Epithelial Barrier Function In Vivo Is Sustained Despite Gaps in Epithelial Layers
Article Outline
Background & Aims: Epithelial cells of the small intestine migrate to the tip of the villus at which they are shed. It is not understood how the intestinal barrier is maintained during this high cell turnover. The aim of this study was to use high-resolution in vivo light microscopy to investigate the mechanism of epithelial shedding and the site of the permeability barrier during cell shedding. Methods: A laparotomy was performed on anesthetized mice, and a segment of small intestine was opened. The exposed epithelial surface of the intestine was imaged by multiphoton microscopy. Nuclei, cytosol, and cell membranes were imaged using the dyes Hoescht 33258, BCECF, a transgenically expressed fluorescent protein, and the membrane dye DiI. The fluorescent caspase substrate PhiPhiLux was used to detect apoptosis. Results: In the epithelial monolayer, gaps were observed that lacked nuclei or cytosol but appeared to be filled with an impermeable substance. Studies with membrane impermeant fluorophores (Lucifer Yellow and Alexa-dextran) showed that the impermeable substance completely fills the void left by the absent cell. Only a fraction of gaps have either ZO-1 staining or cytoplasmic extensions from neighboring cells at the basal pole. Time-lapse studies reveal that cell shedding results in genesis of a gap and that shedding usually occurs prior to detectable cellular activation of caspase 3 or nuclear condensation. Conclusions: Results suggest that epithelial barrier function is sustained at the apical pole of the epithelial layer, despite discontinuities in the cellular layer.
Abbreviations used in this paper: BCECF-AM, 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester , DiI, 1,1′-dilinoleyl-3,3,3′,3′-tetramethylindocarbocyanine, 4-chlorobenzenesulfonate
The intestinal epithelium presents a permeability barrier to the luminal contents that prevents undesirable solutes, microorganisms, and luminal antigens from entering the body.1, 2 However, it remains mysterious how the intestinal barrier is sustained during the high rate of normal cellular turnover in the epithelium. Epithelial cells of the mammalian small intestine arise from stem cells at the base of the crypt and migrate up to the villus at which they are shed. In the mouse, ∼1400 cells are shed from each villus per 24 hours.3 It is remarkable that, despite this loss of ∼1 cell/minute, the functional permeability pore size on the villus is less than .6 nm, some 20,000 times less than the diameter of a villus epithelial cell.4 The simplest explanation is that mechanisms of cell shedding somehow do not disturb the epithelial barrier in healthy tissue.
Tight junctions are unequivocally a site and structure that restricts paracellular flux between adjacent epithelial cells. Furthermore, regulation of the tight junctional complex is a well-established mechanism controlling passive fluxes during sodium and glucose absorption,5 and altered tight junctional structure has been observed in several diseases displaying increased intestinal permeability (for review see Clayburgh et al2). It remains less clear whether loosening of tight junctions is solely responsible for the increased intestinal permeability that has been implicated in the pathogenesis of inflammatory bowel disease, celiac disease, graft vs host following bone marrow transplantation, and the response to intestinal pathogens.6, 7, 8, 9, 10, 11, 12, 13 Deficiency in heparin sulfate at the basolateral surface of enterocytes has been implicated in protein-losing enteropathy and lowered transepithelial resistance.14 Abnormality of cell shedding has been invoked in the pathogenesis of serrated polyps in the colon.15 One hypothesis is that deranged cell shedding could contribute to dysfunction of the epithelial barrier in a variety of intestinal disorders. Our lack of understanding about the process of in vivo villus cell shedding limits our ability to test this hypothesis.
The mechanisms of intestinal epithelial cell shedding remain highly controversial.16 Apoptosis has been suggested as one mechanism of cell death at the villus tip.17, 18 In detached intestinal epithelial cells, apoptosis occurs with activation of the initiator caspases 2 and 9, with subsequent hierarchical activation of executioner caspases such as caspase 3.19 However, it is not clear whether apoptosis is a cause or consequence of the shedding process because apoptotic bodies are rarely observed within the villus epithelium.18, 20 Interestingly, tissue culture models reveal parallels between the shedding of apoptotic cells and epithelial wound closure after physical removal of cells. In both cases, a coordinated purse-string contraction of actin filaments surrounding the monolayer defect helps to close the gap and restore barrier function.21, 22, 23 A major regulator of this contraction is the phosphorylation of myosin II regulatory light chain by myosin light chain kinase and rho-associated kinase.24, 25 Tight junctions may also play a role to maintain the epithelial barrier during physiologic cell shedding. It has been observed by electron microscopy that neighboring cells form a tight junction beneath the extruding cell.26 Finally, a role for subepithelial myofibroblasts has been proposed, wherein they contract following epithelial cell loss or injury, thereby restoring epithelial continuity.27 The contribution of each of these elements to the maintenance of barrier function during physiologic cell shedding remains speculative.21, 28
Studies to date give only limited insight into barrier maintenance during cell shedding in vivo because investigators have been limited to study of either fixed tissue, cells collected from the intestinal lumen, or cell culture models. We have combined fluorescent probes with confocal and multiphoton microscopy to allow real-time study of epithelial architecture, cell shedding process, and barrier function in living mice having an intact circulation to the gut mucosa.29, 30, 31 Results suggest that the previously identified mechanisms cannot fully explain the observed maintenance of epithelial barrier function.
