Human primary T cells or CD4⁺ T cells were isolated from peripheral blood mononuclear cells (PBMC) of buffy coats of anonymous healthy donors (HD) (Policlinico Umberto I, Sapienza University of Rome, Italy) by using a EasySep™ isolation kits (#17951 and #17952, STEMCELL Technology, CAN). HD signed the informed consent and the Ethic Committee of Policlinico Umberto I approved the procedure (ethical code N., 1061bis/2019, 13/09/2019). T cells were cultured in RPMI-1640 containing 5% human serum (Euroclone, UK), L-glutamine, penicillin and streptomycin. The sorted population was > 95% pure, as evidenced by flow cytometry.

Caco-2 epithelial cell line (ATCC, Maryland, USA) was cultured in DMEM supplemented with 10% FBS (Euroclone, UK), L-glutamine, penicillin and streptomycin.

The following antibodies were used: rabbit anti-human vimentin (#5741), rabbit anti-human NF-κB/p65 (#8242), rabbit anti-human phosphoTyr705-STAT3 (#9145) were from Cell Signalling Technologies (MA, USA); mouse anti-human E-cadherin (#610182), mouse anti-human N-cadherin (#610254) were from BD Biosciences (Milan, Italy); rabbit anti-human FAS (#sc-715), mouse anti-human GAPDH (#sc-47724) were from Santa Cruz Biotechnology (Texas, USA); rabbit anti-human ZO-1 (#40-2200), Alexa Fluor 594 phalloidin (#A12381), goat anti-mouse Alexa-flour 488 (#A11070), goat anti-rabbit Alexa Fluor 647 (#A21245) were from Thermo Fisher Scientific (MA, USA). Inflammatory cytokines neutralising Abs (NAbs) used were mouse anti-TNF-α (#MAB610), anti-IFN-γ (#MAB2852), anti-IL6 (#MAB206), anti-IL-17A (#MAB317) and anti-IL-22 (#MAB7822) (R&D Systems, USA). All NAbs were used at 2.5 μg ml^-1 final concentration.

Staphylococcal Enterotoxin B (SEB, #54881) was purchased by Merck (Milan, Italy). 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI, #H-1200) was from Vector Laboratories, Inc. (Burlingame, CA, USA). pSEB_116-132 mimetic peptide (GGVTEHNGNQLDKYRSI) containing D-Ala at both termini to enhance protease resistance was purchased by BIO-FAB research (Rome, Italy). Peptide was > 95% pure and used in culture at 10 μM final concentration. PS1145 (#P6624, Merck) and S31-201 (#sc-204304, Santa Cruz Biotechnology) inhibitory drugs were used at 10 μM and 100 μm, respectively.

T cells (0.5 x 10⁶ ml^-1) were cultured in 24-well plate or 24-well plate inserts when co-cultured with Caco-2 cells (2.5 x 10⁵ ml^-1) and stimulated for different times with 0.1 μg ml^-1 SEB. Secretion of inflammatory cytokines (TNF-α, IL-6, IL-17A, IL-22 and IFN-γ) in culture supernatants was measured by ELISA using human TNF-α (#DY-210), IL-6 (#DY-206), IFN-γ (#DY285), IL-17A (#DY-317) and IL-22 (#DY-782) kits (R&D Systems). The assays were performed in duplicate and data analyses was performed on a Bio-Plex (Bio-Rad, Hercules, CA, USA). The sensitivity of the assays was as follow: IL-6 and IFN-γ = 9.4 pg ml^-1, TNF-α and IL-17A = 15.6 pg ml^-1 and IL-22 = 31.2 pg ml^-1.

