Systemic Lupus Erythematosus Serum Stimulation of Human Intestinal Organoids Induces Changes in Goblet Cell Differentiation and Mitochondrial Fitness

Human intestinal epithelial cells are the interface between potentially harmful luminal content and basally residing immune cells. Their role is not only nutrient absorption but also the formation of a tight monolayer that constantly secrets mucus creating a multi-layered protective barrier. Alterations in this barrier can lead to increased gut permeability which is frequently seen in individuals with chronic extraintestinal autoimmune diseases, such as Systemic Lupus Erythematosus (SLE). Despite recent advances in identifying alterations in gut microbiota composition in SLE patients, not much attention has been given to the epithelial barrier itself. To date, it remains largely unexplored which role and function intestinal epithelial cells have in SLE pathology. Here, we present a unique near-physiologic in vitro model specifically designed to examine the effects of SLE on the epithelial cells. We utilize human colon organoids that are stimulated with serum obtained from SLE patients. Combining bulk and scRNA transcriptomic analysis with functional assays revealed that SLE serum stimulation induced a unique expression profile marked by a type I interferon gene signature. Additionally, organoids exhibited decreased mitochondrial fitness, alterations in mucus composition and imbalanced cellular composition. Similarly, transcriptomic analysis of SLE human colon biopsies revealed a downregulation of epithelial secretory markers. Our work uncovers a crucial connection between SLE and intestinal homeostasis that might be promoted in vivo through the blood, offering insights into the causal connection of barrier dysfunction and autoimmune diseases.


34
The intestine has a substantial surface area that harbors the highest quantity of immune cells in close 35 proximity to the tremendous number of microbes in our body. 1-3 Thus, a functional epithelial barrier is 36 crucial for separating these two compartments to ensure intestinal homeostasis and overall health. 37 Under physiological conditions this function is maintained by constant renewal of the barrier containing 38 a balanced cellular composition. [4][5][6] Intestinal stem cells and transit-amplifying (TA) cells proliferate and 39 give rise to different specialized cells committed to either absorptive or secretory lineage, represented 40 by the two major cell types: colonocytes and goblet cells. While colonocytes mainly contribute to 41 maintaining fluid balance and absorbing nutrients, goblet cells are considered as key players of mucosal 42 barrier integrity. The secreted mucus forms a layer which is not only essential as defense against 43 microbial infiltration, but also acts as a niche for commensals. 5,7-9 Its composition is highly dynamic, 44 can be influenced not only by the abundance of mucus core proteins and antimicrobial peptides 45 secreted to it, but also by factors such as ion concentration, pH, and hydration state 10,11 . The cell type 46 composition and therewith barrier function is continuously influenced by signals from neighboring to be elucidated. SLE manifests with a dysregulation of the immune system and is characterized by 60 elevated type I interferon levels and presence of autoantibodies in serum. [27][28][29][30] The blood is a complex 61 body fluid that serves as transport medium to supply all cells of the body with nutrients and oxygen. 62 The intestine is the most intensely perfused organ. Especially the metabolically highly active epithelial 63 cells are in close contact with the underlying vasculature ensuring swift exchange and efficient nutrient 64 absorption necessary for homeostasis. 31 However, this may also facilitate an influx of circulating 65 cytokines and autoantibodies contained in the blood to the intestine with the potential to alter 66 epithelial cell dynamics and barrier function [32][33][34] . A comprehensive understanding of this 67 interconnection is of great importance to elucidate the role of the intestinal epithelial barrier in 68 autoimmune diseases and for the development of novel therapeutic approaches. 69 In order to fully understand this complex interplay in a multicellular environment, it is necessary to 70 develop a model that can explore the impact of blood components on the intestinal epithelium in near 71 physiological conditions. This model, however, should be strategically designed to only include 72 intestinal epithelial cells. By specifically focusing on these cells, we can eliminate potential interference 73 from other components on opposite sides of the barrier. This includes microbiota on the luminal side, 74 as well as immune and stromal cells on the basolateral side. In this way, we can isolate and observe the 75 direct effects of blood components on intestinal epithelial cells, providing us with a more controlled 76 and precise understanding of these interactions. To this end, the development of organoids as 77 advanced primary cell models has revolutionized the design of human tissue models. These self-78 organized, three-dimensional structures, derived from adult stem cells, surpass traditional cell models 79 while recapitulating the organ architecture and plasticity found in vivo. 35 Organoids have become a 80 notable alternative to animal models, offering a more reliable and human-relevant approach for 81 addressing long-standing medical challenges. The relevance of organoids as a model system has been 82 recognized by the FDA that recently approved non-animal models in the drug development process. 36

