ANCA autoantigen gene expression highlights neutrophil heterogeneity where expression in normal-density neutrophils correlates with ANCA-induced activation.

Abstract

ANCA vasculitis is an autoimmune disease with increased expression of the autoantigen genes, myeloperoxidase (MPO) and proteinase 3 (PRTN3), but their origin and significance of expression is less distinct. To clarify this, we measured MPO and PRTN3 messenger RNA in monocytes, normal-density neutrophils, and in enriched leukocytes from peripheral blood mononuclear cells. Increased autoantigen gene expression was detected in normal-density neutrophils and enriched leukocytes from patients during active disease compared to healthy individuals, with the largest difference in enriched leukocytes. RNA-seq of enriched leukocytes comparing active-remission pairs identified a gene signature for low-density neutrophils. Cell sorting revealed low-density neutrophils contained mature and immature neutrophils depending on the presence or absence of CD10. Both populations contributed to autoantigen expression but the frequency of immature cells in low-density neutrophils did not correlate with low-density neutrophil MPO or PRTN3 expression. Low-density neutrophils were refractory to MPO-ANCA induced oxidative burst, suggesting an alternative role for low-density neutrophils in ANCA vasculitis pathogenesis. In contrast, normal-density neutrophils were activated by MPO-ANCA and monoclonal anti-PR3 antibody. Normal-density neutrophil activation correlated with MPO and PRTN3 mRNA. Increased autoantigen gene expression originating from the mature low-density and normal-density neutrophils suggests transcriptional dysregulation is a hallmark of ANCA vasculitis. Thus, the correlation between autoantigen gene expression and antibody-mediated normal-density neutrophil activation connects autoantigen gene expression with disease pathogenesis.

Graphical Abstract

INTRODUCTION

Anti-neutrophil cytoplasmic autoantibody (ANCA)-associated vasculitis is a systemic autoimmune disorder characterized by damaging inflammation.^1-3 A critical component in the inflammatory cascade is the recognition by ANCA of their target autoantigens,^3-6 neutrophil granule proteins, myeloperoxidase (MPO) and proteinase 3 (PR3). The transcripts, MPO and PRTN3, encoding these autoantigens are produced during granulopoiesis and are low or absent in normal circulating granulocytes. However, in patients with ANCA vasculitis evidence demonstrates elevated MPO and PRTN3 mRNA in circulating peripheral blood cells.^7-12 One explanation for the presence of these transcripts is the premature release of progenitors into the peripheral circulation.¹⁰ Alternatively, defective epigenetic silencing could explain persistent expression or reactivation in mature circulating cells.¹³ In support of the latter, we identified an altered pattern of histone modifications in mature neutrophils from patients with ANCA vasculitis.¹⁴ These studies focused on mature circulating neutrophils; however, other reports have implicated monocytes,¹¹ myeloid progenitors,^{8, 10} and low-density granulocytes⁹ as sources of autoantigen gene expression.

The elevated autoantigen gene expression in ANCA vasculitis could stem from an influx of myeloid progenitors in response to inflammatory signals, or disruption of normal transcriptional controls. If the former, elevated expression may be a consequence of the inflammatory state of the disease; if the latter, elevated expression may predispose to disease. Although not mutually exclusive these options raise at least two questions: 1) What is the source of autoantigen gene expression? 2) Is elevated autoantigen gene expression functionally relevant to disease?

We addressed these questions by measuring expression in purified cell populations from peripheral blood of healthy individuals and patients with ANCA vasculitis during active disease and remission. We immunophenotyped low-density neutrophils (LDNs) and measured autoantigen gene expression in cell subsets within the LDN population. We assessed whether LDNs participate in ANCA-mediated pathological processes by measuring the in vitro response of LDNs to ANCA. In mature normal-density neutrophils (NDNs) we identified an association between autoantigen gene expression and neutrophil activation in response to ANCA, implicating elevated autoantigen gene expression in contributing to disease pathology.

RESULTS

Cell types responsible for elevated autoantigen gene expression

Cross-sectional analysis.

To identify the source of elevated MPO and PRTN3 expression in ANCA vasculitis,^7-12 and investigate the role of expression in pathogenesis we isolated peripheral blood cell types from healthy controls (HC) and patients with ANCA vasculitis during active disease and remission. Representative flow cytometry plots in Supplementary Figure 1 show the purity of monocytes, neutrophils and enriched leukocytes. In monocytes (Figure 1a), there was not a significant difference in MPO expression between HC and active disease, with a small difference between HC and remission (1.5-fold). PRTN3 mRNA in monocytes was low across all groups, suggesting monocytes are not major contributors to PRTN3 mRNA levels in total leukocytes. The expression of MPO was significantly elevated in active patients compared to HC in neutrophils (Figure 1b), and both MPO and PRTN3 were elevated in active patients compared to HC in enriched leukocytes (Figure 1c). In neutrophils and enriched leukocytes, MPO and PRTN3 expression was not different between MPO and PR3 serotypes (Supplementary Figure 2). The difference in expression between HC and active patients was more pronounced in the enriched leukocytes than neutrophils. The median fold-change for MPO and PRTN3 expression between HC and patients with active disease is 2.2 and 1.7 in neutrophils, and 3.5 and 11.9 in enriched leukocytes, respectively.

Figure 1. Robust, differential MPO and PRTN3 expression in enriched leukocytes, where paired analyses identified significant differences in expression between active disease and remission.

(a-c) MPO and PRTN3 expression was measured in monocytes (n = 84) (a), neutrophils (n = 98 MPO; n = 97 PRTN3) (b), and enriched leukocytes (n = 80) (c) from cross-sectional samples of healthy controls (HC, green), ANCA vasculitis patients with active disease (red), and patients in remission (blue). Horizontal bars indicate median and interquartile range. P-values were adjusted for multiple comparisons and only significant p-values < 0.05 are shown. Sample sizes are in parentheses beneath each category. In neutrophils (b) the median fold-change for MPO and PRTN3 expression between HC and patients with active disease is 2.2 (mean FC = 2.5) and 1.7 (mean FC = 3.8), respectively. In enriched leukocytes (c) the median fold-change MPO and PRTN3 expression between HC and patients with active disease is 3.5 (mean FC = 6.3) and 11.9 (mean FC = 10.8), respectively. (d, e) Longitudinal analysis of MPO and PRTN3 expression. MPO and PRTN3 expression was measured in ANCA vasculitis patient pairs at active disease and remission within neutrophils (d) and enriched leukocytes (e) and sub-divided into Static and Dynamic patient groups, based on MPO and PRTN3 expression in WBCs. P-values for this longitudinal cohort are from paired test; only significant p-values < 0.05 are listed. Number of paired samples are listed in parentheses beneath each plot.

Longitudinal analysis.

In total leukocytes we observed a pattern of expression between paired active disease and remission samples (manuscript in preparation) that separates patients into static and dynamic groups. In the static group expression in total leukocytes is within 2 standard deviations of the mean expression in HC. In the dynamic group expression during active disease is greater than 2 standard deviations of the mean expression in HC returning to normal during remission. We tested whether this expression pattern was reflected in neutrophils and enriched leukocytes from active-remission patient pairs that were classified as static or dynamic based on longitudinal MPO and PRTN3 expression in total leukocytes. In neutrophils (Figure 1d) from the dynamic group the expression was not significantly different between active disease and remission. In enriched leukocytes (Figure 1e) the expression between active disease and remission in the dynamic group was significantly different, indicating enriched leukocytes recapitulate the pattern observed in total leukocytes.

