Elevated expression of ABCB5 in ocular surface squamous neoplasia

ATP-binding cassette subfamily B member 5 (ABCB5) is a new member of the ATP-binding cassette superfamily and has been reported as a novel marker for limbal stem cell (LSC), which is essential for corneal homeostasis. ABCB5 expression has also been discovered in the subpopulation of several cancer cells containing the cancer stem cell (CSC). However, the pathogenetic relationship between LSC and CSC and ABCB5 in the ocular surface squamous neoplasm (OSSN) is still entirely unknown. To improve understanding of the role of ABCB5 in OSSN, we performed immunohistochemistry for ABCB5 in nine OSSN case series. While expression of ABCB5 is restricted to the basal epithelial cell layer in the normal limbus, elevated expressions of ABCB5 were clearly observed in all OSSN, and there was some breadth in the range of intensity of ABCB5 expression. Interestingly, the elevated expression patterns of ABCB5 in OSSN could be classified in three categories: perivascular, marginal and diffuse patterns. Our findings demonstrated for the first time that the expression of ABCB5 was upregulated in OSSN and that elevated expression of ABCB5 may be involved in the pathogenesis of OSSN.

Scientific RepoRts | 6:20541 | DOI: 10.1038/srep20541 keratoepithelioplasty (KEP) was performed on the basis of a case-based evaluation ( Table 1). Post-operative chemotherapy with 5-fluorouracil (5-FU) was prescribed for 5 patients (cases [4][5][6][7][8]. For 7 patients of more than 2-year follow-up, local recurrence was observed in 2 eyes (28.6%, one CIS and one SCC). Recurrence was detected in one patient following 2 cycles of post-operative 5-FU administration and was successfully treated with re-excision. Human papilloma virus 16 (HPV 16) was detected in one patient (case 7). From our clinical observations, CIS, CIN and SCC are generally similar in appearance, making it difficult to distinguish between them in a clinical setting. Clinical presentations are summarized in Table 1.
Histopathology. Based on histopathological features, the OSSNs were classified as 1 eye with CIN, 3 eyes with CIS and 5 eyes with SCC. CIN and CIS (Fig. 1B,D,F,H) showed intraepithelial dysplastic change originating from the basal epithelial layer, with intact basement membrane. Dysplastic cells were characterized by bizarre hyperchromatism, high nucleus to cytoplasm ratio and the presence of atypical mitotic figures. In case 1, almost the full thickness of the epithelial layer was occupied by dysplastic cells consistent with the diagnosis of CIN (severe dysplasia) (Fig. 1B). Full thickness dysplastic change was observed without basement membrane invasion in cases 2-4, indicating CIS (Fig. 1D,F,H). In case 3, multiple fronds of neovascularization were observed, correlated with the clinical appearance of a fine hairpin vascular pattern (Fig. 1E,F).
For SCC (cases 5-9), pleomorphic cells were observed with dense cytoplasm, coarse chromatin and high nuclear to cytoplasmic ratio (Fig. 2B,D,F,H,J). The neoplastic cells invaded the basement membrane and reached the subepithelial connective tissue. Loss of cellular polarity was a consistent characteristic of all cases of SCC. Multiple perforating vessels, observed in cases 5 and 7, conformed to the clinical appearance of the papilliform type ( Fig. 2A,B,E,F, respectively). The inflammatory response was usually detected at the connective tissue adjacent to the tumor margin ( Fig. 2B,H,J). Although the clinical presentation of CIS, CIN and SCC was similar in appearance, the pathological findings clearly differed.
Immunohistochemistry. In normal limbal tissues, we observed a well-organized stratified epithelium with intact cellular polarity (Fig. 3A). The palisade of Vogt structure was clearly identified by the finger-like figure extending into the subepithelial connective tissue. In the normal limbus, ABCB5-positive cells were confined to the basal epithelial layer and were hardly observed in the suprabasal epithelial layer (Fig. 3B). An isotype-matched control showed negative expression of ABCB5 in the epithelial layer, with some background staining at the subepithelial connective tissue (Fig. 3C).
In all 9 OSSNs, ABCB5 expression patterns appeared different from those found in normal limbus, most of them showing significant ABCB5 overexpression as compared to normal limbus (Fig. 4). Moreover, the pattern of ABCB5 expression in OSSN was not confined to the basal epithelial layer but also appeared at the suprabasal and superficial cells. Three specific patterns were observed: 1) a perivascular pattern, in which high expression of ABCB5 was observed in a particular area surrounding the vascular structures in cases 3, 5 and 7 (Fig. 4C,E,G, respectively); 2) a marginal pattern, in which high ABCB5 expression was observed at the basal margin of tumors in cases 1, 2 and 7 ( Fig. 4A,B,G, respectively) and 3) a diffuse pattern, in which ABCB5 was diffusely expressed in all epithelial layers in cases 4, 6, 8 and 9 (Fig. 4D,F,H,I). The intensity and pattern of ABCB5 expression differed significantly from normal limbus, but we could find no significant difference in ABCB5 expression pattern among CIN, CIS and SCC.

