FK506 binding protein 51 positively regulates melanoma stemness and metastatic potential

Melanoma is the most aggressive skin cancer; there is no cure in advanced stages. Identifying molecular participants in melanoma progression may provide useful diagnostic and therapeutic tools. FK506 binding protein 51 (FKBP51), an immunophilin with a relevant role in developmental stages, is highly expressed in melanoma and correlates with aggressiveness and therapy resistance. We hypothesized a role for FKBP51 in melanoma invasive behaviour. FKBP51 promoted activation of epithelial-to-mesenchymal transition (EMT) genes and improved melanoma cell migration and invasion. In addition, FKBP51 induced some melanoma stem cell (MCSC) genes. Purified MCSCs expressed high EMT genes levels, suggesting that genetic programs of EMT and MCSCs overlap. Immunohistochemistry of samples from patients showed intense FKBP51 nuclear signal and cytoplasmic positivity for the stem cell marker nestin in extravasating melanoma cells and metastatic brains. In addition, FKBP51 targeting by small interfering RNA (siRNA) prevented the massive metastatic substitution of liver and lung in a mouse model of experimental metastasis. The present study provides evidence that the genetic programs of cancer stemness and invasiveness overlap in melanoma, and that FKBP51 plays a pivotal role in sustaining such a program.

Melanoma is the most aggressive skin cancer; there is no cure in advanced stages. Identifying molecular participants in melanoma progression may provide useful diagnostic and therapeutic tools. FK506 binding protein 51 (FKBP51), an immunophilin with a relevant role in developmental stages, is highly expressed in melanoma and correlates with aggressiveness and therapy resistance. We hypothesized a role for FKBP51 in melanoma invasive behaviour. FKBP51 promoted activation of epithelial-to-mesenchymal transition (EMT) genes and improved melanoma cell migration and invasion. In addition, FKBP51 induced some melanoma stem cell (MCSC) genes. Purified MCSCs expressed high EMT genes levels, suggesting that genetic programs of EMT and MCSCs overlap. Immunohistochemistry of samples from patients showed intense FKBP51 nuclear signal and cytoplasmic positivity for the stem cell marker nestin in extravasating melanoma cells and metastatic brains. In addition, FKBP51 targeting by small interfering RNA (siRNA) prevented the massive metastatic substitution of liver and lung in a mouse model of experimental metastasis. The present study provides evidence that the genetic programs of cancer stemness and invasiveness overlap in melanoma, and that FKBP51 plays a pivotal role in sustaining such a program. Melanoma is a dynamic and heterogeneous skin neoplasia with adverse prognosis in advanced stages. Despite the large body of studies published during the past decade that have contributed to the advance of knowledge in this field, 1-4 the complex biology of this tumour remains obscure. Identifying molecular players with a determinant role in tumour progression may provide useful diagnostic and therapeutic tools and greatly improve treatment outcomes, thereby changing the fate of this unpredictable tumour. We have recently shown that FK506 binding protein 51 (FKBP51), a protein involved in cancer chemoresistance [5][6][7][8][9][10][11] is a promising target for radiosensitization strategies in melanoma. 8 FKBP51 is highly expressed in melanoma. 5,8 Such expression correlates with tumour aggressiveness and is maximal in metastatic lesions. 8 Moreover, FKBP51 overexpression was apparently associated with a melanoma cancer stem cell (MCSC) phenotype. 8 FKBP51 (FKBP5 official symbol) is a large molecular weight component of the family of FK506 binding proteins (FKBPs), classically known as the intracellular receptors for immunosuppressants FK506 and rapamycin. 12,13 FKBPs are multifunctional proteins that modulate several signal transduction pathways. [12][13][14][15] Although the knowledge of FKBP structure and function is far from complete, recent reports reveal that this class of proteins affects gene expression, DNA repair, and DNA replication (reviewed in Yao et al. 15 ). These functions are mediated by histone chaperone activity, regulation of transcription factors, and modifications of chromatin structure. 