Nuclear Antigen–reactive CD4D T Cells Expand in Active Systemic Lupus Erythematosus, Produce Effector Cytokines, And Invade The Kidneys
Mar 16, 2022
Contact: ali.ma@wecistanche.com
Dimas Abdirama et al
Cistanche DHT for kidney disease, click here to get the sample
Systemic lupus erythematosus is a systemic and chronic autoimmune disease characterized by a loss of tolerance towards nuclear antigens with autoreactive CD4D T cells implicated in disease pathogenesis. However, very little is known about their receptor specificity since the detection of human autoantigen-specific CD4D T cells has been extremely challenging. Here we present an analysis of CD4D T cells reactive to nuclear antigens using two complementary methods: T cell libraries and antigenreactive T cell enrichment. The frequencies of nuclear antigen-specific CD4D T cells correlated with disease severity. These autoreactive T cells produce effector cytokines such as interferon-g, interleukin-17, and interleukin-10. Compared to blood, these cells were enriched in the urine of patients with active lupus nephritis, suggesting an infiltration of the inflamed kidneys. Thus, these previously unrecognized characteristics support the role of nuclear antigen-specific CD4D T cells in systemic lupus erythematosus.
Systemic lupus erythematosus (SLE) is a systemic and chronic autoimmune disease characterized by loss of tolerance toward nuclear antigens due to defective disposal of biological waste such as apoptotic material containing ribonucleoproteins and nucleosomes.1 Autoantibody directed against nuclear antigens is characteristic of SLE and usually occurs before the overt disease.2 In addition, autoreactive CD4þ T cells are implicated in the pathogenesis of SLE by promoting autoantibody production by B cells and directly propagating organ damage in inflamed target organs.3 Although the detection of nuclear antigen–reactive T cells has been reported previously, the evidence for their existence remains circumferential until now and it is presently unclear how they interplay in the pathogenesis.4–9 Understanding how these cells participate in autoimmunity will be critical to the design of effective treatment.
The detection of autoantigen-specific T cells is hindered by their extraordinarily low frequencies in circulating blood.10 Previous studies of human naive CD4þ T-cell repertoires have used tetramer-based technology to effectively provide precise information on T-cell responses to specific antigen epitopes; however, this method requires prior knowledge of the human leukocyte antigen types of the donor.11–13 Alternative techniques to simultaneously detect and enumerate rare T-cell populations with reactivity to different antigens without prior knowledge of human leukocyte antigen types were developed by using libraries of amplified polyclonal T cells14 and by enrichment of CD154-expressing T cells after stimulation with an antigen termed ARTE (antigen-reactive T-cell enrichment).15
CD4þ T cells mainly exert their function through secretion of cytokines upon antigenic activation16 and by propagating tissue inflammation such as in lupus nephritis (LN) which represents one of the most serious complications of SLE.17,18
Analysis of the T-cell repertoire present in renal biopsies from patients with LN revealed oligoclonality of kidney-infiltrating CD4þ T cells, indicating accumulation of antigen-specific T cells in inflamed kidneys.19 Our previous study revealed urinary T cells in patients with active LN as a precise biomarker that resembles the phenotype of intrarenal cells.20,21 Although these cells are enriched for CXC chemokine receptor 3 and CC chemokine receptor 5 expressing T cells indicative of a T helper cell 1 (Th1) cell population and contain CD154þ T cells suggestive of recent antigen encounter,22,23 their antigen specificity toward nuclear antigens remains unclear In this study, we present an analysis of CD4þ T cells reactive to nuclear antigens in a total of 17 healthy individuals, 12 subjects with inactive SLE, and 20 subjects with active SLE by using T-cell libraries and ARTE techniques to analyze the frequencies of autoreactive CD4þ T cells in peripheral blood and urine of patients with LN as well as the cytokine production.
