Inhibition of Age-Related Therapy Resistance in Melanoma by Rosiglitazone-Mediated Induction of Klotho

Dermal fibroblasts undergo age-related changes that result in altered secretory profiles. We have recently shown that these age-related changes can drive melanoma progression and therapy resistance. We found that one of the ways in which the aged microenvironment promoted metastasis was via secretion of an inhibitor of canonical Wnt signaling, sFRP2. sFRP2 secreted by aged dermal fibroblasts suppressed β-catenin signaling in melanoma cells. The loss of β-catenin signaling has been associated with an increase in metastatic capacity of melanoma cells, as well as a decrease in sensitivity to vemurafenib. We found this to be true for melanoma cells in an aged microenvironment as well, such that the restoration of β-catenin signaling by anti-sFRP2 antibody in vivo resensitized tumors to vemurafenib and decreased their metastatic ability. In melanoma, the secreted noncanonical Wnt signaling molecule, Wnt5A, also drives the metastatic progression of melanoma (1) and resistance to targeted therapy, in part by inhibiting β-catenin signaling (2). Wnt5A is therefore significantly associated with poorer prognosis in melanoma. We have recently asked what factors might regulate Wnt5A, and we, and others, have identified the antiaging hormone klotho as a regulator of Wnt signaling (3, 4).

Klotho is a circulating serum factor, which is lost during aging. Overexpression of klotho in transgenic mice promotes longevity (5). Conversely, the knockout of klotho results in a complex premature aging phenotype, where mice that are born phenotypically normal develop atherosclerosis, osteoporosis, skin atrophy, and bone degeneration. Klotho has also been shown to decrease Wnt signaling (3). We discovered that Wnt5A is bound to the surface of melanoma cells by heparan sulfate proteogylcans (6), and klotho, which has sialidase activity (7), cleaves Wnt5A from the proteoglycans, such that it cannot be internalized and signaled (4). Once this occurs, Wnt5A mRNA expression also decreases, as Wnt5A increases PKC, which in turn signals to stabilize Wnt5A mRNA. Klotho therefore affects both Wnt5A internalization and mRNA expression, and treatment of Wnt5A-high cells with recombinant klotho significantly inhibits their invasion (4).

In addition to its sialidase activities, klotho plays essential roles in maintaining phosphate homeostasis, inhibiting oxidative stress, suppressing cellular senescence, and modulating insulin receptor signaling (8–11). The loss of klotho is thought to correspond either directly or indirectly to increased insulin resistance, which is associated with diabetes (5). Recently, there has been much interest in using antidiabetic drugs to treat melanoma. Drugs such as metformin and phenformin have been suggested because they affect mitochondrial biogenesis, and changes in mitochondrial biogenesis affect melanoma cell response to vemurafenib and contribute to drug resistance (12, 13).

Another class of drugs used to treat diabetes is the glitazones, which are agonists of the peroxisome proliferator–activated receptor, PPARγ, a nuclear hormone receptor that controls the transcription of multiple genes involved in lipid biogenesis and metabolism (14). Klotho is a transcriptional target of PPARγ, such that PPARγ increases klotho mRNA (15). Wnt5A has recently been shown to inhibit expression of PPARγ via the activation of CAMKII (16). We previously reported that Wnt5A also inhibited klotho, largely at the mRNA level, but at the time did not have a mechanism by which it did so. We hypothesized that Wnt5A may be able to repress klotho expression via CAMKII-mediated suppression of PPARγ. Here, we test and confirm this hypothesis. We further show that using available PPARγ agonists, such as rosiglitazone (trade name Avandia), can increase serum levels of klotho in the blood and in the tumor microenvironment, and decrease Wnt5A. Intriguingly, we also find that rosiglitazone decreases the growth of BRAF-resistant melanomas in aged, but not young, mice.

