Research ArticleCancer

The E3 ligase APC/CCdh1 promotes ubiquitylation-mediated proteolysis of PAX3 to suppress melanocyte proliferation and melanoma growth

See allHide authors and affiliations

Sci. Signal.  01 Sep 2015:
Vol. 8, Issue 392, pp. ra87
DOI: 10.1126/scisignal.aab1995

Sensitizing melanoma to chemotherapeutic drugs

APC/C is an E3 ubiquitin ligase complex that coordinates aspects of the cell cycle by targeting cell cycle regulators, such as cyclins, for destruction. Cao et al. found that Cdh1, a component of this complex, restricted cell proliferation in melanocytes by promoting the degradation of the transcription factor PAX3. However, Cdh1 abundance was decreased in primary and advanced melanoma patient samples compared to normal skin tissue, and the abundance of PAX3 was increased. Restoring Cdh1 abundance in melanoma cells in culture and in xenografts in mice suppressed their proliferation and increased their sensitivity to the chemotherapeutic agent doxorubicin. The findings indicate that inhibition of Cdh1 could enable traditional chemotherapeutic drugs to be effective in melanoma.


The anaphase-promoting complex or cyclosome with the subunit Cdh1 (APC/CCdh1) is an E3 ubiquitin ligase involved in the control of the cell cycle. Here, we identified sporadic mutations occurring in the genes encoding APC components, including Cdh1, in human melanoma samples and found that loss of APC/CCdh1 may promote melanoma development and progression, but not by affecting cell cycle regulatory targets of APC/C. Most of the mutations we found in CDH1 were those associated with ultraviolet light (UV)–induced melanomagenesis. Compared with normal human skin tissue and human or mouse melanocytes, the abundance of Cdh1 was decreased and that of the transcription factor PAX3 was increased in human melanoma tissue and human or mouse melanoma cell lines, respectively; Cdh1 abundance was further decreased with advanced stages of human melanoma. PAX3 was a substrate of APC/CCdh1 in melanocytes, and APC/CCdh1-mediated ubiquitylation marked PAX3 for proteolytic degradation in a manner dependent on the D-box motif in PAX3. Either mutating the D-box in PAX3 or knocking down Cdh1 prevented the ubiquitylation and degradation of PAX3 and increased proliferation and melanin production in melanocytes. Knocking down Cdh1 in melanoma cells in culture or before implantation in mice promoted doxorubicin resistance, whereas reexpressing wild-type Cdh1, but not E3 ligase–deficient Cdh1 or a mutant that could not interact with PAX3, restored doxorubicin sensitivity in melanoma cells both in culture and in xenografts. Thus, our findings suggest a tumor suppressor role for APC/CCdh1 in melanocytes and that targeting PAX3 may be a strategy for treating melanoma.


The anaphase-promoting complex or cyclosome (APC/C, also called APC) is a multisubunit E3 ubiquitin ligase complex (1). Genetic deletion studies show that the APC/C E3 ligase complex has a pivotal role in regulating multiple cell cycle transitions and DNA replication processes (24). The enzymatic activity of APC/C is strictly dependent on its association with coactivators. The best studied APC/C activators are Cdc20 and Cdh1, which directly bind the APC/C core complex to activate its E3 ubiquitin ligase activity and contribute to its substrate recognition and specificity of APC/C (1, 3). Cdc20 activity is largely restricted to mitotic regulation (5), whereas APC/CCdh1 is important to both G1 cell cycle regulation and genomic stability. Mouse modeling studies show that Cdh1 functions as a tumor suppressor (6). Specifically, heterozygous APC/CCdh1 knockout mice are more prone to developing spontaneous epithelial tumors, plasmacytosis, and myelodysplastic syndrome (7). Reduced CDH1 expression is also seen in human lymphoma samples and cell lines (8). Given that APC/CCdh1 controls proliferation, differentiation, and maintenance of genomic integrity both in tissue culture and in vivo in mouse modeling systems (6), it is plausible that Cdh1 might function as a tumor suppressor in melanocytes and melanoma, in part by regulating the proliferation and development of melanocytes.

Melanocyte proliferation and differentiation are regulated in a highly coordinated manner during development, and loss of coordinated proliferation and differentiation contributes to melanocyte malignant transformation. In this regard, many melanocyte lineage–specific transcription factors, including PAX3, SOX10 (SRY-related HMG-box transcription factor 10), and MITF (microphthalmia-associated transcription factor), play crucial roles in the commitment, proliferation, and survival of melanocytes (911). PAX3 overexpression and MITF oncogenic amplification and mutation have been identified in human melanomas (1215).

PAX3 is a member of the paired-box (PAX) family of transcription factors (14, 16). Mutations in PAX3 can produce type I and type III Waardenburg syndrome (17, 18), a pathological condition characterized by melanocyte deficiencies in the skin and inner ear. Mice with a naturally occurring PAX3 loss-of-function mutation (Splotch, Sp/+) displayed severe deficiencies in neural crest–derived melanocyte development (19), leading to pigmentation abnormalities characterized by a white belly patch. Conversely, mice overexpressing PAX3 in melanocytes had hyperpigmented skin, most evident in the ear pinnae and tail, in part due to elevated melanocyte proliferation (20). Collectively, these studies indicate an important role for PAX3 in the proliferation and survival of developing melanocytes, although its role in melanomagenesis requires further investigation.