RESULTS
Nuclear antigen–reactive CD4D T cells are expanded inactive SLE as determined using a T-cell library and ARTE
Given the high theoretical sensitivity to detect rare antigen-specific T cells, libraries of amplified CD4þ T cells were generated and interrogated for reactivity against nuclear antigens. Expanded T-cell libraries from 12 patients with SLE and 6 healthy controls were stimulated with the canonical nuclear antigens SmD1, RNP70, histone, Ro, and La in the presence of antigen-presenting cells followed by the measurement of cell proliferation in response to antigenic stimulation (Supplementary Figure S1A). Because the amplified T-cell blasts responded differently to the antigens, a donor-specific z score was used as a proliferation threshold and the degree of the stimulation index (response of stimulated cells relative to unstimulated cells) to the superantigen Staphylococcus enterotoxin B (SEB) served as the inclusion criterion (Supplementary Figure S1B–D). In addition to nuclear antigens, we included transthyretin as an irrelevant control autoantigen (n ¼ 8 patients with SLE; n ¼ 5 healthy controls) (Figure 1a). The detection of antigen-specific cells was validated by challenging a series of positive-scored microcultures with the respective antigen and analyzing the expression of antigen-specific T-cell activation markers CD154 and interferon (IFN)-g by using fellow cytometry24 (Supplementary Figure S1E). Using Poisson distribution,14 the frequencies of antigen-specific T cells were enumerated. Notably, the median frequencies of some nuclear antigen– reactive CD4þ T cells were increased in patients with active SLE when compared with healthy subjects (SmD1: P ¼ 1; RNP70: P ¼ 0.04; histone: P ¼ 0.35; Ro: P ¼ 0.02; La: P ¼ 0.04) and patients with inactive SLE (SmD1: P ¼ 0.79; RNP70: P ¼ 0.04; histone: P ¼ 0.23; Ro: P ¼ 0.04; La: P ¼ 0.006) (Figure 1b). The total number of SLE-associated autoreactive CD4þ T cells varied among individuals in a range of 0 to 185.5 cells per 1 million cells, with the highest frequencies in patients with active SLE (Figure 1c). Nuclear antigen–reactive CD4þ T cells were detectable in all patients with active SLE. These cells were also detectable in 5 of 6patients with inactive SLE, while only 3 of 6 healthy controls presented circulating autoreactive T cells. Interestingly, patients with active SLE also showed a broader variety of autoreactive T-cell responses, with each patient having reactive T cells against at least 3 nuclear antigens. In comparison, the target portfolio was less variable in patients with inactive SLE and healthy controls. Taken together, circulating nuclear antigen–reactive T cells are not exclusive for SLE, yet inactive SLE, their number is greater and their target structures are more variable.
Because autoreactive T cells play a role in propagating autoimmune responses, we compared the number of autoreactive T cells with disease severity, assessed by the SLE disease activity index (SLEDAI) score. Only the frequencies of CD4þ T cells reactive to RNP70, Ro, and La positively correlated with SLEDAI (RNP70: P ¼ 0.04, r ¼ 0.4865; Ro: P ¼ 0.005, r ¼ 0.6299; La: P ¼ 0.01, r ¼ 0.5651) (Figure 1d). In contrast, the frequency of T cells reactive to transthyretin was found at a fairly similar level across all 3 groups (Figure 1e), thus dismissing a possible experimental bias due to general non–antigen-specific T-cell hyperreactivity inactive SLE.
T-cell libraries offer a series of advantages including high sensitivity to detect very rare cells within a low number of input cells. However, this technique relies on the expansion of cells, which may alter the cell composition and provides only a limited possibility to phenotypically characterize the reactive cells. Therefore, we validated our observations with the ARTE method. The ARTE protocol requires larger amounts of input cells, which is especially challenging in patients with SLE who often present themselves with lymphopenia and anemia. Because of this limitation and also to increase the expected readout of reactive cells, we pooled all 5 canonical nuclear antigens and used this pool to stimulate peripheral blood mononuclear cells (PBMCs) before enrichment of CD154-expressing cells (n ¼ 15 patients with SLE; n ¼ 6 healthy controls). Using a combination of the surface markers CD154 and CD69, T cells reactive to SLE-associated nuclear antigens could be identified within a background population (Figure 1f). Although background expression of CD154 and CD69 was abundantly observed in all individuals, patients with active SLE had significantly increased frequencies of CD154þCD69þ T cells when stimulated with nuclear antigens compared to background stimulation (P ¼ 0.008), whereas healthy individuals and patients with inactive SLE had similar frequencies between antigen-stimulated and nonstimulated cells (healthy individuals: P ¼ 0.44; patients with inactive SLE: P ¼ 0.56) (Figure 1g). Furthermore, by subtracting the background frequencies, the frequencies of nuclear antigen– reactive T cells were higher in patients with active disease than in healthy individuals and patients with inactive disease (P ¼ 0.02 and P ¼ 0.11, respectively) (Figure 1h). The frequency of background-subtracted nuclear antigen–reactive T cells thereby positively correlated with disease activity (P ¼ 0.004, r ¼ 0.5978), suggesting their potential role in SLE pathogenesis (Figure 1i). Moreover, CD4þ T cells reactive to a recall antigen Candida albicans MP65 demonstrated similar frequencies across the donors (Figure 1j), excluding the possibility of a general bias toward higher frequencies of antigen-reactive T cells in patients with active SLE.