15 In accordance with these concepts, FKBP51 plays a relevant role in developmental stages in both mammals [16][17][18] and other organisms. 19 FKBP51 is among the top candidate genes expressed in mesenchymal stem cells, 18 particularly during the mitotically active phase that precedes differentiation into the three mesodermal lineages, namely osteogenesis/chondrogenesis/adipogenesis. 16,17 Similarly, PAS-1 (PASTICCINO1 gene), the FKBP51 homologue in Arabidopsis, plays a critical role in the growth and differentiation of this organism. 19 Recent breakthroughs suggest that genes active in developmental processes are relevant to cancer biology and play a determinant role in reprogramming cancer cells to activate stemness. 20 Notably, cancer stem cells (CSCs) have the ability to self-renew, as well as to differentiate into more mature cancer cells. This conversion is bidirectional within certain limits, resulting in dynamic variation of CSC abundance and heterogeneity of the tumour due to the formation of diverse subpopulations that support tumour growth in different ways. 20 Epithelial-tomesenchymal transition (EMT) is a normal process in embryonic development in which epithelial cells transform into mesenchymal cells, the highly mobile cells that give rise to bone, muscle, connective tissue, and blood vessels. 21 Cancer cells require EMT to spread and metastasize. 22 Induction of the EMT transdifferentiation programme in cancer cells enables them to not only disseminate from primary tumours and form metastases, but also to acquire selfrenewal capability (i.e., CSC phenotype), which is crucial to their subsequent expansion at sites of dissemination. 22,23 We hypothesized a role for FKBP51 in the regulation of the melanoma stemness phenotype and invasive behaviour.

Results
FKBP51 positively regulates EMT hallmarks in melanoma. The transforming growth factor-b (TGF-b) is a key regulator of EMT. [24][25][26] Furthermore, a critical role in melanoma tumour cell plasticity is covered by the type III TGF-b receptor (TbRIII or betaglycan or endoglin). 27 We measured the mRNA and protein levels of TGF-b and TbRIII in FKBP51-depleted or FKBP51-overexpressing SAN melanoma cells. TGF-b protein was investigated with western blot (Figures 1a and c; right) and TbRIII protein was investigated Filter was stained with Crystal Violet. OD value was obtained by reading the eluted Crystal Violet with a spectrophotometer at a wavelength of 570 nm. The increased OD in FKBP51 samples indicated that FKBP51-overexpressing cells invaded matrigel more efficiently than control cells. The data represent the means of three independent experiments performed in triplicate  Figure 1a). Conversely, enhancement of TGF-b expression was observed in FKBP51-overexpressing SAN melanoma cells ( Figure 1c). TbRIII levels were modulated in a similar manner (Figures 1b and d). FKBP51-overexpressing cells showed an increased proliferation (Figure 1e). We performed soft agar colony formation assays to investigate the capability of FKBP51-overexpressing melanoma cells to promote anchorage-independent growth, a typical feature of EMT. 8 Figure 1f shows that FKBP51 strikingly stimulated soft agar growth, as suggested by the colony number and size that were remarkably increased in FKBP51-overexpressing cells. In addition, using Boyden chamber and matrigel, 28 we found that the ability of melanoma cells to migrate through transwell filters and invade matrigel was enhanced by FKBP51-overexpression ( Figure 1g). As expected, invasion was remarkably reduced in FKBP51-silenced melanoma (Supplementary Figure S1). Overall, our findings suggested that FKBP51 promotes activation of EMT features in melanoma. We confirmed that FKBP51 knockdown resulted in TGF-b and TbRIII decrease in A375 melanoma cell line also, using a different small interfering RNA (siRNA) (Supplementary Figure S2).
