A modified cholera toxin B subunit containing an ER retention motif enhances colon epithelial repair via an unfolded protein response

Abstract

Cholera toxin B subunit (CTB) exhibits broad‐spectrum biologic activity upon mucosal administration. Here, we found that a recombinant CTB containing an endoplasmic reticulum (ER) retention motif (CTB‐KDEL) induces colon epithelial wound healing in colitis via the activation of an unfolded protein response (UPR) in colon epithelial cells. In a Caco2 cell wound healing model, CTB‐KDEL, but not CTB or CTB‐KDE, facilitated cell migration via interaction with the KDEL receptor, localization in the ER, UPR activation, and subsequent TGF‐β signaling. Inhibition of the inositol‐requiring enzyme 1/X‐box binding protein 1 arm of UPR abolished the cell migration effect of CTB‐KDEL, indicating that the pathway is indispensable for the activity. CTB‐KDEL's capacity to induce UPR and epithelial restitution or wound healing was corroborated in a dextran sodium sulfate‐induced acute colitis mouse model. Furthermore, CTB‐KDEL induced a UPR, up‐regulated wound healing pathways, and maintained viable crypts in colon explants from patients with inflammatory bowel disease (IBD). In summary, CTB‐KDEL exhibits unique wound healing effects in the colon that are mediated by its localization to the ER and subsequent activation of UPR in epithelial cells. The results provide implications for a novel therapeutic approach for mucosal healing, a significant unmet need in IBD treatment.—Royal, J. M., Oh, Y. J., Grey, M. J., Lencer, W. I., Ronquillo, N., Galandiuk, S., Matoba, N. A modified cholera toxin B subunit containing an ER retention motif enhances colon epithelial repair via an unfolded protein response. FASEB J. 33, 13527‐13545 (2019). www.fasebj.org

ABBREVIATIONS

ATF6

activating transcription factor 6

BiP

binding immunoglobulin protein

CT

cholera toxin

CTA

cholera toxin A subunit

CTA1

catalytic domain of cholera toxin A subunit

CTB

cholera toxin B subunit

DAI

disease activity index

DSS

dextran sodium sulfate

EMEM

Eagle's minimum essential medium

EMT

epithelial‐to‐mesenchymal transition

ER

endoplasmic reticulum

FBS

fetal bovine serum

GM1 ganglioside

monosialotetrahexosylganglioside

H&E

hematoxylin and eosin

IBD

inflammatory bowel disease

IHC

immunohistochemistry

IRE1

inositol‐requiring enzyme 1

KDELR

KDEL receptor

PERK

protein kinase related‐like endoplasmic reticulum kinase

PLA

proximity ligation assay

qRT‐PCR

quantitative RT‐PCR

siRNA

small interfering RNA

siXBP1

X‐box binding protein 1‐targeted small interfering RNAs

T‐PER

Tissue Protein Extraction Reagent

UC

ulcerative colitis

UPR

unfolded protein response

XBP1

X‐box binding protein 1

Cholera toxin (CT) is classified as an AB5 toxin comprising one A subunit [cholera toxin A subunit (CTA)] and a pentameric B subunit [cholera toxin B subunit (CTB)] (1, 2). In the gut, CT binds to GM1 ganglioside (monosialotetrahexosylganglioside), via CTB, located on the plasma membranes of epithelial cells that line the luminal surface (3‐6). CTB contains 5 GM1‐binding sites that lie on the outer edge of each B subunit (7, 8). Once CT is bound to GM1 ganglioside (up to 5 gangliosides at once), it is endocytosed and trafficked via retrograde transport to the endoplasmic reticulum (ER) (9). Once in the ER, the catalytic domain of CTA (CTA1) is dissociated from CTB by protein disulfide isomerase. Subsequently, CTA1 enters the cytosol via the ER‐associated degradation pathway and returns to the inner surface of the plasma membrane to cause water secretion and diarrhea (2, 7, 10). Meanwhile, the fate of CTB after dissociation with CTA1 in the ER is not well understood. We have recently created a recombinant variant of CTB with a C‐terminal ER retention motif [CTB‐KDEL; previously termed CTBp (11)], which was demonstrated to possess GM1‐binding affinity and oral immunogenicity that are virtually equivalent to CTB. Additionally, we found that CTB‐KDEL could mitigate dextran sodium sulfate (DSS)‐induced acute colitis and reduced tumor development in an azoxymethane‐DSS model of colitis‐associated colon cancer in mice (12). These effects were associated with increased TGF‐β expression and activation, although the mechanism by which CTB‐KDEL up‐regulated TGF‐β remained elusive. While investigating the mechanistic principle, we recently discovered that the epithelial wound repair effects observed in the aforementioned studies are unique to CTB‐KDEL and not attained by CTB. The ER retention motif was initially added to CTB to improve recombinant production in tobacco plants (Nicotiana benthamiana), which aided in reducing ER stress that otherwise caused poor production yield (11, 13). It was thought that the KDEL sequence helped prolong CTB‐KDEL's residence time in the ER within plant cells to allow for proper folding and assembly. Additionally, the protein ER retention mechanism involving the KDEL receptor (KDELR) is highly conserved among eukaryotic organisms (14). Thus, the plant‐made protein's C‐terminal ER retention signal sequence KDEL, added to improve the protein's production in planta, could theoretically alter the protein's fate upon entering mammalian cells (2, 9, 15), which in turn might explain the unique colon mucosal healing activity of CTB‐KDEL.

Here, we investigated how the artificial ER retention motif of CTB‐KDEL contributes to the protein's mucosal healing activity in in vitro cell culture, in vivo mouse DSS acute colitis, and ex vivo human inflammatory bowel disease (IBD) colon explant models. The results have led to the discovery that an unfolded protein response (UPR) plays a critical role in the induction of intestinal mucosal healing, providing implications for novel therapeutic approaches for IBD.

MATERIALS AND METHODS

Animals

Eight‐week‐old C57BL/6J female mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Animal studies were approved by the University of Louisville's Institutional Animal Care and Use Committee.

CTB variants

Table 1 summarizes CTB variants used in this study. CTB‐KDEL and the non‐GM1‐binding mutant G33D‐CTB‐KDEL (12) were produced in N. benthamiana and purified to >95% homogeneity with an endotoxin level of <1 endotoxin units/mg, essentially as previously described (11, 16) but with modification in the final chromatography step to selectively purify the proteins with fully intact C terminus (manuscript in preparation). CTB, CTB‐KDE, and E. coli‐produced CTB‐KDEL (eCTB‐KDEL) were produced in E. coli BL21(DE3) according to Hamorsky and Matoba (17) and purified to >95% homogeneity with an endotoxin level of 0.05‐0.1 endotoxin units/mg. We have confirmed that endotoxin levels below 0.1 endotoxin units/mg of CTB variants do not affect experimental outcomes in Caco2 cells (data not shown). Purity and pentamer formation were assessed by SDS‐PAGE under denaturing and nondenaturing conditions, whereas the MW of each CTB variant was verified by mass spectrometry, as previously described by Hamorsky et al. (11).

Table 1.

