medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Anti–GM-CSF autoantibodies promote a “pre-diseased” state in Crohn’s Disease.

2

3 Authors:

4 Arthur Mortha (1,2,4,18)*, Romain Remark (1,5), Diane Marie Del Valle (1,2), Ling-

5 Shiang Chuang (3,6), Zhi Chai (3,6), Inês Alves (15,16), Catarina Azevedo (15), Joana

6 Gaifem (15), Jerome Martin (13, 14), Kevin Tuballes (1,2), Vanessa Barcessat (1,2),

7 Siu Ling Tai (4), Hsin-Hui Huang (2), Ilaria Laface (1,2), Yeray Arteaga Jerez (1), Gilles

8 Boschetti (1,7), Nicole Villaverde (6), Mona D. Wang (4), Ujunwa M. Korie (3,6),

9 Joseph Murray (12), Rok-Seon Choung (12), Takahiro Sato (10), Renee M. Laird (11),

10 Scot Plevy (10), Adeeb Rahman (1,3), Joana Torres (8,9), Chad Porter (11), Mark S.

11 Riddle (11), Ephraim Kenigsberg (3,6), Salomé S. Pinho (15, 16, 17), Judy H. Cho

12 (6,18), Miriam Merad (1,2,3,18)*, Jean-Frederic Colombel (1,8,18)*, Sacha Gnjatic

13 (1,2,3,18)*

14

15

16

17 NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

1 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

18 Institutions:

19 1. Precision Immunology Institute, Icahn School of Medicine at Mount Sinai, New

20 York, NY, USA

21 2. Tisch Cancer Institute, Division of Hematology/Oncology, Department of Medicine,

22 Icahn School of Medicine at Mount Sinai, New York, NY, USA

23 3. Human Immune Monitoring Center at Mount Sinai, New York, NY, USA

24 4. Department of Immunology, University of Toronto, Toronto, CA

25 5. Innate Pharma, Marseille, FR

26 6. Charles Bronfman Institute for Personalized Medicine, Department of Genetics,

27 Icahn School of Medicine at Mount Sinai, New York, NY, USA

28 7. Hépato-Gastroentérologue, Hospices Civils de Lyon, Université Claude Bernard

29 Lyon, FR

30 8. Department of Gastroenterology, Icahn School of Medicine at Mount Sinai, New

31 York, NY, USA

32 9. Gastroenterology Division, Hospital Beatriz Ângelo, Loures, PT

33 10. Janssen R&D, Spring House, PA, USA

34 11. Naval Medical Research Center, Silver Spring, MD, USA

2 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

35 12. Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN, USA

36 13. Université de Nantes, Inserm, CHU Nantes, Centre de Recherche en

37 Transplantation et Immunologie, UMR 1064, ITUN, 44000 Nantes, France

38 14. CHU Nantes, Laboratoire d’Immunologie, CIMNA, 44000 Nantes, France

39 15. i3S – Institute for Research and Innovation in Health, University of Porto, 4200-

40 135 Porto, Portugal.

41 16. Faculty of Medicine, University of Porto, 4200-319 Porto, Portugal.

42 17. Institute of Biomedical Sciences Abel Salazar (ICBAS), University of Porto, 4050-

43 313 Porto, Portugal.

44 18. These authors contributed equally

45 * Corresponding authors

46

47

48

49

50

3 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

51 Grant support: A.M. is supported by a CIHR-Project Grant (388337), a NSERC-

52 Discovery Grant (RGPIN-2019-04521). A.M. is the Tier 2 Canadian Research Chair in

53 Mucosal Immunology and supported by the Tier 2 CRC-CIHR program. S.S.P.

54 acknowledges Portuguese funds through the Portuguese Foundation for Science and

55 Technology (FCT) in the framework of the project POCI-01-0145-FEDER-028772 and

56 support by the Broad Medical Research Program at the Crohn’s and Colitis Foundation

57 of America; the International Organization for the study of Inflammatory Bowel

58 Disease, and the Portuguese Group of Study in IBD (GEDII) for funding. S.S.P. also

59 acknowledges the US Department of Defense, US Army Medical Research Acquisition

60 Activity, FY18 Peer Reviewed Medical Research Program Investigator-Initiated

61 Research Award (award number W81XWH1920053). I.A. thanks FCT for funding

62 [SFRH/BD/128874/2017]. S.L.T. acknowledges funding through the Dr. Edward

63 KETCHUM Graduate Student Scholarships at the University of Toronto and the

64 Canada Graduate Scholarships – Master's (CGS-M) award. M.W. is support by a

65 UROP fellowship by the Department of Immunology at the University of Toronto. J.H.C.

66 acknowledges funding from U01 DK062429, U01 DK062422, R01 DK106593, and the

67 Sanford Grossman Charitable Trust. S.G and J.F.C. are supported through U01

4 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

68 DK124165. S.G. is additionally supported by grants U24 CA224319. Funding and

69 support of the PREDICTS study platform was provided through a Cooperative and

70 Research Development Agreement with direct contributions by Janssen

71 Pharmaceuticals, Prometheus Laboratories and the Naval Medical Research Center

72 (NCRADA number NMR-11-3920). The mass cytometry instrumentation at the Human

73 Immune Monitoring Center was obtained with support from S10 OD023547 and P01

74 CA196521.

75

76 Correspondence:

77 Arthur Mortha, Ph.D.

78 University of Toronto, Department of Immunology, Medical Science Building ,1 King’s

79 College Circle, Room 7326, Toronto M5S 1A8, Ontario Canada

80 [email protected]

81 phone +1 437 777 5418

82

83

84

5 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

85 Disclosure:

86 The University of Toronto and the Mount Sinai Hospital (NY) have collectively filed a

87 patent application listing S.G., A.M., J.F.C., M.M. and R.R. as inventors that is related

88 in part to this publication. S.G. reports consultancy and/or advisory roles for Merck,

89 and OncoMed and research funding from Bristol-Myers Squibb, Genentech, Janssen

90 R&D, Pfizer, Takeda, and Regeneron. JFC reports receiving research grants from

91 AbbVie, Janssen Pharmaceuticals and Takeda; receiving payment for lectures from

92 AbbVie, Amgen, Allergan, Inc. Ferring Pharmaceuticals, Shire, and Takeda; receiving

93 consulting fees from AbbVie, Amgen, Arena Pharmaceuticals, Boehringer Ingelheim,

94 BMS, Celgene Corporation, Eli Lilly, Ferring Pharmaceuticals, Galmed Research,

95 Genentech, Glaxo Smith Kline, Janssen Pharmaceuticals, Kaleido Biosciences,

96 Imedex, Immunic, Iterative Scopes, Merck, Microba, Novartis, PBM Capital, Pfizer,

97 Sanofi, Takeda, TiGenix, Vifor; and hold stock options in Intestinal Biotech

98 Development.

99

100

101

6 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

102 Author Contributions:

103 AM, RR, DMDV, LSC, ZC, IA, CA, JG, JM, KT, SLT, IL, AR, YAJ, GB, NV, MW

104 performed experiments, sample processing data acquisition and analysis. JM, RSC,

105 TS, RML, SP, CP, MSR, EK, SSP, JHC, MM, JFC, AM and SG provided guidance,

106 facilitated sample acquisition or coordinated this study. AM and SG wrote the

107 manuscript.

108

109 Data transparency statement:

110 The data that supports the findings of this study are available from the corresponding

111 authors upon reasonable request.

112

7 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

113 Abstract

114 Background & Aims: Anti–GM-CSF autoantibodies (aGMAb) are detected in ileal

115 Crohn’s Disease (CD) patients. Their induction and mode of action impacting

116 homeostasis during, or prior to disease are not well understood. We aimed to

117 investigate the underlying mechanisms leading to the induction of aGMAb, from

118 functional orientation to recognized epitopes, for their impact on intestinal immune

119 homeostasis and use as predictive biomarker for complicated CD.

120 Methods: Using longitudinally collected sera from active component US personnel, we

121 characterize naturally occurring aGMAb in a subset of CD patients years before

122 disease onset. We employed biochemical, cellular, and transcriptional analysis to

123 uncover a mechanism that governs the impaired immune balance in CD years prior to

124 diagnosis.

125 Results: Neutralizing aGMAb are specific to posttranslational glycosylations on GM-

126 CSF, detectable years prior to diagnosis, and associated with complicated CD at

127 presentation. Glycosylation and production of GM-CSF change in CD patients, altering

128 myeloid homeostasis and destabilizing group 3 innate lymphoid cells. Perturbations in

8 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

129 immune homeostasis precede the inflammation and are detectable in the non-inflamed

130 CD mucosa of patients presenting with anti-GM-CSF autoantibodies.

131 Conclusions: Anti-GM-CSF autoantibodies predict the diagnosis of complicated CD,

132 have unique epitopes, and impair myeloid cell homeostasis across the ILC3-GM-CSF-

133 myeloid cell axis, altering intestinal immune homeostasis long before the diagnosis of

134 disease.

135 Keywords: Crohn’s Disease, innate lymphoid cells, macrophages, GM-CSF,

136 autoantibodies

137

9 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

138 Introduction:

139 Inflammatory bowel disease (IBD), sub-classified into Crohn’s disease (CD) and

140 ulcerative colitis (UC), is an increasingly diagnosed chronic inflammatory pathology of

141 the gastrointestinal tract, affecting 0.3% of the world’s population1. While both CD and

142 UC affect the gastrointestinal tract, their causes remains puzzling and are likely

143 multifactorial2. Genome-wide association studies (GWAS) provide strong support that

144 IBD is a pathology driven by mono and multigenetic variations, but non-genetic and

145 environmental factors are also considered as contributors to the heterogeneity of this

146 disease3,3, 4. Identifying factors contributing to IBD development that may predict

147 disease onset is therefore of high clinical relevance.

148 A key immunologic characteristic of IBD is the break in intestinal homeostasis,

149 commonly manifested through insufficient barrier integrity, decreased immunologic

150 tolerance, or excessive inflammation. The cytokine Granulocyte Macrophage-Colony

151 Stimulating Factor (GM-CSF) reportedly plays a dual role in intestinal inflammation and

152 was shown to have both protective and inflammatory properties in CD5-9. Group 3

153 innate lymphoid cells (ILC3) produce GM-CSF in the gut, which acts on myeloid

154 immune cells to sustain immune homeostasis10. Derived from ILC3, GM-CSF

10 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

155 promotes anti-bacterial and immunomodulatory functions in myeloid cells, to sustain

156 the transcriptional stability of ILC3 via the metabolite retinoic acid (RA), further

157 contributing to the generation and maintenance of immunosuppressive regulatory T

158 cells (Treg)10,11. However, in CD patients, myeloid cells induce differentiation of ILC3

159 into inflammatory group 1 ILC (ILC1) via Interleukin (IL)-12 and IL-2312-14. ILC3 or GM-

160 CSF deficiency impairs anti-microbial immunity and mucosal homeostasis aligning with

161 the identification of mutations in GM-CSF signaling in CD patients6, 10,15. Administration

162 of yeast-produced GM-CSF (sargramostim) improved CD and ameliorated symptoms

163 5, 16. However, larger trials with sargramostim in CD failed to reach statistical

164 significance, possibly due to high placebo group responses and suboptimal study

165 design17-19.

166 Besides genetic defects in GM-CSF signaling, aGMAb may also contribute to CD in a

167 subset of patients, and associate with ileal involvement, disease severity, higher

168 relapse rates and complications20-24. Such autoantibodies can cause pulmonary

169 alveolar proteinosis (PAP), resulting in alveolar macrophage-deficiency and increased

170 lung pathologies25. Why PAP patients do not display intestinal pathologies and why

171 CD patients with aGMAb do not show PAP-associated pulmonary symptoms remains

11 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

172 unknown. Moreover, both the overproduction and absence of GM-CSF significantly

173 increase the susceptibility to develop IBD, emphasizing the heterogeneity in CD

174 pathogenesis, prompting us to investigate the biology of aGMAb prior to diagnosis in

175 CD 5-9, 26.

176 Here, we analyzed >1800 sera, collected longitudinally from active component US

177 military personnel over 10-years including subjects who eventually developed CD

178 (n=220), UC (n=200), and personnel without CD or UC (HD, n=220) 27, along with

179 controls with established CD or PAP. We analyzed titers, isotype profiles, epitopes,

180 and functional consequences of aGMAb, using ELISA, neutralization, and reporter

181 assays, and their effect on immune subsets in non-inflamed ileal CD biopsies by RNA

182 sequencing, to establish mechanisms of pathology prior to disease and at its onset

183 that could be exploited for personalized CD therapies.

