Epigenetic targeting of Hedgehog pathway ...

ARTICLES

Epigenetic targeting of Hedgehog pathway transcriptional output through BET bromodomain inhibition

© 2014 Nature America, Inc. All rights reserved.

Yujie Tang1,2, Sharareh Gholamin2, Simone Schubert1,2, Minde I Willardson3, Alex Lee4, Pratiti Bandopadhayay5–8, Guillame Bergthold5–7, Sabran Masoud1, Brian Nguyen1, Nujsaubnusi Vue1, Brianna Balansay1, Furong Yu1,2, Sekyung Oh1,2, Pamelyn Woo1, Spenser Chen1, Anitha Ponnuswami1, Michelle Monje1, Scott X Atwood4, Ramon J Whitson4, Siddhartha Mitra2, Samuel H Cheshier2, Jun Qi9, Rameen Beroukhim5,8,9, Jean Y Tang4, Rob Wechsler-Reya10, Anthony E Oro4, Brian A Link3, James E Bradner5,8,9 & Yoon-Jae Cho1,2,11 Hedgehog signaling drives oncogenesis in several cancers, and strategies targeting this pathway have been developed, most notably through inhibition of Smoothened (SMO). However, resistance to Smoothened inhibitors occurs by genetic changes of Smoothened or other downstream Hedgehog components. Here we overcome these resistance mechanisms by modulating GLI transcription through inhibition of bromo and extra C-terminal (BET) bromodomain proteins. We show that BRD4 and other BET bromodomain proteins regulate GLI transcription downstream of SMO and suppressor of fused (SUFU), and chromatin immunoprecipitation studies reveal that BRD4 directly occupies GLI1 and GLI2 promoters, with a substantial decrease in engagement of these sites after treatment with JQ1, a small-molecule inhibitor targeting BRD4. Globally, genes associated with medulloblastoma-specific GLI1 binding sites are downregulated in response to JQ1 treatment, supporting direct regulation of GLI activity by BRD4. Notably, patient- and GEMM (genetically engineered mouse model)-derived Hedgehog-driven tumors (basal cell carcinoma, medulloblastoma and atypical teratoid rhabdoid tumor) respond to JQ1 even when harboring genetic lesions rendering them resistant to Smoothened antagonists. Altogether, our results reveal BET proteins as critical regulators of Hedgehog pathway transcriptional output and nominate BET bromodomain inhibitors as a strategy for treating Hedgehog-driven tumors with emerged or a priori resistance to Smoothened antagonists. The Hedgehog (Hh) pathway is an evolutionarily conserved signaling axis that directs embryonic patterning through strict temporal and spatial regulation of cell proliferation and differentiation1. Developmental aberrations in Hh signaling result in dysmorphology, such as cyclopism, holoprosencephaly and limb deformity, when its output is absent or decreased2 and in cancer predisposition, as is seen in nevoid basal cell carcinoma syndrome (Gorlin syndrome)3, when its output is increased or unchecked1,4. In canonical Hh signaling, several morphogens (sonic hedgehog (SHH), Indian hedgehog (IHH) and desert hedgehog (DHH))5,6 have been identified that bind to the multipass cell-surface receptor Patched (PTCH1)1. When not bound by Hh ligand, PTCH1 inhibits the G protein–coupled receptor, SMO7. Once bound by ligand, however, PTCH1 no longer inhibits SMO, allowing SMO to positively regulate mobilization of the otherwise latent zinc finger transcription factor GLI2, residing in the cilia, to the nucleus, where GLI2 transactivates the GLI1 promoter8–10. GLI1 and GLI2 directly transactivate

transcription of Hh target genes, several of which are involved in proliferation, such as MYCN and CCND1 (ref. 11). GLI1 also serves to amplify the output of Hh signaling in a positive feedback loop by activating transcription of GLI2, albeit indirectly12. Ultimately, the transcriptional programs mediated by Hh signaling orchestrate an array of events based on cellular, temporal and spatial context, with perhaps the most phenotypically consequential event being an increase in cell proliferation. Inappropriate activation of Hh signaling results in tumor formation in several tissue lineages, including skin, brain, muscle, breast and pancreas13–15. The tumors most commonly associated with aberrant Hh signaling are basal cell carcinoma (BCC) and medulloblastoma, given their prevalence in individuals with germline mutations in PTCH1 (Gorlin syndrome)3,4. However, the overwhelming majority of Hh-driven BCCs and medulloblastomas activate Hh signaling through sporadic somatic mutations in PTCH1 or other components of the Hh pathway14,16,17. These include activating mutations in SMO

1Department

of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, California, USA. 2Department of Neurosurgery, Stanford University School of Medicine, Stanford, California, USA. 3Department of Cell Biology, Neurobiology and Anatomy, Medical College of Wisconsin, Milwaukee, Wisconsin, USA. 4Program in Epithelial Biology, Stanford University School of Medicine, Stanford, California, USA. 5Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA. 6Pediatric Neuro-oncology, Department of Pediatric Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA. 7Division of Pediatric Hematology/Oncology, Boston Children’s Hospital, Boston, Massachusetts, USA. 8Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA. 9Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA. 10Sanford-Burnham Medical Research, La Jolla, California, USA. 11Stanford Cancer Institute, Stanford University Medical Center, Stanford, California, USA. Correspondence should be addressed to Y.-J.C. ([email protected]). Received 18 March; accepted 29 May; published online 29 June 2014; doi:10.1038/nm.3613

NATURE MEDICINE

ADVANCE ONLINE PUBLICATION

1


ARTICLES or inactivating mutations in SUFU, which negatively regulates Hh output downstream of SMO17,18. Genomic amplification of GLI2, and more rarely GLI1, has also been reported and is associated with a more aggressive clinical course16,19–21. In addition, noncanonical activation of the Hh pathway can occur through loss of SMARCB1, a component of the SWI/SNF chromatin remodeling complex, which results in derepression of transcriptional activity at the GLI1 locus in malignant rhabdoid tumors22. Similarly, the EWS-FLI fusion oncogene responsible for Ewing sarcoma has been shown to directly transactivate the GLI1 promoter23. The identification of SMO as the main pharmacological target of cyclopamine24, a natural compound found in wild corn lily (Veratrum californicum)2, fostered the development of clinically optimized compounds with potent activity against SMO25–27. Some of these compounds have shown clinical efficacy against BCC, medulloblastoma and other cancers28–30. However, emergence of resistance and a priori resistance have been encountered25,29,31, prompting investigations into alternate strategies targeting new sites on SMO and Hh pathway components downstream of SMO32,33 or signaling pathways that cooperate with Hh activation in development and disease 25,34,35. High-throughput screens have also identified scaffolds that regulate GLI processing and its translocation to or from the cilia and nucleus36. However, the effectiveness of these strategies against Hh-driven cancers with MYCN amplification, such as SHH-subtype medulloblastomas, is unclear, as MYCN appears to be epistatic to the targets of many of these drugs. A new class of drugs targeting BET bromodomain proteins (BRD2– BRD4 and BRDT) was described recently37. Bromodomains recognize and bind to E-N-lysine acetylation motifs on open chromatin, such as those found on K27 residues of H3 histone N-terminal tails 38,39. The BET proteins also interact with the positive transcription elongation factor (P-TEFb)40,41 and phosphorylate Ser2 of RNA polymerase II (PolII), facilitating gene transcription at ‘super-enhancer’ sites across the genome42,43. BRD-containing complexes that bind at these super-enhancer sites often localize to promoter regions of key transcription factors such as MYC, and disruption of these complexes by BET inhibitors has produced substantial responses in mice bearing xenografts of treatment-refractory cancers driven by MYC and other previously ‘untargetable’ oncogenes, with limited or no toxicity to normal tissues44–47. Here we aimed to identify whether inhibition of BET bromodomain proteins could provide a strategy for treating Hh-driven tumors, including those resistant to SMO antagonists. We provide evidence that BRD4 is a critical regulator of GLI1 and GLI2 transcription through direct occupancy of their promoters. Furthermore, we show that occupancy of GLI1 and GLI2 promoters by BRD4 and transcriptional activation at cancer-specific GLI promoter–binding sites are markedly inhibited by the BET inhibitor JQ1. In GEMM- and patient-derived tumors with constitutive Hh pathway activation, JQ1 effectively decreases tumor cell proliferation and viability in vitro and in vivo, even when genetic lesions conferring resistance to SMO inhibition (SMOi) are present. Notably, the inhibition of cell proliferation by JQ1 can be rescued by GLI2 expression driven by a plasmid-based cytomegalovirus (CMV) promoter, which, in contrast to endogenous GLI promoters, is not under direct transcriptional regulation by BET proteins. In sum, our study identifies BET proteins as epigenetic regulators of Hedgehog transcriptional output and establishes a rationale for the use of BET inhibitors in cancers with evidence of Hh pathway activation.

2

RESULTS BRD4 is required for ligand-induced Hh transcriptional output The BET protein BRD4 enhances the transcription of key genes involved in embryonic stem cell maintenance42 and oncogenesis43. Therefore, we hypothesized that BRD4 is a transcriptional cofactor for Hh-responsive genes. In the mouse 3T3 cell–based Hh-Light2 reporter line containing a stably integrated Gli-luciferase reporter construct48, ligand-induced activation of Hh-Light2 cells with either Shh-N conditioned medium (CM)49 or Smoothened agonist (SAG)48 resulted in an expected increase in Gli1-luciferase activity and Gli1 mRNA levels, which were both potently inhibited by increasing doses of the BET inhibitor JQ1 (Fig. 1a and Supplementary Fig. 1). Upregulation of other Hh target genes such as Ptch1 and Gli2 was also inhibited by JQ1 (Fig. 1b). In contrast, Smo expression was modestly influenced, and expression of Sufu and Brd4 was not substantially altered by JQ1 (Fig. 1b). Notably, the inhibition of Gli1 expression by JQ1 equaled that by SMO inhibitors (GDC-0449, LDE225 or SANT-1) (Fig. 1c,d). Additionally, shRNA-mediated knockdown of Brd4 in Hh-Light2 cells followed by Shh-N CM or SAG stimulation resulted in marked inhibition of ligand-induced Gli-luciferase activity and Hh target gene expression, directly supporting an essential role of Brd4 in Hh signaling (Fig. 1e,f). To further assess inhibition of Hh transcriptional output by JQ1, we used zebrafish harboring a ptc2:GFP reporter transgene, a welldescribed canonical Hh pathway reporter in zebrafish50,51. Embryos exposed to JQ1 from 2 to 30 hours post fertilization (hpf) showed decreased expression of GFP mRNAs, similar to the results seen in cyclopamine-exposed fish (Fig. 1g). We also assessed whether JQ1 could revert abnormal phenotypes caused by aberrant Hh signaling in a temperature-sensitive transgenic fish line harboring an hsp70l: Shha–enhanced GFP (eGFP) transgene51, which overexpresses Shh and produces a reliable and well-described dysgenic eye phenotype that often includes a ventral coloboma, a structural defect in the eye resulting from improper closure of gaps located between various eye structures during embryonic development52,53. As predicted, heatshocked transgenic fish treated with vehicle alone (DMSO) developed abnormally shaped eyes with diminished diameter relative to their heat-shocked nontransgenic siblings (Fig. 1h). However, fish exposed to JQ1 immediately after heat shock trended toward more normalappearing eyes with statistically significant increases in eye diameter, suggesting that BET inhibition countered the effects of aberrant Hh signaling in vivo in this model (Fig. 1h). BRD4 regulates Hh signaling at Gli1 and Gli2 promoters We next examined the effects of JQ1 on Hh signaling in Sufu−/− mouse embryonic fibroblasts (MEFs) 54 and Hh-Light2 cells overexpressing GLI2. SUFU positively regulates the degradation of GLI proteins54, and thus loss of SUFU activity results in stabilization of GLI and constitutive Hh signaling downstream of SMO. As expected, we observed markedly increased Gli1 mRNA and protein levels in Sufu−/− MEFs, which were substantially downregulated by JQ1 (Fig. 2a,c and Supplementary Fig. 2a,b). We also noted decreased transcription of Gli2, as well as Smo to a lesser extent, after JQ1 treatment, whereas Brd4 mRNA levels remained unchanged (Fig. 2a). In stark contrast to JQ1 treatment, we observed little to no effect on Gli transcripts or Gli1 protein levels in Sufu−/− MEFs after treatment with the SMO inhibitors (LDE225, GDC-0449 or SANT-1) (Fig. 2b,c). Consistent with pharmacological inhibition of Brd4, shRNA-mediated knockdown of Brd4 in Sufu−/− MEFs resulted in