Transcriptome analysis.

We asked whether MPO and PRTN3 gene expression in the enriched leukocytes between active disease and remission was part of larger transcriptional changes. RNA-seq on low-density leukocytes from active-remission pairs revealed the dynamic group (Figure 2a) had 444 differentially expressed genes (DEGs); 314 decreased in remission including MPO and PRTN3. Virtually no differential gene expression was observed in the static group (Figure 2b). The heatmap in Figure 2c shows 30 genes with the largest fold-change from four dynamic active-remission pairs. In five static active-remission pairs (Figure 2d) the expression pattern for these same 30 genes was relatively unchanged. Although the expression of MPO and PRTN3 is of obvious importance, these data indicate global transcriptional changes can occur in ANCA vasculitis.

Figure 2. Transcriptional profile of enriched leukocytes from Active-Remission pairs revealed differential expression of genes functioning in neutrophil degranulation.

(a, b) MA plots indicate differentially expressed genes (DEGs) between active disease and remission in Dynamic (a, n = 4) and Static (b, n = 5) patient pairs, based on MPO and PRTN3 expression in WBCs. Red dots mark DEGs between active disease and remission based on an FDR <0.05. There are 444 DEGs in Dynamic active-remission pairs (a), with 314 genes elevated in active disease and 130 genes elevated in remission. (c, d) Heatmaps show the pattern of expression in the active-remission pairs (pink and teal in legend) for the top 30 DEG in the Dynamic group (c) and the pattern of expression for the same 30 genes in the Static group (d). (e) Top Venn diagram depicts number of genes differentially expressed in ANCA vasculitis that overlap in 4 separate studies. In the dataset reported here (Jones et al) the genes expressed in low-density leukocytes are upregulated in active-remission pairs. In the Cheadle et al and Lyons et al datasets the genes are increased in PBMCs from ANCA vasculitis patients compared to PBMCs from healthy controls. In the Grayson et al dataset from the Rituximab in ANCA-Associated Vasculitis (RAVE) trial, the genes are differentially expressed in whole blood comparing ANCA vasculitis patients who were treatment responders to non-responders. Bottom Venn diagram shows there are 54 genes in common between the dataset reported here and Grayson et al dataset. (f) The network diagram identified with Ingenuity Pathway Analysis illustrates that 25 genes (highlighted in green) of the 54 overlapping genes are in a network involved in cellular compromise, inflammatory response, and infectious disease. (IPA identified a top function of neutrophil degranulation, Benjamini-Hochberg adjusted p-value 2.68e-27.)

We compared the upregulated genes from the active-remission pairs to gene lists from three other expression studies of patients with ANCA vasculitis^8-10 to identify a common expression signature. In Figure 2e the top Venn diagram indicates there are genes shared with the other datasets, but the largest number of overlapping genes is with the Grayson et al. study. The bottom Venn diagram (Figure 2e) shows 54 genes overlap between these two studies, and Ingenuity Pathway Analysis found 25 of the overlapping genes are in a network (Figure 2f). The top function in this network involves neutrophil degranulation. The overlap between DEGs in the enriched leukocytes from our active-remission pairs and DEGs reported by Grayson et al. indicating low-density granulocytes (LDGs) are associated with active ANCA vasculitis prompted us to probe the enriched leukocytes from patients for LDGs.

Heterogeneity within low-density neutrophils

Fluorescence activated cell sorting.

Using cell surface markers reported to identify LDGs in patients with systemic lupus erythematosus (SLE)¹⁵ we isolated CD10-CD15+ and CD10+CD15+ cells from peripheral blood mononuclear cells (PBMCs) (Figure 3a). The nuclear morphology in Figure 3b indicates the CD10-CD15+ cells are immature neutrophils while the CD10+CD15+ cells are mature neutrophils, consistent with CD10 discrimination of immature and mature circulating CD66b+ neutrophils.¹⁶ We used flow cytometry to measure surface expression of the monocyte marker CD14 and the neutrophil marker CD16 (Figure 3c). The surface expression of both CD14 and CD16 was low on CD10-CD15+ cells and significantly greater on CD10+CD15+ cells. These data confirmed CD10-CD15+ cells mark immature neutrophils and CD10+CD15+ mark mature differentiated neutrophils. The combination of nuclear morphology and surface markers indicates the low-density population is comprised of neutrophils (LDNs).^{17, 18}

Figure 3. Cell surface markers and autoantigen gene expression in two populations of low-density neutrophils (LDNs).

(a) Flow cytometry plots illustrate the gating strategy for cell isolation. PBMCs were gated on CD3-CD4- cells (left plot), which were then gated on CD10- CD15+ and CD10+CD15+ cells. (b) Cytospin images show nuclear morphology characteristic of immature cells with myelocytes and metamyelocytes in CD10-CD15+ cells (left) compared to nuclear morphology of mature neutrophils with segmented nuclei in CD10+CD15+ cells (right). Scale bar = 10μm. (c) Cell surface expression is shown for CD14 (left graph) and CD16 (right graph) on CD10-CD15+ and CD10+CD15+ cells from healthy individuals (green), ANCA vasculitis patients with active disease (red), and patients in remission (blue). Sample sizes are listed in parentheses. Bars shown are median and interquartile range. T-test was performed on paired samples and for each the p-value < 0.0001. (d) Expression in CD10+CD15+ and CD10- CD15+ cells is shown for MPO (right graph) and PRTN3 (left graph). The expression values are displayed as the difference in expression in CD10+CD15+ cells minus the expression in total leukocytes (WBC), and the expression in CD10-CD15+ cells minus the expression in total leukocytes from healthy controls (green), ANCA vasculitis patients with active disease (red), and patients in remission (blue). Sample size for CD10+CD15+: MPO n = 16; PRTN3 n = 18, and for CD10-CD15+: MPO n = 21; PRTN3 n = 20. Bars shown are median and interquartile range. The p-values represent univariate t-test adjusted for multiple testing.

We measured MPO and PRTN3 mRNA in sorted CD10-CD15+ and CD10+CD15+ cells (Figure 3d) and detected robust expression in CD10-CD15+ immature LDNs. Importantly, the mean difference in MPO and PRTN3 mRNA between CD10+CD15+ mature LDNs and total leukocytes was significantly different from zero, indicating CD10+CD15+ mature LDNs also express MPO and PRTN3 mRNA. We conclude that both immature and mature LDNs contribute to the overall MPO and PRTN3 expression in total leukocytes.

Elevated autoantigen gene expression in low-density and normal-density neutrophils in patients with ANCA vasculitis

Analysis of expression in enriched LDNs.

