Modeling epidermal melanoma in mice: moving into new realms but with unexpected complexities.

Full text

Page 1

their relationship with quorum-sensing sig-
nals in Gram-negative bacteria. Antimicrob
Agents Chemother 55:1325–31
Chun CK, Ozer EA, Welsh MJ et al. (2004) Inacti-
vation of a pseudomonas aeruginosa quorum-
sensing signal by human airway epithelia. Proc
Natl Acad Sci USA 101:3587–90
Galloway WR, Hodgkinson JT, Bowden SD et al.
(2011) Quorum sensing in gram-negative
bacteria: small-molecule modulation of AHL
and AI-2 quorum sensing pathways. Chem Rev
111:28–67
Kravchenko VV, Kaufmann GF, Mathison JC et al.
(2008) Modulation of gene expression via
disruption of NF-kappaB signaling by a bac-
terial small molecule. Science 321:259–63
Mackness B, Beltran-Debon R, Aragones G et al.
(2010)Human tissue distributionof
paraoxonases 1 and 2 mRNA. IUBMB Life
62:480–2
Pearson JP, Pesci EC, Iglewski BH (1997) Roles
of pseudomonas aeruginosa las and rhl
quorum-sensing systems in control of elas-
tase and rhamnolipid biosynthesis genes.
J Bacteriol 179:5756–67
Precourt LP, Amre D, Denis MC et al. (2011) The
three-gene paraoxonase family: physiologic roles,
actionsand regulation.Atherosclerosis 214:20–36
Steidle A, Sigl K, Schuhegger R et al. (2001)
Visualization of N-acylhomoserine lactone-
mediated cell-cell communication between
bacteria colonizing the tomato rhizosphere.
Appl Environ Microbiol 67:5761–70
Stoltz DA, Ozer EA, Ng CJ et al. (2007)
Paraoxonase-2 deficiency enhances pseudo-
monas aeruginosa quorum sensing in murine
tracheal epithelia. Am J Physiol Lung Cell
Mol Physiol 292:L852–60
Stoltz DA, Ozer EA, Taft PJ et al. (2008)
Drosophila are protected from pseudomonas
aeruginosa lethality by transgenic expression
of paraoxonase-1. J Clin Invest 118:3123–31
Teiber JF, Horke S, Haines DC et al. (2008)
Dominant role of paraoxonases in inactiva-
tion of the pseudomonas aeruginosa quorum-
sensing signal N-(3-oxododecanoyl)-L-homo-
serine lactone. Infect Immun 76:2512–9
Wang EW, Jung JY, Pashia ME et al. (2005)
Otopathogenic Pseudomonas
strains as competent biofilm formers. Arch
Otolaryngol Head Neck Surg 131:983–9
aeruginosa
Zhang LH, Dong YH (2004) Quorum sensing and
signal interference: diverse implications. Mol
Microbiol 53:1563–71
Modeling Epidermal Melanoma in Mice: Moving into
New Realms but with Unexpected Complexities
Journal of Investigative Dermatology (2012) 132, 2299–2302; doi:10.1038/jid.2012.200; published online 14 June 2012
TO THE EDITOR
Histological localization of human and
murine melanomas is very different.
Most human melanomas are epidermal,
invading the dermis during progression,
whereas murine lesions are usually
exclusively dermal. We recently took
a mouse model of dermal melanoma
(Arf?/?::Tyr-NRASQ61K), in which mel-
anocytes are rarely seen in the epider-
mis of normal or lesional skin, and
combined it with an animal carrying
epidermal melanocytes throughout life
due to the overexpression of the Kit
receptor ligand in keratinocytes (K14-
Kitl).Arf?/?::Tyr-NRASQ61K::K14-Kitl
mice developed slow-growing plaques
exhibiting a pagetoid growth pattern
typical of human superficial spreading
melanoma (SSM), and our experiments
suggested a probable epidermal origin
(Walker et al., 2011a). In contrast, Rae
et al. (2012) report the development
of rapidly growing dermal melanomas
inBrafV600E::K14-Kitl
show no epidermal involvement. As
BRAFV600Eis the most common muta-
tion in SSM, modeling BrafV600E-driven
mice, which
epidermal melanomas in mice would
be an important advance. We propose
that this failure to induce epidermal
melanoma is unlikely to be a particular
feature of BrafV600E-driven murine mel-
anoma, and that the contrasting results
illuminate critical aspects of experi-
mental design to be considered when
making mechanistic inferences about
particular engineered mutations. Rae
et al., outlined these differences: (1)
strain background, (2) use of NRASQ61K
versus BrafV600E, (3) we used neonatal
UV radiation (UVR), (4) we used an
additional Arf mutation, and (5) differ-
ent timing of mutation induction. Given
the importance of generating epidermal
melanoma in mice for their utility as
preclinical models, we have examined
these differences in detail.
Strain background can influence
penetrance and other aspects of neo-
plasia, but as epidermal melanocytes
are present in both studies it alone
seems unlikely to explain the contrast-
ing propensities for epidermal involve-
ment. In addition, we know of no
evidence in the melanoma field that
would lead us to suspect that NRASQ61K
and BRAFV600Ewould behave differ-
ently in this context. In a spontaneous
melanoma model (Cdk4R24C/R24C::Tyr-
NRASQ61K::K14-Kitl),
many atypical epidermal melanocytes
(Figure 1a) and the mice developed
plaque-like lesions with atypical mela-
nocytes scattered
epidermis (Walker et al., 2011b and
Figure 1b). Hence, neonatal UVR is
not obligatory for epidermal mela-
noma genesis in K14-Kitl mice. We
also examined the consequence of
NRASQ61Kexpression on melanocyte
proliferation in K14-Kitl mice without
accompanying Arf or Cdk4 mutation. In
Tyr-NRASQ61K::K14-Kitl neonates, we
observed a striking increase in epider-
mal and upper follicular melanocytes
compared with mice carrying Tyr-
NrasQ61Kor K14-Kitl alone (Figure 1c).
In adult Tyr-NRASQ61K::K14-Kitl skin,
melanocytes were frequently almost at
confluence in the basal epidermis,
exhibiting variation in nuclear size
and shape, nesting, and suprabasal
localization (Figure 2a and b) (we did
not perform long term studies on
melanoma development for this geno-
type). NRASQ61K
we observed
throughout the
expression greatly
See related commentary on pg 2135
Abbreviations: mGluR1, metabotropic glutamate receptor-1; SSM, superficial spreading melanoma;
UVR, UV radiation
www.jidonline.org 2299
HY Handoko et al.
Modeling Epidermal Melanoma in Mice

