Autoantibody-positive healthy individuals with lower lupus risk display a unique immune endotype

Abstract

Background:

Autoimmune diseases comprise a spectrum of illnesses and are on the rise worldwide. Although anti-nuclear antibodies (ANA) are detected in many autoimmune diseases, up to 20% of healthy women are ANA+ and most will never develop clinical symptoms. Further, disease transition is higher among ANA+ African Americans compared to European Americans.

Objective:

To determine the immune features that might define and prevent transition to clinical autoimmunity in ANA+ healthy individuals.

Methods:

We comprehensively phenotype immune profiles of African Americans and European Americans who are ANA- healthy, ANA+ healthy, or have systemic lupus erythematosus (SLE) using single cell mass cytometry, next-generation RNA sequencing, multiplex cytokine profiling, and phospho-signaling analyses.

Results:

We found that SLE patients of both races displayed T cell expansion and elevated expression of Type I and II interferon pathways compared to both ANA- and ANA+ healthy individuals. We discovered a unique immune signature that suggests a suppressive immune phenotype and reduced CD11C+ autoimmunity-associated B cells in healthy ANA+ European Americans that is absent in their SLE or even healthy ANA- counterparts, or among African American cohorts. In contrast, ANA+ healthy African Americans exhibited elevated expression of T cell activation markers and higher plasma levels of IL-6 compared to healthy ANA+ European Americans.

Conclusions:

We propose that this novel immune signature identified in ANA+ healthy European Americans protects them from T cell expansion, heightened activation of interferon pathways, and disease transition.

Graphical Abstract

CAPSULE SUMMARY

We comprehensively define the immune phenotype in ANA+ healthy individuals, and provide the first description of a protective immune profile in T cells during pre-clinical autoimmunity.

INTRODUCTION

Autoimmune diseases, such as systemic lupus erythematosus (SLE), are driven by both environmental and genetic factors. Approximately 8% of the population has a classified autoimmune disease, often associated with the presence of anti-nuclear antibodies (ANA) (1, 2). ANAs and other autoantibodies can be detected up to 10 years before autoimmune disease onset. Indeed, ANAs are detected in about 20% of healthy people, particularly females, the elderly, and non-Caucasians (1, 3); however, the presence of autoantibodies alone does not predict the development of clinical symptoms (4–6). Whether a subset of ANA+ individuals possess protective cellular factors and mechanisms that prevent disease transition is unknown.

Significant effort has focused on understanding the mechanisms that drive autoimmune disease. Compared to SLE patients, ANA+ healthy individuals have lower levels of stem cell factor (SCF), B lymphocyte stimulator (BLyS) and Type I IFNs (IFNα and IFNβ), as well as higher levels of the regulatory cytokine, IL-1 receptor antagonist (IL-1RA) (7). However, we recently found that immune pathways are already dysregulated in ANA+ versus ANA- healthy individuals; ANA+ healthy subjects show a modest elevation of pro-inflammatory cytokines in serum (7). The frequencies of monocytes, B cells and T follicular helper cells were also elevated in ANA+ versus ANA- healthy individuals (7, 8). In addition, we discovered that some soluble mediators are elevated in serum approximately 3.5 years before SLE classification and even prior to ANA positivity, including Interleukin-5 (IL-5), IL-6, and Interferon (IFN)-γ (6). Other innate cytokines, IFN-associated chemokines (such as MIG/CXCL9), and BLyS increase ~10 months before SLE classification (6). Nevertheless, we lack a comprehensive evaluation of immunophenotypes and immune function for healthy ANA+ individuals versus SLE patients and ANA- subjects. As a result, it is not fully understood why only a subset of ANA+ healthy individuals will transition to disease.

In addition to ANA positivity and elevated soluble mediators, race and ethnicity contribute to the risk for developing autoimmune disease. For instance, systemic autoimmune rheumatic diseases often show a later onset and milder clinical presentation in those with European American (EA) versus African American (AA) ancestry (9). Genetic, environmental, and socioeconomic factors likely influence the diverse biological mechanisms that contribute to autoimmunity. Autoantibody-dependent IFN activation pathways, BLyS serum cytokine levels, and DNA methylation in naïve CD4+ T cells differ between AA and EA patients with SLE (10–15). Although ancestral backgrounds display distinct genetic factors and gene signatures (15), no studies have examined ANA+ healthy individuals by race. This analysis is needed to identify potentially clinically relevant, but unknown, mechanisms that regulate the transition from an ANA+ healthy status to SLE.

Here, we used an in-depth immune screening platform to identify regulatory and inflammatory immune features that are critical for the development of clinical autoimmunity. We examined populations that are at higher risk (AA) and lower risk (EA) for transitioning to SLE, and compared ANA+ healthy individuals with ANA- controls and SLE patients to discover putative regulatory mechanisms. Unexpectedly, EA ANA+ healthy subjects exhibited a unique immune suppression signature in T cells that was not present in EA ANA- controls or SLE patients, or the AA cohorts. Our results identify the first protective immune profile displayed during development of clinical autoimmune disease, and thus point to potential therapeutic avenues to activate a pathway that could delay or prevent disease transition.

METHODS

Study population and Autoantibody Screening.

