Isolation and Characterization of Cultured Human Conjunctival Goblet Cells

Abstract

purpose. To isolate and characterize goblet cells from normal human conjunctival tissue to determine whether epidermal growth factor (EGF) receptors are present and whether EGF can influence goblet cell proliferation.

methods. Goblet cells were isolated from explant cultures established from normal conjunctival tissue harvested from patients during periocular surgery. The cells were grown in RPMI culture medium supplemented with 10% fetal bovine serum and characterized using morphology, histochemistry, indirect immunofluorescence microscopy, molecular biology, and biochemistry. Proliferation was determined with a MTT proliferation assay after exposing goblet cells, which had been serum deprived for 48 hours, to increasing concentrations of epidermal growth factor (EGF; 0–80 ng/mL) for 24 hours.

results. Goblet cells were isolated from conjunctival explants by scraping nongoblet cells from the culture dish. Human goblet cells exhibited positive reactivity with alcian blue-periodic acid Schiff (PAS) reagent, goblet cell-specific cytokeratin-7, HPA lectin, and MUC5AC, but negative reactivity to the stratified squamous epithelial cell marker, cytokeratin-4. The mRNA for MUC5AC was detected using RT-PCR. The presence of the EGF receptors EGFR, ErbB2, and ErbB3 was confirmed through Western blot analysis of cell lysates. EGF elicited a concentration-dependent increase in goblet cell proliferation of 160% ± 0.5%, 188% ± 0.45%, 293% ± 1.3%, and 220% ± 0.5% of control values with 10, 20, 40, and 80 ng/mL EGF, respectively.

conclusions. Human goblet cells that retain characteristics of goblet cells in vivo can be cultured. EGF receptors are present in human goblet cells, and EGF stimulates their proliferation.

The integrity of the ocular surface is greatly influenced by the levels of mucin present in the tear film. Goblet cells, which are highly specialized epithelial cells, are the primary source of this ocular mucin. ^{1 2} Goblet cells, located in the apical surface of the conjunctiva, are interspersed among its multiple layers of stratified epithelium ³ and are easily recognized by their vast accumulation of secretory vesicles. ⁴ In the human conjunctiva, goblet cells can occur singly or in greater numbers lining epithelial infoldings or crypts ^{5 6} and are thought to have a more rounded appearance than their counterparts in other species and tissues. ⁷ Goblet cells synthesize, store, and secrete complex, high-molecular-weight glycoproteins known as mucins and specifically express MUC5AC, a large gel-forming mucin. ^{8 9} Once mucin is secreted by goblet cells, it has the capability to hydrate and gel, thus keeping the conjunctiva moist. This film produces a protective covering over the ocular surface ^{10 11} shielding it from a variety of pathogens, chemicals, and environmental toxins. ^{12 13} Because mucin is essential in maintaining the health of the ocular surface, aberrations in goblet cell mucin secretion, either underproduction or overproduction, can result in injury to both the cornea and conjunctiva. Mucin deficiency is present in alkali burns, thermal burns, Stevens-Johnson syndrome, neurotrophic keratitis, and ocular cicatricial pemphigoid. ^{14 15} In some diseases, overproduction of mucin owing to excessive goblet cell proliferation and/or secretion is due to the presence of activated T-cells or macrophages. ¹⁶ If left untreated, mucin abnormalities could severely compromise the ocular surface and lead to serious visual impairment. Because treatment is expensive, of long duration, and often unsuccessful, managing these diseases presents a formidable problem. 

The importance of goblet cells as major producers of ocular surface mucin is well established, ^{17 18} with critical emphasis placed on the number of functional goblet cells in the conjunctiva and on the amount and rate by which they synthesize mucin. Our laboratory has worked extensively on rat goblet cells in vivo and has reported that rat goblet cell mucin secretion is mediated by the activation of signal transduction pathways activated by neurotransmitters, such as cholinergic agents, and VIP, and, most recently, that it is mediated by growth factors ^{17 19} . ²⁰ Although secretion from these cells has been widely studied, little is known concerning factors or stimuli that regulate proliferation of conjunctival goblet cells. Our laboratory recently developed a system to isolate and propagate rat conjunctival goblet cells in culture that enables us to study goblet cell biology. ²¹ Yet data obtained in rodent studies are limited and must be extrapolated to be meaningful to humans. Thus, we have developed a system to isolate human goblet cells and to evaluate the effects in goblet cells of growth factors and other agents on short-term processes such as secretion and on long-term processes such as proliferation, migration, differentiation, and apoptosis. 