To address whether the autoantigen genes are differentially expressed in the LDN population between HC and ANCA vasculitis patients we used a neutrophil enrichment procedure to isolate LDNs.¹⁹ Nuclear morphology and immunophenotype indicated the LDNs isolated with the enrichment procedure contained immature and mature neutrophils in HC and patients (Supplementary Figure 3). The median MPO and PRTN3 mRNA in the LDNs was 10.7 and 12.1-fold greater in patients with active disease compared to HC (Figure 4a), and remained significant after adjusting for age (Supplementary Table 1). The expression of MPO and PRTN3 in LDNs was significantly different between patients whose expression in total leukocytes was above (dynamic group) or below (static group) two standard deviations of the mean of HC (Supplementary Figure 4). Expression in LDNs correlated with expression in WBCs (Supplementary Figure 5a), and the percent of LDNs was significantly greater in patients with active disease than in HC (Supplementary Figure 5b). However, the frequency of LDNs only explained 11% and 7% of the variation in MPO and PRTN3 expression in WBCs (Supplementary Figure 5c). Elevated levels of MPO and PRTN3 mRNA were not due to significant differences in expression segregating between MPO-ANCA and PR3-ANCA serotypes (Supplementary Figure 6a, b). Alternatively, the percent of CD10-CD15+ immature neutrophils in the LDN population could account for elevated MPO and PRTN3 expression. We immunophenotyped the LDN population and failed to find a significant correlation between percent of immature cells and expression, with the variation in the percent of CD10-CD15+ immature neutrophils explaining about 7% and 3.5% of the variation in MPO and PRTN3 expression, respectively, in the LDN population (Figure 4b). We conclude that elevated MPO and PRTN3 expression is not solely due to more immature LDNs, or more LDNs, and instead is consistent with transcriptional dysregulation within the LDN population.

Figure 4. Elevated MPO and PRTN3 expression in LDNs unrelated to frequency of immature neutrophils, while elevated expression in NDNs correlated with MPO and PRTN3 in total leukocytes.

(a) LDNs were isolated from PBMCs using a neutrophil enrichment kit on PBMCs. Graphs show the expression of MPO (left) and PRTN3 (right) in LDNs isolated from healthy controls (HC, green), ANCA vasculitis patients with active disease (red), and patients in remission (blue). The median fold-change in expression between HC and patients with active disease is 10.66 (mean FC = 7.01) and 12.07 (mean FC = 9.41) for MPO and PRTN3, respectively. Sample size for MPO mRNA: HC n = 37; Active n = 41; Remission n = 53, and PRTN3 mRNA: HC n = 37; Active n = 40; Remission n = 53. Bars shown are median and interquartile range. P-values were adjusted for multiple comparisons and only significant p-values < 0.05 are listed. (b) Linear regression was used to compare frequency of CD10-CD15+ cells to MPO (left) and PRTN3 (right) expression in LDNs in HC (circles, n = 6) and ANCA vasculitis patients (triangles, n = 33). R-squared and p-value for linear regression are shown in each plot. Gray shading illustrates the 95% confidence limit. Intercept and slope were 12.014 and 0.2664 for MPO, and 10.216 and 0.1647 for PRTN3. Intercept and slope were similar between HC and patients. (c) Normal-density neutrophils (NDNs) were collected from the same samples used to isolate LDNs. Graphs show the expression of MPO (left) and PRTN3 (right) in PMNs isolated from HC (green), ANCA vasculitis patients with Active disease (red), and patients in Remission (blue). The median fold-change in expression between HC and patients with active disease is 2.28 (mean FC = 2.29) and 2.18 (mean FC = 2.40) for MPO and PRTN3, respectively. Sample size for MPO and PRTN3 mRNA: HC n = 39; Active n = 37; Remission n = 52. Bars shown are median and interquartile range. P-values were adjusted for multiple comparisons and only significant p-values < 0.05 are listed. (d) Linear regression was used to compare autoantigen gene expression in NDNs to expression in total leukocytes (WBC) for MPO (left) and PRTN3 (right) HC (circles, n = 27) and ANCA vasculitis patients (triangles, n = 74). R-squared and p-value for linear regression are shown in each plot. Gray shading illustrates the 95% confidence limit. Intercept and slope were 3.0396 and 0.7547 for MPO, and 1.7069 and 0.8626 for PRTN3. Intercept and slope were similar between HC and patients.

Glucocorticoid therapy may explain elevated MPO and PRTN3 expression.²⁰ We examined this possibility and found an apparent effect of prednisone on MPO and PRTN3 expression in LDNs (and enriched leukocytes) (Supplementary Figure 7) that was associated with active disease, but not remission (Supplementary Figure 8a-d; Supplementary Table 2). Expression was generally greater in patients with active disease receiving prednisone treatment at time of sample compared to expression in patients with active disease with no prednisone or no therapy; however, we observed a trend towards greater disease severity, as indicated by higher BVAS, among patients with active disease receiving prednisone (Supplementary Figure 8e). We detected no influence of Rituximab therapy on MPO or PRTN3 expression (data not shown).

Analysis of expression in NDNs.

Since NDNs are abundant in total leukocytes, we tested whether there were differences in MPO and PRTN3 expression between HC and patients in the cohort in which we isolated LDNs. The expression of MPO and PRTN3 was significantly greater in NDNs from patients with active disease compared to HC (Figure 4c), remained significant after adjusting for age (Supplementary Table 1), and was significantly different between NDNs separated into static and dynamic groups (Supplementary Figure 4). The expression of MPO and PRTN3 in NDNs significantly correlated with expression in WBCs (Figure 4d). In contrast to LDNs, prednisone treatment did not influence MPO and PRTN3 expression in NDNs (Supplementary Figure 9; Supplementary Table 2). The abundant NDNs are important in the transcriptional differences between patients and healthy controls and make a significant contribution to total leukocyte autoantigen gene expression.

Antibody mediated in vitro activation of low-density and normal-density neutrophils

We tested whether elevated MPO and PRTN3 expression in LDNs and NDNs influenced their response to MPO-ANCA mediated oxidative burst by measuring intracellular reactive oxygen species (ROS) with dihydrorhodamine 123 (DHR).²¹ To test LDNs, we incubated PBMCs from HC and patients with DHR, treated cells with MPO-ANCA IgG, HC IgG, or no IgG and measured rhodamine fluorescence in LDNs (Figure 5). The histograms in Figure 5a show the mean fluorescence intensity (MFI) for rhodamine in a representative HC and patient. In the HC and patient the rhodamine fluorescence detected in LDNs without exposure to MPO-ANCA (blue histograms) did not increase when samples were exposed to HC IgG or MPO-ANCA (orange and purple histograms). ROS production was induced in LDNs with PMA treatment (green histogram). MPO was detected on the surface of LDNs indicating the target is present for MPO-ANCA to induce an oxidative burst (Supplementary Figure 10a). The rhodamine signal without stimulation did not differ significantly between HC and patients (Figure 5b), and suggests spontaneous activation of LDNs at baseline, similar to NETosis of unstimulated LDGs from patients with ANCA vasculitis.⁹ Elevated autoantigen gene expression and resistance to MPO-ANCA mediated activation appear to be characteristics of LDNs because we detected significantly elevated MPO and PRTN3 expression in LDNs from patients with SLE, similar to a previous report²², and SLE-derived LDNs were resistant to activation by MPO-ANCA (Supplementary Figure 11).