Materials and Methods
Surgical Preparation
Surgical procedure was a modification of published in vivo procedures for rodent stomach.29 Mice (ICR) were housed in a standard 12-hour light/dark cycle with lights on at 0600 hours. Experiments were performed routinely between 1300 and 2000 hours. Mice were anesthetized with thiobutylbarbital 100–150 mg/kg intraperitoneally (IP) (Inactin; Sigma Chemical Co, St. Louis, MO). A tracheotomy was performed to facilitate breathing. A mid-abdominal incision (1–1.5 cm) was made, and a segment of small intestine was exteriorized, flushed with saline, and opened longitudinally along the antimesenteric border by cutting cautery. The anesthetized animal was placed supine on a custom-built chamber on the stage of a multiphoton microscope (Zeiss LSM 510 NLO; Ziess, Jena, Germany) that was heated to 37°C by a circulating water bath. All subsequent drug treatments were made on the anesthetized mice. The opened intestinal segment was placed, luminal surface down, on the inverted microscope stage while bathed in 0.9% NaCl and the villus epithelium imaged with a 40× C-Apo objective. Muscular contraction of the intestine was minimized by application of xylazine directly on the gut. At the end of the experiment, the animal was humanely killed. All procedures were approved by the animal care and use committee of Indiana University.
Microscopy
All images were collected at 20–50 μm below the villus tip in anesthetized animals by a combination of confocal and 2-photon imaging. Autofluorescence images were collected with 2-photon excitation (710 nm) and 380–650 nm emission wavelengths. Nuclear staining was achieved by IP injection of 2 mg/kg Hoescht 33258 (Molecular Probes, Eugene, OR),31 and images were collected with 800-nm excitation and 435–485-nm emission. 2′,7′-Bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (BCECF-AM; 10 μmol/L in saline; Molecular Probes) was loaded into intestinal epithelial cells by direct application to the gut mucosal surface for 15 minutes and confocally imaged at 488-nm excitation and >505-nm emission. Cytosolic YFP fluorescence was imaged at 514-nm excitation and 530–630-nm emission. Plasma membranes were stained by exposure of the gut mucosal surface 15 minutes to 1,1′-dilinoleyl-3,3,3′,3′,-tetramethylindocarbocyanine, 4-chlorobenzenesulfonate (DiI; 5 μL Vybrant cell stain, Molecular Probes) and imaged with 543-nm excitation and 565–615-nm emission. Lucifer Yellow (100 μmol/L in the luminal fluid; Molecular Probes) was used as a membrane impermeable marker of the luminal compartment, imaged with 800-nm, 2-photon excitation and 530–650-nm emission. Dextran MW 10,000 conjugated to Alexa Fluor 647 (2 mg/mL, Molecular Probes) was injected intravenously to image blood vessels and to act as a permeability marker for the extracellular space beneath intestinal epithelial cells. Exposure of the luminal mucosal surface to PhiPhiLux (10 μmol/L; Oncoimmunin, Gaithersburg, MD) identified caspase-3-like activity, imaged at 543-nm excitation and 565–615-nm emission. Confocal reflectance images were collected by reflecting excitation light wavelengths to a confocal detector.
Histochemistry
Murine small intestine was gently flushed with physiologic saline, placed in 10% neutral-buffered formalin saline for 6 hours, processed through to paraffin blocks, and 4-μm sections cut. Great care was taken not to disturb the section of intestine to be sectioned so as to preserve fragile structures. Sections were stained with either alcian blue/diastase-periodic acid schiff for goblet cells or H&E. The human specimen was collected from a right hemicolectomy patient after informed consent was given. The specimen had minimal handling and warm ischemia before fixation. Ethical approval was given by the local research ethics committee at University of Liverpool (Study no. 03/09/182/C [A]).