Epidermal growth factor (EGF) is a multifunctional cytokine that has documented effects on the ocular surface. This growth factor promotes the rapid healing of corneal epithelial and other ocular surface wounds by increasing cell proliferation. ^{22 23} It supports chemotaxis ^{24 25} and assists cell migration by stimulating epithelial cell attachment to a fibronectin matrix. ²⁶ Growth factors generally exert long-term effects that regulate cellular processes such proliferation, migration, differentiation, and protein synthesis. However, recent work has shown that growth factors such as EGF can also regulate short-term responses such as secretion. ²⁷  

The EGF family of growth factors includes 11 members that have a common 40- to 45-amino-acid sequence containing six cysteine residues referred to as the EGF-like domain. ²⁸ They are synthesized as transmembrane precursor molecules that are proteolytically cleaved to release soluble, mature growth factors. EGF and its family members bind to ErbB receptors. The EGF receptors EGFR, ErbB2, ErbB3, and ErbB4 belong to the receptor tyrosine kinase family. ^{29 30 31 32} These proteins have a common structure consisting of an extracellular domain with two ligand binding elements located in two cysteine-rich regions. This is followed by a single transmembrane region that contains tyrosine kinase and a carboxyl terminal tail with multiple autophosphorylation sites. ³⁰ Activation of the tyrosine kinase activity of EGFR, ErbB2, and ErbB3 after ligand binding is thought to play a role in the regulation of cellular proliferation and differentiation. ³³ EGFR, ErbB2, and ErbB3 are expressed in ocular surface epithelium. ³⁴ EGFR appears to be preferentially expressed by basal epithelial cells in the cornea, limbus, and conjunctiva, whereas ErbB2 and ErbB3 are found mainly on the superficial cells and, to a much lesser degree, in the basal cells. ³⁴ To date, these receptors have not been localized in conjunctival goblet cells. 

The focus of the present study was to isolate and characterize goblet cells established from explants of disease-free human conjunctival tissue by using morphologic, histochemical, immunocytochemical, and biochemical markers. Preliminary studies were undertaken to investigate whether the interaction of human goblet cells with the growth factor EGF could modulate their proliferation in vitro. The studies described herein indicate that human goblet cells can be isolated from human conjunctiva and, regardless of gender, age, or the surgical procedure performed on the donor patients, these cells retain characteristics of human goblet cells in vivo. The cultured cells exhibit positivity for known goblet cell markers. These cells contain MUC5AC mRNA and express EGFR, ErbB2, and ErbB3 and are stimulated to proliferate by EGF. 

Materials and Methods

RPMI-1640 culture medium, l-glutamine, penicillin-streptomycin, Hanks’ balanced salt solution (HBSS), and trypsin-EDTA solution were obtained from BioWhittaker (Walkersville, MD); fetal bovine serum from Hyclone Laboratories (Logan, UT); tissue culture flasks, pipettes, and other routine plastics from BD Labware (Franklin Lakes, NJ); glass coverslips from VWR Scientific (San Francisco, CA); and chamber slides (Laboratory Tek) from Nunc, Inc. (Naperville, IL). Recombinant EGF, Taq polymerase, and avian myeloblastosis virus (AMV) reverse transcriptase were from Promega (Madison, WI); G3PDH primers from Clontech (Palo Alto, CA); monoclonal antibody against cytokeratin 7 (CK7) from ICN (San Francisco, CA) and against Ki-67 from Novocastra Laboratories, Ltd. (Newcastle-upon-Tyne, UK); Ulex europaeus agglutinin lectin (UEA-1) and Helix pomatia agglutinin lectin (HPA), directly conjugated with fluorescein isothiocyanate (FITC) or Texas red from Pierce (Rockford, IL); Bandeiraea simplicifolia lectin (BS-1) from Vector Laboratories (Burlingame, CA); monoclonal antibodies to EGFR and ErbB3 from Santa Cruz Biotechnology (Santa Cruz, CA); monoclonal antibody to ErbB2 from NeoMarkers (Fremont, CA); DNase I and extraction reagent (TRIzol) from GibcoBRL-Life Technologies (Rockville, MD); and A431 human epidermoid carcinoma cell extracts, pretreated with EGF and SK-BR-3 human breast carcinoma cell lysates, from Upstate Biotechnology (Lake Placid, NY). All other chemicals, unless otherwise specified, were obtained from Sigma (St. Louis, MO). The cytokeratin 4 antibody was a gift of James Zieske, Schepens Eye Research Institute (Boston, MA). Marcia Jumblatt, University of Louisville School of Medicine (Louisville, KY) provided the antibody to human MUC5AC. ¹⁷  

Isolation and Culture of Cells

Normal human conjunctival tissue was obtained from patients during periocular surgery at the Massachusetts Eye and Ear Infirmary (MEEI) with a protocol that adhered to the tenets of the Declaration of Helsinki and was approved by the MEEI Human Subjects Internal Review Board. This tissue, which would normally be discarded during the surgical procedure, was donated. All tissue was removed after informed consent of the donor. Immediately after excision of the tissue, it was placed into either HBSS or PBS containing 3× (300 μg/mL) penicillin-streptomycin. Tissue was finely minced into 1-mm³ pieces and anchored onto either scored culture dishes or glass coverslips placed within six-well culture dishes. The culture dishes contained just enough medium to cover the bottom of the dish, so that the tissue would receive nutrients through surface tension. More specifically, 0.5 mL of medium was added to each well in the six-well plate before the conjunctival tissue plug was attached. The cell medium that was used to feed explants and culture goblet cells consisted exclusively of RPMI-1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 2 mM l-glutamine, and 100 μg/mL penicillin-streptomycin. Explants were refed every 2 days with 100 to 200 μL of the medium and were grown under routine culture conditions of 95% air and 5% CO₂ at 37°C. The cells were permitted to grow from the conjunctival tissue plug for approximately 72 hours or until adherent, spread-out cells were evident surrounding the plug. At this point, the conjunctival plug was removed to prevent the growth of connective tissue cells, which would overgrow the culture. Cultures displaying exclusively connective tissue morphology were routinely discarded. Goblet cells were further isolated from epithelial cells by scraping contaminating nongoblet cells with a rubber policeman or conventional plastic cell scraper. Goblet cells were identified by the presence of numerous secretory granules within the cells. Nongoblet cells were larger, winglike cells with small nuclei. Immediately after the scraping, contaminating cells were eliminated from the culture dish by rinsing the dish three times with HBSS or PBS, to prohibit reattachment of undesirable cells. Each well was refed with 2 mL of complete RPMI medium. 

