Article Text
Abstract
Objectives Ncf1 polymorphisms leading to low production of reactive oxygen species (ROS) are strongly associated with autoimmune diseases in animal models. The human NCF1 gene is very complex with both functional and non-functional gene copies and genotyping requires assays specific for functional NCF1 genes. We aimed at investigating association and function of the missense single nucleotide polymorphism (SNP), rs201802880 (here denoted NCF1-339) in NCF1 with systemic lupus erythematosus (SLE).
Methods We genotyped the NCF1-339 SNP in 973 Swedish patients with SLE and 1301 controls, using nested PCR and pyrosequencing. ROS production and gene expression of type 1 interferon-regulated genes were measured in isolated cells from subjects with different NCF1-339 genotypes.
Results We found an increased frequency of the NCF1-339 T allele in patients with SLE, 11% compared with 4% in controls, OR 3.0, 95% CI 2.4 to 3.9, p=7.0×10−20. The NCF1-339 T allele reduced extracellular ROS production in neutrophils (p=0.004) and led to an increase expression of type 1 interferon-regulated genes. In addition, the NCF1-339 T allele was associated with a younger age at diagnosis of SLE; mean age 30.3 compared with 35.9, p=2.0×1−6.
Conclusions These results clearly demonstrate that a genetically controlled reduced production of ROS increases the risk of developing SLE and confirm the hypothesis that ROS regulate chronic autoimmune inflammatory diseases.
- NCF1
- NADPH oxidase complex
- SLE
- autoimmunity
- reactive oxygen species
Statistics from Altmetric.com
Introduction
Genetic mapping and positional cloning of genetic polymorphisms associated with chronic autoimmune diseases in animal models have revealed the Ncf1 gene to be of major importance.1 2 The Ncf1 gene encodes the p47phox/Ncf1 protein of the NADPH oxidase (NOX2) complex, which is critical for the induction of reactive oxygen species (ROS).3 ROS were at the time believed to mainly contribute to the chronic inflammation in autoimmune diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA). However, it is now clear that ROS also have important regulatory functions in the immune system (reviewed in ref 4). The finding that genetically controlled low capacity of ROS production by NOX2 leads to autoimmunity was also supported by the observation that chronic granulomatous disease (CGD), caused by mutations in the subunits of NOX2, has autoimmune and lupus-like symptoms.5 6 The question arose whether polymorphisms in the NOX2 genes, leading to decreased ROS production, also are associated with autoimmune diseases. There are now several reports suggesting associations of NOX2 genes with autoimmune diseases such as SLE and RA.7 8 Interestingly, Jacob et al reported an association with SLE to a single nucleotide polymorphism (SNP) in the NOX2 component p67phox/NCF2, where the causal allele was shown to reduce ROS production.9
In the human genome, NCF1 is located in a structurally complex region,10 11 which complicates genotyping and has excluded NCF1 from genome-wide association studies. Close to NCF1 are two NCF1 pseudogenes that encode truncated, non-functional proteins.12 In addition, the NCF1 gene exists in a varying number of copies, which seem to consist of parts of NCF1 and parts of a pseudogene.8 13 We have previously investigated the functional effects of three SNPs in NCF1 and found that the minor allele (T) of an SNP in exon 4 (rs201802880, here denoted NCF1-339) reduced ROS production in transfected cell constructs.8 The nucleotide shift from C to T alters the amino acid from arginine to histidine at a membrane binding site of the Ncf1 protein.14 In light of the findings that genetically encoded low ROS contribute to SLE, we asked if the NCF1-339 SNP was associated with SLE.
Methods
Study populations
Patient characteristics are outlined in online supplementary table 1. All patients included met at least 4 of 11 American College of Rheumatology (ACR) classification criteria.15 The discovery population comprised the Linköping, Uppsala and Lund patients with SLE and 1016 controls from the epidemiological investigation of rheumatoid arthritis (EIRA) case–control study on incidence (72% women, all Caucasians).16 About 480 of the EIRA control genotypes are previously published,8 but an additional 536 EIRA controls were included in this study. The EIRA controls and patients with SLE are sampled from the same area of Sweden and should represent comparable populations. The replication population comprised the Karolinska patients with SLE and 303 controls, matched by age and sex to 303 of the patients with SLE.