Recombinant CTB variants used in this study

Protein name	C‐terminal KDEL sequence	Production system
CTB	No	Escherichia coli BL21(DE3)
CTB‐KDEL	Yes	Nicotiana benthamiana
CTB‐KDE	Lacking the terminal Leu residue	Escherichia coli BL21(DE3)
eCTB‐KDEL	Yes	Escherichia coli BL21(DE3)
G33D‐CTB‐KDEL	Yes	Nicotiana benthamiana

Caco2 wound healing assay

The Caco2 wound healing assay was performed as previously described by Baldauf et al. (12). Briefly, the cells were seeded and grown in complete growth medium [Eagle's minimum essential medium (EMEM) + 20% fetal bovine serum (FBS), 1x penicillin‐streptomycin] to confluence in 6‐well plates (Thermo Fisher Scientific, Waltham, MA, USA). Afterward, the culture medium was discarded, cells were washed with PBS and then incubated in serum‐free medium for 6 h. The cell monolayer was then scratched with a 200‐μl sterile beveled pipette tip to generate two 0.5‐1.0‐mm‐across linear wounds/well and washed with PBS. A test compound (CTB, CTB‐KDEL, CTB‐KDE, or eCTB‐KDEL at 0.1‐10 μM), a positive control [0.2 nM (5 ng/ml) recombinant TGF‐β1; Abcam, Cambridge, MA, USA], a vehicle control (PBS), or 4 μ8C (0.5 μM; MilliporeSigma, Burlington, MA, USA), or any combination thereof, were subsequently added in fresh serum‐deprived medium. Photomicrographs of the wounds were taken 0, 24, and 48 h after the wounding at ×4 magnification. Quantification of the remaining cell‐free area to the initial wound area was measured using the public domain software ImageJ (National Institutes of Health, Bethesda, MD, USA; http://rsbweb.nih.gov), and it was calculated as a mean percentage (n = 2 experimental replicates) per well. The culture medium/supernatants were collected from each well 24 or 48 h after wounding and stored at –80°C until analysis. The culture supernatants were analyzed by a human Cytokine‐Chemokine or TGF‐b1, 2, 3 Magnetic Bead Panel (MilliporeSigma). The panel was analyzed with a Milliplex MAP Kit on a MagPix with Luminex xMAP technology. Each experiment was performed with 4 biologic replicates per construct (n = 4).

Flow cytometry

Flow cytometry was used to assess the binding of CTB and CTB‐KDEL molecules to Caco2 cells according to a well‐established procedure. Briefly, 1 × 10⁵ cells were seeded in EMEM (serum free) in the presence of 1 μM CTB or CTB‐KDEL and incubated for 15 min on ice. Next, cells were washed and blocked with 3% BSA (MilliporeSigma) on ice. Then, cells were incubated with the rat anti‐CTB monoclonal antibody 9F9C7 (1.75 μg/ml), which was produced in‐house from a rat hybridoma cell line, for 1 h at room temperature and washed before incubation with a rabbit Alexa FluorTM 488 anti‐rat IgG (H + L) (A21210; Thermo Fisher Scientific) for 1 h on ice. Finally, cells were washed and analyzed with a FACSCalibur [Becton Dickinson (BD), San Diego, CA, USA], counting 10,000 cells/sample. Data were acquired with a BD FACSCanto II and analyzed with FlowJo v.10 data analysis software, using PBS as a negative control (n = 4).

Caco2 cell immunofluorescence

Cells were seeded 2 × 10⁵ cells/well in Lab‐Tek II chamber slides (Thermo Fisher Scientific) and grown for 72 h in complete growth medium. The medium was subsequently removed, and cells were washed with PBS followed by a test compound (CTB or CTB‐KDEL at 1 μM) or a vehicle control treatment in fresh serum‐deprived medium. Cells were incubated for 24 h at 37°C in humidified 5% CO2, then fixed, permeabilized, and stained. Briefly, after treatment cells were washed with PBS, they were fixed in 0.4 ml of 4% paraformaldehyde in PBS for 15 min at room temperature. Then, cells were permeabilized with 0.4 ml of 0.2% TitonX‐100 in PBS for an additional 15 min at room temperature. Subsequently, cells were blocked (3% BSA; MilliporeSigma) and treated with 4.4 μg/ml of the anti‐CTB monoclonal antibody 9F9C7 and 1 μg/ml of rabbit anti‐KDELR pan pAb (PA5‐67422; Thermo Fisher Scientific) for 3 h at room temperature. After subsequent washing (PBS), an anti‐Rat IgG (H + L) cross‐adsorbed secondary antibody, Alexa Fluor 488 (Thermo Fisher Scientific), and anti‐Rabbit preadsorbed IgG H&L (Alexa Fluor 555) were applied to cells for 3 h at room temperature, followed by additional washing with PBS. Cells were then mounted with coverslips using mounting medium for fluorescence with DAPI (Vectashield; Vector Laboratories, Burlingame, CA, USA). The proximity ligation assay (PLA) (Duolink PLA; MilliporeSigma) was performed according to the manufacturer's protocol using the same fixation, permeabilization, and primary antibody application protocol as previously described using 1 μg/ml of rabbit anti‐KDELR pan pAb (PA5‐67422; Thermo Fisher Scientific) and 1:200 goat anti‐CTB pAb (MilliporeSigma). Slides were analyzed by a wide‐field fluorescence confocal microscope (×60 magnification, z‐stacked images). Colocalization statistical analysis was performed using Imaris software (Bitplane, Belfast, United Kingdom), and values were calculated in PLA signals/cell.

Coimmunoprecipitation

Caco2 cells (1 × 10⁶) were seeded in a 10‐cm² plate and incubated in EMEM with 20% of FBS for 48 h. After changing the medium to EMEM without FBS supplement, 1 μM of G33D‐CTB‐KDEL, CTB or CTB‐KDEL, or a vehicle control (PBS) were added to the cells and incubated for 24 h. Cells were washed with PBS, and cell lysates were prepared in 500 μl of Tissue‐Protein Extraction Reagent (T‐PER) buffer (Thermo Fisher Scientific) supplied with 1x protease, phosphatase inhibitor (Thermo Fisher Scientific). After centrifugation at 13,000 g for 10 min, supernatants were mixed with 4 μg of an anti‐CTB antibody (7A12B3, which was produced in‐house from a rat hybridoma cell line) or anti‐BiP antibody (Cell Signaling Technology, Danvers, MA, USA) or an anti‐inositol‐requiring enzyme 1 (IRE1)α antibody (Cell Signaling Technology). After incubation at 4°C for 24 h, 10 μl of Protein A beads (Santa Cruz Biotechnology, Dallas, TX, USA) were added. After additional incubation at 4°C for 2 h, the mixture was washed with T‐PER buffer. The immunoprecipitated proteins were detected by an immunoblot analysis using 1 μg/ml of the rat anti‐CTB antibody 7A12B3 (produced in‐house), 2 μg/ml of an anti‐KDELR antiserum (Abcam), 2 μg/ml of an anti‐BiP antibody (Cell Signaling Technology), or 2 μg/ml of an anti‐IRE1α antibody (Cell Signaling Technology).

X‐box binding protein knockdown with small interfering RNA

In a 6‐well tissue culture plate, 2 × 10⁵ Caco2 cells were seeded per well in 2 ml of EMEM with 20% FBS and 1x penicillin‐streptomycin. Small interfering RNA (siRNA) was used to silence X‐box binding protein 1 (XBP1). The human XBP1 siRNA and scrambled sequences siRNA (control) were purchased from Santa Cruz Biotechnology. When cells were 60‐80% confluent, 1 μg of siRNAs was transfected by lipofectamine transfection reagents (Santa Cruz Biotechnology). Subsequently, the wound healing assay was performed as previously described, and total RNA was isolated by RNeasy Microarray Tissue Kit (Qiagen, Germantown, MD, USA) to verify XBP1 levels using the XBP‐1 (h) primer set supplied from Santa Cruz Biotechnology.

