Inactive disease in patients with lupus is linked to autoantibodies to type I interferons that normalize blood IFNα and B cell subsets

Summary Systemic lupus erythematosus (SLE) is characterized by increased expression of type I interferon (IFN)-regulated genes in 50%–75% of patients. We report that out of 501 patients with SLE analyzed, 73 (14%) present autoantibodies against IFNα (anti-IFN-Abs). The presence of neutralizing-anti-IFN-Abs in 4.2% of patients inversely correlates with low circulating IFNα protein levels, inhibition of IFN-I downstream gene signatures, and inactive global disease score. Hallmarks of SLE pathogenesis, including increased immature, double-negative plasmablast B cell populations and reduction in regulatory B cell (Breg) frequencies, were normalized in patients with neutralizing anti-IFN-Abs compared with other patient groups. Immunoglobulin G (IgG) purified from sera of patients with SLE with neutralizing anti-IFN-Abs impedes CpGC-driven IFNα-dependent differentiation of B cells into immature B cells and plasmablasts, thus recapitulating the neutralizing effect of anti-IFN-Abs on B cell differentiation in vitro. Our findings highlight a role for neutralizing anti-IFN-Abs in controlling SLE pathogenesis and support the use of IFN-targeting therapies in patients with SLE lacking neutralizing-anti-IFN-Abs.


In brief
Bradford et al. characterize the prevalence of anti-IFNa-Abs in patients with SLE and their association with serum levels of IFNa, clinical parameters, and B cell abnormalities. Patients with SLE harboring autoantibodies that neutralize IFNa show reduced serum IFNa levels and ISG expression, disease severity, and normalized B cell compartments.

INTRODUCTION
Systemic lupus erythematosus (SLE) is a heterogeneous autoimmune disease affecting multiple organ systems. Abnormal B cell proportions including expansion of atypical memory, also known as double-negative (DN) B cells, and autoantibody-secreting plasma cells contribute to autoimmune inflammation and tissue injury. [1][2][3] In addition to B cell dysfunction, $50%-75% of patients with SLE present an upregulation of type I interferon (IFN-I)-stimulated genes (ISGs) that directly correlate with disease severity. The IFN-I family includes IFNb, IFNu, IFNε, IFNk, and 13 additional subtypes of IFNa that bind to the same receptor, IFNAR. 4 We and others have previously shown that a finely tuned IFNa response is required to induce the differentiation of immature B cells into plasma cells that produce antibodies during, for example, viral infection, as well as regulatory B cells (Bregs) that restore homeo-stasis. 5,6 In SLE, chronic IFNa production fuels autoimmunity by promoting the differentiation of monocytes to dendritic cells (DCs), 7,8 which activate autoreactive T cells; the generation of effector and memory CD8 + T cells [9][10][11] ; and the differentiation of B cells into autoantibody-producing plasma cells but not Bregs. 5,12 The pathogenic role of IFNa in SLE is supported by several clinical observations. Patients with monogenic diseases, including complement and FASL deficiency and TREX-1 mutation, which all lead to IFN-I overproduction, display SLE-like symptoms. [13][14][15] Patients treated with IFN-I for cancer and chronic infections develop a lupus-like disease and/or anti-doublestranded DNA (dsDNA) antibodies. 16,17 IFN-a kinoid vaccination induces antibodies that cross-neutralize all IFNa subtypes, which in $50% of immunized SLE patients has shown therapeutic efficacy. 18 IFN-I blockade has also been shown to be beneficial in patients with SLE. 19,20 Neutralizing autoantibodies to IFN-I has been reported to develop in patients treated with IFNa2 or IFNb therapy 21,22 ; in the majority of patients with autoimmune polyendocrinopathy syndrome type I (APS-1) 23,24 or thymoma 25 ; at lower frequencies in rheumatic diseases, including cross-sectional lupus cohorts 26-28 ; and more recently in a subset of patients with lifethreatening COVID- 19. 29,30 Here, we showed that neutralizing autoantibodies against IFNa (anti-IFN-I-Abs) cross-react with all IFNa subtypes in a cross-sectional and longitudinal cohort of patients with SLE and are associated with significantly reduced levels of circulating IFNa levels, disease activity, and restored B cell responses, suggesting a disease-aggravating role for non-neutralizing anti-IFN-Abs.