83
This marks the beginning of a new era in biomedical research. Our research takes a pioneering 84 approach by integrating serum derived from SLE patients into intestinal organoids enabling us to 85 simulate the impact of blood components on the intestinal epithelium while minimizing the 86 interference caused by the intricate intestinal cell milieu, providing a cleaner, more controlled 87 environment for our explorations and analyses. 88 Herein, we report that our organoid model contains all relevant cell types found in colon. We show that 89 SLE serum stimulation can induce alterations in the expression profile, which is dependent on type I 90 interferon signaling, induced through a synergistic effect of all serum contained factors and is specific 91 to SLE. We can show that the secretory lineage is majorly affected, and the mucus layer potentially 92 weakened by alterations in mucus composition, specifically of antimicrobial peptides. A similar impact 93 on the secretory lineage and mucus layer is also observed in Ulcerative Colitis suggesting epithelial 94 barrier dysfunction as hallmark in SLE pathogenesis. Conclusively, our in vitro model emulates a 95 pathological condition where a complex mix of cytokines and other SLE specific factors secreted at the 96 site of inflammation and distributed systemically reaching the highly perfused intestine via the blood 97 can impact the epithelial barrier. This innovative disease model holds the potential to unveil unique 98 insights into disease mechanisms that would otherwise be challenging to investigate. Additionally, it 99 holds the potential as tool to identify innovative treatment approaches and points of intervention. 100 Results 102 1. Organoids show SLE specific response to serum stimulation 103 104 To understand the effect of SLE on the intestinal epithelial barrier, we generated organoid line I and II 105 from the descending colon of two healthy donors, cultured them in conditions that conserved the 106 native cellular diversity 35 and stimulated them for 72h with 5% serum from treatment-naїve SLE 107 patients or sex-and age-matched controls ( Fig. 1A and Table 1). Serum stimulation did not induce any 108 significant phenotypical changes in size or shape (Fig. 1B). There was no evidence for increased 109 cytotoxicity or apoptosis ( Fig. 1C and Suppl. Fig. 1A). Expression profiles of the stimulated organoids 110 were marked by a distinct spread in expression profiles for both stimulation conditions across two 111 tested organoid lines (Fig. 1D). The first principal component (PC1), accounting for 39% of the total 112 variance, displayed a separation of a subgroup of organoids stimulated with SLE sera from the controls 113 in organoid line I. Interestingly, the same SLE sera induced an even more pronounced separation in 114 organoid line II along the PC1 axis. This consistent separation across both organoid lines strongly 115 suggested a donor-independent effect of SLE serum stimulation on epithelial cells. The second principal 116 component (PC2), accounting for 15% or 21% of the total variance, further differentiated the samples 117 within each condition, likely reflecting serum sample-specific responses. 118 Taken together, these results reinforced the inherent complexity of serum stimulated organoids and 119 the need to consider both stimulation and organoid donor-specific effects for downstream analysis and 120 data interpretation. Therefore, we pooled the data from both organoid donors accounting for sex and 121 individual characteristics of the organoid line. In addition, we integrated the response of different cell 122 types given the fact that the cell type composition in both organoid lines varied due to individual 123 proliferation and differentiation dynamics (Suppl Fig. 1B). The combined analysis resulted in 256 124 differentially expressed genes (Fig. 1E). Gene Ontology (GO) analysis of the upregulated genes showed 125 an overrepresentation of terms connected to cell cycle, chromosome organization and replication as 126 well as mitochondrial function and interferon signaling (Fig. 1F). The downregulated genes showed an 127 enrichment of terms connected to secretion, cytoskeleton, and anchoring junctions of the cells (Fig.  128 1G). 129 Intriguingly, the overall results showed that similar pathways were altered in epithelial cells as 130 previously only described in immune cells of SLE patients. 37  We wanted to explore the overrepresentation of terms connected to interferon (IFN) signaling more in 137 depth. Out of the 256 DEG we found 22 genes to be connected to type-I IFN signaling ( Fig. 2A). A 138 significant majority of them were upregulated. They accounted for 32% of the 25 highest upregulated 139 genes underlining the role of type-I IFN in the response induced by SLE serum stimulation. To 140 investigate whether the expression of IFN related genes was altered in both organoid lines by the 141 different patient sera, we selected a range of IFN inducible genes that were represented in several GO 142 terms (Fig. 2B). The representation of their expression in a heatmap revealed an upregulation in all the 143 SLE serum stimulated organoids except for two SLE sera ( Fig. 2B; Suppl. Fig. 2A; Suppl. table 1). On the 144 contrary only two of the control sera showed a similar expression as the SLE sera stimulated organoids 145 indicating the SLE specificity of this response. 146 A common way to analyze the type-I IFN levels and activity in SLE serum samples is to analyze the 147 expression profile of interferon signature genes (IFNSG) in SLE whole blood samples. 38 Given the fact 148 that we were studying epithelial cells we decided to choose the IFNSG based on IFN-α stimulated 149 organoids. Considering only genes relevant in IFNα/β signaling we identified 27 genes which were 150 specific for epithelial cell response to a low dose of IFN-α (Suppl. Fig. 2B). We used this panel of genes 151 (irrespective of their significance of expression) to calculate an IFN score from the normalized counts 152 of SLE and control serum stimulated organoids (Fig. 2C). We could see an overall higher score in 153 organoids stimulated with SLE serum compared to the control irrespective of the organoid line. The IFN 154 score gave us a powerful tool to assess the expression of IFNSG unique to each serum independent of 155 the significance in expression identified by the pooled data. Thus, we were able to integrate and 156 interpret the effect by IFN in a more robust way than just focusing on single gene expression or DEGs 157 which are based on the mean expression. In addition, we ruled out that the seen IFN signature was 158 caused by endogenous expressed IFN by confirming the absence of IFNα1 and IFNβ1 transcripts (Suppl. 159 Fig. 2C). Additionally, we could confirm the activation of IFN signaling by upregulation and 160 phosphorylation of STAT1 which was SLE serum concentration dependent and almost absent in control 161 serum stimulated organoids (Fig. 2D, Suppl. Fig. 2D). Only with SLE serum or IFNα stimulation we could 162 detect IFIT3, Interferon Induced Protein with Tetratricopeptide Repeats 3, abundance (Suppl. Fig. 2D). 163 As a next step we wanted to quantify type I IFN levels of the used serum samples. Since it is challenging 164 to measure it in serum 39,40 , we employed two different approaches analyzing not only the 165 concentration but validating elevated levels by a functional read out. With an IFN-α/β reporter cell line 166 that was stimulated with serum we could detect a significant increase of type I IFN levels in SLE serum 167 (Fig. 2E). We then used a bead-based immunoassay to quantify IFN-α2 levels which confirmed the 168 increased interferon levels in SLE serum (Fig. 2F). To better understand the effects of the serum on the 169 organoids we also checked several other cytokines. We could observe significantly elevated levels of IL-170 18 and the presence of IL-6 in almost all SLE, but not control serum samples, as it has been reported 171 for bigger SLE cohorts 41-45 (Fig. 2G). Even though some of the control sera also showed increased levels 172 of single cytokines only the SLE serum samples showed an overall increase of several cytokines ( Fig.  173 2H). This result did not only indicate the overall inflammatory signature in SLE serum samples but also 174 showed the diversity of the applied samples. 175 Taken together, our data revealed that organoids stimulated with serum derived from SLE patients were 176 characterized by type I IFN signature similar to that of patient derived immune cells. were no differences in mitochondrial mass between both experimental conditions ( Fig. 3C and Suppl. 194 Fig. 3B). This indicated that the upregulated mitochondrial genes were due to an increased 195 mitochondrial activity rather than caused by an increase of mitochondrial mass. 196 We assessed mitochondrial function by a respiratory assay and saw a significant increase of basal 197 respiration and ATP production along with an unaltered maximal respiration upon SLE serum 198 stimulation ( Fig. 3D and Suppl. Fig. 3C). Similar results were seen when CD8 + T cells were stimulated 199 with IFNα. 46 Upon assessing the relative spare respiratory capacity, a notable decrease was observed 200 following SLE serum stimulation, which aligned with previous reports on altered mitochondrial function 201 in CD8 + T cells derived from SLE patients. 46 These results indicated that we were looking at an increased 202 energy demand and a diminished capacity of intestinal epithelial cells to adjust to a dynamic energy 203 demand after SLE serum stimulation. 204 Overall, we could see that the metabolic profile of the organoids was altered upon SLE serum 205 stimulation (Fig. 3E). We were interested in understanding if this alteration was caused by a changed 206 cell type composition of the organoids. Reports from literature suggest that more differentiated cells 207 switch from glycolysis to oxidative phosphorylation as their primary energy source which would be 208 reflected by a shift from the right lower to the left upper quadrant in the energy map. 50-53 However, 209 there is limited knowledge about the metabolic profile of different cell types in human colon. We 210 therefore generated organoids with different cell type composition and observed a metabolic shift with 211 a trend towards higher oxidative phosphorylation upon differentiation (Suppl. Fig. 3D and E). The 212 metabolic shift we observed upon SLE serum differed and showed a trend to higher glycolytic activity 213 compared to the control (Fig. 3E). While the impact of differentiation primarily relied on OCR (oxygen 214 consumption rate), the effects of SLE stimulation influenced both OCR and ECAR (extracellular 215 acidification rate). These results suggested that the changes induced by SLE serum stimulation had a 216 more complex cause than just a shift towards altered differentiated phenotype and might be 217 additionally caused by an overall change in mitochondrial function.  Since the hyperenrichment analysis revealed the terms secretory vesicle and secretory granule 228 amongst the top 7 down regulated gene sets (Fig. 1G, 4B and Suppl. Fig. 4C) indicating an effect on the 229 secretory function of the epithelial barrier we focused on cell type composition changes in the 230 organoids. We checked a panel of cell type marker genes which showed almost exclusively effects on 231 differentiated cells as indicated by a decrease of absorptive and secretory lineage markers (Fig. 4A). Of 232 note, the absorptive cell markers SERPINA1, Serpin Family A Member 1, (also known as AAT) and 233 SCNN1A, Sodium Channel Epithelial 1 Subunit Alpha, (also known as ENaC) are both connected to 234 mucus layer build-up and function. SERPINA1 has antimicrobial functions 56 while SCNN1A is important 235 for ion and fluid regulation 57 thereby playing a role in modifying mucus characteristics. Amongst the 236 secretory cell markers were well known goblet cell markers AGR2, TFF3 and SPINK4 as well as CHGA 237 which marks enteroendocrine cells. AGR2 is essential for the production and processing of gel-forming 238 mucins such as MUC2 58 , whereas TFF3 forms a complex with FCGBP, one of the main components of 239 mucus 59,60 . We were therefore especially interested to understand if we could detect changes in the 240 secretory cells. The number of goblet cells in stained sections was considerable variable between 241 individual organoids, posing challenges for quantification. Analysis of MUC2 staining showed a 242 significant decrease in intensity indicating an alteration in secreted mucus or changes in goblet cell 243 function ( Fig. 4C and D). The quantity of enteroendocrine cells, another type of secretory cell, 244 decreased significantly, as did the amount of CHGA per enteroendocrine cell. (Fig. 4E, F and G). Overall, 245 these results indicated that SLE serum stimulation alters the differentiation dynamics towards the 246 secretory lineage and suggested a change in mucus composition marked by a reduction in MUC2, the 247 major component of mucus, as well as other factors secreted to it. 248

