Xenotoxic damage in inflammatory diseases, including IBD (inflammatory bowel disease), is compounded by reduced activity of the xenobiotic transporter ABCG2 (ATP-binding-cassette G2) during the inflammatory state. An association between the activation of the unfolded protein response pathway and inflammation prompted us to investigate the possibility that reduced ABCG2 activity is causally linked to this response. To this end, we correlated expression of ABCG2 and the unfolded protein response marker GRP78 (glucose-regulated protein of 78 kDa) in colon biopsies from healthy individuals (n=9) and patients with inactive (n=67) or active (n=55) IBD, ischaemic colitis (n=10) or infectious colitis (n=14). In addition, tissue specimens throughout the small bowel from healthy individuals (n=27) and from patients with inactive (n=9) or active (n=25) Crohn's disease were co-stained for ABCG2 and GRP78. In all biopsies from patients with active inflammation, irrespective of the underlying disease, an absolute negative correlation was observed between epithelial ABCG2 expression and GRP78 expression, suggesting that inflammation-dependent activation of the unfolded protein response is responsible for suppression of ABCG2 function. The link between the unfolded protein response and functional ABCG2 expression was further corroborated by live imaging of ABCG2-expressing cells, which showed that various inflammatory mediators, including nitric oxide, activate the unfolded protein response and concomitantly reduce plasma membrane localization as well as transport function of ABCG2. Thus a novel mechanism for explaining xenobiotic stress during inflammation emerges in which intestinal inflammation activates the unfolded protein response, in turn abrogating defences against xenobiotic challenge by impairing ABCG2 expression and function.
- breast cancer resistance protein (BCRP)
- Crohn's disease
- endoplasmic reticulum stress
- inflammatory bowel disease (IBD)
- ulcerative colitis
- unfolded protein response
The intestinal epithelium is a single cell physical barrier between the gut luminal content and mucosa, important for detection  and defence [2,3] against microbes. IECs (intestinal epithelial cells) are constantly exposed to a large variety of xenobiotic substances, including microbes and their products, as well as exogenous (e.g. drugs or food) and endogenous (e.g. bile acids) toxic substances present in the luminal content. IECs utilize various mechanisms to maintain epithelial integrity in the face of a xenobiotic challenge. They are protected against exogenous and endogenous toxic compounds by, among others, proteins from the ABC (ATP-binding-cassette) transporter family. This family includes P-glycoprotein/MDR1 (multidrug resistance 1)/ABCB1, MRP1 (multidrug-resistance protein-1)/ABCC1, MRP2/ABCC2 and BCRP (breast cancer resistance protein)/ABCG2 [4,5]. In particular, the homodimeric xenobiotic transporter ABCG2, which specifically localizes to the IEC apical membrane, is widely considered to be especially important in combating luminal-derived xenobiotic stress [6,7].
Disruption of IEC integrity leads to inflammatory lesions, which are a hallmark of active inflammation in patients with IBD (inflammatory bowel disease). The appropriate response to such a disruption requires rapid mucosal healing by cell division, differentiation and maturation to restore barrier function [8,9]. Active inflammation in patients with IBD has been associated with a dramatically reduced expression of ABC transporters in IECs and the resulting diminished capacity to deal with luminal xenobiotic challenge is thought to interfere with the efficiency of regenerative responses [7,10–12]. Further compounding this problem is that several medications used in IBD are specific substrates of ABCG2. Hence understanding the mechanisms governing ABCG2 expression is highly relevant for devising improved therapy [13,14]. Nevertheless, the molecular mechanisms driving ABCG2 expression during inflammatory responses remain obscure at best.
Generally, excreted and membrane-bound proteins are translated in the ER (endoplasmic reticulum). If ER function is disrupted, proteins can accumulate inside the ER, and this can initiate a specific ER stress syndrome known as the unfolded protein response. Abnormalities in the ER stress response pathways have been associated with the pathogenesis of IBD [15–17]. Proper function of the ABCG2 protein depends on N-glycosylation and the formation of inter-/intra-molecular disulfide bonds [18,19]. These bonds are formed inside the ER by PDIs (protein disulfide-isomerases). Inflammatory conditions, such as increased nitric oxide production, can dramatically reduce the efficiency of these PDIs, which can initiate misfolding of various proteins, leading to ER stress [20,21]. As correct ABCG2 protein folding is necessary for its functionality, we hypothesized that mucosal inflammation in the setting of IBD alters proper ABCG2 protein folding and thereby reduces its apical expression. In the present study we describe a novel mechanism by which ABCG2 protein expression is decreased during active inflammation.
