Supplementary Information for

Supplementary Information for BRCA1-IRIS promotes human tumor progression through PTEN blockade and HIF-1a activation Andrew G. Lia,b, Elizabeth C. Murphyb, Aedin C. Culhanec,d, Emily Powelle, Hua Wanga,b, Roderick T. Bronsonf, Thanh Vonb,g, Anita Giobbie-Hurderc, Rebecca S. Gelmanc,d, Kimberly J. Briggsh, Helen Piwnica-Wormse, Jean J. Zhaob,g, Andrew L. Kungi, William G. Kaelin Jr.h,j, and David M. Livingstona,b a

Department of Genetics, Harvard Medical School, Boston, MA 02115; bDepartment of Cancer Biology, Dana-Farber Cancer Institute, Boston, MA 02215; cDepartment of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, MA 02215; dDepartment of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA 02115; eDepartment of Cancer Biology, The University of Texas MD Anderson Cancer Center, Houston, TX 77030; fDepartment of Pathology, Harvard Medical School, Boston, MA, 02115; gDepartment of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02215; hDepartment of Medical Oncology, DanaFarber Cancer Institute, Boston, MA 02215; iDepartment of Pediatrics, Memorial Sloan Kettering Cancer Center, New York, NY 10065; jHoward Hughes Medical Institute, Chevy Chase, MD 20815 Andrew G. Li Email: [email protected] David M. Livingston Email: [email protected]

1 www.pnas.org/cgi/doi/10.1073/pnas.1807112115

This PDF file includes: Supplementary Materials and Methods Figs. S1 to S6 Tables S1 to S4 Captions for databases S1 to S4 References for SI reference citations Other supplementary materials for this manuscript include the following: Datasets S1 to S4

2

Supplementary Materials and Methods Cell Lines and Reagents. All cell lines were acquired from ATCC. Media used for all cell lines are summarized in Table S1. M231 LM2 cells were provided by Dr. J. Massagué from Memorial Sloan Kettering Cancer Center (1). Generation and growth condition for BC3-p53WT and BC3-p53KD cells were described previously (2). Maintenance of DFBC12-58 cells was described previously (3). All cells were verified to be mycoplasma-free. LY294002 and CHIR98014 were purchased from Sigma Aldrich and Selleck, respectively. siRNA and shRNA Experiments. Transfection of IRIS siRNA or negative control siRNA (scramble sequence, Qiagen) was performed using Lipofectamine 2000 (Life Technologies) in Opti-MEM (Life Technologies) according to the manufacturer’s protocol. Cells carrying shRNA were generated as follows. Sense and antisense oligonucleotides (see Table S2 for target sequences) were annealed, phosphorylated with T4 kinase, and inserted into the AgeI and EcoRI sites of a pLKO vector (4). The resulting plasmids were transfected into 293T cells for virus packaging, and harvested viruses were used for infection. Infected cells were selected for puromycin or hygromycin B resistance and then pooled for subsequent use. Quantitative RT-PCR (qRT-PCR). Total RNA was purified using a RNeasy kit (Qiagen). qRT-PCR was performed using the SYBR Green method (iScript One-Step RT-PCR with SYBR® Green, BIO-RAD) according to the manufacturer’s instructions. Normalization to RPLP0 RNA was routinely performed. Primers used are listed in Table S3. ChIP. ChIP analysis was performed as described previously (5) with minor modifications. The nuclear extract from one confluent 150 mm dish was used for each IP. Primers used in quantitative PCR are described in Table S4. Anchorage-independent Growth Assay. Cells were resuspended in growth medium containing 0.4% Noble agar (Difco); the mixture was plated atop 0.6% soft agar prepared in growth medium. Colonies were counted under the microscope 4 wk post-seeding. All experiments were performed at least twice, and three separate replicates of each culture were analyzed. Cell Migration Assay. Cellular migration was investigated using 24-well cell culture Boyden chambers (BD Biosciences). Cells (5 x 104 per well) were plated atop membranes containing 8 µm pores. Cells that traversed the membrane were stained using 0.5% crystal violet and counted using the ImageJ program. All experiments were performed at least twice, and three separate replicates of each culture were analyzed under each experimental condition. Mammosphere Formation Assay. Mammosphere formation was carried out as described previously (2). All experiments were performed at least twice, and three separate replicates of each culture were analyzed. 3

