Apoptosis in Shed Human Corneal Cells

Abstract

purpose. To determine whether shear forces applied to the corneal epithelium by the repeated insertion and removal of a hydrogel contact lens alter the size and number of cells removed and to determine the contribution of apoptosis to this process.

methods. Human corneal cells were collected from eight healthy subjects by sequential contact lens cytology (20 lens insertions and removals). Collected cells were stained with acridine orange for counting and measurement of cell size. In a separate experiment, collected cells were fixed and stained with TdT-mediated dUTP nick-end labeling (TUNEL) or labeled immediately after collection using annexin V. Hoechst stain and propidium iodide (PI) were used as nuclear counterstains. The proportion of cells labeled with acridine orange, TUNEL, and annexin V was quantified by fluorescence microscopy.

results. The number of cells increased in later collections, and cells were smaller. The mean number of positively stained cells using TUNEL was 57%. Annexin V labeling on unfixed fresh samples showed a mean of 64%, with an increase in later collections. Apoptotic bodies were observed in very few cells. In most cells the nucleus and cytoplasmic membrane were intact. Structures were observed in which nuclei were missing (Hoechst negative) but in which cytoplasm had the size and appearance of whole, nucleated cells. These structures (cell ghosts) increased in number along with the increase in nucleated cells in later collections.

conclusions. The sequential removal of a soft contact lens caused a progressive increase in the number of cells collected from the surface and a progressive decrease in their size. The majority of nucleated cells removed by a contact lens were apoptotic in the sense of being positively labeled by TUNEL and annexin V. Morphologically they differed from classically apoptotic cells, in that cells showed an intact nuclear structure and no discernible apoptotic bodies. They could represent a last stage in a pathway of cell differentiation in which frictional forces induced by the removal of the contact lens activate the apoptotic program and cause the cell to be shed. There is also a pathway in which cells lose their nuclei before leaving the epithelial surface.

Cell shedding and cell death in the corneal epithelium have been explained by several theories, including necrosis and sloughing of cellular debris. ¹ Recent studies have indicated that cell shedding from the corneal surface may be, in part, an apoptotic process. ²  

Blinking applies a shear force to the corneal surface. Friction on the surface of an excised rabbit cornea causes more cells to be released, and it has been suggested that these are apoptotic cells. ² In our study in humans, we used sequential contact lens cytology (CLC) to examine whether the repeated insertion and removal of soft contact lenses leads to changes similar to those reported in the rabbit. Apoptosis was identified using TUNEL ³ and annexin V staining. ^{4 5}  

Methods

Contact Lens Cytology

A −6.0-D hydrogel contact lens (water content 55%, New Vues; Ciba Vision, Duluth, GA) was placed on the cornea and allowed to stabilize for 2 minutes. The lens was removed without decentration onto the conjunctiva. During the period from 9:30 A.M. to 12 noon, the lens was inserted and removed 20 times. The procedure was performed in eight healthy subjects who were not habitual contact lens wearers. Informed consent was obtained from each subject, and the procedures followed the guidelines of the Declaration of Helsinki. After each removal, the contact lens was draped over the convex end of a glass test tube with the front surface of the lens against the glass. Adherent cells were removed by rinsing with approximately 10 ml of solution. The irrigations from the first four contact lens removals ^{1 2 3 4} were collected into one beaker (beaker A). Irrigations from the next four removals ^{5 6 7 8} were collected into beaker B, and so on, up to 20 removals. Cells from each subject were collected in five beakers (A through E). Cells were stained using one of three protocols. 

Acridine Orange.

For counting and measurement of size, suspensions from the five beakers were stained with acridine orange. The contact lens was rinsed with basic tear solution (BTS; 304 mOsm/kg, pH 7.4), ² , containing 116.3 mM NaCl, 18.8 mM KCl, 0.4 mM CaCl₂0.2H₂O, 0.6 mM MgCl₂0.6H₂O, 0.08 mM NaH₂PO₄, and 26.0 mM NaHCO₃. The cell suspension was stained with 10⁻⁴ M acridine orange solution (Sigma, St. Louis, MO), prepared as described earlier. ² Stained cell suspensions were filtered through a 5.0-μm pore diameter polycarbonate filter disc (Poretics, Livermore, CA). The filter disc containing stained cells was transferred to a microscope slide and coverslipped with BTS. Cells were collected from five subjects. 