Most autoAbs to cytokines were either undetectable or produced at low concentrations in patient or control sera. High levels of autoAbs to IFNl were detected in patients and controls ( Figure 1A). We detected a significant increase in autoAbs to IFNa (66 out of 474 patients) and IFNu (59 out of 474 patients) in patients with SLE compared with controls ( Figures 1B and  1C). Reactivity toward IFN-I subtypes was partially overlapping as 12% (n = 43) of patients had autoAbs to both IFNa and IFNu, whereas anti-IFNa or -IFNu single-positive patients comprised 4% each. Interestingly, the levels of anti-IFN-Abs significantly positively correlated with anti-IFNu-Abs ( Figure 1D). Due to the well-established role of IFNa in promoting SLE pathogenesis, we focused our attention on the cohort of patients that displayed anti-IFN-Abs. Of note, anti-IFN-Abs were predominantly of the immunoglobulin G1 (IgG1) subclass ( Figure 1E).
Quantification of serum IFNa levels with the ultrasensitive Simoa method 31 showed that 93% of patients with SLE had IFNa serum levels over the detection limit (0.7 fg/mL) compared with 30% of controls ( Figure S1A). The presence of high titers of anti-IFN-Abs mirrored a significant reduction in the levels of circulating IFNa compared with those who were anti-IFN-Ab negative and with those with low anti-IFN-Ab titers ( Figure 1F).
The capacity of anti-IFN-Abs to neutralize IFNa was assessed using a reporter-cell-line-based neutralization assay as previously described. 32 Serum samples with high anti-IFN-Ab levels were more efficient in blocking all tested subtypes (IFNa2, -5, -6, and -8) of IFNa bioactivity in vitro ( Figures 1G and S1B). Only anti-IFN-Abs with a neutralizing capacity of IC50 >100 negatively correlated with serum levels of IFNa ( Figure 1H).
To gain mechanistic insight into the capacity of neutralizing anti-IFN-Abs to reduce downstream IFN-I signaling, we compared the IFN-I composite score, 33 a cumulative measure of mRNA expression of four individual ISGs, MX1, MCL1, IRF9, and STAT1 (see STAR Methods), in patients with SLE with and without anti-IFN-Abs and controls. IFN-I score was significantly  higher in anti-IFN-Ab-negative and non-neutralizing anti-IFN-Ab patients compared with controls and with patients with neutralizing anti-IFN-Abs. The latter displayed an IFN-I score comparable to controls ( Figures 1I and S1C). We measured anti-IFN-Ab titers longitudinally over an average of 10 years from the first sample collection. All patients tested have autoAbs against 12 subtypes of IFNa (IFNa1, -2, -4, -5, -6, -7, -8, -10, -14, -16, -17, and -21) at high titers ( Figures 1J and S1D).
Neutralizing anti-IFN-Abs are a proxy for persistent low levels of IFNa and are associated with a better clinical outcome We next investigated the effect that the presence of neutralizing anti-IFN-Abs has on disease severity. Patients with at least one neutralizing Ab against an IFNa subtype displayed significantly lower disease activity (as measured by the British Isles Lupus Assessment Group [BILAG] global score [GS]) compared with patients without anti-IFN-Abs in circulation or patients with non-neutralizing anti-IFN-Abs ( Figure 2A). Notably, 5 out of 6 patients with neutralizing anti-IFN-Abs that had active disease (GS R 5) at the time of sampling displayed a consistently reduced GS in the follow-up clinic appointments, suggesting that the generation of neutralizing anti-IFN-Abs precedes amelioration of disease ( Figure S3A). The analysis of organ involvement is depicted in Figure S2. Renal, skin, and musculoskeletal involvement was more common in patients with non-neutralizing Abs than in patients negative for these Abs.
