Purification and Characterization of Mouse Lacrimal Gland Epithelial Cells and Reconstruction of an Acinarlike Structure in Three-Dimensional Culture

Abstract

purpose. To clarify which factors are involved in postnatal lacrimal gland development, the expressions of several growth factors and their receptors were analyzed in purified lacrimal gland epithelial and mesenchymal cells. Reconstruction of acinarlike structures in three-dimensional culture conditions from this epithelial cell population was attempted.

methods. Lacrimal gland epithelial and mesenchymal cells were isolated from newborn mice and purified using nylon mesh and collagenase. By real-time reverse transcription–polymerase chain reaction, immunocytochemistry, and Western blotting, the expressions of several growth factors and their receptors were analyzed. Responses of epithelial cells to the growth factors were analyzed by the addition of these factors to the culture medium.

results. Fibroblast growth factor (FGF)10 and hepatocyte growth factor (HGF) were more intensely expressed in mesenchymal cells than in epithelial cells, and their receptors (FGFR2IIIb and c-Met, respectively) were less intensely expressed, whereas the expression of epidermal growth factor (EGF) and its receptor showed no significant difference between cell types. In the monolayer culture, cell viability was activated by the addition of EGF or HGF to epithelial cells, but no response was observed when FGF10 was added. The epithelial cells formed clusters with lumina when cultured on basement membrane matrix. These clusters contained secretion granules and showed positive immunostaining for aquaporin-5.

conclusions. EGF and HGF were considered to act in an autocrine/paracrine manner in and around the postnatal lacrimal gland, whereas epithelial cells did not respond to FGF10. It was suggested that certain extracellular matrix conditions accommodate these epithelial cells to reconstruct functional acinarlike structures.

Patients with impaired lacrimal gland secretion, such as in Sjögren and other dry eye syndromes, ¹ experience eye irritation and sandy or gritty sensations. Those symptoms, if left untreated, could result in corneal scarring or ulceration and eventual vision loss. Dry eye conditions are typically treated by the application of artificial tears, but this only gives temporary relief, and corneal damage often does not ameliorate. Functions and development of lacrimal glands are sustained by the interactions between epithelial and mesenchymal cells. Further characterization of each kind of cell in vitro will contribute to better understanding of the normal and pathologic states of lacrimal glands. Although several reports have been published on cultured lacrimal gland cells in rabbits, humans, and rats, the purity of epithelial and mesenchymal cells was insufficiently calculated ^{2 3} or not mentioned. ^{4 5 6 7 8 9 10 11 12 13 14} For more precise analysis of lacrimal gland cells, it is necessary to use a highly purified cell population. 

Development of the mouse lacrimal gland begins with an outgrowth of epithelial budding from conjunctival epithelium in the temporal extremity near the lateral canthus at embryonic day (E) 13.5. Subsequently, epithelia extend and acquire the mature branching structure in intraorbital and exorbital lacrimal glands by birth. Although epithelial-mesenchymal interaction is considered to play important roles in those developmental processes, the factors involved in the postnatal period of the lacrimal gland development are still unclear. Several growth factors, such as fibroblast growth factors (FGFs), epidermal growth factor (EGF), and hepatocyte growth factor (HGF), are reported to be involved in the epithelial-mesenchymal interaction during the development of some exocrine glands. 

The FGF family is composed of 23 different molecules divided into seven subfamilies. Each subfamily binds to its own FGF receptor. ¹⁵ Several reports ^{16 17 18} have stated that during the early development of mouse lacrimal glands, FGF10 and FGF7, which bind to the same receptor (FGFR2b), ¹⁵ are secreted by mesenchymal cells and induce initial branching through FGFR2 signaling. Lens-specific expression of FGF10 or FGF7 in transgenic mice induced ectopic glandlike structures within the cornea. ^{16 19} The lacrimal gland is absent or hypoplastic in FGF10 mutant mice ^{16 20} ; however, mice with the null allele of FGF7 develop normal lacrimal glands. ¹⁹  

Therefore, FGF10 is thought to be essential in early lacrimal gland development. Although FGF2 is reported to be expressed in a three-dimensional culture of rabbit lacrimal gland cells, ¹¹ the roles of FGFs in postnatal lacrimal glands are unknown. In mouse salivary gland cells cultured on basement membrane matrix (Matrigel; BD Biosciences, Franklin Lakes, NJ), it has been reported that FGF influenced the morphology of epithelia. ²¹ FGF7 and FGF10 promote branching and duct elongation of the cultured mouse salivary gland in a dose-dependent manner. ^{22 23} Moreover, the FGF10-FGFR2 signal pathway promotes branching formation in lung development. ^{24 25}  

EGF was first found in the salivary gland in 1962 ²⁶ and is known to promote cell proliferation in various kinds of cells. In rabbit lacrimal glands, EGF was strongly expressed in the early stage of three-dimensional culture, ¹¹ and the addition of EGF stimulated acinar cell proliferation. ¹⁰ Although EGF is reported to promote branching and lobule formation in the developmental process of mouse salivary glands, ^{21 22} there have been no reports about its roles in the development of lacrimal glands. 