All experiments were performed in accordance with the Helsinki Declaration and approved by the Institutional Review Board of the Oklahoma Medical Research Foundation. Healthy individuals were recruited through 15 health fairs, and screened initially for 11 serum autoantibody specificities using the Bioplex 2200® system (Bio-Rad Technologies). Positive individuals were also assessed by NOVA Lite indirect immunofluorescence (IIF) and QUANTA Lite ELISA using HEP-2 cells (Inova Diagnostics, Inc., San Diego, CA) following the manufacturer’s recommended protocols and cutoffs (7, 16, 17). IIF testing was performed by CAP-CLIA certified Morris Reichlin, MD Clinical Immunology Laboratory. Briefly, Bioplex 2200 ANA tested autoantibody specificities include dsDNA, chromatin, Ro/SSA, La/SSB, Sm, SmRNP, RNP, Centromere B, Ribosomal P, Scl-70, and Jo-1. All autoantibodies, except anti-dsDNA, were reported in antibody index (AI) units based on a fluorescent intensity range of 0–8. The manufacturer-specified cutoff was used to determine positivity (positive ≥1 AI) for all autospecificites, except for anti-dsDNA where semi-quantitative values were reported as IU/mL with positive ≥ 10 IU/mL. Confirmation assays included INNO-LIA ANA Update EIA testing (Innogenetics NV, Zwijnaarde, Belgium). ANA- and ANA+ healthy individuals had no probable autoimmune rheumatic disease determined by a connective tissue disease screening questionnaire (18), as well as review of medical and medication history. ANA+ healthy individuals were identified as having 1 or more of the 11 tested autoantibodies by Bioplex and no probable autoimmune disease.

ANA+ healthy individuals (n=24) were matched by gender, age ± 5 years and race to ANA- healthy controls (n=24) and SLE patients (n=24) (Supplementary Table 1). SLE patients met at least 4 ACR classification criteria and were assessed for disease activity by SELENA-SLEDAI (19). ANA- healthy controls, ANA+ healthy individuals and SLE patients were divided and analyzed by race, which was self-reported and verified using genetic ancestry informative markers (AIMs) (20). Peripheral blood mononuclear cells (PBMCs) were isolated using Lymphocyte Separation Medium (Mediatech, Inc. Manassas, VA) and stored in freezing media (20% human serum and 10% DMSO in RPMI) in liquid nitrogen until use. Plasma was also collected and stored at −80°C until testing.

Soluble mediator measurement.

Plasma levels of BLyS were assessed using enzyme linked immunosorbent assays (ELISA) per the manufacturer’s protocol and reported in pg/mL (R&D Systems, Minneapolis, MN). All other soluble mediators (sCD40L, CXCL13, eotaxin, G-CSF, GRO-α, ICAM-1, IFN-α, IFN-β, IFN-γ, IL-1α, IL-1β, IL-1RA, IL-10, IL-12p70, IL-13, IL-15, IL-17A, IL-2, IL-21, IL-23, IL-2Rα, IL-4, IL-5, IL-6, IL-7, IL-8, IP-10, leptin, LIF, MCP-1, MCP-3, MIG, MIP-1α, MIP-1β, NGF-β, PAI-1, PDGF-BB, RANTES, resistin, SCF, SDF-1a, sEselectin, sFASL, TGF-β, TNF-α, TNFRI, TNFRII, TRAIL, VCAM-1, and VEGF) were assessed using xMAP (ProcartaPlex™) multiplex assays (Invitrogen/ThermoFischer Scientific, Waltham MA) and run on Bio-plex 200 suspension array reader (Bio-Rad). All data were normalized across plates using a serum control (Cellgro human AB serum, Mediatech) and reported as both FI over serum control and concentration (pg/ml). As previously described (21), soluble mediators were excluded from analysis if ≥50% of measurements reported were below the lowest level of detection resulting in 38 soluble mediators that passed quality control.

CyTOF immunophenotyping.

Assays were performed in the Human Immune Monitoring Center at Stanford University. Antibody clones, staining protocols and gating strategies are previously described (22). Briefly, PBMCs were thawed, washed, and resuspended in CyFACS buffer (PBS supplemented with 2% BSA, 2 mM EDTA, and 0.1% sodium azide), and viable cells were counted by Vicell. Viable cells (1.5 million cells/well) were stained with antibody-polymer conjugate cocktail (Supplementary Table 2). All antibodies were unconjugated and carrier free, and conjugated using polymer and metal isotopes from Fluidigm. Cells were washed and resuspensed in CyFACS buffer. The cells were resuspended in 2 ug/mL Live-Dead (DOTA-maleimide (Macrocyclics) containing natural-abundance indium). The cells were washed and resuspended in 2% paraformaldehyde in PBS and placed at 4°C overnight. The next day, cells were washed and placed in eBiosciences/ThermoFischer permeabilization buffer (1x in PBS) and incubated on ice. Cells were then washed twice in PBS and acquired on CyTOF (Fluidigm, San Francisco, CA). Data analysis was performed using Cytobank by gating on intact cells based on the iridium isotopes from the intercalator, then on singlets by Ir191 vs cell length, then on live cells (Indium-LiveDead minus population) (Supplementary Fig. 1), followed by cell subset-specific gating as shown in Figure 2 by tSNE and manually, as previously published (22).

Figure 2. tSNE analysis pipeline identifies 27 phenotypically distinct populations in PBMCs.

(A) 20 cell surface marker expression is shown using dimensionality reduced t-SNE plots from PBMC data (110,00 cells) derived from 72 samples. Dot plots are individually colored by channel using ArcSinh-5 transformed expression values. (B) A dot plot and density map are shown depicting the density of cells and are numbered according to phenotypic subset. (C) A heatmap summary of the expression values of all 33 cell surface markers are used to distinguish identified cell subsets. Marker values are displayed on a color scale ranging from blue (levels below the mean) through white (levels equal to the mean) to red (levels greater than the mean) using a column Z-score. (D) Density maps depicting EA and AA ANA-, ANA+ and SLE patient PBMC t-SNE plots created using all 33 surface markers are plotted. All plots were derived from cumulative data from 12 individuals per group.

Phosphoepitope flow cytometry.

