Aberrant AKT activation driveswell-differentiated liposarcomaAlejandro Gutierreza,b,1, Eric L. Snyderc,d,e,f, Adrian Marino-Enriquezc, Yi-Xiang Zhangf,g, Stefano Sioleticf,g,Elena Kozakewicha, Ruta Grebliunaitea, Wen-bin Ouc, Ewa Sicinskad,f, Chandrajit P. Rautg,h,George D. Demetrif,g, Antonio R. Perez-Ataydei, Andrew J. Wagnerf,g, Jonathan A. Fletcherc,g,Christopher D. M. Fletcherc, and A. Thomas Looka,b,1

aDepartment of Pediatric Oncology, dDepartment of Medical Oncology and Center for Molecular Oncologic Pathology, fThe Ludwig Center at Dana-Farber/Harvard Cancer Center, and gCenter for Sarcoma and Bone Oncology, The Dana-Farber Cancer Institute, MA 02215; bDivision of Hematology/Oncology,iDepartment of Pathology, Children’s Hospital, Boston, MA 02115; cDepartment of Pathology and hDepartment of Surgery, Brigham and Women’s Hospital,Boston, MA 02115; and eThe Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139

Edited* by Dennis A. Carson, University of California at San Diego, La Jolla, CA, and approved August 15, 2011 (received for review April 21, 2011)

Well-differentiated liposarcoma (WDLPS), one of the most com-mon human sarcomas, is poorly responsive to radiation and che-motherapy, and the lack of animal models suitable for experimen-tal analysis has seriously impeded functional investigation of itspathobiology and development of effective targeted therapies.Here, we show that zebrafish expressing constitutively active Akt2in mesenchymal progenitors develop WDLPS that closely resem-bles the human disease. Tumor incidence rates were 8% in p53wild-type zebrafish, 6% in p53 heterozygotes, and 29% in p53-homozygous mutant zebrafish (P = 0.013), indicating that aberrantAkt activation collaborates with p53 mutation in WDLPS patho-genesis. Analysis of primary clinical specimens of WDLPS, and ofthe closely related dedifferentiated liposarcoma (DDLPS) subtype,revealed immunohistochemical evidence of AKT activation in 27%of cases. Western blot analysis of a panel of cell lines derived frompatients with WDLPS or DDLPS revealed robust AKT phosphoryla-tion in all cell lines examined, even when these cells were culturedin serum-free media. Moreover, BEZ235, a small molecule inhibitorof PI3K and mammalian target of rapamycin that effectively inhib-its AKT activation in these cells, impaired viability at nanomolarconcentrations. Our findings are unique in providing an animalmodel to decipher the molecular pathogenesis of WDLPS, and im-plicate AKT as a previously unexplored therapeutic target in thischemoresistant sarcoma.

Liposarcoma is the most common sarcoma of humans, affect-ing ∼2,000 individuals per year in the United States (1).

These tumors are classified into five histopathologic subtypes,with well-differentiated liposarcoma (WDLPS) accounting for∼50% of cases, and dedifferentiated liposarcoma (DDLPS), aclosely related subtype that appears to arise from further ma-lignant progression of WDLPS, accounting for an additional 9%to 18% of cases (1–3). Liposarcomas are generally thought toarise de novo rather than from preexisting benign lesions, andmost patients lack recognized causative factors. Although com-plete surgical resection can be curative, WDLPS often developsin deep anatomic locations, such as the retroperitoneum ormediastinum, where its propensity to enwrap vital structurestypically makes complete surgical resection difficult or impossi-ble, leading to high morbidity and mortality rates (1, 4). Radia-tion and chemotherapy have limited efficacy in the treatment ofWDLPS (5, 6). Indeed, there are no systemic therapeutic regi-mens known to improve survival when complete surgical re-section is not feasible, underscoring the need for an improvedmolecular understanding of WDLPS to stimulate the develop-ment of effective targeted therapies.The MDM2-p53 pathway plays a prominent role in WDLPS

pathogenesis, with the vast majority of human tumors harboringeither MDM2 amplifications or p53 mutations (6–10). Moreover,individuals with germ-line p53 mutations appear to be at in-creased risk of WDLPS development at a very young age (11).