Collectively, our data generated using the T-cell library and ARTE methods show expansion of autoantigen-specific CD4þ T cells in patients with active SLE. Both techniques are comparable as indicated by similar median stimulation index values between antigen-stimulated and nonstimulated T-cell responses (median stimulation index—T-cell library: 1.857; ARTE: 1.614; P ¼ 0.33), although the ARTE technique revealed higher frequencies of nuclear antigen–reactive CD4þ T cells than did the T-cell library technique (Figure 1k).
Nuclear antigen–reactive T cells in healthy individuals are bona fife autoreactive CD4D T cells
Both T-cell library and ARTE techniques demonstrated the detection of autoreactive T cells in healthy individuals, which represent a possible origin for the development of human autoimmunity.25 However, their number was not significantly different from the background frequencies, leaving it uncertain whether healthy subjects indeed have to circulate nuclear antigen–reactive CD4þ T cells. To unravel the antigen-specificity of autoreactive T cells in healthy subjects, we generated single T-cell clones from CD154þCD69þ CD4þ Tcell precursors isolated from healthy individuals after stimulation of PBMCs with individual nuclear antigens and a control antigen derived from Aspergillus fumigatus lysate. A considerable number of SLE-associated autoantigen-specific T-cell clones responded to their corresponding antigens after antigen restimulation as highlighted by the coexpression of IFN-g with CD154 (number of specific clones—Aspergillus fumigatus: 5 of 5; SmD1: 4 of 8; RNP70: 3 of 11; histone: 5 of 5; Ro: 4 of 9; La: 5 of 7) (Supplementary Figure S2A and B). Moreover, the antigen response (representatively shown for SmD1) was dose-dependent, emphasizing that nuclear antigen–reactive T cells in healthy individuals are bona fife autoreactive CD4þ T cells (Supplementary Figure S2C). Thus, we confirmed the existence of nuclear antigen–reactive T cells in healthy subjects within the background signal of the applied detection methods.
Cytokine production of nuclear antigen–reactive CD4D T cells
We next examined the cytokine production of nuclear antigen–reactive CD4þ T cells. CD154-expressing cells were stained intracellularly for effector cytokines such as IFN-g, interleukin (IL)-17, IL-4, and IL-10 after stimulation with a pool of SLE-associated nuclear antigens (Figure 2a). Frequencies of IFN-g–, IL-17–, and IL-10–producing CD154þCD4þ T cells were significantly increased in patients with active SLE (all cytokines: P ¼ 0.004), whereas the frequency of IL-4–producing autoreactive T cells remained indistinguishable from the background frequency (P ¼ 0.125) (Figure 2b). Moreover, absolute numbers of IFN-g– and IL- 10–producing autoreactive T cells were higher in patients with active SLE than in healthy individuals and patients with inactive SLE (Figure 2b). By subtracting the background frequencies, patients with active SLE had significantly higher frequencies of IFN-g– and IL-10–producing autoreactive T cells than did healthy individuals (IFN-g: P ¼ 0.003; IL-10: P ¼ 0.004) as well as of IFN-g–, IL-17–, and IL-10–producing autoreactive T cells than did patients with inactive SLE (IFN-g: P ¼ 0.002; IL-17: P ¼ 0.002; IL-10: P ¼ 0.005) (Figure 2c). In addition, the frequency of IFN-g– and IL-10– but not of IL-17– and IL-4–producing nuclear antigen– reactive T cells correlated with disease activity. However, only the frequency of IL-4–producing cells showed less correlation with disease activity (IFN-g: P < 0.0001, r ¼ 0.7574; IL-17: P ¼ 0.005, r ¼ 0.5889; IL-4: P ¼ 0.14, r ¼ 0.3717; IL- 10: P ¼ 0.0004, r ¼ 0.6977) (Figure 2d). Judged by the frequency of IFN-g producers, antigen-reactive CD4þ T cells were mainly of the Th1 lineage. In contrast, IL-17, as well as IL-10 producers, were detected only at lower frequencies.26 Compared to autoantibody levels, the frequencies of cytokine-producing CD4þ T cells did not correlate with the concentration of anti-double-stranded DNA and with the serum titer of antinuclear antibodies (Supplementary Figure S3A and B). No other correlation between nuclear antigen–reactive CD4þ T cells and autoantibodies was observed except for La-reactive CD4þ T cells toward La autoantibodies (Supplementary Figure S3C).