184 Methods

185 Human specimen

186 Non-involved intestinal resection and involved intestinal resection samples were

187 obtained from patients with CD undergoing ileal resection surgery at the Mount Sinai

188 Medical Center (New York, NY) after obtaining informed consent. All protocols were

12 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

189 reviewed and approved by the Institutional Review Board (IRB) at the Icahn School of

190 Medicine at Mount Sinai (IRB 08-1236 and IRB HSM 13-00998). Stored pre-diagnosis

191 serum samples were obtained from the Department of Defense Serum Repository,

192 Silver Spring, MD, USA. 220 CD, 200 UC, and 200 healthy controls (HC) samples were

193 provided, sampled at two to three time points prior to diagnosis of disease and one

194 time point post diagnosis of disease27. Blood samples from six adult CD patients were

195 collected at Porto University Centre Hospital, Portugal (CHUP). PAP control sera were

196 kindly provided by Dr. Lanzavecchia (Institute for Research in Biomedicine,

197 Switzerland). All patients were diagnosed with active CD and were naïve for treatment

198 with biologics at the time of blood collection. Blood samples from healthy donors were

199 used as controls. All participants gave informed consent about all clinical procedures

200 and research protocols were approved by the ethical committee of both hospitals.

201 Serum was collected from peripheral blood by centrifugation at 1620 x g for 10 minutes

202 and stored at -80ºC until analysis.

203 Sample collection: Resection samples and biopsies were cleared by a pathologist and

204 placed in complete RPMI media on ice prior to processing. Patients diagnosed with

205 UC or malignancies were excluded from the analysis.

13 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

206 Isolation of PBMC: Buffy coats obtained from the New York City Blood Center and

207 PBMCs collected via a Ficoll gradient. PBMCs were washed, counted and

208 resuspended in complete RPMI followed by consecutive experiments or further used

209 for the enrichment of monocytes using magnetic anti-CD14-beads (purity of monocytes

210 > 98%).

211 Isolation of lamina propria leukocytes: freshly removed ileal resections were washed

212 in ice-cold PBS and scraped off of mucus. Epithelial cells were removed in HBSS free

213 of calcium and magnesium, supplemented with EDTA (5mM) and HEPES (10mM) for

214 20-30 minutes at 37°C at 100rpm. Fibrotic tissue was removed and non-fibrotic

215 mucosa minced and digested using Collagenase IV and DNase I (both Sigma) in HBSS

216 (Ca2+Mg2+) 2% FBS for 30 minutes at 37°C at 100rpm agitation. Cell suspensions were

217 filtered and live cells enriched using Percoll (80%:40%). The interphase was

218 harvested, washed in FACS buffer (PBS, 5mM EDTA and 2% FBS) for subsequent

219 use.

220 Flow Cytometry and Mass Cytometry

14 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

221 Aldefluor staining: freshly isolated PBMCs or lamina propria leukocytes were washed

222 twice in PBS and stained following the manufacturer protocol. Following Aldefluor

223 staining, cells were washed in PBS and processed for surface staining.

224 Surface staining: Fc-receptors were blocked using Fc-block reagent (BD), following a

225 20 min surface staining with directly conjugated monoclonal antibodies. The following

226 antibodies were used and purchased from BD, R+D, Biolegend and Miltenyi: anti-

227 human CD45, anti-human HLA-DR, anti-human CD11c, anti-human CD14, anti-human

228 CD1c, anti-human CD141, anti-human CD127, anti-human CD117, anti-NKp44, anti-

229 CD161, anti-CD3, anti-CD19, anti-RORgt, anti-CD69. Following surface staining, cells

230 were fix using the FOXP3 staining kit to stain for RORgt.

231 Intracellular cytokine staining: for intracellular cytokine staining of GM-CSF and IFN-g,

232 cells were re-suspended in complete RPMI supplemented with 10% FBS, 1% non-

233 essential amino acids, 1% sodium-pyruvate, 1% L-glutamine, 1% penicillin-

234 streptomycin and 1% HEPES. Media was further supplemented with Brefeldin A (for

235 GM-CSF) or Brefeldin A, PMA (Sigma) and Ionomycin (Sigma) (for IFN-g). Cells were

236 incubated for 4h at 37°C and 5% CO2 prior to surface staining. Post surface staining,

237 cells were fixed for 30minutes in Cytofix/CytoPerm (BD), washed in PBS containing

15 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

238 2% FBS, 5mM EDTA (FACS buffer) and 0.5% Saponin (Sigma). Anti-cytokine staining

239 was performed for 30minutes at 4°C in the dark. Cells were washed and samples were

240 analyzed on a BD LSR Fortessa II.

241 Phospho STAT5 staining: PBMCs or isolated LPL were resuspended in PBS and

242 incubated with GM-CSF (10ng/ml) STAT5 phosphorylation was assessed using the

243 BD PhosFlow Protocol for human PBMCs using PermBuffer II. Intracellular staining of

244 pSTAT5 was performed using anti-human pSTAT5-Alexa647.

245 Mass Cytometry: Cells were stimulated at 1-3x106/cells per reaction with recombinant

246 GM-CSF (10ng/ml) for 20-30mins. Following stimulation cells were fixed for 30mins

247 following the PhosFlow Protocol for human PBMCs using PermBuffer II (BD). Post

248 fixation, groups of cells were barcoded using Cell-ID™ 20-Plex Pd Barcoding Kit and

249 combinations of different metal isotope conjugated anti-CD45 antibodies. Pooled

250 samples were stained with markers listed in SUPPLEMENTARY FIGURE 2A and

251 analyzed on a Helios CyTOF system. Dimensionality reduction of CyTOF data was

252 performed on the Cytobank software using the viSNE algorithm.

253 Anti-GM-CSF ELISA, serum GM-CSF ELISA and Lectin ELISA

254 For anti–GM-CSF ELISAs, plates were coated with recombinant human GM-CSF

255 (sargramostim), washed and blocked with TBST/BSA. Wells were incubated with 10-

16 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

256 50µl of serum diluted in TBST followed by three washing steps. Anti–GM-CSF

257 antibodies were detected by pan anti-human IgG HRP or isotype specific secondary

258 antibodies. Substrate reaction was assessed using a plate reader at 550nm. Detection

259 of lectin-binding motifs in the serum GM-CSF from CD patients and HD was performed

260 by a lectin ELISA adapted from 76. Prior to serum capture, serum from CD patients and

261 HD was concentrated using Amicon® Ultra-2 mL Centrifugal Filters, to reach a final

262 concentration of 4X. Microtiter plates (Maxisorp, Nunc) were coated with a mouse anti-

263 human GM-CSF (DuoSet R&D Systems) in PBS buffer, overnight at room temperature

264 (RT). Blocking was performed with Carbo-free blocking solution (Vector Labs) for 1h

265 at RT. Plates were washed 5 times with PBS + 0.05% Tween 20 (PBST) before the

266 CD/HD plasma samples in PBS containing 1% Carbo-free blocking solution (diluent

267 solution) were added and incubated shaking at 200 rpm for 2h at RT. After washing

268 the wells as described above, biotinylated mouse anti-human GM-CSF (DuoSet R&D

269 Systems) was added and incubated for 2h shaking at 200 rpm at RT. Bound GM-CSF

270 was detected using an HRP-conjugated streptavidin (DuoSet R&D Systems) incubated

271 for 20 minutes and Tetramethylbenzidine substrate (DuoSet R&D Systems) was

272 incubated for 20 minutes protected from dark. Reaction was stopped using H2SO4 and

273 the amount of bound lectin was measured at 450 nm using a μQuant Microplate

274 Reader (BioTek, Agilent).

275 Antibody-binding assay: GM-CSF was heat-denaturated in SDS containing buffer and

276 used for anti–GM-CSF ELISAs. Binding-strength of anti–GM-CSF antibodies in patient

277 sera was assessed using wash step with NaCl salt solutions of increasing

278 concentrations (1-4M NaCl). Binding capacity was calculated as % of maximal binding.

17 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

279 Stripping of glycosylation on GM-CSF: GM-CSF was stripped using N-Glycosidase F,

280 α2-3,6,8,9-Neuraminidase, Endo-α-N-acetylgalactosaminidase, β1,4-galactosidase

281 and β-N-Acetylglucosaminidase (SIGMA) and according to the manufacturer’s

282 recommendations.

283 Native PAGE, SDS-PAGE/Western Blot, Lectin Blot: Recombinant and stripped GM-

284 CSF (sargramostim) (8µg/well) were separated on 15% resolving native

285 polyacrylamide gels. Gels were stained with Coomassie G Brilliant Blue to confirm

286 stripping. Proteins were then transferred to nitrocellulose membranes and membranes

287 were blocked with 5% Non-Fat Dry Milk in Tris-Buffer Saline 0.1% Tween-20 (TBST)

288 at 4°C. Membranes were then incubated with serum samples (diluted 1:100 in blocking

289 buffer). Bound anti–GM-CSF antibodies were detected using anti-Human IgG AP at

290 1:1000 in TBST. Images of western blots have been cropped in Figure 1C and

291 supplementary Figure 2B and 2C.

292 Lectin Blot and Western Blot :The glycoprofile of recombinant forms of GM-CSF was

293 evaluated by lectin blot. 2 μg of purified GM-CSF were separated as above and

294 membranes blocked with BSA 4%. Phaseolus Vulgaris Leucoagglutinin (L-PHA),

295 Maackia Amurensis Lectin II (MAL-II), Galanthus Nivalis Lectin (GNA) and Aleuria

296 Aurantia Lectin (AAL) (each 2ug/mL) were used and bands visualized using the

297 Vectorstain Elite ABC kit and ECL reagent. For western blot analysis 5μg of purified

298 GM-CSF separated as above and detected using biotinylated GM-CSF antibody

299 (0,1μg/mL R&D Systems). Target bands were visualized as above.

300

301 Cloning, mutagenesis, expression and purification of GM-CSF

18 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

302 A double-stranded, C-terminal polyhistidine-tag (His-tag) containing GM-CSF DNA

303 fragment was cloned into pIRESpuro2. Mutagenesis of putative glycosylation sites was

304 performed using the Q5® Site-Directed Mutagenesis Kit and constructs of wild type

305 and mutated GM-CSF stabely transfected into HEK293 cells (a kind gift by Dr. Goetz

306 Erhardt). Cytokine secretion was validated using intracellular cytokine staining and

307 ELISA as described above. Recombinant GM-CSF variants were purified using Ni-

308 columns followed by western blot or assessment of bioactivity on U937 cells (a kind

309 gift by Dr. Goetz Erhardt, University of Toronto).

310

311 ILC signatures

312 ILC1: TBX21, IL12RB2, IL12RB1, IL2RB, GZMB, IFNG, FASLG, IL2RA, IRF8, GZMA,

313 IKZF3, EOMES, ZNF683, GZMM

314 ILC3: TNFSF13B, AHR, CCR6, CCL20, IL1R1, ID2, TOX2 ,NR1D1, IL22, LIF, RORC,

315 KIT, IL23R, CSF2, IL7R, MAF

316

317 Statistical Analyses: We used several visualization and clustering tools to describe the

318 association of the investigated cells and analytes, e.g., scatter plot, heatmap, and t-

319 distributed Stochastic Neighborhood Embedding (t-SNE). Continuous variables were

320 summarized by disease groups (CD, UC, HD, and PAP) using standard descriptive

321 statistics: median, interquartile range (IQR), mean, and standard deviation (SD) as

322 appropriate. Antibody titers ≥ 100 were categorized as seropositive, and seronegative

323 otherwise. The differences in analytes between the groups were assessed using

324 Mann-Whitney, Student t-test, Kruskal-Wallis, and Analysis of variance (ANOVA) as

19 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

325 appropriate. The differences in longitudinal antibody values between sampling time

326 points were evaluated using paired t-test or Wilcoxon tests. We classified the trajectory

327 groups based on the longitudinal antibody titers at earliest time point vs. follow-up time

328 points: 1. Positive: titers ≥100 at baseline and follow-up time points; 2. Negative: titers

329 <100 at baseline and follow-up time points; and 3. Increase: titers ≥100 and increasing

330 at least four times compared to titers at previous time point(s). We explored the impact

331 of prior antibody titers on development of severe CD using the Kaplan-Meier method

332 with log-rank hazard ratio calculations. Data in the main figures show the combined

333 data from a set of 320 patients (n=960 samples) used as a training cohort and another

334 300 patients (n=1200 samples) obtained and tested independently as a validation

335 cohort (the data for training and validation cohorts is presented separately in

336 supplemental figures). The model determined by the training cohort was then tested in

337 the validation cohort, which performance metrics included sensitivity, specificity, and

338 area under curve (AUC). The correlation between antibodies to GM-CSF and to ASCA

339 or between GM-CSF autoantibody titers and pSTAT5 levels was assessed using

340 Spearman’s correlation coefficient. The primary analyses were two-sided at the

341 significant level of 0.05. We used the Bonferroni method to adjust for the situation of

342 multiple testing.

