IFN-α activity is universally common across all studied ethnicities. Anti-RBP and anti-dsDNA antibodies were present in relatively similar proportion in the SLE patients across all ethnic backgrounds (Figure 4). European Americans and Hispanics were more commonly double negative for anti-RBP and anti-dsDNA, while African-American and Asian patients were more commonly double positive for both types of autoantibodies. Despite these differences, IFN-α activity was not different between self-reported ethnic backgrounds.
These results demonstrate that high serum IFN-α activity is frequently found in healthy SLE family members as compared to healthy unrelated donors, and high IFN-α activity is clustered in certain families among SLE patients and their first-degree relatives. High IFN-α activity is seen in 18% of healthy first-degree relatives and 47% of SLE patients, suggesting IFN-α is a common heritable risk factor for SLE. Serum IFN-α activity is estimated to have a broad-sense heritability of 48% in this study. No variable for environmental influence is used in the equation, as it cannot be measured in the registries. However, SLE is a rare, noncontagious disease in which environmental factors will likely operate through genetic susceptibilities. Given the large number of families studied and the potentially wide variance in environmental influences between geographically and ethnically distinct families, environmental factors would seem more likely to confound a true correlation than to produce a spurious one. All persons who donated samples for this study were screened in a similar way before donation, including being questioned about any current or recent illness. Subjects with common viral infections could still elude this screening if they do not indicate that they are ill, however these events would likely be distributed equally between controls and family members and should not result in the observed correlations. Studies have suggested that IFN-α activity varies over time in SLE patients in relation to disease activity,3,20 however wide temporal variation in IFN-α activity in the SLE patients and their families would be more likely to obscure true familial correlations and bias toward the null hypothesis.
SLE is a highly complex and heterogeneous disease that is likely caused by multiple convergent risk factors, which are different in different patients. Fifty-three percent of patients did not have high levels of IFN-α, however the healthy family members of these patients usually also had low IFN-α activity. For some SLE patients, it is likely that IFN-α is not a risk factor. This would be expected in a disease as complex and heterogeneous as SLE, and knowledge of individual pathogenic factors for specific patients will be important as we try to rationally individualize therapy.
Anti-RBP and anti-dsDNA autoantibodies are strongly associated with increased IFN-α activity in SLE patients. These autoantibodies are very rare in healthy family members and do not explain the frequent high IFN-α activity in healthy first-degree relatives of SLE patients. Anti-RBP and anti-dsDNA antibodies can be detected in the preclinical phase of SLE,21 and these autoantibodies are relatively specific for SLE and other autoimmune diseases such as Sjogren's syndrome. The absence of anti-RBP and anti-dsDNA autoantibodies in almost all healthy family members helps to exclude subclinical autoimmunity such as Sjogren's syndrome as a confounding factor in the familial correlations. Sjogren's syndrome, associated with activation of the IFN-α pathway,22 is usually accompanied by anti-RBP antibodies directed at Ro or La autoantigens, and Sjogren's may be found with increased frequency in SLE families.10 ANA testing was positive in many healthy family members, and this test is often used as a screening test for SLE and other connective tissue diseases. The fact that IFN-α activity in the healthy family members was completely unrelated to ANA test results also helps to exclude subclinical autoimmune states as the cause of high IFN-α activity in the families, and supports the hypothesis that IFN-α is an independent primary risk factor and not a downstream result of autoimmunity.