Detection of nuclear antigen–reactive CD4D T cells in the urine of patients with active SLE and LN The population of nuclear antigen–reactive Th1 cells is suspected to invade the renal tissue, where they might encounter their almost omnipresent cognate antigen.27 By investigating the T-cell receptor (TCR)-b repertoire of CD4þ T cells isolated from peripheral blood compared with those isolated from the urine of 5 patients with SLE and active LN, we observed oligoclonal skewed TCR repertoires of urine-derived cells (Figure 3a). The cumulative frequency of the 20 most abundant T-cell clones was significantly higher in urine corresponding to a lower diversity index (Supplementary Figure S4A). To test whether these cells also demonstrate reactivity to a set of SLE-associated nuclear antigens, we generated libraries consisting of urinary CD4þ T cells isolated from 3 patients with SLE and active LN (Supplementary Figure S4B). Urinary CD4þ T cells were metabolically and functionally active cells indicated by the ability to proliferate upon stimulation with SEB but with signs of exhaustion shown by a reduced proliferation capacity after polyclonal activation (Supplementary Figure S4C and D). Because of these circumstances, only microcultures with a SEB stimulation index of >4 were included for the enumeration of urinary CD4þ T cells specific to the individual nuclear antigen. In parallel to urinary CD4þ T cells, the frequency of nuclear antigen–reactive CD4þ T cells in peripheral blood was determined, allowing a direct comparison of the respective frequencies within the same individuals (Figure 3b). We found that inactive LN, the frequencies of urinary nuclear antigen–reactive CD4þ T cells were higher than those of peripheral blood. To confirm this observation, we labeled urinary cells isolated from 4 active LN donors with carboxyfluorescein succinimidyl ester and mixed the labeled cells with donor-matched PBMCs to allow a direct comparison of antigen-reactive cells in 1 assay using the ARTE technique. The frequency of CD154þCD69þ T cells in stimulated and nonstimulated samples of donor-matched urinary and peripheral blood cells was used to assess whether nuclear antigen–reactive T cells are enriched in urine as a proxy for the kidney by calculating the stimulation index (Figure 3c). Although the urinary T-cell library and ARTE methods differed in their median stimulation index to the nuclear antigens (stimulation index—urinary T-cell library: 17; ARTE: 2.79; P ¼ 0.4) (Figure 3d), analysis of autoreactivity in peripheral and urinary CD4þ T cells from both methods (in total n ¼ 7) demonstrated an increased nuclear antigen-specific stimulation index in urine (P ¼ 0.0156) (Figure 3e). These observations suggest that nuclear antigen–reactive CD4þ T cells invade and accumulate in the inflamed kidneys. Thus, local tissue inflflammation in SLE might be an antigen-specific phenomenon and directly modulated by the infiltrating autoreactive T cells.
DISCUSSION
A break of tolerance against nuclear antigens is the hallmark of SLE, and besides B cells, autoreactive T cells are thought to play a central role in its pathogenesis. However, as the detection of (auto)antigen-specific, CD4þ T cells still pose a major challenge in autoimmunity research, to date, little is actually known about nuclear antigen–reactive CD4þ T cells in SLE.
Recently, novel methods for the detection and enumeration of antigen-specific CD4þ T cells have been developed. The T-cell library approach exploits prior polyclonal expansion of T cells to enable the detection of rare antigen-specific T-cell clones.14 This strategy offers high sensitivity and allows working with limited amounts of cells. Using this technique, we were able to demonstrate reactive T cells against 5 canonical nuclear antigens in patients with active SLE. The disadvantages of the library-based detection are, besides being labor-intensive, that outgrowth or loss of certain clones during the expansion phase cannot be excluded and the limited options for phenotypic analysis of the cells.
