- Email: [email protected]
Distribution of basement membrane pores in bronchus revealed by microscopy following epithelial removal
Journal of
Structural Biology Journal of Structural Biology 139 (2002) 137–145 www.academicpress.com
Distribution of basement membrane pores in bronchus revealed by microscopy following epithelial removal William J. Howat,*,1 Tony Barab as, James A. Holmes, Stephen T. Holgate, and Peter M. Lackie Respiratory, Cell and Molecular Biology, Division of Infection Inflammation and Repair, School of Medicine, University of Southampton, MP888, Level D, Centre Block, Southampton General Hospital, Southampton SO16 6YD, UK Received 13 February 2002, and in revised form 1 July 2002
Abstract The basement membrane of the bronchial epithelium separates the epithelial and mesenchymal compartments. Basement membrane pores allow cells to cross this boundary. We present a method for preparation of samples of human basement membrane allowing us easy visualisation and characterisation of the distribution and persistence of these pores. Columnar epithelial cells were removed from airway samples with gentle scraping with a circular glass coverslip. In contrast, the underlying basal cells required incubation once in dithiothreitol and twice in ethylenediaminetetraacetic acid. Scanning electron microscopy (SEM) at each stage of the epithelial stripping process showed the selective removal of epithelial cells with eventual visualisation of the pores. Using confocal microscopy on blocks of viable tissue, pores were shown to persist in culture for at least 5 days, despite the presence of viable cells in the submucosa. The distribution of pores in tissues determined by SEM was compared to simulations of three distribution patterns (random, clumped, and distributed). The pattern of pores in the samples was consistent with a random distribution. We suggest that basement membrane pores can be generated by the passage of infiltrating cells into the epithelium providing a network suitable for intraepithelial surveillance. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Cell trafficking; Confocal laser scanning microscopy; Nearest neighbour analysis; Scanning electron microscopy; Transmission electron microscopy
1. Introduction The basement membrane of the bronchial tract serves both as the substrate for the overlying epithelium and as the interface between the epithelium and the underlying mesenchyme. The epithelium itself consists of an upper columnar cell layer of ciliated or secretory cells attached to a basal cell layer, with tight junctions at the apical surface of the columnar cells providing an impermeable paracellular barrier. The integrity of the epithelium and its attachment to the basement membrane is facilitated through a number of adhesion mechanisms including
*
Corresponding author. Fax: +44-12-2349-4919. E-mail address: [email protected] (W.J. Howat). 1 Present address: The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK.
cell-cell desmosomal and cell-matrix hemidesmosomal connections. Infiltrating inflammatory cells and antigen-presenting cells can be found in both the submucosa and the epithelium of the airway in the normal respiratory tract and during inflammation. It had been assumed that infiltrating cell access to the bronchial epithelium was gained by matrix degradation. However, the description of pores in the basement membrane of tissues including intestine (Mahida et al., 1997) and conjunctiva (Scott et al., 1997) and under a variety of both normal and disease conditions indicated that this might not be required. It has recently been demonstrated that the bronchial basement membrane also contains such pores (Howat et al., 2001) with a mean diameter of 1.76 lm, of a size suitable for use by infiltrating cells and penetrating the full thickness of the basement membrane. They would therefore allow cell trafficking between the two
1047-8477/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 7 - 8 4 7 7 ( 0 2 ) 0 0 5 8 9 - 0
138
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
compartments in this key mucosal site where infiltrating cells provide essential immune defence against environmental pathogens. It is neither clear how basement membrane pores are formed nor how long they persist. Protease release from migrating inflammatory cells, particularly metalloproteinases, shows that inflammatory cells are capable of degrading basement membrane and basement membrane components (Okada et al., 1997) and could form pores in so doing. In this case we might expect a random distribution of pores and the persistence of the pores will determine the balance between de novo pore formation and reuse (Kobayashi et al., 1999; Leppert et al., 1995). Alternatively, pores could be formed during the development and growth of the airway when a random or a distributed pattern of pores could be formed using a mechanism such as that seen during feather bud formation and patterning in chick skin (reviewed in Hogan, 1999). Moreover, if there was a functional requirement for a heterogeneous distribution of pores, for example, coordinating with local blood vessels or lymphatics, we might expect that pores would be clumped, with areas of high local density. If as we propose, basement membrane pores are a primary route for infiltrating cell access to the epithelium, the distribution and frequency characteristics of the pores within the airway are critical to understanding infiltrating cell function in the respiratory tract. Since the basement membrane pores are relatively infrequent, making them difficult to see by transmission electron microscopy, and are normally covered by epithelium, it is important to be able to adequately remove the epithelium without damage to the basement membrane to reveal the pores. This permits the examination of large areas of the basement membrane by scanning electron microscopy (SEM)2 to establish the distribution of pores as well as their size and shape. In this report we describe and validate the methods to do this and determine the frequency of the basement membrane pores thus revealed. To determine the distribution of pores across the basement membrane, we have compared the pattern of pores found over the airway with a number of simple models of their distribution.
resection and stored in Leibowitz (L15) medium (Life Technologies, Paisley, UK) at 4 °C overnight. A total of 5 samples taken between June 1999 and January 2000 underwent the epithelial stripping procedure. 2.2. Basement membrane preparation and validation of procedure To ensure complete removal of the epithelium and to minimise tissue disturbance, we adapted our previously reported method (Howat et al., 2001). Tissue was washed in phosphate-buffered saline (PBS) to remove excess mucus. Samples for stripping were first incubated in Ca2þ /Mg2þ -free Hanks balanced salt solution (HBSS) (Life Technologies) for 10 min at 37 °C followed by 30 min of incubation at room temperature in 10 mM dithiothreitol (DTT) (Sigma, Poole, UK) in PBS. After a second incubation in HBSS, the bronchial tissue was incubated in 10 mM ethylenediaminetetraacetic acid (EDTA) (Sigma) in PBS for 30 min at 37 °C and repeated once. After each stage, while in HBSS, a 13-mm round glass coverslip was used to gently stroke the airway surface before fixation for electron microscopy. Tissue samples were collected for SEM before stripping and after each chemical treatment, DTT, EDTA (1), and EDTA (2). 2.3. Scanning EM preparation Tissue samples were fixed for SEM in 3% glutaraldehyde in 0.1 M cacodylate/0.23 M sucrose buffer for a minimum of 1 h. Fixed tissue was transferred to 0.1 M cacodylate/0.23 M sucrose buffer and dehydrated through a graded series of alcohols. The tissue underwent critical point drying with CO2 using a Balzers critical point drier (Balzers, Liechtenstein). Dried specimens were glued onto metal stubs with carbon-coated tabs and sputter-coated with gold/palladium. Scanning EM was performed on a Hitachi S800 (Hitachi, Tokyo, Japan) scanning electron microscope. Micrographs of three random areas of flat denuded basement membrane were taken at the same magnification (500Þ from every sample. 2.4. Transmission EM preparation
2. Methods 2.1. Sample collection Bronchial rings were excised from grossly normal areas of human bronchial tissue following lung cancer 2 Abbreviations used: SEM, scanning electron microscopy; PBS, phosphate-buffered saline; DTT, dithiothreitol; EDTA, ethylenediaminetetraacetic acid; HBSS, Hanks balanced salt solution; TIFF, tagged image file format.