Supplementary Material
Genotyping
The genotyping of NCF1-339 and copy number analysis of NCF1 were performed using a nested PCR strategy and pyrosequencing8 17 with modifications detailed in supplementary methods.
Functional analyses
Patients with RA and SLE with specific NCF1-339 genotypes were recruited at the rheumatology clinics at Karolinska University Hospital and Skåne University Hospital, respectively. Blood was sampled from 2 TT, 4 CT, 8 CC and 3 CCT patients with RA and 5 TT, 12 CT and 13 CC patients with SLE. Polymorphonuclear cells (PMN) and peripheral blood mononuclear cells (PBMC) were isolated according to ref 18 and described in supplementary methods.
ROS production
ROS production by blood cells was analysed by isoluminol or luminol-enhanced chemiluminescence (CL)19 and the Phagoburst assay (Glycotope), detailed in supplementary methods.
Gene expression
Blood was collected in PAXgene blood RNA collection tubes and RNA was isolated according to instructions in the PAXgene blood RNA kit (PreAnalytix, Qiagen). mRNA expression was measured using TaqMan assays (Thermo Fisher Scientific) described in supplementary methods.
Statistical analyses
Statistical analysis was done in the software JMP V.12 (SAS Institute), Prism V.7 (GraphPad), Excel (Microsoft) and StatsDirect V.3 (StatsDirect). Genetic association of genotype, allele and genotype group frequencies was analysed in contingency tables with χ2 tests or Fisher’s exact test. The genetic analyses were stratified according to ancestry, and presented results only include Caucasians. Meta-analyses were performed with Mantel-Haenszel χ2 tests. Mann-Whitney U test and Kruskal-Wallis non-parametric test were used to compare differences in ROS production and gene expression. Distributions of age at diagnosis, ACR classification criteria and Systemic Lupus International Collaborating Clinics/ACR disease damage index (SDI) scores were analysed with histograms and comparisons of frequencies and mean values were analysed using χ2 statistics or Fisher’s exact tests and Mann-Whitney U test and Kruskal-Wallis statistics.
Study approval
Oral and written informed consent was obtained from all subjects. The study protocol for the genetic analyses was approved by the regional ethics review boards in Lund, Linköping, Uppsala and Stockholm and the functional analyses by the regional ethics review boards in Lund and Stockholm.
Results
The T allele of the SNP NCF1-339 is highly enriched in patients with SLE
To investigate if the T allele of NCF1-339 is associated with SLE, we genotyped a discovery population of 570 Swedish Caucasian patients with SLE and 1016 controls from the Swedish EIRA cohort. The T allele was enriched in patients with SLE, with a frequency of 0.11 compared with 0.04 in the controls with an OR of 2.6, 95% CI 2.0 to 3.5 (table 1). As replication, we genotyped a Swedish case–control cohort comprising 403 patients with SLE and 285 controls (Karolinska). The T allele was associated with SLE also in this cohort, with a frequency of 0.10 in patients with SLE compared with 0.02 in the controls, OR 4.6, 95% CI 2.6 to 8.0 (table 1). Meta-analysis of the two study populations gives a T allele frequency of 0.11 in patients with SLE compared with 0.04 in the controls, OR 3.0, 95% CI 2.4 to 3.9, p=7.0×10−20. The complete genotype results are presented in online supplementary table 2.
A higher frequency of patients with SLE have only one NCF1 gene
Carriers of NCF1-CGD have only one NCF1 gene, due to non-allelic homologous recombination between NCF1 and the pseudogenes.11 We analysed NCF1 gene copy number in the SLE study populations and found that a higher frequency of patients with SLE have only one NCF1 gene compared with controls, 1.1% compared with 0.2%, in all cohorts combined, OR 5.0, 95% CI 1.4 to 17.6 (online supplementary table 3). There were no significant differences in frequency of more than two NCF1 genes in patients with SLE compared with controls.