Immunostaining
Tight junctional structures were identified in formalin-fixed paraffin-embedded sections (5 μm) of murine intestine. Sections were incubated with rabbit anti-ZO-1 antibody (Zymed catalog 61-7300). Prior to immunostaining, sections were subjected to proteolytic enzyme digestion for 60 minutes at 37°C, using 0.7% Trypsin (VWR International Ltd, Cat. No. 390414M) in Tris-buffered saline (TBS; 0.05 mol/L Tris, 0.12 mol/L sodium chloride, pH 7.6). After rinsing in tap water followed by deionized water, sections were transferred to an Autostainer (DakoCytomation, Denmask) for staining. Sections were incubated with primary antibody for 40 minutes at room temperature and washed with TBS-Tween (TBS with 0.05% Tween-20); the detection system was ChemMate EnVision HRP (DakoCytomation, Cat. No.K5007), which was used according to the manufacturer’s instructions. Sections were removed from the Autostainer, counterstained with Mayer’s haematoxylin, dehydrated through ethanol, cleared in xylene, and coverslipped using a resinous mountant.
Results
We have used multiphoton and confocal microscopy for real-time study of individual epithelial cell dynamics and barrier function in the small intestine of living mice.29, 30, 31 To observe epithelial architecture in living native tissue, initial studies evaluated cellular autofluorescence in villi of normal, anesthetized mice in response to 2-photon excitation (Figure 1A). Settings were used that had previously been established to measure NAD(P)H fluorescence in response to 2-photon excitation.32 Villus epithelial cells displayed cytosolic, but not nuclear, autofluorescence, and occasional gaps were noted within the annulus of cells in villus cross-sections (Figure 1A, arrows). Simultaneous confocal reflectance imaging, used to identify physical structures that reflect 710-nm laser light,29 reported that this apparently cytoplasm-free gap contained reflective (yet nonfluorescent) material (Figure 1B and 1C).
Figure 1.
Intestinal epithelium is a discontinuous monolayer in the living animal. (A) Autofluorescence of intestinal epithelial cytosol. (B) Confocal reflectance of intestinal epithelial cells. (C) Overlay of images A and B. (D) Nuclear fluorescence with the Hoechst DNA stain. (E) Cytosolic BCECF fluorescence. (F) Confocal reflectance. (G) Overlay of images D, E, and F. Images from transgenic YC3.0 calcium cameleon mice. (H) Nuclear fluorescence. (I) Cytosolic CFP fluorescence. (J) Confocal reflectance. (K) Overlay of images H, I, and J. (L) Nuclear fluorescence with Hoechst DNA stain. (M) Membrane DiI fluorescence. (N) Confocal reflectance. (O) Overlay of images L, M, and N. Bars = 20 μm.
Four vital stains were used to probe annular gaps. Tissue of anesthetized animals was loaded with the cytoplasmic dye BCECF and the nuclear DNA dye Hoechst 33258. Gaps in the annular ring of fluorescent nuclei occurred (Figure 1D, arrow), coincident with lack of cytoplasmic staining (Figure 1E), yet containing reflective material (Figure 1F and 1G). Using mice transgenic for a fluorescent protein (YC3.0 calcium chameleon transgenic mice), results also confirmed that nuclear gaps (Figure 1H, arrows) were coincident with lack of cytoplasmic fluorescent protein (Figure 1I), despite the presence of reflective material (Figure 1J and 1K).
Staining with the membrane dye DiI33 demonstrated that discontinuities in the apical brush border membrane (Figure 1M, arrow) occurred when nuclei were absent (Figure 1L), despite continued presence of reflective material (Figure 1N and 1O). Overall, 6 independent markers report that the small intestinal villus epithelium of living mice is a discontinuous layer, interrupted by gaps having coincident loss of apical brush border membrane, cytosol, and nucleus.
In all analyses, serial optical sections at 1-μm intervals identified gaps that were approximately the same volume as an individual cell, using orthogonal views at the plane of the nuclei (Figure 2A). In contrast to epithelial wound healing models, cells neighboring gaps were not wider or flatter than epithelial cells distant from gaps.21, 34 Focusing in from the villus tip, subcellular resolution of villus epithelial cells was observed to 70-μm depth. This is equivalent to the top 20% of the villi (average length of the mouse villus is 350 ± 22 μm; n = 50 villi). Morphometric analysis in vivo showed that ∼3% of cell positions were gaps, equally distributed along the final 70 μm of the villus tip (Figure 2B). This may underestimate the frequency of gaps in some regions, as en face views of cytosolic fluorescent protein near the apical surface of epithelial cells (Figure 2C) show that the gap diameter is variable, often less than adjacent epithelial cells. In formalin-fixed sections, en face gaps were distinct from goblet cells (Figure 2D). Data demonstrate frequent discontinuities in the epithelial layer of living mice, implying that tight junctions at the apical pole of cells cannot be the only mechanism sustaining intestinal barrier function13 and that another mechanism must be invoked within the gaps.