NATURE MEDICINE

ARTICLES

Shh-N/SAG

h

+

+

–

+

+

+

+

+

M SO M SO G D JQ1 C -0 4 LD 49 E2 S A 25 N T1

+

JQ 1 4 LD 49 E2 SA 25 N T1

+

-0

C

G D

D

+

D

–

M SO M SO

Sm

o

4 rd B

1

0

ch

Anti-Gli1

DMSO

Tg(ptc2:GFP)

EtOH

3

Anti–-tub

Anti–-tub

D

D

g

6

Pt

sh Mo Br ck sh d4Br 1 d4 sh 2 C trl

Anti–-tub

9

li2

Anti-Brd4

DMSO + shCtrl Shh-N + shCtrl Shh-N + shBrd4-1 Shh-N + shBrd4-2 SAG + shCtrl SAG + shBrd4-1 SAG + shBrd4-2

G

d4

d4

sh

Br

Br

trl C

+

-2

+

-1

+

sh

sh

C

trl

–

120 100 80 60 40 20 12

li1

f

10

Anti-Gli1

Shh-N SAG

JQ1 0.6 M

fu Su

rd 4

1

G li1

400 300 200 100 25 20 15 10 5 0 Shh-N/SAG

Fraction of fish with decreased GFP 2/17 (11.8%)

DMSO (veh)

Cyclopamine 25 M

5

d Relative Gli1 mRNA levels

10

G

20

0 Shh-N/SAG

EtOH (veh)

2/13 (15.4%)

JQ1 (0.6 M)

10/19 (52.6%)*

Cyclopamine (25 M)

Heat shock; Tg(hsp70l:Shha-eGFP)

9/12 (75%)*

250

)

M

SO

70

sp

Tg

(h

N

on

tra

l:S

ns

hh

a-

ge

ni

eG

c

si

FP

b

)

JQ

1

(0

D

.6

M

M

.6

JQ

1

(0

D

DMSO

M

SO

)

Eye diameter (mm)

JQ1 0.6 M

DMSO

P = 0.0025 Figure 1 BRD4 is necessary for GLI transcription. (a) Gli-luciferase 200 reporter activity in Hh-Light2 cells treated with Hh ligand (Shh-N CM or SAG) alone or in combination with increasing amount of JQ1. Data 150 represent the mean of triplicates o s.d. (b) qRT-PCR of Hh target genes (Gli1, Gli2 and Ptch1), Hh pathway components (Sufu and Smo) and 100 Brd4 in Hh-Light2 cells treated with Hh ligand (Shh-N CM or SAG) 50 alone or in combination with JQ1 (1 MM). Data represent the mean of triplicates o s.d. (c) qRT-PCR of Gli1 mRNA levels in Hh-Light2 cells n = 12 n = 12 n = 12 n = 12 Heat shock; nontransgenic sibling 0 treated with Hh ligand (Shh-N CM or SAG) alone or in combination with 1 MM of JQ1, GDC-0449, LDE225 or SANT-1. Data represent the mean of triplicates o s.d. (d) Immunoblot detecting Gli1 expression in cell lysates from Hh-Light2 cells treated with Hh ligand (Shh-N CM or SAG) alone or in combination with 1 MM of JQ1, GDC-0449, LDE225 or SANT-1. An anti–B-tubulin (B-tub) immunoblot is shown as a loading control. The immunoblot shown represents a typical result from an experiment performed in duplicate. (e) Gli-luciferase reporter activity in Hh-Light2 cells treated with Hh ligand (Shh-N CM or SAG) in combination with shRNAs against Brd4 (shBrd4-1 and shBrd4-2) or scrambled shRNA (shCtrl). Data represent the mean of quadruplicates o s.d. (f) qRT-PCR of Hh target genes (Gli1, Gli2 and Ptch1), Hh pathway components (Sufu and Smo) and Brd4 after treatment with Hh ligand (Shh-N CM or SAG) alone or in combination with shBrd4-1, shBrd4-2 or shCtrl. Data represent the mean of triplicates o s.d. (g) In situ hybridization detecting GFP mRNA levels in transgenic (Tg) zebrafish (ptc2:GFP) treated with JQ1 (0.6 MM), cyclopamine (25 MM) or vehicle (veh) controls (DMSO or EtOH). The fraction of zebrafish with decreased GFP expression is shown. Fisher’s exact test was used for statistical analysis. *P < 0.05. Scale bar, 100 Mm. (h) Images of a heat-shocked + Tg(hsp70l:Shha-eGFP) zebrafish or a nontransgenic sibling treated with JQ1 (0.6 MM) or DMSO. The eye diameter of each group (n = 12) was measured and is shown. Data represent the group means o s.d. The P value shown was generated using Student’s t test. Scale bar, 100 Mm. JQ1 0.6 M


15

Sm o

30

100

c

0

Shh-N SAG

sh

Relative luciferase units

40

200

B

+

300

Pt ch

+

DMSO JQ1 SAG + DMSO SAG + JQ1 Shh-N + DMSO Shh-N + JQ1

G li2

+

1

0

+

0. 01 0. 05 0. 2

+

400

Shh-N

–

0

10

b Relative mRNA levels

20

0 Shh-N/SAG JQ1 (M)

e

Shh-N SAG

SAG

Relative luciferase units

30

Relative mRNA levels

a

decreased Gli1 and Gli2 mRNA levels (Fig. 2d). It is worth noting that Brd4 knockdown did not abrogate GLI-luciferase activity or Gli expression as effectively as did JQ1 treatment. This result could be explained by incomplete knockdown of Brd4, or it could suggest that other BET proteins (all targets of JQ1) may also contribute to the transcriptional regulation of Gli genes. Indeed, knockdown of either Brd2 or Brd3 resulted in a substantial decrease of Gli mRNA levels in Sufu−/− MEFs (Supplementary Fig. 2c). In Hh-Light2 cells, forced expression of full-length mouse Gli2 (hemagglutinin (HA)-Gli2-FL) or an N-terminally truncated active form of human GLI2 (Myc-GLI2-DN)55 resulted in an increase in Gli1 mRNA levels, which was inhibited by JQ1 but not SMO inhibitors (GDC-0449, LDE225 or SANT-1) (Fig. 2e). Notably, we did not observe any decrease in ectopic GLI2 expression driven by the CMV promoter expression construct after JQ1 treatment, in contrast to the marked decrease in endogenous Gli transcripts (Figs. 1b and 2f). Additionally, upregulation of Ptch1, another Hh target gene, was not inhibited by

NATURE MEDICINE


JQ1, suggesting that not all Hh target genes are directly dependent on Brd4, as Gli genes themselves are (Supplementary Fig. 2d). In Sufu−/− cells, JQ1 decreased Gli1 and Gli2 levels as early as 3 h after treatment, supporting a role for Brd4 as a transcriptional cofactor that directly regulates transactivation of Gli promoters (Supplementary Fig. 2e). Chromatin immunoprecipitation followed by quantitative PCR (ChIP-qPCR) using antibody to Brd4 of regions flanking the transcription start sites of Gli1 and Gli2 promoters confirmed increased Brd4 occupancy at both Gli promoters after SAGmediated activation of Hh signaling in Hh-Light2 cells (Fig. 2g,h). Accordingly, ChIP-qPCR with antibody to PolII showed engagement of both Gli promoters by PolII after SAG stimulation. Notably, both Brd4 and PolII interactions at the Gli promoters were blocked by the addition of JQ1 (Fig. 2g,h). Similarly, in Sufu−/− MEFs, we observed increased baseline occupancy of Gli promoters by Brd4 and PolII relative to that in wild-type (WT) MEFs, which was markedly inhibited by JQ1 (Fig. 2i,j).

3

ARTICLES mice. JQ1 treatment resulted in marked downregulation of Gli mRNA and protein expression with little to no effect on Smo, Sufu or Brd4 (Fig. 3a–d and Supplementary Fig. 3a). Again, we observed a rapid decrease of Gli gene expression after JQ1 treatment (as early as 3 h), supporting a direct effect of BET inhibition on Gli promoters

JQ1 inhibits Ptch-deficient medulloblastoma and BCC We investigated the efficacy of JQ1 in Hh-driven tumors using cell lines derived from autochthonous medulloblastomas (SmoWT-MB and Med1-MB) arising in Ptch+/−; Trp53−/− and Ptch+/−; lacZ mice, respectively32,56, and BCC (ASZ001)57, also derived from Ptch+/−