Immunoblot analysis

When Caco2 cells were 60‐80% confluent in EMEM supplemented with 20% FBS and 1x penicillin‐streptomycin, the medium was changed to EMEM without FBS supplement and treated 1 μM of G33D‐CTB‐KDEL, CTB or CTB‐KDEL, or a PBS vehicle control. After 24‐h incubation, cell lysates were prepared with T‐PER buffer supplied with 1x protease, phosphatase inhibitor (Thermo Fisher Scientific). Cell debris were removed by centrifugation at 13,000 g for 10 min at 4°C. A total of 100 μg of total proteins were loaded for SDS‐PAGE analysis, using a 10% of Bolt Bis‐Tris Plus Gel (Thermo Fisher Scientific) to separate large‐size phosphorylated proteins in 2‐(N‐morpholino) ethanesulfonic acid (MES)‐based running buffer. To analyze the intracellular stability of CTB and CTB‐KDEL, Caco2 cells were incubated 4 h with 2 μM of CTB or CTB‐KDEL, washed, and immediately analyzed or incubated in EMEM for an additional 12 or 24 h before analysis by immunoblot using 1.75 μg/ml of the rat anti‐CTB monoclonal antibody 7A12B3. To detect UPR markers, we used 2 μg/ml of an anti‐BiP antibody (Cell Signaling Technology), an anti‐IRE1α antibody (Cell Signaling Technology), or an anti‐protein kinase related‐like ER kinase (PERK) antibody (Cell Signaling Technology). Loading control was β‐actin, which was detected by 1 μg/ml of an anti‐β‐actin antibody (Santa Cruz Biotechnology). For mouse primary colon epithelial cells, cells were isolated from 3 to 6 cm of the colon from 4 mice. Colons were cut open lengthwise, washed, and then cut into 2‐mm pieces. The tissues were washed 20 times in cold PBS, followed by a 20‐min incubation at room temperature in 50‐ml conical tubes containing 30 ml of prewarmed Ca²⁺, Mg²⁺ free Hank's Balanced Salt Solution (CMF HBSS)/0.1% FBS and 2 mM EDTA on a rocking platform at 20 rmp. The tissues were allowed to settle by gravity for 30 s, and then the supernatant was discarded. To separate epithelial cells, the tissues were resuspended in 10 ml of cold (4°C) CMF HBSS/1.5% FBS and pipetted up and down 3 times viciously. The intestinal pieces were allowed to settle, and then the supernatant was filtered through a 70‐μm filter. This cell separation step followed by filtration was repeated 3 times, creating 4 fractions of filtered supernatants. Fractions 3 and 4 were combined and centrifuged at 300 × g for 5 min at 4°C. Supernatants were carefully discarded, and the pellet was resuspended in 10 ml of cold (4°C) EMEM with 20% FBS followed by a subsequent 200× g centrifugation. Finally, the cells were resuspended in EMEM with 20% FBS, counted, and plated. Thereafter, the mouse primary colon epithelial cells were subjected to the same immunoblot analysis protocol as previously mentioned for Caco₂ cells.

The DSS‐induced acute colitis study

For the acute DSS recovery model of intestinal injury, 10 mice/group, randomly assigned, were used. DSS exposure was initiated on the day mice turned 9 wk old (d 0), using a modified method (12). Body weights were measured at the initiation of DSS exposure as a baseline and every morning thereafter to determine percent change. Animals received 3% DSS (MW 36,000 to 50,000; MP Biomedicals, Santa Ana, CA, USA) in drinking water for 7 d. On the seventh day of DSS exposure, animals were gavaged with PBS, CTB, or CTB‐KDEL after sodium bicarbonate administration, as previously described by Hamorsky et al. (11); the animals were allowed to recover for 7 d, during which they received normal drinking water.

Histology

Colons were removed and washed with PBS. A portion of the distal colon was fixed with paraformaldehyde overnight and stored in 70% ethanol until paraffin embedding, sectioning, and routine hematoxylin and eosin (H&E) staining. Inflammation scoring was performed as previously described by Baldauf et al. (12). Tissue sections from 10 mice were scored in a blinded manner and averaged for each group.

Immunohistochemistry

Colons were removed and washed with PBS (n = 10). A portion of the distal colon was fixed with paraformaldehyde overnight and stored in 70% ethanol until paraffin embedding and sectioning. Sections were deparaffinized with Citrisolv and rehydrated through several ethanol washing steps ending with incubation in distilled water. Antigen retrieval was performed overnight with a 2100 Retriever (Electron Microscopy Sciences, Hatfield, PA, USA) using a pH 6.0 buffer. Tissue sections were blocked for endogenous peroxidase, avidin, biotin, and serum from the animal in which the secondary antibody was raised. Primary antibody (anti‐E‐cadherin 1:100; Abcam) was incubated with the tissue sections for 1 h at room temperature. The Vectastain Elite ABC kit (goat anti‐rabbit; Vector Laboratories) was used to label the primary antibody. E‐cadherin + cells were visualized with the ImmPACT DAB Substrate Kit (Vector Laboratories), dehydrated through an ethanol gradient, and finally incubated with Citrisolv. Sections were scanned using an Aperio ScanScope CS (Leica Biosystems, Buffalo Grove, IL, USA).

RNA isolation

Distal colon tissue (~14 mg) from each animal, initially stored in RNAlater (Qiagen) at –20°C, was placed in Qiazol lysis reagent and homogenized. An RNeasy Microarray Tissue Kit from Qiagen was used to purify the RNA from the tissue homogenate. RNA concentration, quality, and purity were confirmed by spectrophotometer and then stored at –80°C until use.

Quantitative RT‐PCR

First strand cDNA was obtained from reverse transcription of 100 ng RNA using a SuperScript Vilo cDNA Synthesis Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Template cDNA were added to a reaction mixture containing RT² SYBR Green Master Mix (Qiagen) and loaded in RT² Profiler PCR Array Standard 96‐well plates (Qiagen). These plates contain prespotted individual gene expression probes for the detection of genes of interest as well as the housekeeping genes 18S, ACTB, and GAPDH. Gene quantification was carried out on a 7500 Fast Real‐Time PCR System (Thermo Fisher Scientific) with the following conditions: 1 cycle (10 min at 95°C); 40 cycles (15 s at 95°C; 1 min at 60°C). The 7500 Software v.2.0.6 (Thermo Fisher Scientific) was used to determine the C_t for each reaction and derive the expression ratios relative to control. Wound healing pathway analysis was performed with an RT² Profiler PCR Mouse Wound Healing Array (Qiagen) under the same conditions as those previously described.

Treatment and culturing of colon explants obtained from patients with ulcerative colitis

All research involving human tissue was performed in accordance with relevant guidelines and regulations established by the University of Louisville Institutional Review Board Committee. The treatment and culturing of colon explants was performed using an immersion culturing system developed from previously described methods (18‐20). Colorectal tissues were obtained from consenting patients at the time of colectomy. Immediately after excision, colectomy tissue was placed in complete medium [Roswell Park Memorial Institute (RPMI) 1640 supplemented with 2 mM l‐glutamine, 10 mM 4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid (HEPES) buffer, 10 μg/ml gentamicin, 100 U/ml each of penicillin and streptomycin, and 10% FBS (HyClone; Thermo Fisher Scientific)] on ice and transported to the lab. The tissue was immediately divided and placed in organ culture dishes at 37°C in humidified 5% CO₂. The tissues were placed luminal side up in 2‐ml complete growth medium. Cultures were incubated with or without the addition of PBS, CTB, or CTB‐KDEL (1 μM) at 37°C for 24 h. Thereafter, the supernatant was stored in aliquots at –80°C, tissues were washed in complete medium and homogenized for gene expression analysis, formalin‐fixed for histopathology, or frozen in liquid nitrogen for additional analysis as necessary.

Statistics

For all data, outliers were determined by statistical analysis using the Grubb's test (P < 0.05) and excluded from further analysis. Graphs were prepared and analyzed using Prism v.5.0 (Graphpad Software, La Jolla, CA, USA). To compare 2 data sets, an unpaired, 2‐tailed Student's t test was used. To compare 3 or more data sets, 1‐way ANOVA with Bonferroni's multiple comparison posttest or Kruskal‐Wallis test with Dunn's multiple comparison posttest were performed. For body weights and disease activity index (DAI) results, a 2‐way ANOVA with Bonferroni's multiple comparison posttest was employed. A value of P < 0.05 was considered significant.