We obtained goblet cells from every sample listed in Table 1 . However, when organ explant cultures were established, goblet cells did not grow from every piece of tissue. Primary cultures of goblet cells were processed for light and electron microscopy, immunocytochemistry, biochemistry, and molecular biology at various time points after the start of culture. Other cultures were trypsinized and passaged after reaching confluence. The cells were passaged by trypsinization of confluent, adherent cells with 0.05% trypsin in 0.53 mM EDTA (pH 7.4). After the goblet cells were detached from the tissue culture well, they were collected into centrifuge tubes, pelleted, and resuspended in compete RPMI medium. The cells were then seeded into six-well plates (1:1 split ratio), which had been coated with 1% gelatin to facilitate attachment. Five of the goblet cell-enriched cultures listed in Table 1 were successfully passaged, in that 30% to 50% of the cells adhered to the tissue culture plate and grew. In our hands, they were not successfully passaged beyond this point. However, all experiments detailed in this study were performed on cells from primary culture. 

Processing of Human Conjunctival Tissue

Some biopsy specimens of human conjunctival tissue similar to those obtained from donors during surgery for tissue culture were also received from three authors (PADR, CRB, ELC). They were immediately rinsed in PBS, embedded in optimal cutting temperature compound (OCT), frozen, and sectioned on a cryostat according to routine histologic procedures, or RNA was isolated for RT-PCR. 

Histochemistry

To determine whether neutral and/or acidic glycoconjugates, which are well-established markers of goblet cells in vivo, were also present in cultured goblet cells, the cells were fixed with 4% paraformaldehyde and processed with alcian blue-periodic acid Schiff reagent (AB/PAS). Goblet cells examined for lectin histochemistry were grown on either chamber slides (Laboratory-Tek; BD Labware) or glass coverslips in plastic tissue culture wells, rinsed in PBS, fixed in 100% methanol for 15 minutes at room temperature, and returned to fresh PBS. Fixed cells were incubated in blocking buffer that consisted of 1% BSA and 0.2% Triton X-100 in PBS for 30 minutes at room temperature. The cells then were incubated for 1 hour at room temperature with UEA-1 lectin conjugated directly to FITC diluted 1:100 in PBS, BS-1 lectin conjugated to FITC diluted 1:200, or HPA conjugated to Texas red diluted 1:100 in PBS. 

Immunocytochemistry

Methanol-fixed cells were examined for the presence of cytokeratins 4 and 7 and for MUC5AC. Slides with cultured goblet cells were incubated for 30 minutes at room temperature in blocking buffer that contained 1% BSA and 0.2% Triton-X in PBS. The cells then were incubated with the following dilutions of primary antibodies for 1 hour at room temperature: Antibody to cytokeratin 7, which recognizes a goblet cell-specific keratin, ^{35 36} was diluted 1:15 in PBS; antibody to cytokeratin 4, specific for stratified squamous nongoblet epithelial cells, ^{35 36} was diluted 1:10 in PBS; and antibody to human MUC5AC was diluted 1:1000 in PBS. For the investigation of the proliferation profile of cultured goblet cells, an antibody to human Ki-67 nuclear antigen was diluted 1:100 in PBS. The secondary antibodies, conjugated to FITC, rhodamine, or CY3, were diluted 1:150 in PBS and were incubated for 1 hour at room temperature. Slides, coverslips, or dishes were washed three times in PBS, after which they were mounted with 100 mM Tris (pH 8.5 (containing 25% glycerol, 10% polyvinyl alcohol, and 2.5% 1,4-diazobicyclo-[2.2.2]-octane). The cells were viewed by inverted phase-contrast microscope (Eclipse TE 300; Nikon, Tokyo, Japan) equipped for epifluorescence, and cells adherent to glass coverslips or microscope slides were visualized by fluorescence microscope (Eclipse E 800; Nikon). Negative control experiments consisted of substituting PBS for the primary antibody, and the positive control included frozen and/or fixed sections of human conjunctiva containing prominent goblet cells. 

Transmission Electron Microscopy

Medium was removed from subconfluent cultures of goblet cells after which cells were washed two times with cacodylate buffer (pH 7.3). The cells were fixed with cacodylate buffered Karnovsky’s solution, postfixed in 1% osmium tetroxide and embedded in Epon, according to standard transmission electron microscopy techniques. Thin sections, mounted on copper grids, were stained with lead citrate and examined with a transmission electron microscope (model 410; Philips, Eindhoven, The Netherlands). 

