clonogenic cell subpopulations maintain congenital melanocytic … · 2017. 1. 31. · oct4, an...

Clonogenic Cell Subpopulations Maintain CongenitalMelanocytic NeviChristelle Charbel1,2,11, Romain H. Fontaine1,2,11, Natacha Kadlub3,4, Aurore Coulomb-L’Hermine2,5,Thomas Rouille1,2, Alexandre How-Kit6, Philippe Moguelet7, Jorg Tost6,8, Arnaud Picard3,4,Selim Aractingi1,3,9,12 and Sarah Guegan1,2,10,12

Large congenital melanocytic nevi (lCMN) are benign melanocytic tumors associated with an increased risk ofmelanoma transformation. They result predominantly from a post-zygotic somatic NRAS mutation. These lesionspersist and even increase after birth proportionally to the child’s growth. Therefore, we asked here whether cellswith clonogenic and tumorigenic properties persisted postnatally in lCMN. Subpopulations of lCMN cellsexpressed stem cell/progenitor lineage markers such as Sox10, Nestin, Oct4, and ABCB5. In vitro, 1 in 250 cellsfrom fresh lCMN formed colonies that could be passaged and harbored the same NRAS mutation as the originalnevus. In vivo, lCMN specimens xenografted in immunocompromised mice expanded 4-fold. BrdUþ -proliferat-ing and label-retaining melanocytes were found within the outgrowth skin tissue of these xenografts, whichdisplayed the same benign nested architecture as the original nevus. lCMN cell suspensions were not able toexpand when xenografted alone in Rag 2� /� mice. Conversely, when mixed with keratinocytes, these cellsreconstituted the architecture of the human nevus with its characteristic melanocyte layout, lentiginoushyperplasia, and nested architecture. Overall, our data demonstrate that, after birth, certain lCMN cell subtypesstill display features such as clonogenic potential and expand into nevus-like structures when cooperating withadjacent keratinocytes.

Journal of Investigative Dermatology (2015) 135, 824–833; doi:10.1038/jid.2014.437; published online 13 November 2014

INTRODUCTIONCongenital melanocytic nevi (CMN) are benign melanocytictumors resulting from a neural crest–derived proliferation thatcan infiltrate the epidermis, dermis, and hypodermis. Thesenevi concern 1 in 20,000 births (Castilla et al., 1981). Theymay have major medical and socio-aesthetic implications andmoreover a risk toward melanoma transformation. Projected

adult size remains the major criterion for classification(Krengel et al., 2013). Large CMN (lCMN) are associatedwith a higher risk of melanoma (Marghoob et al., 1996; Ruiz-Maldonado, 2004; Krengel et al., 2006). Associated neuro-logical abnormalities and tumors are also reported in thesepatients (Kinsler et al., 2013b). Familial cases have rarely beenreported (de Wijn et al., 2010).

CMN are now considered to result predominantly from apost-zygotic somatic NRAS or BRAF mutation, both mutationsbeing usually exclusive (Papp et al., 2005; Ichii-Nakato et al.,2006; Bauer et al., 2007; Kinsler et al., 2013b; Charbel et al.,2014). Indeed, exome sequencing analysis conducted onlCMN samples by our group has shown that NRAS mutationwas the sole recurrent somatic mutation identified and there-fore sufficient to drive nevus development (Charbel et al.,2014). Cutaneous melanocytic cells are primarily derived fromthe developing neural crest. In mouse, melanocytic precursorcells migrate in the dorsolateral pathway (Mackenzie et al.,1997; Jiao et al., 2006). Migration then continues ventrallybefore the cells colonize the epidermis and developing hairfollicles (Grichnik, 2008).

Pathogenesis of nevi in humans remains uncompletelyunderstood. It was suggested that nevi could arise from bulgemelanocyte stem cells in mice (Krengel 2005; Nishikawa-Torikai et al., 2011; Tanimura et al., 2011). Others have alsoshown that transformed nerve sheath stem cells may lead toformation of occult prenatal nevi (Cramer and Fesyuk, 2012).Barnhill et al. have reported that embryonic stem cells from

ORIGINAL ARTICLE

1Saint Antoine Research Center, U938, Institut National de la Sante et de laRecherche Medicale (INSERM), Paris, France; 2Universite Pierre et Marie Curie-Paris VI, Paris, France; 3Universite Rene Descartes-Paris V, Paris, France;4Department of Maxillofacial and Plastic Surgery, Hopital Necker, Publique-Hopitaux de Paris, Paris, France; 5Department of Pathology, Hopital Trousseau,Publique-Hopitaux de Paris, Paris, France; 6Laboratory for FunctionalGenomics, Fondation Jean Dausset – CEPH, Paris, France; 7Department ofPathology, Hopital Tenon, Publique-Hopitaux de Paris, Paris, France;8Laboratory for Epigenetics & Environment, Centre National de Genotypage,CEA–Institut de Genomique, Evry, France; 9Department of Dermatology,Hopital Cochin, Publique-Hopitaux de Paris, Paris, France and 10Department ofDermatology, Hopital Tenon, Publique-Hopitaux de Paris, Paris, France

Correspondence: Sarah Guegan, INSERM U938-UPMC, Saint-AntoineResearch Center, 27 Rue de Chaligny 75012 Paris, France.E-mail: [email protected]

11These authors contributed equally to this work as first co-authors.12These authors contributed equally to this work as senior co-authors.

Received 25 October 2013; revised 23 September 2014; accepted 26September 2014; accepted article preview online 13 October 2014;published online 13 November 2014

Abbreviations: CMN, congenital melanocytic nevi; DIV, days in vitro; lCMN,large congenital melanocytic nevi; LRC, label-retaining cells; SSM, superficialspreading melanoma

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

http://dx.doi.org/10.1038/jid.2014.437

mailto:[email protected]

the neural crest may migrate along vessels and other tracksthrough angiotropism, resulting in the genesis of CMN(Barnhill et al., 2010). However, such stem cell populationshave not yet been isolated in humans. In CMN, the presenceof a unique NRAS somatic mutation suggests that these tumorsresult from the development of a melanocytic clone. Thesecells then populate subcutaneous, dermal, and epidermalstructures where they organize into a nevus (Grichnik,2008). The analysis of transgenic mice in which constitu-tively activated NRAS is restricted to the melanocyte lineagealso supports this hypothesis. Indeed, these mice developgeneralized, nevus-like dermal melanocytosis (Ackermannet al., 2005). Altogether, these data support clonal prolife-ration in CMN. Of note, after birth, CMN increase in surfaceproportionally to the child’s growth. Such evolution suggestspostnatal potential for proliferation in CMN. Therefore, ourstudy aimed at characterizing nevocytic cells capable ofmaintaining and propagating CMN.

