skintumorigenesisstimulatedbyrafinhibitorsreliesupon raf … · 2013-11-22 · raf functions that...

Molecular and Cellular Pathobiology

Skin Tumorigenesis Stimulated by Raf Inhibitors Relies UponRaf Functions That Are Dependent and Independent of ERK

Eszter Doma, Christian Rupp, Andrea Varga, Florian Kern, Bettina Riegler, and Manuela Baccarini

AbstractRAF inhibitors achieve unprecedented but mainly transient clinical responses in patients with melanoma

whose tumors harbor an activating BRAFmutation. One notable side-effect of RAF inhibitors is the stimulation ofcutaneous skin tumors, arising in about 30% of patients receiving these drugs, which are thought to develop as aresult of inhibitor-induced activation of wild-type Raf in occult precursor skin lesions. This effect raises thepossibility that less manageable tumors might also arise in other epithelial tissues. Here we provide preclinicalevidence supporting this disquieting hypothesis by showing that the RAF inhibitors PLX-4032 (vemurafenib) andGDC-0879 precipitate the development of cell-autonomous, Ras-driven tumors in skin and gastric epithelia. Themagnitude of the effects correlated with the inhibitors' relative abilities to induce ERK activation. Epidermis-restricted ablation of either B-Raf or C-Raf prevented PLX-4032–induced ERK activation and tumorigenesis. Incontrast, GDC-0879 induced ERK activation and tumorigenesis in B-Raf–deficient epidermis, whereas C-Rafablation blocked GDC-0879–induced tumorigenesis (despite strong ERK activation) by preventing Roka–mediated keratinocyte dedifferentiation. Thus, inhibitor-induced ERK activation did not require a specific Rafkinase. ERK activation was necessary, but not sufficient for Rasþ Raf inhibitor-induced tumorigenesis, whereasC-Raf downregulation of Rokawas essential even in the face of sustained ERK signaling to prevent differentiationand promote tumorigenesis. Taken together, our findings suggest that combination therapies targeting ERK-dependent and -independent functions of Raf may be more efficient but also safer for cancer treatment. CancerRes; 73(23); 1–12. �2013 AACR.

IntroductionRAF inhibitors have revolutionized the treatment of met-

astatic patients with melanoma whose cancers harbor theBRAF mutation V600E, which increases kinase activity andleads to the constitutive activation of the MEK/ERK pathway(1–3). The side effects of the drugs are mild but include thedevelopment, in up to 30% of the patients, of keratoacantho-mas or cutaneous squamous cell carcinomas (cSCC). Incultured cells, Raf inhibitors can paradoxically stimulate Rafkinase activity and MEK/ERK activation, particularly in thepresence of activated Ras. The precise mechanism of action ofthe Raf inhibitors is yet unclear; activation/dimerizationinduced by the direct binding of the drugs to the Raf kinasedomain (4, 5), relief of autoinhibitory P-loop phosphorylation(6), and a more complex mechanism involving reactivation ofdesensitized B-Raf (7) have been proposed to explain inhib-itor-induced ERK activation. Inhibitor-induced ERK activa-

tion could potentially account for the development of ther-apy-related tumors in the patients with melanoma. Thesetumor harbor Ras mutations with variably increased frequen-cy (from 17% to 60% in different studies; refs. 8–10). Inaddition, inhibitor treatment accelerates tumor developmentin a mouse model of chemical skin carcinogenesis (8), whichpartially depends on H-Ras and involves a strong inflamma-tory response (11). It is thus yet unclear whether Ras activa-tion, in the absence of other mutations or of inflammation, isenough to license Raf inhibitor-driven tumorigenesis andwhether B-Raf or C-Raf (Raf-1) is essential for the develop-ment of drug-induced epithelial tumors.

To test the effect of Raf inhibitors on tumor-prone epider-mis, we used a transgenic mouse model in which the expres-sion of a dominant active, farnesylated Son of Sevenless (SOS-F)in keratin 5 positive basal keratinocytes leads to the develop-ment of epidermal tumors (K5-SOS-F transgenic mice; ref. 12).This system has the advantage of being cell autonomous andnot dependent on inflammation, which is absolutely necessaryto support the growth of chemically induced skin tumors (11).These 2 features allow us to focus on the effect of the inhibitorson the signaling operating in premalignant cells, reducing theimpact of other cell compartments (inflammatory cells) sup-porting tumorigenesis. In addition, the transgenic modelmimics the situation of spontaneous human cSCC, in that itinduces Ras and Erk activation in the absence of somatic Rasmutations (13), which are comparatively rare event in drug-related tumors (9).

Authors' Affiliation: Department of Microbiology, Immunobiology andGenetics, University of Vienna, Max F. Perutz Laboratories, Vienna, Austria

E. Doma and C. Rupp contributed equally to this work.

Corresponding Author: Manuela Baccarini, Department of Microbiology,Immunobiology and Genetics, University of Vienna, Max F. Perutz Labo-ratories, Doktor-Bohr-Gasse 9, 1030 Vienna, Austria. Phone: 431-427-754607; Fax: 431-427-79546; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-13-0748

�2013 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org OF1

Research. on October 18, 2020. © 2013 American Association for Cancercancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 15, 2013; DOI: 10.1158/0008-5472.CAN-13-0748

http://cancerres.aacrjournals.org/

Materials and MethodsMouse strains and inhibitor treatment

Mouse strains and genotyping have been previouslydescribed (14). Strains were maintained on a 129Sv/Bl6 (F1)background. Animal experiments were authorized by theAustrian Ministry of Science and Communications, followingthe approval by the Ethical Committee for Animal Experimen-tation. GDC-0879, PLX-4032, synthesized and formulatedaccording to published procedures (WO2006/084015 andWO2007/002325), and, in selected experiments GSK1120212(trametinib) and Y-27632 were applied daily by gavage (100mg/kgGDC-0879 in 0.5%Natrosol, citric acid pH3.0; 100mg/kgPLX-4032, high-bioavailability formulation in 5% Klucel, HClpH 3.2, WO2010/114928; 0.5 mg/kg GSK1120212 in DMSO/1%Natrosol; 2 � 15 mg/kg Y-27632 in 0.5% Klucel). The Rokinhibitor Fasudil was added to the drinking water (exchangedevery 3–4 days) at a concentration of 2 mg/mL.