RNA Isolation and RT-PCR

Total RNA from human conjunctiva and cultured goblet cells were isolated with extraction reagent (TRIzol; Gibco) according to the manufacturer’s protocol. Briefly, pieces of human conjunctiva were incubated in 1 mL of the reagent immediately after receipt, and the cultured cells were grown in complete RPMI medium, as described earlier, and removed from the culture vessel by trypsinization in a trypsin-EDTA solution consisting of 0.05% trypsin and 0.53 mM EDTA. The cells then were pelleted by centrifugation, washed, resuspended in 1 mL extraction reagent and then lysed by repeatedly passing them through an 18-gauge needle. The samples were incubated at room temperature for 5 minutes to permit their complete dissociation. Chloroform (0.2 mL) was added and the tubes shaken and centrifuged for 15 minutes at 4°C. The upper aqueous phase containing the RNA was then removed. RNA was precipitated overnight at −20°C in isopropanol. The RNA pellet was washed with 75% ethanol and resuspended in diethyl pyrocarbonate (DEPC)-treated water. 

One microgram of total RNA was treated with 1 U DNase I for 15 minutes at room temperature, followed by addition of EDTA at a final volume of 2.5 mM. DNase I was further inactivated by heating the samples to 65°C for 10 minutes. The treated RNA was then used for cDNA synthesis. Reverse transcription was performed in a buffer containing 5 mM MgCl₂, 10 mM Tris-HCl (pH 9), 50 mM KCl, 0.1% Triton X-100, 1 mM dNTPs, 1 U/μL RNasin RNase inhibitor, 15 U/μg AMV reverse transcriptase, and 0.5 μg oligo-dT primers at 42°C for 60 minutes. The reverse transcriptase was denatured by incubating the samples for 5 minutes at 95°C followed by 5 minutes at 0°C. 

The cDNA was amplified by PCR with 0.75 U Taq polymerase (Promega) in buffer containing 1.5 mM MgCl₂. PCR was performed with primers for human MUC5AC. ³⁷ Primer sequences were as follows: MUC5AC sense primer, 5′-TCC ACC ATA TAC CGC CAC AGA-3′; and antisense primer, 5′-ATG GTG CCG TTG TAA TTT GTT GT-3′ and G3PDH sense primer, 5′-ACC ACA GTC CAT GCC ATC AC-3′ and antisense, 5′-TCC ACC ACC CTG TTG CTG TA-3′. G3PDH was used to control for RNA quality. The amplification reaction was performed in a thermal cycler (PCR Sprint; Hybaid, Middlesex, UK). The conditions were 5 minutes at 94°C, followed by 37 cycles of denaturation for 30 seconds at 94°C, amplification for 1 minute at 55°C, and extension for 1 minute at 72°C. Amplified cDNA was analyzed by electrophoresis on a 1% agarose gel in buffer containing 89 mM Tris borate (pH 8.3) and 2 mM EDTA and viewed by ethidium bromide staining. Absence of genomic DNA was confirmed by omitting reverse transcriptase and performing the PCR. No bands were obtained when this control was performed. 

Electrophoresis and Immunoblot Analysis

Goblet cells, grown under routine conditions in complete RPMI culture medium supplemented with 10% fetal bovine serum, 2 mM l-glutamine, and 100 μg/mL penicillin-streptomycin, were scraped and collected into homogenization buffer containing 30 mM Tris-HCl (pH 7.5), 10 mM EGTA, 5 mM EDTA, 1 mM dithiothreitol (DTT), 10 mg/mL phenylmethylsulfonyl fluoride (PMSF), and 5 U/mL aprotinin and then further lysed by sonication. For determination of the presence of the EGF receptor types EGFR, ErbB2, and ErbB3 in human goblet cells, the lysates were separated by SDS-PAGE of 6% gels and transferred to nitrocellulose membranes, as described by Towbin et al. ³⁸ To detect EGF receptor subtypes, the membranes were blocked for 1 hour at room temperature in 5% dried milk in TBST consisting of 10 mM Tris-HCl (pH 8.0), 500 mM NaCl, and 0.05% Tween-20 and then incubated with antibodies to EGFR, ErbB2, or ErbB3 at dilutions of 1:80, 1:20, and 1:300, respectively, for 1 hour at room temperature. The nitrocellulose membranes were washed three times with TBST and then incubated with a 1:2500 dilution of IgG conjugated to horseradish peroxidase in TBST for 1 hour. The membranes were washed three times, after which the EGF receptor subtypes present on human goblet cells were visualized using the enhanced chemiluminescence method. Homogenized A431 cell lysates and SK-BR-3 human breast carcinoma lysates known to contain ErbB3 were used as the positive control. 