Page 2

increased melanocyte density in the
K14-Kitl epidermis (Figure 2c), whereas
Cdk4R24Cmutation (or p19Arf loss) did
not, although it greatly exacerbated
neoplastic progression. Notwithstand-
ing the lack of epidermal melanoma,
it is as yet unclear whether BrafV600E
cooperates with Kitl to evoke epidermal
melanocyte proliferation, although it
does increase epidermal pigmentation
(Rae et al., 2012).
Mice overexpressing both the meta-
botropic glutamate receptor-1 (mGluR1)
and K14-Kitl also develop exclusively
dermal melanoma (Abdel-Daim et al.,
2010). Of note, BrafV600Eand mGluR1
were activated in adults, whereas in
our study NRASQ61Kis active during
embryonic and neonatal development.
Oncogene induction during embryo-
genesis can adversely effect develop-
ment and predispose to tumorigenesis.
With melanocytes migrating through
the epidermis and developing hair
follicles, the neonatal period may be
particularly important in murine mela-
noma, explaining why neonatal but not
adult UVR exposure induces melanoma
(Noonan et al., 2001). In the Mt-Ret
mice (Kato et al., 2004) antibody-
mediated depletion of neonatal mela-
nocytes (except
stem cells which at the first hair cycle
regenerate follicular
greatly diminishes spontaneous mela-
noma development. We tracked K14-
Kitl neonatal melanocytes and found
that they remain there into adulthood
(Walker et al., 2011a). Hence, the
presence of engineered or induced
mutations in neonatal epidermal mela-
nocytes could potentially be important
for epidermal lesion development in
K14-Kitl mice. If so, activation of
BrafV600E
in K14-Kitl neonates may
bulgemelanocyte
melanocytes)
Trp1/K14/DAPI
Trp1/K14/DAPI
Kit/DAPI Kit/DAPIKit/DAPI Kit/DAPI
Wild type
Tyr-NrasK14-Kitl
Tyr-Nras::K14-Kitl
a
b
c
Figure 1. K14-Kitl mice drive proliferation of epidermal melanocytes in the context of transgenic melanocyte-specific NrasQ61Kmutation.
(a) Immunofluorescence (IF) and hematoxylin and eosin (H&E) staining of non-lesion skin from Cdk4R24C/R24C::Tyr-Nras::K14-Kitl mice. IF staining for Trp1 (red)
and K14 (green) fully described in Walker et al., 2011a, and accompanying supplementary methods. Red staining melanocytes can be seen mainly at the basal layer,
but also suprabasally. The epidermis is green. In the H&E panel, we stained bleached formalin-fixed paraffin-embedded sections. Yellow arrows indicate melanocytes
(cells with a clear area surrounding a basophilic nuclei) in both the basal and suprabasal epidermis. There are also melanocytes within the papillary dermis.
Strain background mixed C57BL/6::FVB, but at least four crosses down C57BL/6. Age of animal: 119 days. (b) Representative IF and H&E staining of a flat
slow-growing lesion developing on a Cdk4R24C/R24C::Tyr-Nras::K14-Kitl mouse. In the H&E panel, yellow arrows indicate melanocytes of varying nuclear size
and shape, and sometimes in small nests, within the epidermis. Strain background mixed C57BL/6::FVB, but at least four crosses down C57BL/6. Age of animal:
119 days. (c) IF staining of skin from neonatal mice at postnatal day 6 (P6). Primary antibody is anti-Kit (ACK2 rat monoclonal). Dotted yellow line denotes the
epidermal basal layer. Genotypes, from left to right are as follows: wild type, Tyr-NrasQ61K, K14-Kitl, and Tyr-NrasQ61K::K14-Kitl. Note the dramatic increase in
melanocytes in the epidermis and upper hair follicle resulting from the cooperation of NrasQ61Kand keratinocyte Kitl in neonatal mouse skin. Large round cells in
the dermis are mast cells, which also stain for Kit but can be clearly differentiated from melanocytes. Strain background C57BL/6. Bars¼20mM.
2300 Journal of Investigative Dermatology (2012), Volume 132
HY Handoko et al.
Modeling Epidermal Melanoma in Mice