A previous report suggests a heritable tendency toward having anti-RBP and anti-dsDNA autoantibodies, as evidenced by familial clustering of specific anti-RBP and anti-dsDNA autoantibodies in SLE families.19 In persons genetically predisposed to high serum IFN-α activity, development of these autoantibodies may exacerbate this tendency. Nucleic-acid immune complexes formed by these autoantibodies could enter cells via Fc receptors and subsequently activate endosomal toll-like receptor signaling pathways resulting in IFN-α production, as has been suggested by in vitro models.23 If the impact of increased IFN-α activity on SLE risk is dose-dependent, then such an exacerbation of IFN-α overproduction would increase the chance that SLE will develop. This suggests a ‘two hit’ model of IFN-α-related SLE disease pathogenesis in which persons with a genetic predisposition to high IFN-α activity carry some risk of SLE, and subsequent development of either anti-RBP or anti-dsDNA autoantibodies would exacerbate this tendency and further increase risk of SLE (Figure 5). In this way, anti-RBP and anti-dsDNA autoantibodies may function as ‘microenvironmental’ risk factors, which themselves are at least partially genetically determined. Longitudinal study of SLE sera before and after the onset of disease could help to confirm this hypothesis, and establish the sequence of events in early SLE pathogenesis.
Genes with relevance to the immune system show strong evidence for evolution as they passed from chimpanzees to humans,24 and many also show evidence for ongoing recent selection in human populations.25 As a heritable immune system feature, increased endogenous production of IFN-α could have been positively selected through evolution due to heavy selective pressure from microorganisms. Survival benefit due to improved host defense could explain the persistence of a high IFN-α phenotype, despite a possible increased risk of autoimmunity.
A number of reports support the existence of a basal endogenous IFN-α activity level.26 Variation in basal IFN-α signaling among individuals has been suggested in a microarray gene expression study in which 15 genes known to be regulated by type I IFN showed a 10-fold variation on average between the 75 different healthy persons studied, and were among the most highly variable of all gene transcripts measured.27 Genetic variability in endogenous IFN-α signaling has been suggested by the association of SNPs in the IFN-α pathway genes IRF5 and TYK2 with SLE.12,13 Subsequently, three functional variations in IRF5 have been discovered (a splice site variation, an insertion/deletion and an alternate poly-adenylation site), which define both risk and protective haplotypes in the gene.28,29 The impact of these polymorphisms on the activation of the IFN-α pathway in vivo is not known, however. Genetic study of IFN-α activity in SLE will likely require molecular phenotyping methods as described in this study to elucidate complex interacting genetic risk factors, and we are currently actively investigating the influence of known SLE-associated genetic polymorphisms on IFN-α activity in vivo.
Serum and plasma samples were obtained from the Hospital for Special Surgery (HSS) Lupus Family Registry and the Lupus Multiplex Registry and Repository (LMRR) at the Oklahoma Medical Research Foundation. The HSS cohort was used as an initial test cohort, and then samples from LMRR were used to validate the findings. There were no significant differences between the two cohorts in familial IFN-α activity or in the measures of familial clustering, and data from the two cohorts are presented in aggregate. All serum and plasma samples were carefully stored and were not freeze-thawed before use for this study. A total of 226 samples from the HSS Lupus Family Registry were studied, including 111 SLE patients and 115 healthy family members. Clinical data are available for all samples in the registry, and serologic data are available for all of the SLE-affected individuals. A total of 445 samples from the LMRR were studied, including 155 SLE patients and 290 healthy family members. All samples in the registry, including healthy and affected family members, have carefully collected clinical data available including serologies, SLE disease manifestations and other comorbid illnesses. Healthy unrelated control samples were obtained from healthy blood donors (n = 40), as well as from unrelated individuals who had donated to the two registries as controls (HSS n = 33, LMRR n = 32). All the families in both registries have been genotyped at a number of genetic markers. If the reported family structure is not compatible with the observed distribution of genetic markers, the family is excluded from the study. The study was approved by the institutional review boards at both institutions, and informed consent was obtained from all subjects in the study.