Fixed tissue was transferred to 0.1 M cacodylate/ 0.23 M sucrose buffer before 2 h postfixation in 2% osmium tetroxide in cacodylate/sucrose buffer. The tissue was then incubated in 1.5% aqueous uranyl acetate for 30 min at room temperature before dehydration through a series of graded alcohols and embedding in Spurr resin following a standard protocol. A suitable area of epithelium in transverse section was selected and ultrathin sections were cut using a diamond knife on an Ultracut ultramicrotome (Leica, Milton Keynes, UK) set to
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
90 nm. Sections were dried onto copper grids, stained with lead citrate, and examined on a Hitachi H7000 (Hitachi, Tokyo, Japan) transmission electron microscope. 2.5. Confocal microscopy Unfixed, stripped bronchial tissue was glued to a petri dish with cyanoacrylate adhesive to ensure stability. The basement membrane was labeled with a dextran-rhodamine conjugate (10 lg/ml) (Molecular Probes, Leiden, Netherlands) diluted in PBS, for 1 h at room temperature, and washed three times with PBS. The sample was viewed from above the lumenal surface with a Leica TCS 4D confocal laser scanning microscope (Leica) using a 50 water immersion lens (N.A. ¼ 0.75) and a scan thickness of 2.9 lm. Images of the same area were taken over 5 consecutive days. Between scanning, samples were cultured in M199 medium (Life Technologies) supplemented with 2% Ultroser G (Life Technologies). At 1- and 2-day time points, the vital dye fluorescein diacetate (Jones and Senft, 1985) was used to assess viability of cells within the lamina propria.
139
samples. Coordinates for the points (‘‘pores’’) in the patterns were generated using Visual Basic with a randomly seeded random number generation function providing sets of coordinates within the analysis area (Random pattern). ‘‘Clustered’’ patterns of points were produced by generating 3 points randomly distributed up to 18 lm away from an initial random ‘‘seed’’ point. A ‘‘distributed’’ pattern was generated by discarding points that were within 9 lm of any existing point. Since the number of points per unit area affects the nearest neighbour distances and varied between
2.6. Pore counts and analysis The area of stripped basement membrane measured by image analysis (Scion Image PC, Scion, MD) and basement membrane pores were counted on each micrograph. Pore counts were represented as pores/mm2 of basement membrane. The minimum and maximum diameter of each pore was determined using a magnifying eyepiece with a 20-mm scale bar and converted back to original size. Statistical analysis between the minimum and maximum diameter was performed using an independent T test (SPSS, Chicago, IL). Electron micrographs were scanned and saved as TIFF format images. The position of each pore was marked and the image calibrated from the integral scale bar provided by the microscope and scanned with each image. Nearest neighbour distances were calculated from the relative coordinates of each pore, using a specially written Microsoft Visual Basic computer program using simple geometry to find for each pore, the distance from the first, second, and third nearest pores. Nearest neighbour distances for pores within an 8-lm ‘‘guard’’ region around the edge of the image were excluded but these pores could be used to provide nearest neighbour values for other pores. The three nearest neighbour measurements for each pore were saved in a Microsoft Excel spreadsheet for further analysis. 2.7. Simulated model distribution patterns Simulated patterns for pores distribution were used to compare with the real distributions seen in tissue
Fig. 1. Transmission electron micrograph of human bronchial epithelium. (A) The stratified structure of the bronchial epithelium is illustrated with upper ciliated cells (open star), attached to a lower layer of basal cells, themselves bound to the upper basement membrane layer of the lamina densa (arrowhead). (B) A basement membrane pore (arrows) located under the epithelium crosses the full depth of the basal lamina, ending at the intersection with the lamina propria. Scale bar, 10 lm (A), 5 lm (B).
140
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
micrographs, the same numbers of points per sample as those recorded from the tissue samples were used to generate random model sets of data. These were then analysed in the same way as the coordinates from the samples. Frequency histograms of nearest neighbour distances were generated, grouping values from all tissue samples or simulation replicates. Kolmogorov– Smirnov statistics were applied to test for the normality of the nearest neighbour distances. Kruskal–Wallis nonparametric tests were used to detect differences in distribution between the real and the simulated patterns, using the SPSS statistical program v9 (SPSS).
3. Results 3.1. Morphology of the bronchial epithelium Transmission EM of normal bronchial tract illustrated the stratified structure of the epithelium (Fig. 1A). Two layers of epithelial cells were seen, basal epithelial cells resting on the basement membrane and tall suprabasal ciliated and secretory columnar cells. The lamina densa and lamina reticularis of the basement membrane could be seen as the matrix layer below the epithelium. Basement membrane pores could be observed as discrete channels running through lamina densa and lamina reticularis ending at the intersection
Fig. 2. Scanning electron micrographs of human bronchial epithelium following each stage of the epithelial stripping procedure. (A) A mat of cilia can be seen before the stripping treatment commences. (B) A single scraping step detaches a large proportion of ciliated columnar cells, leaving occasional cells (arrowheads) attached to remaining basal cells (arrow). (C) After the first incubation with DTT and resultant scraping step the basement membrane (open star) can be revealed as basal cells are lost. (D) After a further two incubations in EDTA with scraping, all basal cells are removed and a denuded basement membrane with basement membrane pores (arrowheads) can be found. Scale bar, 10 lm.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
with lamina propria (Fig. 1B). Due to their relatively low frequency they are rarely seen in transverse sections and it was necessary to devise a methodology that allowed enumeration of the pores and their distribution seen en face. 3.2. Bronchial epithelial removal To effectively remove the epithelium while maintaining the morphology of the basement membrane, a nonenzymatic epithelial stripping procedure was developed, adapted from the method of Mahida et al. (1997). For bronchial tissue, the original protocol used by Mahida removed neither the basal nor the suprabasal layers of the bronchial epithelium and was indistinguishable from untreated bronchial tissue (Fig. 2A). In order to remove cells, it was necessary to add a scraping step using a 13-mm round glass coverslip at the beginning of the procedure and after each chemical treatment.
141
This effected the complete removal of all epithelial cells from the basement membrane, except in deeper crevices and submucosal glandular tracts. Before treatment, a mat of ciliated epithelial cells could be observed (Fig. 2A). After the initial scraping step prior to incubation in DTT, a high proportion of columnar epithelial cells could be removed (Fig. 2B) leaving a layer of basal cells adhering to the basement membrane. This layer and the remaining columnar cells could be progressively detached following incubation in DTT (Fig. 2C) and two further incubations in EDTA, such that no epithelial cells could be observed after the final EDTA incubation and scraping step (Fig. 2D). The basement membrane thus revealed showed the characteristic ridges and grooves with numerous basement membrane pores. The effectiveness of this stripping procedure combined with the scraping steps allowed us to use one less EDTA incubation than previously described (Howat et al., 2001).
