Access the full text.
Sign up today, get DeepDyve free for 14 days.
Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 doi:10.1093/brain/awy109 BRAIN 2018: 141; 2263–2271 | 2263 REPORT Distinct HLA associations of LGI1 and CASPR2-antibody diseases 1, 1, 2, 1 2 Sophie Binks, * James Varley, * Wanseon Lee, * Mateusz Makuch, Katherine Elliott, 3 4 1 5 6 Jeffrey M. Gelfand, Saiju Jacob, M. Isabel Leite, Paul Maddison, Mian Chen, 3 1 1 1 7 Michael D. Geschwind, Eleanor Grant, Arjune Sen, Patrick Waters, Mark McCormack, 7 6 2, 1, Gianpiero L. Cavalleri, Martin Barnardo, Julian C. Knight * and Sarosh R. Irani * *These authors contributed equally to this work. The recent biochemical distinction between antibodies against leucine-rich, glioma-inactivated-1 (LGI1), contactin-associated pro- tein-2 (CASPR2) and intracellular epitopes of voltage-gated potassium-channels (VGKCs) demands aetiological explanations. Given established associations between human leucocyte antigen (HLA) alleles and adverse drug reactions, and our clinical observation of frequent adverse drugs reactions in patients with LGI1 antibodies, we compared HLA alleles between healthy controls (n = 5553) and 111 Caucasian patients with VGKC-complex autoantibodies. In patients with LGI1 antibodies (n = 68), HLA-DRB1*07:01 was strongly represented [odds ratio = 27.6 (95% conﬁdence interval 12.9–72.2), P = 4.1 10 ]. In contrast, patients with CASPR2 antibodies (n = 31) showed over-representation of HLA-DRB1*11:01 [odds ratio = 9.4 (95% conﬁdence interval 4.6– 19.3), P = 5.7 10 ]. Other allelic associations for patients with LGI1 antibodies reﬂected linkage, and signiﬁcant haplotypic associations included HLA-DRB1*07:01-DQA1*02:01-DQB1*02:02, by comparison to DRB1*11:01-DQA1*05:01-DQB1*03:01 in CASPR2-antibody patients. Conditional analysis in LGI1-antibody patients resolved further independent class I and II associ- ations. By comparison, patients with both LGI1 and CASPR2 antibodies (n = 3) carried yet another complement of HLA variants, and patients with intracellular VGKC antibodies (n = 9) lacked signiﬁcant HLA associations. Within LGI1- or CASPR2-antibody patients, HLA associations did not correlate with clinical features. In silico predictions identiﬁed unique CASPR2- and LGI1- derived peptides potentially presented by the respective over-represented HLA molecules. These highly signiﬁcant HLA associations dichotomize the underlying immunology in patients with LGI1 or CASPR2 antibodies, and inform T cell speciﬁcities and cellular interactions at disease initiation. 1 Oxford Autoimmune Neurology Group, Nufﬁeld Department of Clinical Neurosciences, University of Oxford, Level 3, West Wing, John Radcliffe Hospital, Oxford, OX3 9DS, UK 2 Wellcome Centre for Human Genetics, University of Oxford, Oxford, OX3 7BN, UK 3 UCSF Department of Neurology, 675 Nelson Rising Lane, San Francisco, CA 94158, USA 4 Centre for Rare Diseases and Queen Elizabeth Neuroscience Centre, University Hospitals Birmingham, UK 5 Department of Neurology, Queen’s Medical Centre, Derby Road, Nottingham NG7 2UH, UK 6 Transplant Immunology and Immunogenetics Laboratory, Oxford Transplant Centre, Churchill Hospital, Oxford, UK 7 Department of Molecular and Cellular Therapeutics, the Royal College of Surgeons in Ireland, Dublin, Ireland Correspondence to: Professor Sarosh R. Irani Oxford Autoimmune Neurology Group West Wing, Level 3 John Radcliffe Hospital Received September 8, 2017. Revised January 29, 2018. Accepted February 9, 2018. Advance Access publication May 18, 2018 The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 2264 | BRAIN 2018: 141; 2263–2271 S. Binks et al. Oxford OX3 9DU UK E-mail: email@example.com Keywords: human leucocyte antigen; leucine-rich glioma-inactivated 1; contactin-associated protein 2; voltage-gated potassium channel; major histocompatibility complex Abbreviations: HLA = human leucocyte antigen; VGKC = voltage-gated potassium channel immunosuppressants (McCormack et al., 2011; Yip et al., Introduction 2015), and have essential antigen-presenting functions (Trowsdale and Knight, 2013), we hypothesized that The discovery of autoantibodies against leucine-rich, HLA associations existed in patients with LGI1 antibodies. glioma-inactivated 1 (LGI1), contactin-associated protein Indeed, recently, HLA-DRB1*07:01, HLA-DQB1*02:02 2 (CASPR2) (Irani et al., 2010; Lai et al., 2010) and, and HLA-DRB4 were found to be present in varying pro- more recently, intracellular epitopes of voltage-gated potas- portions of patients with LGI1 antibodies in two cohorts sium channels (VGKCs) (Lang et al., 2017), have redeﬁned totalling 40 patients, from Korea and the Netherlands (Kim the immunology of the VGKC-complex (Thieben et al., et al., 2017; van Sonderen et al., 2017). 