KIT Is a Frequent Target for Epigenetic Silencing in Cutaneous … · 2017. 1. 31. · To investigate a causal relationship between KIT promoter hypermethylation and transcriptional

KIT Is a Frequent Target for Epigenetic Silencing inCutaneous MelanomaChristina Dahl1, Cecilie Abildgaard1, Rikke Riber-Hansen2, Torben Steiniche2, Johanne Lade-Keller2 andPer Guldberg1

The receptor tyrosine kinase KIT and its ligand, stem cell factor (SCF), are essential for the proliferation andsurvival of normal melanocytes. In melanomas arising on mucosal, acral, and chronically sun-damaged skin,activating KIT mutations have been identified as oncogenic drivers and potent therapeutic targets. Through aninitial whole-genome screen for aberrant promoter methylation in melanoma, we identified the KIT promoter as atarget for hypermethylation in 43/110 melanoma cell lines, and in 3/12 primary and 11/29 metastatic cutaneousmelanomas. Methylation density at the KIT promoter correlated inversely with promoter activity in vitro andin vivo, and the expression of KIT was restored after treatment with the demethylating agent 5-aza-20-deoxycytidine. Hypermethylation of KIT showed no direct or inverse correlations with well-documentedmelanoma drivers. Growth of melanoma cells in the presence of SCF led to reduced KIT expression andincreased methylation density at the KIT promoter, suggesting that SCF may exert a selection pressure for the lossof KIT. The frequent loss of KIT in cutaneous melanoma by promoter hypermethylation suggests that distinct KITsignaling pathways have opposing roles in the pathogenesis of melanoma subtypes.

Journal of Investigative Dermatology (2015) 135, 516–524; doi:10.1038/jid.2014.372; published online 25 September 2014

INTRODUCTIONThe growth, survival, migration, and differentiation ofmelanocytes are complex processes that are controlled byparacrine and autocrine cytokine networks (Imokawa, 2004).One of the important paracrine growth factors for themelanocytic lineage is stem cell factor (SCF). Duringembryonic development, SCF is critical for the survival andproliferation of neural crest–derived melanoblasts, and itserves as guidance cues that direct the migration of thesecells to their final destination in the hair follicle and epidermis(Lin and Fisher, 2007).

Signal transduction by SCF occurs through KIT, a class IIItyrosine kinase receptor that is expressed on several cell types,including hematopoietic progenitors, mast cells, melanoblasts,and differentiated melanocytes (Lennartsson and Rönnstrand,2012). Ligand binding causes KIT to homodimerize, leading tothe activation of its intrinsic kinase activity throughautophosphorylation of tyrosine residues. KIT has a numberof potential tyrosine phosphorylation sites, which interact withmultiple downstream signaling pathways, including the

phosphatidylinositol 3-kinase, MAP kinase, and Src familykinase pathways (Lennartsson and Rönnstrand, 2012). One ofthe downstream targets of these pathways is the melanocytemaster regulator the microphthalmia transcription factor(MITF) (Levy et al., 2006). Germline mutations in KIT, SCF,and MITF are associated with a range of pigmentationdisorders (Lin and Fisher, 2007), highlighting the importanceof the KIT/SCF system and its signaling to MITF in controllingvarious cellular activities in the melanocytic system.

A large body of evidence has implicated aberrant KITsignaling in the development and progression of melanoma.Several studies based on immunohistochemical evaluationhave shown that KIT is expressed in normal melanocytes andbenign nevi, but it is lost with progression to invasive andmetastatic forms (Montone et al., 1997; Shen et al., 2003; Zhuand Fitzpatrick, 2006). Consistent with these data, KITexpression is lost in a great proportion of melanoma-derivedcell lines (Lassam and Bickford, 1992; Natali et al., 1992;Zakut et al., 1993), and lack of KIT expression correlates witha higher metastatic potential of melanoma xenografts in nudemice (Gutman et al., 1994). Furthermore, forced KITexpression in KIT-deficient melanoma cell lines retards thegrowth of these cells in nude mice and confers susceptibilityto SCF-induced growth arrest and apoptosis in vitro (Huanget al., 1996). Although all of the above observations supporteda tumor-suppressing role of KIT in melanoma, this view hasmarkedly changed. Most importantly, genome-wide screenshave uncovered KIT amplifications and activating mutations ina large proportion of melanomas on palms, soles andsubungual sites (acral melanomas), mucosal membranes,and chronically sun-damaged skin (Curtin et al., 2006). The

See related commentary on pg 337ORIGINAL ARTICLE

1Danish Cancer Society Research Center, Copenhagen, Denmark and 2Instituteof Pathology, Aarhus University Hospital, Aarhus, Denmark

Correspondence: Per Guldberg, Danish Cancer Society Research Center,Strandboulevarden 49, DK-2100 Copenhagen, Denmark. E-mail:[email protected]

Received 22 April 2014; revised 7 August 2014; accepted 12 August 2014;accepted article preview online 1 September 2014; published online 25September 2014

Abbreviations: ESTDAB, The European Searchable Tumour Line Database;MeDIP, methylated DNA immunoprecipitation; MITF, microphthalmiatranscription factor; MS-MCA, methylation-sensitive melting-curve analysis;SCF, stem cell factor

516 Journal of Investigative Dermatology (2015), Volume 135 & 2015 The Society for Investigative Dermatology

http://dx.doi.org/10.1038/jid.2014.372mailto:[email protected]

majority of KIT mutations occur in the juxtamembrane regionof the receptor, which is also a target for mutations in 80% ofgastrointestinal stromal tumors and smaller proportions ofsome hematological malignancies. These mutations lead toligand-independent activation of KIT and its downstreamsignaling pathways (Lennartsson and Rönnstrand, 2012),and they constitute potent therapeutic targets for thetyrosine kinase inhibitor imatinib mesylate, with clinicalactivity observed in patients with metastatic gastrointestinalstromal tumor (Heinrich et al., 2003) and in patients withmetastatic KIT-mutated extracutaneous melanomas (Guo et al.,2011; Hodi et al., 2013). Collectively, these later findings haveestablished KIT as a bona fide oncogene in melanoma.

