ERR alpha is critical for the growth of ER-negative breast cancer Stein, 1

Supplementary Legends

Table S1. Primers used for quantitative PCR.

Table S2. MCF-7 microarray expression values for significantly regulated

by either ERRspPGC1 (vehicle treated, PGC-1α L2L3M versus ERRspPGC1

infected) or estradiol treatment (PGC-1α L2L3M infected, vehicle versus

estradiol). Expression data from each condition, used to generate Figures 1A

and 1C, are included. P-values and fold changes are indicated for each

comparison and the probeset classification (for the hierarchical clustering in

Figure 1A and direct comparison in Figure 1C) is listed.

Table S3. ontology analysis of significantly regulated genes by either ERRα

or by estradiol treatment. The 620 ERRα regulated probesets (differentially expressed in ERRspPGC1 versus PGC-1α L2L3M infected cells) and 99 ERα

regulated probesets were clustered into functional categories.

(GO) overrepresentation analysis was performed in Onto-Express, and p-values

are indicated shaded in light gray (21, 22). Affymetrix HG133A was used as a blank set; probability distribution of GO terms was assumed to be

hypergeometric; FDR correction was applied to correct for multiple

comparisons. Complete GO analysis data listed in the supplementary file: GO

Term Analysis.xlsx. ERR alpha is critical for the growth of ER-negative breast cancer Stein, 2

Figure S1. Knockdown of ERRα by siRNA does not alter ERα expression level. MCF-

7 cells were infected with adenovirus expression a control siRNA (siCON) or siRNA directed to ERRα (siERRα). Two days after infection, cells were harvested and total cell lysate separated by SDS-PAGE. GAPDH, shown as a loading control, and ERRα were detected as indicated in materials and methods. ERα was detected using mouse monoclonal anti-ERα (D12, Santa Cruz).

Figure S2. Infection of PGC-1α or β-galactosidase adenoviruses followed by estrogen treatment. A, Western blot analysis of the levels of PGC-1α (H-

300, Santa Cruz Biotechnology, Inc) and ERRα in whole cell extract from the indicated MCF-7 treatment groups with GAPDH shown as a loading control.

Both PGC-1α WT and ERRspPGC1 induce ERRα expression while the negative control PGC-1α L2L3M adenoviral infection does not. B, Real-time PCR of adenovirus encoded PGC-1α and the ERα target gene PR (progesterone receptor) using 36B4 as an internal control.

Figure S3. Hierarchical clustering of genes differentially expressed in any of the

MCF-7 array conditions. Average probeset intensity is indicated for biological replicates in each condition (n=3). Probesets were selected from log transformed data by ANOVA (p<0.05) with the additional criteria that standard deviation across samples >0.5, that the probeset was identified as “present” in

>20% of the samples (dChip), and that expression was consistent within ERR alpha is critical for the growth of ER-negative breast cancer Stein, 3

replicates ((standard deviation/mean)<0.5 for each of the 8 replicate groups).

Unsupervised clustering analysis was performed in dChip. While non-nuclear receptor interacting PGC-1α L2L3M regulated few genes relative to the βgal control samples, both wild-type PGC-1α and ERRspPGC1 significantly affected patterns. Expression of wild-type PGC-1α and ERRspPGC1 similarly affected the majority of genes, but differentially regulated genes can be appreciated particularly in the presence of estradiol. This is in accordance with reports that show PGC-1α coactivation of estrogen stimulated ERα, and led

us to focus the remainder of our analysis on genes regulated by ERRspPGC1

versus PGC-1α L2L3M in the presence and absence of estradiol. The complete

dataset has been deposited according to microarray guidelines (GSE11266).

Figure S4. Real-time PCR of additional genes from each class found in the microarray. MCF-7 cells were infected with the indicated PGC-1α adenovirus (or

βgal control) and treated with 10nM estradiol (or vehicle). VEGF (vascular endothelial growth factor) belongs to ERRspPGC1/ERRα-induced Class I. PS2

(trefoil factor 1, TFF1) is a group II gene induced by both ERRα and estradiol

treatment. Group III member MCM5 (minichromosome maintenance complex

component 5) is induced by estradiol, but this induction is blunted by activation of ERRα. Estrogen-regulated genes are represented in group IV and include

SGK3 (serum/glucocorticoid regulated kinase family member 3). Finally, VDR

(vitamin D receptor) is repressed by PGC-1α coactivation of ERRα, but non- ERR alpha is critical for the growth of ER-negative breast cancer Stein, 4

nuclear receptor-mediated functions of PGC-1α may also be involved as PGC-

1α L2L3M also represses VDR expression below βgal infected cells. Quantitative

PCR was performed using 36B4 as an internal normalization control and data are

represented ± SEM for a representative experiment.

Figure S5. Microvessel density was not statistically different at the time of

maximal xenograft size between tumors derived from MDA231 siCON versus

MDA231 siERRα cells. When tumors grew to maximal allowable volume, or if they

became ulcerated, they were harvested and flash frozen in liquid nitrogen.

Tumors were embedded in Optimal Cutting Temperature (Tissue Tek, Sakura

Finetek) and 10uM sections prepared. Sections (three per tumor) were

incubated with anti-CD31 antibody (BD Biosciences) at room temperature for 60

minutes (dilution 1:200) and Cy2-conjugated donkey anti rat (Jackson

Immunoresearch) was used as a secondary antibody. After the slides were

scanned at low magnification (40X), MetaMorph (Molecular Devices

Corporation) was used to stitch together the images for each tumor section.

Image J (National Institutes of Health) was used to binarize and filter the image.

The number of vessels were counted and divided by the total area of the tumor, which was calculated based on a calibration of 1 pixel = 4.35μM.