Monoclonal Antibody (3G5)–Defined Ganglioside: Cell Surface Marker of Corneal Keratocytes

Abstract

purpose. To evaluate the anti-ganglioside monoclonal antibody 3G5 as a marker of corneal keratocytes.

methods. 3G5 expression on keratocytes was investigated by immunofluorescence microscopy. Studies were performed on frozen sections of normal human, bovine, porcine, rabbit, rat, and mouse corneas and on repairing rabbit cornea. In vitro studies were performed on tissue-cultured human, bovine, porcine, mouse, and rabbit keratocytes.

results. 3G5 stained frozen sections of human, bovine, porcine, rat, and rabbit cornea but not mouse cornea and the staining pattern followed the distribution of stromal keratocytes but did not stain epithelium or endothelium. Subconfluent human and bovine keratocyte cultures were 3G5 negative. Almost 100% of the human and bovine cells that were maintained at confluence without replacement of serum-containing culture medium for 2 weeks became 3G5 positive. The 3G5 antigen was constitutively expressed on cultured rabbit and porcine keratocytes under all conditions examined. Mouse keratocyte cultures did not express 3G5. The 3G5 antigen was not present on myofibroblastic cells in the repairing area of a full-thickness wound in rabbit cornea that had been healing for 20 days. The area surrounding the healing wound expressed 3G5 antigen in an altered distribution, whereas 3G5 antigen was distributed in the expected pattern in areas that were distant from the wound. When rabbit keratocytes were induced to express the myofibroblast marker α-smooth muscle actin by treatment with TGFβ1 in vitro, the pattern of 3G5 staining was altered.

conclusions. The 3G5 antigen is a useful marker for the identification of corneal keratocytes and for documenting their response to environmental stimuli associated with wound repair.

The keratocyte is a highly specialized cell that is sandwiched between orthogonally arranged layers of collagen lamellae in the corneal stroma. The keratocyte plays important roles in corneal wound healing and tissue repair and is known to undergo phenotypic transformations in wounds, ¹ due to the influence of growth factors and cytokines. These phenotypic subtypes are defined by their morphology and anatomic location and have been extensively studied in tissue-culture models. However, there is a paucity of useful biochemical markers for identification of these cells in situ or in cell culture. 

It has been reported that corneal stromal keratocytes, like pericytes of the microvasculature, can promote vascular endothelial cell differentiation in vitro, ² suggesting that keratocytes may have pericyte-like properties. We have demonstrated that the monoclonal antibody 3G5 is a marker of microvascular pericytes ³ in vivo and in vitro. We report herein the characterization of 3G5 antigen expression in keratocytes and its utility as a marker of keratocytes in vivo and in vitro. The 3G5 antigen has been shown to be a ganglioside that migrates as a monosialo-ganglioside with mobility between the gangliosides G_M1 and G_M2 on thin-layer chromatograms, ^{3 4 5} except in the brain, where additional more complex gangliosides also bind the antibody. ⁶ Ganglioside appearance on cells is often associated with processes of differentiation and is associated with cell adhesive behavior. ^{7 8} Experimentally increasing the mol percent of gangliosides in plasma membranes has been shown to increase curvature of these membranes. ⁹ This may be an indication of the function of 3G5 antigen as it is principally expressed on cell types that are characterized by membrane processes (i.e., neurons, pericytes, glomerular podocytes). Thus, 3G5 may be involved in the regulation and maintenance of cell shape. 

Materials and Methods

Studies on Human Tissues

All studies using human tissues were in accordance with the tenets of the Declaration of Helsinki and in accordance with the policies of the institutional review board for human subjects. 

Studies on Animal Tissues

All procedures were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and in accordance with the policies of the institutional animal care and use committee. 

Antibodies

Monoclonal antibody 3G5 was prepared as an ascites fluid, as previously described. ³ The hybridoma cell line has been deposited at the American Type Culture Collection (CRL-1814; Manassas, VA). FITC-conjugated monoclonal antibody to smooth muscle actin was purchased from Sigma-Aldrich (St. Louis, MO). The MOPC 104E IgM myeloma protein was purchased from Sigma-Aldrich. 

