Assessing Campylobacter jejuni Extracellular Vesicle–Host Interaction Using a Microfluidic Platform with Caco-2 Spheroides-on-Chip

Assessing Campylobacter jejuni Extracellular Vesicle-Host Interaction Using a Microfluidic Platform with Caco-2 Spheroides-on-Chip Silvia Tea Calzuolaa,b*, Jeanne Malet-Villemagnec, Debora Pinamontid, Francesco Rizzottoc, Céline Henryc, Christine Péchauxe,f, Jean Baptiste Blondéb, Emmanuel Royb, Marisa Manzanod, Goran Lakisicc, Sandrine Truchetc, Jasmina Vidicc* a UMR7646 Laboratoire d'hydrodynamique (LadHyX), Ecole Polytechnique, Palaiseau, France b Eden Tech, Paris, France c Micalis Institute, INRAE, AgroParisTech, Université Paris-Saclay, Jouy en Josas, France d Department of Agriculture Food Environmental and Animal Sciences, University of Udine, 33100, Udine, Italy e Université Paris-Saclay, INRAE, GABI, Jouy en Josas, France. f MIMA2 Imaging Core Facility, Microscopie et Imagerie des Microorganismes, Animaux et Aliments, INRAE, Jouy en Josas, France. Corresponding Authors * Silvia Tea Calzuola – Email: [email protected]; * Jasmina Vidic – Email: [email protected] https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 2 ABSTRACT Campylobacter jejuni is a foodborne pathogen that adheres to and invades the epithelial cells of the human intestinal tract. The extracellular vesicles (EVs) of C. jejuni have an important impact during pathogenicity but their role in invasion of the host intestinal epithelial cells remain largely unknown. In vitro models lack the complexity of tissue and fail to accurately replicate the dynamic interactions between EVs and human intestinal epithelial cells, while animal infection models have species-specific differences that limit their translational relevance and are associated with ethical concerns. To bridge this gap, we propose a microfluidic platform integrated with an impedimetric sensor to monitor C. jejuni EV interactions with human intestinal epithelial Caco-2 cells. When cultured in this microfluidic device, Caco-2 cells underwent spontaneous 3D morphogenesis and were spatially organised into spheroid-like structures. By combining impedance spectroscopy and microscopic observation, the platform allowed direct monitoring of cellular spatial growth and sensitive detection of their interaction with EVs. Although functional assays revealed that the C. jejuni EVs killed Caco-2 cells cultured on plates, 3D Caco-2 spheroids showed significantly greater resistance to both C. jejuni secretome and EVs. The physiological relevance of the model was confirmed by TEER measurements that suggested that Campylobacter EVs difused paracellularly. The developed microfluidic device is a promising platform for investigating host‒microbe interactions and may have a broad impact on biomedical research on gastroenteritis. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 3 Introduction Campylobacter is the most common zoonotic cause of diarrheal diseases worldwide. Species from the genus Campylobacter are gram-negative, microaerophilic spiral-shaped polar rods whose characteristic corkscrewlike motility is mediated by an unsheathed flagellum at one or both ends. The most frequently reported Campylobacter species associated with human diseases are Campylobacter jejuni and Campylobacter coli, with C. jejuni responsible for over 80% of Campylobacteriosis 1 . The main source of contamination is undercooked poultry meat, followed by unpasteurized milk, unwashed vegetables, and water. Contamination of meat occurs mainly from the fecal content to the meat during the slaughtering process because Campylobacter colonizes the gastrointestinal tract of birds and warm-blooded mammals (such as pigs and sheep) without causing any symptoms 2 . In contrast, the ingestion of only 500 C. jejuni cells is sufficient to trigger disease in humans 3 . The clinical symptoms of human Campylobacteriosis include diarrhea (frequently bloody), abdominal pain, fever, headache, nausea, and vomiting, which can last up to two weeks. Severe complications, such as Guillain–Barré syndrome, inflammatory bowel disease, and arthritis, occur in approximately 1% of cases 4 . Although C. jejuni is one of the most common foodborne pathogens worldwide, the mechanisms of its pathogenesis and host colonization have not yet been fully clarified, primarily due to difficulties in establishing relevant animal models. Indeed, unlike humans, animals are naturally resistant to C. jejuni colonization and disease, and in vivo tests often fail to accurately predict clinical responses during Campylobacteriosis 5, 6 . Alternatively, in vitro host-microbe cocultures provide practical platforms for investigating specific questions. The investigation of Campylobacter infection in conventional cell culture models, such as plates and Transwells, has suggested that several virulence factors are involved in bacterial-host interactions 7-11 . The impact of C. jejuni extracellular vesicles (EVs) and their protein cargo on epithelial cells has been of particular interest 11-14 because the bacterium lacks classical secretion systems and relies on vesicles to secrete biomolecules 6 . However, 2D platforms cannot allow dynamic studies and are ineffective in explaining how bacterial cells or their vesicles cross the epithelial barrier to invade the host. Recent studies have shown that forces such as shear stress and mechanical strain in