HGF was found to be a growth-stimulating factor for hepatocytes ²⁷ and is also called a scatter factor. HGF promoted branching in a three-dimensional culture of salivary glands. ^{28 29} It has been reported that salivary epithelial cells express c-Met, an HGF receptor, and that salivary mesenchymal cells express HGF. ²⁹ In humans, HGF is synthesized in lacrimal gland mesenchymal cells and is secreted as a component of normal tears. ³⁰ c-Met is localized in the cornea. Therefore, HGF is considered to modulate corneal epithelial cell proliferation, motility, and differentiation. 

Three-dimensional cultures of purified lacrimal gland epithelial cells have been studied by several groups. ^{2 4 5 6 9 10 11} Those studies showed that epithelial cells reconstructed acinarlike structures on basement membrane matrix (Matrigel; BD Biosciences) or collagen gel, suggesting that the formation of an acinarlike structure was possible even without mesenchymal cells. However, in those studies, the influence of contaminating mesenchymal cells was not excluded. 

In this study, we successfully obtained lacrimal gland epithelial and mesenchymal cells with a purity of 99.5%, using the differential adhesiveness to culture vessels and adequate selective medium for each kind of cells. Our unique method facilitated the examination of functioning molecules in each kind of cell. We examined the expressions of FGF10, EGF, HGF, and their receptors and clarified that EGF and HGF upregulated epithelial viability in an autocrine and a paracrine manner, respectively. We also showed that after efficient expansion in a monolayer culture, highly purified lacrimal gland epithelial cells reconstructed a functional acinarlike structure on basement membrane matrix (Matrigel; BD Biosciences), in mimicry of in vivo lacrimal gland morphology and function. 

Materials and Methods

Cell Culture

C57BL/6 mice 1 to 4 days old were killed by inhalation of diethylether, and the exorbital lacrimal glands were dissected. All experiments were performed in accordance with the National Defense Medical College Animal Care and Use Committee and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. After connective tissue was removed, the glands were decapsulated with fine forceps and digested by 1% collagenase for 3 minutes at 37°C with vigorous shaking. Digested cells were washed and suspended in Dulbecco modified Eagle medium (DMEM; Invitrogen, Carlsbad, CA)/nutrient mixture F-12 Ham supplemented with10% newborn calf serum (NCS), penicillin, streptomycin, and fungizone (DF-10% NCS). Cells were filtered through a 40-μm mesh nylon cell strainer (BD Biosciences, Franklin Lakes, NJ). Cells that passed through the strainer were incubated in DF-10% NCS for 90 minutes. After extensive washing to remove nonadherent cells, the attached cells were maintained in DF-10% NCS and were considered to be mesenchymal cells. Cells retained on the strainer were incubated in DF-10% NCS for 90 minutes. Floating cells were collected and digested with 0.05% trypsin and 0.02% EDTA for 5 minutes at 37°C. The cells were plated on type IV collagen-coated culture vessels and maintained in epidermal keratinocyte medium (CnT-07; CELLnTEC Advanced Cell Systems, Bern, Switzerland). We regarded these cells as lacrimal gland epithelial cells. Culture media were exchanged twice a week. 

Retinas from the same mice were also cultured as a positive control in immunocytochemistry for neurofilaments and glial fibrillary acidic protein (GFAP). Eyes were enucleated and cut into halves at the equator. Retinas were isolated with fine forceps, triturated by pipetting, plated, and cultured in DMEM-10% NCS. As a positive immunocytochemical control for α-smooth muscle actin (SMA) and von Willebrand factor (VWF), aortas were dissected from the same mice, minced, trypsinized, and cultured in DF-10% NCS. 

Immunocytochemistry

Cultures were fixed with absolute methanol for 20 minutes at −20°C or with phosphate-buffered 4% paraformaldehyde for 10 minutes at room temperature. Primary antibodies were reacted for 60 minutes at room temperature or overnight at 4°C. Primary antibodies used in this study were anti-pancytokeratin (pan-CK) antibody (422061; Nichirei, Tokyo, Japan), anti-GFAP antibody (N1506; DAKO, Glostrup, Denmark), anti-VWF antibody (A0082; DAKO), and anti–epidermal growth factor receptor (EGFR) antibody (ab2430; Abcam, Cambridge Science Park, UK) from rabbit; anti-vimentin antibody (sc7557; Santa Cruz Biotechnology, Santa Cruz, CA) and anti–hepatocyte growth factor receptor (c-Met) antibody (AF527; R&D Systems, Minneapolis, MN) from goat; anti–smooth muscle actin (SMA) antibody (asm-1; Labvision, Fremont, CA), anti-neurofilament antibody (01-20082; American Research Products, Belmont, MA), and anti-EGF antibody (MON 8001; MONOSAN, Uden, The Netherlands) of mouse monoclonal IgG, anti–cytokeratin 8 (CK-8) antibody (35βH11; DAKO) of monoclonal mouse IgM, and anti-CD34 antibody (Abcam) and anti–ZO-1 antibody (MAB1520; Chemicon, Temecula, CA) from rat. 