Preparation of Frozen Sections of Cornea

Normal (transplant quality) human donor corneas (donor age, 39–70 years) were obtained from the New England Eye and Tissue Transplant bank. Bovine and porcine eyes were obtained from local abattoirs, rabbit eyes were obtained from PelFreez Co. (Rogers, AR), and Wistar rats and CD1 strain mice were obtained from Charles River Laboratories (Wilmington, MA). Corneas were snap frozen in liquid nitrogen and then embedded in optimal cutting temperature (OCT) compound (Tissue Tek, Elkhart, IN). Five to 8-μm thick cross sections of corneal tissue were cut with a cryotome (model 3050; Leica, Deerfield, IL). The sections were mounted on glass slides, air dried, and stored at −20°C until used for immunostaining. 

Rabbit Model of Corneal Wound Healing

New Zealand White rabbits (2.5 kg; Charles River Laboratories) were used in all experiments. General anesthesia was administered, and corneas were treated with topical application of 0.5% proparacaine hydrochloride drops (Squibb, Princeton, NJ). Penetrating keratectomy was performed by excising a central corneal button using a 2.0-mm trephine, according to the method previously described. ^{10 11}  

Immunofluorescent Staining of Tissue Sections

Slides were recovered from the −20°C freezer and allowed to equilibrate to room temperature. A wax pencil was then used to circle each tissue section on the slide. Fifty microliters of PBS containing 1% BSA and 3G5 ascites fluid at 1:100 dilution was then added. The sections were allowed to incubate for 30 minutes at room temperature. The sections were then washed by gently pouring PBS over the slide, being careful to pour it onto the slide where no section was present and allowing the PBS to flow over the sections. The slide was dried carefully before incubating each section with 50 μL PBS containing 1% BSA and goat anti-mouse IgM-rhodamine isothiocyanate (RITC; Cappel/ICN, Costa Mesa, CA) at 1:100 dilution for 30 minutes, covered in foil, at room temperature. The slide was then mounted with mounting fluid containing 5 μg/mL of Hoechst 33258 (Sigma-Aldrich). 

Negative control antibody for 3G5 was the MOPC 104E IgM myeloma protein, which has no known antigen specificity. Immunofluorescent staining with MOPC 104E was performed as described for 3G5. 

Isolation of Human, Bovine, Porcine, and Murine Corneal Keratocytes

Corneal tissue was minced and cultured with 10% fetal calf serum (FCS) in DMEM in a humidified incubator with an atmosphere of 10% CO₂ in air. Under these culture conditions, keratocytes can be seen migrating out of the whole corneal explants. Endothelial cells are not mitotically active under these conditions and are lost on passaging. Epithelial cells grow poorly under these conditions and are lost by dilution on passaging. By the second passage, all cells have a morphology that is consistent with the keratocyte. On reaching confluence, third-passage explant cell populations were cryopreserved under liquid nitrogen in growth medium containing 10% dimethyl sulfoxide (DMSO). Cells were recovered from liquid nitrogen and cultured in DMEM+10% FCS for immunostaining experiments. 

For serum-free isolation and maintenance of human keratocytes, the procedure described in the following section for rabbit keratocyte isolation was used. 

Isolation and Propagation of Rabbit Corneal Keratocytes

Corneas from New Zealand White rabbits were dissected from the endothelial layer, and the stromal cells were isolated as previously reported. Central corneal tissue was excised with a 9-mm trephine and placed in 0.25% trypsin in PBS (Life Technologies-Gibco, Gaithersburg, MD) and incubated overnight at 4°C. The epithelial layer was then gently scraped off the stromal button with a scalpel. The stromal buttons were then cut into small pieces and dissociated with 5 mg/mL collagenase (Worthington Biochemicals, Lakewood, NJ) in MEM containing 10% FCS. After enzymatic digestion, the stromal cells were pelleted by centrifugation, resuspended in MEM+10% FCS and plated for subculture. For serum-free isolation, FCS was omitted from the collagenase digestion step and the plating medium. 