As a transcription factor, PAX3 directly promotes MITF transcription by binding a proximal region of the MITF promoter in a SOX10-dependent manner (21) and synergizes with CREB (adenosine 3′,5′-monophosphate response element–binding protein) to activate the transcription of MITF (12). Critically, both PAX3 and MITF bind the promoters of various melanocyte-specific genes, including those encoding tyrosinase-related protein-1 (Tyrp-1) and dopachrome tautomerase (DCT), to up-regulate their transcription (22, 23). Furthermore, both T-box 2 (TBX2) (24), an anti-senescence transcription factor, and BRN2 (POU3F2), a proinvasive regulator of MITF expression, are direct targets of PAX3 in melanocytes and melanoma (25). The interaction between MITF and PAX3 modulates the stem cell state in the melanocyte lineage (23), but the clinical relevance of this interaction in melanomagenesis requires further investigation. In contrast to the well-characterized downstream targets of PAX3 in the melanocyte lineage, upstream regulation of PAX3 is not yet adequately defined. We previously showed that the transforming growth factor–β (TGF-β)/SMAD signaling pathway, which is inhibited by ultraviolet (UV) irradiation, represses PAX3 transcription in melanocytes (12), whereas the fibroblast growth factor 2 (FGF2)–signal transducer and activator of transcription 3(STAT3) signaling axis activates PAX3 transcription (20). Here, we investigated the role of APC/CCdh1 in melanocytes and how PAX3 protein stability is physiologically regulated in melanocytes and, in turn, is aberrantly increased in pathological settings such as melanoma.


Mutations in APC/CCdh1 are identified in melanomas

Multiple groups have demonstrated that APC/CCdh1 functions as a tumor suppressor, playing a crucial role in the control of proliferation, differentiation, and maintenance of the genomic integrity (3, 26, 27). However, it remains largely unclear whether APC/CCdh1 contributes to melanoma development and progression. We therefore analyzed the genetic status of APC/C components in melanomas collected in The Cancer Genome Atlas (TCGA), National Institutes of Health (NIH), and the original exome sequencing data from the Broad Institute (28). Specifically, 9.9% (34 of 344) and 12.6% (17 of 135) of the melanoma samples reported by TCGA or Broad, respectively, have an APC/C component mutation (table S1). Moreover, 2.6% (13 of 344) and 1.5% (8 of 135) of melanoma sequencing results from the TCGA or Broad database, respectively, detected CDH1 mutation(s), and 3.8% (13 of 344) and 5.59% (2 of 135), respectively, detected CDC27 mutations. Notably, we found that 35.3% (18 of 51) of mutations showed loss of heterozygosity (LOH) silent mutations (table S1). Interestingly, 81.8% (9 of 11) of CDH1 mutations and 19% (4 of 21) of CDC27 mutations were C-T or CC-TT—mutations that are closely related to UV-induced DNA-damaging events (table S1). These data indicate that some APC/CCdh1 mutations might associate with UV irradiation in melanomagenesis.

Loss of APC/CCdh1 occurs with development and progression of melanoma and increased PAX3 abundance

To address a possible role of Cdh1 in melanocyte melanomagenesis, we tested the expression of Cdh1 in melanocyte transformation and melanoma progression and whether overexpression of Cdh1 affects the viability of melanocytes. To this end, Cdh1 expression was analyzed by immunohistochemical staining in normal human skin and melanoma tissue array slices. Notably, in normal skin, the Cdh1-specific antibody showed strong nuclear staining in all epidermal cells, including individual melanocytes, but the abundance of Cdh1 was decreased in nevi and melanoma specimens (Fig. 1A). Semiquantitative analysis (assessed by H score) indicated that the abundance of Cdh1 was less in primary melanoma compared to normal skin and was inversely correlated with pathological progression (Fig. 1B). Because we previously demonstrated that the PAX3 protein is overexpressed in melanoma samples and correlates with the clinical characteristics of melanoma patients (12), we further determined whether there was an intrinsic connection between increased PAX3 and decreased Cdh1 in melanomas. Indeed, there was an inverse correlation between Cdh1 and PAX3 immunohistochemical staining in melanoma samples (Fig. 1C and fig. S1B). Double immunohistochemical staining showed that Cdh1 staining occurred in cells that were also positive for the melanocyte-specific marker, MART1 (29), indicating that Cdh1 was expressed in melanocytes (fig. S1A). These results together indicate that loss of the Cdh1 tumor suppressor might cause the aberrant up-regulation of PAX3 in melanoma.

Fig. 1 Correlation between Cdh1 and PAX3 abundance in melanoma.

(A and B) Immunohistochemistry (A) and quantification (B) of Cdh1 staining (brown) in human skin and melanoma tissue spanning the benign to metastatic pathological stages. (C) Immunohistochemical analysis of the correlation between Cdh1 and PAX3 staining in patient primary melanoma tissue.

APC/CCdh1 promotes PAX3 ubiquitylation and subsequent degradation

Given that APC/CCdh1 mainly functions as an E3 ubiquitin ligase and there is an inverse correlation between APC/CCdh1 and PAX3 in melanomas, we started our investigation to examine whether PAX3 is a putative APC/CCdh1 substrate. In support of this notion, we identified several evolutionarily conserved D-boxes in the PAX3 primary protein sequence (fig. S2A). Moreover, depletion of endogenous Cdh1 in mouse or human melanocytes using multiple independent short hairpin RNAs (shRNAs) markedly increased the abundance of PAX3 protein (Fig. 2, A and B), but not that of PAX3 mRNA (fig. S2, B to E). These findings indicate that APC/CCdh1 may promote PAX3 proteolysis in melanocytes. Indeed, we found that the ectopic expression of wild-type Cdh1, but not an E3 ligase–deficient (ΔC-box mutant) Cdh1 (30), markedly increased ubiquitylation of Flag-tagged PAX3 in transfected melanocytes (Fig. 2, C and D) and human embryonic kidney (HEK) 293T cells (fig. S2F) and subsequently decreased PAX3 protein abundance (Fig. 2, E and F), but not PAX3 mRNA expression (fig. S2, G and H), in melanocytes.