Caco-2 cells (2.5 × 10⁵ ml⁻¹) were co-cultured with T cells (0.5 x 10⁶ ml⁻¹) in 24 trans-well plate inserts for different times at 37°C. At the end of incubation, cells were lysed for 30 min on ice in 1% Nonidet P-40 lysis buffer (0.15 M NaCl, 0.02 M Tris-HCl pH 7.5, 0.001 mM EGTA and MgCl₂, 0.05 M NaF, 0.01 M Na₄P₂O₇) containing proteases and phosphatases inhibitors (10 μg ml^-1 leupeptin, 10 μg ml^-1 aprotinin, 1 mM NaVO₄, 1 mM Pefablock-SC). Proteins were resolved by SDS-PAGE and blotted onto nitrocellulose membranes. Blots were incubated with anti-E-cadherin, anti-ZO-1, anti-N-cadherin or anti-vimentin (1:1000 dilution) or anti-GAPDH (1:400 dilution) Abs, extensively washed and after incubation with horseradish peroxidase (HRP)-labelled goat anti-rabbit (#NA934V, 1:5000 dilution) or HRP-labelled goat anti-mouse (#NA931V, 1:5000 dilution) developed with the enhanced chemiluminescence’s detection system (GE Healthcare Life Sciences, Italy) and analysed by ChemiDoc XRS⁺ (Bio-Rad Laboratories, Italy). Protein levels were quantified by densitometric analysis using the ImageJ 1.50i program (NIH, USA)

Caco-2 (2.5 x 10⁵ ml^-1) were co-cultured with T cells (0.5 x 10⁶ ml⁻¹) unstimulated or stimulated with 0.1 μg ml^-1 SEB in 24-well inserts plates. After 72 hours, Caco-2 cells were stained with 10 μg ml-1 propidium iodide (PI) and the percentage of PI-positive cells was quantified by flow cytometry (FACScalibur, BD Biosciences, Mountain View, CA).

Caco-2 cells (2.5 × 10⁵ ml⁻¹) were co-cultured with T cells (0.5 x 10⁶ ml⁻¹) in 24 trans-well plate inserts for different times. After fixing (2% paraformaldehyde) and permeabilization (0.1% saponin), E-cadherin and ZO-1 staining was performed by using mouse anti-human E-cadherin (1:100 dilution) followed by goat anti-mouse Alexa flour 488 (1:100 dilution), and rabbit anti-human ZO-1 (followed by goat anti-rabbit Alexa flour 647 (1:100 dilution). Alexa fluor 594 phalloidin (1:1000 dilution) was used to stain filamentous actin (F-actin) and DAPI was used to stain nuclei. Observations were performed with a Nikon Eclipse Ti2 confocal microscope (63X oil objective). The same acquisition settings were used to process the Z stack images by NIS Elements AR 5.30 software (Nikon Europe B.V.).

Caco-2 cells were treated as indicated and the nuclear translocation of RelA/NF-κB and pSTAT3 was analysed by using anti- RelA (1:300 dilution) and anti- pSTAT3 (1:300 dilution) Abs followed by goat anti-rabbit Alexa-flour 488 (1:100 dilution). Nuclei were stained with DAPI. Images of RelA and pSTAT3 localization were obtained with a computer-controlled Nikon Eclipse 50i epifluorescence microscope and a plan achromat microscope objective 100XA/1.25Oil OFN22 WD 0.2 and QImaging QICAM Fast, 1394 Digital Camera, 12-bit-Mono (Minato, Tokyo, Japan). The fluorescence intensities of nuclear RelA and pSTAT3, overlapping with DAPI, were quantified by using Fiji ImageJ software and intensities ratio were calculated for each cell. At least one hundred cells were examined quantitatively for each condition in three independent experiments.

Total RNA was extracted using Trizol (Thermo Fisher Scientific) from 2.5 × 10⁵ Caco-2 cells, reverse-transcribed into cDNA, and real-time PCR was performed by Luna^® Universal qPCR Master Mix (NEB, M3003E). The relative quantification was performed using the comparative cycle threshold (C_T) method. Specific primer/probe sets ( Table 1 ) to quantify the expression level of ZO-1, CDH1, SNAIL-1, TWIST-1, ZEB-1 and GAPDH mRNA levels were purchased by Metabion International AG (Planegg, Germany).