Standard fellow cytometry is limited by the number of detectable events, which usually restricts the analysis of samples to the population with frequencies above 0.01%. Our observation using the T-cell library revealed that the frequency of circulating autoreactive CD4þ T cells for a particular autoantigen in patients with a disease flare was w20 cells in a million cells, representing a frequency of 0.002%; therefore, the detection of autoantigen-specific CD4þ T cells using standard, nonmanipulated fellow cytometry is almost impossible. Bacher et al. introduced pre-enrichment of CD4þ T cells that are reactive to particular antigens before acquisition on fellow cytometry, known as the ARTE method.15 Adapting this approach for the detection of nuclear antigen-specific CD4þ T cells, we were able to confirm our findings and demonstrate the expansion of nuclear antigen– reactive CD4þ T cells in active SLE. Notably, the frequency of T cells reactive with transthyretin and C. Albicans was indifferent between healthy controls and patients with SLE with varying disease activity. Therefore, the observed expansion of autoreactive T cells is likely not a result of general hyperreactivity inactive SLE but of the expansion of antigen-specific T cells.
Prior reports have already demonstrated the existence of autoreactive T cells in SLE, primarily using proliferation, cytokine production, or upregulation of activation markers, albeit without being able to quantify a clear population of these cells.4–6,8 The best evidence on the frequency of nuclear antigen-specific T cells exist for reactivity against the U1 small ribonucleoprotein, a characteristic autoantigen target in mixed connective tissue disease and found only in a fraction of patients with SLE. U1 small nuclear ribonucleoprotein– reactive CD4þ T cells were determined using limiting dilution and Enzyme-Linked Immuno Spot Assay (ELISPOT) and were reported to occur at a frequency of (40 to 250) - 106 T cells and (20 to 60) 106 PBMCs, respectively,7,9 which is of the similar magnitude as the frequency of autoreactive cells in our study. In our work, we investigated the reactivity against several characteristic SLE-associated nuclear antigens. Interestingly, although all patients with active LN showed increased frequencies of anti-nuclear antigen–reactive CD4þ T cells, patients differed in their target portfolio, and the reactivity was aimed against a broader set of antigens compared to patients with inactive SLE. The biological significance of this is presently unknown. Different CD4þ T cell reactivities may associate with certain clinical manifestations of different reactivities that may be redundant for the disease pathogenesis.
Similar to other autoimmune diseases, we were also able to demonstrate the existence of autoreactive CD4þ T cells in healthy individuals,28,29 albeit at much lower frequencies compared to patients with active SLE. These cells were undetectable by comparing the number of activated cells with or without antigen stimulation, even with such highly sensitive techniques as T-cell libraries or ARTE methods. However, by single-cell cloning, we were able to demonstrate that the “background” signal indeed contained autoantigen-specific CD4þ T cells. Which specific events lead to the break of tolerance and increasing frequencies of autoreactive T cells remains speculative; nevertheless, our data demonstrate that expansion of autoreactive clones does play a part in the pathogenesis of SLE.
Circulating nuclear antigen–reactive CD4þ T cells mainly produced IFN-g and, to a lesser extent, IL-17 and IL-10. This cytokine profile together with the weak/no correlation with autoantibody production suggests that the role of nuclear antigen–reactive CD4þ T cells may not be the exclusive provider of B cell help. IFN-g has been shown to be indispensable in the pathogenesis of LN.30,31 Th1 cells have been shown to be recruited into the inflamed renal tissue in SLE,20,32 making nuclear antigen–reactive T cells prime candidates to invade the kidneys, where they would encounter their ubiquitous cognate autoantigen. Urinary T cells have been previously described to reflect disease activity and mirror the phenotype of intrarenal cells in LN20,33; thus, we used urinary T cells as a proxy for renal T cells. Urinary T cells revealed a restricted TCR variability, which is in line with observations in kidney biopsies,19,34 indicating enrichment of certain T-cell clones in the inflamed kidney tissue. Among urinary T cells, we were able to detect nuclear antigen–reactive T cells, and their frequency was enriched compared to circulation T cells. Consequently, it seems likely that nuclear antigen–reactive cells directly participate in the propagation of local organ damage.