These assays were performed at the Human Immune Monitoring Center at Stanford University. Briefly, PBMCs were suspended at 0.1×10⁶ viable cells and stimulated with T cell receptor (TCR) stimuli (CD3/CD28 Dynabeads) or B cell receptor (BCR) stimuli (anti-human IgG, anti-human IgM and H₂O₂) and incubated at 37°C for either 30 minutes (TCR) or 4 minutes (BCR). The PBMCs were then fixed with paraformaldeyde, and permeabilized with methanol. Cells were bar-coded using a combination of Pacific Orange and Alexa-750 dyes (Invitrogen, Carlsbad, CA) and then stained with the following antibodies (all from BD Biosciences, San Jose, CA): CD3 Pacific Blue, CD4 PerCP-Cy5.5, CD20 PerCp-Cy5.5, CD33 PE-Cy7, CD45RA Pacific Orange, p38 FITC, pPLCγ2 PE, pSTAT-5 PE-TX-Red, and pERK1/2- APC. Cells were collected (100,000/sample) using DIVA 6.0 software on an LSRII flow cytometer (BD Biosciences). Data analysis was performed using FlowJo v9.3 by gating on live cells based on forward versus side scatter profiles, then on singlets using forward scatter area versus height, followed by cell subset-specific gating of the 90^th percentile. Fold change was calculated over basal phospho-protein levels to assess differences following BCR and TCR stimulation.

Cell Sorting and bulk RNA-sequencing.

PBMCs from 36 subjects were stained with antibodies for CD3 (UCHT1), CD19 (SJ25-C1), HLA-DR (G46–6), CD14 (61D3), CD16 (3G8), CD56 (NCAM16.2), CD66b (G10F5) and CD66b-CD19-CD3+ T cells, CD66b-CD3-CD19+ B cells, CD66b-CD3-CD19-CD56-HLA-DR+CD14+CD16+/− monocytes were sorted using a FACS Aria III (BD Biosciences). RNA was isolated using TRIZOL reagent (Invitrogen/Thermo Fischer), purified with Direct-zol RNA microprep kit (Zymo Research, Irvine, CA), and quantitated using 2100 Bioanalyzer (Agilent, Santa Clara, CA). QuantSeq 3’ mRNA-sequencing library prep kit for Illumina (FWD) (Lexogen, Vienna, Austria) was used to create cDNA libraries, amplify and sequence 3 million reads/sample using NextSeq 550 (Illumina, San Diego, CA). Libraries were evaluated for poor sample quality using FASTQC (version 0.11.8), and adaptors and contaminating non-specific reads removed using BBTools (version38.56) (23, 24). Reads were aligned to the GENCODE release 28 transcriptome using STAR (version 2.5.3a) (25, 26). BAM files were then converted to a gene count matrix using StringTie (version 1.3.6) (27). Counts were normalized using the DESeq2 package (version 1.24.0) in R and transformed using variance stabilizing transformation for heatmap visualizations (28, 29). All low expressed genes were removed prior to analysis. A pre-selected gene list of 500 genes was used for differential expression analysis between disease groups by race. Significant differentially expressed genes were calculated between two groups of either ANA-, ANA+, or SLE using a Kruskal-Wallis test with P-value <0.05 considered significant.

Virus IgG Detection ELISAs.

The CMV IgG, EBV-EA IgG, EBV-VCA IgG, HSV1 IgG and HSV2 IgG levels were measured according to manufacturer’s instructions using 1:21 dilution of sera (Zeus Scientific, Inc., Branchburg, NJ) (30). The anti-EBNA-1 IgG ELISA was also performed according to the manufacturer’s specifications using a 1:101 dilution of sera (EuroImmun, Eubeck, Germany). Samples determined as equivocal were rerun to determine positive, negative or equivocal status.

Data Analysis and Statistics.

tSNE analysis were performed using Cytobank (31). FCS files were uploaded to Cytobank and gated off live intact singlet cells. To generate tSNE plots, 22,000 events were used per sample with data for 33 cell surface markers. Concatenated files of 110,000 cells were used for representative tSNE images and cell subset profiling. Frequencies of cell subsets were exported from tSNE for cell number calculations and analysis. Traditional bivariant gating was performed in Cytobank. Cytokine data were non-normally distributed; therefore, continuous data were analyzed using the Kruskal-Wallis test with Wilcoxon-Mann-Whitney two-tailed test for two-group comparisons. The q values were calculated using the qvalue R package (version 3.3.3) to correct for multiple comparisons and estimate the false discovery rate to control for the expected proportion of incorrectly rejected null hypotheses. All analyses, heatmaps and plots were performed and generating using GraphPad Prism 6.0 for Windows (GraphPad Software, San Diego, CA) or TIBCO Spotfire 6.0.1 (TIBCO Software Inc., Boston, MA). The 3D bar graphs were generated in R version 3.2.2 using the latticeExtra, RColorBrewer, and gridExtra packages.

RESULTS

European American and African American ANA+ healthy individuals have distinct autoantibody specificities

We recruited and screened 1035 healthy subjects for autoantibodies, using both indirect immunofluorescence and luminex bead-based assays that measure common lupus, Sjogren’s, systemic sclerosis and myositis autoantibodies, as previously described (7, 16). Approximately 25.6% of the cohort were ANA+, with an ANA titer ≥120 defined by indirect immunofluorescence.

Using Bioplex 2200 ANA testing, 41 EA (7.32% of total EA) and 12 AA individuals (7.84% of total AA) were ANA+, having at least 1 of 11 autoantibody specifications, yet without a diagnosed autoimmune rheumatic disease. Autoantibody specificity varied between EA and AA ANA+ healthy individuals. In EA ANA+ healthy individuals, anti-ribonucleoprotein (anti-RNP) was the primary autoantibody (41.5% in EA versus 20.0% in AA), followed by antibodies against centromere B (17.0%), dsDNA (14.6%), Ro (14.6%) and La (12.2%). Anti-dsDNA antibody was the most prevalent autoantibody in AA ANA+ healthy individuals, with 50.0% of subjects testing positive, followed by anti-RNP (20.0%), anti-La (20.0%) and anti-Ro (8.3%).

