Technical AdvanceImmunology Open Access | 10.1172/jci.insight.174976

Elze Rackaityte,1 Irina Proekt,2 Haleigh S. Miller,1,3 Akshaya Ramesh,4 Jeremy F. Brooks,5 Andrew F. Kung,1,3 Caleigh Mandel-Brehm,1 David Yu,2 Colin R. Zamecnik,4 Rebecca Bair,4 Sara E. Vazquez,1,2 Sara Sunshine,1 Clare L. Abram,6 Clifford A. Lowell,6 Gabrielle Rizzuto,7 Michael R. Wilson,4 Julie Zikherman,5 Mark S. Anderson,2 and Joseph L. DeRisi1,8

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Rackaityte, E. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Proekt, I. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Miller, H. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Ramesh, A. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Brooks, J. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Kung, A. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Mandel-Brehm, C. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Yu, D. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Zamecnik, C. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Bair, R. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Vazquez, S. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Sunshine, S. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Abram, C. in: JCI | PubMed | Google Scholar

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Lowell, C. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Rizzuto, G. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Wilson, M. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Zikherman, J. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by Anderson, M. in: JCI | PubMed | Google Scholar |

1Department of Biochemistry and Biophysics,

2Diabetes Center, School of Medicine,

3Biological and Medical Informatics Program,

4Weill Institute for Neurosciences, Department of Neurology, School of Medicine,

5Division of Rheumatology, Rosalind Russell and Ephraim P. Engleman Rheumatology Research Center, Department of Medicine, and

6Department of Laboratory Medicine, UCSF, San Francisco, California, USA.

7Human Oncology & Pathogenesis Program and Department of Pathology & Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, USA.

8Chan Zuckerberg Biohub, San Francisco, California, USA.

Address correspondence to: Joseph L. DeRisi, 1700 4th St., QB3 Room 404, San Francisco, California 94158-2330, USA. Phone: 415.418.3647; Email: joe@derisilab.ucsf.edu.

Find articles by DeRisi, J. in: JCI | PubMed | Google Scholar |

Published November 7, 2023 - More info

Published in Volume 8, Issue 23 on December 8, 2023
JCI Insight. 2023;8(23):e174976. https://doi.org/10.1172/jci.insight.174976.
© 2023 Rackaityte et al. This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Published November 7, 2023 - Version history
Received: August 21, 2023; Accepted: October 25, 2023 View PDF Abstract

Autoimmunity is characterized by loss of tolerance to tissue-specific as well as systemic antigens, resulting in complex autoantibody landscapes. Here, we introduce and extensively validate the performance characteristics of a murine proteome-wide library for phage display immunoprecipitation and sequencing (PhIP-seq) in profiling mouse autoantibodies. This library was validated using 7 genetically distinct mouse lines across a spectrum of autoreactivity. Mice deficient in antibody production (Rag2–/– and μMT) were used to model nonspecific peptide enrichments, while cross-reactivity was evaluated using anti-ovalbumin B cell receptor–restricted OB1 mice as a proof of principle. The PhIP-seq approach was then utilized to interrogate 3 distinct autoimmune disease models. First, serum from Lyn–/– IgD+/– mice with lupus-like disease was used to identify nuclear and apoptotic bleb reactivities. Second, serum from nonobese diabetic (NOD) mice, a polygenic model of pancreas-specific autoimmunity, was enriched in peptides derived from both insulin and predicted pancreatic proteins. Lastly, Aire–/– mouse sera were used to identify numerous autoantigens, many of which were also observed in previous studies of humans with autoimmune polyendocrinopathy syndrome type 1 carrying recessive mutations in AIRE. These experiments support the use of murine proteome-wide PhIP-seq for antigenic profiling and autoantibody discovery, which may be employed to study a range of immune perturbations in mouse models of autoimmunity profiling.