Anti–-tub

e

G li2

G li1

Sm o

B rd 4

G li2

G li1

0

HA-Gli2-FL 8

10

Anti–-tub

Relative DNA levels

20

Anti-Myc Myc-GLI2-DN +

+

SO JQ C 1 -0 4 LD 49 E2 SA 25 N T1

SO

+

–1

D

G

N SA

+310

Gli1–TSS-2 6 4 2

0 Anti-Brd4 Anti-PolII SAG JQ1

i

Gli2 promoter –290

Gli1–TSS-1

M

1

M

25

+

D

9 44 -0

C

LD

+

G

D

–

D

+

T-

+

E2

+

D

M

SO

SO M D

+

1

Gli2-DN/FL +

TSS-2

Hh-Light2

Anti–-tub

30

h

+250

TSS-1

Anti-HA

40

–

–1

A TG

HA-Gli2-FL

JQ

Relative Gli1 mRNA levels

Gli1 promoter –100

Myc-GLI2-DN

50

0 Gli2-DN/FL

0.5

0

g

f Hh-Light2

1.0

B rd 4

0.5

0

+

+

+

– –

+ –

+ +

+ – –

+ + –

+ + +

j

ATG

Hh-Light2 Gli2–TSS-2

4 3 2 1

0 Anti-Brd4 Anti-PolII SAG JQ1

+ – –

+ + –

+ + +

+ – –

+ + –

+ + +

Gli1–TSS-1

Gli2–TSS-1

Gli1–TSS-2

Gli2–TSS-2

20

Gli2–TSS-1 Relative DNA levels

5

TSS-2

12 Relative DNA levels

TSS-1

Relative DNA levels


Anti-Gli1

shCtrl shBrd4-1 shBrd4-2

Sufu–/– MEF

G li2

1.0

1.5

D M SO

0.5

1.5

d

JQ 1

1.0

GDC-0449 LDE225 SANT-1

G li1



1.5

DMSO JQ1

Sufu–/– MEF

DMSO JQ1

c

Sufu–/– MEF 2.0


b

Sufu–/– MEF

0. 5 M JQ 1 G 1 D C M -0 44 9 LD 1 M E2 25 SA 1 M N T1 1 M

a

15 10 5

0 Anti-Brd4 Anti-PolII MEF JQ1

+ WT –

+

+ –/–

Sufu –

–/–

Sufu +

+ WT –

+ + Sufu–/– Sufu–/– – +

9 6 3

0 Anti-Brd4 Anti-PolII MEF JQ1

+ WT –

+

+

Sufu –

–/–

Sufu–/– +

+ WT –

+ + Sufu–/– Sufu–/– – +

Figure 2 JQ1 inhibits Hh target gene activation at the level of the GLI1 and GLI2 promoters. (a) qRT-PCR showing Gli1, Gli2, Smo and Brd4 mRNA levels in Sufu−/− MEFs treated with JQ1. Data represent the mean of triplicates o s.d. (b) Gli1 and Gli2 mRNA levels in Sufu−/− MEFs treated with DMSO, JQ1, GDC-0449, LDE225 or SANT-1. Data represent the mean of triplicates o s.d. (c) Immunoblot detecting GLI1 expression in cell lysates from Sufu−/− MEFs treated with DMSO, JQ1, GDC-0449, LDE225 or SANT-1. An anti–B-tubulin immunoblot is shown as a loading control. The immunoblots in c and f represent a typical result from each experiment performed in duplicate. (d) qRT-PCR showing Gli1, Gli2 and Brd4 mRNA levels in Sufu−/− cells expressing shBrd4-1, shBrd4-2 or shCtrl. Data represent the mean of triplicates o s.d. (e) qRT-PCR showing Gli1 mRNA levels in Hh-Light2 cells transiently transfected with HA-Gli2-FL or Myc-GLI2-DN and their responses to JQ1, GDC-0449, LDE225 or SANT-1. Data represent the mean of triplicates o s.d. (f) Anti-HA and anti-Myc immunoblots on cell lysates from Hh-Light2 cells transfected with HA-Gli2-FL or Myc-GLI2-DN and treated with DMSO, JQ1, GDC-0449, LDE225 or SANT-1. An anti–B-tubulin immunoblot is shown as a loading control. (g–j) Schematic of regions flanking the Gli1 and Gli2 promoter transcription start sites (TSS) analyzed by ChIP-qPCR of Brd4 and PolII occupancies in Hh-Light2 cells treated with SAG and JQ1 (g,h) and in Sufu−/− MEFs treated with JQ1 (i,j). Data represent the mean of triplicates o s.d. Except where indicated, cells were treated with 1 MM of JQ1, GDC-0449, LDE225 or SANT-1.

4


NATURE MEDICINE

ARTICLES

60

6

11

16

24

JQ1 (h)

0

6

24

48

Su fu

80

g Med1-MB

1,500

DMSO JQ1 0.1 M JQ1 0.2 M JQ1 0.5 M JQ1 1 M JQ1 2 M JQ1 5 M JQ1 10 M

1,000 500

80

SmoWT-MB

60 EdU+ (%)

f

h

0.6

EdU+/DAPI+

Sm

o

0

40 20

0.4 0.2

0

0 0

20

40 Time (h)

60

80

Med1-MB

0

SO JQ 48 h D 14 M SO 8 h JQ 24 1 h 24 h

40 Time (h)

Anti–-tub 0

D M

0 20

JQ1 (h)

24 D JQ1 h C -0 24 4 LD 49 h E2 24 25 h 24 h

50 0

Anti–-tub

0.5

SO

100

Anti-Gli1

D M

150

Med1-MB

Anti-Gli1

DMSO JQ1 0.001 µM JQ1 0.01 µM JQ1 0.1 µM JQ1 1 µM JQ1 10 µM GDC-0449 0.01 µM GDC-0449 0.1 µM LDE225 0.01 µM LDE225 0.1 µM

SmoWT-MB

d

G


Relative luciferase units (× 1,000)

e 200

1.0

G li1

fu Su

4

Sm o

rd B

G li2

0

c SmoWT-MB

B rd 4

0.5

DMSO JQ1

Med1-MB

G li2

1.0

1.5

Relative luciferase units ( 1,000)

1.5

b Relative mRNA levels

2.0

DMSO JQ1 GDC-0449 LDE225

SmoWT-MB

G li1


a

Figure 3 JQ1 inhibits Hh pathway activity and cell viability and proliferation in Ptch-mutated medulloblastoma (SmoWT-MB and Med1-MB) cells. (a,b) qRT-PCR of the expression of Hh pathway target genes (Gli1 and Gli2), components (Smo and Sufu) and Brd4 in SmoWT-MB and Med1-MB cells treated with JQ1 (1 MM), GDC-0449 (0.1 MM) or LDE225 (0.1 MM). Data represent the mean of triplicates o s.d. (c,d) Immunoblots detecting Gli1 expression in response to JQ1 treatment over time. An anti–B-tubulin immunoblot is shown as a loading control. The immunoblots represent a typical result from each experiment performed in duplicate. (e,f) Cell viability detection over time with increasing doses of JQ1 or SMO inhibitors. Data represent the group means o s.d. (g,h) Proliferative index in response to JQ1 (1 MM) or SMO inhibitors (GDC-0449 or LDE225 at 0.1 MM) as measured by EdU incorporation. Data represent the group means o s.d.

(Supplementary Fig. 3b). Accordingly, ChIP-qPCR using antibodies to Brd4 and PolII showed potent inhibition of Brd4 and PolII occupancy at Gli promoters in all cell lines after exposure to JQ1 (Supplementary Fig. 3c,d). In Med1-MB and SmoWT-MB cells, JQ1 treatment resulted in dose-responsive decreases in cell viability to a much greater extent than those observed in Hh-Light2 or Sufu−/− MEFs (Supplementary Fig. 4a). Potent growth inhibition was achieved (half-maximum inhibitory concentration (IC50) ~50–150 nM; Supplementary Fig. 4b,c) with marked decreases of proliferation (Fig. 3e–h), induction of apoptosis (Supplementary Fig. 4d,e) and, in Med1-MB cells, an increased fraction of cells in G1 and a decreased fraction of cells transitioning through S phase (Supplementary Fig. 4f). Notably, in SmoWT-MB cells, the inhibitory effects of JQ1 on Gli expression, cell viability and proliferation were equivalent to those of SMO inhibitors (GDC-0449 or LDE225) (Fig. 3a,e,g and Supplementary Figs. 3a and 4d), and these effects were enhanced when we exposed cells to both JQ1 and GDC-0449 in combination (Supplementary Fig. 4g). Using microarray analysis, we assessed changes in global gene expression in JQ1-treated SmoWT-MB cells compared with DMSO- and GDC0449-treated cells. We observed a substantial overlap between significantly differentially expressed genes (P < 0.0001) or gene sets (P < 0.0001; Supplementary Dataset) by JQ1 and GDC0449 in both cell lines compared with DMSOtreated controls, including the anticipated GLI target genes Gli2, Ptch1, Ccnd1, Ccnd2, Hhip and Cdk6 (Supplementary Fig. 5a–c). We next compared JQ1-induced gene expression profiles with gene sets derived from previously published ChIP-chip studies, which indexed gene promoters with Gli1-binding sites in normal granule neuron precursor cells (GNPs) and Ptch+/− medulloblastoma cells58. Specifically, we analyzed for enrichment of ChIP-chip peaks associated with GNPs, medulloblastoma, the overlap of both and peaks associated with GNPs alone or medulloblastoma alone

NATURE MEDICINE


(Supplementary Table 1). Gene set enrichment analysis (GSEA) revealed that only genes with Gli1 promoter–binding sites associated with medulloblastoma were significantly enriched (P < 0.0001) in JQ1-treated cells (Supplementary Fig. 5d). These results confirm the disruption of Gli1-mediated transcription by JQ1 and the preferential targeting of Gli1 transcriptional activity in tumor cells43. Ectopic GLI2 expression rescues growth inhibition by JQ1 We tested whether knockdown of Brd4 could phenocopy the effects of JQ1 in Hh-driven medulloblastoma cells. As expected, knockdown of Brd4 resulted in decreased Gli expression (Fig. 4a,b) and cell proliferation (Fig. 4c,d), suggesting that the inhibitory effect of JQ1 was through targeting of Brd4. Furthermore, to directly assess whether BET inhibition blocked cell proliferation in Hh-driven tumor cells through targeting of Gli transcription, we used plasmid-based expression of GLI2 (Myc-GLI2-DN; Fig. 2f) in SmoWT-MB cells and monitored its ability to rescue the inhibition of proliferation by JQ1 (Fig. 4c). Notably, ectopic expression of GLI2 inhibited the effects of JQ1 on 5-ethynyl-2`-deoxyuridine (EdU) incorporation, resulting in levels of EdU incorporation that were nearly equivalent to levels in DMSO-treated control cells (Fig. 4e,f). This result indicates that inhibition of proliferation by JQ1 is mediated largely through inhibition of Gli transcription and, intriguingly, that Brd4-independent transcriptional targets of Gli transcription factors are sufficient to overcome BET inhibition. SMOi-resistant Hh-driven tumors are inhibited by JQ1 Given the documented mechanisms of resistance to current, clinically available SMO inhibitors25,31 and the potential of BET inhibitors as a strategy to overcome this resistance, we examined the efficacy of JQ1 in Hh-driven cancers with either acquired or a priori resistance to SMO inhibitors (Fig. 5a). We analyzed the efficacy of JQ1 and SMO inhibitors (GDC-0449 and LDE225) against medulloblastoma cells

5

ARTICLES b 1.5

80

0.5

60 40

DMSO

or Myc-GLI2-DN; GFP

f

JQ1 0.1 µM

JQ1 0.1 µM

DMSO

0.4

sh C trl sh Br d4 -1 sh Br d4 -2

2

1 sh

+10 mM EdU

JQ1 0.5 µM

50.3%


4-

4-

C t rl

16 h

24 h

shCtrl shBrd4-1 shBrd4-2

0

sh

d4 Br

G li1

0 sh

Empty vector; GFP

Med1-MB

0.2

20

B rd 4

G li2

G li1

e

0.8

0.6 Edu+/DAPI+

1.0

0

0

d

Br d

0.5

SmoWT-MB

100

Br d

1.0

c

Med1-MB

EdU+ (%)