Figure 5. Impaired in vitro activation of LDNs, while in vitro activation of NDNs correlated with autoantigen gene expression.

Intracellular reactive oxygen species (ROS) were determined by fluorescence of rhodamine. Histograms (a) show fluorescence signal in LDNs from HC (left) and Patient (right) following no stimulation (light blue), activation with HC IgG (orange), activation with MPO-ANCA IgG (purple), or PMA (green). Red histogram illustrates background autofluorescence of LDNs. (b) Geometric mean fluorescence intensity (MFI) indicates the baseline intracellular ROS in LDNs with no stimulation for HC (circles, n = 6) and ANCA vasculitis patients (triangles, n = 9). (c and d) Intracellular ROS were determined by fluorescence of rhodamine in NDNs. Geometric MFI values are shown for NDNs activated with HC IgG (circles) or (c) MPO-ANCA IgG (triangles) or (d) anti-PR3 monoclonal antibody (triangles). (c) HC, n = 13; patients with active disease, n = 11; patients in remission, n = 15. (d) HC, n = 11; patients with active disease, n = 3; patients in remission, n = 10. Significant p-values shown were adjusted for multiple comparisons. P-values for NDNs from HC, Active patients, or Remission patients activated with HC IgG compared to the same samples activated with MPO-ANCA IgG were all <0.0003 (not shown). (Comparison of oxidative burst with monoclonal anti-PR3 antibodies and PR3-ANCA IgG are shown in Supplementary Table 5.) (e and f) Linear regression was used to compare MFI in NDNs activated with (e) MPO-ANCA IgG to MPO expression in NDNs from HC (circles, n = 11) and ANCA vasculitis patients (triangles, n = 29), and (f) in NDNs activated with anti-PR3 monoclonal antibody to PRTN3 expression in NDNs from HC (circles, n = 11) and ANCA vasculitis patients (triangles, n = 12). Gray shading illustrates the 95% confidence limit. (g) Intracellular ROS were determined by fluorescence of rhodamine in neutrophils from mouse strain C57BL/6 (circles, n = 4) and 129-S6 (triangles, n = 4). Plotted values are the ratio of the median fluorescence intensity (FI) with anti-mouse MPO IgG to median fluorescence intensity without anti-mouse MPO IgG.

In contrast, NDNs were activated in response to MPO-ANCA without in vitro priming (Figure 5c), and in vitro primed NDNs were activated with monoclonal anti-PR3 antibody (Figure 5d). NDNs responded modestly to HC IgG, with no significant differences between HC and patients; however, MPO-ANCA IgG activated NDNs from patients significantly more than NDNs from HC (Figure 5c). We tested whether MPO and PRTN3 expression was associated with activation by MPO-ANCA or monoclonal anti-PR3 antibody. The rhodamine fluorescence significantly correlated with MPO and PRTN3 expression (Figure 5e and f, respectively) which supports the concept that transcriptional dysregulation of the autoantigen genes contributes to ANCA/antibody-mediated activation of neutrophils. The significant correlation between autoantigen gene expression and activation does not appear to be a consequence of increased MPO and PR3 protein (Supplementary Figure 10b and d) because surface MPO and PR3 protein on NDNs did not correlate with and MPO and PRTN3 expression (Supplementary Figure 10b-e). Using the DHR assay, in vitro activation of mouse neutrophils with anti-mMPO antibodies was greater in neutrophils from 129S6/SvEv mice than neutrophils from C57BL/6 mice (Figure 5g). 129SvEv mice have increased Mpo expression¹³ and more severe anti-mMPO induced glomerulonephritis.²³ These results confirm elevated expression can precede inflammation and suggest transcriptional differences in autoantigen genes are a potential predisposing factor in disease pathogenesis.

DISCUSSION

It is unclear how elevated ANCA autoantigen gene expression impacts disease. We measured expression in cells likely to be responsible for elevated MPO and PRTN3 mRNA. RNA-seq of enriched leukocytes identified genes that overlapped with previous transcriptome analyses of ANCA vasculitis including a report implicating low-density granulocytes.⁹ We detected elevated MPO and PRTN3 expression in LDNs that contained immature and mature neutrophils. The MPO and PRTN3 expression is more robust in LDNs compared to the more numerous mature NDNs; however, the level of autoantigen mRNA did not correlate with the percentage of immature neutrophils in the LDN population. Although immature LDNs contributed to autoantigen gene expression,^{8, 10} increased MPO and PRTN3 mRNA was also detected in mature LDNs and NDNs suggesting altered transcriptional regulation. This interpretation agrees with results that showed relatively little MPO and PRTN3 expression in patients with a “left-shift”¹² and epigenetic data that identified a transcriptionally permissive chromatin environment at the autoantigen genes in patients with ANCA vasculitis.^{13, 14, 24}

The consequence of the elevated autoantigen gene expression in LDNs is unclear, but LDNs/LDGs have been identified in various inflammatory conditions.^{15, 25-28} Evidence implicates them in the pathogenesis of ANCA vasculitis, lupus, and rheumatoid arthritis based on proinflammatory characteristics and propensity to form neutrophil extracellular traps (NETs).^{9, 19, 29} Despite elevated autoantigen gene expression, LDNs failed to respond to in vitro activation with MPO-ANCA.³⁰ The failure of MPO-ANCA to stimulate the oxidative burst of LDNs suggests alternative roles for LDNs in pathogenesis of ANCA vasculitis. LDNs could be involved in the immunopathogenesis of ANCA vasculitis via their enhanced capacity to form NETs and expose new/cryptic autoantigen epitopes. In SLE LDGs have proinflammatory activity on T cells.³¹ If LDNs in ANCA behave similarly they could contribute to the expansion of a previously reported population of suppression-resistant T cells.³² Increased LDNs could precede ANCA vasculitis as a consequence of a prodromal condition as reported in diabetic mice infected with Staphylococcus aureus.³³ Addressing these possibilities, or whether LDNs are a consequence of increased granulopoiesis,³⁴ warrant future studies to delineate a role for LDNs in ANCA vasculitis separate from mediating vascular damage via ANCA-induced oxidative burst.

Normal-density neutrophils are central to ANCA vasculitis pathology. We showed MPO-ANCA induced activation was significantly greater in NDNs from ANCA vasculitis patients than from HC and the response correlated with MPO expression. Likewise, anti-PR3 antibody induced NDN activation positively correlated with PRTN3 mRNA. Increased expression could contribute to disease pathology by increasing the autoantigen abundance. However, when interpreting the consequence of elevated MPO and PRTN3 expression it is important to consider new protein synthesis³⁵ because transcript levels alone may not predict protein levels. Instead of increasing MPO and PR3 proteins, or positive correlation between protein and mRNA, elevated gene expression could be accompanied by alterations in protein dynamics, such as turnover or misfolding.^{36, 37} The fate of newly synthesized MPO and PR3 may result in different phenotypic consequences. Alternatively, global transcriptional changes or changes to the neutrophil proteome³⁸ could alter other neutrophil functions involved in pathogenesis.