Texas red- or fluorescein-conjugated secondary antibody (Vector Laboratories, Burlingame, CA) against the respective immunoglobulin of the primary antibody was reacted for 30 minutes at room temperature. Nuclear counterstaining was performed with 4′,6-diamidino-2-phenylindole (DAPI). 

On some culture specimens, we performed double immunostaining for CK-8/vimentin, pan-CK/ZO-1, pan-CK/SMA, and SMA/vimentin. Primary antibodies were reacted together, and secondary antibodies were reacted separately in an appropriate order. Primary antibodies were omitted for the negative control. Samples were observed and photographed with a universal microscope (BioZero; Keyence, Osaka, Japan). 

Western Blot Analysis

The culture was washed with Dulbecco phosphate-buffered saline (PBS) without calcium and magnesium and dissolved in loading buffer for Laemmli’s sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Protein content was assayed by DC protein assay (Bio-Rad Laboratories, Hercules, CA). The cell lysate containing 3 μg protein was subjected to SDS-PAGE. Proteins were transferred to a membrane (Immobilon-P; Millipore, Bedford, MA). Blots were immunoreacted with anti–c-Met (1:1000; rabbit polyclonal SP260; Santa Cruz Biotechnology) antibody and horseradish peroxidase-conjugated goat anti–rabbit immunoglobulin (554021; 1:2500; BD Biosciences), and the protein bands were visualized by a chemiluminescence detection system (Super Signal West Pico; Pierce Biotechnology, Rockford, IL). 

Quantification of Gene Expression of Growth Factors and Their Receptors by Real-Time Reverse Transcription–Polymerase Chain Reaction

Total RNA was extracted from cultured epithelial and mesenchymal cells (RNeasy kit; Qiagen, Oosterhout, Netherlands) according to the manufacturer’s protocol. RNA was reverse transcribed to cDNA (SuperScript First-Strand Synthesis System; Invitrogen) for reverse transcription–polymerase chain reaction (RT-PCR) with the use of random hexamers as primers. Probes (TaqMan; Applied Biosystems, Foster City, CA) for the detection of gene expression for FGF10, FGFR2IIIb, FGFR2IIIc, HGF, c-Met, EGF, EGFR, and 18S ribosomal RNA as an endogenous control and other reagents required for real-time RT-PCR were purchased from Applied Biosystems. Real-time RT-PCR was performed using probes (TaqMan; Applied Biosystems) and cDNA templates from the epithelial and mesenchymal cultures with a PCR system (7900HT Fast Real-Time; Applied Biosystems). Data were analyzed by relative quantification study software in the automatic mode. Statistical analysis was performed with the Wilcoxon t-test. 

Effects of Growth Factors

Epithelial culture on day 3 was incubated in 50 ng/mL human recombinant epidermal growth factor (EGF; R&D Systems), 10 ng/mL mouse recombinant fibroblast growth factor 10 (FGF10; Wako, Osaka, Japan), or 50 ng/mL human recombinant hepatocyte growth factor (HGF; R&D Systems) in antibiotics/antimycotics containing defined keratinocyte serum-free medium without supplements (KSFM; 10744–019; Invitrogen) and incubated for 4 days. The culture was labeled with 1 μM 5′-bromo-2′-deoxyuridine (BrdU) for 18 hours before assay. The activity of lactate dehydrogenase (LDH) released from dead cells into the culture medium was measured with a cytotoxicity detection kit (Roche Applied Science, Mannheim, Germany) to evaluate cell death. The culture was incubated in new KSFM containing a reagent (XTT Cell Proliferation Kit II; Roche Applied Science) that measured mitochondrial dehydrogenase activity of metabolically active cells until the proper color was developed in the medium. After the reagent assay, BrdU incorporation in the DNA-synthesizing cells was measured with a labeling and detection kit (BrdU Labeling and Detection Kit III; Roche Applied Science). In other experiments, we added FGF10 and HGF together in the culture medium and performed LDH and XTT assays. All kits were used according to the manufacturers’ instructions. We evaluated the effect of an EGFR inhibitor 4-(3-chloroanilino)-6,7-dimethoxyquinazoline (AG1478; Calbiochem, Darmstadt, Germany) on cultured epithelial cells. AG1478 was added to the culture alone or together with EGF or HGF at a concentration of 2.5 μM. After 4 days, cell viability was measured by XTT assay. Data were statistically analyzed by nonrepeated measures ANOVA with Bonferroni correction, Student-Newman-Keuls test, and unpaired t-test. 