Fig. 2 APCCdh1 negatively regulates PAX3 abundance in melanocytes.

(A and B) Immunoblot (IB) analysis of whole-cell lysates (WCL) derived from mouse melan-a cells (A) or human primary melanocytes (HPM) (B) infected with species-specific Cdh1 shRNAs. (C and D) Immunoprecipitation (IP) for hemagglutinin (HA) and immunoblot for Flag in human primary melanocytes (C) or melan-a (D) cells transfected with HA-ubiquitin, the indicated Flag-PAX3, and wild-type (WT) or mutant Myc-Cdh1 constructs and treated with MG132 (25 μM, 6 hours). (E and F) Immunoblot of whole-cell lysates derived from melan-a (E) or human primary melanocytes (F) transfected with PAX3 and WT or mutant Cdh1. (G and H) As in (A) and (B) in the presence of MG132 (10 μM, 10 hours) where indicated. (I to N) Immunoblot analysis of melan-a or human primary melanocytes infected with shRNAs against Cdc27 (I and J), APC10 (K and L), or Cdc20 (M and N). (O and P) Immunoblot analysis in human primary melanocytes (O) or melan-a cells (P) infected with Cdh1 shRNA or a negative control (Scr) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing cycloheximide (CHX). Blots are representative of three experiments.

Consistent with a critical role for Cdh1 in governing PAX3 stability, we further observed that the Cdh1 knockdown–induced increase in PAX3 abundance was largely abolished after the 26S proteasome was inhibited by the addition of MG132 (N-carbobenzyloxy-l-leucyl-l-leucyl-l-leucinal) to the culture medium (Fig. 2, G and H). These results further confirm that the PAX3 protein stability is negatively regulated by APC/CCdh1 in melanocytes. It is also worth mentioning that Cdh1 depletion–induced PAX3 up-regulation also stimulated a subsequent increase in the abundance of the melanocyte transcription factor MITF (Fig. 2, A and B), further advocating a critical biological role of APC/CCdh1 in melanocyte expansion, in part by governing the PAX3 protein stability. Consistent with the notion that APC/CCdh1 suppresses the PAX3/MITF signaling axis, we found that PAX3 and MITF protein abundance was markedly increased by the depletion of other APCCdh1 E3 ligase complex components, including Cdc27 (Fig. 2, I and J) and APC10 (Fig. 2, K and L), but not Cdc20, a close homolog of Cdh1 (Fig. 2, M and N).

To further confirm the potential role of APC/CCdh1 in controlling PAX3 stability, we monitored the PAX3 protein abundance in mouse melan-a cells and human primary melanocytes before and after depletion of endogenous Cdh1 in the presence of cycloheximide to stop new protein synthesis. Notably, we found that the half-life of PAX3 protein was significantly longer after the depletion of Cdh1 in melanocytes (Fig. 2, O and P).

APC/CCdh1 interacts with PAX3

Next, we explored whether APC/CCdh1 physically interacts with PAX3 to regulate its stability. Using immunofluorescence, we found that Cdh1 and PAX3 colocalized in both mouse and human melanocytes (fig. S3, A and B) and melanoma cell lines (fig. S3, C and D). A direct or indirect interaction between endogenous (Fig. 3, A and B) and ectopic proteins (Fig. 3C) was detected in immunoprecipitates derived from melan-a and human primary melanocyte whole-cell extracts. Ectopic, Flag-tagged PAX3 interacted with (Fig. 3C) and was degraded by (Fig. 3, D and E) Cdh1, but not by the close homolog of Cdh1, Cdc20. In vitro glutathione S-transferase (GST) pull-down assays further revealed a direct physical interaction between recombinant, tagged PAX3 and Cdh1 (Fig. 3F).

Fig. 3 Cdh1, but not Cdc20, specifically interacts with PAX3 to promote its degradation in a D-box–dependent manner.

(A and B) Immunoblotting for endogenous Cdh1 and PAX3 in whole-cell lysates after immunoprecipitation in melan-a (A) or human primary melanocytes (B) pretreated with MG132 (10 μM, 10 hours). Immunoglobulin G (IgG) served as the immunoprecipitation control. (C) Immunoprecipitation and immunoblotting in whole-cell lysates (WCL) from 293T cells transfected with Flag-PAX3 and HA-Cdh1 or HA-Cdc20, pretreated with MG132 (10 μM, 10 hours). (D and E) Immunoblotting of whole-cell lysates from melan-a (D) or human primary melanocytes (E) transfected with PAX3 and the indicated HA-Cdc20 or HA-Cdh1 constructs. (F) GST pull-down analysis to assess GST-Cdh1 interaction with Flag-PAX3 in transfected 293T cells. (G and H) Immunoblotting for Flag (PAX3), HA (Cdh1), and cyclin B in melan-a cells transfected with Flag-tagged WT (G) or 3D-box mutant (H) PAX3 with increasing abundance of ectopic HA-tagged Cdh1. (I) Immunoprecipitation and immunoblotting in whole-cell lysates from 293T cells transfected with HA-Cdh1 and the indicated PAX3 construct, pretreated with MG132 (10 μM, 10 hours). (J) Immunoprecipitation and immunoblotting analysis for endogenous Cdh1 in whole-cell lysates from B16 cells transfected with the indicated Flag-PAX3 construct. (K) GST pull-down analysis to assess GST-Cdh1 interaction with WT or D-box mutant His-PAX3 in transfected 293T cells. Blots are representative of three experiments.