To understand the stability of the anti-IFN-Abs, we assessed the kinetics of anti-IFN-Ab production, circulating IFNa levels, and disease activity in a longitudinal cohort (30 patients with SLE) over a 10-year period (cohort's demographics is presented in Table S2). The presence of high titers of neutralizing anti-IFN-Abs mirrored a reduction of serum pan-IFNa protein to undetectable levels. The prolonged presence of neutralizing anti-IFN-Abs together with a consistently low IFNa concentration also paralleled a persistent inactive clinical score (Figure 2B). Patients with non-neutralizing anti-IFN-Abs in circulation present with high levels of serum IFNa and a more severe disease activity ( Figure 2C). We also observed reduced titers of anti-dsDNA autoAbs in patients with neutralizing anti-IFN-Abs but no changes in C3 levels between the different groups ( Figures S3B and S3C).
Follow-up analysis of organ involvement showed that both the negative and non-neutralizing groups experienced more disease flares in the renal, musculoskeletal, skin, and hematological systems compared with patients with neutralizing anti-IFN-Abs (Figure S3D). One individual in the neutralizing anti-IFN-Ab group maintained a B score in renal activity; however, this patient had consistently high Ab titers and neutralizing capacity with undetectable serum IFNa for the entire duration and displayed inactive disease in all other organ systems.
The bioactivity of IFNa from the sera of non-neutralizing anti-IFN-Ab and anti-IFN-Ab-negative patients was similar, confirming that non-neutralizing anti-IFN-Abs do not neutralize circulating IFNa ( Figure S3E). These results suggest that nonneutralizing anti-IFN-Abs may stabilize circulating IFNa levels as previously suggested for other cytokines. 34-36 Patients lacking anti-IFN-Abs present active disease over time ( Figure 2D). Neither the titers of anti-IFN-Abs nor IFNa serum levels were affected by treatment regime (Figures S3F-S3H).

Restored B cell populations in patients with SLE with neutralizing anti-IFN-Abs
Patients with SLE are known to present with a variety of B cell abnormalities, including increased frequencies of immature, DN B cells and plasmablasts and a decrease in Bregs. [1][2][3] Previous work by us and others has demonstrated that the level of exposure to IFNa determines immature B cell fate. 5,6,37 Whereas exposure of immature B cells to low-moderate concentrations of IFNa simultaneously expand both Bregs and plasmablasts, high concentrations of IFNa (observed in patients with SLE) biases B cell differentiation toward pro-inflammatory plasmablasts and plasma cells. 6 To evaluate whether the presence of neutralizing anti-IFN-Abs is associated with a normalization of the B cell frequencies and their responses, we quantified ex vivo B cell subset frequencies in patient groups defined by the presence or absence of neutralizing and non-neutralizing anti-IFN-Abs and controls (Table S3). Anti-IFN-Ab-negative patients showed a significant increase in immature, DN (CD27 À IgD À ) and plasmablast(CD27 + IgD À CD38 hi ) B cells and a reduced frequency of unswitched memory (USM; CD27 + IgD + ) and classswitched memory (CD27 + IgD À CD38 low ) B cells compared with controls ( Figures 3A and 3B; gating strategy in Figure S4A).
Patients with neutralizing (IC50 > 100) anti-IFN-Abs have similar B cell subset frequencies to controls except for classswitched memory (CD27 + IgD À CD38 lo ) B cells. In contrast, patients with non-neutralizing (IC50 < 100) anti-IFN-Abs display the same degree of altered subset frequencies as anti-IFN-Abnegative patients ( Figure 3C). We show no differences in the frequencies of T follicular helper cell (T FH ) subsets (circulating [cT FH ] or activated [aT FH ]) between patients and controls ( Figures S5A and S5B). No differences were detected in CD4 + CXCR5 À PD-1 + T peripheral helper cells (TPH) frequencies, previously described to be expanded in patients with SLE and to be drivers of disease activity 38 between controls and any group of patients with SLE ( Figure S5C). This supports a direct role of anti-IFN-Abs in normalizing B cell subset frequencies rather than indirectly via modifications to the T FH or TPH compartment.