Page 3

produce lesions with some degree of
epidermal involvement. In the study of
Dankort et al. (2009) neonatal BrafV600E
activation induces dermal nevi that
frequently exhibit epidermal involve-
ment without K14-Kitl, suggesting that
BrafV600Eis not absolutely incompatible
with the presence of atypical epidermal
melanocytes in mice.
The means of engineering muta-
tions can also influence phenotypic
outcome. In inducible Braf models,
themutation is
endogenous locus in adult mice by
Cre-mediated recombination. In ‘‘trans-
genics’’ the endogenous locus is un-
affected but multiple transgene copies
integrate randomly in the genome and
are expressed throughout embryonic
development. Despite this, BrafV600E
(Dhomen et al., 2009; Rae et al.,
2012) or mGluR1 (Abdel-Daim et al.,
2010) activation in adults induces
melanoma rapidly, more rapidly than,
forinstance,the
NRASQ61K(Ackermann et al., 2005)
we used. Transgenic Tyr-BRAFV600E
age of onset is very dependent upon
transgene expression level (Goel et al.,
2009). Interestingly, BrafV600Eexpres-
sion results in perinatal lethality when
insertedintothe
transgenic Tyr-
recombined into the endogenous locus
during embryonic development (Dan-
kort et al., 2009; Dhomen et al., 2010),
whereas ‘‘transgenic’’
BRAFV600Eexpression does not (Goel
et al., 2009). As the latter were gener-
ated using classical pronuclear injec-
tion, it is theoretically possible that
embryonic lethality in some lines could
have been missed without a strict
analysis; or perhaps, the retention of
endogenous wild-type Braf alleles in
the transgenic enables embryonic de-
velopment. There are other potentially
importantdifferences.
genic overexpresses a single human
BRAFV600Ecomplementary DNA driven
by the exogenous tyrosinase gene pro-
moter. In contrast, the inducible models
use the endogenous locus that exists in
several differentially spliced isoforms
using different transcriptional start sites.
Hence, the transgenic and inducible
BrafV600E
genes are regulated very
differently.Even
models can give different results. One
inducible BrafV600E::Tyr-CreER
develops only nevi (Dankort et al.,
2009) while the model used by Rae
et al. (2012) develops aggressive mela-
noma. These differences may be due to
embryonic
Thetrans-
ostensiblysimilar
model
different genetic backgrounds and/or
different Tyr-CreERstrains used by the
respective groups. It is conceivable that
subtle genetic engineering design fea-
tures may also be relevant in epidermal
melanoma induction.
Notably, in our epidermal melanoma
model, we also see some dermal melano-
cyte proliferation (Walker et al., 2011a),
which is to be expected as NRASQ61K
and Kitl are expressed in both epidermal
and follicular melanocytes and keratino-
cytes, respectively. Presumably, in all
mouse models oncogene activation per-
turbs melanocytes in various parts of the
hair follicle, which move into the dermis
where they expand and form dermal
nevi and occasionally melanomas. This
may be exacerbated by keratinocyte
Kitl—possibly this is why its expression
exacerbates BrafV600E-driven dermal mel-
anoma genesis, rather than a direct effect
of Kitl on ectopically located dermal
melanocytes. Unfortunately, we cannot
explain why epidermal melanocytes are
not transformed,
complications, particularly involving the
timing and means of oncogene activa-
tion, may be at play. These must be
understood to enable appropriate experi-
mental design to model SSM in mice.
although technical
Adult K14-Kitl skin
Kit
Kit
Trp1
Trp1
Adult K14-Kitl::Tyr-Nras skin
a
b
25
20
15
5
0
Kitl (T)
Nras::Kitl (T)
Nras::Kitl::Cdk4 (T)
Nras::Cdk4 (T)
(H)(H) (H)
(H)
Epidermal melanocyte count
10
c
Figure 2. Genotype dictates melanocyte density within the adult murine epidermis. (a) Immunofluorescence (IF) staining for Kit and Trp1 of dorsal skin
from an 8-week-old K14-Kitl mouse, with corresponding hematoxylin and eosin (H&E) staining on the right panel. Note epidermal melanocytes along the basal
layer (indicated by arrows). Mice in (a) and (b) are 12-week-old littermates. Strain background C57BL/6. (b) IF staining for Kit (green) and Trp1 (red) of dorsal
skin from a Tyr-NrasQ61K::K14-Kitl mouse with H&E staining on the right panel. Arrows point to epidermal melanocytes, which are mainly at the basal layer,
but can also be seen at all levels of the epidermis and in clusters (dotted arrow). (c) Plot of the average number of epidermal melanocytes per field (y axis) in
adult mice. Counts were based on counting Trp1-staining cells (T), and from standard melanocyte morphology in H&E sections (H). For all counts at least
10 sections from at least two mice per genotype were assessed. All animals were 12 to 16 weeks old. Strain background C57BL/6, except for animals carrying
Cdk4R24Cwhich are mixed C57BL/6::FVB, but at least four crosses down C57BL/6. Bar¼20mM.
www.jidonline.org 2301
HY Handoko et al.
Modeling Epidermal Melanoma in Mice