Induction of Myofibroblast Phenotype in Rabbit Keratocytes In Vitro

Passaged keratocytes were plated in 16-chamber glass slides (Fisher Scientific-Nunc, Pittsburgh, PA) to approximately 25% of confluence and treated with 10 ng/mL TGFβ1 for 3 days. The cells were then fixed in 10% buffered formalin (Sigma-Aldrich) for 1 hour at 4°C and immunostained with monoclonal antibody 3G5 and monoclonal antibody to smooth muscle actin. Monoclonal antibody 3G5 was used at a dilution of 1:100 in PBS-1%BSA and was applied to the cells for 30 minutes at room temperature, after which unbound antibody was washed off with three changes of PBS. An FITC-conjugated monoclonal antibody to smooth muscle actin (Sigma-Aldrich) at a dilution of 1:500 was then applied to the cells simultaneously with goat anti-mouse IgM-RITC (Cappel/ICN) at 1:100 dilution for 30 minutes at room temperature. The cells were washed as before, mounted under a coverslip, and viewed with a fluorescence microscope (Eclipse E400; Nikon, Tokyo, Japan). Images were recorded with a digital camera (Spot; Diagnostic Instruments Inc., Sterling Heights, MI). 

Time Course Study of 3G5 Reexpression

Keratocytes were seeded into six-well cluster dishes in DMEM containing 10% FCS. The medium was not replaced for the whole duration of the experiment. Duplicate wells were trypsinized at various time points and stained for 3G5 immunofluorescence as will be described later. Cells in cluster wells were first washed with PBS without calcium once and trypsin-EDTA was then added to the well/flask and placed in a 37°C incubator for a few minutes until the cells lifted off. An equal amount of DMEM with glutamine and 10% FBS was added. The contents were then centrifuged at 1000 rpm for 10 minutes. The supernatant was subsequently taken off, and the pellet was resuspended in 1 mL of prechilled (4°C) PBS. Ten microliters of this mixture was then taken out for a cell density count. Another 4 mL of PBS was added and centrifuged again for 10 minutes at 1000 rpm. The supernatant was again taken off, and the pellet was resuspended in 200 μL of prechilled PBS containing 0.1% BSA, 0.01% sodium azide (PBS-SA), and monoclonal antibody 3G5 at a dilution of 1:100. The suspension was allowed to incubate on ice for 30 minutes. Immediately after the incubation period, 5 mL of prechilled PBS-SA was added and then centrifuged for 10 minutes at 1000 rpm. The supernatant was then taken off, and the pellet was resuspended in 200 μL of prechilled PBS-SA containing goat anti-mouse IgM-RITC (Cappel/ICN) at 1:100 dilution and incubated for 30 minutes on ice, while covered with foil. After the incubation, the cells were washed twice, as just described. The final pellet was then resuspended in 10% buffered formalin (Sigma-Aldrich) with 2 mM Ca and 0.01% sodium azide. The suspension was kept chilled for 1 hour before 20 μL was mounted on a slide. Cell counts were obtained by random sampling of the slide. All cells were counted under phase-contrast optics, and the same field was counted under rhodamine fluorescence optics until 100, or more, fluorescent cells were scored. 

Results

Frozen cross-sections of human, rabbit, rat, mouse, bovine, and porcine corneas were stained with the 3G5 monoclonal antibody by indirect immunofluorescence. Staining in the stroma was observed in human, rabbit, rat, bovine, and porcine sections (Table 1) . No staining was seen in mouse cornea (Table 1) . The appearance of the 3G5 staining of keratocytes in human cornea sections is shown in Figure 1 . No staining of corneal epithelium or endothelium was seen in any of the above species tested (Fig. 1 , Table 1 ). 