To establish whether B cells from patients with SLE with anti-IFN-Abs have regained the capacity to differentiate into Bregs (hereafter defined as IL-10 + B cells), we stimulated peripheral blood mononuclear cells (PBMCs) from patients with SLE and controls with CpGC for 72 h to induce IFNa production by plasmacytoid DCs (pDCs) and IL-10 + B cell differentiation, as previously shown by our group. 39 There was a significant decrease in IL-10 + B cell frequencies in anti-IFNa-autoAb-negative and nonneutralizing anti-IFN-Ab patients but not in patients with neutralizing anti-IFN-Abs compared with controls ( Figure 3D).

IFNa-induced immature and plasmablast B cell expansion is inhibited by IgG from patients with SLE with neutralizing anti-IFN-Abs
In response to viral infections, pDCs rapidly produce IFNa that drives B cell maturation into plasma cells producing Abs against viral antigens. 5 In view of the recent findings showing the detrimental effect of neutralizing anti-IFN-Abs in patients with COVID-19, it is important to understand the impact of neutralizing anti-IFN-Abs on ''nascent'' IFNa produced by challenged pDCs and how this affects healthy B cell differentiation. PBMCs from controls were stimulated with CpGC and cultured, respectively, with purified total IgG from patients with SLE with no Abs (negative), with non-neutralizing anti-IFN-Abs, and with neutralizing anti-IFN-Abs. Healthy allogeneic IgG was used as a control. An Fc blocking reagent was included to remove the IgG-mediated activation of FcR-expressing immune cell subsets. Inclusion of the Fc blocking reagent did not alter frequencies of immature B cells, plasmablasts, or Blimp1 + or IL-10 + B cells compared with CpGC stimulation alone ( Figure S5D).
IgG from patients containing neutralizing anti-IFN-Abs significantly downregulated ISG expression in cultured CpGC-stimulated control PBMCs, confirming their ability to inhibit IFNa downstream signaling ( Figure 4A). IgG from patients with neutralizing anti-IFN-Abs reduced the levels of IFNa in culture supernatants, whereas non-neutralizing anti-IFN-Abs increased IFNa concentrations ( Figure 4B). Control IgG or IgG from patients with SLE lacking anti-IFN-Abs show no effect.
Addition of control IgG, or IgG from anti-IFN-Ab-negative patients with SLE, did not impair the CpG-induced expansion of immature B cells and plasmablasts. IgG isolated from patients with neutralizing anti-IFN-Abs significantly reduced the expansion of immature B cells and plasmablasts, with the latter also confirmed by a reduced Blimp1 expression, compared with non-neutralizing anti-IFN-Abs ( Figures 4C and 4D). IgG from patients with non-neutralizing anti-IFN-Abs increased the frequencies of immature B cells and plasmablasts (and Blimp1 + B cells), suggesting that these autoAbs stabilize IFNa and enhance B cell responses to IFNa. Only IgG from patients with neutralizing anti-IFN-Abs halted the CpGC-driven IL-10 + B cell expansion, further confirming their neutralization capacity and the requirement of optimal IFNa levels for Breg differentiation ( Figure 4E).

DISCUSSION
In summary, we report that a subset of patients with SLE harbor neutralizing anti-IFN-Abs that can modulate B cell responses and are associated with a better disease outcome. This is in contrast to patients with non-neutralizing low titers of anti-IFN-Abs, which appear to stabilize IFNa in the blood and expand circulating frequencies of DN memory B cells and plasmablasts. It has been previously shown that CD11c + DN B cells are pathogenic in SLE. Although we have not specifically measured this population, it is interesting that the DN B cells were reduced in patients with neutralizing anti-IFN-Abs. Future work with a larger cohort of patients quantifying frequencies of CD11c + Tbet + DN B cells and their association with the development of neutralizing versus non-neutralizing anti-IFN-Abs are warranted. The association of non-neutralizing anti-IFN-Abs with high IFNa concentrations is intriguing. It has been previously suggested that in certain cases, including more recently in patients with COVID-19, 40 circulating autoAbs can increase the half-life of the molecule they bind, possibly through the uptake and release of immune complexes by the neonatal Fc receptor on endothelial cells. 41,42 In addition, autoAb binding may change the conformation of IFNa and lead to more efficient binding to the receptor.