In summary, we here demonstrate that nuclear antigen–reactive CD4þ T cells are expanded inactive SLE; they are phenotypically mainly IFN-g–producing Th1 T cells and invade inflamed target organs such as the kidney.
METHODS
Detailed methods are given in Supplementary Material.
Subjects and blood and urine samples
Blood samples were obtained from 17 healthy individuals, 12 patients with inactive SLE (SLEDAI score <10), and 20 patients with active SLE (SLEDAI score >10). In addition, urine samples were obtained from 12 patients with active SLE and LN. All subjects had given their informed consent on the basis of the ethical approval obtained by the institutional review board and ethics authorities atCharité – Universitätsmedizin Berlin (EA 1/342/12, EA 1/098/07, and 1/036/16). PBMCs and urinary cells were prepared using a Ficoll-Hypaque density gradient.
Antibodies and fellow cytometry
Depending on the experiments, several surface molecules were stained on antigen-stimulated cells in different combinations of the following monoclonal antibodies: anti-CD3, -CD4, -CD8, -CD14,-CD20, -CD69, and -CD154; to discriminate live and dead cells, the LIVE/DEAD Aqua kit (Life Technologies Ltd., Paisley, UK)was used. To stain the intracellular cell compartment, cells were fixated with2% (v/v) paraformaldehyde for 15 minutes at room temperature and permeabilized with BD FACS Permeabilizing Solution 2 (BD Biosciences, San Jose, CA); the following antigens were stained intracellularly by using a standard protocol: CD154, IFN-g, IL-2, IL-4, IL-10, and IL-17. Samples were acquired on BD FACSCanto II and BD LSRFortessa(BD Biosciences) fellow cytometers at the Flow Cytometry Core Facility Deutsches Rheuma-Forschungszentrum Berlin by using BDFACSDiva software (BD Biosciences). Flow cytometry data were analyzed using FlowJo software (FlowJo v10, Three Star, Ashland, VA).
Antigen and antigen pool preparation
We used the following antigens to stimulate CD4þ T cells in the T-cell library assay: 0.5 mg/ml of human SNRPD1 (SmD1) recombinant protein (Biorbyt Ltd., Cambridge, UK), 50 ng/ml of humanSNRP70 (RNP70) recombinant protein (Abcam Plc., Cambridge, UK), 0.5 mg/ml of natural human histone protein (Abcam Plc.), 0.5mg/ml of human SS-A/Ro recombinant protein (kindly provided byEuroimmun AG, Lübeck, Germany), 0.5 mg/ml of human SS-B/Larecombinant protein (kindly provided by Orgentec DiagnostikaGmbH, Mainz, Germany), 1 mg/ml of SEB (Sigma-Aldrich ChemieGmbH, Steinheim, Germany), 1 mg/ml of PepTivator Candidaalbicans MP65 (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), and 0.5 mg/ml of human transthyretin recombinant protein (ATGen, Seongnam, South Korea).
Peripheral and urinary T-cell libraries
The protocol to generate amplified peripheral and urinary T-cell libraries was adopted and adapted as previously described.14 Brieflfly, 200,000 peripheral CD4þ T cells or 500 to 2000 urinary CD4þ T cells were isolated using anti-human CD4 MicroBeads (Miltenyi Biotec GmbH) according to manufacturer’s instructions. The purity of CD4þ T-cell fraction was routinely checked using fellow cytometry on the basis of CD3þCD4þ expression, where the purity always reached >99% and 70% to 75% for peripheral blood and urinary samples, respectively. In parallel, antigen-presenting cells were prepared by collecting CD3 cells from PBMCs after depletion with anti-human CD3 MicroBeads (Miltenyi Biotec GmbH) and cryopreserved. After 1 to 2 weeks of culture, fractions of amplified CD4þ T cells were distributed into 96-well plates, depending on the number of antigens to be analyzed. Before stimulation with the antigens, cells were rested for at least 4 days. On the day of stimulation, antigen-presenting cells were thawed and distributed into the T-cell culture in a ratio of at least 1 antigen-presenting cell to 100 CD4þ T cells. CD4þ T-cell libraries for negative control (unstimulated cells) and positive control (cells stimulated with SEB) were always included in the experiments. CD4þ T cells were stimulated with the antigen for 4 days, and proliferation was determined using a standard [3 H]-thymidine protocol.