Invasion Assay. Local invasion of M231 cells was investigated using 24-well cell culture Boyden chambers (BD Biosciences). Cells (1 x 105 per well) were plated atop Matrigelcoated membranes containing 8 µm pores. Cells that penetrated the membrane were stained using 0.5% crystal violet and counted using the ImageJ program. All experiments were performed at least twice, and three separate replicates of each culture were analyzed under each experimental condition. Colony Formation Assay. Cells were seeded on plastic plates in growth medium. Colonies were stained using 0.5% crystal violet 2 wk post-seeding and counted using the ImageJ program. All experiments were performed at least twice, and three separate replicates of each culture were analyzed. Antibodies, Immunoblotting, and Immunoprecipitation (IP). The following antibodies were used: E-cadherin (610404, BD Transduction Lab), N-cadherin (610920, BD Transduction Lab), b-Catenin (610153, BD Transduction Lab), V5 (A190-120A, Bethyl), BRCA1 (A300-000A, Bethyl), HA (901501, BioLegend), Vimentin (IF01, Calbiochem), AKT (#9272, Cell Signaling), phospho-AKT (#9275, Cell Signaling), phospho-b-Catenin (#9561, Cell Signaling), GSK-3b (#9315, Cell Signaling), phosphoGSK-3b (#5558, Cell Signaling), HDAC1 (#5356, Cell Signaling), PTEN (#9188, Cell Signaling), Slug (#9585, Cell Signaling), Snail (#3879, Cell Signaling), ARNT (NB100110, Novus), HIF-1a (NB100-134, Novus), Myc (sc-40, Santa Cruz), Actin (A5441, Sigma), and Vinculin (V4505, Sigma). Western blot and IP were carried out as described previously (5) with specific antibodies. Anti-human IRIS antibody was raised in rabbits immunized with the intron-11 peptide (N-GIGTRFLCLPQSIYRSELNVYAFGEHILQISKYS-C) fused to glutathione Stransferase. IRIS-specific polyclonal rabbit antibody was affinity-purified using a synthetic peptide of the same sequence as the immunogen. Plasmids and Mutagenesis. Myc-FL IRIS and Myc–Tr IRIS (lacking all sequences encoded by the BRCA1 intron 11) were generated by PCR, using full-length IRIS cDNA as the template, and were cloned into the pLX304 (Addgene) (6) vector through Gateway cloning (Life Technologies). The WT HIF-1a cDNA was also cloned into the pLX304 vector. The STA (Ser551Ala, Thr555Ala and Ser589Ala) HIF-1a construct was generated with a QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent) according to the manufacturer’s protocol. All constructs were confirmed as accurate by sequencing analyses. Protein Half-life Experiments. Protein half-life was estimated in cycloheximide-chase experiments. Specifically, cultures of M231 cells transduced with various hairpins and/or protein expression vectors were treated with 50 µg/ml cycloheximide (Calbiochem) and whole cell extracts of the indicated culture were collected at indicated time points. To mimic a hypoxic condition, cells were pretreated with 1 mM DMOG (Sigma Aldrich) for 12 hr. Band intensities in Western blots were quantified using the ImageJ program. 4