Together, the results thus far suggest that APC/CCdh1 promotes ubiquitylation and subsequent proteolysis of the PAX3 protein through direct interactions. Consistent with this hypothesis, we found that PAX3 protein abundance was significantly reduced in a Cdh1 dose-dependent manner (Fig. 3G). Furthermore, consistent with previous reports indicating a critical role of the D-box within the substrates of Cdh1 to mediate Cdh1-dependent proteolysis (3033), we found that deletion of the three identified D-boxes (fig. S5A) in PAX3 prevented Cdh1-mediated degradation of PAX3 (Fig. 3H), in part by abolishing its interaction with Cdh1 (Fig. 3, I and K). These results further suggest that the D-boxes of PAX3 are necessary for mediating the interaction with full-length Cdh1. Furthermore, using coimmunoprecipitation assays, we found that PAX3 interacted with the C-terminal WD40 repeat domain of Cdh1 (fig. S4, A and B), which is the region where most of the well-characterized Cdh1 substrates interact. These results further indicate that PAX3 is a putative ubiquitin substrate of APC/CCdh1.

APC/CCdh1 promotes the ubiquitylation of PAX3 in a D-box–dependent manner

In keeping with the notion that APC/CCdh1 interacts with PAX3 in a D-box–dependent manner, we found that ectopic expression of Cdh1 in B16 melanoma cells promoted the degradation of transfected full-length PAX3, but not PAX3 mutants that are deficient in associating with Cdh1 (Fig. 4A). Likewise, wild-type Cdh1, but not Cdh1 mutants that are deficient in associating with PAX3, promoted PAX3 degradation in cells (Fig. 4B and fig. S4, A to C).

Fig. 4 APC/CCdh1 promotes the ubiquitylation of PAX3 in a D-box–dependent manner.

(A) Immunoblot analysis of whole-cell lysates from 293T cells transfected with HA-Cdh1 and the indicated full-length (FL) or mutant Flag-PAX3 construct. (B) Immunoblot of whole-cell lysates from 293T cells transfected with Flag-PAX3 together with the indicated HA-Cdh1 constructs. EV, empty vector. WT, amino acids 1 to 496; N/F1, amino acids 1 to 395; N/F2, amino acids 1 to 305; N, amino acids 1 to 155. (C and D) Pull-down (IP) for Flag-PAX3 and immunoblotting for HA-ubiquitin in transfected mouse melan-a (C) or human primary melanocytes (D) infected with control or Cdh1 shRNA and pretreated with MG132 (25 μM, 6 hours). (E) As in (C) in control or Cdh1-depleted melan-a cells transfected with HA-ubiquitin and full-length or D-box mutant Flag-PAX3, pretreated with MG132. (F) In vitro ubiquitination assay using WT or 3D-box mutant His-PAX3. (G) Immunoprecipitation with Ni-NTA–conjugated agarose beads followed by immunoblotting as indicated in 293T cells transfected with His-ubiquitin, Flag-PAX3, and WT or mutant HA-Cdh1 (or an empty vector), and pretreated with MG132. Blots are representative of three experiments.

Given that APC/CCdh1 is a multisubunit E3 ubiquitin ligase complex (1), and to further pinpoint the critical role of APC/CCdh1 in promoting PAX3 ubiquitylation, we found that depleting Cdh1 reduced the abundance of ubiquitylated species of PAX3 in both mouse melan-a cells and human primary melanocytes (Fig. 4, C and D). Furthermore, unlike wild-type PAX3, depletion of Cdh1 did not affect the ubiquitylation of a PAX3 mutant in which the region containing the three D-boxes was deleted (Fig. 4E and fig. S5A). Furthermore, we found that the purified APC/CCdh1 holoenzyme promoted the in vitro polyubiquitylation of recombinant His-tagged wild-type, but not D-box mutant, PAX3 (Fig. 4F), indicating that Cdh1 promotes PAX3 ubiquitylation in cells largely in a D-box–dependent manner. Further relating our findings to the melanoma disease setting, we found that several melanoma-derived Cdh1 mutants failed to bind PAX3 compared to wild-type Cdh1 (fig. S5, B and C), which led to reduced ubiquitylation of PAX3 (Fig. 4G). Thus far, together, the results support the notion that loss of Cdh1 abundance or function (through mutation) in melanoma enables stabilization of the PAX3 protein, which may facilitate melanomagenesis.

APC/CCdh1 inhibits melanocyte proliferation and melanin production by suppressing PAX3

We observed that ectopic expression of Cdh1 significantly inhibited the proliferation of melanocytes (Fig. 5A) and induced a G1/G0 cell cycle phase arrest (Fig. 5B). Alternatively, in human primary melanocytes depleted of endogenous PAX3, overexpressing Flag-tagged PAX3 promoted cell proliferation (Fig. 5C) and the progression of cells into S phase (Fig. 5D). However, ectopic overexpression of Cdh1 in these cells inhibited PAX3-induced cell proliferation and cell cycle phase shifts (Fig. 5, C and D). Together, these results suggest that Cdh1 may antagonize the cellular function of PAX3 protein that promotes melanocyte proliferation.