TUNEL.

The contact lens was rinsed with 269 mOsm/kg phosphate-buffered saline (PBS; pH 7.4), containing 120 mM NaCl, 2.7 mM KCl, and 10 mM phosphate buffer (Sigma). The cell suspension in each beaker was divided into four equal aliquots. One aliquot from beaker D was used for the positive control, and one from beaker E for the negative control. 

The cell suspensions were centrifuged (3000 rpm, 1500g, 15 minutes) and stored at +4°C. Cells were fixed in paraformaldehyde 4% (Fischer Chemical, Springfield, NJ) in PBS (pH 7.4) centrifuged (2500 rpm, 1000g, 10 minutes), air dried onto glass slides, and stained with TUNEL (Boehringer–Mannheim, Mannheim, Germany). Specimens were counterstained with Hoechst 33342 (Sigma). In the positive control, collected cells were incubated with DNase I (Boehringer–Mannheim) for 15 minutes at room temperature, and in the negative control cells were incubated with fluorescein-tagged dNTP without TdT enzyme. Cells were collected from three subjects. 

Annexin V.

PBS (269 mOsm/kg; pH 7.4) containing 120 mM NaCl, 2.7 mM KCl, and 10 mM phosphate buffer, was used to rinse the contact lens for collection of cells. Annexin V staining was performed using Annexin-V-FLUOS (Boehringer–Mannheim) containing annexin V labeled with fluorescein isothiocyanate (FITC) and mixed propidium iodide (PI) in HEPES buffer (10 mM HEPES-NaOH [pH 7.4], and 140 mM NaCl, 5 mM CaCl₂). Hoechst 33342 was added to the solution. The solutions were then placed on the filter disc containing filtered cells, coverslipped, and immediately viewed by fluorescence microscopy. A 2.5-ml sample from beaker D was used for the negative control by adding FITC, PI, and Hoechst without annexin V to the sample. Cells were collected from three subjects. 

Fluorescence Microscopy

Examination and cell counting were performed at 400× (40× objective, 10× eyepiece) using a microscope (Aristoplan microscope; Wild Leitz, Wetzlar, Germany) equipped with appropriate filter combinations. The longest dimension of a cell was measured using a reticle in the eyepiece of the microscope. 

Results

Acridine Orange

As the number of insertions and removals increased the mean number of cells increased (Wilcoxon signed rank test, n = 5, P < 0.05), and the mean cell size decreased (P < 0.05; Fig. 1 ). 

TUNEL

TUNEL green intensely fluorescence–positive cells were found in all collections (Fig. 2A ). Their nuclei were labeled with Hoechst (Fig. 2B) . Some cells showed a weaker green fluorescence, but their nuclei stained strongly with Hoechst. These were regarded as TUNEL negative during cell counting. 

The mean TUNEL-positive percentages for the three subjects were 71% ± 10%, 57% ± 9%, and 43% ± 9%, giving a grand mean of 57% of the total for all shed cells in all collections. The percentages in individual beakers are shown in Figure 3A . Almost all TUNEL-positive cells had intact nuclei. Only a few cells with nuclear budding or fragmentation of the nucleus were observed (Figs. 2C 2D) . Most cells had a polygonal shape and occurred singly. Occasionally, groups of two or more attached cells were observed. In the negative control no TUNEL-positive nuclei could be detected, whereas in the positive control all nuclei were TUNEL positive. 

Annexin V

The majority of shed cells (64% on average) showed bright green annexin-positive staining of the cytoplasmic membranes (Fig. 2E) . They were rounded in shape and not sharply polygonal. There was an increase in the percentage (Fig. 3B) and absolute number (Fig. 4B ) of annexin-positive cells in later collections. On average, a total of 64% were labeled Hoechst positive, PI positive, and annexin V positive (H+ P+ A+). These cells had intact nuclei and cytoplasm, although some cells showed marginated condensation of chromatin. On average 23% of nucleated cells were H+ P+ A−. An annexin negative cell is shown in Figure 2F (lower cell) along with an annexin-positive cell (upper cell). These H+ P+ A− cells had not turned annexin positive when observed 24 hours later. Only a few cells (2%) were H+ P− A+ (Figs. 2G 2H) . These cells had fragmented nuclei and, in most cells, the annexin stain was not as bright as in PI-positive cells. 