Secretory lineage differentiation is affected 24h after stimulation 249
To better understand the underlying mechanism, we stimulated organoids for 24h with SLE or control 250 serum. This gave us the opportunity to confirm that we were facing an effect on the differentiation 251 process rather than cell type loss. Sequencing analysis indicated that even brief stimulation could 252 trigger a distinct response, evidenced by the separate clustering in the PCA analysis (Fig. 4H). Amongst 253 the DEGs we observed an upregulation of genes connected to IFN signaling and several mitochondrial 254 encoded genes indicating their distinct role in initiating the SLE serum-induced changes (Fig. 4I).

255
Functional analysis showed a stimulation time-dependent decrease of relative spare respiratory 256 capacity similar to that observed after 72 h stimulation ( Fig. 4J and Fig. 3D). Additionally, transcription 257 factors, SPDEF, ATOH1 and HES1, all important for the induction of differentiation, were down regulated 258 indicating a delayed or altered differentiation towards specialized cells (Fig. 4I). 259 In summary, these results support the hypothesis that SLE serum stimulation induced alterations in 260 differentiation, particularly impacting the secretory lineage. Short-term serum stimulation induced 261 changes in transcription factor expression necessary for differentiation which upon long-term 262 stimulation leads to alterations in cell type composition of the organoids. Ultimately, the consequence 263 will accumulate in an altered mucus layer that in vivo has the potential to trigger alterations in the 264 protective function of the mucus and influence the microbiome. 265 6. Effects on epithelial cells depend on IFNAR1, although not being 266 exclusively linked to IFNα activity.