SmoWT-MB

G li2


1.5


a

29.5%

8h

EdU staining FACS analysis

JQ1 0.5 µM 23.0% Vector

GFP+ 250K

51.7%

47.5%

45.7%

200K

SSC-A

72.5%

Myc-GLI2-DN

13.1%

150K 100K 50K 53.1%

36.7%

27.7%

0 100

101

102

103

104

105

Vector

GFP GFP–

Figure 4 On-target validation of the inhibitory effects of JQ1 in Hh-driven 55.1% 35.4% 27.9% tumor cells. (a–d) qRT-PCR of Gli1, Gli2 and Brd4 mRNAs (a,b) and proliferative index Myc-GLI2-DN determined by EdU incorporation (c,d) in shBrd4-1–, shBrd4-2 or shCtrl-expressing medulloblastoma tumor cells (SmoWT-MB and Med1-MB). Data represent the group means o s.d. (e) The schematic workflow of the GLI2-overexpressing rescue experiment in SmoWT-MB cells. (f) FACS analysis of EdU incorporation in SmoWT-MB cells transfected with empty vector or Myc-GLI2-DN followed by JQ1 treatment (0.1 or 0.5 MM). GFP-expressing plasmid was used for co-transfection to mark the transfected (GFP +) cells. SSC-A, side scatter.

carrying an aspartate-to-glycine substitution at amino acid residue 477 in Smo that results in decreased sensitivity to SMO antagonists (SmoD477G-MB) (Fig. 5b)32; patient-derived SUFU-mutated primary SHH-subtype medulloblastoma cells (RCMB025); patient-derived primary atypical teratoid rhabdoid tumor (ATRT) cells (CHB_ATRT1 and SU_ATRT2) with derepression of GLI1 transcription through loss of SMARCB1 (also called SNF5 or INI1) (ref. 22); and patientderived MYCN-amplified primary SHH-subtype medulloblastoma cells (RCMB018). Cell viability (Fig. 5b–f, top), Gli and GLI levels (Fig. 5b–f, bottom) and EdU incorporation (Supplementary Fig. 6a–c) were markedly decreased in response to JQ1 in all of these

6

cells, and we observed little or no effect with the SMO inhibitors GDC-0449 and LDE225. Additionally, we examined Myc, MYC, Mycn and MYCN expression in SmoWT-MB, SmoD477G-MB, RCMB025, CHB_ATRT1 and RCMB018 cells and found that Mycn and MYCN expression was consistently inhibited by JQ1 (Fig. 5f, bottom and Supplementary Figs. 4h and 6d–f), suggesting that JQ1 targets at least two important driver oncogenes (GLI and MYCN) in these tumors. In vivo inhibition of Hh-driven tumors by JQ1 To support a therapeutic role for BET inhibition in Hh-driven tumors, we assessed the in vivo efficacy of JQ1 against medulloblastomas and


NATURE MEDICINE

ARTICLES

0

GLI1 MYC

0.5

25

1

E2

44 D

G

G

LD

D

M

E2

9

0 SO

25

1 LD

D

JQ

SO

0.5

BRD4

1.0

-0

1.0

C


1.5

GLI2 MYCN

JQ

Relative cell viability

0.5

BRD4

2.0

0

1.0

1.5

M

9 D

GLI2

RCMB018 1.5

JQ D M 1 S 1 O G J JQ1 0 µ D Q C 1 1 M 0 G 0 µM G DC 449 .1 µ D -0 1 M C 4 0 -0 4 µ LD 44 9 1 M E 9 0 µM LD 225 .1 µ LD E 10 M E2 225 µ 25 1 M 0. µM 1 µM

0

JQ D 1 MS 1 O G J JQ 0 µ D Q 1 M C 1 1 G 04 0 µM D G C 49 .1 µ D -0 1 M C 4 0 -0 4 µ LD 44 9 1 M E 9 0 µM LD 225 .1 µ LD E 10 M E2 225 µ 25 1 M 0. µM 1 µM


0.5

25

1

1.0

GLI1

Relative mRNA levels M

E2

SO

0

G

SU_ATRT2 1.5

BRD4

0.5

D

D

GLI2

1.0

25

9


GLI1

1.5

LD

1

SO M D

LD

44

Relative mRNA levels 9

25 E2

44

JQ

-0

BRD4

2.5 2.0 1.5 1.0 0.5 0

G

D

C

D

M

1

0 SO


0.5

0

MYCN amplification

f

Figure 5 JQ1 inhibits Hh pathway activity and cell viability and proliferation in SMOi-resistant Hh-driven tumors. (a) Schematic depicting mechanisms of resistance to Smoothened antagonists in Hh-driven cancers. (b–f, top) Cell viability in SMOi-resistant medulloblastoma cells (SmoD477G-MB; b), patient-derived SUFU mutant medulloblastoma cells (RCMB025; c), patient-derived ATRT cells (CHB_ATRT1 and SU_ATRT2; d,e) and patient-derived MYCN-amplified medulloblastoma cells (RCMB018; f) treated with increasing doses of JQ1, GDC-0449 or LDE225. Data represent the group means o s.d. (b–f, bottom) qRT-PCR of Gli1, GLI1, Gli2, GLI2, Brd4 and BRD4 (plus MYC and MYCN levels for RCMB018) in SmoD477G-MB (b), RCMB025 (c), CHB_ATRT1 (d), SU_ATRT2 (e) and RCMB018 (f) cells in response to JQ1 (1 MM), GDC-0449 or LDE225 (0.1 MM for SmoD477G-MB and 1 MM for the other groups). Data represent the mean of triplicates o s.d.

b Percentage survival

Ctrl (n = 5) JQ1 (n = 5)

800 600 400

P = 0.0160

200

c

Ctrl (n = 5) JQ1 (n = 5)

Tumor volume (mm3)

a

100

50 P = 0.0017 0

0 0

7

14 21 Treatment (d)

28

cells with vehicle control, JQ1 (50 mg per kg body weight per day i.p.) or GDC-0449 (100 mg per kg body weight per day orally (p.o.)). We observed marked decreases in the growth of SmoD477G-MB flank allografts in response to JQ1 but not GDC-0449, whereas SmoWT-MB flank allografts responded to both GDC-0449 and JQ1 (Fig. 6c,d). To evaluate the efficacy of JQ1 against BCCs in vivo, we used an allograft model of Ptch+/−; K14-creER2; p53flox/flox–derived mouse BCC cells34. JQ1 treatment (50 mg per kg body weight per day i.p.)

d 10,000

Ctrl (n = 8) JQ1 (n = 8) GDC-0449 (n = 8)

10,000 8,000 6,000

Tumor volume (mm3)

BCCs. We treated flank and intracranial allografts of Med1-MB cells stably expressing a firefly luciferase reporter in immunodeficient NSG mice with either JQ1 (50 mg per kg body weight per day intraperitoneally (i.p.)) or vehicle control. We observed a significant reduction in flank tumor growth in JQ1-treated mice, as well as an increase in overall survival in JQ1-treated mice harboring intracranial allografts (Fig. 6a,b and Supplementary Fig. 7a). Additionally, we treated medulloblastoma flank allografts of SmoWT-MB or SmoD477G-MB

Average radiance (× 1,000,000)

JQ1 vs. Ctrl, P < 0.0001 GDC-0449 vs Ctrl, P < 0.0001

4,000 2,000

8,000 6,000

10

20 30 Treatment (d)

40

50

JQ1 vs. Ctrl, P < 0.0001 JQ1 vs. GDC-0449, P < 0.0001

4,000 2,000 0

0 0

Ctrl (n = 8) JQ1 (n = 8) GDC-0449 (n = 8)

0

5 10 Treatment (d)

15

0

e

5 10 Treatment (d)

15

Ctrl (n = 4)

Figure 6 JQ1 inhibits Hh-driven tumor growth in vivo. (a,b) Med1-MB cells transduced with a lentiviral 15 JQ1 (n = 5) BMS-833293 (n = 10) luciferase reporter were used for flank (a) or cerebellum (b) injections of NSG mice, which were then 10 randomized for treatment with either JQ1 (50 mg per kg body weight daily i.p.) or vehicle. (a) Tumor growth JQ1 vs. Ctrl, P = 0.0346 of Med1-MB allografts assessed by IVIS imaging and presented as the average radiance. (b) Survival curve 5 of the mice injected with Med1-MB cells in the cerebellum. (c,d) Tumor growth, assessed by caliper measurement, after SmoWT-MB (c) and SmoD477G-MB (d) cells were injected into the flanks of NSG mice 0 (both flanks of each mouse) followed by treatment with JQ1 (50 mg per kg body weight daily i.p.), GDC-0449 −5 (100 mg per kg body weight daily p.o.) or vehicle. (e) Tumor growth, assessed by caliper measurement, 0 10 20 30 of SMOi-naive mouse BCC tumors generated under the dermis of NSG mice that were treated with JQ1 Treatment (d) (50 mg per kg body weight daily i.p.), BMS-833293 (100 mg per kg body weight daily i.p.) or vehicle. Data on tumor growth represent the group means o s.e.m. Two-way analysis of variance (ANOVA) was used for comparing tumor growth curves. Log-rank test was used for comparing survival curves. The results shown are from two separate experiments testing JQ1 and BMS-833293, independently, but are presented on the same graph. Average change in BCC length (mm)


1.0

GLI2

0.5

JQ

GLI1

1.0

C

Brd4

JQ

Gli2

CHB_ATRT1

44

0

e

1.5

-0

0.5

D

Gli1

1.5


1.0

C

0


0.5

RCMB025

JQ D M 1 S 1 O G J JQ1 0 µ D Q C 1 1 M G -04 0. µM G DC 49 1 µ D -0 1 M C 4 0 -0 4 µ LD 44 9 1 M E 9 0 µM LD 225 .1 µ LD E 10 M E2 225 µ 25 1 M 0. µM 1 µM

1.0

Derepression of GLI1 transcription

d

1.5

JQ M 1 SO JQ 10 µ G JQ 1 1 M G DC 1 0 µM D . G C-0 044 1 µ D C 44 9 1 M -0 9 µ 44 0. M LD 9 0 1 µ LD E2 .01 M LD E2 25 µM E2 25 1 µ 25 0.1 M 0. µM 01 µM


SmoD477G-MB SmoWT-MB

Hh target genes (GLI1, MYCN, CCND1, and so on)

GLI2

SUFU mutation

c

SmoD477G-MB 1.5

SUFU

JQ D M 1 S 1 O G J JQ1 0 µ D Q C 1 1 M G -04 0. µM G DC 49 1 µ D -0 1 M C 4 0 -0 4 µ LD 44 9 1 M E 9 0 µM LD 225 .1 µ LD E 10 M E2 225 µ 25 1 M 0. µM 1 µM