Goblet Cell Proliferation

Goblet cells were deprived of serum for 48 hours and then exposed to increasing concentrations of EGF (0, 10, 20, 40, and 80 ng/mL in serum-free RPMI) for 24 hours. Proliferation was then determined in triplicate with a colorimetric nonradioactive, MTT proliferation assay. Cleavage of MTT, the pale yellow tetrazolium salt 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrzolium bromide, by viable, growing mitochondria forms a dark blue formazan product that can be completely solubilized in acidic isopropanol and detected with a microplate reader. ³⁹ After exposure of human goblet cells to EGF, MTT was added for a final concentration of 1.2 mM and allowed to incubate at 37°C for at least 4 hours. An equal volume of acidic isopropanol was then added to the cells, after which the culture dish was wrapped in foil and maintained at −20°C overnight. The next morning, after trituration of the cells, samples were placed in a microtiter plate and the absorbance read at 570 nm with an ELISA plate reader. In separate experiments, primary cultures of goblet cells isolated from two patients were grown on glass coverslips and subjected to the same protocol of serum starvation and stimulation with increasing concentrations of EGF as described earlier. After stimulation with EGF, cells on coverslips were fixed in methanol and processed for Ki-67 immunocytochemistry. Five predetermined fields that encompassed the entire area of the coverslip were scored for labeled and unlabeled cells. By calculating the number of labeled cells in comparison with the total number of cells counted, the percentage of stimulation due to EGF was calculated. 

Statistical Analysis

Data are expressed as the mean ± SEM. Student’s t-test for paired values was used to analyze data. P ≤ 0.05 was considered to be significant. 

Results

Growth and Morphology of Cultured Human Goblet Cells

Tissue samples were received from 22 patients (male and female) ranging in age from 21 to 84 years, who were undergoing the various types of periocular surgical procedures listed in Table 1 . All tissue received for isolation of goblet cells was normal tissue located adjacent to that which was diseased. 

As early as 24 hours after establishment of the organ culture, regardless of the patient’s age and gender or the surgical procedure, adherent, round cells were visible surrounding the anchored piece of conjunctival tissue. By 48 hours, it was possible to identify whether the cells were of an epithelial or fibroblastic lineage, based on their morphology. After 72 hours in culture, the conjunctival tissue was removed, to inhibit growth of fibroblasts and contamination with other stromal cell types. At this juncture, our early cultures were subjected to AB/PAS staining to identify potential goblet cells. Based on our experience with the use of these morphologic and histochemical markers to identify rat goblet cells in organ cultures, nongoblet cells were scraped away from the culture dish with a rubber policeman. Human goblet cells, unlike those observed in rat conjunctival organ cultures, appeared to grow out as sheets of relatively (approximately 90%) pure cells from either one or multiple areas of the anchored piece of conjunctival tissue. The morphology of human goblet cells in primary culture (Fig. 1) showed their variation in both size and shape. Smaller cells were present nearest the area occupied by the tissue plug, whereas larger, more mature goblet cells were found at the periphery of the culture. Goblet cells in primary culture were evaluated for proliferation at random, with Ki-67 used as a marker of cell division. All cultures tested showed the presence of dividing cells (data not shown). 

Human goblet cells were difficult to passage. Most cultures were lost after trypsinization of cells and subsequent plating on untreated plastic tissue culture vessels. For passaging, the cells were routinely seeded in vessels that had been coated with 1% gelatin. With this substrate, a 30% to 50% plating efficiency was observed. Human goblet cells derived from organ cultures could not be successfully subcultivated beyond the first passage. 

Characterization of Cultured Human Goblet Cells

Although AB/PAS was used as a preliminary marker to identify the morphology of human goblet cells, we routinely used this histochemical reaction to evaluate the purity of our cultures. Human goblet cells in culture reacted positively with AB/PAS, exhibiting the presence of both acidic (blue) and neutral (red) glycoconjugates in the cells themselves as well as in the strands and layers of mucin deposited over the cell cultures (Fig. 2A) . Human goblet cell-enriched cultures, unlike those of the rat, were almost immediately covered and often obscured with thick layers of what we interpreted to be secreted mucins. In addition, reactivity with AB/PAS was retained in passaged cell cultures (data not shown). AB/PAS reactivity of goblet cells in culture was compared with that of goblet cells in vivo (Fig. 2B) . A similar reactivity of goblet cells was observed both in cultures as well as in tissue sections of human conjunctiva. 

Lectins, which bind specific carbohydrate residues, have been widely used for the localization of specific carbohydrates. We used lectins as histochemical markers in the identification of human goblet cells and tested their reactivity to the following lectins: UEA-1 which recognizes the l-fucose moiety of glycoproteins in the secretory granules of goblet cells in a variety of species, but is absent in most human conjunctival goblet cells; HPA which recognizes l-galactosamine within the secretory granules of goblet cells including human goblet cells and BS-1, which recognizes the n-galactosyl groups of glycoproteins in stratified epithelial cells. Human goblet cells in culture reacted only with HPA, which was localized within the cells secretory granules (Fig. 3A) . These data differ from cultured rat goblet cells, which react with UEA-1. The reaction of conjunctival goblet cells with HPA in conjunctival tissue sections was similar to that of cells in culture (Fig. 3B) . Human goblet cells displayed no reactivity in vivo or in vitro to UEA-1 or to BS-1 (data not shown). 