267
The importance of type I IFN in SLE pathogenesis and its potential as target for treatment was 268 highlighted by the approval of anifrolumab, a type 1 interferon receptor (IFNAR1) antagonist in 2021. 61 269 Our novel disease model was able to show the involvement of type I IFN in altering epithelial cell 270 signatures as suggested by the increased IFNSG expression and activation of IFN signaling cascade ( Fig.  271 2A-E). To verify this, we utilized the inhibitory potential of anifrolumab on IFN signaling. We validated 272 that anifrolumab was able to inhibit phosphorylation of STAT1, a downstream target of IFNAR1 273 activation (Suppl. Fig. 5A) by organoid stimulation with up to 100 ng/ml IFNα2b. Analysis of the 274 expression profile showed no significant difference of organoids treated with anifrolumab in addition 275 to serum comparing SLE to CTRL serum stimulation (Fig. 5A). To further validate the successful 276 inhibition of IFNAR1 signaling we checked normalized counts of DEGs specific for epithelial IFN signaling 277 identified before ( Fig. 5B and 2B). We could see that anifrolumab was able to abolish the increased 278 expression of these genes not only in IFN stimulated organoids but more importantly also in 279 combination with SLE serum stimulation making them indistinguishable to the control condition ( Fig.  280 5B). As a next step we wanted to assess if anifrolumab was also able to revert the mitochondrial 281 dysfunction. The respiratory assay confirmed that with blocking IFNAR1 SLE serum stimulated 282 organoids had a similar relative spare respiratory capacity as in the control condition (Fig. 5C). 283 Additionally, to confirm that the effect we were looking at was highly specific to SLE, we analysed the 284 effect of serum from patients suffering from another systemic autoimmune disease, granulomatosis 285 with polyangiitis (GPA). We could show that organoids stimulated with serum from GPA patients, a 286 disease not driven by type I interferon, showed no distinctive expression pattern (data not shown) and 287 no mitochondrial dysfunction underlining the dependency on type I interferon (Suppl. Fig. 5B). 288 To confirm that the effect we observed was specific to SLE serum, rather than simply a consequence of 289 IFN independently, we stimulated organoids with IFNα in combination with control serum (IFNα-290 serum). However, this stimulus was not able to induce the same complex response as seen for SLE 291 serum. Overlapping the DEGs from both comparisons, SLE compared to control serum and IFNα-serum 292 compared to IFNα-serum+anifrolumab, showed only a small overlap of almost exclusively IFN related 293 genes highlighting the unique response of the organoids to SLE serum (Suppl. Fig. 5C). 294 Collectively, these findings suggest that the response of the organoids to SLE serum is dependent on 295 IFNAR1, but not caused solely by its activation by IFNα. This implies that the unique composition 296 present in the SLE serum plays a significant role in provoking the response, in conjunction with the 297 dependency on IFNAR1. 298 299 7. Colon organoid Single cell sequencing reveals the presence of all 300 major colonic cell types found in vivo 301 To gain a deeper understanding of the cell types that could potentially exhibit higher sensitivity to 302 serum stimulation, as well as to examine the cellular composition of the organoids, we conducted 303 single-cell RNA sequencing. This approach allowed us to analyze expression patterns of individual cells 304 and gain insights into their specific responses to serum stimulation. We could assign nine clusters to 305 the cells analyzed which showed a distinct cell cycle distribution and expression profile specific to their 306 cell type ( Fig. 6 A-C; Suppl. Fig. 6A and B). Trajectory analyses identified lineage transitions and showed 307 connectivity between progenitor and differentiated populations ( Fig. 6D and Suppl. Fig. 6C). As 308 anticipated, stem cells gave rise to transit-amplifying cells (TA1, TA2 and TA3), the secretory lineage 309 (GC) and the absorptive lineage comprising three colonocyte clusters (eCL, cCL and ncCL). 310 We identified two distinct clusters of stem cells (SC1 and SC2) marked by the expression of classical 311 stem cell markers LGR5, OLFM4, RGMB, and SMOC2 (Fig. 6A, B). The key characteristic that sets apart 312 these two clusters is their distinctive cell cycle distribution. While the majority of cells in SC2 are in S 313 phase, a significant portion of cells in SC1 are in G1 phase (Fig. 6C). Transition to cycling TA cells (TA1) 314 was marked by the transition to exclusively G2/M phase and an enrichment of the proliferation markers 315 MKI67 and TOP2A expression ( Fig. 6B and C). Furthermore, we identified two additional TA clusters, 316 TA2 and TA3 (Fig. 6A). All three TA clusters showed gene-expression gradients along with distinct cell 317 cycle distributions reflecting active proliferation and transition to progenitors of absorptive and 318 secretory lineage ( Fig. 6B and C; Suppl. Fig. 6C). Cell transition to G1 phase marked all identified 319 differentiated cell types, early colonocytes (eCL), canonical colonocytes (cCL), non-canonical 320 colonocytes (ncCL) and goblet cells (GC) (Fig. 6C). 321 Interestingly, we could identify two different types of colonocytes emerging from early colonocytes 322 (Fig. 6D). Canonical colonocytes were marked by the expression of classical absorptive markers SLC26A3, CA2 and FABP1. These markers were absent in the non-canonical colonocytes which were 324 marked by high expression of MMP7, LYZ, IL32 and the IFN response genes ISG15, ISG20 and IFI27 ( Fig.  325 6B). Even though MMP7 and LYZ were lately described to be present in deep crypt secretory cells 62,63 , 326 our cell population lacked other reported markers of these newly described cells. In addition, we could 327 observe the expression of HES1 an exclusive marker of the absorptive lineage (Fig. 6B). 328 Since the clustering showed two distinct populations, we wanted to understand the major features 329 separating them. Analysis of the significantly enriched markers (padj<0.1 and log2fc >0.25) suggested 330 different functions of the two cell types. Canonical colonocyte showed an enrichment of lipid 331 metabolism related pathways and expressed several ion transporters. Specifically SLC26A3 and CA2 332 which in vivo are expressed by fully mature colonocytes, the central players of absorption 13 suggesting 333 that also in vitro their function would focus on lipid processing and absorption. In contrast, we could 334 observe an enrichment of pathways connected to immune defense in non-canonical colonocytes 335 indicating cytokine-driven responses to maintain barrier integrity (Fig. 6E). and SPDEF all known to be mucus core proteins or genes involved in mucus biosynthesis 64 (Fig. 6G). In 343 contrast to that cluster GC2 was characterized by a higher expression of ZG16 and TFF3 which function 344 as AMP and regulator of mucus viscosity respectively 65,66 . Cluster GC4 showed only low expression 345 levels of typical goblet cell markers. With further analysis we were able to identify a small fraction of 346 cells within this GC4 expressing CHGA and POU2F3 indicating the presence of EECs and Tuft cells 347 respectively (Suppl. Fig 6F). Interestingly, similar as reported previously for human colon tissue 67-69 we 348 were able to see differences in mucin gene expression patterns between the goblet cell cluster 349 indicating their distinct functionality also in vitro (Fig. 6H). or AQP5 which were significantly downregulated (Fig.7A). However, we could see that each cell type had a specific response to SLE serum stimulation. In SC1, but not in SC2, stem cell markers SMOC2 and 369 OLFM4 were significantly downregulated. In contrast, in canonical colonocytes