Autoantibody discovery across monogenic, acquired, and COVID-19-associated autoimmunity with scalable PhIP-seq

  1. Sara E Vazquez
  2. Sabrina A Mann
  3. Aaron Bodansky
  4. Andrew F Kung
  5. Zoe Quandt
  6. Elise MN Ferré
  7. Nils Landegren
  8. Daniel Eriksson
  9. Paul Bastard
  10. Shen-Ying Zhang
  11. Jamin Liu
  12. Anthea Mitchell
  13. Irina Proekt
  14. David Yu
  15. Caleigh Mandel-Brehm
  16. Chung-Yu Wang
  17. Brenda Miao
  18. Gavin Sowa
  19. Kelsey Zorn
  20. Alice Y Chan
  21. Veronica M Tagi
  22. Chisato Shimizu
  23. Adriana Tremoulet
  24. Kara Lynch
  25. Michael R Wilson
  26. Olle Kämpe
  27. Kerry Dobbs
  28. Ottavia M Delmonte
  29. Rosa Bacchetta
  30. Luigi D Notarangelo
  31. Jane C Burns
  32. Jean-Laurent Casanova
  33. Michail S Lionakis
  34. Troy R Torgerson
  35. Mark S Anderson
  36. Joseph L DeRisi

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This work presents a series of enhancements to the PhIP-seq method of autoantibody discovery, with the goal of improving scaling to larger cohorts and increasing disease specificity. The strength of the paper is the validation of the high throughput format, although results from screening patient samples confirm or only modestly extend previous data.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Phage immunoprecipitation sequencing (PhIP-seq) allows for unbiased, proteome-wide autoantibody discovery across a variety of disease settings, with identification of disease-specific autoantigens providing new insight into previously poorly understood forms of immune dysregulation. Despite several successful implementations of PhIP-seq for autoantigen discovery, including our previous work (Vazquez et al., 2020), current protocols are inherently difficult to scale to accommodate large cohorts of cases and importantly, healthy controls. Here, we develop and validate a high throughput extension of PhIP-seq in various etiologies of autoimmune and inflammatory diseases, including APS1, IPEX, RAG1/2 deficiency, Kawasaki disease (KD), multisystem inflammatory syndrome in children (MIS-C), and finally, mild and severe forms of COVID-19. We demonstrate that these scaled datasets enable machine-learning approaches that result in robust prediction of disease status, as well as the ability to detect both known and novel autoantigens, such as prodynorphin (PDYN) in APS1 patients, and intestinally expressed proteins BEST4 and BTNL8 in IPEX patients. Remarkably, BEST4 antibodies were also found in two patients with RAG1/2 deficiency, one of whom had very early onset IBD. Scaled PhIP-seq examination of both MIS-C and KD demonstrated rare, overlapping antigens, including CGNL1, as well as several strongly enriched putative pneumonia-associated antigens in severe COVID-19, including the endosomal protein EEA1. Together, scaled PhIP-seq provides a valuable tool for broadly assessing both rare and common autoantigen overlap between autoimmune diseases of varying origins and etiologies.

Article activity feed

  1. Version published to 10.7554/elife.78550 on eLife
  2. eLife

    Author Response

    Public Evaluation Summary:

    This work presents a series of enhancements to the PhIP-seq method of autoantibody discovery, with the goal of improving scaling to larger cohorts and increasing disease specificity. The strength of the paper is the validation of the high throughput format, although results from screening patient samples confirm or only modestly extend previous data.

    We thank the reviewers for their feedback and agree that the validation of our high throughput, easily accessible approach is a strength of this work. We appreciate that the reviewers expressed uncertainty about whether there were sufficient advances to qualify this paper as a Research Advance. In addition to a point-by-point rebuttal, we quantify and enumerate the advances, improvements, and novel findings disclosed in this manuscript, relative to our original eLife paper.

    1. Demonstration of the importance of adequate healthy control cohorts in PhIP-Seq design. Using scaled protocols, we demonstrate the importance of using large control cohorts to filter out non-specific hits, as well as to detect rare but specific disease-associated antigens such as PDYN. To our knowledge, we are the first to demonstrate and discuss the consequences of PhIP-Seq dataset interpretation in the absence of sufficient controls. These findings are especially important in light of recent, high-impact papers using few to no controls (Mina et al. Science 2019, Gruber et al. Cell 2020, among others) to make conclusions about novel autoantibodies in the context of specific diseases.