Some technical aspects impacted this study. Although we limited positive selection of most cell types to reduce the influence of processing on transcription, expression in monocytes and sorted cells could be affected by selection with antibodies and cell sorting.^{39, 40} The LDN analyses were limited to cross-sectional comparisons with few active-remission pairs. Because the LDN yield was low we measured their oxidative burst in PBMCs and the mixture of cells may dampen the response. Besides the limitation of measuring LDN activation in PBMCs, LDNs could contribute to disease pathology other than through oxidative burst, especially given their smaller frequency compared to NDNs. Oxidative bust of NDNs induced by MPO-ANCA was much greater compared to PR3-ANCA, as reported previously,⁴¹ prompting us to switch to a monoclonal anti-PR3 antibody. The level of oxidative burst measured by rhodamine fluorescence was not greater in NDNs from patients with active disease than patients in remission. The DHR assay measures intracellular ROS, and more extracellular ROS production may occur from NDNs of patients with active disease. ROS production in NDNs from patients with active disease or remission could be equivalent and NDNs from patients with ANCA vasculitis are in a pathological state receptive to an imbalance in immune tolerance.

Our investigation of autoantigen gene expression in two populations of neutrophils resulted in discovering differential consequences of aberrant expression. On one hand, LDNs are resistant to the direct pathological processes induced by ANCA. Further studies are needed to determine if LDNs in ANCA vasculitis are a response to chronic inflammation or if LDNs contribute to alternative mechanisms of disease pathogenesis. On the other hand, the pathological response of NDNs to MPO-ANCA or anti-PR3 antibody correlated with MPO and PRTN3 gene expression, providing further rationale to develop therapies that suppress autoantigen gene expression.

METHODS

Human Subjects

Patients with ANCA vasculitis and healthy volunteers were recruited, gave informed written consent, and participated according to the guidelines of the University of North Carolina Office of Human Research Ethics/Institutional Review Board (IRB study #97-0523). ANCA serotypes were determined by indirect immunofluorescence and antigen-specific PR3 and MPO ELISA.⁴² Patients were diagnosed according to the Chapel Hill Consensus Conference.^{43, 44} Disease activity was assigned based BVAS⁴⁵ and chart review covering three months before and after sample date. Active disease was defined as a BVAS ≥ 3 with clinical and/or laboratory evidence of disease. Remission was defined as a BVAS of 0 and no clinical or laboratory evidence of disease activity within 3 months of sample. Patient and HC demographics are summarized in Table 1. Patients with drug-induced forms of ANCA and overlapping disease were excluded. If more than one active and remission sample was collected per patient, samples with the most complete data or highest total leukocyte expression were retained.

Table 1:

Demographics of ANCA Vasculitis Patients and Healthy Controls in 3 Cohorts

Characteristic		Cell Type Expression Cohort		Flow Sorting Cohort		Low- and Normal- Density Neutrophil Cohort
		Patient	HC	Patient	HC	Patient	HC
Sample size		82	23	30	8	150	65
Age (years)	Mean ± SD	60 ± 15	50 ± 15	58 ± 16	43 ± 14	58 ± 18	42 ± 17
	Median (IQR)	61 (52, 71)	56 (41,58)	58 (46, 71)	46 (29,56)	63 (46, 70)	31 (25,55)
Gender	Female	32 (39%)	11 (48%)	9 (30%)	32 (39%)	65 (43%)	38 (56%)
	Male	50 (61%)	12 (52%)	21 (70%)	50 (61%)	85 (57%)	26 (44%)
Ethnicity	Caucasian	69 (84%)	23 (100%)	23 (77%)	7 (88%)	135 (90%)	50 (78%)
	African American	9 (11%)	0	5 (17%)	0	10 (6.7%)	2 (1.8%)
	Hispanic	2 (2.4%)	0	1 (3%)	0	2 (1.3%)	7 (9.2%)
	Asian	1 (1.2%)	0	0	1 (12%)	2 (1.3%)	1 (1.8%)
	Asian/Pacific Islander	0	0	0	0	0	2 (3.7%)
	Other	1 (1.2%)	0	1 (3%)	0	1 (0.7%)	3 (5.5%)
Serotype	MPO	44 (54%)		14 (47%)		61 (41%)
	PR3	36 (44%)		16 (53%)		86 (57%)
	Both	1 (1.2%)		0		2 (1.3%)
	Negative	1 (1.2%)		0		0
	MPO; Membranous	0		0		1 (0.7%)
Diagnosis	GPA	34 (41%)		13 (43%)		67 (45%)
	MPA	30 (37%)		11 (37%)		58 (39%)
	Lim	10 (12%)		4 (13%)		19 (13%)
	EGPA	5 (6%)		1 (3%)		4 (2.7%)
	Unknowna	3 (4%)		1 (3%)		2 (1.3%)
Activity Status	Active	49 (60%)		12 (40%)		52 (35%)
	Remission	33 (40%)		14 (47%)		70 (47%)
	Unclearb	0		4 (13%)		6 (4.0%)
	Unknownc	0		0		18 (12%)
	Not Determinedd	0		0		4 (2.7%)
Medications at sample date
No Medications		34 (42%)		7 (23%)		38 (25%)
Prednisone		23 (28%)		6 (20%)		32 (21%)
Methylprednisolone		3 (4%)		1 (3%)		5 (3.3%)
Cyclophosphamide		5 (6%)		1 (3%)		4 (2.7%)
Rituximabe	<3 months	14 (17%)		7 (23%)		18 (12%)
	>3 months	14 (17%)		10 (33%)		42 (28%)
Mycophenolate mofetil		8 (10%)		6 (20%)		11 (7.3%)
Azathioprine		4 (5%)		1 (3%)		11 (7.3%)
Plasmapheresis		2 (2.4%)		1 (3%)		3 (2.0%)
Methotrexate		0		0		3 (2.0%)
Plaquenil		2 (2.4%)		1 (3%)		1 (0.7%)

HC, healthy control; SD, standard deviation; IQR, interquartile range; MPO, myeloperoxidase; PR3, proteinase 3; GPA, granulomatosis with polyangiitis; MPA, microscopic polyangiitis; Lim, renal-limited small vessel vasculitis; EGPA, eosinophilic granulomatosis with polyangiitis

^aUnknown: Clinic data not entered in database at time of submission

^bUnclear: Unable to assign activity status after chart review

^cUnknown: Sample collected without clinic notes for chart review

^dNot Determined: Chart review not complete

^eRituximab: Sample was collected either within 3 months or after 3 months of treatment

Isolation of immune cell types

Peripheral blood was collected in sodium heparin tubes and processed within 4 hours of blood draw. HetaSep™ (Stem Cell Technologies; Vancouver, BC) was used to deplete red blood cells. To isolate neutrophils and PBMCs, total leukocytes were placed over Histopaque 1077 (Sigma; St. Louis, MO) and centrifuged (400g, 30 minutes, no brake). Using magnetic microbeads, CD14+ monocytes were positively selected from PBMCs (EasySep™ Stem Cell Technologies), and enriched leukocytes were isolated from the CD14- PBMCs with a human CD4 T cell enrichment cocktail (Stem Cell Technologies).^{46, 47} Neutrophils were isolated from red blood cell/granulocyte pellet after Histopaque separation and red cell lysis, and lysed for RNA or resuspended in Hank’s Buffered Saline (HBSS, Gibco; Gaithersburg, MD) for in vitro activation experiments. Low-density neutrophils (LDNs) were enriched from PBMCs using the human neutrophil isolation kit (Stem Cell Technologies).