T-type genotypes have a higher frequency in patients with SLE
The effect of the T allele needs to be considered in relation to the number of NCF1 gene copies. The CT, TT and CTT genotypes are enriched in patients with SLE, but CCT is not, indicating that additional C alleles can compensate the functional effects mediated by the T allele (online supplementary table 2). To estimate the total genetic effects of the NCF1-339 association, we grouped the genotypes with less than two C alleles: C, T, TT, CT, CTT (denoted T type) and the remaining genotypes: CC, CCT, CCCC, CCCT and CCTT (denoted C type) and compared the frequency in SLE cases and controls. The T-type group had a higher frequency in patients with SLE compared with controls (table 2). Meta-analysis of the combined study populations gives an OR of 3.7, 95% CI 2.7 to 4.9, p=1.8×10−18.
The SLE-associated T allele of NCF1-339 reduces extracellular ROS production
To investigate if the NCF1-339 T allele has an effect on ROS production, we measured extracellular and intracellular ROS production in primary cells from patients with SLE with two NCF1 genes and CC, CT and TT NCF1-339 genotypes. To capture the effect of the T allele on different NOX2 activation pathways, three stimuli were used: phorbol 12-myristate 13-acetate (PMA), the chemotactic peptide N-Formyl-Met-Leu-Phe (fMLF) and phagocytosis inducing serum-opsonised zymosan (SOZ). The PMA-stimulated extracellular ROS were delayed and reduced in PMN from patients with TT genotypes compared with CC (figure 1A–C). There was also a T allele-dependent reduction in extracellular ROS in fMLF and non-stimulated PMNs (figure 1D,F). No genotype-dependent differences could be observed when cells were stimulated with SOZ (figure 1E), in PBMCs (online supplementary figure 1), or in intracellular ROS production for any stimuli (online supplementary figure 2). In order to verify these results and to investigate ROS production from different types of cells, we used a flow-cytometry assay to measure ROS accumulated intracellular, after stimulation with PMA or opsonised Escherichia coli. The T allele significantly reduced PMA-stimulated ROS in PMN, but not in monocytes or when stimulated with opsonised E. coli (figure 2). The T allele also reduced the responsiveness of PMNs to stimulation with PMA.
Supplementary Material
There was no difference in age, disease activity, as determined by the SLE Disease Activity Index,20 or corticosteroid use that could explain the observed differences in ROS production (online supplementary table 4 and supplementary figure 3). There was a difference in hydroxychloroquine (HCQ) usage between the genotype groups and ROS were higher in patients on HCQ treatment compared with patients without. However, there was no difference when HCQ treatment was analysed within the CT genotype group (online supplementary figure 3).
To investigate the ROS-reducing effect of the T allele in a different autoimmune patient group, and to see if additional NCF1 genes can restore the reduction of ROS caused by the T allele, we measured extracellular and intracellular ROS in PMN and PBMCs from patients with RA with CC, CT, TT and CCT NCF1-339 genotypes. The T allele reduced PMA-stimulated extracellular ROS in both PMNs and PBMCs, but no difference in intracellular ROS was detected (online supplementary figure 4). There was also no significant difference in ROS production in patients with CCT genotypes compared with patients with CC genotypes.
The T allele of Ncf1-339 increases the expression of type 1 interferon-regulated genes
We have previously shown that genetically encoded NOX2 deficiency, in both mice and patients with CGD, increased the expression of type 1 interferon-regulated genes (IRG).21 To investigate if the same is seen for the NCF1-339 T allele, we measured the expression of selected IRGs in whole blood from patients with SLE and RA included in the ROS study. Patients with RA with CT genotypes had a significantly increased expression of five IRGs, IFI44L, ISG15, OAS1, IRF7 and STAT1, compared with the patients with CC genotypes (figure 3). In the SLE cases, however, there were no expression differences between the genotype groups for IRF7, ISG15 and IFI44L. The difference in fold change between patients with SLE was much larger compared with patients with RA, reflective of a stronger interferon (IFN) signature in patients with SLE.