WM858, WM793, WM164, and 1205LU cells were maintained in MCDB153 (Sigma)/L-15 (Cellgro; 4:1 ratio) supplemented with 2% FBS and 1.6 mmol/L CaCl2 (Tumor growth media). WM983B cells were maintained in DMEM (Invitrogen), supplemented with 5% FBS, 100 units/mL penicillin and streptomycin, and 4 mmol/L l-glutamine. The PLX4720-resistant cell lines were cultured in media same as the parent lines in the presence of PLX4720 (Chemietek). FS5 and FS4 cells were maintained in RPMI (Invitrogen), supplemented with 10% FBS, 100 units/mL penicillin and streptomycin, and 4 mmol/L l-glutamine. Yumm1.7 melanoma cells derived from the BRAFV600E/PTEN^−/−/CDKN2A^−/− mouse model of melanoma and human fibroblasts were maintained in DMEM, supplemented with 10% FBS, 100 units/mL penicillin and streptomycin, and 4 mmol/L l-glutamine. Yumm1.7 cells were cultured to resistance in 10 μM PLX4720. Keratinocytes were maintained in keratinocyte SFM supplemented with human recombinant Epidermal Growth Factor 1-53 (EGF 1-53) and Bovine Pituitary Extract (Invitrogen). All the cell lines were cultured at 37°C in 5% CO₂. Cell stocks were fingerprinted using the AmpFLSTR Identifiler PCR Amplification Kit from Life Technologies TM at The Wistar Institute Genomics Facility. Although it is desirable to compare the profile with the tissue or patient of origin, our cell lines were established over the course of 40 years, long before acquisition of normal control DNA was routinely performed. However, each STR profile is compared with our internal database of over 200 melanoma cell lines, as well as control lines, such as HeLa and 293T. STR profiles are available upon request. Cell culture supernatants were mycoplasma tested using a Lonza MycoAlert assay at the University of Pennsylvania Cell Center Services.

Cells were treated with 100 ng/mL of recombinant Wnt5A (R&D Systems; cat. no. 645WN010CF) for 16 hours or 10 ng/mL of recombinant klotho (R&D Systems; cat. no. 5334-KL-025) for 48 hours. KN93 (Cayman; cat. no. 13319) was used at a final concentration of 10 μM and rosiglitazone (Sigma-Aldrich; cat. no. R2408-50MG) at 10 μM for 48 hours unless otherwise stated. BRAF-resistant subclones were seeded overnight in the absence of plx4720 and then treated as indicated.

Western blotting was done as described in ref. 17. Briefly, total protein lysate of 50 to 65 μg was run on 4% to 12% NuPAGE Bis Tris gel (Invitrogen; cat. no. NW 041222BOX), transferred onto PVDF membrane using iBlot 2, Life technologies, and blocked in 5%milk/TBST. To analyze level of secretory klotho, conditioned media from fibroblasts were run on 4% to 12% NuPAGE Tris gel, and total protein lysate from the fibroblasts was used to probe for HSP90 as loading control. Primary antibodies were used at the following concentrations: biotinylated Wnt5A (500 ng/mL; R&D Systems; cat. no. BAF645), Klotho (1:500; Abcam; cat. no. ab98111), HSP90 (1:4,000; Cell Signaling Technology; cat. no. 4877S), CAMKII (1:500; Cell Signaling Technology; cat. no. 3362S), CAMKII alpha/delta (p Thr286; 1:500; NBP1-51462; Novus), pERK (1:500; Cell Signaling Technology; cat. no. 4370S), and ERK (1:500; Cell Signaling Technology; cat. no. 4696S). All primary antibodies were diluted in 5% milk/TBST and incubated overnight at 4°C. The membranes were washed in TBST and probed with the corresponding horseradish peroxidase (HRP)–conjugated secondary antibody (0.2–0.02 μg/mL of anti-mouse, streptavidin, or anti-rabbit). Proteins were visualized using ECL prime (Amersham) or Luminata Crescendo (Millipore).

Organotypic three-dimensional (3D) skin reconstructs were generated as previously described (18). Briefly, 6.4 × 10⁴ fibroblasts were plated in each insert on top of the acellular layer (BD, cat. no. 355467 and Falcon, cat. No. 353092) and incubated for 45 minutes at 37°C in a 5% CO₂ tissue culture incubator. The tissue culture trays were filled with DMEM containing 10% FBS and incubated for 4 days. Reconstructs were then incubated for 1 hour at 37°C in HBSS containing 1% dialyzed FBS (wash media). Washing media were removed and replaced with reconstruct media. Keratinocytes (4.17 × 10⁵) and melanoma cells (8.3 × 10⁴) were added to the inside of each insert. Media were changed every other day until day 18, reconstructs were then harvested, fixed in 10% formalin, paraffin embedded, sectioned, and stained.

