RESEARCH ARTICLE OPEN ACCESS Guillain-Barré Syndrome Following Zika Virus Infection Is Associated With a Diverse Spectrum of Peripheral Nerve Reactive Antibodies Alexander J. Davies, PhD,* Cinta Lleixà, PhD,* Ana M. Siles, PhD, Dawn S. Gourlay, BSc, Georgina Berridge, BSc, Correspondence Dr. Rinaldi Wanwisa Dejnirattisai, PhD, Carolina Ramı́rez-Santana, PhD, Juan-Manuel Anaya, MD, PhD, simon.rinaldi@nhs.net Andrew K. Falconar, PhD, Claudia M. Romero-Vivas, PhD, Lyda Osorio, MD, PhD, Beatriz Parra, PhD, Gavin R. Screaton, DPhil, FMedSci, Juthathip Mongkolsapaya, DPhil, Roman Fischer, PhD, Carlos A. Pardo, MD, Susan K. Halstead, PhD, Hugh J. Willison, MD, PhD, Luis Querol, MD, PhD,† and Simon Rinaldi, MBChB, PhD† Neurol Neuroimmunol Neuroinflamm 2023;10:e200047. doi:10.1212/NXI.0000000000200047 Abstract Background and Objectives Recent outbreaks of Zika virus (ZIKV) in South and Central America have highlighted sig- nificant neurologic side effects. Concurrence with the inflammatory neuropathy Guillain-Barré syndrome (GBS) is observed in 1:4,000 ZIKV cases. Whether the neurologic symptoms of ZIKV infection are immune mediated is unclear. We used rodent and human live cellular models to screen for anti-peripheral nerve reactive IgG and IgM autoantibodies in the sera of patients with ZIKV with and without GBS. Methods In this study, 52 patients with ZIKV-GBS were compared with 134 ZIKV-infected patients without GBS and 91 non-ZIKV controls. Positive sera were taken forward for target identification by immunoprecipitation and mass spectrometry, and candidate antigens were validated by ELISA and cell-based assays. Autoantibody reactions against glycolipid antigens were also screened on an array. Results Overall, IgG antibody reactivities to rat Schwann cells (SCs) (6.5%) and myelinated cocultures (9.6%) were significantly higher, albeit infrequent, in the ZIKV-GBS group compared with all controls. IgM antibody immunoreactivity to dorsal root ganglia neurones (32.3%) and SCs (19.4%) was more frequently observed in the ZIKV-GBS group compared with other controls, whereas IgM reactivity to cocultures was as common in ZIKV and non-ZIKV sera. Strong axonal-binding ZIKV-GBS serum IgG antibodies from 1 patient were confirmed to react with neurofascin 155 and 186. Serum from a ZIKV-infected patient without GBS displayed strong myelin-binding and putative antilipid antigen reaction characteristics. There was, however, no significant association of ZIKV-GBS with any known antiglycolipid antibodies. Discussion Autoantibody responses in ZIKV-GBS target heterogeneous peripheral nerve antigens sug- gesting heterogeneity of the humoral immune response despite a common prodromal infection. *These authors contributed equally to this work as co–first authors. †These authors contributed equally to this work as co–senior authors. From the Nuffield Department of Clinical Neurosciences (A.J.D., S.R.), University of Oxford, John Radcliffe Hospital, UK; Neuromuscular Diseases Unit (C.L., A.M.S., L.Q.), Neurology Department, Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona, Spain; Centro para la Investigación Biomédica en red en Enfermedades Raras–(CIBERER) Madrid (C.L., A.M.S., L.Q.), Spain; Institute of Infection (D.S.G., S.K.H., H.J.W.), Immunity & Inflammation, University of Glasgow, University Place, UK; Target Discovery Institute (G.B., R.F.), NDM Research Building, University of Oxford, Old Road Campus, UK; Wellcome Centre for Human Genetics (W.D., G.R.S., J.M.), Nuffield Department of Medicine, University of Oxford, UK; Center for Autoimmune Diseases Research (CREA) (C.R.-S., J.-M.A.), Universidad del Rosario, Bogotá, Colombia; Departamento de Medicina (A.K.F., C.M.R.-V.), Universidad del Norte, Barranquilla, Colombia; Grupo de Epidemiologı́a y Salud Poblacional (GESP) (L.O.,), School of Public Health, Universidad del Valle, Cali, Colombia; Department of Micro- biology (B.P.), School of Basic Sciences, Universidad del Valle, Cali, Colombia; Dengue Hemorrhagic Fever Research Unit (J.M.), Office for Research and Development, Siriraj Hospital, Faculty of Medicine, Mahidol Univeristy, Bangkok, Thailand; Department of Neurology (C.A.P.), Johns Hopkins University School of Medicine, Baltimore, MD; and LifeFactors (J.-M.A.), Rionegro, Colombia; Division of Emerging Infectious Disease (W.D.), Research Department, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand. Go to Neurology.org/NN for full disclosures. Funding information is provided at the end of the article. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary ALCAM = activated leukocyte cell adhesion molecule; ANOVA = analysis of variance; ANXA2 = annexin A2; AMAN = acute motor axonal form of GBS; AXL = Axl receptor tyrosine kinase; CNTN1 = contactin 1; DENV = dengue virus; DPYSL2 = dihydropyrimidinase-related protein 2; DRG = dorsal root ganglia; Fc = fragment crystalizable region (of antibody); GBS = Guillain-Barré syndrome; GFRA1 = glial cell-derived neurotrophic factor family receptor alpha-1; hiPSC = human induced pluripotent stem cell; HRP = horse radish peroxidase; IgG = immunoglobulin G; IgM = immunoglobulin M; ITGA7 = integrin alpha 7; L1CAM = neural cell adhesion molecule L1; MS/MS = tandem mass spectrometry; MOPS = 3-(N-morpholino) propanesulfonic acid; NCAM1 = neural cell adhesion molecule 1; NEP = neprilysin; NF155 = neurofascin 155 isoform; NF186 = neurofascin 186 isoform; NF200 = neurofilament heavy chain (200 kDa); NFASC = neurofascin; NHS = normal human serum; NKCC1 = solute carrier Na-K-2Cl cotransporter 1; NrCAM = neuronal cell adhesion molecule; PRX = periaxin; SC = Schwann cell; SEM = standard error of the mean; TGFBR3 = transforming growth factor receptor type 3; VINC = vinculin; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications; ZIKV-OND = ZIKV infection and other inflammatory neurologic diseases. Zika virus (ZIKV) is a member of the Flavivirus family that is uncomplicated infections.10 More recently, a higher incidence transmitted by Aedes spp. mosquito vector species and in hu- of antiganglioside antibodies was identified in a Brazilian co- mans typically results in asymptomatic or mildly symptomatic hort of patients with ZIKV-GBS compared with un- infections. An outbreak of ZIKV infections in French Polynesia complicated ZIKV controls,11 suggesting an association in 2013, however, was followed by a spike in the diagnoses of the between nerve reactive autoantibodies and the peripheral disabling, acute inflammatory neuropathy Guillain-Barré syn- neurologic complications of ZIKV. drome (GBS). Early case-control studies established 1 excess GBS case for every 4,000 people infected with ZIKV.1 By 2015, The peripheral neurology associated with ZIKV infection may the ZIKV pandemic reached South America, where in Colombia, also occur via direct viral toxicity.12 ZIKV directly infected the GBS incidence rates were increased by 210%–350% com- peripheral neurons and induced cell death in both mouse and pared with the pre-ZIKV period.2,3 The overall risk estimates cell culture models,13 as well as infecting sensory neurons and range from 1 to 8 GBS cases per 10,000 ZIKV infections.4 Schwann cells (SCs) in dorsal root ganglia (DRG) explants Recent studies estimate that the total number of ZIKV infections from interferon receptor 1 knockout mice leading to myelin during the 2015–2016 outbreak was much higher than originally fragmentation and later axonal degeneration.14 However, the reported, suggesting that previous incidence rates of GBS per presence of mononuclear inflammatory cells in a biopsy of the infection may have been substantially overestimated.5 fascicular sural nerve in a male patient with ZIKV-GBS, consistent with classic GBS, and the lack of viral transcript The onset of GBS post-ZIKV infection (ZIKV-GBS) oc- detection by PCR suggest that neurotropism is unlikely to be curred within a median of 6 days after a transient febrile a major causative factor in the demyelination and axonal de- illness, although the peak incidence of GBS cases was ob- generation observed within the nerve.15 served 3 weeks after the peak of ZIKV diagnoses.1,2 Whether this represents an infectious, parainfectious, or postinfectious The antibody response is thought to be