Effects of lifestyle and environmental factors on intestinal epithelial cell morphogenesis

Cigarette smoking is emerging as an important life style factor that affects intestinal function. Epidemiological studies indicate that cigarette smoking influences the course of inflammatory bowel disorders and is a risk factor for colorectal cancer and hyperplastic intestinal polyp development. The mechanism by which cigarette smoke exerts its impact on the intestinal epithelium is unclear. These may involve indirect mechanisms via alterations in the immune system and/or direct actions of cigarette smoke on the intestinal epithelium. Little is known about the direct effects of cigarette smoke on intestinal epithelial cell behavior. In this study, we employed a 3D human intestinal epithelial Caco-2 cell spheroid model to investigate the effects of cigarette smoke extract (CSE) on intestinal epithelial barrier function, cell polarity and morphogenesis. We found that CSE did not affect the polarity and barrier function of pre-established lumen-forming Caco-2 spheroids. However, when exposed to CSE during the process of three-dimensional morphogenesis, cells showed dose-dependent impairment of growth and de novo development of apical-basal cell polarity and tight junctions, resulting in defective self-assembly into single lumen-forming spheroids. These effects penetrated into future cell generations. The administration of the glutathione precursor N-acetyl-L-cysteine (NAC) completely neutralized the inhibitory effects of CSE in a manner that involved glutathione-dependent and –independent mechanisms. Our data demonstrate that CSE has direct inhibitory effects on intestinal epithelial cell polarity development and morphogenesis, and suggests a potential clinical application of NAC in the treatment of smoking-related intestinal disorders.

Effects of interferon-gamma and energy metabolism pathways on intestinal epithelial gland morphogenesis in three-dimensional culture 5 136

Abstract
Cigarette smoking is emerging as an important life style factor that affects intestinal function.
Epidemiological studies indicate that cigarette smoking influences the course of inflammatory bowel disorders and is a risk factor for colorectal cancer and hyperplastic intestinal polyp development. The mechanism by which cigarette smoke exerts its impact on the intestinal epithelium is unclear. These may involve indirect mechanisms via alterations in the immune system and/or direct actions of cigarette smoke on the intestinal epithelium. Little is known about the direct effects of cigarette smoke on intestinal epithelial cell behavior. In this study, we employed a 3D human intestinal epithelial Caco-2 cell spheroid model to investigate the effects of cigarette smoke extract (CSE) on intestinal epithelial barrier function, cell polarity and morphogenesis. We found that CSE did not affect the polarity and barrier function of pre-established lumen-forming Caco-2 spheroids. However, when exposed to CSE during the process of three-dimensional morphogenesis, cells showed dose-dependent impairment of growth and de novo development of apical-basal cell polarity and tight junctions, resulting in defective self-assembly into single lumen-forming spheroids. These effects penetrated into future cell generations.
The administration of the glutathione precursor N-acetyl-L-cysteine (NAC) completely neutralized the inhibitory effects of CSE in a manner that involved glutathione-dependent and -independent mechanisms. Our data demonstrate that CSE has direct inhibitory effects on intestinal epithelial cell polarity development and morphogenesis, and suggests a potential clinical application of NAC in the treatment of smoking-related intestinal disorders.