typical absorptive 370 markers were unaltered while tight junction proteins TJP1 and CLDN7 were downregulated. Both are 371 important for barrier integrity and their absence is associated with a delay in mucosal repair or an 372 initiation of inflammation in vivo. 72,73 This specifically suggests an alteration in the epithelial barrier and 373 lays the ground for a self-perpetuating chronic inflammation a seen in SLE. 374 Next, we wanted to understand if we could detect a change in goblet cell differentiation as indicated 375 by the bulk RNA sequencing data. We observed a noticeable decline in the levels of goblet cell markers 376 such as TFF3, ZG16, MUC2, and FCGBP within the TA population (TA1, TA2, and TA3) (Fig. 7B). This 377 suggested that the stimulation with SLE serum led to diminished differentiation towards the secretory 378 lineage, which in turn suggested a potential reduction of cells differentiating towards goblet cells. Our 379 findings were corroborated by the results from the bulk RNA sequencing analysis, wherein we noticed 380 a reduction in the expression of early goblet cell markers during the initial phase of stimulation (24h), 381 and a subsequent decrease in late secretory markers with extended stimulation (72h) (Fig. 4A, B, I). 382 Consistent with these results, we could detect a minor reduction in the number of cells in the goblet 383 cell cluster ( Fig. 7C and Suppl. Fig. 6A). Interestingly, within the diminished goblet cell population we 384 observed a significant upregulation of FCGBP, an IgGFc-binding protein, which is crucial for structural 385 integrity of the mucus layer 60,74 . Additionally, WFDC2 which was only recently identified to be required 386 for barrier integrity and is downregulated in goblet cells of UC patients 75 was also downregulated upon 387 SLE serum stimulation (Fig. 7D). Further analysis showed generalized trend towards a decrease across 388 all goblet cell subclusters, with a significant reduction in GC4 (Suppl. Fig. 7C). GC4 lacked the expression 389 of classical goblet cell markers and contained CHGA expressing cells suggesting a reduction in 390 enteroendocrine cells as identified by staining (Fig. 4F). Additionally, GC4 showed no DEGs while the 391 DEGs of the other GC clusters were mainly downregulated ribosomal genes connected to translation 392 and protein translocation. This might indicate an impact on goblet cells function that highly depends 393 on a constant translation and protein trafficking. Apart from the reduced number of goblet and 394 enteroendocrine cells, these findings suggest an altered mucus synthesis. 395 Since mucus properties can easily be modified by an altered abundance of mucins and AMPs 76,77 which 396 is not restricted to goblet cells we checked known markers in the absorptive lineage. In the early 397 colonocyte population we observed a downregulation of MUC12 a membrane bound mucin which is 398 part of the protective glycocalyx 12 . In addition, we observed an increase of lysozyme (LYZ) and decrease 399 in REG4 in the canonical colonocytes (Fig. 7D). Both are known to be secreted into mucus and are 400 important for microbial defense 78,79 . A similar alteration was also evident within the non-canonical 401 colonocytes where we observed an alteration in the expression of MMP7, RETNLB and PLA2G2A, genes 402 that are essential components of barrier function, especially RETNLB and PLA2G2A with their 403 bactericidal properties 80-83 (Fig. 7D). Additionally, in non-canonical colonocytes and GC2 we were able 404 to detect the induction of IFI27, an interferon inducible gene, similar as reported for whole blood 405 samples from SLE patients 84-86 (Fig. 7D) and as observed by the induction of IFNSG in our bulk RNA 406 sequencing data (Fig. 2C). Interestingly, non-canonical colonocytes and GC2 both expressed AMPs 407 which are known to be altered by proinflammatory cytokines 87-89 suggesting a similar connection in 408 our model. 409 Overall, these results revealed that SLE serum stimulation of epithelial cells lead to distinct To highlight the relevance of this disease model and provide context regarding intestinal involvement 419 in SLE, we conducted an expression analysis of intestinal biopsies derived from descending colon of 420 healthy controls (n=5) and SLE patients (n=3). We could identify 256 DEGs which were representing the 421 integrated response of all present cell types in the biopsies. In our further analysis we focused on genes 422 that were expressed by epithelial cells. Overall, we could observe a downregulation of several genes 423 connected to absorption and ion transport (SLC1A1, SLC13A2, SLC16A9, SLC25A20, SLC25A34). BEST2, 424 a marker for goblet cells, was significantly downregulated while markers for enteroendocrine cells 425 (SGC5, GCG, VWA5B2) were upregulated suggesting an alteration in the secretory lineage (Fig. 7E). 426 Additionally, we observed reduced expression of ion transporters SLC26A3, SCNN1A and SCNN1B while 427 other markers for colonocytes were unaltered. This indicated an alteration in cell function rather than 428 a loss of colonocytes. Those three ion transporters are the major players in water absorption in 429 descending colon and are able to modulate mucus viscosity, ion content and ultimately alter its 430 structural composition via changes in sodium absorption and bicarbonate secretion influencing the 431 mucosal pH. 13,90 In addition to that several factors important in the maintenance of the intestinal stem 432 cell niche and differentiation (WNT5A, RSPO2, GREM1, BMP5) were significantly downregulated 433 suggesting changes in the physiological proliferation and differentiation dynamics. We then checked 434 the expression of the top secretory cell markers identified with single cell sequencing. Surprisingly, we 435 could see clustering of two SLE patient samples that showed a reduction in the selected markers ( Fig.  436 7F). 437 The evidence we have gathered collectively suggests that the impact of SLE is not confined to immune 438 cells, but also affects intestinal epithelial cells. data from patients included in this study were provided also by the I3PT Biobank and the DTI 465 Foundation (Barcelona, Spain) they were processed following standard operating procedures with the 466 appropriate approval of the Ethics and Scientific Committees. 467 All research procedures were conducted adhering to the principles stipulated in the WMA Declaration 468 of Helsinki. All participants in the study gave written informed consent before their inclusion and were 469 assured that their identities would remain confidential in relation to this study. 470 11. Organoid generation and culture 471 Crypt isolation was performed as previously published 91 . The mucosa of approximately 1 cm 2 was 472 separated from the underlying tissue and cut into small pieces, washed with cold PBS before incubation 473 in 10 mM EDTA/PBS for 1 h at 4°C under constant rocking. The biopsies were then transferred to 5 ml 474 cold PBS and crypts were liberated by pipetting 20 times. This was repeated twice, and the pooled 475 fractions were centrifuged at 250g for 5 min at 4°C. The pellet was once washed in base medium (Suppl. 476 Table 2) before the crypts were resuspended in 45 µl 80% Matrigel (#356231, Corning) in expansion 477 medium (Suppl. Table 2) and plated in small drops in a 24-well plate. After polymerization 500 µl growth 478 medium (Suppl. 300 µm diameter. After establishing the organoid line organoids were passaged at a ratio of 1:6-1:10 481 every 4-5 days by a 45s incubation in Accutase at 37°C followed by mechanical dissociation in 6 ml base 482 medium with a 10 ml pipet equipped with a 200 µl pipet tip. Organoid fragments were centrifuged at 483 150g for 5 min at 4°C and plated as stated above. Medium was changed every 2-3 days. 484