Xenograft Studies and Histological Analysis. For subcutaneous or primary tumor growth, M231 cells carrying shIRIS or control shRNA were mixed with 40% Matrigel (BD Biosciences) in PBS, and 2 x 106 cells were inoculated subcutaneously or into the mammary fat pads of 6-wk-old, intact female nude mice (Charles River Laboratories). Subsequent tumor size was measured with a caliper, and tumor volume was calculated using the equation: Volume = Length x Width2/2. All mice in mammary fat pad injection experiments were euthanized, and their tissues were fixed 7 wk post-injection. Fixed tissues were used for histological analysis in search of evidence of metastasis. For metastasis assays, intact female SCID-beige mice (6-wk-old, Charles River Laboratories) were inoculated intravenously or by intracardiac injection of M231 cells (2 x 106 per mouse) stably transduced with various hairpins and/or protein expression vectors. All mice were euthanized and fixed 3 wk post-transplantation for further histopathological analysis to assay for metastasis. Mice carrying BC3-p53KD cells (1 x 106 per animal) were euthanized and fixed 7 wk post-transplantation. Histopathological analysis was performed blindly by experts in the DF/HCC Rodent Histopathology Core facility. Metastatic burden was analyzed and quantified as described previously (7). This method involves the quantitation of metastatic tumor deposits present in a standard number of sections of each organ and tissue destined to be evaluated. Typically, metastases were sought in lung, heart, brain, eye, bone, bone marrow, liver, kidney, adrenal glands, pancreas, spleen, muscle, lymph nodes, mesentery, mediastinum, fat, intestine, stomach, ovary, uterus, and connective tissues. Analysis of RNA-Seq Data. The Kallisto estimates of p220 and IRIS abundance were extracted from the dataset cgc-05-0002 on the Institute for Systems Biology Google Cloud (https://bigquery.cloud.google.com). Access to controlled data (FASTQ) was approved through the Database of Genotypes and Phenotypes (dbGaP), and analyses were performed in accordance with Data Use Certification and the local IRB protocol 16368. Expression levels were defined in a log2+1 format. An empirical Bayes batch effect correction algorithm, ComBat (8), was applied using the sva package in Bioconductor to retain the disease-related information and to remove the platform effects. See Dataset S1 for adjusted expression data. A general cutoff of 0.1 TPM was used for filtering of low transcript counts (9). Cancer types having fewer than 3 cases with IRIS transcript counts of >0.1 TPM were considered as baseline level IRIS-expressing groups. Cancer types including at least 3 cases with >0.1 TPM IRIS levels were considered elevated IRIS-expressing cancer types. The exactTest in Bioconductor package edgeR (10) was used to compare IRIS transcript levels in each elevated IRIS-expressing type to those in all baseline level IRIS-expressing types combined. P values were adjusted for multiple testing with Benjamini & Hochberg FDR correction. The code to reproduce analysis is available on https://github.com/aedin. Due to low levels of IRIS abundance estimated by Kallisto in all TCGA cases, a direct counting approach was used to examine the relationship between IRIS and PTEN expression in the TCGA breast cancer dataset. Reads that map to the intron 11 sequence of the IRIS transcript in public alignments (BAM files downloaded from the UCSC Cancer Genomics Hub server) were counted using featureCounts in the Rsubread R package (11). Counts were normalized over the total number of reads using the median 5

ratio method (Eqn 5) (12) and the DESeq2 package in Bioconductor (primary functions, estimateSizeFactors, fpkm). The library size (total number of reads) was estimated from the Level-3 exon data. The PTEN mutation and mRNA expression data were extracted from the processed Level-3 TCGA data. All cases with no read in any of BRCA1 exons were excluded from the analysis. Cases with homozygous BRCA1 deletion or early nonsense mutations were removed from the analysis. Any cases subjected to a reporting bias were also removed from the analysis. See Dataset S4 for patient ID numbers and expression data.

6

Supplementary Figures

Fig. S1. IRIS functions in human cancer cells. (A) Dot plot showing p220 RNA-Seq data in various sporadic tumors in the TCGA data collection. The order of cancer types in the plot is the same as in Fig. 1A. R: Pearson correlation coefficient between p220 and IRIS RNA abundance in each cancer type. Refer to Dataset S3 for R values. (B) Immunostaining of E-cadherin in A549 and ACHN cells before and after IRIS depletion. (Scale bar, 20 µm.) (C) Overexpression of FL IRIS, but not Tr IRIS, rescued the growth of endogenous IRIS-depleted M231 cells in soft agar (n = 9). (Scale bar, 200 µm.) ns, not significant. (D) Colony formation on plastic surfaces by M231, A549, and ACHN cells expressing control or IRIS specific hairpin (n = 6). (Scale bar, 200 µm.)

7

Fig. S2. Primary breast tumor growth is not dependent on IRIS and additional results of metastasis after intracardiac tumor cell injection. (A) Mouse mammary fat pad injection followed by mammary tumor growth of M231 cells that express a control or an IRIS hairpin. (B) Subcutaneous tumor cell injection followed by local mouse tumor growth of M231 cells that express a control or an IRIS hairpin. (C) Quantitation of pulmonary metastatic deposits detected in the mammary fat pad injection experiments (n = 5). ns, not significant. (D) Summaries of metastatic deposit quantitation in mice injected and analyzed in the IC model (scramble: n = 11; shLacZ: n = 12; shIRIS1: n = 12; shIRIS2: n = 13), as depicted in Fig. 2. (Left) Number of lung tumor nodules detected in each mouse. (Center) Number of sites (excluding lung) in which tumor was detected in each mouse. (Right) number of tumor nodules detected in each mouse, excluding those in lung. ND, not detected.