Fig. 5 Ectopic overexpression of Cdh1 inhibits melanocyte proliferation.

(A and B) Relative cell number (A) and cell cycle profiles (B) in human primary melanocytes infected with HA-tagged Cdh1 or a control viral vector. *P < 0.01 by t test. (C and D) As in (A) and (B) in cells overexpressing both HA-Cdh1 and HA-PAX3. *P < 0.01 by t test. (E) Immunoblotting in whole-cell lysates from human primary melanocytes before and 1 to 6 hours after irradiation with UVB. Blot is representative of three independent experiments. Data are means ± SD of three experiments.

Because most Cdh1 and APC/C component mutations found in melanomas were of the CC-TT or C-T UV-induced signature (table S1), we determined the response of Cdh1 after UVB exposure in melanocytes. Consistent with previous reports that acute UV irradiation triggers proteolysis of Cdh1 (34), and that PAX3 protein abundance is up-regulated after UVB exposure in melanocytes (12), we found that Cdh1 protein abundance was decreased and that of PAX3 (and a downstream target, MITF) was increased after UVB exposure in melanocytes (Fig. 5E). These results further indicate the biological connections between Cdh1 and PAX3 in melanocytes.

To further identify the potential role of APC/CCdh1 in regulating melanocyte function, we investigated the role of APC/CCdh1 in melanin production. In this regard, we found that silencing Cdh1 in B16 mouse melanocytes significantly increased PAX3 abundance (Fig. 6A), melanin production (Fig. 6B), and tyrosinase activity (Fig. 6C), as well as the expression of MITF (fig. S6A), TYR that encodes tyrosinase (fig. S6B), and several other melanosome-associated and MITF target genes (fig. S6C). A similar increase in these markers of the PAX3-MITF-tyrosinase signaling pathway was observed after depleting Cdh1 in human melanocytes at both the protein or enzymatic (Fig. 6, D and E) and mRNA (fig. S6, D to F) levels. On the contrary, ectopic expression of Cdh1 significantly inhibited tyrosinase abundance and activity and melanin production in both mouse B16 melanoma cells (Fig. 6, G to I) and human A375 melanoma cells (Fig. 6, J to L). Therefore, these results support a physiological role of Cdh1 in suppressing MITF/tyrosinase-mediated melanin production. Consistent with this model, additional depletion of endogenous PAX3 in Cdh1-depleted early-passage human primary melanocytes prevents the induction of MITF, tyrosinase, and melanin production (fig. S6, G to I), suggesting that PAX3 is at least one of the targets suppressed by Cdh1 to limit activity of MITF and tyrosinase in melanocytes.

Fig. 6 APC/CCdh1 functions in melanin production.

(A to C) Immunoblot (A), melanin production (B), and tyrosinase activity (C) in murine B16 melanoma cells infected with lentiviral shRNA against Cdh1 or a negative control (shScr). Tyr, tyrosinase. (D to F) As in (A) to (C) in human A375 melanoma cells. (G to I) Immunoblot (G), melanin production (H), and tyrosinase activity (I) in B16 cells transfected with HA-tagged empty vector, WT Cdh1, or mutated Cdh1 (ΔC-box). (J to L) As in (G) to (I) in A375 cells. Data are means ± SD from three independent experiments.

APC/CCdh1 silencing results in enhanced survival of melanoma cells treated with doxorubicin

To further evaluate the physiological role of Cdh1 and the pathological role of its loss in melanoma, we investigated whether the loss of Cdh1 contributed to the proliferation and chemosensitivity of human melanoma cell lines in culture and in xenografts in mice. Depleting endogenous Cdh1 (Fig. 7A) increased the proliferation of A375 melanoma cells on plastic dishes (Fig. 7B and fig. S7A) and in soft agar (Fig. 7C), but this was prevented by simultaneously knocking down PAX3 (Fig. 7, B and C), confirming a role for PAX3 contributing to cell proliferation downstream of Cdh1 loss.

Fig. 7 Cdh1 deficiency promotes melanomagenesis.

(A) Immunoblot analysis of whole-cell lysates from A375 cells infected with the indicated lentiviral constructs. (B and C) A375 cells were subjected to colony formation (B) and soft agar (C) assays of A375 cells generated in (A) cultured for 21 days. Data are means ± SD from three independent experiments. (D and E) Tumor mass of A375 (D) or UACC62 (E) melanoma xenografts that were infected with shRNA against CDH1 or a negative control before subcutaneous inoculation. Tumor mass was measured at day 34 after injection. Data are means ± SEM from 10 tumors. *P < 0.05 (Student’s t test). (F and G) Tumor mass of A375 (F) or B16 (G) melanoma xenografts that were infected with targeted or control shRNA as described in (D) and (E). Data are means ± SEM from 10 tumors. *P < 0.05, **P < 0.001 (Student’s t test).