the context of host-microbe interactions play crucial roles in bacterial pathogenicity 15-17 . These effects are difficult to assess via conventional 2D in vitro models. Therefore, there is strong interest in developing alternative experimental 3D platforms to mimic key aspects of the human gut epithelium for studying C. jejuni diseases. Organs-on-a-chip are microfluidic cell culture devices designed to replicate organ-level physiology by reconstructing tissue-tissue interfaces, mechanical signals, fluid flow, and the biochemical cellular microenvironment 18 . Microfluidic technologies provide a significant advantage in bridging the gap between in vitro experimental models and in vivo pathophysiology. For instance, the topology and peristaltic cycles simulated by guts-on-a-chip are essential for reproducing the virulence of Shigella flexneri infections reported in clinical studies 19 , as well as the infection dynamics of Entamoeba histolytica 20 . Through a similar organ-onchip model, Kim et al. 21 showed that peristaltic deformation is the main mechanism responsible for preventing the uncontrolled growth of commensal bacteria in the gut, aligning with clinical observations 22 . Here, we present a microfluidic platform with integrated microelectrodes that allows investigation of Campylobacter EV interactions with human intestinal epithelial cells. To achieve this goal, we designed a twochannel, user-friendly microfluidic system. When Caco-2 cells were cultured on one side of the porous membrane, they underwent spontaneous 3D morphogenesis. C. jejuni secretome was characterized for the presence of virulence factors, and EVs were purified using density gradient ultracentrifugation. By comparing the cytotoxicity of EVs secreted by C. jejuni on Caco-2 cells in 2D in vitro models and within a https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 4 microfluidic device, we demonstrated the crucial role of spatial organization of epithelial cells in order to withstand bacterial EV-secreted virulence factors. Materials and methods Materials and reagents Flexdym TM was supplied by Eden Tech (France). Poly(methyl methacrylate) (PMMA) was purchased from Oracal Polikarbonati (Serbia). Polyester membranes were supplied by GVS Corporation (Italy). Columbia agar and brain-heart-infusion (BHI) broth or agar, horse defibrinated blood, Dulbecco’s modified Eagle’s medium (DMEM), foetal calf serum (FCS), L-glutamine (GlutaMax), phosphate buffer solution (PBS), and 2.5% trypsin were purchased from Gibco via Thermo Fisher Scientific (France). The Matrigel Basement Membrane Matrix was purchased from BD Scientific (France). Hoechst blue 33342, propidium iodide, and Alexa Fluor 488conjugated wheat germ agglutinin (WGA-AF488) were purchased from Thermo Fisher Scientific (France). Odixanol, uranyl acetate, glutaraldehyde, dimethyl sulfoxide (DMSO), 3-(4,5dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT), ammonium bicarbonate, iodoacetamide, dithiolthreitol (DTT), and hydrogen peroxide (H 2 O 2 ) were purchased from Sigma (France). Cell line, Campylobacter strain, and culture The human colorectal carcinoma cell line (Caco-2, ATCC HTB-37) was routinely cultured in DMEM containing GlutaMAX, 100 IU/ml penicillin, 100 μg/ml streptomycin and 20% FCS at 37°C in a humidified 10% CO 2 atmosphere following the ATCC recommendations. The cells were routinely maintained in T25 flasks and passaged twice per week. The atypical aerotolerant strain Campylobacter jejuni strain Bf 23 was cultured on Columbia agar media supplemented with horse blood (5%) at 42°C under microaerophilic conditions in a candle jar for 16 h. Overnight cultures in BHI supplemented with horse blood (0.4%) were used to recover the supernatant for EV purification. Purification and size characterisation of EVs C. jejuni cells grown in BHI supplemented with horse blood until the saturation phase (10 mL) were removed by filtration through a 0.20 µm membrane filter (Whatman, GE Healthcare Life Sciences). The supernatant was ultracentrifuged at 120,000 × g for 2 h in an SW41 rotor using a Beckman XL-90 ultracentrifuge to pellet the EVs. The resulting vesicle pellet was suspended in PBS and then added to a three-layer iodixanol discontinuous gradient of 45%, 26% and 10% (w/v) in SM buffer (50 mM Tris pH 7.5, 100 mM NaCl, and 10 mM MgCl 2 ). The tubes were ultracentrifuged at 200,000 × g for 5 h in an SW55 rotor. The resulting fractions containing EVs were pooled (1.8 ml). The aliquots were used immediately or stored at 4°C and used within one week. The total protein concentration of the purified secretome and EVs was determined via the Bradford assay (Bio-Rad, France). The hydrodynamic diameter of the EVs was determined via dynamic light scattering (DLS) via a Zetasizer Pro (Malvern, France). The scattering intensity data were processed via instrumental software to obtain the size distribution of the particles. A total of 15 scans with an overall duration of 5 min were performed for each sample. Measurements were performed at 20°C. To determine the number of released vesicles and their sizes, the samples were characterised via nanoflow cytometry (nFCM) via a NanoAnalyzer (NanoFCM Co., Ltd., Nottingham, UK) according to the manufacturer’s instructions. MemGlow staining was performed to observe the purity of the EV samples by measuring fluorescence positivity in one channel (green = 