Introduction

Autoimmune diseases arise from a complex breakdown in immune tolerance and are frequently characterized by the presence of autoantibodies and autoreactive T cells. Autoimmunity spans a breadth of different clinical subtypes and patterns, with almost any organ system or tissue being susceptible. Defining the specificity and origins of the autoimmune response is key for developing methods to diagnose, prevent, and treat this family of diseases. On a mechanistic level, a key tool for unraveling autoimmunity has been the use of mouse models of human autoimmune diseases. A combination of autoimmune-susceptible mouse strains or mouse lines with models of human genetic defects has played a key role in our understanding in the pathogenesis of an array of autoimmune diseases. For example, use of genetically altered mice has allowed for our understanding of how the monogenic autoimmune diseases autoimmune polyendocrine syndrome type 1 (APS1) (1) and immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX) (2, 3) are linked to defects in thymic central tolerance or the function of T regulatory cells, respectively. Given the highly controlled nature of mouse modeling for both environmental and genetic influences, it remains an essential tool for dissecting autoimmunity. An important aspect to this work is defining the autoimmune response in these mouse models. In this regard, a typical approach has been to search for autoantibodies from affected mice in targeted assays such as Western blotting and indirect immunofluorescence. Recently, there has been rapid development of new approaches to identifying autoantibody specificities that broadly cover the entire proteome in the human setting (4–9). Thus, a similar approach in a mouse model could serve as an important method to further define and unravel autoimmunity.

Phage display immunoprecipitation and sequencing (PhIP-seq) is a powerful tool to identify antibody targets, originally described by Larman, Elledge, and colleagues (4, 10). Since 2011, it has been used to discover novel antibody autoreactivities in a wide range of human syndromes, including paraneoplastic diseases (6, 7) and inborn autoimmune syndromes (8, 9, 11). In PhIP-seq, libraries of long oligonucleotides encoding overlapping peptides are synthesized as DNA oligomers and cloned into the T7 phage genome. These libraries are expressed fused to gene 10 of the surface-exposed capsid protein on lytic T7 phage and used as bait for antibodies in patient sera. Complex, multiantigen immunoprecipitants are deconvoluted by sequencing the enriched phage-encoded peptides to identify multiple antibody targets in a single reaction.

Due to the programmable nature of PhIP-seq, any proteome may be comprehensively encoded in a phage library in principle. Library designs may also be highly customized, including coverage of specific protein isoforms, putative coding regions, and other features. Here, we present the construction and validation of a murine proteome-wide PhIP-seq library, based on the GRCm38.p5 Mus musculus genome, composed of over 480,000 peptides, representing over 76,000 protein sequences. Taking advantage of genetic manipulations available in the mouse model, library performance was evaluated across 7 mouse strains, including Rag2–/–, those expressing genetically modified immunoglobulin M transmembrane domains (μMT mice), C57BL/6J (wild-type strain, B6), OB1, Lyn–/–, polygenic nonobese diabetic (NOD), and Aire–/–. Mice lacking mature B cells (Rag2–/– and μMT) were used to determine proteome-wide background binding in immunoprecipitations (IPs), while serum from B6, OB1, Lyn–/–, NOD, and Aire–/– mice was used to identify strain-specific autoreactivities. Building on our previous identification of autoantibodies against perilipin-1 (Plin1) in Aire–/– mice (9), we identified the binding epitope of these antibodies and demonstrate their relationship to immune cell infiltrates in adipose tissue in affected mice. Taken together, these results demonstrate the utility of the approach for a broad assessment of the array of autoimmune specificities in various mouse models.

Results

Design and construction of murine proteome-wide library. To construct a Mus musculus proteome-wide library, the GRCm38.p5 reference proteome sequences, including all isoforms, were downloaded from the NCBI and divided into 62–amino acid peptide tiles with 19–amino acid overlaps (Figure 1A). The library was supplemented with several positive and negative control peptides, including those derived from human glial fibrillary acid protein (GFAP), human tubulin, GFP, and others. The resulting library of 482,672 peptides was synthesized (Agilent) as a DNA oligomer pool (Figure 1A) and cloned into T7 phage fused in frame with gene 10, which encodes the capsid protein of T7 phage. The complete peptide design file and further details are freely available as a companion to this manuscript on protocols.io (see Methods). The synthesized oligomer library and the packaged library were sequenced by next-generation sequencing with 173 and 79 million paired-end 147–base pair reads on an Illumina NovaSeq 6000, respectively, which resulted in an approximate 360× and 163× coverage of the library, respectively. Alignment of the reads from sequencing of the pooled DNA oligomer library yielded 89.2% identical matches, with greater than 99.9% of all the expected peptides represented. Sequencing of the packaged T7 phage library yielded an alignment rate of 78.9%, yet representation remained high at greater than 99% of the expected peptides (Figure 1B).