RESULTSImmunohistochemical and flow cytometric analyses indicatethe presence of minor lCMN populations expressing stem cellmarkers

lCMN (n¼ 4 specimens) and superficial spreading melanoma(SSM; n¼ 4) were compared (Figure 1). The lesions weredissected in a gridded manner and various sectors analyzedper lesion: the number of positive cells was counted andreported to the total number of DAPIþ cells per percentage oftotal cells per 100mm2 surface areas, both in the epidermisand dermis (three per sector and per skin region).

The expression of melanocytic/melanoma marker such asMITF and HMB45 was first analyzed. MITF staining in thelesions demonstrated a higher level of differentiated melano-cytic infiltrate in lCMN as compared with SSM (Figure 1a;lCMN mean±SEM: 7.07±0.51; ***Pp0.0001). Both SSM andlCMN harbored HMB45þ cells; these cells were predomi-nantly found in the epidermis and upper dermis in lCMN witha histochemical maturation (Supplementary Figure S1a online).

Neural crest stem cell marker expression was then eval-uated. Sox10, a key regulator of pigment cell formation, isexpressed throughout all stages of melanocyte differentiation(Sommer, 2011), as well as in CMN and melanoma (Shakhovaet al., 2012). Nestin is expressed in nevi and melanoma(Sharma et al., 2010). In lCMN, Sox10 was predominantlyexpressed in the dermis. When compared with SSM, asignificantly lower number of Sox10þ cells was found inlCMN (Figure 1b; SSM mean±SEM: 7.69±0.75%; lCMNmean±SEM: 5.34±0.53%; *Pp0.05). The fact that lCMNharbor more MITFþ cells and less Sox10þ cells comparedwith melanoma probably relates to a higher ratio ofdifferentiated cells in nevi. Nestin was highly expressedboth in lCMN and SSM without any statistical difference(Figure 1c).

Oct4, an embryonic stem cell transcription factor mainlyassociated with an undifferentiated phenotype and adult germcell tumors, may be expressed in the basal layer of human skinepidermis and in melanoma (Tai et al., 2005; Cheli et al.,2011). Oct4-MelanA double stainings were performed.

Oct4þMelan-Aþ cells were detected in both SSM andlCMN without any statistical difference (Figure 1d).

Chemoresistance mediator ABCB5 is expressed in humanmalignant melanoma initiating cells (Schatton et al., 2008;

MITF

Nestin

Ki67

Sox10/DAPI

Oct4/Melan-A/DAPI

ABCB5/Melan-A/DAPI

SSM /CMN

SSM /CMN

SSM0

0

0

5

10

15

20

0

5

10

15

20

0

5

10

15

20

20

30

40

50

10

0

2

4

6

8

10

5

10

15

20

MIT

F+ c

ells

(% o

f tot

al c

ells

)S

ox10

+ c

ells

(% o

f tot

al c

ells

)N

estin

+ c

ells

(% o

f tot

al c

ells

)A

BC

B5/

Mel

an-A

+ c

ells

(% o

f tot

al c

ells

)K

i67+

cel

ls(%

of t

otal

cel

ls)

ICMN

SSM ICMN

SSM ICMN

SSM ICMN

SSM ICMN

SSM ICMN

***

*

***

***

SSM /CMN

SSM /CMN

SSM /CMN

SSM /CMN

Oct

4/M

elan

-A+ c

ells

(% o

f tot

al c

ells

)

Figure 1. Human lCMN express initiating stem cell markers. Immuno-

histochemical stainings of superficial spreading melanoma (SSM) and human

large congenital melanocytic nevi (lCMN): MITF (a), Sox10 (b), Nestin(c),

Oct4/Melan-A (d), ABCB5/Melan-A (e), and KI67(f). MITFþ , ABCB5þ , KI67þ

cells were stained with brown/red AEC chromogen. Melan-Aþ and Sox10þ

cells were stained using Alexa488 secondary antibody. Oct4þ cells were

stained using Cy3 secondary antibody. Scale bar¼ 100mm. Histograms show

the number of positive cells per total cells in both the epidermis and dermis.

(lCMN n¼5; melanoma, n¼ 4). *Represents differences between SSM and

lCMN. Bars: SEM. *Pp0.05, **Pp0.001, ***Pp0.0001 in Student’s t-test.

Red dots delineate the edge of the lesions.

C Charbel et al.Clonogenic Cells in Large Congenital Nevi

www.jidonline.org 825

http://www.jidonline.org

Sharma et al., 2010). As expected, double staining revealeda higher level of ABCB5þMelan-Aþ cells in SSM whencompared with lCMN (Figure 1e; SSM mean±SEM: 33.64±6.33%; lCMN mean±SEM: 8.79±1.23%; ***Pp0.0001).Expression of CD44, a well-accepted cancer stem cell surfacemarker, has been reported in benign and malignant melano-cytic lesions (Leigh et al., 1996). Here, benign lCMN as wellas malignant SSM were featured by a similar CD44 expression(Supplementary Figure S1b online).

As expected, the number of proliferating cells estimatedthrough Ki67 labeling was strikingly lower in lCMN as com-pared with SSM (Figure 1f; SSM mean±SEM: 13.66±1.03%;lCMN mean±SEM: 8.64±0.69%; ***Pp0.0001). Of note, inlCMN, Ki67 expression was intense and localized mostly tothe basal layer of the epidermis where melanocytes mainlyreside usually within nevocytic nests, indicating a proliferativeactivity of melanocytes. This observation could possiblyexplain why lCMN have a higher risk of transformationcompared with small congenital or acquired nevi.