525/40 bandpass filter). The particle concentration and size distribution were calculated via nFCM software (NF Profession V1.08). https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 5 Electron microscopy To visualise vesicle release from C. jejuni, bacterial cells in the saturation phase were fixed with 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) at room temperature for 1 h and gradually dehydrated as described previously(26). The samples were then gradually substituted in a mixture of propylene oxide epon and embedded in Epon (Delta Microscopy, Labège, France). Thin sections (70 nm) of bacterial preparation were collected onto Formvar/carbon-coated 200-mesh copper grids (Agar Scientific) and counterstained with lead citrate for transmission electron microscopy (TEM) visualization. Purified vesicles were directly adsorbed onto copper grids. After the samples were deposited, the grids were washed twice for 1 min with PBS and negatively stained by floating on a 10-μl drop of 2% (w/v) uranyl acetate for 1 min. The grids were air-dried before observation under a Philips EM12 electron microscope at an exciting voltage of 80 kV. Images were acquired with a charge-coupled device camera (AMT). In addition, after fixation, C. jejuni cells were mounted on aluminum stubs (50 mm diameter) with carbon adhesive discs (Agar Scientific, Oxford Instruments SAS, Gomez-la-Ville, France) and visualized via field emission gun scanning electron microscopy (SEM) under high vacuum conditions with a Hitachi SU5000 instrument (Milexia, Verrières-le-Buisson, France). Electron microscopy analyses were performed at the Microscopy and Imaging Platform MIMA2 (INRAE, Jouy-en-Josas, France). Proteome analysis Proteomic analysis was performed on the C. jejuni secretome containing EVs. Proteins were resuspended in Laemmli buffer and subjected to SDS‒PAGE (short migration). The bands on the gel were cut, and the proteins were reduced with DTT (10 mM) for 30 min at 56°C and alkylated with iodoacetamide (final concentration 55 mM) for 45 min at room temperature in the dark. In-gel digestion was conducted with 50 mM ammonium bicarbonate (pH 8.0) overnight at 37°C with 300 ng of trypsin (Promega) per sample. Peptides were extracted with 5% formic acid in water/acetonitrile (v/v). The supernatant and extracted tryptic peptides were dried and resuspended in 100 µL of 0.1% (v/v) formic acid and 2% (v/v) acetonitrile. Mass spectrometry was performed by coupling an Orbitrap Fusion™ Lumos™ Tribrid™ (Thermo Fisher Scientific) with an UltiMate™ 3000 RSLCnano System (Thermo Fisher Scientific) at the PAPPSO platform, Jouy en Josas, as previously described 24 . Four microlitres were injected, the mass range was m/z 400–1500, the analysed charge states were set to 2–5, the dynamic exclusion was set to 60 s, the intensity threshold was fixed at 1 × 104, and (ii) MS/MS was performed via HCD (30% collision energy) in an Orbitrap (AGC target = 5.0 × 103 max. injection time = 100 ms). Data analysis was performed via the C. jejuni Bf subsp. jejuni NCTC 11168 (ATCC 700819) database (NCBI, version 2021, 1572 entries) and the X! TandemPipeline (open source software developed by PAPPSO, version 0.4.5). Protein identification was performed with a precursor and a fragment mass tolerance of 10 ppm. The enzymatic cleavage rules were set to trypsin digestion (“after Arg and Lys, unless Pro follows directly after”), and no semi-enzymatic cleavage rule was allowed. The fix modification was set to cysteine carbamidomethylation, and the potential modification was set to methionine oxidation. A peptide E value < 0.01 with a minimum of 2 peptides per protein and a protein E value of <10 −4 were used as criteria to filter identified proteins. DNA analysis A NanoFCM instrument equipped with two channels (green = 525/40 bandpass filter, red = 670/30 bandpass filter) was used to estimate the DNA distribution in EVs of different sizes. Freshly filtered (0.2 µm) PBS was analyzed as a background signal and subtracted from the sample measurements. The LIVE/DEAD https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 6 BacLight kit (Thermo Fisher Scientific, France) was used to label DNA within or attached to vesicles. The stained EVs were diluted with PBS, resulting in particle counts in the optimal range of 2,500-12,000 events. Infection assay on plates The MTT test was performed to determine the toxicity of the C. jejuni secretome or EVs to Caco-2 cells. Caco2 cells were plated at a density of 30,000 cells per well on 96-well plates. After 48 h, the cells were incubated with EVs at an MOI of 10 (according to vesicle enumeration via nanoflow cytometry) or with supernatant containing 0.08 mg/mL total protein and incubated for 24 h. Then, freshly prepared MTT at a final concentration of 0.8 mg/ml was added, and the mixture was incubated at room temperature for 1 h. Subsequently, the cell layer was dried, and the MTT formazan produced by conversion of the water-soluble MTT was solubilized in DMSO as previously described(27). Cell survival was quantified by measuring absorbance values at 560 nm and corrected for a background signal by subtracting the signal measured at 670 nm via Tecan Spark®. Cell survival was expressed as the percentage of cells treated with only PBS (20%). Chip design and fabrication The gut-on-a-chip with integrated electrodes is composed of five layers: two electrode layers, two microfluidic channels, and a porous membrane that divides the apical from the basal section of the gut-on-a-chip ( Figure 1 ). The geometry of the channels and electrodes was designed using Autodesk Inventor. The outer layers comprised 2 mm thick PMMA slides that were the substrate for the 50 nm-thick gold electrodes. Two “Lshaped” electrodes (L: 2.25 cm; W: 1 mm) were deposited on each PMMA slide. Four circular through-holes were drilled on the PMMA upper layer, serving as inlets and outlets for the apical and basal channels. A plot cutter was used to cut out the channels (L: 2.5 cm; W: 1 mm) from the 250 µm-thick Flexdym sheets, forming the two middle fluidic layers. A polyester track-etched membrane with a pore size of 5 µm and a pore density of 4·10 5 pores/cm 2 separated the apical and basal fluidic channels and served as the substrate for cell culture. The five layers are aligned and assembled layer-by-layer, and the assembled device is then placed at 100°C under moderate weight for 1 hour to ensure leak-free bonding. We exploited the unique bonding properties of Flexdym to bond all the layers of the chip without the need for adhesives or surface treatments. For microscopic investigations, a gut-on-a-chip without electrodes is fabricated via hot embossing. An aluminum mould with a channel design (L:2.5 cm; W:1 mm; H:250 µm) was produced via CNC machining and used to fabricate 750 µm-thick Flexdym chips via hot embossing (3 minutes, 150°C, 0.7 bar) with the Sublym machine (Eden Tech). The same polyester track-etched membrane was used to separate the apical and basal channels. The three layers were aligned, assembled, and bonded at 100°C for 1 hour under moderate weight. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 7 Figure 1: Multilayer chip design with integrated membrane and electrodes. (A) Exploded view showing the materials of the microfluidic device: clear PMMA with 3D printed gold electrodes, Flexdym, and polyester tracketched membrane (B) Top view of the chip. (C) Cross-section showing the apical and basal microchannels. The microchannels are 2 cm long, 1 mm wide, and 250 µm high. (D) Photograph of the assembled organ chip (with connectors). (E) Simplified electrical model. Infection assay on microfluidic chips Caco-2 cells were used as a model of intestinal epithelial cells. Before cell seeding, the microfluidic channels and connectors were sterilized with 70% ethanol. Once the ethanol was fully evaporated, the devices were rinsed with sterile PBS. For microfluidic cell culture, Caco-2 cells were suspended at a concentration of 3∙10 6 cells/mL in a Matrigel solution at a ratio of 1:3 with the culture medium. Two hundred microliters of the cell suspension were then loaded into the apical channel. After seeding, the medium of the cells was changed once per day. DMEM with 10% FCS, 1% L-glutamine, and 3% penicillin‒streptomycin was used for microfluidic culture. The epithelial cells inside the chip were infected after 48 hours of culture with the C. jejuni secretome (0.08 mg/mL total protein), purified EVs (MOI of 10). PBS (20%) was used as a negative control. The positive control was 1 mM H 2 O 2, which killed all the cells. Impedance Spectroscopy The impedance spectra were acquired with a PalmSens4 potentiostat/galvanostat (PalmSens BV). An alternating current with a 0.1 V amplitude was applied between the apical and basal electrodes across a frequency range from 10 kHz to 100 kHz, resulting in 53 impedance measurements. Each frequency scan lasted approximately 2 minutes, and the measurement was repeated three times. Before each measurement, the cells were gently rinsed, and the apical and basal channels were filled with fresh culture medium to ensure consistency in the medium composition with respect to the baseline. Paracellular transport activity The trans-epithelial electrical resistance (TEER) values were measured to assess the confluency and evaluate the cytotoxic effect of Campylobacter EVs on intestinal epithelial cells. For the TEER experiment, Caco-2 cells were seeded at a density of 5 × 104 on Transwell plates (12 mm-diameter wells, polystyrene membranes with 0.4 μm https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 8 pores, Costar, Corning, MI, USA) and then maintained in DMEM. The culture media were refreshed every 2 days. Cells were allowed to grow and every day, DMEM was replaced with PBS, and resistance values (Ω) were measured for all culture inserts, including the blank, using Epithelial voltohmmeter (World Precision Instruments, Sarasota, FL, USA). The TEER values were calculated according to the following equation: TEER = (R (Cells) − R (Blank)) × tissue culture inserts surface area (Ω.cm2). Campylobacter EVs were added to the apical side of the culture insert, and TEER values were measured after 24h. Image acquisition and analysis Images were acquired with an Axio-Observer Z1 inverted fluorescence microscope (Zeiss) equipped with an AxioCam MRm digital camera and fluorescence filters using a ×10/0.25 P A-Plan objective or a ×20/0.8 Apochromat objective (all Zeiss). Microfluidic chips were imaged 4 hours after seeding to check the starting cell confluency in each device. Caco-2 cells were also imaged on day 2 before infection and on day 3, i.e., 24 h post infection (p.i.). Live cell nuclei were stained with Hoechst blue 33342 (1:100), dead cell