Fig. 3. Confocal microscope images of bronchial basement membrane denuded of epithelium. Three individual scanning planes of basement membrane of unfixed tissue following labeling with a dextran-rhodamine conjugate. At Day 0 (A–C) basement membrane labeling highlights one of three pores (arrow) present in each image plane (separation 2.9 lm) whereas the normal morphological characteristics of the basement membrane (arrowhead) disappear. After 5 days in culture (D), the same three pores are still apparent and appear unchanged. Scale bar, 10 lm.
142
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
In all samples treated by this method the detached epithelial cells remained viable. Both columnar and basal cells were observed to adhere to laminin-coated tissue culture flasks and could be grown for a minimum of 3 days (data not shown). 3.3. Characteristics of bronchial basement membrane pores Labeling of the basement membrane with a rhodamine–dextran conjugate followed by confocal microscopy illustrated the same morphological features as found by scanning EM. Pores were seen as unstained regions within the basement membrane (Fig. 3B). Confocal optical slices at 2.9-lm separation demonstrated that basement membrane pores extended through the lamina densa in contrast to the normal morphological features of the basement membrane (Figs. 3A–C). Repeated scanning of the same area, maintained in tissue culture medium over a 5-day period, showed exactly the same pattern of pores on Day
0 and Day 5 (Figs. 3A and D). Viable cells were seen within the lamina propria close to the pores on Days 1 and 2. Detailed measurement of pore frequency using three scanning EM micrographs from each of five samples revealed a pore density of 737 pores/mm2 (range 342– 1391, 95% confidence interval ¼ 148 pores/mm2 ). Measurement of the long and short axes of each pore in the samples determined that the pores were, on average, oval in shape. Mean long axis 1.9 lm (range 0.7–4.3 lm, 95% confidence interval 0.12 m, n ¼ 144) and mean short axis 1 lm (range 0.4–2.7 m, 95% confidence interval ¼ 0.06 lm, n ¼ 144) with the mean of the long and short axes 1.5 lm (95% confidence interval ¼ 0.08 lm, n ¼ 288). Thus the mean area of each pore could also be calculated using the formula pab (a ¼ short arm radius, b ¼ long arm radius) as 6.1 lm2 (range 1–30 lm2 , 95% confidence interval ¼ 0.82 lm2 , n ¼ 288). Statistical analysis by independent T test shows that the long and short axes measurements were significantly different (P < 0:001).
Micrographs
Random
Clustered
Distributed
Fig. 4. Frequency histograms of nearest neighbour distances from sample micrographs and from simulated pore distributions ‘‘Random,’’ ‘‘Clustered,’’ and ‘‘Distributed.’’ Nearest neighbour (NN) 1 to NN 3 illustrated.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
143
3.4. Distribution pattern of pores
4. Discussion
All results were ranked in order of distance; nearest neighbour (NN) 1, nearest pore; NN 2, 2nd nearest pore; NN 3, 3rd nearest pore. By combining the data from all micrographs and testing normality with the Kolmogorov–Smirnov Z statistic, the NN 2 and NN 3 histograms showed ‘‘bell-shaped’’ curves and showed no significant difference compared to a normal distribution (P ¼ 0:124 and 0.169, respectively). In contrast, NN 1 showed a slight skewing to the right (P ¼ 0:026) (Fig. 4). To clarify the distribution characteristics of the pores in the samples, three model simulations were generated using the same number of pores as found in the micrographs: (1) ‘‘random,’’ where the pores were randomly generated across the area of the image, (2). ‘‘clustered,’’ where the pores were randomly generated in clusters of three, and (3) ‘‘distributed,’’ where an ‘‘exclusion zone’’ was generated, enforcing a minimum spacing between pores (Fig. 4). The clustered pattern was strikingly different from that generated from the micrographs with a cumulative frequency in the NN1 dataset of 312 pores within the first 10 lm of the clustered pattern and 81 pores in the micrographs. The random simulation produced a cumulative frequency most similar to the micrographs (57 pores) while the distributed simulation had the least number of pores within 10 lm (8 pores) (Fig. 4). To assess these differences, the Kruskal–Wallis nonparametric test was applied to the NN1 dataset (Fig. 5). This demonstrated that all simulations were significantly (P < 0:001) different from the real data and provided a mean rank for each simulation. Comparison of the mean ranks demonstrated that the random simulation was most similar to the real data (Fig. 5).
This report demonstrates that basement membrane pores can be revealed with the use of a novel epithelial stripping method and remain patent in vitro after 5 days in culture and that pores are not clustered across the basement membrane surface, but are scattered randomly. Such a random scattering is consistent with pore formation by cellular degradation mechanisms. The bronchial epithelium is a stratified structure of basal and overlying columnar cells. It was noticeable that removal of the columnar epithelium was accomplished only with gentle scraping with a coverslip while more vigorous chemical means had to be used to remove the underlying basal cell layer. This corresponds to the formation of so-called ‘‘creola bodies,’’ where columnar epithelial cells are found dislodged in the lumen of asthmatic patients (Beasley et al., 1989). The relative ease with which columnar cells can be removed, reflecting differences in adhesion may be explained, at least in part, by evidence showing that columnar cells adhere to the basement membrane only through attachments to underlying basal cells (Evans and Plopper, 1988). Basal cells, in contrast, were effectively removed after one incubation with DTT and two incubations with EDTA and three further coverslip scrapings. Since both adherens junctions and desmosomes are stabilised by divalent cations, the removal of Ca2þ and Mg2þ from the cell culture media and use of the Ca2þ -chelating agent EDTA assisted in the weakening of the basal cell attachment. Furthermore, it has recently been demonstrated that wounding to confluent sheets of epithelial cells disrupts established Ca2þ -independent desmosomes, reverting them to Ca2þ dependence (Wallis et al., 2000). Therefore, coverslip scraping may both confer Ca2þ -dependence to otherwise Ca2þ -independent desmosomes and allow greater penetration of DTT and EDTA to the basal layer by disturbing columnar cell confluence. This confirms our results that chemical treatment alone will not remove the bronchial epithelium. Following epithelial removal, the exposed basement membrane pores were confirmed as oval in shape, with a mean long axis measurement of 1.88 lm, short axis of 1.03 lm (mean long and short 1.45 lm), and mean area of 6.62 lm2 . These measurements were consistent with those described previously (Howat et al., 2001), confirming that the changes to the methodology have had no effect on these characteristics of pores. The normal immunological surveillance of the respiratory tract requires passage of antigen-presenting cells from the vascular system through the basement membrane and into the epithelium. At that point, antigenic material can be collected for future presentation to T and B cells in the lymph nodes. Therefore, access to the epithelium from the submucosa is an essential feature
Fig. 5. Comparison of frequency histograms of NN 1 dataset for real data and all simulations. Mean rank of each simulation presented in legend.