Organoid stimulation 485
Organoids were passaged as described above with the modification of excluding bigger sized fragments 486 by using a 70 µm strainer after the dissociation. In addition, fragments were counted and 400 487 fragments/µl were seeded. Organoids were grown in expansion medium for four days before 488 dissociating them into single cells by incubation for 90 s in Accutase followed by mechanical 489 dissociation as described above. Bigger fragments were removed by using a 40 µm strainer. Then 800 490 cells/µl were seeded in 10 µl Matrigel mix in 96-well plate. After polymerization 100 µl growth medium 491 were added containing Y-27632. Organoids were stimulated with 5% serum or 5% control serum in 492 stimulation medium (Suppl.

505
Organoids from two wells were pooled in base medium. The organoid pellet was resuspended in 100 µl 506 harvesting solution (Cultrex, Biotechne) and incubated for 45 min at 4°C. The released organoids were 507 washed with ice-cold PBS, then 4% PFA was added, and organoids were fixed for 45 min at 4°C followed 508 by a final wash. The organoids were then resuspended in 4% low-melt point agarose. Paraffin 509 embedding was performed following standard protocols. For immunohistochemistry 3 µm sections 510 were rehydrated, heat-induced antigen retrieval in 10 mM sodium citrate acid and 0.05% Tween 20 (pH 511 6) and quenching 50 mM NH4Cl was performed. Sections were then blocked with 5% goat serum for 1 512 h. Primary antibodies anti-cleaved

Serum analysis 538
Serum samples were analyzed using a bead-based multiplex assay (BioLegend, LEGENDplex, 740809) 539 following the manufacturer's protocol. 5000 events (bead populations A + B) were recorded using a BD 540 FACS Aria Fusion flow cytometer. Cytokine concentration was calculated based on a standard curve 541 using Biolegend's LegendPlex data analysis software. Cytokine levels that were below the detection 542 limit were set to half the detection limit to calculate statistical significance. Cytokine levels for each 543 sample can be found in Suppl.

551
Oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) were measured using 552 Seahorse XFe96 Analyzer (Agilent) and mitochondrial stress test was performed as previously 553 described 97 . Organoids were imaged before the assay, total area determined using the OrgaQuant 554 tool 92 and then used for normalization. 555 21. Bulk RNA sequencing 556 Total RNA was isolated from stimulated organoids using RNeasy Micro Kit (Qiagen) and quantified using 557 Qubit High Sensitivity Assay (Invitrogen). 558

559
The process of bulk RNA sequencing was carried out at Novogene using an Illumina Novaseq HiSeq 560 Pair-Ended 150bp, with a sequencing depth of 9G, equivalent to 30 million reads for each sample. 561 Similarly, single-cell RNA sequencing (scRNAseq) was executed, achieving a sequencing depth of 90G 562 for every sample. 563 23. Bulk RNA-seq data analysis 564 RNA-seq data was preprocessed using the nf-core/rnaseq pipeline (version 3.8.1) 98 . In brief, reads were 565 aligned using STAR 99 to reference genome version GRCh38 with genome annotation version GRCh38 566 (release 106, Ensembl), and gene expression was quantified using Salmon 100 . 567 Differentially expressed genes were identified using DESeq2 101 . Genes were included in differential 568 expression analysis if they were detected with more than 1 count-per-million in at least n samples, 569 where n is the number of samples in the smallest group of samples in the comparison. An adjusted p-570 value (FDR) < 0.1 was applied as threshold to consider genes significantly differentially expressed. 571 similarity of terms based on shared genes, a distance matrix was calculated containing the pairwise 584 distances between terms. As distance metric, binary distance was used, i.e., the proportion of non-585 shared genes amongst the union of genes from a given pair of terms. The resulting distance matrix was 586 used as input for multidimensional scaling to visualize the similarity of terms in two dimensions. 587

588
IFN score was calculated as z-score-based standardized score 109 , using the transformed count data 589 obtained through variance stabilizing transformation (DESeq2) as input and the following genes as and fragment size confirmed using using High Sensitivity DNA Kit (Agilent). 598