Histological analysisHematoxylin and eosin (H&E) staining and immunohis-

tochemistry were performed as described (14). For BrdUincorporation, mice were injected with BrdU (12.5 mg/g bodyweight) 1 hour before tissue isolation.

Cell cultureMouse keratinocytes were isolated and maintained as

described (14). HaCaT cells (15) were obtained from the DKFZand SCC cell lines (16) were kindly provided by Gian-PaoloDotto and maintained as previously described (16). COS7 cellswere bought from Genlantis and maintained according to themanufacturer's instruction. The cell lines were not authenti-cated in our lab but were used shortly after resuscitation ofearly passages (<5) of the original material received from theprovider. A total of 70% confluent cells were treated withinhibitors for 1 hour at 32�C or at 37�C for the COS7 cells.

Site-directed mutagenesisPoint mutation was introduced with the QuickChange Site-

Directed Mutagenesis Kit (Stratagene) using pcDNA FLAG-tagged C-Raf as a template and the primers 50-gtggctgttctgcg-caaaacacaccatgtgaacattctg-30 (forward) and 50-cagaatgttca-catggtgtgttttgcgcagaacagccac-30 (reverse). The mutation wasverified by sequencing.

Immunoblotting and immunoprecipitationFor ex vivo immunoblotting and immunoprecipitation, ker-

atinocytes were prepared as previously described (17), butinstead of plating the cells were subjected to differentialcentrifugation to enrich for specific epidermal layers. TheK5þ/K10þ layers were pooled and processed further forimmunoblotting and immunoprecipitation. Cultured cells orepidermis were lysed with Laemmli buffer at 95�C for 5minutes. Lysates were immunoblotted using the followingprimary antibodies (1:1000): a-A-Raf, a-B-Raf, and a-pCofilin(Santa Cruz); a-C-Raf and a-SOS (BD Biosciences); a-pMEK,a-tMEK, a-pERK, and a-panERK (Cell Signaling); a-Cofilin,pRSK, pELK, cyclin D1, and cFos (Abcam); a-GAPDH (Milli-pore) andFLAG (Bethyl). Immunoprecipitationwas performed

as previously described (18) using a-B-Raf (Santa Cruz),a-C-Raf (Cell Signaling), or Roka antibodies (BD Biosciences)coupled to ProteinG-Sepharose beads (Invitrogen). Immuno-precipitation of FLAG-tagged C-Raf was carried out using Anti-FLAG M2 affinity gel (Sigma-Aldrich).

Ras activation assayRas-GTP was pulled down from keratinocyte lysates using

immobilized C-Raf RBD according to the manufacturer's rec-ommendation (Cytoskeleton Inc.).

Statistical analysisValues are expressed as mean (�SD) of at least 3 indepen-

dent experiments. P values were calculated using the 2-tailedStudent t test. A P value <0.05 is considered statisticallysignificant.

ResultsWe have treated the K5-SOS-F transgenic mice with 2

selective Raf inhibitors: PLX-4032 (vemurafenib; refs. 1–3, 19), currently used in the clinical treatment of melanoma;and GDC-0879 (20, 21), which is not used in the clinic but hashigher affinity for Raf and reportedly increases proliferation inmouse epidermis (5). The estimated IC50 values of the inhibi-tors at 1 mmol/L ATP concentration are similar for wild-typeB-Raf and C-Raf; in both cases, GDC-0879 is more potent withan IC50 of 34 or 180 nmol/L against B- or C-Raf, respectively,compared with the 530 and 1100 nmol/L IC50 of the toolcompound PLX-4720, the precursor of PLX-4032 (5). Theinhibitors, applied daily by gavage, could be detected in mouseplasma (Fig. 1A); in particular, the plasma exposures to PLX-4032were comparable to those necessary to achievemelanomaregression in patients (1, 2). Both drugs caused a range of skinphenotypes in K5-SOS-F transgenics, including the growth ofpreexisting tumors and the thickening of the epidermis of tail,eyelids, and ears (Fig. 1B and C). These phenotypes werealready evident after 5 to 7 days of inhibitor treatment. Themice rapidly lost weight (around 20% in 1 week; Fig. 1D) andhad to be euthanized. Necropsy revealed papilloma develop-ment in the squamous epithelia of esophagus and forestomachin 13 of 13 animals/inhibitor (Fig. 1E), likely causingweight lossand death. These epithelia express K5, but they do not developtumors in the absence of inhibitor treatment. All phenotypeswere most marked in GDC-0879-treated animals and corre-lated with an increase in proliferation (% BrdUþ cells) and adecrease in differentiation (% of epidermal layers positive forK10, expressed by suprabasal keratinocytes) in the affectedtissues, and with a strong increase in ERK phosphorylation(Fig. 2A–C). Neither inhibitor caused tumor development,toxicity, or any other overt adverse effects in WT mice, indi-cating that the phenotypes depended on concomitant Rasactivation. Some thickening of the tail, accompanied byincreased ERK phosphorylation and proliferation but not bydedifferentiation, could be observed at late time points inGDC-0879–treated WT animals (Supplementary Fig. S1). ERKactivation in vivo was paralleled by MEK/ERK activation incultured wild type and K5-SOS-Fþ premalignant mouse kerat-inocytes treated with low concentrations of the inhibitors,

Doma et al.