For a positive control for cell culture experiments with FGF10, an organ culture of embryonic lacrimal gland was performed as follows. A cross-linked gelatin hydrogel that permits the controlled release of growth factors (MedGel; MedGel, Kyoto, Japan) was cut into 4-mm² pieces and soaked in PBS for control or in 10 μg/mL FGF10 solution at 4°C overnight. Explant culture of E14.5 lacrimal glands from the P6 5.0 LacZ reporter line of transgenic mice was performed. ¹⁷ In this transgenic mouse, β-galactosidase is expressed in the lens, conjunctival, and lacrimal gland epithelia. Explants containing growing lacrimal glands were excised and placed on membrane filters of 0.8-μm pore size (Millipore) supported by stainless steel grids. PBS- or FGF10-soaked gelatin hydrogel (MedGel; MedGel) was put on the explants, which were then incubated in DMEM-10% NCS supplemented with nonessential amino acids (ICN Biomedicals, Aurora, OH) and antibiotics/antimycotics. The organ culture was maintained for 4 days, then fixed and stained with X-gal ¹⁷ to visualize the epithelium. 

Three-Dimensional Culture

For three-dimensional culture, epithelial cells grown to subconfluence in a monolayer culture were trypsinized, plated onto gelated growth factor-reduced basement membrane matrix (Matrigel; BD Biosciences), and maintained in medium (CnT-07; CELLnTEC Advanced Cell System). 

After 2 weeks of culture, epithelial cell masses that formed in the gel were fixed by phosphate-buffered 4% paraformaldehyde at 4°C for 10 minutes. The samples were dehydrated in a graded ethanol series and embedded in paraffin. Sections were made for histologic and immunohistochemical observations. Some sections were stained with hematoxylin-eosin. 

For immunohistochemistry, antigens were retrieved by the treatment of sections in 1% trypsin in PBS for 30 minutes at room temperature. Anti–aquaporin-5 (AQP5) antibody (sc-9891; Santa Cruz Biotechnology) was reacted overnight at 4°C, and fluorescein-conjugated secondary antibody (Vector Laboratories) was reacted. Sections were counterstained with DAPI. 

Some epithelial cell masses that formed in the three-dimensional gel culture were observed by a transmission electron microscope (model 1010; JEOL, Tokyo, Japan). The sample was fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.4; Karnovsky fixative) for 30 minutes at 4°C and embedded in embedding resin (Epon 812). Ultrathin sections were cut and stained as previously described. ³¹  

Results

Purification of Lacrimal Gland Cells

In each culture condition, mesenchymal and epithelial cells proliferated well. Epithelial cells formed a confluent monolayer of polygonal cells within 1 week (Fig. 1A) . We assessed the purity of the cultured epithelial and mesenchymal cells by immunocytochemistry. Double immunostaining for pan-CK and ZO-1 revealed a sharp outline of ZO-1 around pan-CK–positive cytoplasm, showing the presence of a tight junction (Fig. 1B) . Double immunostaining for vimentin and CK-8, which is reported to be strongly expressed in rat lacrimal glands, ⁹ showed that almost all cells were CK-8 positive and that few were vimentin positive (Fig. 1C) . 

Some cells were double immunopositive cells for SMA and pan-CK (Fig. 1D) , indicating the presence of myoepithelial cells that might have proliferated from the original glandular epithelial cells. For further characterization of possibly contaminating cells in the epithelial culture, we double immunostained for vimentin and SMA. Cells that were SMA negative and vimentin positive were mesenchymal cells (fibroblasts), cells positive for both SMA and vimentin were myofibroblasts, and cells that were SMA positive but vimentin negative were considered myoepithelial cells or vascular smooth muscle cells (Fig. 1E) . We also examined the presence of cells from vascular, neural, and bone marrow origins. VWF, NF, GFAP (Figs. 1F 1G 1H) , or CD34 (data not shown) were not expressed by the cultured cells but did show positive staining in the control aortic and retinal cells (Figs. 1F 1G 1H , insets). 

We counted the numbers of nuclei and vimentin-positive cells on microscope images and found that 0.5% of cells in the epithelial culture were vimentin positive. We concluded that the purity of epithelial cells in the culture was 99.5%; the culture was available for further study as a purified epithelial culture. 

We also assessed the purity of the mesenchymal culture. Phase-contrast images and immunocytochemistry for vimentin and pan-CK are shown (Figs. 1I 1J 1K) . All cells were immunopositive for vimentin but negative for pan-CK. No NF- or GFAP-positive cells were detected (data not shown). The purity of mesenchymal cells was nearly 100%. 

Expression of Growth Factors and Their Receptors

Gene expression of growth factors FGF10, HGF, and EGF and their respective receptors FGFR, c-Met, and EGFR were examined by real-time RT-PCR. Protein expression of c-Met, EGF, and EGFR was examined with Western blotting and immunocytochemistry. 

Real-Time RT-PCR Assay for FGF10 and FGFR

Gene expression of FGF10 was remarkable in the mesenchymal culture but less apparent in the epithelial culture. Mesenchymal cells expressed high levels of FGFR-2IIIc and nearly undetectable levels of FGFR2IIIb. Conversely, epithelial cells expressed higher levels of FGFR2IIIb than FGFR2IIIc. Because FGFR2IIIb has high affinity for the FGF7 family, including FGF10, FGF10 is considered to work by paracrine action (Fig. 2A) . 