8

Fig. S3. Effects of endogenously overexpressed IRIS in PDX-derived cells. (A) IRIS was required for anchorage-independent growth of BC3-p53KD cells (n = 9). (Scale bar, 200 µm.) (B) Colony formation on plastic surfaces by BC3-p53KD cells before and after IRIS depletion (n = 9). (Scale bar, 200 µm.) (C) Primary and secondary mammosphere formation of BC3-p53KD cells expressing various hairpins (n = 6). (Scale bar, 160 µm.) (D) Morphology of IRIS-depleted and undepleted BC3-p53KD and BC3-p53WT cells. (Scale bar, 10 µm.)

9

Fig. S4. IRIS regulates the PTEN/AKT/HIF-1a pathway. (A) IRIS depletion in Hs578t cells resulted in the depletion of HIF-1a protein. (B) Western blot showing expression levels of various proteins in BC3-p53WT cells before and after IRIS depletion. (C) Western blot showing expression levels of various proteins in IRIS-depleted cells before and after exogenous myc-tagged IRIS overexpression (indicated by arrows). N.B. intron 11-specific IRIS antibody failed to identify truncated IRIS. (D) Relationship between the presence of PTEN and IRIS-associated HIF-1a abundance. (E) Morphology of HME cells overexpressing either FL IRIS or Tr IRIS mutant. (Scale bar, 20 µm.) (F) Overexpression of FL IRIS, but not Tr IRIS, enabled HME for Matrigel growth (n = 6). (Scale bar, 40 µm.) ns, not significant. (G) Overexpression of FL IRIS, but not Tr IRIS, promoted mammosphere formation of HME cells (n = 6). (Scale bar, 40 µm.) ns, not significant. (H) Overexpression of FL IRIS, but not Tr IRIS, suppressed E-cadherin expression in MCF7 cells, as shown by Western blot.

10

Fig. S5. HIF-1a is a downstream mediator of IRIS functions. (A) Endogenous IRIS and HIF-1a were each required for anchorage-independent growth of M231 cells. (Scale bar, 200 µm.) (B) HIF-1a depletion inhibited the transmembrane migration of M231 cells in either normoxic or hypoxic condition. (Scale bar, 400 µm.) (C) Endogenous IRIS and HIF-1a were each required for the maintenance of the mesenchymal morphology of M231 cells. (Scale bar, 40 µm.) (D) V5-tagged WT or STA HIF-1a, and IRIS expression were detected by Western blotting in various M231 cultures in indicated conditions. (E) Half-lives of WT and STA HIF-1a in the absence and presence of DMOG. (F) Overexpression of STA, but not WT, HIF-1a rescued the anchorage-independent growth of IRIS-depleted M231 cells. (Scale bar, 200 µm.) (G) Representative staining pictures of transmembrane cells expressing indicated hairpins and/or cDNAs. (Scale bar, 400 µm.) (H) Mesenchymal morphology reappeared in IRIS-depleted, STA HIF-1aoverexpressing M231 cells. (Scale bar, 40 µm.) (I) Local invasion of IRIS-depleted and control M231 cells (n = 6). (Scale bar, 400 µm.)

11

Fig. S6. IRIS regulates PTEN transcription. (A) IRIS depletion in Hs578t cells resulted in higher PTEN mRNA levels. Values are normalized mean ± SD (n = 6). (B) IRIS binding to the PTEN promoter in Hs578t cells before and after IRIS depletion. Values are normalized mean ± SD (n = 6). (C) Diagram of the human PTEN promoter. (Upper line) Schematic representation of the human PTEN genomic locus (GenBank accession number AF067844). The exons are shown as black boxes 1-9. Five pairs of primers (arrowheads) were used to identify the specific IRIS-binding region by ChIP-qPCR analysis. (Bottom line) Enlargement of the region directly upstream of the first intron. Open ovals show the binding sites for transcription activators (red) and repressors (blue). Residue positions are numbered relative to the transcription start site (+1). Blue arrows indicate the specific IRIS-binding region in the PTEN promoter. (D) PTEN depletion enabled IRIS-depleted M231 cells to grow in soft agar. (Scale bar, 200 µm.) Neg, negative control hairpin. (E) Representative transmembrane cell staining of M231 cells expressing indicated hairpins. (Scale bar, 400 µm.) (F) PTEN depletion restored the mesenchymal morphology of IRIS-depleted M231 cells. (Scale bar, 40 µm.)