A reporter cell assay is used to measure the ability of patient and family member sera to cause IFN-α-induced gene expression. The reporter cells (WISH cells, ATCC #CCL-25, Manassas, VA, USA) are an epithelial-derived cell line that is highly responsive to IFN-α. Cells are cultured in minimal essential media (Gibco, Invitrogen, Carlsbad, CA, USA) with Earle's salts, 10% fetal bovine serum, 10 mm 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer, 2mm l-glutamine and penicillin/streptomycin. WISH cells are plated in 96-well culture plates at a density of 5 × 105/ml and incubated with 50% patient plasma or serum for 6 h. Recombinant IFN-α is used as a positive control, and healthy sera and culture media are used as negative controls. After 6 h, the cells are lysed. The ability of sera to cause IFN-induced gene expression is largely abrogated by the addition of monoclonal anti-IFN-α antibody, confirming that IFN-α is the major active type I IFN causing the IFN-induced gene expression. This is consistent with results from the previous study,14 in which anti-IFN-α antibodies induced a dose-response decrease in the type I IFN-induced gene response in the reporter cells, while anti-IFN-β, anti-IFN-γ and anti-IFN-λ antibodies had no effect on type I IFN-induced gene expression driven by SLE serum. In a set of samples that show low IFN-α activity, the addition of recombinant IFN-α to the sample results in IFN-α activity proportional to the amount of IFN-α added, excluding any frequent significant endogenous inhibitors of IFN-α in the samples. Data from these inhibitor and agonist control experiments are shown in Supplementary Figure 1. For the agonist experiments to rule out inhibitors, SLE patient samples were used, as there are reports of SLE patients having anti-IFN-α antibodies,30 and it would seem that functional inhibitors would therefore be more common in this group. WISH cells do not express the toll-like receptors TLR3, TLR7, TLR8 or TLR9 to any significant degree (data not shown), so immune complexes containing DNA or RNA in serum should not confound the assay by causing IFN-α generation in the WISH cells. Also, preincubation of WISH cells with cycloheximide does not decrease the IFN-α-induced gene response,14 suggesting that IFN-α which is already present in the samples is driving the IFN-α-induced gene expression in the reporter cells. WISH assay was done on 50 serum samples from SLE patients who had previously been studied for IFN-α-induced gene expression in their PBMCs,3 and serum IFN-α activity by WISH assay was highly correlated with IFN-α-induced gene expression in PBMC (Spearman's r = 0.524, P = 0.0001, data not shown). Additionally, 29 of 30 replicates done on different aliquots of the same serum sample, but performed on different days produced the same high vs low IFN-α activity result, suggesting that day-of-assay did not play a role in the familial clustering seen in the paper (data not shown). Efforts were also made to run samples from members of the same family on different days.
Total cellular mRNA is purified from the WISH cell lysates using the Qiagen Turbocapture 96-well RNA purification kit (Qiagen, Valencia, CA, USA) per manufacturer protocol. cDNA is made from the mRNAs using the Invitrogen oligo-dT primer and Superscript III reverse-transcriptase system. Invitrogen RNAse-Out RNAse inhibitor is used in this step to prevent degradation of the RNA.
Ten microliters of a 1:40 dilution of the cDNA made from total cellular mRNA is then quantified using real-time PCR with the Biorad SYBR Green fluorophore system with a Biorad iCycler thermal cycler (Biorad, Hercules, CA, USA). Forward and reverse primers for the genes MX1, PKR and IFIT1, which are known to be highly inducible by IFN-α,7 are used in the reaction (MX1: F = TACCAGGACTACGAGATTG, R = TGCCAGGAAG GTCTATTAG; PKR: F = CTTCCATCTGACTCAGGTTT, R = TGCTTCTGACGGTATGTATTA; IFIT1: F = CTCCT TGGGTTCGTCTATAAATTG, R = AGTCAGCAGCCAG TCTCAG). The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is also amplified to control for background gene expression (GAPDH: F = CAACGGATTTGGTCGTATT, R = GATGGCAACAA TATCCACTT). Each sample and control is run in duplicate. Melt curves are analyzed to ensure the specificity of the amplified product. Standard curves are generated for each PCR experiment using known dilutions of cDNA to control for variable PCR efficiency at varying concentrations of substrate.