FKBP51 regulates expression of melanoma stemness markers. Normal stem cells have high drug efflux capability because of the expression levels of ATP binding cassette (ABC) transporters, which actively pump drugs out of the cell. ABC, subfamily B, member 5 (ABCB5) 29 and ABC, subfamily G, member 2 (ABCG2) 30 are members of this superfamily of proteins, and are known molecular determinants of cancer stem/initiating cells of melanoma. 29,30 We investigated if FKBP51 modulated MCSC phenotype. To this end, we silenced FKBP51 with siRNA and measured mRNA levels of ABCB5 using real-time PCR. FKBP51 siRNA produced a 10-fold decrease in FKBP51 mRNA levels (black histogram) and a threefold decrease in ABCB5 mRNA (grey histogram), compared with cells transfected with a nonsilencing (NS) RNA (Figure 2a, left). We confirmed that FKBP51 knockdown resulted in ABCB5 decrease in several melanoma cell lines, using a different siRNA (Supplementary Figure S3). Conversely, in SAN melanoma cells overexpressing FKBP51, a sevenfold increase in ABCB5 mRNA (grey histogram) was measured (Figure 2b; left), compared with wild-type (WT) or empty-vector stably transfected cell (EV) mRNA levels. Similarly, ABCB5 expression levels, measured by western blot, decreased and increased in accordance with FKBP51 levels (Figures 2a and b; right). In the same way, ABCG2 mRNA was upregulated in melanoma cells that over expressed FKBP51 (Figure 2c, upper). Most interestingly, flow cytometry demonstrated that the cell count of ABCG2positive cells was increased 10-fold when FKBP51 was overexpressed (Figure 2c, lower). Furthermore, ABCG2 þ sorted cells from two different cell lines, SAN and A375 (Figure 2d, upper and lower, green histogram), expressed very high levels of FKBP51 mRNA compared with unsorted cells (Figure 2c, pink histogram). These results suggested that FKBP51 upregulates the melanoma stemness phenotype. Additional clones were analysed for TGF-b, TbRIII, and stemness marker expression (Supplementary Figure S4). To confirm the stemness feature of ABCG2 þ sorted cells, we performed a genome-wide study of cancer stem cell gene expression on these cells using a stem cell real-time PCR array (RT2 profiler PCR Array-Human Stem Cell signalling pathway; Qiagen Sciences, Philadelphia, PA, USA). ABCG2 þ cells expressed increased levels (more than twofold increase) of 23 stem cell gene transcripts, including Notch, Wnt, and bone morphogenetic protein (Supplementary Figure S5). As FKBP51 has shown to positively regulate EMT, and ABCG2 þ melanoma cells contained high FKBP51 levels, we hypothesized that ABCG2 þ cells expressed EMT genes. To verify this hypothesis, we measured, the expression of TGF-b, endoglin, and other EMT genes (VIM, TWIST1, SNAI1, SNAI2, CDH2) 31-34 in sorted ABCG2 þ melanoma cells. These cells exhibited expression levels of EMT genes enhanced by 5-to 40-fold compared with ABCG2 À cells ( Figure 2e). As expected, CDH1 expression was reduced. These findings suggested that MCSC and EMT genetic programs were simultaneous active in ABCG2 þ cells. The strict relationship between MCSCs and metastasis was further reinforced by the finding that deparaffinized melanoma metastasis samples contained ABCG2 levels significantly higher than primary melanoma samples. A naevus sample was arbitrarily chosen as reference sample (expression ¼ 1) (Figure 2f). This finding is in accordance with the concept that metastasis requires activation of MCSC genes to spread and generate secondary tumours. 20,22,23 Invading melanoma cell coexpresses stemness markers and FKBP51. To address the relationship between FKBP51, stemness, and metastasis in melanoma patients, we performed double immunochemical staining of FKBP51 and nestin, an intermediate filament expressed in the cytoplasm of neuroepithelial stem cells 35 in a series of 10 primary cutaneous and 20 brain melanoma metastases. We found melanoma cells with nuclear FKBP51 and membranecytoplasmic nestin in the vertical (invasive, intradermal) growth phase of primary melanomas. Figure    Post-mortem organ examination showed a dramatic difference between the morphologically preserved organs dissected from FKBP51 siRNA-treated mice, and severely compromised organs from untreated or NS RNA-treated mice (Figure 4b, left). The histological examination confirmed that melanoma metastasis extensively substituted the lungs and liver of untreated or NS RNA-treated animals (Figure 4b, right). In contrast, few tumoural aggregates could be visualized in the lungs and livers of FKBP51 siRNA-treated mice (Figure 4b, right). Most interestingly, invading cells coexpressed FKBP51 and nestin (Figure 5a). A quantitative estimate of organ invasion was performed and is shown in Figure 5b. Measurement of FKBP51 transcript levels in lungs and livers confirmed reduced melanoma colonization of FKBP51 siRNA-treated mice.