nuclei were stained with propidium iodide (PI, 2 µM), and cell membranes were stained with WGA-AF488 (1:500). The culture medium used in the microfluidic experiments was supplemented with 25 mM HEPES before live imaging to improve the buffering capacity of DMEM and reduce cell stress during image acquisition. The acquired images were processed via ZEN Lite 3.9 image acquisition software (Zeiss). Results Campylobacter extracellular vesicles (EVs) Like other bacteria, C. jejuni releases EVs during all growth phases, which contributes to its function and interaction with the host 6, 11, 25 . In this study, we used the clinical strain C. jejuni Bf isolated from a French patient in 1994, which is highly virulent despite lacking the gene encoding cytolethal distending toxin 26 . Ultrathin sections of C. jejuni cells in the stationary phase examined by transmission electron microscopy (TEM) revealed vesicle-like structures protruding from bacterial cells ( Figure 2A ). Scanning electron microscopy (SEM) also showed the presence of small spherical structures surrounding the bacterial cells or attached to them ( Figure 2B ). We isolated the EVs by collecting and filtering the culture broth to separate the extracellular medium from the bacterial cells. Crude EVs were pelleted by ultracentrifugation from the secretome and then purified by density gradient ultracentrifugation ( Figure 2C ). All the fractions were visualized via negative-staining TEM, and those containing EVs were pooled and dialyzed against PBS to remove iodixanol. As shown in Figure 2D , the purified EVs were round, with diameters ranging from 20 to 200 nm. Additionally, we explored the hydrodynamic diameter (R H ) of EVs by performing DLS measurements. Before EV purification, the C. jejuni secretome contained a population of particles with R H between 10 nm and 30 nm, which most probably corresponded to secreted proteins and their aggregates, and a population with a mean R H of 212 nm, which corresponded to the expected diameter of the EVs ( Figure 2E ). After the density gradient purification step, only the 212 nm diameter population was detected, strongly suggesting that the vesicles were separated from other entities secreted in the extracellular medium. DLS measurements tend to overestimate the sizes of lipidic vesicles because they are solvated and move freely in aqueous solution. Therefore, we stained EVs with MemGlow membrane dye and estimated their size, purity, and amount by nanoflow cytometry. The purity of the EVs was 95%, and they ranged in size from 50 to 100 nm, with a median size of ~70 nm (Fig. 2F). Following the protocol in Fig. 2C, the concentration of the purified EVs ranged from 1∙10 8 to 6∙10 8 vesicles/mL. Thereafter, we verified whether EVs contained DNA using the Syto9/Propidium Iodide (PI) assay, which is commonly used to distinguish live and dead bacteria. As PI is not membrane permeable in contrast to https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 9 Figure 2: Characterisation of C. jejuni secretome and extracellular vesicles (EVs). (A) Transmission electron microscopy (TEM) image of C. jejuni at stationary phase with vesicle-like structures indicated by the arrow. (B) Scanning electron microscopy image of C. jejuni. (C) Secretome and EVs purification protocol. (D) Negative staining TEM of purified EVs. (E) Dynamic light scattering (DLS) measurement of EVs’ hydrodynamic diameter in the secretome and after purification. (F) EVs size distribution measured by nanoflow cytometry. (G) Syto9/Propidium Iodide (PI) assay showing the presence of external (PI) or internal DNA (Syto9) in the vesicles and their respective size. Scale bar: 500 nm. Syto9, it was possible to detect external and internal DNA. Interestingly, DNA was mostly present in EVs with diameters close to the median size, whereas smaller EVs contained no nucleic acid (Fig. 2G). DNA is a versatile molecule that may influence the innate and cell-mediated immune response of host cells and plays a https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 16 cell nuclei were stained with Hoechst blue and observed to verify cell adhesion and distribution across the apical microchannel (Figs. 6A-6D). The channel coverage was between 50% and 60% in all the chips seeded, and the cell morphology was comparable to that observed in the flask culture ( Figures 6E-6H , and Figure S4). After only 24 h, the Caco-2 cells had aggregated, forming rounded structures with diameters ranging from 50 to 100 µm, displaying smooth profiles and a morphology consistent with that of healthy Caco-2 spheroids reported in the literature 37, 38 . The cells were cultured for 48 hours and then infected as in the impedance spectroscopy experiments in Figure 5 . Finally, live imaging was performed 24 hours post infection (p.i.), as shown in Figures 6I-6L . As expected, the cells in the chip treated with PBS (negative control) remained alive, and the spheroids preserved their characteristic morphology, whereas in the chip exposed to H 2 O 2 (positive control) most cells died, and the spheroids assumed a granular appearance. Remarkably, in agreement with the impedance measurements, no cytotoxic effect was observed in the Caco-2 spheroids exposed to bacterial secretome in the microfluidic platform ( Figure 6K , and Figure S4). The chip incubated with EVs showed spheroids with irregular profiles, similar to those in the positive control, but with living core cells and some dead cells at the outermost layers. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 17 Figure 6: Effect of C. jejuni purified extracellular vesicles (EVs) and secretome on Caco-2 cells in the gut-on-chip. Live fluorescence Images of the Caco-2 cells 4 h, and 24 h after seeding in the gut-on-chip and 24 post infection (p.i.) (corresponding to 72 h of culture), with EVs (MOI of 10) and secretome (20% v/v). Cells after infection are compared to the negative control (PBS 20% v/v) and hydrogen peroxide (H2O2, 1 mM) as positive control. Live cells’ nuclei labelled with Hoechst blue 3324, dead cells nuclei with propidium iodide (PI), and membrane protein labelled with WGA. Scale bar: 50 µm. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 18 These findings strongly suggest that the spatial 3D organization of Caco-2 cells obtained in the microfluidic devices protects them from the cytotoxicity of C. jejuni-secreted biomolecules. The differences in cytotoxicity observed between traditional 2D cell culture and spheroids-on-chip suggest that Campylobacter EV absorption by Caco-2 cells occurs via diffusion, driven by the concentration gradient and facilitated by the lipophilic nature of the vesicles, which can easily dissolve in the membrane. This passive diffusion process may be transcellular or through the tight junctions between cells. To assess tight junctions in Caco-2 cells, we performed TEER measurements ( Figure 7 ). After seeding, Caco-2 cells proliferated on the membrane and reached confluence after 5 days (Fig. S5), as indicated by the stabilization of TEER values ( Figure 7A ). By day 10, further cell proliferation led to hyper-confluence (Fig. S5), resulting in an additional increase in TEER ( Figure 7A ). Therefore, we added EVs on day 5, considering that the high TEER (>100 Ω/cm²) corresponded to intact tight junctions and low paracellular permeability. Strikingly, Caco-2 cells forming tight junctions were resistant to the cytotoxic effect of Campylobacter EVs ( Figure 7B ). This suggest that EVs indeed diffuse through the tight junction between Caco-2 cells. The efficiency of such vesicle absorption is low because of the small area of the paracellular space in confluent cell layers. TEER measurements strongly suggest the pertinence of a device carrying intestinal spheroids for the investigation of hostCampylobacter EVs interactions. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 19 Figure 7: TEER (transepithelial electrical resistance) measurements of C. jejuni purified extracellular vesicles (EVs) challenged Caco-2 cells. (A) Optical microscopy images of Caco-2 cell cultures grown on Transwell® membranes. Scale bar: 300 µm. (B) Real-time TEER of Caco-2 (5×104 cells/well) in DMEM. (C) Confluent Caco-2 monolayer at day 5 was treated with EVs at MOI of 10. Signals corresponding to the medium containing vesicles and the medium without vesicles are shown as red and orange lines, respectively. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 20 Discussion We report an advanced microfluidic device equipped with built-in electrodes for cultivating intestinal epithelial cells and directly monitoring their growth and interactions with bacterial extracellular vesicles. We demonstrated that Caco-2 epithelial cells formed spheroids when cultured within the microfluidic chip that presented significantly greater resistance to the C. jejuni EVs than Caco-2 cells organized in 2D monolayers. Intestinal epithelial cells are organized into a highly structured monolayer with a crypt-villus architecture. The Caco-2 monolayer model, frequently used to assess the mechanisms of pathogenesis of various agents and the effects of therapeutics, cannot fully represent the gut-specific architecture. To better mimic the higher complexity of the in vivo environment, spheroids and organoids, as 3D culture systems, are employed as models that reflect the functional properties of organs and maintain cellular characteristics39, 40. Intestinal spheroids derived from intestinal epithelial cells have been shown to form more realistic tissue-like structures than 2D epithelial monolayers 41, 42. Spheroids are simpler than organoids but still provide key physiological characteristics for research, as they closely mimic the structure and function of the intestinal epithelium, allow cells to interact in all directions, and form tight junctions. Our device possesses several noteworthy features. (i) It integrates Flexdym, a polymer that is costeffective, reusable, and suitable for production in low-resource settings. These characteristics promote wider adoption and accessibility of microfluidic applications, overcoming some traditional barriers such as long fabrication protocols, costly facilities and single-use prototypes. (ii) The Flexdym surface is covered with a Matrigel matrix, mimicking the basement membrane, which is necessary for the adherence and survival of epithelial cells. In addition, we showed that Matrigel, in addition to being both cytocompatible and compatible with Flexdym, did not interfere with the secretome or EVs. (iii) The seeding of human intestinal cell lines results in the formation of spheroids after 24 hours under static conditions. The chip substrate is covered with densely packed epithelial spheroids, providing an in vitro model of the intestinal