The amount of PCR product of the IFN-α-induced gene is normalized to the amount of product for the housekeeping gene GAPDH in the same sample. The relative expression of each of the three tested IFN-induced genes is calculated as a fold increase compared to its expression in WISH cells cultured with media alone. Healthy unrelated donor sera are tested in the WISH assay to establish a normal value for IFN-α activity, and the mean and standard deviation (s.d.) of IFN-α-induced gene relative expression induced by healthy donor sera in the WISH assay are calculated. The ability of patient and family member serum samples to cause IFN-induced gene expression in the reporter cells is then compared to the mean and s.d. induced by healthy unrelated donor serum. The number of s.d. of relative expression above healthy donors for each gene is calculated. Number of s.d. above normal is used to quantify IFN-α activity instead of raw relative expression data because some genes are more highly induced than others, resulting in the more highly inducible genes being overrepresented in aggregate relative expression data. The relative expression values for each of the three genes individually showed a strong correlation with the combined sum of the number of s.d. above healthy unrelated donors (Spearman's r-values: IFIT-1 = 0.710, MX-1 = 0.770, PKR = 0.822, all P-values <0.0001), confirming coordinate regulation of the three transcripts and that each measured gene is accurately represented in the s.d. analysis and is contributing to the combined sum of s.d. value (Supplementary Figure 1). The three transcripts chosen for measurement in this study were selected to represent coordinate activation of the IFN-α pathway as would be expected after ligation of the type I IFN receptor, instead of measurement of only one IFN-α-induced transcript which may not always accurately represent pathway activation.
Experimental samples are also analyzed categorically using the following method, which has been developed in our laboratory for use with real-time PCR gene expression data and validated in a cohort of SLE patients.7 Patients and family members are considered to have high serum IFN-α activity if either of the two following criteria are met: (1) two of the three tested IFN-α-induced genes are expressed >1 s.d. above healthy donors, and at least one is greater than or equal to 2 s.d. above healthy donors or (2) one IFN-α-induced gene is expressed greater than 4 s.d. above healthy donor sera. If neither criterion is met, the sample is considered to have low IFN-α activity. Fisher's exact and χ2 statistical analyses are used for categorical data such as presence or absence of autoantibodies, high vs low IFN-α, and so on. Mann–Whitney t-test is used to compare quantitative IFN-α activity data.
ANA testing was done with standard clinical laboratory methods by applying sera to Hep2 cells, and subsequent fluorescent staining for bound antibodies. Anti-RBP antibodies were detected using ELISA assays for anti-Ro, anti-La, anti-Sm and anti-RNP antibodies. dsDNA antibodies were detected by standard clinical lab methods using the Crithidia luciliae assay. In this assay, sera are applied to fixed C. luciliae organisms, followed by fluorescent staining for bound antibodies.
Family members are first classified as SLE affected or unaffected. The unaffected individuals are then categorized by their closest relationship to an SLE-affected individual in the family. Unaffected family members were classified by their most direct relationship to an SLE patient (for example in an SLE family with multiple affected generations, sometimes a person could be both an SLE mother and an SLE grandmother – in this case the person is categorized as an SLE mother). Each person in each registry is represented only once. Familial clustering is detected using the Fisher's exact test with categorical IFN data from SLE patients and their families. Because high IFN-α activity is uncommon in healthy individuals, the following screening algorithm was used. All SLE patients and their corresponding first-degree relatives were examined for high or low IFN-α activity. If one or more of the first-degree relatives had high IFN-α, the nuclear family was categorized as high IFN-α. The nuclear family IFN-α status was then compared to that of the SLE patient. Each SLE patient was counted once, and in nuclear families with more than one SLE patient, each patient was categorized individually. Families with at least two healthy first-degree relatives with IFN-α data were included (n = 166 nuclear families, mean of 2.76 available first-degree relatives, s.d. = 0.83). Numbers in each category