2004; Vincent et al., 2004). Patient stratiﬁcation by these To extend these early observations, and given the hy- antigenic targets has shown that the ‘double-negative’ pothesis that the VGKC complex may be the initiating VGKC-complex antibodies, those without LGI1 or immunizing agent, we sought to compare and contrast CASPR2 reactivities, are observed across all ages, in HLA-associations in a sizeable cohort of clinically well- healthy controls and in a variety of syndromes, many of characterized patients with antibodies against LGI1, which are not immune-mediated (Graus and Gorman, CASPR2, both LGI1 and CASPR2, and VGKCs, and in 2016; van Sonderen et al., 2016; Lang et al., 2017). In silico to identify peptides that may be presented by these contrast, patients with LGI1 or CASPR2 antibodies often HLA molecules. have clinically-indistinguishable late-onset forms of limbic encephalitis and neuromyotonia with associated dysautono- mia, sleep disturbances, pain and seizures (Irani et al., Materials and methods 2010; Lai et al., 2010; Klein et al., 2013; Gadoth et al., 2017). While these features occur at different rates in LGI1- versus CASPR2-antibody cohorts, only faciobrachial dys- Patients tonic seizures (FBDS) robustly predict LGI1 reactivity One hundred and eleven Caucasian patients were identiﬁed (Irani et al., 2011; Gadoth et al., 2017; Thompson et al., from previous studies (n = 51) (Irani et al., 2011, 2013; Lang 2018). Furthermore, these two autoantibodies are both et al., 2017), referrals to the Oxford Autoimmune Neurology often of the IgG4 subclass and frequently co-exist in pa- Group (n = 46) or from the Autoimmune Encephalopathy tients with the ultra-rare Morvan’s syndrome (Irani et al., Clinic, University of California San Francisco (n = 14). These 2012; Arino et al., 2016). The striking overlaps of these ˜ patients had serum antibodies against LGI1 only (n = 68), CASPR2 only (n = 31), both LGI1 and CASPR2 (n =3) or rare neurological features and autoantibodies, and the fre- intracellular aspects of VGKCs (n = 9), as determined by pre- quent co-expression of their antigenic targets within mam- viously described antigen-speciﬁc cell-based assays (Irani et al., malian CNS-membrane complexes (Irani et al., 2010; Binks 2010; Lang et al., 2017). Clinical phenotypes, including infor- et al., 2018), suggest they are involved in autoimmuniza- mation relating to past medical history and adverse drug reac- tion. Indeed, this has been reported in abattoir workers tions (Table 1), were evaluated via direct patient and relative with autoantibodies against VGKC-complexes and, less interviews and case-note reviews. All patients provided written so, CASPR2 (Meeusen et al., 2012). The nature of the informed consent (REC16/YH/0013 or the IRB 10-04905 available complexes, antigen presentation mechanisms and approvals). the available T cell repertoires are likely to determine which antigen dominates the ensuing T–B cell response. If so, Genotyping, HLA imputation, human leucocyte antigen (HLA) variants, intimately related veriﬁcation and multi-locus to antigen presentation, may play critical roles in distin- guishing the aetiology of these syndromes. haplotype-block construction Previously, high rates of adverse drug reactions were The Inﬁnium Global Screening Array-24 + v1.0 BeadChip with observed in patients with LGI1 antibodies, typically second- Illumina Inﬁnium HTS custom markers were used for genotyp- ary to antiepileptic drugs (AEDs) and, less so, corticoster- ing. We proceeded to impute HLA alleles using SNP2HLA oids (Irani et al., 2011, 2013; Thompson et al., 2018). As at eight classical loci (HLA-A, HLA-B, HLA-C, HLA-DPA, HLA variants have been implicated in several adverse drug HLA-DPB, HLA-DQA1, HLA-DQB1 and HLA-DRB1) at reactions, including those associated with AEDs and two-ﬁeld resolution (Jia et al., 2013; Neville et al., 2017). Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 HLA comparisons across the VGKC-complex BRAIN 2018: 141; 2263–2271 | 2265 Table 1 Clinical features of patients with antibodies to VGKC complex proteins: LGI1, CASPR2, both LGI1 and CASPR2 or intracellular aspects of VGKCs LGI1 CASPR2 LGI1 and VGKC LGI1 versus CASPR2 CASPR2 (P-value)* Number of patients 68 31 3 9 ND Median age at onset age (range) 63 (41–85) 68 (19–82) 56 (52–65) 43 (33–71) ND Female (%) 20 (29) 2 (6) 1 (33) 4 (44) ND Clinical syndrome (%) Epilepsy 8 (12) 2 (6) 0 (0) 4 (44) ND Encephalitis 58 (85) 18 (58) 1 (33) 2 (22) ND Morvan’s 1 (2) 3 (10) 2 (67) 0 (0) ND Isolated neuromyotonia 0 (0) 2 (6) 0 (0) 0 (0) ND Other 1 (2) 6 (19) 0 (0) 3 (33) ND Clinical features (%) Any seizure 66 (97) 25 (81) 2 (67) 7 (78) 0.01 Faciobrachial dystonic seizures 47 (69) 0 (0) 1 (33) 0 (0) 50.0001 Generalized seizure 26 (38) 10 (32) 2 (67) 6 (67) NS Amnesia 58 (85) 23 (74) 3 (100) 5 (56) NS Neuromyotonia 1 (2) 8 (26) 2 (67) 0 (0) 0.0003 Neuropathic pain 5 (7) 14 (45) 2 (67) 0 (0) 50.0001 Autoimmune features (%) Other autoimmune disease 19 (28) 7 (23) 0 (0) 0 (0) NS Atopy 8 (12) 2 (6) 2 (67) 1 (11) NS Adverse effects of corticosteroids 32 (47) 5 (16) 0 (0) 1 (11) 0.004 Drug rash 24 (35) 1 (3) 0 (0) 0 (0) 0.0004 Other features Mean change in mRS (range) 1.6 (3 to 4) 1.5 (0 to 4) 1.7 (0 to 3) 1 (0 to 2) ND Tumour (%) 9 (13) 4 (13) 2 (67) 1 (11) ND Live cell-based assays were used for LGI1 and CASPR2 antibody determination (Irani et al., 2010), and ﬁxed assays to detect