The present study was conducted with the approval of the Ethics Committee of Erasmus MC University Medical Centre, Rotterdam. Consent was obtained from all patients participating in the study.
FFPE (formalin-fixed paraffin-embedded) colonic biopsies were collected from nine healthy individuals and from 76 patients: 25 with CD (Crohn's disease), 36 with UC (ulcerative colitis), ten with ischaemic colitis and five with infectious colitis. Biopsies throughout the small intestine were collected from 17 other healthy individuals and 34 CD patients. For mRNA isolation, freshly frozen biopsies were collected from six healthy individuals and from 15 patients: eight with inactive IBD and seven with active IBD.
The demographic characteristics, number of biopsies per group, duration of disease and medication use are described in Supplementary Tables S1–S3 (at http://www.BiochemJ.org/bj/441/bj4410087add.htm).
Immunohistochemistry and scoring
FFPE sections were immunohistologically stained using the primary antibodies BXP-21 anti-ABCG2 (Santa Cruz Biotechnology) and BiP (immunoglobulin heavy-chain-binding protein) anti-GRP78 (glucose-regulated protein of 78 kDa; Cell Signaling Technology) according to the manufacturers' protocol (described in the Supplementary Experimental section at http://www.BiochemJ.org/bj/441/bj4410087add.htm). The expression of ABCG2 and GRP78 in IECs was scored by the intensity over the whole slide by two independent observers, and discrepancies were re-assessed to come to a final agreement. Individual scores were used for statistical analyses, whereas they were combined into three subgroups for graphic representation. ABCG2-positive staining at the apical membrane of the epithelial cells was defined as positive if >50% of the IECs were positive, mild if 15–50% or negative if <15%. Immunoreactivity for GRP78 was also defined in terms of positive IECs, but now the percentages were: >70% defined as positive, 20–70% as mild and <20% as negative.
mRNA isolation and qPCR (quantitative PCR)
mRNA was isolated from colonic biopsies, which were taken from CD and UC patients, using the NucleoSpin Extract II mRNA isolation kit (Macherey Nagel). Gene expression of GAPDH (glyceraldehyde-3-phosphate dehydrogenase), IL8 (interleukin 8), IL1β, ABCG2, villin and the ER-stress-associated gene CHOP [C/EBP (CCAAT/enhancer binding protein) homologous protein]  were measured using qPCR (IQ5, Bio-Rad). Villin and ABCG2 are both IEC-specific, thus ABCG2 mRNA was corrected for villin to prevent underestimation in biopsies with fewer IECs. A detailed protocol and the primer sequences involved are presented in the Supplementary Experimental section and Supplementary Table S4 (at http://www.BiochemJ.org/bj/441/bj4410087add.htm).
HEK-293T [human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40); A.T.C.C.; CRL-1573] cells were cultured according to routine procedures. Transient transfection was performed using polyethylenimine (Polysciences). SNAP (S-nitroso-N-acetyl-DL-penicillamine; Sigma–Aldrich) was used as a nitric oxide donor. Tunicamycin from Streptomyces sp. (Sigma–Aldrich) was used as a positive control for ER stress induction. The cell culture supernatant nitrate concentrations were measured using Griess reagents (Cayman Chemical), according to the manufacturer's protocol. Additional cell culture methods are described in the Supplementary Experimental section.
ER stress and unfolded protein response activity was assessed by identifying mature XBP1 mRNA and its spliced variant as described previously  (described in the Supplementary Experimental section).
To investigate ABCG2 functionality, we created a GFP (green fluorescent protein)–ABCG2 fusion construct (described in the Supplementary Methods).