epithelium. (iv) The microfluidic platform has embedded electrodes that allow real-time measurements with decreased signal noise. Since C. jejuni adhesion and invasion involve bacterial crossing of the epithelial barrier and entry into underlying tissues, developing a disease model that can modulate physiological complexity is crucial for assessing its mechanisms. Considering that bacteria of the genus Campylobacter do not have a classical secretion system, they are expected to rely on the release of membrane vesicles to transport different cargoes, including DNA and proteins involved in interactions with the host6. Indeed, the observed cytotoxicity of EVs but not of the secretome on peripheral cells of the spheroids suggests that C. jejuni virulence factors are mainly transported by EVs. We found that EVs were loaded with factors that promote intestinal C. jejuni adhesion and the invasion of host cells such as flagellar hook proteins FlgK, FlgE, and the major outer membrane protein (MOMP) PorA. Importantly, the mechanism employed by Campylobacter to cross the epithelial layer is still not fully understood, and two transmigration routes have been hypothesized. C. jejuni has been shown to invade epithelial cells at the basal pole by opening cell junctions using proteases such as HtrA, thereby creating a pathway between cells 43. Additionally, the bacterium can enter at the apical pole and exit at the basal pole44. Our findings support the paracellular route, as EVs seem to diffuse through the tight cellular junctions. The enhanced resistance of Caco-2 cells self-organizing in spheroids to C. jejuni vesicles containing virulence factors highlights the potential of microfluidic devices for studying gastrointestinal bacterial infections and underscores the importance of integrating 3D models into future in vitro studies. In addition to infection, the intestine is susceptible to a range of pathologies, such as tumors and Crohn’s disease. The microfluidic device can be further expanded to incorporate other cell types (i.e., fibro blasts and mesenchymal stromal cells) and may be a good platform for studying intestinal barrier dysfunctions, exposure to drugs, metabolic stress, diarrheal toxins, or the role of the microbiota. All these investigations can ultimately be performed with https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 21 primary cells derived from patients to establish personalized precision medicine. Declarations Data availability The data supporting this article have been included as additional files. Additional file 1 (.pdf): Fig. S1, Effect of fetal calf serum concentration in culture medium on Caco-2 viability; Fig. S2, Delta of the impedance modulus in chips pre and post infection with secretome and EVs compared to the respective negative control (f = 10, 20, 30 kHz); Fig. S3, Preliminary test of the effect of C. jejuni secretome on Caco-2 cultured in the gut-on-chip. Additional file 2 (.xlsx): Identification by LC-MS/MS of proteins present in C. jejuni secretome. Competing interests S.T.C. and J.B.B. were employed by Eden Tech, the company that sells Flexdym™, and E.R. is a co-owner of Eden Tech. However, the authors declare that this affiliation did not influence the study, and they received no financial benefits or revenues from this work. Funding This work was supported in part by the European Union (grant agreement no. 101135402, Mobiles project https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 22 and grant agreement no. 872662, IPANEMA project) and the French National Agency for Research (ANR21-CE21-009 SIENA and ANR-21-CE42 ELISE projects, to J.V.). Author contributions S.T.C. and J.V. conceptualization, methodology, formal analysis, and writing of the original draft; S.T.C., F.R., D.P., S.T., G.L., C.H., C.P., and J.M.V. investigation and data curation; S.T.C. and J.M.V. visualization. J.V., E.R., J.-P.B. project administration, and funding acquisition. All authors review and editing. Acknowledgements We acknowledge Dr. Vasa Radonic and Dr. Ivana Podunavac (BioSense, Serbia) for providing the electrode layers of the microfluidic device. We thank the MicrobAdapt team for fruitful discussions and the MIMA2 platform for access to electron microscopy equipment (MIMA2 Imaging Core Facility, INRAE, Jouy en Josas, doi.org/10.15454/1.5572348210007727E12). https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 23 https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 24 https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0 25 References (1) Igwaran, A.; Okoh, A. I. Human campylobacteriosis: A public health concern of global importance. Heliyon 2019, 5 (11). (2) Authority, E. F. S. The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals and food in 2017. Efsa Journal 2019, 17 (2). (3) Black, R. E.; Levine, M. M.; Clements, M. L.; Hughes, T. P.; Blaser, M. J. Experimental Campylobacter jejuni infection in humans. Journal of infectious diseases 1988, 157 (3), 472-479. (4) Kreling, V.; Falcone, F. H.; Kehrenberg, C.; Hensel, A. Campylobacter sp.: Pathogenicity factors and prevention methods—new molecular targets for innovative antivirulence drugs? Applied Microbiology and Biotechnology 2020, 104, 10409-10436. (5) Mousavi, S.; Bereswill, S.; Heimesaat, M. M. Novel clinical Campylobacter jejuni infection models based on sensitization of mice to lipooligosaccharide, a major