144
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
for the respiratory tract immune responses. Previous research showed that basement membrane pores allowed infiltrating cells access to the epithelium (Howat et al., 2001). Using confocal microscopy we were able to demonstrate in the current study that pores were not a transient feature of the basement membrane, but persisted throughout a 5-day culture period, unrestricted and available for cellular transmigration. Although cells were shown to be viable within the submucosa by staining with the vital stain fluorescein diacetate (a compound fluorescent only after processing in live cells) (Jones and Senft, 1985), these cells did not ‘‘repair’’ the pore in the basement membrane. This shows that basement membrane pores can persist in the basement membrane, thus potentially allowing access of multiple cells to/from the epithelium. Despite the data on number, size, and persistence of basement membrane pores, there is little information on the spatial distribution of these pores within the basement membrane itself. This knowledge could potentially indicate the processes through which pores are formed. Therefore, three simulations of pore distribution were developed and were compared to the distribution from sample micrographs. Distributed mimicked a developmental formation of pores, where pores are formed during development and growth. Similar developmental patterns are found during feather bud patterning in chicks where the development of feather buds is dependent on the ratio of activator molecules against inhibitor molecules. For example, only where an activator morphogen is at a greater concentration than the corresponding inhibitor, a feather bud is formed (reviewed in Hogan, 1999). Thus, pore formation could be spatially determined at the same time as basement membrane is formed. In contrast, the random simulation pattern mimicked a process of pore formation by cellular degradation of the extracellular matrix when cells randomly arrive at the basement membrane. Such a method of pore formation would be expected to lead to a random formation of pores. Conversely, coordinate migration of more than one cell into the epithelium from the same area of submucosa (e.g., an adjacent endothelium) could produce an aggregation of pores in one area. Therefore, a clumping simulation was generated. Following comparison of the data from the samples with the simulations it was clear from the mean ranks that the sample data were most similar to the random simulation. This concurs with the evidence that inflammatory cells can degrade basement membrane components using matrix metalloproteinases, particularly MMP9 (gelatinase B) (Kobayashi et al., 1999; Leppert et al., 1995; Okada et al., 1997), and other serine proteases (Heck et al., 1990). It was also clear that there was no evidence of the aggregation of pores in an area, thus inferring that once formed, a pore is used by further
cells. This is consistent with the persistence of pores in our in vitro studies. The random distribution of pores across the basement membrane surface at the observed frequency would facilitate the surveillance function of antigenpresenting dendritic cells, since on average no cell would be greater than 9 lm from a pore (mean distance of NN1 ¼ 18 lm). Consequently, the intraepithelial movement of dendritic cells required to encounter antigen would be minimised. Dendritic cells have been shown to migrate through 3-dimensional collagen lattices following the directional paths in the lattice in preference to the degradation of the lattice (Gunzer et al., 1997). Therefore, although dendritic cells may have the ability to degrade basement membrane components (Kobayashi et al., 1999), we consider that it is more likely that they use basement membrane pores, where they are present, as an easier route of access into the epithelium. In summary, we have demonstrated a method allowing the luminal aspect of the basement membrane to be visualised, without enzymatic digestion, following coverslip scraping of the overlying epithelial cells. The basement membrane pores thus revealed showed an oval morphology, persisted within the basement membrane for up to 5 days, and were scattered randomly across the membrane. This implicates cellular degradation mechanisms in the formation of pores and provides a distribution of pores across the membrane suitable for antigen-presenting cell surveillance, giving easy access to all of the epithelial basement membrane.
Acknowledgments The authors thank the staff of the Pathology Department and Biomedical Imaging Unit at Southampton General Hospital for their help and advice. This work was funded by the Wessex Medical Trust (Hope).
References Beasley, R., Roche, W.R., Roberts, J.A., Holgate, S.T., 1989. Cellular events in the bronchi in mild asthma and after bronchial provocation. Am. Rev. Respir. Dis. 139, 806–817. Evans, M.J., Plopper, C.G., 1988. The role of basal cells in adhesion of columnar epithelium to airway basement membrane. Am. Rev. Respir. Dis. 138, 481–483. Gunzer, M., Kampgen, E., Brocker, E.B., Zanker, K.S., Friedl, P., 1997. Migration of dendritic cells in 3D-collagen lattices. Visualisation of dynamic interactions with the substratum and the distribution of surface structures via a novel confocal reflection imaging technique. Adv. Exp. Med. Biol. 417, 97–103. Heck, L.W., Blackburn, W.D., Irwin, M.H., Abrahamson, D.R., 1990. Degradation of basement membrane laminin by human neutrophil elastase and cathepsin-G. Am. J. Pathol. 136, 1267–1274. Hogan, B.L.M., 1999. Morphogenesis. Cell 96, 225–233.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145 Howat, W.J., Holmes, J.A., Holgate, S.T., Lackie, P.M., 2001. Basement membrane pores in human bronchial epithelium: a conduit for infiltrating cells? Am. J. Pathol. 158, 673–680. Jones, K.H., Senft, J.A., 1985. An improved method to determine cell viability by simultaneous staining with fluorescein diacetate-propidium iodide. J. Histochem. Cytochem. 33, 77–79. Kobayashi, Y., Matsumoto-Kobayashi, M., Kotani, M., Makino, T., 1999. Possible involvement of matrix metalloproteinase-9 in Langerhans cell migration and maturation. J. Immunol. 163, 5989–5993. Leppert, D., Waubant, E., Galardy, R., Bunnett, N.W., Hauser, S.L., 1995. T-cell gelatinases mediate basement membrane transmigration in vitro. J. Immunol. 154, 4379–4389. Mahida, Y.R., Galvin, A.M., Gray, T., Makh, S., McAlindon, M.E., Sewell, H.F., 1997. Migration of human intestinal lamina propria
145
lymphocytes, macrophages and eosinophils following the loss of surface epithelial cells. Clin. Exp. Immunol. 109, 377–386. Okada, S., Kita, H., George, T.J., Gleich, G.J., Leiferman, K.M., 1997. Migration of eosinophils through basement membrane components in vitro: Role of matrix metalloproteinase-9. Am. J. Respir. Cell Mol. Biol. 17, 519–528. Scott, R.A.H., Lauweryns, B., Snead, D.M.J., Haynes, R.J., Mahida, Y., Dua, H.S., 1997. E-cadherin distribution and epithelial basement membrane characteristics of the normal human conjunctiva and cornea. Eye 11, 607–612. Wallis, S., Lloyd, S., Wise, I., Ireland, G., Fleming, T.P., Garrod, D., 2000. The a isoform of protein kinase C is involved in signalling the response of desmosomes to wounding in cultured epithelial cells. Mol. Biol. Cell 11, 1077–1092.