    2. Design, validation and documentation of accessible, benchtop protocols for scaled PhIP-Seq. These protocols enable parallel testing of 600-800 samples without contamination or batch effects. Using a substantially expanded, multi-cohort set of patients with APS1, we validate the quality of the protocol and apply this protocol to numerous other disease contexts. Importantly, our protocols are documented (protocols.io) with each step tested for optimal quality, and are easily accessible without the need for robotics or specialized equipment.

    3. Machine Learning for disease classification using phage-based immunoprofiling. We show that large, well-controlled PhIP-Seq datasets lend well to machine learning approaches and enable unsupervised classification of disease status. To our knowledge, this is the first successful application of an unsupervised machine learning approach to phage-based immunoprofiling data. We demonstrate that PhIP-Seq data enables APS1 disease classification in 97% of cases (compare even to the 95% sensitivity seen in current testing for anti-IFN antibodies in the setting of suspected APS1). This finding, while applied to only one large cohort, demonstrates that PhIP-Seq data, when appropriately controlled, can have substantial value outside of simply a single-antigen discovery platform. The combination of machine learning and phage-based immunoprofiling will likely have extensive applications beyond APS1 including the discovery of novel diagnostic tests and biomarkers.

    4. Novel IPEX antigen BTNL8. We discovered and validated anti-BTNL8 antibodies in 42% of IPEX patients, suggesting that this may be a major autoantigen in IPEX. BTNL8 is a cell surface-expressed protein in intestinal gamma-delta T-cells, raising the novel question of a possible role for autoantibodies in directly regulating gut epithelial immune homeostasis (see discussion, lines 540-551). This is the first report, not only of BTNL8, but of any antigen discovery by PhIP-seq immunoprofiling in IPEX patients. Given the importance of this discovery, we sought to validate the presence of these autoantibodies in an additional validation cohort. We were successful, and present these findings in the new Figure 5., highlighting the generalizability of our findings to IPEX patients.

    5. BEST4 autoantibodies in IPEX and RAG-hypomorphic patients. We discovered anti-BEST4 antibodies in 15% of patients with IPEX, as well as in 2 patients with RAG1/2 mutations, demonstrating a connection between the intestinal autoimmunity seen in both IPEX and RAG1/2 deficiency. Of note, one of the 2 positive RAG1/2 deficient patients with anti-BEST4 antibodies is known to have very-early-onset IBD (VEO-IBD), a rare sub-phenotype in RAG-hypomorphs (and other primary immune deficiencies). Given the severity of VEO-IBD and how little is known about why certain patients with immune dysregulation develop this phenotype, these findings mark an important scientific advance and provide an essential clue into etiology. Furthermore, given that IPEX is driven by dysfunctional Treg cells, the commonality of these findings in both IPEX and hypomorphic RAG indicate a potential role for Treg dysfunction in hypomorphic RAG.

    6. Expansion of scaled PhIP-Seq to interrogate severe COVID-19 pneumonia, Kawasaki disease (KD), and Multisystem Inflammatory Syndrome in Children (MIS-C). Importantly, in MIS-C we find no evidence for any of the previously reported autoantigens described in Gruber at al (Cell, 2020) – a study which made strong conclusions about autoantibodies despite featuring only 4 PhIP-Seq control samples. Our results highlight the importance of scaling and appropriate control groups, and caution against overinterpretation of reported disease-specific autoantigens in PhIP-Seq (or other expanded antigen screening technologies such as near-proteome wide fixed protein arrays) which utilize smaller control cohorts, often without orthogonal validation experiments.

    7. Anti-CGNL1 antibodies in KD/MIS-C. We discovered and validated autoantibodies to CGNL1 in KD and MIS-C. It is possible that these antibodies represent a subset of specificities within anti-endothelial cell antibodies, given the endothelial expression of CGNL1 as well as its implications in cardiovascular disease.

    Reviewer #2 (Public Review):

    The authors update PhIP-seq into a high throughput format with the goal to accommodate screening of large numbers of human patient sera for the presence of novel autoantibodies and screening of more control sera to better determine standards for positivity of experimental samples. The high throughput protocol is detailed in an associated web-based format and validated in the paper using sera from patients with inherited immunodeficiencies and patients with MIS-C, Kawasaki syndrome, and COVID19. These are strengths of the work, and the high throughput PhIP-seq format will be useful to other investigators doing similar screenings. Yet, the findings do not significantly extend our knowledge of the range of autoantibodies in these illnesses, and many of the autoantibodies detected using PhIP-seq linear epitopes are not validated with other strategies, limiting significance of the results. The data from MIS-C and Kawasaki cohorts are confounded by an undetermined number of IVIG treated subjects, and limited numbers of control samples, including sera from patients with febrile illnesses that contain autoantibodies that are not discussed in the context of findings from the experimental groups.