Immunohistochemistry (IHC) was performed on paraffin-embedded sections as described previously (18). Briefly, paraffin-embedded sections were rehydrated through a xylene and alcohol series rinsed in H₂O and washed in PBS. Antigen retrieval was performed by steaming for 20 minutes in target retrieval buffer (Vector Labs) and steamed for 20 minutes. Samples were blocked in a peroxidase blocking buffer (Thermo Scientific) followed by Protein block (Thermo Scientific) and incubated in appropriate primary antibody diluted in antibody diluent (S0809; Dako) at 4°C overnight in a humidified chamber. Samples were then incubated in biotinylated anti-rabbit or polyvalent secondary antibody (Thermo Scientific) followed by streptavidin–HRP solution. Samples were then washed in PBS and incubated in 3-Amino-9-Ethyl-l-Carboazole (AEC) chromogen. Finally, samples were washed in H₂O, incubated in Meyer's hematoxylin for 1 minute, rinsed in cold H₂O, and mounted in Aquamount. Patient samples were collected under Institutional Review Board exemption approval for protocol #EX21205258-1.

Cells were seeded onto glass cover slips and incubated overnight. These cells were then treated with fibroblast-conditioned media, KN93, or rosiglitazone with the indicated dose for 48 hours. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% TritonX-100, blocked with 1% BSA, and stained for primary antibodies as previously described (18). When staining for klotho, cells were not permeabilized. For immunofluorescence staining in the 3D skin reconstructs, the paraffin-embedded sections were rehydrated, processed for antigen retrieval, and blocked as described for IHC. Primary antibodies were used at the following concentrations: Klotho (1:150), PPARγ (1:100; Abcam; cat. No. ab19481), Wnt5A (1:50), HMB45 (1:50; Abcam; cat. No. ab787) and incubated at 4°C overnight. Cells were washed in PBS and incubated with the appropriate secondary antibody (1:2,000; Invitrogen) for 1 hour at room temperature and mounted in Prolong Gold anti-fade reagent containing DAPI (Invitrogen). Images were captured on a Leica TCS SP5 II scanning laser confocal system.

shRNA against human klotho was obtained from Sigma-Aldrich, KL MISSION shRNA Bacterial Glycerol Stock. Klotho shRNA clones used were: TRCN0000158823 and TRCN0000160036. Lentiviral production was performed as described previously (18). Briefly, 293T cells were cotransfected with Klotho shRNA vector and lentiviral packaging plasmids (pCMV-dR8.74psPAX2, pMD2.G). The supernatant containing virus was harvested at 36 and 60 hours, combined and filtered through a 0.45 μm filter, and stored in −80°C till further use. For transduction, the cells were layered overnight with lentivirus containing 8 μg/mL polybrene. The cells were allowed to recover for 24 hours and then selected using 1 μg/mL puromycin.

KL siRNA and negative control siRNA (100 nmol/L; Ambion) were transfected into cells using Lipofectamine 2000 (Thermo Fisher Scientific) as per the manufacturer's protocol.

Melanoma cells treated with indicated conditions were seeded onto a 384-well plate at 3,000 cells/well in triplicate (PerkinElmer cell carrier). For analyzing mitochondrial mass, cells were incubated with MitoTracker Green FM (M7514, ThermoFisher Scientific) at 1 μM concentration for 1 hour. To analyze production of superoxide by mitochondria, cells were incubated with MitoSOX Red reagent (M36008; ThermoFisher Scientific) at a final concentration of 5 μM for 10 minutes, washed extensively, and were imaged using PerkinElmer Operetta. The fluorescent signal was quantified using Harmony 3.0 software. Readings were normalized to cell number which was determined by Hoescht staining (Hoechst 33342; Invitrogen). Cells seeded on coverslips were also stained with MitoTracker Red CM-H2Xros (M7513; Thermofisher Scientific) at 1 μM concentration for 1 hour, fixed with formaldehyde 4%, and mounted in Prolong Gold anti-fade reagent. Images were captured on a Leica TCS SP5 II scanning laser confocal system. Fluorescence was normalized to cell number.

The RNAseq and Clinical dataset for skin cutaneous melanoma (19) was downloaded from The Cancer Genome Atlas (TCGA; http://cancergenome.nih.gov/). Patient ages were grouped into 10-year age ranges. Patient sample information is included in Supplementary Table S1.

RNA was extracted using Trizol (Invitrogen) and RNeasy Mini kit (Qiagen) as previously described (18). cDNA was prepared using the cDNA synthesis kit (Bio-Rad; cat. No. 1708891). Gene expression was quantified using SYBR green method of qPCR, and mRNA levels were compared with standard curves. qPCR was performed on an ABI StepOnePlus sequence detection system using fast conditions, and samples were normalized against the 18S gene, using Universal 18S primers (Invitrogen; cat. no. AM1718). Expression was calculated using the standard curve method according to the manufacturer's protocol (Perkin Elmer). The sequences for primers are as follows: Klotho forward primer: GCTCTCAAAGCCCACATACTG and Klotho reverse primer: GCAGCATAACGATAGAGGCC; Wnt5a forward primer: AGGGCTCCTACGAGAGTGCT and Wnt5a reverse primer: GACACCCCATGGCACTTG.