Research over the past 30 years has identified the maingenetic drivers in cutaneous melanoma (reviewed in Dahl andGuldberg, 2007; Tsao et al., 2012). In an effort to generate, ona whole-genome scale, a profile of epigenetic alterations incutaneous melanoma, we initially compared promotermethylation patterns of four well-characterized melanomacell lines with that of normal cultured melanocytes. Thisanalysis led to the identification of KIT as a frequent target forepigenetic silencing in cutaneous melanoma.

RESULTSWhole-genome promoter methylation profiling in melanoma celllines and cultured melanocytes

Genome-wide promoter methylation profiling was performedin normal cultured human epidermal melanocytes and fourhuman melanoma cell lines (ESTDAB-019, ESTDAB-013,ESTDAB-024, and SK-MEL-28). We used the well-establishedmethylated DNA immunoprecipitation (MeDIP) approach,which uses an anti-5-methylcytosine antibody to provideenrichment of DNA fragments containing 5-methylcytosine(Weber et al., 2005). The enriched DNA was hybridized tohuman promoter microarrays containing 244 k 60-mer probes,which cover 5.5 kb upstream to 2.5 kb downstream of thetranscription start site of B21,000 defined human RefSeqs.This approach allows unbiased analysis of gene promoters,including non-CpG island promoters, which can also be targetsfor silencing by DNA methylation in tumorigenesis (Han et al.,2011). As our focus in this study was on DNA methylationchanges with a potential impact on gene expression, werestricted our analysis to probes within proximal promoterregions, defined here as 300 nucleotides upstream and 200nucleotides downstream of the transcription start site. Owingto the small sample size, we used M-values as the method forestimating methylation levels (Du et al., 2010).

When comparing the methylation profile of normal mela-nocytes with those of melanoma cells, we identified 663promoter regions that were hypermethylated in at least one ofthe four melanoma cell lines and 54 that were hypermethy-lated in all four cell lines (Supplementary Table S1 online).Examples of the array-based methylation profiling of thesegenes are shown in Figure 1a, illustrating the increasedmethylation levels around the transcription start site inmelanoma cells compared with normal melanocytes. Tovalidate the findings of the MeDIP analysis, we selected fourgenes for promoter methylation and gene expression analyses:

DDIT4L, NID1, PPP1R3C, and RRAD. Two of these genes(DDIT4L and PPP1R3C) have previously been shown to beaberrantly hypermethylated in melanoma (Furuta et al., 2006;Koga et al., 2009; Gao et al., 2013). The methylation status ofeach promoter was determined using methylation-sensitivemelting-curve analysis (MS-MCA), which measures methy-lation content as a function of the melting temperature (Tm) ofan amplicon generated from a bisulfite-treated template(Worm et al., 2001). Analysis of DDIT4L, NID1, PPP1R3C,and RRAD confirmed the higher methylation levels in all fourmelanoma cell lines compared with normal melanocytes(Figure 1b and data not shown). Quantitative RT-PCR analysisof the same four genes revealed an inverse correlation betweenpromoter hypermethylation and gene expression, with higherexpression levels in melanocytes and low to undetectableexpression in the cell lines (Figure 1c).

KIT is frequently hypermethylated and silenced in melanoma celllines

The KIT promoter was enriched in the MeDIP approach(Figure 2a; Supplementary Table S1 online) and was selectedfor more in-depth analysis. As shown in Figure 2b, ESTDAB-019 and SK-MEL-28 had an MS-MCA profile of the KITpromoter similar to that of the fully methylated control,melanocytes had a profile corresponding to the unmethylatedcontrol, and ESTDAB-013 and ESTDAB-024 had a compositeprofile. Bisulfite pyrosequencing confirmed that the KITpromoter was densely methylated in ESTDAB-019 and SK-MEL-28, unmethylated in melanocytes, and intermediatelymethylated in ESTDAB-013 and ESTDAB-024 (Figure 2c anddata not shown). Next, using MS-MCA, we determined themethylation status of the KIT promoter in 106 additionalmelanoma cell lines (Supplementary Table S2 online). Over-all, 43 of the 110 cell lines (39%) showed increased KITpromoter methylation levels.

Expression of KIT was analyzed in 70 of the 110 melanomacell lines using RT-PCR (Supplementary Table S2 online).Expression of KIT protein at the cell surface was confirmed byFACS analysis in selected cell lines (Supplementary Figure S1online). There was a strong correlation between KIT expressionand KIT promoter methylation, with increased methylationlevels in 19 out of 20 cell lines showing loss of KIT expression(P¼3�10�8; Fisher’s exact test). However, the correlationwas not absolute, as KIT mRNA was detected in 11 out of50 cell lines with a hypermethylated KIT promoter (22%). Tofurther characterize the association between methylationstatus and transcriptional activity of the KIT promoter, wedetermined the methylation density at 34 individual CpG sitesusing bisulfite pyrosequencing and measured expression levelsusing quantitative RT-PCR in 12 cell lines showing variousMS-MCA profiles of KIT hypermethylation (Figure 2d). Asshown in Figure 2e, KIT was expressed only in those three celllines with the lowest KIT promoter methylation density(ESTDAB-013, EST146, and EST168). Furthermore, KIT expres-sion was lost in all of the 14 cell lines with fully hypermethy-lated KIT promoters (Figure 2d and e, and data not shown),suggesting that methylation density at the KIT promoter is amain determinant of KIT expression.