We further identified cells as goblet cells using immunocytochemical localization of cytokeratin-7 an intermediate filament associated specifically with goblet cells. Goblet cells in culture and in tissue sections exhibited intense staining for cytokeratin-7, whereas other cells displayed no staining for this intermediate filament (Figs. 4A 4B) . Our cultures were also evaluated for the presence of cytokeratin-4, an intermediate filament associated with stratified squamous epithelial cells. Occasional large, winglike squamous cells that were cytokeratin 4-positive were evident in some of our cultures, but were found to be short-lived and would spontaneously detach from the culture dish (Fig. 4C) . We next performed double immunolabeling of human goblet cells in vivo and in vitro to determine whether HPA and cytokeratin-7 are present in the same goblet cell to aid in their identification and characterization. HPA and cytokeratin-7 often colocalized in some of the goblet cells both in vitro and in vivo (Figs. 5A 5B) . 

Additional immunocytochemical studies provided evidence that the secretory granules within the cytoplasm of human goblet cells stained intensely for the goblet cell-specific mucin, MUC5AC (Figs. 6A 6B) . Immunocytochemistry of human conjunctival sections also verified the specific localization of this mucin in goblet cells (6C) and not in adjoining epithelial cells. In complementary studies, total RNA was isolated from primary cultures of human goblet cells, pooled from three patients ranging in age from 72 to 84 years, and human conjunctiva (positive control). The RNA was reverse transcribed and analyzed for the message of MUC5AC. As shown in Figure 7 , the message for MUC5AC was detected in human goblet cell-enriched cultures. No products were detected in the negative controls not containing cDNA or only containing RNA. 

When cultured goblet cells were examined using transmission electron microscopy (Fig. 8) , a typical goblet cell morphology was observed. En face sections revealed an elongated body containing an apically placed nucleus and several storage vesicles observed in many cells, whereas others were more round in appearance. 

Detection of EGFR, ErbB2, and ErbB3 in Human Goblet Cells

Cultured human goblet cells were examined for the presence of EGF receptors, a receptor commonly involved in epithelial cell proliferation. Proteins from cell lysates of primary cultures of goblet cells were analyzed by Western blot methods using antibodies against EGFR, ErbB2, and ErbB3. A band of 170 kDa was observed in human cultured goblet cells indicating the presence of EGFR (Fig. 9A) . A band of the same molecular weight was detected in cultured rat goblet cells and A431 cells, our positive controls. A431 cells overexpress EGFR accounting for the intense band detected. A major band of 185 kDa was detected in human cultured goblet cells indicating the presence of ErbB2 (Fig. 9B) , a band of similar molecular weight was identified in rat conjunctival goblet cells and A431 cells, our positive controls, and a band of 160 kDa was detected in cultured human goblet cells, indicating the presence of ErbB3 (Fig. 9C) . A band of similar molecular weight was identified in rat conjunctival goblet cells and in SK RB3 cells, our positive control. 

Proliferation of Goblet Cells in Response to EGF

Goblet cells were grown from conjunctival tissue obtained from five individuals undergoing periocular surgery. Because these studies were executed using primary cultures of goblet cells often with uneven patterns of growth, after serum-starvation, the bottom of each culture dish was marked with a square (same measurements for each dish) to ensure that an equal surface area was covered by experimental and control cultures. EGF elicited a concentration-dependent increase in goblet cell proliferation of 160% ± 0.5%, 188% ± 0.45%, 293% ± 1.3% and 220% ± 0.5% over control with 10, 20, 40, and 80 ng/mL, respectively (Fig. 10A) . Selected cultures isolated from two patients who had been exposed to EGF were also processed for immunocytochemistry using an antibody to Ki-67, a nuclear antigen that detects cells engaged in the cell cycle. Control cultures exhibited no fluorescence for Ki-67 (data not shown), whereas goblet cells exposed to EGF had labeled cells corresponding to increasing in proliferation (Fig. 10B) . Quantification of the effect of EGF on proliferation using Ki67 indicated an increase in proliferation with the same concentration of EGF as obtained with the MTT assay (Fig. 10C) . 

Discussion

Our data indicate that we successfully cultured human conjunctival goblet cells that retain their in vivo characteristics. We have previously adapted this specialized culture methodology to successfully isolate and grow rat conjunctival goblet cells in vitro. ²¹ These cells retained morphologic, histochemical, immunocytochemical, and functional markers in primary cultures. Explant or organ cultures have also been successfully used by a number of investigators to propagate a variety of ocular surface-derived cells, including keratinocytes ⁴⁰ and different types of epithelia. ^{41 42} Even though explant cultures do not consistently yield high numbers of specific cell types, they have distinct advantages over methods that use harsh chemical digestion to liberate primary cells. These explant cultures require minuscule amounts of tissue from which cultured cells are allowed to grow and differentiate under conditions of a normal host milieu. 