To further identify whether APC/CCdh1-regulated melanocyte function correlates with sensitivity to chemotherapies in melanoma cells, we examined the effect of Cdh1 depletion on the effectiveness of doxorubicin (35) in A375 and UACC62 melanoma cells. Doxorubicin causes DNA damage by intercalating DNA. In A375 cells, depleting endogenous Cdh1 expectedly increased the abundance of PAX3 and that of Polo-like kinase 1 (PLK1), the abundance of which coincides with and promotes G2/M transition in the cell cycle (fig. S7B). Although doxorubicin suppressed the abundance of PLK1 in control cells, it did not suppress PLK1 abundance in cells transfected with Cdh1 shRNA. Additionally, loss of Cdh1 increased the abundance of γ-H2AX, a marker of DNA damage, which further increased when coincidently exposed to doxorubicin (fig. S7B), suggesting that Cdh1 may have an additional role in promoting an effective response to DNA damage, possibly through suppression of the cell cycle to enable DNA repair or cell death, but this requires further investigation. Loss of Cdh1 conferred resistance to doxorubicin, as determined by cell proliferation and cell cycle profile in cultured A375 cells (fig. S7, C and D) as well as xenograft tumor mass in nude mice (Fig. 7D). Together with the loss of doxorubicin-induced suppression of PLK1 (fig. S7B), these results suggest that cell cycle control in naïve cells and cell cycle arrest and death in doxorubicin-treated cells are impaired in the absence of Cdh1. These observations in cultured A375 melanoma cells (fig. S7, B to D) were reproduced in UACC62 melanoma cells both in culture (fig. S7, E to G) and in xenografts (Fig. 7E). Furthermore, we found that loss of endogenous Cdh1 in melanoma cells may confer chemoresistance through loss of its suppression of PAX3 because additional depletion of endogenous PAX3 in Cdh1-depleted A375 (Fig. 7F) and B16 cells (Fig. 7G and fig. S7H) largely restored doxorubicin sensitivity in xenografts. Collectively, our data suggest that restoring Cdh1 or targeting PAX3 might aid therapeutic efficacy in melanoma patients.


Down-regulation of Cdh1 coincides with malignant progression of lymphoma (36), colorectal cancer (26, 37), and breast cancer (26, 37). Analysis of cancer gene expression databases confirmed these studies and further revealed that CDH1 expression is repressed in other solid tumors, including ovary, liver, prostate, and brain cancers (38). Here, we demonstrated that APC/CCdh1 abundance is down-regulated not only in melanoma but also in nevi, a precancerous stage, suggesting that APC/CCdh1 might suppress nevi formation. Our analyses implicate APC/CCdh1 as a critical regulator of melanocyte cell cycle progression. However, further investigation is required to determine whether this cell cycle–related role of APC/CCdh1 contributes to malignant transformation of melanocytes. In addition, it is yet unclear how APC/CCdh1 expression is repressed in melanoma growth and perhaps melanocyte malignant transformation. Various mechanisms could be involved in this repression process, including by CDH1 silent mutations, which we found in the TCGA and Broad databases, or by degradation of Cdh1 triggered by the ubiquitin ligase SCFβ-TrCP (32).

The major functionality of APC/CCdh1 in cell cycle regulation and tumor suppression in other cancers is largely attributed to its E3 ubiquitin ligase activity. Most Cdh1 substrates, including cyclin B, Aurora A and B, TPX2, Cdc6, and Cdc20 (3), are overexpressed in cancers that have chromosomal instability. PAX3 functions largely as a transcription factor that plays crucial roles in commitment, proliferation, and survival of melanocytes. It was previously reported that PAX3 is associated with type І Waardenburg syndrome (21, 39). Several independent groups, including ours, have demonstrated the up-regulation of PAX3 in melanoma (12, 14, 15, 40) and that increased PAX3 expression contributes to melanoma cell proliferation (12). Notably, previous reports demonstrated that knockdown of PAX3 by antisense oligonucleotide is detrimental to the survival of melanoma cells (14, 40). However, it was not clear how PAX3 expression is suppressed by upstream signaling and how PAX3 abundance is aberrantly increased in melanoma (12, 14, 15, 40). Here, we identified PAX3 as an APC/CCdh1 substrate in melanocytes and that APC/CCdh1-regulated cell cycle arrest may be mediated by direct Cdh1-induced, ubiquitin-mediated degradation of PAX3 in melanocytes. Our results further indicated that loss of APC/CCdh1 in melanoma contributes to the resistance to the DNA-damaging chemotherapeutic doxorubicin. Because DNA repair pathways can enable tumor cells to survive such chemotherapeutics, it is plausible that exploring the addition of DNA repair inhibitors to DNA-damaging chemotherapeutic drugs (41) might be efficacious in melanomas with low Cdh1 abundance. Collectively, our findings suggest that restoring Cdh1 or targeting PAX3 might be therapeutically beneficial in melanoma patients.



HA-Cdh1, HA-Cdc20, Myc-Cdh1, and GST-Cdh1 were described previously (30). HA-PAX3 and GST-PAX3 were generated as described previously (25). shRNA constructs targeting human Cdh1 (RHS4533-EG51343), mouse Cdh1 (RHS4534-EG56371), human Cdc20 (RHS4533-EG991), mouse Cdc20 (RHS4534-EG107995), human APC10 (RHS4533-EG10393), mouse Apc10 (RHS4534-EG68999), human Cdc27 (RHS4533-EG996), human PAX3 (RHS4533-EG5077), and mouse Pax3 (RHS4534-EG18505) were purchased from Open Biosystems. Site-directed mutagenesis to generate various PAX3 D-box mutants was performed using the QuikChange XL Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer’s instructions.