critical for viral etiology has been debated. Electrophysiologic studies per- control in ZIKV infection.16 We therefore sought to address formed on 37 patients with ZIKV-GBS identified in French the potential for humoral immunity in ZIKV-GBS. In this Polynesia showed characteristics consistent with the acute study, we screened sera from a large cohort of patients with motor axonal form of GBS (AMAN).1 However, follow-up ZIKV-GBS and healthy and disease controls for nerve cell studies at 4 months were more suggestive of distal de- reactive immunoglobulins. We used a range of animal and myelination.6 In a Colombian series, 70% of ZIKV-GBS cases humanized nerve cell cultures we have previously shown ca- were classified as acute inflammatory demyelinating poly- pable of detecting nerve cell reactive antibodies in a range of radiculoneuropathy, suggestive of immune-mediated injury.7 autoimmune diseases17-20 and proteomic-based antigen de- ZIKV has also been linked to a number of central and pe- tection methods. Our results show that peripheral nerve re- ripheral neurologic complications, all of which may have an active autoantibodies are indeed a feature of a subset of immune-mediated etiology.8 patients who developed GBS following ZIKV infection, tar- geting a broad range of peripheral nerve–related antigens. A potential role for ZIKV-induced immunity is supported by similarities between the ZIKV E protein and human com- plement component C1q in silico and the presence of anti- Methods C1q antibodies in the sera of animals infected with the virus.9 Higher-titer antibody responses to ZIKV have been reported Full details of this section in eMethods, as well as additional in patients who developed GBS compared with those with data in eFigures and eTables, links.lww.com/NXI/A760, are 2 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN available in the Supplement. Any additional methodologic anti-human complement C3c-specific antibodies. Confocal details or data not shown are available on reasonable request images were assessed for IgG or IgM antibody reactivities to the corresponding authors. by a blinded observer. Patient Cohorts and Sample Acquisition Assessment of Serum-Induced Demyelination Serum samples were collected from patients with ZIKV in- Myelinating cocultures were incubated for 1 hour in complete fection both with and without neurologic complications and neurobasal media supplemented with fluoromyelin red. from other infectious, healthy, and convalescent controls from Serum-free myelination medium was then supplemented with the Cucuta, Cali, and Barranquilla regions of Colombia, the or without human anti-ZIKV patients’ sera at a 1:100 dilution United Kingdom, and Spain (Figure 1 and eTable 1, links. together with 20% normal human serum (NHS) as a source of lww.com/NXI/A760). Patients with ZIKV-GBS were classi- complement and added to the cocultures. For the assessment fied according to previously published electrodiagnostic cri- of demyelination in postfixed cultures, a preset of 5 × 5 grid of teria,7 with full characterizations previously reported for positions (or 7 × 7 for sparsely myelinated coverslips) at ×20 Cucuta21 and Cali cohorts.3 Serum samples were collected magnification (0.5 digital zoom) was used for automated ac- after treatment, which was given in approximately 60% of quisition across each coverslip. patients with ZIKV-GBS (typically infusions of IV immunoglobulin).3,21 Samples were aliquoted and stored at Glycoarray −80°C. This study was conducted and reported in accordance Serum samples were also screened on a glycolipid microarray with the STROBE guidelines.22 as previously described.25 For this study, sera were screened against a panel of 16 single glycolipids (GM1, GM2, phos- Standard Protocol Approvals, Registrations, phatidylserine, GM4, GA1, GD1a, GD1b, GT1a, GT1b, and Patient Consents GQ1b, GD3, SGPG, LM1, cholesterol, GalC, and sulfatide) Sample collection was performed in compliance with Act and 120 heteromeric 1:1 (vol/vol) complexes, printed in 008430/1993 of the Ministry of Health of the Republic of duplicate. Colombia and the institutional review boards of the Uni- versidad del Rosario and Universidad del Valle in Immunoprecipitation and Mass Spectrometry Colombia. Additional samples from healthy participants, Human sera showing moderate or strong reactivity against those with non–immune-mediated neurologic disorders, rat DRG neurons or myelinating cocultures were used for and nodal/paranodal antibody-positive patients were immunoprecipitation (IP) experiments using the same tar- obtained for use from local collections in Oxford, United get cell type, as previously reported.23 hiPSC lines from 4 Kingdom (South Central–Oxford A Research Ethics donors were used to generate myelinating cocultures to give Committee approval number 14/SC/0280) or Barcelona, 4 biological replicates per serum sample. IP eluates from Spain (The Sant Pau Biomedical Research Institute ethical myelinating cocultures were prepared by chloroform:meth- approval number IIBSP-AUT-2016-69). All patients gave anol precipitation and in-solution trypsin digestion. The written informed consent to participate in the study. resultant peptides were then subjected to high-resolution C18 reversed-phase column chromatography and analyzed Rat DRG Neuron and SC Culture and Nerve by data-dependant MS/MS on a ThermoFisher Fusion Section Immunocytochemistry Lumos mass spectrometer. Significant data points were de- Rat DRG neurons and SCs were extracted in accordance with termined with a permutation-based false discovery rate cal- Schedule 1 of the UK Home Office Animals (Scientific Pro- culation (threshold = 0.1). Raw data are available via cedures) Act 1986 and Animal Ethics’ Committee of Hospital ProteomeXchange with identifier PXD028476. de la Santa Creu i Sant Pau. Cells were cultured, and immu- nocytochemistry was performed as previously described.23 Protein Electrophoresis and Western Blot Fluorescence signal intensities were scored on a 0–3 scale by 2 The lysates from myelinating cocultures, hiPSC-derived independent researchers. Immunohistochemistry on ma- sensory neuron monocultures, and primary rat SCs were caque peripheral nerve tissue slides (Inova Diagnostics, Inc., subjected to electrophoresis in acrylamide gels using MOPS San Diego, CA) was performed and analyzed as previously running buffer under denaturing, nonreducing conditions and described.20 then transferred to a nitrocellulose membrane. These blots were probed with using patients’ sera at a 1:2,500 dilution Myelinating Cocultures and followed by HRP-conjugated Fc-specific anti-human IgG Immunofluorescence Labeling secondary antibody. The bands detected by Western blotting Myelinating cocultures were prepared using human in- were aligned in a parallel gel stained with Pierce Imperial duced pluripotent stem cell (hiPSC)-derived sensory Protein Stain and excised for mass spectrometry (MS). neurons and neonatal rat SCs as previously described.17,24 Sera from ZIKV-exposed patients were diluted to 1:100 and ELISA added to live cocultures for 1 hour at 37°C. Fixed cultures Recombinant human nidogen-1, laminins 111, 121, 211, were probed with anti-human IgG, anti-human IgM, or 221, 411, 421, 511, and 521, prosaposin, and vinculin were Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 3 Figure 1 Diagram Illustrating the Patient Cohorts and the Experimental Workflow With Their Serum Samples The numbers of patients’ serum samples collected in the different categories of ZIKV-GBS (patients with ZIKV who developed GBS), ZIKV-OND (patients with ZIKV who dis- played other neurologic diseases), ZIKV-CON (patients with ZIKV who never developed clinical neurologic disease), DENV-CON (patients with DENV who never developed clini- cal neurologic disease), ALS (patients with amyotrophic lat- eral sclerosis, a progressive nervous system disease that affects nerve cells in the brain and spinal cord, causing loss of muscle control), dysferlin (patients with dysferlinopathy, a muscle disease that has a slow progression of muscle weakness and atrophy), and healthy controls (n, number of patient samples). The patients’ serum samples were initially screened for their autoantibody