Single-cell RNA-seq data analysis 599
Single-cell RNA-seq data was preprocessed using the nf-core/scrnaseq pipeline (version 2.2.0) 105 . In 600 brief, read alignment, filtering and counting was performed using cellranger using a reference package 601 created from reference genome version GRCh38 with genome annotation version GRCh38 (release 106, 602 Ensembl). Further analysis steps were performed using Seurat 106 . Genes that were detected in fewer 603 than 4 cells (from the whole dataset consisting of 34255 cells) were excluded. Cells were included in 604 further analysis if the number of detectable genes was more than 200, the percentage of mitochondrial 605 reads was less than 25%, and the percentage of ribosomal reads was more than 5%. Doublets were 606 excluded using DoubletFinder 107 . 607 For further downstream analyses using the Principal Components (e.g., UMAP, nearest-neighbor graph 608 construction for clustering), the first 60 Principal Components were considered. For further 609 downstream analyses with a parameter for number of nearest neighbors (e.g., UMAP, nearest-neighbor 610 graph construction for clustering), a value of 300 and 50 was used for the analysis on all cells and Goblet 611 cells, respectively. For clustering, a cluster resolution of 1 and 0.5 was used for the analysis on all cells 612 and Goblet cells, respectively. For differential expression between two groups of cells, Wilcoxon Rank 613 Sum test was used and genes with an adjusted p-value < 0.1 (Bonferroni correction) were called 614 significant. Trajectory analysis was performed using the Slingshot package 108 with cluster 0 set to be 615 the cluster of origin (parameter "start.clus"). 616

Statistical analyses 617
All statistical analysis not connected to bulk or single-cell RNA sequencing were performed in GraphPad 618 Prism 9 Software (GraphPad Software, Inc.). Analyses were performed with unpaired t-test or ANOVA 619 with Holm-Šidák's multiple comparisons test. In the case of organoid size an average of 100 organoids 620 was calculated and the resulting values were compared using chi-square test. To calculate statistical 621 significance for the cytokine measurements undetected cytokines were considered as half of the 622 detection limit. If not stated with exact p values, p values below 0.05 were considered significant and 623 mean with standard deviation is shown 624 whereof one could be assigned to colonocyte progenitors (eCL) and the other two as differentiated 644 colonocytes. Among these two distinct clusters, we discovered a canonical colonocyte population that 645 was highlighted by the expression of markers such as SLC26A3, CA2, and FABP1. Interestingly, the 646 second colonocyte population was marked by a unique expression profile with increased expression of 647 REG4, MMP7, LYZ and the cytokines IL18, IL23, IL33 (Suppl . Table 10). For these genes it has been shown 648 that they can be induced by pro-inflammatory cytokines in vitro or in vivo. 12,80,82,115-118 In addition, we 649 could observe the presence of interferon inducible genes IFI27, ISG15 and ISG20. These findings 650 indicate that the expression profile of this non-canonical colonocyte population was shaped by the 651 cytokine cocktail contained in the serum and identifies a cell population with a specialized function 652 connect to immune defense. Deeper analysis is needed to classify the relevance and function of this 653 cell type in vivo. Further studies will be required to fully understand the implications of these findings 654 and their possible relevance in the generation of complex in vitro models. Nevertheless, given that 655 most cell types of the descending colon were represented in our model, we could utilize it to study the 656 effects of SLE on the intestinal epithelial barrier. 657 SLE is a multifaceted autoimmune disease which central player is IFN-α 119-121 . The induction of IFNSG 658 in immune cells has been used as a reliable tool to identify increased serum IFN-α levels as it is the case 659 for the majority of the patients. 46 Our study shows that serum of SLE patients is able to induce a similar 660 IFNSG in epithelial cells in vitro. Thus, intestinal organoids represent a novel tool to assess type-I IFN 661 levels in patient serum. Additionally, we could show that changes induced on the epithelial cells are 662 dependent on type I IFN signaling and can be inhibited by the co-stimulation with anifrolumab an 663 inhibitor of IFNAR1. We further confirmed type I interferon dependency by the absence of DEGs after 664 the stimulation of organoids with serum derived from Granulomatosis with Polyangiitis patients. 665 However, IFN-α in combination with control serum was not able to reproduce the phenotype induced 666 by SLE serum indicating a synergistic effect of type I IFN with other cytokines contained in SLE serum.