antibodies against the intracellular aspects of VGKCs (Lang et al., 2017). Other diagnoses included movement disorders (n = 4, CASPR2, generalized chorea, hemifacial spasm, cervical dystonia and cerebellar ataxia), psychogenic amnesia (n = 2, VGKC antibodies), widespread non-neuropathic pain (n = 1, VGKC antibodies), axonal neuropathy (n = 1, CASPR2), psychosis (n = 1, CASPR2) and stroke (n = 1 with LGI1 antibodies). Two patients with antibodies against intracellular VGKC epitopes had epilepsy secondary to structural lesions. Autoimmune diseases in LGI1-antibody patients: [n = 19: diabetes (n = 1), heparin-induced thrombocytopaenia (n = 1), hyper- and hypothyroidism and Hashimoto’s thyroiditis (n = 8), multiple sclerosis (n = 1), myasthenia gravis (n = 1), neuromyelitis optica (n = 1), optic neuritis (n = 1), pernicious anaemia (n = 1), psoriasis (n = 6), Raynaud’s disease (n =1), and ulcerative colitis (n = 1)] and CASPR2-antibody patients [n = 7: congenital adrenal hyperplasia, hypothyroidism, pernicious anaemia, pemphigus, polymyalgia rheumatica, psoriasis and Raynaud’s disease (all n = 1)]. Corticosteroid-related complications, sometimes multiple, in LGI1-antibody patients [n = 32: marked weight gain (n = 12), behavioural disturbance (n = 5) and diabetes (n =5), or worsened diabetes (n = 1), insomnia (n = 4), fracture (n = 3), myopathy or muscle weakness (n = 3), skin thinning/easy bruising (n = 3), mania/hypomania (n = 2), poor wound healing or abscess (n = 2), ophthalmic infections (n = 2; keratitis and ophthalmic shingles), perforated abdominal viscus (n = 2), and one each of: avascular necrosis of the hip (AVN), cerebral venous sinus thrombosis, high INR and steroid-induced psychosis] and in CASPR2-antibody patients [n = 5: marked weight gain (n = 1), rash (n = 2), striae/thin skin/bruising (n =2), and hallucinations (n = 1)]. Tumours in LGI1-antibody patients (n = 9) were: basal cell carcinoma (n = 3), other skin – type not known (n = 2), bladder (n = 1), breast (n = 1), prostate (n = 1), dysplastic colonic polyp (n = 1) and in 4 CASPR2-antibody patients were: pancreatic (n = 1), prostate (n = 2), thymic cyst (n = 1). NS = not signiﬁcant; ND = not done; mRS = modiﬁed Rankin scale (as Thompson et al., 2018). *Statistical comparisons with Fisher’s exact test throughout. To complement this, DRB1, DRB4 and DQ alleles underwent DPA1-DPB1 (Stephens et al., 2001; Stephens and Donnelly, intermediate-resolution HLA-typing using PCR-sequence- 2003). Further details on genotyping and imputation are pro- speciﬁc primers (SSP), updated from Bunce et al. (1995). vided in the Supplementary material. PCR-SSP deﬁned the ﬁrst-ﬁeld plus a string of second-ﬁeld possibilities: the highest frequency allele in Caucasians was considered most likely. For all discordant data, the PCR-SSP Statistical analyses ﬁrst-ﬁeld was accepted as the ﬁnal result. HLA alleles from 5553 Caucasian healthy controls (from Oxford Biobank) For each antibody group, Fisher’s exact test (two-tailed) was were available from imputation using the same platforms, used to compare the HLA allele and haplotype carrier frequen- and conﬁrmed in 70 individuals within the same laboratory cies between patients and the healthy control dataset. Hochberg’s method was used to correct for multiple compari- by PCR-SSP (Neville et al., 2017). Probable haplotype blocks were calculated on the basis of a Bayesian algorithm using sons. Corrected P-values 5 0.05 were considered signiﬁcant, PHASE V2 software with 10 000 iterations for three haplotype and are presented. Odds ratios (ORs) were calculated using the blocks: HLA-C-B, HLA-DRB1-DQA1-DQB1, and HLA- median-unbiased estimation method. Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 2266 | BRAIN 2018: 141; 2263–2271 S. Binks et al. HLA binding predictions Patients with LGI1 or CASPR2 antibodies have strong and distinct The NetMHCIIpan 3.1 server model based on artiﬁcial neural networks (Andreatta et al., 2015) evaluated HLA HLA allelic proﬁles haplotype binding afﬁnities for 15-amino acid-long consecu- We proceeded to analyse HLA associations, as summarized tive overlapping peptides from full-length LGI1 and CASPR2 sequences (UniProt accession numbers O95970 in Fig. 1 and Table 2. Consistent with previous smaller re- and Q9UHC6, respectively). Predicted peptide afﬁnities ports (Kim et al., 2017; van Sonderen et al., 2017), almost all (nM) were compared to 200 000 random peptides of the LGI1-antibody-positive patients carried HLA-DRB1*07:01 same length to generate rank values: this measure is less (91%, compared to 26% in healthy controls) [OR 27.6 susceptible to the intrinsic capacity of some HLA alleles to (95% conﬁdence interval, CI 12.9–72.2), P =4.1 10 ]. generate high-afﬁnity predictions, and rank values (%) 53 Further, 13% (9/68) were homozygous for DRB1*07:01, were considered strong binders. As expected, consecutive 15- compared to 2% (115/5553) healthy controls [OR 7.3 merpeptideswith highrankvaluesoftensharedacore (95% CI 3.3–14.4), P =3 10 ]. Alleles recognized to be sequence. part of haplotypes involving HLA-DRB1*07:01 (Gonza´lez- Galarza et al., 2015) were over-represented, namely HLA- DQA1*02:01, HLA-DQB1*02:02, HLA-DQB1*03:03 and HLA-DPB1*11:01. Additionally, associations were found Results with two HLA class I alleles, HLA-B*57:01 [OR = 3.7 (95% CI 2.0–6.5); P = 0.014] and HLA-C*06:02 [OR = 3.9 Clinical differences between patients (95% CI 2.4–6.3); P =4.6 10 ]. After conditioning on the commonest allele, HLA-DRB1*07:01, two other DQ alleles stratiﬁed by VGKC-complex autoan- reached statistical signiﬁcance consistent with evidence of an tibody targets independent association, HLA-DQA1*01:03 [OR = 4.4 Table 1 summarizes the clinical features of the 111 (95% CI 2.2–8.1); P =4 10 ] and HLA-DRB1*01:03 patients, subgrouped by their autoantibody speciﬁcities. In [OR = 14.7 (95% CI 3.6–51.5), P =0.04]. agreement with previous studies, onset ages were typically In striking contrast, analysis of the CASPR2-antibody around 60 years, and patients with LGI1 or CASPR2 group identiﬁed a single risk allele; HLA-DRB1*11:01, antibodies most frequently had encephalitis or epilepsy. which was present in 48% of CASPR2-antibody patients FBDS were exclusive to patients with LGI1 antibodies compared to 4% of patients with LGI1 antibodies and 9% (P5 0.0001) who had more seizures (P = 0.01) than of healthy controls [OR 9.4 (95% CI 4.6–19.3); patients with CASPR2 antibodies, where peripheral nerve P = 5.7 10 ]. One CASPR2-antibody patient was homo- zygous for HLA-DRB1*11:01. Interestingly, the four pa- features of neuromyotonia (P = 0.0003) and neuropathic tients with non-immune conditions and CASPR2 pain (P5 0.0001) were preferentially associated. As antibodies (Table 1) did not carry HLA-DRB1*11:01, expected, the nine patients with antibodies to intracellular giving it a 56% (15/27) frequency in the remainder. No VGKC epitopes had heterogeneous, often non-immune, clin- additional alleles were observed after conditioning on ical syndromes. By contrast, likely non-immune syndromes HLA-DRB1*11:01. were noted in only one patient with LGI1 antibodies Intriguingly, of the three patients with co-existent (stroke) and in four with CASPR2 antibodies (axonal neur- CASPR2 and LGI1 antibodies, only one carried HLA- opathy, cervical dystonia, hemifacial spasm and psychosis). DRB1*07:01 and none carried HLA-DRB1*11:01. Of greater relevance to a HLA study, patients with LGI1 However, all three carried HLA-B*44:02, HLA-C*05:01, or CASPR2 antibodies often had co-existent autoimmune HLA-DQA1*03:01 and HLA-DQB1*03:01, a different conditions (28% and 23%, respectively), including complement of alleles to the patients with antibodies to Hashimoto’s thyroiditis (n = 8), psoriasis (n = 7) and perni- either LGI1 or CASPR2 (Supplementary Fig. 1). There cious anaemia (n = 2). Moreover, the LGI1-antibody cohort were no signiﬁcant ﬁndings within the group with intracel- was distinctive for a 47% rate of adverse drug reactions lular VGKC antibodies (Supplementary Fig. 1). from corticosteroids (P = 0.004; 16% with CASPR2 anti- bodies) and a signiﬁcantly higher rate of drug-induced rashes (35% versus 3% in CASPR2, P = 0.0004). The re- Haplotype-speciﬁc distinctions ported rashes were secondary to AEDs [n = 13: including between patients with LGI1 and carbamazepine (n = 6), phenytoin (n = 4), lamotrigine CASPR2 antibodies (n = 2) and valproate (n = 1)], antibiotics [n = 6: penicillins (n = 5) and metronidazole (n = 1)] and immunosuppressants Next, to understand the en bloc allelic inheritance and in vivo [n = 5: azathioprine (n = 2), corticosteroids (n = 2) and relevance of HLA combinations that may present LGI1 and methotrexate (n = 1)]. Thus, the LGI1- and CASPR2-anti- CASPR2 antigens, we explored associations involving HLA body groups displayed differing clinical autoimmune fea- haplotypes (Fig. 1B and full analysis in Supplementary Figs tures suggesting divergent immunogenetic pathways. 2–4). We noted that HLA-DQA1*02:01, HLA-DQB1*02:02, Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 HLA comparisons across the VGKC-complex BRAIN 2018: 141; 2263–2271 | 2267 Figure 1 HLA allele and haplotype associations in patients with LGI1 and CASPR2 antibodies. Bar chart depicting allele (A) and haplotype (B) associations and their frequency in patients with antibodies to LGI1 (n = 68, red denotes signiﬁcant associations) and CASPR2 (n = 31, blue denotes signiﬁcant associations), together with the frequency of these alleles or haplotypes in 5553 healthy controls (black bars). HC = healthy controls. HLA-DQB1*03:03 and HLA-DPB1*11:01 show evidence of with psoriasis harboured the psoriasis risk allele C*06:02 linkage disequilibrium with HLA-DRB1*07:01 (r values (Arakawa et al., 2015), suggesting the extended haplotypes 0.64, 0.49, 0.13, 0.10 and D 1, 0.95, 0.8 and 1, respectively) may explain these speciﬁc co-morbidities. Finally, from the (Supplementary material). This was reﬂected in the most fre- nine LGI1- and four CASPR2-antibody patients with a quent HLA class II haplotypes found in patients with LGI1 tumour, there were no signiﬁcant HLA differences com- antibodies, namely HLA-DRB1*07:01-DQA1*02:01- pared