RESULTS

The C‐terminal KDEL sequence is essential for the colon epithelial repair activity of CTB‐KDEL

To first determine if the previously described colon mucosal healing effect is unique to CTB‐KDEL, we employed a human colon epithelial cell line Caco2 model of wound healing (12). Cell migration (i.e., wound closure) over 48 h was tested using 3 μM of CTB‐KDEL, CTB, or PBS. As shown in Figure 1 A–C, the results revealed that CTB had no effect beyond the natural healing response noted in the PBS‐treated group, whereas CTB‐KDEL treatment significantly enhanced wound closure both at 24 and 48 h after wounding similarly to the recombinant TGF‐β positive control (Fig. 1 B). This enhanced cell migration effect was observed in a broad concentration range of CTB‐KDEL, from 3 μM down to 0.1 μM. On the other hand, CTB had no effect at the same concentrations tested (Fig. 1 C). We have previously demonstrated in the same model that CTB‐KDEL's cell migration effect was mediated by TGF‐β because an anti‐TGF‐β antibody completely inhibited the effect (12). Thus, TGF‐β levels in the culture supernatants were measured using a multiplex bead array. In contrast to CTB, CTB‐KDEL significantly increased TGF‐β1and TGF‐β2 levels 24 h after wounding (Fig. 1 D). To rule out the possibility that the difference observed between CTB‐KDEL and CTB was the result of differential binding to Caco2 cells, we performed a flow cytometry analysis where Caco2 cells were treated with 1 μM CTB‐KDEL or CTB. Both CTB‐KDEL and CTB showed high and similar binding, with >75% of Caco2 cells showing positive (Fig. 1 E). Consistent with these results, CTB‐KDEL and CTB have shown comparable GM1‐binding affinity in ELISA, surface plasmon resonance, and flow cytometry analysis using the mouse macrophage cell line RAW264.7 (11). Together, these results indicate that CTB‐KDEL, but not CTB, facilitates TGF‐β–mediated epithelial wound healing in the Caco2 model. Because CTB‐KDEL and CTB equally bound to the epithelial cell, the discrepancy can be attributed to distinct intracellular events, which are likely triggered upon their internalization and retrograde transportation.

Figure 1.

CTB‐KDEL, but not CTB, enhances cell migration and TGF‐β levels in a human colon epithelial cell wound healing model. Caco2 cells were grown to confluence and scratched with a pipette tip. Cells were then incubated with PBS, CTB, CTB‐KDEL, or TGF‐β1. The in vitro wound closure was recorded over 48 h and original magnification, ×4 images were acquired with the EVOS FL imaging system (Thermo Fisher Scientific) and mean percentage closure was determined by ImageJ software. A) Representative photomicrographs of Caco2 cells treated with PBS, CTB‐KDEL, or CTB at 0 and 24 h after scratch. Scale bars, 1 mm. B) Analysis of “wound closure” induced by 3 μM CTB‐KDEL or CTB, or 0.2 nM (5 ng/ml) TGF‐β1 over 48 h, compared with a PBS vehicle control. C) Percent increase in wound closure by varying concentrations of CTB or CTB‐KDEL compared with PBS after 24 h. D) TGF‐β1 and TGF‐β2 protein concentrations in Caco2 cell supernatants. E) Caco2 cell flow cytometry analysis. Cells (1 × 10⁵) were treated with 1 μM CTB or CTB‐KDEL, or PBS for 15 min on ice, then fixed and stained with a rat anti‐CTB monoclonal antibody. F) Percent increase in wound closure by 1 μM of CTB‐KDEL, CTB‐KDE, or CTB compared with PBS after 24 h. Bars represent means ± sem of 4 independent analyses. One‐way ANOVA with Bonferroni's multiple comparison tests was used to compare all pairs of groups (B–E). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. ^# P < 0.05, ^## P < 0.01, ^### P < 0.001 when comparing matching concentrations between CTB and CTB‐KDEL (i.e., 0.3 μM CTB is significantly different (P < 0.05) from 0.3 μM CTB‐KDEL).

The previous results led us to hypothesize that the C‐terminal KDEL sequence is essential to CTB‐KDEL's wound healing activity. First, in order to rule out the possibility that the unique activity is not attributed to the N. benthamiana production system, CTB‐KDEL was produced in Escherichia coli and tested in the Caco2 cell wound healing assay. Indeed, eCTB‐KDEL significantly promoted wound healing 24 h after wounding in contrast to the non‐KDEL counterpart (data not shown). To further delineate the importance of the ER retention motif, a variant lacking the C‐terminal Leu residue (CTB‐KDE) was made. As shown in Figure 1 F, CTB‐KDE failed to promote wound healing unlike CTB‐KDEL in a Caco2 cell wound healing assay.

We next explored if the ER retention motif of CTB‐KDEL has any influence on the protein's residence within epithelial cells. Because of the known function of KDELRs (21‐23), it was hypothesized that CTB‐KDEL could localize and remain in the ER for an extended period after retrograde transport (15). To test this hypothesis, Caco2 cells were treated with CTB‐KDEL or CTB and then washed and incubated for 0, 12, and 24 h, followed by an immunoblot analysis to detect CTB and CTB‐KDEL in the cell lysates. As shown in Figure 2 A, CTB‐KDEL was continuously detected in the cell lysates over 24 h, whereas CTB showed decreasing levels, with only a marginal amount detected at 24 h after treatment. To discern the location of CTB‐KDEL and CTB in the cells, we used confocal microscopy, in which cells were treated with CTB‐KDEL or CTB for 24 h and then fixed and stained for CTB, the ER, the KDELR, and nuclei. As depicted in Figure 2 B, CTB‐KDEL was detected much more prevalently inside Caco2 cells at 6 and 24 h after treatment. Because CTB‐KDEL and CTB showed similar binding to the surface of Caco2 cells (Fig. 1 E), it is likely that the difference in their amounts detected within Caco2 cells occurred after internalization. Confocal microscopy showed the colocalization of CTB‐KDEL and KDELR 24 h after treatment, whereas CTB was hardly detected (Fig. 3 A). This was also corroborated by a PLA, which reveals protein‐protein interactions that are within a 40‐nm proximity (24) (Fig. 3 B, C). The interaction of CTB‐KDEL, CTB‐KDE, or CTB with KDELR in Caco2 cells was also confirmed by coimmunoprecipitation (Fig. 3 D). Although CTB‐KDEL showed clear binding to the KDELR, such interaction was not observed with CTB or CTB‐KDE. Collectively, these results demonstrate that CTB‐KDEL has a prolonged residence within Caco2 cells because of its interaction with the KDELR.

Figure 2.

Stability of CTB and CTB‐KDEL in Caco2 cells. A) After 4‐h incubation with 2 μM of CTB or CTB‐KDEL, Caco2 cells were washed with EMEM, and then, cells were lysed at different timings (0, 12, and 24 h). Anti‐CTB antibodies showed CTB and CTB‐KDEL proteins in Western blot (anti‐CTB) and actin antibodies detected actin proteins (anti‐β‐actin). B) Immunofluorescence analysis of CTB‐KDEL or CTB intracellular localization within the ER. CTB‐KDEL accumulates within Caco2 cells in contrast to CTB. Caco2 cells were treated with 1 μM PBS, CTB, or CTB‐KDEL for 6 or 24 h. Cells were fixed/permeabilized and stained with anti‐CTB mAb detected by Alexa FluorTM 488 (green), ER‐selective red dye (red), and DAPI (blue). Slides were analyzed by a wide‐field fluorescence confocal microscope. Original magnification, ×60 z‐stacked images. Scale bars, 20 μm.

Figure 3.

CTB‐KDEL binds to the KDELR. A) Immunofluorescence analysis of CTB‐KDEL or CTB intracellular localization within the ER. Caco2 cells were treated with 1 μM of CTB or CTB‐KDEL, or a vehicle control (PBS) for 24 h. Cells were fixed/permeabilized and stained with an anti‐CTB monoclonal antibody detected by Alexa Fluor 488 (green), anti‐KDEL polyclonal antibodies (red), and DAPI (blue). Scale bars, 10 μm. B) PLA of CTB‐KDEL or CTB protein‐protein interaction with KDELR within the ER. Caco2 cells were treated with 1 μM of CTB or CTB‐KDEL, or a vehicle control (PBS) for 24 h. Scale bars, 10 μm. A, B) Slides were analyzed by a wide‐field fluorescence confocal microscope. Original magnification, ×60 z‐stacked images. C) PLA signals/cell (n = 3). D) Coimmunoprecipitation (Co‐IP) of KDELR with CTB or CTB‐KDEL. Caco2 cell lysates were incubated with 4 μg of CTB or CTB‐KDEL for 24 h. A total of 10% of mixture was presented in total lanes. CTB antibodies and Protein A beads were loaded for immunoprecipitation. After 3 wash steps, beads in sample buffer were boiled at 95°C for 10 min; then, supernatants were loaded in IP lanes.

