Main

Initially discovered as chemotherapeutic drug-efflux pumps, ATP-binding cassette (ABC) superfamily of transporters represents the largest family of transmembrane proteins.1 On the basis of the organization and sequence of the nucleotide-binding domain, this superfamily is classified into seven distinct families (A–G) in humans.2 Although the conserved nucleotide-binding domains of these transporters drive transport, the more variable transmembrane domains create the translocation pathway, providing safe passage for multiple, different substrates.1, 2 The ABC superfamily normally functions in the ATP-dependent transport of structurally diverse molecules and participate in tissue differentiation and survival in various organ systems as well.1, 2 There are 48 ABC genes in the human genome.1 Although the function of only some of these transporters is known, a number of them are implicated in multidrug resistance and are recognized causes for failure of cancer chemotherapy.1 This is the case of ATP-dependent drug efflux transporter P-glycoprotein (P-gp or ABCB1) whose expression on tumor cells has been identified to be a well-recognized mechanism of cancer multidrug resistance by decreasing intracellular drug accumulation.1

The ABC transporter superfamily includes ABCB5 and ABCF2. ABCB5 with its several distinct forms is expressed in various tissues, including testis, mammary tissue, melanocytes and retinal pigmented epithelium, and has also been shown to have enhanced expression in several cancer subtypes, including breast cancer, malignant melanoma, colorectal cancer and hepatocellular carcinoma, when compared with their normal tissue counterpart.1, 3, 4, 5, 6, 7, 8 The β form of ABCB5 (ABCB5.a) specifically is a half-transporter that has been shown to be expressed in melanocytes and melanoma.1 It has been shown to be a determinant of membrane potential and regulator of cell fusion in physiological skin progenitor cells.9 In melanoma, gene expression profiling of ABC transporters has revealed a high mRNA expression of melanocyte-specific ABCB5 and ABCF2 in melanoma cell lines.10 Furthermore, ABCB5 has recently been shown to be highly expressed in a melanoma stem cell-enriched population, especially CD133+ melanoma stem cells.3 These ABCB5+ melanoma stem cells were associated with melanoma progression and believed to mediate resistance to doxorubicin.3 Ablation of this melanoma stem cell population through ABCB5 has been shown to inhibit tumor initiation and growth in preclinical xenotransplantation models, suggesting that it might be of clinical utility as a potential therapeutic target.3

Nestin, an intermediate filament protein expressed in the cytoplasm of neuroepithelial cells, is known to be a marker for neural progenitor cells in the central nervous system.11, 12, 13, 14 Recently described as a melanocytic stem cell marker,12 nestin has been shown to have a significantly higher expression in melanomas compared with nevi and in metastatic melanomas compared with primary melanomas.12, 13, 14

The primary aim of this study was to ascertain immunohistochemical expression of the ABC transporters ABCB5 and ABCF2 and, the stem cell marker, nestin in a spectrum of benign and malignant nevomelanocytic proliferations that ranged from banal nevi to invasive primary cutaneous melanomas to assess whether they played a role in the development of malignancy. In addition, expression of these markers was compared with established melanoma prognosticators to ascertain their utility as independent prognosticators.

Materials and methods

Study Samples

The study was approved by the Boston University School of Medicine institutional review board (IRB No. H-30441). Archival materials were retrieved from the files of the Skin Pathology Laboratory, Boston University School of Medicine (Boston, MA, USA). A total of 85 cases fixed with formalin and embedded in paraffin were chosen. These included cases of primary cutaneous malignant melanoma (n=24), malignant melanoma in situ (n=31) and benign nevi (n=30), all from chronic/intermittent sun-exposed sites. Histological sections of all cases were reviewed and the diagnoses confirmed by the dermatopathologist (MM). All patient data were de-identified. Data regarding patient demographics (patient age and gender) and histopathology (melanoma prognostic factors, including Clark's level, Breslow's thickness, mitoses, ulceration, lymphovascular invasion and regression) were extracted from patient's medical records (Table 1).