Page 4

CONFLICT OF INTEREST
The authors state no conflict of interest.
ACKNOWLEDGMENTS
We thank Shin-ichi Nishikawa, Takahiro Kunisada,
and Vince Hearing for the antibodies. We are
grateful to the reviewers of the manuscript for
providing additional ideas that we have included.
This work was funded by the NH&MRC of Australia
and the Cancer Council of Queensland. GJW is the
recipient of a Senior Research Fellowship from the
Cancer Council of Queensland. NFB was supported
in part by NIH grant 1P30AR057212.
Herlina Y. Handoko1, Neil F. Box2and
Graeme J. Walker1
1Skin Carcinogenesis Laboratory, Queensland
Institute of Medical Research, Herston,
Queensland, Australia and2Charles C. Gates
Center for Regenerative Medicine and Stem
Cell Biology, University of Colorado Denver,
Denver, Colorado, USA
E-mail: Graeme.Walker@qimr.edu.au
REFERENCES
Abdel-Daim M, Funasaka Y, Komoto M et al (2010)
Pharmacogenomics of metabotropic glutamate
receptor subtype 1 and in vivo malignant
melanoma formation. J Dermatol 37:635–46
Ackermann J, Frutschi M, Kaloulis K et al (2005)
Metastasizing melanoma formation caused by
expression of activated N-RasQ61Kon an INK4a-
deficient background. Cancer Res 65:4005–11
Dankort DP, Curley D, Cartlidge RA et al (2009)
Braf(V600E) cooperates with Pten loss to induce
metastatic melanoma. Nat Genet 41:544–52
Dhomen N, Reis-Filho JS, da Rocha Dias S et al
(2009) Oncogenic Braf induces melanocyte
senescence and melanoma in mice. Cancer
Cell 15:294–303
Dhomen N, Da Rocha Dias S, Hayward R et al
(2010) Inducible expression of (V600E)Braf
usingtyrosinase-driven
results in embryonic lethality. Pigment Cell
Melanoma Res 23:112–20
Cre recombinase
Goel VK, Ibrahim N, Jiang G et al (2009)
Melanocytic nevus-like hyperplasia and mel-
anoma in transgenic BRAFV600Emice. Onco-
gene 28:2289–98
Kato M, Takeda K, Kawamoto Y et al (2004) c-Kit-
targeting immunotherapy
melanoma in a mouse model. Cancer Res
64:801–6
forhereditary
Noonan FP, Recio JA, Takayama H et al (2001)
Neonatal sunburn and melanoma in mice.
Nature 413:271–22
Rae J, Viros A, Hayward R et al (2012)V600EBraf::
Tyr-CreERT2::K14-Kitl mice do not develop
superficial spreading-like melanoma: kerati-
nocyte Kit ligand is insufficient to ‘‘trans-
locate’’
epidermis. J Invest Dermatol 132:488–91
V600EBraf-driven melanoma to the
WalkerGJ,Soyer HP,Handoko HYetal
(2011a) Superficial spreading-like melanoma
in Arf?/?::Tyr-NrasQ61K::K14-Kitl mice: kerati-
nocyte Kit ligand expression sufficient to
‘‘translocate’’ melanomas from dermis to
epidermis. J Invest Dermatol 131:1384–7
Walker GJ, Soyer HP, Terzian T et al (2011b)
Modeling melanoma in mice. Pigment Cell
Melanoma Res 24:1158–76
The Glycan-Binding Protein Galectin-9 Has Direct
Apoptotic Activity toward Melanoma Cells
Journal of Investigative Dermatology (2012) 132, 2302–2305; doi:10.1038/jid.2012.133; published online 10 May 2012