Cell viability was determined using CellTiter 96 Aqueous One Solution Cell Proliferation Assay according to the manufacturer's protocol. Briefly, cells were seeded in flat-bottom 96-well plates and allowed to adhere overnight. The next day, cells were treated with the indicated drugs for indicated time period. Following treatments, cells were incubated with MTS dye (20 μL/well) for 2 hours. Absorbance was determined at 490 nm using an EL800 microplate reader (BioTek). The percent cell viability was calculated by converting the experimental absorbance to percentage of control and plotted versus drug concentration. The values were then analyzed using a nonlinear dose-response analysis in GraphPad Prism.

All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC; IACUC #112503X_0) and were performed in an Association for the Assessment and Accreditation of Laboratory Animal Care–accredited facility. YUMM1.7 parental or YUMM1.7_BR (2.5 × 10⁵ cells) were injected subcutaneously into young (6-week-old) and aged (52-week-old) C57/BL6 mice (Charles River). Tumor sizes were measured every 3 to 4 days using digital calipers, and tumor volumes were calculated using the following formula: volume = 0.5 × (length × width²). When resulting tumors reached 200 mm³, mice were fed with either AIN-76A chow or AIN-76A chow containing 417 μg/kg PLX4720. Time-to-event (survival) was determined by a 5-fold increase in baseline volume (∼1,000 mm³) and was limited by the development of skin necrosis. For treatment with recombinant mouse Klotho (R&D systems; cat. no. 1819KL) or rosiglitazone, mice were injected intraperitoneally with recombinant Klotho, 0.02 μg/kg every 2 days, or with rosiglitazone, 10 μg/kg every other day. To analyze klotho levels in serum, serum was isolated from blood collected by submandibular bleeding, and Klotho levels were analyzed by Western blotting. At the end of experiment, mice were euthanized, tumors were harvested, and a portion of the tumors were embedded in paraffin and sectioned.

For in vitro studies, a Student t test or Wilcoxon rank-sum test (Mann–Whitney) was performed for two-group comparison. Estimate of variance was performed, and parameters for the t test were adjusted accordingly using Welch's correction. ANOVA or Kruskal–Wallis test with post-hoc Bonferroni's or Holm–Sidak's adjusted P values was used for multiple comparisons. For in vivo studies, repeated measures ANOVA was calculated between samples. The Holm–Sidak correction was performed. For other experiments, Graphpad/Prism6 was used for plotting graphs and statistical analysis. Data was represented as ±SEM. Significance was designated as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Klotho expression is decreased in aged human skin (Fig. 1A). We have recently shown that aged dermal fibroblasts can drive the metastasis and therapeutic resistance of melanoma largely through secreted factors. To investigate if the loss of klotho from fibroblasts contributes to the increased aggression and therapy-resistant phenotype of melanoma cells in the aged microenvironment, we assayed dermal fibroblasts derived from nonmelanoma, healthy donors who were either young (25–35 years old) or aged (55–65 years old) for expression of klotho. All fibroblast experiments were normalized to cell number. As expected, aged dermal fibroblasts demonstrated a decrease in klotho expression as compared with young fibroblasts, as measured by RT-PCR, Western blot analysis, (Fig. 1B and C), and immunofluorescent analysis (Supplementary Fig. S1A). Further, klotho expression was also decreased in the serum of aged mice, as measured by ELISA (Supplementary Fig. S1B). In skin reconstructs built with aged and young fibroblasts, young fibroblasts continued to express klotho, whereas aged fibroblasts did not (Supplementary Fig. S1C). Melanoma cells, surprisingly, recapitulated the age-related klotho expression of fibroblasts: if young fibroblasts were present in the skin reconstruct, melanoma cells in the proximity also expressed klotho. If aged fibroblasts were present, melanoma cells expressed lower levels of klotho (Fig. 1D). TCGA analysis of melanoma samples revealed that KL is indeed decreased in melanoma patients as they age. The difference is most significant under 40 (the age where klotho is lost in the serum) and over 55 (where the incidence of melanoma increases rapidly); however, expression decreases steadily by decade (Supplementary Fig. S1D, patient information provided in Supplementary Table S1).