Figure 1

Design and validation of murine PhIP-seq library. (A) GRCm38.p5 annotated proteins were downloaded from Refseq and 62–amino acid (62-aa) tiles were chosen to cover the 76,217 proteins with 482,672 peptides with a 19-aa overlap. The tiles contained necessary cloning sites for expression in a T7 phage display system. (B) Representation of designed oligonucleotides after oligonucleotide synthesis and cloning. (C) Sum of all fold changes (FCs) above the mean read counts in mock IP in each experimental sample (mock IP) or mouse strain (Rag2–/–, μMT, OB1, B6, and Lyn–/– IgD+/–) by PhIP-seq. Exact P value is reported, and each dot corresponds to a mouse or mock-IP replicate. Kruskal-Wallis test with Tukey’s HSD post hoc test.

Evaluation of library performance and background binding in Rag2–/– and μMT mice. The performance of the packaged library was first benchmarked utilizing a commercial antibody with a known specificity. Previously, we have utilized a commercial anti-GFAP polyclonal antibody as a positive control for human PhIP-seq libraries, due to its consistent IP performance (8, 11–13). The murine PhIP-seq library contains both human and mouse GFAP sequences and binding to sequences of both species was expected with the commercial antibody. Antibody bound to a mix of protein A and protein G magnetic beads was used to IP phage from the murine library, followed by either sequencing or further amplification in Escherichia coli. As observed previously with the human PhIP-seq library (8, 13), additional rounds of phage IP followed by amplification in E. coli resulted in increased enrichment of the target sequences. Using 2 rounds, approximately 20% of resulting phage encoded peptides derived from either human GFAP or mouse Gfap (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.174976DS1). Using 3 rounds, the enrichment approached 50%, or 1 × 108–fold greater than the amount of the same phage in the starting library (Supplemental Figure 1A). Given the significant enrichment of Gfap peptides relative to nonspecific peptides, all subsequent experiments utilized 3 rounds of IP and amplification.

The library was next evaluated across 5 mouse strains on the B6 background. Two strains (Rag2–/– and μMT) lacking IgG (14, 15) were utilized to evaluate background binding. When compared with a mock IP control lacking serum, sera from both Rag2–/– and μMT mice failed to significantly enrich phage from the murine library, as expected (Figure 1C and Supplemental Figure 1B). At the individual peptide level, fewer than 10 peptides were consistently enriched (at least 2 of 3 mice) by sera from either mouse strain, presumably through nonspecific interactions (Supplemental Figure 1C). For subsequent experiments, the mean frequency of each phage across mock IP, Rag2–/–, and μMT was determined and used to calculate fold-change and z scores for experimental samples (murine background model, MBM).

B cells in OB1 mice harbor physiological B cell receptor (BCR) rearrangements in the IgH and Igκ loci that encode for a clonal IgG1 receptor with specificity for the chicken ovalbumin (OVA) protein; thus, their sera contains predominantly high-affinity anti-OVA IgG1 antibodies (16). Unlike Rag2–/– and μMT mice, sera from OB1 mice yielded a moderate amount of enrichment consistent with the restricted B cell repertoire of this genetic strain (Figure 1C and Supplemental Figure 1B). In contrast, sera from wild-type B6 mice yielded significant enrichment, with 3.2-fold more enrichment than OB1 sera (Figure 1C and Supplemental Figure 1B). Finally, Lyn–/– IgD+/– mice were also examined, which exhibit polyclonal B cell activation and develop lupus-like disease (17, 18). Consistent with a greater degree of autoreactivity in these mice, Lyn–/– IgD+/– sera yielded the largest number of significantly enriched peptides, with higher fold-change than sera from wild-type B6 mice (Figure 1C and Supplemental Figure 1B).

Proteome-wide PhIP-seq identifies OB1 reactivity to known epitopes in mouse proteome. Previously, the recognition site for immunoglobulin binding to the OVA protein in the OB1 strain was determined as DKLPGFGDSI by alanine mutagenesis, where the Phe-Gly-Asp (FGD) sequence was essential for BCR binding (16). Although chicken OVA sequences were not included in this murine PhIP-seq library, over 2,000 peptides in the library contain sequence similarity (P < 0.0001; see Methods) to the OVA target sequence, and of those, 241 contain the critical FGD core sequence. Furthermore, an additional 2,028 peptides contain the core FGD, but lack similarity in the regions flanking this motif (see Methods and Supplemental Table 1). Using sera from OB1 mice, 193 peptides were enriched greater than 4-fold above the average of MBM, with a z score of greater than 3 (Figure 2A). To investigate similarity of enriched peptides to the known OB1 BCR recognition epitope, a scoring system for each peptide was developed using the sum of Smith-Waterman alignment scores to FGD across 4–amino acid sliding windows and weighted if an exact FGD match was present (see Methods). Most peptides enriched by OB1 sera exhibited high epitope similarity scores, indicating specific enrichment of peptides containing the known OB1 BCR binding site (Figure 2A). Visual inspection of the most enriched peptides revealed significant sequence similarity to the known OVA target sequence, including the core FGD (Figure 2B). We noted a Hnrnpa2b1 peptide containing 2 FGD repeats among the most highly enriched, indicating that valency within the peptide may contribute to greater enrichment.