SMO mutation

b

SMO

E2

PTCH1

LD

SHH IHH DHH

-0

a

NATURE MEDICINE


7

ARTICLES


resulted in significant growth inhibition of BCCs but was not as effective as the clinically optimized SMO inhibitor BMS-833293 (ref. 34) (Fig. 6e). Nonetheless, in all Hh-driven tumor models tested, we observed reduction of Gli mRNA levels after JQ1 treatment regardless of whether allografts were sensitive or resistant to SMO inhibition (SMOi) (Supplementary Fig. 7b–f). Together these results demonstrate in vivo efficacy of JQ1 against Hh-driven tumors, even those with acquired or a priori resistance to clinically available SMO inhibitors. DISCUSSION We have shown that BRD4 and other BET proteins are critical regulators of GLI1 and GLI2 transcription and that BET inhibition provides a new therapeutic strategy against Hh-driven tumors. Notably, as BET proteins regulate the far-downstream transcriptional output of Hh signaling, BET inhibition was effective against tumor cells that evade Smoothened antagonists through mutation of SMO or amplification of nodes downstream of SMO. Our study is clinically relevant for patients who have a priori resistance to SMO inhibitors and in cases in which the emergence of resistance develops after an initial response to such therapy. By acting directly on the GLI1 and GLI2 promoters, BET inhibition circumvents all SMOi resistance mechanisms that have been reported so far, which include mutations of SMO or SUFU or amplifications in GLI2 or MYCN16,17,20,25,31. The response to JQ1 observed in MYCN-amplified SHH medulloblastoma cells (RCMB018), in terms of both decreased cell viability and MYCN levels, is similar to the results of a recent study showing the efficacy of BET inhibitors in MYCN-amplified neuroblastoma46. However, in Hh-driven tumors, it is likely that decreased MYCN levels in response to JQ1 treatment reflect the role of GLI in directly transactivating the MYCN promoter, in addition to the role of JQ1 in directly regulating expression of Mycn and MYCN. Given the importance of Hh signaling in normal development, it will be essential to understand and anticipate potential toxicities of BET inhibitor therapies as they enter into clinical trials. We observed developmental anomalies at very high doses of JQ1 in our zebrafish studies (data not shown), consistent with those seen in Brd4 heterozygous mice, which display a multitude of defects that overlap with cyclopamine-treated or Hh-deficient mice59,60. Of note, however, Brd4 heterozygous mice develop craniofacial but not overt axial skeletal phenotypes59, unlike cyclopamine-exposed embryos2,60, suggesting lineage-specific differences of Hh pathway dependency on Brd4. Our finding that plasmid-driven GLI2 expression can rescue the proliferation defect induced by JQ1 supports the existence of GLI-responsive promoters that do not require BRD4 for their transactivation. Notably, such genes appear to be either individually or collectively sufficient to mediate part, if not all, of the oncogenic phenotype associated with Hh-GLI signaling. Investigating how BRD4 regulates normal Hh-mediated biological processes and documenting BRD4-related changes that occur during Hh-mediated oncogenic transformation could potentially elucidate factors essential for tumor development that are independent of normal development. Our analysis of gene expression changes in JQ1-treated medulloblastoma supports observations by Lee et al.58, who identified marked shifts in Gli1 occupancy across the genome in medulloblastoma compared to GNPs. An unbiased characterization of Brd4 binding across the genome in GNPs and medulloblastomas will clarify whether the genomic footprint of Brd4 overlaps with Gli occupancy in the oncogenic state relative to the normal developmental state. Related to this point, emerging evidence suggests BET proteins

8

converge on super-enhancer sites across the genome and that these super-enhancer sites help transactivate promoters of key regulators of cellular identity in normal and pathogenic contexts42,43. Whether GLI transactivates super enhancer–related promoters and, accordingly, whether super-enhancer sites are positioned over GLI promoters is currently under active investigation. METHODS Methods and any associated references are available in the online version of the paper. Accession codes. Gene expression profiling data has been deposited into the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) with accession code GSE58185. Note: Any Supplementary Information and Source Data files are available in the online version of the paper. ACKNOWLEDGMENTS This work was supported the St. Baldrick’s Foundation Scholar Award (Y.-J.C.), a Beirne Faculty Scholar Endowment (Y.-J.C.), US National Institutes of Health (NIH) grant U01-CA176287 (Y.-J.C.), an Alex’s Lemonade Stand Foundation Young Investigator Award (Y.T.), the Damon-Runyon Cancer Research Foundation (J.Q. and J.E.B.), NIH R01-CA159859 (R.W.-R.) and a pilot project grant from the Medical College of Wisconsin Cancer Center–Advancing a Healthier Wisconsin (B.A.L.). R.W.-R. is the recipient of a Leadership Award (LA1-01747) from the California Institute of Regenerative Medicine. We thank the Stanford Functional Genomics Facility (SFGF) and the Protein and Nucleic Acid (PAN) facility for their assistance in generating gene expression microarray data and reagents. We thank M. Scott (Stanford), C. Rudin (Memorial Sloan-Kettering Cancer Center), P. Beachy (Stanford), A. Sweet-Cordero (Stanford), P.-T. Chang (University of California, San Francisco), R. Karlstrom (University of Massachusetts, Amherst) and J.K. Chen (Stanford) for reagents, helpful suggestions and/or critical reading of the manuscript. AUTHOR CONTRIBUTIONS Y.T. and Y.-J.C. conceived the project and wrote the manuscript. Y.T., B.N., S.M., B.B., N.V., S.S., S.C., A.P., S.O. and F.Y. performed all molecular biology experiments. J.Q. and J.E.B. synthesized and supplied JQ1 for all studies. A.E.O., S.X.A., R.J.W., A.L. and J.Y.T. generated and prepared GEMM-derived BCC cells, and R.W.-R. generated and provided patient-derived medulloblastoma cells. P.B., G.B., R.B. and Y.-J.C. performed all informatics analyses. S.G., A.L., Y.T., S.M., P.W., M.M., S.H.C. and S.S. performed JQ1 mouse in vivo studies. M.I.W. and B.A.L. performed all zebrafish studies. COMPETING FINANCIAL INTERESTS The authors declare competing financial interests: details are available in the online version of the paper. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Goodrich, L.V., Johnson, R.L., Milenkovic, L., McMahon, J.A. & Scott, M.P. Conservation of the hedgehog/patched signaling pathway from flies to mice: induction of a mouse patched gene by Hedgehog. Genes Dev. 10, 301–312 (1996). 2. Binns, W., Shupe, J.L., Keeler, R.F. & James, L.F. Chronologic evaluation of teratogenicity in sheep fed Veratrum californicum. J. Am. Vet. Med. Assoc. 147, 839–842 (1965). 3. Gorlin, R.J., Vickers, R.A., Kellen, E. & Williamson, J.J. Multiple nasal-cell nevi syndrome. An analysis of a syndrome consisting of multiple nevoid basal-cell carcinoma, jaw cysts, skeletal anomalies, medulloblastoma, and hyporesponsiveness to parathormone. Cancer 18, 89–104 (1965). 4. Johnson, R.L. et al. Human homolog of patched, a candidate gene for the basal cell nevus syndrome. Science 272, 1668–1671 (1996). 5. Echelard, Y. et al. Sonic hedgehog, a member of a family of putative signaling molecules, is implicated in the regulation of CNS polarity. Cell 75, 1417–1430 (1993). 6. Marigo, V. et al. Cloning, expression, and chromosomal location of SHH and IHH: two human homologues of the Drosophila segment polarity gene hedgehog. Genomics 28, 44–51 (1995). 7. Alcedo, J., Ayzenzon, M., Von Ohlen, T., Noll, M. & Hooper, J.E. The Drosophila smoothened gene encodes a seven-pass membrane protein, a putative receptor for the hedgehog signal. Cell 86, 221–232 (1996).


NATURE MEDICINE


ARTICLES 8. Huangfu, D. & Anderson, K.V. Cilia and Hedgehog responsiveness in the mouse. Proc. Natl. Acad. Sci. USA 102, 11325–11330 (2005). 9. Haycraft, C.J. et al. Gli2 and Gli3 localize to cilia and require the intraflagellar transport protein polaris for processing and function. PLoS Genet. 1, e53 (2005). 10. Liu, A., Wang, B. & Niswander, L.A. Mouse intraflagellar transport proteins regulate both the activator and repressor functions of Gli transcription factors. Development 132, 3103–3111 (2005). 11. Oliver, T.G. et al. Transcriptional profiling of the Sonic hedgehog response: a critical role for N-myc in proliferation of neuronal precursors. Proc. Natl. Acad. Sci. USA 100, 7331–7336 (2003). 12. Regl, G. et al. Human GLI2 and GLI1 are part of a positive feedback mechanism in basal cell carcinoma. Oncogene 21, 5529–5539 (2002). 13. Mao, J. et al. A novel somatic mouse model to survey tumorigenic potential applied to the Hedgehog pathway. Cancer Res. 66, 10171–10178 (2006). 14. Xie, J. et al. Mutations of the PATCHED gene in several types of sporadic extracutaneous tumors. Cancer Res. 57, 2369–2372 (1997). 15. Thayer, S.P. et al. Hedgehog is an early and late mediator of pancreatic cancer tumorigenesis. Nature 425, 851–856 (2003). 16. Cho, Y.J. et al. Integrative genomic analysis of medulloblastoma identifies a molecular subgroup that drives poor clinical outcome. J. Clin. Oncol. 29, 1424–1430 (2011). 17. Pugh, T.J. et al. Medulloblastoma exome sequencing uncovers subtype-specific somatic mutations. Nature 488, 106–110 (2012). 18. Taylor, M.D. et al. Mutations in SUFU predispose to medulloblastoma. Nat. Genet. 31, 306–310 (2002). 19. ten Haaf, A. et al. Expression of the glioma-associated oncogene homolog (GLI) 1 in human breast cancer is associated with unfavourable overall survival. BMC Cancer 9, 298 (2009). 20. Northcott, P.A. et al. Subgroup-specific structural variation across 1,000 medulloblastoma genomes. Nature 488, 49–56 (2012). 21. Buczkowicz, P., Ma, J. & Hawkins, C. GLI2 is a potential therapeutic target in pediatric medulloblastoma. J. Neuropathol. Exp. Neurol. 70, 430–437 (2011). 22. Jagani, Z. et al. Loss of the tumor suppressor Snf5 leads to aberrant activation of the Hedgehog-Gli pathway. Nat. Med. 16, 1429–1433 (2010). 23. Zwerner, J.P. et al. The EWS/FLI1 oncogenic transcription factor deregulates GLI1. Oncogene 27, 3282–3291 (2008). 24. Chen, J.K., Taipale, J., Cooper, M.K. & Beachy, P.A. Inhibition of Hedgehog signaling by direct binding of cyclopamine to Smoothened. Genes Dev. 16, 2743–2748 (2002). 25. Buonamici, S. et al. Interfering with resistance to smoothened antagonists by inhibition of the PI3K pathway in medulloblastoma. Sci. Transl. Med. 2, 51ra70 (2010). 26. Lee, M.J. et al. Hedgehog pathway inhibitor saridegib (IPI-926) increases lifespan in a mouse medulloblastoma model. Proc. Natl. Acad. Sci. USA 109, 7859–7864 (2012). 27. Zhang, Y., Laterra, J. & Pomper, M.G. Hedgehog pathway inhibitor HhAntag691 is a potent inhibitor of ABCG2/BCRP and ABCB1/Pgp. Neoplasia 11, 96–101 (2009). 28. LoRusso, P.M. et al. Phase I trial of hedgehog pathway inhibitor vismodegib (GDC0449) in patients with refractory, locally advanced or metastatic solid tumors. Clin. Cancer Res. 17, 2502–2511 (2011). 29. Rudin, C.M. et al. Treatment of medulloblastoma with hedgehog pathway inhibitor GDC-0449. N. Engl. J. Med. 361, 1173–1178 (2009). 30. Von Hoff, D.D. et al. Inhibition of the hedgehog pathway in advanced basal-cell carcinoma. N. Engl. J. Med. 361, 1164–1172 (2009). 31. Yauch, R.L. et al. Smoothened mutation confers resistance to a Hedgehog pathway inhibitor in medulloblastoma. Science 326, 572–574 (2009). 32. Kim, J. et al. Itraconazole and arsenic trioxide inhibit Hedgehog pathway activation and tumor growth associated with acquired resistance to smoothened antagonists. Cancer Cell 23, 23–34 (2013). 33. Kim, J., Lee, J.J., Gardner, D. & Beachy, P.A. Arsenic antagonizes the Hedgehog pathway by preventing ciliary accumulation and reducing stability of the Gli2 transcriptional effector. Proc. Natl. Acad. Sci. USA 107, 13432–13437 (2010). 34. Atwood, S.X., Li, M., Lee, A., Tang, J.Y. & Oro, A.E. GLI activation by atypical protein kinase C I/L regulates the growth of basal cell carcinomas. Nature 494, 484–488 (2013).