Introduction
The intestinal epithelium and its homeostasis is subject to influences from many life-style and environmental factors. Epidemiological studies indicate that cigarette smoking has a strong influence on the intestinal epithelium. Cigarette smoking influences the course of Crohn's disease and ulcerative colitis 1,2 , two predominant inflammatory bowel disorders that, at the epithelial level, are characterized by impaired intestinal barrier function [3][4][5] . Impaired intestinal barrier function is also associated with the smoking-related lung disease chronic obstructive pulmonary disease (COPD) 6,7 . Furthermore, cigarette smoking is a risk factor for hyperplastic intestinal polyp development, as well as for colorectal cancer [8][9][10] , which is one of the leading causes of death in the Western world. Animal studies support the correlation between cigarette smoking and its negative effects on the intestinal epithelium. Exposure of normal laboratory mice to cigarette smoke led to impaired intestinal barrier disruption 7,11 , increased apoptosis and autophagy in the ileum 12,13 , decreased cystic fibrosis transmembrane conductance regulator (CFTR) activity in the intestinal epithelium 14 , increased bacterial translocation, intestinal villi atrophy 11 and small/large intestine-specific alterations in tight junction protein expression 11,15 . Furthermore, cigarette smoke exposure has been reported to aggravate intestinal inflammation in trinitrobenzenesulfonic acid-induced colitis in rats 16 . The mechanism by which cigarette smoke exerts its impact on the intestinal epithelium is unclear 17 . Many cell types contribute to intestinal epithelial homeostasis, including intestinal epithelial cells that are constantly renewed through intestinal stem cell activity, complex communities of symbiotic microorganisms in the gut lumen (the microbiota), and various immune cell populations in the lamina 138 propria. Not surprisingly therefore, the mechanisms by which cigarette smoking influences intestinal epithelial homeostasis have been proposed to involve alterations of the immune system 18 and/ or the intestinal microbiota 15,[19][20][21][22] . Little is known, however, about the potential direct effects of cigarette smoke on intestinal epithelial cell behavior 23 . In order to investigate potential direct effects of cigarette smoke on intestinal epithelial cell behavior, a reductionist in vitro approach is required in which other cell types that are normally present in the intestine (e.g., bacteria, immune cells) are excluded. Several in vitro model systems to study intestinal epithelial cell biology are available. Of these, twodimensional (2D) cell culture, in which cells are grown on flat polystyrene or glass substrata, is most commonly used. However, these substrata are unnatural and do not allow cells to organize themselves in a three-dimensional environment as they do in vivo. Moreover, 2D cell cultures are poor predictors of cellular toxicity and toxigenic responses [24][25][26][27] . To overcome these and other drawbacks of 2D cell culture, three-dimensional (3D) cell culture systems have been developed. In 3D cultures single epithelial cells are mixed with extracellular matrix/ basement membrane components, which allows the proliferating epithelial cells to establish natural cellmatrix and cell-cell attachments and to self-organize to form a multicellular polarized architecture that best reflects the physiological in vivo situation [28][29][30][31][32] . 3D cell culture systems therefore provide a suitable approach bridging the gap between 2D cell culture and in vivo animal models/humans 33 . In this study, we employed a recently developed 3D in vitro intestinal epithelial cell culture system 34-38 to investigate the effects of cigarette smoke extract on intestinal epithelial cell polarity, barrier function and morphogenesis.
Effects of interferon-gamma and energy metabolism pathways on intestinal epithelial gland morphogenesis in three-dimensional culture 5 139

Materials and Methods
Three-dimensional Caco-2 cell culture. Threedimensional (3D) culture of human Caucasian colon carcinoma-2 (Caco-2) cells (ATCC) was performed as described previously 36 . Briefly, Caco-2 cells were grown in culture medium (Dulbecco's modified Eagle's medium (DMEM) supplemented with glutamine, 10% fetal bovine serum (FBS) and antibiotics). Cells were grown to ~85% confluence in a 25 cm 2 tissue culture flask, detached with trypsin and ethylene diamine-tetraacetic acid (EDTA) dissolved in phosphate-buffered saline solution (PBS), re-suspended in culture medium supplemented with 2% (v/v) Matrigel, and plated at a concentration of 1.5*10 4 cells/ml in wells of a 8-well plate pre-coated with 100% (v/v) Matrigel for up to 7 or 8 days 36 . With this protocol, Caco-2 cells developed spheroids consisting of a monolayer of cells with functional apical-basal cell polarity and tight junctions that surround a central apical lumen [36][37][38] .
In vitro exposure of cells to CSE. Cigarette smoke extract (CSE) was prepared as described 39
Measurement of glutathione levels. Cells were seeded in a 6-well plate cultured for 3 days to form a monolayer. Cells were incubated with CSE in the presence or absence of NAC and/or BSO for 24 h. Glutathione levels were determined spectrophotometrically assessing the reaction with glutathione reductase, using the Hitachi u-29000 spectrophotometer at 30 o C for 10 minutes, measuring every 60 seconds.
Statistical analyses. The percentage of lumenforming cell clusters under different treatment conditions were compared using χ-square tests, sphere sizes were compared using t-tests.