In order to better evaluate the percentages of cells expres-sing stem cell markers in lCMN, FACS was performedon CD31�CD45�DAPI� sorted lCMN cell suspensions.Single-positive cells were present in these cell suspensionswith 2.97% ABCB5þ cells (Supplementary Figure S2a and bonline), 2.79% CD133þ cells, and 1.53% CD166þ cells(Supplementary Figure S2b online). Neural crest nerve growthfactor Receptor CD271 expressed by human neural crest–derived tissues can characterize human melanoma–initiatingcells (Boiko et al., 2010). CD271 expression has not beenpreviously reported in congenital nevi. In our study, 1.05% oflCMN cells were CD271þ (Supplementary Figure S2a and bonline). There was no co-expression of these stem cell mar-kers. Altogether, these immunostainings and flow cytometricanalyses demonstrate the presence within lCMN of cellsexpressing several stem cell markers, thus reflecting lCMNheterogeneity.

In vitro characteristics of spheroid colonies obtained from lCMN

Non-adherent cell culture using optimized sphere mediumwas performed to better characterize the nevocytic cellsuspensions. Sphere-forming and clonogenic assays allowevaluation of both cell self-renewal and differentiation at thesingle-cell level (Schatton et al., 2008; Pastrana et al., 2011).Melanospheres generated from primary melanoma andmelanoma cell lines have extensively been analyzed (Fanget al., 2005; Perego et al., 2010; Cheli et al., 2011) and servedas controls in this study.

lCMN cell suspensions, primary melanoma cells, andMel501 melanoma cells were cultured in three differentoptimized media: N2, B27, or B27þ insulin (See Materialsand Methods section for details). Primary and Mel501 mela-noma cells and all lCMN cell suspensions (n¼5) gave rise tospheroid colonies in the 3 media tested (Supplementary FigureS3a online). The B27þ insulin medium was chosen as the mostadequate (Supplementary Figure S3b online). After 7 and 13days of culture (DIV7 and DIV13), the number and size oflCMN colonies were significantly lower as compared withthose of Mel501 colonies (***Pp0.0001 in Figure 2a), as

expected in a benign proliferation. Similar results were obtainedwith primary melanoma colonies (Supplementary Figure S3bonline). Mel501 spheres showed a constant increase in sizeand number between DIV7 and DIV13 (Figure 2a; yPp0.05,yyyPp0.0001). On the other hand, lCMN colonies reached aplateau on DIV7 and showed a very slight increase in theirnumber but not in their size between DIV7 and DIV13(Figure 2a, yPp0.05; Supplementary Figure S3c online).

To further assert the replicative abilities of CMN spheres, thesewere enzymatically dissociated on DIV13, replated de novo atvery low density in optimized sphere medium, and studied overfive passages in vitro. lCMN cells constantly formed spheresafter each passage (Figure 2b). A special care was given to celldigestion, absence of aggregation being constantly verified ateach passage. Sphere number and diameter evaluated on DIV7and 13 did not vary throughout the experiment (graphs inFigure 2b). On DIV13, lCMN spheres expressed Melan-A aswell as the stem cell markers Nestin and Oct-4 (Figure 2c).Melan-A and/or Nestin expression was also found in all adherentcultured cells (Supplementary Figure S4a), showing that in vitroculture allowed melanocyte selection. Of note, the percentagesof Melan-A-positive cells were previously evaluated on DIV0 (atleast 60% of Melan-Aþ cells; Supplementary Figure S4b online).

The percentage of lCMN cells giving rise to colonieswas calculated using an in vitro limiting dilution assay(see Materials and Methods section and Liu et al., 2013).The frequencies of cells displaying clonogenic potentialwere 1 in 302 cells on DIV7 and 1 in 254 cells on DIV13,respectively (n¼3 specimens; Figure 3a).

Soft agar colony formation assay that monitors anchorage-independent growth was used in order to confirm sphereclonality and assess anchorage-independent growth. On DIV13, individually seeded lCMN cells were able to form coloniesin soft agar only when cultured in optimized sphere medium(Supplementary Figure S4c online; photomicrographs inFigure 3b). Colony diameters measured 60–70mm as in theclassical sphere forming assay (n¼5; graphs in Figure 2a andb, and 3b). Therefore, lCMN harbored cells capable ofdividing and forming spheres, as well as soft agar coloniesat a frequency of one in 250 cells.

Finally, the NRAS mutational status of these spheroidcolonies was ascertained by pyrosequencing: NRAS exon 3genotyping was performed on five lCMN specimens as well ason the in vitro colonies arising after adequate culture of thecorresponding cell suspensions (Figure 3c). The in vitrocolonies always bore the same c.181C4A, p.Q61K orc.182A4G, p.Q61R NRAS mutation as the original nevus.Altogether, these data support the existence of NRAS-mutatedcells capable of clonal proliferation within lCMN.

In vivo potential to induce benign tumors

In order to assess the tumorigenic potential of the nevocyticcells, xenotransplantations were performed using immuno-compromised mice. Standardized specimens (3-mm punchbiopsies) of lCMN were xenografted in brown adipose tissuein Rag2� /� mice. Grafts were harvested 3 and 7 months posttransplantation, and tumor volume was measured (n¼8grafted specimens; Figure 4a). Seven months post grafting,


826 Journal of Investigative Dermatology (2015), Volume 135

lCMN tissues had increased 4-fold when compared with theoriginal biopsies (graph in Figure 4a). Hematoxylin-eosinstaining (Figure 4b) revealed a xenograft architecture (‘‘graftednevus’’) similar to the original nevus (‘‘pregraft’’) and a

surrounding pigmented outgrowth tissue (‘‘graft neoepi.’’)enclosing a cystic structure. Anti-human Melan-A and anti-Sox10 labeling demonstrated in the grafted nevus as well asin the developing neoepidermis the presence of human

Mel501

DIV 0 DIV 7 DIV 13

DIV

7 Num

ber

of c

olon

ies

per

0.25

mm

2D

iam

eter

of c

olon

ies

(μm

)

DIV

13

Pas

sage

P3

Num

ber

of c

olon

ies

per

0.25

mm

2

Dia

met

er o

f col

onie

s (μ

m)

