Investigation of the Corneal Filament in Filamentary Keratitis

Abstract

purpose. To date, no studies have elucidated the composition of the corneal filament in detail. In this study, an immunohistochemical technique was used to clarify the exact composition of the corneal filament in filamentary keratitis. In addition, the mechanisms responsible for filament formation were identified.

methods. Filaments were obtained from 13 patients with filamentary keratitis with a background of penetrating keratoplasty, aqueous tear deficiency, and severe ocular surface disorders, who were receiving treatment at an outpatient facility. From those tissues, transverse and longitudinal frozen sections were prepared and subjected to an indirect fluorescent immunohistochemical analysis with primary antibodies, including cytokeratins (CK1, -4, -6, -10, -12, and -13), mucins (MUC1, -4, -5AC, and -16), keratinization-related proteins (transglutaminase [TGase]-1 and filaggrin), cell proliferation marker Ki67, and markers of infiltration cells (HLA-DR and neutrophil-elastase). TUNEL staining was used for the detection of apoptosis. Fluorescent images of the sections were inspected with a fluorescence microscope.

results. The filaments were composed of CK12-positive cells and had a roll-formed central core. They were covered with MUC5AC- and -16-positive mucins including CK4- and -13-positive cells and neutrophil-elastase–positive cells. The filaments also included broken cells and DNA fiber-form postlesional nuclei that were positive for TUNEL staining. However, those areas stained weakly for CK6 and HLA-DR; faintly for CK1, CK10, MUC1, and MUC4; and not at all for Ki67, TGase-1, and filaggrin.

conclusions. The results of this research have the potential to open new pathways toward understanding the mechanism that generates the filament in filamentary keratitis, as well as new treatments in the future.

Filamentary keratitis is a chronic disorder of the cornea and, rarely, the conjunctiva, which appears as a variable-length tail extending from the ocular surface epithelium when viewed by slit lamp biomicroscope. The filaments can be stained by rose Bengal, lissamine green, and fluorescein. Patients with filamentary keratitis sometimes experience severe ocular pain. Filamentary keratitis is related to dry eye, especially aqueous-tear–deficient (AQD) dry eye, and it is reportedly associated with various kinds of ocular surface diseases or conditions, such as superior limbic keratoconjunctivitis, acute viral keratoconjunctivitis, postcataract surgery or penetrating keratoplasty (PKP), recurrent erosion, neurotrophic keratopathy, vernal keratoconjunctivitis, prolonged use of an eye patch, ptosis, and large-angle strabismus. ^{1 2 3 4 5 6} Filaments are known to be composed of mucin, which is stained by periodic acid-Schiff (PAS) or Alcian blue, and degenerating and regenerating epithelial cells. ^{7 8 9} However, only histologic studies by light- or electron microscopy, not by immunohistochemical examination, have been performed on this condition. ^{7 8 10 11 12} To date, reported studies regarding filaments are limited to only basic pathologic examination. Therefore, in our present study we used an immunostaining technique to accurately obtain the specific composition of the filaments. In addition, we identified the mechanisms responsible for filament formation. 

Methods

Enrollment of Patients

The study involved 13 patients with filamentary keratitis (Table 1)who were receiving treatment for removal of the filaments at an outpatient facility. This research was approved by the Committee for Ethical Issues on Human Research at Kyoto Prefectural University of Medicine and adhered to the tenets of the Declaration of Helsinki. 

Sample Collection

Filaments were obtained from 13 patients with filamentary keratitis after the ocular surface of each patient was examined and photographed with a slit lamp biomicroscope (SL-1600; Nidek Co., Ltd., Aichi, Japan). After anesthesia with topical appreciation of 0.4% oxybuprocaine hydrochloride (Benoxil; Santen Pharmaceutical Co., Ltd., Osaka, Japan), the larger firmaments were collected with forceps. The other filaments were then scraped away with a glass stick. Each patient reported that most of the pain had subsided within a few days after removal of the filament. A normal donor cornea, including conjunctiva (keratoconjunctiva), isolated at 5.5 hours postmortem was obtained from SightLife (Seattle, WA). Filaments and keratoconjunctiva were embedded in OCT compound (Tissue-Tek; Sakura Fine Technical Co., Ltd., Tokyo, Japan) and immediately snap frozen with liquid nitrogen for immunostaining. 