are analyzed using Fisher's exact test (all Fisher's exact results are calculated using the two-sided sum of small P's method). Odds ratio (OR) for concordance in categorical IFN-α activity between SLE patients and their nuclear families was calculated using standard procedure with an odds ratio calculator (http://www.hutchon.net/ConfidOR.htm), with input variables being the number of families with of each of the following IFN-α activity patterns: patient↑ (high)/family↑, patient↑/family↓ (low), patient↓/family↑ and patient↓/family↓. Relative recurrence risk ratios (λ) were calculated as follows: λ = proportion of high IFN-α healthy SLE relatives/proportion of high IFN-α healthy unrelated,16 and the restricted relative recurrence risk ratio (adapted from Rybicki et al.17) is calculated as λ* = proportion of high IFN-α relatives of a high IFN-α SLE proband/proportion of high IFN-α healthy unrelated donors. Calculations of λ and λ* were performed using the data from all available first-degree and second-degree relatives. Spearman's r is calculated for the correlation of the variances in IFN-α activity (expressed quantitatively as the number of s.d. above healthy donors) between SLE patients and the average of the variance in available first-degree relatives. In this calculation, families with two or more healthy first-degree relatives with IFN-α activity data were used. Heredity in the broad sense (H2) is calculated as follows: H2 = ρ/t, where ρ is the coefficient of correlation for the quantitative trait, and t the coefficient of relationship, which is assumed to be 0.5 for all first-degree relatives.31 ORs for concordance of IFN-α activity in SLE-affected sib pairs were calculated using the following method from Olson et al.,32 OR = 4n1n3/ , where n1 is the number of concordant affected, n2 the number of discordant and n3 the number of concordant unaffected. Ninety-five percent confidence interval was calculated as recommended: ±1.96√(n2n3+n1n2+4n1n3)/(n1n2n3). OR for the concordance of both autoantibodies and IFN-α activity in the SLE-affected sib pairs is calculated using the standard OR formula to measure the degree that concordance of one variable influences the concordance of the other. Input data for the dual concordance ORs are the proportion of pairs with the following IFN-α/autoantibody classifications: concordant/concordant, concordant/discordant, discordant/concordant or discordant/discordant, respectively at these two variables. In the SLE-affected sib pairs, P-values for the dependence of IFN-α concordance on autoantibody concordance calculated using Fisher's exact test, with the number of sib pairs in each of the above four concordance categories as the variables in the quadratic. In the model, which accounts for both autoantibodies simultaneously as equal variables, the sib is classified as autoantibody positive if either anti-RBP or anti-dsDNA autoantibodies are present, and autoantibody negative in the absence of both. For example, Sib 1: high IFN-α, anti-RBP +/anti-dsDNA−; Sib 2: high IFN-α, anti-RBP-/anti-dsDNA + would be considered to be concordant for both IFN-α activity and autoantibodies in this model, while this pair would be considered discordant for both antibodies when each antibody is considered separately. Sib pair sample size prevents modeling the two autoantibodies in an additive fashion.
We acknowledge Karen Onel and Kenan Onel for their contribution in establishing the Hospital for Special Surgery Family Lupus registry, and Marie Flesch for her assistance in obtaining materials from the Lupus Multiplex Registry and Repository at Oklahoma Medical Research Foundation. The Hospital for Special Surgery Family Lupus Registry was established with support from the S.L.E. Foundation, Inc. and the Toys-R-Us Foundation. The Lupus Multiplex Registry and Repository is supported by NIH NIAMS AR-5-2221 to JBH TBN received salary support from NIH T32 AR07517 and NIAID Clinical Research Loan Repayment grants. This work was supported by the NIH (AI059893) and research grants from the Alliance for Lupus Research, the Lupus Research Institute and the Mary Kirkland Center for Lupus Research to MKC.
Disclosures and Support: TB Niewold – NIH T32 AR07517, NIAID Clinical Research Loan Repayment; J Hua – patent pending for interferon assay; TJA Lehman – none; JB Harley – Lupus Multiplex Registry and Repository – NIH NIAMS AR-5-2221; MK Crow – patent pending for interferon assay, NIH R01 AI05983-01, Research Grants from Alliance for Lupus Research, Mary Kirkland Center for Lupus Research and Lupus Research Institute; Support for HSS Family Lupus Registry – Toys 'R Us Foundation and SLE Foundation.