Structural Biology Journal of Structural Biology 139 (2002) 137–145 www.academicpress.com
Distribution of basement membrane pores in bronchus revealed by microscopy following epithelial removal William J. Howat,*,1 Tony Barab as, James A. Holmes, Stephen T. Holgate, and Peter M. Lackie Respiratory, Cell and Molecular Biology, Division of Infection Inflammation and Repair, School of Medicine, University of Southampton, MP888, Level D, Centre Block, Southampton General Hospital, Southampton SO16 6YD, UK Received 13 February 2002, and in revised form 1 July 2002
Abstract The basement membrane of the bronchial epithelium separates the epithelial and mesenchymal compartments. Basement membrane pores allow cells to cross this boundary. We present a method for preparation of samples of human basement membrane allowing us easy visualisation and characterisation of the distribution and persistence of these pores. Columnar epithelial cells were removed from airway samples with gentle scraping with a circular glass coverslip. In contrast, the underlying basal cells required incubation once in dithiothreitol and twice in ethylenediaminetetraacetic acid. Scanning electron microscopy (SEM) at each stage of the epithelial stripping process showed the selective removal of epithelial cells with eventual visualisation of the pores. Using confocal microscopy on blocks of viable tissue, pores were shown to persist in culture for at least 5 days, despite the presence of viable cells in the submucosa. The distribution of pores in tissues determined by SEM was compared to simulations of three distribution patterns (random, clumped, and distributed). The pattern of pores in the samples was consistent with a random distribution. We suggest that basement membrane pores can be generated by the passage of infiltrating cells into the epithelium providing a network suitable for intraepithelial surveillance. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Cell trafficking; Confocal laser scanning microscopy; Nearest neighbour analysis; Scanning electron microscopy; Transmission electron microscopy
1. Introduction The basement membrane of the bronchial tract serves both as the substrate for the overlying epithelium and as the interface between the epithelium and the underlying mesenchyme. The epithelium itself consists of an upper columnar cell layer of ciliated or secretory cells attached to a basal cell layer, with tight junctions at the apical surface of the columnar cells providing an impermeable paracellular barrier. The integrity of the epithelium and its attachment to the basement membrane is facilitated through a number of adhesion mechanisms including
*
Corresponding author. Fax: +44-12-2349-4919. E-mail address: [email protected] (W.J. Howat). 1 Present address: The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK.
cell-cell desmosomal and cell-matrix hemidesmosomal connections. Infiltrating inflammatory cells and antigen-presenting cells can be found in both the submucosa and the epithelium of the airway in the normal respiratory tract and during inflammation. It had been assumed that infiltrating cell access to the bronchial epithelium was gained by matrix degradation. However, the description of pores in the basement membrane of tissues including intestine (Mahida et al., 1997) and conjunctiva (Scott et al., 1997) and under a variety of both normal and disease conditions indicated that this might not be required. It has recently been demonstrated that the bronchial basement membrane also contains such pores (Howat et al., 2001) with a mean diameter of 1.76 lm, of a size suitable for use by infiltrating cells and penetrating the full thickness of the basement membrane. They would therefore allow cell trafficking between the two
1047-8477/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 7 - 8 4 7 7 ( 0 2 ) 0 0 5 8 9 - 0
138
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
compartments in this key mucosal site where infiltrating cells provide essential immune defence against environmental pathogens. It is neither clear how basement membrane pores are formed nor how long they persist. Protease release from migrating inflammatory cells, particularly metalloproteinases, shows that inflammatory cells are capable of degrading basement membrane and basement membrane components (Okada et al., 1997) and could form pores in so doing. In this case we might expect a random distribution of pores and the persistence of the pores will determine the balance between de novo pore formation and reuse (Kobayashi et al., 1999; Leppert et al., 1995). Alternatively, pores could be formed during the development and growth of the airway when a random or a distributed pattern of pores could be formed using a mechanism such as that seen during feather bud formation and patterning in chick skin (reviewed in Hogan, 1999). Moreover, if there was a functional requirement for a heterogeneous distribution of pores, for example, coordinating with local blood vessels or lymphatics, we might expect that pores would be clumped, with areas of high local density. If as we propose, basement membrane pores are a primary route for infiltrating cell access to the epithelium, the distribution and frequency characteristics of the pores within the airway are critical to understanding infiltrating cell function in the respiratory tract. Since the basement membrane pores are relatively infrequent, making them difficult to see by transmission electron microscopy, and are normally covered by epithelium, it is important to be able to adequately remove the epithelium without damage to the basement membrane to reveal the pores. This permits the examination of large areas of the basement membrane by scanning electron microscopy (SEM)2 to establish the distribution of pores as well as their size and shape. In this report we describe and validate the methods to do this and determine the frequency of the basement membrane pores thus revealed. To determine the distribution of pores across the basement membrane, we have compared the pattern of pores found over the airway with a number of simple models of their distribution.
resection and stored in Leibowitz (L15) medium (Life Technologies, Paisley, UK) at 4 °C overnight. A total of 5 samples taken between June 1999 and January 2000 underwent the epithelial stripping procedure. 2.2. Basement membrane preparation and validation of procedure To ensure complete removal of the epithelium and to minimise tissue disturbance, we adapted our previously reported method (Howat et al., 2001). Tissue was washed in phosphate-buffered saline (PBS) to remove excess mucus. Samples for stripping were first incubated in Ca2þ /Mg2þ -free Hanks balanced salt solution (HBSS) (Life Technologies) for 10 min at 37 °C followed by 30 min of incubation at room temperature in 10 mM dithiothreitol (DTT) (Sigma, Poole, UK) in PBS. After a second incubation in HBSS, the bronchial tissue was incubated in 10 mM ethylenediaminetetraacetic acid (EDTA) (Sigma) in PBS for 30 min at 37 °C and repeated once. After each stage, while in HBSS, a 13-mm round glass coverslip was used to gently stroke the airway surface before fixation for electron microscopy. Tissue samples were collected for SEM before stripping and after each chemical treatment, DTT, EDTA (1), and EDTA (2). 2.3. Scanning EM preparation Tissue samples were fixed for SEM in 3% glutaraldehyde in 0.1 M cacodylate/0.23 M sucrose buffer for a minimum of 1 h. Fixed tissue was transferred to 0.1 M cacodylate/0.23 M sucrose buffer and dehydrated through a graded series of alcohols. The tissue underwent critical point drying with CO2 using a Balzers critical point drier (Balzers, Liechtenstein). Dried specimens were glued onto metal stubs with carbon-coated tabs and sputter-coated with gold/palladium. Scanning EM was performed on a Hitachi S800 (Hitachi, Tokyo, Japan) scanning electron microscope. Micrographs of three random areas of flat denuded basement membrane were taken at the same magnification (500Þ from every sample. 2.4. Transmission EM preparation
2. Methods 2.1. Sample collection Bronchial rings were excised from grossly normal areas of human bronchial tissue following lung cancer 2 Abbreviations used: SEM, scanning electron microscopy; PBS, phosphate-buffered saline; DTT, dithiothreitol; EDTA, ethylenediaminetetraacetic acid; HBSS, Hanks balanced salt solution; TIFF, tagged image file format.