Cigarette smoke extract does not perturb the integrity of preformed Caco-2 cell spheroids
Caco-2 cells in three-dimensional culture, consisting of the basement membrane extract (Matrigel), selfassembled to form spheroids over a period of 7 days that comprised a monolayer of cells that surrounded a single central (apical) lumen. The cells in these monolayers were highly polarized, with a dense actin network at the luminal side (reflecting the presence of actin-based microvilli 36,38 ), and tight junction proteins at the apex of the lateral plasma membrane between neighboring cells 36,38 ( Figure 1A). The functionality of these tight junctions was confirmed by the exclusion of basolaterally-administered FITC-labeled 4 kDa dextran (FD4) from the apical lumen ( Figure 1C). Treatment of the spheroids with ethylene glycol tetraacetic acid (EGTA), a calcium

144
formed multicellular, single lumen-forming spheroids as a function of time (c.f., Figure 1A). Thus, 7 days after plating cell clusters displayed a typical donutshaped morphology, as readily visualized by light microscopy (Figure 2) 34-36 . In contrast, when the culture medium of Caco-2 cells was supplemented with CSE during this 7-day period, cell clusters were smaller in diameter and very few donut-shaped clusters were observed at day 7 ( Figure 2; for quantification see Figure 4A and B). Together these data demonstrate that CSE led to reduced sphere size and inhibited the self-assembly of Caco-2 cells to form single-lumen spheroids.

Cigarette smoke extract impairs the de novo establishment of epithelial cell polarity
Epithelial morphogenesis is critically dependent on the ability of the cells to polarize, that is, to develop apical and basolateral plasma membrane domains bordered by tight junctions 41 , and to orientate their polarity with respect to the extracellular matrixcontaining environment 42,43 . To investigate the effect of CSE on cell polarity and tight junctions, Caco-2 cells were cultured for 7 days in normal 3D culture medium or normal 3D culture medium supplemented with CSE, fixed with PFA and processed for fluorescence (immuno)labeling to visualize the apical actin network, the tight junction protein ZO-1 and cell nuclei. Most cells cultured in normal culture medium formed single lumen-forming spheroids consisting of a monolayer of cells (evidenced by nuclei staining; Figure 3A). Actin was predominantly present at the lumen surface of the cells ( Figure 3A), which reflects the presence of actin-dense apical microvilli and apical surface polarity. ZO-1 appeared predominantly with a punctate pattern at the apex of the lateral plasma membrane facing the lumen ( Figure 3A), which is in agreement with the position of the tight junctions 36,38 . In contrast, cells cultured in medium

146
supplemented with CSE showed impaired cell polarity and lumen formation ( Figure 3B and 4A). Cell nuclei staining revealed that the cells had failed to assemble in monolayers (c.f., Figure 2) and, instead, developed into solid clusters ( Figure 3B) of smaller size ( Figure 4B). Actin labeling was predominantly observed at the culture medium-facing plasma membrane as well as at the lateral side of the cells ( Figure 3B). ZO-1 appeared to be randomly distributed along the cell surface without a clear polarized localization ( Figure 3B). In some cases, the cells showed a punctate ZO-1 staining at the lateral sides of the cells, but facing the surrounding medium ( Figure 3B, arrowhead), indicative of misoriented cell polarity. Quantification of the data demonstrated that CSE exposure caused a decrease (~50 and ~80% with 7.5 and 10% CSE, respectively) in the percentage of single lumen-forming Caco-2 spheroids ( Figure 4A) and in cell cluster size ( Figure 4B).
In previous studies it has been shown that the recruitment of laminin, which is a major basement membrane component present in matrigel, to the cell surface is an essential step in epithelial cell polarity orientation and lumen formation in 3D culture 42,44 . We observed predominant laminin accumulation around the polarized lumen-forming Caco-2 clusters under control conditions ( Figure 3C), as well as around the non-polarized non-lumen-forming Caco-2 cell clusters exposed to CSE ( Figure 3D). This indicates that CSE exposure does not inhibit the recruitment of laminin to the epithelial cell surface.
Together, these data indicate that CSE exposure impairs de novo development of Caco-2 cell polarity and tight junctions downstream of laminin recruitment, resulting in defective epithelial morphogenesis.