to non-tumour patients (Supplementary Table 2). DQB1*02:02 [OR = 5.2 (95% CI 3.2–8.6); P =2.3 10 ], DRB1*07:01-DQA1*02:01-DQB1*03:03 [OR = 3.1 (95% DRB4 analysis CI 1.7–5.5); P = 0.02] and DPA1*02:01-DPB1*11:01 To extend a previous report (van Sonderen et al., 2017), [OR = 4.8 (95% CI 2.5–8.5); P =3.8 10 ]. In addition, the HLA-DRB1 paralogue HLA-DRB4 was sequenced and LGI1-antibody status was associated with a HLA class I observed to be absent or a null allele in 61% (19/31) of the haplotype, HLA-C*06:02-B*57:01 [OR = 3.6 (95% CI 1.9– 6.2); P =8.8 10 ]. By contrast, only one HLA class II CASPR2-antibody cohort and 44% (4/9) of the intracellu- haplotype was associated with CASPR2 antibodies: lar VGKC-antibody cohort, consistent with the healthy DRB1*11:01-DQA1*05:01-DQB1*03:01 [OR = 7.4 (95% control frequencies. However, only 16% (11/68) of the CI 3.5–15.2), P =5.7 10 ]. LGI1-antibody cohort showed an absent or null HLA- Given these signiﬁcant and distinct allelic and haplotypic DRB4 allele, and these 11 patients all carried HLA- HLA associations, for each serologically-deﬁned group, we DRB1*07:01 (Supplementary Table 3). evaluated their value in explaining sub-phenotypes (limbic encephalitis or epilepsy; peripheral or CNS), long-term out- Predictions of HLA-binding peptides comes or adverse drug reactions (Supplementary Table 1), and found no signiﬁcant HLA allele or haplotype associ- These robust HLA class II associations strongly implicate ations. However, within LGI1-antibody patients, ﬁve of six CD4 T cells in the pathogenesis of both LGI1- and patients with antibiotic-induced rashes carried HLA- CASPR2-antibody-associated diseases. To locate potentially B*57:01 known to associate with risk of rash to abacavir high-afﬁnity peptides that complex with HLA class II het- and ﬂucloxacillin (Yip et al., 2015), and four of six patients erodimers, and may interact with patient T cells, in silico Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 2268 | BRAIN 2018: 141; 2263–2271 S. Binks et al. Figure 2 Peptides derived from full-length LGI1 and CASPR2 predicted to bind MHC-dimers encoded by over-represented HLA haplotypes. Rankings and position of peptides derived from full-length sequences of LGI1 (A and B) and CASPR2 (C and D). The haplotypes correspond to Fig. 1B and when in bold they relate to those observed in patients with antibodies to the corresponding protein. Red circles denote the LGI1-antibody cohort and blue the CASPR2-antibody cohort. Grey circles and italicized haplotypes relate to peptides from the other antigenic protein (i.e. CASPR2 in A and B; and LGI1 in C and D). Rank describes the predicted peptide afﬁnities (IC , nM) by comparison to 200 000 random peptides of the same length. Dotted lines represent the 3% cut-off for peptide rank. Within B and D, circles represent the highly-ranked peptides across the full-length sequences of LGI1 or CASPR2: black circles represent peptides with some predicted promiscuity across LGI1- and CASPR2-antibody HLA variants, whereas pink circles highlight peptides that are not predicted to cross-react. modelling was used and focused on all the class II haplo- As expected for the shared core sequences between con- types identiﬁed above (Fig. 2). secutive 15-mers, many highly ranked peptides were from Overall, many peptides from both LGI1 and CASPR2 tightly clustered locations within the full-length protein ranked highly for potential binding to several HLA-DR, (Fig. 2B, D and Supplementary Table 4). Most peptides HLA-DP and HLA-DQ variants (Fig. 2A and C), likely within these clusters showed potential to bind the HLA consistent with the varied intrinsic properties of different variants observed in both the LGI1- and CASPR2-antibody HLA molecules. Furthermore, for HLA-DRB1*07:01 and cohorts (Fig. 2B and D, black circles). This included one HLA-DRB1*11:01, which pair with the invariant DRA previously identiﬁed peptide (Kim et al., 2017) and argues chain, and for HLA-DQA1*02:01-DQB1*02:02 heterodi- against its role in disease speciﬁcity. However, 9/13 LGI1- mers, peptide ranks showed little difference between derived peptides and 7/13 from the CASPR2 sequence LGI1- and CASPR2-derived peptides, suggesting a lack of showed binding potential that was more restricted to the antigen selectivity. Also, no highly-ranked peptides were variants associated with the corresponding antibody cohort identiﬁed to bind the LGI1-antibody-associated heterodimer (Fig. 2B and D, pink circles). From LGI1, 4/9 core peptides HLA-DQA1*02:01-DQB1*03:03, making it an unlikely were predicted to bind with high afﬁnity (540 nM), typic- candidate molecule for LGI1 peptide presentation. By con- ally to HLA-DRB1*07:01, although interestingly the high- trast, the CASPR2-antibody-associated HLA-DQA1*05:01- est afﬁnity peptide was predicted to bind HLA- DQB1*03:01 heterodimer was predicted to bind some DPA1*02:01-DPB1*11:01 (Supplementary Table 4). From high-ranking peptides from the CASPR2 sequence only, CASPR2-derived peptides, 7/7 were predicted to bind with suggesting CASPR2 speciﬁcity. high