Caco-2 cells were cultured for 8 h with the supernatant of primary T cells unstimulated or stimulated for 72 h with 0.1 μg ml^-1 SEB. ChIP assays were performed as previously described (35). Briefly, after fixing in 1% formaldehyde, Caco-2 cells were lysed and chromatin sheared by sonication. Lysates were incubated at 4°C with anti-RelA (1:100 dilution), anti pSTAT3 (1:100 dilution), or anti-FAS (1:100 dilution) Abs as control, and immune complexes were collected with sperm–saturated Protein-A Sepharose beads. After protein-DNA cross-links reversion, DNA was extracted by phenol-chloroform and analysed by real-time PCR with Luna^® Universal qPCR Master Mix. Specific primer/probe sets for the human SNAIL-1, TWIST- and ZEB-1 promoters were used ( Table 1 ). Specific enrichment was calculated by using the C_T: 2^{(controlChIP−controlInput)}/2^{(specificChIP−specificInput)} as previously described (36).

Structures and complexes modelling were performed as previously described (23). The following experimentally determined structures was used to this purpose: SEB (PDB: 1SEB) (37); CD28 extracellular domain (PDB: 1YJD) (38); TCR (TRAV22/TRBV19) (PDB:4C56) (39). Briefly, full-length proteins were modelled with AlphaFold2 (40) and docked with ClusPro 2.0 (41), HADDOCK (42), and MultiLZerD (43). Protein sequence manipulations, superpositions and modelling were carried out using PyMod 3.0 (44).

The sample size was chosen based on previous studies to ensure adequate power. Statistical analysis (mean and standard error of mean, SEM) was performed through Prism 8.0 (GraphPad Software, San Diego, CA), by using Student’s t test or one-way ANOVA with the Fisher’s LSD test for multiple comparisons. For all tests, P values < 0.05 were considered significant.

We have recently demonstrated that staphylococcal SAgs may trigger inflammatory cytokine production by binding the TCR and CD28 in the absence of APCs expressing MHC-II- and/or B7 (23). Consistently with these data and with the dose-response production of inflammatory cytokines ( Supplementary Figure S1A ), stimulation of T cells with 0.1 μg ml^-1 SEB induced a significant secretion of IFN-γ ( Figure 1A ), IL-6 ( Figure 1B ), TNF-α ( Figure 1C ), IL-17A ( Figure 1D ) and IL-22 ( Figure 1E ) after 48-72 hours (h). Since inflammatory cytokines produce by SEB-stimulated T cells have been described to alter the homeostasis and integrity of intestinal epithelial cells (45), we co-cultured SEB-activated inflammatory T cells with Caco-2 cells in trans-well plates in time-course experiments. Caco-2 is a cell line derived from a human colorectal adenocarcinoma that acquires in vitro the morphological and functional features of mature enterocytes of the small intestine, representing a good model of intestinal barrier (46, 47). Thereafter, both Caco-2 cell viability ( Figure 1F ) and intestinal epithelial barrier integrity ( Figure 1G ) were assessed. While no change in Caco-2 cell viability was observed ( Figure 1F ), confocal microscopy analysis revealed changes in F-actin organization in Caco-2 cells co-cultured with SEB-activated T cells ( Figure 1G ). For instance, unstimulated Caco-2 cells (CTR) displayed an extensive actin network with F-actin mainly organised into cortical bundles closely associated with cell-cell contacts. Neither SEB alone nor unstimulated T cells (T) significantly affected the steady-state of actin cytoskeleton organization as well as Caco-2 cell morphology. On the contrary, within 24 h of co-culture, Caco-2 cells exhibited para-cellular gaps, which further increased after 48-72 h ( Figure 1G ). Moreover, after 48 h of co-culture, SEB-activated T cells induced changes in F-actin dynamics in Caco-2 cells with the formation of tight parallel bundles and/or stress-like fibers (white arrows), which further increased after 72 h ( Figure 1G ).

