Identification of novel, clinically correlated autoantigens in the monogenic autoimmune syndrome APS1 by PhIP-Seq

The identification of autoantigens remains a critical challenge for understanding and treating autoimmune diseases. Autoimmune polyendocrine syndrome type 1 (APS1), a rare monogenic form of autoimmunity, presents as widespread autoimmunity with T and B cell responses to multiple organs. Importantly, autoantibody discovery in APS1 can illuminate fundamental disease pathogenesis, and many of the antigens found in APS1 extend to common autoimmune diseases. Here, we performed proteome-wide programmable phage-display (PhIP-Seq) on sera from an APS1 cohort and discovered multiple common antibody targets. These novel autoantigens exhibit tissue-restricted expression, including expression in enteroendocrine cells and dental enamel. Using detailed clinical phenotyping, we find novel associations between autoantibodies and organ-restricted autoimmunity, including between anti-KHDC3L autoantibodies and premature ovarian insufficiency, and between anti-RFX6 autoantibodies and diarrheal-type intestinal dysfunction. Our study highlights the utility of PhIP-Seq for interrogating antigenic repertoires in human autoimmunity and the importance of antigen discovery for improved understanding of disease mechanisms.


tissue-restricted expression, including expression in enteroendocrine cells and dental enamel. 48
Using detailed clinical phenotyping, we find novel associations between autoantibodies and organ-49 restricted autoimmunity, including between anti-KHDC3L autoantibodies and premature ovarian 50 insufficiency, and between anti-RFX6 autoantibodies and diarrheal-type intestinal dysfunction. 51 Our study highlights the utility of PhIP-Seq for interrogating antigenic repertoires in human 52 autoimmunity and the importance of antigen discovery for improved understanding of disease 53 mechanisms. 54 between the T and B cell compartments (Lanzavecchia, 1985;Meyer et al., 2016). 68 Identification of the specificity of autoantibodies in autoimmune diseases is important for 69 understanding underlying disease pathogenesis and for identifying those at risk for disease (Rosen 70 & Casciola- Rosen, 2014). However, despite the long-known association of autoantibodies with 71 specific diseases in both monogenic and sporadic autoimmunity, many autoantibody specificities 72 remain undiscovered. Challenges in antigen identification include the weak affinity of some 73 autoantibodies for their target antigen, as well as rare or low expression of the target antigen. One 74 approach to overcome some of these challenges is to interrogate autoimmune patient samples with 75 particularly high affinity autoantibodies. Indeed, such an approach identified GAD65 as a major 76 autoantigen in type 1 diabetes by using sera from people with Stiff Person Syndrome (OMIM 77 #184850), who harbor high affinity autoantibodies (Baekkeskov et al., 1990). We reasoned that 78 PhIP-Seq interrogation of APS1, a defined monogenic autoimmune syndrome with a broad 79 spectrum of high affinity autoantibodies, would likely yield clinically meaningful targets -80 consistent with previously described APS1 autoantibody specificities that exhibit strong, clinically  Winqvist et al., 1993). 84 The identification of key B cell autoantigens in APS1 has occurred most commonly 85 through candidate-based approaches and by whole-protein microarrays. For example, lung antigen 86 BPIFB1 autoantibodies, which are used to assess people with APS1 for risk of interstitial lung 87 disease, were discovered first in Aire-deficient mice using a combination of targeted 88 immunoblotting, tissue microscopy, and mass spectrometry (Shum et al., 2013(Shum et al., , 2009  technology in uncovering shared antigens across APS1 cohorts, it is likely that many shared 97 antigens remain to be discovered, given that these arrays do not encompass the full coding potential 98 of the proteome. 99 Here, we took an alternate approach to APS1 antigen discovery by employing Phage Together, our results demonstrate the applicability of PhIP-Seq to antigen discovery, substantially 117 expand the spectrum of known antibody targets and clinical associations in APS1, and point 118 towards novel specificities that can be targeted in autoimmunity. 119