Figure 4. Effect of NAC and BSO on cigarette smoke extract-induced impairment of Caco-2 epithelial cell polarity and morphogenesis. A)
Effects of various treatments (see text) on the percentage of single lumen-forming spheroids in comparison to the control condition (7 days 3D culture in normal culture medium, set to 1); B) Effects of various treatments (see text) on cell cluster size in comparison to the control condition (7 days 3D culture in normal culture medium, set to 1); C) Effects of pre-exposition to CSE and/or NAC (see text) on the percentage of single lumen-forming spheroids in comparison to the control condition (non-pre-exposed cells, set to 1); D) Effects of pre-exposition to CSE and/or NAC (see text) on cell cluster size in comparison to the control condition (non-pre-exposed cells, set to 1); E) Effects of various treatment on intracellular glutathione levels. Continuous data (i.e. size and glutathione expression are presented as (normalized) mean ± SD of at least three independent experiments, dichotomous data (i.e. spheroid formation), is represented as normalized fraction, error bars are ranges.

148
The inhibitory effect of cigarette smoke extract on epithelial cell polarity and morphogenesis shows penetrance to next generations In order to determine the (ir)reversibility of the effects of CSE on Caco-2 cell polarization and morphogenesis, we pre-exposed Caco-2 cells to CSE (or control culture medium) for 24 h and subsequently cultivated the cells in 3D in the absence of CSE for 7 days. The cells were then fixed and (immuno)labeled to visualize nuclei, actin and ZO-1, and the percentage of single lumen-forming spheroids was determined. As shown in Figure 4C, CSE-pre-exposed cells showed impaired epithelial polarization and morphogenesis when subsequently grown in 3D culture medium in the absence of CSE, similar to non-pre-exposed cells that were grown in 3D culture medium in the presence of CSE. Notably, CSE pre-exposure did not significantly reduce cell cluster size in next generation cells grown in the absence of CSE ( Figure 4D). These data demonstrate the penetrance of the inhibitory effect of CSE on Caco-2 cell polarization and morphogenesis to next generations of cells, which was independent of cell growth.

N-acetyl-L-Cysteine neutralizes the inhibitory effects of cigarette smoke extract on cell growth, cell polarity and morphogenesis
Cigarette smoke extract contains many compounds that are toxic for cells. These include among others reactive oxygen species and potential carcinogens, such as polycyclic aromatic hydrocarbons. Exposure of Caco-2 cells to CSE did not lead to cell death (as analyzed by lactate dehydrogenase leakage tests; data not shown). Many cells, including Caco-2 cells, produce glutathione, a low-molecular-weight thiol, in order to detoxify toxic compounds 45,46 . Exposure of Caco-2 cells to CSE resulted in increased (~2-fold) levels of intracellular glutathione when compared to cells that were grown in normal culture medium 149 ( Figure 4D). When Caco-2 cells were grown in 3D culture medium supplemented with CSE together with the glutathione synthesis inhibitor L-buthione-S,R-sulfoximine (BSO) 47 , which completely abolished glutathione production in the cells (Figure 4E), no cells were found attached to the culture plate after 24 h (data not shown), which is indicative for impaired cell viability. When Caco-2 cells were grown in 3D culture medium supplemented with only BSO, normal development of epithelial cell polarity and morphogenesis was observed, indistinguishable to cells cultured in normal medium ( Figure 4A, B). These data suggest that CSE stimulated the production of glutathione and glutathione synthesis was required for the cells to survive and form cell clusters in the presence of CSE. We hypothesized that the synthesis of glutathione was sufficient to maintain cell survival when exposed to CSE, but insufficient to prevent the CSE-mediated inhibition of epithelial sphere size, polarity development and morphogenesis. Therefore, we included the glutathione precursor N-acetyl-L-cysteine (NAC) in the CSE-containing culture medium. NAC completely abolished the inhibitory effects of CSE. Thus, Caco-2 cells when cultured in CSE-containing medium supplemented with NAC developed normal-sized single lumenforming spheroids, similar to cells cultured in normal medium only ( Figure 4A). Surprisingly, however, the addition of NAC did not result in increased intracellular glutathione levels ( Figure 4D), whereas the neutralizing effect of NAC was abolished in the presence of the glutathione synthesis inhibitor BSO ( Figure 4A). This suggested that the neutralizing effect of NAC was via a glutathione-independent mechanism, yet requiring endogenous (i.e., NACindependent) glutathione synthesis. Interestingly, whereas combined exposure of cells to CSE and BSO inhibited cell viability (see above), the combined exposure of cells to CSE, BSO and NAC did not, 150 which also indicates a glutathione-independent effect of NAC treatment. Further, in the presence of NAC, glutathione synthesis was not essential to allow cell survival while glutathione synthesis was essential but not sufficient to prevent the inhibition of epithelial cell growth, polarity development and morphogenesis in response to CSE. Interestingly, NAC was not able to restore the polarity and morphogenesis phenotype of CSE-pre-exposed cells ( Figure 4C and D), indicating that NAC was required at the time of CSE exposure to neutralize the effects of CSE.