reactions against different neuronal cell culture types and from which the sera from patient A to D’s serum samples were then tested for their reactions in additional assays. coated at 1 μg/mL (100 μL per well) in Maxisorp Statistical Analysis ELISA plates overnight at 4°C. The ganglioside GM3 The results were analyzed in Prism v9.1.0 (GraphPad). Sta- ELISA was performed as previously described.26 For the tistical comparisons of the proportions of seropositive pa- detection of anti-ZIKV antibodies, Maxisorp plates were tients among the ZIKV groups were performed using coated with the anti-flavivirus mouse monoclonal anti- contingency analyses with the application of a 2-tailed Fisher body 4G2 and performed as previously described.27 exact test for individual group-group comparisons and the χ 2 Bound human IgG antibodies were detected through se- test for comparing antibody prevalence between all groups. quential reaction steps using Fc-specific HRP-conjugated Analysis of confocal microscopy images was performed by the secondary antibodies and substrate. Total human serum Student t test for comparison of 2 groups and ANOVA for IgG levels were interpolated from standard curves comparison of 3 or more groups, with correction for multiple according to the manufacturer’s instructions (Invitrogen, comparisons where indicated. Cat. 88-50550-86). Data Availability MS proteomics data were deposited to the ProteomeXchange Transfected Cell-Based Assays Consortium via the PRIDE partner repository with the data HEK293 cells were transfected overnight using JetPEI with set identifier PXD028476. A summary of MS data and all mammalian expression vectors encoding human AHNAK2, other anonymized data is available on reasonable request to ANXA2, CD44S, CNTN1, CASPR1, GFRA1, ITGA6, the corresponding authors. ITGA7, MAG, NEP, NFASC, NKCC1, PRX, and TGBR3 or using Lipofectamine 2000 for human ALCAM, AXL, DPYSL2, GAS6, L1CAM, NCAM1, and NrCAM. Sera Results were diluted 1:100, and IgG antibody binding was assessed Patient Cohorts using fluorescent-conjugated anti-human IgG secondary Serum samples were obtained from a total of 218 Colombian antibodies. participants: 52 patients with GBS and confirmed ZIKV 4 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Table 1 Statistical Analysis of Serum Immunoreactivity to DRG Neurons and SC Cultures Ig class ZIKV-GBS (n = 31) ZIKV-OND (n = 12) ZIKV-CON (n = 77) CON (n = 43) p Value (χ2) DRG neurons IgG 5 (16.1%) 2 (16.7%) 3 (3.9%) 3 (7%) 0.1193 DRG neurons IgM 10 (32.3%) 2 (16.7%) 3 (3.9%) 0 <0.0001 (***) Schwann cells IgG 2 (6.5%) 0 0 0 0.0348 (*) Schwann cells IgM 6 (19.4%) 0 6 (7.8%) 0 0.0121 (*) Any reactivity IgG/IgM 12 (38.7%) 4 (33.3%) 8 (10.4%) 3 (7%) 0.0004 (***) *p < 0.05, ***p < 0.001.8 Abbreviations: DRG = dorsal root ganglia; GBS = Guillain-Barré syndrome; SC = Schwann cell; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications; ZIKV-OND = ZIKV infection and other inflammatory neurologic diseases. infection (ZIKV-GBS), 17 patients with ZIKV infection and the GBS subtype Miller-Fisher syndrome and another with other inflammatory neurologic diseases (ZIKV-OND), 117 anti-GM1 IgG antibodies, typically present in the AMAN ZIKV-infected patients without neurologic complications subtype. (ZIKV-CON), and a control group (CON) formed by 23 patients with dengue virus (DENV) infections (DENV- Screening Against Rat DRG Neurons and SCs CON) from Cali and 9 healthy controls from Cucuta. An IgM autoantibodies that were moderate or strongly reactive to additional 38 serum samples (5 patients with amyotrophic rat DRG were observed in proportionally more sera (32.3%: lateral sclerosis, 5 dysferlin patients, and 28 healthy controls) p ≤ 0.0001) of patients with ZIKV-GBS compared with pa- from Spanish participants and 21 healthy controls from the tients with ZIKV-OND (16.7%), patients with ZIKV-CON United Kingdom were also used (Figure 1). (3.9%), and those from other control (CON) groups (0.0%) (Table 1). On the other hand, moderate or strong IgG au- Overall, the patients with ZIKV-GBS were slightly older and toantibody reactivities against DRG neurons were not statis- less often female (eTable 1, links.lww.com/NXI/A760) and tically different (p = 0.1193) between these groups. With the showed a modest but significantly higher total serum IgG level SCs, a greater proportion of moderate or strong IgG (median 4.162 mg/mL; 95% CI 3.473–4.651) than ZIKV- (p = 0.0348) and IgM (p = 0.0121) antibody reactivities were CON (median 3.221 mg/mL; 95% CI 3.004–3.615) and observed in the sera of patients with ZIKV-GBS compared DENV-CON (median 3.210 mg/mL; 95% CI 2.242–3.703) with other patients with uncomplicated ZIKV and control and healthy controls (median 3.051 mg/mL; 95% CI groups (Table 1). 2.086–3.529), but not ZIKV-OND samples (median 3.601 mg/mL, 95% CI 2.821–3.927) (1-way ANOVA, F(4, 222) = Overall, there was a greater chance of observing any anti- 4.481, p = 0.0017) (eFigure 1, links.lww.com/NXI/A760). peripheral nerve cell IgG or IgM antibody reactivity in the ZIKV capture ELISA showed the presence of anti-ZIKV IgG sera of patients with ZIKV-GBS (38.7%) and ZIKV-OND in all ZIKV-GBS, ZIKV-OND, and ZIKV-CON patients’ sera (33.3%), compared with noncomplicated ZIKV (ZIKV- tested, as well as the DENV patients’ sera, likely due to the CON) (10.4%) and non-ZIKV control (CON) (7.0%) cross-reactivity of DENV and ZIKV patients’ antibodies27 groups (p = 0.0004, χ 2 test). In addition, we observed that (eFigure 1, links.lww.com/NXI/A760). IgG and IgM antibody binding to DRG neurons or SCs was significantly more common in the sera of patients with Microarray Screening for Glycolipid ZIKV-GBS than in the sera of the ZIKV-CON group Complex Antibodies (p = 0.0042, Fisher exact test), whereas there was no significant ZIKV-associated patients’ sera were screened against a panel difference in the proportions of grouped anti-ZIKV sera and of glycolipids using a combinatorial glycoarray for the pres- non-ZIKV sera (p > 0.05, Fisher exact test) (see eFigure 3, ence of IgG (eFigure 2A, links.lww.com/NXI/A760) and IgM links.lww.com/NXI/A760, for multiple comparisons). (eFigure 2B, links.lww.com/NXI/A760) anti-glycolipid anti- bodies. Overall, the glycolipid array revealed only weak anti- Screening Against Myelinating Cocultures body binding intensities against the panel of 136 unique IgM antibody binding to myelinating cocultures was fre- glycolipid targets, including known gangliosides implicated in quently detected with both the sera of patients with ZIKV- GBS, none of which were found to be statistically associated GBS and the control group (Table 2 and eFigure 1, links. with post-ZIKV GBS. Two ZIKV-GBS samples were found to lww.com/NXI/A760). IgM antibody reactivity was ob- contain IgG antibodies against various gangliosides at elevated served against the abaxonal membranes of myelinating SCs, levels, as typically seen in GBS and its clinical variants. This axons, the processes of the nonmyelinating SCs, and out- included a single ZIKV-GBS sample with anti-disialosyl IgG pouchings of myelin (eFigure 4A and 4B, links.lww.com/ (GQ1b, GT1a, and GD3) antibodies normally associated with NXI/A760). Overall, however, there was no significant Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 5 Table 2 Frequency and Statistical Analysis of Myelinating Coculture Immunoreactivity by Pattern Myelinating coculture reactivity Myelin/abaxonal Myelin outpouchings Schwann cell Nodal/axonal Any reactivity Ig class IgG IgM IgG IgM IgG IgM IgG IgM IgG IgM Patient cohort ZIKV-GBS (n = 52) 4 (7.7%) 2 (3.8%) 1 (1.9%) 7 (13.5%) 0 4 (7.7%) 1 (1.9%) 1 (1.9%) 5 (9.6%) 12 (23.1%) ZIKV-CON (n = 117) 1 (0.9%) 1 (0.9%) 0 25 (21.4%) 0 6 (5.1%) 0 0 1 (0.9%) 32 (27.1%) ZIKV-OND (n = 17) 0 0 0 1 (5.9%) 0 2 (11.8%) 0 0 0 3 (17.6%) CON (n = 53) 0 0 0 7 (13.2%) 0 2 (3.8%) 0 0 0 8 (15.1%) 2 χ p Value 0.0213 (*) 0.2773 0.3066 0.2497 — 0.5885 0.3066 0.3066 0.0031 (**) 0.3347 *p < 0.05, **p < 0.01. Abbreviatioms: GBS = Guillain-Barré syndrome; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications; ZIKV-OND = ZIKV infection and other inflammatory neurologic diseases. difference in the proportions of the sera of patients with (5/52; 9.6%) compared all controls (1/187; 0.5%) (p = ZIKV-GBS and controls in terms of the frequency or pat- 0.0021, Fisher exact test) (see Table 2 for χ 2 tests on multiple tern of positive IgM antibody reactivity (p = 0.3374, Fisher comparisons). exact test) (see Table 2 for multiple comparisons). IgM antibody labeling of myelin outpouchings varied in in- Nerve Section Immunoreactivity tensity between the samples (eFigure 4C and 4D, links. We also analyzed the reactivity of ZIKV-GBS (n = 30) and lww.com/NXI/A760); however, immunofluorescence in- ZIKV-CON (n = 40) samples from the Barranquilla and tensity at the myelin outpouchings normalized to the in- Cucuta cohorts against macaque peripheral nerve. Overall, ternode (eFigure 4E and 4F, links.lww.com/NXI/A760) IgG and IgM reactivity against nerve tissue was frequently was again not significantly different between the serum detected in patients with ZIKV-GBS (40%) and patients with samples of patients with ZIKV-GBS, patients with ZIKV- ZIKV-CON (25%), although not statistically significant. We CON, or the healthy control group (1-way ANOVA, did, however, observe that 8 (26.7%) patients with ZIKV-GBS F(2,106) = 0.7417, p = 0.4788) (eFigure 4G, links.lww. showed strong reactivity against nonmyelinating SCs (4 pa- com/NXI/A760). IgG antibody reactivity was observed tients with IgG, 2 patients with IgM, and 2 patients with both against different topographic domains within the cocul- IgG and IgM antibodies), whereas only 1 (2.5%) patient with tures. Sera from 4 of 52 (7.7%) patients with ZIKV-GBS ZIKV-CON showed immunoreaction of IgG (IgG, p = contained IgG antibodies that bound the abaxonal mem- 0.0371; IgM, p = 0.0299) (data not shown). brane of a subset of myelinating SCs (Table 2), while showing no apparent reactivity against nonmyelinating SCs Combined Screening Results or axons (eFigure 5A and 5B, links.lww.com/NXI/A760). Overall, 125 samples were tested against both isolated pri- mary cells and myelinating cocultures. Only, 1/13 (16.0%) IgG antibodies from the serum of 1 patient with ZIKV-GBS IgG-positive sera (ZN005A) and 8/50 (16.0%) IgM-positive (male, 16–20 years old) were deposited on noncompact sera were identified in both systems. The kappa statistics for myelin at paranode-like regions of the node of Ranvier and inter-test agreement were 0.129 (IgG) and 0.084 (IgM). Schmidt-Lanterman incisures (eFigure 5C, links.lww.com/ NXI/A760). IgG antibodies from the serum of another ZIKA- Assessment of Antibody-Mediated GBS patient (patient A, male, 41–45 years old) labeled the Axonopathy and Demyelination axons of cultured neurons and was particularly concentrated We next investigated the pathologic potential of IgG auto- at the nodal axolemma (eFigure 5D, links.lww.com/NXI/ antibodies in the 2 serum samples that reacted strongly A760). These patterns of IgG antibody reactivity were never against the cell cocultures: patient A (eFigure 5D, links.lww. seen with the control sera. The serum of 1 patient with ZIKV- com/NXI/A760) and patient B (eFigure 5E, links.lww.com/ CON (patient B, female, 41–45 years old) also contained IgG NXI/A760). The axonal and nodal binding of IgG antibodies antibodies, which gave an intense pattern of immunofluo- present in patient A’s serum was exclusively of the IgG1 rescence over the abaxonal membranes of all myelinating SCs, subclass (data not shown). The addition of patient A’s serum and at the nodal microvilli (eFigure 5E, links.lww.com/NXI/ at a 1:100 dilution did not result in any axonal degeneration or A760), but no IgG antibody labeling occurred against sensory other visible morphologic nerve injury in the presence of neuron or SC monocultures (data not shown). Overall, IgG complement in 20% NHS compared with that caused by a antibody reactivity against myelinating cocultures was signif- known complement-fixing anti-ganglioside IgM-positive icantly more common in the sera of patients with ZIKV-GBS control antibody (data not shown). Antibodies reactive to 6 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Figure 2 Complement-Dependent Demyelinating Activity of ZIKV Patient Sera The autoantibodies in the sera of ZIKV-CON patient B (BQ008) reacted with myelin internodes and fixed complement (C3c deposition; green) over myelin internodes (MBP; red) when complement was present from 20% normal human sera (NHS) (A) but not when only NHS was added to the cells (B). Coun- terstaining for NF200 (blue, axons) and MBP (red) is also shown. (C) Fluoromyelin labeling (red) from live imaging of a myelinated coculture before (above) and 16 hours after (below) incubation with patient B’s serum at a 1:50 dilution in the presence of complement in 20% NHS, while (D) shows the results of time-lapse stills of the receding myelin internode junctions shown from right to left over time (arrows). (E) shows the overlaid example images of live myelin staining at baseline (red) and after 24 hours of incubation (silver) with patient B’s serum with or without 20% NHS containing complement. (E.b) shows numerous myelin internodes are either lost completely or extensively fragmented after 24 hours (shown by arrows), while (E.a) without NHS containing complement the distribution of the myelin was essentially unchanged over 24 hours. (F) The mean internodal length was significantly reduced at 24 hours compared with baseline in the ZIKV-CON + NHS group only (****p < 000.1, 2-way ANOVA with Sidak correction); ns = not significant. MBP = myelin basic protein; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications. myelinating SCs in patient B’s serum were also predominantly presence of complement (Figure 2, C and D). After 24 hours, of the IgG1 antibody subclass, with some IgG4 antibody multiple myelin internodes were either lost completely or subclass reaction against either SC processes or axons (data significantly fragmented compared with baseline (0 hour) not shown). (Figure 2E). ZIKV-CON serum treatment of the myelinating cocultures in the presence of complement in the 20% NHS, The addition of patient B’s sera at a 1:100 dilution to cocul- but not ZIKV-CON serum or NHS alone, led to a significant tures in the presence of 20% NHS resulted in serum com- loss of myelin coverage and shortening of internode length plement deposition (C3c) around myelin internodes and (Figure 2F), thereby confirming the effect seen using the myelin fragmentation without apparent axonal degeneration time-lapse imaging (Figure 2D). In a confirmatory experi- (Figure 2, A and B). Time-lapse imaging of fluoromyelin- ment (eFigure 6A, links.lww.com/NXI/A760), the addition stained internodes revealed retraction of the myelin from the of myelin-binding IgG from patient B serum BQ008 (1:50) to node within hours of exposure to the anti-ZIKV serum in the the cocultures for 24 hours also resulted in smaller myelin Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 7 Figure 3 Identification of a ZIKV Patient’s Autoantibodies Against Myelin Antigen: Immunoprecipitation and Peptide Mass Spectrometry Immunofluorescent photomicrographs of (A) a positive control patient’s serum with autoanti- bodies against the known nerve antigen NF155 and (B) ZIKV-CON patient B’s serum autoantibody reac- tions against unknown antigens enriched in abaxonal myelin, with higher magnifications of the nodal regions shown as insets (far right). Scale bars, 20 μm. Counterstaining for NF200 (blue) and MBP (red) is also shown. (C) Proteins identified in im- munoprecipitates from ZIKV-CON patient B serum and NF155+ serum-treated coculture lysates. The t test of enriched proteins presented as a volcano plot. Significantly enriched membrane proteins are shown as red squares. NFASC protein isoforms from a human and rat database are grouped together (blue circle). AHNAK2 = desmoyokin; CD44 = P-glycoprotein 1; ITGA6 = integrin alpha 6. MBP = myelin basic protein; NEP = neprilysin; SLC12A2 = solute carrier 12/Na-K-2Cl cotransporter A2; TGBR3 = transforming growth factor receptor 3; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications. fragments (F[3,214] = 6.871, p = 0.0002, 2-way ANOVA; samples BQ010 (M, 36–40 years) and BQ024 (F, 46–50 interaction between serum sample and complement with the years). Neuronal density was not affected (eFigure 6D, links. Sidak