Regions of chromosome 12q13-15 are often amplified in well-differentiated and dedifferentiated liposarcomas, typically in-volving MDM2, CDK4, and HMGA2, along with several othergenes (6, 10, 12, 13); JUN can also be amplified in WDLPS casesthat have a dedifferentiated component (14). Further dissectionof WDLPS molecular pathogenesis has been greatly impeded bythe lack of animal models suitable for experimental analysis.Oncogenic signal transduction through the PI3K-AKT path-

way, which is widely dysregulated in human cancer, is normallydown-regulated by the PTEN tumor suppressor (15). Individualswith germ-line PTEN-inactivating mutations frequently developmultiple lipomas (benign adipocytic neoplasms) (16), and AKTactivation has been described in human liposarcomas (17), sug-gesting that the PI3K-AKT pathway is involved in adipocytetransformation. Here we show that expression of constitutivelyactive Akt2 in zebrafish mesenchymal progenitors inducesWDLPS, thus being unique in providing an animal model forfuture investigation of this disease. Moreover, we also show thatAKT pathway inhibition impairs viability in human cell linesderived from patients with WDLPS and DDLPS, thus implicat-ing AKT as a previously unexplored therapeutic target in thesechemoresistant sarcomas.

ResultsExpression of Constitutively Active Akt2 Induces Well-DifferentiatedLiposarcoma. To test the hypothesis that Akt is a WDLPS onco-gene that collaborates with p53 inactivation during adipocytetransformation, we in-crossed zebrafish harboring heterozygousp53M214K mutations, which encode a transactivation-defectivep53 protein (18), and all resultant embryos were microinjected atthe one-cell stage with a rag2:myr-mAkt2 expression construct(Fig. 1A). This construct encodes a myristoylated, constitutivelyactive mouse Akt2 transgene (19) driven by a zebrafish rag2

Author contributions: A.G., E.L.S., A.M.-E., W.-b.O., J.A.F., C.M.D.F., and A.T.L. designedresearch; A.G., E.L.S., A.M.-E., Y.-X.Z., S.S., E.K., R.G., and W.-b.O. performed research;E.L.S., S.S., E.S., C.P.R., G.D.D., A.J.W., J.A.F., and C.M.D.F. contributed new reagents/ana-lytic tools; A.G., E.L.S., A.M.-E., Y.-X.Z., S.S., E.K., W.-b.O., C.P.R., G.D.D., A.R.P.-A., J.A.F.,C.M.D.F., and A.T.L. analyzed data; and A.G. and A.T.L. wrote the paper.

Conflict of interest statement: C.P.R. is a consultant for Novartis and participates in clinicaltrials of Novartis. G.D.D. is a consultant for Novartis, Pfizer, Ariad, Johnson & Johnson,PharmaMar, Genentech, Infinity Pharmaceuticals, EMD-Serono, Glaxo Smith Kline, Am-gen, Daiichi-Sankyo, ArQule, Enzon, Millenium/Takeda; is a member of the scientificadvisory board of Plexxikon, ZioPharm, Nereus, N-of-One, and Kolltan Pharmaceuticals;and participates in clinical trials of Novartis, Pfizer, Ariad, Johnson & Johnson, Pharma-Mar, and Infinity Pharmaceuticals. A.J.W. is a consultant for Novartis, Roche/Genentech,Sanofi, Pfizer, EMD-Serono, and participates in clinical trials supported by Novartis,Roche/Genentech, Pfizer, and Exelixis.

*This Direct Submission article had a prearranged editor.1To whom correspondence may be addressed. E-mail: thomas_look@dfci.harvard.edu orAlejandro_Gutierrez@dfci.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1106127108/-/DCSupplemental.

16386–16391 | PNAS | September 27, 2011 | vol. 108 | no. 39 www.pnas.org/cgi/doi/10.1073/pnas.1106127108

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

promoter fragment that drives ectopic expression in mesenchy-mal progenitors (20). Zebrafish injected with rag2:myr-mAkt2developed externally visible solid tumors between 1 and 4 mo ofage; the tumor incidence rates were 29% in p53-homozygousmutants, 6% in p53 heterozygotes, and 8% in their p53 wild-typesiblings (P = 0.01) (Fig. 1 B and C). Histologic analyses revealedthat 91% of these tumors consisted of locally invasive masses ofadipocytes showing considerable variation in cell size, togetherwith scattered atypical stromal cells with hyperchromatic nucleiand lipoblasts characterized by multivacuolated cytoplasm andlarge hyperchromatic pleomorphic nuclei, findings that are di-agnostic of WDLPS in humans (Fig. 1 D–F) (2), whereas theremaining 9% of tumors were osteosarcomas, as described be-low. Immunohistochemical analysis revealed strong reactivity forphospho-AKT(Ser473) in the tumor cells of rag2:myr-mAkt2transgenic zebrafish but not in the normal fat of control rag2:GFP transgenic fish, indicating expression of the constitutivelyactive Akt2 transgene (Fig. 1 G–I). Moreover, none of the con-trol p53-homozygous mutant zebrafish injected with rag2:GFP(n = 60) developed tumors by 6 mo of age.In addition, one p53-homozygous mutant injected with rag2:

myr-mAkt2 developed a rigid mass at the base of the dorsal fin

that demonstrated increased lobulation and vascularity com-pared with the WDLPS tumors (Fig. 2 A and B). Histologicanalysis revealed that this mass consisted predominantly of os-teoid interspersed with large malignant cells characterized bypleomorphic nuclei, features that are diagnostic of osteosarcomain humans (Fig. 2 C and D) (2).