Cell culture, transfection, and infection

Cell culture conditions, including transfection, have been described previously (30). Human primary melanocytes were cultured in Medium 254 (M-254-500, Life Technologies) with the addition of human melanocyte growth supplement (S-002-5, Life Technologies). All other cell lines were grown in DMEM with 10% fetal bovine serum (FBS) in humidified incubators supplemented with 5% CO2. Lentiviral shRNA virus packaging and subsequent infection of various cell lines were performed according to the protocol described previously (42).


Anti-PAX3 (P03442) antibody was purchased from Aviva Antibody Corporation; anti-Mitf (C5) was purchased from Thermo Fisher. Anti-MART1 antibody (M2-9E3) was purchased from Novus. Anti–cyclin A (H-432), anti-Plk1 (F-8), anti-APC10 (B-1), anti-Cdc20 (E-7), and polyclonal anti-HA (Y-11) antibodies were purchased from Santa Cruz Biotechnology. Anti-tubulin (T-5168) was purchased from Bethyl Laboratories. Polyclonal anti-Flag antibody (F-2425), monoclonal anti-Flag (F-3165) antibody, anti-Flag agarose beads (A-2220), anti-HA agarose beads (A-2095), peroxidase-conjugated anti-mouse secondary antibody (A-4416), and peroxidase-conjugated anti-rabbit secondary antibody (A-4914) were purchased from Sigma. Monoclonal anti-HA antibody (MMS-101P) was purchased from Covance. Polyclonal anti-Cdh1 antibody (34-2000) was purchased from Invitrogen. Monoclonal anti-Cdh1 antibody (CC43) was purchased from Calbiochem.

Semiquantitative analysis (H score)

All specimens were stained with anti-Cdh1 (34-2000, Invitrogen) and/or anti-PAX3 antibodies (P03442, Aviva Antibody Corporation) in duplicate. For each sample, the Cdh1 and PAX3 expression were scored, as described before (12), as negative (0), weak (1), moderate (2), or strong (3), and the percentage of cells with each expression level were graded as 0 for 0 to 5%, 1 for 6 to 25%, 2 for 26 to 75%, and 3 for 76 to 100%. The values of the two scores were averaged to avoid the possible labeling variation within the lesion (43). A nonparametric rank correlation was used to analyze the correlation between Cdh1 and PAX3 expression.

Immunoblots and immunoprecipitation

Cells were lysed as described previously (44). The protein concentrations were measured with the bicinchoninic acid reagent (Bio-Rad) on a Beckman Coulter DU-800 spectrophotometer. Proteins were separated by 10% SDS–polyacrylamide gel electrophoresis (PAGE) and immunoblotted with indicated antibodies. For immunoprecipitation, 1000 μg of lysates was incubated with the appropriate antibody (1 to 2 μg) overnight at 4°C followed by 1-hour incubation with Protein A Sepharose beads (GE Healthcare). Immunocomplexes were washed five times with lysis buffer before being resolved by SDS-PAGE and immunoblotted with indicated antibodies.

GST pull-down assays

Binding to immobilized GST proteins was performed as described previously (33, 45).

Ubiquitylation assays

Ubiquitylation assays were performed as described (30). Briefly, Cdh1 control and knockdown cells were infected with Flag-PAX3 wild type or mutant and HA-ubiquitin lentivirus. Thirty-six hours after infection, 10 μM MG132 was added to block proteasome degradation, and cells were harvested in lysis buffer containing protease inhibitors. Whole-cell lysates (2 mg) were incubated with Flag-conjugated agarose beads for 4 hours, followed by washing four times with lysis buffer. Then, the washed pellets were boiled in SDS-containing sample buffer and resolved by SDS-PAGE.

In vitro ubiquitylation assays

The in vitro ubiquitylation assay was performed as described previously (35). Briefly, recombinant Cdh1 proteins were expressed in baculovirus system (Invitrogen) and purified using the Ni-NTA beads (Qiagen). APC/C was immunoprecipitated from the HeLa cell lysates using anti-Cdc27 antibody. Immunopurified APC/C was bound to recombinant Cdh1 protein and then subjected to the ubiquitylation reaction. APC/C-bound antibody beads were mixed with a reaction buffer [20 mM tris-Cl (pH 7.5), 150 mM NaCl, 1 mM dithiothreitol, 10% glycerol] containing purified E1 (80 μg/ml, BIOMOL), UbcH10 and UbcH5a (50 μg/ml each, Wako), ubiquitin (1.25 mg/ml, Sigma), adenosine triphosphate (ATP)–regenerating system [10 mM creatine phosphate, 2 mM ATP, 1 mM MgCl2, 0.1 mM EGTA, and rabbit creatine phosphokinase type I (39 U/ml)], and purified GST-PAX3 wild-type or mutant as substrate. Samples were immunoblotted to detect polyubiquitylation by anti-PAX3 antibody.


Immunohistochemical studies were performed on normal skin tissue or melanoma specimens in tissue slices purchased from Pantomics Inc. (cat. no. MEL961). Primary antibody included anti-Cdh1 antibody (1:100, cat. no. 34-2000, Life Technologies), anti-Pax3 (P03442, Aviva Antibody Corporation), and anti-MART1 antibody (M2-9E3, Novus). The immunohistochemical studies were performed as described previously (46) with slight modification by using the Dako DAB or AEC Detection Kit (Dako EnVision+ System, HRP) and Polink DS-MR-Hu A1 kit (D20-6A, Golden Bridge International) according to the kit instruction.