NATURE MEDICINE


35. Kenney, A.M., Widlund, H.R. & Rowitch, D.H. Hedgehog and PI-3 kinase signaling converge on Nmyc1 to promote cell cycle progression in cerebellar neuronal precursors. Development 131, 217–228 (2004). 36. Hyman, J.M. et al. Small-molecule inhibitors reveal multiple strategies for Hedgehog pathway blockade. Proc. Natl. Acad. Sci. USA 106, 14132–14137 (2009). 37. Filippakopoulos, P. et al. Selective inhibition of BET bromodomains. Nature 468, 1067–1073 (2010). 38. Dhalluin, C. et al. Structure and ligand of a histone acetyltransferase bromodomain. Nature 399, 491–496 (1999). 39. Jacobson, R.H., Ladurner, A.G., King, D.S. & Tjian, R. Structure and function of a human TAFII250 double bromodomain module. Science 288, 1422–1425 (2000). 40. Jang, M.K. et al. The bromodomain protein Brd4 is a positive regulatory component of P-TEFb and stimulates RNA polymerase II–dependent transcription. Mol. Cell 19, 523–534 (2005). 41. Yang, Z. et al. Recruitment of P-TEFb for stimulation of transcriptional elongation by the bromodomain protein Brd4. Mol. Cell 19, 535–545 (2005). 42. Whyte, W.A. et al. Master transcription factors and mediator establish superenhancers at key cell identity genes. Cell 153, 307–319 (2013). 43. Lovén, J. et al. Selective inhibition of tumor oncogenes by disruption of superenhancers. Cell 153, 320–334 (2013). 44. Delmore, J.E. et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 146, 904–917 (2011). 45. Mertz, J.A. et al. Targeting MYC dependence in cancer by inhibiting BET bromodomains. Proc. Natl. Acad. Sci. USA 108, 16669–16674 (2011). 46. Puissant, A. et al. Targeting MYCN in neuroblastoma by BET bromodomain inhibition. Cancer Discov. 3, 308–323 (2013). 47. Bandopadhayay, P. et al. BET bromodomain inhibition of MYC-amplified medulloblastoma. Clin. Cancer Res. 20, 912–925 (2014). 48. Chen, J.K., Taipale, J., Young, K.E., Maiti, T. & Beachy, P.A. Small molecule modulation of Smoothened activity. Proc. Natl. Acad. Sci. USA 99, 14071–14076 (2002). 49. Maity, T., Fuse, N. & Beachy, P.A. Molecular mechanisms of Sonic hedgehog mutant effects in holoprosencephaly. Proc. Natl. Acad. Sci. USA 102, 17026–17031 (2005). 50. Lewis, K.E., Concordet, J.P. & Ingham, P.W. Characterisation of a second patched gene in the zebrafish Danio rerio and the differential response of patched genes to Hedgehog signalling. Dev. Biol. 208, 14–29 (1999). 51. Shen, M.C. et al. Heat-shock–mediated conditional regulation of hedgehog/gli signaling in zebrafish. Dev. Dyn. 242, 539–549 (2013). 52. Macdonald, R. et al. Midline signalling is required for Pax gene regulation and patterning of the eyes. Development 121, 3267–3278 (1995). 53. Lee, J. et al. An ENU mutagenesis screen in zebrafish for visual system mutants identifies a novel splice-acceptor site mutation in patched2 that results in colobomas. Invest. Ophthalmol. Vis. Sci. 53, 8214–8221 (2012). 54. Chen, M.H. et al. Cilium-independent regulation of Gli protein function by Sufu in Hedgehog signaling is evolutionarily conserved. Genes Dev. 23, 1910–1928 (2009). 55. Roessler, E. et al. A previously unidentified amino-terminal domain regulates transcriptional activity of wild-type and disease-associated human GLI2. Hum. Mol. Genet. 14, 2181–2188 (2005). 56. Goodrich, L.V., Milenkovic, L., Higgins, K.M. & Scott, M.P. Altered neural cell fates and medulloblastoma in mouse patched mutants. Science 277, 1109–1113 (1997). 57. Aszterbaum, M. et al. Ultraviolet and ionizing radiation enhance the growth of BCCs and trichoblastomas in patched heterozygous knockout mice. Nat. Med. 5, 1285–1291 (1999). 58. Lee, E.Y. et al. Hedgehog pathway–regulated gene networks in cerebellum development and tumorigenesis. Proc. Natl. Acad. Sci. USA 107, 9736–9741 (2010). 59. Houzelstein, D. et al. Growth and early postimplantation defects in mice deficient for the bromodomain-containing protein Brd4. Mol. Cell. Biol. 22, 3794–3802 (2002). 60. Cooper, M.K., Porter, J.A., Young, K.E. & Beachy, P.A. Teratogen-mediated inhibition of target tissue response to Shh signaling. Science 280, 1603–1607 (1998).

9

ONLINE METHODS


Ethics statement. All studies were performed under approval and oversight by the Institutional Review Board committees of Stanford University, Boston Children’s Hospital and Rady Children’s Hospital/Sanford-Burnham Medical Research Institute. Cell lines and drug reagents. Mouse BCC (ASZ001), 293T and Hh-Light2 cells were derived and maintained as previously described24,34,57. RCMB025 and RCMB018 cells were derived from primary surgical resections of two medulloblastoma cases at Rady Children’s Hospital and were further characterized by whole-genome sequencing as having a SUFU mutation and MYCN amplification, respectively61. CHB_ATRT1 cells were derived from tumor obtained at the time of primary surgical resection of a posterior fossa ATRT at Boston Children’s Hospital. SU_ATRT2 cells were derived from tumor obtained at the time of surgical resection of an intraventricular ATRT at Lucile Packard Children’s Hospital/Stanford University Medical Center. Med1-MB cells, generated from a spontaneous tumor arising in a Ptch+/−; lacZ mouse, were kindly provided by M. Scott (Stanford). SmoWT-MB and SmoD477G-MB cells isolated from either parental SmoWT or SmoD477G mouse Ptch+/−; p53−/− MB hindflank allografts were kindly provided by C. Rudin (Memorial Sloan-Kettering Cancer Center). Sufu−/− MEFs (conditional deletion of exons 4–8 (ref. 54)) were kindly provided by P.-T. Chang (University of California, San Francisco). SAG, SANT-1, GDC-0449 (S1082, Vismodegib, HhAntag691) and LDE225 (S2151, NVP-LDE225, Erismodegib) were purchased from SelleckChem.com. Shh-N CM was kindly provided by P. Beachy (Stanford). JQ1 was synthesized as previously described44. RNA extraction and qRT-PCR. RNA was extracted using QIAzol Lysis Reagent (79306, Qiagen, Venlo, Netherland) per the manufacturer’s instructions. Reverse transcription was performed with 1 Mg total RNA using the High Capacity cDNA Reverse Transcription Kit (4368813, Invitrogen). Real-time qPCR was performed using 2× Maxima SYBR Green qPCR Master Mix (#K0251, Thermo Scientific) on an Eppendorf Mastercycler PCR machine. The qPCR primers used are listed in Supplementary Table 2. Cell cycle, proliferation, viability and apoptosis assays. For cell cycle analysis, cells were fixed in 70% ethanol for 30 min at 4 °C. After two washes with cold PBS, fixed cells were resuspended in staining buffer (200 Ml PBS + 10 Ml 1 mg ml−1 propidium iodide + 2 Ml 100 mg ml−1 RNase A) and incubated at 37 °C for 45 min. Cells were washed once with cold PBS and filtered through a 70-MM mesh (ELKO Filtering Co., Miami, FL, USA). Filtered cells were centrifuged and resuspended in 500 Ml PBS for FACS analysis. Proliferation assays were performed by culturing cells in the presence of 10 MM EdU for 6–8 h. The EdU+ population was determined using either the Click-iT EdU Alexa Fluor 488 Flow Cytometry Assay Kit (C35002, Invitrogen, CA, USA) or the Click-iT EdU Alexa Fluor 594 Imaging Kit (C10339, Invitrogen, CA, USA). Cells were counterstained with DAPI (D8417, Sigma, MO, USA), and the proliferation index was calculated as EdU+/DAPI+ cells. Apoptosis was analyzed using the BD Pharmingen FITC Annexin V Apoptosis Detection kit I (Cat# 556547, BD Biosciences, CA, USA) per the manufacturer’s instructions. Cell viability was assessed using CellTiter-Glo (G7573, Promega, WI, USA) according to the manufacturer’s instructions. Cells were plated at 5,000 cells per well in 96-well plates and treated with drugs as indicated, and data were collected on a TECAN Infinite 200 plate reader. The drug synergy between JQ1 and GDC-0449 was calculated using CalcuSyn software (Biosoft, Cambridge, UK). A combination index less than 1 was considered as synergistic. All FACS data were collected on a BD Fortessa analyzer (BD Biosciences, CA, USA), and data analyses were performed using Flowjo software (Tree Star, OR, USA). GLI2 overexpression. The Myc-GLI2-DN (17649, pCS2-MT-GLI2-$N) plasmid was purchased from Addgene (Cambridge, MA, USA). The 3×HA-Gli2-FL plasmid was kindly provided by P. Beachy (Stanford). Plasmid transfection was performed using Turbofect transfection reagent (#R0531, Thermo Scientific) according to the manufacturer’s instructions. Cells were treated with drugs 24 h after transfection as indicated.

