Mucin Granules are in Close Contact with Tubular Elements of the Endoplasmic Reticulum

Abstract

Live cell imaging methods were used to characterize goblet cells expressing a MUC5AC domain fused to enhanced green fluorescent protein that labels the granule lumen. Golgi complex and endosome/lysosome elements largely resided in the periphery of the granular mass. On the contrary, a tubular meshwork of endoplasmic reticulum (ER) was in close contact with the mucin granules. This meshwork could be identified in fixed native human bronchial goblet cells labeled with an anti-calreticulin antibody. The potential biological significance of this ER network for mucin secretion is discussed.

G el-forming mucins, the protein family that endows mucus secretions in the gastrointestinal, urogenital, tracheobronchial, and auditory, and visual systems with their characteristic viscoelastic properties, are secreted from specialized cells known as mucous or goblet cells. Much of our current notion on the cellular organization of the mucous/goblet cells derives from electron microcopy studies of chemically fixed cells (for reviews see Neutra et al. 1984; Specian and Oliver 1991). Thus these cells are characterized by the presence of numerous, closely apposed mucin granules that occupy most of the apical cytoplasm, leaving the nucleus constrained in the basal pole of the cell. A layer of microtubule/filament-rich cytoplasm encircles the granule mass, forming a cuplike structure known as theca, whereas actin filaments separate the granule mass from the apical membrane (Neutra et al. 1984; Specian and Oliver 1991). This compact organization would help to extrude large numbers of granules during massive compound exocytosis, which involves fusion of granules between themselves and with the plasma membrane, leading to the formation of the typical “omega” structures viewed with the electron microscope (Neutra et al. 1984). In contrast, cryofixation (Puchelle et al. 1991) and video-enhanced microscopy with live cells (Lethem et al. 1993) also suggest the existence of apocrine- and merocrine-type mechanisms of secretion, respectively. Most important, freeze-substitution studies (e.g., Shimomura et al. 1996) suggested the goblet-like morphology was an aberration (i.e., mucin granules were not tightly packed in apposition to one another and different organelles and cytoskeletal elements filled the intergranular space). We have recently reported studies in live mucous/goblet cells expressing a mucin-enhanced green fluorescent protein (GFP) fusion protein showing mucous cells are rather columnar and their granules are often separated by cytoplasmic space (Perez-Vilar et al. 2005). However, granule apposition and goblet-like morphology was often observed when these cultures were fixed and processed for standard electron microscopy. Thus we speculated that chemical fixation and processing alter mucous cell/mucin granule morphology.

We report here the spatial relationship between mucin granules and other organelles, including the Golgi complex, endosomes/lysosomes and the endoplasmic reticulum (ER) in HT29-SHGFP-MUC5AC/CK mucous cells. These cells, which have been characterized previously (Perez-Vilar et al., 2005), were obtained by transduction of human colon adenocarcinoma HT29-18N2 cells (e.g., Phillips et al. 1995) with a retrovirus expressing a fusion protein comprising a signal peptide, a His-tag, GFP, and the CK domain of MUC5AC at the C termini. Morphological studies demonstrated formation of typical mucous/goblet cells with abundant mucin granules. The granules stored large quantities of the fusion protein (SHGFP-MUC5AC/CK) together with endogenous MUC5AC. Addition of ATP, an established mucin secretagogue, resulted in granule discharge within minutes (Perez-Vilar et al. 2005). These results strongly suggested that these cells differentiated into mucous cells and, accordingly, were a suitable model system to study mucous/goblet cells.

Live HT29-SHGFP-MUC5AC/CK cells grown for 6–14 days in serum-free media (Phillips et al. 1995) were observed on the microscope stage of a Zeiss LSM 510 (University of North Carolina M. Hooker Microscopy Facility; Chapel Hill, NC) at 37C using 488-nm laser excitation for GFP and, when needed, 510 nm for markers tagged/labeled with DsRFP2 or rhodamine (Perez-Vilar et al. 2005). For detecting trans-Golgi elements, HT29-SHGFP-MUC5AC/CK cells were transfected with plasmid pGOLR and fugene-6 (Boehringer; Ingelheim, Germany), selected in G418-containing medium (500 μg/ml) and positive cells isolated by fluorescence activated cell sorting (University of North Carolina Flow Cytometry Facility; Chapel Hill, NC). The expression vector pGOLR was made by replacing the enhanced cyan fluorescent protein-encoding sequence in pGOLGI-ECFP (BD Bioscience; Palo Alto, CA) with a red fluorescent protein (DsRFP2)-encoding DNA sequence obtained from pDsRFP2 (Invitrogen Co.; Carlsbad, CA) by standard PCR. The resulting cells coexpressed SHGFP-MUC5AC/CK and GOLR, which contained the N-terminal, sorting region of human galactosyltransferase and, accordingly, was retained in the trans-Golgi compartment. As shown in Figure 1A, GOLR-containing trans-Golgi elements (magenta signal) were clearly separated from the granules containing SHGFP-MUC5AC/CK (green signal). Although we have not yet succeeded in expressing a cis-Golgi protein marker in live HT29-18N2 cells, we have studied instead the intracellular location of GM130, a well-characterized cis-Golgi protein (Nakamura et al. 1997), in cells fixed with paraformaldehyde, permeabilized with triton X100 and sequentially stained with mouse monoclonal anti-human GM130 (BD Bioscience) followed by rhodamine-conjugated secondary antibody (Jackson Immunochemicals Inc.; West Grove, PA) (Perez-Vilar et al. 2005). GM130 was detected in tubulovesicular structures around the granular mass (Figure 1B), although it must be noticed that the immunofluorescence protocol affected mucin granule morphology, producing a loss of its characteristic round geometry.