FKBP51 binds to chromatin. Several lines of evidence support a role for FKBPs as transcriptional coregulators (reviewed in Yao et al. 15 ). As we have found that FKBP51 modulated expression of several genes at transcriptional level, we hypothesized a role for FKBP51 as a transcriptional coregulator. To investigate whether the immunophilin interacted with DNA, we performed chromatin  Figure 6a. We found enriched DNA amplification bands in anti-Flag-immunoprecipitated samples, which suggested that FKBP51, similar to the zincfinger protein, binds to chromatin. We then attempted to address whether the FKBP51-IP chromatin included a promoter of a gene, whose expression appeared to be regulated by the immunophilin. Among the regulated genes, we choose ABCG2, because the transcript of this gene was found very abundant in melanoma metastasis samples from patients. We constructed primers to amplify different regions of ABCG2 promoters. Our results showed enrichment in DNA fragments, size selected on an agarose gel, corresponding to three different regions of ABCG2 promoter: the first localized at 3450 bp ( À 3450) from the transcription start site (TSS), the second at À 3219, and the third at À 939 in FKBP51immunoprecipitated chromatin (Figure 6b). Differently, no enrichment of À 2897, À 2060, and À 127 regions was found, which is consistent with the specificity of FKBP51 binding (Figure 6b). This result suggests an interaction between FKBP51 and ABCG2 promoter. Because results from stem cell array (shown in Supplementary Figure S5, only a single small subcapsular nodule is visible in liver biopsy from FKBP51 siRNA-treated mouse; NS RNA mouse liver is completely substituted by multiple nodules FK506 binding protein 51 and melanoma progression S Romano et al gene 8) have suggested that ABCG2 þ cells contained increased levels of the general transcriptional co-activator, the E1A binding protein 300 (p300), we investigated whether FKBP51 interacted with this protein, which reinforced our hypothesis that immunophilin takes part in transcriptional complexes. A representative result of three independent experiments is shown in Figure 7a. We found that FKBP51 co-immunoprecipitated with p300, and, conversely, p300 co-immunoprecipitated with FKBP51, providing evidence that the two proteins interact with each other. In addition, we found DNA enrichment corresponding to the À 3450 sequence of ABCG2 promoter in p300-immunoprecipitated chromatin, which suggested that FKBP51 and p300 participated in the same transcriptional complex (Figure 7b). Interestingly, such enrichment was not observed in FKBP51-silenced cells, which suggests FKBP51 favours interaction between p300 and DNA. Overall, these results are consistent with our hypothesis that FKBP51 may act as a transcriptional coregulator.

Discussion
Melanoma often starts as a single skin tumour. Following primary surgical resection, this tumour can remain clinically quiescent for years but, unpredictably, malignant melanocytes can spread to distant locations and cause metastasis. The most common sites of melanoma metastases are in the subcutaneous tissue, lymph nodes, lungs, liver, brain, and bone. In most cases, there is no cure for advanced melanoma. The change during melanoma progression, that causes the switch from a locally growing tumour to a metastatic invader, remains obscure. In the present study, we show that FKBP51 plays a major role in promoting activation of genes involved in melanoma progression. FKBP51 increased expression of TGF-b and, consistent with the notion that this cytokine is the master regulator of EMT, 24-26 immunophilin promoted EMT features and improved migration and invasiveness of melanoma cells. In addition, FKBP51 regulated expression of MCSC markers at transcriptional level and the amount of malignant melanocytes with a cancer stem cell phenotype resulted significantly  increased when FKBP51 was overexpressed. Most interestingly, melanoma cells with cancer stem cell phenotype expressed higher levels of EMT gene transcripts in comparison with the rest of the melanoma cells. Conversely, metastatic melanoma expressed CSC markers at higher levels in comparison with primary melanoma. These findings suggest a strict relationship between MCSC phenotype and melanoma invasiveness, in accordance with a previous study. 29 Notably, appearance of stemness markers on melanoma cells can occur within reversible changes in a broad range of markers, including EMT markers, that turn on and off within tumour in a context of phenotypic heterogeneity and plasticity. 