RNA isolation and quantitative RT-PCR

RNA was isolated from monocytes and enriched leukocytes using the AllPrep DNA/RNA Mini Kit (Qiagen; Germantown, MD), from neutrophils using STAT-60 (Tel-Test “B”; Friendswood, TX), and from LDNs with Ambion RNAqueous-Micro total RNA isolation kit (Invitrogen; Carlsbad, CA). Quantitative detection ofMPO and PRTN3 mRNA from monocytes, neutrophils, and enriched leukocytes was performed with TaqMan® RNA-to-Ct™ 1-Step kit (Applied Biosystems; Waltham, MA). For sorted CD10-CD15+, CD10+CD15+, and LDNs total RNA was converted to cDNA with iScript™ Reverse Transcription Supermix (BioRad; Hercules, CA) and cDNA was used for quantitative amplification with TaqMan® Universal PCR Master Mix (Applied Biosystems). Quantitative RT-PCR was performed on an ABI PRISM 7900HT using primer and probe assays listed in Supplementary Table 3. Quantitative detection of MPO and PRTN3 mRNA levels was determined by 2^−ΔΔCt calculations and reported relative to standard curves as described previously.¹⁴

RNA sequencing

RNA-seq was performed on enriched leukocytes from nine longitudinal ANCA vasculitis patient pairs. RNA-seq libraries were prepared with Kapa mRNA Stranded kit (KapaBiosystems; Wilmington, MA) and pooled libraries were sequenced on Illumina HiSeq4000. Sequenced reads were filtered using Trimmomatic⁴⁸, aligned to hg19 using tophat⁴⁹, and counted using HTseqcount.⁵⁰ Differentially expressed genes were identified with DESeq2 in R (version 3.5.0, 2018-04-23)^{51, 52} and analyzed using Ingenuity Pathway Analysis (IPA) version 01-08 (Qiagen Bioinformatics; Redwood City, CA).

Flow cytometry sorting and analysis

PBMCs and LDNs were counted using a hemacytometer. PBMCs were washed and resuspended at 5.0x10⁷ cells/ml in FACS buffer (HBSS, 2% FBS) and stained against surface markers (Supplementary Table 4). Cells were sorted using a Becton Dickinson FACS Aria II equipped with FACSDiva 7 software (BD Biosciences; San Jose, CA) under BSL 2 conditions. Immunophenotyping of LDNs used surface markers listed in Supplementary Table 5. LDNs were incubated with antibodies for 30 minutes on ice, washed twice with FACS buffer, fixed using a 1X 1-step Fix/Lyse solution (eBioscience; San Diego, CA) for 20 minutes at room temperature, and re-suspended in FACS buffer. Samples were acquired using the Attune NxT flow cytometer. Data were analyzed by FlowJo (version 10.5.3 Tree Star, Inc; Ashland, OR).

Cytospins

Cytospins were performed on sorted CD10-CD15+, CD10+CD15+, and enriched LDNs using a Shandon Cytospin 2 Cytocentrifuge. Cells were stained with Shandon Kwik-Diff solutions (ThermoFisher; Waltham, MA). Images were acquired using Olympus BX43F microscope and cellSens Entry 1.11 software, or for LDNs the Zeiss AxioImager-M2 microscope and AxioVison 4.8 software.

Preparation of Human Immunoglobulins

Human immunoglobulin G (IgG) was prepared from plasma samples from an active biopsyproven MPO-ANCA patient as well as from a HC using a HiTrap Protein G HP column (GE Healthcare; Pittsburgh, PA) in an FPLC system. IgG isolation was confirmed using Coomassiestained gels.

Neutrophil Oxidative Burst Assay

Purified NDNs or PBMCs were re-suspended at 1 x 10⁶ and 2 × 10⁶ cells/mL respectively in HBSS and stained with Dihydrohodamine (DHR)-123 (Sigma-Aldrich) in a final concentration of 2.9μM for 15 minutes at 37°C. Cells without priming were stimulated with either control IgG or MPO-ANCA IgG at a final concentration of 250μg/mL for 20minutes at 37°C. NDNs were primed with TNFα (2ng/mL final concentration) for 15 minutes at 37°C prior to stimulation with PR3-ANCA or monoclonal anti-PR3 antibody 4A5⁵³ (diluted 1:1000) (Supplementary Table 5). Cells were washed twice with FACS buffer and re-suspended in cold FACS buffer. NDNs were stained, as described above, with anti-human CD16, and PBMCs were stained with anti-human CD3, CD4, CD10, CD15 antibodies (Supplementary Table 6). Samples were acquired using the LSRII or Attune NxT flow cytometers using gating strategies shown in Supplementary Figures 12 and 13. ROS production was quantified by measuring rhodamine fluorescence using FlowJo.

Mice

C57BL/6J (B6) mice were purchased from Jackson Laboratories (Bar Harbor, ME), and 129S6 mice from Taconic Farms (Germantown, NY). All animal experiments were approved by the UNC IACUC, and performed according to NIH Guide for Care and Use of Laboratory Animals. Mouse neutrophils were purified from bone marrow as described.²³ Oxidative burst was performed as described above except with anti-mMPO antibodies and Ly6G (clone 1AB, Biolegend).

Statistical analysis

Data were analyzed with SAS software version 9.4 (SAS Institute; Cary, NC) and R version 3.5.0, and figures were made with GraphPad Prism 7 (La Jolla, CA), IPA and ggplot2 package in R.⁵⁴ Log2 transformation was used to normalize data distributions. Multiple comparisons used linear regression or Wilcoxon rank sum test depending on data distributions, with Bonferroni adjustments. R-square and p-values were reported for correlation of expression groups. Paired data were analyzed using paired student's T-test. A p-value less than 0.05 was considered significant in the statistical analyses.

Supplementary Material

1

Figure S1. Flow cytometric analysis of isolated monocytes, neutrophils, and low-density leukocytes. Representative flow cytometry plots show purity of isolated cell populations. (a) Purity of monocytes was determined by percent of CD14 positive cells, with percent positive shown in plot. (b) Purity of neutrophils was determined by percent of CD16 positive cells, with percent positive shown in plot. The purity low-density leukocytes obtained following CD4 T cell enrichment of PBMCs is show in c and d. The low-density leukocytes were enriched for CD4+ cells but also contained CD3-CD4- cells. Both examples contain a CD3-CD4- population.

Figure S2. Autoantigen gene expression in enriched leukocytes and neutrophils divided by disease activity status and serotype. Patient disease status was classified as active or remission according to criteria described in Methods section on Human subjects. ANCA serotypes were determined by indirect immunofluorescence and antigen-specific PR3 and MPO ELISA. (a, b) The level of MPO and PRTN3 mRNA was measured in enriched leukocytes. (c, d) The level of MPO and PRTN3 mRNA was measured in neutrophils. The log2 transformed values for expression were compared between samples from patients with either MPO-ANCA or PR3-ANCA serotypes separated by disease activity. The p values were calculated using Welch two sample t-test. The *p values were calculated using Wilcoxon rank sum test when distribution of one group failed normality test (Shapiro-Wilks test).