These data suggest a potential role of inflammatory cytokines produced by SEB-stimulated T cells in altering F-actin dynamic and intestinal barrier integrity.

Since F-actin remodelling observed in Caco-2 cells exposed to SEB-activated inflammatory T cells resembled those occurring during epithelial-mesenchymal transition (EMT) (48, 49), we analysed the expression of E-cadherin (E-CAD) and ZO-1 (epithelial markers) as well as of N-cadherin (N-CAD) and vimentin (mesenchymal markers). The kinetic analysis pointed out that both ZO-1 and E-CAD were strongly down-regulated after 72 h of co-culture of Caco-2 cells with T cells activated with SEB ( Figures 2A, B ). On the contrary, no detectable expression of N-CAD and vimentin was observed ( Figure 2A ). The reduction of both E-CAD and ZO-1 protein content in Caco-2 cells cultured with SEB-activated T cells was also accompanied by the inhibition of ZO-1 ( Figure 2C ) and E-CAD mRNA levels (CDH1, Figure 2D ), which started to decrease within 24 h and reached a peak of inhibition after 72 h ( Figure 2E ).

The transcriptional downregulation of ZO-1 and E-CAD observed in Caco-2 cells co-cultured with SEB-stimulated T cells prompted us to analyse the expression of the EMT-transcription factors (TFs) SNAIL, TWIST and ZEB, known epithelial marker repressors (50). Interestingly, we found a strong up-regulation of SNAIL-1, TWIST-1 and ZEB-1 expression in Caco-2 cells co-cultured for 72 h with SEB-stimulated T cells ( Figure 2F ), suggesting that these effects might be linked to E-CAD and ZO-1 downregulation. Of note, the comparison of Caco-2 cells co-cultured with total T cells or CD4⁺ T cells evidenced no significant differences in cell-cell junction downregulation and EMT-TF upregulation ( Supplementary Figure S1B ), supporting our previous data showing that most of the inflammatory cytokines produced by T cells following SEB stimulation derive from CD4⁺ T cells (23).

Finally, to assess whether Caco-2 cell dysfunctions were mediated by the inflammatory cytokines produced by SEB-stimulated T cells, Caco-2 cells were cultured with anti-TNF-α, anti-IFN-γ, anti-IL-6, anti-IL-17A and anti-IL-22 neutralising antibodies. Indeed, Caco-2 cells express all the receptors for these inflammatory cytokines and are susceptible to their activity (51–54). The neutralization of each cytokine did not significantly affect neither ZO-1 and CDH1 downregulation ( Supplementary Figure S2A ) nor SNAIL-1, TWIST-1 and ZEB-1 up-regulation ( Supplementary Figure S2B ) induced in Caco-2 cells by SEB-activated T cells. However, when a cocktail of all neutralising Abs (NAbs) was added in culture, we observed a significant impairment of F-actin remodelling as well as E-CAD and ZO-1 delocalization in Caco-2 cells co-cultured with SEB-stimulated T cells ( Figure 3A ). Moreover, neutralization of inflammatory cytokines also restored ZO-1 and CDH1 gene expression ( Figure 3B ) and inhibited SNAIL-1, TWIST-1 and ZEB-1 expression ( Figure 3C ) induced in Caco-2 cells by SEB-activated T cells.

Thus, inflammatory cytokines produced by SEB-stimulated T cells disrupt Caco-2 intestinal epithelial barrier and cell-cell adhesion.