Figure 1

Spatial relationship between mucin granules and other organelles in HT29-SHGFP-MUC5AC/CK mucous cells. (A) Projection of 30 consecutive xy-confocal images of a pair of representative live mucous cells stably coexpressing SHGFP-MUC5AC/CK (green signal) and the trans-Golgi fusion protein GOLR (magenta signal). (B) Mucous cells were processed for immunofluorescence with anti-GM130 as described in the text. Shown is a confocal xy-image showing location of SHGFP-MUC5AC/CK (green signal) and GM130 (magenta signal) in a representative mucous cell. (C) Mucous cells were incubated in the presence of WGA-rhodamine for 2 hr and then visualized live with a confocal microscope. Bar = 2 μm.

To identify endosomes/lysosomes, live SHGFP-MUC5AC/CK mucous cells were incubated for up to 4 hr at 37C with wheat germ agglutinin-rhodamine (0.5 μg/ml; Vector Laboratories, Burlingame, CA) in regular media before changing the media to confocal buffer (HBSS supplemented with 20 mM HEPES, pH 7.2, 1% [v/v] FBS, 4.5 g/l glucose, essential and nonessential amino acids). As shown in Figure 1C, the internalized wheat germ agglutinin could not be detected in the SHGFP-MUC5AC/CK-labeled mucin granules, although it was found at the cell surface and in tubulovesicular structures scattered in the cytoplasm of goblet cells. Altogether, these results suggest that, in live mucous cells, Golgi complex elements and endosomes/lysosomes were largely excluded from the intergranular cytoplasm.

To identify the ER in live mucous cells, we generated HT29-SHGFP-MUC5AC/CK cells stably expressing the fluorescent protein DsRFP-ER using the expression plasmid pDsRFP2-ER (Invitrogen Co.) and the transfection agent fugene-6 as directed by the manufacturer (Boehringer Ingelheim). This red fluorescent fusion protein contains two ER retention signals, including the calreticulin N-terminal region and a C-terminal KDEL sequence. As shown in Figure 2A, DsRFP-ER was clearly distributed within the cytoplasm, but absent from the mucin granules, as expected for an ER luminal protein. At high magnification, however, a network of thin, DsRFP-ER-containing tubules encircling the granules was evident (Figures 2B and 2C). That only midsections of the cells were imaged for these studies indicated that these ER tubules were different from the cortical ER that surrounds the granular mass. The thin nature of the ER meshwork, however, made it difficult to prevent photobleaching of DsRFP-ER2 during imaging and, ultimately, its visualization when compared with the cortical ER. Independent evidence for the granule-associated ER network was obtained by fluorescence recovery after photobleaching analysis of live HT29-18N2-SHGFP-MUC5AC/CK cells (Perez-Vilar et al. 2005). Selected regions of cells maintained at 37C at the microscope stage were irreversibly photobleached at high laser power (80% laser power/100% transmittance) for 0.07–0.1 sec using the 488-nm line of the confocal microscope. Recovery of the fluorescent signal inside the bleaching region was imaged at 80% laser power/0.5% transmittance. Images were analyzed with Zeiss LSM software(Carl Zeiss; Verkauf, Germany). SHGFP-MUC5AC/CK diffuses throughout the whole ER with an apparent diffusion constant of ≃1.5 μm²/sec (Perez-Vilar et al. 2005). In contrast, mucin granules are independent structures and, accordingly, bleaching of an intact granule does not affect fluorescence of the surrounding granules. We took advantage of this difference to visualize the ER in the region occupied by granules.

Figure 2

The mucin granule-associated endoplasmic reticulum (ER) tubular network in live HT29-SHGFP-MUC5AC/CK mucous cells. Low magnification xy-image of HT28-SHGFP-MUC5AC/CK cells coexpressing SHGFP-MUC5AC/CK (green signal) and DsRFP2-ER (magenta signal) (A). A three-dimensional projection of a portion of one of these cells comprising several granules is shown in (B) (signal associated with both proteins) and (C) (only DsRFP2-ER-associated signal). The three-dimensional projection was made from seven consecutive xy-images separated by 0.6 μm. (D) The area occupied by SHGFP-MUC5AC/CK-containing mucin granules in a representative live mucous cell (saturated white signal; a low magnification of the same cell is shown in the upper left corner insert) was photobleached (E) and the cells imaged over time with the confocal microscope. Notice that after 5 min (F), a network of endoplasmic reticulum elements was observed. Consecutive xy-images separated by 0.5 μm of a representative mucous cell 5 min after photobleaching the granular mass (G-I). Bars: A = 1 μm; B,C = 2 μm; D (inset) = 5 μm; D-I = 2 μm.