3,23 Several mechanisms, for example, drug resistance, 5,6,8,38 immune evasion, 39 and vasculogenic mimicry, 40 confer aggressiveness to these cells. In this perspective, FKBP51 could be an important regulator of melanoma plasticity. The finding that FKBP51 participates in transcriptional complexes supports this conclusion. Moreover, the interaction with p300 acetyltransferase is also in accordance with a role for FKBP51 in the chromatin changes required for melanoma cell reprogramming. The translational significance of our findings was strengthened by immunochemical evaluation of primary and metastatic melanoma specimens, suggesting that coexpression of FKBP51 and the stem cell marker nestin is a distinctive feature of the invading melanoma cell, which is equipped for extravasation, easily adaptable to the colonized milieu, and capable of subverting the microenvironment of distant sites. In fact, in primary melanoma, malignant melanocytes coexpressing FKBP51 and nestin become clearly visible in the intradermal, invasive growth phase; the same double-stained cells formed neoplastic emboli in dermal vessels of skin lesions. In advanced melanoma, malignant melanocytes in vascular and perivascular compartments were FKBP51/nestin double positive. Finally, malignant melanocytes totally subverting the brain tissue showed the same immunochemical feature. In an experimental model of melanoma metastasis, FKBP51 and nestin were coexpressed in malignant melanocytes that colonized murine organs. The important role of FKBP51 in melanoma progression was reinforced by studies in this mouse model of metastatic melanoma using FKBP51siRNA as a therapeutic approach. In vivo imaging and anatomohistopathological examination inequivocally showed that FKBP51 targeting dramatically reduced metastases in mice.
In conclusion, our study suggests that FKBP51 may be a suitable novel melanoma biomarker that can improve the risk classification of primary lesions. In addition, we provide a promising target for development of compounds that can prevent metastasis after surgical treatment, forecasting that prognosis of this unpredictable, insidious tumour, which is lethal in advanced stages, may change in the very near future.

Materials and Methods
Cell culture and transfection. The melanoma cell lines SAN, A375, G361, and SK-MEL-3 were cultured as described. 8 To create FKBP51-overexpressing SAN melanoma cells, a p3XFLAG-CMV-14 expression vector (Sigma Aldrich, St. Louis, MO, USA) carrying the FKBP51 gene was transfected using Metafectene (Biontex, Munich, Germany), according to the manufacturer's recommendations. A void p3xFlag-CMV vector was also transfected to generate control cells. To generate stable populations, cells were selected using 500 mg/ml G418 (GIBCO, Invitrogen, Carlsbad, CA, USA) at 24 h post transfection and grown until colony formation. Different clones were expanded and analysed by western blot using M2 anti-FLAG antibody (Sigma Aldrich) and real-time PCR using specific primers for FKBP51 (see real-time PCR section for primer sequences); three of these clones were chosen for further manipulation. For siRNA transfection, 24 h before transfection, cells were seeded into six-well plates at a concentration of 2 Â 105 cells/ml to obtain 30-60% confluence at the time of transfection. Then, cells were transfected with specific short-interfering oligoribonucleotide (siRNA) or with a NS oligoribonucleotide (NS RNA) as control at a final concentration of 50 nM using Metafectene according to the manufacturers' recommendations. NS RNA (AllStars neg control siRNA) and siRNA for FKBP51 (5 0 -ACCUAAUGCUGAGCUUAUA-3 0 ) were purchased from Qiagen. In experiments shown in Supplementary Figure S2 and S3, the sequence 5 0 -GCCGAUGAUUGGAGACAAA-3 0 (Dharmacon Thermo Scientific, Chicago, IL, USA) of FKBP51 siRNA (siRNA 2) was used.
FKBP51 cloning. A DNA fragment encoding the complete FKBP51 protein was amplified from PBMC cDNA, with the 5 0 forward: GGAATTCCGGCTGAAGGG TTAGCGGAGCAC and the 3 0 reverse: CGGGATCCCGTACGTGGCCCTCAGGTT TCTCTTC. The amplified fragments were cloned into EcoRI/BamHI sites of p3XFLAG-CMV-14. Half of the ligation was transformed into E.Coli Top-ten highcompetence bacteria. After 16-20 h of growth at 371C, 10-25 colonies/dish were obtained and expanded in L-Broth (5 ml) containing 50 mg/ml ampicillin. Clones containing inserts were subjected to digestion with EcoRI and BamHI, and to triple digestion with EcoRI, HindIII, and BamHI, to obtain FKBP51 cleavage in three differently sized fragments. Clones were sequence-verified by Primm (Milano, Italy).