Aberrant AKT Pathway Activation in Primary HumanWell-Differentiatedand Dedifferentiated Liposarcomas. To test whether AKT activa-tion is also involved in human liposarcoma pathogenesis, weperformed immunohistochemical analysis for phospho-AKT(Ser473) on clinical specimens from 58 patients with well-dif-ferentiated and dedifferentiated liposarcomas. These studiesdemonstrated AKT activation in a subset of the human WDLPSor DDLPS cases (Fig. 3 A–F), including 22% of the pureWDLPS (n = 23) and 46% of pure DDLPS (n = 13) tumorsanalyzed (Fig. 3J). We also included human tumors containingboth well-differentiated and dedifferentiated liposarcoma com-ponents in our analysis (n = 22), which revealed phospho-AKTpositivity in 32% of the well-differentiated components and 45%in the dedifferentiated components of these cases (Fig. 3J).Given that phospho-AKT is a labile epitope in clinical speci-

C

Control Well-Differentiated Liposarcoma

H&

Ep

ho

sph

o-A

KT

D E F

IHG

0 2 4 6 80

10

20

30

p53 wt/wt (n=25)p53 mut/wt (n=52)p53 mut/mut (n=21)

Age (months)

Sol

id T

umor

Inci

denc

e (%

)

P = 0.013A BX

tp53

rag2 promoter- myr mAkt2

wt/mut tp53wt/mut

Inject:

Assess Tumor Onset Genotype

Fig. 1. ConstitutiveAkt activationdrivesWDLPS in the zebrafish. (A) Experimental design. (B) Solid tumor incidence inp53wild-type, heterozygous, orp53M214K

homozygous mutant siblings injected with a rag2:myr-mAkt2 transgene at the one-cell stage. P value calculated via log-rank test. (C) Representative rag2:myr-mAkt2-injected zebrafish, which developed what appear to be two independent solid tumors. The animal shown was p53-homozygous mutant. (Scale bar, 5mm.) Note that the image shown in (C) consists of merged adjacent photomicrographs. (D) Control H&E-stained zebrafish section. Arrow points to normalsubcutaneous adipocytes. (E) Low-magnification view of an H&E section through the zebrafish shown in C, demonstrating a locally invasive mass consisting ofwell-differentiated adipocytes with significant variation in cell size. (F) High-magnification view of H&E section demonstrating a representative lipoblastscattered throughout these tumors, characterized by a multivacuolated cytoplasm and large hyperchromatic pleomorphic nuclei. (G) Phospho-AKT immuno-histochemistry on a control zebrafish section. Arrow points to normal subcutaneous adipocytes, which lacked detectable pAKT staining. (H and I) Phospho-AKTimmunohistochemistry on tumor sections from the zebrafish shown in C, revealing strong immunoreactivity for phospho-AKT in tumor cells of rag2:myr-mAkt2-transgenic zebrafish. (Scale bars, 100 μm.)

Gutierrez et al. PNAS | September 27, 2011 | vol. 108 | no. 39 | 16387

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

mens, we also performed immunohistochemistry for phospho-S6(Ser235/236), a downstream target of the AKT pathway (21).Similar results (Fig. 3 G–I) were obtained, with 17% of pureWDLPS (n = 23), 41% of well-differentiated WDLPS/DDLPScomponents (n = 22), and 47% of DDLPS (pure DDLPS, n =13; dedifferentiated WDLPS/DDLPS components, n = 21),

demonstrating evidence of S6 activation (Fig. 3K). We found noimmunohistochemical evidence of AKT or S6 phosphorylation inlipomas or in normal adipose tissue (Fig. 3 D and G, and Fig. S1).To determine whether AKT activation is aberrant in human

WDLPS and DDLPS, we took advantage of a panel of cell linesderived from patients with these sarcomas to evaluate phos-phorylation of AKT and of its downstream target GSK3β after4 h of growth in serum-free conditions. Strikingly, Western blotanalysis revealed persistent phosphorylation of AKT and of itsdownstream target GSK3β in all eight cell lines examined, evenunder such serum-starved conditions. In contrast, serum starva-tion resulted in silencing of AKT and GSK3β phosphorylation incontrol SU-CCS-1 clear cell sarcoma cells (Fig. 4).