/CMN

5

4

3

§

§

§§§

***

*** ***

***2

1

300

200

2.5

2.0

1.5

1.0

0.5

0.0P1 P2 P3 P4 P5

80

60

40

20

0P1 P2 P3 P4 P5

100

0

0Mel501 /CMN

DIV 7

Mel501 /CMN

DIV 13

Mel501 /CMN

DIV 7

Mel501 /CMN

DIV 13

DIV 7 DIV 13

NestinMelan-A / DAPI Oct4 / DAPI

P1 P2 P3 P4 P5 P1 P2 P3 P4 P5

DIV 7 DIV 13

Figure 2. Large congenital melanocytic nevi (lCMN) melanocytic cells are capable of forming sphere colonies and self-renew. (a) Photomicrographs (left panel)

and histograms (right panel) of Mel501 melanoma and lCMN colonies on DIV7 and DIV13 (lCMN n¼ 5). Scale bar¼100mm. y Represents the difference between

DIV7 and 13 in the same group; * represents difference between Mel 501 and lCMN. Bars: SEM. yPp0.05, yyyPp0.0001, ***Pp0.0001 in Student’s t-test.

(b) Photomicrographs (top panel) and histograms (bottom panel) for lCMN at passage 3. Scale bar¼ 100mm. (c) Immunocytochemistry of lCMN spheres. Nestinþ

cells (middle panel) were stained with AEC chromogen. Melan-Aþ (left panel) and Oct4þ cells (right panel) were stained, respectively, with Alexa 488 and Cy3

secondary antibody. Scale bar¼ 50mm.




Melan-Aþ cells (Figure 4c) and Sox10þ cells (Figure 4d)organized in melanocytic nests, thus reproducing the originalnevus structure. Double labeling with anti-human Melan-A/anti-BrdU or anti-Sox10/anti-BrdU antibodies was performedin mice that received a single BrdU injection 30 minutesbefore euthanasia (n¼3 grafted specimens). In these graftednevi, Melan-Aþ /BrdUþ cells were present in both the graftedpart of the nevus and neoepidermis (Figure 4e). In anothergroup of mice in which BrdU was injected 1 month beforeeuthanasia, the presence of few human Sox10þ /BrdUþ cellscorresponding to Label-Retaining Cells (LRCs) could be con-firmed (Figure 4f). Quantitations of double-positive cells didnot show any statistical difference between the grafted nevusand the neoepidermis. Xenografts displayed 4.5% of Melan-Aþ /BrdUþ -proliferating cells and 6.5% of Sox10þ /BrdUþ

label–retaining cells (Histograms in Figures 4e and f). Thepercentage of Sox10þ /BrdUþ label–retaining melanocyteswas 4% (histogram in Supplementary Figure S5a online).These data demonstrate the presence of quiescent melanocyticcells within lCMN with features of label-retaining cells.

To fully determine the ability of these cells to reproduce anevus, other transplantation experiments using lCMN suspen-sions were conducted. One million fresh nevocytic cellsuspensions were mixed with Matrigel then grafted into theright flank of Rag2� /� mice (n¼5 injections from fivespecimens). Seven months post grafting, Matrigel was har-vested. Inside the Matrigel, low numbers of DAPIþ cells andvery few isolated human Melan-A and Sox-10þ cells were stilldetectable (Figure 5a, white arrows), indicating the persistenceof isolated long-term surviving human melanocytes. Theabsence of nevi development in Matrigel injected withnevocytes alone contrasted with the expanding grafted nevi.This strongly encouraged us to evaluate melanocytes’ need foranother cell type in order to expand––namely keratinocytes.To test this hypothesis, human non-tumorigenic keratino-cytes (the HaCaT cell line) were mixed with the lCMNcell suspension in order to mimic the required physio-logical environment. Different concentrations of sortedCD31� /CD45� cell suspensions (1.5 million, 5,00,000 and4,00,000 cells) were injected with HaCaT keratinocytes into

100 100

75

50

37

25

0

75

50

3725

0% N

egat

ive

cultu

res

DIV

7

% N

egat

ive

cultu

res

DIV

13

0 100

Number of plated cells

DIV 0

Aga

r

Sample

1 4 mo Female Trunk

Trunk

Trunk

Trunk

Trunk

20–30 NRAS3

NRAS3

NRAS3

NRAS3

NRAS3

Q61K Q61K

Q61K

Q61R

Q61R

Q61R

NRAS3

NRAS3

NRAS3

NRAS3

NRAS3

Q61K

Q61R

Q61R

Q61R

20–30

10–20

40–60

20–30

Male

Male

Male

Female

5 y

4 mo

7 mo

3 mo

2

3

4

5

Age (months) Sex Tissue site PAS (cm)1 Nevusmutation2

In vitro coloniesmutation2

DIV 13

200 300

302 254

400 500

80

60

40

20

0DIV 3 DIV 5 DIV 7 DIV 9 DIV 13

Dia

met

er o

f col

onie

s (μ

m)

0 100

Number of plated cells

200 300 400 500

Figure 3. In vitro characterization of large congenital melanocytic nevi (lCMN)-initiating cells: percentage of initiating cells and the soft agar assay. (a) lCMN

cells were plated with decreasing concentrations from 500–10 cells per well. Colony numbers were counted on DIV7 and 13 (n¼ 3 specimens). 1/302- and 1/254-

initiating cells were capable of forming colonies on DIV7 and 13, respectively. (b) Soft agar assay assessed on lCMN cells (n¼ 5). Individually seeded lCMN cells

were able to form colonies. Photomicrographs taken on DIV0 and DIV13 (left panel). Scale bar¼ 100mm. The histogram (right panel) represents the diameter of

colonies measured from DIV3 to DIV13. (c) Clinical and mutational characteristics of 5 lCMN and derived colonies. 1PAS, projected adult size; 2pyrosequencing

was used to screen NRAS exon 3 mutations.