Cancer Res; 73(23) December 1, 2013 Cancer ResearchOF2




μmol/L μmol/L ,

μmol

/L

μmol

/L%

Cha

nge

in b

ody

wei

ght

A

B

C D

E

Figure 1. Raf inhibitors accelerate tumor formation. A, the concentration of GDC-0879 or PLX-4032 in the plasma ofmicewas determined at the indicated timepoints after application of the inhibitors via gavage and used to determine AUC. The plasma exposures to PLX-4032 were comparable to those necessary toachievemelanoma regression in patients (1; 2). The total plasmaexposurewashigher for PLX-4032 than forGDC-0879. BandC, thickening of the epidermis ofthe tail and eyelids in K5-SOS-Fþ mice treated with vehicle, GDC-0879, or PLX-4032 for 8 consecutive days. D, inhibitor-treated K5-SOS-Fþ micerapidly lose weight. Mice (n¼ 6) were weighed every day and euthanized as soon as the weight loss reached 20%. �,P < 0.01. E, massive hyperproliferation inthe esophagus of inhibitor-treated K5-SOS-Fþmice. Top, H&E-stained sections; middle, expression of the basal keratinocyte marker K5 in the suprabasallayers of the epithelia; bottom, increased proliferation in the esophagus of inhibitor-treated mice. Proliferation was assessed by BrdU incorporation. Positivecells are stained in brown. Scale bars, 100 mm.

Raf Inhibitors and Skin Carcinogenesis

www.aacrjournals.org Cancer Res; 73(23) December 1, 2013 OF3




whereas high concentrations reduced phosphorylation (Fig.3A). Inhibitor treatment similarly increased ERK phosphory-lation in human keratinocytes and squamous cell carcinomacell lines (Supplementary Fig. S2; refs. 15 and 16). GDC-0879and, to a lesser extent, PLX-4032 induced B-Raf-C-Raf hetero-dimerization in primary keratinocytes (Fig. 3B). More im-

portantly, heterodimerization could be clearly detected inepidermal lysates of inhibitor-treated animals (Fig. 3C andD). GDC-0879 was more effective than PLX-4032 in inducingex vivo MEK/ERK activation and Raf heterodimerization, ingood correlation with the more pronounced effects achievedin vivo at lower GDC-0879 plasma exposures (Fig. 1A).

A

B

C

Figure 2. Raf inhibitor treatment is associated with increased proliferation, ERK phosphorylation, and dedifferentiation in vivo. A–C, thickening of the tailepidermis in K5-SOS-Fþ mice treated with vehicle, 100 mg/kg GDC-0879, or 100 mg/kg PLX-4032 orally for 8 consecutive days. A, H&E-stainedsections. B, increased proliferation and ERK phosphorylation. C, reduced differentiation in the epidermis of inhibitor-treated K5-SOS-Fþ mice.Proliferation was assessed by BrdU incorporation. Keratin 10 (K10) was used as a marker of differentiating, suprabasal keratinocytes. BrdU, pERK, andK10 positive cells are stained in brown. The plots show the percentage of BrdU- and pERK-positive cells and the percentage of K10þ layers in vehicleand inhibitor-treated mice (mean � SD; n ¼ 3; �, P < 0.01). Dotted lines, the boundary between epidermis and dermis. Scale bars, 100 mm.

Doma et al.





Importantly, the MEK inhibitor trametinib reduced both Rasand RasþRaf inhibitor-induced tail tumor formation, indicat-ing that activation of the ERK pathway is necessary for thisprocess (Supplementary Fig. S3).The existing literature predominantly indicates that

inhibitor-induced ERK activation in cultured cells dependson C-Raf (4, 5, 7). Raf inhibitors might simply induceconformational changes that promote the folding of theRaf kinase domain and therefore dimerization (4, 5), orpreclude autoinhibitory P-loop phosphorylation (6). Alter-natively, there might be a more complex mode of action,involving the inhibition of B-Raf activity and release ofinactive B-Raf from a cytosolic complex, which, in Ras-transformed cells, leads to B-Raf plasma membrane local-ization and formation of a Ras/B-Raf/C-Raf complex (7). Toinvestigate which of these models is operating in vivo duringRas- and Raf inhibitor-driven tumorigenesis, we combined

the K5-SOS-F transgenic mice with epidermis-restrictedB-Raf or C-Raf ablation (K5-SOS-F;b-rafKOep and K5-SOS-F;c-rafKOep, knockout in the epidermis). Using these mice,we have previously demonstrated that both the develop-ment and the maintenance of Ras-driven tumors absolutelyrequire the presence of C-Raf, not as an Erk activator butrather as an endogenous inhibitor of the cytoskeleton-basedkinase Roka (Rock2), which induces keratinocyte differen-tiation (14). The effect of B-Raf ablation is milder, but it stillleads to an impressive reduction of tumor number and sizethat correlates with reduced ERK activation in the epider-mis (22). Both K5-SOS-F;c-rafKOep and K5-SOS-F;b-rafKOepmice were more resistant to treatment with Raf inhibitors.Neither strain experienced dramatic weight loss or death.Epidermis-restricted B-Raf or C-Raf ablation protectedK5-SOS-Fþ animals from the effects of PLX-4032 (Supple-mentary Fig. S4). However, treatment with GDC-0879, a

A C

B D

(μmol/L)

×g,, ×g

×g×g

×g

(μmol/L)