CTB‐KDEL induces a UPR and IRE1‐XBP1 signaling in CACO2 cells

It is well known that accumulation of aberrant proteins in the ER can lead to a UPR and subsequently activate various intracellular signaling pathways (25‐27). Such aberrant proteins would usually be unfolded and misfolded polypeptides of endogenous origin, but adventitious proteins such as CTB may also induce UPR (15, 28). There are 3 branches of UPR known in mammalian cells, including IRE1, PERK, and activating transcription factor 6 (ATF6; a and β isoforms) (29, 30). Among these, the IRE1‐XBP1 pathway has been linked to wound healing and mitigation of DSS colitis (31‐33). Therefore, we tested whether CTB‐KDEL induces UPR and IRE1‐XBP1 signaling in Caco2 cells. A quantitative RT‐PCR (qRT‐PCR) analysis showed that, when cells were treated with PBS, CTB, or CTB‐KDEL for 24 h, all 3 sensors of the UPR pathway (ATF6, PERK, IRE1), as well as spliced XBP1 and TGF‐B1, were significantly up‐regulated in CTB‐KDEL‐treated cells when compared with PBS or CTB, or both (Fig. 4 A). Congruent with the qRT‐PCR results, an immunoblot analysis revealed significantly increased IRE1α, phosphorylated (p‐) IRE1α, PERK, and binding immunoglobulin protein (BiP) levels in CTB‐KDEL–treated cells, but not in PBS, CTB, or G33D‐CTB‐KDEL–treated cells (Fig. 4 B, C).

Figure 4.

CTB‐KDEL induces the UPR. Caco2 cells were grown to confluence and scratched. Cells were then incubated with PBS, CTB, or CTB‐KDEL, or any combination thereof. A) qRT‐PCR analysis of UPR gene expression in Caco2 cells 6 h or 24 h after treatment. One‐way ANOVA with Bonferroni's multiple comparison tests was used to compare all pairs of groups. Means ± sem are shown for each group (n = 4). B) Immunoblot analysis of ER stress and UPR‐related proteins. Caco2 cells were incubated for 24 h with PBS, 1 μM of G33D‐CTB‐KDEL (a non‐GM1 binding CTB‐KDEL mutant), CTB, or CTB‐KDEL. A total of 200 μg of total protein were loaded for immunoblot analysis. β‐Actin was used as a loading control. A representative image from 3 independent analyses is shown. C) The intensities of Western blot bands were measured by ImageJ program. All graphs showed relative values and error bar means sd (n = 3). D) Coimmunoprecipitation (Co‐IP) of BiP with CTB or CTB‐KDEL. For 1 or 24 h, Caco2 cells were incubated with 1 μM CTB and 1 μM CTB‐KDEL. A total of 20% of intracellular extracts were loaded in total lanes (total 20%). Intracellular extracts were incubated with 4 μg of BiP antibodies for 2 h at 4°C. A total of 10 μl of Protein A beads were employed to precipitate the binding complex of BiP. After washing, samples were loaded in IP lanes. Anti‐CTB antibodies showed CTB and CTB‐KDEL proteins in Western blot (anti‐CTB) and BiP antibodies detected BiP proteins (anti‐BiP). E) Co‐IP of IRE1α with BiP. For 24 h, Caco‐2 cells were incubated with PBS buffer, 1 μM G33D CTB mutants (G33D CTB‐KDEL), 1 μM CTB, and (continued on next page) 1 μM CTB‐KDEL. A total of 10% of intracellular extracts were loaded in total lanes (total 10%). Intracellular extracts were incubated with 4 μg of IRE1α antibodies for 2 h at 4°C. A total of 10 μl of Protein A beads were employed to precipitate the binding complex of IRE1 α. After washing, samples were loaded in IP lanes. Anti‐IRE1 α antibodies showed IRE1 α proteins in Western blot (anti‐IRE1α) and BiP antibodies detected BiP proteins (anti‐BiP).

The onset of UPR is preceded by the dissociation of BiP from the UPR sensors (ATF6, PERK, IRE1), which is caused by the attraction of the molecular chaperone by aberrant proteins in the ER (34, 35). Accordingly, we examined the binding of BiP with CTB‐KDEL in Caco2 cells. A coimmunoprecipitation assay using an anti‐BiP antibody revealed that, when the cells were incubated with CTB or CTB‐KDEL for 1 h, both proteins formed a complex with BiP (Fig. 4 D). However, after 24‐h incubation, only CTB‐KDEL was shown to interact with the ER chaperone (Fig. 4 D). These results indicate that the CTB domain has the capacity to interact with BiP and that the retention of CTB‐KDEL in the ER (by the KDELR) prolonged their interaction. To confirm this observation, we monitored the amounts of IRE1‐BiP complex in a coimmunoprecipitation assay using an anti‐IRE1a antibody after incubating Caco2 cells with G33D‐CTB‐KDEL, CTB or CTB‐KDEL, or PBS (Fig. 4 E). The results revealed that the association between BiP and IRE1α was significantly reduced by CTB‐KDEL, but not by PBS, G33D‐CTB‐KDEL, or CTB.

Collectively, the previous findings strongly suggest that CTB‐KDEL's prolonged residence in the ER induces a UPR in Caco2 cells, which may be in part caused by the protein's interaction with the ER chaperone BiP. To test the impact of the UPR on CTB‐KDEL's wound healing effect, we treated Caco2 cells with CTB‐KDEL in the presence or absence of 4μ8C, an inhibitor of IRE1‐mediated XBP1 splicing [and hence inhibits the activation IRE1‐XBP1 signaling pathway (36)], after scratching the cell monolayer. The analysis revealed that 4μ8C completely blocked CTB‐KDEL's cell migration effects; whereas CTB‐KDEL treatment significantly enhanced wound closure, CTB‐KDEL + 4μ8C and 4μ8C‐treated groups showed no effect compared with the PBS control group (Fig. 5 A). Furthermore, as shown in Figure 5 B, CTB‐KDEL treatment significantly increased TGF‐β1 and TGF‐β2 levels when compared with PBS, CTB‐KDEL + 4μ8C, or 4μ8C treatment, indicating that CTB‐KDEL increases TGF‐β1 and TGF‐β2 levels and subsequent cell migration via the IRE1‐XBP1 signaling pathway. These results are further supported by an XBP1 knockdown experiment, where Caco2 cells were transfected with control siRNAs or XBP1‐targeted siRNAs (siXBP1), and then cells were scratched and treated with CTB‐KDEL or CTB. The results clearly showed that siXBP1 treatment significantly reduced CTB‐KDEL‐induced up‐regulation of XBP1 levels as well as cell migration activity (Fig. 5 C, D).

Figure 5.