NCF1-339 T allele is associated with a lower age at diagnosis
We analysed the clinical data from the SLE study populations to investigate if the NCF1-339 T-type genotypes are associated with clinical phenotypes or disease severity. Patients with T-type genotypes had a younger age at diagnosis compared with patients with C-type genotypes, mean age 30.3 compared with 35.9, p=2.0×10−6 (table 3). There were no consistent differences in the distribution of ACR criteria or in SDI score between C-type and T-type genotypes, although renal disorder was more common in the T-type group in the Karolinska patients with SLE (p=0.03), with the same tendency seen in Lund and Linköping (online supplementary table 5).
Discussion
We report that an amino acid replacement in NCF1, leading to a lower capacity of inducing oxidative burst, is strongly associated with SLE. The OR is 3.7, which makes it one of the strongest identified genetic associations with SLE.22 23 The strength of the association is indicated by a 6-year earlier disease onset in patients with T-type genotypes. The NCF1 gene is highly complex, which has excluded SNPs in NCF1 in genome-wide association studies. NCF1 genotyping requires specialised methods to exclude the non-functional NCF1-pseudo genes, and capture all functional gene copies. During the finalisation of this paper, a similar study was published confirming a strong effect of the NCF1-339 T allele on SLE and also on Sjögren’s syndrome and RA,24 thus two independently performed reports show that NCF1-339 is one of the strongest SNPs, outside the HLA region, associated with autoimmune diseases.
The NCF1-339 T allele leads to a shift from Arg to His at position 90, which is located in the phox domain of NCF1 and mediate binding to the cellular membrane. We and others have previously shown that mutating position 90 reduces the ROS response8 and the binding efficiency of NCF1 to the membrane.25 Here we wanted to investigate the functional effects of the NCF1-339 T allele on primary cells relevant for the pathogenesis of SLE. We demonstrate that the NCF1-339 T allele reduced the capacity of the NOX2 complex to produce ROS in PMNs from both patients with SLE and RA. In patients with RA, a significant reduction is seen also in PBMCs.
The high frequency of a ROS-reducing allele in patients with SLE, found in this study, goes in line with the previously reported association of a ROS-reducing SNP in the NCF2 gene with SLE,9 as well as with data from animal models. Mice with a loss-of-function mutation in Ncf1 develop spontaneous SLE-like disease, including production of autoantibodies and increased expression of type 1 IRGs.21 Furthermore, patients with CGD have an increased expression of type 1 IRG in whole blood.21 In this study, we found that the NCF1-339 T allele increased the expression of type 1 IRG in whole blood from patients with RA. Taken together, these findings show that ROS are important for regulation of this pathway. However, we were not able to detect an increased type 1 IRG expression in patients with SLE with the NCF1-339 T allele, instead we saw a dramatic increase in expression of IRG in most of the patients, reflective of the IFN signature. A likely explanation could be that ROS are involved in regulating the initiation of the type 1 IFN pathway, but once the pathway is activated the strong IFN signature in SLE overshadows the genetic effect of the T allele. However, it is also possible that the IFN effect caused by the NCF1-339 T allele seen in patients with RA and CGD is not present in patients with SLE.