Most of the inflammatory cytokines produced by T cells following SEB bivalent binding to the TCR and CD28 exert their functions by activating two main TFs, RelA/NF-κB and STAT3 (23, 35, 55). Consistently, a significant nuclear translocation of RelA/NF-κB ( Figures 4B, C ) and Tyr705 phosphorylated STAT3 (pSTAT3) (56) ( Figures 4D, E ) was observed in Caco-2 cells cultured for 8 h with the inflammatory milieu of SEB-stimulated T cells ( Figure 4A ). Treatment of Caco-2 cells with the NF-κB inhibitor PS1145 (57) or the selective STAT3 inhibitor S31-201 (58) restored both ZO-1 and CDH1 gene expression ( Figure 4F ) and impaired the up-regulation of SNAIL-1, TWIST-1 and ZEB-1 ( Figure 4G ) induced by the inflammatory milieu of SEB-activated T cells.

Altogether, these data highlight a critical role of RelA/NF-κB and STAT3 in inducing epithelial barrier dysfunctions and associated EMT-TFs up-regulation following exposure of Caco-2 cells to SEB-activated inflammatory T cells.

Several putative binding sites for RelA/NF-κB and STAT3 have been identified within the promoters of SNAIL-1, TWIST-1 and ZEB1 (59–63). In particular, SNAIL-1 promoter contains three putative RelA/NF-κB binding sites, with NF-κB2 and NF-κB3 sites mainly involved in promoting SNAIL-1 expression (63), and one STAT3 binding site (62) near the NF-κB3 site ( Figure 5A ). Two STAT3 (59) and one NF-κB consensus sequence (63) were identified in the promoter of TWIST-1 ( Figure 5C ). Similarly, ZEB-1 promoter contains one NF-κB (61) and two STAT3 binding sites (60) ( Figure 5E ). To assess whether the up-regulation of EMT-TFs induced by the inflammatory milieu of SEB-activated T cells could depend on RelA/NF-κB and/or pSTAT3, we performed chromatin immunoprecipitations (ChIPs) in Caco-2 cells cultured with the supernatants of SEB-activated T cells using specific oligonucleotide probes ( Table 1 ). The results evidenced a specific recruitment of both pSTAT3 and RelA on the promoters of SNAIL-1, with RelA that preferentially bound NF-κB2 but not NF-κB3 site ( Figure 5B ). Similarly, selective and specific recruitment of both RelA and pSTAT3 was also observed on TWIST-1 ( Figure 5D ) and ZEB-1 promoters ( Figure 5F ).

Altogether these data evidence that SEB activates T cells to secrete inflammatory cytokines, which in turn promote the activation of RelA/NF-κB and STAT3 and their cooperation in up-regulating EMT-TFs expression.

We have recently demonstrated that SEB may bind the TCR and CD28 in a bivalent manner, thus stimulating the production of inflammatory cytokines independently of APCs (23). Based on the proposed complex between SEB the TCR and CD28, we identified a solvent exposed loop of SEB (residues 116-132) interacting with CD28, and evolutionarily well conserved in other SAgs ( Figure 6A ). We then tested the capability of a mimetic peptide of this region (N-GGVTEHNGNQLDKYRSI-C; hereinafter pSEB_116-132) to dampen SEB inflammatory activity. Treatment of T cells with pSEB_116-132 mimetic peptide significantly impaired inflammatory cytokine production induced by SEB ( Supplementary Figure S2 ). Moreover, when Caco-2 cells were co-cultured with SEB-stimulated T cells in the presence of pSEB_116-132 mimetic peptide, F-actin remodelling as well as E-CAD and ZO-1 down-regulation were impaired ( Figure 6B ). Finally, pSEB_116-132 mimetic peptide also restored ZO-1 and CDH1 gene expression ( Figure 6C ) and inhibited the up-regulation of SNAIL-1, TWIST-1 and ZEB-1 mRNA levels ( Figure 6D ).

Altogether these data disclose a novel pSEB_116-132 mimetic peptide that by competing with SEB for its binding to the homodimer interface of CD28 is able to dampen inflammatory cytokine production and associated intestinal epithelial cell dysfunctions.