    In summary, the paper is solid technically, with the high throughput strategy seemingly well validated; however, the advance here is primarily a technical one.

    We thank the reviewer and agree that the technical advance here is substantial and will be of value to other investigators doing similar screenings – as well as to investigators who previously did not have access to this technology due to high requirements for robotics and specialized equipment in previous iterations of the protocol. As such, we feel that this, combined with the demonstration of how to appropriately control PhIP-Seq experiments, should be considered a valuable research advance alone -- even in the absence of the extensive validation and novel findings on 5 additional disease contexts, summarized in greater detail above.

    IVIG status is discussed in lines 417-423. Briefly, the large majority of MISC samples are confirmed to be IVIG free at the time of blood draw. All of our KD samples are confirmed IVIG-free.

    While pediatric febrile illness samples could conceivably contain autoantibodies, we believe that this is best group for comparison given that these samples are taken from age-matched, acutely ill patients, thus providing a control group that is as clinically similar to MIS-C as possible. In addition, we included adult healthy sera and adult COVID19 sera as secondary control groups. Of note, this matching is much more extensive (and substantially larger in number) than the recent study in Cell (Gruber at el 2020), which for PhIP-Seq used only 4 healthy, COVID19-negative samples to compare to 9 MISC samples.

    Reviewer #3 (Public Review):

    This paper presents a rigorously performed series of studies to improve the ability of the PhIP-seq method to discover autoantibodies against peptide antigens that span the whole peptidome at scale, and increase the ease of validation and definition of disease specificity. The paper is an extension of a recent paper from the DeRisi and Anderson groups done on APS1 patients, which defined and validated a novel series of tissue-specific autoantigens in APS1. The current studies show that the authors can find the antibodies they previously defined, and using larger numbers of disease and control samples, can expand some what they detect. They then use the new method to look at multiple additional processes in which autoimmunity has been demonstrated/postulated.

    The dataset may be of use to others interested in defining novel autoantibodies. The findings really did not share significant new insights into the processes they studied,. As the authors note, they were unable to detect the antibodies (~10% of patients) recognizing type I IFNs in severe COVID-19, where these had been demonstrated effectively using ELISA previously. Unlike APS1, where their findings about uncommon tissue specific autoantibody responses across a population with known genetic deficiency and heterogeneous phenotypes could really illustrate the power of the method and approach, that elegance and powerful and novel conclusion is not as evident here.

    The trade-off between sensitivity, specificity, and screening power of antigen discovery tools is present in every assay. We do not feel that the comparison of our assay to a single protein ELISA assay is appropriate (nor particularly relevant for the conclusions drawn in this manuscript) given the inherent difference in nature and goals of the two assays. It has long been understood that PhIP-Seq does not have sensitivity for all protein antigens, including post-translationally modified and conformational antigens, which we state for readers in lines 190-193, within the discussion section, as well as in our previous work.

  3. eLife

    Evaluation Summary:

    This work presents a series of enhancements to the PhIP-seq method of autoantibody discovery, with the goal of improving scaling to larger cohorts and increasing disease specificity. The strength of the paper is the validation of the high throughput format, although results from screening patient samples confirm or only modestly extend previous data.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    The authors have modified protocols for Phage Immunoprecipitation sequencing or PhIP-seq to allow much larger throughput and have examined value of this platform for auto-antigen discovery. Overall the manuscript is technically sound. The finding of shared auto-antigens in Kawasaki Disease and MIS-C was of interest.

  5. eLife

    Reviewer #2 (Public Review):

    The authors update PhIP-seq into a high throughput format with the goal to accommodate screening of large numbers of human patient sera for the presence of novel autoantibodies and screening of more control sera to better determine standards for positivity of experimental samples. The high throughput protocol is detailed in an associated web-based format and validated in the paper using sera from patients with inherited immunodeficiencies and patients with MIS-C, Kawasaki syndrome, and COVID19. These are strengths of the work, and the high throughput PhIP-seq format will be useful to other investigators doing similar screenings. Yet, the findings do not significantly extend our knowledge of the range of autoantibodies in these illnesses, and many of the autoantibodies detected using PhIP-seq linear epitopes are not validated with other strategies, limiting significance of the results. The data from MIS-C and Kawasaki cohorts are confounded by an undetermined number of IVIG treated subjects, and limited numbers of control samples, including sera from patients with febrile illnesses that contain autoantibodies that are not discussed in the context of findings from the experimental groups.

    In summary, the paper is solid technically, with the high throughput strategy seemingly well validated; however, the advance here is primarily a technical one.