How ROS affect the type 1 IFN pathway is not known, but the influence of oxidation is far upstream, affecting STAT1 in monocytes21 or possibly the release of IFN-α from plasmacytoid dendritic cells.26 ROS produced by monocytes can also inhibit interacting cells, for example, NK cells or classical αβ T cells during antigen presentation.27 28 ROS can be released within the immunological synapse or by transfer of exosomes containing NOX2 to CD4 T cells.29 These findings all point to the crucial role that ROS play in controlling the immune response and preventing excessive activation of immune effector cells, such as T cells and NK cells. Furthermore, we show that the NCF1-339 T allele reduces the burst capacity of PMNs, which mainly consist of neutrophils. This is possibly due to that neutrophils produce large amount of ROS detectable in vitro and does not exclude that the regulatory ROS come from other cell types. Neutrophil-derived ROS are a prerequisite for the formation of neutrophil extracellular traps (NETs), suggested to be pathogenic in SLE. However, NETs have also been reported to protect against chronic inflammation by absorption of inflammatory cytokines.30 In addition, in patients with CGD, NETs are also formed as a result of mitochondrial-derived ROS, which seem to be more inflammatory compared with nuclear DNA in regular NETs.31
The previously reported SLE-associated SNP in NCF2 only affects intracellular ROS, stimulated by phagocytosis inducing stimuli, via the Fcγ receptor (FcγR).9 The NCF1-339 T allele, on the other hand, has no effect on ROS induced via the FcγR with opsonised zymosan or E. coli, and a strong effect on extracellular ROS. These two SLE-associated SNPs seem to affect two different NOX2 activation pathways. However, it is methodologically difficult to separate extracellular from intracellular ROS production. We used two methods to measure ROS inside the cell with various results, likely explained by the different experimental set-up. In the Phagoburst assay, extracellular H2O2 produced by NOX2 in the plasma membrane could diffuse across the membrane into the cell and react with the detection probe.32 In the intracellular CL assay, extracellular ROS are immediately scavenged by added superoxide dismutase and catalase, ensuring that only intracellular ROS are measured. The difficulties to separate intracellular from extracellular ROS are also reflected in the in vivo situation, where ROS produced by NOX2 in the plasma membrane could diffuse or be transferred into the cell and affect intracellular pathways. The fact that we see a genotype-dependent difference in burst using the Phagoburst assay, but not using the intracellular CL assay, suggests that extracellular H2O2 has the capacity to alter intracellular systems. The reasons why the T allele has no effect on FcγR-stimulated ROS could be because NCF1 is less important for the intracellular ROS induced by particulate stimuli and released during phagocytosis.25 NCF1 and the NOX2 subunit NCF4 are both keeping the cytosolic part of NOX2 attached to the membrane, which is required for full activation. The two membrane-binding sites of NCF1, including Arg90, mediate stronger binding to the plasma membrane, whereas the binding sites of NCF4 have specificity for phagosomal membranes. Mutations in Ncf4 and Ncf1 have in fact been shown to affect autoimmune diseases differentially.33
We saw no difference in ROS production in subjects with the NCF1-339 CCT genotype compared with CC, despite the T allele, which demonstrates how important it is to consider NCF1 copy number when analysing the genetic and functional effects of the NCF1-339 association. The functional effects of the NCF1-339 genotypes with additional NCF1 genes are harder to predict than the standard two allele genotypes. One additional NCF1 gene does not increase ROS production because there is no increase in the other NOX2 subunits.34 The CCT genotype could have a similar functional effect as CC, because of the two functional C alleles; however, the NCF1 protein with His90 (encoded by the T allele) could also compete with NCF1 proteins with Arg90 for binding to NOX2 subunits, and thus lead to reduced ROS production.
Taken together, the SLE-associated NCF1-339 SNP, leading to amino acid variability in the NCF1 protein and a lower capacity to induce oxidative burst, is so far one of the strongest loci associated with an autoimmune disease and confirm the hypothesis that ROS regulate chronic autoimmune inflammatory diseases.
Acknowledgments
We would like to thank Lars Klareskog and Leonid Padyukov for providing the EIRA control samples.
References
Footnotes
Contributors LMO and RH designed the research. LMO performed the major part of the experimental work and statistical analyses. ÅCJ performed the Phagoburst assay analysis and BG helped with preparing the patient samples. AJ, SS, LR, DL, JW, CS, ES, IG and AAB contributed patient samples and clinical data. LMO and RH wrote the manuscript. All authors revised and approved the final manuscript.
Funding The work was supported by grants from the Knut and Alice Wallenberg Foundation, the Swedish Rheumatism Association, the Swedish Medical Research Council, the Swedish Research Council, the Swedish Foundation for Strategic Research and the following foundations: King Gustaf V’s 80-Year Fund, Alfred Österlund, Greta and Johan Kock, Anna-Greta Crafoord and Swedish Heart-Lung. The work was also supported by the Stockholm County Council and Skåne University Hospital (ALF) and the FOREUM Foundation. The research leading to these results received further funding from the European Union Innovative Medicine Initiative project BeTheCure.
Competing interests None declared.
Ethics approval The regional ethics review boards in Lund, Linköping, Uppsala and Stockholm.
Provenance and peer review Not commissioned; externally peer reviewed.