Figure 2.
Frequency and appearance of epithelial cell gaps. (A) En face and associated orthogonal views of villus from cameleon transgenic mouse (green; fluorescent protein) additionally stained with HOESCHT 33258 (blue) as in Figure 1. En face gap shown in crosshairs is confirmed by associated orthogonal views optically slicing thru epithelial layer. Orthogonal views along green and red axes are presented in the green and red boxes. (B) Percentage of epithelial cell positions that lack cells, as a function of distance from the villus tip. Serial optical sections were taken at 1-μm intervals 0–70 μm from the villus tip. Cell positions were visualized by nuclear staining and confocal reflectance. Gaps were defined as a region ∼10 μm in diameter in the x, y, and z dimension with coincident loss of nucleus (Hoechst 33258 stain) and cytosol (assayed by cytosolic BCECF, autofluorescence, or CFP). Thirty-two villi were counted from 10 mice. (C) En face view of villus cells of cameleon transgenic mouse imaged as in Figure 1 demonstrates variability in diameters and angularity of the boundary of cell-free zones (yellow arrows) vs epithelial cells (green: fluorescent protein; red: reflectance). (D) Formalin-fixed sections of mouse tissue. Histologic stain for mucins positively distinguishes goblet cells (red arrows) from gaps (yellow arrow) when villus viewed en face.
It has been suggested that tight junction formation beneath shedding cells can sustain epithelial barrier function.26 In live tissue studies using cells expressing a cytosolic fluorescent protein, we tested whether such a mechanism explained observations. In a population of 22 gaps, 45% had lamellipodia from neighboring cells that contacted each other under the epithelial gaps (Figure 3A shows CFP fluorescence of a representative cell, which is overlaid with confocal reflectance [red] in Figure 3B). In the remaining 55% of gaps, there was no discernable cytoplasm from neighboring cells in the gaps (Figure 3C and 3D). Using fixed tissue, we immunostained the tight junction protein ZO-1 to determine whether tight junctions were present under shedding cells. In a population of 73 shedding cells, 9% were observed to be bounded by neighboring cells contacting at the basal pole in a V-shaped formation with a spot of ZO-1 immunoreactivity at the apex of the V (Figure 3E). The remaining cells had no identifiable ZO-1 immunoreactivity near the basement membrane (Figure 3F). Thus, in both fixed and live tissue, evidence suggests that tight junctions from neighboring cells can only be used in a fraction of instances to help reseal the breach in the epithelial layer.
Figure 3.
Features at basal pole of gaps and shedding cells. A–D are images from cameleon transgenic mice showing CFP fluorescence (A,C) or CFP fluorescence in green overlaid with confocal reflectance in red (B,D, respectively). Yellow arrow, apical pole of cell; yellow arrowhead, basal pole. (A and B) Example of gap with cytoplasmic extensions across basal pole. (C and D) Example of gap without cytoplasm at basal pole. (E and F) Formalin-fixed sections of mouse tissue immunostained for ZO-1 (brown peroxidase) and counterstained with hematoxylin. (E) Red arrow indicates location of ZO-1 staining at base of shedding cell. (F) Red arrow shows absence of ZO-1 staining at base of shedding cell.
To test whether gaps present a luminal permeability barrier, we added the cell impermeable dye Lucifer Yellow35 to luminal fluid and identified cell-free regions by lack of Hoescht-stained nuclei. The majority (229 of 235, 97%) of nucleus gaps (Figure 4C, arrow) did not permit entry of Lucifer Yellow (Figure 4A), presumably because of being filled with the reflective material (Figure 4B). In the remaining 3% of gaps (6 of 235), entry of Lucifer Yellow occurred (Figure 4A, arrowhead) but never extended deeper than the line of nuclei from adjacent cells. Deeper entry of Lucifer Yellow appeared limited by the confocal reflective substance (Figure 4B and 4D, arrowhead). We noted one additional category in which Lucifer Yellow fluorescence extended into the epithelial layer identified by confocal reflectance. In these cases, nuclei were present but were more centrifugally positioned than neighbors (Figure 4G, arrowhead), suggesting that the cells were in the process of being shed. Lucifer Yellow permeation (Figure 4E) extended into the space enveloping the exiting nuclei, suggesting that cell membrane integrity had been compromised during cell shedding. Thus, 45% of all Lucifer Yellow intrusions into the epithelial layer (5 of 11) were into regions with exiting nuclei. Conversely, no centrifugal nuclei were observed that sustained impermeability to Lucifer Yellow.
Figure 4.