TO THE EDITOR
In recent years, a regulatory role has
emerged for the glycan-binding protein
galectin-9 (Gal-9) in normal physiology
and pathology (reviewed by Wiersma
et al. (2011). In melanoma and other
malignancies, the available data suggest
that Gal-9 has a tumor-suppressor func-
tion, with loss of Gal-9 being closely
associated with metastatic progression
(Kageshita et al., 2002; Irie et al., 2005;
Yamauchi et al., 2006; Liang et al.,
2008). In particular, melanocytic nevi
and primary melanoma lesions highly
express Gal-9, whereas metastatic mel-
anoma lesions have no or minimal
expression of Gal-9 (Kageshita et al.,
2002). Furthermore, ectopic expression
of Gal-9 abrogates the formation of
metastases by Gal-9-deficient B16F10
murine melanoma cells (Nobumoto
et al., 2008). Similarly, treatment of
Gal-9-deficient B16F10 cells with a
recombinant form of Gal-9, designated
Gal-9(0), strongly reduced metastasis
formation (Nobumoto et al., 2008).
This anti-metastatic activity of Gal-9(0)
on B16F10 has been attributed mainly
to inhibition of melanoma cell adhesion
to endothelial cells and/or extracellular
matrix components, such as collagen
type I (collagen-I; Nobumoto et al.,
2008). The data presented in the current
letter suggest that within the 1-h time
frame of adhesion-type assays, treat-
ment with Gal-9(0) triggers early apop-
totic cellular changes. In line with earlier
findings, Gal-9(0) inhibits the adhesion
of B16F10 and 7 human melanoma cell
lines to collagen-I (Figure 1a). However,
the morphology of Gal-9(0)-treated cells
that had adhered to collagen-I-coated
wells resembled that of dying cells
(Figure 1b; illustrated for B16F10).
Subsequent analysis of this melanoma
cell line panel, as well as primary patient–-
derived malignant melanoma cells for
the early apoptotic marker phosphatidyl
serine (PS), revealed that treatment with
Gal-9(0) induced B90% cell death with-
in the time frame used in the adhesion
assay (Figure 1c). Early apoptotic PS
exposure was followed by apoptotic
cell death within 24h of treatment, as
evidenced by loss of viability (Supple-
mentary Figure S1a online), the presence
of late apoptotic Annexin-V/PI double-
positive cells (Supplementary Figure S1b
andc online),
in DNA fragmentation (Supplementary
Figure S1d online). In primary human
melanocytes, Gal-9(0) also triggered
PS exposure, albeit to a lesser extent
(Figure 1c; B55%). More importantly,
the viability of these normal cells was
not negatively affected after 24h (Sup-
plementary Figure S1a online). Thus,
Gal-9(0) induces rapid apoptotic cell
death in melanoma cells, but not in
andan increase
Abbreviations: collagen-I, collagen type I; Gal-9, galectin-9; PS, phosphatidyl serine
2302 Journal of Investigative Dermatology (2012), Volume 132
VR Wiersma et al.
Melanoma Cell Death by Gal-9