Because we have previously shown that exogenous klotho affects Wnt5A internalization and expression, we asked whether differences in the amount of klotho secreted from young versus aged fibroblasts affected Wnt5A accordingly in melanoma cells. To confirm that any effects observed were fibroblast-related, we treated melanoma cells with media from young and aged fibroblasts, as well as with unconditioned media with and without 10% FBS. As expected, in the melanoma cells treated with serum-free media, Wnt5A was elevated (Fig. 1E). Consistent with our hypothesis, Wnt5A expression was low in melanoma cells treated with both serum-containing media and media from young fibroblasts. However, when we tested the serum-containing media for klotho, we found that it contained klotho (data not shown). In contrast, aged fibroblast media resulted in an elevated level of Wnt5A expression (Fig. 1E; Supplementary Fig. S1E). Melanoma cells in aged fibroblast-conditioned media also have less klotho mRNA as compared with the cells in young fibroblast-conditioned media (Supplementary Fig. S1F).

We next observed Wnt5A regulation by the aged microenvironment in artificial skin reconstructs made with young or aged fibroblasts, into which we put identical melanoma cell lines. As noted above (Fig. 1D), melanoma cells recapitulate the klotho expression of the fibroblasts, such that melanoma cells in reconstructs made with young fibroblasts express klotho, and melanoma cells in reconstructs made with aged fibroblasts express less klotho. Consistent with our hypothesis that klotho modulates Wnt5A expression, in skin reconstructs made with aged fibroblasts and melanoma cells, Wnt5A was increased in melanoma cells, compared with identical skin reconstructs built with young fibroblasts (Fig. 1F, additional cell line, Supplementary Fig. S1I and S1J). To determine whether this differential Wnt5A expression could be altered by manipulating klotho in young and aged fibroblasts, we knocked down KL using sh_RNA against klotho (sh_klotho) in young fibroblasts (Supplementary Fig. S1G and S1H) and observed an increase in Wnt5A in melanoma cells exposed to sh_klotho fibroblast media. Conversely, treating aged fibroblasts with recombinant klotho (rklotho) decreased Wnt5A in melanoma cells exposed to rklotho-supplemented aged media (Fig. 1G). We confirmed this result in skin reconstructs built using young fibroblasts treated with either a control (scrambled) shRNA or with a klotho shRNA. In the reconstructs built with the klotho knockdown fibroblasts, the levels of klotho in the melanoma cells decrease (Fig. 1H) and the level of Wnt5A increases (Fig. 1I), mimicking the aged fibroblast conditions. Together, these data indicate that the loss of klotho in the aged microenvironment, specifically in dermal fibroblasts, increases Wnt5A in melanoma cells. We have shown extensively that Wnt5A is involved in multiple aspects of tumor progression. Therefore, reconstitution of Wnt5A upon klotho loss during aging may be another mechanism leading to the increased aggression of melanoma in the elderly.

We previously showed that Klotho acts as a sialidase, cleaving Wnt5A away from the surface of melanoma cells. We next wanted to determine the mechanism by which Wnt5A can regulate klotho. Klotho is a target of PPARγ, a regulator of fatty acid storage and glucose metabolism (15). PPARγ is phosphorylated and degraded by the activation (phosphorylation) of CAMKII, which in turn is a direct target of Wnt5A (16). Thus, we hypothesized that Wnt5A regulates klotho expression via CAMKII-mediated degradation of PPARγ (outlined in Fig. 2A). Indeed, treatment of melanoma cells with rWnt5A increased CAMKII phosphorylation (Fig. 2B) and decreased PPARγ and klotho (Fig. 2C). Conversely, when aged fibroblasts were treated with the CAMKII inhibitor KN93, mRNA expression of klotho was increased (Fig. 2D), as was protein expression of PPARγ (Fig. 2E). Even at the highest doses, KN93 was not toxic to fibroblasts (Supplementary Fig. S2A and S2B). To determine whether KN93 had similar effects on melanoma cells, we treated melanoma cells with KN93 (10 μmol/L, 48 hours) and demonstrated that PPARγ and klotho were restored (Fig. 2F and G). Consistently, exploration of PPARγ expression levels in the TCGA database revealed that, like klotho, PPARG levels were decreased in older melanoma patients (Supplementary Fig. S2C; aged vs. young at different decades of life).