PI3K-AKT-Mammalian Target of Rapamycin Pathway Inhibition Im-pairs the Viability of Human Liposarcoma Cells. To determinewhether human WDLPS and DDLPS cells are dependent onaberrant AKT pathway activation, we treated four cell lines (twoWDLPS and two DDLPS) with BEZ235, a dual-specificity in-hibitor of PI3K and both mammalian target of rapamycin(mTOR) complexes (22) that effectively silences AKT pathwayactivation in these cells (Fig. 5A). BEZ235 treatment for 72 hdecreased the viability of all cell lines tested, with IC50 valuesranging from 13 to 75 nM (Fig. 5B). Treatment with rapamycin,an mTORC1 inhibitor, had less of an effect on viability (Fig. 5C),suggesting that the PI3K-AKT pathway plays both mTORC1-dependent and -independent roles in the pathobiology ofWDLPS and DDLPS. Analysis of cell-cycle profiles revealed thatBEZ235 treatment induced G1 arrest at nanomolar concen-trations (Fig. 5D), whereas apoptosis was induced only at a 1-μMconcentration (Fig. 5E).

A Lipoma

H&

E

pAK

T

WDLPS DDLPS

B C

D E F

J K

pS6

G H I

Pure WDLPS (n=23)

Pure DDLPS (n=13)

0

20

40

60

80

100 Negative Intermediate High

pAK

T IH

C S

tain

ing

(%)

Well-Diff (n=22)

Dediff (n=22)

Cases with WDLPS & DDLPS components

0

20

40

60

80

100

pS6

IHC

Sta

inin

g (%

)

Pure WDLPS (n=23)

Pure DDLPS (n=13)

Negative Intermediate High

Well-Diff (n=22)

Dediff (n=21)

Cases with WDLPS & DDLPS components

Fig. 3. AKT pathway activation in primary human well-dif-ferentiated and dedifferentiated liposarcomas. (A–C) H&Estaining of human lipoma, WDLPS, and DDLPS specimens. (D–F)Immunohistochemistry for Ser473-phosphorylated AKT in arepresentative lipoma and AKT-positive WDLPS and DDLPSclinical specimens. (Scale bar, 100 μm.) (G–I) Immunohisto-chemistry for phospho-S6 ribosomal protein (Ser235/236) ina representative lipoma, as well as AKT-positive WDLPS andDDLPS clinical specimens. (J and K) Quantitation of phospho-AKT and phospho-S6 immunohistochemistry in pure WDLPS,pure DDLPS, and in human tumors with mixed well-differen-tiated and dedifferentiated liposarcoma components.

A B

C D

Fig. 2. Osteosarcoma development in a rag2:myr-mAkt2-injected, p53-ho-mozygous mutant zebrafish. (A and B) A p53-homozygous mutant zebrafishinjected with the rag2:myr-mAkt2 expression construct developed a solidlobulated mass at the base of the dorsal fin. Note that the image shown in Aconsists of merged adjacent photomicrographs. (Scale bars, 1 mm.) (C and D)H&E-stained sections at low and high magnification, respectively, demon-strate a mass consisting predominantly of osteoid matrix interspersed withlarge malignant cells with pleomorphic nuclei, features that in humans arediagnostic of osteoblastic osteosarcoma. (Scale bars, 100 μm.)

16388 | www.pnas.org/cgi/doi/10.1073/pnas.1106127108 Gutierrez et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

DiscussionWe have demonstrated that expression of activated Akt2 in mes-enchymal progenitors drives WDLPS in transgenic zebrafish, andthat nearly one-third of clinical specimens from primary cases ofhuman WDLPS and DDLPS showed immunohistochemical evi-dence of AKT pathway activation. Moreover, treatment with thePI3K-AKT pathway inhibitor BEZ235 inhibited viability in all celllines derived from patients with WDLPS and DDLPS that wetested. Taken together, our findings implicate a central role foroncogenic AKT signaling in themolecular pathogenesis of humanWDLPS and DDLPS, and suggest the need for clinical trials ofAKT pathway inhibitors in patients with unresectable disease, forwhom there are currently no known effective therapies.Our analyses of primary human tumors revealed an increased

frequency and intensity of staining for phospho-AKT and phospho-S6 ribosomal protein in dedifferentiated liposarcomas com-pared with their well-differentiated counterparts. These findingssuggest the intriguing possibility that AKT activation may definea subset of WDLPS, which is particularly prone to dediffer-entiation, a process that may be caused in part by the acquisitionof additional oncogenic abnormalities further potentiating sig-naling through the AKT pathway. However, we cannot rule outthe possibility that this apparent difference may simply be relatedto the greater difficulty of detecting phosphorylated epitopes inWDLPS sections, in which most of the tumor mass consists oflarge fat vacuoles within malignant yet well-differentiated adipo-cytes, whereas the densely cellular DDLPS tumors have a muchgreater number of cellular elements in which AKT phosphoryla-tion can be assessed per section. Further studies will be requiredto establish the mechanisms underlying this observation.Recent work has revealed that 18% of myxoid/round-cell lip-