Real-Time RT-PCR Assay for HGF and c-Met

Mesenchymal cells produced higher levels of HGF than epithelial cells. Conversely, c-Met expression was higher in epithelial cells than in mesenchymal cells (Fig. 2B) . 

Western Blot of c-Met

c-Met was detected only in epithelial cells by Western blot (Fig. 2C) . c-Met expression in mesenchymal cells was below detectable levels. The result was consistent with gene expression analysis by real-time RT-PCR in which epithelial cells expressed higher levels of c-Met than mesenchymal cells. 

Immunocytochemistry for c-Met

Cultured epithelial cells were immunostained for anti–c-Met. Most of the cells were positively stained in the cytoplasm and showed strong linear staining along the cell boundary (Fig. 2D) . 

Results of gene expression, Western blot analysis, and immunocytochemistry were consistent. Mesenchymal cells synthesized HGF, and epithelial cells expressed its receptor c-Met. We concluded that HGF secreted from mesenchymal cells binds to its receptor on the epithelial cell membrane to work in a paracrine manner. 

Real-Time RT-PCR Assay for EGF and EGFR

In contrast to those for FGF10, HGF, and their receptors, the expression levels of EGF and EGFR did not show obvious differences between epithelial and mesenchymal cells (Fig. 2E) . 

Immunocytochemistry for EGF and EGFR

Immunocytochemistry for EGF revealed uniformly stained signals in the cytoplasm. Although the stain was weak, there was a distinct difference from the negative control (Fig. 2F) . Epithelial cells immunostained for EGFR (Fig. 2G)showed spotty signals on the diffusely positive cell layer. 

Cellular Reactivity to the Growth Factors

The reactivity of epithelial cells to growth factors was tested on the basis of cell viability, cell death, and DNA synthesis (Fig. 3A) . Cultures treated with EGF and HGF had significantly increased cell viability and DNA synthesis but reduced cell death, indicating a reduced ratio of cell death in the cell population. The increase in DNA synthesis in EGF- and HGF-treated cultures was much higher than that of cell viability, suggesting the enhancement of DNA synthesis in these cultures. EGF and HGF had two different effects on the prevention of cell death and the enhancement of cell growth, both of which contributed to increased numbers of viable cells. On the other hand, the addition of FGF10 did not show much effect on cell viability, cell death, or DNA synthesis. 

Although FGFR2IIIb, the FGF10 receptor, is strongly expressed in epithelial cells, FGF10 did not stimulate cell proliferation. We examined whether another growth factor might enhance its action in concert with FGF10. We added HGF at indicated concentrations with or without 10 ng/mL FGF10 and measured cell viability (Fig. 3B , left). HGF enhanced cell viability in a dose-dependent manner, but the addition of FGF10 did not influence cell viability. Then we raised the concentration of FGF10 to 100 ng/mL with no HGF or with 5 ng/mL HGF (Fig. 3B , right). The addition of HGF increased cell viability at each concentration of FGF10, but no significant change of cell viability was observed by the addition of FGF10. FGF10 did not reduce cell death when added alone or together with HGF (data not shown). We concluded that FGF10 did not modify the growth-promoting activities of HGF in this culture system. 

Because FGF10 stimulates the early development of lacrimal glands, ¹⁷ we excised E14.5 lacrimal glands and organ-cultured them with FGF10-soaked gelatin hydrogel. The lacrimal gland underwent no branching in this stage. Staining with X-gal revealed that explants cultured with FGF10–gelatin hydrogel yielded more branches than did those with control gelatin hydrogel (Fig. 3C) . This result suggested that the failure of growth stimulation by FGF10 was not caused by inactivation. 

To clarify whether EGF works in an autocrine or a paracrine manner, epithelial cultures were incubated with AG1478 alone or with EGF or HGF, and cell viability was measured (Fig. 3D) . AG1478 is a highly potent and specific inhibitor of EGFR tyrosine kinase. ³² AG1478 added alone inhibited cell viability. The addition of EGF with AG1478 did not augment cell viability at all, indicating that AG1478 completely inhibited the effects of EGF. On the other hand, the enhancement of cell viability by HGF was not eliminated by the addition of AG1478. The increase in cell viability in culture with HGF and AG1478 together relative to that with AG1478 alone was comparable to the increase in the culture with HGF alone relative to that in the control. We concluded that the decrease in cell viability after the addition of AG1478 resulted from the inhibition of EGFR from the effect of EGF secreted from epithelial cells themselves in an autocrine manner. 

Cells Cultured on Basement Membrane Matrix

Epithelial cells cultured on basement membrane matrix (Matrigel; BD Biosciences) formed aggregates of several cells within 3 days. The aggregates gradually increased in size, and the cells formed spherical cell clusters within 1 week. The clusters expanded in size thereafter (Figs. 4A 4B 4C) . Paraffin sections of the cells from the three-dimensional culture on day 14 showed that the clusters had an acinarlike structure (Fig. 4D) . The lumina were filled with acellular fluid weakly stained with eosin. Immunohistochemistry revealed that most of the cells showed strong positive immunostaining for AQP5, a water-channel protein expressed in lacrimal acinar and ductal cells, ^{33 34 35 36} on the cell membrane (Fig. 4E) . 