C Dahl et al.Epigenetic Silencing of KIT in Cutaneous Melanoma

www.jidonline.org 517

http://www.jidonline.org

To investigate a causal relationship between KIT promoterhypermethylation and transcriptional silencing, we treatedfour nonexpressing melanoma cell lines with the demethylat-ing agent 5-aza-20-deoxycytidine. RT-PCR analysis showedthat 5-aza-20-deoxycytidine restored KIT expression in a dose-dependent manner (Figure 2f and data not shown). Further-more, FACS analysis showed that KIT was re-expressed at thecell surface in KIT-negative cells after treatment with 5-aza-20-deoxycytidine (Supplementary Fig S1b online). Collectively,these results suggest that KIT promoter hypermethylation is a

frequent cause of transcriptional silencing of KIT in melanomacell lines. Loss of the transcription factor AP-2 has beensuggested as a mechanism by which KIT expression is lostduring melanoma progression (Huang et al., 1998). However,AP-2 was expressed in all of the 27 melanoma cell linestested, including 11 lacking KIT expression (data not shown).

KIT hypermethylation in uncultured melanomas

A previous study found KIT promoter hypermethylation in tworhabdomyosarcoma cell lines, which was considered as an

ESTDAB-019 ESTDAB-013 ESTDAB-024 SK-MEL-28 HEMn

+4

PPP1R3C

PPP1R3C

NID1

NID1

+2

+1

0

–1

PPP1

R3C

–2

–4

0.35

0.3

0.25

–dF

/dT

Rel

ativ

e ex

pres

sion

0.2

0.15

0.1

0.05

0

1.4

1.2

1

0.8

0.6

0.4

0.2

0DDIT4L NID1 PPP1R3C RRAD

74 76 78 80 82 84 86 88 70 72 74 76 78 80 82

ESTDAB-019

ESTDAB-019

ESTDAB-013

ESTDAB-013

ESTDAB-024

ESTDAB-024

SK-MEL-28

SK-MEL-28

HEMn

HEMn

WGA

IVM

0

0.1

0.2

0.3

0.4

0.5

Temperature (°C) Temperature (°C)

93.3

815

Mb

93.3

819

Mb

93.3

824

Mb

93.3

828

Mb

93.3

832

Mb

93.3

836

Mb

234.

2939

Mb

234.

2943

Mb

234.

2947

Mb

234.

2952

Mb

234.

2956

Mb

234.

2960

Mb

+4

+2

+1

0

–1

–2

–4

Figure 1. Genome-wide screen for aberrant promoter methylation in melanoma cell lines. (a) Plots from the Agilent DNA Analytics software showing the

methylation patterns for the PPP1R3C and NID1 promoters in four melanoma cell lines and normal human melanocytes (HEMn). For each sample, Cy5-labeled

methylation-enriched DNA and Cy3-labeled input genomic DNA were competitively hybridized to an Agilent Human Promoter Array. Each point represents a

probe plotted at its genomic position (hg18 build; x axis) relative to its log2(Cy5/Cy3) (M-value; y axis). The gene is represented by horizontal bars. (b) Analysis of

the methylation status of the PPP1R3C and NID1 promoters by MS-MCA. Bisulfite-treated DNA was amplified in the presence of SYBR Green I, and the melting

characteristics of the PCR products were determined by continuous fluorescence monitoring during a temperature transition. Shown is the negative derivative of

fluorescence over temperature (� dF/dT) versus temperature. Universal methylated DNA (IVM) and whole-genome amplified (WGA) DNA provided positivecontrols for methylated and unmethylated alleles, respectively. (c) Expression levels of DDIT4L, NID1, PPP1R3C, and RRAD in melanoma cell lines and normal

melanocytes determined by qRT-PCR and normalized to PBGD expression. Expression in melanocytes was set at 1. The data represent the meanþ SD of threemeasurements. MS-MCA, methylation-sensitive melting-curve analysis; qRT-PCR, quantitative RT-PCR.


518 Journal of Investigative Dermatology (2015), Volume 135

in vitro artifact as aberrant KIT methylation was not detectedin uncultured tumor specimens (Enguita-German et al., 2011).To determine whether KIT promoter hypermethylation inmelanoma cell lines was an in vitro culture phenomenon,we first took advantage of five frozen tumor biopsies fromwhich some of our cell lines had been established. In all cases,cell line and corresponding tumor tissue had similar KITpromoter MS-MCA profiles. Most important, for the two celllines with hypermethylated KIT (ESTDAB-019 and ESTDAB-023), large fractions of hypermethylated KIT alleles weredetected in the corresponding uncultured specimens

(Figure 3a and Supplementary Table S2 online), demonstratingthat KIT hypermethylation had occurred in vivo.

We next investigated the methylation status of the KITpromoter region in frozen surgical biopsies from benign nevi(N¼ 2) and primary (N¼12) and metastatic (N¼ 29) cuta-neous melanomas using MS-MCA and confirmed the resultsfor selected samples using bisulfite pyrosequencing(Figure 3b). KIT promoter hypermethylation was detected in3 of the primary melanomas (25%) and 11 of the metastaticmelanomas (38%), and in none of the nevi (SupplementaryTable S3 online). Finally, to investigate whether KIT promoter

–4

55.2171 Mb

ESTDAB-019

HEMn

100

75

50

25

0

–25

5 10 15 20 25 50 35 40 45

SK-MEL-28

100

125

75

50

25

0

80

Met

hyla

tion

dens

ity (

%)

100

120

60

40

20

0

HE

Mn

ES

TD

AB

-019

ES

TD

AB

-013

ES

TD

AB

-023

ES

TD

AB

-024

ES

TD

AB

-027

ES

TD

AB

-127

ES

TD

AB

-146

ES

TD

AB

-151

ES

TD

AB

-168

SK

-ME

L-28

IVM

0.8

Rel

ativ

e ex

pres

sion

1

0.6

0.4

0.2

0

0 0.1 1 2

KIT

GAPDH

0.5

HE

Mn

ES

TD

AB

-019

ES

TD

AB

-013

ES

TD

AB

-023

ES

TD

AB

-024

ES

TD

AB

-027

ES

TD

AB

-127

ES

TD

AB

-146

ES

TD

AB

-151

ES

TD

AB

-168

SK

-ME

L-285 10 15 20 25 50 35 40 45

ESTDAB-013

100

125

75

50

25

0

5 10 15 20 25 50 35 40 45

ESTDAB-013 ESTDAB-024 SK-MEL-28 HEMn

55.2175 Mb 55.2180 Mb 55.2184 Mb 55.2188 Mb 55.2192 Mb

–2

–1

0

+1

+2

+4 0.5ESTDAB-019

ESTDAB-013

ESTDAB-024

SK-MEL-28

HEMn

WGA

IVM

0.4

0.3

0.2–dF

/dT

0.1

074 76 78 80

Temperature (°C)