CTB‐KDEL's epithelial wound healing activity is mediated through IRE1‐XBP1 signaling. A) Caco2 cells were grown to confluence and scratched with a pipette tip. Cells were then incubated with PBS, CTB, CTB‐KDEL, or 4 μ8C, or any combination thereof. Wound closure over 24 h was analyzed. Mean percentage closure was determined by ImageJ software (n = 4). B) TGF‐β concentrations in Caco2 cell supernatants from wound healing assay in A. Means ± sem are shown for each group (n = 4). C) Impact of XBP1 knockdown on CTB‐KDEL's wound healing activity. Caco2 cells (n = 6) were transfected with control siRNAs (siControl) or siXBP1. After 72 h, cells were scratched and treated with 1 μM of CTB or CTB‐KDEL. The wound closure was measured at 24 h. D) The transcript levels of XBP1 were measured by qRT‐PCR, using 18S rRNA as a reference. One‐way ANOVA with Bonferroni's multiple comparison tests was used. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

CTB‐KDEL induces a UPR in primary colon epiethelial cells and facilitates colonic wound healing after DSS‐induced acute colitis in mice

To determine if the previously described mechanism by which CTB‐KDEL induces cell migration in vitro is applicable to intestinal wound healing in vivo, we isolated and treated mouse primary colon epithelial cells with PBS, CTB‐KDEL, CTB, or G33D‐CTB‐KDEL for 24 h. An immunoblot analysis revealed that CTB‐KDEL significantly increased IRE1α, p‐IRE1α, PERK, and BiP levels compared with all other treatment groups (Fig. 6 A, B). Albeit to a lesser extent, CTB also significantly increased the levels of IRE1α and BiP when compared with PBS; however, CTB did not increase the p‐IRE1α level. Next, we employed a mouse acute DSS colitis model to compare the wound healing efficacies of CTB‐KDEL and CTB. Female C57bl/6 mice were given 1 dose of PBS, 3 mg CTB‐KDEL or 3 mg CTB via gavage after a 7‐d 3% DSS exposure period (d 7), at which the onset of colonic epithelial damage had taken place (37). As shown in Figure 7 A, CTB‐KDEL–dosed mice showed a significantly more rapid recovery from DSS‐induced weight loss than PBS and CTB‐dosed groups. This trend was noted as early as 3 d after administration (d 10) through d 14, at which the mice were euthanized. The accelerated recovery in body weight was accompanied by a significant protection from colon shrinkage as well as a significant decrease in DAI and histopathological crypt scores (Fig. 7 B). Furthermore, a complete blood count analysis of terminal blood samples (collected on d 14) showed that CTB‐KDEL treatment significantly reduced the DSS‐induced escalation of monocyte, basophil, and eosinophil levels compared with PBS‐dosed mice. Neutrophil levels were significantly higher in PBS and CTB‐dosed mice than those of the healthy control animals, but not in CTB‐KDEL‐dosed animals (Fig. 7 C). Similar results were also obtained in male C57bl/6 mice, in which CTB‐KDEL, but not CTB, significantly facilitated recovery from acute DSS colitis on the basis of DAI, colon length, and crypt scores, suggesting that there is no sex‐related difference in the therapeutic effects of CTB‐KDEL (Fig. 7 D).

Figure 6.

Immunoblot analysis of UPR‐related proteins. Mouse primary colon epithelial cells were incubated for 24 h with PBS, or 1 μM of G33D‐CTB‐KDEL (a non‐GM1 binding CTB‐KDEL mutant), CTB or CTB‐KDEL. A) A total of 200 μg of total protein were loaded for immunoblot analysis. β‐Actin was used as a loading control. A representative image from 3 independent analyses is shown. B) The band intensities of Western blot bands were measured by ImageJ. All graphs show relative values. Bars represent means ± sem (n = 3).

Figure 7.

Effects of orally administered CTB or CTB‐KDEL in an acute DSS colitis model. Mice (female and male C57BL/6J, 9‐wk old) were exposed to 3% DSS for 7 d and orally administered with PBS, CTB, or CTB‐KDEL on d 7. Colon tissues were isolated after a 7‐d recovery for analyses. Means ± sem are shown for each group; n = 10 animals/group. A) Femalemicepercent change of body weights. Animals were weighed daily and just prior to the initiation of DSS exposure. Percent change was based on the initial body weight. 2‐way ANOVA with Bonferroni's multiple comparison tests. ∗P < 0.05 between DSS‐exposed, CTB‐KDEL, and PBS‐administered groups; ∗∗∗∗P < 0.0001 when DSS+3ug CTB‐KDEL is compared to DSS PBS; ^# P < 0.05 between DSS‐exposed, CTB‐KDEL, and CTB‐administered groups; ^## P < 0.01, ###P < 0.001 when DSS+3μg CTB‐KDEL is compared to DSS+3μg CTB. B) Top‐right: female mice DAI scores calculated from body weight loss, fecal consistency, and occult blood scores at the time of euthanization. Bottom left: colon length measured at euthanization. Bottom right: colon crypt scoring from paraffin embedded tissue sections were scored after staining with H&E. Scoring was based on 0‐4 scale. Top‐right: colon length. C) Female mice complete blood count analysis performed at the time of euthanization. Means ± sem are shown for each group (n = 5). D) Male mice (male C57BL/6J, 9‐wk old) DAI scores, colon length, and colon crypt score. Means ± sem are shown for each group (n = 10). Bonferroni's multiple comparison tests was used to compare all pairs of groups. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001).

Additional histopathological examination on H&E‐stained distal colon tissues of the female mice revealed that CTB‐KDEL treatment protected from ulceration, inflammation, and loss of colonic epithelial surface and crypts, in contrast to PBS and CTB treatment (Fig. 8 A). Moreover, immunohistochemistry (IHC) staining for an epithelial marker, E‐cadherin (38, 39), on the same tissues used for histopathological examination, clearly revealed that CTB‐KDEL‐treated mice had an epithelial surface that has recovered (or maintained) near‐normal thickness, with ongoing crypt regeneration noted throughout the tissue (Fig. 8 A). Consistent with these findings, a qRT‐PCR analysis revealed that CTB‐KDEL administration significantly up‐regulated multiple genes associated with epithelial repair, including Cdh1, Wnt5a, and Col5a3 (Fig. 8 B) (40‐43). Collectively, these results demonstrate CTB‐KDEL's unique ability to mitigate DSS colitis and facilitate epithelial repair in mouse colons. Given that CTB‐KDEL activated a UPR in primary mouse colon epithelial cells (Fig. 6), the enhanced epithelial repair was likely mediated via the UPR, or specifically the IRE1 pathway, as demonstrated in the Caco2 model (Fig. 5).

Figure 8.

CTB‐KDEL and CTB effects on mouse colon histologic alterations induced by DSS colitis. CTB‐KDEL treatment protected mice from ulceration, inflammation, and loss of colonic epithelial surface and crypts. A) Top: representative ×4 (left) and ×10 (right) photomicrographs of H&E‐stained distal colon tissues from each treatment group. Bottom: IHC staining of E‐cadherin positive cells in distal colon tissue. Representative photomicrographs are shown. B) qRT‐PCR analysis of cytokine gene expression in mouse colon tissue. Means ± sem are shown for each group (n = 4).

CTB‐KDEL induces a wound healing response in human IBD colon tissues

We next explored whether the UPR‐mediated epithelial cell migration effects in Caco2 cells and epithelial recover activity in mice of CTB‐KDEL could translate into a therapeutic effect in human colon tissues ex vivo. A sigmoid colon tissue of a 57‐y‐old male patient with ulcerative colitis (UC) who underwent total colectomy was sectioned into 9 pieces, which were cultured with PBS, CTB‐KDEL, or CTB (n = 3 for each treatment) for 24 h, and then UPR, TGF‐β, and wound healing signaling were analyzed by pathway‐focused qRT‐PCR arrays. Consistent with results observed in Caco2 cells (Fig. 4), CTB‐KDEL‐treated tissue showed a significant up‐regulation of the IRE1 gene ERN1, PERK gene ETF2AK3, and TGF‐B1, whereas there was no significant difference between PBS‐ and CTB‐treated tissues (Fig. 9 A). Additionally, CTB‐KDEL significantly up‐regulated 21 out of 84 wound healing‐related genes in the qRT‐PCR array by >2‐fold compared with PBS, whereas CTB significantly enhanced only 1 gene (MAPK1; Fig. 9 B). Of note, one of the 21 genes significantly increased by CTB‐KDEL was the E‐cadherin gene CDH1, which was also up‐regulated in the acute DSS model (Fig. 8 B). Congruent with the qRT‐PCR results, histopathological analysis showed that CTB‐KDEL‐treated tissues (n = 3) had early stage crypt formations and relatively low neutrophil infiltration in the mucosa, whereas PBS‐ and CTB‐treated tissues did not have such distinct histologic features (Fig. 9 C). Because CTB‐KDEL‐treated tissue was cut adjacent from PBS or CTB‐treated ones, these observations likely reflect the effects of treatment.