Figure 3. Effect of the inhibitors on mouse primary keratinocytes and epidermis. A, immunoblot analysis of lysates from primary wild type (WT) andK5-SOS-Fþ keratinocytes treated with different concentrations of GDC-0879 or PLX-4032. B, GDC-0879 and, to a lesser extent, PLX-4032 promote Rafheterodimerization in WT and K5-SOS-Fþ primary keratinocytes. C, differential centrifugation for the production of K5/K10þ-enriched epidermallysates. Epidermal lysates were centrifuged at progressively increasing g, and the pellets were subjected to immunoblotting. The pellets of the first twocentrifugation steps were enriched in K5þ basal cells and K10þ suprabasal cells. Active transglutaminase 1 (Tgase1), a marker of progressive differentiation,was present in increasing amounts in the pellets of the last three centrifugation steps. All the proteins at study were present in the first two pellets, whichwere therefore pooled and used to determine the effect of inhibitor treatment on signaling in vivo. D, GDC-0879 and, to a lesser extent, PLX-4032 promoteRaf heterodimerization in K5-SOS-F epidermis from systemically inhibitor treated mice. The presence of A-, B-, and C-Raf, total (tMEK), phosphorylatedMEK (pMEK), total (tERK), and phosphorylated ERK (pERK) was detected by immunoblotting. GAPDH is shown as a loading control. D, DMSO. �, longerexposure.






more potent Raf activator, induced the development, albeitdelayed, of large tail tumors in 8 of 8 K5-SOS-F;b-rafKOepmice (Fig. 4A and B). Thus, B-Raf is required for Ras-inducedtumorigenesis (22), and specifically tumorigenesis inducedby Ras þ PLX-4032, but this requirement is alleviated bytreatment with GDC-0879. In contrast, 22 of 22 K5-SOS-F;c-rafKOep mice were resistant to GDC-0879 treatment andmerely showed moderate thickening of the tail epidermisand lengthening of claws (Fig. 4A and B and SupplementaryFig. S5A). Importantly, Ras is activated at similar levels inK5-SOS-F, K5-SOSF;c-rafKOep, and K5-SOS-F;b-rafKOep pri-mary keratinocytes (Fig. 4C).

Spontaneous regression of therapy-related cutaneous SCCupon discontinuation of inhibitor treatment has beenobserved in the clinic (9). In the K5-SOS-F transgenic micetreated with PLX-4032 (clinically relevant situation), tailtumors continued to develop after inhibitor treatmentstopped (Supplementary Fig. S6A). This result was expectedbecause the SOS-F transgene can drive the tumors indepen-dently of inhibitor treatment. However, the mice regainedsome weight (Supplementary Fig. S6B) and the thickening ofthe esophageal epithelia brought about by the Raf inhibitorstopped (Supplementary Fig. S6B). In addition, the interrup-tion of GDC-0879 treatment, which induces tail tumors in

D

F

VehicleGDC-0879PLX-4032

K5-SOS-F; C-Raf KOep

020406080

120

% K

10 p

os. l

ayer

s

100

0

20

40

60

80

% B

rdU

pos

. cel

ls *

0

20

40

60

80

%pC

of.p

os.c

ells

010

30

50

70

*

%pE

RK

pos.

cells

K5-SOS-F; B-Raf KOep

0

20

40

60

80

% B

rdU

pos

. cel

ls *

% K

10 p

os. l

ayer

s

020406080

120100

*

0

20

40

60

80

%pC

of.p

os.c

ells

*

010

30

50

70

%pE

RK

pos.

cells

*

E

K10

pC

ofil

in

GDC-0879

H&

E

K5-SOS-F; B-Raf KOep K5-SOS-F; C-Raf KOep

Brd

UpE

RK

Time (d)1 8 1511 19 30 45

2

2.5

Tail

volu

me

(cm

3)

1.5

1

0.5

0

*

* *

*

*

K5-SOS-F K5-SOS-F;B-Raf KOep K5-SOS-F;C-Raf KOep

Vehicle

GDC-0879

PLX-4032

K5-SOS-FB-Raf KOep C-Raf KOep

Vehicle GDC-0879 Vehicle GDC-0879 Vehicle GDC-0879

K5-SOS-F

SOS-FSOS

Ras

B-Raf

C-Raf

Ras-GTP

WT WT B- C-Raf KOep

A B

C

Treatment: Genotype:

Figure 4. C-Raf, but not B-Rafablation, prevents tumordevelopment in inhibitor-treatedK5-SOS-Fþ mice. A, tail tumorsdeveloped in K5-SOS-F;b-rafKOep mice treated with GDC-0879, but not with PLX-4032 for 21days. K5-SOS-F;c-rafKOep micedisplayed slight epidermalthickening, but did not develop anytumors. B, the plot shows thevolume of tumor-affected tails(mean � SD; n ¼ 8/genotype;P < 0.01). C, the K5-SOS-Ftransgene drives Ras activationin f/f and KOep primarykeratinocytes. Ras-GTP waspulled down from keratinocytelysates using immobilized C-RafRBD. The presence of SOS,SOS-F, B-Raf, C-Raf, and Ras wasdetected by immunoblotting. D(top), H&E-stained sections of thetail of GDC-0879–treated mice.Middle and bottom, proliferation(% of BrdU positive cells) and ERKphosphorylation in the epidermis ofinhibitor-treated K5-SOS-Fþ miceof different genotypes. Note theincreased proliferation and ERKactivation in the epidermis ofGDC-0879–treated K5-SOS-F;c-rafKOep mice. E, reduceddifferentiation (% of K10 positiveepidermal layers) and cofilinphosphorylation (pCofilin) inK5-SOS-F; b-rafKOep treatedwith GDC-0879. The plots in Fshow a quantification of tissuesobtained from GDC-0879 andPLX-4032–treated mice of theindicated genotypes (mean � SD;n¼3;P<0.01). Scale bars, 100mm.

Doma et al.





the otherwise tumor-refractory K5-SOS-F;b-rafKOep mice,stopped the growth of inhibitor-induced tumors by reducingproliferation and increasing differentiation (SupplementaryFig. S6C and S6D).

