The two-layer (upper and bottom) microfluidic chip used in this study was designed using LayoutEditor on a chrome mask. The sizes and shapes of the chip on the two layers were identical, and one was bonded to the opposite one with a polyester, porous, semipermeable membrane (Corning) in between. The master mold (silicon wafer) was created by the standard photolithography process. The polydimethylsiloxane (PDMS, Dow Corning) elastomer was fabricated by replica molding from the SU-8 (SU8-2010, MicroChem)/silicon master based on a ratio of 10:1 with the curing agent. After pouring PDMS on the mold, the mixture was degassed in a vacuum chamber for air bubble removal and cured at 80°C for 1 h in the oven. The microfluidic chip was cut out by razor blade, punched for the fluidic connection ports, and then bonded to a glass slide (24 mm × 60 mm) by oxygen plasma. Ag/AgCl electrodes were produced by immersing platinum wire and silver wire into potassium chloride electrolyte and inserted into the designed electrode side of the chip.

Patients with IBD and healthy control individuals were enrolled from the First Affiliated Hospital of Zhengzhou University from December 2021 to March 2022. The inclusion criteria were patients with moderate to severe active IBD who met the diagnostic criteria for IBD. The exclusion criteria were patients with other intestinal organic diseases, other infectious diseases, and pregnant and lactating women. This study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval number: L2018-Y150), and the patients signed an informed consent form. We collected 5 mL of peripheral blood from each participant into individual serum collection tubes. Serum samples were obtained by centrifuging at 4°C and 3000 rpm for 10 min after coagulation.

Serum exosomes were extracted using an exosome extraction kit (KGPE001-10, KeyGEN BioTECH), in which 250 μL of extracellular vesicle separation reagent was added to the serum, gently mixed well, and allowed to stand for 2 h at 4°C. Next, the samples were centrifuged 12,000×g at 4°C for 20 min. The supernatant was discarded to obtain the extracted exosomes, which were stored at −80°C until use. The exosomes were identified through particle size analysis using a nanoparticle tracking analyzer (NTA, ZetaView^®, Particle Metrix) and image observation using transmission electron microscopy (TEM, HT7700, HITACHI). Biomarkers for exosomes including TSG101, CD63, and Alix proteins were also detected by Western blot.

Human umbilical vein endothelial cells (HUVECs) were ordered from Procell Life Sciences Co., Ltd. (Wuhan, Hubei, China) and were cultured in Ham’s F-12K medium complemented with 0.1 mg/mL heparin, 0.03–0.05 mg/mL ECGs, and 10% FBS at 37°C in a humidified atmosphere with 5% (v/v) CO₂.

TEER was measured using a lock-in amplifier (HF2LI, Zurich Instruments, Switzerland) operated by a customized LabVIEW program. Before each measurement, the medium inside the chip was refreshed to reach room temperature to minimize measurement errors caused by the changes in temperature and medium conductivity. In the microfluidic chip, two electrodes were located above and below the PC membrane, respectively. The impedance values between any two electrodes were recorded with an AC signal of ∼0.8 V applied to one terminal and the other terminal grounded. The response current was amplified by a current amplifier (HF2TA, Zurich Instruments, Switzerland) with an amplification of 10⁸. Afterward, the amplified signals were fed to the HF2LI for demodulation, with the oscillator frequency set to 10 kHz and the signal sampling rate set to 7K Sa/s. Based on the equivalent resistive circuit and the Gaussian elimination, the TEER value was calculated using Eq. 1, discussed in Section 3.1.

An IBD mouse model was induced by DSS (Mw 36,000–50000) treatment. Male C57BL/6 J mice (6–8 weeks, 20–25 g) were ordered from the Experimental Animal Center of Zhengzhou University (Experimental Animal Use License No. SCXK (Yu) 2020 0008) and housed in a specific pathogen-free (SPF) environment at 22°C–26°C and a relative humidity of 40%–70%. All the experiments were performed in strict accordance with good animal practice. The experimental mice were randomly assigned to four groups (n = 6 per group): Control, a negative control group; IBD model, DSS-induced colitis group; IBD model + HC-exos, DSS-induced colitis group treated with exosomes derived from human control; and IBD model + IBD-exos, DSS-induced colitis group treated with exosomes derived from patients with IBD. Mice in IBD model drank 2% DSS water, and those in the control group drank normal water freely. On days 3, 5, and 7, mice in IBD model + HC-exos and IBD model + IBD-exos groups were injected through the tail vein with total protein from 1 mg of exosomes derived from human controls and from patients with IBD. Daily diarrhea, bloody stools, and weight were monitored to calculate the Disease Activity Index (DAI) scores. The experimental mice were euthanized when they showed significant weight loss, with symptoms of curling up, erect hair, slow movement, decreased appetite, and loose stools. The lengths of the colons of these mice were measured. Hematoxylin–eosin (HE) staining was used to identify the severity of the disease, and immunohistochemistry (IHC) was used to detect the corresponding protein changes in the colon tissues.