As our data above suggested that activation of PPARγ should increase klotho and regulate Wnt5A, we treated aged fibroblasts with the PPARγ agonist rosiglitazone, which increased PPARγ and klotho (Fig. 2H and I). Melanoma cells treated with rosiglitazone demonstrated a decrease in the secreted form of Wnt5A in the lysate (Supplementary Fig. S2D, upper band). As in fibroblasts, we observed increases in both PPARγ and klotho in melanoma cells treated with rosiglitazone (Supplementary Fig. S2E). Like KN93, rosiglitazone was not toxic to either fibroblasts or melanoma cells at high doses (Supplementary Fig. S2F and S2G). To translate these findings in vivo, we used Yumm1.7 melanoma cells derived from the BRAF^V600E/PTEN^−/−/CDKN2A^−/− mouse model of melanoma and injected them into C57BL/6 mice. When tumors were palpable, we treated one set of aged mice with 10 μg/kg of rosiglitazone, second set with 0.02 μg/kg of recombinant klotho, and a third set with PBS/DMSO vehicle control. Tumor burden was significantly reduced as compared with control by treatment with either rosiglitazone or recombinant klotho (Fig. 2J). Treatment of aged mice with rosiglitazone or recombinant klotho decreased Wnt5A and increased klotho levels in the tumor (Fig. 2K). Together, these data suggest that using rosiglitazone as an agent to increase PPARγ and klotho can decrease Wnt5A and inhibit tumor growth in an aged microenvironment.

One of the first line therapies for melanoma is the drug vemurafenib, which targets a mutant BRAF protein (BRAF^V600E). Our previous work demonstrated that Wnt5A can promote resistance to targeted therapy in melanoma (2, 17), and we support this by showing that knockdown of Wnt5A can increase the sensitivity of melanoma cells to PLX4720, the tool compound for Vemurafenib (Supplementary Fig. S3A and S3B). Further, Wnt5A is more highly secreted by aged as compared with young fibroblasts (Supplementary Fig. S3C), which is in keeping with our previous data showing that melanomas in an aged microenvironment are more resistant to vemurafenib, largely due to changes in Wnt signaling (18). Importantly, it has been shown that Wnt5A can signal to activate ERK, and CAMKII, also a target of Wnt5A, activates the Raf signaling pathway further. This represents a way to maintain activation of the MAPK pathway, regardless of mutational status. To determine if manipulating this pathway in melanoma cells could affect resistance of melanoma cells to BRAF inhibition, we created a resistant subclone of Yumm1.7 cells that we could use in vivo. These subclones are as resistant in vitro to PLX4720 (the tool compound for vemurafenib) as are WM983B BRAF inhibitor–resistant subclones (Supplementary Fig. S4A). Next, we analyzed the Yumm1.7, WM983B, WM793, and their resistant subclones for Wnt5A and PO₄-ERK. Each of these cell lines bears a different set of mutations (Supplementary Table S2). In each case, both Wnt5A and PO₄-Erk were elevated (Fig. 3A). This was also concomitant with the loss of KL and PPARG expression (Supplementary Fig. S4B–S4F). We then asked whether increasing klotho via rosiglitazone reduced PO₄-Erk signaling, via downregulation of Wnt5A. PO₄-Erk is decreased in BRAF-resistant cell lines after treatment with rosiglitazone (Fig. 3B), and this is concomitant with the decrease in Wnt5A (Supplementary Fig. S4G). KN93 is similarly able to cause decreases in both PO₄-Erk and Wnt5A (Supplementary Fig. S4H). We then asked whether transcriptionally silencing klotho would prevent rosiglitazone from decreasing ERK activity and decreasing Wnt5A. Knocking down klotho in young fibroblasts increased PO₄-Erk and Wnt5A (Supplementary Fig. S4I). In melanoma cells treated with rosiglitazone, there is a reduction of both Wnt5A and PO₄-ERK. However, first knocking down klotho results in activated Erk, and sustained Wnt5A signaling, even in the presence of rosiglitazone (Fig. 3C; Supplementary Fig. S4J). These data suggest that rosiglitazone is mediating its effects on the Erk pathway via klotho-mediated suppression of Wnt5A.

We have previously shown that tumors in an aged microenvironment are more resistant to targeted therapy. Therefore, we examined whether increasing klotho in the microenvironment could affect response to PLX4720. We took aged fibroblasts and treated them with rosiglitazone. We then exposed BRAF-resistant melanoma cells to the media from the treated fibroblasts, in the presence or absence of PLX4720. Western analysis of PO₄-ERK (Fig. 3D) demonstrates that rosiglitazone treatment of the fibroblasts also reduces PO₄-ERK levels in melanoma cells. We have shown that mitochondrial ROS marks response to PLX4720, such that increases in mito-Sox correlate to increased resistance (Supplementary Fig. S5A and S5B). As with PO₄-ERK, mito-sox activity is lowered by rosiglitazone in melanoma cells incubated with aged fibroblasts, both in the presence and absence of PLX4720 (Fig. 3E; Supplementary Fig. S5C–S5E).