The GFP-tagged proteins were examined using a confocal microscope (LSM510META, Zeiss). For the transport assay, nuclear Hoechst 33342 (4 μM, Sigma-Aldrich) accumulation was determined at 30 s intervals over a total of 12 min. The ER was stained with ER–RFP (red fluorescent protein) using ER-Tracker (Molecular Probes), according to the manufacturer's instructions. We analysed the GFP-tagged proteins with a 488 nm laser, Hoechst with a 405 nm laser and the ER–RFP with 543 nm laser. A specific ABCG2 transport inhibitor, Ko143 (2 μM, Sigma–Aldrich), was used to indentify ABCG2 transport specificity.
SDS/PAGE and fluorescent immunoblotting
To check the length of fusion proteins was correct, whole cell lysates were analysed on immunoblots from a standard SDS/PAGE gel, using antibodies against ABCG2 (BXP-21; Santa Cruz Biotechnology) and GFP (polyclonal, rabbit IgG; Invitrogen) (described in the Supplementary Experimental section).
Statistics and software
Statistical analyses were performed using the SPSS 11.0 software package for Windows. Data on different patient groups were compared using the Mann–Whitney U test with a Kruskal–Wallis post-hoc test. Using univariate and multivariate multinomial logistic regression analysis, we examined the association between colitis and GRP78 and ABCG2 expression, adjusting for age and duration of disease. A two-tailed P value of <0.05 was accepted as statistically significant. Images were composed using Adobe Photoshop CS3.
Colonic inflammation reduces apical ABCG2 expression in IECs
To investigate ABCG2 expression during human intestinal inflammation, ABCG2 expression in colonic biopsies was qualitatively assessed employing immunohistochemistry. As expected, IECs in biopsies of control subjects as well as of IBD patients in remission (with no active inflammation) showed strong specific staining for ABCG2 at the apical surface (Figure 1A). In contrast, apical IEC ABCG2 staining was dramatically reduced in patients with active IBD, in the colon as well as in the small intestine, and also in biopsies from ischaemic colitis and infectious colitis patients (Figure 1A). Thus diverse colonic inflammatory responses are uniformly characterized by the down-regulation of apical ABCG2 levels, suggesting that processes common to these alternative modes of inflammation underlie the absence of apical ABCG2 during intestinal inflammation.
In apparent agreement, a quantitative study of ABCG2 levels in the human colon revealed significantly less ABCG2 expres-sion in active colitis (24% positive expression, 13 out of 54), ischaemic colitis (10% positive staining, one out of ten) and infectious colitis (0% positive staining, zero out of 14) as compared with either inactive colitis (80% positive expression, 53 out of 67) or control tissues (100% positive expression, nine out of nine) (all P<0.001; see Figure 1B). There was no significant difference in ABCG2 levels between active CD (27%, six out of 22) compared with active UC (22%, seven out of 32) and inactive CD (67%, 18 out of 27) compared with inactive UC (88%, 35 out of 40). As a control, we examined left- and right-sided biopsies from IBD patients and controls to investigate a confounding possible differential expression of ABCG2 based on position in the colon, but this analysis demonstrated that only the degree of inflammation is an important factor guiding apical ABCG2 levels. Thus inflammation itself is associated with a marked reduction in apical ABCG2 expression in the human colon.
Small bowel inflammation is associated with down-regulation of apical ABCG2 expression
Mirroring the findings obtained in the human colon, small bowel biopsy specimens obtained from CD patients with active disease by double balloon endoscopy exhibit markedly less ABCG2 staining when compared with biopsies obtained from patients with inactive disease or no intestinal inflammatory phenomena (see Figure 1A). In active CD, there was significantly less small bowel ABCG2 staining (75% negative staining, 19 out of 25) as compared with inactive CD (11% negative staining, one out of nine) and controls (8% negative staining, two out of 25) (both P<0.05; see Figure 1C). We found no significant differences between inactive CD and controls. In all cases, current drug use did not influence the expression of ABCG2. We conclude that human intestinal inflammation is accompanied by reduced expression of the xenobiotic transporter ABCG2 and experiments were initiated to identify the underlying molecular mechanisms.