343

344 Results

345 CD-associated aGMAb are distinct from those in PAP and UC.

20 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

346 To assess the prevalence and characteristics of aGMAb in IBD, serum IgG titers

347 against sargramostim, a yeast-produced recombinant GM-CSF, were tested in a

348 cohort of patients with active CD (n=81) or UC (n=37), healthy donors (HD, n=43) and

349 PAP patients (n=12). Among IBD patients, 40% of CD and 14% of UC patients

350 displayed detectable levels of IgG aGMAb (titers >1/100) (FIG.1A). Titers detected in

351 CD but not UC patients were significantly higher compared to HD, but lower when

352 compared to PAP patients (FIG.1A). Only 3/53 sera from IBD patients with aGMAb

353 reacted with other known autoantigens, emphasizing that the vast majority of aGMAb

354 responses were antigen-specific (FIG.S1A). Sera showing broad non-specific antigen

355 reactivity were excluded. Even though aGMAb titers differed between PAP and IBD

356 patients, adjusted titers demonstrated comparable, relative avidity to GM-CSF

357 (FIG.S1B).

358 We examined the isotypes and IgG subclasses of aGMAb in IBD vs. PAP patients.

359 PAP-associated aGMAb were enriched in IgG1 and IgG4, isotypes virtually absent in

360 IBD-associated aGMAb, were enriched in IgG2 and IgA. IgM and IgG3, but not IgE,

361 were detectable in both PAP and IBD, with higher average IgM aGMAb in CD patients

362 (FIG.1B). IBD-associated aGMAb were a specific marker for CD with ileal involvement

21 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

363 and complicated disease, confirming previous studies (FIG.S1C-D and TAB.S1)21-24.

364 Measuring aGMAb thus allows the discrimination of a subgroup of CD patients

365 amongst all IBD patients.

366 Another major difference between IBD- and PAP-associated aGMAb was found in their

367 epitopes. Peptide epitopes in PAP patients have previously been described in GM-

368 CSF, but synthetic overlapping 20-mer linear peptides covering GM-CSF failed to react

369 with CD-associated aGMAb (data not shown)25. Suspecting conformational epitopes,

370 binding of aGMAb to denaturated GM-CSF was virtually absent using CD sera

371 (FIG.S1E). Unlike PAP sera, CD sera did not react with GM-CSF in the absence of

372 posttranslational modifications: only PAP sera bound bacterially produced

373 recombinant GM-CSF, while CD sera required a eukaryotic GM-CSF product in order

374 to react (data not shown). Analysis of yeast-produced GM-CSF by gel electrophoresis

375 revealed three bands at ~19.5 kDa, ~16.5 kDa and ~14.5 kDa (FIG.S1F). The highest

376 band carried posttranslational glycosylations that were lost when enzymatically

377 stripped (FIG.S1F). GM-CSF expressed in HEK293 cells lost this band pattern when

378 all glycosylation sites were mutated to alanine (FIG.S1G). Posttranslational

379 modifications of GM-CSF have previously been suggested as ideal antibody-

22 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

380 recognition sites28. We assessed whether sera from CD patients would react with the

381 different forms of yeast-derived GM-CSF. While PAP sera bound all bands of GM-CSF,

382 CD-associated aGMAb bound larger glycosylated GM-CSF (FIG.1C, S1E-F). When

383 sargramostim was enzymatically stripped of glycans, seroreactivity of CD to the 19.5

384 kDa band was lost but remained to the 16.5 kDa (FIG.1C), while PAP sera also

385 recognized the 14.5 kDa band, which reflected the fully deglycosylated form of GM-

386 CSF ( FIG.S1G)25. Similarly to IgG, IgA aGMAb from CD patients showed high

387 specificity to glycosylation (FIG.1C). We confirmed specificity of CD-associated

388 aGMAb using recombinant human GM-CSF produced in human cells, indicating that

389 reactivity was not exclusive to yeast-specific modifications (data not shown). To

390 evaluate commonalities in the glycan composition, yeast- and mammalian cell-derived

391 GM-CSF, were resolved via SDS-PAGE and the glycoprofile of GM-CSF assessed via

392 lectin blot. To determine commonalities in the glycosylation of GM-CSF, membranes

393 were probed with L-PHA, MALII, GNA and AAL (FIG.S1H). Mammalian cell-derived

394 GM-CSF displays a 25kDa glycoform reacting positive with L-PHA and AAL implicating

395 complex branched and fucosylated N-glycans (FIG.S1I). The 16-17kDa band of

396 mammalian cell-derived GM-CSF reacted with MALII and GNA, revealing high-

23 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

397 mannose N-glycans and hybrid N-glycans with terminal sialylation (FIG.S1I). Yeast-

398 derived GM-CSF reacted only with GNA suggesting high-mannose N-glycans as

399 shared glyco-modification (FIG.S1I). These results demonstrate that mammalian-

400 derived GM-CSF shares high mannose N-glycans with yeast-derived GM-CSF and

401 suggest that CD-associated aGMAb selectively recognize these structures on correctly

402 folded, native GM-CSF. To determine if glycosylations on GM-CSF differed between

403 CD and HD, serum GM-CSF was captured and incubated with lectins. GNA binding

404 was significantly higher on GM-CSF from CD patients, suggesting an increase in high

405 mannose N-glycans on GM-CSF in CD (FIG.1D). While serum GM-CSF levels did not

406 differ in CD and HD, decreased AAL binding indicated a decrease in the core fucose

407 residues in CD (FIG.1D and FIG.S2A). The increase in mannose and decrease in the

408 core-fucose structures suggest a CD-specific glycol-signature on GM-CSF.

409

410 Neutralizing capacity of CD-associated aGMAb

411 The enriched abundance of binding of aGMAb to GM-CSF in CD patients (TAB.S1 and

412 FIG.1A-B) supports the hypothesis that aGMAb possess neutralizing capacities20. We

413 next determined whether aGMAb abrogate GM-CSFR signaling in monocytes,

24 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

414 dendritic cells (DC) and plasmacytoid DC (pDC) of HC. Blood leukocytes stimulated

415 with recombinant human GM-CSF ex vivo for 20min in the presence of PAP-sera (n=9),

416 CD-sera negative for aGMAb (n=20), or CD-sera positive for aGMAb (n=20). Post

417 stimulation, phosphorylation of STAT5 as measured by mass cytometry (TABLE.S2).

418 Reduced STAT5 phosphorylation was observed in the presence of PAP-sera and CD-

419 sera positive for aGMAb (FIG.1E and FIG.S2B-C). Stimulation of with GM-CSF did not

420 result in pSTAT5 in T/B/NK cells across all groups (FIG.S2B). IL-3 stimulation on

421 basophils was likewise unaffected by aGMAb+ CD-sera, suggesting unaltered

422 common beta chain signaling (FIG.S2D). Decreased pSTAT5 correlated with aGMAb

423 titers demonstrating neutralizing activity on circulating precursors of intestinal antigen-

424 presenting cells (APC) (FIG.1F).

425

426 CD-associated aGMAb precede the onset of complicated disease.

427 The presence of neutralizing aGMAb in CD patients suggests that isotypes may

428 contribute to disease pathophysiology and possibly etiology. In order to address this

429 hypothesis, we tested sera obtained from the Department of Defense Serum

430 Repository, prospectively collected from military service members during their annual

25 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

431 routine medical examinations29. Some of these service members were eventually

432 diagnosed with either UC or CD during the course of their service. Three to four

433 longitudinal serum samples spanning up to 10 years, obtained prior, at and post

434 diagnosis from 220 CD, 200 UC, and 200 matched individuals remaining healthy (HD),

435 were analyzed in two independent runs (TAB.S3). Total IgG and IgA anti–GM-CSF

436 ELISAs were performed for each time point of collection. Healthy military service

437 members and service members diagnosed with UC had a similar 5-10% detection rate

438 of anti-GM-CSF IgG, with low mean titers below the limit of significance (between 1/25

439 and 1/50), and nearly no anti–GM-CSF IgA detection (0-1%), without significant

440 change by time point (FIG.2A-B and FIG.S3A-D). In contrast, IgG and IgA aGMAb

441 were already found 6 years prior to CD diagnosis in 21% and 7% of samples, with

442 additional patients seroconverting and mean titer increasing from 1/190 to 1/320

443 towards the time of diagnosis (FIG.2A-C and FIG.S3). At the time of diagnosis, IgA

444 aGMAb were exclusively elevated in 12% of CD, while IgG were significantly more

445 frequent (25%) compared to HD and UC (FIG.2A-B and FIG.S3C). Nearly all CD

446 patients with detectable aGMAb 6 years prior to diagnosis maintained or increased

447 their titers until the time of diagnosis (FIG.2C and FIG.S3E-F). Most (75%) anti–GM-

26 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

448 CSF IgA co-occurred with IgG, while IgG to GM-CSF was more frequent and detected

449 in 63% of cases in the absence of IgA.

450 Anti-GMAb in CD associated with ileal/ileocolonic involvement and complications

451 within 100 days of diagnosis, with a 2.8 risk hazard ratio of having penetrating and/or

452 stricturing disease or surgery at, or soon after clinical diagnosis (FIG.2D and FIG.S4A-

453 D). Presence of IgA up to 6 years prior to diagnosis provided a predictor for CD

454 development with >97% specificity and with sensitivity increasing from 15% to 21% as

455 time to diagnosis decreased (AUC 0.6). The detection of aGMAb did not correlate with

456 date of birth, sex, race, or year of sample acquisition, rendering this biomarker

457 universal across patients (data not shown). Anti-Saccharomyces cerevisiae antibodies

458 (ASCA) a common serological marker for IBD were shown to present prior to diagnosis

459 using similar cohorts3, 30,31. Anti-GMAb preceded the occurrence of anti-ASCA IgA and

460 showed no significant correlations 2000 days prior to diagnosis (FIG.2E and

461 TABLE.S4).

462

463 The CD mucosa shows impaired homeostatic functions in GM-CSF-responsive

464 myeloid cells.

27 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

465 GM-CSF engages the heterodimeric GM-CSFR, composed of the GM-CSF binding

466 alpha chain CD116 (CSF2RA) and the signal transducing common beta chain CD131

467 (CSF2RB) activating the transcription factor STAT5. To understand impairments in

468 GM-CSFR signaling, we determined CD116 and CD131 expression across most

469 hematopoietic cells in the inflamed (INF) and non-inflamed (NI) CD mucosa

470 (TAB.S2B). While CD16+ or CD14+ monocytes, CD141+ DC, CD1c+ DC, pDC and

471 neutrophils differed in their abundance, CD116 and CD131 expression and STAT5

472 phosphorylation remained unchanged between INF and NI CD tissues (FIG.3A-D and

473 FIG.S5A-B). These findings suggest unperturbed GM-CSFR expression and

474 responsiveness in the INF and NI CD mucosa.