We identified EA (n=12) and AA (n=12) individuals that were ANA+ and healthy by Bioplex as defined above and matched them to healthy ANA- controls (n=24) and SLE patients (n=24) according to age (±5 years), sex, and race (Supplementary Table 1). All ANA+ subjects, except for one RNP+ AA ANA+ individual were also positive by IIF and/or ELISA (Supplementary Table 3). All control participants completed a connective tissue disease screening questionnaire (CSQ), to assess whether participants had possible autoimmune rheumatic disease. No probable disease was found in ANA+ subjects or ANA- controls (18). Autoantibody specificities of the ANA+ healthy individuals were selected to reflect the EA or AA cohort (Supplementary Table 4). To reduce variability associated with gender, and given that there were fewer ANA+ healthy males (11.3%), all individuals selected for this study were female.

The selected EA SLE patients had a higher prevalence of anti-Ro/SSA antibodies (33.3%), and AA SLE patients had a higher prevalence of anti-RNP antibodies (50.0%) (Supplementary Tables 5&6). As had been previously reported (9), AA SLE patients had a more active clinical disease presentation, evidenced by higher prevalence of renal disease (58.3% versus 16.7%) and a higher average SELENA-SLEDAI score (5.5 versus 2.8 in European American patients). At the time of this study, there were no significant differences in proteinuria (0.0% EA versus 8.3% AA) between groups, and no EA or AA SLE patients had lymphopenia, increased DNA binding, CNS issues, or hematuria.

European American ANA+ healthy individuals have reduced numbers of T cells, NK cells and autoimmunity-associated B cells

To identify biologic mechanisms that regulate autoimmune disease progression, we collected PBMCs and plasma from matched subjects of EA and AA ancestry for ANA- controls, ANA+ healthy individuals and SLE patients (Figure 1, six groups of 12 individuals). PBMCs were assessed using (1) mass cytometry (CyTOF), to detect differences in immune cell frequencies, (2) phospho-flow cytometry, to detect differences in signaling responses, and (3) RNA sequencing, to detect differences in gene expression in T cells, B cells, and monocytes (Figure 1A). In addition, we assessed the plasma levels of 51 soluble mediators associated with innate immunity, adaptive immunity, regulation, growth, adhesion, and migration (Figure 1B). Serum was used to assess prior environmental exposure to herpesviruses.

Figure 1. Schematic workflow for ANA+ healthy individual biologic analysis.

The workflow is broadly divided into 2 steps. First, 72 samples consisting of EA and AA ANA- healthy individuals, ANA+ healthy individuals, and SLE patients were matched by age, race, and gender. (A) PBMCs were collected and used for immunophenotyping by mass cytometry, T-cell receptor and B-cell receptor signaling analysis by phospho-flow, and gene expression analysis by 3’ QuantSeq. (B) Plasma was collected for soluble mediator analysis of 51 different metabolites using multiplex bead-based assays and ELISAs, and serum was used for viral IgG ELISAs. Immunophenotyping markers are colored according to cell association/pathway: T cell (blue), B cell (orange), myeloid cell (green), NK cell (purple) and chemokine receptors (red). Soluble mediators are also grouped by cell association/pathway using color: B cells (red), T cell (blue), Th1 (purple), Th2 (orange), regulatory (green), adhesion (red), precursor/growth factors (light blue), apoptotic (purple), myeloid/neutrophil (orange), and adipose (black).

We investigated cell lineages that are essential for the development of ANA positivity and autoimmune disease transition by race. Briefly, we designed a CyTOF panel of 33 metal isotype-tagged monoclonal antibodies specific for cell lineage markers that discriminate the major immune cell subsets and subpopulations (22). Additional markers were included to distinguish the activation status and homing properties of specific cell subsets. Cell frequencies were visualized both by a high-dimensionality reduction method (tSNE) and by a standard hand-gating scheme that identified 55 different cell subsets (Figure 2). The tSNE analysis incorporated over 110,000 cells and distinguished 27 phenotypically distinct clusters (Figure 2A and B). Marker expression of gated phenotype clusters is summarized by median intensity in a heatmap (Figure 2C). For major immune cell subsets, tSNE and manual gating found similar differences in frequencies between ANA- controls, ANA+ healthy individuals, and SLE patients (Supplementary Tables 7–14).

B cell (CD3-CD19+) and monocyte (CD3-CD19-HLA-DR+CD11c+CD14+) frequencies were elevated in EA ANA+ healthy individuals compared to ANA- healthy controls (Supplementary Fig. 2, Supplementary Tables 7–10) (7). The frequency of CD8+ T cells (CD3+CD56-CD8+) was elevated in SLE patients compared to ANA+ healthy individuals in both the EA and AA groups (Supplementary Fig. 2). Some cell populations, including dendritic cells (DCs) (CD3-CD19-HLA-DR+CD11c+CD14-), plasmacytoid DCs (pDCs) (CD3-CD19-HLA-DR+CD11c-CD123+), and NK cells (CD3-CD19-CD56+), showed decreased frequencies in AA SLE patients compared to ANA- or ANA+ healthy controls. (Supplementary Fig. 2).

To determine the source of these different frequencies in EA and AA ANA+ healthy individuals, total cell subsets/mL were back-calculated from cell frequencies (Supplementary Table 11–14). Cell numbers (total cells/mL) were reduced in EA ANA+ healthy individuals compared to both ANA- healthy controls and SLE patients (Figure 3A). Although the cell numbers were reduced in AA ANA+ healthy individuals versus SLE patients, they were similar to ANA- healthy controls (Figure 3A). ANA titers did not correlate with decreases in cell numbers (Supplementary Fig. 3).

Figure 3. Calculated cell numbers indicate elevated T cells in SLE patients and suppressed T cells in EA ANA+ healthy individuals.