Intestinal inflammation acts to reduce ABCG2 via a post-transcriptional mechanism
To delineate the molecular events reducing the capacity of the intestine to cope with xenobiotic challenge, we first determined whether genomic or post-transcriptional processes are involved in this effect. To this end, we measured ABCG2 mRNA levels in freshly frozen colon biopsies from healthy individuals, and inactive and active IBD patients. Importantly, ABCG2 mRNA levels in IECs did not differ between these three groups (all P> 0.8). The expression, however, of the pro-inflammatory cytokines IL8 and IL1β were 10- and 8-fold higher (P<0.05) in biopsies with active mucosal inflammation (Figure 2), and thus technical issues do not underlie the absence of intestinal inflammation-induced effects on apical ABCG2 expression. Hence we are forced to conclude that down-modulation of ABCG2 in human intestinal inflammation is brought about by a post-transcriptional process.
Activation of the unfolded protein response accompanies human intestinal inflammation
As it is becoming increasingly clear that activation of the unfolded protein response accompanies inflammatory processes and that ER stress may interfere with proper production of complexly folded proteins (the ‘half’ transporter ABCG2 transverses the membrane six times), we examined whether the ER stress pathway is activated in intestinal inflammation and to which extent this activation mirrors the down-regulation of ABCG2. To this end, we employed the expression of the bona fide ER stress marker GRP78. Representative images of the GRP78 immunohistochemical analyses of the same biopsies as stained for ABCG2 are shown in Figure 3(A). There was a significant increase in GRP78 expression in active colitis (32% positive expression, 16 out of 54), ischaemic colitis (90% positive staining, nine out of ten) and infectious colitis (79% positive staining, 11 out of 14), compared with inactive colitis (0% positive expression, zero out of 67) and control tissues (0% positive expression, zero out of nine) (all P<0.001; see Figure 3B). There are no significant differences in GRP78 expression between active CD and active UC or between inactive CD and inactive UC. In addition there were no differences in GRP78 expression in left- and right-sided colon biopsies taken from the same patient. We conclude that colonic inflammation is accompanied by the ER stress response, which is absent in the non-inflamed colon.
In the small bowel biopsies of active CD patients, there was increased GRP78 expression (29% positive expression, six out of 21) compared with inactive CD patients (0% positive staining, zero out of nine) and controls (8% positive expression, two out of 25) (P<0.05; see Figures 3A and 3C). In all actively inflamed tissue samples, elevated GRP78 expression was evident, irrespective of disease type or location. ER stress during intestinal inflammation was further confirmed by measuring the mRNA expression of the ER-stress-associated gene CHOP in colonic biopsies. In biopsies taken from actively inflamed mucosa, a significant increase in CHOP mRNA expression was measured (P<0.01; see Figure 2B). Moreover, an increased ratio of spliced XBP1 mRNA was detected in patients with active intestinal inflammation (Figure 2C). We conclude that activation of the unfolded protein response and intestinal inflammation are highly associated processes and we initiated experiments to investigate whether ER stress explains the down-regulation of ABCG2 in intestinal inflammation.
Impaired ABCG2 function in ER-stressed cells
To investigate the effect of activated ER stress on ABCG2 protein expression and function, we created a GFP–ABCG2 fusion protein. To this end, ABCG2 cDNA was cloned into a pEGFP-C1 vector, which led to the production of GFP–ABCG2 after transfection (Figure 4A). The nitric oxide donor SNAP was used to ascertain whether nitric oxide was able to induce ER stress in 293t-AB (GFP–ABCG2-transfected HEK-293T) cells. The Griess assay revealed that SNAP application induced robust increases in nitrate concentration, a good proxy measure for nitric oxide production (see Figure 4B). The validity of the assay was demonstrated by experiments showing an increase in the ratio of unspliced XBP1 mRNA over spliced XBP1 mRNA 9 h after the application of 0.625 mM SNAP (see Figure 4C). In agreement with this, administration of tunicamycin, a well-established ER-stress-response inducer, increased the XBP1 splicing ratio to approximately the same level as SNAP application (see Supplementary Figure S1B at http://www.BiochemJ.org/bj/441/bj4410087add.htm). As is consistent with high ABCG2 transport activity, administration of Hoechst 33342 did not stain the nuclei of the 293t-AB cells (Figure 5A), also demonstrating the functionality of the ABCG2–GFP fusion protein in xenobiotic transport. Nevertheless, 16 h after addition of the ER stress inducers SNAP or tunicamycin, the nuclei of 293t-AB cells became Hoechst 33342-positive (Figure 5A and Supplementary Movies S1, S2 and S3 at http://www.BiochemJ.org/bj/441/bj4410087add.htm). As shown in Figure 5(B), Hoechst 33342 also stained the nucleus blue in 293t-AB cells in the presence of the specific ABCG2 inhibitor Ko143 (see Supplementary Movie S4 at http://www.BiochemJ.org/bj/441/bj4410087add.htm). No altered GFP–ABCG2 protein function was found when SNAP or tunicamycin was added 48 h after transfection (see Supplementary Figure S1C). Hence, in this overexpressing model system, ER stress provokes a very strong decrease in ABCG2 function, in agreement with a scheme in which inflammation-dependent induction of ER stress is causatively related to a reduced capacity to deal with xenobiotic challenge.