Restricted permeation of luminal Lucifer Yellow into epithelial layer. Lucifer Yellow (100 μmol/L) was added to the fluid bathing the mucosal surface to image all compartments accessible to luminal fluids. Images were compensated for limited bleed over of Hoechst 33258 fluorescence into the Lucifer Yellow channel. (A and E) Lucifer Yellow fluorescence in intervillus space does not permeate the epithelial layer or enter most cell-free gaps (arrow). Limited permeation of some gaps was observed (arrowheads; see text). (B and F) Confocal reflectance of 800-nm light. (C and G) Nuclear fluorescence of Hoechst 33258-stained DNA used to define gaps (arrow/arrowhead). (D and H) Overlay images. Bars = 20 μm.
To localize the permeability barrier relative to the basal pole of the epithelial layer, we injected a 10-kilodalton fluorescent dextran (conjugated to Alexa Fluor 647) intravenously into mice expressing a cytosolic fluorescent protein. As shown in Figure 5A, dextran fluorescence was observed in the lateral intercellular spaces between adjacent epithelial cells up to the level of the tight junctions (gap identified by lack of CFP fluorescence as shown in Figure 5B). The dextran also permeated into the perimeter of gaps but did not extend as far toward the apical membrane. As shown for 3 representative gaps in Figure 5C–E, this is most clearly seen in en face views near the apical surface, at which dextran fluorescence surrounding gaps is minimal compared with that observed in lateral spaces between adjacent cells.
Figure 5.
Permeation of gaps from the serosal fluid compartment. Fluorescent dextran (10,000 mw, conjugated to Alexa Fluor 647) was injected intravenously into transgenic YC3.0 calcium cameleon mice to image all compartments accessible to serosal fluids. Panels A, C, D, and E show dextran fluorescence alone, overlaid as red channel with CFP fluorescence (green channel) in panels B, F, G, and H, respectively. Asterisks indicate location of gap in epithelium. (A and B) Arrow indicates apical pole and arrowhead the basal pole of epithelium. Dextran permeates into lateral intercellular spaces and perimeter surrounding basal portion of gap. (C and F) En face views of villus epithelium near apical boundary of cells. Dextran fluorescence is excluded from the region surrounding gaps, although it can be readily detected in lateral space between adjacent cells. D, G and E, H are 2 other representative gaps to reinforce conclusions from C, F.
To monitor the biogenesis of cell-free gaps, we made time-lapse images of Hoechst 33258-stained villus tips every 1–3 minutes in anesthetized animals (Figure 6A). Nuclei were shed from the monolayer at a speed of 0.83 ± 0.06 μm/minutes (n = 53 cells), initiating departure at apparently random times (Figure 6B, upper graph). Time-lapse series revealed that resultant gaps were not fully filled in by migration of neighboring cells for at least 60 minutes (data not shown), explaining the heterogeneity of gap size noted earlier in Figure 2C.
Figure 6.
Biogenesis of gaps: time-lapse studies of cell shedding. (A) Time-lapse images of a Hoechst 33258-stained cell being shed (arrow). (B) Upper graph of internuclear distance between 3 shed cells and their immobile neighbors. Lower graph of nuclear fluorescence intensity in same cells over same time course. Panels C and D are fixed sections of mouse tissue. (C) Cells being shed from the murine villus tip stained by H&E. (D) Stained for mucins (arrow in inset shows goblet cell). (E) Fixed section of human small intestine stained with H&E. Bars = 20 μm.
We sought to test whether intact cells or cell fragments were shed during the biogenesis of gaps. Apoptosis has been suggested as the mechanism of cell loss at the villus tip.18, 19, 20 Measurement of nuclear DNA fluorescence intensity provided a tool to track condensation of chromatin, a classic morphologic feature of apoptosis.36 In the in vivo time-lapse experiments, no condensation occurred during or prior to cell shedding (Figure 6B, lower graph), suggesting that this hallmark of the final stages of apoptosis was not activated early in cell shedding. In contrast, Hoechst 33258-stained nuclei already shed into the luminal space were 1.92- ± 0.51-fold brighter than cells residing within the epithelial layer (mean ± SEM, n = 20 in each category, P < .001, n = 20). Histologic staining of murine small intestine confirmed that only cells late in the shedding process have fragmented nuclei or condensed chromatin (Figure 6C and 6D). Regions devoid of nuclei also occurred beneath cells being shed from the villus tip (Figure 6C and 6D) and were distinct from goblet cells (Figure 6D, insert). Similar histologic outcomes were found in human small intestine (Figure 6E).