Cells numbers were calculated from cell subsets using frequencies and total cell counts. Cell numbers for (A) Total cells/mL, (B) B cells (CD3-CD19+), (C) Monocytes (HLA-DR+CD11c+CD14+CD16±), (D) Autoimmunity-associated B cells (CD3-CD19+IgD-CD27-CD11c+), (E) CD4+ T cells (CD3+CD56-CD4+CD8-), (F) CD8+ T cells (CD3+CD56-CD4-CD8+), (G) CD4+ Memory T cells (CD3+CD56-CD4+CD45RA-), (H) CD8+ Memory T cells (CD3+CD56-CD8+CD45RA-), (I) CD4+ Central Memory T cells (CD3+CD4+CCR7+CD45RA-), (J) CD8+ Central Memory T cells (CD3+CD8+CCR7+CD45RA-), (K) CD4+ Effector Memory T cells (CD3+CD4+CCR7-CD45RA-), (L) CD8+ Effector Memory T cells (CD3+CD8+CCR7-CD45RA-), (M) CD4+ Naïve/Effector T cells (CD3+CD56-CD4+CD45RA+), (N) CD8+ Naïve/Effector T cells (CD3+CD56-CD8+CD45RA+), (O) NK cells (CD3-CD19-CD56+) and (P) NKT cells (CD3+CD19-CD56+) are shown. Black correlates to ANA- healthy controls, blue to ANA+ healthy individuals, and red to SLE patients. *p<0.05, **p<0.01, ***p<0.001. Kruskal-Wallis test with two-tailed Mann-Whitney for multiple comparisons. Mean± standard deviation (SD) shown.

Although the frequencies of monocytes and B cells were elevated in EA ANA+ healthy individuals compared to controls, the number of total monocytes and B cells were similar in the six groups (Figure 3B–C). However, autoimmunity-associated B cells (ABCs), characterized as CD11c+CD27-IgD-, were reduced in EA ANA+ healthy individuals relative to controls (Figure 3D, Supplementary Table 13&14).

T cell numbers were significantly increased in both AA and EA SLE patients compared to ANA+ healthy individuals (Figure 3E–F, Supplementary Table 13&14). Specifically, both CD4+ and CD8+ T cells in memory, naïve, and effector T cell subsets such as Th1-type (CXCR3+) and Th17-type (CCR6+CD161+) were elevated in SLE patients compared to ANA+ healthy individuals (Figure 3E–N, Supplementary Table 13&14). Further, NK cell subsets were also increased in SLE patients compared to ANA+ healthy individuals in both races, whereas NKT cells were only elevated in EA SLE patients (Figure 3O–P, Supplementary Table 13&14). These data further support the involvement of T cells and NK cells in SLE pathogenesis.

Intriguingly, NK cell and T cell populations were decreased in ANA+ versus both ANA- healthy controls and SLE patients in the EA cohort but not the AA cohort. These results imply that reduced levels of T cells, NK cells, and ABCs protect EA ANA+ healthy individuals from transitioning to SLE and might contribute to the reduced risk of EA versus AA populations.

Cell-surface activation markers show increased expression in SLE patients

To uncover potential functional differences, we assessed all 55 immune cell populations for surface expression of activation markers (CD86, HLA-DR, CD38, ICOS), inhibitory receptors (CD85j, CD94, CD33, PD-1), and chemokine receptors (CCR6, CCR7, CXCR3, CXCR5) (Supplementary Tables 15–16). CD85j was more highly expressed on conventional DCs, monocytes, and CD4+ T cells from EA ANA+ versus ANA- healthy individuals (Figure 4A–D, Supplementary Tables 15–16). The activation marker CD86 showed higher expression on B cells, primarily transitional B cells and non-switched memory B cells, from SLE patients versus ANA+ healthy individuals of EA ancestry (Figure 4E–H). pDCs, NK cells, and NKT cells showed higher expression of activation markers in SLE patients versus ANA+ and ANA- controls of AA ancestry (Figure 4I–K, Supplementary Tables 15–16). Thus, activation markers were elevated in SLE patients, but not ANA+ healthy individuals for both EA and AA cohorts.

Figure 4. Cell subset marker expression identify activated cell subsets in SLE patients and elevated regulatory marker expression in ANA+ healthy individuals.

All 55 cell subsets were manually gated and assessed for frequencies of activation and regulatory surface markers. (A) t-SNE of t-SNE plots were used to illustrate specific expression differences among immune cell populations. (A) CD85j expression on EA ANA-, ANA+ and SLE patients is shown using t-SNE plots with red indicating elevated expression and blue low expression, along with dot plot of significant differences in (B) cDCs, (C) monocytes and (D) CD4+ T cells. CD86 expression of B cells in EA subjects is shown by (E) t-SNE plot, and dot plots in (F) B cells, (G) transitional B cells, and non-switched memory B cells. Differences in pDCs in AA subjects is shown by (I) CD38 expression in t-SNE plot, and dot plots depicting (J) CD86 expression in EA subjects and (K) CD38 expression in AA subjects. B cell CD24 expression of AA subjects is shown via t-SNE plot and dot plot (M,L). *p<0.05, **p<0.01. Kruskal-Wallis test with two-tailed Mann-Whitney for multiple comparisons. Mean± SD shown.

Some receptors associated with immune regulation were significantly decreased in AA ANA+ healthy individuals and SLE patients compared to AA ANA- controls. These include B cell expression of CD24, which is associated with regulatory B cell subsets (Figure 4L,M) as well as T cell, monocyte, and DC expression of the regulatory receptor CD94, which binds to HLA-E (Supplementary Tables 15–16). These data support a potentially important role for inhibitory receptors in attenuating pre-clinical autoimmunity.

