Supplement Materials and Methods

Tissue Microarray analysis of TOP2A and EZH2 (DFCI cohort)

TSA-plus Fluorescence Immunohistochemistry

A total of 131 formalin-fixed, paraffin-embedded prostate cancer tissues were obtained from

consented patients at Brigham and Women’s Hospital (Supplement Figure 5) and utilized to

construct the 3 TMA’s, each sample is represented by up to 3 tissue microarray cores for tumor. A

multiplexed tyramide signal amplification method was performed on 4-µm sections of the TMA

for detection of EZH2 and TOP2A protein. Alpha-methylacyl-CoA racemase (AMACR) staining

was also implemented to mask for tumor epithelial gland. The staining approach consists of a multi-

step protocol of sequential TSA-amplified immunofluorescence labels for EZH2, TOP2A and

AMACR, and a 4’,6-diamidino-2-phenylindole (DAPI) counterstain. Briefly the sections were

deparaffinized and hydrated; prior to each immunofluorescence labeling, EZH2, TOP2A and

AMACR antigens are retrieved with a single microwave step. Each labeling cycle consists of

application of a primary antibody, and a secondary antibody conjugated to horse radish peroxidase

(HRP), and TSA conjugated to a fluorophore. The slides were consecutively incubated with the

antibody against EZH2 (Clone 6A10, Leica NCL-L-EZH2) at a dilution of 1:200, TOP2A (Clone

3F6, Leica, NCL-TOPOIIA) at a dilution of 1:100 and AMACR (P504S, Clone 13H4, ZETA

Corporation #Z2001) at a dilution of 1:250 for 30 mins respectively. TSA reagents were obtained

from PerkinElmer, Inc. TSA conjugated fluorescein was used FITC for EZH2; CY3 for TOP2A

and CY5 for AMACR, respectively. Prostate cancer tissue from RP specimens was used as positive

controls for EZH2, TOP2A and AMACR; Omission of the primary antibody was utilized as a

negative control. Additionally, tonsil was used as a positive control for EZH2 and TOP2A, skeletal

muscle as negative/weak control for EZH2 and adipose tissue as negative/weak control for TOP2A.

Single staining of each antibody and DAPI only counterstained slides were used for spectral library

construction.

Spectral Imaging

Each TSA-FIHC stained TMA was scanned on a Vectra 2 multispectral slide analysis system from

Perkin Elmer using the 20x objective to obtain high power field images of individual cores.

Exposure times for the mercury-halide light source at 10% power output were optimized for each

target-dye combination; nuclei-DAPI, EZH2-FITC, TOP2A-CY3, AMACR-CY5. Using DAPI

filter for autofocus, the full slide TMA was surveyed at 4x to identify and align cores that were

accepted for automated 20x image acquisition. Single stained control slides were imaged with the

multiplex filters for use in generating the spectral library and blank control for autofluorescence.

Using inForm image analysis software from Perkin Elmer, the spectral library was built and applied

to unmix the fluorescence spectra. An inForm algorithm was configured for trainable tissue

segmentation, DAPI based cell segmentation and AMACR based Tumor phenotyping. Tissue

segmentation was implemented to distinguish glands from stroma. Gland tissue was classified and

binned into “tumor” and “other” based on pathology review and use of AMACR positive

expression as a tumor-specific mask, employed for phenotyping with the Nuance software tool.

The normalized total expression intensity was recorded on a per pixel basis. The reported mean for

a given cell was the average intensity of all the normalized total pixel values in each nucleus. The

algorithm was then applied to all the images contained within the TMA. The processed batch was

reviewed and edited for the quality of tissue segmentation and tumor phenotyping, and status of

each core matched to the definition from the TMA map. Case level statistics for each patient across

the 3 TMA’s were calculated nuclear EZH2 were compared to nuclear TOP2A mean expression

levels in all cases that were AMACR positive.

BTaylor et al. (2010) Taylor et al. (2010)

Taylor et al. (2010) Taylor et al. (2010)

Supplement Figure 1: Taylor et al. (2010) data indicate that concurrent TOP2A and EZH2 expression identify primary PCa patients with a more aggressive disease. (A) Unsupervised clustering analysis of Taylor et al. (2010) data demon-strates TOP2A+/EZH2+ patients (green) tightly cluster apart from other patients (purple) based on their differentially expressed genes (DEGs). (B) Unique DEGs between TOP2A+/EZH2+ and other patients was validated by PCA. (C) Kaplan-Meier analysis reveals that patients with concurrent high TOP2A and EZH2 expression (TOP2A+/EZH2+) have a faster progression to biochemical recur-rence when compared to other conditions (right).

TOP2A+/EZH2+

Other

A

CPC2 (0% variance)

PC

1 (9

9.9%

var

ianc

e)

0.5

0

-0.5

-19 -19 -19-1.0

Time (months)

Pro

porti

on b

ioch

emic

al re

curr

ence

-free

0 20 40 60 80 100 120 140

0

0.2

0.4

0.6

0.8

1.0

0

0.2

0.4

0.6

0.8

1.0

Time (months)

0 20 40 60 80 100 120 140

Pro

porti

on b

ioch

emic

al re

curr

ence

-free

P = 0.119

TOP2A+/EZH2+ (events = 52)Other (events = 79) P = 0.0282

TOP2A+/EZH2+ (event = 52)

TOP2A+ (event = 6)TOP2A-/EZH2- (event = 24)EZH2+ (event = 49)

0 0.5 1.0 1.5 2.0 2.5 3.0

E2F1_UP.V1_UP

GCNP_SHH_UP_LATE.V1_UP

CSR_LATE_UP.V1_UP

PRC2_EZH2_UP.V1_UP

VEGF_A_UP.V1_DN

RPS14_DN.V1_DN

Normalized enrichment score

GSEA Oncogenic SignaturesTCGA (2015)Taylor et al. (2010)

Supplement Figure 2: GSEA for oncogenic signatures in primary human PCa revealed statistically significant overlapping for gene signatures involving VEGF and E2F1 signaling from the TCGA (2015) and Taylor et al. (2010) datasets. (P < 0.05 and FDR < 0.15; Supplement Table 3).