Matrigel at a ratio of one melanocyte for two keratinocytesinto the right flank of Rag2� /� mice (n¼ 4 injections fromfour specimens). In parallel, HaCaT cell suspensions wereinjected in the left flank of mice and used as positive controls.Interestingly, 3 months after xenotransplantation, a pigmentedtumor developed in the right flank of two of these four mice(Supplementary Figure S5b online). These tumors mimicked

the morphology of a nevus, with an external layer of well-organized keratinocytes, in which Melan-Aþ and Sox-10þ

cells were found (Figure 5b, white arrows). These melanocyticcells migrated through the multilayered epithelial structure butwere mainly present in its basal layer following a lentiginouspattern (Figure 5b, black arrowheads). Interestingly, somemelanocytes formed nests or theques (Figure 5b, black

Punch Pregraft

Postgraft

HE

2

1

Grafted neoepi. (2)

HE

Grafted nevus (1)

800700600500400

200100

0

107.2%

473.4%

Gra

fted

tum

or v

olum

e in

mm

3

(% o

f pun

ch v

olum

e)

3 Months

Grafted nevus

Graftednevus

Brd

U 3

0′

8 10

8

6

4

2

030d.

Brd

U/S

ox10

+ c

ells

(% o

f tot

al c

ells

)

6

4

2

0

30′ B

rdU

/ M

elan

-A+ c

ells

(% o

f tot

al c

ells

)

Grafted neoepi.

Graftednevus

Graftedneoepi.

7 Months

300

/CMN

Melan-A /DAPI

Melan-A / DAPI / BrdU Sox10 / DAPI / BrdU

Sox10/ DAPI

Grafted nevus Grafted neoepi.

Brd

U 3

0 da

ys

Grafted nevus Grafted neoepi. Grafted nevus Grafted neoepi.

Graftedneoepi.

Figure 4. In vivo xenotransplantation of large congenital melanocytic nevi (lCMN) tissue in immunocompromised mice. Biopsies of lCMN grafted in Rag2� /�

mice were collected after 3 or 7 months (n¼ 8 grafted specimens). (a) 4-fold expansion of the tumors after 7 months. (b) Histology of initial nevus (‘pregraft’) and of

whole-graft tissue 7 months post grafting with the formation of a neoepidermis (‘grafted neoepi.’). Scale bar¼ 100mm, except in postgraft where scale bar¼ 1 mm.

Immunofluorescence stainings of original graft (‘grafted nevus’) and graft neoepidermis (‘graft neo-epi.’): Melan-A (c), Sox10 (d) (Alexa 488-labeled secondary

antibodies), and Brdu/Melan-A (e), Brdu/Sox10 (f) co-labeling 30min or 30 days after BrdU injection, respectively (Cy3-labeled secondary antibody). Melan-Aþ /

BrdUþ or Sox10þ /BrdUþ cells are indicated with arrows. Higher magnifications are shown for each staining. Histograms show the number of double-positive

cells per total cells. Scale bar¼ 100mm.




arrows), a histological feature characteristic of nevus archi-tecture. In these epithelial tumors, cystic contents (Figure 5b,black star), dependent on the keratinocyte activity,were present. In order to confirm the proliferative abilityof these nevocytic cells, immunostaining using anti-Ki67antibody was performed (Supplementary Figure S5c online).No Ki67þ cells were found in Matrigel containing onlymelanocytes. In contrast, Matrigel containing both melano-cytes and keratinocytes displayed numerous Ki67þ cells.Altogether, these results support the postnatal persistence oflCMN cells retaining tumorigenic potential and displayingfeatures of propagating cells when cooperating withkeratinocytes.

DISCUSSIONWe here report evidence for postnatal persistence of clono-genic and tumorigenic nevi cells in CMN. Our immunostain-ing results indicate heterogeneity in lCMN tissue with minorcell populations expressing various stemness markers. Weconfirmed by flow cytometry the expression of ABCB5,CD166, and CD271 in a low number of lCMN cells. Tworecent studies also show that CMN cells express pluripotencymarkers (Kinsler et al., 2013a; Shakhova et al., 2012).However, these results do not give proof of a hierarchybetween cells. Of note, FACS analyses revealed that stem/propagating cell markers such as ABCB5 or CD271 werenever co-expressed on the same cells conversely to whathas been reported in melanoma. In contrast, ABCB5 wasexpressed on differentiated cells expressing Melan-A, asalready reported in melanoma (Frank et al., 2003, 2005;Quintana et al., 2010; Schatton et al., 2010). These data also

indicate that in various types of melanocytic proliferations,cell membrane phenotype may differ from normal melano-cytes’ phenotype.

Proliferative and clonogenic abilities of postnatal lCMNcells were then analyzed in vitro. We were able to demon-strate such features as lCMN cells were able to grow inspheres, could be passaged at least five times, and formedcolonies in soft agar. Interestingly, nevocytic cell proliferationin sphere medium was significantly lower than that oftheir melanoma counterparts. Moreover, nevus spheres tendedto reach a plateau in contrast with melanoma spheres, whichgrew permanently. Such kinetics may indicate that nevus cellshave a limited ability to proliferate, followed by a possible‘‘quiescence or senescence’’ phenomenon. Of note, thesoft agar assays yielded colonies only when optimizedsphere medium was used (Supplementary Figure S3d online).There too, lCMN colony diameter remained unchanged bet-ween DIV7 and DIV13, whereas fully transformed Mel501melanoma cells formed continuously growing colonies.Beyond clonality, growth in soft agar is usually consideredas evidence for transformation. However, normal diploidcells such as articular chondrocytes are also capable of growthin soft agar without any transforming event (Skantze et al.,1985). All these data indicate that lCMN contain cells thathave the ability to grow clonally and give progeny in electiveconditions and that appear to be partially transformed. Indeed,genotyping experiments showed that these cells harboredan NRAS mutation, similar to the one found in the originalnevus. Overall, CMN colonies could be maintained for atleast 84 days in vitro. Similarly, tumor-propagating cellshave been reported in other benign cutaneous proliferations

HE

HE

/CMN Theques

Melan-A / Melan-A /

Sox10 / Sox10 /

DAPI DAPI

DAPI DAPI

Lentiginous hyperplasia

Melan-A / Sox10 /DAPI DAPI

Mel

anoc

ytes

Mel

anoc

ytes

+ H

aCaT

Figure 5. In vivo xenotransplantation of large congenital melanocytic nevi (lCMN) cells in immunocompromised mice. lCMN cell suspensions were grafted into

Rag2� /� mice. (a) No tumor was formed after 7 months; Melan-A and Sox10þ cells persisted in the Matrigel (n¼ 5 injections from five specimens, white arrows).