Exosomal proteins were extracted using SDT buffer lysis (4% SDS, 100 mM Tris HCl, 1 mM DTT, pH7.6), and the protein concentration was determined using the BCA protein quantification kit (Bio-Rad, USA). The proteins were digested with pancreatin and processed using the ultrafiltration-assisted sample preparation (FASP) method. The peptide segments obtained after treatment were desalinated using a C18 chromatographic column, concentrated by vacuum centrifugation, and dissolved in 40 µL 0.1% formic acid.

Identification and analysis were conducted using liquid chromatography tandem mass spectrometry (LC-MS/MS). Each peptide sample was separated using a nanoliter flow rate high-performance liquid chromatography system, with mobile phase buffer A containing 0.1% formic acid aqueous solution and B containing 0.1% formic acid acetonitrile aqueous solution (84% acetonitrile). Using a 95% A-liquid equilibrium chromatographic column, each sample was loaded into a C18 chromatographic column using an automatic sampler and separated by a C18-A2 analytical column at a flow rate of 300 nL/min. After chromatographic separation, each sample was subjected to mass spectrometry analysis on a timsTOF Pro mass spectrometer. The detection method was positive ions, the ion source voltage was set to 1.5 kV, and TOF was used for detection and analysis in both mass spectrometry and tandem mass spectrometry. The scanning range of the mass spectrometry was set to 100–1700 m/z. The data parallel accumulation serial fragmentation (PASEF) mode was used to data collection. After collection, the mother ions were collected in 10 PASEF modes, with a cycle window time of 1.17 s and secondary mass spectrometry with a charge number of 0–5. The dynamic exclusion time of tandem mass spectrometry scanning was set to 24 s to avoid repeated scanning of the parent ions and generate original mass spectrometry detection data. The mass spectrometry raw data were identified and quantitatively analyzed using the LFQ (label-free quantification) algorithm in MaxQuant software (version 1.6.14).

Subcellular localization was analyzed using CELLO (http://cello.life.nctu.edu.tw/). The structural domain prediction used InterProScan software. Gene Ontology (GO) annotation passed through the GO database, and Blast2Go was used (https://www.blast2go.com/). The software annotates the GO function of all differentially expressed proteins. Cytoscape (http://www.cytoscape.org/Version 3.2.1) was used for visualization.

T-tests were used for inter-group differences, and Fisher’s exact tests were used for functional enrichment analysis. P-values <0.05 indicated statistically significant differences.

This work developed an in-house customized microfluidic chip that emulated the endothelial barrier microenvironment and enabled TEER monitoring. The chip was designed to comprise a semipermeable membrane that permitted the formation of an endothelial cell barrier and four electrodes to realize the calculation of resistances (Figure 1A). The circular central chamber was 5 mm in diameter and 140 μm in depth, with separate inlets and outlets for the top and bottom layers. The device consisted of three layers, including the top layer for the perfusion of medium to maintain nutrition and pressure during cell culture, the PC membrane for endothelial cellular barrier formation, and the bottom layer to mimic the extracellular matrix using hydrogel. This chip allowed easy real-time imaging and facilitated the performance of immunocytochemistry in situ and cell retrieval after the experiments for further analysis.

In this device, it was possible to directly measure the transendothelial or transepithelial electrical resistance (TEER), a strategy widely used for non-invasive and label-free monitoring of the tightness of the endothelial barrier in vitro. In the chip, four electrodes were inserted into the two channels, with two on each side of the PC membrane. In six different measurement configurations, we can directly derive isolated TEERs independent of the channel characteristics. The four electrodes were inserted into top and bottom channels some distance from the cellular barrier to avoid hampering the visual inspection of the cells, while the prestored medium in the electrode channels independent from the inlets and outlets reduced the influence of medium composition changes on the apparent TEER. From the equivalent circuit shown in Figure 1B, resistance can be determined by Gaussian elimination, including the resistance of the cellular barrier and membrane ( R m ). The TEER ( Ω c m 2 ) is determined by the normalization of the resistance of the R m ( Ω ) to the culture area A c u l t ( c m 2 ) using Eq. 1. The TEER of the system prior to cell seeding was regarded as the baseline, which should be subtracted. T E E R = R m ∙ A c u l t = 1 4 R 1 → 2 + R 1 → 3 + R 2 → 4 + R 3 → 4 − R 1 → 4 − R 2 → 3 ∙ A c u l t (1) R i → j depicted the resistance between two electrodes, while i and j referred to the electrode numbers in Figure 1. The culture area on the membrane is π 5 2 2 = 6.25 π   m m 2 . This method directly measures TEER in the microfluidic chip without the need for an integrated electrode close to the cell barrier. Hence, large variations caused by non-biological sources in chips filled with culture media were eliminated.