NATURE MEDICINE

Western blot analysis. Cells were lysed with RIPA buffer (sc-24948, Santa Cruz Biotechnology) for 30 min on ice, and lysates were cleared by centrifugation at 13,000 r.p.m. for 15 min at 4 °C. Supernatants were incubated with 4× Laemmli sample buffer (#161-0747, Bio-rad) at 95 °C for 5 min. The samples were then separated with SDS-PAGE gel and immunoblotted with the indicated antibodies: anti-HA (ab18181, Abcam; 1:5,000 dilution), anti–c-Myc (sc-789, Santa Cruz Biotechnology; 1:1,000 dilution), anti-GLI1 (#2643, Cell signaling; 1:1,000 dilution) and anti–B-tubulin (ab6046, Abcam; 1:5,000 dilution). Lentiviral infection. shRNA lentiviral constructs against mouse Brd2, Brd3 and Brd4 (The RNAi Consortium mouse collection) were kindly provided by A. Sweet-Cordero (Stanford), and shRNA insertion sequences were confirmed by Sanger sequencing. To produce shRNA lentiviruses, 293T cells were transfected with a lentiviral vector and packaging plasmids (pDelta 8.92 + VSV-G). Titers were collected 48 h after transfection and concentrated by polyethylene glycol precipitation. The precipitated lentivirus was resuspended in PBS and aliquoted for storage at −80 °C. For shRNA lentivirus infection, cells were incubated with shRNA lentivirus for 16 h. At 48 h after infection, puromycin was added to select virally infected cells for further experiments. Dual-luciferase reporter assay. Hh-Light2 cells were cultured until confluent and treated with drugs as indicated. Dual-luciferase reporter assays were performed using the Dual-Luciferase Reporter Assay System 10-Pack (E1960, Promega, WI, USA) according to manufacturer’s instructions, and data were collected on a TECAN Infinite 200 plate reader. ChIP-qPCR. Cells were fixed with 1% formaldehyde for 10 min at room temperature before adding glycine to stop the fixation. The cells were then harvested, snap frozen and stored at −80 °C before use. For each ChIP experiment, chromatin isolated from 106 to 107 cells was sonicated and immunoprecipitated with 3–5 Mg of the indicated antibody and 100 Ml Dynabeads protein G. Beads were washed five times with RIPA buffer and one time with Tris + EDTA containing 50 mM NaCl. Bound complexes were eluted by heating at 65 °C with occasional vortexing for 30 min, and crosslinking was reversed by overnight incubation at 65 °C. INPUT DNA was also treated for crosslink reversal. Immunoprecipitated DNA and INPUT DNA were then purified by RNaseA/proteinase K treatment, phenol:chloroform extraction and ethanol precipitation. qPCR was performed using 2× Maxima SYBR Green qPCR Master Mix (#K0251, Thermo Scientific) on an Eppendorf Mastercycler PCR machine. The ChIP-qPCR primer sequences used are listed in Supplementary Table 2. Gene expression microarray analysis. All gene expression profiling data has been deposited into the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) with accession code GSE58185. Gene expression data were generated from total RNA derived from biological duplicates of SmoWT-MB cells treated with control (DMSO), JQ1 (1 MM) or GDC-0449 (0.1 MM) for 6 h. RNA was hybridized to Illumina MouseWG-6 v2.0 (SmoWT-MB) expression bead arrays per the manufacturer’s instructions. Rank-invariant normalized data were generated using GenomeStudio v1.9.0 and converted to .gct file format, which was then collapsed to gene symbols using the GSEA desktop application (http://www.broadinstitute.org/gsea/index.jsp). Differentially expressed genes were visualized using the GENE-E desktop application (http://www.broadinstitute.org/cancer/software/GENE-E/), and the top 5,000 differentially expressed genes between drug-treated and control-treated cells were used for agglomerative hierarchical clustering using Pearson correlation and the average linkage metric across samples and genes. Comparative marker selection analysis between JQ1- or GDC-0449– and DMSO-treated cells was performed in GenePattern using the default settings. Genes with a P value less than 0.05 and a q value less than 0.1 were considered to be significantly differentially expressed. We performed C2 analysis to determine the significance of the overlap between genes that were downregulated by JQ1 and GDC-0449. To identify gene sets differentially expressed after treatment with JQ1 or GDC-0449 (compared to DMSO-treated controls), GSEA was performed as previously described62 using the C2cpg gene set (MSigDB). Gene sets with a nominal P value less than 0.05 and q value less than 0.25 were considered

doi:10.1038/nm.3613


significant. We performed Fisher’s exact test to determine the significance of the overlap between gene sets that were downregulated by JQ1 and GDC-0449. GSEA was also performed using gene sets (.gmt files) derived from Puissant et al.46, Atwood et al.34 and Lee et al.58 (Supplementary Table 1). Briefly, genes associated with Gli1 ChIP-chip peaks in normal GNPs and medulloblastoma (listed in Supplementary Table 1a,b from Lee et al.58) were converted to gene sets (Lee_Gli1_GNP and Lee_Gli1_MB). We then used the Venn diagram function in GENE-E to generate gene sets of overlapping and distinct genes between these lists (Lee_Gli1_GNP_only, Lee_Gli1_MB_only and Lee_Gli1_GNP_ MB_overlap). In vivo mouse studies. In vivo efficacy studies were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee at Stanford University and Children’s Hospital Research Center Oakland. SMOinaive BCC allografts were derived from BCC tumors generated in Ptch+/−; K14creER2; p53flox/flox mice as previously described34. Tumors were treated with vehicle control, BMS-833293 (Bristol Myers Squib Hedgehog inhibitor) (100 mg per kg daily i.p.) or JQ1 (50 mg per kg daily i.p.) until euthanasia was required when the size of vehicle-treated tumors exceeded the limit in our animal care guidelines. Tumor size was measured with calipers every 3–4 d. Tumors were also harvested for RNA analysis. SMOi-resistant mouse BCCs were generated by treating SMOi-naive BCC allografts with BMS-833293 in a cyclical fashion and then with JQ1 as described above. The tumors were treated with JQ1 (50 mg per kg daily i.p.) or vehicle for 7 d before harvesting for RNA analysis. For in vivo medulloblastoma studies, SmoWT-MB, SmoD477G-MB and GFP-luciferase–transduced Med1-MB cells were used for flank or cerebellum injections. 2 × 106 cells were injected into the flank of each 4- to 6-week-old NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mouse (The Jackson Laboratories). 0.5 × 106 cells were used for cerebellum injection, as previously described 32. After engraftments were confirmed, mice were randomized into treatment and control groups and treated with vehicle control, GDC-0449 (100 mg per kg body weight daily p.o.) or JQ1 (50 mg per kg body weight daily i.p.) until euthanasia was required. Tumor growth was measured with calipers or monitored by IVIS imaging on a Xenogen IVIS2000 (Perkin-Elmer). At the end of

doi:10.1038/nm.3613

treatment, tumors were harvested in RNAlater for RNA analysis. Survival data were recorded for the cerebellum-injected mice using Med1-MB cells. Zebrafish studies. All fish studies were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee at the Medical College of Wisconsin. Zebrafish embryos from an outcross of Tg(GBS-ptch2: eGFP)+/− with TL (Tübingen long-fin) WT were exposed to JQ1 at concentrations ranging from 0.25 to 5 MM. A 0.6 MM working dose was determined to be optimal for in vivo studies, as it caused no phenotype, in contrast to 0.75 MM, which caused elevated cell death and dysmorphology. The ptch2:GFP reporter fish were then exposed to JQ1, cyclopamine (25 MM) or vehicle control (DMSO or EtOH for JQ1 and cyclopamine, respectively) from 2 to 30 hpf and then fixed with 4% paraformaldehyde for in situ hybridization using a GFP probe and fast red. All fish with GFP positivity were scored for intensity of staining. For Shh overexpression experiments, embryos from a Tg(hsp70l:ShhaeGFP)+/− × TL WT cross were collected and heat shocked at 38 °C for 15 min at the eight-somite stage. After heat shock, the embryos were immediately placed in JQ1 at a 0.6 Mm or 6.0 Mm concentration. DMSO was used as a negative control. At 12 hpf, GFP-positive and GFP-negative embryos were sorted, and embryos were transferred to equivalent concentrations of fresh drug or DMSO. Images were captured using a Nikon Coolpix digital P520 camera fitted to a Lieca MZLIII stereo microscope at 30 and 56 hpf. Embryos were scored for eye size (dorsal axis length) at 56 hpf. Statistical analyses. Two-way ANOVA was used for comparing tumor growth curves. Log-rank test was used for comparing survival curves. C2 or Fisher’s exact test was used for statistical analyses of contingency table data. Student’s t test was used for all the other comparisons. 61. Kool, M. et al. Genome sequencing of SHH medulloblastoma predicts genotyperelated response to Smoothened inhibition. Cancer Cell 25, 393–405 (2014). 62. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550 (2005).

NATURE MEDICINE

Supplemental Figures and Legends

Epigenetic targeting of Hedgehog transcriptional output through BET bromodomain inhibition Yujie Tang1,2, Sharareh Gholamin2, Simone Schubert1,2, Minde I. Willardson3, Alex Lee4, Pratiti Bandopadhayay5,6,7, Guillame Bergthold5,6, Sabran Masoud1, Brian Nguyen1, Nujsaubnusi Vue1, Brianna Balansay1, Furong Yu1,2, Sekyung Oh1,2, Pamelyn Woo1, Spenser Chen1, Anitha Ponnuswami1, Michelle Monje1, Scott X. Atwood4, Ramon J. Whitson4, Siddhartha Mitra2, Samuel H. Cheshier2, Jun Qi8, Rameen Beroukhim5,7,8, Jean Y. Tang4, Rob Wechsler-Reya9, Anthony E. Oro4, Brian A. Link3, James E. Bradner5,7,8, Yoon-Jae Cho1,2,10* 1Department

of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA USA 2Department of Neurosurgery, Stanford University School of Medicine, Stanford, CA USA 3Department of Cell Biology, Neurobiology and Anatomy, Medical College of Wisconsin, Milwaukee, WI USA 4Program in Epithelial Biology, Stanford University School of Medicine, Stanford, CA USA 5Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, MA USA 6Pediatric Neuro-oncology, Department of Pediatric Oncology, Dana-Farber Cancer Institute and Division of Pediatric Hematology/Oncology, Boston Children’s Hospital, Boston, MA USA 7Broad Institute of MIT and Harvard, Cambridge, MA USA 8Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA USA 9Sanford-Burnham Medical Research, La Jolla, CA USA 10Stanford Cancer Institute, Stanford University Medical Center, Stanford, CA USA *To whom correspondence should be addressed: Yoon-Jae Cho, MD Department of Neurology and Neurological Sciences Stanford University School of Medicine 1201 Welch Road, MSLS Bldg, Rm P213 Stanford, CA 94305 Ph: 650-725-0955 Fax: 650-498-6262 Email: [email protected]

Nature Medicine: doi:10.1038/nm.3613

Supplemental Figure 1. Quantitative RT-PCR showing Gli1 mRNA levels in Hh-Light2 cells treated with Hh ligand (Shh-N CM or SAG) alone or in combination with increasing doses of JQ1. Data represent mean of triplicates ± SD. Supplemental Figure 2. (a)Quantitative RT-PCR of Gli1, Gli2, Brd4 and Smo in wild-type (WT) MEFs and Sufu

MEFs. Data represent mean of triplicates ± SD. (b) Quantitative RT-

PCR of Gli1 mRNA in Sufu

MEFs in response to increasing doses of JQ1. Data represent

mean of triplicates ± SD. (c) Quantitative RT-PCR of Gli1, Gli2, Brd2 and Brd3 mRNAs in Sufu