Figure S3. Cytospins and flow cytometry identify mature neutrophils within the enriched population of LDNs. (a) Representative nuclear morphology of LDNs isolated from healthy control, a patient with active disease and a patient in remission. Mature neutrophils with segmented nuclei are denoted by arrows. Images were acquired at 40X; scale bar denotes 20μm. (b) Top 3 flow panels show the gating scheme used to immunophenotype LDNs enriched from PBMCs. The left panel shows gating on singlets, which were then gated on granulocytes (center panel), and finally gated on CD3-CD4- cells (right panel). The CD3-CD4- cells were immunophenotyped for CD10 and CD15, shown in the left column of 3 flow plots from healthy control (top), active patient (middle), and remission (bottom). Percent of CD10-CD15+ and CD10+CD15+ are shown in plot. CD3-CD4- cells were also immunophenotyped for CD16 (middle column of 3 flow plots) and CD14 (right column of 3 flow plots).

Figure S4. Autoantigen gene expression in LDNs and NDNs subdivided based on total leukocyte expression. Patients were divided into Static or Dynamic groups based on MPO and PRTN3 expression in total leukocytes. Patients were classified as Static if expression in total leukocytes was within 2 standard deviations of the mean expression in healthy controls. Patients were classified as Dynamic if expression in total leukocytes was greater than 2 standard deviations of the mean expression in healthy controls. The level of MPO mRNA (a) and PRTN3 mRNA (b) in LDNs from the Static Group was compared to the level in LDNs from the Dynamic group. Similarly the level of MPO mRNA (a) and PRTN3 mRNA (b) in NDNs from the Static Group was compared to the level in NDNs from the Dynamic group. Autoantigen gene expression in both LDNs and NDNs was significantly greater in the Dynamic group compared to the Static group. Red dots denote patients with active; blue dots denote patients in remission.

Figure S5. Autoantigen expression in LDNs and their frequency in PBMCs. (a) Linear regression was used to compare autoantigen gene expression in LDNs to expression in total leukocytes (WBC) for MPO (left) and PRTN3 (right) in HC (black circles, n = 32) and ANCA vasculitis patients (black triangles, n = 81). R-squared and p-value for linear regression are shown in each plot. Gray shading illustrates the 95% confidence limit. (b) LDNs were enriched from PBMCs and counted manually. The graph shows the log2 transformed percent of LDNs in PBMCs from healthy controls (HC, green, n = 26), ANCA vasculitis patients with Active disease (red, n = 33), and patients in Remission (blue, n = 46). The median fold-change in percent LDNs between HC and patients with active disease is 2.36 (mean FC = 2.25) and the median fold-change in percent LDNs between HC and patients in remission is 1.93 (mean FC = 2.01). Bars shown are median and interquartile range. (c) Linear regression was used to compare percent of LDNs in PBMCs to autoantigen gene expression in LDNs for MPO (left) and PRTN3 (right) in HC (black circles, n = 32) and ANCA vasculitis patients (black triangles, n = 81). R-squared and p-value for linear regression are shown in each plot. Gray shading illustrates the 95% confidence limit.

Figure S6. Autoantigen gene expression in low-density and normal-density neutrophils divided by disease activity status and serotype. Patient disease status was classified as active or remission according to criteria described in Methods section on Human subjects. ANCA serotypes were determined by indirect immunofluorescence and antigen-specific PR3 and MPO ELISA. (a, b) The level of MPO and PRTN3 mRNA was measured in low-density neutrophils from HC (green triangles), patients with Active disease and MPO-ANCA (red squares) or PR3-ANCA (red circles), and patients in Remission and MPO-ANCA (blue squares) or PR3-ANCA (blue circles). (c, d) The level of MPO and PRTN3 mRNA was measured in normal-density neutrophils from HC (green triangles), patients with Active disease and MPO-ANCA (red squares) or PR3-ANCA (red circles), and patients in Remission and MPO-ANCA (blue squares) or PR3-ANCA (blue circles). The log2 transformed values for expression were compared between samples from patients with either MPO-ANCA or PR3-ANCA serotypes separated by disease activity. The p values were calculated using Welch two sample t-test. The *p values were calculated using Wilcoxon rank sum test when distribution of one group failed normality test (Shapiro-Wilks test).

Figure S7. Effect of prednisone treatment on autoantigen gene expression in enriched leukocytes and low-density neutrophils. MPO and PRTN3 gene expression were measured in (a, b) enriched leukocytes and (c, d) low-density leukocytes from patients with active disease or remission receiving no therapy (NoRx); no prednisone, but another therapy (NoPred); or prednisone not excluding other therapies (Pred) at time of sample collection. (a) Mean MPO mRNA in enriched leukocytes from active and remission samples is 6.69 for NoRx, 6.45 for NoPred, 8.23 for Pred. (b) Mean PRTN3 mRNA in enriched leukocytes from active and remission samples is 5.47 for NoRx, 5.53 for NoPred, 7.98 for Pred. (c) Mean MPO mRNA in low-density neutrophils from active and remission samples is 12.10 for NoRx, 12.97 for NoPred, 14.12 for Pred. (d) Mean PRTN3 mRNA in low-density neutrophils from active and remission samples is 9.87 for NoRx, 10.98 for NoPred, 12.02 for Pred. Effect of prednisone treatment on autoantigen gene expression was determined by ANOVA test or Kruskal-Wallis (KW) test when distribution of one group failed normality test (Shapiro-Wilks test). (e) Percent of low-density neutrophils in PBMCs was compared among the 3 prednisone treatment groups. Sample sizes for each of the three prednisone treatment groups are listed in parentheses.

Figure S8. Prednisone treatment impacts autoantigen gene expression in enriched leukocytes and low-density neutrophils in patients with active disease. MPO and PRTN3 gene expression were measured in (a, b) enriched leukocytes and (c, d) low-density neutrophils from patients with active disease (red circles) or remission (blue circles) receiving no therapy (NoRx); no prednisone, but another therapy (NoPred); or prednisone not excluding other therapies (Pred) at time of sample collection. Effect of prednisone treatment on autoantigen gene expression was determined by ANOVA test or Kruskal-Wallis (KW) test when distribution of one group failed normality test (Shapiro-Wilks test). Sample sizes for each of the three prednisone treatment groups are listed in parentheses. (a) Mean MPO mRNA in active samples is 6.88 for NoRx, 6.69 for NoPred, 9.39 for Pred; in remission samples is 6.49 for NoRx, 6.16 for NoPred, and 6.78 for Pred. (b) Mean PRTN3 mRNA in active samples is 5.87 for NoRx, 5.84 for NoPred, 8.85 for Pred; in remission samples is 5.07 for NoRx, 5.19 for NoPred, and 6.75 for Pred. (c) Mean MPO mRNA in active samples is 12.26 for NoRx, 13.06 for NoPred, 14.48 for Pred; in remission samples is 11.95 for NoRx, 12.93 for NoPred, and 13.58 for Pred. (d) Mean PRTN3 mRNA in active samples is 9.77 for NoRx, 11.38 for NoPred, 12.43 for Pred; in remission samples is 9.97 for NoRx, 11.16 for NoPred, and 11.39 for Pred. Autoantigen expression in healthy controls (HC, green circles) is included to illustrate elevated expression in patients regardless of prednisone treatment compared to expression in HC. (e) BVAS for patients with active disease from low-density neutrophil cohort (in c and d) was compared among the 3 prednisone treatment groups. Mean BVAS is 6.0 for NoRx, 7.5 for NoPred, and 10.4 for Pred. Welch’s ANOVA test was used to determine effect of prednisone because variances among the group were unequal, as determined by Bartlett’s test.