Table 1 Demographics for cases of malignant melanoma in study

Immunohistochemistry

Sections were deparaffinized and peroxidase quenched with 0.3% H2O2 for 5 min. Antigen retrieval was performed in a steamer containing pre-heated antigen retrieval buffer solution. Following washing with Tween-20, sections were incubated with rabbit polyclonal antibody for 30, 60 and 20 min for ABCB5, ABCF2 and nestin, respectively, and washed in Tween-20. Dual-link HRP polymer was applied for 30 min and sections were again washed with Tween-20. The color was developed by 5 min incubation with DAB solution and sections were counterstained with hematoxylin. Primary antibodies against ABCB5 (rabbit affinity isolated polyclonal antibody; dilution 1:160; Sigma-Aldrich, St Louis, MO, USA), ABCBF2 (rabbit affinity isolated polyclonal antibody; dilution 1:400; Sigma-Aldrich) and nestin (MAB; dilution 1:100; Millipore, Billerica, MA, USA) were performed. Appropriate positive and negative controls were included. Normal prostate tissue with strong granular cytoplasmic staining served as the positive control for ABCB5 antibody, normal pancreatic tissue sections with strong cytoplasmic and membranous staining of islet cells and acinar cells served as the positive control for ABCF2 and normal skin with cytoplasmic expression in endothelial cells and bulge area of the hair follicle served as the positive control for nestin.

Immunohistochemical expression was evaluated with a semiquantitative scoring system by percentage of positive cells and intensity and using the following scale for percentage positivity: 0=<10%, 1+=10–25%, 2+=26–50%, 3+=51–75% and 4+=>76–100%; and the following scale for intensity: 1+=weak/equivalent to the staining of epidermis, 2+=moderate staining and 3+=strong staining. Proportion scoring was performed only if intensity of tumor cell staining was more than that of overlying epidermis. Histological sections of all cases were reviewed by two board-certified dermatopathologists (initial sign-out on all by a dermatopathologist; cases were then re-reviewed and the diagnoses confirmed by the senior author). All stained slides were initially reviewed and scored by the first author (NS) and re-reviewed by the dermatopathologist (MM) in a blinded manner with respect to diagnosis to ensure consistency of interpretation.

Statistical Analysis

The Pearson correlation coefficient was used to assess the association of ABCB5 and nestin levels in each of the three types of lesions. Mean expression levels for ABCB5 and nestin were compared among lesion types using one-way ANOVA with linear contrasts (for malignant melanoma in situ and malignant melanoma groups combined compared with nevi and malignant melanoma in situ compared with malignant melanoma.

Results

ABCB5

Positive staining for ABCB5 staining was considered by ascertaining cytoplasmic expression and any nuclear staining was considered background artifact. In normal skin, weak ABCB5 cytoplasmic expression was observed in keratinocytes in the uninvolved epidermis, endothelial cells, eccrine glands, smooth muscle of vessel wall, bulb of hair follicle, arrector piloris muscle and nuclear staining of sebocytes, which served as positive internal controls in each case where they could be visualized.

Mean cumulative score (percentage of positive cells staining+intensity of staining) for ABCB5 expression based on lesion type was as follows: 5.3 (s.d. 1.9) for malignant melanoma, 4.4 (s.d. 0.6) for malignant melanoma in situ and 3.8 (s.d. 0.6) for benign nevi (Figures 1 and 2). Differences achieved statistical significance in between all three groups (analysis of variance (ANOVA) showed P=0.0047), between melanoma (malignant melanoma in situ and malignant melanoma together) and nevi (P=0.0011), and between malignant melanoma and malignant melanoma in situ (P=0.05).

Figure 1
figure 1

Tumoral expression of ABCB5 and nestin in benign nevi (a–c), malignant melanoma in situ (d–f) and malignant melanoma (g–i). a, d, g=hematoxylin and eosin (H&E) ( × 10); b, e, h=ABCB5 ( × 10); and c, f, i=nestin ( × 10). ABC, ATP-binding cassette.

Figure 2
figure 2

Comparison of mean cumulative score of ABCB5 in malignant melanoma, malignant melanoma in situ and benign nevi. *Statistically significant (P=0.0047). ABC, ATP-binding cassette.

ABCF2

ABCF2 staining was considered positive by ascertaining cytoplasmic expression and any nuclear staining was considered background artifact. In normal skin, weak ABCF2 cytoplasmic expression was observed in the uninvolved epidermis (higher intensity in stratum corneum compared with basal layer), lymphocytes, endothelial cells, eccrine ducts, sebocytes, bulb of hair follicles and perinuclear staining in arrector pili muscle, which served as positive internal controls in each case where they could be visualized.

The staining of ABCF2 was either equivalent to or less than that of the epidermis in all three groups. Proportionate staining was to be performed only if the intensity of staining of lesional cells was greater than the epidermis. As a result, the cumulative score for ABCF2 staining was 1+ in all three groups. There was no difference in the pattern/intensity and proportion of the staining in malignant melanoma, malignant melanoma in situ and nevi.

Nestin

Nestin staining was also considered positive by ascertaining cytoplasmic expression and any nuclear staining was considered background artifact. In normal skin, nestin expression was observed in endothelial cell cytoplasm and bulge area of the hair follicle, which served as positive internal control in each case where they could be visualized.