Electron microscopic observation of the cell clusters on basement membrane matrix (Matrigel; BD Biosciences) for 1 week showed that squamous or cuboidal cells formed lumina filled with colloidal amorphous material. Cell membranes facing the lumina were rich in microvilli (Fig. 4F) . Higher magnification (Fig. 4G)revealed basement membrane outlining the clusters. Tight junctions, interdigitations, and lumina of canaliculi were formed between cells, which contained numerous vesicles containing materials considered secretory granules. 

Discussion

Three-Step Purification Method and Immunocytochemical Analysis of Lacrimal Gland Cells

Despite numerous studies on the characterization of lacrimal and salivary gland cells of adult primates, rabbits, and rodents, ^{2 3 5 6 7 8 9 10 11 13 14 37 38 39 40 41 42} no reports have been published in newborn mice. 

In this study, we successfully obtained lacrimal gland epithelial and mesenchymal cells with a purity of 99.5% from newborn mice by three steps. For the first step, collagenase-digested lacrimal gland tissue was filtered. Most epithelial cells, still maintaining acinar or ductal structures, were retained on the mesh, whereas mesenchymal cells were dispersed and passed through the mesh. For the second step, we purified the epithelial and mesenchymal cells by using their difference in the time required for adhesion to culture vessels. Mesenchymal cells adhered firmly sooner than epithelial cells. ⁴⁰ For the third step, epithelial or mesenchymal culture was maintained in a different medium that preferentially stimulated the proliferation of one of either type of cells. After 7 days of culture, the purity of cells in each culture was confirmed by immunocytochemical analyses. 

Immunocytochemical characterization of cultured lacrimal and salivary gland cells has been reported by several groups. As epithelial markers, some cytokeratins (CK-5, -6, -8, -17) are positive, but CK-18 is negative ^{8 9} in rabbit and rat lacrimal gland epithelial cells. Other studies show that ZO-1, ^{40 41 42} claudin-1, ^{40 41 42} E-cadherin, ^{40 41} occludin, ^{13 42} and connexin32, ¹² expressed in the intercellular junction, were shown in lacrimal gland and salivary epithelial cells and that vimentin ^{39 43} was detected in mesenchymal cells. Thus, we chose cytokeratins as epithelial markers, ZO-1 as a marker of tight junctions, and vimentin as a mesenchymal marker. Furthermore, we used SMA as a muscle marker, VWF as a vascular marker, GFAP as a glial marker, NF as a neuronal marker, and CD34 as a bone marrow-derived cell marker to identify the origins of contaminating cells and to calculate the purity of lacrimal gland epithelial cells. Although there is the possibility of the presence of lymphocytes in lacrimal epithelial preparations, ³ lymphocytes do not adhere to the culture substrate and are thought to be washed away by the exchange of culture medium. Therefore, we did not examine lymphocyte contamination. 

Because of the presence of cytokeratins and ZO-1 but not vimentin, we considered that the immunocytochemical features of our lacrimal gland epithelial cells in a monolayer culture were consistent with those mentioned in previous reports. Our three-step purification method was proven to be an efficient method for obtaining highly purified lacrimal gland cells from newborn mice. 

Previous studies ^{39 40} reported that salivary gland epithelial cells proliferated well with passages in media containing several growth factors or on a 3T3 feeder, but lacrimal gland epithelial cells in our culture conditions did not proliferate sufficiently after replating. Further improvement of culture conditions is necessary. 

Responses of Lacrimal Gland Cells to Growth Factors

It has been reported that FGF10 is essential and sufficient to initiate the organogenesis ^{16 17 20} of mouse lacrimal glands. On the other hand, there have been numerous reports about the involvement of other growth factors such as EGF, FGFs, and HGF in the growth and development of exocrine glands in mouse, rat, and rabbit. ^{10 16 17 18 19 20 21 22 23 28 29 44 45 46}  

In our experiment, FGF10 was expressed in lacrimal gland mesenchymal cells, and its receptor, FGFR2IIIb, was expressed in epithelial cells at an early postnatal stage of mice. FGF10 could stimulate branching formation in organ culture from E14.5 lacrimal glands, as previously reported in the budding stage at E13.5. ¹⁷ However, unexpectedly, FGF10 failed to stimulate epithelial cell proliferation in cell culture in spite of the expression of FGFR2IIIb. It is possible that FGF10 works in concert with autocrine or paracrine factors. If FGF10 enhanced EGF effects, the addition of FGF10 would increase cell viability in the control medium in which EGF works in an autocrine manner. In our experiments using postnatal lacrimal gland epithelial cells, we observed no significant difference in cell viability between control and FGF10 media. We concluded that FGF10 had no influence on the action of EGF or other autocrine factors. The addition of FGF10 with HGF, which is a paracrine factor, also failed to upregulate the action of HGF. Because of the possibility that the FGF10 preparation in this experiment was inactive, we confirmed that the same FGF10 could stimulate the development of E14.5 lacrimal glands. Lacrimal gland epithelial cells show different responses to FGF10 at different developmental stages and in different culture conditions, though they are continuously expressing its receptor, FGFR2IIIb. At present, we have not clarified the function of FGF10 in postnatal lacrimal glands. Although we detected no effects of FGF10 on cell proliferation in a monolayer culture, FGF10 may affect other cellular behavior such as glandular morphogenesis and functional differentiation. 