Cultured cells were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 15 min at room temperature. After three washes in PBS for 5 min each, the specimens were blocked in blocking buffer [1× PBS with 5% FBS, 1% bovine serum albumin (BSA), and 0.3% Triton X-100] for 1 hour. The blocking buffer was then aspirated, and primary antibodies of PAX3 (1:100) and Cdh1 (1:100) diluted in antibody dilution buffer (1× PBS with 1% BSA and 0.3% Triton X-100) were applied and incubated for 1 hour at room temperature. Specimens were then rinsed three times in PBS for 5 min each, and secondary antibodies conjugated with Alexa Fluor 488 and Alexa Fluor 568 (Invitrogen) were subsequently applied and incubated in the dark at room temperature for 1 hour. After washing in PBS, coverslips were mounted onto glass slides with ProLong Gold antifade reagent with DAPI (4′,6-diamidino-2-phenylindole) (Invitrogen). All images were taken using the Nikon deconvolution wide-field epifluorescence microscope (Cellular Imaging Core, Boston University).

Real-time reverse transcription polymerase chain reaction

Quantitative reverse transcription polymerase chain reaction procedures were followed as described previously (12). Gene-specific primer sets are as reported for PAX3: forward, 5′-AGCCGCATCCTGAGAAGTAA-3′; reverse, 5′-CTTCATCTGATTGGGGTGCT-3′. Primers used in this study including MITF-M, TYR, TYRP1, DCT, and PMEL were described previously (47).

Measurement of melanin contents

The melanin content was measured as described previously (48). Briefly, the cells were harvested, pelleted, and washed in PBS. Half of the cells were used for protein amount determination. The other half was responded in 400 μl of 1 N NaOH solution and incubated at room temperature for 2 hours; 100 μl of the solution was then loaded into a 96-well plate. The melanin content was determined by measuring the absorbance of the solution between 405 and 490 nm and compared with the melanin standard (Sigma). The melanin content was normalized to the protein content.

Tyrosinase activity assays

Tyrosinase activity was determined as described previously (49). Briefly, cells were sonicated in protein lysis buffer. After protein quantification, 250 μg of total protein lysates in a total volume of 100 μl of lysis buffer was loaded onto 96-well plate followed by the addition of 100 μl of 1 mM l-DOPA (3,4-dihydroxyphenylalanine). Then, the plate was incubated at 37°C for 1 hour, and the absorbances at 475 nm were recorded. The tyrosinase activity was expressed as the percentage of the control.

Statistical analysis

All data were presented as means ± SD of at least three independent experiments by Student’s t test for between-group differences. P < 0.05 was considered as statistically significant.


Fig. S1. Correlation and colocalization between Cdh1 and PAX3 abundance in normal skin and melanoma tissue.

Fig. S2. Depletion of endogenous Cdh1 does not affect PAX3 mRNA expression.

Fig. S3. Detection of colocalization between Cdh1 and PAX3 in melanocytes and melanoma cell lines.

Fig. S4. APC/CCdh1 E3 ligase does not target MITF in melanocytes.

Fig. S5. Interaction with PAX3 is impaired by the expression of melanoma-derived Cdh1 mutants.

Fig. S6. Depletion of endogenous Cdh1 in mouse and human melanocytes increases MITF and tyrosinase signaling.

Fig. S7. Cdh1 silencing decreases doxorubicin sensitivity in melanoma cells.

Table S1. Genetic mutations of APC/C components in TCGA melanomas and exome sequencing data.


Acknowledgments: We thank the members of Wei and Cui laboratories for proofreading the manuscript. Funding: This work was supported by the NIH (1RO1CA137098 to R.C. and GM089763 to W.W.); the American Cancer Society (RSG-09-022-01-CNE); the Harry J. Lloyd Charitable Trust (to R.C.); a Wellcome Trust grant (078327 to E.V.S.); the National Basic Research Program of China (2010CB530400 to Y.W.); the National Natural Science Foundation of Major International Cooperation Projects in China (81220108027 to Y.W.); the Program for Changjiang Scholars and Innovative Research Team in University (IRT1270 to Y.W.); the Joint Research Fund for Overseas Chinese, Hong Kong and Macao Young Scientists of the National Natural Science Foundation of China (81428025 to R.C.); and the Longhua Medical Project (to R.C.). C.R.G. is funded by the Ludwig Institute for Cancer Research, and R.C. and W.W. are American Cancer Society Research Scholars. Author contributions: J.C., X.D., L.W., W.W., and R.C. designed experiments and wrote the manuscript. J.C., X.D., and L.W. performed most of the experiments with help from H.W., J.Z., and Z. Xu. X.X. provided all tissue samples and diagnosed all samples. P.S.G. and E.V.S. provided melan-a and other cell lines and critical technical assistance and advice. Z. Xuan analyzed mutations in melanomas from TCGA, Broad database, and cited report. K.T.F. analyzed patient tumor specimens. C.R.G., D.V.F., and P.H. helped design experiments, interpret results, and write the manuscript. W.W., Y.W., and R.C. guided and supervised the project. All authors discussed the results and commented on the manuscript. Competing interests: K.T.F. is a consultant to GlaxoSmithKline, Novartis, and Roche. W.W. is a consultant for Cell Signaling Technology. Other authors declare that they have no competing interests. Data and materials availability: The Broad database of exome sequencing is publicly available at
View Abstract

Navigate This Article