  6. eLife

    Reviewer #3 (Public Review):

    This paper presents a rigorously performed series of studies to improve the ability of the PhIP-seq method to discover autoantibodies against peptide antigens that span the whole peptidome at scale, and increase the ease of validation and definition of disease specificity. The paper is an extension of a recent paper from the DeRisi and Anderson groups done on APS1 patients, which defined and validated a novel series of tissue-specific autoantigens in APS1. The current studies show that the authors can find the antibodies they previously defined, and using larger numbers of disease and control samples, can expand some what they detect. They then use the new method to look at multiple additional processes in which autoimmunity has been demonstrated/postulated.

    The dataset may be of use to others interested in defining novel autoantibodies. The findings really did not share significant new insights into the processes they studied,. As the authors note, they were unable to detect the antibodies (~10% of patients) recognizing type I IFNs in severe COVID-19, where these had been demonstrated effectively using ELISA previously. Unlike APS1, where their findings about uncommon tissue specific autoantibody responses across a population with known genetic deficiency and heterogeneous phenotypes could really illustrate the power of the method and approach, that elegance and powerful and novel conclusion is not as evident here.

  7. ScreenIT

    SciScore for 10.1101/2022.03.23.485509: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    EthicsIRB: APS1: North America - 1: All patient cohort data was collected and evaluated at the NIH, and all APECED/ APS1 patients were enrolled in a research study protocol approved by the NIH Institutional Review Board Committee and provided with written informed consent for study participation (protocol #11-I-0187, NCT01386437).
    Consent: APS1: North America - 1: All patient cohort data was collected and evaluated at the NIH, and all APECED/ APS1 patients were enrolled in a research study protocol approved by the NIH Institutional Review Board Committee and provided with written informed consent for study participation (protocol #11-I-0187, NCT01386437).
    Sex as a biological variablenot detected.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.

    Table 2: Resources

    Antibodies
    SentencesResources
    #2272S) and/or anti-FLAG (Cell Signaling Technologies, #1493S) antibodies were used as positive controls for immunoprecipitations.
    anti-FLAG
    suggested: None
    Software and Algorithms
    SentencesResources
    We applied a logistic regression classifier on log-transformed PhIP-Seq RPK values from APS1 patients (n = 128) versus healthy controls (n = 186) using the scikit-learn package (Pedregosa et al 2011) with a liblinear solver and L1 regularization.
    scikit-learn
    suggested: (scikit-learn, RRID:SCR_002577)
    North America - 2: All patients underwent informed consent with research study protocols approved by the UCSF Human Research Protection Program (IRB# 10-02467).
    UCSF Human Research Protection Program
    suggested: None

    Results from OddPub: Thank you for sharing your data.

    Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:
    A practical limitation of this technique is that non-specific phage may also be amplified, in addition to a wide array of autoreactive, but non-disease related peptides. We tested whether we could detect global differences between case and control cohorts as a measure of autoreactivity. As each APS1 patient is known to have multiple, high-affinity antibodies to self-proteins (Fishman et al., 2017; Landegren et al., 2016; Meyer et al., 2016; Vazquez et al., 2020) we reasoned that this would be an ideal cohort to determine whether a global autoreactive state was discernible. As expected, each individual sample exhibits a spectrum of enriched genes, regardless of disease status (Figure 2A), indicating that measures of simple enrichment are inadequate for discrimination of cases from controls. We and others have shown that PhIP-Seq can robustly detect disease-associated antigens by comparing antigen-specific signal between disease and control cohorts (Larman et al., 2011; Mandel-Brehm et al., 2019; O’Donovan et al., 2020; Vazquez et al., 2020). In this dataset, encompassing 186 control samples and 128 APS1 samples, we further evaluated the importance of control cohort size. We iteratively down sampled the number of healthy control samples in our dataset to 5, 10, 25, 50, 100, or 150 (out of n=186 total control samples). The number of apparent hits was determined in each condition, where a gene-level hit was called when the following criteria were met: 1) at least 10% of APS1 samp...

    Results from TrialIdentifier: We found the following clinical trial numbers in your paper:

    IdentifierStatusTitle
    NCT01386437RecruitingNatural History of Individuals With Immune System Problems T…
    NCT03394053RecruitingThe Mechanistic Biology of Primary Immunodeficiency Disorder…
    NCT03610802RecruitingSend-In Sample Collection to Achieve Genetic and Immunologic…

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    Results from scite Reference Check: We found no unreliable references.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  8. Version published to 10.1101/2022.03.23.485509v1 on bioRxiv