02

8

10

12

14

16G

ene

Expr

essi

on L

evel

sCDKN1A

02

8

10

12

14FGFR1

02

6

8

10

12PMP22

****

*

Both High Other Both High Other Both High Other

Supplement Figure 3: Genes that have been demonstrated to be up-regulated in indolent primary PCa are down regulated in TCGA (2015) patients with concur-rent up-regulation of TOP2A and EZH2 (TOP2A+/EZH2+) (Mann-Whitney; * P < 0.05; *** P < 0.001).

0 20 40 60 80 100 120 140 0 50 100 150 200 0 50 100 150 200

Mayo Clinic validation study Thomas Jefferson University dataset JHMI post-RP dataset

0

0.2

0.4

0.6

0.8

1.0

Pro

babi

lity

of B

CR

-Fre

e S

urvi

val

Time (months)

0

0.2

0.4

0.6

0.8

1.0

Pro

babi

lity

of B

CR

-Fre

e S

urvi

val

Time (months)

0

0.2

0.4

0.6

0.8

1.0

Pro

babi

lity

of B

CR

-Fre

e S

urvi

val

Time (months)

A B C

P = ns (8.99e-01)

TOP2A+/EZH2+ (events = 2)TOP2A-/EZH2- (events = 31)

P = ns (4.21e-01)

TOP2A+/EZH2+ (events = 6)TOP2A-/EZH2- (events = 45)

P = ns (5.94e-01)

TOP2A+/EZH2+ (events = 5)TOP2A-/EZH2- (events = 44)

Supplement Figure 4: Concurrent high TOP2A and EZH2 expression in primary prostatic tumors (TOP2A+/EZH2+) is not associated to a shorter time to biochemical recurrence when compared to TOP2A-/EZH2- patients included in the (A) Mayo Clinic Validation study, (B) Thomas Jefferson University dataset or the (C) Johns Hopkins University post-RP dataset.

sKO cells dKO cells

DM

SO5 μM

GSK

126

1 μM

Eto

posi

de5 μM

EPZ

6438

1 μM

Eto

posi

de

0

5

10

15

20

SubG

1 (%

)

** **

0

20

40

60

80

100

Rel

ativ

e C

olon

y Ar

ea (%

)

**

**

0

20

40

60

80

100

Cel

l Cyc

le D

istri

butio

n (%

)

SubG1

G1

S

G2M

>4NsKO cells

0

20

40

60

80

100

Cel

l Cyc

yle

Dis

tribu

tion

(%)

SubG1

G1

S

G2M

>4NdKO cells

Control GSK126(5 μM)

EPZ6438(5 μM)

Etoposide (1 μM)

sKO

dKO

Control GSK126(5 μM)

EPZ6438(5 μM)

Etoposide (1 μM)

Control GSK126(5 μM)

EPZ6438(5 μM)

Etoposide (1 μM)

Control GSK126(5 μM)

EPZ6438(5 μM)

Etoposide (1 μM)

A B

Supplement Figure 5: (A) Combination therapy induces loss of clonogenicity specific to dKO cells (top: representative experiment; unpaired t test; ** P < 0.01, triplicate, mean ±SEM). (B) Cell cycle analysis indicates that combination therapy targeting both EZH2 (GSK126 or EPZ6438, 5 μM) and TOP2A (Etoposide, 1 μM) induces G2M arrest in both sKO and dKO cell lines but results in a significant increase in apoptosis as indicated by SubG1 accumulation only in dKO (unpaired t test; ** P < 0.01, triplicate, mean ±SEM).

Clear Renal Cell Carcinoma(TCGA)

Papillary Kidney Carcinoma(TCGA)

Time (months)

Reg

ress

ion-

free

surv

ival

Reg

ress

ion-

free

surv

ival

0

0.2

0.4

0.6

0.8

1.0

0

0.2

0.4

0.6

0.8

1.0

0 20 40 60 80 100 120

Time (months)0 20 40 60 80 100

P = 1.54e-05

P = 0.0358

A

B

0

25

50

75

100

Met

asta

tic S

tagi

ng (%

)

M0M1

0

25

50

75

100

Met

asta

tic S

tagi

ng (%

)

M0M1

TOP2A+/EZH2+(N = 94)

Other(N = 196)

TOP2A+/EZH2+(N = 226)

Other(N = 308)

TOP2A+/EZH2+ (events = 94)Other (events = 196)

TOP2A+/EZH2+ (events = 226)Other (events = 308)

Supplement Figure 6: TOP2A and EZH2 co-expression can select for meta-static disease progression and shorter regression-free survival in the TCGA’s (A) Clear Renal Cell Carcinoma and the (B) Papillary Kidney Carcinoma datasets. (Abbreviations: no distant metastasis, M0; distant metastasis, M1)