multiple comparison test; p = 0.0002) (eFigure 6B, lww.com/NXI/A760). links.lww.com/NXI/A760) and a greater proportion of frag- mented internodes (p < 0.0001, Fisher exact test) after the Antigen Identification in addition of 20% NHS (eFigure 6C, links.lww.com/NXI/ Myelinating Cocultures A760), which was not observed with DRG/SC reactive ZIKV- The nodal/axonal binding ZIKV-GBS sera of patient A GBS sample BQ004 (M, 41–45 years) (eFigure 6A, links.lww. (eFigure 5D, links.lww.com/NXI/A760) showed a similar com/NXI/A760) and 2 other nonreactive ZIKV-CON binding pattern to sera from a separate study containing anti- 8 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Figure 4 Removal of ZIKV-CON Patient B’s Serum IgG Autoantibody Target Antigen(s) in Myelinated Sensory Neuron Cultures by Lipid Extraction (A) Immunofluorescent photomicro- graphs of ZIKV-CON patient B’s serum IgG autoantibodies showing strong staining with untreated (control) abaxonal membranes of all myelinating SCs, which was particularly intense at the Schwann cell microvilli overlying the node of Ranvier (arrowhead) while (B) weakly staining these same cells after solvent treatment (solvent). The control reactions for neuronal (NF200) and my- elin (MBP) markers remained un- changed. For this study, the serum- treated fixed cultures were incubated with a mixture of chloroform, methanol, and water at a ratio of 4:8:3 for 1 hour on ice to remove glycolipids. The white boxes indicate regions of fluorescence profiling per internode (5 × 20 μm). Ar- rows indicate weak paranode-like IgG antibody binding after lipid extraction. (C) Fluorescence intensity profiles of the human IgG autoantibodies and MBP immunolabeling across the internodes of control and solvent-treated cultures. Data are presented as mean (± SEM in gray) of 2 internodes per field of view, 12–14 fields of view from 2 independent experiments. (D) Quantification of IgG: MBP fluorescence ratio. Student un- paired t test, p < 0.0001, t = 5.053, n = 24–28 internode profiles per group. MBP = myelin basic protein; ZIKV = Zika virus; ZIKV-CON = ZIKV-infected patients without neurologic complications. CNTN1 antibodies.28 Both sera maintained a similar IgG monocultures. Three potential antigens were identified as antibody binding pattern when reacted with neuronal enriched in the IP: NFASC, NCAM1, and NCAM2, con- monocultures, thereby confirming the neuronal origin of both firming that NFASC was a target antigen of patient A’s serum. the unknown ZIKV-GBS autoantigen and CNTN1. By con- trast, no antibody labeling was observed with healthy control The abaxonal SC binding by ZIKV-CON patient B’s serum serum (data not shown). In accordance with the nodal IgG was compared with a patient with anti-NF155 antibodies by antibody labeling observed in the myelinating cell cocultures, IP-MS, which showed a similar pattern of IgG autoantibody we tested for the nodal/paranodal candidate neurofascin binding (Figure 3, A–C). A total of 28 proteins were signifi- (NF) in cell-based assays. Heterologous expression of NF155 cantly enriched by patient B’s sera including immunoglobulin and NF186 isoforms in an HEK cell-based assay confirmed complement, and intracellular proteins, and 7 membrane- their antigen-specific targeting by IgG1 antibodies in this related proteins: ANXA2, TGFBR3, NKCC1, NEP, CD44, ZIKV-GBS patient’s sera (data not shown) who subsequently ITGA6, and AHNAK2 (Figure 3C). died due to their illness. This anti–pan-NFASC protein iso- form IgG antibody autoreactivity pattern was not detected in Sera from 2 of the patients with ZIKV-GBS (patient C and any of the other ZIKV-GBS patients’ or control sera. patient D), which showed moderate or strong antibody re- activity against primary DRG neurons, and 1 from a healthy Immunoprecipitation and Mass control were also taken forward for IP-MS analysis. Antigen- Spectrometry (IP-MS) bound IgG antibodies from patient C’s sera and patient D’s sera We used a proteomic approach to identify candidate antigens or secondary IgG anti-IgM antibodies bound to patient D’s from our live cell culture screening by isolating antigen-bound IgM anti-neuronal autoantibodies were isolated from DRG IgG antibodies from serum-treated cultures using protein G neuron lysates by IP and processed for MS. In these experi- bead IP followed by high-resolution C18 reversed-phase ments, 6 potential autoantigens (ALCAM, DPYSL2, NCAM1, chromatography and then MS analyses. We first validated CNTN1, L1CAM, and VINC) were identified as enriched by this approach using patient A’s serum in neuronal cell these anti-ZIKV sera compared with the control sera. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 9 Evaluation of Candidate Antigens affected by prior neuraminidase treatment of the cell cocul- Candidate antigens identified by IP-MS were evaluated by tures (p = 0.3605, t = 0.9340, n = 12 internode profiles per either ELISA or heterologous expression in live cell-based group, Student unpaired t test) compared with anti-GD2, anti- assays. ELISAs using human recombinant prosaposin and GD3 anti-disialosyl ganglioside antibody–positive controls (p < GM3 (alone or in combination), sulfatide, multiple human 0.001 and p < 0.0001, unpaired Student t test; data not shown), laminin isoforms, and nidogen-1 failed to confirm these as suggesting that the target antigen was not a sialylated glycolipid. targets of patient B’s serum IgG autoantibodies. Neither heterologous expression in HEK cells nor ELISAs Discussion revealed binding of ZIKV-CON patient B’s serum IgG anti- bodies to any of the large number of other candidate proteins Our study found that the humoral immune response in ZIKV- (see eMethods, links.lww.com/NXI/A760, for the list of GBS cases was diverse and heterogeneous. We found neither a plasmids). Furthermore, neither the IgG or IgM antibodies common antigen nor a common pattern of serum immunore- from ZIKV-GBS patient C or D reacted against any of the activity against peripheral nerve structures despite all these pa- candidate antigens by heterologous expression in these cell- tients with ZIKV-GBS sharing a common prodromal infection. based assays. We additionally analyzed IgG and IgM auto- antibody reactivity of anti-ZIKV sera against GAS6 and AXL Consistent with a recent study of a modest cohort of patients proteins, which have been described as the receptor ZIKV with ZIKV-associated GBS from Brazil,11 we observed that a virus entry to human neural cells and NrCAM, which shares 6 small number of patients with GBS had elevated anti- common peptide sequences with the ZIKV polyprotein. glycolipid IgG antibodies. However, we did not find a sig- However, none of these sera reacted against these 3 candidate nificant statistical association of ZIKV-GBS patient cases with antigens or members of the AXL-GAS6 complex. any anti-glycolipid antibody signature, thereby supporting the results from a recent cohort study from Northeast Brazil.30 In To account for the potential loss of antigen during the IP this study, we did not perform any additional screening for process, we subjected the lysates of myelinating cell cocultures Campylobacter. However, in the Cucuta cohort, IgG anti- to electrophoresis on an acrylamide gel, and after electro- bodies for Campylobacter jejuni were measured and found to blotting onto nitrocellulose membranes, the reactions of pa- be present in 24.14% of patients with GBS and in 16.22% of tient B’s serum autoantibodies were tested against these nerve controls, with no significant difference.21 Thus, in accordance cell antigens. Two bands of relative molecular weights of with our glycoarray data, a humoral response to C jejuni is approximately 62 and 50 kDa were strongly detected by the possible in a minority of patients but unlikely to be a key serum IgG antibodies in the myelinated cell cocultured lysate, driver of neurologic disease in ZIKV-GBS. but not in the neuronal or Swann cell lysates (data not shown). Manual filtering for membrane proteins associated Analysis of the ZIKV polyprotein showed that it contained with SCs or myelination identified 3 potential target antigens: peptide sequences that were also found in human proteins and PRX, ITGA7, and GFRA1. However, no IgG antibody