82 84 86 88

5-aza-CdR (µM)

Figure 2. Evaluation of the methylation status of the KIT promoter region in melanoma cell lines and normal melanocytes. (a) MeDIP analysis showing the

pattern of CpG methylation at the KIT promoter. See legend to Figure 1a for details. (b) MS-MCA of KIT promoter methylation. (c) Plots from the PyroMark Q24

software showing methylation levels at individual CpG sites (shaded regions) in the KIT promoter determined by bisulfite pyrosequencing. The sequence at the

bottom refers to the nucleotide dispensation, and the y axis represents the intensity of the fluorescent signal. (d) Methylation density at the KIT promoter in

melanoma cell lines and HEMn. The data shown represent the results obtained for 34 CpG sites analyzed by bisulfite pyrosequencing. (e) KIT expression levels in

melanoma cell lines and HEMn determined by qRT-PCR and normalized to PBGD expression. The data represent the meanþ half-range of two measurements.(f) Re-expression of KIT induced by 5-aza-20-deoxycytidine (5-aza-CdR) in melanoma cells carrying a hypermethylated KIT promoter. HEMn, human melanocytes;

MeDIP, methylated DNA immunoprecipitation; MS-MCA, methylation-sensitive melting-curve analysis.




hypermethylation correlated with the loss of KIT expressionin vivo, we examined DNA from archived formalin-fixedparaffin-embedded tissues from 57 primary cutaneous mela-nomas and compared the results with an immunohistochem-ical analysis of KIT expression. Overall, KIT promoterhypermethylation was detected in 19 out of 20 KIT-negativeand 3 out of 37 KIT-positive tumors (P¼4.5�10� 11; Fisher’sexact test).

Association of KIT promoter hypermethylation with knownmelanoma driversMolecular cancer drivers often display patterns of mutualexclusivity across tumors, reflecting their partially redundantfunctions as individual components of the same oncogenicsignaling pathways (Vogelstein and Kinzler, 2004). Toinvestigate whether KIT promoter hypermethylation wasmutually exclusive to known melanoma drivers, wecompared the KIT methylation status in 105 melanoma celllines with the status of 15 common genetic (BRAF, NRAS,TP53, PTEN, INK4A (p16), ARF (p14), CCND1, MYC, CDK4,and MITF) and epigenetic (APC, IGFBP7, PYCARD, RARB,and RASSF1A) drivers, which have all been previously charac-terized in the same series of cell lines (Dahl et al., 2013). KITpromoter hypermethylation showed direct correlations withother DNA methylation events, consistent with a CpG islandmethylator phenotype (Issa, 2004; Tanemura et al., 2009), but

no statistically significant correlations with any of the geneticdrivers (Supplementary Table S4 online).

Exposure to SCF triggers epigenetic silencing of KIT inmelanoma cells

Previous work has shown that although SCF is required tosupport the proliferation and survival of normal melanocyticcells, it may inhibit the growth of KIT-expressing melanomacells (Funasaka et al., 1992; Zakut et al., 1993). Furthermore,Huang et al. (1996) showed that forced expression of KIT inthe KIT-negative human melanoma cell line A375SMrendered these cells susceptible to SCF-induced cell-cyclearrest and apoptosis. To recapitulate these studies, but avoid-ing the possible confounding effects of KIT overexpression, weexamined the effect of SCF on ESTDAB-013 melanoma cells,which express high levels of KIT and display a relatively lowdensity of KIT promoter hypermethylation (Figure 2d and e).We reasoned that these cells had a propensity to epigeneti-cally silence KIT when a selection pressure was re-established.Initial FACS analysis showed that KIT was expressed on thesurface of the vast majority of ESTDAB-013 cells (Figure 4a).After 3 days of treatment with 200 ng ml� 1 SCF, there was noincrease in the number of apoptotic cells, as determined byflow-cytometric analysis of annexin V levels (data not shown).However, after being cultured in the presence of SCF for3 weeks, these cells showed reduced expression of KIT, as

ESTDAB-023 biopsy

ESTDAB-023 cell line

PBL

IVM

Mel #7

200

150

100

50

0

5 10 15 20 25 30 35 40 45

–dF

/dT

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0

Temperature (°C)

Epi56(hypermethylated KIT promoter)

Epi96(unmethylated KIT promoter)

76 78 80 82 84 86 88

Mel #58

Mel #67

200

150

100

50

0

5 10 15 25 30 35 40 4520

200

150

100

50

0

5 10 15 20 25 30 35 40 45

Figure 3. Evaluation of KIT promoter methylation in uncultured melanomas. (a) MS-MCA of the KIT promoter in the ESTDAB-023 melanoma cell line and

corresponding uncultured tumor. Universal methylated DNA (IVM) and DNA from peripheral blood leukocytes (PBL) served as controls. (b) Bisulfite

pyrosequencing showing methylation levels at the KIT promoter in uncultured melanoma metastases. (c) Immunohistochemical analysis of KIT expression in

primary melanomas with (Epi56, left) or without (Epi96, right) KIT promoter hypermethylation. MS-MCA, methylation-sensitive melting-curve analysis. Scale

bar¼ 50mm.



determined by FACS (Figure 4a) and quantitative RT-PCR(Figure 4b) analysis, as well as an increase in methylationdensity at the KIT promoter (Figure 4c and d). These datasuggest that SCF can induce epigenetic downregulation of KITin melanoma cells in vitro.