475 GM-CSF stimulation controls essential myeloid functions including RA production. We

476 next assessed the production of RA by APCs using ALDEFLUOR staining on HLA-

477 DR+CD11c+ APCs from the INF and NI CD mucosa. APCs including CD14+ MP,

478 CD141+ DC and CD1c+ DC showed decreased ALDEFLUOR staining specifically in

479 the INF mucosa (FIG.3E-F, FIG.S5C-D). Monocytes and precursor DC continuously

480 infiltrate the intestinal mucosa to differentiate into DC and MP and require GM-CSF for

481 RA production10, 15,32, 33. GM-CSF-stimulation of blood-derived CD14+ monocytes

28 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

482 confirmed a GM-CSF-dependent increase in ALDEFLUOR staining comparable to

483 staining in APCs from the NI CD mucosa (FIG.S5C-E). APCs isolated from a patient

484 carrying a CSF2RBMUT allele revealed the requirement for GM-CSF in maintaining APC

485 function even in NI tissues (FIG.3G-H). To determine if aGMAb alter GM-CSF-driven

486 gene expression in the NI CD mucosa, expression of STAT5A, STAT5B, IRF5, CCR2,

487 CCL22, CCL17, VDR, SOCS3, ALDH1A1, ALDH1A2, ALDH1A3, TNF, NOS2, IL1B

488 was investigated in total tissue RNAseq data. Subdivision of patients based on their

489 aGMAb status revealed decreased ALDH1A1, ALDH1A3, CCL22, IL23A, VDR,

490 CCL17, IRF5, CCL2, NOS2 and IL1B levels and increased CCR2, TNF and ALDH1A2

491 expression in the presence of aGMAb (FIG.3I). These data suggest that GM-CSF

492 regulates homeostatic gene expression incl. RA production in the NI CD mucosa,

493 implicating an aGMAb-driven perturbation of homeostasis even prior to inflammation.

494

495 T cells and ILC3 contribute to the pool of GM-CSF in the non-inflamed and inflamed

496 CD mucosa.

497 Assessment of spontaneously released GM-CSF by CD45+ cells was comparable

498 between INF and NI CD tissues (FIG.4A), in line with unchanged serum GM-CSF

29 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

499 levels in CD patients (FIG.S2A). A characterization of GM-CSF-producing cells in the

500 NI CD ileal mucosa revealed that 80% were NKp44+CD3- innate lymphoid cells (ILC),

501 while the remaining 20% composed of CD3+ T cells, or cells lacking both markers

502 (FIG.4B). Interestingly, the number of GM-CSF-producing T cells increased in the INF

503 CD mucosa while the number of GM-CSF+NKp44+ cells decreased (FIG.4B-C). GM-

504 CSF-secreting NKp44+ cells co-expressed CD117, CD127, CD161, CD69 and the

505 transcription factor Retinoic acid-related Orphan Receptor (ROR) gamma (g) t

506 (FIG.4D), identifying them as natural cytotoxicity receptor (NCR)+ group 3 ILC

507 (NCR+ILC3)34. Analyzing NCR+ILC3 numbers and cytokine release revealed lower

508 levels of GM-CSF production and GM-CSF producers in the INF mucosa (FIG.4E-F

509 and FIG.S5F). Despite of increased GM-CSF+ T cell frequencies, a lower per cell

510 output was observed in these cells (FIG.4C and FIG.S5F). We next determined if

511 NCR+ILC3 increasingly differentiated into group 1 ILCs (ILC1) as reported in CD12, 14.

512 ILC1s were increased in the INF mucosa releasing higher levels of interferon(IFN)-g

513 (FIG.4G)35, 36. These findings demonstrate changes in the source and levels of GM-

514 CSF in the INF CD mucosa, accompanied by decreased NCR+ILC3 and increased

515 ILC1 counts (FIG.4E-G). The aberrantly glycosylated GM-CSF observed in CD sera

30 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

516 (FIG.1D), inspired the analysis of glycogene-expression in GM-CSF+ leukocytes of the

517 INF and NI CD mucosa. Published scRNA-Seq data from CD patients was analyzed

518 for the expression of in T cells, NCR+ILC3s and ILC137.

519 Glycogene signatures differed in T cells and NCR+ILC3s between INF and NI mucosa,

520 implicating the production of differential glycovariants of secreted proteins including

521 GM-CSF. T cells and NCR+ILC3 in the INF mucosa displayed an upregulation of

522 mannose-related (ALG6, ALG8, ALG9), fucosylation-related (FUT2, FUT7) and α2,3-

523 sialylation-related glycogenes, when compared to the NI mucosa, in line with previous

524 reports (FIG.4H-I)38-42. Noteworthy, substrate availability and modulation of

525 glycosyltransferases expression may collectively, or individually result in aberrantly

526 glycosylated proteins including GM-CSF. The decreased availability of NCR+ILC3-

527 derived GM-CSF was reported to reduce myeloid RA production in mice, aligning with

528 our findings in CD patients (FIG.3E-G and FIG.S5C-F)10, 43. Importantly, myeloid-

529 derived RA prevents the accumulation of inflammatory ILC1, suggesting that aGMAb

530 perturb the equilibrium of tissue-resident ILCs in pre-diagnostic CD patients through

531 the modulation of GM-CSF-dependent myeloid RA production29,30. To determine if the

532 presence of aGMAb coincided with altered NCR+ILC3 and ILC1 gene signatures, bulk

31 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

533 mRNA-Seq data from NI ileal CD biopsies (n = 191) and non-IBD controls (n = 121)

534 was re-analyzed for ILC1 and ILC3 signature gene expression. Samples were

535 displayed based on their PC-NCR+ILC3 and PC-ILC1 signatures gene expression

536 profiles and revealed groups with high NCR+ILC3 and ILC1 (3H1H), low NCR+ILC3

537 and ILC1 (3L1L) and low NCR+ILC3 and high ILC1 signatures (3L1H) (FIG.4J). We

538 next evaluated aGMAb titers across the most differential PC signatures and found

539 aGMAb enriched in groups low in NCR+ILC3 signature gene expression (FIG.4J-K),

540 suggesting an association of aGMAb with lower NCR+ILC3 signatures in NI CD

541 tissues.

542

543 Unmodified GM-CSF as a potential way to restore homeostatic functions of GM-CSF.

544 Enzymatically stripped and genetically engineered GM-CSF, lacking posttranslational

545 glycosylations remained biologically active and initiated phosphorylation of STAT5

546 (FIG.S6A-C). We hypothesize that GM-CSF devoid of glycosylations would render the

547 cytokine capable of escaping neutralization by CD-associated aGMAb. Freshly

548 isolated PBMCs stimulated with glycosylated and deglycosylated GM-CSF in the

549 presence or absence of aGMAb were analyzed for STAT5 phosphorylation. The

32 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

550 neutralizing effects of CD-associated aGMAb were recapitulated for glycosylated GM-

551 CSF but reverted when enzymatically stripped GM-CSF was used (FIG.S6A-D).

552

553 DISCUSSION

554 Here, we report the presence of aGMAb in the sera of CD patients years before

555 diagnosis, and propose that these antibodies contribute to the pathophysiology of CD.

556 We demonstrated IgG2- and IgA-skewed aGMAb isotypes in CD patients, suggesting

557 an origin within the intestinal mucosa. Anti–GM-CSF autoantibodies were associated

558 with ileal disease location, were present up to 6 years before diagnosis in

559 asymptomatic subjects developing CD, and predicted complications at disease

560 presentation (HR=2.9 by log rank, p<0.001). IgA aGMAb, were an exclusive hallmark

561 of CD, and blocked GM-CSFR signaling by binding to GM-CSF glycovariants. This in

562 turn impaired communication of ILC3 and myeloid cells in the NI CD mucosa at steady-

563 state, resulting in significantly lower expression of an ILC3-specific gene signature in

564 the presence of aGMAb. Furthermore, GM-CSF target gene expression in the NI

565 mucosa of CD patients is reduced in the presence of aGMAb, identifying a subgroup

566 of individuals at high risk of developing complicated CD through alterations of the

33 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

567 intestinal immune balance. Importantly, GM-CSF in CD patients is aberrantly

568 glycosylated via an altered gene expression of glycosyltransferases in GM-CSF-

569 producing ILC3 and T cells, suggesting immunogenicity of post-translational epitopes

570 recognized by aGMAb. Together, these results support a novel mechanism of disease

571 pathophysiology that may be exploited for developing personalized CD-preventive and

572 therapeutic strategies.

573 Myeloid cells of the INF CD mucosa displayed markedly reduced GM-CSF-dependent

574 RA production, associated with reduced NCR+ILC3 numbers, lower GM-CSF output

575 and altered glycogene expression. Abrogating GM-CSF-mediated signaling altered

576 myeloid cell function and increased ILC1 signature gene expression even in NI tissues,

577 suggesting a role for aGMAb in abrogating homeostasis13, 14, 34, 36, 44. Enrichment in

578 lower ILC3 signatures in aGMAb+ CD patients suggest a GM-CSF-dependent myeloid

579 regulation of the ILC3/ILC1 balance. Autoreactive aGMAb thus promote the

580 accumulation of inflammatory ILC1 by disrupting the NCR+ILC3-myeloid cell circuit.

581 We suggest that CD-associated aGMAb promote immune imbalance during the years-

582 long “pre-diagnostic” period of CD by altering GM-CSF-dependent homeostasis, a

34 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

583 hypothesis supported by our data that demonstrates reduced GM-CSF-dependent

584 gene expression even in the NI CD mucosa of aGMAb+ patients.

585 There is rising interest in identifying the preclinical phase of CD where prevention

586 strategies may eventually be pursued45. Measuring aGMAb adds to the list of predictive

587 biomarkers which have been recently identified, such as microbial antibodies46 and

588 intestinal permeability47. One specific interest is that aGMAb are associated with

589 complicated CD disease at onset and thus may one day help identify relevant

590 candidates for disease prevention at its earliest phase.

591 The differential glycosylation of GM-CSF and altered expression

592 suggest that new (glyco)epitopes on GM-CSF may promote the development of

593 aGMAb. Engineering of GM-CSF variants with functional activity but capable of

594 circumventing aGMAb may provide a way to re-establish homeostasis, or delay

595 disease progression. Our data calls for revisiting the use of GM-CSF in clinical trials

596 with modified versions of active GM-CSF not prone to antibody neutralization, with

597 careful pre-selection of patients based on their aGMAb profiles. Our findings

598 demonstrate an intriguing mechanism for the development of complicated CD and

599 allow the identification of these patients using a predictive serological biomarker.

35 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

600 Collectively, these findings open new roads for the precise diagnosis, classification,

601 and personalized treatment of CD patients.

602

36 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

603 REFERENCES

604 1. Ng SC, Shi HY, Hamidi N, et al. Worldwide incidence and prevalence of

605 inflammatory bowel disease in the 21st century: a systematic review of

606 population-based studies. Lancet 2018;390:2769-2778.

607 2. Ananthakrishnan AN, Bernstein CN, Iliopoulos D, et al. Environmental triggers

608 in IBD: a review of progress and evidence. Nature Reviews Gastroenterology &

609 Hepatology 2018;15:39-49.

610 3. Silverberg MS, Satsangi J, Ahmad T, et al. Toward an integrated clinical,

611 molecular and serological classification of inflammatory bowel disease: report

612 of a Working Party of the 2005 Montreal World Congress of Gastroenterology.

613 Canadian Journal of Gastroenterology 2005;19 Suppl A:5A-36A.

614 4. Verhelst X, Dias AM, Colombel JF, et al. Protein Glycosylation as a Diagnostic

615 and Prognostic Marker of Chronic Inflammatory Gastrointestinal and Liver

616 Diseases. Gastroenterology 2020;158:95-110.

617 5. Han X, Gilbert S, Groschwitz K, et al. Loss of GM-CSF signalling in non-

618 haematopoietic cells increases NSAID ileal injury. Gut 2010;59:1066-1078.

37 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

619 6. Hirata Y, Egea L, Dann SM, et al. GM-CSF-facilitated dendritic cell recruitment

620 and survival govern the intestinal mucosal response to a mouse enteric bacterial

621 pathogen. Cell Host & Microbe 2010;7:151-163.

622 7. Lang RA, Metcalf D, Cuthbertson RA, et al. Transgenic mice expressing a

623 hemopoietic growth factor gene (GM-CSF) develop accumulations of

624 macrophages, blindness, and a fatal syndrome of tissue damage. Cell

625 1987;51:675-686.

626 8. Sainathan SK, Hanna EM, Gong Q, et al. Granulocyte macrophage colony-

627 stimulating factor ameliorates DSS-induced experimental colitis. Inflammatory

628 Bowel Diseases 2008;14:88-99.

629 9. Song C, Lee JS, Gilfillan S, et al. Unique and redundant functions of NKp46+

630 ILC3s in models of intestinal inflammation. Journal of Experimental Medicine

631 2015;212:1869-1882.

632 10. Mortha A, Chudnovskiy A, Hashimoto D, et al. Microbiota-dependent crosstalk

633 between macrophages and ILC3 promotes intestinal homeostasis. Science

634 2014;343:1249288.