re- that are known to be targets of inflammatory neuropathy auto- activity to these particular antigens was observed using patient antibodies.31 However, although all ZIKV patient samples tested B’s serum on transiently transfected cell-based assays. contained anti-ZIKV IgG antibodies, anti-nerve reactive auto- antibodies showed a more discrete distribution, suggesting that Chemical and Enzymatic Characterization of any nerve-related auto-antigenicity was probably not a conse- Target Antigens quence of humoral immunity to ZIKV antigens per se. Despite The intense labeling of the patient B’s serum autoantibodies the high level of antigenic similarity between flaviviruses, we did in the lipid-rich membrane and nodal microvilli of myelinating not observe any selective nerve cell reactive autoantibodies in the SCs (Figure 4A) led us to further investigate the target anti- DENV-infected control (DENV-CON) sera, in keeping with the gen characteristics using a biochemical approach. Solvent rarity of GBS associated with recent DENV infections.32 Al- extraction of lipids from serum-treated and fixed myelinated though we observed 1 case of anti–pan-NFASC protein isotype cocultures with a mixture of chloroform, methanol, and water IgG1 autoantibodies in our Colombian cohort (ZIKV-GBS pa- as described previously29 significantly reduced the amount of tient A), prodromal infection is not a consistent feature of this patient B’s serum IgG antibodies, which bound to the rare and potentially fatal peripheral neuropathy subtype.33 abaxonal SC membranes (Figure 4, B–D) (p < 0.0001, Stu- dent t test), whereas the reactivity of commercial antibodies By using a comprehensive series of peripheral nerve cell cul- directed against NF200 and MBP antigens was retained ture models and antigen expression systems, we reveal a (Figure 4, B and C). Solvent extraction did not affect labeling surprisingly heterogeneous autoantibody response with po- of human IgG autoantibodies to the GPI-linked protein tential nerve reactivity in patients’ sera. Overall, IgG reactivity CNTN1 (p = 0.256, t = 1.223, df = 8, Student unpaired t test, to rat SCs and the myelinating cocultures was significantly n = 5 coverslips per group; data not shown). Limited IgG higher, albeit infrequent, in the ZIKV-GBS patient group antibody labeling remained at the paranodes (Figure 4B). compared with all controls. IgM antibody reactivity against rat Binding of patient B’s IgG antibodies was not, however, DRG neurons and rat SCs was also significantly higher in 10 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN patients with ZIKV-GBS than in the control group. Moreover, cell-based assays, reinforcing the challenges of antigen identifi- we also observed a higher proportion of ZIKV-GBS sera cation in the inflammatory neuropathies.37 Nevertheless, the reacting against nonmyelinating SCs in peripheral nerve selective patterns of immunoreactivity, such as those against sections, compared with the ZIKV-CON group. On the other nodal structures, axons, or myelin internodes in the myelinating hand, IgM antibody immunoreactivity to myelin out- cocultures, were not observed in any control samples. This pouchings, a feature reminiscent of the myelin foldings, suggests that the near-native antigen conformation in the cell swellings, or tomaculae sometimes seen in certain hereditary coculture system is a high-fidelity substrate for screening po- neuropathies,34 was remarkably common in both the ZIKV- tentially pathogenic patients’ serum IgG autoantibodies.17-19 GBS patient and control groups. The significance of these structures in the coculture system is unclear. A limitation of our study is that our antigen identification protocols were optimized for proteins and not for lipids. In fact, Over 50% of our ZIKV-GBS cohorts received IVIg treatment solvent-based lipid extraction significantly reduced reactivity of at some point before sampling. We therefore asked whether patient B’s IgG autoantibody reactions against myelinating SCs artificially elevated serum IgG levels may lead to an increase in in the coculture system, suggesting that a lipidomic approach false positives on the cell-based assays. Although overall IgG may be required. It is also possible that certain protein antigens levels were on average slightly higher in the ZIKV-GBS group, were not detectable, either due to poor IP or lack of repre- IgG levels in the ZIKV-CON and ZIKV-GBS groups fell sentation in the peptide database used to compile our list of within the same range (1.3–7.9 mg/mL), with the exception candidate antigens. An alternative approach is offered by high- of a single outlier (Cali-021) that did not receive a positive throughput antibody screening techniques such as the newly score on the coculture screen, and there was no correlation described Rapid Extracellular Antigen Profiling technique,38,39 between overall IgG levels and cell culture positivity. Al- or chip-based proteome profiling,40 which has recently offered though this finding supports the selectivity of detection of insights into the autoantibody repertoire of patients with nerve reactive positivity in the live cell culture screen despite chronic inflammatory demyelinating polyradiculoneuropathy.41 IVIg treatment, we cannot exclude a loss in sensitivity of detection leading to false negatives in post-IVIg samples with Our findings mirror our recent screening of a separate cohort originally low level reactivity.35 of 100 patients with GBS, where we report a similarly het- erogeneous reactivity of their IgG and IgM antibodies to Only 1 patient’s serum sample was identified as strongly im- nerve-related antigens.20 The narrow definition of the present munoreactive in both myelinating cocultures and DRG/SC GBS cohort, in which all patients associated with the same assays. The low concordance suggests key antigenic differ- infectious insult, suggests that the heterogeneity of the hu- ences between the acute primary culture and long-term moral immune responses in GBS may be a core feature of the myelinating culture assays. Efforts to reduce the background disease and not necessarily due to the heterogeneity of the immunofluorescence signal, as well as further dilution series patient cohorts or prodromal infections. of serum samples, may improve the fidelity of the live culture systems to parse GBS from non-GBS nerve immunoreactivity. Acknowledgment The authors thank Professors Peter Brophy and Sarosh Irani One patient (ZIKV-CON patient B), with an uncomplicated for kindly providing the anti-periaxin antibody and healthy ZIKV infection, displayed striking IgG antibody reactivity human control sera, respectively. The authors also thank against the abaxonal membrane of myelinating SCs that was Professor Robert MacLaren, Dr. Tina Storm, Dr. Ahmed particularly enriched at the nodal villi. These IgG1 subclass Salman, and Dr. Adan Pinto Fernandez for their generous autoantibodies induced profound demyelination in our in assistance with the IncuCyte time-lapse imaging. The authors vitro model system via a complement-dependent pathway, acknowledge the Department of Medicine at the Universitat despite the patient lacking neurologic symptoms. This ap- Autònoma de Barcelona. C. Lleixá and L. Querol are members parent functional discrepancy might be accounted for by a of the European Reference Network for Neuromuscular number of factors: the use of rodent SCs in the coculture Diseases–Project ID No. 870177. pCSC-SP-PW-Nep (aka: system and a potential lack of cross-reactivity to human SCs; pBOB-NEP) was a gift from Inder Verma. CD44S pBabe- by the target antigen(s) remaining hidden from circulating puro was a gift from Bob Weinberg. anxA2-GFP was a gift antibodies in situ; or by a functional loss of complement from Volker Gerke and Ursula Rescher. ITGA6-bio-His was a signaling in this patient, although the latter is rare and usually gift from Gavin Wright. pcDNA3.1 Flag YFP hNKCC1 accompanied by a spectrum of autoimmune presentations,36 (NT17) was a gift from Biff Forbush. pBabe neo TGFBR3- which were lacking in this case. HA was a gift from Kevin Janes. The authors thank the anonymous reviewers for their thorough reviews and helpful By using IP-MS in sera-treated myelinating