Discussion
The ability of intestinal epithelial cells to selfassemble and form polarized monolayers in conjunction with basement membrane remodeling is a crucial process in the generation of epithelial tissue architecture and epithelial cell homeostasis, including the wound healing response. Here, we show for the first time that cigarette smoke extract directly perturbs the de novo multi-dimensional organization of intestinal epithelial cells in 3D culture, whereas no effect is observed in pre-formed intestinal epithelial cultures. As part of the underlying mechanism, we demonstrate that cigarette smoke impairs cell growth and the de novo development of intestinal epithelial cell polarity (i.e., the segregation of apical and basolateral plasma membrane domains bordered by tight junctions), which is an essential step in the 3D organization of columnar epithelial cells in vitro and in vivo 29 .
Previous studies have suggested that cigarette smoke may exert its negative effects on intestinal epithelium homeostasis indirectly via altering the immune system or microbiota 17 . While not excluding these possibilities, our study now indicates that also direct effects of cigarette smoke components on the intestinal epithelial cells that interfere with epithelial morphogenesis should be considered. The precise molecular pathways via which CSE 151 impairs Caco-2 cell polarity development remains to be elucidated. CSE impairs epithelial sphere size: after 7 days of CSE exposure the size of cell clusters was similar to that of 3 day-old non-exposed cell clusters. Three day-old non-exposed Caco-2 cells already polarize and organize into single lumenforming spheroids 34 (our unpublished data). Therefore, the reduced size of CSE-exposed cell clusters, as such, is not likely to underlie the cell polarity defects. This is further supported by the notion that the effects of CSE on cluster size and morphogenesis did not follow the same trend as a function of CSE dose ( Figure 4A, B). Nevertheless, as cell growth and differentiation are intimately linked processes, it is possible that the reduced growth rate in CSE-exposed cells affects cell polarity development and morphogenesis.
Intestinal epithelial cell differentiation and polarity development depends on cell-cell interactions mediated by adherens junctions and tight junctions [48][49][50] , although the necessity for cell-cell interactions for intestinal epithelial cell polarity development has been debated 51,52 . In airway epithelial cells, exposure to smoke-concentrated medium resulted in loss of cell polarity as a result of perturbed function of p120-catenin function 53 , a protein that plays a role in maintaining the integrity of adherens and tight junctions 54 . Our data demonstrate that ZO-1 is mislocalized in CSE-exposed intestinal epithelial cells, and ZO-1 has been reported to couple the assembly of tight junctions to adherens junctions 55 . Further studies are needed to investigate whether defective adherens and tight junction-mediated cell-cell adhesions underlie the CSE-mediated cell polarization defects in intestinal epithelial cells as reported in this study. Notably, the inhibitory effects of CSE on cell polarization and morphogenesis was passed on to next generation of cells, even when to were no longer exposed to CSE. Given that smoking has 152 been demonstrated to induce epigenetic changes 56,57 , and that epithelial cell organization has been linked to epigenetic regulators 58,59 , it is conceivable that epigenetic changes underlie the polarization and morphogenesis defects caused by CSE. While the effects of cigarette smoking on epigenetic modifications in blood cells has been investigated 56 , the effects of smoking on epigenetic modifications in the intestinal epithelium remain unknown. The ability of the intestinal epithelial cells to (re) establish cell polarity and (re)organize themselves into polarized monolayers in response to cues from the extracellular matrix (that is, the basement membrane) is a crucial step in epithelial repair processes 60 . Loss of epithelial cell polarity and consequent defects in multicellular organization has also been associated with the development of cancer 58,61,62 , and cigarette smoking is a risk factor for the development of colorectal cancer. Based on our results we speculate that, in vivo, exposure to cigarette smoke components