Figure 9.

Effects of CTB‐KDEL and CTB in human colon tissue isolated from a 57‐y‐old male patient with UC who underwent colectomy. CTB‐KDEL induces UPR, TGF‐β, and wound healing signaling in human colon colectomy tissue. The sigmoid colon tissue was divided into 9 adjacent sections and cultured with PBS, or 1 μM of CTB‐KDEL or CTB for 24 h. A) qRT‐PCR analysis of UPR and wound healing gene expression. One‐way ANOVA with Bonferroni's multiple comparison tests was used to compare all pairs of groups. Means ± sd are shown for each group (n = 3). B) Wound healing pathway‐focused qRT‐PCR analysis of gene expression in human patient UC colectomy tissue. Volcano plots show P value (y axis) and fold change (x axis) for all 84 genes in the wound healing pathway upon treatment with PBS vs. CTB‐KDEL (left) and PBS vs. CTB (right). Dots represent ≥2‐fold up‐regulation (red), ≥2‐fold down‐regulation (green), or <2‐fold change (black). The area corresponds to >2‐fold increase with values of P < 0.05 highlighted with a red box. Values were calculated by the Qiagen Gene Globe Design and Analysis Hub (http://www.qiagen.com/geneglobe); n = 3/treatment. C) Human patient UC colectomy tissue H&E stain. Representative ×4 (left) and ×10 (right) photomicrographs of H&E‐stained sigmoid colon tissues from each group. Arrows point to crypt formations.

CTB‐KDEL's wound healing effects were further validated in additional IBD colon tissues. Altogether, colectomy tissues from 5 patients were treated with CTB‐KDEL or vehicle control (PBS) ranging from 20 to 72 y old, including 2 patients with Crohn's disease (a 57‐y‐old male and a 60‐y‐old female). Albeit to varying degrees, CTB‐KDEL significantly up‐regulated multiple wound healing‐associated genes by over 2‐fold in all 5 patients' tissues (Fig. 10 A). In contrast, there was no gene in the wound healing pathway analyzed that was significantly down‐regulated by CTB‐KDEL in any of the tissues (data not shown). The most notable effect was observed in a sigmoid colon tissue from a 20‐y‐old male patient with UC, in which CTB‐KDEL significantly enhanced as many as 79 out of 84 wound healing‐associated genes analyzed (Fig. 10A). As observed in the mouse DSS model (Fig. 8 B), CTB‐KDEL significantly increased CDH1, TGF‐B1, and WNT5A (Fig. 10 B). The robust up‐regulation of wound healing genes was accompanied with distinct crypt formations and relatively low neutrophil infiltration in the mucosa of CTB‐KDEL–treated tissues in contrast to those treated with PBS. Ki67 IHC analysis corroborated the histopathological observation, which depicts live, proliferating crypt formation in CTB‐KDEL treat tissues (n = 3) (Fig. 10 C). Taken together, the previous results provide evidence that CTB‐KDEL can exhibit wound healing activity in human IBD colon tissues, highlighting the protein's therapeutic potential.

Figure 10.

Effects of CTB‐KDEL and CTB in human colon tissue from patients with IBD who underwent colectomy. A) Wound healing pathway‐focused qRT‐PCR analysis of the number of genes up‐regulated in treated human UC or Crohn's disease (CD) patient colectomy tissues by CTB‐KDEL compared with the PBS vehicle control. The graph shows the number of genes ≥2‐fold up‐regulated. Values were calculated by the Qiagen Gene Globe Design and Analysis Hub (http://www.qiagen.com/geneglobe); n = 3/patient. B) qRT‐PCR analysis of TGF‐B1, CDH1, and WNT5A gene expression 24 h after treatment in the tissue from the 20‐y‐old (20yo) male patient with UC. Means ± sd are shown for each treatment (n = 3). A 2‐tailed, unpaired Student's t test was used to compare groups. ∗∗P < 0.01, ∗∗∗P < 0.001. C) Representative ×4 (left) and ×10 (right) H&E‐stained and Ki67 IHC photomicrographs of the 20‐yo male patient UC colectomy tissue.

DISCUSSION

Here, we aimed to delineate the mechanism by which CTB‐KDEL exhibits unique colon epithelial wound repair activity that has not been previously observed with CTB. Our data reported herein demonstrated that it is ascribed to the C‐terminal ER retention motif, which was added initially to improve the yield of the cholera vaccine antigen by means of accumulation in the ER of host plant cells upon transient overexpression (2). However, the KDELR is highly conserved among eukaryotic organisms, including plant and mammalian cells, signifying that CTB‐KDEL could interact with the receptor within human intestinal epithelial cells. Once bound to KDEL‐containing proteins, the receptor transports proteins from the Golgi apparatus to the ER (9, 14, 15). It has been previously shown that KDELR‐mediated retention of proteins in the ER can lead to a UPR in human cells (44, 45). Meanwhile, a UPR has been linked to TGF‐β activation in hepatocytes, skin fibroblasts, and lung epithelial cells (46‐48). Consequently, it is deemed reasonable that CTB‐KDEL induces TGF‐β signaling via a UPR in intestinal epithelial cells, leading to the new function of mucosal wound healing in colitis models. Our data provide multiple lines of evidence to support the above notion. First, a variant of CTB‐KDEL that lacks the terminal Leu residue (CTB‐KDE) failed to induce cell migration in Caco2 cells (Fig. 1). Second, CTB‐KDEL colocalized with KDELR and remained longer than CTB within the ER of Caco2 cells (Figs. 2 and 3). Previous studies investigating the intracellular trafficking of CT and CTB have shown that CTB's retrograde transportation is triggered upon binding to cell surface GM1‐ganglioside, which allows the proteins to reach deep endomembrane compartments and the ER (15, 49). As shown in Figures 2 and 3, both CTB and CTB‐KDEL were found in the ER; however, the latter was more consistently retained and localized within the ER, which was likely because of its capacity to interact with the KDELR. Third, CTB‐KDEL, but not CTB, increased the levels of UPR sensors by 1.5‐2‐fold in Caco2 and mouse primary colon epithelial cells, as well as in a human UC colon tissue (Figs. 4, 6, and 9). Finally, genetic and chemical inhibition of IRE1‐XBP1 signaling completely abolished the TGF‐β up‐regulation and cell migration effects of CTB‐KDEL in Caco2 cells (Fig. 5). Combined with our previous data showing that GM1 binding is essential for CTB‐KDEL's wound healing effect in Caco2 cells (12), these data strongly suggest that the C‐terminal KDEL sequence rendered CTB‐KDEL with a unique new wound healing activity in the colon epithelial cells.