Mean cumulative score for nestin expression based on lesion type was as follows: 5.3 (s.d. 0.8) for malignant melanoma, 4.6 (s.d. 0.8) for malignant melanoma in situ and 4.6 (s.d. 0.6) for benign nevi (Figures 1 and 3). There was no significant difference among the three groups in mean nestin expression in a one-way ANOVA (P=0.3128).

Figure 3
figure 3

Comparison of mean cumulative score of nestin in malignant melanoma, malignant melanoma in situ and benign nevi. P=not significant for all.

We also attempted to correlate the expression of ABCB5 and nestin in the different groups (Table 2). There was a significant correlation in their expression in malignant melanoma in situ and benign nevi, with a much higher correlation in nevi in comparison with malignant melanoma in situ. There was no significant correlation in their expression in malignant melanoma.

Table 2 Correlation of ABCB5 and nestin expression by lesion type

Correlation of ABCB5 and Nestin with Known Melanoma Prognosticators

Further, we attempted to correlate the expression of ABCB5 and nestin with age, gender, Clark's level, Breslow's thickness, presence of radial/vertical growth phase, mitoses, ulceration, lymphovascular invasion and regression. In ulcerated lesions, ABCB5 expression was lower than in non-ulcerated lesions (Table 3). After controlling for the presence of ulceration, the expression of ABCB5 did not significantly correlate with any of the known prognostic factors. Likewise, nestin expression was lower in ulcerated and lesions with mitoses >1 (P<0.001 and <0.026, respectively). After controlling for the presence of ulceration and mitoses, nestin expression also did not significantly correlate with any of the known prognostic factors.

Table 3 Comparison of mean cumulative score of ABCB5 and nestin in ulcerated and non-ulcerated cases of malignant melanoma and in cases of malignant melanoma with mitoses <1 and >1

Discussion

A role for ABCB5 in the pathogenesis and development of chemoresistance has been demonstrated previously.1, 3 Findings from this study extend the functionality of ABCB5 to include melanomagenesis as we found significantly increased expression of ABCB5 in in situ and invasive melanoma as a group compared with benign nevi. While Sharma et al11 and Schatton et al5 also observed overexpression of ABCB5 in primary melanoma, in terms of primary lesions, both study groups were limited to nevi and melanoma.5, 11 Our observation of a higher level of expression of ABCB5 in invasive compared with in situ melanoma and in situ melanoma compared with benign nevi indicate that ABCB5 is a key player in the stepwise development of melanoma.

In light of our findings, we then attempted to ascertain the utility of ABCB5 as an independent prognosticator. We found that in patients with invasive melanoma, expression of ABCB5 did not correlate with age, sex, Clark's level, presence of radial/vertical growth phase, mitoses, lymphovascular invasion and regression. In ulcerated lesions, while expression of ABCB5 was lower than in non-ulcerated lesions, after controlling for the presence of ulceration, the expression of ABCB5 did not significantly correlate with any of the known melanoma prognosticators. Although this diminishes the utility of ABCB5 as an independent prognosticator, ulceration is an established, independent prognosticator in primary cutaneous melanoma. Thus, an alteration in expression relative to this parameter implies biological significance. However, possible limiting factors of this study include the cross-sectional nature and moderately small size of this study set. Thus, biological relevance can only be ascertained in a longitudinal study with an increased number of cases.

From a clinical perspective, the significance of ABCB5 expression appears to be intimately related to drug resistance.3, 9 It has been shown that ABCB5 marks cells with the CD133 immunophenotype, a marker defining progenitor cells among human melanocytes. In the same study, the authors demonstrated that ABCB5 determines as a regulator of membrane potential the propensity of this subpopulation to undergo cell fusion.9 In physiological progenitor cells, ABCB5 functions to maintain membrane hyperpolarization, associated with multidrug resistance phenotype of cancer cells including melanoma.3 It is believed that ABCB5 mediates resistance to chemotherapeutic agents such as doxorubicin through drug efflux leading to decreased intracellular drug accumulation.3 Blocking ABCB5 significantly reverses resistance of melanoma cell lines to doxorubicin, identifying ABCB5 as a novel drug transporter and mediator of chemoresistance in malignant melanoma. A similar role of ABCB5 has also been implicated in other cancers such as hepatocellular carcinoma8 and colorectal carcinoma.7 More recently, using serial human-to-mouse xenotransplantation experiments, Schatton et al5 identified a subpopulation of cancer stem cells called malignant-melanoma-initiating cells based on their expression of ABCB5. They established that ABCB5+ cancer cells, comprising 2–20% of the entire tumor cell population, had enhanced tumorigenicity and possessed an exclusive capacity to self-renew and to give rise to more differentiated melanoma cells. More importantly, using tissue microarrays, a positive correlation was demonstrated between neoplastic progression in human melanoma patients and the abundance of this cancer stem cell subset.5 Using flow cytometric analyses, the same group identified a subpopulation enriched for human malignant-melanoma-initiating cells defined by the expression of ABCB5. Using the technique of suppression subtractive hybridization to investigate the molecular signature of melanoma progression in cell lines, another group found preferential expression of the ABCB5 gene by metastatic melanomas as opposed to the primary tumor.15 On the basis of all of this evidence, it appears that ABCB5 in human malignant melanoma provides a unique direct link between cancer stem cells and cancer therapeutic resistance to select chemotherapeutic agents. Our results support a role for ABCB5 in neoplastic progression and the stepwise development of melanoma.16