Phospho-signaling dysregulation in T cell signaling pathways of ANA+ healthy individuals

To assess the relationship between the T cell and B cell phenotypes of these cohorts and T-cell receptor (TCR) and B-cell receptor (BCR) signaling, we evaluated pERK1/2, p38, pPLC_γ2 and pSTAT5 in PBMCs by flow cytometry following treatment with TCR (anti-CD3/CD28) or BCR (anti-IgM/IgG F(ab’)2) stimuli (Figure 1A, Supplementary Table 17). Innate immune cells (non-B/T cell subsets) had elevated basal levels of pERK1/2 and p38 in EA ANA+ healthy individuals and SLE patients, respectively, compared to ANA- controls (Supplementary Fig. 4 and 5, Supplementary Table 18–19). No significant differences were identified in either the EA or AA cohort following BCR stimulation (Supplementary Fig. 5). Following TCR stimulation, memory CD8+ T cells from EA ANA+ healthy patients had elevated pERK1/2 compared to SLE patients and ANA- controls (Supplementary Fig. 4 and 5, Supplementary Table 18). In AA SLE patients, basal levels of pERK1/2 and p38 were increased in CD4+ and CD8+ T cells (Supplementary Fig. 3–4 and Table 18). AA SLE patient memory CD4+ and CD8+ T cells appear to have elevated basal levels of phosphorylated proteins that were not amplified, but slightly decreased, in response to TCR stimulation. These data suggest that these cells were already activated and maximally utilizing this signaling cascade.

Soluble mediator levels trend lower in EA ANA+ healthy individuals

To examine peripheral levels of cytokines, chemokines and other soluble mediators, we assessed the levels of 51 soluble mediators in plasma (Figure 1B). Plasma soluble mediator levels trended lower in EA ANA+ versus ANA- healthy individuals (Figure 5A, Supplementary Table 20–23). In the AA cohort, ANA+ healthy individuals and SLE patients had elevated IL-6 levels compared to ANA- controls (Figure 5B). Monocyte chemoattractant protein 3 (MCP-3), produced primarily by monocytes, along with the B cell activating sCD40L and apoptosis associated sFas ligand (sFasL), were increased in EA SLE patients compared to ANA+ healthy individuals (Figure 5C–E). AA SLE patients also had elevated IFN-associated soluble mediators, including IP-10, MIG, BLyS and TNFRII, along with selectins (sEselectin), TNF-related apoptosis-inducing ligand (TRAIL), and IL-2 receptor alpha (IL-2Rα) (Figure 5F–L). SCF was the only cytokine elevated in both EA and AA SLE patients compared to both ANA- and ANA+ healthy controls (Figure 5M). Thus, SCF production is a critical indicator of SLE autoimmune disease that is independent of race.

Figure 5. SCF distinguishes SLE patients from ANA+ healthy individuals in both European and African Americans.

Pro-inflammatory soluble mediators were measured by multiplex or ELISA. (A) A heatmap summary of the plasma levels for each individual are shown. Soluble mediator levels are displayed on a color scale ranging from blue (protein levels below the mean) through white (protein levels equal to the mean) to red (protein levels greater than the mean) using a column z-score. Significant cytokines included, (B) IL-6, (C) MCP-3, (D) sCD40L, (E) sFasL, (F) IP-10, (G) MIG, (H) BLyS, (I) TNFRII, (J) sEselectin, (K) TRAIL, (L) IL-2Rα, and (M) SCF. *p<0.05, **p<0.01. Kruskal-Wallis test with two-tailed Mann-Whitney for multiple comparisons. Mean±SD shown.

T cell immune suppression signature in EA ANA+ healthy individuals

To determine the mechanisms underlying the reduced numbers of T cells in EA ANA+ healthy individuals, we sorted T cells, monocytes and B cells and assessed RNA expression of 500 genes involved in cell regulation, HLA inhibition, cytokine regulation, apoptosis, STAT and cytokine pathways, adhesion, and cell activation (Figure 1A, full list in Supplementary Table 24). We found that IFN inducible genes and HLA Class I genes were markedly downregulated in T cells from ANA+ healthy individuals compared to those from ANA- controls and SLE (Figure 6A). Further, STAT1 upregulation and a pyroptosis signature, characterized by elevated CASP1, distinguished T cells from EA ANA+ healthy individuals versus ANA- controls and SLE patients (Figure 6A). Compared to the EA cohorts, the AA cohorts showed distinct and fewer changes in T cells from ANA+ versus ANA- healthy individuals, including upregulation of the activation element CD69 (Figure 6A). T cells from SLE patients of both races demonstrated increased expression of IFN inducible genes, HLA genes, and other pro-inflammatory cytokine genes (Figure 6A). Overall, these data suggest that immune suppression occurs in T cells of EA ANA+ healthy individuals, which may help to prevent clinical autoimmune onset.

Figure 6. EA ANA+ healthy individuals exhibit a gene expression signature in T cells consistent with virus-induced immune evasion.

RNA-sequencing of PBMCs for 36 subjects were used to assess differences in inflammatory and regulatory associated gene expression among groups. (A) A heatmap summary of individual gene expression in T cells is shown. Significant differences between groups are indicated by dots between columns. Genes significantly upregulated are indicated by a red dot and significantly downregulated by a blue dot. Dots between column one and two indicate a significant difference between ANA- and ANA+, between column two and three indicate significant differences between ANA+ and SLE, and after column three indicate significant differences between ANA- and SLE. Correlations between T cell numbers and gene expression were conducted between significant T cell, B cell and monocyte genes. Significant correlations were observed between (B) CD4+ T cell numbers and STAT4 T cell gene expression in European Americans, and (C) STAT1 and (D) IFNGR2 T cell gene expression and CD8+ T cell numbers of African Americans. All other significant correlations are recorded in table (E).

To determine the genetic pathways associated with T cell suppression in EA ANA+ individuals, we examined whether T cell numbers correlated with gene expression patterns in T cells, B cells and/or monocytes (Supplementary Fig. 6 and 7). Expression of STAT4, which is driven by Type II IFN, was positively associated with CD4+ T cell numbers in EA individuals (Figure 6B). In AA individuals, gene expression of both Type I and II IFN pathways (STAT1, IFNGR2) was positively correlated with CD4+ and CD8+ T cell numbers (Figure 6C–E). Regulation and activation of B cells and monocytes likely contribute to T cell expansion, and the expression of IFN inducible genes, CD85 regulatory molecules (LILRB2), CD86, and BLyS (TNFSF13B) positively correlated with T cell numbers (Figure 6E). Further, TGFBR1 expression negatively correlated with T cell numbers in AA individuals. These data suggest that activation of Type I and II IFN pathways in T cells and B cells, and reduced TGFβ signaling in monocytes, coincide with T cell dysregulation and clinical autoimmune disease.