Epithelial and mesenchymal cells expressed c-Met and HGF, respectively, as reported in salivary glands, ²⁹ and the addition of HGF enhanced epithelial cell proliferation. HGF secreted from lacrimal gland mesenchyme appeared to work in a paracrine manner to stimulate the epithelial growth in newborn mice. 

We showed that lacrimal gland epithelial cells from newborn mice expressed EGF and EGFR and confirmed that EGF worked in an autocrine manner in the experiment with the addition of EGF and EGFR inhibitor. According to these results, EGF is considered to stimulate cell growth in newborn mouse lacrimal glands. 

This is the first report on the requirement of HGF and EGF for epithelial proliferation in newborn mouse lacrimal glands. Figure 5summarizes the action of growth factors and their receptors in lacrimal gland epithelial cells. 

Three-Dimensional Culture of Highly Purified Lacrimal Gland Epithelial Cells

It is well known that glandular epithelial cells in a three-dimensional culture ^{2 4 5 6 11 14 42 47} express cell features mimicking in vivo glands compared with those in a monolayer culture. ^{13 38 40 41} In previous reports on three-dimensional cultures, the epithelial cells were resected from glands and plated on gel directly after dissociation. Therefore, the influence of possibly contaminating cells in the epithelial culture cannot be excluded. In this study, we successfully reconstructed the acinarlike structures with our highly purified lacrimal gland epithelial cells. Epithelial cells in the cluster displayed strong immunostaining for AQP5 and showed ultrastructures, such as secretory granules and intercellular canaliculi, that suggested mature functions. The acinarlike cell clusters obtained by our method functionally and morphologically resembled mature acini of lacrimal glands. We are conducting further experiments to observe the physiological responses of the epithelial cells in a three-dimensional culture, including the evaluation of effects of FGF10. 

Our methods for the purification and culture of lacrimal gland cells may provide new in vitro models to study not only the physiological and morphologic characteristics of the lacrimal gland but also the basis of the cellular/molecular pathogenesis of lacrimal gland disorders. 

 Supported by National Defense Medical College.

 Submitted for publication June 28, 2008; revised October 6 and November 26, 2008; accepted March 9, 2009.

 Disclosure: Y. Ueda, None; Y. Karasawa, None; Y. Satoh, None; S. Nishikawa, None; J. Imaki, None; M. Ito, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Phase-contrast images (A, I) and immunocytochemistry (B–H, J, K) of methanol-fixed lacrimal gland epithelial (A–H) and mesenchymal (I–K) cells cultured for 7 days. (A) Polygonal cells form a confluent monolayer of epithelium. (B) Immunocytochemistry for pan-CK (Texas red) and ZO-1 (fluorescein). Sharp linear staining with fluorescein-labeled ZO-1 is shown around Texas red–stained pan-CK–positive polygonal cells. (C) Immunocytochemistry for CK-8 (Texas red) and vimentin (fluorescein). Almost all cells forming a monolayer are immunostained for CK-8, and a few are immunostained for vimentin (arrows). (D) Immunocytochemistry for pan-CK (Texas red) and SMA (fluorescein). Arrows: myoepithelial cells double positive for pan-CK and SMA. (E) Immunocytochemistry for SMA (Texas red) and vimentin (fluorescein). Arrows: myofibroblasts immunostained for SMA and vimentin. (F) Immunocytochemistry for VWF. Inset: cultured aortic cells as a positive control. Although diffuse spotty immunostaining is seen in the cytoplasm in the inset, no positive stain is observed in epithelial cells. Immunocytochemistry for NF (G) and GFAP (H). Insets: cultured neural retina as a positive control. Stringlike positive staining of NF or GFAP is not shown among epithelial cells. (I–K) Phase-contrast images (I) and immunocytochemistry for vimentin (J) and pan-CK (K) of mesenchymal cells. All cells stained positive for vimentin (J) but negative for CK-8 (K; compare with B and D). Scale bars, 100 μm.

Figure 2.