We tested whether caspase activation occurred within intact cells prior to initiation of cellular shedding. Preliminary experiments confirmed that sheets of intestinal epithelial cells freshly isolated from murine small intestinal villi activate caspase 3 as indicated by PhiPhiLux (PPL), a fluorogenic substrate for caspase 3-like activities (data not shown).37 Addition of PPL to the luminal solution bathing villi confirmed that bright nuclei (Figure 7A’, arrowhead) colocalized with areas of brighter PPL fluorescence in the lumen (Figure 7A”, arrowhead): Note that the PPL substrate is also fluorescent, so lumen was labeled with background PPL fluorescence. Thus, activation of caspase and nuclear condensation confirmed apoptosis of previously shed cells. In contrast, during shedding, the most common observation (10 of 12) was that cells remained unstained by PPL (Figure 7A–C, arrow). However, 2 examples were noted of cells with PPL-positive cytoplasm both apical and basal to the annular ring of nuclei during the process of cell shedding (Figure 7D–G), in which both the nucleus and surrounding cytoplasm with activated caspase is expelled into the lumen. We conclude that full caspase-3 activation is an uncommon event prior to cell shedding,38 and later stages of apoptosis are only observed in cells already shed. All observations suggest that shedding of intact cells is a mechanism leading to annular gaps.
Figure 7.
Caspase activation is occasionally seen during cell shedding. Time course studies of tissues stained with Hoechst 33258 and exposed to 10 μmol/L PhiPhiLux in the luminal fluid. A” PhiPhiLux fluorescence; A’ nuclear fluorescence of Hoechst 33258-stained DNA; A overlay of images A” and A’. Nuclei with condensed chromatin (A’, arrowhead) are in the lumen, surrounded by brighter PhiPhiLux (A”). (B and C) Overlay images collected 2 and 5 minutes later. The arrow indicates a cell being shed but remaining PhiPhiLux negative (A–C). (D–G) Time course of shedding for 1 cell (at arrow) with PhiPhiLux-positive cytoplasm.
Discussion
Using confocal and multiphoton microscopy, we have performed the first high-resolution imaging of intestinal epithelial morphology and epithelial cell dynamics in vivo. We were surprised to demonstrate discontinuities in the intestinal epithelium using 6 separate imaging modalities (nuclear DNA stain, autoflourescence of NAD(P)H, cellular uptake and enzymatic conversion of BCECF/AM to yield the fluorescent product BCECF, apical membrane staining with DiI, transgenic fluorescent protein expression in the cytosol, and confocal reflectance). Equivalent gap-like structures have also been seen in intestinal villi with scanning electron microscopy as has been described previously.39 The simplest explanation for the observations was that cell packing in the epithelial layer is imperfect, such that approximately 3% of the epithelial layer lacks a cell (Figure 2B). This led us to question how and where barrier function is maintained around such gaps and to ask whether physiologic cell shedding can generate gaps.
Our in vivo observations cannot be completely explained by known mechanisms for maintaining intestinal barrier function. The model proposed by Madara in 199026 requires that tight junctions be formed underneath cells that undergo shedding. We observed that 45% of shedding events had cytoplasmic extensions from neighboring cells that appeared to make a continuous structure beneath the shedding cell and that only 9% of shedding events showed basal ZO-1 staining. By both criteria, only a fraction of gaps generated by cell shedding could be sealed in a manner consistent with the Madara model. Furthermore, the site of luminal impermeability in a fully formed gap is at the apical boundary (Figure 4), not the basal pole. Because gaps are not rapidly closed by entry of adjacent cells, it also seems clear that barrier function is not solely dependent on mechanisms that require cell migration, shape changes, or mechanical forces that restore epithelial continuity. Thus, although mechanisms of epithelial wound closure, trefoil peptide stimulated epithelial migration, or myofibroblast contraction may contribute to restoration of epithelial continuity, they cannot readily explain how the epithelial barrier is sustained across the gaps we observe.24, 25, 27, 40, 41
Remarkably, the apparently cell-free gaps are not empty. They are filled with a substance that so far has only been positively identified as a material that reflects laser light. The material is impermeable to aqueous dyes added from either the luminal or serosal compartments, is stable for at least an hour (observed duration of gaps), and must enter gaps synchronous with the cell shedding process. Our studies have not revealed the identity of the plugging substance. We speculate that the impermeable extracellular material is secreted by neighboring epithelial cells and/or myofibroblasts (the latter known to secrete basement membrane protein matrix42, 43, 44). A plausible candidate is heparin sulfate. Congenital heparin sulfate deficiency is known to lead to protein-losing enteropathy and diarrhea.14, 45
Our observations also provide some additional information about mechanisms of cell shedding. We observe that cells that are being shed are largely intact and have not progressed into late stages of apoptosis prior to loss from the monolayer. A number of investigators have suggested that apoptosis is the reason for cell shedding.18, 46 Our results suggest that the majority of shed cells do not have an apoptotic morphology or sufficient caspase activation to be detected by PhiPhiLux. Therefore, we suggest that, if apoptosis is the cause of cell shedding, it must be early events in the apoptotic cascade that drive cell detachment and expulsion.47 We also observe that cells are shed at a speed that is at least 2-fold greater than the diurnally varying migration up the mouse villus (maximum speed 0.3 μm/min).48 We propose that loosening of contacts between neighboring intestinal cells and the basement membrane leads to forceful cell expulsion simultaneous with the secretion of an extracellular substance that fills the void left by the departing cell.