cocultures and DRG suggestions to improve this article for publication. neurons, we revealed a number of candidate antigens related to SC membranes, myelin formation, or neural cell adhesion. Study Funding However, we could not confirm any candidate antigen reactivity This work is supported by the Medical Research Council UK by these patients’ IgG or IgM antibodies by either the ELISA or (MR/P008399/1) (S. Rinaldi); UKRI Future Leaders Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 11 Fellowship (MR/V02552X/1) (A.J. Davies); Fondo de Investigaciones Sanitarias (FIS), Instituto de Salud Carlos III, Appendix (continued) Spain, and FEDER under grant FIS19/01407 and INT20/ Name Location Contribution 00080; Wellcome Trust Grants 092805 and 202789 (S.K. Halstead and H.J. Willison); and by the GBS-CIDP Foun- Cinta Lleixà, PhD Neuromuscular Diseases Drafting/revision of the Unit, Neurology manuscript for content, dation International. C.A. Pardo, L. Osorio, and B. Parra are Department, Hospital de including medical writing part of the Neurovirus Emerging in the Americas Study la Santa Creu i Sant Pau, for content; major role in Universitat Aut ònoma de the acquisition of data; (NEAS) funded by NIH (R01-NS110122). A.K. Falconar, Barcelona; Centro para study concept or design; C.M. Romero-Vivas, L. Osorio, B. Parra, C.A. Pardo, D.S. la Investigaci ón and analysis or Biom édica en red en interpretation of data Gourlay, S.K. Halstead, and H.J. Willison were also members Enfermedades of a European Union’s Horizon 2020 Research and In- Raras–(CIBERER) Madrid novation Program ZikaPLAN grant agreement 734584 under which many of the patients with ZIKV-GBS, ZIKV-OND, Ana M. Siles, PhD Neuromuscular Diseases Major role in the Unit, Neurology acquisition of data and ZIKV-CON, and DENV-CON were clinically diagnosed and Department, Hospital de la analysis or interpretation their serum samples were collected and initially assessed be- Santa Creu i Sant Pau, of data. Universitat Autònoma de fore being transferred to the United Kingdom for inclusion in Barcelona this study. Dawn S. Gourlay, Institute of Infection, Major role in the BSc Immunity & Inflammation, acquisition of data and Disclosure University of Glasgow, analysis or interpretation S. Rinaldi has received (1) a speaker’s honoraria from Fre- University Place of data senius, Alnylam, and Excemed and payments to provide ex- Georgina Target Discovery Institute, Major role in the pert medicolegal advice on inflammatory neuropathies and Berridge, BSc NDM Research Building, acquisition of data and University of Oxford, Old analysis or interpretation their treatment, (2) a complimentary registration and prize Road Campus of data money from the Peripheral Nerve Society, and (3) a travel Wanwisa Wellcome Centre for Major role in the bursary from the European School of Neuroimmunology and Dejnirattisai, Human Genetics, Nuffield acquisition of data and is a member of the GBS and Associated Inflammatory Neu- PhD Department of Medicine, analysis or interpretation University of Oxford; of data ropathies (GAIN) patient charity medical advisory board. He Division of Emerging also runs a not-for-profit nodal/paranodal antibody testing Infectious Disease (W.D.), Research Department, service at the Nuffield Department of Clinical Neurosciences, Faculty of Medicine Siriraj John Radcliffe Hospital, Oxford, United Kingdom, in part- Hospital, Mahidol nership with Clinical Laboratory Immunology of Oxford University, Bangkok, Thailand University Hospitals. A.J. Davies, C. Lleixà, A.M. Siles, D.S. Gourlay, G. Berridge, W. Dejnirattisai, C. Ramı́rez-Santana, Carolina Center for Autoimmune Study concept or design Ramı́rez- Diseases Research (CREA), J.M. Anaya, A.K. Falconar, C.M. Romero-Vivas, L. Osorio, B. Santana, PhD Universidad del Rosario, Parra, G.R. Screaton, J. Mongkolsapaa, R. Fischer, C.A. Pardo, Bogotá, Colombia S.K. Halstead, H.J. Willison, and L. Querol report no conflicts Juan-Manuel Center for Autoimmune Study concept or design; of interest. Go to Neurology.org/NN for full disclosure. Anaya, MD, PhD Diseases Research (CREA), additional contributions: Universidad del Rosario, provision of samples for Bogotá, Colombia; study Publication History LifeFactors, Rionegro, Colombia Previously published at medRxiv doi: 10.1101/2021.10.28.21265167. Received by Neurology: Neuroimmunology & Neuroinflammation Andrew K. Departamento de Drafting/revision of the December 5, 2022. Accepted in final form September 1, 2022. Falconar, PhD Medicina, Universidad del manuscript for content, Norte, Barranquilla, including medical writing for Submitted and externally peer reviewed. The handling editor was Colombia content; major role in the Marinos C. Dalakas, MD, FAAN. acquisition of data; study concept or design; analysis or interpretation of data; additional contributions: provision of samples for Appendix Authors study Name Location Contribution Claudia M. Departamento de Study concept or design Romero-Vivas, Medicina, Universidad del Alexander J. Nuffield Department of Drafting/revision of the PhD Norte, Barranquilla, Davies, PhD Clinical Neurosciences, manuscript for content, Colombia University of Oxford, John including medical writing Radcliffe Hospital for content; major role in Lyda Osorio, MD, Grupo de Epidemiologı́a y Study concept or design the acquisition of data; PhD Salud Poblacional (GESP), study concept or design; School of Public Health, and analysis or Universidad del Valle, Cali, interpretation of data Colombia 12 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN 5. Moore SM, Oidtman RJ, Soda KJ, et al. Leveraging multiple data types to estimate the size of the Zika epidemic in the Americas. PLoS Negl Trop Dis. 2020;14(9):e0008640. Appendix (continued) doi: 10.1371/journal.pntd.0008640. 6. Uncini A, Shahrizaila N, Kuwabara S. Zika virus infection and Guillain-Barre Name Location Contribution syndrome: a review focused on clinical and electrophysiological subtypes. J Neurol Neurosurg Psychiatry. 2017;88(3):266-271. doi: 10.1136/jnnp-2016- Beatriz Parra, Department of Study concept or design 314310. PhD Microbiology, School of 7. Uncini A, Gonzalez-Bravo DC, Acosta-Ampudia YY, et al. Clinical and nerve Basic Sciences, Universidad conduction features in Guillain-Barre syndrome associated with Zika virus in- del Valle, Cali, Colombia fection in Cucuta, Colombia. Eur J Neurol. 2018;25(4):644-650. doi: 10.1111/ ene.13552. Gavin R. Wellcome Centre for Analysis or interpretation 8. Munoz LS, Parra B, Pardo CA, Neuroviruses Emerging in the Americas Study. Screaton, DPhil, Human Genetics, Nuffield of data; additional Neurological implications of Zika virus infection in adults. J Infect Dis. 2017;216(suppl FMedSci Department of Medicine, contributions: provision of 10):S897-S905. doi: 10.1093/infdis/jix511. University of Oxford samples for study 9. Koma T, Veljkovic V, Anderson DE, et al. Zika virus infection elicits auto-antibodies to C1q. Sci Rep. 2018;8(1):1882. doi: 10.1038/s41598-018-20185-8. Juthathip Wellcome Centre for Human Major role in the 10. Lynch RM, Mantus G, Encinales L, et al. Augmented Zika and Dengue neutralizing Mongkolsapaya, Genetics, Nuffield acquisition of data antibodies are associated with Guillain-Barre syndrome. J Infect Dis. 2019;219(1): DPhil Department of Medicine, 26-30. doi: 10.1093/infdis/jiy466. University of Oxford; Dengue 11. Rivera-Correa J, de Siqueira IC, Mota S, et al. Anti-ganglioside antibodies in patients Hemorrhagic Fever Research with Zika virus infection-associated Guillain-Barre Syndrome in Brazil. PLoS Negl Unit, Office for Research and Trop Dis. 2019;13(9):e0007695. doi: 10.1371/journal.pntd.0007695. Development, Siriraj 12. Medina MT, England JD, Lorenzana I, et al. Zika virus associated with sensory Hospital, Faculty of Medicine, polyneuropathy. J Neurol Sci. 2016;369:271-272. doi: 10.1016/j.jns.2016.08.044. Mahidol Univeristy, Bangkok, 13. Oh Y, Zhang F, Wang Y, et al. Zika virus directly infects peripheral neurons Thailand and induces cell death. Nat Neurosci. 