Cell Ghosts

Structures resembling cells, but without nuclei, were observed in all collections. On many occasions, they were as abundant as nucleated cells (Fig. 3C) . These cell-like structures incorporated labels diffusely but were deemed H− P− A− and were not included in the nucleated cell count. A few cell ghosts were H− P− A+, making up 8% of the nonnucleated structures. In all subjects the highest percentage of H− P− A+–labeled structures occurred in beaker A (Fig. 3B ; Cell Ghosts). The increase in cell ghosts in subsequent beakers (Fig. 3C) was due almost entirely to H− P− A− structures. Measurement of a sample of cell ghosts from one subject showed an average longest dimension of 36.2 ± 7.6 μm (n = 43), whereas the longest dimension of nucleated cells in the same sample was 35.4 ± 8.5 μm (n = 103). The difference between these means was not statistically significant (t = 0.48, P > 0.05, unpaired t-test). 

Discussion

It has been proposed that cells can be shed from the corneal surface by apoptosis. ⁶ This conclusion was based on the presence in rabbits of occasional TUNEL-positive cells on the surface of the normal epithelium, as well as in collections of cells irrigated from the precorneal film. It was suggested that there are two discrete categories of shed cells: apoptotic cells that are TUNEL positive, and terminally differentiated cells that are ethidium positive and calcein negative. 

The sequential insertion and removal of a soft contact lens caused a progressive increase in the number of cells removed from the surface along with a progressive decrease in their size (Fig. 1) . The decrease in cell size could be explained by the collection of smaller subsurface cells, yet it is unlikely that enough surface cells could have been removed to expose underlying cells as small as 10 μm. The increase in cell number could be explained by the loss of essential lubricants as the lens is removed repeatedly from the cornea, with the increased friction causing removal of more cells. Neither of these is an explanation, however, for both the increased number of cells and the reduction in cell size. We therefore pursued the possibility that the process was due to an apoptotic mechanism, as was suggested in another study. ²  

Approximately half the nucleated cells removed from the surface were TUNEL positive (Fig. 3A) . A puzzling feature of these cells was that there was no morphologic evidence of classic apoptosis in the form of blebs, and only a few showed fragmented nuclei (Figs. 2C 2D) . In the annexin assay, more than half the nucleated cells were positive (Fig. 3B) , and again the typical appearance of cells (Fig. 2E) gave no indication of blebbing of the cell membrane characteristic of classic apoptosis. Only 2% of cells had labeling characteristics that suggested classically apoptotic cells (H+ P− A+). 

How can the difference be resolved between the positive evidence for apoptosis from TUNEL and annexin and the almost total absence of morphologic changes characteristic of classic apoptosis? One possibility is that the assays are providing a misleading number of false-positive results. For example, the TUNEL assay has been reported to label cells undergoing necrosis. ⁷ However, in our study an increase in permeability, such as occurs in necrosis, did not in itself cause cells to be labeled with annexin. If H+ P+ A+ cells were positive to annexin because of membrane disruption, then H+ P+ A− cells should have been annexin positive for the same reason. Yet, 23% of nucleated cells in our study were in the H+ P+ A− category. 

The finding of TUNEL-positive and H+ P+ A+ cells suggests that the molecular changes associated with apoptosis occur in morphologically intact cells. This has been reported in the epithelium of the small intestine ³ and in Henle’s layer of hair follicles, ⁸ both of which are tissues with a high turnover rate similar to that in the corneal epithelium. In epidermal keratinocytes the apoptotic program is arrested before fragmentation of the cell. ⁹ Because the appearance of this apoptotic cell type is different from classic apoptotic cells, we used the term nonclassic apoptosis in this study. Mechanical forces may activate this nonclassic apoptotic program through, for example, tension and compression of the cytoskeleton, which have been shown to be capable of communicating extracellular forces to nuclear DNA. ¹⁰ Also, apoptosis may be induced by disruption of cell–matrix interactions. ¹¹  

An additional pathway for cell loss was found that has not received attention in previous publications. These are nonnucleated (Hoechst negative) cell-like structures, referred to here as cell ghosts. The increase in their number, along with the increase in nucleated cells (Fig. 3C) , suggests that they were affected by the process of CLC in a manner similar to nucleated cells. These cell-like structures were indistinguishable in size from nucleated cells from the same sample, and they appear to be cells that have lost their nuclei before leaving the corneal surface. 

In the model (Fig. 2K) , cell ghosts are shown to be downstream from dead cells in the nonapoptotic pathway, but we cannot rule out the possibility that the apoptotic pathways may also lead to ghost cells. However, there were few annexin-positive cell ghosts in later beakers (Fig. 3B) . On average, only 8% of cell ghosts were annexin positive, compared with 64% of nucleated cells. From beaker A through beaker E nucleated cells increased in annexin-positive cells, whereas cell ghosts increased in annexin-negative cells. Thus, although we cannot rule it out, we do not have evidence to support the contention that cell ghosts arise from an apoptotic pathway. 

Under the conditions of our experiment there appeared to be three pathways by which a cell could leave the corneal surface (Figs. 2I 2J 2K) . The majority of corneal cells were shed by nonclassic apoptosis, in the sense that no condensed apoptotic bodies could be observed. However, this could be a function of the shear stress placed on the surface by the removal of the contact lens. A substantial number of cells left the surface by a nonapoptotic pathway as dead cells or cell ghosts. Few cells left the surface by classic apoptosis. 

 Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1998.

 Supported by The Futura Foundation, Inger Holms Medical Foundation, Aase Bye and Trygve J. B. Hoff’s Medical Foundation, Rasmussens Stiftelse, and the National Institutes of Health (EY03039).

 Submitted for publication March 18, 1999; revised December 3, 1999 and April 11, 2000; accepted May 22, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

The number and size of nucleated corneal cells collected from five subjects in 20 removals of a contact lens (mean ± SD).

Figure 2.

 

Immunocytochemical staining of shed human corneal epithelial cells collected by CLC and viewed by fluorescence microscopy. (A) TUNEL-positive corneal epithelial cell nucleus. (B) The same nucleus as in (A) stained with Hoechst. (C) The nuclei of two TUNEL-positive cells; the top cell shows a fragmented nucleus. (D) Hoechst staining of the nuclei in (C). (E) Annexin V staining of the cytoplasmic membrane of a corneal epithelial cell. Double exposure to show the PI-positive nucleus. (F) Two cells with PI-positive nuclei. The plasma membrane of the upper cell stained annexin V–positive. The lower cell showed no annexin V staining. (G) Hoechst-positive small cell with a fragmented nucleus. This cell stained PI negative. (H) The same cell as in (G) stained annexin V positive. Bars, 10 μm. (I, J, and K) A model for cells collected from the corneal surface by CLC (blue circles). Cells on the corneal surface are mainly viable cells. This is shown in the rectangle on the left in which green represents viable cells stained positively with calcein. ⁶ There are several ways by which a cell can leave the corneal surface. Three pathways are shown: (I) A few cells leave the surface by classic apoptosis in which blebs are attached to the cell in the initial stage. (J) The majority are shed by nonclassic apoptosis in the sense that no condensed apoptotic bodies are observed. (K) Some cells leave the cornea as dead cells; these are shown with PI-positive nuclei. Initially, the remains of intracellular esterase activity can be identified by the bright green punctate distribution of calcein labeling. In the final stage, the cells lose their nuclei and become cell ghosts.

Figure 3.

 

Corneal cells collected by CLC in three human subjects. Beakers A to E. (A) Percentage TUNEL-positive nucleated cells. (B) Percentage annexin V–positive nucleated cells and cell ghosts. (C) Number of nucleated cells and cell ghosts.

Figure 4.

 

Number of labeled nucleated corneal cells collected by CLC in three human subjects. Beakers A to E. (A) Hoechst positive (nucleated cells), (B) annexin positive, and (C) PI-positive (dead) cells.