It is important to emphasize that there is considerable variation between mammalian species in the mechanism of cell shedding and maintenance of barrier function (for review, see Mayhew et al16). For example, in the guinea pig, apoptotic fragments are pinched off, leaving junctional complexes intact. Remarkably, nuclei are rarely found in the lumen.17 In reindeer and seals, cell fragments are lost from the villus either by extrusion or phagocytosis by neighboring macrophages. There is also evidence of necrotic cell death at the villus tip with breach of the epithelial monolayer.49 Our evidence suggests that the novel mechanism we have identified coexists alongside other mechanisms.
Our observations have implications for the mechanisms by which simple columnar epithelia may create and sustain a barrier between 2 body compartments. Our results suggest that defects in the sealing of epithelial gaps might be a new pathogenic mechanism to be considered in inflammatory and infective disease in which permeation of luminal antigens, toxins, or organisms occurs.12, 13
The authors thank Drs Roger Tsien and Varda Lev-Ram for the gift of calcium cameleon transgenic mice (creation of the mice was supported by NIH RO1 NS027177); Drs Hannah Carey, John Cuppoletti, Jörg Galle, Marcus Loeffler, Chris Potten, Mark Pritchard, and Jerrold Turner and Penny Ottewell for helpful discussions; and Mike Hershman for collection of the human specimens.
References
- . The future of GI and liver research: editorial perspectives: I. Visions of epithelial research . Am J Physiol Gastrointest Liver Physiol . 2003;284:G547–G550
- . A porous defense (the leaky epithelial barrier in intestinal disease) . Lab Invest . 2004;84:282–291
- . Stem cells: attributes, cycles, spirals, pitfalls and uncertainties. Lessons for and from the crypt . Development . 1990;110:1001–1020
- . Permeability of the rat small intestinal epithelium along the villus-crypt axis (effects of glucose transport) . Gastroenterology . 2000;119:1029–1036
- . Physiological regulation of epithelial tight junctions is associated with myosin light-chain phosphorylation . Am J Physiol . 1997;273:C1378–C1385
- . Intestinal permeability in children with Crohn’s disease and coeliac disease . Br Med J (Clin Res Ed) . 1982;285:20–21
- . Intestinal permeability in patients with Crohn’s disease and their healthy relatives . Gastroenterology . 1989;97:927–931
- . The gut mucosal barrier in bone marrow transplantation . Bone Marrow Transplant . 1990;5:373–377
- . CD45RO expression on circulating CD19+ B cells in Crohn’s disease correlates with intestinal permeability . Gastroenterology . 1995;108:132–137
- . IL-4 directly modulates function of a model human intestinal epithelium . J Immunol . 1994;153:2122–2129
- . Inflammatory bowel disease and adenomas in mice expressing a dominant negative N-cadherin . Science . 1995;270:1203–1207
- . Increased intestinal permeability precedes the onset of Crohn’s disease in a subject with familial risk . Gastroenterology . 2000;119:1740–1744
- . Intestinal epithelial responses to enteric pathogens (effects on the tight junction barrier, ion transport, and inflammation) . Gut . 2003;52:439–451
- . Heparan sulfate depletion amplifies TNF-α-induced protein leakage in an in vitro model of protein-losing enteropathy . Am J Physiol Gastrointest Liver Physiol . 2005;288:G1015–G1023
- . Emerging concepts in colorectal neoplasia . Gastroenterology . 2002;123:862–876
- . Epithelial integrity, cell death and cell loss in mammalian small intestine . Histol Histopathol . 1999;14:257–267
- . A novel mechanism for disposing of effete epithelial cells in the small intestine of guinea pigs . Gastroenterology . 1993;105:1089–1097
- . Regulation of cell number in the mammalian gastrointestinal tract (the importance of apoptosis) . J Cell Sci . 1994;107:3569–3577
- . Apoptotic signaling during initiation of detachment-induced apoptosis (“anoikis”) of primary human intestinal epithelial cells . Cell Growth Differ . 2001;12:147–155
- . Caspase activation during spontaneous and radiation-induced apoptosis in the murine intestine . J Pathol . 2001;195:285–292