perturbs repair processes in response to epithelial damage or lesions (such as occur in inflammatory bowel disease) by impairing the ability of intestinal epithelial cells to repolarize and reassemble into monolayers, thereby preventing disease remission and eventually contributing to aberrant organization of the intestinal epithelium and onset of polyp formation and/or carcinogenesis. In this regard it is of interest that especially Crohn's ileitis is worsened by smoking, and that the ileum is also the site where the first lesions in Crohn's disease are found 12 . It is tempting to extrapolate the results of this study to other polarized tissues. Indeed, cigarette smoking is also a significant risk factor for intraoral epithelial wound healing after oral surgery 63 , corneal epithelial wound healing 64 , for lung epithelial wound healing 65,66 and for the outcome of plastic surgery patients 67,68 . Possibly, cigarette smoke-induced defects in epithelial cell polarity development 153 and morphogenesis underlie smoking-associated impaired wound healing also in these typically polarized tissues. Our preliminary data indicate that CSE has similar effects on morphogenesis of MDCK cells, which are tubular kidney epithelial cells (our unpublished results), supports this possibility and warrant further studies. Importantly, the inhibitory effects of cigarette smoke extract on intestinal epithelial cell polarity development and morphogenesis were efficiently neutralized by NAC. Interestingly, NAC-involving therapy was reported to accelerate mucosal healing in a rodent model of colitis 69 . However, the mechanisms by which NAC exerts its beneficial effects on intestinal mucosal healing remain elusive. Whereas NAC is typically used as a precursor for glutathione synthesis, we did not observe elevated levels of intracellular glutathione in NAC-treated Caco-2 cells. Yet, the neutralizing effect of NAC was completely abolished by the glutathione synthesis inhibitor BSO. These data suggest that the neutralizing effect of NAC was not through NAC-stimulated glutathione synthesis, but critically dependent on endogenous (that is, non-NACmediated) glutathione synthesis. Of interest in this respect, NAC was reported to promote neuronal PC12 cell survival independently of glutathione but via stimulating transcriptional activity 70 . Moreover, NAC was recently demonstrated to stimulate mammalian target of rapamycin (mTOR)-mediated protein synthesis in intestinal epithelial IPEC cells independent of glutathione synthesis 71 . Given the role of mTOR signaling in polarity of Sertoli cells 72 , further studies are warranted to explore the role of mTOR in NAC-mediated neutralization of CSEmediated impairment of intestinal epithelial polarity development and morphogenesis. Our data provide further insight into the cell biological mechanisms that may underlie beneficial effects of NAC in the clinical treatment of smoking-related intestinal 154 disorders. The cigarette smoke component(s) that are responsible for the detrimental effects on 3D intestinal epithelial cell morphogenesis remain to be determined. Cigarette smoke contains free radicals and many carcinogens, including polycyclic aromatic hydrocarbons 73 . In a study that examined the influence of functional glutathione S-transferases (GST) gene polymorphisms on the association between smoking and colorectal cancer, Koh and colleagues reported a role for GST enzymes in the detoxification of colorectal carcinogens in tobacco smoke 9 . We did not obtain evidence of CSE-induced production of reactive oxygen species in the Caco-2 cell spheroids (our unpublished results). Moreover, menadione, a synthetic compound also known as 2-Methyl-1,4-naphthoquinone or vitamin K3 that induces reactive oxygen species, did not mimic the effects of cigarette smoke extract Moreover, the effects of cigarette smoke extract were not efficiently prevented by the reactive oxygen species inhibitor resveratrol (our unpublished results). Given the clear requirement for glutathione synthesis for the neutralizing effect of NAC in our experiments, it is conceivable that cigarette smoke components that are substrates for GST are responsible for the effects of CSE on intestinal epithelial morphogenesis as demonstrated in this study, and possibly involved in the in vivo etiology of smoking-related intestinal epithelium dysfunction.