Figure S9. Effect of prednisone treatment on autoantigen gene expression in normal-density neutrophils. MPO and PRTN3 gene expression were measured in normal-density neutrophils in the cohort of samples from Figure 1 (a, b) and the cohort of samples from Figure 4 (c, d) receiving no therapy (NoRx); no prednisone, but another therapy (NoPred); or prednisone not excluding other therapies (Pred) at time of sample collection. (a) Mean MPO mRNA in active and remission samples is 4.56 for NoRx, 4.60 for NoPred, 5.22 for Pred. (b) Mean PRTN3 mRNA in active and remission samples is 2.68 for NoRx, 2.59 for NoPred, 3.37 for Pred. (c) Mean MPO mRNA in active and remission samples is 6.31 for NoRx, 6.05 for NoPred, 6.71 for Pred. (d) Mean PRTN3 mRNA in active and remission samples is 3.62 for NoRx, 3.66 for NoPred, 4.09 for Pred. Effect of prednisone treatment on autoantigen gene expression was determined by ANOVA test. Sample sizes for each of the three prednisone treatment groups are listed in parentheses.

Figure S10. MPO and PR3 proteins on the surface of low-density and normal-neutrophils. (a) Representative histogram from flow cytometry analysis of MPO on the surface of low-density neutrophils. Red histogram is geometric mean fluorescence intensity (MFI) for unstained low-density neutrophils. Blue histogram is MFI for low-density neutrophils stained with anti-MPO antibody conjugated to phycoerythrin fluorophore. (b) Boxplot compares MPO surface expression on normal-density neutrophils (used in in vitro activation study reported in Figure 5c, d) between healthy controls (HC, black circles, n = 5) and patients (black triangles, n = 16). (c) Scatterplot shows a positive, non-significant correlation between MPO surface expression and MPO mRNA levels in normal-density neutrophils. (d) Boxplot compares PR3 surface expression on normal-density neutrophils (used in in vitro activation study reported in Figure 5e, f) between healthy controls (HC, black circles, n = 10) and patients (black triangles, n = 7). (c) Scatterplot shows a positive, non-significant correlation between PR3 surface expression and PRTN3 mRNA levels in normal-density neutrophils. P value in (b) and (d) was calculated using the Wilcoxon rank-sum test. P value and r-squared in (c) and (e) were calculated from linear regression using linear model, lm, function in R.

Figure S11. Elevated autoantigen gene expression in low-density neutrophils in patients with SLE and the response of SLE LDNs to MPO-ANCA. MPO (a) and PRTN3 (b) gene expression were measured in low-density neutrophils from healthy controls (HC, green circles), patients with Active ANCA vasculitis (red) or Remission (blue), patients with SLE (orange circles) and patients with ANCA vasculitis with Undetermined disease activity status (purple diamonds). HC, Active and Remission expression values are from Figure 4a and b. LDNs for these samples were isolated using StemCell Technologies human neutrophil isolation kit catalog #19257. The manufacturer replaced this kit. The new catalog #17957 was used to isolate LDNs from patients with SLE (orange circles) and patients with ANCA vasculitis (purple diamonds). No significant differences were detected between MPO or PRTN3 expression from patients with SLE and ANCA vasculitis patients with Active disease, Remission, or Undetermined. MPO and PRTN3 expression were significantly different between HC and SLE, Benjamini-Hochberg adjusted p values 0.00112 and 0.00361, respectively. Sample size for MPO mRNA: HC n = 37; Active n = 41; Remission n = 53; SLE n = 13; ANCA = 12 and PRTN3 mRNA: HC n = 37; Active n = 40; Remission n = 53; SLE n = 13; ANCA = 12. (c) Intracellular reactive oxygen species (ROS) were determined by fluorescence of rhodamine. Representative histograms show fluorescence signal from left to right in LDNs from HC, patient with ANCA vasculitis, patient with SLE, and in NDNs from patient with ANCA vasculitis. Histograms represent rhodamine fluorescence with no stimulation (light blue), activation with HC IgG (orange), activation with MPO-ANCA IgG (purple), or PMA (green). Red histogram illustrates background autofluorescence of LDNs. Treatment with MPO-ANCA IgG only produced a shift in rhodamine fluorescence distinct from treatment with HC IgG in NDNs, not in LDNs from patients with ANCA vasculitis or SLE.

Figure S12. Gating scheme to measure Reactive Oxygen Species (ROS) in LDNs. (a) The gating scheme to isolate LDNs from total PBMCs: the left panel shows gating on singlets, which were then gated on granulocytes (center panel), and finally gated on CD3-CD4- cells (right panel). (b) The CD3-CD4- cells were immunophenotyped for CD10 and CD15. Representative plot shows the percentage of CD10-CD15+ and CD10+CD15+ within the CD3-CD4- population. (c) The CD3- D4- cells were analyzed for rhodamine fluorescence in the FITC channel, which is proportional to intracellular ROS.

Figure S13. Gating scheme to measure Reactive Oxygen Species (ROS) in human NDNs and mouse neutrophils. (a) The left panel shows gating on singlets from human NDNs which were then gated on granulocytes left panel. (b) Granulocytes were gated on CD16+ NDNs and DHR fluorescence was measured in the FITC channel. DHR fluorescence is proportional to intracellular ROS and a representative histogram of DHR fluorescence in the FITC channel is shown (left panel). (c) The left panel shows gating on singlets of mouse neutrophils which were then gated on granulocytes left panel. (b) Granulocytes were gated on Ly6G+ neutrophils and DHR fluorescence was measured in the FITC channel. DHR fluorescence is proportional to intracellular ROS and a representative histogram of DHR fluorescence in the FITC channel is shown (left panel).

Table S1: Comparison of Log2 LDN expression between HC and active patients

Table S2a: ANOVA test for prednisone treatment on autoantigen gene expression; samples grouped by Disease Activity Status (Active or Remission) or ANCA Serotype (MPO-ANCA or PR3-ANCA)

Table S2b: TukeyHSD post-hos test from multiple comparison of groups with ANOVA p-value < 0.05

Table S3: Primers and Probe for TaqMan quantitative real-time PCR

Table S4: Cell surface markers and Antibodies for flow sorting PBMCs

Table S5. Comparison of oxidative burst between PR3-ANCA IgG and anti-PR3 monoclonal antibodies

Table S6: Cell surface markers and Antibodies to immunophenotype LDNs