Western blot and immunoprecipitation. Whole-cell lysates were homogenized in modified RIPA buffer as described 9 and assayed in western blot as described. 9 Primary antibodies against FKBP51 (F-13; goat polyclonal; Santa Cruz Biotechnology, Santa Cruz, CA, USA), TGF-b (V; rabbit polyclonal; Santa Cruz Biotechnology), and Smad 2/3 (H465, rabbit polyclonal Santa Cruz Biotechnology) were used diluted 1 : 500; anti-FKBP51 (rabbit polyclonal; Novus Biologicals, Littleton, CO, USA), anti-FKBP51 (mouse polyclonal; Abnova, Taipei, Taiwan), anti-FKBP12 (N-19, goat polyclonal; Santa Cruz Biotechnology), anti-actin (I-19, goat polyclonal; Santa Cruz Biotechnology), anti-G3PDH (D16H11; rabbit monoclonal; Cell Signaling, Danvers, MA, USA), and anti-KAT3B/p300 (RW109; mouse monoclonal, Novus Biologicals) were used 1 : 1000. Anti-ABCB5 (rabbit polyclonal; Rockland, Gilbertsville, PA, USA) was used 1 : 5000. For immunoprecipitation (IP), 500 mg of total lysate was precleared for 1 h. Then, 3 mg anti-KAT3B/p300 (Novus Biologicals) or anti-Flag (M2, mouse monoclonal, Sigma Aldrich) was added to total lysate, and kept in rotation at 41C , and 1.25% of Noble Agar (Difco) was poured first (7 ml); after solidifying, this was followed by a layer containing 2 ml of the same solution and 5 Â 104 cells. The plates were placed in the incubator and after 10 days, colonies were photographed and counted. The invasion assay was performed in accordance with Albini and Benelli. 28 Briefly, this method is a simple modification of Boyden chamber assay because of addition of matrigel, which mimics the physiological barrier to tumoural cells, allowing to evaluate both migratory and invasive capabilities of tumoural cells. Transwell chambers, with 8 mm polyester filters, were used for this assay (Corning Incorporated Life Sciences, Tewksbury, MA, USA) and coated with 100 ml of basement membrane matrix, matrigel (Becton Dickinson, Franklin Lakes, NJ, USA). The matrigel was diluted with cold PBS 1 Â , applied to the filters, dried in humidified atmosphere at 371C for at least 5 h, and reconstituted with serum-free medium. The coated filters were placed in Boyden chambers, forming two medium-filled compartments separated by such microporous membrane. A nutrient gradient was created between the two chambers (1% FCS medium in the upper and 10% FCS medium in the lower compartment). Melanoma cells (75 Â 103) were placed in the upper compartment and, after a 48-h incubation time, the membrane between the two compartments was fixed and stained with 0.25% Crystal Violet in methanol. Cells at the lower surface of the filter were photographed; then, the Crystal Violet was eluted in 1% SDS and read in a spectrophotometer at a wavelength of 570 nm.
Immunohistochemistry. The procedures followed for immunohistochemistry studies on human samples were in accordance with the ethical standards of the local institutional responsible committee on human experimentation and with the Helsinki Declaration of 1975, as revised in 1983. Immunostaining for FKBP51 and nestin was performed using a dual chromogen method (peroxidase with brown chromogen followed by alkaline phosphatase (AP) with red chromogen). Serial sections from paraffin-embedded tumours were cut and deparaffinized as previously described. 8  . For each run, sections from normal (reactive) human lymph node and glioma were respectively used as positive control for FKBP51 and nestin. For negative controls, nonimmune serum (1 : 500) in Tris-buffered saline (TBS) buffer was used instead of the primary antibodies. The immunohistochemical evaluation was performed independently by two consultant pathologists (SS and MM). The cells were judged positive for FKBP51 when a brown signal confined to the nucleus and for nestin when a red cytoplasmatic/membrane staining was observed. In all cases, 10 fields ( Â 40 magnification) were assessed, evaluating a minimum of 100 cells in the selected (representative) areas. Sections were counterstained with Mayer's haematoxylin.