afﬁnity, distributed across the variants within the Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 HLA comparisons across the VGKC-complex BRAIN 2018: 141; 2263–2271 | 2269 Table 2 Signiﬁcant allele (top) and haplotype (bottom) associations in patients with LGI1 (n = 68) or CASPR2 (n = 31) antibodies Antibody Allele / haplotype OR 95% 95% Fisher’s Corrected group CI lower CI upper exact test (Hochberg) P-value 5 2 LGI1 B*57:01 3.7 2.0 6.5 5.2 10 1.0 10 7 5 LGI1 C*06:02 3.9 2.4 6.3 1.6 10 4.6 10 6 3 LGI1 DPB1*11:01 4.9 2.6 8.7 8.4 10 2.0 10 17 15 LGI1 DQA1*02:01 8.9 5.2 16.1 3.0 10 8.7 10 17 14 LGI1 DQB1*02:02 8.1 4.9 13.7 9.5 10 2.8 10 8 5 LGI1 DQB1*03:03 4.7 2.8 7.7 3.6 10 1.0 10 28 26 LGI1 DRB1*07:01 27.6 12.9 72.2 1.4 10 4.1 10 8 6 CASPR2 DRB1*11:01 9.4 4.6 19.3 2.0 10 5.7 10 11 9 LGI1 DRB1*07:01-DQA1*02:01-DQB1*02:02 5.2 3.2 8.6 4.7 10 2.3 10 4 2 LGI1 DRB1*07:01-DQA1*02:01-DQB1*03:03 3.1 1.7 5.5 4.4 10 2.1 10 5 4 LGI1 DPA1*02:01-DPB1*11:01 4.8 2.5 8.5 1.0 10 3.8 10 4 3 LGI1 C*06:02-B*57:01 3.6 1.9 6.2 1.3 10 8.8 10 6 5 CASPR2 DRB1*11:01-DQA1*05:01-DQB1*03:01 7.4 3.5 15.2 1.1 10 5.7 10 Corrected P-values indicate comparison between disease and healthy controls. CASPR2-antibody-associated haplotype (Supplementary sub-phenotypes, outcomes or, in contrast to a previous ob- servation, the presence of associated tumours (van Table 4). Sonderen et al., 2017). Also, the 9–27% frequencies of these HLA variants in healthy Caucasians are far higher than disease prevalence, implicating additional loci, envir- Discussion onmental or stochastic inﬂuences in disease manifestation. This study is the ﬁrst comparative HLA analysis of LGI1- Furthermore, our data also provide several intriguing in- and CASPR2-autoantibody mediated diseases, and shows sights into the immunopathogenesis of these diseases. First, marked and strikingly different HLA associations for they extend the frequent HLA associations in IgG4-related these patients, at both allelic and haplotypic levels. Given diseases (Huijbers et al., 2015), but here, exceptionally, the frequently overlapping clinical features in patients with with no DQ5 association. Second, the presence of domin- LGI1 and CASPR2 antibodies, and their co-expression in ant HLA class II associations implicates extracellular anti- VGKC complexes, these ﬁndings indicate that dichotomous gen processing and CD4 T cells in disease initiation predisposing HLA variants govern the generation of LGI1 (Trowsdale and Knight, 2013), but the LGI1 antibody versus CASPR2 antibodies. Furthermore, they strongly im- class I associations found here, and HLA-B*44:03 and plicate T cells in disease initiation and the candidate HLA- HLA-C*07:06 reported in seven patients previously (Kim binding peptide partners generated by our in silico data et al., 2017), are compatible with a role for intracellular may help identify these interacting T cells. antigen processing, including viruses and drugs. These class While HLA-DRB1*07:01 and linked class II alleles, I differences between studies may be explained by ethnicity, including the haplotype HLA-DRB1*07:01-DQA1*02:01- sample size and relatively weak associations (Kim et al., DQB1*02:02, showed very strong associations with 2017; van Sonderen et al., 2017). Indeed, this extended LGI1-antibody patients, this was not observed among haplotype and the related complex linkage disequilibrium CASPR2-antibody patients in whom we found clear asso- in this region of the genome warrant further analysis. ciations with HLA-DRB1*11:01 only. Among LGI1-anti- Furthermore, our original hypothesis of adverse drug reac- body patients, DRB1*11:01 was observed at around tion-related HLA variants may relate to the linked adverse healthy control rates, DRB4 was less frequently detected drug reaction-related class I and II HLA variants (HLA- than DRB1*07:01, homozygosity for HLA-DRB1*07:01 DRB1*07:01, HLA-DQA1*02:01, HLA-B*57:01) (Yip was recognized, and other independent associations et al., 2015). These and future observations in patients involved HLA class I alleles HLA-B*57:01 and HLA- with LGI1 antibodies may inform the genetic basis of C*06:02. Albeit limited by their intrinsic rarity, intri- more common adverse drug reactions. Third, the HLA guingly, the three patients with both LGI1 and CASPR2 similarities between tumour and non-tumour LGI1-anti- antibodies had yet another complement of HLA alleles. body cases suggests the absence of a unique paraneoplastic Perhaps this implicates further divergence in molecular signature, in contrast to Lambert-Eaton myasthenic syn- mechanisms responsible for the generation of both auto- drome (Wirtz et al., 2005). Perhaps this implies tumours antibody speciﬁcities within an individual. However, the in patients with LGI1 antibodies largely reﬂect the age- HLA associations do not appear to distinguish between matched background rate, rather than a distinct immune Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 2270 | BRAIN 2018: 141; 2263–2271 S. Binks et al. mechanism, although a paucity of tumours classically asso- Award (204969/Z/16/Z). J.V. is supported by