CMV and EBV seroconversion is not associated with immune suppression

Our data so far show that ANA+ EA individuals have reduced T cell numbers, decreased plasma soluble mediators, dysregulated T cell signaling, and altered expression of HLA class I, Type I IFN, apoptosis and STAT1 associated genes in T cells. These features are reminiscent of virus immune-evasion strategies, and virus-induced immune suppression, often seen with members of the Herpesviridae family, specifically human cytomegalovirus (HCMV) (32, 33). Further, environmental factors, particularly herpesvirus infection and reactivation, most notably Epstein-Barr virus (EBV), are associated with the development of SLE (34). To determine whether persistent infection or reactivation of herpesviruses were involved in T cell suppression observed in EA ANA+ healthy individuals, we examined the IgG responses against HCMV, EBV early antigen (EBV-EA), EBV viral capsid antigen (EBV-VCA), EBV nuclear antigen (EBNA-1), herpes simplex virus 1 (HSV-1) and HSV2.

Consistent with previous reports, there was a higher frequency of SLE patients with antibodies directed against EBV-EA, in conjunction with VCA- and EBNA-1-directed antibodies, suggesting current or recent reactivation of EBV (Supplementary Fig. 8A). The frequency of ANA+ healthy individuals with antibodies to EBV-EA was higher when compared to healthy ANA- controls, but reduced when compared to SLE patients. However, EBV, CMV and HSV infection did not correlate with T cell numbers in EA individuals (Supplementary Fig. 8B–D). No correlations were observed between reduced systemic soluble mediators and viral positivity; however, IFN-β levels were higher among CMV+ individuals (Supplementary Fig. 8E). CD4+ and CD8+ CD57+ T cells, a marker of replicative senescence, were also elevated among CMV+ healthy subjects and SLE patients, which may influence disease symptoms and outcome.

Suppressed immune phenotype does not correlate with ANA specificity

Although we strongly attempted to perfectly match all subjects within this study, some differences remained. Differences were primarily found within the AA SLE patients, who had a trend higher in age, disease activity and slightly more DNA-specific autoantibodies (Supplementary Table 4–5, Supplementary Fig. 9) compared to the EA SLE patients and controls. In an attempt to control for these differences, linear regression analyses were done with age and SLEDAI in SLE patients between all significant differences in the study. When all SLE patients were assessed, a positive correlation was observed with CD4+ memory T cell numbers and CD4+ Th17-type T cells and SLEDAI (Supplementary Fig. 10) suggesting these cell subset increases may be influenced by higher SLEDAI in AA SLE patients. Further, AA SLE patients also had HLA-DR^Hi expression on transitional and naïve B cells that positively correlate with SLEDAI (Supplementary Fig. 10). sEselectin was the only finding that positively associated with age in AA SLE patients (Supplemental Fig. 11).

Finally, ANA+ healthy subjects or SLE patients were grouped by ANA specificity, either DNA-specific or RNA-specific autoantibodies. No significant differences were observed in either the ANA+ healthy individuals or the SLE patients in the significant findings of this study.

DISCUSSION

Understanding the early cellular events that occur during pre-clinical autoimmunity in the absence of confounding medication use is valuable for delineating important pathways in lupus pathogenesis. Studying healthy individuals with lupus-associated autoantibodies, in particular, is crucial for identifying not only early cellular dysregulation, but also regulatory pathways that may prevent clinical disease development and progression in most individuals. Our study provides the most comprehensive analysis of ANA+ healthy individuals to date.

Individuals with AA ancestry are at a higher risk for developing autoimmune diseases, like SLE, than those with EA ancestry (9). We discovered a unique and potentially protective immune profile associated with asymptomatic autoimmunity in EA individuals. Unlike EA ANA- healthy controls, EA ANA+ healthy individuals had a suppressed immune profile with decreased T cell, NK cell, and NKT cell numbers, a trend lower in cytokine levels, dysfunctional T cell signaling, and altered T cell transcriptional profiles. Although lymphopenia is common in SLE patients and often associated with disease activity (35), no EA ANA+ healthy subjects demonstrated or reported lymphopenia or leukopenia. T cell numbers also did not correlate with ANA titers.

Although the suppressed immune features in EA ANA+ healthy are reminiscent of certain viral infections (32, 33), no correlations were found with herpesvirus seroconversion or seroreactivation. However, an expanded panel to assess comprehensive infection history and chronic low-level viral infection could uncover possible viral involvement in immune suppression. Whether these EA ANA+ individuals are more susceptible to infection is unknown, but heightened TCR phospho-signaling suggests a normal or elevated T cell immune response to stimuli. Reduced T cell numbers may arise from a number of mechanisms, but we observed altered gene expression of apoptosis pathways and altered cytokine profiles, which could contribute to the decreases in circulating cell populations (36).

Suppression may be a form of regulation in response to early autoreactivity or a pathogenic result of unseen immune activation in EA ANA+ healthy individuals. However, elevated levels of the inhibitory receptor CD85j, which is expressed on DCs, monocytes and T cells, and a lack of elevated activation markers on B cells and pDCs in EA ANA+ healthy individuals suggest that enhanced regulatory pathways contribute to a suppressed immune phenotype. Further, disruption of regulatory pathways, such as PD-1, is associated with autoimmune disease development, and upregulation of these exhaustion receptors are associated with better disease outcomes (37–39). The absence of this suppression signature in AA ANA+ healthy individuals may be attributed to the known stronger inflammatory responses in individuals of African ancestry (40), which could contribute to more ANA+ subjects transitioning to SLE, or progressing more rapidly to SLE.