Discussion
Based on the immunohistochemical studies, restricted expression of ABCB5 was found in normal limbal epithelium at the basal epithelial layer, aligning with the recent study by Ksander et al. 17 . It was known that OSSN develops predominantly in the limbal region and is histologically composed of the group of dysplastic cells originating from the basal epithelial layer, suggesting the possibility of transdifferentiation from limbal stem cell. However, this concept has not been experimentally proved. Within the limitations of small samples, the present study demonstrated the particular subpopulation in OSSN, using the immunohistochemical study of ABCB5. Interestingly, this aberrant pattern of ABCB5 expression was observed in all OSSNs. Although three specific patterns were detected, ABCB5 expression in OSSN presented most often as a non-specific pattern through the whole thickness of the tumor (Fig. 4). We could find no relationship between patterns of expression and tumor progression. Further tumor growth studies using in vitro colony-forming assay and in vivo xenotransplant experiments will be required to identify cancer stem cells or the cell lineage of OSSN. Grimm et al. have demonstrated that ABCB5 is significantly overexpressed in human oral squamous cell carcinoma (OSCC) as compared to normal oral squamous epithelium, which is confined to the basal cell layer 18 . In addition, ABCB5 expression was also associated with tumor progression and survival rate in cases of OSCC 18 . As for OSCC, ABCB5 overexpression was also found in OSSN. Given the limited sample and the various expression patterns of ABCB5 observed in this study, the relationship between ABCB5 expression and tumor progression  could not be properly evaluated. Additional studies, including a greater number of clinical cases, are therefore needed to clarify this relationship.
During cancer therapy, ABCB5 has been demonstrated to be involved in chemoresistant stem cell, as well as drug efflux transportation of doxorubicin in human melanoma 12,19,20 and chemoresistance to 5-Fluorouracil (5-FU) in colorectal cancer 21 . Mitomycin C, 5-FU and interferon are currently considered to be the main options for treating OSSN. Recurrence following 5-FU therapy has rarely been reported, and re-administration of the same agent remains effective for treatment in cases of recurrence 22,23 . In our series, there were 2 OSSNs with tumor recurrence; one received postoperative adjuvant 5-FU while the other did not. Both were successfully managed by means of repeated surgical en bloc excision. The interesting questions are whether ABCB5 determines drug efflux in OSSN and why multidrug resistance is rarely found in cases of OSSN where ABCB5 is highly expressed. We hypothesized that ABCB5 might not be involved in the chemoresistance mechanism of tumor cells in OSSN, but more work is needed to conclusively establish this.
Taken together, the clinical findings, histopathology and immunohistochemical study demonstrated first of all that the elevated ABCB5 expression pattern in all OSSNs suggests up-regulation and misarrangement of specific subpopulations containing some undifferentiated stem cell-like phenotype. Our findings provide an initial basis for further exploration of the particular subpopulation in relation to the pathogenesis and advanced treatment of OSSN.

Materials and Methods
Subjects. Nine pathological specimens were obtained from OSSN patients who underwent surgery at Kyoto Prefectural Hospital University of Medicine (KPUM). The research protocol was approved by the institutional review boards (IRBs) for Human Studies of Kyoto Prefectural University of Medicine, and prior informed consent was obtained from all patients in accordance with the tenets set forth in the Declaration of Helsinki for research involving human subjects. Demographic (age, gender and laterality), clinical presentation, treatment and follow-up data were reviewed. Anterior segment photographs were collected and analyzed by two cornea specialists (C.Z. and T.I.). From histopathology, the OSSNs were classified into three groups (CIS, CIS and SCC). CIN was  no basement membrane invasion. SCC was defined by anaplastic change of stratified squamous epithelial cells extending into the subepithelial connective tissue. For normal control, three corneoscleral rims were obtained from US corneal donors (SightLife Eye Bank, USA).

Immunohistochemistry (IHC).
Immunohistochemical studies were performed according to our previously described method 24,25 . Briefly, pathological specimens were immediately embedded in OCT compound and stored at − 80 °C following the operation. Eight μ m frozen sections were placed on silanized slides (Dako, Glostrup, Denmark) and allowed to dry before fixing with 100% acetone at 4 °C for 10 minutes. After washing with phosphate-buffered saline (PBS) and 0.15% TRITON X-100 (Dow Chemical Company, MI, USA) for 10 minutes each, slides were subsequently blocked with 2% bovine serum albumin (BSA; Sigma-Aldrich Co. LLC., MO, USA), diluted by goat serum (Jackson ImmunoResearch Laboratories, Inc., PA, USA) at room temperature (RT) for 1 hour to avoid nonspecific antibodies. The sections were then incubated with primary antibodies (Rabbit anti-human ABCB5 polyclonal antibody NBP1-50547 × 500; Novus Biologicals, CO, USA) at RT for 30 minutes and washed with 0.15% TRITON X-100 for 10 minutes (x 2), followed by PBS for 10 minutes. Isotype control at the same concentration as the primary antibodies was performed to exclude non-specific staining. Appropriated secondary antibodies (Goat anti-rabbit 488; MP Biomedicals, CA, USA) were incubated for 30 minutes. After repeat washing (as in the primary antibody process), all slides were mounted in propidium iodide (PI; Vectashield; Vector Laboratories Inc., CA, USA) and then examined under a confocal microscope (Fluoview FV1000; Olympus Co., Tokyo, Japan).