 

(A) Relative expression of FGF10 (left), FGFR2IIIb (middle), and FGFR2IIIc (right) to 18S rRNA by real-time reverse transcription–polymerase chain reaction. FGF10 is expressed more in mesenchymal cells than in epithelial cells (P < 0.05). Its receptor FGFR2IIIb is expressed exclusively in epithelial cells (P < 0.05). Mesenchymal cells express another FGF receptor, FGFR2IIIc, which does not bind with FGF10 and is detected at lower levels in epithelial cells than in mesenchymal cells (P < 0.05). (A, B, E) Symbols represent results from an independent culture experiment. (B) Quantification of gene expression of HGF (left) and c-Met (right) by real-time reverse transcription–polymerase chain reaction. Mesenchymal cells express higher levels of HGF than epithelial cells (P < 0.05). Conversely, epithelial cells express higher levels of c-Met than mesenchymal cells (P < 0.05). (C) Western blot of c-Met, which was detected only in epithelial cells. (D) Immunocytochemistry on epithelial cells for c-Met. Cells are outlined by strong immunopositive stain for c-Met. In many cells, the cytoplasm is also immunostained. (inset) Negative control. (E) Quantification of gene expression of EGF (left) and EGFR (right). Both expression levels show no significant difference between epithelial and mesenchymal cells, in contrast to FGF10, HGF, and their receptors. Immunocytochemistry for EGF (F) and EGFR (G). (F) Cytoplasm is uniformly immunostained. (G) Spotty immunostaining is observed on the diffusely immunostained cell layer. Inset: negative control. Scale bars, (D, F, G) 100 μm.

Figure 3.

 

(A) Evaluation of cell death (LDH), cell viability (XTT), and DNA synthesis (BrdU) of epithelial cells cultured with growth factors. EGF (dotted bars) and HGF (black bars) reduce cell death and augment cell viability and DNA synthesis. FGF10 (gray bars) was not significantly different from control (white bars). (B) Epithelial cell viability cultured in combinations of various concentrations of FGF10 and HGF. Left: cell viability at various concentrations of HGF with 0 or 10 ng/mL FGF10. Cell viability increased by the addition of HGF in a dose-dependent manner. No significant differences in cell viability were observed with and without FGF10 at each concentration of HGF. Right: cell viability at various concentrations of FGF10 with 0 or 5 ng/mL HGF. No significant difference in absorbance was observed among the three concentrations of FGF10 at each concentration of HGF. Addition of HGF raised cell viability in each concentration of FGF10. (C, upper) Budding of the lacrimal gland at the stage before treatment (E14.5). Lacrimal glands organ-cultured with PBS- (middle) and FGF10- (lower) soaked gelatin hydrogel cultured for 4 days. The epithelium is blue after X-gal staining. The lacrimal gland with FGF10-gelatin hydrogel developed more branches than with PBS-gelatin hydrogel. L, lens. Dotted line: area of tissue resected. Asterisks: conjunctival epithelium. Arrows: distal ends of growing lacrimal gland epithelium. Scale bar, 0.5 mm. (D) Cell viability of epithelial culture fed with EGFR inhibitor AG1478 alone or together with EGF or HGF. AG1478 alone decreased cell viability and completely eliminated the effect of EGF. Hepatocyte growth factor increased cell viability even in the presence of AG1478. (A, B, D, bars) Averages and standard deviations of four measurements.

Figure 4.

 

Figures of lacrimal epithelial cells in three-dimensional culture. (A–C) Dissection microscopic images on day 3 (A), day 7 (B), and day 14 (C) in three-dimensional culture. Cells formed aggregates (A) and then clusters with lumina (B). Clusters gradually increased in size (C). (D) Paraffin section stained with hematoxylin-eosin. Acinarlike structures with lumina filled with amorphous material weakly stained with eosin. (E) Immunohistochemistry for AQP5. Many cells had strong immunoreactivity for AQP5 on the cell membrane (arrows). (F, G) Images of three-dimensional culture by transmission electron microscopy. (F) The lumen (L) was filled with amorphous colloidal materials. The cell surface facing the lumen was rich in microvilli (arrowheads). (G) Higher magnification. Basement membrane (arrows) formed between basement membrane matrix and the cell layer. The cells contained vesicles (V). Tight junctions (TJ) and interdigitation (ID) appear between cells. Canaliculus (C) is surrounded by cell membrane with microvilli. Scale bars: (A–C) 250 μm; (D, E) 100 μm; (F) 5 μm; (G) 2 μm.

Figure 5.

 

A schematic model describing the interaction of growth factors between the epithelium and mesenchyme in the lacrimal gland. Epithelial cells expressed EGFR, c-Met, and FGFR2IIIb. Epithelial growth factor secreted from epithelial cells bound to EGFR on epithelial cells. HGF released from mesenchymal cells bound to c-Met on epithelial cells. Binding of EGFR and c-Met to their ligands activated their downstream signal transduction pathways to suppress cell death and stimulate cell proliferation, resulting in an increase in cell number. FGF10 secreted from mesenchymal cells did not bind to FGFR2IIIc on mesenchymal cells but did bind to FGFR2IIIb on epithelial cells. However, FGF10 did not enhance epithelial cell proliferation. In spite of the certain expression of FGF10 and its receptor, their roles remain unknown.

The authors thank Kazuyo Kuroki and Yuko Ogura for their technical assistance with the animal experiments and Yayoi Ichiki for her assistance with transmission electron microscopy.