Amplified T-cell blasts responded differently to the antigens displayed by highly heterogeneous scintillation CPM; thus, a normalization of the proliferation threshold determined as donor-specific z score was necessary. The z score was calculated as 5 differences of 75th and 25th percentile above the median CPM of unstimulated microcultures. Cells in microcultures stimulated with SEB served as a positive control. Of note, microcultures with a SEB stimulation index of <5 for peripheral blood samples, or 4 for urinary samples, were excluded because it indicated poor cell viability. Enumeration of the precursor frequency of antigen-specific CD4þ T cells was calculated using numbers of negative microcultures according to Poisson distribution and expressed per 1 million cells, as described previously.14 The use of Poisson distribution is based on the probability that at least 1 CD4þ T cell is present in a single microculture of T-cell libraries when the radioactive signal was detected after antigen-specific stimulation.
Peripheral and urinary ARTE methods
We followed the protocol that uses the CD154 MicroBead Kit (Mil-tenyi Biotec GmbH).35 Briefly, cells were stimulated for 7 hours with antigens and then labeled with anti-human CD154-biotin antibodies and anti-biotin microbeads according to the manufacturer’s instructions. Labeled cells were loaded onto calibrated MS columns (Miltenyi Biotec GmbH) to enrich CD154-expressing cells. Labeling of cells with carboxyfluorescein diacetate N-succinimidyl ester (SigmaAldrich Chemie GmbH) was done following a standard protocol.
Generation of single-cell clones
Single-cell clones were generated from antigen-reactive cells, which were enriched by CD154 expression following the ARTE protocol.CD154-expressing cells were filtered with a 30-mm preparation filter (Miltenyi Biotec GmbH) and resuspended in 1 ml of 20% (v/v)cold phosphate-buffered saline, 0.5% bovine serum albumin, and 2mM ethylenediaminetetraacetic acid. Cells were single cells sorted for the CD154þCD69þ phenotype, and single sorted cells were cultured in single wells of 96-well plates in the presence of feeder cells. Cell cultures were maintained for 3 to 4 days until several clones were visible. One day before restimulation, antigen-presenting cells were thawed and distributed into the T-cell culture in a ratio of at least 1 antigen-presenting cell to 100 CD4þ T cells. Cells were stimulated with 1 mg/ml of antigen, and unstimulated clones served as a negative control.
Next-generation sequencing-based TCR repertoire analysis
Genomic DNA was isolated from peripheral CD4þ T cells and urinary cells using the AllPrep DNA/RNA Micro Kit (Qiagen, Venlo, The Netherlands). Recombined TCR-b locus was amplified following the protocol as described previously.36 Reads were processed using IMSEQ,37 and equal clonotypes were clustered and further analyzed.
Statistics
Statistical tests were performed with GraphPad Prism software (Prism 8, GraphPad Software Inc., La Jolla, CA). Bonferroni adjustment for multiple comparisons was applied. Mann-Whitney U test and Wil-Coxon signed-rank test were used in the experiments with independent and dependent data set, respectively. After Bonferroni adjustment for multiple comparisons, the calculation of critical P values was corrected because it is based on the number of planned comparisons. P values <0.01 were considered statistically significant with the following indication: *P < 0.01, **P < 0.005, and ***P < 0.0005. Correlation analyses were performed using Spearman rank correlations by showing the absolute SLEDAI values instead of ranked values to better provide an understanding of the correlation between the number of cells and disease activity. For correlation analysis, P values <0.5 were considered statistically significant with the following indication: *P < 0.05, **P < 0.01, and ***P < 0.001. When background frequencies were higher than the frequencies of nuclear antigen–reactive CD4þ T cells, the frequencies without background were defined as zero. A log(x þ 1) transformation was applied to the data set when it included zero values. Supplemental patient data and statistical data are available in Supplementary Tables S1 and S2.
ACKNOWLEDGMENTS
We thank Toralf Kaiser and Jenny Kirsch (Flow Cytometry and CellSorting Facility, Deutsches Rheuma-Forschungszentrum Berlin) forassistance with flflow cytometry and cell sorting. Support for thesestudies was provided by grants from the DeutscheForschungsgemeinschaft within the Sonderforschungsbereich 650 toGR and by grants from Deutsche Gesellschaft für Nephrologie andClinical Scientist Program of Charité – Universitätsmedizin Berlin andBerlin Institute of Health to PE.