The cellular source of these anti-IFN-Abs remains unknown. It is plausible to speculate that anti-IFN-Abs could be produced either by a pool of memory B cells that, upon IFNa . Ex vivo frequencies of (A) immature (Imm; CD24 hi CD38 hi ) and mature (Mat; CD24 int CD38 int ) B cells gated within the naive (CD27 À IgD + ) subset; (B) naive (N; CD27 À IgD + IgM + ) unswitched memory (USM; CD27 + IgD + ) and double-negative (DN; CD27 À IgD À ) B cells gated within the total CD19 + population; and (C) class-switched memory B cells (CSMs) and plasmablasts/plasma cells (PB/PC) gated within the CD27 + IgD À subset. All values are given as the percentage of total CD19 + population (gating strategy in Figure S3A). (D) Representative contour plots and graphs show frequencies of IL-10 + B cells within the total CD19 + population following 72 h in vitro CpGC stimulation of PBMCs isolated from patients with SLE with neutralizing anti-IFN-Abs (n = 7) or non-neutralizing anti-IFN-Abs (n = 12), anti-IFN-Ab-negative patients with SLE (n = 16), and healthy individuals (n = 13). *p < 0.05, **p < 0.01, ***p < 0.001 by non-parametric Kruskal-Wallis test with Dunn's multiple comparison. Error bars are shown as mean ± SEM. Data are representative of at least 3 independent experiments.
6 Cell Reports Medicine 4, 100894, January 17, 2023 Report ll OPEN ACCESS challenge, such as following infection, induce the production of anti-IFN-Abs. However, our findings showing a persistent presence of autoAbs matching dramatically reduced levels of IFN-I and clinical score suggest a role for long-lived plasma cells in the production of these Abs. Due to the reduced disease severity afforded by the presence of high titers of neutralizing anti-IFN-Abs, none of these patients were treated with rituximab, which would abrogate circulating IFNa-specific memory B cells.
Our findings are relevant in the current COVID-19 pandemic, where anti-IFN-I-Abs and impaired IFN signaling have been associated with higher susceptibility for serious illness. 29 When administering anti-IFN-I blockade therapy (e.g., anifrolumab, a human monoclonal Ab to IFN-I subunit 1), measuring levels of anti-IFN-Abs in patient sera would be clinically more practical than measuring the IFN-I PBMC gene signature for prescreening patients. Anifrolumab has been now approved as a therapy for patients with SLE with moderate and severe disease. It would be important to pre-screen patients to establish the presence and neutralization capacity of anti-IFN-Abs and exclude these patients from this treatment.

Limitations of the study
The scale of our analysis of B cells/PBMCs from these patients was limited by restricted sample availability due to the COVID-19 pandemic.
As discussed, the cellular source of neutralizing and nonneutralizing anti-IFN-Abs remains to be determined. Unfortunately, Report ll OPEN ACCESS this type of analysis requires a substantial amount of peripheral blood, which we are unable to obtain both because patients with SLE are frequently lymphopenic and our ethics only permit us to draw 25 mL blood per clinic visit.
The pathogenic role of non-neutralizing autoAbs through stabilization of IFNa levels in the circulation was suggested through indirect evidence; this has yet to be formally proven. Our study was also unable to discriminate autoAb avidity from concentration.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Professor Claudia Mauri (c.mauri@ucl.ac.uk).

Materials availability
This study did not generate new unique reagents.
Data and code availability d Data reported in this paper will be shared by the lead contact upon request. d This paper does not report original code. d Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.  43,44 Patients were positive for antinuclear antibody (ANA or anti-double-stranded DNA (dsDNA) antibodies.
Exclusion criteria were an age under 18, history of treatment with rituximab, participation in any interventional trial and pregnancy. Patients with severe CNS lupus, congestive heart failure, a history of cancer, severe glomerulonephritis, a history of recurrent or active infections such as HIV, tuberculosis, hepatitis B/C viruses and a history of demyelinating disease, for example, multiple sclerosis or optic neuritis, were also excluded.