Figure 2

Identification of autoreactive epitopes recognized by ovalbumin-specific BCR-transgenic (OB1) mice. (A) Log10(fold change) and z score over murine background model (mean of Rag2–/–, μMT, and mock IP) of peptides enriched by PhIP-seq in OB1 mice colored by alignment score to known epitope and essential FGD motif. (B) Multiple sequence alignment of top OB1-enriched peptides. (C) Logo plot of multiple sequence alignment of 193 peptides enriched by OB1 sera. (D) Sum log10(fold change) over MBM of OB1 peptides enriched by sera from Rag2–/–, μMT, OB1, B6, or Lyn–/– IgD+/– mice. Exact adjusted P value is reported, and each dot corresponds to 1 mouse. Kruskal-Wallis test with Tukey’s HSD post hoc test.

Enriched peptides containing the core sequence were aligned, revealing over representation of N-terminal Gly (Figure 2C), suggesting that this residue may contribute to recognition. Overall, 80% of the significantly enriched peptides contain the FGD core recognition sequences. Compared with Rag2–/–, μMT, wild-type, and Lyn–/– IgD+/– mice, these 193 significantly enriched peptides from the OB1 mice were highly specific to this strain (Figure 2D).

Lyn–/– IgD+/– mice exhibit lupus-like autoantibody reactivity to nuclear and apoptotic antigens. Deficiency of the Src family kinase Lyn results in widespread autoantibody production and lupus-like disease because Lyn plays a nonredundant negative regulatory role downstream of the BCR by mediating immunoreceptor tyrosine-based inhibitory motif–dependent inhibitory signaling; such autoimmunity is accelerated in Lyn–/– IgD+/– mice due to impaired quiescence of autoreactive B cells regulated by IgD misexpression (17, 19). Lyn–/– IgD+/– mice were selected for interrogation with the murine PhIP-seq library due to the profound defect in self-tolerance in this established model of lupus. Sera from Lyn–/– IgD+/– mice (n = 6) and wild-type B6 mice (n = 6) were used to IP phage from the mouse PhIP-seq library. Sequencing libraries were prepared after 3 rounds of IP and amplification. The resulting read counts for each peptide were normalized to the total number of reads in each library and then averaged across all technical replicates for each mouse. High correlations (Pearson’s r > 0.75) were observed between technically replicated samples (Supplemental Figure 1D). A stringent set of requirements was then used to identify peptide enrichments that were specific to the mutant mice relative to controls. A total of 508 peptides, derived from 425 proteins (Supplemental Table 2 and Supplemental Figure 2A), had a fold-change of at least 2-fold, a z score of least 3, and were not enriched in any of the control B6 mice IPs. Approximately 37% of mutant-specific peptide enrichments corresponded to proteins known to be components of or related to nuclear proteins (Figure 3A), consistent with the high degree of anti-nuclear staining observed in these mice (Supplemental Figure 3) and in patients with lupus (20). To validate these findings, Lyn-specific PhIP-seq reactivities were compared to results from a 96-autoantigen array probed with sera from Lyn–/– mice (Supplemental Table 3). Among protein autoantigens represented in the array (n = 63), 32 proteins were significantly enriched in Lyn–/– IgG above wild-type controls (see Methods), 20 of which were also identified by PhIP-seq in Lyn–/– IgD+/– mice (Supplemental Table 3). Orthogonally validated Lyn-specific reactivities were those to nuclear antigens such as nuclear proteins (SP100), complement (C1q), collagen (collagen VI) and laminin (Lama1), as well as additional nuclear proteins (Supplemental Figure 2, B–D, and Supplemental Table 3). In addition to small ribonucleoproteins (sRNPs), PhIP-seq identified proteins related to their generation that were not included in the targeted autoantigen array (Figure 3B). This overrepresentation of sRNP autoimmune targets in Lyn–