Culturing conjunctival goblet cells has been difficult. Risse March et al. ⁴⁰ were unable to find any evidence of goblet cell migration from conjunctival explants. They hypothesized that their culture conditions were not favorable to goblet cell growth, because it was performed in serum-free medium whereas the culture medium in the present study contained 10% serum. In addition, they obtained biopsy specimens from the bulbar conjunctiva, which contains few goblet cells. Other investigators have been able to identify goblet cells grown in cultures from conjunctival explants. ⁴³ However, goblet cells were identified only after several passages of keratinocytes, implying that keratinocytes and goblet cells arise from a common progenitor. ⁴³  

Little information exists concerning the human goblet cell in vitro, despite its critical importance to the ocular surface. This may be largely because normal human tissue is scarce. It is often in various states of viability dependent on the length of time it has been removed from the eye, and is very fragile when exposed to routinely used chemical isolation techniques. Furthermore, goblet cell function, except for secretion, cannot be studied in intact tissue or tissue pieces because of contamination from other cell types present. In this regard, we used a modified organ culture system to isolate goblet cells from fragments of normal human conjunctival tissue obtained from patients undergoing various types of conjunctival surgery. We show herein that human goblet cells, even though they cannot be passaged more than one time, can be successfully isolated from very small tissue samples. These cells regardless of donor age, sex, or surgical procedure undertaken, exhibit markers characteristic of human goblet cells in vivo. Human goblet cells react positively with AB/PAS and HPA lectin. They do not react with BS-1, a lectin normally found in stratified squamous cells or with UEA-1, a lectin found in rat but not human goblet cells. They express positive immunofluorescence for the intermediate filament cytokeratin-7 and the mucin MUC5AC, both specific markers of goblet cells in vivo. They were not positive for cytokeratin 4 which is found in conjunctival stratified squamous cells. Furthermore, cultured goblet cells have mRNA for MUC5AC. In addition, the cultured goblet cells expressed the EGF receptors EGFR, ErbB2, and ErbB3, and after exposure to increasing concentrations of EGF, their proliferation increased accordingly. Cultured human goblet cells will be useful to identify other stimuli of proliferation as well as to characterize goblet cells biochemically or molecularly. 

Conjunctival goblet cells due to their secretion of mucin are crucial to the integrity of the ocular surface. When goblet cells are lost, various mucin-deficient ocular surface diseases arise. ^{44 45 46 47} In this study, we have shown by AB/PAS reagent reactivity that cultured human goblet cells are associated with both acidic and neutral types of mucin, similar to those found in vivo in human conjunctiva as well as in tissue of other species. ^{21 48 49 50} By immunocytochemistry, we further identified the presence of the goblet cell-specific mucin MUC5AC within the secretory granules of cultured human conjunctival goblet cells and verified these results by demonstrating the presence of MUC5AC mRNA within these same cultured cells. 

Little is known about goblet cell proliferation, either in vitro or in vivo. EGF has been shown to exert profound effects on proliferation of ocular surface epithelium during wounding. ^{22 23 24} EGF and its receptors have also been detected by immunocytochemistry and immunoblot analysis in the corneal and conjunctival epithelium of rodents and humans. ^{51 52 53} In the current study we showed, with the use of Western blot analysis, that cultured human goblet cells possess receptors for EGFR, ErbB2, and ErbB3. Moreover these same cells, when exposed to increasing concentrations of EGF, respond by proliferating. 

In conclusion, this study describes a practical and reproducible method of isolating human conjunctival goblet cells from human donors. Cells isolated in this manner can be used in primary culture to further our understanding of goblet cell morphology, growth kinetics, and response to various stimuli as well as many aspects of human goblet cell function. Although the number of goblet cells isolated from each biopsy specimen is limited, the ability to obtain enriched goblet cell culture opens up the possibility of creating human goblet cell lines that can be carefully characterized and used for both in vitro and in vivo studies. These may be useful tools by which ocular surface disease caused by conjunctival goblet cell deficiency may be eradicated. 

 Supported by National Eye Institute Grant EY09057, Merit Award from Research to Prevent Blindness, and a grant from The Center for Alternatives to Animal Testing (CAAT).

 Submitted for publication June 6, 2002; revised December 2, 2002, and January 23, 2003; accepted February 7, 2003.

 Disclosure: M.A. Shatos, None; J.D. Ríos, None; Y. Horikawa, None; R.R. Hodges, None; E.L. Chang, None; C.R. Bernardino, None; P.A.D. Rubin, None; D.A. Dartt, 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.

 

Table 1.

 

Patient Information for Biopsy Specimens Used as Sources of Goblet Cells

Age	Sex	Surgical Procedure	Sample Received from Surgeon
40	Male	Conjunctivoplasty	Lateral eyelid margin
76	Male	Lateral tarsal strip	Lateral tarsal conjunctiva
76	Male	Punctoplasty	Punctal conjunctiva
74	Male	Lateral tarsal strip	Lateral tarsal conjunctiva
21	Female	Internal ptosis repair	Superior forniceal conjunctiva
88	Female	Punctoplasty	Punctal conjunctiva
85	Female	Medial spindle	Medial palpebral conjunctiva
66	Female	Lateral tarsal strip	Lateral tarsal conjunctiva
84	Female	Lateral tarsal strip	Lateral tarsal conjunctiva
56*	Male	Carunculectomy	Caruncle
74*	Female	Lateral tarsal strip	Lateral tarsal conjunctiva
36*	Female	Punctoplasty	Punctal conjunctiva
65*	Female	Punctoplasty	Punctal conjunctiva
76*	Female	Punctoplasty	Punctal conjunctiva
60	Male	Internal ptosis repair	Superior forniceal conjunctiva
60	Female	Internal ptosis repair	Superior forniceal conjunctiva
81	Male	Medial spindle	Medial palpebral conjunctiva
84	Male	Lateral tarsal strip	Lateral tarsal conjunctiva
61	Female	Internal ptosis repair	Superior forniceal conjunctiva
71	Male	Lateral tarsal strip	Lateral tarsal conjunctiva
80	Male	Lateral tarsal strip	Lateral tarsal conjunctiva
73	Male	Lateral tarsal strip	Lateral tarsal conjunctiva

Figure 1.