DISCUSSIONStudies from various research disciplines have demonstratedan extensive heterogeneity of melanoma at the clinical,cellular, and molecular levels. Significant progress towardunderstanding this heterogeneity has been obtained throughdetailed genome-wide studies of somatic, genetic, and epige-netic alterations, which have uncovered distinct molecularprofiles across melanoma stages and subtypes (Whitemanet al., 2011; Tsao et al., 2012). One of the notable mole-cular differences lies within the profile of key driveroncogenes. Although BRAF and NRAS mutations are presentin the majority of cutaneous and conjunctival melanomas,mutations in two members of the Gaq family of guanosinetriphosphatases, GNAQ and GNA11, are common andmutually exclusive in uveal melanomas and blue nevi (VanRaamsdonk et al., 2010), and KIT mutations are found nearlyexclusively in melanomas arising on mucosal, acral, andchronically sun-damaged skin (Curtin et al., 2006). Themolecular background for this genetic variation remainsunclear but may be related to differences in the inherent pro-perties of precursor cells, differences in tissue microenviron-ment, or both (Whiteman et al., 2011).

Given the well-established function of mutated KIT as apotent melanoma oncogene, the finding that this receptor isepigenetically silenced in a large proportion of cutaneousmelanomas was unexpected. Specifically, we found that aCpG island in the promoter of KIT was hypermethylated inmore than one-third of melanoma cell lines and biopsies.Furthermore, methylation density at this region correlatedinversely with KIT expression in vitro and in vivo, and KITexpression could be restored by pharmacological DNAdemethylation. Several mechanisms have been described thatcan mediate the downregulation of KIT, including dysregu-lated expression of specific microRNAs (Felicetti et al., 2008;Igoucheva and Alexeev, 2009; Siemens et al., 2013) or the AP-2 transcription factor (Huang et al., 1998), consistent with theidea that loss of KIT is a consequence rather than a cause ofmelanoma progression. The frequent epigenetic silencing ofKIT owing to promoter hypermethylation more directlyimplicates the loss of KIT in melanoma pathogenesis, sub-stantiating previous suggestions that KIT may have a tumor-suppressive function in cutaneous melanoma. Furthermore,the strong correlation with transcriptional silencing suggeststhat KIT promoter hypermethylation represents the mainmechanism responsible for stably inherited repression of KITduring melanoma progression.

The well-documented melanoma drivers display clearpatterns of mutual exclusivity according to their functions inthe canonical oncogenic signaling pathways (Dahl et al.,2013). Epigenetic silencing of KIT did not correlate with any

103

103

102

102101

101

100

100 103102101100

103

102

101

100

A1

CD

117

(c-K

IT)

A2 A1 A2

+ SCF

120

100

80

60

40

20

40

50

30

Met

hyla

tion

dens

ity (

%)

20

10

0

0+ SCF

+ SCF

*

*

Rel

ativ

e ex

pres

sion

A3 A4 A3 A4

No staining

0.5ESTDAB-013

ESTDAB-013 + SCF

WGA

IVM

0.4

0.3

0.2

0.1

074 76 78 80 82 84 86 88

Temperature (°C)

No staining

A1: 99% A1: 52%A3: 1% A3: 48%

–dF

/dT

Figure 4. Epigenetic downregulation of KIT expression in melanoma cells in response to SCF. (a) FACS analysis showing the expression of KIT in

ESTDAB-013. Cells were grown in the presence or absence of SCF (200mM) for 3 weeks. (b) qRT-PCR analysis of KIT expression in ESTDAB-013 cells grown in thepresence or absence of SCF. Expression levels were normalized to PBGD expression, and expression in untreated cells was set at 100%. The data represent

the meanþ SD of two independent experiments. (c) MS-MCA and (d) bisulfite pyrosequencing showing increased levels of methylation at the KIT promoter inESTDAB-013 cells after 3 weeks in culture in the presence of SCF. SCF, stem cell factor.




of the known genetic melanoma drivers, suggesting that thetumor-suppressive role of KIT is associated with a hithertounknown independent molecular pathway. Furthermore, asKIT is expressed in large proportions of metastatic melanomasand melanoma cell lines, it may not have a classical‘‘gatekeeper’’ function in the context of cutaneousmelanoma. The growth-suppressive function of KIT in somemelanomas should probably best be viewed in the context ofmicroenvironmental cues and intracellular signaling. Indeed,growth of KIT-expressing melanoma cells in the presence ofSCF led to reduced KIT expression and increased methylationdensity at the KIT promoter, suggesting that SCF imposes aselection pressure for the loss of KIT. In this respect, our datasupport early studies showing that ectopic expression of KIT inmelanoma cells sensitizes these cells to SCF (Huang et al.,1996).

In the adult skin, production of SCF by keratinocytes,endothelial cells, and fibroblasts is a main regulator ofmelanocyte homeostasis, supporting the recruitment, prolif-eration, and survival of melanocytes and their precursors(Grichnik, 2008; White and Zon, 2008). The loss ofrequirement for ligand-dependent KIT activation during thetransformation of melanocytic cells may at least in part beattributed to the redundant pathway activation through geneticmodification, such as mutation of BRAF or NRAS. Themechanism conferring sensitivity of melanoma cells to SCFremains unknown, but it may be caused by a synthetic lethalrelationship between SCF-induced and oncogenic signaling. Itis known that growth-factor stimulation may have biphasiceffects, with hyperstimulation of the RAS/MAPK pathwayleading to cell cycle arrest, senescence, or death (Serranoet al., 1997). Furthermore, as SCF-KIT interaction mediatesmelanocyte adhesion to keratinocytes and thereby preventsmelanocyte proliferation (Haass and Herlyn, 2005), there maybe dual selection for melanocytes with an oncogenic mutationto lose KIT, both to escape from control by keratinocytes andto avoid overstimulation of the MAPK pathway.

The opposing functions of KIT in melanoma development,with growth-promoting effects in normal melanocytic cellsand extracutaneous melanomas versus growth-suppressiveeffects in cutaneous melanomas, may have therapeutic impli-cations. The identification of KIT as an oncogenic driver inextracutaneous melanomas (Curtin et al., 2006) has providedan important opportunity for targeted therapy of melanoma.However, clinical KIT inhibitors effectively target wild-typeKIT, as demonstrated by skin hypopigmentation in patientstreated with imatinib (Tsao et al., 2003), and therefore couldcontribute to the progression of early-stage melanomas thatare growth-inhibited owing to intrinsic expression of KIT.Therefore, caution should be taken against the use of KITinhibitors for the treatment of melanomas of cutaneous originor with unknown KIT status.