Scale bar¼ 100mm. (b) lCMN cell suspensions were mixed with HaCaT keratinocytes and injected (n¼ 4 injections from four specimens). After 3 months,

hematoxylin–eosin staining revealed melanocytic theques (black arrows) and a lentiginous melanocytic hyperplasia (black arrowheads) within keratinocyte

cysts (black star). Scale bar¼200mm, except in theques and lentiginous hyperplasia hematoxylin–eosin stainings where scale bar¼ 100mm. Melanocytic cells

were visualized using anti-Melan-A and anti-Sox10 antibodies (Alexa 488 secondary antibodies; white arrows). Higher magnifications are shown for each staining.

Scale bar¼ 50mm.



such as papilloma and hemangioma (Khan et al., 2008; Xuet al., 2011; Lapouge et al., 2012).

Beyond in vitro studies, maintaining properties were studiedin vivo. Grafted biopsies from lCMN had expanded 4-fold inimmunosuppressed mice after 7 months. This increaseresulted from an outgrowth of additional nevus at theperiphery of the graft. These results indicate that a populationof ‘‘committed’’ propagating cells was responsible for theexpansion of a nevus-like structure. The presence of label-retaining human melanocytes was found in the nevus out-growth. Of note, Herlyn et al. grafted adult acquired nevi inmice, and persistence of melanocytes was found only beforeday 88 post grafting (Herlyn et al., 1986), whereas this wasconstant after 7 months in our study. Herlyn’s data togetherwith ours suggest that long-standing tumorigenic cells are aprivilege of lCMN, probably because of differences inmutational profile (Ichii-Nakato et al., 2006). Indeed, NRASmutation is a hallmark of lCMN as opposed to small/intermediate CMN and acquired nevi that frequently harbora BRAF mutation (Charbel et al., 2014).

When seeded in Matrigel, melanoma cells are able to growin immunosuppressed mice even when injected at low or verylow numbers (Quintana et al., 2008; Boiko et al., 2010;Perego et al., 2010). Applying this technique to lCMN cellsuspension disclosed a significant difference: a small minorityof cells from the initial nevi suspensions was able to persistafter several months without evidence of proliferation. Thesecells appeared always as isolated and interspersed cells. Thedifference of behavior in Rag2� /� mice between isolatedmelanocytes (in Matrigel) and nevus grafts prompted us toquestion cooperation between nevi melanocytes andkeratinocytes. When a non-transformed human keratinocytecell line was co-injected with lCMN cell suspensionin Matrigel, a large and solid pigmented tumor developed.These specimens showed a multilayered pigmented epithelialstructure containing proliferating melanocytes organized inmelanocytic nests or in a lentiginous pattern. Therefore, lCMNcontained cells capable of reconstituting a nevus-like structurewhen mixed with keratinocytes. These results are inaccordance with an older study. The authors grafted onathymic mice a CMN coculture of epidermal cells, includingmelanocytes and keratinocytes, together with a fibroblastmatrix. Up to 34 weeks, melanocytes were present only in a‘‘darkened’’ epidermis without true nests. However, after thisdelay, the dermis presented many round multinucleated cellscorresponding to CMN reconstitution (Cooper et al., 1999).These results suggested that melanocytic nests originated inthe epidermis and then migrated below. Our model of lCMNsuspensions mixed with keratinocytes has shown thatkeratinocytes have a major role in melanocytic behavior innevi, allowing the development and maintenance of melano-cytic nests. Mediators of this cross talk remain to be deter-mined. Keratinocytes control melanocyte migratory propertiesand dermal invasion potential by modulating E-cadherinexpression (Gontier et al., 2002, 2004). Moreover, keratino-cytes have a role in maintaining the melanocyte reservoir.The melanocyte stem cells reside in the hair follicle bulge–subbulge area, the lower permanent portion of the hair

follicle, where they serve as a melanocyte reservoir for theskin and hair pigmentation. There, they directly adhere to hairfollicle stem cells (Nishimura, 2011). Forced expression ofstem cell factor /KITL, which is constitutively expressed in thehuman epidermis, in the basal epidermis of mice using theKrt14 promoter allowed survival of melanocytes in mouseepidermis (Kunisada et al., 1998). Certain inflammatory factorssuch as leukotriene C4 also induce chronic growth stimulationof human adult melanocytes (Medrano et al., 1993). On theother hand, it has been shown that different fibroblast-secretedmelanogenic peptides could modulate melanocyte numbers inmonolayer cultures (Cario-Andre et al., 2006; Salducci et al.,2014). In vitro results on monolayers and skin reconstructs,using growth factors and conditioned medium suggested thatmurine fibroblasts secrete soluble factors that can stimulatenormal human melanocytes. Medium from human fibroblastsstressed by medium starvation could increase normalmelanocyte proliferation in vitro. Moreover, among thesegrowth factors, both human and murine stem cell factorinduced an increase in melanocyte numbers (Cario-Andreet al., 2006). Therefore, the role of fibroblasts as well as ofother epithelial and hematopoietic cells as potential promotersof CMN tumor formation should also be investigated.

In conclusion, we demonstrated that lCMN contain a sub-population of cells able to persist after birth, self-renew, andinduce a nevus-like structure when cooperating with kerati-nocytes. This could explain nevus secondary resurgences,which have been described by surgeons following nevusexcision, as well as the natural expansion of nevi during thechild growth. On the other hand, the genomic analysis ofthese cells is in line with our previous report on the role ofNRAS mutation as driver mutation in these benign melanocy-tic proliferations (Charbel et al., 2014). Future studies shouldfocus on better characterizing these lCMN propagating cells.Finally, a clinical valorization is expected, as the isolated cellscould also serve as tools for drug screening in CMN and/ormelanoma.

MATERIALS AND METHODSDetails can be found in Supplementary Materials and Methods online.

Human tissue samples

The study was approved by the institutional independent ethics

committee (Comite de Protection des Personnes Ile de France Paris

V) and complied with the Declaration of Helsinki principles. Patients

and patients’ guardians gave their written informed consent prior to

participation. Human lCMN samples were obtained from the maxillo-

facial department of Armand Trousseau hospital. Seventeen patients

with compound lCMN whose ages varied between 2 months and 5

years were included prospectively in the study on the basis of

adequate clinical data. Classification was done using the projected

adult size of the largest lesion (Krengel et al., 2013). Four SSM from

Tenon Hospital were used as positive controls.