Previous work in ANA+ healthy individuals has found elevated CD86 expression on B cells and elevated Tfh and Treg frequencies compared to ANA- controls (8). Although this differs from our results, sample recruitment differed between studies, with ANA+ healthy subjects being referred to rheumatology clinics for positive ANA (using IIF with ANA ≥1:160) in the prior study (8). ANA testing was performed for various reasons including non-inflammatory arthritis/arthralgia (41%), family history of autoimmune disease (7%), uriticarial/non-specific rash (7%), and recurrent miscarriages or a child with neonatal lupus (13%), with some subjects already taking hydroxychloroquine (8.2%). In the current study, asymptomatic subjects were recruited through health fairs and ANA positivity was determined by Bioplex and confirmed by either ELISA or IIF. Further, almost half of healthy individuals were ethnic minorities in the previous study, which may contribute to a more activated profile (41). The importance of expanding and assessing ANA+ healthy individuals longitudinally cannot be understated, as determining if ANA positivity is maintained and whether EA ANA+ healthy individuals with a suppressed immune profile will ever develop disease, or whether this signature changes to a more activated profile seen in higher risk subjects, is of significant interest.

ANA positivity can be determined by a number of methods, and more recently testing ANA positive subjects by various ANA assays has identified discordant results (42–44). To aid in interpretation and future discussion, we assessed our subjects by three different methods (Bioplex, ELISA, and IIF) with all subjects ANA+ by Bioplex and one other method, with exception of one RNP+ AA individual who was only positive by Bioplex (specificity was confirmed by INNOLIA). The discordance of ANA testing results has led to questions of whether healthy subjects with Bioplex ANA positivity, which is specific indicator for RNA- and DNA- specific autoantibodies common to rheumatic diseases, but who are IIF- are false positives or subjects captured early in the disease process. To address this, Pérez et al followed 411 healthy Mediterranean subjects that were Bioplex+ and IIF- for 3 years (45). At follow up, 76% of subjects were positive by IIF and 87% had developed a classified autoimmune disease, suggested Bioplex has a greater sensitivity for autoimmune-specific antibody detection. Thus, although other ANA+ healthy studies have used ELISA or IIF as indicators of positivity, we utilized Bioplex as a primary indicator of early autoimmune specific ANA positivity (8, 17). Longitudinal studies assessing these results in other ethnic populations will assist in the understanding and importance of changes in ANA positivity by different methodology prior to disease diagnosis or lack thereof.

Other notable differences in ANA+ healthy individuals are derived from their resemblance to SLE patients, or lack thereof. We noted SCF, BLyS, IL-12p40 and Type I IFNs levels were significantly elevated only in SLE patients compared to ANA+ healthy individuals (7). We found that SCF was the only cytokine significantly elevated in both EA and AA SLE patients in this study. Stem cell factor is most commonly known as a niche component for hematopoietic stem cell renewal, and for driving the development and survival of mast cells (46–48). More recently, c-kit (the receptor for SCF), was found to be expressed by NK cells and DCs in the periphery suggesting an important role for SCF in these cell types (49, 50). Dendritic cells produced more IL-6 following SCF/c-kit signaling, which drives expansion of Th2 and Th17 immune responses, an environment commonly associated with SLE (51). Further, we found SLE patients had strong IFN signatures among various cell types, and elevated IFN associated soluble mediators were more visible in AA patients. IL-6, IFNγ and Th2 cytokines were previously found to be the first soluble mediators to increase in healthy subjects who transition to SLE (6). We found IL-6 was already elevated in AA ANA+ healthy individuals, and increased T cells numbers correlated with elevated gene expression in Type I and II IFN signaling pathways in SLE patients, suggesting dysregulation of these cytokines contributes to autoimmune pathogenesis.

The current study is not without its limitations. The cross-sectional study design allows us to only capture the immune profile of these subjects at one timepoint. A secondary cohort with a larger sample size to confirm the findings of this study and a longitudinal assessment of ANA+ healthy individuals will be important in verifying the findings and the changes in this immune profile over time and up to the point of transition. Further, the use of established SLE patients with controlled disease in this study may contribute to differences in the immune profile versus patients with new-onset or active disease.

Collectively, our findings highlight the importance of race on early autoimmune profiles, and identify a novel immune endotype with hallmarks of suppression in EA ANA+ healthy individuals. The racial differences in early autoimmune regulation likely influence which individuals may transition to SLE or other classified autoimmune diseases, and offer potential pathways to target for disease prevention.

CLINICAL IMPLICATIONS.

A protective immune signature identified in pre-clinical autoimmunity may be pivotal in finding novel autoimmune disease therapeutic targets and in identifying those at greatest risk for transition to clinical disease.

ABBREVIATIONS

SLE

systemic lupus erythematosus

ANA

anti-nuclear antibodies

SCF

stem cell factor

BLyS

B lymphocytes stimulator

IFN

interferon

IL-

interleukin

IL-1RA

IL-1 receptor antagonist

EA

European American

AA

African American

AI

autoantibody index

AIMs

ancestry informative markers

PBMCs

peripheral blood mononuclear cells

ELISA

enzyme linked immunosorbent assay

TCR

T cell receptor

BCR

B cell receptor

tSNE

t-distributed stochastic neighboring embedding

CMV

cytomegalovirus

EBV

Epstein barr virus

HSV

Herpes simplex virus

CSQ

connective tissue disease screening questionnaire

ABCs

autoimmunty-associated B cells

NK

natural killer

MCP-3

monocyte chemoattractant protein

sFASL

soluble Fas ligand

TRAIL

TNF-related apoptosis-inducing ligand

OMRF

Oklahoma Medical Research Foundation

HIMC

Human Immune Monitoring Center