In support for such a model, all images from 293t-AB cells with ER stress revealed a reduced apical membrane expression of GFP–ABCG2, with intracellular expression remaining. In order to ascertain whether GFP–ABCG2 remained in the ER after stress induction, we stained the ER with ER–RFP. When ER stress was present, some GFP–ABCG2 co-localized with ER–RFP, whereas the majority was present in other cellular compartments (Figure 5C). As shown in Supplementary Figure S2 (at http://www.BiochemJ.org/bj/441/bj4410087add.htm) the localization and expression of non-ER-dependent GFP-tagged proteins [GFP–β-catenin and GFP–H2B (histone 2B)] was not changed during ER stress conditions. Thus protein folding difficulties seems incompatible with functional expression of ABCG2.
The results shown in the present study are consistent with a model in which intestinal inflammation drives an epithelial misfolded protein response, which in turn interferes with the ABCG2 expression and function, leading to a diminished capacity of the epithelial compartment to cope with xenobiotic stress. Colonic and small bowel biopsies from IBD patients with active disease and those from patients with ischaemic and infectious colitis all show reduced apical ABCG2 expression in IECs compared with controls and IBD patients in remission. The decline of ABCG2 expression in the presence of active inflammation in IBD patients corresponds with the findings of two studies done exclusively on UC patients [7,22]. In addition, we found that ABCG2 expression is not affected by medication use, patient age or location of the inflammation.
Expression of ABCG2, similar to that of many other proteins, can be regulated at various levels. In addition to a decline in ABCG2 protein expression, others have described a reduction in ABCG2 mRNA expression in actively inflamed intestine [7,22]. We corrected the ABCG2 mRNA expression to villin to prevent an underestimation of ABCG2 expression, since both genes are specific for IECs. Our results show no reduction of ABCG2 mRNA expression in biopsies with active mucosal inflammation. However, since the sample size used in the present study is smaller than those in previous studies, we cannot completely rule out the contribution of reduced mRNA expression to the decline in ABCG2 protein expression. As such, our findings merely imply that the decline in mRNA expression probably does not solely account for the complete loss of ABCG2 protein expression discernable on our immunohistochemistry images and that post-transcriptional mechanisms must be involved. The observation that unfolded protein responses are both general and specific to intestinal inflammation suggest a causal link between ER stress and diminished apical ABCG2 levels. Indeed, ABCG2 is a ‘half’ transporter, spanning the membrane six times, functionally critically dependent on homodimerization, N-linked glycosylation and various disulfide bonds and thus is probably an early victim of the ER stress responses [18,19]. Disrupted protein-folding mechanisms that can lead to ER stress during active inflammation have been linked to IBD [16,23,24], and may well involve the reduced capacity of the IEC compartment to deal with the xenobiotic stress and thus in turn interfere with demanding regenerative responses that an adequate response to IBD-related intestinal challenges requires.