Immunohistochemistry

Sections of 5-mm thickness were labeled with the following anti-

bodies: anti-human MITF (1:250), anti-human Nestin (1:100), anti-

human ABCB5 (1:50), anti-human KI67 (1:100), anti-human Oct4




(1:100), anti-human Melan-A (1:100), and anti-human Sox10

(1:100).

In vitro assays

Fresh nevocytic cell suspension deriving from patients were seeded,

passaged, and/or plated in soft agar at a density of 50,000 cells per

well in DMEM/F12 supplemented with EGF (20 ng ml� 1), basic

fibroblast growth factor (40 ng ml� 1), B27 supplement, and insulin

(5mg ml� 1). Spheres were counted and the diameter was measured.

The number and frequency of initiating cells were determined using

limiting dilution analysis (Liu et al., 2013). For immunocytochemistry,

sphere colonies were stained with anti-Melan-A (1:100), anti-Oct4

(1:50), and anti-Nestin (1:100) antibodies.

DNA extraction and genotyping

Samples were screened for the c.181C4A, p.Q61K or c.182A4G,

p.Q61R NRAS mutation. DNA was extracted using standard techni-

ques and pyrosequencing performed as previously described (Charbel

et al., 2014).

In vivo assays

All animal experiments were performed according to approved

experimental protocols following European Community Council

guidelines. For lCMN tissue transplantations, standardized specimens

were xenografted in brown adipose tissue in Rag2� /� mice. For

human nevocytic cell transplantations, CD31�CD45� sorted cells

were resuspended in 25% of Matrigel either alone or mixed with

human keratinocyte cell line HaCat at a ratio of 1:2 and injected

subcutaneously into the flank of Rag2/� mice. Xenografts were

collected 3 months or 7 months after grafting. Xenografted mice

were given i.p injections of BrdU prior to euthanasia. BrdU labeling

was detected using anti-BrdU antibody (1:50).

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSWe are grateful to all the patients and families who participated in this study.We thank Delphine Muller and Tatiana Ledent (Saint-Antoine Research CenterAnimal Facility), Anne-Marie Faussat (IFR65), and Coralie Guerin (InsermU970, PARCC, European Georges Pompidou Hospital, Paris, France) fortechnical support and Francois Delhommeau, Heather Etchevers, ThierryJaffredo, and Raphael Scharfmann for helpful discussions. This work wasfunded by the Societe Francaise de Dermatologie and the Societe deRecherche Dermatologique. CC was supported by the Fondation ReneTouraine, Lebanese CNRS, and AREMPH.

SUPPLEMENTARY MATERIAL

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

REFERENCES

Ackermann J, Frutschi M, Kaloulis K et al. (2005) Metastasizing melanomaformation caused by expression of activated N-RasQ61K on an INK4a-deficient background. Cancer Res 65:4005–11

Barnhill RL, Chastain MA, Jerdan MS et al. (2010) Angiotropic neonatalcongenital melanocytic nevus: how extravascular migration of melano-cytes may explain the development of congenital nevi. Am J Dermato-pathol 32:495–9

Bauer J, Curtin JA, Pinkel D et al. (2007) Congenital melanocytic nevi frequentlyharbor NRAS mutations but no BRAF mutations. J Invest Dermatol 127:179–82

Boiko AD, Razorenova OV, van de Rijn M et al. (2010) Human melanoma-initiating cells express neural crest nerve growth factor receptor CD271.Nature 466:133–7

Cario-Andre M, Pain C, Gauthier Y et al. (2006) In vivo and in vitro evidence ofdermal fibroblasts influence on human epidermal pigmentation. PigmentCell Res 19:434–42

Castilla EE, da Graca DM, Orioli-Parreiras IM (1981) Epidemiology ofcongenital pigmented naevi: I. Incidence rates and relative frequencies.Br J Dermatol 104:307–15

Charbel C, Fontaine R, Malouf G et al. (2014) NRAS mutation is the solerecurrent somatic mutation in large congenital melanocytic nevi. J InvestDermatol 134:1067–74

Cheli Y, Giuliano S, Botton T et al. (2011) Mitf is the key molecular switchbetween mouse or human melanoma initiating cells and their differen-tiated progeny. Oncogene 30:2307–18

Cooper ML, Spielvogel RL, Hansbrough JF et al. (1999) Reconstitution of the histo-logic characteristics of a giant congenital nevomelanocytic nevus employingthe athymic mouse and a cultured skin substitute. J Invest Dermatol 97:649–58

Cramer SF, Fesyuk A (2012) On the development of neurocutaneous units–implications for the histogenesis of congenital, acquired, and dysplasticnevi. Am J Dermatopathol 34:60–81

de Wijn RS, Zaal LH, Hennekam RC et al. (2010) Familial clustering of giantcongenital melanocytic nevi. J Plast Reconstr Aesthet Surg 63:906–13

Fang D, Nguyen TK, Leishear K et al. (2005) A tumorigenic subpopulation withstem cell properties in melanomas. Cancer Res 65:9328–37

Frank NY, Pendse SS, Lapchak PH et al. (2003) Regulation of progenitor cellfusion by ABCB5 P-glycoprotein, a novel human ATP-binding cassettetransporter. J Biol Chem 278:47156–65

Frank NY, Margaryan A, Huang Y et al. (2005) ABCB5-mediated doxorubicintransport and chemoresistance in human malignant melanoma. CancerRes 65:4320–33

Gontier E, Cario-Andre M, Lepreux S et al. (2002) Dermal nevus cells fromcongenital nevi cannot penetrate the dermis in skin reconstructs. PigmentCell Res 15:41–8

Gontier E, Cario-Andre M, Vergnes P et al. (2004) The role of E-cadherinin nevogenesis: an experimental study using epidermal reconstructs.Exp Dermatol 13:326–31

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

Herlyn D, Elder DE, Bondi E et al. (1986) Human cutaneous nevi transplantedonto nude mice: a model for the study of the lesional steps in tumorprogression. Cancer Res 46:1339–43