Thus midoptical sections of mucous cells were imaged in areas containing ER and mucin granules, with both organelles extending above and below the focal plane. A focal plane containing the nuclear envelope was also selected to better identify the ER after the bleaching. Because the intraluminal ER SHGFP-MUC5AC/CK signal was weaker than the granule signal, it was necessary to saturate the fluorescent signal in the granules to visualize the ER (Figure 2D). The entire granular mass was bleached at high laser power (Figure 2E), and the cell scanned every 30 sec afterwards. Within seconds after the bleaching pulse, a fluorescent network of interconnected tubules that was continuous with the nuclear envelope was visible in the region occupied by the bleached granules (Figure 2F). These results indicated that unbleached SHGFP-MUC5AC/CK from neighboring areas of the ER had mixed with the photobleached molecules within the ER tubular meshwork, as expected for connected regions. Figures 2G-2I show three consecutive, 0.55-μm apart, xy images of a live mucous cell 5 min after photobleaching the granular mass, showing the tubular nature of the ER associated with the granules.

To investigate whether the mucin-granule-associated ER network was present in native goblet/mucous cells, freshly isolated human bronchial epithelia from chronic obstructive pulmonary disease subjects, provided by the University of North Carolina Cystic Fibrosis Center under the auspices of protocols approved by the Institutional Committee on the Protection of the Rights of Human Subjects, were fixed with 4% (w/v) paraformaldehyde in PBS, pH 7.4, and embedded in paraffin. Sections (≃7-μm thick) were double labeled with a rabbit polyclonal anti-calreticulin antibody (Affinity Bioreagents; Golden, CO) and a mouse anti-MUC5AC monoclonal antibody (Lab Vision Co.; Fremont, CA) following procedures described earlier (Ribeiro et al. 2005). Images were captured on the DeltaVision RT Imaging System (Applied Precision, LLC; Seattle, WA) as a z-series with 0.15-μm focal steps through the entire specimen with a 64 X 1.4 numerical aperture objective and standard FITC and rhodamine filters. The entire dataset was deconvolved for each wavelength using DeltaVision's contrained iterative three-dimensional deconconvolution algorithm to improve image contrast. MetaMorph software (Universal Imaging Co.; Downington, PA) was used to assemble the figures using a single representative plane from the deconvolved z-series. Using this system, the mucin granule-associated ER tubular meshwork in superficial goblet cells was clearly revealed (Figure 3). Similarly to cultured goblet cells, the calreticulin-containing meshwork (Figures 3B, 3D, 3E and 3G) was visualized as thin tubules that appeared to encircle the mucin granules stained with anti-MUC5AC antibodies (Figures 3C, 3D, 3F, and 3G).

Figure 3

The mucin granule-associated endoplasmic reticulum (ER) tubular network in fixed human bronchial goblet cells. Representative histological section of human bronchial tissue, immunostained with anti-calreticulin (green color) and anti-MUC5AC (magenta color), showing numerous goblet cells (A). The goblet cell in the top right corner is also shown below. Distribution of calreticulin (B,D,E,G) and MUC5AC (C,D,F,G). (E-G) A higher magnification of a region of the goblet cell marked with arrowheads, which indicate the position of some of the ER tubules in close apposition with the granules. Bar: A = 10 μm; B-G = 3 μm.

The close association of an ER network with mucin granules in goblet cells from native human bronchial epithelia is in agreement with the studies with live cultured mucous cells. Because HT29 cells were originally derived from a human colon adenocarcinoma, these results suggest that the mucin granule-associated ER meshwork is a general characteristic of mucous/goblet cells irrespective of the tissue these cells were derived from. In addition, the existence of ER structures closely associated with secretory granules in pancreatic acinar cells (Gerasimenko et al. 2002) suggests that such a morphological feature is not exclusive of mucous/goblet cells.

Although the function of this ER meshwork is not known, its close association with the mucin granules might indicate a role for this specialized ER compartment in granule homeostasis or exocytosis. The requirement for locally generated Ca²⁺ signals during granule-regulated secretion has been recently suggested by two independent studies. Thus Nguyen et al (1998) provided evidence that the mucin granule lumen could supply the Ca²⁺ required for its own exocytosis, assuming the existence of an inositol 1,4,5-triphosphate (InsP₃)-activated Ca²⁺ channel in the granular membrane. Studies in rat SPOC cells by Rossi et al. (2004) also suggested the existence of an InsP₃-independent release of Ca²⁺ from granules, or in close proximity, the ER. The close association between ER meshwork and granules found in the present study supports the view that the primary source of intracellular Ca²⁺ mobilization on mucin secretagogue signaling is the ER meshwork surrounding the granules.

In conclusion, our studies have revealed a close spatial association between mucin granules and thin extensions of the ER that appear to form a tubular meshwork surrounding the granules in live cultured mucous cells and fixed native human bronchial goblet cells. Future studies are warranted to address the functional role of this specialized ER network on Ca²⁺-dependent mucin secretion.