Nestin, an intermediate filament protein expressed in the cytoplasm of neuroepithelial cells and a marker for neural progenitor cells in the central nervous system, has been described as a marker of melanocytic stem cells.11, 12, 13, 14 Previous studies have shown that there is an increased expression of this protein in malignant vs benign nevomelanocytic proliferations and that there appears to be a stepwise increase in the proportion of lesions expressing nestin from banal nevi to primary melanomas to metastatic melanoma.12, 13 More recently, nestin staining in stage I and II melanoma patients significantly predicted poor survival with lower rates in cases with nestin positivity in both tumoral and endothelial cells.14 Consistent with these findings, we observed enhanced expression of nestin in invasive melanoma compared with in situ melanoma and nevi, although differences did not quite achieve statistical significance. From a scientific perspective, nestin expression in nevi suggests that it may not be necessarily associated with tissue dedifferentiation and indicates that nestin cannot be used to distinguish benign from malignant melanocytic lesions. We observed that expression of nestin, like that of ABCB5, did not correlate with age, sex, Clark's level, presence of radial/vertical growth phase, mitoses, lymphovascular invasion and regression. In ulcerated lesions and lesion with mitotic count >1, nestin expression was lower than that in non-ulcerated lesions. However, after controlling for the presence of ulceration and mitotic count, nestin expression did not significantly correlate with any of the known prognosticators. Of the three sample groups studied, a significant correlation between expression of ABCB5 and nestin was observed by us in nevi and malignant melanoma in situ but not in invasive melanoma, the precise significance of which is not, albeit as yet, clear.

Although it has a structure different from other members as it contains a pair of nucleotide-binding domains without transmembrane domains, ABCF2 is yet another member of the ABC transporter family whose function in cancer cells is still unclear.17, 18, 19 While one study suggested that ABCF2 may contribute to the chemoresistant cancer phenotype as it was overexpressed in chemoresistant clear-cell ovarian carcinoma, another on breast cancer suggested that expression of ABCF2 might indicate good prognosis.18, 19 Recently, Heimrl et al10 observed a striking upregulation of ABCF2 at the transcriptional mRNA level in primary and metastatic melanoma cell lines compared with normal human epidermal primary melanocytes, raising the possibility that it may play a role in the development and progression of malignant melanoma. However, similar findings at the translational level were not replicated in our study as tumor cells in all three groups exhibited equivalent or less protein expression than the epidermis, which, in keeping with the manufacturer's observations, showed very strong ABCF2 staining. Thus, our findings suggest that ABCF2 is not related to the pathogenesis of melanocytic proliferations.

ABCG2 is another ATP transporter expressed in various cancer stem cells lines and has been linked to prognosis, disease-free survival and response to chemotherapy in acute myelogenous leukemia. Its role in chemoresistance has also been studied in breast carcinoma, non-small-cell lung carcinoma, oral squamous cell carcinoma and advanced testicular germ cell tumors.20 Contradictory results have been found regarding its expression in melanoma cell lines.21, 22 Deichmann et al21 studied 18 melanoma resection specimens and could not demonstrate ABCG2 mRNA upregulation by reverse transcriptase polymerase chain reaction, while Monzani et al22 found in vitro and in vivo ABCG2 expression in all the cells expressing CD133 by reverse transcriptase polymerase chain reaction and immunohistochemistry.

Although of limited value as an independent prognosticator, in this cross-sectional study the gradual increase in expression of ABCB5 from benign nevus to in situ to invasive melanoma suggests that it plays a role in melanomagenesis. On the basis of our findings, a prospective study with follow-up data is required to ascertain the utility of ABCB5 as a therapeutic target.