12

Supplementary Tables Table S1. Cell growth conditions. Cell line Growth medium 293T DMEM (Mediatech) + 10% FBS (Gemini) 786-O RPMI 1640 (Mediatech) + 10% FBS A375 DMEM + 10% FBS A549 RPMI 1640 + 10% FBS ACHN DMEM + 10% FBS BT549 RPMI 1640 + 10% FBS Caki2 McCoy's 5A (Mediatech) + 10% FBS Capan2 McCoy's 5A + 10% FBS DU145 RPMI 1640 + 10% FBS G361 DMEM + 10% FBS HME MEGM (Lonza) Hs578t DMEM+10% FBS + 0.01 mg/ml Insulin (Sigma) IMR32 DMEM + 10% FBS Jurkat RPMI 1640 + 10% FBS MCF7 DMEM + 10% FBS MDA-MB-231 DMEM + 10% FBS MDA-MB-468 RPMI 1640 + 10% FBS NCI-H596 RPMI 1640 + 10% FBS PANC1 DMEM + 10% FBS PC3 RPMI 1640 + 10% FBS SU-DHL-5 RPMI 1640 + 10% FBS SU-DHL-8 RPMI 1640 + 10% FBS T98G DMEM + 10% FBS U2OS DMEM + 10% FBS

13

Table S2. Target Sequences for RNAi. siRNA IRIS1 agaagtgagctaaatgttt IRIS2 tgtgtttgccccagtctat scramble LacZ HIF1A1 HIF1A2 PTEN

shRNA tttgtgtttgccccagtctat aattactggtggacttacttc cctaaggttaagtcgccctcg cgctaaatactggcaggcgtt tgctctttgtggttggatcta ccagttatgattgtgaagtta acttgaaggcgtatacagga

14

Table S3. Primer Sequences Used in qRT-PCR. Forward primer PTEN agaaagacttgaaggcgtatacaggaaca RPLP0 atcaacgggtacaaacgagtcctg

Reverse primer gataagttctagctgtggtgggttatggt aaggcagatggatcagccaagaag

15

Table S4. Primer Sequences Used in ChIP. Forward primer VEGFA promoter tggcctcagttccctggcaacatctg SNAI1 promoter gataagggaaggaacgggtgctcttg SNAI2 promoter tctcaagagcctaagagcagagcttgt PTEN promoter aagtagttccgactgtggcccgtgtat Non-specific region aaggcagatggatcagccaagaag

Reverse primer agtgactgggagggaagaggacctgttgga cattgacgagggaaacgcacatcact gtcctaggttggtgagaaattggtgct gagtggaaagtacggaacggtaggaa atcaacgggtacaaacgagtcctg

16

Supplementary Datasets Dataset S1. ID numbers and expression data associated with patient cases in the TCGA datasets. Dataset S2. P values for IRIS expression in TCGA tumors. Dataset S3. Correlation tests between p220 and IRIS abundance in each cancer type in TCGA. Dataset S4. ID numbers and expression data associated with tumors in the TCGA breast cancer dataset.

17

Supplementary References 1. Minn AJ, et al. (2005) Genes that mediate breast cancer metastasis to lung. Nature 436:518-524. 2. Powell E, et al. (2016) p53 deficiency linked to B cell translocation gene 2 (BTG2) loss enhances metastatic potential by promoting tumor growth in primary and metastatic sites in patient-derived xenograft (PDX) models of triple-negative breast cancer. Breast Cancer Res 18:13. 3. Wang Y, et al. (2015) CDK7-dependent transcriptional addiction in triple-negative breast cancer. Cell 163:174-186. 4. Wee S, et al. (2008) PTEN-deficient cancers depend on PIK3CB. Proc Natl Acad Sci USA 105:13057-13062. 5. Li AG, et al. (2007) An acetylation switch in p53 mediates holo-TFIID recruitment. Mol Cell 28:408-421. 6. Yang X, et al. (2001) A public genome-scale lentiviral expression library of human ORFs. Nat Methods 8:659-661. 7. Sinkevicius KW, et al. (2014) Neurotrophin receptor TrkB promotes lung adenocarcinoma metastasis. Proc Natl Acad Sci USA 111:10299-10304. 8. Johnson WE, Li C, Rabinovic A (2007) Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8:118-127. 9. Everaert C, et al. (2017) Benchmarking of RNA-sequencing analysis workflows using whole-transcriptome RT-qPCR expression data. Sci Rep 7:1559. 10. Robinson MD, Smyth GK (2008) Small-sample estimation of negative binomial dispersion, with applications to SAGE data. Biostatistics 9:321-332. 11. Liao Y, Smyth GK, Shi W (2014) featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30:923930. 12. Anders S, Huber W (2010) Differential expression analysis for sequence count data. Genome Biol 11:R106.

18