Ichii-Nakato N, Takata M, Takayanagi S et al. (2006) High frequency ofBRAFV600E mutation in acquired nevi and small congenital nevi, but lowfrequency of mutation in medium-sized congenital nevi. J Invest Dermatol126:2111–8

Jiao Z, Zhang ZG, Hornyak TJ et al. (2006) Dopachrome tautomerase (Dct)regulates neural progenitor cell proliferation. Dev Biol 296:396–408

Khan ZA, Boscolo E, Picard A et al. (2008) Multipotential stem cellsrecapitulate human infantile hemangioma in immunodeficient mice.J Clin Invest 118:2592–9

Kinsler VA, Anderson G, Latimer B et al. (2013a) Immunohistochemical andultrastructural features of congenital melanocytic naevus cells support astem cell phenotype. Br J Dermatol 169:374–83

Kinsler VA, Thomas AC, Ishida M et al. (2013b) Multiple congenitalmelanocytic nevi and neurocutaneous melanosis are caused by post-zygotic mutations in codon 61 of NRAS. J Invest Dermatol 133:2229–36

Krengel S (2005) Nevogenesis-new thoughts regarding a classical problem.Am J Dermatopatho 27:456–65

Krengel S, Hauschild A, Schafer T (2006) Melanoma risk in congenitalmelanocytic naevi: a systematic review. Br J Dermatol 155:1–8

Krengel S, Scope A, Dusza SW et al. (2013) New recommendations for thecategorization of cutaneous features of congenital melanocytic nevi. J AmAcad Dermatol 68:441–51



http://www.nature.com/jid

http://www.nature.com/jid

Kunisada T, Lu SZ, Yoshida H et al. (1998) Murine cutaneous mastocytosisand epidermal melanocytosis induced by keratinocyte expression oftransgenic stem cell factor. J Exp Med 187:1565–73

Lapouge G, Beck B, Nassar D et al. (2012) Skin squamous cell carcinomapropagating cells increase with tumour progression and invasiveness.EMBO J 31:4563–75

Leigh CJ, Palechek PL, Knutson JR et al. (1996) CD44 expression in benign andmalignant nevomelanocytic lesions. Hum Pathol 27:1288–94

Liu M, Miller CL, Eaves CJ (2013) Human long-term culture initiating cellassay. Methods Mol Biol 946:241–56

Mackenzie MA, Jordan SA, Budd PS et al. (1997) Activation of the receptortyrosine kinase Kit is required for the proliferation of melanoblasts in themouse embryo. Dev Biol 192:99–107

Marghoob AA, Schoenbach SP, Kopf AW et al. (1996) Large congenitalmelanocytic nevi and the risk for the development of malignantmelanoma. A prospective study. Arch Dermatol 132:170–5

Medrano EE, Farooqui JZ, Boissy RE et al. (1993) Chronic growth stimulation ofhuman adult melanocytes by inflammatory mediators in vitro: implica-tions for nevus formation and initial steps in melanocyte oncogenesis.Proc Natl Acad Sci USA 90:1790–4

Nishikawa-Torikai S, Osawa M, Nishikawa S (2011) Functional characteriza-tion of melanocyte stem cells in hair follicles. J Invest Dermatol131:2358–67

Nishimura EK (2011) Melanocyte stem cells: a melanocyte reservoir in hairfollicles for hair and skin pigmentation. Pigment Cell Melanoma Res24:401–10

Papp T, Schipper H, Kumar K et al. (2005) Mutational analysis of the BRAFgene in human congenital and dysplastic melanocytic naevi. MelanomaRes 15:401–7

Pastrana E, Silva-Vargas V, Doetsch F (2011) Eyes wide open: a critical reviewof sphere-formation as an assay for stem cells. Cell Stem Cell 8:486–98

Perego M, Tortoreto M, Tragni G et al. (2010) Heterogeneous phenotype ofhuman melanoma cells with in vitro and in vivo features of tumor-initiating cells. J Invest Dermatol 130:1877–86

Quintana E, Shackleton M, Sabel MS et al. (2008) Efficient tumour formationby single human melanoma cells. Nature 456:593–8

Quintana E, Shackleton M, Foster HR et al. (2010) Phenotypic heterogeneityamong tumorigenic melanoma cells from patients that is reversible andnot hierarchically organized. Cancer Cell 18:510–23

Ruiz-Maldonado R (2004) Measuring congenital melanocytic nevi. PediatrDermatol 21:178–9

Salducci M, Andre N, Guere C et al. (2014) Factors secreted by irradiated agedfibroblasts induce solar lentigo in pigmented reconstructed epidermis.Pigment Cell Melanoma Res 27:502–4

Schatton T, Murphy GF, Frank NY et al. (2008) Identification of cells initiatinghuman melanomas. Nature 451:345–9

Schatton T, Schutte U, Frank NY et al. (2010) Modulation of T-cell activationby malignant melanoma initiating cells. Cancer Res 70:697–708

Shakhova O, Zingg D, Schaefer SM et al. (2012) Sox10 promotes the formationand maintenance of giant congenital naevi and melanoma. Nat Cell Biol14:882–90

Sharma B, Manglik V, Elias EG (2010) Immuno-expression of humanmelanoma stem cell markers in tissues at different stages of the disease.J Surg Res 163:e11–5

Skantze KA, Brinckerhoff CE, Collier JP (1985) Use of agarose culture tomeasure the effect of transforming growth factor beta and epidermalgrowth factor on rabbit articular chondrocytes. Cancer Res 45:4416–21

Sommer L (2011) Generation of melanocytes from neural crest cells. PigmentCell Melanoma Res 24:411–21

Tai MH, Chang CC, Kiupel M et al. (2005) Oct4 expression in adult humanstem cells: evidence in support of the stem cell theory of carcinogenesis.Carcinogenesis 26:495–502

Tanimura S, Tadokoro Y, Inomata K et al. (2011) Hair follicle stem cellsprovide a functional niche for melanocyte stem cells. Cell Stem Cell8:177–87

Xu D, O TM, Shartava A et al. (2011) Isolation, characterization, and in vitropropagation of infantile hemangioma stem cells and an in vivo mousemodel. J Hematol Oncol 4:54




clonogenic cell subpopulations maintain congenital melanocytic … · 2017. 1. 31. · oct4, an...

Documents