Immunostaining and Light Microscopic Analysis

Filament and frozen sections of keratoconjunctiva (8 μm) for immunostaining and light microscopic analysis were placed transversely and longitudinally on glass slides and subjected to hematoxylin and eosin (HE) staining or indirect immunostaining analysis. In brief, the sections were fixed with Zamboni’s fixative or acetone (4°C, 5 minutes), immersed for 1 hour in blocking solution (1% BSA in 0.01 M PBS), and treated with primary antibody solutions (Table 2)and normal mouse IgG1, IgG2a, IgG2b (DakoCytomation, Kyoto, Japan), and goat IgG (Santa Cruz Biotechnology Inc., Santa Cruz, CA) as negative controls. After a 1-hour incubation, the sections were washed with 0.01 M PBS and then treated with fluorescent secondary antibody solutions (Alexa-488- or 594-labeled anti-mouse IgG or anti-goat IgG; Invitrogen, Carlsbad, CA). After a 1-hour incubation, the sections were washed with 0.01 M PBS, and the nuclei were stained with propidium iodide or DAPI and mounted with medium containing 3% anti-photo-bleaching reagent (DABCO; Wako Pure Chemical Industries, Ltd., Osaka, Japan). Unless otherwise stated, all incubations were performed at room temperature. Some filaments were embedded in paraffin wax by standard procedures. The paraffin-embedded sections (4 μm) were stained with HE. Fluorescent and HE-stained images of the sections were taken by microscope with a cooled CCD camera (Olympus Corp., Tokyo, Japan). A commercial fluorometric TUNEL system (DeadEnd; Promega, Madison, WI) was used for analysis of apoptosis. 

Results

Slit Lamp Examination

Slit lamp examination of the 13 patients with filamentary keratitis revealed filaments of various sizes. Almost all the big filaments found on the patients’ corneas were located behind the upper or lower eye lids (Fig. 1E , Table 1 ). Whole images of these filaments were observable when fingers were used to open patients’ lids (Figs. 1A 1B 1C 1D 1F) . 

Light Microscopic Analysis

HE staining of paraffin-embedded sections showed that the filament consisted of a core composed of eosinophilic cells that had spindle-shaped cellular cytoplasm and nuclei. The surrounding parts were basophilic fibers and faint basophilic acellular areas that included basophilic segments and polymorphic nucleic cells (Fig. 2) . The light microscopic images of HE-stained cryosections of each filament and of normal cornea and conjunctiva showed similar patterns to the patterns shown in the immunostaining images (Figs. 3A1 3A2 3A3 3A4 3A5) . The histologic images of the cryosections were similar to those of the paraffin-embedded sections. The filaments had eosinophilic cores, acellular areas, and basophilic segments. The normal cornea and conjunctiva had eosinophilic epithelium, and a pale goblet cell (Fig. 3A2 , arrow) was found in the conjunctival epithelium. 

Immunostaining

CK12 was expressed on all layers of normal corneal epithelium. CK4 was expressed on normal conjunctival epithelium and the superficial layer of normal corneal epithelium. Normal conjunctival epithelium was immunostained against CK13 (Fig. 3) . In samples that had cores on HE-stained paraffin-embedded and frozen sections (Figs. 2 , 3A3 3A4 3A5 ), the core areas strongly stained red, positive for CK12. In the areas surrounding the filaments, the cellular components strongly stained green, positive for CK4 and -13. A longitudinal section of the filament showed a long core of eosinophilic cells at the center of the filament under HE staining (Fig. 3D) . By immunostaining, the red-stained CK12-positive corneal epithelium was found at the center of the filament. These red cores were of an inosculated, ropy, braided shape. The CK4- and -13-positive conjunctival epithelium, which were stained green, were found in the area around the corneal cells. The side of the attachment site had round or elliptical nuclei stained blue, but the side of the free extremity had a fiber form stained blue by DAPI (Figs. 3D 3E 3F 3G 3H 3I 3J) . MUC1 and -4 were slightly positive in the conjunctival epithelium, MUC5AC was positive in conjunctival epithelium at the goblet cells, and MUC16 was positive in both the corneal and conjunctival epithelia (Figs. 4A 4B 4C 4D 1, A–D2). We detected a considerable amount of MUC5AC (goblet-cell–derived, gel-forming mucin) in the acellular area of the filaments, MUC16 (membrane-bound mucin in the corneal and conjunctival epithelium) in almost all areas (Figs. 4C 4D) , but a limited amount of MUC1 and -4 in the filament itself (Figs. 4A 4B) . Moreover, MUC5AC and -16 stained positive in the longitudinal section, and the side of the free extremity had a fiber form that was stained red by PI (Figs. 4E 4F) . Infiltrating cells were positive for neutrophil elastase or HLA-DR. The surrounding areas stained weakly for CK6, which was expressed at wound healing or in a hyperproliferative situation; stained faintly for CK1 and -10, which are major cytokeratins in the epidermis of the skin; and stained not at all for keratinization-related protein (TGase-1 and filaggrin; Fig. 5 ). The nuclei of the core areas and PI-positive fibrous material around the core were TUNEL positive, but were negative without the rTdT enzyme. These nuclei did not stain for the cell proliferation marker Ki67 (Fig. 6) . 

Negative control experiments were used for immunostaining under the same conditions used for each antibody, and they showed no positive staining. 

A summary of the histological data is shown in Table 3 . 

Discussion

Wright ⁸ showed by PAS, Alcian blue, and red oil staining that most of the filaments of filamentary keratitis are composed of mucus with epithelial squamous, lipids, and foreign matter taken up secondarily. Lambert ^{13 14} reported a theory about the formation of filaments, and that theory has subsequently been included in various textbooks. ^{2 3} As a normal epithelial surface degenerates due to desiccation, some cells die and fall off, thus leaving a defect. Mucin may adhere to this high-energy pit, and eventually epithelium may grow over the mucin to form a filament. However, Thiel et al. ¹⁰ and Zaidman et al. ¹¹ presented another theory pertaining to filamentary keratitis by transmission electron microscopy of a patient with a brain stem disease, which showed that groups of inflammatory cells and fibroblasts were present just below the basal epithelium. It appears that these cells had disrupted the epithelial basement membrane and Bowman’s layer. These findings support the theory that filamentary keratitis is associated with the damage of basal epithelial cells, epithelial basement membrane, or both. These findings regarding the components of the filament were derived from classic staining and electron microscopic examination, although they were not enough to completely elucidate the detail and origins of the components. The mechanisms of filament generation that are different from our proposal may be due to these different methods and samples. Therefore, to further elucidate the components of the filament, we examined the filament with an immunostaining technique. 

HE staining of the paraffin-embedded sections clearly showed the filament to consist of a core composed of cells that had spindle-shaped, eosinophilic cytoplasm and nuclei. The surrounding parts were basophilic fibers and faint basophilic acellular areas that included basophilic segments and inflammatory cells. Our study revealed that the filament core was composed of corneal epithelium, a finding supported by positive staining for CK12, which is specifically expressed in corneal epithelium. On the other hand, the surrounding portion of the filament was composed mainly of degenerating conjunctival epithelial cells, which had markers typical of conjunctival epithelial cells ¹⁵ and did not have markers typical of corneal epithelium. It is also reasonable to speculate that the conjunctival epithelial cells surrounding the filament were mainly derived from the bulbar conjunctival epithelium that is next to the peripheral corneal epithelium and the palpebral conjunctival epithelium that is always in contact with the corneal surface. At the interface between the tear film and ocular surface epithelium, it was demonstrated that membrane-associated mucins, including MUC1, -4, and -16, and a major mucin of the secretory class, the goblet-cell–derived gel-forming MUC5AC, were all present, ^{16 17 18 19 20} and our study demonstrated that MUC5AC and -16 were the major mucins in the filament. MUC16 is expressed in the superficial corneal epithelium and conjunctival epithelium, and it was found in every part of the filaments in our study. CK1 and -10, although typical of epidermal keratinocytes, have also been described in corneal differentiated cells. ²¹ Previously, we have shown that the conjunctival epithelium in Sjögren’s syndrome expresses the keratinization marker or keratinization-related proteins CK1, CK10, TGase-1, and filaggrin. ²² However, in this study we found that those proteins were expressed in the filaments in only a limited amount. 

It is also known that filamentary keratitis is often associated with ocular surface inflammation. Our study reaffirmed that association and also demonstrated the existence of neutrophil and HLA-DR-positive cells in the filament. The structural rigidity of the filament, which is assumed to be the result of the twisted form of its core that is additionally supported by the surrounding mucin and the fiber of DNA from postlesional nuclei, is resistant to the condition of repeated friction by the eyelid. Moreover, although it has not been noted in previous reports, the DNA fiber, which exists mainly in the free extremity of the filament and is mixed with mucin, may be essential for the generation of the filament. 

Since the large filaments that we were able to collect for this experiment were located behind the upper or lower eye lids, and the core from the corneal epithelium was seen to be of an inosculated, ropy, braided shape, it was thought that the filaments were formed by some mechanical energy. On the ocular surface, we speculate that the mechanical energy was the blinking of the eyelids and/or movement of the eye behind the eyelids. 

In summary, although our new theory about the mechanism behind filament generation diverges from that previously reported, ^{11 13 14} our immunohistologic study of a large number of filament samples obtained from the cornea have led us to the following conclusion. We hypothesize that filament generation starts from an injury to the surface epithelium of the cornea due to various disease conditions. Subsequently, friction cause by blinking and/or movement of the eye develops between the palpebral conjunctiva and the injured epithelium and produces the filament core. Further frictional stress is exerted by the eyelid, and the core then entwines with mucin, conjunctival epithelium, fiber of DNA from postlesional nuclei, and inflammatory cells, thus building up the filament. This phenomenon is sometimes associated with inflammation and the detachment of basal cells from the Bowman’s layer due to blink- or eye-movement–related mechanical friction (Fig. 7) . We believe that the results of this research will open new pathways toward understanding the mechanism that generates the filament in filamentary keratitis, as well as new methods of treatment in the future. 

 Supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Culture, and Sports of Japan.

 Submitted for publication September 25, 2008; revised February 23, 2009; accepted June 5, 2009.

 Disclosure: H. Tanioka, None; N. Yokoi, None; A. Komuro, None T. Shimamoto, None; S. Kawasaki, None; A. Matsuda, None; S. Kinoshita, 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.

 

Patients and Samples

Patient No.	Filament Location	Number of Filaments Collected	Age	Sex	Underlying Cause of Filament	Current Treatment at Time of Filament Removal	Photos
1	Upper	2	85	F	OCP	AT	Fig. 2
2	Upper	1	50	F	PKP	AT+OFLX+BETA	Figs. 1A 1B 3A 3B 3C 3, 4A–D3, 5A–G1, 6A–C1
3	Upper	2	84	F	OCP	AT+OFLX+FLM+BETA+PSL	Figs. 1C 1D 3A 3B 3C 4, 4A–D4, 5A–G2, 6A–C2
4	Lower	1	60	F	ATD	AT+LVFX+FLM	Figs. 1E 1F 3A 3B 3C 5, 4A–D5
5	Lower	1	54	F	ATD	AT+LVFX+FLM+HYA	Figs. 3D 3E 3F 3G 3H 3I 3J 4E 4F
6	Lower	2	88	M	ATD	AT+LVFX+FLM	Figs. 5A 5B 5C 5D 5E 5F 5G 3, 6A–C3
7	Upper	2	77	M	PKP	LVFX+FLM
8	Upper	1	51	M	VS	LVFX+BETA+OFLXO
9	Lower	1	78	F	ATD	AT+LVFX+FLM
10	Lower	1	69	F	ATD	AT+LVFX+FLM
11	Lower	1	71	F	ATD	AT+LVFX+FLM
12	Lower	1	58	F	ATD	AT+OFLX+FLM
13	Lower	1	49	F	ATD	AT+LVFX+FLM

 Upper, behind upper eye lid; Lower, behind lower eye lid; OCP, ocular cicatricial pemphigoid; ATD, aqueous tear deficiency; VS, vitreous surgery; AT, artificial tear; OFLX, 0.3% ofloxacin eye drop; BETA, 0.1% betamethasone sodium phosphate eye drop; FLM, 0.02% fluorometholone eye drop; PSL, prednisolone tablets; LVFX, 0.5% levofloxacin eye drop; HYA, 0.3% sodium hyaluronate eye drop; OFLXO, 0.3% ofloxacin eye ointment.

Table 2.

 

Antibodies

Group	Antigen	Final Concentration μg/mL	Type of Antibody	Immunized Animal	Company	Annotation
Cytokeratins	CK1	0.025	Mo	M	Novocastra	Major cytokeratin in skin epidermis
	CK4	4.5	Mo	M	Novocastra	Major cytokeratin in nonkeratinizing mucosal epithelium
	CK6	1.0	Mo	M	Novocastra	Expressed at wound healing or hyperproliferative situation
	CK10	0.30	Mo	M	Novocastra	Major cytokeratin in skin epidermis
	CK12	2.0	Po	G	Santa Cruz	Major cytokeratin in cornea
	CK13	2.3	Mo	M	Novocastra	Major cytokeratin in nonkeratinizing mucosal epithelium
Mucin	MUC1	3.4	Mo	M	ZYMED	A membrane-bound mucin
	MUC4	1.0	Mo	M	ZYMED	A membrane-bound mucin
	MUC5AC	0.55	Mo	M	Novocastra	Secreted mucin/goblet cell mucin
	MUC16	0.85	Mo	M	Abcam	A membrane-bound mucin
Infitration cells	HLA-DR	1.7	Mo	M	Dako Cytomation	Marker of macrophage, Langerhans cell
	Neutrophil elastase	0.70	Mo	M	Dako Cytomation	Marker of neutrophil
Proliferation	Ki67	0.93	Mo	M	Dako Cytomation	Cell proliferation marker
Keratinization-related proteins	Transglutaminase-1	5.0	Mo	M	Blogenesis	Enzyme that catalyzes the cross-linking of envelope component proteins
	Filaggrin	0.25	Mo	M	Blogenesis	Serves as matrix for keratin 1/10 aggregation

 Mo, monoclonal; Po, polyclonal; M, mouse; G, goat; Novacastra: Novocastra Laboratories, Ltd., Benton Lane, UK; ZYMED: Zymed Laboratories, Inc., South San Francisco, CA; Abcam: Abcam Co. Ltd., Tokyo, Japan; Biogenesis: Biogenesis Ltd., Poole, UK.

Figure 1.

 

Slit lamp examination of filamentary keratitis. The photos are of representative cases of filamentary keratitis. (A, B) Post-PKP (case 2); (C, D) ocular cicatricial pemphigoid (case 3) and (E, F) ATD plus blepharoptosis (case 4). Arrows: large filaments.

Figure 2.

 

Paraffin-embedded HE staining of a filament. A filament taken from the patient with optical cicatricial pemphigoid + blepharoptosis (case 1) was fixed with 10% neutral buffered formalin, embedded in paraffin, sectioned, and stained with HE (hematoxylin eosin) (A–D). The filament consisted of a core composed of eosinophilic cells that have spindle-shaped cellular cytoplasm and nuclei (arrows; B). (C) The surrounding parts were basophilic fibers (arrows) and faint basophilic acellular areas that included basophilic segments (arrowheads); (D) many cells had polymorphic nuclei (arrows).

Figure 3.

 

The distribution of the marker of corneal and conjunctival epithelium. (A, D) HE staining of the frozen sections (arrow indicates a goblet cell); (B, E, G, H) CK12/CK4/nuclei (red/green/blue); (C, F, I, J) CK12/CK13/nuclei (red/green/blue); (A–C) transverse sections; (D–J) longitudinal sections of filament; (G, H) magnified view of (E); and (I, H) magnified view of (F). (1) Corneal epithelium and (2) conjunctival epithelium. Filaments in (3) a post-PKP eye (case 2), (4) an ocular cicatricial pemphigoid eye (case 3), and (5) an ATD eye (case 4). Staining was positive for CK12 on the corneal epithelium (B1) and core area of the filaments (B3-5, C3-5, E, F); for CK4 on the conjunctival epithelium (B2), the superficial epithelium of the cornea (B1), and the area around the filaments (B3-5); and for CK13 on the conjunctival epithelium (C2) and the area around the filaments (C3-5). These red cores were of an inosculated, ropy, braided shape (G, J). Blue: nuclear and DNA fibers.

Figure 4.

 

The distribution of mucins. (A) MUC1/nuclei (green/red); (B) MUC4/nuclei (green/red); (C, E) MUC5AC/nuclei (green/red); (D, F) MUC16/nuclei (green/red). (1) Corneal epithelium and (2) conjunctival epithelium. Filaments in (3) a post-PKP eye (case 2), (4) an ocular cicatricial pemphigoid eye (case 3), and (5) an ATD eye (case 4). Staining for MUC1 and -4 was slightly positive on conjunctival epithelium (A2, B2); for MUC5AC was positive on conjunctival epithelium (C2) at the goblet cells and the area around the filaments (C3-5, E); and for MUC16 was positive on both corneal (D1) and conjunctival (D2) epithelia and almost all areas of the filaments (D3-5, F).

Figure 5.

 

Immunostaining of the marker of inflammation cells, keratins, and keratinization-related protein. (A) Neutrophil elastase; (B) HLA-DR; (C) CK1; (D) CK6; (E) CK10; (F) filaggrin; and (G) TGase-1. Filaments in (1) a post-PKP eye (case 2), (2) an optical cicatricial pemphigoid eye (case 3), and (3) an ATD eye all stained green and the nuclei stained red (case 6). The surrounding areas stained for neutrophil-elastase (A) and HLA-DR (B), stained weakly for CK6 (D), stained faintly for CK1 (C) and CK10 (E), and did not stain at all for keratinization-related protein (F, G).

Figure 6.

 

TUNEL staining and immunostaining with Ki67. (A1–3) TUNEL staining: The nuclei of the core areas and PI-positive fibrous material around the core stained positively for TUNEL, but (B1–3) stained negatively without the rTdT enzyme. (C0) Some human corneal epithelial cells stained for Ki67 (arrows). Filaments of (C1) a post-PKP eye (case 2), (C2) an optical cicatricial pemphigoid eye (case 3), and (C3) an ATD eye (case 6) did not stain clearly for Ki67.

Table 3.

 

Results of Histological Examination

Stain/Antigen	Number	Filament*									Corneal Epithelium	Conjunctival Epithelium
HE (Paraffin Section)	1	Core Eosinophilic Spindle-Shaped Cells	Basophilic Fibers	Faint Basophilic Acellular Areas	Basophilic Segment	Polymorphic Nuclei Cells
HE (Cryosection)	12	Eosinophilic	Basophilic Fibers	Faint Basophilic	Basophilic	Eosinophilic/Basophilic
CK12	12	12	−	−	−	−	++	−
CK4	12	−	−	−	12	−	+ (S)	++
CK13	12	−	−	−	12	−	−	++
Muc1	4	−	−	−	4	−	−	+
Muc4	11	−	−	−	11	−	−	+/−
Muc5AC	12	−	−	12	12	−	−	++ (goblet)
Muc16	12	12	−	12	12	−	++ (S)	++ (S)
Neutrophil elastase	11	−	−	−	−	11
HLA-DR	4	−	−	−	−	4
CK1	4	4	−	−	−	−
CK6	4	4	−	−	4	−
CK10	4	−	−	−	4	−
Filaggrin	3	−	−	−	−	−
Transglutaminase-1	3	−	−	−	−	−
TUNEL	3	3	3	−	3	3
Ki67	3	−	−	−	−	−

*  Data shown in the Filament are the number that is positive. −, negative; +/−, slightly positive; +, moderately positive; ++, strongly positive; S, superficial.

Figure 7.

 

Schema for proposed mechanism of filament generation.

The authors thank John Bush for reviewing the article and the Northwest Lion’s Eye Bank foundation for helping to obtain fresh human corneal tissues.