Interestingly, GDC-0879 induced proliferation and ERKphosphorylation to similar extents in B- or C-Raf–deficientK5-SOS-Fþ epidermis (Fig. 4D and F). Moreover, GDC-0879maintained ERK signaling in cells deficient in both B-Raf and

A

B

Figure 5. GDC-0879 treatment activates ERK signaling in K5-SOS-Fþ, B-Raf–or C-Raf–deficient epidermis. A, immunohistochemical analysis ofthe phosphorylation of a cytosolic (pRSK) and a nuclear (pELK) target of ERK, as well as of the expression of ERK-inducible gene products(cFOS, cyclin D1). Positive cells are stained in brown. B, the plots show the percentage of positive cells observed in mice of the indicated genotypes(mean � SD; n ¼ 3; �, P < 0.05; ��, P < 0.01). Scale bars, 100 mm.






C-Raf, likely because of stimulation of signaling through A-Raf (Supplementary Fig. S7). C-Raf-deficiency, however, wassufficient to reverse RasþGDC-0879 induced de-differentia-tion. K10-negative cells were confined to the basal layer, andreinstatement of differentiation correlated with the presenceof pCofilin (Fig. 4E and F) and the absence of pSTAT3 andnuclear Myc staining (Supplementary Fig. S5B), a sign thatRoka signaling was activated in the epidermis of inhibitor-treated K5-SOS-F;c-rafKOep mice (14). Proliferation, differ-entiation, ERK, and cofilin phosphorylation were indistin-guishable in vehicle and PLX-4032-treated K5-SOS-F;c-raf-KOep and K5-SOS-F;b-rafKOep mice (Supplementary Fig. S4Aand S4B). To exclude the possibility that the ablation of B- orC-Raf has different effects on downstream ERK signaling, wetested the phosphorylation of cytosolic (RSK) and nuclear(ELK1) ERK targets, as well as the expression of ERK-

induced gene products (cFos, cyclin D1) in the epidermisof vehicle- or inhibitor-treated K5-SOS-F;c-rafKOep and K5-SOS-F;b-rafKOep mice (Fig. 5). In all cases, GDC-0879–induced ERK downstream signaling was activated at com-parable levels in the B- and C-Raf–deficient epidermis. Toobtain a more quantitative measure of these changes, wemonitored signaling in K5þ/K10þ epidermal lysates (Fig.6A). This method revealed a reduction in the amount ofC-Raf, B-Raf, and Roka in inhibitor-treated mice, irrespec-tive of their genotype; in contrast, the amount of cofilin wasincreased. The basis of these changes is currently unknown.Irrespective of this, the experiment shows similar ERKdownstream signaling in inhibitor-treated B- and C-Raf–deficient epidermis, although there was a tendency towardincreased ERK signaling in vehicle-treated C-Raf KO tissue(Figs. 5 and 6A; ref. 22); in contrast, phosphorylation of the

A

IP: C-Raf

Rok-α

B-Raf

C-Raf

− + − +B-Raf KOep

K5-SOS-F

GDC-0879

C-Raf

Rokα

IP: Rokα

B-Raf

− + − +B-Raf KOep C-Raf KOep

K5-SOS-F

GDC-0879− +− + − +B-Raf KOep C-Raf KOep

K5-SOS-F

GDC-0879

pRSK

tERK

pERK

B-Raf

C-Raf

cFos (p40)

Cyclin D1

Rokα

pCofilin

tCofilin

GAPDH

cFos (p62)

Rokα

C-Raf

IP: FLAG-C-Raf

pERK

pCofilin

tCofilin

+−− + GDC-0879WT R401H WT R401H

COS-7

B-Raf

Input

B

C

Rokα

Figure6. GDC-0879activates ERKsignalingwithout interferingwith theC-Raf/Roka interaction. A, immunoblot analysis of epidermal lysates (K5þ/K10þcells)frommice of the indicated genotypes treatedwithGDC-0879 for 8 days. B- andC-Raf, phosphorylatedERK (pERK), total (tERK), phosphorylatedRSK (pRSK),cFos p62 and p40, cyclin D1, Roka, phosphorylated cofilin (pCofilin), and total (tCofilin) were detected by immunoblotting. GAPDH is shown as aloading control. The upregulation of cFos is most clearly seen in the p40 product. B, complex formation between C-Raf and Roka is increased in B-Raf KOepidermis and is not affected by GDC-0879 treatment in vivo. Roka (top) or C-Raf (bottom) were immunoprecipitated from epidermal lysates obtainedfrom mice of the indicated genotypes, untreated or treated with GDC-0879 for 8 days. The presence of Roka, B-Raf, and C-Raf was visualized byimmunoblotting. Note the reduced amount of C-Raf i.p. loaded in lane 2 of the top panel. C, Raf dimerization is dispensable for the C-Raf/Roka interaction.FLAG-tagged wild-type C-Raf or a dimerizationmutant incapable of Raf dimerization was expressed in COS-7 cells. GDC-0879–treated or control cells werelysed and the presence of Roka and B-Raf in the FLAG-C-Raf immunoprecipitates as well as ERK and cofilin phosphorylation in the lysates was determinedby immunoblotting.

Doma et al.