 

Phase-contrast micrograph of a representative primary culture of goblet cells growing from human conjunctival tissue in RPMI medium supplemented with 10% FBS, 2 mM l-glutamine, and 100 μg/mL penicillin-streptomycin. Magnification, ×200.

Figure 2.

 

Photomicrographs of human goblet cells in culture and in vivo after histochemical staining with AB/PAS. Goblet cells stained intensely with AB/PAS indicating the presence of both acidic (blue) and neutral (pink) glycoproteins associated with both cultured cells (A) and human conjunctival tissue (B). Cultured goblet cells also had mucin strewn over their surface. Magnification, ×200.

Figure 3.

 

Photomicrographs of human goblet cells in culture and in vivo after histochemical staining with the lectin HPA. Histochemistry confirmed that HPA, which recognizes l-galactosamine within the secretory vesicles of goblet cells, was present to various degrees in both cultured cells (A) and in tissue sections of human conjunctiva (B). Magnification, ×200.

Figure 4.

 

Photomicrographs of human goblet cells in culture and in vivo. Human goblet cells in culture (A) and in conjunctival tissue sections (B) stained intensely for cytokeratin-7, an intermediate filament associated specifically with conjunctival goblet cells, but not with other types of conjunctival epithelial cells. Human goblet cells in culture (C) did not stain for cytokeratin-4. Magnification: (A, C) ×600; (B) ×200.

Figure 5.

 

Photomicrographs of human goblet cells in culture and in vivo. Immunolocalization of HPA lectin and cytokeratin-7 in human goblet cells. Double labeling for these markers indicated that the presence of HPA (shown in red) and cytokeratin-7 (shown in green) in the same cell both in culture (A) and in vivo (B). Magnification: ×200.

Figure 6.

 

Photomicrographs of human goblet cells in culture and in vivo. Immunolocalization of MUC5AC indicated that human goblet cells in culture (A, B) and those located in conjunctival tissue were positive for this mucin (C). In cultured goblet cells, MUC5AC was specifically localized within secretory granules, shown as punctate staining in (A) and (B). Magnification: (A, C) ×200; (B) ×600.

Figure 7.

 

Presence of MUC5AC in cultured goblet cells. Ethidium bromide-stained gel of RT-PCR performed with either MUC5AC or G3PDH (housekeeping gene) primers on RNA isolated from human conjunctiva (Conj) from two separate patients as well as cultured goblet cells. MW, molecular weight. PCR was performed without reverse transcriptase—that is, with RNA alone (negative control 1) or with water replacing cDNA (negative control 2).

Figure 8.

 

Transmission electron micrograph of a representative goblet cell grown from human conjunctiva and sectioned en face, showing a rounded but slightly elongated body, a small nucleus, and many storage vesicles. Magnification, ×6000.

Figure 9.

 

Western blot analysis demonstrating the presence of EGF receptors on cultured human goblet cells. EGFR (A) is localized in both rat (lane 1) and in human goblet cells (lane 2). The human epidermoid carcinoma cell line, A431 pretreated with EGF and known to contain large amounts of EGF receptors, is the positive control (lane 3). ErbB2 (B) is present in both rat (lane 1) and human (lane 2) cultured goblet cells. ErbB2 in A431 cells, a positive control, is shown in lane 3. ErbB3 (C) is also present in both rat (lane 2) and in human (lane 3) goblet cells. The human breast carcinoma cell line, which contains ErbB3 receptors, a positive control, is also shown (lane 1).

Figure 10.

 

Effect of EGF on goblet cell proliferation. (A) Proliferation was determined in triplicate with a nonradioactive, colorimetric MTT assay. Data shown are obtained in goblet cell-enriched cultures of five patients and are expressed as the mean ± SEM. Significance (P ≤ 0.5) was determined by Student’s t-test for paired comparisons. (B) Localization of Ki-67 nuclear antigen in cultured human goblet cells that had been exposed to 20 ng/mL EGF for 24 hours. Arrows: Ki67-positive cells. Magnification, ×200. (C) Quantification of Ki67 staining. Total number of cells and cells labeled with Ki67 were counted. Cells of two patients were counted, and data are expressed as a percentage of the total. *Statistical significance compared with no EGF.

The authors thank Patricia Pearson, LiLi Chen, and Rachel Tarko for expert technical assistance; J. Wayne Streilein and Michael Young for the use of their respective tissue culture and microscopy facilities.