38 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

635 11. Castro-Dopico T, Fleming A, Dennison TW, et al. GM-CSF Calibrates

636 Macrophage Defense and Wound Healing Programs during Intestinal Infection

637 and Inflammation. Cell Rep 2020;32:107857.

638 12. Vonarbourg C, Mortha A, Bui VL, et al. Regulated expression of nuclear

639 receptor RORgammat confers distinct functional fates to NK cell receptor-

640 expressing RORgammat(+) innate lymphocytes. Immunity 2010;33:736-751.

641 13. Bernink JH, Krabbendam L, Germar K, et al. Interleukin-12 and -23 Control

642 Plasticity of CD127(+) Group 1 and Group 3 Innate Lymphoid Cells in the

643 Intestinal Lamina Propria. Immunity 2015;43:146-160.

644 14. Bernink JH, Peters CP, Munneke M, et al. Human type 1 innate lymphoid cells

645 accumulate in inflamed mucosal tissues. Nature Immunology 2013;14:221-229.

646 15. Chuang LS, Villaverde N, Hui KY, et al. A Frameshift in CSF2RB Predominant

647 Among Ashkenazi Jews Increases Risk for Crohn's Disease and Reduces

648 Monocyte Signaling via GM-CSF. Gastroenterology 2016;151:710-723 e2.

649 16. Dabritz J. Granulocyte macrophage colony-stimulating factor and the intestinal

650 innate immune cell homeostasis in Crohn's disease. American Journal of

651 Physiology-Gastrointestinal and Liver Physiology 2014;306:G455-465.

39 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

652 17. Dieckgraefe BK, Korzenik JR. Treatment of active Crohn's disease with

653 recombinant human granulocyte-macrophage colony-stimulating factor. Lancet

654 2002;360:1478-1480.

655 18. Korzenik JR, Dieckgraefe BK, Valentine JF, et al. Sargramostim for active

656 Crohn's disease. New England Journal of Medicine 2005;352:2193-2201.

657 19. Roth L, MacDonald JK, McDonald JW, et al. Sargramostim (GM-CSF) for

658 induction of remission in Crohn's disease: a cochrane inflammatory bowel

659 disease and functional bowel disorders systematic review of randomized trials.

660 Inflammatory Bowel Diseases 2012;18:1333-1339.

661 20. Han X, Uchida K, Jurickova I, et al. Granulocyte-macrophage colony-stimulating

662 factor autoantibodies in murine ileitis and progressive ileal Crohn's disease.

663 Gastroenterology 2009;136:1261-1271, e1-3.

664 21. Nylund CM, D'Mello S, Kim MO, et al. Granulocyte macrophage-colony-

665 stimulating factor autoantibodies and increased intestinal permeability in Crohn

666 disease. Journal of Pediatric Gastroenterology and Nutrition 2011;52:542-548.

40 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

667 22. Gathungu G, Kim MO, Ferguson JP, et al. Granulocyte-macrophage colony-

668 stimulating factor autoantibodies: a marker of aggressive Crohn's disease.

669 Inflammatory Bowel Diseases 2013;19:1671-1680.

670 23. Jurickova I, Collins MH, Chalk C, et al. Paediatric Crohn disease patients with

671 stricturing behaviour exhibit ileal granulocyte-macrophage colony-stimulating

672 factor (GM-CSF) autoantibody production and reduced neutrophil bacterial

673 killing and GM-CSF bioactivity. Clinical and Experimental Immunology

674 2013;172:455-65.

675 24. Dabritz J, Bonkowski E, Chalk C, et al. Granulocyte macrophage colony-

676 stimulating factor auto-antibodies and disease relapse in inflammatory bowel

677 disease. American Journal of Gastroenterology 2013;108:1901-1910.

678 25. Piccoli L, Campo I, Fregni CS, et al. Neutralization and clearance of GM-CSF

679 by autoantibodies in pulmonary alveolar proteinosis. Nature Communications

680 2015;6:7375.

681 26. Griseri T, Arnold IC, Pearson C, et al. Granulocyte Macrophage Colony-

682 Stimulating Factor-Activated Eosinophils Promote Interleukin-23 Driven Chronic

683 Colitis. Immunity 2015;43:187-199.

41 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

684 27. Porter CK, Riddle MS, Gutierrez RL, et al. Cohort profile of the PRoteomic

685 Evaluation and Discovery in an IBD Cohort of Tri-service Subjects (PREDICTS)

686 study: Rationale, organization, design, and baseline characteristics. Contemp

687 Clin Trials Commun 2019;14:100345.

688 28. Miyajima A, Otsu K, Schreurs J, et al. Expression of murine and human

689 granulocyte-macrophage colony-stimulating factors in S. cerevisiae:

690 mutagenesis of the potential glycosylation sites. EMBO Journal 1986;5:1193-

691 1197.

692 29. Perdue CL, Eick-Cost AA, Rubertone MV. A Brief Description of the Operation

693 of the DoD Serum Repository. Mil Med 2015;180:10-2.

694 30. Plevy S, Silverberg MS, Lockton S, et al. Combined serological, genetic, and

695 inflammatory markers differentiate non-IBD, Crohn's disease, and ulcerative

696 colitis patients. Inflammatory Bowel Diseases 2013;19:1139-1148.

697 31. Torres J, Petralia F, Sato T, et al. Serum Biomarkers Identify Patients Who Will

698 Develop Inflammatory Bowel Diseases Up to 5 y Before Diagnosis.

699 Gastroenterology 2020.

42 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

700 32. Bujko A, Atlasy N, Landsverk OJB, et al. Transcriptional and functional profiling

701 defines human small intestinal macrophage subsets. Journal of Experimental

702 Medicine 2018;215:441-458.

703 33. Richter L, Landsverk OJB, Atlasy N, et al. Transcriptional profiling reveals

704 monocyte-related macrophages phenotypically resembling DC in human

705 intestine. Mucosal Immunology 2018;11:1512-1523.

706 34. Cella M, Fuchs A, Vermi W, et al. A human natural killer cell subset provides an

707 innate source of IL-22 for mucosal immunity. Nature 2009;457:722-725.

708 35. Abraham C, Cho JH. Inflammatory bowel disease. New England Journal of

709 Medicine 2009;361:2066-2078.

710 36. Glatzer T, Killig M, Meisig J, et al. RORgammat(+) innate lymphoid cells acquire

711 a proinflammatory program upon engagement of the activating receptor NKp44.

712 Immunity 2013;38:1223-1235.

713 37. Martin JC, Chang C, Boschetti G, et al. Single-cell analysis of Crohn's disease

714 lesions identifies a pathogenic cellular module associated with resistance to

715 anti-TNF therapy. Cell 2019;178:1493-1508 e20.

43 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

716 38. Dias AM, Pereira MS, Padrao NA, et al. Glycans as critical regulators of gut

717 immunity in homeostasis and disease. Cell Immunol 2018;333:9-18.

718 39. Pereira MS, Duraes C, Catarino TA, et al. Genetic Variants of the MGAT5 Gene

719 Are Functionally Implicated in the Modulation of T Cells Glycosylation and

720 Plasma IgG Glycome Composition in Ulcerative Colitis. Clin Transl

721 Gastroenterol 2020;11:e00166.

722 40. Fujii H, Shinzaki S, Iijima H, et al. Core Fucosylation on T Cells, Required for

723 Activation of T-Cell Receptor Signaling and Induction of Colitis in Mice, Is

724 Increased in Patients With Inflammatory Bowel Disease. Gastroenterology

725 2016;150:1620-1632.

726 41. Pereira MS, Alves I, Vicente M, et al. Glycans as Key Checkpoints of T Cell

727 Activity and Function. Front Immunol 2018;9:2754.

728 42. Dias AM, Dourado J, Lago P, et al. Dysregulation of T cell receptor N-

729 glycosylation: a molecular mechanism involved in ulcerative colitis. Hum Mol

730 Genet 2014;23:2416-27.

44 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

731 43. Samarakoon A, Shim YA, Dosanjh M, et al. CD45 regulates GM-CSF, retinoic

732 acid and T-cell homing in intestinal inflammation. Mucosal Immunol

733 2016;9:1514-1527.

734 44. Croxatto D, Micheletti A, Montaldo E, et al. Group 3 innate lymphoid cells

735 regulate neutrophil migration and function in human decidua. Mucosal

736 Immunology 2016;9:1372-1383.

737 45. Torres J, Halfvarson J, Rodriguez-Lago I, et al. Results of the Seventh Scientific

738 Workshop of ECCO: Precision medicine in IBD- prediction and prevention of

739 inflammatory bowel disease. J Crohns Colitis 2021.

740 46. Choung RS, Princen F, Stockfisch TP, et al. Serologic microbial associated

741 markers can predict Crohn's disease behaviour years before disease diagnosis.

742 Aliment Pharmacol Ther 2016;43:1300-10.

743 47. Turpin W, Espin-Garcia O, Bedrani L, et al. Analysis of Genetic Association of

744 Intestinal Permeability in Healthy First-degree Relatives of Patients with Crohn's

745 Disease. Inflamm Bowel Dis 2019;25:1796-1804.

746

747

45 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

748 FIGURE LEGENDS

749

750 FIGURE 1. Characterization of aGMAb in CD patients.

751 A) Reciprocal titers for total serum IgG aGMAb in sera of HD, PAP, CD and UC

752 patients. B) Isotype profiles of aGMAb in PAP and CD patients. Horizontal rows

753 represent patients and vertical rows indicate isotypes. C) Western blots probed with

754 polyclonal sera from CD and PAP patients shows binding of anti-GM-CSF IgG and IgA

755 to glycosylated (PT-modified) and stripped GM-CSF D) Levels of GNA, AAL and L-

756 PHA binding to GM-CSF from healthy donors (black) and Crohn’s Disease (red)

757 patients, normalized for the total levels of GM-CSF in each sample. Schematic

758 representation of N-glycan highlighting lectin recognition. E) Plots show quantification

759 of pSTAT5 signal in DC, MP and pDC either unstimulated or stimulated with GM-CSF

760 pre-incubated with serum form the indicated patient groups. F) Loss in pSTAT5

761 correlates with aGMAb titers. One-way analysis of variance (ANOVA) Bonferroni’s

762 multiple comparison test was performed. Mann-Whitney test *p-value <0.05

763

764 FIGURE 2. CD-specific aGMAb precede the onset of disease by years.

46 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

765 A) and B) show reciprocal titers of aGMAb (IgG and IgA) in combined serum samples

766 (training and validation cohort) at two time points prior to and one time point post

767 diagnosis. C) Trajectory of aGMAb titers in CD patients. Blue lines indicate aGMAb+,

768 black line aGMAb- patients and red lines indicate sero-converters. D) Kaplan-Meier

769 analysis with hazard ratio for developing complications after diagnosis in aGMAb+ (red)

770 and aGMAb- patients 6 years prior to disease. E) Correlations of aGMAb with ASCA

771 IgG and ASCA IgA antibodies a different time points prior to diagnosis.

772

773 FIGURE 3. The inflamed CD mucosa shows impaired homeostatic functions in GM-

774 CSF-responsive myeloid cells.

775 LPL from NI and INF ileal mucosa were analyzed . A) ViSNE analysis shows

776 distribution of leukocyte populations in NI and INF tissues indicating supervised

777 annotation of populations. B) Stimulation of cells in A) with GM-CSF, followed by

778 pSTAT5 measurement. Signal intensity is visualized in t-SNE plots (blue=low,

779 red=high pSTAT5). C) Representative histograms show pSTAT5 levels in the indicated

780 myeloid cell populations. D) Heat map shows pSTAT5 intensity across myeloid

781 populations from A) and B). E) DC subset and MP identification by floc cytometry. F)

47 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

782 Mean fluorescence intensity (MFI) of ALDEFLOUR staining in CD45+CD11c+HLA-DR+

783 cells. ALDEFLUOR staining in CD45+CD11+HLA-DR+ cells in CD control and one

784 CSF2RBMUT carrier from G) NI and H) INF tissue biopsies. I) Gene expression analysis

785 of total tissue RNA from NI ileal tissues CD patients positive (n=14) of negative (n=27)

786 for aGMAb. Heat map shows fold change in gene expression in aGMAb+ patients

787 compared to aGMAb- CD patients.