Fixed tissue was transferred to 0.1 M cacodylate/ 0.23 M sucrose buffer before 2 h postfixation in 2% osmium tetroxide in cacodylate/sucrose buffer. The tissue was then incubated in 1.5% aqueous uranyl acetate for 30 min at room temperature before dehydration through a series of graded alcohols and embedding in Spurr resin following a standard protocol. A suitable area of epithelium in transverse section was selected and ultrathin sections were cut using a diamond knife on an Ultracut ultramicrotome (Leica, Milton Keynes, UK) set to
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
90 nm. Sections were dried onto copper grids, stained with lead citrate, and examined on a Hitachi H7000 (Hitachi, Tokyo, Japan) transmission electron microscope. 2.5. Confocal microscopy Unfixed, stripped bronchial tissue was glued to a petri dish with cyanoacrylate adhesive to ensure stability. The basement membrane was labeled with a dextran-rhodamine conjugate (10 lg/ml) (Molecular Probes, Leiden, Netherlands) diluted in PBS, for 1 h at room temperature, and washed three times with PBS. The sample was viewed from above the lumenal surface with a Leica TCS 4D confocal laser scanning microscope (Leica) using a 50 water immersion lens (N.A. ¼ 0.75) and a scan thickness of 2.9 lm. Images of the same area were taken over 5 consecutive days. Between scanning, samples were cultured in M199 medium (Life Technologies) supplemented with 2% Ultroser G (Life Technologies). At 1- and 2-day time points, the vital dye fluorescein diacetate (Jones and Senft, 1985) was used to assess viability of cells within the lamina propria.
139
samples. Coordinates for the points (‘‘pores’’) in the patterns were generated using Visual Basic with a randomly seeded random number generation function providing sets of coordinates within the analysis area (Random pattern). ‘‘Clustered’’ patterns of points were produced by generating 3 points randomly distributed up to 18 lm away from an initial random ‘‘seed’’ point. A ‘‘distributed’’ pattern was generated by discarding points that were within 9 lm of any existing point. Since the number of points per unit area affects the nearest neighbour distances and varied between
2.6. Pore counts and analysis The area of stripped basement membrane measured by image analysis (Scion Image PC, Scion, MD) and basement membrane pores were counted on each micrograph. Pore counts were represented as pores/mm2 of basement membrane. The minimum and maximum diameter of each pore was determined using a magnifying eyepiece with a 20-mm scale bar and converted back to original size. Statistical analysis between the minimum and maximum diameter was performed using an independent T test (SPSS, Chicago, IL). Electron micrographs were scanned and saved as TIFF format images. The position of each pore was marked and the image calibrated from the integral scale bar provided by the microscope and scanned with each image. Nearest neighbour distances were calculated from the relative coordinates of each pore, using a specially written Microsoft Visual Basic computer program using simple geometry to find for each pore, the distance from the first, second, and third nearest pores. Nearest neighbour distances for pores within an 8-lm ‘‘guard’’ region around the edge of the image were excluded but these pores could be used to provide nearest neighbour values for other pores. The three nearest neighbour measurements for each pore were saved in a Microsoft Excel spreadsheet for further analysis. 2.7. Simulated model distribution patterns Simulated patterns for pores distribution were used to compare with the real distributions seen in tissue
Fig. 1. Transmission electron micrograph of human bronchial epithelium. (A) The stratified structure of the bronchial epithelium is illustrated with upper ciliated cells (open star), attached to a lower layer of basal cells, themselves bound to the upper basement membrane layer of the lamina densa (arrowhead). (B) A basement membrane pore (arrows) located under the epithelium crosses the full depth of the basal lamina, ending at the intersection with the lamina propria. Scale bar, 10 lm (A), 5 lm (B).
140
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
micrographs, the same numbers of points per sample as those recorded from the tissue samples were used to generate random model sets of data. These were then analysed in the same way as the coordinates from the samples. Frequency histograms of nearest neighbour distances were generated, grouping values from all tissue samples or simulation replicates. Kolmogorov– Smirnov statistics were applied to test for the normality of the nearest neighbour distances. Kruskal–Wallis nonparametric tests were used to detect differences in distribution between the real and the simulated patterns, using the SPSS statistical program v9 (SPSS).
3. Results 3.1. Morphology of the bronchial epithelium Transmission EM of normal bronchial tract illustrated the stratified structure of the epithelium (Fig. 1A). Two layers of epithelial cells were seen, basal epithelial cells resting on the basement membrane and tall suprabasal ciliated and secretory columnar cells. The lamina densa and lamina reticularis of the basement membrane could be seen as the matrix layer below the epithelium. Basement membrane pores could be observed as discrete channels running through lamina densa and lamina reticularis ending at the intersection
Fig. 2. Scanning electron micrographs of human bronchial epithelium following each stage of the epithelial stripping procedure. (A) A mat of cilia can be seen before the stripping treatment commences. (B) A single scraping step detaches a large proportion of ciliated columnar cells, leaving occasional cells (arrowheads) attached to remaining basal cells (arrow). (C) After the first incubation with DTT and resultant scraping step the basement membrane (open star) can be revealed as basal cells are lost. (D) After a further two incubations in EDTA with scraping, all basal cells are removed and a denuded basement membrane with basement membrane pores (arrowheads) can be found. Scale bar, 10 lm.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
with lamina propria (Fig. 1B). Due to their relatively low frequency they are rarely seen in transverse sections and it was necessary to devise a methodology that allowed enumeration of the pores and their distribution seen en face. 3.2. Bronchial epithelial removal To effectively remove the epithelium while maintaining the morphology of the basement membrane, a nonenzymatic epithelial stripping procedure was developed, adapted from the method of Mahida et al. (1997). For bronchial tissue, the original protocol used by Mahida removed neither the basal nor the suprabasal layers of the bronchial epithelium and was indistinguishable from untreated bronchial tissue (Fig. 2A). In order to remove cells, it was necessary to add a scraping step using a 13-mm round glass coverslip at the beginning of the procedure and after each chemical treatment.
141
This effected the complete removal of all epithelial cells from the basement membrane, except in deeper crevices and submucosal glandular tracts. Before treatment, a mat of ciliated epithelial cells could be observed (Fig. 2A). After the initial scraping step prior to incubation in DTT, a high proportion of columnar epithelial cells could be removed (Fig. 2B) leaving a layer of basal cells adhering to the basement membrane. This layer and the remaining columnar cells could be progressively detached following incubation in DTT (Fig. 2C) and two further incubations in EDTA, such that no epithelial cells could be observed after the final EDTA incubation and scraping step (Fig. 2D). The basement membrane thus revealed showed the characteristic ridges and grooves with numerous basement membrane pores. The effectiveness of this stripping procedure combined with the scraping steps allowed us to use one less EDTA incubation than previously described (Howat et al., 2001).