In healing full-thickness corneal wounds, keratocytes undergo a phenotypic transformation, acquiring a fibroblastic morphology. Some of these express the smooth muscle actin marker and are often referred to as “myofibroblasts.” To determine whether 3G5 binds to corneal keratocytes once they have undergone this transformation, frozen sections of 20-day healing, full-thickness corneal wounds were stained with 3G5 and FITC-labeled monoclonal antibody to smooth muscle actin (Fig. 2) . Staining with the antibody to smooth muscle actin was seen in a subset of repair cells, as expected (Fig. 2C) . However, 3G5 did not stain any of the repair cells, although 3G5 staining was apparent in areas adjacent to the wound in a pattern that was different from that seen in normal corneal sections (compare Fig. 2B with Fig. 1 ). Corneal tissue that was distal to the wound stained with 3G5 in a pattern that was indistinguishable from normal cornea (not shown). In Figures 2B and 2C the epithelium and endothelium appeared to be stained with 3G5 and anti-smooth muscle actin, This appearance was due to autofluorescence, as fluorescence of the epithelium and endothelium can be seen in unstained frozen sections. 

The presence of 3G5-antigen in human and rabbit keratocytes in vitro was also determined by immunofluorescence analysis (Fig. 3) . Human keratocytes that had been isolated and propagated in the presence of serum did not stain with 3G5 antibody (Fig. 3A) . Human keratocytes were also found to be 3G5 negative when isolated and maintained in serum-free conditions (not shown). However, when we maintained confluent cultures of serum-exposed keratocytes for a week without replacement of the growth medium, 3G5-positive cells were observed in the culture (Fig. 3B) . In contrast, 3G5 antigen was present on rabbit keratocytes, regardless of whether they were isolated and maintained in serum-containing medium (Fig. 4A) or serum-free medium (Fig. 3C) . Similarly, porcine keratocytes were found to express 3G5 antigen constitutively in vitro, whereas bovine keratocytes were negative for 3G5-antigen expression (Table 1) . The 3G5 antigen could also be induced in cultured bovine cells by not refeeding the cells for 2 weeks. The expression of 3G5 antigen by rabbit keratocytes that had been transformed into myofibroblasts in vitro by addition of TGFβ1 was similarly investigated (Fig. 4) . In the absence of TGFβ1, rabbit keratocytes were positive for 3G5 antigen (Fig. 4A) but not smooth muscle actin (Fig. 4B) . Rabbit keratocytes grown in the presence of TGFβ1 were negative for 3G5 antigen (Fig. 4C) and positive for smooth muscle actin (Fig. 4D) . 

As the 3G5 antigen appeared to be inducible in human keratocyte cultures, the time course of antigen expression was investigated, comparing human to rabbit keratocytes in vitro (Fig. 5) . Human keratocyte cultures were 3G5 negative up to 3 days after plating in serum-containing medium. However, at 5 days after plating, 3G5-positive cells were detected, and the proportion of 3G5-positive cells rose steadily, reaching approximately 100% by day 16 after plating. In contrast, rabbit keratocytes were 100% 3G5 positive at plating, and 3G5 antigen was constitutively expressed by 100% of these cells for the duration of the study (14 days). Bovine keratocytes behaved in a manner similar to human keratocytes in vitro. They were 3G5 negative when subconfluent, and at 5 days without feeding they were 0.06% 3G5 positive and at 15 days were 100% 3G5 positive. Porcine keratocytes in vitro behaved similar to the rabbit keratocytes, in that they constitutively expressed the 3G5 antigen. 

Discussion

We have demonstrated that the 3G5 antigen is a novel cell surface marker of keratocytes. The cell-type specificity is contextual, in that within the cornea it is keratocyte specific. The 3G5 antigen has been shown to be present on several other cell types, including capillary pericytes, ^{3 12} renal glomerular podocytes, ⁴ neurons, a subpopulation of T lymphocytes, ⁵ and pancreatic islet cells. ¹³ The range of cell types that express the 3G5 antigen does not follow any pattern of lineage and is not simply related to differentiation of cells. The expression of the 3G5 antigen is probably dictated principally by its function, which remains unknown. 