788

789 FIGURE 4. Innate and adaptive sources of intestinal GM-CSF in CD patients

790 A) shows GM-CSF+CD45+ cells the adjacent dot plot their frequency in NI and INF CD

791 mucosa. B) GM-CSF+CD45+ cells were analyzed for CD3 and NKp44 expression in NI

792 and INF CD mucosa. Numbers in gate represents percentages. C) Plots show

793 percentages of NKp44+GM-CSF+ and CD3+GM-CSF+ cells. D) Expression of CD127,

794 CD161, RORgt and CD69 on NKp44+ CD117+ cells. E) GM-CSF production by

795 NCR+ILC3 (CD45+CD3-CD4-CD127+CD161+NKp44+CD117+), F) NCR+ILC3

796 frequencies and G) INFg-producing ILC1 were quantified. Numbers in gates represent

797 percentages. H) and I) show gene expression analysis of glycosyltransferase genes in

798 the indicated population from INF and NI CD patients. J) Scatter plot shows PC1 values

48 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

799 of ILC1-PC and NCR+ILC3-PC derived from PCA analysis of signature ILC1 and ILC3

800 genes against a sub-dataset (195 NI ileal CD and 121 NI ileal non-IBD samples) of the

801 bulk mRNAseq samples from the MSCCR. Blue dots and red dots indicate anti-GM-

802 CSF status. K) The top 25 samples with either the lowest and highest NCR+ILC3 or

803 ILC1 signature gene expression were tested for aGMAb titers. Reciprocal titers are

804 shown in plots.

805

806 FIGURE S1. Specificity and affinity of anti–GM-CSF autoantibodies in CD patients.

807 A) ELISA specific to cytokines (G-CSF, IL-2 and GM-CSF), nuclear antigens and

808 unrelated autoantigens was performed using sera from PAP patients and IBD patients.

809 Sera simultaneously reacting against GM-CSF and other antigens were excluded from

810 the study. B) Binding avidity of aGMAb was determined using anti–GM-CSF ELISA

811 with adjusted PAP and IBD sera. Post incubation of plates with anti–GM-CSF sera,

812 wells were washed with buffers containing increasing amounts of NaCl. Signal

813 detected by the ELISA plate reader post washing with high salt buffer was normalized

814 to signals obtained in regular ELISAs. The binding strength of PAP sera (black) and

815 IBD sera (red) are displayed. C) Anti–GM-CSF ELISA were performed using CD, UC

49 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

816 and HC sera. Secondary antibodies recognizing IgG, IgA and IgM were used to identify

817 enrichment of aGMAb in CD patients. D) Sera from CD patient were tested for their

818 association with one of the three behavioral stages described in the Montreal

819 Classification. E) Bar graph shows titers of aGMAb ELISA on native and denaturated

820 GM-CSF using sera from PAP and CD patients. F) Native PAGE of GM-CSF

821 (sargramostim) and stripped GM-CSF stained with Coomassie Brilliant Blue. G)

822 Western blot of sargramostim, recombinantly expressed human GM-CSF purified from

823 HEK293 cells stably secreting wild type human GM-CSF, or human GM-CSF mutated

824 to lack glycosylations. H) Scheme adjacent to plot shows exemplified positioning of the

825 evaluated N-linked glycosylations. I) Characterization of N-glycosylation of yeast (Y)-

826 and mammalian (M)-produced GM-CSF by lectin blot for L-PHA, MALII, GNA and AAL.

827 Western blot for GM-CSF confirms the position and molecular weight of the

828 recombinant protein.

829

830 FIGURE S2. Neutralization of GM-CSF by aGMAb from CD patients.

831 A) Total serum GM-CSF concentration in healthy donor (HD) or CD patients. B)

832 PBMCs were isolated from a buffy coat. For every tested serum, 2 million PBMCs were

50 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

833 seeded into the well of a 96 well plate. Cells were either left unstimulated or stimulated

834 with rhGM-CSF for 20 minutes. GM-CSF-stimulated samples were pre-incubated with

835 either serum from aGMAb- CD patients, aGMAb+ CD patients or PAP patients. Cells

836 were fixed after stimulation, barcoded prior to surface staining, then pooled and stained

837 for surface marker and intracellular pSTAT5. Phosphorylation of STAT5 was analyzed

838 in the indicated populations. Heat maps show signal intensity of anti-pSTAT5 staining

839 in yellow color code for individual patients (lanes) within the indicated population (row).

840 C) Representative pSTAT5 staining in monocytes either left untreated (grey), or

841 stimulated with GM-CSF in the presence of control CD serum (blue) or CD serum

842 containing aGMAb (red). D) PBMCs were stimulated in vitro with rhIL-3 in the presence

843 or absence of the indicated sera for 20 minutes. Cells were washed, barcoded, pooled

844 and then stained with surface antibodies follow by Intracellular staining with antibodies

845 against phosphorylated STAT5. Mass cytometry was performed and intensity of

846 pSTAT5 was visualized using heat maps. Scatter plots show quantification of IL-3

847 mediated STAT5 phosphorylation in Basophils in the presence aGMAb- or aGMAb+

848 CD sera.

849

51 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

850 FIGURE S3. Anti–GM-CSF autoantibodies precede the onset of Crohn‘s Disease.

851 A-D) Anti–GM-CSF ELISA was performed using serum samples obtained from CD

852 patients, UC patients and HD. Sera were obtained at two (training cohort) and three

853 (validation cohort) time points prior to diagnosis of disease and one time point after

854 diagnosis of disease. Titers of anti–GM-CSF IgG and IgA were determined for each

855 time point. Trajectory of aGMAb titers in CD, UC and HD across different time points

856 are displayed in E) and F). Blue line indicate patient tested positive at the earliest time

857 point of collection. Red lines indicate sero-converter, while black lines indicate patients

858 without aGMAb.

859

860 FIGURE S4. Anti–GM-CSF autoantibodies determine disease location and disease

861 severity in two independent cohorts.

862 A-F) Serum samples described in supplementary table 3 were analyzed for the

863 association of IgG and IgA with disease location, obstruction, penetrance, surgery,

864 perianal involvement and complications. A trainings cohort was established A-C) and

865 compared to a validation cohort D-F).

866

52 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

867 FIGURE S5. The inflamed CD mucosa shows intact GM-CSFR expression but

868 reduced homeostatic, GM-CSF-dependent myeloid functions.

869 A) and B) show CD116 and CD131 expression intensity across all leukocyte

870 populations in NI A) and B) INF tissues identified by t-SNE analysis. Scale adjacent to

871 plots indicates signal intensity. C) Histogram of representative ALDEFLOUR staining

872 on intestinal macrophages from the NI (red) and INF (blue) CD mucosa. D) Plots show

873 percentages of ALDEFLUOR staining+ MP, CD141+DC and CD1c+DC. E) Blood

874 CD14+ monocytes were cultured in GM-CSF or M-CSF. Cells were analyzed for RA

875 production using ALDEFLUOR staining 5 days later. F) Plots show quantification of

876 GM-CSF+ NCR+ILC3 cells, relative abundance of NCR+ILC3s among all ILCs,

877 IFNg+CD45+CD3-CD4-CD127+CD161+NKp44- ILC1/ex-NCR+ILC3 cells and GM-

878 CSF+CD3+ cells. Datasets shown are representative of 9-20 individual ileal CD

879 resections. Statistical analysis was performed using paired student’s t-test. P values

880 are indicated adjacent to datasets.

881

882 FIGURE S6. Deglycosylated GM-CSF escapes aGMAb neutralization.

53 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

883 A) Purified CD14+ monocytes were stimulated with rhGM-CSF (sargramostim) or

884 stripped rhGM-CSF for 20min. Cells were analyzed for pSTAT5 levels. B) U937

885 myelomonocytic cells were analyzed for their expression of CD116 and CD131.

886 Histograms show surface stained cells (blue) and unstained controls (grey). C) U937

887 cells were stimulated with rhGM-CSF (sargramostim), purified fully glycosylated GM-

888 CSF or mutated GM-CSF lacking all posttranslational glycosylation sites. Following

889 stimulation, pSTAT5 levels were analyzed. D) Plots show quantification of pSTAT5

890 signal intensity in monocytes and DC either stimulated with GM-CSF or stripped GM-

891 CSF for 20 minutes pre-incubated with serum form the indicated patient groups. One-

892 way analysis of variance (ANOVA) Bonferroni’s multiple comparison test was

893 performed.

894

895 Supplementary Table 1.

896 Available patient information for serum samples used in FIGURE.1.

897 Supplementary Table 2.

898 Table shows antibodies and isotope conjugates used in mass-cytometry analysis of

899 peripheral blood mononucleated cells (PBMC) and lamina propria leukocytes (LPL).

54 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

900 Scheme adjacent to tables demonstrate the workflow for pSTAT5 staining in these

901 samples.

902 Supplementary Table 3.

903 Available patient information for serum samples used in Figure.3, Figure S3 and Figure

904 S4.

905 Supplementary Table 4.

906 ELISAs for ASCA-specific antibodies were performed on serum samples collected at

907 three different time points (time point 1 and 2 = prior to disease diagnosis, time point 3

908 post disease diagnosis). ASCA-specific IgG and IgA were measured.

909

910 Graphical abstract: Establishing a pre-diseased state through aGMAb in CD

911 Model of CD development in aGMAb+ individuals. Scheme shows cellular crosstalk in

912 healthy intestinal tissue (left). NCR+ILC3 and T cells produce GM-CSF that engages

913 the GM-CSFR on myeloid cells to trigger the production of RA. Retinoic acid in turn

914 stabilizes NCR+ILC3 and prevents excessive differentiation of NCR+ILC3 into IFN-g

915 producing ILC1. The middle scheme displays the scenario in the presence of aGMAb.

916 GM-CSF produced by T cells and NCR+ILC3 is aberrantly glycosylated and may drive

55 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

917 the activation of glycosylation-specific B cells. B cell-derived aGMAb neutralize GM-

918 CSF and reduce GM-CSFR signaling, leading to a decreased production of retinoic

919 acid. Consequently, NCR+ILC3s differentiation into IFNg producing ILC1. The right

920 scheme resembles the immune landscape in full-blown CD. The inflamed CD mucosa

921 is characterized by a decrease in NCR+ILC3 numbers, lower levels of GM-CSF

922 produced by NCR+ILC3, reduced levels of RA from myeloid cells and excessive

923 differentiation into IFNg producing ILC1 and T cells. The presence of aGMAb may

924 interrupt the NCR+ILC3 – myeloid cell circuit, destabilizing homeostasis towards the

925 scenario of full-blown CD.

926

56 medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

6 4 0 2 2 1 3 0

6 4 0 0 4 0 0 0

5 3 0 0 0 0 0 0

5 4 0 0 0 0 2 0

4 3 2 2 4 0 1 0

4 3 0 3 0 0 1 0

**** **** 4 3 0 0 2 0 0 0 Figure 1. 4 3 0 3 0 1 2 0 CD 4 3 0 0 2 3 0 0 PAP 4 3 0 0 3 0 0 0

4 3 0 0 0 0 0 0 PAP (n=12)

4 3 0 0 0 1 0 0

#NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM! 4 0 3 0 0 2 0 0 C A B 4 0 3 0 0 3 1 0

4 2 1 0 0 3 2 0

6 4 2 3 0 1 2 3 0 ** 4 1 1 3 0 0 3 0

4 0 2 0 0 0 0 0

4 0 3 1 0 2 3 0

4 0 0 3 0 2 0 0

10 4 3 0 3 0 2 3 0

4 0 1 0 0 0 1 0

ns 3 0 0 0 0 1 1 0

3 0 2 3 0 3 3 0

3 0 0 0 0 0 0 0

3 0 2 0 0 3 0 0

3 0 0 0 0 2 0 0

3 0 0 0 0 3 3 0

3 2 0 0 0 0 1 1

3 0 0 0 0 1 0 0

3 0 0 0 0 3 2 0

3 0 0 1 0 2 0 0

3 1 0 0 0 0 2 0

3 1 0 2 0 3 0 0

3 0 0 0 0 0 3 0

3 0 0 0 0 0 0 0

3 0 2 0 0 0 3 0

3 0 0 0 0 0 0 0

5 3 0 0 0 0 0 -1 -1 PT modified GM-CSF (IgG)

3 0 0 0 0 0 2 0

3 0 0 0 0 0 0 0

3 0 2 0 0 0 0 0 30

2 0 1 0 0 0 2 0

10 2 1 0 0 0 0 2 0

2 0 0 0 0 2 3 0

2 0 0 0 0 0 -1 -1 2 0 0 0 0 0 -1 -1 Crohn’s 1 0 0 0 0 0 -1 -1

1 0 0 0 0 0 -1 -1

1 0 0 0 0 0 -1 -1

1 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1 Disease (n=81)

0 0 0 0 0 2 -1 -1

0 0 0 0 0 0 -1 -1 4 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

10 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 1 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

3 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

10 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

#NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM!