Fig. 3. Confocal microscope images of bronchial basement membrane denuded of epithelium. Three individual scanning planes of basement membrane of unfixed tissue following labeling with a dextran-rhodamine conjugate. At Day 0 (A–C) basement membrane labeling highlights one of three pores (arrow) present in each image plane (separation 2.9 lm) whereas the normal morphological characteristics of the basement membrane (arrowhead) disappear. After 5 days in culture (D), the same three pores are still apparent and appear unchanged. Scale bar, 10 lm.
142
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
In all samples treated by this method the detached epithelial cells remained viable. Both columnar and basal cells were observed to adhere to laminin-coated tissue culture flasks and could be grown for a minimum of 3 days (data not shown). 3.3. Characteristics of bronchial basement membrane pores Labeling of the basement membrane with a rhodamine–dextran conjugate followed by confocal microscopy illustrated the same morphological features as found by scanning EM. Pores were seen as unstained regions within the basement membrane (Fig. 3B). Confocal optical slices at 2.9-lm separation demonstrated that basement membrane pores extended through the lamina densa in contrast to the normal morphological features of the basement membrane (Figs. 3A–C). Repeated scanning of the same area, maintained in tissue culture medium over a 5-day period, showed exactly the same pattern of pores on Day
0 and Day 5 (Figs. 3A and D). Viable cells were seen within the lamina propria close to the pores on Days 1 and 2. Detailed measurement of pore frequency using three scanning EM micrographs from each of five samples revealed a pore density of 737 pores/mm2 (range 342– 1391, 95% confidence interval ¼ 148 pores/mm2 ). Measurement of the long and short axes of each pore in the samples determined that the pores were, on average, oval in shape. Mean long axis 1.9 lm (range 0.7–4.3 lm, 95% confidence interval 0.12 m, n ¼ 144) and mean short axis 1 lm (range 0.4–2.7 m, 95% confidence interval ¼ 0.06 lm, n ¼ 144) with the mean of the long and short axes 1.5 lm (95% confidence interval ¼ 0.08 lm, n ¼ 288). Thus the mean area of each pore could also be calculated using the formula pab (a ¼ short arm radius, b ¼ long arm radius) as 6.1 lm2 (range 1–30 lm2 , 95% confidence interval ¼ 0.82 lm2 , n ¼ 288). Statistical analysis by independent T test shows that the long and short axes measurements were significantly different (P < 0:001).
Micrographs
Random
Clustered
Distributed
Fig. 4. Frequency histograms of nearest neighbour distances from sample micrographs and from simulated pore distributions ‘‘Random,’’ ‘‘Clustered,’’ and ‘‘Distributed.’’ Nearest neighbour (NN) 1 to NN 3 illustrated.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
143
3.4. Distribution pattern of pores
4. Discussion
All results were ranked in order of distance; nearest neighbour (NN) 1, nearest pore; NN 2, 2nd nearest pore; NN 3, 3rd nearest pore. By combining the data from all micrographs and testing normality with the Kolmogorov–Smirnov Z statistic, the NN 2 and NN 3 histograms showed ‘‘bell-shaped’’ curves and showed no significant difference compared to a normal distribution (P ¼ 0:124 and 0.169, respectively). In contrast, NN 1 showed a slight skewing to the right (P ¼ 0:026) (Fig. 4). To clarify the distribution characteristics of the pores in the samples, three model simulations were generated using the same number of pores as found in the micrographs: (1) ‘‘random,’’ where the pores were randomly generated across the area of the image, (2). ‘‘clustered,’’ where the pores were randomly generated in clusters of three, and (3) ‘‘distributed,’’ where an ‘‘exclusion zone’’ was generated, enforcing a minimum spacing between pores (Fig. 4). The clustered pattern was strikingly different from that generated from the micrographs with a cumulative frequency in the NN1 dataset of 312 pores within the first 10 lm of the clustered pattern and 81 pores in the micrographs. The random simulation produced a cumulative frequency most similar to the micrographs (57 pores) while the distributed simulation had the least number of pores within 10 lm (8 pores) (Fig. 4). To assess these differences, the Kruskal–Wallis nonparametric test was applied to the NN1 dataset (Fig. 5). This demonstrated that all simulations were significantly (P < 0:001) different from the real data and provided a mean rank for each simulation. Comparison of the mean ranks demonstrated that the random simulation was most similar to the real data (Fig. 5).
This report demonstrates that basement membrane pores can be revealed with the use of a novel epithelial stripping method and remain patent in vitro after 5 days in culture and that pores are not clustered across the basement membrane surface, but are scattered randomly. Such a random scattering is consistent with pore formation by cellular degradation mechanisms. The bronchial epithelium is a stratified structure of basal and overlying columnar cells. It was noticeable that removal of the columnar epithelium was accomplished only with gentle scraping with a coverslip while more vigorous chemical means had to be used to remove the underlying basal cell layer. This corresponds to the formation of so-called ‘‘creola bodies,’’ where columnar epithelial cells are found dislodged in the lumen of asthmatic patients (Beasley et al., 1989). The relative ease with which columnar cells can be removed, reflecting differences in adhesion may be explained, at least in part, by evidence showing that columnar cells adhere to the basement membrane only through attachments to underlying basal cells (Evans and Plopper, 1988). Basal cells, in contrast, were effectively removed after one incubation with DTT and two incubations with EDTA and three further coverslip scrapings. Since both adherens junctions and desmosomes are stabilised by divalent cations, the removal of Ca2þ and Mg2þ from the cell culture media and use of the Ca2þ -chelating agent EDTA assisted in the weakening of the basal cell attachment. Furthermore, it has recently been demonstrated that wounding to confluent sheets of epithelial cells disrupts established Ca2þ -independent desmosomes, reverting them to Ca2þ dependence (Wallis et al., 2000). Therefore, coverslip scraping may both confer Ca2þ -dependence to otherwise Ca2þ -independent desmosomes and allow greater penetration of DTT and EDTA to the basal layer by disturbing columnar cell confluence. This confirms our results that chemical treatment alone will not remove the bronchial epithelium. Following epithelial removal, the exposed basement membrane pores were confirmed as oval in shape, with a mean long axis measurement of 1.88 lm, short axis of 1.03 lm (mean long and short 1.45 lm), and mean area of 6.62 lm2 . These measurements were consistent with those described previously (Howat et al., 2001), confirming that the changes to the methodology have had no effect on these characteristics of pores. The normal immunological surveillance of the respiratory tract requires passage of antigen-presenting cells from the vascular system through the basement membrane and into the epithelium. At that point, antigenic material can be collected for future presentation to T and B cells in the lymph nodes. Therefore, access to the epithelium from the submucosa is an essential feature
Fig. 5. Comparison of frequency histograms of NN 1 dataset for real data and all simulations. Mean rank of each simulation presented in legend.