Taken together, these data suggested that targeting klotho in both the melanoma cells and the microenvironment in aged mice should have effects on BRAFi-resistant melanomas. To test this hypothesis in vivo, we treated Yumm1.7BR-resistant cells with rosiglitazone or klotho, in the presence and absence of PLX4720, and demonstrated that these treatments decreased Wnt5A (Supplementary Fig. S5F). Yumm1.7BR cells were injected subcutaneously into aged mice. Treatment with PLX4720 alone did not affect tumor growth compared with controls, whereas treatment with rosiglitazone alone, and the combination of rosiglitazone and PLX4720, impeded the rate of tumor growth (Fig. 3F). The combination was not more effective than rosiglitazone alone. To determine if these results differed in young mice, we repeated the identical experiments in 8-week-old mice. In young mice, Yumm1.7BR cells re-acquired sensitivity to PLX4720, in keeping with previous data from our laboratory that indicated that melanoma cells are more sensitive to BRAFi in a young microenvironment (18). Intriguingly, treating young mice with rosiglitazone alone accelerated tumor growth; however, rosiglitazone + PLX4720 decreased tumor burden compared with PLX4720 alone (Fig. 3G). To assay whether klotho was differentially increased in Yumm1.7BR cells in rosiglitazone-treated aged versus young mice, we examined the serum of mice for klotho expression and stained the tumors for klotho. In aged mice, klotho expression is low in Yumm 1.7BR tumors, and in serum, and klotho increases in aged mice upon treatment with rosiglitazone (Fig. 3H, left). In young mice however, klotho is high at baseline in both the tumor and mouse serum, and elevated even further by rosiglitazone treatment (Fig. 3H, right). These data indicate that rosiglitazone may be a unique therapy for older patients, and while it cannot re-sensitize melanoma cells to vemurafenib, it can still reduce tumor burden, both in BRAF inhibitor–sensitive and –resistant cells in aged animals. Rosiglitazone may be a viable alternate strategy to reduce tumor burden in older patients who have developed resistance to BRAF inhibitors. Our data suggest also that because PO₄-Erk is reduced by direct treatment of melanoma cells with rosiglitazone, the right dosing schedule of rosiglitazone and PLX4720 may provide additional benefit in combination with BRAF inhibitors in young patients as well, which will be the subject of further study in our laboratory.

The role of klotho in aging has been well documented. The loss of klotho has also been implicated in the progression of multiple tumor types, including lung (20), cervical (21), and gastric cancer (22). In addition, we have shown that treating melanoma cells with recombinant klotho can reduce their invasion by inhibiting Wnt5A signaling, which is elevated in metastatic melanoma cells. We have also shown that, in addition to promoting metastasis, Wnt5A can increase resistance to vemurafenib. Maintaining klotho in aged patients then could have the twin benefits of decreasing metastasis and inhibiting therapy resistance. Because klotho is lost during aging, and we have recently shown that aging promotes melanoma metastasis and therapy resistance (18), here we asked whether klotho loss in the aged microenvironment, specifically in fibroblasts, promotes Wnt5A expression and affects therapy resistance. In this study, we confirm that klotho is decreased in the aged skin microenvironment, and manipulation of klotho in aged and young dermal fibroblasts affects Wnt5A expression in melanoma cells. We also show, for the first time, that Wnt5A can regulate klotho and that it does so by CAMKII-mediated suppression of PPARγ. When Wnt5A is absent, PPARγ activation increases transcription of klotho.

Because some antidiabetic drugs (e.g., metformin) have activity against melanoma (23), and because some of these medications are PPARγ agonists, there has been much interest in the use of PPARγ agonists to treat cancer. However, the studies performed so far provide conflicting data as to whether PPARγ agonists are beneficial. In colon cancer, studies indicate that malignancy can be reversed by rosiglitazone, pioglitazone, and other thiazolidinediones (TZD; refs. 24–26), but phase II clinical trials do not support this (27). Similarly, in breast cancer, late-stage tumors are not affected by TZD. However, a more recent study has shown that if the glitazone is administered earlier, to less aggressive tumors, i.e., HER2⁺ breast cancer, then it does have protective effects (28). In the current study, we identify yet another parameter that may affect response to glitazones, and that is age. We show that rosiglitazone can increase levels of klotho in the serum of aged mice bearing melanoma, and that this translates to an inhibition of tumor growth similar to treating mice directly with klotho. Treating BRAF inhibitor–resistant cells with rosiglitazone reduces ROS in the mitochondria and kills tumor cells. Although we cannot claim that rosiglitazone re-sensitizes cells to vemurafenib, because at the doses used there was no added benefit to rosiglitazone + PLX4720 versus rosiglitazone alone, it is quite clear that rosiglitazone may prove to be effective against BRAFi-resistant melanoma, when other options have failed. Similarly, increasing klotho with rosiglitazone inhibits the growth of both BRAFi-sensitive and BRAFi-resistant melanoma equally effectively in aged animals.