667
IL-6, IL-10, IL-18 and IL-23 levels were elevated, and their receptors expressed by the organoids, hence 668 the observed response reflects most likely the integration of all their effects. It was impressive to see 669 that cytokines with a concentration <3 pg/ml were able to induce such a pronounced phenotype 670 demonstrating their potency. In addition, this underlines the artificial environment created using high 671 levels of a cytokine or even a cytokine cocktail for stimulation questioning their relevance in modeling 672 pathological conditions in vitro. While they have their relevance in understanding underlying signaling 673 pathways, novel near-physiologic in vitro models like the one developed in this study will be of high 674 importance in the future to mimic complex diseases. Our model provides intriguing avenues for 675 exploration into the pathogenesis of SLE. 676 So far the focus of SLE research has been the dysregulation of the immune system, but studies in mouse 677 intestine (inclusive our own non-published data) and lung cells show that epithelial cells can be affected 678 as well by type I IFN leading to detrimental changes in the barrier. 54, 55 We could show here that this 679 holds also true for human intestinal cells. Additionally, we observed that mitochondrial pathways and 680 function in the intestinal organoids were affected. SLE serum stimulation caused an increased 681 metabolic activity marked by higher basal respiration and reduced relative spare respiratory capacity 682 highlighting the capacity of serum to induce metabolic changes. A similar mitochondrial dysfunction 683 was recently also reported in CD8+ T cells derived from SLE patients. 46 In addition, our data suggested 684 that the alteration induced by SLE serum was different from the metabolic shift induced by 685 differentiation. We could see that differentiation led to an increase of oxidative phosphorylation, even 686 in the absence of external energy source promoting this shift in vivo 52,71,122 . In contrast SLE serum 687 stimulation induced only a slight increase in oxidative phosphorylation but was more dominated by an 688 increase in glycolysis. In UC patients a similar shift towards increased glycolytic activity is accompanied 689 by an impaired goblet cell differentiation. In contrast SLE serum stimulation induced only a slight 690 increase in oxidative phosphorylation but was more dominated by an increase in glycolysis. This shift 691 could mark an overall alteration of mitochondrial function or might reflect the altered cell type 692 composition of the organoids. Since in UC patients a similar shift towards increased glycolytic activity 693 is accompanied by an impaired goblet cell differentiation 71 . Thus, this shift could mark an overall 694 alteration of mitochondrial function or might reflect the altered cell type composition identified upon 695 SLE serum stimulation. 696 We were intrigued to see that the most common alteration which was persistent in all our experiments 697 was the involvement of the secretory cell lineage. The downregulation of several markers connected 698 to secretion, goblet and enteroendocrine cells identified with bulk RNA sequencing aligned with our 699 results from microscopy and single-cell RNA sequencing which revealed a diminished goblet and 700 enteroendocrine cellpopulation and a decrease or delay in the differentiation towards secretory cells. 701 Additionally, our results indicated a functional change of differentiated cells resulting in an altered 702 mucus layer. Those changes were not restricted to the goblet cell population but extended to the 703 absorptive cell lineage. We classified numerous misregulated genes, many of which were integral to 704 mucus composition (AGR2, SPINK4, MUC12, FCGBP) and antimicrobial defense (PLA2G2A, RETNLB, 705 REG4, TFF3, and WFDC2). Our data aligned with the altered expression pattern of several mucus related 706 genes and the reported metaplastic, colonic expression of LYZ associated with UC 707 pathogenesis 12,13,65,123-125 . We observed downregulation of MUC12, a membrane-bound mucin that has 708 a protective role for colonocytes. 126 FCGBP, important for mucus integrity 12,123,124,127 , and MMP7 which 709 is associated with LPS-induced barrier permeability and connected to SLE pathogenesis 81,128 itself were 710 upregulated. The additionally identified downregulation of the sodium channel SCNN1A and the 711 aquaporin AQP5 suggests an alteration in ion composition and hydration of the mucus 57,129 . With 712 scRNA-seq we were able to identify downregulation of tight junction proteins CLDN3, CLDN7 and TJP1 713 that indicate a change in intestinal epithelial permeability. Furthermore, we could show that non-714 canonical colonocytes and a subcluster of goblet cells were specifically responsive to IFN contained in 715 the serum by the expression of IFI27 indicating a cell type specific response. 716 These findings show that SLE serum stimulation led to an alteration in both the secretory and 717 absorptive lineage that resulted in a potential structural weaking of the mucus barrier. Not only the 718 differentiation to goblet cells and therewith a reduction in their number was affected but also several 719 secreted factors necessary for a functional mucus layer. The alteration in expression of AMPs suggested 720 changes in intestinal antimicrobial capacity of the mucus which is pivotal to maintain the sterility of the 721 inner mucus layer and therewith reducing the microbial-epithelial interaction in vivo. 130 It is important 722 to mention that in UC it is suggested that mucus alterations are causative for disease pathogenesis and 723 precede the active disease, indicating the importance of the mucus layer for barrier function and 724 intestinal homeostasis. 13 Also other diseases like cystic fibrosis highlight the importance of a 725 physiological mucus composition in order to maintain organ function and underline that even minor  control or SLE serum respectively). F: IFN-α2 levels in SLE serum confirmed via multiplex ELISA, 783 validating the increased interferon levels. Grey line shows ½ the detection limit. n=10 and n=8 for 784 control or SLE serum respectively. G: Multiplex ELISA data highlighting elevated IL-18 and IL-6 levels in 785 SLE patient sera compared to controls. In only one control IL-6 was detectable compared to six out of 786 eight SLE serum samples. Half the detection limit concentration was used for calculation if the cytokine 787 was undetectable. Grey line shows ½ the detection limit. n=10 and n=8 for control or SLE serum 788 respectively. H: Radar plot illustrating the distinct cytokine composition in each serum sample, 789 quantified by Multiplex ELISA. Each line color represents one cytokine, levels are shown as fraction, 790 and normalized to the highest and lowest value for each cytokine. Green and pink labels represent 791 control and SLE samples respectively. Unless otherwise indicated, an unpaired t-test was used for 792 statistical analysis. Unless otherwise specified, all experiments depicted in this figure were analyzed 793 with n= 10 control serum (samples a to j) and n=8 SLE serum (samples 1 to 8). Energy map demonstrating altered metabolic profile of organoids following SLE serum stimulation, with 808 a shift toward higher glycolytic activity. Each dot corresponds to the average of two technical replicates 809 of stimulated organoids, n= 10 control serum and n=8 SLE serum. Statistical tests performed with 810 unpaired t-test. 811 organoid normalized to the number of nuclei. Each dot corresponds to a single organoid, revealing a 818 decrease in mucus intensity. Control n=29 organoids and SLE=40 organoids analyzed using unpaired t-819 test. D: Fluorescence microscopy images of organoids after 72h stimulation with either control serum 820 otherwise specified, all experiments depicted in this figure were analyzed with n= 10 control serum 839 (samples a to j) and n=8 SLE serum (samples 1 to 8). 840 with anifrolumab in addition to SLE or CTRL serum stimulation. The plot shows no significant difference 843 in expression profiles between these treatment groups, emphasizing the role of type I interferon and 844 inhibition of IFNAR1 signaling. n= 4 control serum + anifrolumab (samples a, c, d and g) and n=4 SLE 845 serum (samples 2 to 5   genes were identified as unique to the epithelial cell response to a low dose of IFN-α, which were 901 completely downregulated when blocking IFNAR1 with anifrolumab. n=3 IFNα+control serum, and n=3 902 IFNα+control serum+anifrolumab. C. Graph depicting normalized counts for IFNα-1 and IFNα-2 in 903 serum stimulated organoids of line I and II. Absence of main interferon transcripts validates that the 904 observed IFN signature was not due to endogenous IFN expression. n= 10 control serum (samples a to 905 j) and n=8 SLE serum (samples 1 to 8). D. Western blot illustrating the levels of STAT1 and IFIT3 in 906 organoids from SLE and control conditions after 48h of stimulation at two different serum 907 concentrations (5% and 30%). In the IFNα-stimulated organoids, STAT1 and IFIT3 levels were assessed 908 after 1h and 48h of stimulation at a concentration of 100ng/ml. Each sample represents two pooled 909 wells of organoids stimulated with serum from one donor. 910 Suppl. Fig. 3  organoids. Unpaired t-test was performed. 965