MATERIALS AND METHODSMelanoma cell lines, melanocytes, and reagents

The 110 human melanoma cell lines used in this study (listed in

Supplementary Table S2 online) have been described and character-

ized previously (Guldberg et al., 1997; Worm et al., 2004; Jönsson

et al., 2007; Dahl et al., 2013). The majority of these cell lines

(N¼ 105) were obtained from The European Searchable Tumour LineDatabase (ESTDAB) (http://www.ebi.ac.uk/ipd/estdab). SK-MEL-28

cells were purchased from the ATCC. Melanoma cells were

routinely cultured as monolayers in RPMI 1640 medium containing

10% fetal bovine serum and antibiotics at 37 1C and 5% CO2.Primary human melanocytes (Invitrogen, Carlsbad, CA) were

maintained in Medium 254 containing Human Melanocyte Growth

Supplement 2 (Invitrogen) at 37 1C and 5% CO2. SCF and 5-aza-20-deoxycytidine were purchased from Sigma-Aldrich (St. Louis, MO).

Pathologic specimensFresh-frozen biopsy specimens from benign nevi and melanomas

were obtained from the Department of Pathology, Rigshospitalet,

Copenhagen, Denmark (Worm et al., 2004). The clinicopathological

characteristics of these biopsies are listed in Supplementary Table S3

online. Formalin-fixed, paraffin-embedded sections from primary

melanomas were obtained from the Institute of Pathology, Aarhus

University Hospital, Aarhus, Denmark. All tumor samples were

removed as part of the patient’s treatment, and the study was

approved by the local ethics committees. All tumor samples were

removed as part of the patients’ treatment, and only tissue sections

that were not needed for diagnosis were used in the study. In

accordance with Danish law and approval by the Danish Ethics

Committee, patient consent was not required for the retrospective

analysis of archival tissue biopsies.

DNA isolation and bisulfite treatment

Genomic DNA was isolated using the Qiagen Mini Prep kit (Qiagen

GmbH, Hilden, Germany (cultured cells and fresh-frozen specimens))

or the Qiamp DNA FFPE Tissue Kit (Qiagen GmbH (formalin-fixed,

paraffin-embedded sections)) and quantified using a NanoDrop ND-

1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE).

Bisulfite conversion of DNA was carried out using the EZ DNA

Methylation-Goldt Kit (Zymo Research, Irvine, CA) according to the

manufacturer’s protocol.

MeDIP and microarray analysis

MeDIP assays were performed essentially as described (Weber et al.,

2005). Eight micrograms of genomic DNA extracted from melanoma

cells and cultured epidermal melanocytes were sonicated using a

Bioruptor (Diagenode, Liège, Belgium) to generate 200- to 600-bp

fragments. Fragmented DNA (1.6mg) was used as a reference sample,whereas the remaining was subjected to immunoprecipitation.

DNA was denatured for 5 minutes at 95 1C and immunoprecipitatedfor 4 hours at 4 1C with 10mg of monoclonal mouse antibody against5-methylcytosine (Clone 33D3; 1 mg ml� 1; Eurogentec, Liège,

Belgium) bound to pan-mouse IgG Dynal magnetic beads

(Invitrogen). The efficiency of the MeDIP was evaluated by real-

time quantitative PCR using previously validated targets (Dahl et al.,

2013). Primer sequences are listed in Supplementary Table S5 online.

Real-time quantitative PCR was performed using the Roche Light-

Cycler 2.0 and the FastStart DNA MasterPLUS SYBR Green I Kit

(Roche, Basel, Switzerland). Immunoprecipitated methylated DNA

was labeled with Cy5 fluorophore and the input genomic DNA was

labeled with Cy3 fluorophore (Agilent Technologies, Santa Clara,

CA). Equal amounts of labeled DNA from the enriched and the

reference samples were combined (2.5–4mg each) and hybridized to



http://www.ebi.ac.uk/ipd/estdab

the Human Promoter Array 244K (Agilent Technologies). Arrays were

then washed and scanned with an Agilent DNA microarray scanner.

Data extraction and normalization were performed using the Feature

extraction software, version 9.5.3.1 (Agilent technologies). DNA

analysis was performed using the DNA analytics program (Agilent

technologies), the UCSC database (http://genome.ucsc.edu), and the

Galaxy tool (http://galaxy.psu.edu).

MS-MCA and pyrosequencing

Methylation-specific melting curve analysis (MS-MCA) (Worm et al.,

2001) was performed using the LightCycler 1.1 and 2.0 instruments

(Roche) and the FastStart DNA Master SYBR Green I Kit (Roche).

Pyrosequencing was performed on a PyroMark Q24 platform, using

PyroMark Gold Q24 Reagents (Qiagen, Valencia, CA). Data analysis

was performed with the PyroMark Q24 software. Primer sequences

are listed in Supplementary Table S5 online. Enzymatically methy-

lated DNA (CpGenome Universal Methylated DNA; Millipore,

Billerica, MA) was used as a methylation-positive control. DNA from

peripheral blood leukocytes from a healthy donor and unmethylated

DNA prepared by whole-genome amplification (WGA; GenomePlex,

Sigma-Aldrich) were used as negative controls for methylation.

Reverse transcription PCR and quantitative reverse transcriptionPCR

Total RNA was isolated from melanoma cells and melanocytes using

the RNeasy kit (Qiagen) and quantified using a Nanodrop spectro-

photometer (Nanodrop Technologies). cDNA was synthesized from

2mg of RNA using random hexamers, oligo-dT primers, and Super-script III Reverse Transcriptase (Invitrogen). Conventional PCR was

carried out using a block thermocycler (GeneAmp PCR System 9600;

Perkin-Elmer, Norwalk, CT) and the HotStar Taq DNA Polymerase

(Qiagen). PCR products were analyzed in a 2% agarose gel. Real-time

quantitative PCR was performed using the Roche LightCycler 2.0 and

the FastStart DNA MasterPLUS SYBR Green I Kit (Roche). Primer

sequences are listed in Supplementary Table S5 online.