osarcomas harbor activating mutations in PIK3CA, encoding thecatalytic subunit of class IA PI3K (13). Myxoid/round-cell lip-osarcomas are characterized by t(12;16)(q13-14;p11) transloca-tions, resulting in expression of TLS-CHOP fusion proteins,whereas these tumors lack the 12q amplifications characteristicof WDLPS/DDLPS, leading most investigators to believe thatthese liposarcoma subtypes are biologically distinct (1, 2). S6K1,a direct target of mTORC1 downstream of PI3K-AKT, has re-cently been shown to be required for the earliest stages of adi-pogenesis (23), providing one plausible mechanism to explainselection for AKT pathway activation in diverse liposarcomasubtypes. Nevertheless, the fact that expression of activated AKTin p53-mutant zebrafish drives development of WDLPS but notmyxoid/round-cell liposarcomas indicates that AKT activationand p53 mutation can be early events in WDLPS pathogenesis,whereas expression of the TLS-CHOP fusion may be required inaddition to PI3K-AKT pathway activation in the genesis ofmyxoid/round-cell liposarcomas.

Current knowledge of the molecular pathogenesis of WDLPShas been driven by genetic analyses of human tumors, which haverevealed that almost all cases harborMDM2 amplification orTP53mutations (7–10). Evidence also suggests that individuals withgerm-line TP53 mutations are at increased risk of WDLPS de-velopment at a very young age (11). By demonstrating that acti-vatedAkt2 and p53mutations collaborate in the zebrafish, we havenow experimentally demonstrated the long-suspected role of p53as a tumor suppressor in WDLPS. These tumors are also charac-terized by recurrent amplifications of distinct regions of chromo-some 12q13-15 (6, 10, 12), but until now it has not been possible toidentify which of the involved genes are WDLPS oncogenesdriving the selection for these amplifications, andwhich aremerelynonpathogenic “passengers.” Furthermore, although dediffer-entiated liposarcoma is thought to arise because of further ma-lignant transformation of WDLPS, it has previously been im-possible to directly test the ability of candidate genetic lesions todrive this transformation in a physiological context. The zebrafishmodel we describe now provides a platform for experimental

vinculin

GSK3βp-GSK3β

AKTp-AKT

WDLPS DDLPS

1 1 2 3 4 5 6 7 8Serum +

Control

Fig. 4. Aberrant AKT activation in human cell lines derived from patientswith WDLPS and DDLPS. Western blot analysis for phosphorylation of AKTand its downstream target GSK3β was performed in a panel of cell linesderived from patients with WDLPS (samples 1–3) or DDLPS (samples 4–8),grown in serum-free conditions for 4 h before analysis. Control is the SU-SSC-1clear-cell sarcoma cell line, shown in the presence and absence of serum.Sample 1 was run on both Western blots as a control for interblot variability.

0 1 3 10 30 100 300 10000

20

40

60

80

100G0/G1SG2/M

BEZ235 (nM)

% o

f Cel

ls

0.1 1 10 100 10000

BEZ235 (nM)

0

0.2

0.4

0.6

0.8

1.0

1.2

Rel

ativ

e C

ell N

umbe

r

A

B

D

C

E

0 30 100 300 10000

5

10

15

% o

f Cel

ls

BEZ235 (nM)

Early ApoptosisLate Apoptosis

vinculin

GSK3β

p-GSK3β

AKT

p-AKT

0 1 3 10 30 100 300 1000

0.1 1 10 100 10000Rapamycin (nM)

0

0.2

0.4

0.6

0.8

1.0

1.2

Rel

ativ

e C

ell N

umbe

r

LP6 BEZ235 (nM)

0

Controls

+ -

12

LP66

12

LP66

Fig. 5. PI3K-AKT-mTOR pathway inhibitors impair viability in humanWDLPSand DDLPS cells. (A) Western blot analysis of the LP6 cell line, derived froma patient with DDLPS, demonstrating that BEZ235 effectively inhibits phos-phorylation of AKT and its downstream target GSK3β at nanomolar concen-trations. Positive and negative controls are the SU-SSC-1 cell line grown in thepresence and absence of serum, respectively. (B) Viability of four cell linesderived frompatientswithWDLPS (cells 1 and 2) or DDLPS (cells 6 and LP6)wasassessed after 72 h of BEZ235 treatment using the CellTiter-Glo luminescentcell viability assay. Values represent mean± SEM (n = 3 replicates). IC50 valueswere 19, 32, 14, and75nM, respectively, for cells 1, 2, 6, and LP6. (C) Viability ofWDLPS/DDLPS cell lines after 72hof rapamycin treatment. (D) Effect ofBEZ235treatment on cell cycle distribution of LP6 cells examined by flow cytometryanalysis at 24 h. Data shown are representative of two independent experi-ments. (E) Effect of BEZ235 treatment on apoptosis of LP6 cells, assessed byAnnexin V and 7-AAD double-staining. Data shown are representative of twoindependent experiments.