4 0 2 0 0 0 0 0

3 0 0 0 0 0 0 0

3 0 0 0 0 0 3 1

3 0 3 0 0 0 -1 -1

2 0 2 0 0 0 2 0

2 0 0 0 0 0 -1 -1

1 0 0 0 0 0 -1 -1

1 0 0 0 0 0 -1 -1 15

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

2 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 Ulcerative 10 0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1

0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 Colitis (n=37) 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 1 0 0 0 0 0 0 -1 -1 10 0 0 0 0 0 0 -1 -1 3 -1 -1 -1 -1 -1 -1 -1 Reciprocal Titer (IgG) Titer Reciprocal 2 -1 -1 -1 -1 -1 -1 -1 2 -1 -1 -1 -1 -1 -1 -1 2 -1 -1 -1 -1 -1 -1 -1 2 -1 -1 -1 -1 -1 -1 -1 1 0 0 0 0 0 -1 -1 1 -1 -1 -1 -1 -1 2 0 0 0 0 0 0 0 -1 -1 0 -1 -1 -1 -1 -1 0 0 0 0 0 0 0 0 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 0 0 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 Healthy stripped GM-CSF (IgG) 0 0 0 0 0 0 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 10 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 30 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 Donors (n=43) 0 -1 -1 -1 -1 -1 -1 -1 100% 40% 14% 5% 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 0 0 0 0 0 -1 -1 0 0 0 0 0 0 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1 -1 -1 0 0 -1 0 -1 -1 -1 -1 -1 0 0 10 0 -1 -1 -1 -1 -1 0 0 PAP CD UC HD Reciprocal Titer IgE IgA IgG (n=12) (n=81) (n=37) (n=43) IgM IgG1 IgG2 IgG3 IgG4 0 106 15 Crohn‘s Disease D Healthy Donors L-PHA GNA AAL 30 PT modified GM-CSF (IgA) 3 0.3 15 L-PHA GNA AAL 2 0.2 10

1 0.1 5 15 Absorbance 450 Absorbance Absorbance 450 Absorbance 450 Absorbance 0 0 0

Dendritic cells Monocytes pDC Dendritic cells Monocytes Data**** 1 F E **** **** 40 40 CD control 40 20 *** 40 *** ** 20 CD anti-GM-CSF 30 30 30 30 unstimulated PAP anti-GM-CSF CD control 20 r= -0.6153 20 r= -0.7883 20 20 1010 CD anti-GM-CSF pSTAT5 pSTAT5

pSTAT5 10 10 pSTAT5 pSTAT5 PAP anti-GM-CSF 10 10 r= -0.7857 r= -0.8469 0 0 0 0 00 101 102 103 104 105 106 101 102 103 104 105 106 titer titer medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

Figure 2.

A IgG **** C 104 104 D **** 100 Seronegative at d-2175 103 103 IgG IgA 105 Seropositive at d-2175 **** 2 102

IgG 10 104 * ** 1 1 3 10 10 10 50 0 0 102 10 10 CD CD Log Rank 101 10-1 10-1 HR: 2.8 (1.5-5.4) -7000 -2000 -1000 0 -7000 -2000 -1000 0 p<0.0001 0 % complication-free 10 104 104 0

Titer anti-GM-CSF 21% 22% 25% 5% 6% 6% 6% 8% 10% 0 1000 2000 3000 4000 5000 10-1 103 103 Follow-up (days) d+0 d+0 d-740 d-720 2 2 d-2175d-1300d+100 d-3500 **** d-3500 10 10 B IgA CD UC HD **** 1 1 E (n=220) (n=200) (n=200) 10 10 105 d-2160: Spearman r=0.56, **** d-2160: Spearman r=0.17, ns **** 100 100 d-1230: Spearman r=0.57, **** d-1230: Spearman r=0.23, *

IgA 4 * * 10 UC UC d+84: Spearman r=0.70, **** d+84: Spearman r=0.26, ** 10-1 10-1 104 104 103 -7000 -2000 -1000 0 -7000 -2000 -1000 0 IgG IgA IgG

4 IgA 4 IgA

10 10 IgG 102 103 103 Anti-GM-CSF positive 3 3

1 CSF 10 10 Anti-GM-CSF increased CSF CSF

10 - - Anti-GM-CSF negative 102 102 0 2 2

10 GM 10 10 GM - 7% 9% 12% 1% 1% 0% 1% 0% 0% - Titer anti-GM-CSF10 -1 101 101 101 101

d+0 d+0 0 0 Titer anti-GM-CSF

d-740 d-720 10 10 Titer anti-GM-CSF d-2175d-1300d+100 d-3500 d-3500 0 0

Titer anti Titer 10 10 HD anti Titer HD CD UC HD -1 -1 10 10 0 50 100 0 20 40 60 80 (n=220) (n=200) (n=200) -7000 -2000 -1000 0 -7000 -2000 -1000 0 ASCA IgG (EU) ASCA IgA (EU) Days prior to diagnosis medRxiv preprint doi: https://doi.org/10.1101/2021.08.23.21262143; this version posted August 26, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

Figure 3. A B PBS GM-CSF C D NI NI NI CD14+ Monocytes CD16+ Monocytes PBS CD14+ Monocytes GM-CSF

+

tSNE2 CD16 Monocytes

CD1c+ DC pSTAT5 INF INF CD141+ DC tSNE1 INF CD141+ DC CD1c+ DC

pDC - + - + GM-CSF tSNE2 NI INF

pSTAT5

tSNE1 CCL17 E G H I CCL171.4 F ALDH1A31.2 SOCS31 NON-INFLAMED INFLAMED NI 20.7 ALDH1A10.8 10 MUCOSA MUCOSA 70.4 CCL22 3 IL1B 8 VDR 6.9 78 22 CD control IL23A 6 IRF5 Cy5.5 - CCL2 INF 29.5 4 MUT NOS2

BV421 53.6 CD CSF2RB

- STAT5A PerCP BV605 STAT5A - - CCL17 STAT5B 1.4

DR 2

- ALDH1A3 CCR2 1.2

MFI (ALDEFLUOR) x10 MFI(ALDEFLUOR) CCR2 41.6 22 78 staining control SOCS3 1 TNF HLA CD11c CD11c CD141 CD141 0 ALDH1A1 0.8 ALDH1A2 CD11c - CD14 - APC CD1c - NI INF Neg Fold Change ALDEFLUOR fold change CCL22 BV605 Pe-Cy7 ALDEFLUOR in NI anti-GM-CSF+ IL1B

mucosa VDR

IL23A

IRF5

CCL2

NOS2

STAT5A

STAT5B

CCR2

TNF

ALDH1A2 Neg Fold Change % NCR+ILC3 in GMCSF producer% T cells in GMCSF producer A Figure 4. B C P = 0.0045 P = 0.0001 100 50 D 1 100 50

+ 7.9 0.9 30.3 8.3 + + cells cells 8080 4040 + Pe 8.7 82.5 4.7 56.7 - cells of cells

CSF 60 CSF 30 cells of cells - - 0.5 + 60 30 + CSF - - 40 20 CSF 40 20 CD161-Pe-Cy7 CD127-BV570 - of all CD45

NI INF all GM 20 all GM 10 CD3 CD3 Alexa700 GM 0 20 10 % GM % CD3

NI INF % NKp44 CD45-APC-Cy7 NKp44 - Pe 00 00 E G J NI INF NI INF F + PC-NCRILC3ILC3-PC1-Ab-only ILC3_PC1, CD uninvolvedILC3-PC1 inflamedileum, n=195 NI 10000 inflamed non inflamed 64.2 non inflamed RORgt-APC CD69-APC 16 5000

24.4 NI NI 0

Cy5.5 -2000-2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 30003000 PC-ILC1 NCR+ILC3 - INF

-5000 29.4 -5000 PerCP APC

26.1 Cy5.5 - – - - Pe -10000 -

g 10.2 14.3 NI ileal CD tissue INF INF 3L1L 3L1H CD117 PerCP IFN CD117 CD117 NKp44 -15000 Anti-GM-CSF negative GM-CSF - Pe CD117 - NKp44 - APC 3H1L Anti-GM-CSF positive NKp44-Pe

PerCP-Cy5.5 -20000 H I ALG9 FUT2 FUT7 + K ALG6 ALG9 ALG8 CD4 FUT2 ALG6 NCR+ILC3 ILC1 3 T cells 3 6 P < 0.0001 6 activated NCR+ILC3 NI NI type 3 ILC1 4 4

activated NCR+ILC3 2 2 INF INF type 3 ILC1

Glycosyltransferases Glycosyltransferases x10 (IgG) titer Reciprocal 0 x10 (IgG) titer Reciprocal 0 high low high low FIGURE S1. * ns ns 100 ns ns ns ns ns ns A B C 10000 *** *** ** D 10000

3 1 1 1 1 1 1 1

2 1 1 1 1 1 2 1

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 m

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 u

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 F 2 1 1 1 1 1 1 1 75

2 1 1 1 1 1 1 1 m

2 1 1 1 1 1 1 1 i S

2 1 1 1 1 1 1 1

2 1 1 1 1 1 2 1 x

2 1 1 1 1 1 1 2 C

2 1 1 1 1 1 2 1 -

a 1000 2 1 1 1 1 1 1 1 1000

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 M

2 1 1 1 1 1 1 1 100 m

2 1 1 1 1 1 1 1 100

2 1 1 1 1 1 1 1 f G

2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 o

2 1 1 1 1 1 1 1

IBD o 2 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 t e

2 1 1 1 1 1 1 1

2 1 1 1 1 1 2 1 IBD patients

2 1 1 1 1 1 1 1 g

g 50

2 1 1 1 1 1 1 1 IBD patients

2 2 2 2 2 2 2 2 a patients n

2 1 1 1 1 1 1 1 i t

2 1 1 1 1 1 1 1 m

2 1 1 1 1 1 1 1 CSF d

2 1 1 1 1 1 1 1 n u - 2 1 1 1 1 1 1 1 PA100P patients

2 1 1 1 1 1 1 1 n e F 2 1 1 1 1 1 1 1 i 75 100

2 1 2 2 2 1 1 1 m c

2 1 1 1 1 1 1 1

i 80 b S

2 2 2 2 1 2 2 2 r

2 1 1 1 1 1 1 1 x

2 1 1 1 1 1 1 1 C e

2 1 1 1 1 1 1 1 - a

2 1 1 1 1 1 1 1

2 2 1 1 1 1 1 1 P

2 1 2 1 1 1 1 1 M

2 1 1 1 1 1 1 1 m

2 1 1 1 1 1 1 1 25

to GM to

1 1 1 1 1 1 1 1 f G

1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 o

o

#NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM! #NUM! t

4 2 1 1 1 1 1 2 e

Reciprocal Titer Reciprocal Reciprocal Titer Reciprocal 4 1 1 1 1 1 1 1 binding max. of % PAP

3 1 1 1 1 1 1 1 10 g

g 50 3 1 1 1 1 1 2 1 60 IBD patients 10

3 1 1 1 1 1 1 1 a n

3 1 1 1 1 1 1 1 i

PAP t 3 1 1 1 1 1 2 1 3 1 1 1 1 1 1 1 patients 3 1 1 1 1 1 1 1 d n 3 1 1 1 1 1 1 1 PAP patients

2 1 1 1 1 1 1 1 n e

2 1 1 1 1 1 1 1 patients i

c 0 GM NY TP53 SSX4 MAGEA10 MAGEA3 IL G b 0 r

e 0 1 2 3 4 - - 1 2 3 4 P CSF 2 -

- 25 Concentration NaCl (M) ESO 1 CSF NaCl (M) 1

UC CD UC CD UC CD B1 B2 B3 B1 B2 B3 B1 B2 B3 - 1 0 IgG IgA IgM 0 1 2 3 4 IgG IgA IgM Concentration NaCl (M) non-stricturing, non-fistulating stricturing Y = Yeast fistulating E 10107 7 native G H M = Mammalian 10106 6 Y M Y M Y M Y M Y M 5 denaturated MALII 10105 I 10104 4 10103 3 25 kDa L-PHA 2 2 1010 25 kDa 10101 1 15 kDa GNA Reciprocaltiters (IgG) 1001 AAL 15 kDa

CD PAP F 20kDa

mutated 15kDa fully glycosylated-CSF 10kDa Sargramostim GM L-PHA MALII GNA AAL Anti-GM-CSF GM-CSF stripped -CSF GM-CSF GM FIGURE S2.