The exosomes isolated from patient serum were identified by particle size analysis, image observation, and biomarker protein detection. The average particle size was 119.8 ± 63.7 nm, and 94.2% particles were distributed at approximately 101.6 nm (Figure 2A). TEM image observation showed the intact exosomes at magnification of 60 k (Figure 2B). The Western blot results demonstrated the positive detection of biomarkers for exosomes, including TSG101, CD63, and Alix proteins (Figure 2C). These data verified the characteristic of exosomes, which were then used in the subsequent experiments.

A HUVEC cell line was used in this study to mimic the on-chip endothelial cellular barrier. As shown in Figure 3A, first, Matrigel was injected into the bottom chamber and incubated for 4 h for complete gelation to mimic the extracellular matrix. Then, HUVEC cells were loaded into the top chamber at a density of 10×10⁶ cells/mL and left to attach and grow on the porous membrane overnight to form the cellular barrier. The cells were maintained in a perfusion flow regime at a flow rate of 8 uL/hour using culture medium. The following day, medium containing exosomes was perfused for treatment. To confirm the treatment time, the exosomes were labeled using PKH67, a green fluorescent dye that binds to the lipid components of the membrane structure, to trace the exosomes. In the conventional culture plate, exosomes (green) were endocytosed by HUVEC cells (blue) after 24 h (Figure 3B) because the labeled exosomes were discarded if not endocytosed. The on-chip endothelial cellular barrier was then exposed to 100 μg of total exosome protein for 24 h. After that, resistances between the electrodes were measured to calculate TEER. The exosomes derived from IBD serum significantly reduced the TEER levels (p < 0.01), indicating that exosomes derived from IBD serum impacted the endothelial barrier (Figure 3C).

To validate the effect of serum exosomes on the progression of IBD in vivo, a mouse IBD model induced by DSS was established. Exosomes derived from control or IBD patients were injected through the tail vein every other day from the third day after DSS treatment. Mice in the IBD group, regardless of exosome treatment or not, showed higher DAI scores than those in the normal control group at day 5, indicating the effects of DSS induction. To day 7, mice in the IBD group treated with exosomes derived from IBD patients showed significantly higher DAI scores than all the other three groups. The levels continue to increase until day 8, and the mice showed much more severe illness (Figure 4A). All the mice were euthanized on day 8, and the lengths of colons in the IBD group treated with IBD exosomes were significantly shorter than those of the control group (Figure 4B). HE staining also showed that the IBD group treated with IBD exosomes had the most serious disease (Figure 4C). Additionally, S100A8, which plays an important role in inflammation and angiogenesis (Li et al., 2012), detected by IHC showed greatly enhanced expression in the colon tissues of the IBD exosome-treated group (Figure 4D). These data supported that exosomes derived from IBD patients significantly promoted disease progression.

Although the impact of serum exosomes on IBD progress was validated, the corresponding regulatory protein cargoes were still not clear. Therefore, we further profiled the differential proteome of IBD serum exosomes. In total, 5765 peptides and 891 proteins were identified, including transthyroxine (TTR) and ß-2-glycoprotein 1 (APOH), which were consistent with the common proteins in serum exosomes listed in the ExoCarta (http://www.exocarta.org) public database. Quantitative analysis screened a total of 55 differential (38 upregulated and 17 downregulated) proteins in IBD serum exosomes compared with control (Figure 5A). Meanwhile, 38 proteins were screened as only present in control exosomes (not identified in IBD), and 42 proteins were screened as only present in IBD exosomes (not identified in control) (Figure 5B). Among these identified differential proteins, many immunoglobulins were significantly upregulated, showing that immune factors, especially abnormally expressed immunoglobulins, contributed to IBD pathogenesis. Transferrin receptor protein 1 (TFRC) was significantly upregulated by more than five-fold in IBD exosomes. Transferrin receptor levels are reportedly increased in inflammatory tissues, while HIF a pathway agonists may benefit such patients by increasing non inflammatory tissue-mediated iron absorption and reducing mucositis (Fagundes et al., 2022). Annexin A5 (ANXA5), a calcium and phospholipid-binding protein, was significantly downregulated in IBD exosomes. ANXA5 reportedly has potential inhibitory effects on inflammation through binding to aged red blood cells, activated platelets, endothelial microparticles, and tumor blood vessels (Domeij et al., 2013), effectively alleviating TNBS-induced colitis by inhibiting inflammatory cell infiltration (Zhang et al., 2021), which may be a potential target in the treatment of IBD.

Analyses of biological processes, cellular components, and molecular functions were also conducted on the screened differential proteins. As shown in Figure 5C, these differential proteins were mainly enriched in biological processes such as intracellular processes, stimulus responses, biological regulation, metabolic processes, intracellular component biosynthesis, and signaling; the cell components included organelles, extracellular areas, and cell membranes; and the molecular functions involved binding, catalytic activity, molecular function regulation, structural molecular activity, transporter activity, molecular transduction activity, antioxidant activity, and transcriptional regulation.