Gutierrez et al. PNAS | September 27, 2011 | vol. 108 | no. 39 | 16389

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

studies to dissect molecular pathogenesis and discover noveltherapeutic targets in this chemoresistant tumor.

Materials and MethodsZebrafish Husbandry, Mutant Lines, and Imaging. Zebrafish husbandry wasperformed as previously described (24), in accord with protocols approved bythe Dana-Farber Cancer Institute Animal Care and Use Committee. Thep53M214K-mutant zebrafish line was previously described (18). Zebrafishimages were obtained using a Nikon SMZ1500 microscope, Nikon DS2MBWccamera, and NIS-Elements F Package Ver. 3.00 (Nikon Instruments Inc.).

Expression Constructs. The rag2:myr-mAkt2 construct was generated byplacing a myristoylated murine Akt2 transgene (19), which is constitutivelyactivated as a result of constitutive membrane localization, downstream ofa zebrafish rag2 promoter fragment (25) in a modified pBluescript vector,wherein the rag2:myr-mAkt2 construct is flanked by recognition sequencesfor I-SceI meganuclease.

Generation of Transgenic Zebrafish. Circular rag2:myr-mAkt2 plasmid DNA(30 pg) was microinjected along with I-SceI meganuclease (New EnglandBiolabs) into one-cell stage zebrafish embryos from the AB wild-type strain,as previously described (26).

Zebrafish Paraffin Embedding and Sectioning. Zebrafish were killed in tricaineanesthetic, fixed in 4% paraformaldehyde at 4 °C for 2 d, decalcified with0.25 M EDTA (pH 8.0) for 2 d, dehydrated in alcohol, cleared in xylene, andembedded in paraffin. Tissue sections from paraffin-embedded tissueblocks were placed on charged slides, deparaffinized in xylene, rehydratedthrough graded alcohol solutions, and stained with H&E or analyzed byimmunohistochemistry.

Patient Samples. Human well-differentiated liposarcoma, dedifferentiatedliposarcoma, lipoma, and normal fat specimens were removed at surgery andcollected from patients treated at Brigham andWomen’s Hospital, who gaveinformed consent for use of anonymized surgical specimens for researchpurposes after all clinically relevant evaluations were performed, with ap-proval of the Partners Health Care Institutional Review Board. The diagnosisof well-differentiated liposarcoma/atypical lipomatous tumor, dediffer-entiated liposarcoma, or lipoma was made by institutional pathologists andreviewed by E.L.S. and C.D.M.F. to ensure diagnostic accuracy based on cri-teria of the World Health Organization (2).

Immunohistochemistry. Immunohistochemistry on human samples was per-formed on a tissue microarray containing three 0.4-mm cores from eachindividual tumor, and on selected whole tumor sections. Zebrafish immu-nohistochemistry was performed on slides of whole zebrafish sections. Slideswere deparaffinized and pretreated with 10 mM citrate (pH 6.0) in a steampressure cooker (Decloaking Chamber, BioCare Medical) according to themanufacturer’s instructions, followed by washing in distilled water. Slideswere pretreated with Peroxidase Block (Dako) for 5 min, followed by serum-free protein block (Dako) for 20 min. Primary rabbit antibody to Ser473phospho-AKT (#4058; Cell Signaling Technology) or to Ser235/236 phospho-S6 ribosomal protein (#4858; Cell Signaling Technology) was applied ata 1:50 dilution in Antibody Diluent (Dako) and incubated at 4 °C overnight(phospho-AKT) or at room temperature for 1 h (phospho-S6). Anti-rabbithorseradish peroxidase-labeled polymer (Dako) was applied for 30 min.Immunoperoxidase staining was performed with diaminobenzidine (DAB)+chromogen kit (Dako), according to the manufacturer’s instructions.

Immunohistochemistry was independently scored by two pathologists(E.L.S. and S.S.), who then reviewed the few discordant cases togetherto arrive at a consensus score. Tumors were scored as positive if >10% ofsarcoma cells exhibited evidence of specific staining for phospho-AKT orphospho-S6. Positive tumors were further subclassified based on intensity ofstaining into either high (strong staining) or intermediate (weak-to-moderatestaining) categories. Normal adipose tissue was used as a negative control.

Inhibitors. BEZ235 and rapamycin were purchased from AXONMedchem andCalbiochem, respectively.