Immunohistochemistry

Immunohistochemistry was carried out on melanoma tissue micro-

arrays, as described previously (Lade-Keller et al., 2013). The sections

were incubated at room temperature for 30 minutes with a primary

antibody against KIT diluted 1:500 (polyclonal anti CD117 antibody,

Dako, Glostrup, Denmark). Bound primary antibody was visualized

using the Super Sensitive Polymer-HRP IHC kit (BioGenex, Fremont,

CA) and a novared chromogen (Novared, Vector Laboratories,

Petersborough, UK), and slides were counterstained with hematoxylin.

All slides were digitalized using Zeiss Mirax Scan (Zeiss, Birker�d,

Denmark) and evaluated on a computer screen using the Arrayimager

software (Visiopharm, H�rsholm, Denmark). All tissue microarrays

cores from each patient were evaluated together as a single sample.

Positivity was defined as discrete staining with a predominantly

membranous pattern. Internal positive control was basal cells and

keratinocytes in normal epidermis. Negative internal control was

normal keratinocytes in the outer layers of the epidermis. Liver tissue

was used as an external negative control. The percentage of positively

stained tumor cells was scored semiquantitatively and data were

subsequently dichotomized into two groups: the KIT-negative group

(0% tumor cells stained) and the KIT-positive group (40% tumor cellsstained).

FACS analysisSurface expression of KIT on melanoma cells was determined using

a phycoerythrin-conjugated anti-KIT/CD117 antibody (AC126-PE;

Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) and an isotype

control antibody (PE Mouse IgG1 k Isotype Control; BD Biosciences,San Jose, CA). The cell samples were analyzed using an FC500 MPL

Flow Cytometer (Beckman Coulter) and the CXP analytical software

(Beckman Coulter, Brea, CA).

Apoptosis

Apoptosis was measured using the Annexin V-FITC Apoptosis Detec-

tion Kit (BD Biosciences). Cells were harvested and washed twice in

cold phosphate-buffered saline and resuspended in binding buffer.

A total of 5� 105 cells were stained with Annexin V-FITC andpropidium iodide. After 30 minutes of incubation, the cell samples

were analyzed using an FC500 MPL Flow Cytometer (Beckman

Coulter).

Database accession

The array data discussed in this publication have been deposited

in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and are

accessible through the GEO Series accession number GSE53801

(http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE53801).

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSWe thank Vibeke Ahrenkiel and Ahmad Abdul-Al for technical assistance. Thisstudy was supported by grants from the Danish Cancer Society, the NeyeFoundation, and the Danish Cancer Research Foundation.

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http://www.nature.com/jid

REFERENCES

Curtin JA, Busam K, Pinkel D et al. (2006) Somatic activation of KIT in distinctsubtypes of melanoma. J Clin Oncol 24:4340–6

Dahl C, Christensen C, Jönsson G et al. (2013) Mutual exclusivity analysis ofgenetic and epigenetic drivers in melanoma identifies a link betweenp14ARF and RARbeta signaling. Mol Cancer Res 11:1166–78

Dahl C, Guldberg P (2007) The genome and epigenome of malignantmelanoma. APMIS 115:1160–75

Du P, Zhang X, Huang CC et al. (2010) Comparison of Beta-value and M-valuemethods for quantifying methylation levels by microarray analysis. BMCBioinformatics 11:587

Edgar R, Domrachev M, Lash AE (2002) Gene Expression Omnibus: NCBI geneexpression and hybridization array data repository. Nucleic Acids Res30:207–10

Enguita-German M, Gurrea M, Schiapparelli P et al. (2011) KIT expressionand methylation in medulloblastoma and PNET cell lines and tumors.J Neurooncol 103:247–53

Felicetti F, Errico MC, Bottero L et al. (2008) The promyelocytic leukemiazinc finger-microRNA-221/-222 pathway controls melanomaprogression through multiple oncogenic mechanisms. Cancer Res 68:2745–54

Funasaka Y, Boulton T, Cobb M et al. (1992) c-Kit-kinase induces a cascadeof protein tyrosine phosphorylation in normal human melanocytes inresponse to mast cell growth factor and stimulates mitogen-activatedprotein kinase but is down-regulated in melanomas. Mol Biol Cell 3:197–209



http://genome.ucsc.eduhttp://galaxy.psu.eduhttp://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE53801http://www.nature.com/jidhttp://www.nature.com/jidhttp://www.jidonline.org

Furuta J, Nobeyama Y, Umebayashi Y et al. (2006) Silencing of Peroxiredoxin2 and aberrant methylation of 33 CpG islands in putative promoterregions in human malignant melanomas. Cancer Res 66:6080–6

Gao L, Smit MA, van den Oord JJ et al. (2013) Genome-wide promotermethylation analysis identifies epigenetic silencing of MAPK13 in primarycutaneous melanoma. Pigment Cell Melanoma Res 26:542–54

Grichnik JM (2008) Melanoma, nevogenesis, and stem cell biology. J InvestDermatol 128:2365–80

Guldberg P, Thor Straten P, Birck A et al. (1997) Disruption of the MMAC1/PTEN gene by deletion or mutation is a frequent event in malignantmelanoma. Cancer Res 57:3660–3

Guo J, Si L, Kong Y et al. (2011) Phase II, open-label, single-arm trial ofimatinib mesylate in patients with metastatic melanoma harboring c-Kitmutation or amplification. J Clin Oncol 29:2904–9

Gutman M, Singh RK, Radinsky R et al. (1994) Intertumoral heterogeneity ofreceptor-tyrosine kinases expression in human melanoma cell lines withdifferent metastatic capabilities. Anticancer Res 14:1759–65