Increasing rosiglitazone increased klotho levels, decreased Wnt5A, and decreased tumor burden in aged mice. However, this was not the case in young mice, and we were fascinated to find that treatment with rosiglitazone differed in young versus aged mice. First, we have recently shown that changes in the aged microenvironment (e.g., increases in ROS and sFRP2) can drive increased resistance to targeted therapy (18). In that study, placing PLX4720-sensitive tumors in aged mice increased their ability to resist therapy. Here, we show that the opposite is also true, i.e., that placing resistant tumors in young mice, slightly, but significantly, increases their sensitivity to PLX4720. Further, the combination of PLX4720 and rosiglitazone is effective in young mice, and unlike in aged mice, rosiglitazone appears to sensitize tumors to PLX4720. We have shown in this study that treatment of melanoma with rosiglitazone lowers PO₄-Erk levels, which may be enough in the young microenvironment to further sensitize melanomas to PLX4720. However, rosiglitazone alone is ineffective in treating resistant tumors in young animals, and actually increases tumor burden in young mice. The mechanism for this is not entirely clear, but may be due to a klotho threshold present in cells, which determines their susceptibility to PLX4720 and other MAPK-targeting agents. In this case, endogenous levels of klotho in young mice may partially re-sensitize resistant tumors to targeted therapy, and increasing klotho with rosiglitazone in aged animals where klotho is endogenously low can decrease the growth of resistant tumors. However, raising klotho further in young animals does not serve to reduce tumor burden, presumably because klotho levels in young animals already surpass the threshold. In addition, increasing klotho too much may increase proliferation of the tumor, as klotho is known to increase cellular proliferation (29). If this is the case, as suggested by the accelerated tumor growth in the rosiglitazone-treated young mice, the age-specific nature of this therapy might serve to explain some of the observed discrepancies in studies analyzing the role of PPARγ agonists in the therapy of breast and colon cancer. Age may play an important role in determining which patients should receive rosiglitazone as therapy for melanoma, and because klotho is secreted in the serum, it should be relatively simple to determine which patients might benefit from rosiglitazone therapy and which would not. Understanding the role that the aged microenvironment plays in modulating cell fates like metastasis and therapy resistance is critical for guiding therapy in older patients.

A.E. Aplin reports receiving commercial research grants from Pfizer Inc. A. Roesch reports receiving commercial research grants from Novartis. A.T. Weeraratna is a consultant/advisory board member for Phoremost Technologies. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Behera, A.T. Weeraratna

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Behera, A. Kaur, M.R. Webster, S. Kim, A. Ndoye, G.M. Alicea, J. Wang, K. Ghosh, P. Cheng, S. Lisanti, K. Marchbank, V. Dang, M. Levesque, R. Dummer, M. Herlyn, A.E. Aplin, C. Caino, A.T. Weeraratna

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Behera, A. Kaur, M.R. Webster, C.H. Kugel III, J. Wang, K. Ghosh, P. Cheng, X. Xu, A.T. Weeraratna

Writing, review, and/or revision of the manuscript: R. Behera, M.R. Webster, C.H. Kugel III, R. Dummer, M. Herlyn, A.E. Aplin, A. Roesch, D.C. Altieri, A.T. Weeraratna

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.H. Kugel III, P. Cheng, M. Levesque, R. Dummer, A.T. Weeraratna

Study supervision: A.T. Weeraratna

We thank Dr. Gideon Bollag of Plexxikon for PLX4720. We thank Dmitry Gourevitch of the Wistar Histology Core and Fred Keeney of the Wistar Imaging facility.

This work was supported in part by RO1 CA174746-01 (A.T. Weeraratna and R. Behera), P01 CA 114046-06 (A.T. Weeraratna and M. Herlyn), T32 CA 9171-36 (M.R. Webster and C.H. Kugel III), an MRF Established Investigator award (A.T. Weeraratna), an ACS-IRG award (A.T. Weeraratna), a Miriam and Sheldon Adelson Research Foundation Award (A.E. Aplin and M. Herlyn), P50 CA174523-01 (A.T. Weeraratna, X. Xu, and M. Herlyn), and R01-CA1826635 (A.E. Aplin). Core facilities at The Wistar Institute are supported by the CCSG grant P30 CA010815.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.