Western Blotting. Whole-cell lysates were prepared using lysis buffer (1%Nonidet P-40, 50mM Tris-HCl pH 8.0, 100mM sodium fluoride, 30mM sodiumpyrophosphate, 2 mM sodium molybdate, 5 mM EDTA, and 2 mM sodiumorthovanadate) containing protease inhibitors (10 μg/mL aprotinin, 10 μg/mLleupeptin, and 1 mM phenylmethylsulfonyl fluoride). The lysates werethen rocked overnight at 4 °C. Lysates were cleared by centrifugation at16,100 × g for 30 min at 4 °C, and protein concentrations were determinedwith a Bio-Rad protein assay (Bio-Rad Laboratories). Equal amounts of pro-tein were separated by SDS/PAGE, blotted to nitrocellulose membranes(Schleicher and Schuell) and then stained with the following antibodies: AKT(Cell Signaling Technology; #9272; 1:500), phospho-AKT(Ser473) (Cell SignalingTechnology; #9271; 1:500), phospho-GSK3β(Ser9) (Cell Signaling Technology;#9336; 1:1,000), GSK3 (Santa Cruz; #sc-7291; 1:500), and vinculin (Sigma-Aldrich; #V4505; 1:500). The hybridization signals were detected by chem-iluminescence (ECL, Amersham Biosciences) and captured using a LAS1000-pluschemiluminescence imaging system (Fujifilm).

Cell Viability Assays. Liposarcoma cells were plated in 96-well plates at 2,000cells per well in 100 μL of medium containing 15% FBS. After 24 h, cells wereexposed to increasing concentrations of compounds. Each concentration wastested in triplicate. Cell viability was determined after 72 h using the Cell-Titer-Glo Luminescent Cell Viability Assay Kit (Promega) with a modificationto the manufacturer’s protocol wherein the CellTiter-Glo reagent was di-luted 1:3 with PBS. The relative luminescence units (RLU) were measuredusing the FLUOstar Optima plate reader (BMG Labtech GmbH) and relativecell number was calculated by normalization to the RLU of the controltreated cells. The inhibitory concentrations 50 (IC50) were calculated usingSigmoidal dose-response (variable slope) curve fitting with Prism version 5.0(GraphPad Software).

Cell Cycle and Apoptosis Analyses. Human liposarcoma cells were exposed toinhibitors or 0.1% DMSO for 24 h and harvested. For cell cycle analysis, cellswere washed with ice-cold PBS, fixed in 70% ethanol at 4 °C overnight,and stained in PBS containing 10 μg/mL RNase A and 20 μg/mL propidium io-dide (Sigma) in the dark. DNA content analysis was performed by flowcytometry (FACScan; Becton Dickinson) with CellQuest and ModFIT LT soft-ware (Becton Dickinson).

For apoptosis analysis, cells were exposed to BEZ235 or 0.1% DMSO for38 h and harvested. Annexin V and 7-aminoactinomycin D (7-AAD) stainingwas performed using the PE Annexin V Apoptosis Detection Kit I (#558763;BD Pharmingen) according to the manufacturer’s instructions. Stained cellswere quantitated as viable (Annexin V−/7-AAD−), early apoptotic (AnnexinV+/ 7-AAD−), or late apoptotic (Annexin V+/ 7-AAD+) by flow cytometry(FACScan; BD Biosciences) with CellQuest software (BD Biosciences).

Statistical Analyses. Differences in sarcoma incidence between zebrafish ofdifferent p53 genotypes were assessed by the log-rank test. Differences incategorical data were assessed via Fisher’s exact test.

ACKNOWLEDGMENTS. We thank G. Molind and L. Zhang for zebrafishhusbandry, J. Testa for the myristoylated mouse Akt2 transgene used inthese studies, and D. E. Fisher for the SU-CCS-1 cell line. This work wasconducted with support from Harvard Catalyst j The Harvard Clinical andTranslational Science Center (National Institutes of Health Award UL1 RR025758 and financial contributions from Harvard University and its affiliatedacademic health care centers), as well as the Ludwig Center at Dana-Farber/Harvard Cancer Center and Harvard Medical School; A.G. is supported byNational Institutes of Health Grant 1K08CA133103 and is a scholar of theAmerican Society of Hematology-Amos Medical Faculty Development pro-gram; and A.M.-E. is supported by Fundacion Alfonso Martin Escudero.