Haass NK, Herlyn M (2005) Normal human melanocyte homeostasis as aparadigm for understanding melanoma. J Investig Dermatol Symp Proc10:153–63

Han H, Cortez CC, Yang X et al. (2011) DNA methylation directly silencesgenes with non-CpG island promoters and establishes a nucleosomeoccupied promoter. Hum Mol Genet 20:4299–310

Heinrich MC, Corless CL, Demetri GD et al. (2003) Kinase mutations andimatinib response in patients with metastatic gastrointestinal stromaltumor. J Clin Oncol 21:4342–9

Hodi FS, Corless CL, Giobbie-Hurder A et al. (2013) Imatinib for melanomasharboring mutationally activated or amplified KIT arising on mucosal,acral, and chronically sun-damaged skin. J Clin Oncol 31:3182–90

Huang S, Jean D, Luca M et al. (1998) Loss of AP-2 results in downregulation ofc-KIT and enhancement of melanoma tumorigenicity and metastasis.EMBO J 17:4358–69

Huang S, Luca M, Gutman M et al. (1996) Enforced c-KIT expression rendershighly metastatic human melanoma cells susceptible to stem cell factor-induced apoptosis and inhibits their tumorigenic and metastatic potential.Oncogene 13:2339–47

Igoucheva O, Alexeev V (2009) MicroRNA-dependent regulation of cKit incutaneous melanoma. Biochem Biophys Res Commun 379:790–4

Imokawa G (2004) Autocrine and paracrine regulation of melanocytes inhuman skin and in pigmentary disorders. Pigment Cell Res 17:96–110

Issa JP (2004) CpG island methylator phenotype in cancer. Nat Rev Cancer4:988–93

Jönsson G, Dahl C, Staaf J et al. (2007) Genomic profiling of malignantmelanoma using tiling-resolution array CGH. Oncogene 26:4738–48

Koga Y, Pelizzola M, Cheng E et al. (2009) Genome-wide screen of promotermethylation identifies novel markers in melanoma. Genome Res19:1462–70

Lade-Keller J, Riber-Hansen R, Guldberg P et al. (2013) E- to N-cadherin switchin melanoma is associated with decreased expression of phosphatase andtensin homolog and cancer progression. Br J Dermatol 169:618–28

Lassam N, Bickford S (1992) Loss of c-kit expression in cultured melanomacells. Oncogene 7:51–6

Lennartsson J, Rönnstrand L (2012) Stem cell factor receptor/c-Kit: from basicscience to clinical implications. Physiol Rev 92:1619–49

Levy C, Khaled M, Fisher DE (2006) MITF: master regulator of melanocytedevelopment and melanoma oncogene. Trends Mol Med 12:406–14

Lin JY, Fisher DE (2007) Melanocyte biology and skin pigmentation. Nature445:843–50

Montone KT, Van Belle P, Elenitsas R et al. (1997) Proto-oncogene c-kitexpression in malignant melanoma: protein loss with tumor progression.Mod Pathol 10:939–44

Natali PG, Nicotra MR, Winkler AB et al. (1992) Progression of humancutaneous melanoma is associated with loss of expression of c-kit proto-oncogene receptor. Int J Cancer 52:197–201

Serrano M, Lin AW, McCurrach ME et al. (1997) Oncogenic ras provokespremature cell senescence associated with accumulation of p53 andp16INK4a. Cell 88:593–602

Shen SS, Zhang PS, Eton O et al. (2003) Analysis of protein tyrosinekinase expression in melanocytic lesions by tissue array. J Cutan Pathol30:539–47

Siemens H, Jackstadt R, Kaller M et al. (2013) Repression of c-Kit by p53 ismediated by miR-34 and is associated with reduced chemoresistance,migration and stemness. Oncotarget 4:1399–415

Tanemura A, Terando AM, Sim MS et al. (2009) CpG island methylatorphenotype predicts progression of malignant melanoma. Clin Cancer Res15:1801–7

Tsao AS, Kantarjian H, Cortes J et al. (2003) Imatinib mesylate causeshypopigmentation in the skin. Cancer 98:2483–7

Tsao H, Chin L, Garraway LA et al. (2012) Melanoma: from mutations tomedicine. Genes Dev 26:1131–55

Van Raamsdonk CD, Griewank KG, Crosby MB et al. (2010) Mutations inGNA11 in uveal melanoma. N Engl J Med 363:2191–9

Vogelstein B, Kinzler KW (2004) Cancer genes and the pathways they control.Nat Med 10:789–99

Weber M, Davies JJ, Wittig D et al. (2005) Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normaland transformed human cells. Nat Genet 37:853–62

White RM, Zon LI (2008) Melanocytes in development, regeneration, andcancer. Cell Stem Cell 3:242–52

Whiteman DC, Pavan WJ, Bastian BC (2011) The melanomas: a synthesis ofepidemiological, clinical, histopathological, genetic, and biologicalaspects, supporting distinct subtypes, causal pathways, and cells of origin.Pigment Cell Melanoma Res 24:879–97

Worm J, Aggerholm A, Guldberg P (2001) In-tube DNA methylation profilingby fluorescence melting curve analysis. Clin Chem 47:1183–9

Worm J, Christensen C, Gr�nbæk K et al. (2004) Genetic and epigeneticalterations of the APC gene in malignant melanoma. Oncogene 23:5215–26

Zakut R, Perlis R, Eliyahu S et al. (1993) KIT ligand (mast cell growthfactor) inhibits the growth of KIT-expressing melanoma cells. Oncogene8:2221–9

Zhu YI, Fitzpatrick JE (2006) Expression of c-kit (CD117) in Spitz nevus andmalignant melanoma. J Cutan Pathol 33:33–7



Kit Is A Frequent Target For Epigenetic Silencing In Cutaneous Melanoma��Introduction��Results��Discussion��Materials And Methods��References��

Documents

KIT Is a Frequent Target for Epigenetic Silencing in Cutaneous … · 2017. 1. 31. · To investigate a causal relationship between KIT promoter hypermethylation and transcriptional