the ciated with paraneoplastic neurological syndromes may Association of British Neurologists and Guarantors of limit our interpretation. Finally, our in silico predictions Brain. M.D.G. has received research funding from the suggest that HLA-DQA1*02:01-DQB1*03:03 is unlikely NIH/NIA (grant R01 AG AG031189), Quest Diagnostics, to mediate presentation of LGI1-derived peptides, whereas Inc., the Tau Consortium and the Michael J. Homer Family the HLA-DQA1*05:01-DQB1*03:01 heterodimer may be Fund. M.D.G. has served as a consultant for Quest implicated in the CASPR2-antibody phenotype. Also, the Diagnostics, Inc. promiscuity of both CASPR2 and LGI1 peptides for some HLA variants, including HLA-DRB1*07:01 (Kim et al., 2017), may explain why immunization with the same Conﬂicts of interest VGKC complexes may generate two distinct disease enti- ties, and underlie the observed co-existence of both antibo- S.R.I. and P.W. are co-applicants and receive royalties on dies at rates far higher than expected by chance (Irani et al., patent application WO/2010/046716 entitled ‘Neurological 2012). However, this in silico approach is inherently lim- Autoimmune Disorders’. The patent has been licensed to ited by the possibility that high afﬁnity peptides are more Euroimmun AG for the development of assays for LGI1 effectively deleted through central tolerance. Nevertheless, and other VGKC-complex antibodies. Other authors taken together, the range of antigen-restricted peptides report no conﬂicts of interest. derived herein, and the relative HLA variant frequencies in disease versus control populations, generate hypothesis- driven approaches to expand disease-speciﬁc T cells in vitro Supplementary material and complement recent clinical and laboratory observations which strongly implicate T cell dependence of antibody- Supplementary material is available at Brain online. mediated diseases (Makuch et al., 2018; Wilson et al., 2018a, b). In summary, the distinct HLA associations in patients References with LGI1 and CASPR2 autoantibodies, together with dif- fering clinical features relating to autoimmunity, support an Andreatta M, Karosiene E, Rasmussen M, Stryhn A, Buus S, Nielsen immunological dissociation in generation of these clinically- M. Accurate pan-speciﬁc prediction of peptide-MHC class II binding overlapping autoantibody-mediated syndromes. The dom- afﬁnity with improved binding core identiﬁcation. Immunogenetics 2015; 67: 641–50. inant class II HLA involvement combined with in silico Arakawa A, Siewert K, Sto¨ hr J, Besgen P, Kim SM, Ru¨ hl G, et al. predictions, offers potential to better understand the likely Melanocyte antigen triggers autoimmunity in human psoriasis. J Exp initiating T–B cell interactions. Further work should focus Med 2015; 212: 2203–12. on the environmental factors that inﬂuence the presentation Arino H, Armangue T, Petit-Pedrol M, Sabater L, Martinez- of peptides in genetically predisposed individuals. Hernandez E, Hara M, et al. Anti-LGI1-associated cognitive impair- ment: presentation and long-term outcome. Neurology 2016; 87: 759–65. Binks SNM, Klein CJ, Waters P, Pittock SJ, Irani SR. LGI1, CASPR2 Acknowledgements and related antibodies: a molecular evolution of the phenotypes. J Neurol Neurosurg Psychiatry 2018; 89: 526–34. We thank the High-Throughput Genomics Group at the Bunce M, O’Neill CM, Barnardo MC, Krausa P, Browning MJ, Wellcome Centre for Human Genetics (funded by Morris PJ, et al. Phototyping: comprehensive DNA typing for Wellcome grant reference 090532/Z/09/Z) for the gener- HLA-A, B, C, DRB1, DRB3, DRB4, DRB5 & DQB1 by PCR with 144 primer mixes utilizing sequence-speciﬁc primers (PCR- ation of Genotyping data. SSP). Tissue Antigens 1995; 46: 355–67. Gadoth A, Pittock SJ, Dubey D, McKeon A, Britton JW, Schmeling JE, et al. Expanded phenotypes and outcomes among 256 LGI1/ CASPR2-IgG positive patients. Ann Neurol 2017; 82: 79–92. Funding Gonza´lez-Galarza FF, Takeshita LY, Santos EJ, Kempson F, Maia MH, da Silva AL, et al. Allele frequency net 2015 update: new S.R.I. is supported by the Wellcome (104079/Z/14/Z), features for HLA epitopes, KIR and disease and HLA adverse BMA Research Grants- Vera Down grant (2013) and drug reaction associations. Nucleic Acids Res 2015; 43: D784–8. Margaret Temple grant (2017), Epilepsy Research UK Graus F, Gorman MP. Voltage-gated potassium channel antibodies: (P1201) and by the Fulbright UK-US commission (MS- game over. Neurology 2016; 86: 1657–8. Society research award). S.B., J.V., S.R.I. and J.C.K. and Huijbers MG, Querol LA, Niks EH, Plomp JJ, van der Maarel SM, the research were supported by the National Institute for Graus F, et al. The expanding ﬁeld of IgG4-mediated neurological autoimmune disorders. Eur J Neurol 2015; 22: 1151–61. Health Research (NIHR) Oxford Biomedical Research Irani SR, Alexander S, Waters P, Kleopa KA, Pettingill P, Zuliani L, Centre (the views expressed are those of the author(s) et al. Antibodies to Kv1 potassium channel-complex proteins leu- and not necessarily those of the NHS, the NIHR or the cine-rich, glioma inactivated 1 protein