1. Dalal KM, Antonescu CR, Singer S (2008) Diagnosis and management of lipomatous

tumors. J Surg Oncol 97:298–313.2. Fletcher CDM, Unni KK, Mertens F eds (2002) Pathology and Genetics of Tumours of

Soft Tissue and Bone (IARC Press, Lyon).3. Kransdorf MJ (1995) Malignant soft-tissue tumors in a large referral population:

Distribution of diagnoses by age, sex, and location.AJR Am J Roentgenol 164:129–134.4. Dei Tos AP, Pedeutour F (2002) Atypical lipomatous tumor/well differentiated lip-

osarcoma; Dedifferentiated liposarcoma. Pathology and Genetics of Tumors of Soft

Tissue and Bone: World Health Organization Classification of Tumors., ed

Fletcher CDM (IARC press, Lyon), pp 35–39.5. Jones RL, Fisher C, Al-Muderis O, Judson IR (2005) Differential sensitivity of lip-

osarcoma subtypes to chemotherapy. Eur J Cancer 41:2853–2860.6. Conyers R, Young S, Thomas DM (2011) Liposarcoma: Molecular genetics and thera-

peutics. Sarcoma 2011:483154.7. Leach FS, et al. (1993) p53 Mutation and MDM2 amplification in human soft tissue

sarcomas. Cancer Res 53(10, Suppl):2231–2234.

16390 | www.pnas.org/cgi/doi/10.1073/pnas.1106127108 Gutierrez et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021

8. Pilotti S, et al. (1997) Distinct mdm2/p53 expression patterns in liposarcoma sub-groups: Implications for different pathogenetic mechanisms. J Pathol 181:14–24.

9. Schneider-Stock R, et al. (1998) MDM2 amplification and loss of heterozygosity at Rband p53 genes: no simultaneous alterations in the oncogenesis of liposarcomas.J Cancer Res Clin Oncol 124:532–540.

10. Italiano A, et al. (2008) HMGA2 is the partner of MDM2 in well-differentiated anddedifferentiated liposarcomas whereas CDK4 belongs to a distinct inconsistent am-plicon. Int J Cancer 122:2233–2241.

11. Debelenko LV, et al. (2010) p53+/mdm2- atypical lipomatous tumor/well-differenti-ated liposarcoma in young children: an early expression of Li-Fraumeni syndrome.Pediatr Dev Pathol 13:218–224.

12. Pedeutour F, et al. (1999) Structure of the supernumerary ring and giant rod chro-mosomes in adipose tissue tumors. Genes Chromosomes Cancer 24:30–41.

13. Barretina J, et al. (2010) Subtype-specific genomic alterations define new targets forsoft-tissue sarcoma therapy. Nat Genet 42:715–721.

14. Snyder EL, et al. (2009) c-Jun amplification and overexpression are oncogenic in lip-osarcoma but not always sufficient to inhibit the adipocytic differentiation pro-gramme. J Pathol 218:292–300.

15. Liu P, Cheng H, Roberts TM, Zhao JJ (2009) Targeting the phosphoinositide 3-kinasepathway in cancer. Nat Rev Drug Discov 8:627–644.

16. Marsh DJ, et al. (1999) PTEN mutation spectrum and genotype-phenotype correla-tions in Bannayan-Riley-Ruvalcaba syndrome suggest a single entity with Cowdensyndrome. Hum Mol Genet 8:1461–1472.

17. Hernando E, et al. (2007) The AKT-mTOR pathway plays a critical role in the de-

velopment of leiomyosarcomas. Nat Med 13:748–753.18. Berghmans S, et al. (2005) tp53 mutant zebrafish develop malignant peripheral nerve

sheath tumors. Proc Natl Acad Sci USA 102:407–412.19. Tan Y, et al. (2008) A novel recurrent chromosomal inversion implicates the homeo-

box gene Dlx5 in T-cell lymphomas from Lck-Akt2 transgenic mice. Cancer Res 68:

1296–1302.20. Langenau DM, et al. (2007) Effects of RAS on the genesis of embryonal rhabdo-

myosarcoma. Genes Dev 21:1382–1395.21. Hay N, Sonenberg N (2004) Upstream and downstream of mTOR. Genes Dev 18:

1926–1945.22. Maira SM, et al. (2008) Identification and characterization of NVP-BEZ235, a new

orally available dual phosphatidylinositol 3-kinase/mammalian target of rapamycin

inhibitor with potent in vivo antitumor activity. Mol Cancer Ther 7:1851–1863.23. Carnevalli LS, et al. (2010) S6K1 plays a critical role in early adipocyte differentiation.

Dev Cell 18:763–774.24. Westerfield M (1994) The Zebrafish Book: A Guide for the Laboratory Use of Zebrafish

(Brachydanio rerio) (University of Oregon Press, Eugene, OR); 2.1 Ed.25. Jessen JR, Jessen TN, Vogel SS, Lin S (2001) Concurrent expression of recombination

activating genes 1 and 2 in zebrafish olfactory sensory neurons. Genesis 29:156–162.26. Gutierrez A, et al. (2011) Pten mediates Myc oncogene dependence in a conditional

zebrafish model of T cell acute lymphoblastic leukemia. J Exp Med 208:1595–1603.

Gutierrez et al. PNAS | September 27, 2011 | vol. 108 | no. 39 | 16391

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

28, 2

021