Investigation of APS1 serum autoantibodies by PhIP-Seq 123
Individuals with APS1 develop autoantibodies to many known protein targets, some of 124 which exhibit tissue-restricted expression and have been shown to correlate with specific 125 autoimmune disease manifestations. However, the target proteins for many of the APS1 tissue-126 specific manifestations remain enigmatic. To this end, we employed a high-throughput, proteome-127 wide programmable phage display approach (PhIP-Seq) to query the antibody target identities contains approximately 700,000 unique phage, each displaying a 49 amino acid proteome segment. 134 As previously described, phage were immunoprecipitated using human antibodies bound to protein 135 A/G beads. In order to increase sensitivity and specificity for target proteins, eluted phage were 136 used for a further round of amplification and immunoprecipitation. DNA was then extracted from 137 the final phage elution, amplified and barcoded, and subjected to Next-Generation Sequencing 138 ( Figure 1A). Finally, sequencing counts were normalized across samples to correct for variability 139 in sequencing depth, and the fold-change of each gene was calculated (comprised of multiple 140 unique tiling phage) as compared to mock IPs in the absence of human serum (further details of 141 the protocol can be found in the methods section). 142  From a cohort of 67 APS1 serum samples, a total of 39 samples were subjected to PhIP-143   Seq investigation, while the remaining 28 samples were obtained at a later time point and reserved  144   for downstream validation experiments (for clinical data, refer to Supplemental Table 1). In 145 addition, 28 non-APS1 anonymous blood donor serum samples were subjected to PhIP-Seq, and positives, a conservative set of criteria were used as follows. We required a minimum of 2/39 156 APS1 samples and 0/28 non-APS1 control samples to exhibit normalized gene counts in the 157 immunoprecipitation (IP) with greater than 10-fold enrichment as compared to the control set of 158 18 mock-IP (beads, no serum) samples. This simple, yet stringent criteria enabled detection of a 159 total of 23 known autoantibody specificities ( Figure 1B). Importantly, many of the well-validated 160 APS1 antigens, including specific members of the cytochrome P450 family (CYP1A2, CYP21A1, 161 CYP11A1, CYP17A1), lung disease-associated antigen KCRNG, as well as IL17A, IL17F, and 162 IL22, among others were well represented ( Figure 1B). In contrast, the diabetes-associated 163 antigens GAD65 and INS did not meet these stringent detection criteria and only weak signal was 164 detected to many of the known interferon autoantibody targets known to be present in many people 165 with APS1, perhaps due to the conformational nature of these autoantigens ( Figure Ziegler et al., 1996). 168 Three known autoantigens that were prevalent within our cohort were selected to determine 169 how PhIP-Seq performed against an orthogonal whole protein-based antibody detection assay. A Importantly, and in contrast to PhIP-Seq, this assay tests for antibody binding to full-length protein 174 ( Figure 1C). By RLBA, these three antigens were present in and specific to both the initial 175 discovery APS1 cohort (n=39) as well as the expanded validation cohort (n = 28), but not the non-176 APS1 control cohort (n = 61). Together, these results demonstrate that PhIP-Seq detects known 177 APS1 autoantigens and that PhIP-Seq results validate well in orthogonal whole protein-based 178

assays. 179
To determine whether the PhIP-Seq APS1 dataset could yield higher resolution information 180 on antigenic peptide sequences with respect to previously reported targets, the normalized 181 enrichments of all peptides belonging to known disease-associated antigens CYP11A1 and SOX10 suggesting peptide-level commonalities and convergence among the autoreactive antibody 186 repertoires across individuals. These data suggest that people with APS1 often target similar, but 187 not identical protein regions. 188

Identification of novel APS1 autoantigens 190
Having confirmed that PhIP-Seq analysis of APS1 sera detected known antigens, the same 191 data were then investigated for the presence of novel, previously uncharacterized APS1 192 autoantigens. We applied the same positive hit criteria as described for known antigens, and 193 additionally increased the required number of positive APS1 samples to 3/39 to impose a stricter 194 limit on the number of novel candidate autoantigens. This yielded a list of 82 genes, which included 195 10 known antigens and 72 putative novel antigens (Figure 2). 196 The most commonly held hypotheses regarding the nature and identity of proteins targeted 197 by the aberrant immune response in APS1 are that targeted proteins (1) tend to exhibit AIRE-198 dependent thymic expression and (2) have restricted expression to one or few peripheral organs 199 and tend not to be widely or ubiquitously expressed. We investigated whether our novel antigens 200 were also preferentially tissue-restricted. In order to systematically address this question, tissue-201 specific RNA expression was assessed using a consensus expression dataset across 74 cell types 202 and tissues (Uhlen et al., 2015). For each gene, the ratio of expression in the highest tissue as 203 compared to the sum of expression across all tissues was calculated, resulting in higher ratios for 204 those mRNAs with greater degrees of tissue-restriction. Using this approach, the mean tissue-