All participants underwent a structured examination by a rheumatologist. BILAG SLE criteria were recorded. 45 Disease duration was defined as the time (years) from the first point at which an SLE diagnosis was documented in the patient records, until inclusion into this cohort. Disease activity was assessed by the British Isles Lupus Assessment Group (BILAG), a standardized disease activity assessment. Blood tests are performed as part of routine clinic visits and include: anti-dsDNA (double stranded DNA) autoantibody titers, complement C3 levels, complete blood counts, urea/electrolytes/serum creatinine, leukocyturia and haematuria, and a dip stick test for protein with a protein:creatinine ratio requested if + or more is recorded. Fever was defined as a body temperature above 38.5 C, weight loss as a loss of at least 5% of body weight, and cytopenia as leukopenia <3 G/L or thrombocytopenia <100 G/L. Leukopenia related to drugs or benign ethnic causes were not scored in the BILAG.
To provide numerical scores, we used a previous weighting system that assigned a score of 9 to active manifestations (grade A in the BILAG), 3 to grade B manifestations, 1 to grade C manifestations, and 0 to grade D and E manifestations. We used the sum of these scores as a summary index (possible range 0-72). 45 Low lupus disease activity was defined as a BILAG global score of %5 with no activity in major organ systems and no hemolytic anemia or gastrointestinal activity, without new lupus disease activity compared with the previous assessment, and with corticosteroid treatment up to 7.5 mg/day of prednisone (treatment with an immunosuppressant and/or hydroxychloroquine (HCQ) were allowed). 46 In the case of multiple serum samples at different dates for the same patient, only the oldest one was included and established as day 0. The kinetics of anti-IFN-a-autoantibody levels over time were determined in all the available serum samples of patients who tested positive for anti-IFN-Ab more than once.
Demographics, clinical characteristics, routine laboratory testing and therapeutic regimen (reported in Tables 1, S2 and S3) were collected from electronical medical files of the visit to the clinic recorded on the day blood was drawn (Day 0). Healthy controls from UCLH and UCL were enrolled after informed consent.

Cell and cell lines Primary cells
Prior to experiments, PBMCs from healthy controls and SLE patients were stored in liquid nitrogen in cryovials containing 10% DMSO and 90% fetal calf serum (FCS) and were thawed in warm RPMI 1640 (Sigma-Aldrich) supplemented with 10% FCS and 100 IU/mg penicillin/streptomycin (Sigma-Aldrich). For primary cell cultures, PBMCs were seeded in 96-well plates at a density of 5 3 10 6 cells/mL in RPMI 1640 supplemented with 10% FCS and 100 IU/mg penicillin/streptomycin. PBMCs were stimulated with 1mM CpGC ODN 2395 (InvivoGen), then incubated for 72 hrs at 37 C and 5% CO 2 . For IgG cultures, PBMCs from healthy donors were cultured at 5 3 10 6 cells/mL with 1mM CpGC, sodium azide-free Fc blocking reagent (Miltenyi) and 200 mg/mL IgG isolated from healthy donors or SLE patients. Cell lines HEK293 cells were thawed and plated in DMEM (Lonza) containing 10% FCS and Antibiotic/Antimycotic mix (Corning) into 10mL tissue culture plates. Cells were incubated at 37 C and 5% CO 2 for 72h. Cells were washed with PBS and detached with warm trypsin (Corning) for 1 min. Trypsin was inactivated with the medium and cells pelleted, then seeded at 250,000 cells per 3mL well of a 6 well plate in DMEM containing 10% FCS and Antibiotic/Antimycotic mix. Following overnight incubation at 37 C 5% CO 2 cells were transfected with 4mg DNA, 8mL lipofectamine (Invitrogen) and 250mL OptiMem reduced serum media (Gibco).
HEK-Blue cells were thawed and plated in DMEM containing 10% heat-inactivated FCS and Antibiotic/Antimycotic mix in 10mL culture plates. Cells were incubated at 37 C and 5% CO 2 and maintained and subcultured in growth medium supplemented with 30 mg/mL blasticidin and 100 mg/mL zeocin (Invitrogen). Cells were passaged upon reaching a 70-80% confluency.