Experimentally increasing the mol percent of gangliosides in plasma membranes has been shown to increase curvature of these membranes. ⁹ This may be an indication of the function of the 3G5 ganglioside, as it is principally expressed on cell types that are characterized by membrane processes (i.e., neurons, pericytes, and glomerular podocytes). Thus, 3G5 may be involved in the regulation and maintenance of cell shape. This hypothesis would explain our observations of 3G5 expression in the cornea. In cross sections of corneal tissue, 3G5-positive keratocytes had a spindlelike morphology with membrane processes. However, keratocyte-derived myofibroblasts in vivo and in vitro had a flattened/spread morphology, without processes and did not express the 3G5 ganglioside. 

Why the 3G5 antigen is not present on mouse corneal keratocytes is currently inexplicable and represents an unfortunate limitation of the marker, as its nonexpression in mouse cornea rules out studies using mouse models. 3G5 antigen is expressed in some mouse tissues that we have studied, such as Con-A–stimulated T lymphocytes (Nayak RC, unpublished observation). Although our survey of mouse tissues for 3G5 expression was by no means exhaustive, it was sufficient to conclude that the antigen is not expressed in a lineage-specific manner across species. It was surprising to observe that rabbit keratocytes in tissue culture expressed the 3G5 antigen constitutively, regardless of whether serum was present in the growth medium, whereas human keratocytes did not express the 3G5 antigen, regardless of whether the human cells were isolated and maintained in the presence or absence of serum. The induction of 3G5 antigen expression by starvation of cells in serum-containing medium by not refeeding cultures suggests that a serum factor that negatively regulates 3G5 antigen expression is depleted from the medium by the cells over time. This tissue culture system may therefore be a useful model for studying factors controlling the redifferentiation pathway of 3G5 expression in human keratocytes. 

Species differences in the regulation of 3G5 expression suggest that there are differences in mechanisms that control phenotypic transitions and, by inference, species-specific differences in the molecular mechanisms involved in corneal wound healing. Consequently, all animal models of wound healing may not be equivalent or equally appropriate for investigating drug effects on wound healing, as the presence of the molecular target of the drug may depend on the species being investigated. 

The current model of the stromal reaction to injury encompasses a keratocyte transition from a quiescent state to an activated, migratory, collagen-secreting cell phenotype that is often referred to as a corneal fibroblast. ¹ Tissue-cultured keratocytes are generally thought of as being corneal fibroblasts. ¹ However, the constitutive expression of 3G5 by rabbit and porcine tissue cultured keratocytes indicate that they retain some phenotypic traits of the quiescent keratocyte (i.e., 3G5 expression). Furthermore, the 3G5 antigen is not expressed by human skin fibroblasts in vivo or in vitro. ^{12 14} Consequently, consideration of the keratocyte and corneal fibroblast as being distinctly different cell types may be conceptually inappropriate, as the corneal fibroblast and myofibroblast may be more appropriately considered products of the functional and phenotypic plasticity of the corneal keratocyte. 

Keratocan expression has been reported to be a specific marker of corneal keratocytes. ^{15 16} Keratocan is a secreted product, however, and in tissue culture systems anti-keratocan antibodies may be difficult to use as a marker because subconfluent cells migrate in the culture dish. The 3G5 antigen is likely to be a very useful marker of keratocytes as it is a protease-resistant cell surface antigen and does not suffer from the potential caveat of the cells becoming physically dislocated from the biochemical marker. Another potential cell surface marker of human keratocytes, CD34, was recently reported. ¹⁷ In this study, we used only human cornea sections, and consequently it is not known whether tissue-cultured keratocytes express CD34. The CD34 antigen is a protein, that is a potential disadvantage for use in methodologies such as cell sorting, which is performed on trypsinized cells. Because the 3G5 antigen is a ganglioside (acidic glycolipid) it is trypsin insensitive, and the 3G5 antibody binds well to unfixed trypsinized cells, making it well suited to applications such as fluorescence-activated sorting of living cells. ³  