Rok downstream target cofilin was abolished by GDC-0879in B-Raf–deficient epidermis, but was maintained in C-Raf–deficient epidermis (Fig. 6A).We next investigated whether Raf inhibitors had any influ-

ence on the interaction between Roka and C-Raf by immu-noprecipitating Roka from epidermal lysates of GDC-0879or vehicle-treated mice. We found that inhibitor treatmentdid not affect the amount of C-Raf co-precipitating with Roka,but that more C-Raf bound to Roka in B-Raf–deficient lysates;B-Raf could not be detected in Roka immunoprecipitatesfrom epidermal lysates of vehicle- or inhibitor-treated mice(Fig. 6B). This result was confirmed by assessing the amountof Roka and B-Raf in C-Raf immunoprecipitates from thesame epidermal lysates. In addition, a Raf dimerization-defi-cient C-Raf mutant (4) failed to form heterodimers with B-Rafbut it still bound to Roka in a manner similar to wild-typeC-Raf in both untreated and Raf inhibitor-treated cells. Con-sistently, ERK but not cofilin phosphorylation was affectedby GDC-0879 treatment or by the expression of the Rafdimerization-deficient C-Raf mutant (Fig. 6C). These resultsshow that C-Raf homodimer formation is not required forRoka binding or inhibition. They are in line with our previousdata showing that C-Raf/Roka interaction is a nonkinasefunction of C-Raf mediated by the N-terminal regulatorydomain of C-Raf (17, 23, 24) and further support the hypothesisthat Ras þ Raf-inhibitor–induced tumors requires both Raf-mediated ERK activation and a kinase-independent Rokainhibition by C-Raf.To causally link blockade of tumorigenesis andmaintenance

of differentiation to the lack of C-Raf's inhibition of Roka, wetreated K5-SOS-F;c-rafKOep mice with both Raf and Rokainhibitors (Figs. 7 and S8). Medium-term (18 days) treatmentwith the selective and potent Rok inhibitor Y-27632, whichreactivates tumor development in K5-SOS-F;c-rafKOep epider-mis (14), promoted proliferation in the GDC-0879 treatedK5-SOS-F;c-rafKOep forestomach (Supplementary Fig. S8A);and longer treatment with another Rok inhibitor, Fasudil,inhibited differentiation and induced proliferation as well asthe development of squamous cell carcinomas of the forest-omach. This was accompanied by loss of cofilin phosphoryla-tion (Fig. 7A–C). Surprisingly, cotreatment with Fasudil didnot induce tumor formation in the epidermis. This seems tobe because of a side-effect of the inhibitor in the epidermis,namely the concomitant inhibition of Rok signaling and ERKphosphorylation (Supplementary Fig. S8B), sporadicallyobserved in other tissues (25–27). ERK inhibition by Fasudilis not observed in the forestomach (Fig. 7B), reinforcingthe concept that both ERK activation and the dimming ofRok signaling are required for the formation of Ras plus Rafinhibitor-driven epithelial tumors.

DiscussionOur results provide formal proof that Raf inhibitors increase

Ras-dependent ERK activation in vivo, and show that Rasactivation suffices, in the absence of other mutations andtumor-promoting inflammation, to render keratinocytes moreprone to forming tumors upon treatment with Raf inhibitors.

The development of drug-induced tumors in the K5-SOS-Fþesophagi and forestomachs, where cells express K5 but do notgive rise to tumors unless treated, indicates that Raf inhibitorsexpand the spectrum of epithelial tumors induced by Rasactivation. Relevant for drug use in patients, both KRAS andBRAF mutations can be detected in high-grade Barrett'sintraepithelial neoplasias and in Barret's adenocarcinoma (28).

Our results support a model in which inhibitor-driven Rafand ERK activation are necessary and sufficient for hyperpro-liferation of the epidermis; in the presence of SOS-F, differen-tiation is suppressed, leading to tumor formation (Supplemen-tary Fig. S9). This is consistent with the assumption that thelevel of ERK activation is rate limiting for tumorigenesis (1), aconcept that advocates the combination therapies using RafandMEK inhibitors (29) currently being tested in clinical trials(30, 31). In support of this, PLX-4032, a weaker promoter of Rafactivation and ERK phosphorylation, does not lead toincreased proliferation in wild-type epidermis (SupplementaryFig. S1); similarly, reducing ERKactivation by cotreatmentwithTrametinib (Supplementary Fig. S3) or by ablation of either B-Raf or C-Raf (Supplementary Fig. S4) abrogates SOS-F–andPLX-4032–induced ERK activation and tumorigenesis; thus, B-and C-Raf are equally important in the context of PLX-4032-induced pathway activation.

In contrast to PLX-4032, GDC-0879 can induce sustainedERK activation and proliferation in WT epidermis, andincreases Ras-driven tumorigenesis in the absence of B-Raf(Fig. 4). Taking into account that GDC-0879–induced ERKactivation is reduced in cells expressing a C-Raf dimerizationmutant (Fig. 6C; ref. 4), the most likely explanation for thisfinding is that GDC-0879, but not PLX-4032, can facilitate Rafactivation in the context of C-Raf homodimers or of hetero-dimers containing A-Raf, as previously reported (5). In supportof this, GDC-0879 but not PLX-4032 is able to induce ERKactivation in keratinocytes lacking both B-Raf and C-Raf(Supplementary Fig. S7).

Mechanistically, the fact that GDC-0879 sustains ERK acti-vation in the absence of bothB- orC-Raf in vivo indicates thatB-and C-Raf are redundant in this context and therefore thatconformational changes leading to kinase maturation/dimer-ization involving A-Raf (4, 5), rather than a more complexmechanism involving reactivation of desensitized B-Raf (7),are responsible for Raf/ERK activation and keratinocyte pro-liferation. Because Raf heterodimerization is not readilydetected in other cell types (5, 7), these data suggest a potentialexplanation for the selective effects of the drugs on the squa-mous epithelia.

The increase in ERK activation and proliferation is necessarybut not sufficient for the development of tumors; concomitantdedifferentiation, induced by SOS-F and requiring C-Raf asendogenous inhibitor of the Roka signaling pathway, is alsoindispensable (Supplementary Fig. S9). This concept is clearlyillustrated by the lack of tumor development in SOS-Fþ,C-Raf–deficient epidermis treated with GDC-0879, despiteincreased ERK activation and proliferation (Figs. 4 and 5); andby the induction of squamous cell carcinomas in the forest-omach of K5-SOS-F;c-rafKOep mice treated with both Raf andRoka inhibitors (Fig. 7).