144
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145
for the respiratory tract immune responses. Previous research showed that basement membrane pores allowed infiltrating cells access to the epithelium (Howat et al., 2001). Using confocal microscopy we were able to demonstrate in the current study that pores were not a transient feature of the basement membrane, but persisted throughout a 5-day culture period, unrestricted and available for cellular transmigration. Although cells were shown to be viable within the submucosa by staining with the vital stain fluorescein diacetate (a compound fluorescent only after processing in live cells) (Jones and Senft, 1985), these cells did not ‘‘repair’’ the pore in the basement membrane. This shows that basement membrane pores can persist in the basement membrane, thus potentially allowing access of multiple cells to/from the epithelium. Despite the data on number, size, and persistence of basement membrane pores, there is little information on the spatial distribution of these pores within the basement membrane itself. This knowledge could potentially indicate the processes through which pores are formed. Therefore, three simulations of pore distribution were developed and were compared to the distribution from sample micrographs. Distributed mimicked a developmental formation of pores, where pores are formed during development and growth. Similar developmental patterns are found during feather bud patterning in chicks where the development of feather buds is dependent on the ratio of activator molecules against inhibitor molecules. For example, only where an activator morphogen is at a greater concentration than the corresponding inhibitor, a feather bud is formed (reviewed in Hogan, 1999). Thus, pore formation could be spatially determined at the same time as basement membrane is formed. In contrast, the random simulation pattern mimicked a process of pore formation by cellular degradation of the extracellular matrix when cells randomly arrive at the basement membrane. Such a method of pore formation would be expected to lead to a random formation of pores. Conversely, coordinate migration of more than one cell into the epithelium from the same area of submucosa (e.g., an adjacent endothelium) could produce an aggregation of pores in one area. Therefore, a clumping simulation was generated. Following comparison of the data from the samples with the simulations it was clear from the mean ranks that the sample data were most similar to the random simulation. This concurs with the evidence that inflammatory cells can degrade basement membrane components using matrix metalloproteinases, particularly MMP9 (gelatinase B) (Kobayashi et al., 1999; Leppert et al., 1995; Okada et al., 1997), and other serine proteases (Heck et al., 1990). It was also clear that there was no evidence of the aggregation of pores in an area, thus inferring that once formed, a pore is used by further
cells. This is consistent with the persistence of pores in our in vitro studies. The random distribution of pores across the basement membrane surface at the observed frequency would facilitate the surveillance function of antigenpresenting dendritic cells, since on average no cell would be greater than 9 lm from a pore (mean distance of NN1 ¼ 18 lm). Consequently, the intraepithelial movement of dendritic cells required to encounter antigen would be minimised. Dendritic cells have been shown to migrate through 3-dimensional collagen lattices following the directional paths in the lattice in preference to the degradation of the lattice (Gunzer et al., 1997). Therefore, although dendritic cells may have the ability to degrade basement membrane components (Kobayashi et al., 1999), we consider that it is more likely that they use basement membrane pores, where they are present, as an easier route of access into the epithelium. In summary, we have demonstrated a method allowing the luminal aspect of the basement membrane to be visualised, without enzymatic digestion, following coverslip scraping of the overlying epithelial cells. The basement membrane pores thus revealed showed an oval morphology, persisted within the basement membrane for up to 5 days, and were scattered randomly across the membrane. This implicates cellular degradation mechanisms in the formation of pores and provides a distribution of pores across the membrane suitable for antigen-presenting cell surveillance, giving easy access to all of the epithelial basement membrane.
Acknowledgments The authors thank the staff of the Pathology Department and Biomedical Imaging Unit at Southampton General Hospital for their help and advice. This work was funded by the Wessex Medical Trust (Hope).
References Beasley, R., Roche, W.R., Roberts, J.A., Holgate, S.T., 1989. Cellular events in the bronchi in mild asthma and after bronchial provocation. Am. Rev. Respir. Dis. 139, 806–817. Evans, M.J., Plopper, C.G., 1988. The role of basal cells in adhesion of columnar epithelium to airway basement membrane. Am. Rev. Respir. Dis. 138, 481–483. Gunzer, M., Kampgen, E., Brocker, E.B., Zanker, K.S., Friedl, P., 1997. Migration of dendritic cells in 3D-collagen lattices. Visualisation of dynamic interactions with the substratum and the distribution of surface structures via a novel confocal reflection imaging technique. Adv. Exp. Med. Biol. 417, 97–103. Heck, L.W., Blackburn, W.D., Irwin, M.H., Abrahamson, D.R., 1990. Degradation of basement membrane laminin by human neutrophil elastase and cathepsin-G. Am. J. Pathol. 136, 1267–1274. Hogan, B.L.M., 1999. Morphogenesis. Cell 96, 225–233.
W.J. Howat et al. / Journal of Structural Biology 139 (2002) 137–145 Howat, W.J., Holmes, J.A., Holgate, S.T., Lackie, P.M., 2001. Basement membrane pores in human bronchial epithelium: a conduit for infiltrating cells? Am. J. Pathol. 158, 673–680. Jones, K.H., Senft, J.A., 1985. An improved method to determine cell viability by simultaneous staining with fluorescein diacetate-propidium iodide. J. Histochem. Cytochem. 33, 77–79. Kobayashi, Y., Matsumoto-Kobayashi, M., Kotani, M., Makino, T., 1999. Possible involvement of matrix metalloproteinase-9 in Langerhans cell migration and maturation. J. Immunol. 163, 5989–5993. Leppert, D., Waubant, E., Galardy, R., Bunnett, N.W., Hauser, S.L., 1995. T-cell gelatinases mediate basement membrane transmigration in vitro. J. Immunol. 154, 4379–4389. Mahida, Y.R., Galvin, A.M., Gray, T., Makh, S., McAlindon, M.E., Sewell, H.F., 1997. Migration of human intestinal lamina propria
145
lymphocytes, macrophages and eosinophils following the loss of surface epithelial cells. Clin. Exp. Immunol. 109, 377–386. Okada, S., Kita, H., George, T.J., Gleich, G.J., Leiferman, K.M., 1997. Migration of eosinophils through basement membrane components in vitro: Role of matrix metalloproteinase-9. Am. J. Respir. Cell Mol. Biol. 17, 519–528. Scott, R.A.H., Lauweryns, B., Snead, D.M.J., Haynes, R.J., Mahida, Y., Dua, H.S., 1997. E-cadherin distribution and epithelial basement membrane characteristics of the normal human conjunctiva and cornea. Eye 11, 607–612. Wallis, S., Lloyd, S., Wise, I., Ireland, G., Fleming, T.P., Garrod, D., 2000. The a isoform of protein kinase C is involved in signalling the response of desmosomes to wounding in cultured epithelial cells. Mol. Biol. Cell 11, 1077–1092.