MEFs expressing shBrd2s, shBrd3s or shCtrl. Data represent mean of triplicates ±

SD. (d) Quantitative RT-PCR showing Ptch1 mRNA levels in Hh-Light2 cells transiently transfected with HA-Gli2-FL or Myc-GLI2-DN and their response to JQ1, GDC-0449, LDE225 or SANT-1. Data represent mean of triplicates ± SD. (e)Quantitative RT-PCR of Gli1 and Gli2 mRNAs showing the time course of decreased Gli1 and Gli2 mRNA levels in response to JQ1 (1 M). Data represent mean of triplicates ± SD. Supplemental Figure 3. In SmoWT-MB, Med1-MB and ASZ001 cells, (a) Quantitative RTPCR of Gli1 mRNA levels in response to increasing doses of JQ1, data represent mean of triplicates ± SD. (b) Quantitative RT-PCR showing time course of Gli1 and Gli2 mRNA levels in response to JQ1 treatment (1 M) , data represent mean of triplicates ± SD. (c,d) ChIPqPCR analysis of Brd4 and PolII occupancies on Gli1 and Gli2 promoter regions, data represent mean of triplicates ± SD. Supplemental Figure 4. (a) Comparison of JQ1 dose-response measuring cell viability in Hh-driven tumor cells (SmoWT-MB and Med1-MB) and normal cells (Hh-Light2 and Sufu MEF). Data represent group means ± SD. JQ1 dose-response curves measuring cell viability from day1 to day3 in SmoWT-MB (b) and Med1-MB (c) cells, data represent group means ± SD. Annexin V/DAPI staining FACS analysis measuring cell apoptosis in response to JQ1 or SMO inhibitors in SmoWT-MB (d) and Med1-MB (e) cells. (f) In Med1-MB cells, PI staining FACS analysis measuring cell cycle distribution in response to JQ1. (g) Combination treatment using increasing doses of JQ1 and GDC-0449 in SmoWT-MB cells to test the drug


synergy. Data represent group means ± SD. (h) Quantitative RT-PCR analysis of Myc and Mycn mRNA levels in response to JQ1 (1 M) or GDC-0449 (0.1 M) treatment in SmoWTMB cells. Data represent mean of triplicates ± SD. Supplemental Figure 5. (a) Hierarchical clustering of samples and differentially regulated genes following 6-hour treatment of SmoWT MB cells with DMSO, JQ1 (1 M), or GDC-0449 (0.1 M). Red indicates increased gene expression and blue reduced gene expression and displayed as row normalized data. Venn diagram and statistical analysis of the overlapping differentially expressed genes (b) or enriched gene sets (c) between JQ1- and GDC-0449treated SmoWT MB cells. Chi-square test and Fisher’s exact test were used for statistical analysis. (d) GSEA analysis of JQ1-treated SmoWT-MB cells using “LEE_GLI1” gene sets. Supplemental Figure 6. In SmoD477G-MB, RCMB025 and CHB_ATRT1 cells, (a) proliferative index was measured by EdU incorporation in response to JQ1, GDC-0449 or LDE225. (b) Myc/MYC and Mycn/MYCN expressions were measured by qRT-qPCR in response to JQ1 or SMO inhibitors. Data represent group means ± SD. Supplemental Figure 7. (a) Tumor size of JQ1 or vehicle treated Med1-MB flank allografts at day 21 post treatment. Quantitative RT-PCR of Gli1, Gli2, Brd4 and Smo in Med1-MB (b), SmoWT-MB (c), SmoD477G (d), SMOi-naïve mBCC (e) and SMOi-resistant mBCC (f) allografts after JQ1 treatment. Data represent group means ± S.E.M. Student’s t-test was used for statistical analysis. * P < 0.05, ** P < 0.01, *** P < 0.001.


Supplementary Figure 1 Tang et al. 2014

Hh-Light2 Relative Gli1 mRNA level

200

Shh-N SAG

150 100 50 0

Shh-N/SAG

-

+

+

+

+

+

JQ1 ( M)

0

0

0.01

0.05

0.2

1


Supplementary Figure 2 Tang et al. 2014 a

b Sufu-/- MEF

250

o Sm

1.0

0.5


G li2

G li1

0

1

Myc-GLI2-DN HA-Gli2-FL

6 4 2

+ N T1

LD E2

44

1 D G

+

SA

+ 9

+

-0

M

+

C

-

25

0

D

Br d2 / Relative mRNA levels

JQ1 0 h JQ1 3 h JQ1 6 h JQ1 24 h

0.1

8

SO

Br d3

G li2

G li1

Sufu-/- MEF 1.5

0.01

Sufu-/- MEF

Gli2-DN/FL

0

0

JQ

0.5

e

0

SO


shCtrl shBrd2-1 shBrd2-2 shBrd3-1 shBrd3-2

1.0

0.5

M

Br d4

d

Sufu-/- MEF

1.5

1.0

JQ1 ( M)

Relative Ptch1 mRNA levels

c

G

G

li2

200 5 4 3 2 1 0

1.5

D

300


WT MEF Sufu-/- MEF

350

li1


400

Supplementary Figure 3 Tang et al. 2014 Med1-MB cells

SMOWT-MB cells

ASZ001 cells

a 1.5

0.5

0

JQ1 ( M)

0

0.01

0.1

1

0.5

GDC-0449 24 h

0.5

0

0.01

0.05

0.2

1

1.5

JQ1 0 h JQ1 2.5 h JQ1 6 h JQ1 24 h

1.0

0.5

G li1

G li2

G

G li1

li2

li1

0

0

0 G

0

JQ1 ( M)

JQ1 0 h JQ1 3 h JQ1 7.5 h JQ1 24 h

1.0

0.5

Ak 1

1.0

1.0

G li2

Ctrl JQ1 3 h JQ1 6 h JQ1 24h GDC-0449 3 h GDC-0449 6 h


1.5

1.5


1.0

0

b


0.5

1.5



1.0

JQ DM S 1 0. O 0 JQ 1 1 G 0. M D C JQ 1 -0 M G 44 1 1 D 9 C -0 0.0 M G 449 1 D C 0 M LD -04 .1 E2 49 M 2 1 LD 5 0 E2 .01 M 2 LD 5 0 M E2 .1 25 M 1 M


1.5

c

1.0

0.5

1.0

0.5

1.5

+

+

-

+

+ -

Anti-Brd4 Anti-Pol II JQ1

+ +

Gli1-TSS-1 Gli1-TSS-2

1.0

0.5

0

0

0



Relative DNA levels

Relative DNA levels


Relative DNA levels

1.5

1.5

+

+

-

+

+ -


+ +

+

+

-

+

+ -

+ +

d

1.0

0.5

0 Anti-Brd4 Anti-Pol II JQ1

1.5


1.0

0.5

+

-

+

+ -


+ +



1.0

0.5

0

0

+

1.5

Relative DNA levels


Relative DNA levels

Relative DNA levels

1.5

+

+

-

+

+ -

+ +


+

+

-

+

+ -

+ +


b

0.8 0.6 0.4 0.2 10 -2

10 -1 10 0 JQ1 ( M)

10 1

1.0

Day 1 Day 2 Day 3

0.8

Relatvie cell viability

Hh-Light2 Sufu-/- MEF Med1-MB SmoWT-MB

1.0



1.2

1.2

1.2

0 10 -3

c

0.6 0.4 0.2 0 10 -4

10 2

10 -3

10 -2

10 -1

10 0

10 1

Day 1 Day 2 Day 3

0.8 0.6 0.4 0.2 0 10 -3

10 2

JQ1 ( M)

10 -2

10 -1 10 0 JQ1 ( M)

10 1

10 2

e DMSO

JQ1

g

JQ1

1.0


h

%100

G1 S G2/M

%80 %60 %40 %20

50

100

200

500

1000

2.5

5

10

20

40

80

-

+

+

+

+

+

1.5

1.0

0.5

yc n

JQ1

2000

DMSO JQ1 GDC-0449

0

%0

DMSO


0.2

M

Percentage of total cells

AV

f

DAPI

0.4

JQ1 (nM) GDC-0449 (nM) Synergistic

DAPI


LDE225

0.6

0

AV

GDC-0449

JQ1 GDC-0449 JQ1+GDC-0449

0.8

yc

DMSO

M

d

1.0


SMOwt-MB_DMSO_1 SMOwt-MB_DMSO_2 SMOwt-MB_GDC0449_1 SMOwt-MB_GDC0449_2 SMOwt-MB_JQ1_1 SMOwt-MB_JQ1_2 PTCH1 HHIP

GLI2

CDK6

CCND1

BMP7

CCND2

b

c Differentially expressed genes

Differentially expressed gene sets

JQ1 GDC0449

4638

3446

*

3404

148

*p < 0.0001 Chi-square test

49 *

38

*p < 0.0001 Fisher’s exact test

d

Enrichment score (ES)

“LEE_GLI1_MB” gene set (Genes associated with MB Gli1 ChIP-chip peaks)

DMSO


JQ1

Gene set LEE_GLI1_MB LEE_GLI1_MB_ONLY LEE_GLI1_GNP_MB_OVERLAP LEE_GLI1_GNP

NOM p-val 0 0 0.02059925 0.46063653

FDR q-val 0 0 0.00653859 0.4634985

FWER p-val 0 0 0.016 0.798

Supplementary Figure 6 Tang et al. 2014

c

RCMB025 (SUFU mutated)

0


G D

25 E2 LD

D

1.5 1.0 0.5

YC N M

YC N M

YC

0

M

M yc n

M yc

0

9

M SO

25

0.5


YC

0.5

1.0

2.0

M

1.0



DMSO JQ1 GDC-0449



f 1.5

1.5

-0 44

EdU+% G D

C

D

E2 2

E2

M SO

0

LD

-0 C

0

G

D

10

e

d

20

0.1

5

44 9

1 JQ

D

M

SO

0

0.2

LD

10

30

9

20

0.3

-0 44

30

40

JQ 1

EdU+ / DAPI+

EdU+%

40

CHB_ATRT1 (SMARCB1/SNF5 mutated)

0.4

C

SmoD477G-MB 50

JQ 1

b

a

Supplementary Figure 7 Tang et al. 2014 b

Med1-MB flank allograft

Med1-MB flank allograft 1.5


1.5 1.0

P = 0.0281 0.5

**

*

Ctrl (n = 4) JQ1 (n = 4)

**

1.0

0.5

0

0

Ctrl (n = 5)

Br d4

G li2

G li1

JQ1 (n = 5)

c

d SmoWT-MB flank allograft

SmoD477G-MB flank allograft

***

**

**

1.6

0.8 0.4

***

1.2

***

Ctrl (n = 4) JQ1 (n = 4)

***

0.8 0.4

0

e

Br d4

li2

Sm o

G

li1

Br d4

o Sm

G li2

G li1

0 G

1.2

Ctrl (n = 4) JQ1 (n = 4)



1.6

f

***

***

Ctrl (n = 3) JQ1 (n = 3)

*

0.8 0.4 0

1.6 1.2

SMOi-resistant mBCC allograft

***

**

Ctrl (n = 2) JQ1 (n = 4)

***

0.8 0.4


Br d4

G li2

Br d4

Sm o

li2 G

G

li1

0 G li1

1.2

SMOi-naive mBCC allograft Relative mRNA levels


1.6

Sm o

Tumor size (cm 3)

2.0

Sm o

a