Roka inhibition is a nonkinase function of C-Raf, which canbe carried out both by a kinase-dead mutant (17) and bytruncated C-Raf lacking the kinase domain (23, 24). Inhibition

is mediated by the physical interaction of the C-Raf regula-tory domain with the Roka kinase domain (23). This interac-tion is not affected by Raf-inhibitor–induced Raf hetero- or

A

B

C

Figure 7. The Rok inhibitor Fasudil allows tumorigenesis in K5-SOS-Fþ, C-Raf–deficient gastric epithelia. K5-SOS-F;c-rafKOepmice were treated with GDC-0879 for 50 days. Where indicated, Fasudil was added to the drinking water. Squamous cell carcinoma developed in the forestomach of mice treated withGDC-0879þ Fasudil. A, H&E-stained sections showingmassive proliferation and carcinoma development inGDC-0879þ Fasudil-treatedmice. B, increasedBrdU incorporation and ERK phosphorylation in the forestomach of GDC-0879 þ Fasudil-treated mice. C, differentiation is decreased and cofilinphosphorylation is ablated in the forestomach of Fasudil-treated mice. Positive cells are stained in brown. Scale bars, 100 mm.

Doma et al.





homodimerization, but it is increased in SOS-Fþ, B-Raf–defi-cient epidermis; whether B-Raf ablation frees up an additionalpool of C-Raf for Roka interaction or whether more indirectmechanisms are responsible for this effect is currently unclear.Collectively, our results verify the disquieting hypothesis

that Raf inhibitors dramatically exacerbate Ras-driven tumor-igenesis in vivo, and extend the range of target tissues to gastricepithelia. Drug-induced epidermal tumors develop in micewith epidermis-restricted B-Raf ablation, whereas C-Raf abla-tion prevents tumorigenesis despite increased ERK activationby enforcing keratinocyte differentiation. These data suggeststhat combination therapies targeting kinase and nonkinasefunctions of Raf may be more efficient and safer for thetreatment of skin tumors.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: E. Doma, C. Rupp, F. Kern, A. Varga, M. BaccariniDevelopment of methodology: E. Doma, C. Rupp, F. Kern, A. Varga, M. Baccarini

Acquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): E. Doma, C. Rupp, F. Kern, A. Varga, B. RieglerAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): E. Doma, C. Rupp, F. Kern, A. Varga, B. RieglerWriting, review, and/or revision of the manuscript: E. Doma, C. Rupp,A. Varga, M. BaccariniAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): M. BaccariniStudy supervision: M. Baccarini

AcknowledgmentsThe authors thank the animal house team for technical support, M. Sibilia for

the gift of K5-SOS-F mice, and J. Raguz, I. Rauch, B. Cseh, A. Baum, O. Schaaf, S.Steuerer, and I. Waizenegger for discussions and support.

Grant SupportThisworkwas supported by grant 827500 of the Austrian Research Promotion

Agency (FFG; M. Baccarini). A. Varga was the recipient of an FEBS long-termFellowship.

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

Received March 12, 2013; revised October 3, 2013; accepted October 5, 2013;published OnlineFirst October 15, 2013.

References1. Bollag G, Hirth P, Tsai J, Zhang J, Ibrahim PN, Cho H, et al. Clinical

efficacy of a RAF inhibitor needs broad target blockade in BRAF-mutant melanoma. Nature 2010;467:596–9.

2. Flaherty KT, Puzanov I, Kim KB, Ribas A, McArthur GA, Sosman JA,et al. Inhibition of mutated, activated BRAF in metastatic melanoma.N Engl J Med 2010;363:809–19.

3. Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, Larkin J,et al. Improved survival with vemurafenib in melanoma with BRAFV600E mutation. N Engl J Med 2011;364:2507–16.

4. Poulikakos PI, ZhangC, Bollag G, Shokat KM, RosenN. RAF inhibitorstransactivate RAF dimers and ERK signalling in cells with wild-typeBRAF. Nature 2010;464:427–30.

5. Hatzivassiliou G, Song K, Yen I, Brandhuber BJ, Anderson DJ,Alvarado R, et al. RAF inhibitors prime wild-type RAF toactivate the MAPK pathway and enhance growth. Nature 2010;464:431–5.

6. Holderfield M, Merritt H, Chan J, Wallroth M, Tandeske L, Zhai H, et al.RAF inhibitors activate the MAPK pathway by relieving inhibitoryautophosphorylation. Cancer Cell 2013;23:594–602.

7. Heidorn SJ, Milagre C, Whittaker S, Nourry A, Niculescu-Duvas I,Dhomen N, et al. Kinase-dead BRAF and oncogenic RAS co-operate to drive tumor progression through CRAF. Cell 2010;140:209–21.

8. Su F, Viros A, Milagre C, Trunzer K, Bollag G, Spleiss O, et al. RASmutations in cutaneous squamous-cell carcinomas in patients treatedwith BRAF inhibitors. N Engl J Med 2012;366:207–15.

9. Oberholzer PA, Kee D, Dziunycz P, Sucker A, Kamsukom N, Jones R,et al. RAS mutations are associated with the development of cutane-ous squamous cell tumors in patients treatedwithRAF inhibitors. JClinOncol 2012;30:316–21.

10. Arnault JP, Mateus C, Escudier B, Tomasic G, Wechsler J, Hollville E,et al. Skin tumors induced by sorafenib; paradoxic RAS-RAF pathwayactivation andoncogenicmutations ofHRAS, TP53, and TGFBR1.ClinCancer Res 2012;18:263–72.