and contactin-associated pro- Department of Health). J.C.K. is supported by Arthritis tein-2 in limbic encephalitis, Morvan’s syndrome and acquired Research UK (20773) and a Wellcome Investigator neuromyotonia. Brain 2010; 133: 2734–48. Downloaded from https://academic.oup.com/brain/article/141/8/2263/4999724 by DeepDyve user on 19 July 2022 HLA comparisons across the VGKC-complex BRAIN 2018: 141; 2263–2271 | 2271 Irani SR, Michell AW, Lang B, Pettingill P, Waters P, Johnson MR, Stephens M, Donnelly P. A comparison of bayesian methods for haplotype reconstruction from population genotype data. Am J et al. Faciobrachial dystonic seizures precede Lgi1 antibody limbic Hum Genet 2003; 73: 1162–9. encephalitis. Ann Neurol 2011; 69: 892–900. Stephens M, Smith NJ, Donnelly P. A new statistical method for Irani SR, Pettingill P, Kleopa KA, Schiza N, Waters P, Mazia C, et al. haplotype reconstruction from population data. Am J Hum Genet Morvan syndrome: clinical and serological observations in 29 cases. 2001; 68: 978–89. Ann Neurol 2012; 72: 241–55. Thieben MJ, Lennon VA, Boeve BF, Aksamit AJ, Keegan M, Irani SR, Stagg CJ, Schott JM, Rosenthal CR, Schneider SA, Pettingill Vernino S. Potentially reversible autoimmune limbic encephalitis P, et al. Faciobrachial dystonic seizures: the inﬂuence of immuno- with neuronal potassium channel antibody. Neurology 2004; 62: therapy on seizure control and prevention of cognitive impairment 1177–82. in a broadening phenotype. Brain 2013; 136: 3151–62. Thompson J, Bi M, Murchison AG, Makuch M, Bien CG, Chu K, Jia X, Han B, Onengut-Gumuscu S, Chen WM, Concannon PJ, Rich et al. The importance of early immunotherapy in patients with facio- SS, et al. Imputing amino acid polymorphisms in human leukocyte brachial dystonic seizures. Brain 2018; 141: 348–56. antigens. PLoS One 2013; 8: e64683. Trowsdale J, Knight JC. Major histocompatibility complex genomics Kim TJ, Lee ST, Moon J, Sunwoo JS, Byun JI, Lim JA, et al. Anti- and human disease. Annu Rev Genom Hum Genet 2013; 14: LGI1 encephalitis is associated with unique HLA subtypes. Ann 301–23. Neurol 2017; 81: 183–92. van Sonderen A, Roelen DL, Stoop JA, Verdijk RM, Haasnoot GW, Klein CJ, Lennon VA, Aston PA, McKeon A, O’Toole O, Quek A, Thijs RD, et al. Anti-LGI1 encephalitis is strongly associated with et al. Insights from LGI1 and CASPR2 potassium channel complex HLA-DR7 and HLA-DRB4. Ann Neurol 2017; 81: 193–8. autoantibody subtyping. JAMA Neurol 2013; 70: 229–34. van Sonderen A, Schreurs MW, de Bruijn MA, Boukhrissi S, Lai M, Huijbers MG, Lancaster E, Graus F, Bataller L, Balice-Gordon Nagtzaam MM, Hulsenboom ES, et al. The relevance of VGKC R, et al. Investigation of LGI1 as the antigen in limbic encephalitis positivity in the absence of LGI1 and Caspr2 antibodies. previously attributed to potassium channels: a case series. Lancet Neurology 2016; 86: 1692–9. Neurol 2010; 9: 776–85. Vincent A, Buckley C, Schott JM, Baker I, Dewar BK, Detert N, et al. Lang B, Makuch M, Moloney T, Dettmann I, Mindorf S, Probst C, Potassium channel antibody-associated encephalopathy: a poten- et al. Intracellular and non-neuronal targets of voltage-gated potas- tially immunotherapy-responsive form of limbic encephalitis. Brain sium channel complex antibodies. J Neurol Neurosurg Psychiatry 2004; 127: 701–12. 2017; 88: 353–61. Wilson R, Makuch M, Kienzler AK, Varley J, Taylor J, Woodhall M, Makuch M, Wilson R, Al-Diwani A, Varley J, Kienzler AK, Taylor J, et al. Condition-dependent generation of aquaporin-4 antibodies et al. N-methyl-D-aspartate receptor antibody production from from circulating B cells in Neuromyelitis Optica. Brain 2018a; germinal center reactions: therapeutic implications. Ann Neurol 141: 1063–74. 2018; 83: 553–61. Wilson R, Menassa DA, Davies AJ, Michael S, Hester JO, Kuker W, McCormack M, Alﬁrevic A, Bourgeois S, Farrell JJ, Kasperavicˇiut eD _ , et al. Seronegative antibody-mediated neurology after immune Carrington M, et al. HLA-A 3101 and carbamazepine-induced hyper- checkpoint inhibitors. Ann Clin Transl Neurol 2018b; 17: 305–6. sensitivity reactions in Europeans. N Engl J Med 2011; 364: 1134–43. Wirtz PW, Willcox N, van der Slik AR, Lang B, Maddison P, Meeusen JW, Klein CJ, Pirko I, Haselkorn KE, Kryzer TJ, Pittock SJ, Koeleman BP, et al. HLA and smoking in prediction and prognosis et al. Potassium channel complex autoimmunity induced by inhaled of small cell lung cancer in autoimmune Lambert-Eaton myasthenic brain tissue aerosol. Ann Neurol 2012; 71: 417–26. syndrome. J Neuroimmunol 2005; 159: 230–7. Neville MJ, Lee W, Humburg P, Wong D, Barnardo M, Karpe F, et al. Yip VL, Alﬁrevic A, Pirmohamed M. Genetics of immune-mediated High resolution HLA haplotyping by imputation for a British popu- adverse drug reactions: a comprehensive and clinical review. Clin lation bioresource. Hum Immunol 2017; 78: 242–51. Rev Allergy Immunol 2015; 48: 165–75.
Brain – Oxford University Press
Published: Aug 1, 2018
Keywords: human leukocyte antigens; antibodies; lgi1 protein; peptides; hla-drb1 gene
Access the full text.
Sign up today, get DeepDyve free for 14 days.