B CD anti-GM-CSF unstimulated CD + GM-CSF Crohn‘s Disease + GM-CSF A Healthy Donors D Ctrl + IL-3 CD anti-GM-CSF +IL-3

T/B/NK C Basophils 40 PAP + GM-CSF ns 30

20

pSTAT5 10

Unstimulated Basophils in pSTAT5 0 CD + GM-CSF + anti-GM-CSF -CSF -3 DC DC DC pDC pDC pDC CD + GM-CSF GM Baso Mono Baso Mono Baso Mono - 3 T/B/NK T/B/NK T/B/NK CD - anti Ctrl + IL + IL FIGURE S3.

TRAINING COHORT VALIDATION COHORT **** **** **** **** **** A IgG **** B IgA **** C IgG **** D IgA ** 5 *** 105 ** 105 * 105 ** 10 ** ** **

IgA 4

IgA 4 4 4 IgG IgG 10 10 10 10

3 103 103 103 10

102 102 102 102

101 101 101 101

100 100 100 100 Titer anti-GM-CSF anti-GM-CSF Titer

Titer anti-GM-CSF anti-GM-CSF Titer 14% 13% 15% 16% 5% 7% 6% 6% 7% 7% 7% 11% 6% 5% 7% 9% 2% 2% 1% 0% 0% 0% 0% 0% Titer anti-GM-CSF anti-GM-CSF Titer Titer anti-GM-CSF anti-GM-CSF Titer 26% 29% 33% 4% 6% 5% 5% 9% 8% 8% 12% 16% 0% 0% 0% 1% 0% 0% 10-1 10-1 10-1 10-1

UC 0 HD 0 UC 0 HD 0 UC 0 ±0 HD 0 ±0 UC 0 ±0 HD 0 ±0

UC -750 ±45 UC -750 ±45 CD -634CD +127 ±98 ±87 UC -727 ±39 HD -724 ±48 CD -634CD +127 ±98 ±87 UC -727 ±39 HD -724 ±48 CD -2160 ±59CD +84 ±102 HD -712 ± 44 CD -2160 ±59CD +84 ±102 HD -712 ± 44 UC -1465 ±62 HD -1458 ±64 UC -1465 ±62 HD -1458 ±64 CD -1230 ±126UC -2978 ±570 HD -3122 ±622 CD -1230 ±126UC -2978 ±570 HD -3122 ±622 CD -2198CD -1496 ±253 ±222 UC -3962 ±966 HD -4152 ±677 CD -2198CD -1496 ±253 ±222 UC -3962 ±966 HD -4152 ±677 Crohn's D. Ulcer. C. Healthy Crohn's D. Ulcer. C. Healthy Crohn's D. Ulcer. C. Healthy Crohn's D. Ulcer. C. Healthy (n=120) (n=100) (n=100) (n=120) (n=100) (n=100) (n=100) (n=100) (n=100) (n=100) (n=100) (n=100) E F

Crohn's Disease (CD) Ulcerative Colitis (UC) Healthy Donors (HD) Crohn's Disease (CD) Ulcerative Colitis (UC) Healthy Donors (HD) 10000 10000 10000 10000 10000 10000

1000 1000 1000 1000 1000 1000 CSF IgG CSF CSF IgG CSF 100 100 - 100 100 100 - 100 Titer Titer Titer Titer Titer 10 10 Titer 10 10 10 10 GM GM - - 1 1 1 1 1 1 anti anti 0.1 0.1 0.1 0.1 0.1 0.1 -7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0

10000 10000 10000 10000 10000 10000 Anti-GM-CSF positive Anti-GM-CSF positive 1000 1000 1000 Anti-GM-CSF increased 1000 1000 1000 Anti-GM-CSF increased

100 100 100 Anti-GM-CSF negative 100 100 100 Anti-GM-CSF negative CSF IgA CSF CSF IgA CSF Titer - Titer Titer Titer Titer Titer - 10 10 10 10 10 10

1 1 1 GM 1 1 1 GM - - 0.1 0.1 0.1 0.1 0.1 0.1

-7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0 anti -7000 -2000 -1000 0 -7000 -2000 -1000 0 -7000 -2000 -1000 0 anti FIGURE S4.

Crohn'sTraining Disease only (max cohort IgG or IgA) ns ns Crohn's Disease only (max IgG or IgA) ns Crohn's Disease only (max IgG or IgA) A * **** ** B C ns ns ns *** * ** ns ns 42% 16% 53% 19% 18% 5% 21% 10% *** ** 50% 37% 59% 30% 36% 13% 26% 12% 10000 30% 64% 38% 100% 12% 28% 13% 100% 10000 10000 1000 1000 1000

100 100 100

Reciprocal Titer 10 Reciprocal Titer 10 Reciprocal Titer 10

1 1 1 L1: Ileum (terminal) L2: Colon L1 (28%)L2 (16%)L3 (39%) L1 (28%)L2 (16%)L3 (39%)

L3: Ileocolon Surgery (1%) Surgery (1%) Unknown (17%) Unknown (17%) Penetrating (7%) Penetrating (7%) Obstructing (21%) Obstructing (21%) IgG IgA Perianal No (88%) Perianal No (88%) IgG IgA Perianal Yes (12%) Perianal Yes (12%) Complications No (72%) Complications No (72%) Crohn's Disease only (max IgG or IgA) Non-obstr, non-penetr (72%) Non-obstr, non-penetr (72%) Complications Yes (28%) Complications Yes (28%) Validation cohort Crohn's Disease only (max IgG or IgA) ns ns Crohn's Disease only (max IgG or IgA) IgG IgA ns D ns ns * ns E F **** ns ** *** *** 10% 20% 32% 14% 10% 11% 29% 4% 36% 15% 23% 12% 18% 6% 20% 4% ns ns ** * 10000 10000 14% 31% 42% 0% 4% 23% 41% 0% 10000 1000 1000 1000 100 100 100

Reciprocal Titer 10

Reciprocal Titer 10

Reciprocal Titer 10

1 1 1

L1 (11%)L2 (33%)L3 (30%) L1 (11%)L2 (33%)L3 (30%)

Surgery (3%) Surgery (3%) Unknown (26%) Unknown (26%) Perianal No (90%) Perianal No (90%) ObstructingPenetrating (13%) (12%) ObstructingPenetrating (13%) (12%) Perianal Yes (10%) Perianal Yes (10%) IgG IgA IgG IgA ComplicationsComplications Yes (28%) No (72%) ComplicationsComplications Yes (28%) No (72%) Non-obstr, non-penetr (72%) Non-obstr, non-penetr (72%) IgG IgA FIGURE S5. A B

CD116 NI CD131 NI CD116 INF CD131 INF tSNE2 tSNE2

tSNE1 tSNE1 tSNE1 tSNE1 C D E MUCOSA CD14+ MONOCYTES

AF % pos CD141 AF % pos CD1c DC DC MP + 50 50 50 + NON-INFLAMED + GM-CSF CD1c CD14 CD141 + + + 25 25 INFLAMED 25 M-CSF

staining control staining control 0 0 00 % ALDEFLUOR %

% ALDEFLUOR % NI INF NI INF ALDEFLUOR % NI INF ALDEFLUOR ALDEFLUOR % GM-CSF in CD3+ P = 0.0016 P = 0.0227 P = 0.0049 P = 0.0006 88 F 40 100 30 ILC3 + + ILC3 + 66

30 CD3

NCR 20 + - NCR

+ 20 50 44 CSF -

CSF 10 ILC1/ex - 22 10 + g % GM % NCR+ILC3 of all ILC % GM 0 0 0 00 NI INF NI INF % INF NI INF NINI INFINF FIGURE S6. A C Monocytes U937 cells

unstimulated GM-CSF GM-CSF (stripped) unstimulated

Sargramostim

pSTAT5 Recombinant fully glycosylated Unstained control GM-CSF B Stained U937 cells Recombinant GM-CSF mutated pSTAT5 D pSTAT5 dg GM Monocytes Monocytes pSTAT5Dendritic dg Dendritic cells cells 60 40 55 40 * ** 55 30 50 30

CD116 50 20 45 20

45 pSTAT5 10 40 10

40 0 35 0 CD control + GM-CSF CD control + GM-CSF (stripped)

CD anti-GMIBD serum-CSF rhGM-CSFIBD +serum GM dgGM-CSF-CSF CD anti-GM-CSF + GM-CSF (stripped) IBD serum anti-GMIBD serumrhGM-CSF anti-GM dgGM-CSF CD131 IBD serum rhGM-CSFIBD serum dgGM-CSF

IBD serum anti-GMIBD serumrhGM-CSF anti-GM dgGM-CSF TABLE S1. TABLE S2.

A B PBMC LPL LPL isotope antigen isotope antigen PBMC Pd102Di Pd102Di Rh103Di viability Rh103Di Viability In115Di CD45_115 harvesting cells harvesting cells Pd104Di Pd104Di La139Di FceR1a Pd105Di Pd105Di Pr141Di CD45_141 Pd106Di Pd106Di Nd142Di CD19 GM-CSF stimulation GM-CSF stimulation Pd108Di Pd108Di Nd144Di CD141 Pd110Di Pd110Di Nd145Di CD4 Nd144Di CD141 Nd146Di CD8 fixation fixation Nd148Di CD16 Nd148Di CD16 Nd150Di pSTAT5 Sm149Di CD66 CD45 barcoding paladium barcoding Eu151Di CD123 Nd150Di p-STAT5 Sm152Di CD66b Eu151Di CD123 Eu153Di CD1c Eu153Di CD1c pool CD45 barcoding Tb159Di CD45_159 Sm154Di CD163 Gd160Di CD14 Tb159Di CD11c Dy162Di CD64 Gd160Di CD14 surface stain pool Ho165Di CD45_165 Dy162Di CD32 Tm169Di CD45_169 Ho165Di CD116 intracellular stain surface stain Er170Di CD3 Er166Di CD24 Yb172Di CD11b Er167Di CD38 Yb173Di CD56 Er168Di CD206 analysis intracellular stain Yb174Di HLA-DR Tm169Di CD131 Lu175Di CD45_175 Er170Di CD3 Ir191Di DNA Yb173Di CD56 analysis Ir193Di DNA Yb174Di HLADR Ir191Di DNA Ir193Di DNA TABLE S3.

Training cohort Validation cohort CD UC HC CD UC HC N subjects 120 100 100 100 100 100 N samples 360 300 300 400 400 400 Mean age 30.8 29.11 29.18 31.9 32.2 34.39 Sex (M) 79% 100% 100% 78% 80% 97% % complication 28.3% - - 28% - - L1 27.5 11% - - L2 15.8 33% - - L3 39.1 30% - - Unknown location 17.5 26% - - TABLE S4.

Healthy Controls Ulcerative Colitis Crohn’s Disease

ASCA-IgG (-2000 days) Spearman r = 0.40; p < 0.0001 **** Spearman r = 0.35; p = 0.0003 Spearman r = 0.56; p < 0.0001 **** *** ASCA-IgG (-500 days) Spearman r = 0.38; p < 0.0001 **** Spearman r = 0.31; p = 0.0019 ** Spearman r = 0.57; p < 0.0001 **** ASCA-IgG (+100 days) Spearman r = 0.48; p < 0.0001 **** Spearman r = 0.29; p = 0.0034 ** Spearman r = 0.70; p < 0.0001 **** ASCA-IgA (-2000 days) Spearman r = 0.18; n.s. Spearman r = 0.18; n.s. Spearman r = 0.17; n.s. ASCA-IgA (-500 days) Spearman r = 0.15; n.s. Spearman r = 0.18; n.s. Spearman r = 0.23; p = 0.0115 * ASCA-IgA (+ 100 days) Spearman r = N/A; n.s. Spearman r = 0.17; n.s. Spearman r = 0.26; p = 0.0049 **