Our link from inflammation to unfolded protein responses remains unclear, but inflammatory responses are associated with an excessive production of nitric oxide, which in turn is predicted to disturb the function of redox-state sensitive disulfide-bonding chaperones such as PDIs, that are necessary for proper protein folding . In all biopsies with active inflammation (i.e. IBD and non-IBD related), the GRP78 expression is significantly higher than in the biopsies of IBD patients in remission and in those of healthy controls. Furthermore, expression of CHOP mRNA (an ER stress marker) is increased in biopsies with active mucosal inflammation. In addition to this, biopsies with active inflammation reveal an increased amount of spliced XBP1 mRNA expression. Thus factors common to multiple types of inflammation must be involved here and, as nitric oxide production is one of the few cellular responses described to generally accompany inflammation, it is tempting to attribute an important role to excessive nitric oxide generation in inflammation-related induction of ER stress responses [23,25–27]. Indeed, we observed elevated XBP1 splicing  following application of the nitric oxide donor, suggesting that nitric oxide can directly affect protein-folding mechanisms  and induce ER stress. In the present study we show that the reduction in ABCG2 expression is associated with inflammation, which fits with the observations that ABCG2 expression is reduced in IBD-unrelated inflammatory conditions in other organs [10,12,28]. Other IBD-associated proteins are also reduced in protein expression during active intestinal inflammation [29–31]. Further research is needed to investigate the role of impeded protein folding and the effect of this on cellular processes. Noteably, ABCG2-mediated efflux of toxic components is thought to protect the epithelium of the gut and the underlying tissue to xenobiotic challenges. Importantly, as xenobiotic stress is directly linked to genotoxicity, our results may also help to explain the still poorly understood relationship between intestinal inflammation and colorectal cancer . Decreased ABCG2 expression may have other deleterious consequences as well. Sulfasalazine, a derivate of mesalazine [5-ASA (5-aminosalicylic acid)] that is commonly used in the treatment of IBD patients, is actively excluded from the epithelial compartment by ABCG2 . Rodents with reduced IEC ABCG2 expression have increased 5-ASA trough levels . In addition to its anti-inflammatory effect, accumulation of sulfasalazine can be toxic for epithelial cells [35,36]. Thus reduced expression and function of ABCG2 may therefore greatly enhance the sensitivity of IECs to sulfasalazine and limit the use of this compound for the treatment of IBD. Hence understanding the mechanisms governing ABCG2 expression are highly relevant for efforts with respect to rational design of novel therapy.
In conclusion, IEC ABCG2 protein expression is decreased in the inflamed bowel, probably as a result of incorrect protein folding due to inflammation-driven ER dysfunction. Thus a novel paradigm explaining xenobiotic stress during inflammation emerges in which intestinal inflammation activates the unfolded protein response, in turn abrogating the defences against xenobiotic challenge by impairing ABCG2 expression and function.
Jasper Deuring designed and performed the research, and wrote the paper. Colin de Haar analysed the data, performed extensive drafting of the paper and supervised the study. Chantal Koelewijn helped to design the research. Ernst Kuipers and Maikel Peppelenbosch critically reviewed the paper prior to submission. Janneke van der Woude critically reviewed the paper prior to submission and supervised the study.
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
We thank Dr R. Smits (Department of Gastroenterology and Hepatology, Erasmus Medical Centre, Rotterdam, The Netherlands) for providing the GFB–β-catenin and GFP–H2B fusion constructs. We also thank Dr H. van Dekken and Dr K. Biermann for pathological assessment of the intestinal cross-sections.
Abbreviations: ABC, ATP-binding-cassette; 5-ASA, 5-aminosalicylic acid; CD, Crohn's disease; CHOP, CCAAT/enhancer-binding protein-homologous protein; ER, endoplasmic reticulum; FFPE, formalin-fixed paraffin-embedded; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFP, green fluorescent protein; GRP78, glucose-regulated protein of 78 kDa; HEK-293T, human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40); IBD, inflammatory bowel disease; IEC, intestinal epithelium cell; IL, interleukin; MRP, multidrug-resistance protein; PDI, protein disulfide-isomerase; qPCR, quantitative PCR; RFP, red fluorescent protein; SNAP, S-nitroso-N-acetyl-DL-penicillamine; 293t-AB, HEK-293T cells transiently transfected with GFP–ABCG2; UC, ulcerative colitis; XBP1, X-box-binding protein 1
- © The Authors Journal compilation © 2012 Biochemical Society