- . An epithelial cell destined for apoptosis signals its neighbors to extrude it by an actin- and myosin-dependent mechanism . Curr Biol . 2001;11:1847–1857
- . Intestinal epithelial restitution. Characterization of a cell culture model and mapping of cytoskeletal elements in migrating cells . J Clin Invest . 1992;89:1501–1511
- . A novel cytoskeletal structure involved in purse string wound closure and cell polarity maintenance . J Cell Biol . 1993;121:565–578
- . PKC-dependent regulation of transepithelial resistance (roles of MLC and MLC kinase) . Am J Physiol . 1999;277:C554–C562
- . A differentiation-dependent splice variant of myosin light chain kinase, MLCK1, regulates epithelial tight junction permeability . J Biol Chem . 2004;279:55506–55513
- . Maintenance of the macromolecular barrier at cell extrusion sites in intestinal epithelium (physiological rearrangement of tight junctions) . J Membr Biol . 1990;116:177–184
- . Villus contraction aids repair of intestinal epithelium after injury . Am J Physiol . 1989;257:G274–G283
- . Single-cell epithelial defects close rapidly by an actinomyosin purse string mechanism with functional tight junctions . J Physiol . 2002;545:485–499
- . Dynamic regulation of gastric surface pH by luminal pH . J Clin Invest . 1999;103:605–612
- . Endogenous cyclo-oxygenase activity regulates mouse gastric surface pH . J Physiol . 2002;544:871–882
- . Functional studies of the kidney of living animals using multicolor two-photon microscopy . Am J Physiol Cell Physiol . 2002;283:C905–C916
- . Quantitative NAD(P)H/flavoprotein autofluorescence imaging reveals metabolic mechanisms of pancreatic islet pyruvate response . J Biol Chem . 2004;279:31780–31787
- . Vital imaging and ultrastructural analysis of individual axon terminals labeled by iontophoretic application of lipophilic dye . J Neurosci Methods . 1999;93:13–20
- . Modulation of barrier function during Fas-mediated apoptosis in human intestinal epithelial cells . Gastroenterology . 2000;119:1524–1536
- . Tubular lysosomes accompany stimulated pinocytosis in macrophages . J Cell Biol . 1987;104:1217–1222
- . DFF, a heterodimeric protein that functions downstream of caspase-3 to trigger DNA fragmentation during apoptosis . Cell . 1997;89:175–184
- . Assessment of caspase activities in intact apoptotic thymocytes using cell-permeable fluorogenic caspase substrates . J Exp Med . 2000;191:1819–1828
- . Control of adhesion-dependent cell survival by focal adhesion kinase . J Cell Biol . 1996;134:793–799
- . Ultrastructure of cell loss in intestinal mucosa . J Ultrastruct Res . 1977;60:272–277
- . Impaired defense of intestinal mucosa in mice lacking intestinal trefoil factor . Science . 1996;274:262–265
- . Trefoil factors (initiators of mucosal healing) . Nat Rev Mol Cell Biol . 2003;4:721–732
- . Cytokines modulate fibroblast phenotype and epithelial-stroma interactions in rat intestine . Gastroenterology . 1997;112:826–838
- . Adult human colonic subepithelial myofibroblasts express extracellular matrix proteins and cyclooxygenase-1 and -2 . Am J Physiol . 1997;273:G1341–G1348
- . Myofibroblasts. II. Intestinal subepithelial myofibroblasts . Am J Physiol . 1999;277:C183–C201
- . Congenital enterocyte heparan sulphate deficiency with massive albumin loss, secretory diarrhoea, and malnutrition . Lancet . 1996;347:1299–1301
- . Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation . J Cell Biol . 1992;119:493–501
- . Apoptosis and colorectal cancer . Gut . 2004;53:1701–1709
- . Cell migration velocities in the crypts of the small intestine after cytotoxic insult are not dependent on mitotic activity . Cell Tissue Kinet . 1986;19:601–610
- . Further evidence of species variation in mechanisms of epithelial cell loss in mammalian small intestine (ultrastructural studies on the reindeer (Rangifer tarandus) and seal (Phoca groenlandica)) . Cell Tissue Res . 1998;291:513–523
Supported in part by a grant from the HVC Foundation (Grant no. 10).The authors have no competing financial interests.
PII: S0016-5085(05)01117-0
doi:10.1053/j.gastro.2005.06.015
© 2005 American Gastroenterological Association. Published by Elsevier Inc. All rights reserved.