Accumulation of aberrant proteins in the ER promotes the dissociation of BiP from the 3 main ER transmembrane sensors, IRE1, PERK, and ATF6, which triggers their dimerization and phosphorylation for the activation of sensor signal transduction leading to the UPR (29, 50, 51). Our data demonstrated that CTB‐KDEL indeed interacted with BiP, which in turn induced the dissociation of BiP from IRE1 complex (Fig. 4). Therefore, it is likely that the binding of CTB‐KDEL to BiP plays a role in the initiation of UPR seen with CTB‐KDEL treatment. IRE1 represents the most evolutionarily conserved branch of the UPR that is made up of 2 isoforms, IRE1α and IRE1β, of which IRE1β is unique to the epithelium of the digestive and respiratory tract (52). Although there has been no direct evidence for TGF‐β activation via IRE1‐XBP1 signaling in colon epithelial cells, IRE1‐XBP1 signaling has been linked to TGF‐β activation in hepatocytes and fibroblasts, skin wound healing, colon epithelial cell prosurvival signaling, and protection from DSS‐induced colitis, as previously mentioned (31, 33, 46, 47, 53). In this regard, CTB‐KDEL significantly increased IRE1α expression 6 h after treatment and all 3 arms of UPR sensors (ATF6, PERK, and IREI), as well as the IRE1 signal transducer XBP1, 24 h after treatment (Fig. 4). Furthermore, inhibition of IRE1 signaling by the chemical inhibitor 4m8C completely blocked the wound healing activity of CTB‐KDEL as well as CTB‐KDEL‐mediated TGF‐β1 and TGF‐β2 induction (Fig. 5), indicating that CTB‐KDEL's wound healing activity is mediated through IRE1 signal transduction. Much work, however, remains to be done to understand the mechanistic details of how the IRE1 signal leads to TGF‐β activation. Of note, UPR activation is closely linked to ER stress and apoptotic pathways (52). However, we did not observe any cytotoxic response to CTB‐KDEL up to 10 mM (0.61 mg/ml) in Caco2 cells (Fig. 1), suggesting that the protein did not overstimulate UPR beyond the threshold for a cell death response. In our preliminary study using polarized T84 cell monolayers, we found that the levels of UPR marker genes induced by CTB‐KDEL were significantly lower than the well‐known UPR inducer thapsigargin (data not shown). It is possible that the amount of cell surface GM1 ganglioside or the KDELR, or both, were limiting factors, whereby the receptor(s) became saturated before the concentration of CTB‐KDEL reached a point that overload the ER. In turn, this could also explain the apparent lack of a dose‐dependent effect in cell migration activity or toxicity over 2 logs of CTB‐KDEL concentration (0.1‐10 μM) in Caco2 cells (Fig. 1 C). Further investigation is necessary to understand CTB‐KDEL's impacts on UPR, ER stress and survival in epithelial cells.

In the mouse DSS acute colitis model, a single oral administration of as little as 3 μg of CTB‐KDEL significantly enhanced recovery from colitis (Figs. 7 and 8). The therapeutic effects were characterized by enhanced epithelial regeneration, demonstrated by E‐cadherin IHC and up‐regulated Cdh1 and Wnt5a levels, suggesting that the rapid resolution of epithelial injury is the primary mechanism of CTB‐KDEL's efficacy in this model. It should be noted that CTB failed to show such effects and was essentially futile in this model. This is insharp contrast to previous findings by Boirivant et al. (19, 54), which showed that oral administration of CTB can resolve 2,4,6‐trinitrobenzenesulfonic acid‐induced colitis via T helper 1 cell inhibition. Although a direct comparison cannot be made because of 2 different colitis models, the discrepancy in CTB's efficacies could be in part explained by the different dose amounts and timings employed in those 2 studies; whereas Boirivant et al. (19, 54) administered 4 daily doses of 100 μg CTB immediately after the administration of 2,4,6‐trinitrobenzenesulfonic acid, we dosed one 3 μg dose of CTB‐KDEL or CTB at the end of DSS exposure. This points to the possibility that the low dose of CTB was not sufficient to show a therapeutic effect in the DSS model. Conversely, whether CTB‐KDEL at high dose levels exhibits Tcell‐inhibitory effects like CTB remains to be determined.

In DSS‐induced colitis model, the epithelium receives a barrage of mucosal insults, both physical and chemical, that result in the loss of the epithelial barrier and damage to the mucosa (55). Thus, the initial step in injury repair occurs through a rapid migration response of the epithelial sheet (termed restitution) (56, 57). The process occurs independently of proliferation and results in depolarization of intestinal epithelial cells surrounding the wounded area (57). This depolarization leads to an epithelial‐to‐mesenchymal transition (EMT) where cells adopt a migratory phenotype induced by increased TGF‐β levels (58, 59). Previously, we showed that CTB‐KDEL induces multiple TGF‐β‐dependent EMT pathways after treating mice with CTB‐KDEL (12). However, the TGF‐β expression level in the CTB‐KDEL‐dosed animals was not significantly high, although up‐regulated 3‐fold, at the time of euthanization, suggesting that the cytokine had likely passed its peak expression point. In fact, we observed a significant up‐regulation of Serpine1, an inhibitor of TGF‐β, in the CTB‐KDEL‐dosed group (data not shown). Additionally, we detected up‐regulated Cdh1 gene in CTB‐KDEL–treated mice (Fig. 8), which is indicative of a late‐phase wound healing involving epithelial proliferation and maturation (40, 41, 43, 60, 61). E‐cadherin expression is inversely correlated with TGF‐β levels because of TGF‐β–induced class switching of E‐cadherin to N‐cadherin, which depolarizes cells and allows them to become a migratory phenotype (59, 61, 62). Thus, the elevated Cdh1 levels and strong E‐cadherin positive epithelial cell staining that lined the mucosa of CTB‐KDEL–treated mice indicate the re‐polarization of the epithelial cells, increased tightening of mucosal barrier, and improved mucosal barrier integrity, which are critical steps during epithelial repair (61). In addition to EMT, TGF‐β1 stimulates and increases expression of WNT5a (60). WNT5a has been shown to induce new crypt formation and reestablish epithelial homeostasis after injury (60). The H&E‐stained tissue sections clearly revealed the regeneration and formation of new crypts in the CTB‐KDEL‐dosed group, corroborating the function of up‐regulated Wnt5a expression in the colon mucosa. Taken together, these data strongly support the notion that oral administration of CTB‐KDEL can facilitate colon epithelial restitution and wound healing, at least in the conditions tested in the acute DSS colitis model.

In addition to the acute DSS model, we demonstrated the mucosal healing potential of CTB‐KDEL in an explant culture model using human patient IBD colectomy tissues (Figs. 9 and 10). Although the number of patient colon tissues tested in this study is limited, all 5 patient specimens showed up‐regulation of wound healing–related genes in response to CTB‐KDEL treatment, in contrast to PBS or CTB. Of note is that, consistent with the findings in the DSS model, CTB‐KDEL significantly activated UPR, TGF‐β signaling pathways, CDH1, and WNT5a (Fig. 9). Additionally, new crypt formations were observed in CTB‐KDEL‐treated tissues (Figs. 9 and 10). We acknowledge, however, that the wound healing response induced by CTB‐KDEL substantially varied among the different patients' tissues tested, for which the underlying mechanism remains unknown. The least wound healing response was observed with the 72‐y‐old male patient with UC, which may be explained by an age effect; recently, it has been reported that aging individuals experience delayed and impaired healing of the gastric mucosa because of reduced angiogenic capacity (63). Nonetheless, these results overall support the premise that CTB‐KDEL could induce mucosal healing in patients with IBD, warranting a further study using additional tissue specimens. As there is no cure available for IBD, mucosal healing is currently regarded as the standard treatment goal in IBD therapy (64‐67). Current therapeutic options can indirectly achieve mucosal healing in only ~50% of patients, and their direct effect on epithelial repair remains elusive (68‐70). Thus, the results presented herein provide implications for the unique therapeutic potential of CTB‐KDEL that may address a significant unmet need in IBD treatment.

In summary, the data herein reveal that CTB‐KDEL exhibits unique colon mucosal wound healing effects that are mediated by its localization in the ER via KDELR and subsequent activation of a UPR and TGF‐β signaling in colon epithelial cells. Further investigation of the role of the UPR in TGF‐β activation in colon epithelial cells may shed light on the mechanistic link to epithelial restitution and repair, facilitating the development of a new agent for the treatment of inflammatory diseases of the mucosa.

AUTHOR CONTRIBUTIONS

J. M. Royal, Y. J. Oh, and N. Matoba wrote the manuscript; J. M. Royal, Y. J. Oh, and M.J. Grey performed experiments; J. M. Royal, Y. J. Oh, M. J. Grey, W. I. Lencer, and N. Matoba analyzed data; J. M. Royal and Y. J. Oh contributed equally to the work; and N. Matoba conceived and designed experiments, secured funding, and supervised the study.

Royal, J. M. , Oh, Y. J. , Grey, M. J. , Lencer, W. I. , Ronquillo, N. , Galandiuk, S. , Matoba, N. A modified cholera toxin B subunit containing an ER retention motif enhances colon epithelial repair via an unfolded protein response. FASEB J. 33, 13527–13545 (2019). www.fasebj.org