There is a clear need for a variety of keratocyte markers to understand the factors that control phenotypic plasticity in the keratocyte. When the corneal stroma is damaged, keratocytes at the wound edge become activated, migrate into the damaged area, and acquire a fibroblastic phenotype. As wound healing progresses, many of these cells transiently acquire the myofibroblast marker smooth muscle actin, ^{18 19} whose expression is regulated by TGFβ2. ²⁰ As wound healing approaches completion, a partial reversal of this differentiation pathway occurs as cells return to the quiescent state, but it remains to be learned whether the corneal keratocyte can ever fully return to its uninjured state. ¹ Tissue culture of isolated corneal keratocytes in the presence of serum seems to mimic the transition to a fibroblastic phenotype, which can be prevented or reversed to some degree by isolation and culture in serum-free conditions. ^{21 22}  

Incomplete redifferentiation toward the ontogenetic keratocyte phenotype may be at least partially responsible for the persistence of corneal haze after wound healing. ^{1 23 24 25 26 27} The availability of a panel of relevant markers is, therefore, of fundamental importance to understanding the factors that regulate keratocyte differentiation and how that relates to the problem of corneal wound healing and the development of corneal haze. It is also crucial to enable development of a rational approach to tissue engineering of an artificial cornea for transplantation. 

 Supported by the Massachusetts Lions Eye Research Fund Inc., National Eye Institute Grants P30 EY13078, EY09828, NIH Grant EY014406 (PJF-N), and a Research to Prevent Blindness (RPB) award to the Department of Ophthalmology, University of Arizona College of Medicine. MEF was a Research to Prevent Blindness Stein professor and currently holds the Walter G. Ross Chair in Ophthalmic Research.

 Submitted for publication March 12, 2003; revised October 7, 2003; accepted November 17, 2003.

 Disclosure: B.M. Stramer, None; M.G.K. Kwok, None; P.J. Farthing-Nayak, None; J.-C. Jung, None; M.E. Fini, None; R.C. Nayak, 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.

 

Species-Specific Expression of 3G5 Antigen by Corneal Cells in Frozen Sections and in Tissue Cultured Keratocytes

Species	3G5 Antigen Expression
		Frozen Section					Tissue Culture Keratocyte
		Epithelium	Keratocyte	Endothelium
Human	−	+	−	−*
Rabbit	−	+	−	+
Rat	−	+	−	ND
Mouse	−	−	−	−
Bovine	−	+	−	−*
Porcine	−	+	−	+

*  3G5-antigen expression is inducible. ND, not done.

Figure 1.

 

3G5 antibody binding to a frozen section of human cornea. The section was stained with 3G5 (red) and Hoechst stain (blue). Ep, epithelium; En, endothelium. Scale bar, 112 μm.

Figure 2.

 

Immunofluorescent detection of 3G5 and anti-smooth muscle actin antibody binding in a healing, full-thickness corneal wound in rabbits. (A) Photomicrograph of Hoechst-stained healing corneal wound 20 days after wounding. (B) Same field as in (A) stained with 3G5 antibody. (C) Same field as in (A) showing repair cells in wound stained with antibody to smooth muscle actin. Scale bar, 100 μm.

Figure 3.

 

Immunofluorescent detection of 3G5 antibody binding to human and rabbit keratocytes in vitro. (A) Subconfluent human keratocytes. (B) Hyperconfluent, starved human keratocytes. (C) Rabbit keratocytes isolated and maintained in serum-free conditions. Scale bar, 20 μm.

Figure 4.

 

3G5 and anti-smooth muscle actin expression in TGFβ1-treated rabbit keratocytes. (A, B) No TGFβ1 treatment; (C, D) TGFβ1 treated; (A, C) 3G5 Immunofluorescence; (B, D) anti-smooth muscle actin immunofluorescence. Scale bar, 50 μm.

Figure 5.

 

3G5 antigen expression in human and rabbit keratocyte cultures.