2017;20(9):1209-1212. doi: 10.1038/ nn.4612. Roman Fischer, Target Discovery Institute, Analysis or interpretation 14. Volpi VG, Pagani I, Ghezzi S, Iannacone M, D’Antonio M, Vicenzi E. Zika virus PhD NDM Research Building, of data replication in dorsal root ganglia explants from interferon Receptor1 knockout mice University of Oxford, Old causes myelin degeneration. Sci Rep. 2018;8(1):10166. doi: 10.1038/s41598-018- Road Campus 28257-5. 15. Nascimento OJM, da Silva IRF. Guillain-Barre syndrome and Zika virus outbreaks. Carlos A. Pardo, Department of Neurology, Study concept or design; Curr Opin Neurol. 2017;30(5):500-507. doi: 10.1097/WCO.0000000000000471. MD Johns Hopkins University additional contributions: 16. Culshaw A, Mongkolsapaya J, Screaton G. The immunology of Zika virus. F1000Res. School of Medicine, provision of samples for 2018;7:203. doi: 10.12688/f1000research.12271.1. Baltimore study 17. Davies AJ, Fehmi J, Senel M, Tumani H, Dorst J, Rinaldi S. Immunoadsorption and plasma exchange in seropositive and seronegative immune-mediated neuropathies. Susan K. Institute of Infection, Major role in the J Clin Med. 2020;9(7):2025. doi: 10.3390/jcm9072025. Halstead, PhD Immunity & Inflammation, acquisition of data and 18. Rinaldi S, Davies A, Fehmi J, et al, Australian and New Zealand MOG Study Group. University of Glasgow, analysis or interpretation Overlapping central and peripheral nervous system syndromes in MOG antibody- University Place of data associated disorders. Neurol Neuroimmunol Neuroinflamm. 2021;8(1):e924. doi: 10.1212/NXI.0000000000000924. Hugh J. Willison, Institute of Infection, Study concept or design 19. Ramanathan S, Tseng M, Davies AJ, et al. Leucine-rich Glioma-inactivated 1 versus MD, PhD Immunity & Inflammation, and analysis or contactin-associated protein-like 2 antibody neuropathic pain: clinical and biological University of Glasgow, interpretation of data comparisons. Ann Neurol. 2021;90(4):683-690. doi: 10.1002/ana.26189. University Place 20. Lleixa C, Martin-Aguilar L, Pascual-Goni E, et al. Autoantibody screening in Guillain- Barre syndrome. J Neuroinflammation. 2021;18(1):251. doi: 10.1186/s12974-021- Luis Querol, MD, Neuromuscular Diseases Drafting/revision of the 02301-0. PhD Unit, Neurology manuscript for content, 21. Anaya JM, Rodriguez Y, Monsalve DM, et al. A comprehensive analysis and immu- Department, Hospital de la including medical writing nobiology of autoimmune neurological syndromes during the Zika virus outbreak in Santa Creu i Sant Pau, for content; major role in Cucuta, Colombia. J Autoimmun. 2017;77:123-138. doi: 10.1016/j.jaut.2016.12.007. Universitat Autònoma de the acquisition of data; 22. Gallo V, Egger M, McCormack V, et al. STrengthening the reporting of OBservational Barcelona; Centro para la study concept or design; studies in epidemiology–molecular epidemiology (STROBE-ME): an extension of Investigación Biomédica and analysis or the STROBE statement. Eur J Clin Invest. 2012;42(1):1-16. doi: 10.1111/j.1365- en red en Enfermedades interpretation of data 2362.2011.02561.x. Raras–(CIBERER) Madrid 23. Siles AM, Martinez-Hernandez E, Araque J, et al. Antibodies against cell adhesion molecules and neural structures in paraneoplastic neuropathies. Ann Clin Transl Simon Rinaldi, Nuffield Department of Drafting/revision of the Neurol. 2018;5(5):559-569. doi: 10.1002/acn3.554. MBChB, PhD Clinical Neurosciences, manuscript for content, 24. Clark AJ, Kaller MS, Galino J, Willison HJ, Rinaldi S, Bennett DLH. Co-cultures with University of Oxford, John including medical writing stem cell-derived human sensory neurons reveal regulators of peripheral myelination. Radcliffe Hospital for content; major role in Brain. 2017;140(4):898-913. doi: 10.1093/brain/awx012. the acquisition of data; 25. Halstead SK, Kalna G, Islam MB, et al. Microarray screening of Guillain-Barre syn- study concept or design; drome sera for antibodies to glycolipid complexes. Neurol Neuroimmunol Neuro- and analysis or inflamm. 2016;3(6):e284. doi: 10.1212/NXI.0000000000000284. interpretation of data 26. Willison HJ, Veitch J, Swan AV, et al. Inter-laboratory validation of an ELISA for the determination of serum anti-ganglioside antibodies. Eur J Neurol. 1999;6(1):71-77. doi: 10.1046/j.1468-1331.1999.610071.x. 27. Dejnirattisai W, Supasa P, Wongwiwat W, et al. Dengue virus sero-cross-reactivity drives antibody-dependent enhancement of infection with Zika virus. Nat Immunol. References 2016;17(9):1102-1108. doi: 10.1038/ni.3515. 1. Cao-Lormeau VM, Blake A, Mons S, et al. Guillain-Barre Syndrome outbreak asso- 28. Fehmi J, Davies AJ, Antonelou M, et al. Contactin-1 antibodies link autoimmune neu- ciated with Zika virus infection in French Polynesia: a case-control study. Lancet. ropathies to nephrotic syndrome. SSRN. 2020. ssrn.com/abstract=3739819http:// 2016;387(10027):1531-1539. doi: 10.1016/S0140-6736(16)00562-6. dx.doi.org/10.2139/ssrn.3739819. 2. Dos Santos T, Rodriguez A, Almiron M, et al. Zika virus and the Guillain-Barre 29. Svennerholm L, Fredman P. A procedure for the quantitative isolation of brain syndrome–case series from seven countries. N Engl J Med. 2016;375(16):1598-1601. gangliosides. Biochim Biophys Acta. 1980;617(1):97-109. doi: 10.1016/0005- doi: 10.1056/NEJMc1609015. 2760(80)90227-1. 3. Parra B, Lizarazo J, Jimenez-Arango JA, et al. Guillain-barre syndrome associated with 30. Leonhard SE, Halstead S, Lant SB, et al. Guillain-Barre syndrome during the Zika Zika virus infection in Colombia. N Engl J Med. 2016;375(16):1513-1523. doi: virus outbreak in Northeast Brazil: an observational cohort study. J Neurol Sci. 2021; 10.1056/NEJMoa1605564. 420:117272. doi: 10.1016/j.jns.2020.117272. 4. Keegan LT, Lessler J, Johansson MA. Quantifying Zika: advancing the epidemiology 31. Lucchese G, Kanduc D. Zika virus and autoimmunity: from microcephaly to Guillain- of Zika with quantitative models. J Infect Dis. 2017;216(suppl 10):S884-S890. doi: Barre syndrome, and beyond. Autoimmun Rev. 2016;15(8):801-808. doi: 10.1016/ 10.1093/infdis/jix437. j.autrev.2016.03.020. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 13 32. Sanchez OA, Portillo KM, Reyes-Garcia SZ, England JD, Medina MT. Characterization 37. Querol L, Siles AM, Alba-Rovira R, et al. Antibodies against peripheral nerve antigens of adult patients with Guillain-Barre syndrome during the arboviral infection outbreaks in in chronic inflammatory demyelinating polyradiculoneuropathy. Sci Rep. 2017;7(1): Honduras. J Neurol Sci. 2021;427:117551. doi: 10.1016/j.jns.2021.117551. 14411. doi: 10.1038/s41598-017-14853-4. 33. Fehmi J, Davies AJ, Walters J, et al. IgG1 pan-neurofascin antibodies identify a severe 38. Freeth J, Soden J. New advances in cell microarray technology to expand applications yet treatable neuropathy with a high mortality. J Neurol Neurosurg Psychiatry. 2021; in target deconvolution and off-target screening. SLAS Discov. 2020;25(2):223-230. 92(10):1089-1095. doi: 10.1136/jnnp-2021-326343. doi: 10.1177/2472555219897567. 34. Sander S, Ouvrier RA, McLeod JG, Nicholson GA, Pollard JD. Clinical syndromes 39. Wang EY, Dai Y, Rosen CE, et al. REAP: a platform to identify autoantibodies that associated with tomacula or myelin swellings in sural nerve biopsies. J Neurol Neu- target the human exoproteome. bioRxiv. 2021. doi: 10.1101/2021.02.11.430703. rosurg Psychiatry. 2000;68(4):483-488. doi: 10.1136/jnnp.68.4.483. 40. Jeong JS, Jiang L, Albino E, et al. Rapid identification of monospecific monoclonal 35. Gruter T, Ott A, Meyer W, et al. Effects of IVIg treatment on autoantibody testing in antibodies using a human proteome microarray. Mol Cell Proteomics. 2012;11(6): neurological patients: marked reduction in sensitivity but reliable specificity. J Neurol. O111.016253. doi: 10.1074/mcp.O111.016253. 2020;267(3):715-720. doi: 10.1007/s00415-019-09614-4. 41. Moritz CP, Tholance Y, Stoevesandt O, Ferraud K, Camdessanche JP, Antoine JC. 36. Vassallo G, Newton RW, Chieng SE, Haeney MR, Shabani A, Arkwright PD. Clinical CIDP antibodies target junction proteins and identify patient subgroups: an auto- variability and characteristic autoantibody profile in primary C1q complement deficiency. antigenomic approach. Neurol Neuroimmunol Neuroinflamm. 2021;8(2):e944. doi: Rheumatology (Oxford). 2007;46(10):1612-1614. doi: 10.1093/rheumatology/kem207. 10.1212/NXI.0000000000000944. 14 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN