Immunohistochemical Localization of NQO1 in Epithelial Dysplasia and Neoplasia and in Donor Eyes

Abstract

purpose. To examine the expression of NAD(P)H:quinone oxidoreductase 1 (NQO1, DT-diaphorase), a potential bioactivating enzyme for mitomycin C in corneal and conjunctival epithelial dysplasia and neoplasia and in normal tissues from human donor eyes, by immunohistochemistry.

methods. Formalin-fixed, paraffin-embedded sections of human donor eyes and tissue sections with histologic diagnoses of corneal and conjunctival epithelial dysplasia and neoplasia from the Eye Pathology Laboratory, University of Colorado Health Sciences Center were analyzed. Detection of NQO1 in tissues was performed using standard immunohistochemical techniques with monoclonal antibodies against NQO1 and immunoperoxidase staining.

results. All 20 tumors stained positive for NQO1. In seven eyes from four donors, positive staining for NQO1 was detected in all epithelial and endothelial layers, in fibroblasts, in all retinal layers except the photoreceptor outer segments, and in the fascicles and arachnoid of the optic nerve. Only minimal staining was detected in the photoreceptor outer segments and the optic nerve pia and dura. Immunostaining was markedly reduced in all tissues in both eyes from donor 5. Genetic analysis confirmed that this individual was homozygous for a polymorphism in NQO1 (NQO1*2).

conclusions. NQO1 was detected by immunohistochemistry in every examined section of corneal and conjunctival epithelial dysplasia and neoplasia, suggesting that NQO1 may play a role in the bioactivation of mitomycin C in these tumors. However, the presence of NQO1 in the corneal, conjunctival, and ciliary epithelium; the retinas; and the optic nerves of donor eyes may indicate the potential for mitomycin C toxicity, particularly at higher doses.

Topical mitomycin C is widely used as a surgical adjuvant in glaucoma filtration surgery ¹ and is effective in the treatment of pterygium, primary acquired melanosis, conjunctival melanoma, ocular cicatricial pemphigoid, and corneal and conjunctival epithelial dysplasia and neoplasia and in optic nerve decompression surgery for pseudotumor cerebri. ^{2 3 4 5 6 7 8 9 10 11} It is under investigation for the treatment of proliferative vitreoretinopathy. ¹² Complications associated with topical ophthalmic mitomycin C therapy include corneal epitheliopathy, ulcers and perforation, scleral melting, bleb leaks, hypotony, endophthalmitis, and neuroretinitis. ^{13 14 15 16 17} Mitomycin C is a bifunctional alkylating agent that must undergo bioreductive activation to exert an antitumor effect. ¹⁸ NAD(P)H:quinone oxidoreductase (EC 1.6.99.2, NQO1) or DT-diaphorase is a cytosolic obligate two-electron reductase that can induce bioactivation of mitomycin C in vitro and plays an important role in its bioactivation in vivo. ^{18 19 20 21} Previous immunohistochemical studies have localized NQO1 in human lung epithelial cancers and in normal respiratory epithelium, vascular endothelium, and adipocytes, ²² but to our knowledge there has been no report describing the distribution of NQO1 in the human eye. In addition, a polymorphism in NQO1 (NQO1*2) has been characterized recently. ²³ Among whites, the prevalence of the NQO1*2 polymorphism is approximately 4% to 7%, but it may be as high as 15% to 20% in Hispanic and Asian populations. ^{23 24} Tissues and cell lines derived from individuals homozygous for the NQO1*2 polymorphism have markedly reduced NQO1 activity and protein. ²⁵  

Corneal and conjunctival epithelial neoplasia and dysplasia, including carcinoma in situ, is thought to be a precursor for invasive squamous cell carcinoma, the most common primary malignancy of the conjunctiva. ⁸ Although the standard treatment for these lesions is surgical excision, recurrent and diffuse disease has been treated successfully with topical mitomycin C. ^{6 7 8 9} The purpose of this study was to use immunohistochemistry to determine the presence of NQO1 in corneal and conjunctival epithelial dysplasia and neoplasia and in normal tissues from human donor eyes. The role of the NQO1*2 polymorphism on NQO1 expression in normal tissues of the eye was also examined. 

Methods

All investigations were performed according to the tenets of the Declaration of Helsinki. 

Tissue Specimens

The archives of the University of Colorado Eye Pathology Laboratory were examined, and 20 cases of corneal and conjunctival epithelial dysplasia and neoplasia were located in 19 patients. Paraffin-embedded tissue specimens were investigated in every case. Human donor eyes from the Rocky Mountain Lions Eyebank were processed for histopathologic analysis by formalin fixing, opening, dehydration, and paraffin embedding of pupil–optic nerve sections. Genotyping for the NQO1*2 polymorphism was performed using a polymerase chain reaction–restriction fragment length polymorphism (PCR-RFLP) assay and genomic DNA, as previously described. ²³ Genomic DNA was extracted from formalin fixed, paraffin-embedded tissue using a template preparation kit (High Pure PCR; Boehringer–Mannheim, Indianapolis, IN). 

Anti-NQO1 and Control Monoclonal Antibodies

Anti-NQO1 monoclonal antibody–secreting hybridomas (clonesA180 and B771) were derived from a BALB-c mouse immunized with purified human recombinant NQO1 protein. Antibodies from these hybridoma clones react with both wild-type and mutant (NQO1*2) human NQO1 proteins. A control (nonspecific IgG–secreting) hybridoma (clone C100) was derived from a BALB-c mouse. All hybridoma cell lines were grown in spinner flasks in RPMI medium containing 50 U/ml penicillin, 50 mg/ml streptomycin, 1% (wt/vol) l-glutamine (Life Technologies, Gaithersburg, MD), and 10% (vol/vol) fetal bovine serum (Hyclone, Logan, UT) in 5% CO₂ at 37°C to a concentration of 10⁶ cells/ml. Hybridoma tissue culture supernatants were prepared by centrifugation at 3000g for 10 minutes and then stored at −80°C. Before use, supernatants were centrifuged at 10,000g for 5 minutes. 

Immunohistochemistry

The immunohistochemical techniques have been described in detail previously. ²² In brief, deparaffinized, rehydrated tissue sections were exposed to hydrogen peroxide to eliminate endogenous peroxide activity. The sections were incubated with hybridoma tissue culture supernatant for monoclonal antibodies to NQO1 or control antibodies. Slides were then exposed to avidin-biotinylated complex with horseradish peroxidase. The chromogens were diaminobenzidine or 9-aminoethylcarbazole with hydrogen peroxide. The slides were counterstained with hematoxylin, and immunostaining was scored as 0 (no staining), trace, 1+, 2+, or 3+. 

Results

Immunohistochemical Localization of NQO1 in Human Corneal and Conjunctival Epithelial Dysplasia and Neoplasia

Immunohistochemical staining was performed on multiple sections of 20 lesions from 19 patients. Eleven lesions were completely confined to the epithelial layer, whereas nine lesions showed microinvasion histologically. NQO1 was detected in every specimen (Table 1) . There was no difference in the mean staining intensity between the epithelial lesions and the microinvasive lesions. Adjacent histologically normal epithelium and vascular endothelium also stained positive for NQO1, but no immunostaining for NQO1 was detected in the underlying connective tissue. The pattern of NQO1 immunostaining varied among specimens (Figs. 1A 1B, 1C ), with some showing uniform staining, and others showing a variable or mosaic pattern. One specimen was remarkable for only minimal staining in the tumor, with heavy staining in attached normal epithelium (Fig. 1D) . In 41 of 43 negative control specimens, there was no immunostaining for NQO1, whereas two negative controls showed trace to 1+ (false-positive) staining. These data demonstrate that NQO1 was expressed in corneal and conjunctival epithelial dysplasia and neoplasia. 

Immunohistochemical Localization of NQO1 in Tissues from Human Donor Eyes

Nine eyes from five donors were examined. Results of immunohistochemical analysis of the donor eyes are given in Figure 2 and Table 2 . Seven eyes (four donors) showed positive immunostaining for NQO1 in tissues throughout the eyes. Staining was particularly strong in epithelial, endothelial, and retinal tissues. Minimal staining was observed in subconjunctival, scleral, and optic nerve connective tissues and in photoreceptor outer segments. Repeated attempts to genotype these eyes were unsuccessful. 

In contrast, the two eyes from a donor genotyped as homozygous for the NQO1*2 polymorphism did not demonstrate significant immunostaining in the keratocytes, conjunctival or choroidal endothelium, Tenon’s capsule fibroblasts, trabecular meshwork cells, cells lining Schlemm’s canal, or the arachnoid of the optic nerve. 

The presence of NQO1 in tissues from eye donors was confirmed by immunoblot analyses of fresh-frozen cornea, lens, and optic nerve. A protein band at 31 kDa corresponding with purified human NQO1 was detected in all samples examined (data not shown). 

Discussion

Using activity assays and immunoblot analysis, NQO1 has been shown to be present in many human normal and neoplastic tissues. ^{26 27} The biologic role of NQO1 in these tissues has not been established. Experiments with purified human NQO1 and cell lines with high levels of NQO1 suggest that NQO1-catalyzed reduction of quinones to hydroquinones may result in either activation or detoxification, depending on the structure of the hydroquinone generated. ¹⁹ Recent work with endogenous quinones (coenzyme [Co]Q10 and α-tocopherolquinone) suggests that NQO1 may function as an antioxidant enzyme. Reduction of CoQ10 andα -tocopherolquinone by NQO1 results in the formation of hydroquinones with excellent antioxidant properties. ^{28 29}  

There is increasing evidence implicating NQO1 in the bioactivation of mitomycin C. Biochemical studies have shown that purified NQO1 can induce bioactivation of mitomycin C to a species capable of cross-linking DNA. ^{18 19} In studies using the National Cancer Institute’s anticancer drug screening panel of tumor cell lines and in mouse xenograft studies a strong correlation between NQO1 activity and sensitivity to mitomycin C was reported. ^{30 31} Transfecting Chinese hamster ovary and human colon carcinoma cells with human NQO1 leads to increased mitomycin C cytotoxicity. ^{20 32}  

The presence of NQO1 in corneal and conjunctival epithelial dysplasia and neoplasia suggests that the clinically observed antitumor activity of mitomycin C may be due to bioactivation by NQO1 within these cells. However, some clones of cells with low or no NQO1 activity may have decreased sensitivity to mitomycin C, and this may help explain the variable response seen clinically. ^{6 7 8 9} In experiments with cultured human colon and gastric cell lines, treatment with mitomycin C resulted in selection of clones with decreased NQO1 activity. ²⁰ The presence of NQO1 in attached normal epithelium and vascular endothelium suggests that mitomycin C may also be bioactivated in these tissues, resulting in the clinically observed toxicities of hyperemia, epithelial defects, and punctate epithelial keratopathy. 

The intracellular location of NQO1 helps explain the distribution of NQO1 in the donor eyes. Highly cellular tissues such as epithelium, endothelium, and retina demonstrated strong NQO1 staining, whereas connective tissues with extensive extracellular matrix, such as subepithelial conjunctival tissue, corneal stroma, and the optic nerve pial septae and dura mater showed much less staining. 

On the ocular surface, the effectiveness of mitomycin C in treating pterygium, primary acquired melanosis, and ocular cicatricial pemphigoid may arise from bioactivation by NQO1 and intracellular toxic effects on tumors and degenerations in corneal and conjunctival epithelial cells and melanocytes. At the same time, the bioactivation of mitomycin C by NQO1 in tissues found to have strong immunostaining, including corneal endothelium, keratocytes, lens epithelium and cortex, scleral fibroblasts, and vascular endothelium, may explain the complications of mitomycin C treatment, such as corneal edema, necrotizing keratitis, cataract, and scleral melting. ^{13 14 15 16 17}  

An important ophthalmic use of topical mitomycin C is as a surgical adjunct in glaucoma filtering surgery. ¹ Topical mitomycin reduces postoperative intraocular pressure and reduces the frequency of bleb failure. ¹ The mechanism is thought to be inhibition of the proliferation of Tenon’s capsule fibroblasts ^{1 13} ; however, recent data indicate that a direct toxic effect on the ciliary epithelium may also occur. ^{13 33} The presence of NQO1 in Tenon’s fibroblasts and the ciliary epithelium suggests that intracellular bioactivation of mitomycin C can occur at these locations. The finding of NQO1 in vascular endothelium and ciliary epithelium is consistent with mitomycin’s known side effects and complications, which include avascular blebs, blebitis, endophthalmitis, and hypotony. ^{13 14} Because the eyes from the donor with the NQO1*2 polymorphism did not show immunostaining at Tenon’s fibroblasts, individuals with this polymorphism may be less responsive to mitomycin C during trabeculectomy. A clinical study of this possibility may be warranted. 

The effects of mitomycin C on the retina have recently been examined. ^{12 14 34 35} A case report of exudative vitreitis, peripapillary retinitis, and macular edema from the overuse of topical mitomycin C has been published, ¹⁴ and experiments investigating the toxicity of mitomycin C in the retina and retinal pigment epithelium have also been published. ^{34 35} Recently, the treatment of experimental vitreoretinopathy in the rabbit with mitomycin C has been investigated. ¹² The potential for bioactivation and direct toxicity of mitomycin within the cells of the retina is suggested by the presence of NQO1 throughout the layers of the retina. During extraocular therapy, mitomycin could enter the eye through thinned sclera, inadvertent intraocular injection of subconjunctival doses, or by direct diffusion from other compartments. ¹⁴  

Topical mitomycin has recently been used on the optic nerve. A small case series on the use of topical mitomycin C as a surgical adjunct in optic nerve sheath decompression for pseudotumor cerebri has been reported. ¹⁰ Formation of a bleb or fistula was associated with successful outcome, and mitomycin appeared to reduce scarring around the nerve after a second operation. Patients had no notable side effects from the mitomycin. ¹⁰ However, a study in rabbits showed decreased amplitudes of electroretinograms, without identifiable histopathologic lesions of the optic nerve, after topical administration of mitomycin C on the optic nerve sheath. ³⁶ A recent case report suggests that posterior scleritis occurred after topical mitomycin treatment during optic nerve decompression surgery. ¹¹ In the present study, the identification of NQO1 in the optic nerve axon fascicles raises the possibility of neurotoxicity through bioactivation within the fascicles. However, the absence of NQO1 in the pial septae and dura mater suggests that other tissues may be involved in mitomycin’s role in optic nerve sheath fenestration. 

Although the evidence linking the antitumor activity of mitomycin C with bioactivation by NQO1 is mounting, it is clear that other reductases can also cause bioactivatation of mitomycin C such as NADPH:cytochrome P450 reductase, NADH:cytochrome b5 reductase and xanthine dehydrogenase. In addition, bioactivation of mitomycin C has been shown to be influenced by intracellular pH and oxygen concentrations. ^{37 38} A direct correlation between these immunohistochemical observations of NQO1 expression and the antitumor activity and toxicity of mitomycin C cannot be made until further studies have been performed. The location within the eye of other enzymes that compete with NQO1 for bioactivation of mitomycin C, must be identified. Penetration of mitomycin C in sufficient concentrations for therapeutic results or toxic effects on the tissues examined in this study should be investigated further. Because NQO1 is a highly inducible enzyme, factors such as diet and oxidative stress may lead to higher levels of NQO1 in some tissues. ³⁹ Determining whether NQO1 levels can be manipulated in tissues of the eye to affect mitomycin C bioactivation awaits further investigations. Finally, the small number of donor eyes in this study implies that caution should be used in extrapolating these results to populations. 

The minimal staining for NQO1 in the eyes of the donor genotyped as homozygous for the NQO1*2 polymorphism confirms previous reports, showing a similar loss of NQO1 activity and protein in cell lines and tissues from individuals homozygous for the NQO1*2 polymorphism. ²⁵ Determining whether individuals who are homozygous for the mutant NQO1 allele demonstrate alterations in their clinical response to mitomycin C and experience the same toxic effects awaits further study. 

Bioactivation of mitomycin C by NQO1 could occur in ocular tissues, contributing to its clinical efficacy and to the toxicities that have been observed. Particularly noteworthy is the presence of NQO1 in the ciliary epithelium, retina, and optic nerve, providing potential for toxic effects in these structures. Further ophthalmic clinical trials of mitomycin C should take into account its potential toxic effects on these tissues. In addition, when topical mitomycin C therapy is considered for ophthalmic disease, it may be useful to examine the role of the NQO1*2 polymorphism in treatment response and toxicities. 

 Supported by Grant HHS CA51210 from the US Department of Health and Human Services, Grant EPA R82 5281 010 from the US Environmental Protection Agency, and an unrestricted grant from Research to Prevent Blindness.

 Submitted for publication September 1, 1999; revised November 17, 1999; accepted December 9, 1999.

 Commercial relationships policy: N.

 

Table 1.

 

Immunostaining for NQO1 in 20 Cases of Corneal and Conjunctival Epithelial Dysplasia and Neoplasia

Case	Diagnosis	Staining
1	SCC	1+ to 3+
2	SCC	1+ to 2+
3	CIN	2+ to 3+
4	SCC	2+
5	CIN	2+
6	SCC	3+
7	SCC	2+ to 3+
8	SCC	1+ to 2+
9	SCC	1+ to 2+
10	CIN	1+ to 2+
11	CIN	1+ to 3+
12	CIN	1+
13	CIN	2+ to 3+
14	CIN	3+
15	SCC	1+
16	CIN	2+ to 3+
17	CIN	2+ to 3+
18	CIN	2+ to 3+
19	CIN	1+
20	SCC	2+

 CIN, intraepithelial dysplasia and neoplasia; SCC, microinvasive squamous cell carcinoma.

Figure 1.

 

Immunostaining for NQO1in corneal and conjunctival epithelial dysplasia and neoplasia. (A) Relatively uniform staining for NQO1 in one tumor. (B) In another tumor, basal epithelial cells, and cords and islands of more superficial cells stained for NQO1, whereas other epithelial cells did not. (C) A third tumor shows staining for NQO1 in basal epithelial cells and in isolated superficial cells, but not in most superficial cells. (D) Another specimen shows only minimal staining in the tumor, whereas an abrupt change to uniform staining is observed in attached histologically normal epithelium. Immunoperoxidase and hematoxylin; original magnification, ×80.

Figure 2.

 

Immunostaining for NQO1in human donor eyes. (A) Corneal epithelium and keratocytes stained strongly. (B) Conjunctiva, with intense staining in epithelium and endothelium, but minimal staining in subepithelial connective tissue. (C) Anterior chamber angle structures, showing immunostaining of the cells lining the trabecular meshwork and Schlemm’s canal. (D) Ciliary body, with staining in nonpigmented epithelial cells. (E) Lens, with intense staining in epithelium, moderate staining in cortical fibers, and absent staining in the capsule. (F) Retina, showing intense staining in the nerve fiber layer, moderate staining in the inner and outer plexiform layers, intense staining in the photoreceptor inner segments, and minimal staining in the photoreceptor outer segments. (G) Retinal pigment epithelium and choroid. (H) Optic nerve, with staining in axon fascicles, but only minimal staining in pial septae. (I) Cornea from a donor genotyped as homozygous mutant for the NQO1*2 polymorphism, showing minimal uptake. Immunoperoxidase and hematoxylin; original magnification (A, B, E, G, H), ×320; (C, D, F, I), ×80.

Table 2.

 

Immunostaining for NQO1 in Human Donor Eyes

Tissue	Nongenotyped			Homozygous Mutants, *
		Number Positive (%)	Strength	Number Positive (%)	Strength
Cornea
Epithelium	7 (100)	2+ to 3+	2 (100)	Tr to 3+
Keratocytes	7 (100)	1+ to 2+	0 (0)	—
Endothelium	7 (100)	1+ to 3+	2 (100)	Tr to 1+
Conjunctiva
Epithelium	7 (100)	1+ to 3+	2 (100)	Tr to 2+
Vascular endothelium	7 (100)	Tr to 2+	0 (0)	—
Tenon’s Capsule	7 (100)	Tr to 2+	0 (0)	—
Lens
Epithelium	7 (100)	1+ to 3+	2 (100)	Tr to 2+
Cortex	6 (86)	Tr to 2+	2 (100)	Tr to 2+
Ciliary body
Epithelium	7 (100)	1+ to 2+	2 (100)	Tr
Muscle	6 (86)	Tr to 2+	1 (50)	Tr
Retina
Nerve fiber layer	7 (100)	1+ to 3+	2 (100)	1+ to 2+
Ganglion cell layer	7 (100)	Tr to 3+	2 (100)	Tr to 2+
Inner plexiform layer	7 (100)	Tr to 2+	2 (100)	Tr to 1+
Outer plexiform layer	7 (100)	Tr to 3+	2 (100)	Tr to 1+
Photoreceptor inner segments	7 (100)	1+ to 3+	2 (100)	1+ to 2+
Photoreceptor outer segments	3 (42)	Tr to 1+	1 (50)	1+ to 2+
Retinal pigment epithelium	7 (100)	Tr to 2+	1 (50)	1+
Choroid
Vascular endothelium	7 (100)	1+ to 2+	0 (0)	Tr
Optic nerve
Axon fascicles	7 (100)	Tr to 3+	2 (100)	Tr to 1+
Pial septae	2 (28)	Tr	0 (0)	—
Arachnoid	5 (72)	1+ to 2+	0 (0)	—
Dura	0 (0)	—	0 (0)	—
Sclera
Fibroblasts	7 (100)	1+ to 2+	1 (50)	2+
Trabecular meshwork	6 (86)	1+ to 2+	0 (0)	—
Schlemm’s canal	6 (86)	Tr to 2+	0 (0)	—

 Nongenotyped, n = 7; homozygous mutants, n = 2. Tr, trace.

*  Homozygous mutant for the NQO1

*  2 polymorphism.

The authors thank Mary Jo Garascia for meticulous preparation of the histologic sections.