bacterial factor triggering innate immune responses in human campylobacteriosis. Microorganisms 2020, 8 (4), 482. (6) Malet-Villemagne, J.; Vidic, J. Extracellular vesicles in the pathogenesis of Campylobacter jejuni. Microbes and Infection 2024, 26 (8), 105377. (7) Luk, C. Y. M.; Lee, S. A.; Naidovski, N.; Liu, F.; Tay, A. C. Y.; Wang, L.; Riordan, S.; Zhang, L. Investigation of Campylobacter concisus gastric epithelial pathogenicity using AGS cells. Frontiers in Microbiology 2024, 14, 1289549. (8) MacCallum, A. J.; Harris, D.; Haddock, G.; Everest, P. H. Campylobacter jejuni-infected human epithelial cell lines vary in their ability to secrete interleukin-8 compared to in vitro-infected primary human intestinal tissue. Microbiology 2006, 152 (12), 3661-3665. (9) Callahan, S. M.; Dolislager, C. G.; Johnson, J. G. The host cellular immune response to infection by Campylobacter spp. and its role in disease. Infection and Immunity 2021, 89 (8). (10) Montanari, M.; Guescini, M.; Gundogdu, O.; Luchetti, F.; Lanuti, P.; Ciacci, C.; Burattini, S.; Campana, R.; Ortolani, C.; Papa, S. Extracellular vesicles from campylobacter jejuni CDT-treated Caco-2 cells inhibit proliferation of tumour intestinal Caco-2 cells and myeloid U937 cells: detailing the global cell response for potential application in anti-tumour strategies. International Journal of Molecular Sciences 2022, 24 (1), 487. (11) Le, L. H. M.; Elgamoudi, B.; Colon, N.; Cramond, A.; Poly, F.; Ying, L.; Korolik, V.; Ferrero, R. L. Campylobacter jejuni extracellular vesicles harboring cytolethal distending toxin bind host cell glycans and induce cell cycle arrest in host cells. Microbiology Spectrum 2024, 12 (3), e03232-03223. (12) Sharafutdinov, I.; Tegtmeyer, N.; Rohde, M.; Olofsson, A.; Rehman, Z. u.; Arnqvist, A.; Backert, S. Campylobacter jejuni surface-bound protease HtrA, but not the secreted protease nor protease in shed membrane vesicles, disrupts epithelial cell-to-cell junctions. Cells 2024, 13 (3), 224. (13) Elmi, A.; Nasher, F.; Jagatia, H.; Gundogdu, O.; Bajaj‐Elliott, M.; Wren, B.; Dorrell, N. Campylobacter jejuni outer membrane vesicle‐ associated proteolytic activity promotes bacterial invasion by mediating cleavage of intestinal epithelial cell E‐cadherin and occludin. Cellular microbiology 2016, 18 (4), 561-572. (14) Chiba, M.; Miri, S.; Yousuf, B.; Esmail, G. A.; Leao, L.; Li, Y.; Hincke, M.; Minic, Z.; Mottawea, W.; Hammami, R. Dual bacteriocin and extracellular vesicle-mediated inhibition of Campylobacter jejuni by the potential probiotic candidate Ligilactobacillus salivarius UO. C249. Applied and Environmental Microbiology 2024, 90 (8), e00845-00824. (15) Zalewska-Piątek, B.; Olszewski, M.; Lipniacki, T.; Błoński, S.; Wieczór, M.; Bruździak, P.; Skwarska, A.; Nowicki, B.; Nowicki, S.; Piątek, R. A shear stress micromodel of urinary tract infection by the Escherichia coli producing Dr adhesin. PLoS pathogens 2020, 16 (1), e1008247. (16) Park, A.; Jeong, H.-H.; Lee, J.; Kim, K. P.; Lee, C.-S. Effect of shear stress on the formation of bacterial biofilm in a microfluidic channel. BioChip Journal 2011, 5, 236-241. (17) Sherman, E.; Bayles, K.; Moormeier, D.; Endres, J.; Wei, T. Observations of shear stress effects on Staphylococcus aureus biofilm formation. Msphere 2019, 4 (4), 10.1128/msphere. 00372-00319. (18) Calzuola, S. T.; Newman, G.; Feaugas, T.; Perrault, C. M.; Blondé, J.-B.; Roy, E.; Porrini, C.; Stojanovic, G.; Vidic, J. Membrane-based microfluidic systems for medical and biological applications. 2024. (19) Grassart, A.; Malardé, V.; Gobaa, S.; Sartori-Rupp, A.; Kerns, J.; Karalis, K.; Marteyn, B.; Sansonetti, P.; Sauvonnet, N. Bioengineered human organ-on-chip reveals intestinal microenvironment and mechanical forces impacting Shigella infection. Cell host & microbe 2019, 26 (3), 435-444. e434. (20) Boquet-Pujadas, A.; Feaugas, T.; Petracchini, A.; Grassart, A.; Mary, H.; Manich, M.; Gobaa, S.; Olivo-Marin, J.-C.; Sauvonnet, N.; Labruyère, E. 4D live imaging and computational modeling of a functional gut-on-a-chip evaluate how peristalsis facilitates enteric pathogen invasion. Science Advances 2022, 8 (42), eabo5767. (21) Kim, H. J.; Huh, D.; Hamilton, G.; Ingber, D. E. Human gut-on-a-chip inhabited by microbial flora that experiences intestinal peristalsislike motions and flow. Lab on a Chip 2012, 12 (12), 2165-2174. (22) Ghoshal, U. C.; Ghoshal, U. Small intestinal bacterial overgrowth and other intestinal disorders. Gastroenterology Clinics 2017, 46 (1), 103-120. (23) Bronnec, V.; Turoňová, H.; Bouju, A.; Cruveiller, S.; Rodrigues, R.; Demnerova, K.; Tresse, O.; Haddad, N.; Zagorec, M. Adhesion, biofilm formation, and genomic features of Campylobacter jejuni Bf, an atypical strain able to grow under aerobic conditions. Frontiers in microbiology 2016, 7, 1002. (24) Auger, S.; Henry, C.; Péchoux, C.; Suman, S.; Lejal, N.; Bertho, N.; Larcher, T.; Stankic, S.; Vidic, J. Exploring multiple effects of Zn0. 15Mg0. 85O nanoparticles on Bacillus subtilis and macrophages. Scientific reports 2018, 8 (1), 12276. (25) Omole, Z.; Dorrell, N.; Elmi, A.; Nasher, F.; Gundogdu, O.; Wren, B. W. Pathogenicity and virulence of Campylobacter jejuni: What do we really know? Virulence 2024, 15 (1), 2436060. https://doi.org/10.26434/chemrxiv-2024-jf41x-v2 ORCID: https://orcid.org/0009-0000-5737-0537 Content not peer-reviewed by ChemRxiv. License: CC BY-NC 4.0