Identification of novel antigens common to many individuals 210
Identified autoantigens were ranked by frequency within the cohort. Five antigens were 211 positive in ten or more APS1 samples, including two novel antigens. In addition, the majority of 212 antigens found in 4 or more APS1 sera were novel ( Figure 3A).   Table 2). Using a whole-protein radiolabeled 226 binding assay (RLBA) for validation, all seven proteins were immunoprecipitated by antibodies in 227 both the PhIP-Seq APS1 discovery cohort (n=39), as well as in the validation cohort of APS1 sera 228 that had not been interrogated by PhIP-Seq (n=28). Whereas an expanded set of non-APS1 controls 229    In this cohort, 82% (55/67) of APS1 sera exhibited an RFX6 signal that was at least 3 296 standard deviations above the mean of non-APS1 control signal due to the extremely low RFX6 297 signal across all non-APS1 controls by RLBA ( Figure 3B). Using a more stringent cutoff for 298 RFX6 positivity by RLBA at 6 standard deviations above the mean, 65% of APS1 samples were The extent and frequency of intestinal dysfunction in people with APS1 has only recently 307 been clinically uncovered and reported, and therefore still lacks unifying diagnostic markers as 308 well as specific intestinal target antigen identities (Ferré et al., 2016). This investigation of APS1 309 sera revealed several antigens that are expressed in the intestine, including RFX6, GIP, PDX1, and 310 NKX6-3. We chose to further study whether autoimmune response to RFX6+ cells in the intestine 311 was involved in APS1-associated intestinal dysfunction. Using a publicly available murine single- RFX6 is also expressed in the pancreas, we also examined the association of anti-RFX6 antibodies 339 with APS1-associated type 1 diabetes. We observed that 6/7 APS1-associated type 1 diabetes  3B). However, due to small sample size, an expanded cohort would be needed to determine the 342 significance of this observation. Together, these data suggest that RFX6 is a common, shared 343 autoantigen in APS1 that may be involved in the immune response to intestinal enteroendocrine 344 cells as well as pancreatic islets. Future studies will help to determine whether testing for anti-345  Testing for defined autoantibody specificities provides substantial clinical benefit for 390 prediction and diagnosis of autoimmune disease. A primary goal of this study was to identify 391 autoantigens with potential clinical significance; consistently, our analyses focused primarily on 392 antigens that appeared across multiple samples, rather than autoantigens that were restricted to 393 individual samples. Using conservative inclusion criteria, we discovered 72 novel autoantigens 394 that were shared across a minimum of 3 APS1 samples, of which 7/7 were successfully validated 395 at the whole protein level. Overall, we have expanded the known repertoire of common APS1 396 antigens, confirming that the antibody target repertoire of common antigens in APS1 is larger than 397 previously appreciated. Interestingly, our data also suggest that the size of the commonly 398 autoantibody-targeted repertoire of proteins is dramatically lower than the number of genes 399 (~4000) that exhibit AIRE-dependent thymic expression. While we report many novel antigens, we also acknowledge that the relationship between 452 autoantibody status and disease is often complicated. This concept can be illustrated by examining 453 the well-established autoantibody specificities in autoimmune diabetes (Taplin & Barker, 2009). 454 First, islet autoantibodies (GAD65, ZNT8, etc.) can be found within non-autoimmune sera, where 455 they are thought to represent an increased risk of developing disease as compared to the antibody-456 negative population. Second, not all patients with autoimmune diabetes are autoantibody positive. 457 In sum, while autoantibodies can be extremely useful for risk assessment as well as for diagnosis, 458 they often lack high sensitivity and specificity; both of these caveats can result in difficulties 459 detecting strong clinical associations. For example, anti-ACP4 antibodies are highly prevalent in 460 our cohort, but they exhibit only a trending association with dental enamel hypoplasia despite the 461 strong biological evidence that ACP4 dysfunction leads to enamel hypoplasia (Seymen et al., 2016;462 C. E. Smith et al., 2017). Our data in humans is currently insufficient to determine whether immune 463 responses to novel antigens such as ACP4 are pathogenic, indirectly linked to risk of disease, or 464 instead simply represent a B-cell bystander effect. To better address these questions, we propose 465 that future studies in mouse models could elucidate whether immune response to specific proteins, 466 including ACP4, can result in the proposed phenotypes. 467 As the spectrum of diseases with potential autoimmune etiology continues to expand, the 468 characteristic multiorgan autoimmunity in APS1 provides an ideal model system to more broadly 469 approach the question of which proteins and cell types tend to be aberrantly targeted by the immune 470 system. The data presented here has illuminated a collection of novel human APS1 autoimmune 471 targets, as well as a novel antibody-disease association between RFX6 and diarrheal-type intestinal 472 dysfunction, a highly prevalent disorder in APS1 that has until now lacked clinically applicable 473 predictive or diagnostic markers. In sum, this data has significantly expanded the known

PhIP-Seq Analysis 501
Sequencing reads from fastq files were aligned to the reference oligonucleotide library and peptide 502 counts were subsequently normalized by converting raw reads to percentage of total reads per 503 sample. Peptide and gene-level enrichments for both APS1 and non-APS1 sera were calculated by 504 determining the fold-change of read percentage per peptide and gene in each sample over the mean 505 read percentage per peptide and gene in a background of mock-IP (A/G bead only, n = 18). 506 Individual samples were considered positive for genes where the enrichment value was 10-fold or 507 greater as compared to mock-IP. For plotting of multiple genes in parallel (Figures 1 & 2), 508 enrichment values were z-scored and hierarchically clustered using Pearson correlation.  Table 3). The cpms of immunoprecipitated protein was quantified using a 96-well 572 Microbeta Trilux liquid scintillation plate reader (Perkin Elmer). d.