11. Mueller MM. Inflammation in epithelial skin tumours: old stories andnew ideas. Eur J Cancer 2006;42:735–44.

12. Sibilia M, Fleischmann A, Behrens A, Stingl L, Carroll J, Watt FM, et al.The EGF receptor provides an essential survival signal for SOS-dependent skin tumor development. Cell 2000;102:211–20.

13. Dajee M, Lazarov M, Zhang JY, Cai T, Green CL, Russell AJ, et al.NF-kB blockade and oncogenic Ras trigger invasive human epidermalneoplasia. Nature 2003;421:639–43.

14. Ehrenreiter K, Kern F, Velamoor V, Meissl K, Galabova-Kovacs G,Sibilia M, et al. Raf-1 addiction in Ras-induced skin carcinogenesis.Cancer Cell 2009;16:149–60.

15. Boukamp P, Petrussevska RT, Breitkreutz D, Hornung J, MarkhamA, Fusenig NE. Normal keratinization in a spontaneously immortal-ized aneuploid human keratinocyte cell line. J Cell Biol 1988;106:761–71.

16. Rheinwald JG, Beckett MA. Tumorigenic keratinocyte lines requiringanchorage and fibroblast support cultures from human squamous cellcarcinomas. Cancer Res 1981;41:1657–63.

17. Ehrenreiter K, Piazzolla D, Velamoor V, Sobczak I, Small JV, Takeda J,et al. Raf-1 regulates Rho signaling and cell migration. J Cell Biol 2005;168:955–64.

18. Chakrabandhu K, Herincs Z, Huault S, Dost B, Peng L, Conchonaud F,et al. Palmitoylation is required for efficient Fas cell death signaling.EMBO J 2007;26:209–20.

19. Joseph EW, Pratilas CA, Poulikakos PI, Tadi M, Wang W, Taylor BS,et al. The RAF inhibitor PLX4032 inhibits ERK signaling and tumor cellproliferation in a V600E BRAF-selective manner. Proc Natl Acad SciU S A 2010;107:14903–8.

20. Hoeflich KP, Herter S, Tien J, Wong L, Berry L, Chan J, et al. Antitumorefficacy of the novel RAF inhibitor GDC-0879 is predicted byBRAFV600E mutational status and sustained extracellular signal-reg-ulated kinase/mitogen-activated protein kinase pathway suppression.Cancer Res 2009;69:3042–51.

21. Hansen JD,Grina J,NewhouseB,WelchM, TopalovG, LittmanN, et al.Potent and selective pyrazole-based inhibitors of B-Raf kinase. BioorgMed Chem Lett 2008;18:4692–5.

22. Kern F, Doma E, Rupp C, Niault T, Baccarini M. Essential, non-redundant roles of B-Raf and Raf-1 in Ras-driven skin tumorigenesis.Oncogene 2013;32:2483–92.

23. Niault T, Sobczak I, Meissl K, Weitsman G, Piazzolla D, Maurer G,et al. From autoinhibition to inhibition in trans: the Raf-1 regu-latory domain inhibits Roka kinase activity. J Cell Biol 2009;187:335–42.

24. Wimmer R, Cseh B, Maier B, Scherrer K, Baccarini M. Angiogenicsprouting requires the fine tuning of endothelial cell cohesion by theRaf-1/Roka complex. Dev Cell 2012;22:158–71.

25. ChenXY,DunJN,MiaoQF, ZhangYJ. Fasudil hydrochloride hydrate, aRho-kinase inhibitor, suppresses 5-hydroxytryptamine-induced pul-monary artery smooth muscle cell proliferation via JNK and ERK1/2pathway. Pharmacology 2009;83:67–79.






26. Hata Y, Miura M, Nakao S, Kawahara S, Kita T, Ishibashi T. Antiangio-genic properties of fasudil, a potent Rho-Kinase inhibitor. Jpn JOphthalmol 2008;52:16–23.

27. Wang N, Guan P, Zhang JP, Li YQ, Chang YZ, Shi ZH, et al. Fasudilhydrochloride hydrate, a Rho-kinase inhibitor, suppresses isoproter-enol-induced heart failure in rats via JNK and ERK1/2 pathways. J CellBiochem 2011;112:1920–9.

28. Sommerer F, Vieth M, Markwarth A, Rohrich K, Vomschloss S, May A,et al. Mutations of BRAF and KRAS2 in the development of Barrett'sadenocarcinoma. Oncogene 2004;23:554–8.

29. Nissan MH, Solit DB. The "SWOT" of BRAF inhibition in melanoma:RAF inhibitors, MEK inhibitors or both? Curr Oncol Rep 2011;13:479–87.

30. GlaxoSmithKline. A Rollover study to provide continued treatmentwith GSK2118436 to subjects with BRAF mutation-positive tumors[cited 2013 Aug]. Available from: http://clinicaltrials.gov/show/NCT01231594

31. Hoffmann-La-Roche. A studyof vemurafenib andGDC-0973 inpatientswith BRAF-mutation positive metastatic melanoma [cited 2013 Aug].Available from: http://clinicaltrials.gov/show/NCT01271803

Doma et al.





Published OnlineFirst October 15, 2013.Cancer Res Eszter Doma, Christian Rupp, Andrea Varga, et al. ERKUpon Raf Functions That Are Dependent and Independent of Skin Tumorigenesis Stimulated by Raf Inhibitors Relies

Updated version

10.1158/0008-5472.CAN-13-0748doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2013/10/15/0008-5472.CAN-13-0748.DC1

Access the most recent supplemental material at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/early/2013/11/22/0008-5472.CAN-13-0748To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-13-0748

http://cancerres.aacrjournals.org/content/suppl/2013/10/15/0008-5472.CAN-13-0748.DC1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/early/2013/11/22/0008-5472.CAN-13-0748


skintumorigenesisstimulatedbyrafinhibitorsreliesupon raf … · 2013-11-22 · raf functions that...

Documents